Proposal for Pharmacologically Distinct Conformers of PDE4 Cyclic AMP Phosphodiesterases

Cell. Signal. Vol. 9, No. 3/4, pp. 227–236, 1997 Copyright  1997 Elsevier Science Inc.

ISSN 0898-6568/97 $17.00 PII S0898-6568(96)00173-8

TOPICAL REVIEW

Proposal for Pharmacologically Distinct Conformers of PDE4 Cyclic AMP Phosphodiesterases John E. Souness* and Sudha Rao Rhoˆne Poulenc Rorer Ltd., Dagenham Research Centre, Rainham Road South, Dagenham, Essex RM10 7XS

ABSTRACT. cAMP-specific phosphodiesterase inhibitors display a range of activities in vitro and in vivo which suggest they may be useful in the treatment of inflammatory diseases. However, these compounds elicit a number of side-effects which may limit their therapeutic potential. Certain side-effects of PDE4 inhibitors such as emesis and gastric acid secretion are associated with their actions at a high-affinity rolipram binding site (HARBS). In contrast, a number of anti-inflammatory actions of PDE4 inhibitors are better correlated with inhibition of PDE4 catalytic activity than with displacement of [3H] rolipram from HARBS. This suggests that native PDE4s in different cell-types can be discriminated pharmacologically. Although known to be associated with PDE4, the nature of HARBS is uncertain. The majority of evidence suggests it represents particular conformational states of PDE subtypes with which rolipram interacts with high potency (KD z2 nM) (High-affinity PDE4, HPDE4). Rolipram is generally moderately or weakly active (IC50-200 nM-2000 nM) in inhibiting catalytic activity of the majority of crude, partially-purified or recombinant PDE4 preparations (Low-affinity PDE4, LPDE4). Solubilization or V/GSH treatment of particulate eosinophil PDE4, cAMP-dependent kinase activation of RNPDE4D3 and membrane association of HSPDE4A4 increase the potencies of some (e.g., rolipram) but not other (e.g., trequinsin) inhibitors. In eosinophils, the changes in enzyme properties brought about by solubilization result in a close correlation between the potency order of compounds in inhibiting cAMP hydrolysis and displacing [3H] rolipram from HARBS. The identification of distinct pharmacological PDE4 forms may have therapeutic consequences since it may be possible to synthesize potent inhibitors of LPDE4 with low affinity for HARBS which should, theoretically, be less emetic. Most inhibitors synthesized to date (rolipram, denbufylline nitraquazone, etc.) display high-affinity for HARBS but are much weaker in inhibiting cAMP hydrolysis. Other compounds (RP 73401, trequinsin, CDP 840) display slightly higher potency against LPDE4 or do not discriminate between the two putative PDE4 forms. Recently, inhibitors have been synthesized which are considerably more active against LPDE4 than HPDE4. Such compounds with appropriate pharmacokinetic properties may retain anti-inflammatory activity but have a reduced capacity to cause nausea and emesis and, consequently, have a wider therapeutic window than compounds currently undergoing clinical evaluation. cell signal 9;3/4:227–236, 1997.  1997 Elsevier Science Inc. KEY WORDS. RP 73401, Rolipram, cAMP, Phosphodiesterase, Conformers

INTRODUCTION The therapeutic potential of cyclic AMP-specific phosphodiesterase (PDE4) inhibitors has attracted considerable attention, primarily because of their dampening effects on the functions of several inflammatory/immunocompetent cells [1]. Most interest to date has focused on the potential anti* Author for correspondence. Abbreviations: PDE4–cAMP-specific phosphodiesterase; HARBS, highaffinity rolipram binding site; HPDE4, high-affinity PDE4; LPDE4, lowaffinity PDE4; V/GSH, vanadyl/glutathione complex; MBP, major basic protein; ECP, eosinophil cationic protein; EDN, eosinophil-derived neurotoxin; .O 22, superoxide anion; PGE 2, prostaglandin E2; LTB4 , leukotriene B4 ; TNFa, tumour necrosis factor alpha; IL-2, interleukin-2; LPS, lipopolysaccharide; IBMX, 3-isobutyl-1-methylxanthine; HSPDE4, Homo sapiens PDE4; RNPDE4; Rattus norvegicus PDE4; Staphylococcal aureus enterotoxin A, Staph. A Received 9 July 1996; and accepted 12 September 1996.

asthma effects of these compounds but, more recently, the potent suppression of tumour necrosis factor-alpha (TNFa) production from mononuclear phagocytes by PDE4 inhibitors [2, 3] has opened the possibility of treating the wide-number of pathological conditions associated with over-elaboration of this pro-inflammatory cytokine such as auto-immune diseases (arthritis, multiple sclerosis, etc.) [4, 5], certain viral diseases (eg. AIDS) [6] and bacterial or parasitic infections (eg. septic shock, cerebral ischaemia) [5, 7]. Indeed, PDE4 inhibitors ameliorate disease progression in animal models of arthritis [8], multiple sclerosis (experimental allergic encephalomyelitis) [9, 10] and septic shock [11]. Although the therapeutic potential of PDE4 inhibitors in chronic inflammatory disorders seems considerable, the utility of these compounds may be limited by side-effects.

228

J. E. Souness and S. Rao TABLE 1. Potencies of PDE4 inhibitors against monocyte PDE4 and in competing for HARBS in the microsomal and cytosolic fractions from guinea-pig brain.

Compound R-(2)-rolipram (6)-Rolipram WAY PDA 641 Ro 20-1724 Ibudilast S-(1)-rolipram Denbufylline Compound A IBMX RP 73401 CDP 840 [30] Trequinsin

Displacement of [3H] rolipram binding (Kiannann: mM)

PDE4 inhibition (IC50: mM)

Microsomes**

Cytosol

PDE4/HARBS*

0.29 0.31 0.30 2.4 1.3 0.75 0.20 0.02 14 0.0012 0.007 0.4

0.0009 0.0017 0.0015 0.017 0.010 0.013 0.0041 0.00046 0.84 0.0004 0.026 1.7

0.0006 0.0013 0.0014 0.017 0.0071 0.0063 0.0023 ND 0.43 0.0004 0.055 1.3

322 238 200 141 130 58 49 43 17 3.0 0.27 0.23

cAMP PDE activity was measured in the cytosolic fraction of human monocytes as described in [34]. [3 H]-(6)-rolipram binding was measured in the microsomal and cytosolic fractions of guinea-pig brain [34]. The results represent the means of 2–4 experiments. Compound A is a Syntex nitraquazone (Figure 2) analogue-1-(3-Nitrophenyl)-3-(4-pyridyl-1methyl)pyrido[2,3-d]pyrimidine-2,4(1H,3H)-dione (patentWO 93/19068). It is noteworthy that, based on inhibitor displacement of [3 H]-(6)-rolipram, cytosolic and membrane-bound HARBS are almost identical. * PRE4/HARBS ratio based on microsome rolipram binding data.

Nausea and vomiting have been reported following administration of several PDE4 inhibitors to human subjects [1]. The mechanism(s) by which PDE4 inhibitors induce these side-effects is/are uncertain but studies demonstrating the potentiation of apomorphine-induced emesis in dogs [12] by Ro 20-1724 suggest that the nausea and vomiting are likely to be produced, at least in part, via the emesis centres in the brain. Local effects may also contribute to the gastro-intestinal side-effects observed. For example, rolipram (Figure 1) is a very potent stimulators of acid secretion [13] from gastric parietal cells which may, by producing local irritation, exacerbate gastro-intestinal disturbances. The obvious therapeutic goal is to optimise on the antiinflammatory actions of PDE4 inhibitors while formulating strategies to limit their side-effects. The identification of PDE4 sub-type-selective inhibitors offers a rational though uncertain way forward. In the near future, such compounds are likely to be identified and reports on their in vivo actions are awaited with interest. Recently, evidence for the existence of distinct, non-interconvertable, conformational states of PDE4 in different cell-types emerged [14–16]. The proposal was based on pharmacological data on compounds exhibiting different relative potencies against human (HS) recombinant PDE4A and in displacing [3H] (6)rolipram from its high-affinity receptor in brain cytosol [14–16]. The former activity was suggested to represent a low-affinity binding site on PDE4 (LPDE4) and the latter a high-affinity binding site (HPDE4). In a range of studies with compounds which show great selectivity for high-affinity PDE4 or which exhibit slightly greater potency on low-affinity PDE4 or do not discriminate between the two sites (Table 1), it was concluded that suppression of certain inflammatory cell functions is associated, by and large, with actions on LPDE4, whereas the functional responses in other cells

(eg. parietal cells, neutrophils, airways smooth muscle) are more closely linked with actions on HPDE4 [14–17]. The emetic activity of PDE4 inhibitors has also been linked to HPDE4 [16, 18]. Thus, it was proposed, if compounds could be identified which potently inhibit LPDE4 but only weakly interact with HPDE4, they may exhibit an improved therapeutic window compared to the present generation of compounds. The aim of the current review is to evaluate the evidence for the so-called high- and low-affinity forms of PDE4 and assess the possible molecular basis for their existence. Before considering the evidence for distinct conformational states of PDE4, it is pertinent to summarize briefly the diverse properties of different PDE4 preparations and, in particular, the inconsistencies in inhibitor interactions. DIVERGENT PROPERTIES OF PDE4 FROM DIFFERENT CELLS AND TISSUES Properties of PDE4 PDE4 is considered to be a relatively poorly defined enzyme due to its being present in only trace amounts within cells, its apparent localization in different subcellular compartments and its disposition to proteolysis during purification procedures. The properties of PDE4 vary depending upon the cell/tissue source and the method of preparation. For example, partially-purified PDE4 from cardiac [19] and smooth muscle [20] are cytosolic (apparently) and display linear (Michaelis–Menton) kinetics, whereas eosinophil PDE4 is tightly membrane-bound and displays complex kinetics [21]. Although artifacts produced during enzyme isolation may contribute to the diverse properties in different preparations, the identification of four distinct genes encoding PDE4 subtypes in rat (RNPDE4A-D) and man

PDE4 Cyclic AMP Phosphodiesterases

229

FIGURE 1. Structures of RP 73401 and rolipram.

(HSPDE4A-D) offer a molecular basis for the diverse properties of PDE4. Furthermore, multiple splice variants of individual PDE4 subtypes have been identified which differ in their membrane association properties (eg RNPDE4A/ RD1) [22] and susceptibility to phosphorylation by cAMPdependent protein kinase (PKA) (eg RNPDE4D) [23].

FIGURE 2. Structures of quinazolines and trequinsin.

Variable Potencies of Rolipram Rolipram, a compound which has assumed a position of great importance in the study of PDE4, was discovered 20 years ago [24]. Several studies on partially-purified PDE4 preparations from smooth muscle and heart indicated that this archetypal inhibitor is a competitive, non-stereo-selective and relatively weak inhibitor (low mM) [19, 20, 25]; however, in contrast to this early work, experiments on a number of PDE4 preparations have demonstrated IC50 values of rolipram in the sub mM range [25–28]. Indeed, rolipram displays a wide-range of inhibitory potencies against PDE4 preparations from a variety of sources (Table 2). This does not appear to be due to a diversity of PDE4 subtypes being present in the different PDE4 preparations since the limited information published to date suggest that, with the exception of PDE4C, which has a limited tissue distribution [29] and against which it is approximately 10-fold less potent, rolipram does not discriminate between subtypes [30, 31]. It should be pointed out that the systems used to express a particular PDE4 sub-type can influence the potency of rolipram. For example, it is at least 10-fold more potent against HSPDE4A expressed in yeast (IC50 , 0.1 mM) compared to COS cells (IC50: z1 mM) [32, 33]. Although the

reason(s) for these differences are uncertain, it is possible that post-translational processing (folding) of PDE4 differs between the different expression systems and this might influence the inhibitory potency of rolipram. In marked contrast to rolipram, RP 73401 (piclamilast, Fig. 1), a recently discovered and very potent inhibitor [27], displays similar IC50 values against PDE4 isolated from a variety of cells and tissues either in crude sub-cellular fractions or following partial purification (Table 2). Hence, possible changes to the properties of PDE4 caused by cell/tissue disruption and preparative procedure, which appear to have a major impact on the inhibitory activity of rolipram, have little influence on the potency of RP 73401, suggesting perhaps that the two compounds interact differently with PDE4. Relationship Between Functional Effects of Inhibitors and PDE4 Inhibition In some [21, 34], but not all [35, 36] cells, poor correlations have been noted between the IC50 values of inhibitors against PDE4 and their potencies in eliciting functional responses. This is exemplified in airways smooth muscle in

TABLE 2. Inhibitory potencies of RP 73401 and rolipram against PDE4 from different cells and tissues.

Source of PDE4 Pig aortic PDE4 Bovine tracheal PDE4 Eosinophil PDE4 No treatment Solubilization V/GSH Frozen* Macrophage PDE41 No treatment Human monocytes PDE41 RBL cells1

Nature of enzyme preparation

IC50 (nM) RP73401

rolipram

1.0 1.1

1470 1967

Particulate

2.0 1.9 2.6 1.0 1.0

230 14 20 685 205

Cytosolic Cytosolic

1.5 1.3

313 300

Partially-purified Partially-purified Tightly membrane-bound

* Particulate enzyme prepared from cells stored at 2808C. Data are from [27, 35] and 1 unpublished observations.

230

J. E. Souness and S. Rao

TABLE 3. Potency differences between RP 73401 and rolipram in intact cells are related to whether the functional responses

to PDE4 inhibitors are better linearly correlated to low-affinity PDE4 or HARBS. IC50 /EC50 (nM) Functional response Guinea-pig trachea [27] Relaxation of methacholine-induced contraction Guinea-pig eosinophils [27] Inhibition of LTB4-induced MBP release Enhancement of isoprenaline-induced cAMP accumulation Human monocytes [35] Inhibition of LPS-induced TNFa release Enhancement of PGE2-induced cAMP accumulation Murine splenocytes1 Inhibition of Staphylococcal enterotoxin A-induced IL-2 release

RP 73401

Correlation coefficient (r)

rolipram

Fold-difference

FR vs PDE4

FR vs HARBS

34

99

3.0

0.23*

0.97*

115

602

5.2

ND

ND

93

297

3.2

0.79

0.98

6.9

490

50

0.93

0.65

9.2

503

71

0.95

0.67

0.46

540

1173

0.95**

0.39

1 Souness J. E., Sardar N. and Withnall M. T. (unpublished). * Data from [34]. ** Effects of PDE inhibitors on IL-2 release correlated with their inhibition of low-affinity PDE4 from CTLL cells. FR, functional response; ND, not determined.

which RP 73401 and rolipram display similar potencies in relaxing tissue toned with contractile agonist, even though the former is .1000-fold more potent than the latter against partially-purified PDE4 isolated from this tissue [27]. Therefore, PDE4 activity as isolated from some cells and tissues does not always faithfully represent the native form(s) of the enzyme. Proteolysis or loss of a post-translational modification (eg dephosphorylation) may offer possible explanations for such anomolies. PHARMACOLOGICAL EVIDENCE FOR THE EXISTENCE OF MULTIPLE, NON-INTERCONVERTABLE PDE4 FORMS Association Between Functional Responses and Competition for a High-Affinity Rolipram Binding Site (HARBS) The inhibitory potency (IC50) of rolipram against PDE4 usually lies in the range 0.1 mM–.2.0 mM. This is the case for PDE4 measured in crude subcellular fractions [35], partially purified preparations [19, 20], or recombinant protein expressed in mammalian cells [32], E. coli [37], yeast [33] and insect cells [38]. However, studies reported approximately 10 years ago demonstrated high-affinity (KD z 2 nM) rolipram binding in brain (membranes and cytosolic fraction) [39]. In vitro and in vivo studies have shown marked rolipram stereoselectivity at this site with the R-(2)- enantiomer being 15–20 fold more potent than the S-(1)- enantiomer [39, 40]. This contrasts with the slight stereoselectivity observed with rolipram on cAMP hydrolysis in several PDE4 preparations [25, 33, 35]. Close correlations have been reported between the central (CNS) actions (antagonism of reserpine-induced hypothermia in mice, induction of head twitches in rats, etc.) of PDE4 inhibitors and their potencies in displacing [3H] rolipram from brain membranes [40–43].

Certain peripheral actions of PDE inhibitors also appear to be linked to an interaction with the high-affinity rolipram binding site: In guinea-pig trachea, although a poor correlation exists between PDE4 inhibition and functional antagonism of histamine-induced contraction, a very strong correlation between inhibitor relaxation and displacement of [3H] rolipram from a high-affinity binding site in brain has been documented [34]. As in in vitro studies, no correlation exists between inhibition of PDE4 activity and suppression of histamine-induced bronchoconstriction in anaesthetised guinea-pigs. In contrast, an excellent relationship is obtained when suppression of histamine-induced bronchoconstriction is correlated with PDE4 inhibitor displacement of [3H] rolipram from brain membranes [34] (Table 3). A very close correlation also exists between the potencies of compounds in potentiating isoprenaline-induced cAMP accumulation in guinea-pig eosinophils and in displacing [3H] rolipram from brain membranes [25, Fig. 3]. The discrepancy between the airways smooth muscle relaxant effects of RP 73401 and rolipram and their PDE4 inhibitory potencies was highlighted above. The 3-4-fold difference in potency between these two compounds in airways smooth muscle (Fig. 4) and eosinophils [27], which contrasts so markedly with the PDE4 inhibition data, is in very good agreement with their competition for HARBS [27]. Close associations have also been reported between the rank order potencies of compounds in stimulating acid secretion in rabbit parietal cells [15] and inhibiting human neutrophil degranulation with competition for HARBS [36]. Functional Effects of PDE4 Inhibitors Are Not Always Associated with HARBS Not all functional responses to PDE inhibitors are closely associated with HARBS. Indeed, in human monocytes, PDE inhibitor suppression of lipopolysaccharide (LPS)-

PDE4 Cyclic AMP Phosphodiesterases

231

FIGURE 3. PDE4 inhibitor-induced cAMP accumulation in

guinea-pig eosinophils or human monocytes as functions of PDE4 inhibition or displacement of [3H] rolipram binding to guinea-pig brain membranes. Rolipram binding displacement data are expressed as apparent Ki values (Kiapp), cAMP data as EC50 values and PDE4 inhibition data as IC50 values. Regression analysis demonstrates that potentiation of PGE2-induced cAMP accumulation in human monocytes is better correlated with inhibition of monocyte cytosolic PDE4 (r 5 0.95) (Panel A) than displacement of [3H] rolipram to HARBS (r 5 0.67) (Panel B). In contrast, a poorer correlation is observed between potentiation of isoprenaline (1 mM)-induced cAMP accumulation in guinea-pig eosinophils and inhibition of membrane-bound eosinophil PDE4 (r 5 0.79) (Panel C) than competition for HARBS (r 5 0.96) (Panel D). Compounds: 1. RP 73401; 2. R-(2)-rolipram; 3. (6)-rolipram; 4. denbufylline; 5. ibudilast; 6. S-(1)-rolipram; 7. Ro 20-1724; 8. IBMX; 9. trequinsin; 10. compound A; 11. compound B; 12. compound C; 13. compound D; 14. AH-21-132 (benafentrine); 15. dipyridamole. For chemical structures of compounds A–B see ref 35.

induced TNFa release and potentiation of PGE2-induced cAMP accumulation are poorly correlated with competition for HARBS and a much better relationship is observed with inhibition of monocyte PDE4 [35, 36] (Table 3, Fig. 3). In murine splenocytes, the relationship between suppression of Staphylococcus aureus enterotoxin A (Staph. A)induced IL-2 production and competition for HARBS falls away completely, whereas a significant correlation exists with PDE4 inhibition (Table 3). A much greater (50-fold) potency difference is observed between RP 73401 and rolipram on monocyte cAMP accumulation and suppression of TNFa release [35] than is observed in eosinophils and airways smooth muscle and, in murine splenocytes, RP 73401 is approximately 1170-fold more potent than rolipram in inhibiting Staph. A-induced IL-2 generation (Fig. 4b). Stereoselectivity of Rolipram Actions in Intact Cells The information presented above would point to the presence of pharmacologically distinct PDE4 forms existing in

FIGURE 4. Inhibition of methacholine-induced contraction of

guinea-pig trachealis (panel A) and Staph A.-induced IL-2 release from murine splenocytes (panel B) by RP 73401 (d) and rolipram (s). Only a 3-fold potency difference is observed in relaxation of airways smooth muscle [data from 34], but a greater than 1100-fold potency difference is observed in suppression of IL-2 release from splenocytes.

different cell-types. However, the issue is clouded somewhat when the stereoselectivity of rolipram on intact cell functional responses is investigated. As alluded to previously, only slight rolipram selectivity is observed against several PDE4 preparations [33, 35, 36], whereas the R-(2)-enantiomer has approximately 15–20-fold greater affinity for HARBS than the S-(1)-enantiomer [39]. It would be predicted that a greater enantiomeric potency difference of rolipram would be observed in cells where pharmacological effects of PDE4 inhibitors are better correlated with competition for HARBS than inhibition of cAMP hydrolysis. This is indeed the case for the CNS actions of rolipram where 15–20-fold stereoselectivity is observed [40–43]. However, this is not always the rule in peripheral cells. For example, the stereoselectivity of rolipram in eosinophils

232

J. E. Souness and S. Rao

and parietal cells, where pharmacological effects of PDE4 inhibitors are closely associated with HARBS [15, 25], is similar to that observed in human monocytes and murine splenocytes, where functional responses to PDE4 inhibitors are more closely related to inhibition of cAMP hydrolysis [25, 36, unpublished observations]. Surprisingly, in human neutrophils in which PDE4 inhibitor suppression of degranulation is better associated with HARBS, only 3-fold rolipram stereoselectivity is observed [36]. Thus it would appear that the magnitude of rolipram stereoselectivity is not predictive of whether the pharmacological effects of PDE4 inhibitors are better associated with competition for HARBS or inhibition of cAMP hydrolysis.

coordinating Zn 21 [48]. Substitution of His505 and His506 residues with Asn on HSPDE4A abolishes catalysis but does not affect high-affinity rolipram binding [46]. Substitutions of other conserved histidines around the catalytic site result in losses in cAMP PDE and rolipram binding activities [45]. Rolipram binding in brain is highly dependent on divalent cations such as Mg21 or Mn21 [39] and depletion of divalent cations with EDTA greatly reduces the affinity of rolipram [39]. This demonstrates that treatments which decrease PDE catalytic activity also compromise binding of rolipram. Although substantial levels of rolipram binding (Bmaxz 20 pmol/g for cytosolic and particulate fractions) are measured in brain [39], specific binding of rolipram to peripheral cells or tissues is low [15, 49] or not detectable [35, 49].

DOES HARBS REPRESENT ONE OF MULTIPLE CONFORMERS OF PDE4? HARBS is Associated with PDE4

Studies on Eosinophil PDE4

It is now clear that HARBS is located on PDE4. High-affinity rolipram binding, whose characteristics are similar to those in brain, has been measured on recombinant HSPDE4A and HSPDE4B [33, 44]. The characteristics of binding differ slightly on the two PDE4 subtypes. Only a single high-affinity (Kdz2 nM), non-interacting (nH 5 1) binding site for rolipram was demonstrated on HSPDE4A, although a low-affinity binding site (Kd 5 40 nM) was detected; however, similar studies performed on HSPDE4B revealed two non-interacting high-affinity rolipram-binding sites (Kds 5 0.4 and 6 nM) [44]. Studies on recombinant PDE4 suggest that the rolipram binding site and catalytic site are distinct entities [33]. Firstly, the affinity of rolipram for its binding site on HSPDE4A is approximately 100-fold greater than for inhibition of catalytic activity [33]. Furthermore, the rank order of potency of structurally dissimilar compounds in displacing [3H] rolipram does not correlate with their inhibition of cyclic AMP hydrolysis [33]. The nature of HARBS remains uncertain, although several pieces of evidence point to it representing a distinct conformer of PDE4 rather than an allosteric site. The stoichiometry of high-affinity rolipram binding on HSPDE4A is considerably lower than the one mole of rolipram per mole of enzyme predicted for a distinct, ‘allosteric site’ [45]. cAMP (IC50: 11 mM) inhibits high-affinity rolipram binding to HSPDE4A indicating that the binding site is at or close to the catalytic site [45]. The importance of the quaternary structure of PDE4 on high-affinity rolipram binding is unclear. In contrast to full-length HSPDE4A, truncated HSPDE4A (Met 322-886) does not bind rolipram with high affinity although full catalytic activity and low-affinity rolipram binding are retained [46]. A longer HSPDE4A truncate (Met 265-886) exhibits catalytic and rolipram binding activities similar to the full-length enzyme. The cAMP PDE inhibitory potency of rolipram against the Met 265-886 truncate is only slightly lower than against the longer constructs [46]. Histidine residues are critical for cAMP PDE catalytic activity of HSPDE4A and RNPDE4D1 [46, 47], possibly by

cAMP PDE in guinea-pig eosinophils is tightly membranebound, exhibits complex kinetics and is inhibited by rolipram (IC50z200 nM) [21]. Solubilization with deoxycholate and salt as well as treatment of membranes with vanadyl/glutathione complex (V/GSH) markedly alter the kinetic properties of the enzyme: the Vmax for cAMP hydrolysis is greatly increased while, paradoxically, the affinity of cAMP for PDE4 is reduced [26]. These treatments increase the potency of (6)-rolipram by greater than 10fold, the IC50 values against the solubilized and V/GSHtreated enzymes being 14 nM and 20 nM, respectively [26]. This potency of rolipram against eosinophil PDE4 approaches that observed for HARBS. The IC50s of several other PDE4 inhibitors, namely Ro 20-1724, denbufylline and ibudilast are also increased by these treatments [26, 50]. However, the inhibitory effects of compounds such as RP 73401, trequinsin and dipyridamole are unaffected [26, 27]. The stereoselectivity of rolipram is also increased by solubilization and V/GSH [25]. In untreated membranes, the R-(2) enantiomer is only approximately 3-fold more potent than the S-(1) enantiomer, but solubilization of eosinophil PDE4 or treatment of membranes with V/GSH increases the enantiomeric selectivity markedly (15–20-fold). This level of stereoselectivity is similar to that observed on the high-affinity rolipram binding site in brain [39]. Indeed, a strong correlation exists for displacement of [3H] rolipram from brain membranes by a range of PDE inhibitors and inhibition of solubilised eosinophil PDE4 [25], tempting speculation that HARBS represents a particular conformation of PDE4 that can be induced by solubilization or treatment with V/ GSH. Similar data have recently been reported for the PDE4 isolated from guinea-pig peritoneal macrophages [49]. To explain these results, two similar, though slightly different explanations can be offered: In the first, two-site model [51, 52], solubilization and V/GSH is proposed to convert PDE4 from a form(s) against which rolipram interacts with low affinity to one against which it interacts with high affinity [26, 51]. The influence that one inhibitor binding site, designated Sc, exerts on cAMP hydrolysis is unaffected by the conformation of PDE4, whereas that of

PDE4 Cyclic AMP Phosphodiesterases

the other, Sr, is dependent on the conformational state [51, 52]. Thus, the potencies of compounds postulated to bind with high affinity to Sr and low affinity to Sc (rolipram, denbufylline, Ro 20-1724, ibudilast) are altered by treatments which change the conformation of the enzyme, whereas those with higher affinity for Sc or which do not discriminate between the two sites (RP 73401, trequinsin, dipyridamole) are unaffected by such treatments. In a second, one-site model, the altered potencies of rolipram and similar compounds are suggested to simply reflect their different affinities for distinct conformational states of PDE4 [51, 52]. In this alternative paradigm, membrane-bound PDE4 exists in eosinophil membranes in two (or more?) conformational states, offering a possible explanation for the complex kinetics observed [21]. The rolipram group of inhibitors acting competitively at or near the catalytic site, would have higher affinity for one conformational state than the other, whereas RP 73401 and similar compounds would not discriminate between the different forms. Solubilization and V/GSH are proposed to increase the proportion of PDE4 existing in the ‘high-affinity’ state, so increasing the potencies of the rolipram group of compounds. Since the conformation of PDE4 would not influence the affinities of RP 73401 and similar compounds, their inhibitory potencies would not be affected by solubilization or V/GSH.

PKA-Dependent Activation of RNPDE4D3 As alluded to above, four different rat and human genes encoding PDE4 subtypes, with highly conserved central sequences corresponding to the catalytic domain but with less homologous N- and C-termini, have been identified [30, 53, 54]. Northern blotting and reverse-transcription polymerase chain reaction (RT-PCR) studies have demonstrated that transcripts of the four PDE4 variants of different sizes are differentially expressed between tissues [55, 56]. The molecular size of purified PDE4, as revealed by separation on polyacrylamide gel electrophoresis, varies greatly [56, 57]. Alternative splicing of newly transcribed nuclear RNA is responsible, at least in part, for producing the heterogeneity in the sizes of PDE4 subtypes [23]. Several PDE4D splice variants with divergent N-termini have been detected in mammalian cells [23]. A long form of PDE4D (RNPDE4D3) with an extended N-terminus contains concensus sequences for phosphorylation by cAMP-dependent protein kinase (PKA) [23, 57] (Table 4). Phosphorylation of RNPDE4D3 by PKA results in activation which is important in attenuating signalling by stimuli of adenylate cyclase in thyroid FRTL-5 cells [57]. Interestingly, the potencies of nitroquazone analogues, RS-25344 and RS-33793 (Figure 2), against the activated PDE4D3 are 60-300-fold greater than against enzyme in its basal state [28]; however, the potency of trequinsin (Figure 2) is unaffected by PKA-induced phosphorylation and activation [28]. Whether PDE4D3 is the only subtype splice variant which is susceptible to this form of regulation is uncertain. Other PDE4 subtype splice variants possess XRRXS consensus motifs towards their

233 TABLE 4. PDE4 clones containing consensus motifs

(XRRXS) for phosphorylation by PKA. Transcript name HSPDE4A4 HSPDE4A5 RNPDE4A5 HSPDE4B1 HSPDE4C1 HSPDE4D3 RNPDE4D3B HSPDE4D4

Consensus motif(s)

Position

Reference

QRRES QRRES QRRES QRRES QRRES FRRHS QRRES FRRHS QRRES QRRES

30–34 136–140 141–145 129–133 115–119 9–13 50–54 9–13 50–54 66–70

32 30 65 44 29 30, 37 23 30

Identification of PKA consensus motifs [66] by Lasergene (DNA Star Inc, Madison, USA).

N-termini (Table 4), but no reports on PKA-induced activation or effects on inhibitor potencies have, to the author‘s knowledge, been published. These results bear marked similarities to the effects of solubilization and V/GSH on the potencies of inhibitors against the eosinophil PDE4. V/GSH was originally shown to be a stimulator of adipocyte PDE3, an enzyme known to be activated by phosphorylation [58]. Although the mechanism by which V/GSH activates PDE4 is unclear, it is tempting to speculate that it mimics the effects of phosphorylation. Eosinophils express predominantly if not exclusively PDE4D [27] and, although it is uncertain whether this represents a long-form splice-variant, it may also be regulated by phosphorylation in a manner similar to RNPDE4D3 in thyroid FRTL-5 cells. The compounds (rolipram, denbufylline, Ro 20-1724, ibudilast) where increased potency is observed when eosinophil PDE4 is solubilized or exposed to V/GSH [26, 50] are generally those which exhibit much greater potencies in displacing [3H] rolipram from brain membranes (HPDE4) than against cAMP PDE catalytic activity (LPDE4), whereas the compounds that are unaffected are those that display similar activities against the two sites (Table 1).

FIGURE 5. Structures of recently identified PDE4 inhibitors

with reduced affinities for HARBS.

234

This may also be the case for PKA-induced phosphorylation and activation of recombinant RNPDE4D3. Quinazoline diones such as nitraquazone and its analogues generally show much greater affinities for HARBS than LPDE4 [59] (Table 1). It is therefore consistent with the eosinophil data that the potencies of nitraquazone analogues, RS 25344 and RS 33793, are increased by PKA treatment, whereas that of trequinsin, which is slightly more potent against LPDE4, is not affected by exposing RNPDE4D3 to PKA [28]. HSPDE4B2B expressed in yeast or SF9 insect cells is phosphorylated , predominantly on Ser487 and Ser489 [60]. The amino acid sequence around these serine residues conforms to the consensus motif (PXSP) for mitogen-activated protein kinase (MAP kinase) [60]. Non-phosphorylated HSPDE4B2B (Met 151-528) expressed in E-coli is phosphorylated on Ser487 by MAP kinase but this has little impact on activity or inhibition by rolipram [60]. Influence of Subcellular Localization of PDE4A4B on Rolipram Potency. A HSPDE4A4 construct with an extended N-terminus with no censensus sequence for PKA phosphorylation (HSPDE4A4B) expressed in COS cells is located predominantly (90%) in the particulate fraction [61]. Like the eosinophil enzyme, it is not released from membranes by Triton-X-100 or high NaCl concentrations [61]. Rolipram displays partial competitive inhibition (Ki 5 37 nM, Ki 5 2300 nM) against the membrane-bound enzyme. In contrast, a HSPDE4A4 construct lacking this N-terminal extention (HSPDE4A4C) localizes entirely to the cytosolic fraction of COS cells [61]. This shorter form displays a high Vmax (11-fold higher than HSPDE4A4B) and is inhibited less potently (Ki 5 1600 mM) by rolipram. Thus factors other than phosphorylation can have a major impact on the inhibitor affinity of rolipram. It will be interesting to determine the rolipram binding properties of the two splice variants of HSPDE4A. CONCLUSIONS It goes without saying that rational drug discovery is dependent upon structure activity relationships against isolated enzymes or receptors being translated into whole cell responses and then in vivo. This is not always the case for PDE4 inhibitors where misleading results in the test tube can be caused by artifacts produced during enzyme preparation (eg. proteolysis, loss of post-translational modification etc). It is also important to remember that factors other than intrinsic PDE4 inhibitory activity will influence the absolute potencies of inhibitors in eliciting functional responses in intact cells. For example, whereas the IC50 values of RP 73401 against PDE4 preparations from a variety of sources are very similar, its potencies in eliciting whole-cell responses vary greatly, RP 73401 is 192-fold more potent in inhibiting Staphylococcus aureus enterotoxin-induced IL-2 release from murine spleenocytes than LTB4-stimulated

J. E. Souness and S. Rao

MBP release from guinea-pig eosinophils (Table 3). This may be due to differences in the proportion of total PDE4 activity required in various cell-types to be inhibited for elevation of intracellular cAMP to occur. This, in turn, will be dependent upon the relative intracellular rates of cAMP synthesis and hydrolysis (turnover). Alternatively, the magnitude of the response to the activating signal will also influence the potency of PDE4 inhibitors. Indeed, certain contractile agonists, such as methacholine, can functionally antagonise the relaxant responses to agents that elevate cAMP in airways smooth muscle [62]. Unequal uptake of PDE4 inhibitors may also be a factor to consider when comparing their potencies on the functional responses in different cells and tissues. Unfortunately, little documented information is available to assess its importance. Having taken this into consideration, it appears that, inspite of anomalies (eg. stereoselectivity of rolipram), functional effects of PDE4 inhibitors in a variety of cell-types can be categorized pharmacologically into two main groups, those more closely associated with inhibition of PDE4 catalytic activity and those associated with competition for HARBS. The molecular significance of HARBS is still far from clear. If it only represents a PDE4 conformer against which rolipram displays high affinity, why is it so difficult to measure in peripheral cells and tissues, even where close associations exist between pharmacological effects of PDE4 inhibitors and HARBS. Is it purely a question of the levels of high-affinity conformer PDE4 molecules being much lower in these peripheral tissues compared to brain or, if HARBS represents one of multiple PDE4 comformers, is it better preserved in brain than elsewhere following cell disruption? If so, what factors influence the stability of the high-affinity form(s) of PDE4? This latter hypothesis may explain why it is so difficult to demonstrate rolipram potency in the low nM range. The increased inhibitor potencies caused by solubilization and V/GSH, in the case of eosinophil PDE4, PKA, in the case of RNPDE4D3, and membrane association, in the case of HSPDE4A4, may indicate conversion of a low-affinity form to a high-affinity form. Perhaps, in their native states, these PDE4s do have high affinity for rolipram, which is lost upon cell disruption, possibly due to the loss of a post-translational modification such as phosphorylation (mimicked by V/GSH?). Thus, extending the argument further, HARBS may simply represent the native post-translationally-modified, splice variant of PDE4 which is highly sensitive to the actions of rolipram, so explaining the close association between HARBS and PDE4 inhibitor-elicited cAMP accumulation in eosinophils. Such an hypothesis remains unproven and detailed studies to determine the effects of V/GSH, solubilization and PKA on rolipram binding have not been reported. Although it is prudent not to leap to conclusions regarding the nature of HARBS, it is possible that the pharmaceutical industry has already, inadvertantly, embarked on a mission to identify compounds which discriminate between splice variants of one or more PDE4 subtype. For reasons outlined in the Introduction, intensive effort has been in-

PDE4 Cyclic AMP Phosphodiesterases

vested in trying to identify compounds which potently inhibit LPDE4 but have low affinity for HARBS (HPDE4?). For example, SB 207499 (Figure 5), which is generally equipotent with rolipram on functional responses in several inflammatory cells, is far less effective in stimulating acid secretion in parietal cells [17]. This compound, although 3-fold more potent than rolipram against HSPDE4A, is 28fold less potent in displacing [3H] rolipram from its binding site. The HPDE4/LPDE4 ratio of 1.1 differs considerably from that for rolipram (IC50/IC50: 0.013) [14]. Recently, triaryl ethanes, such as CDP 840 [31], as well as Pfizer oxindoles and benzimidazoles [63, 64] (Figure 5), have been identified which display an even greater degree of selectivity for LPDE4. Whether compounds with such a profile of activities whose pharmacokinetic properties are appropriate for an oral medication will display an acceptable therapeutic window remains to be seen.

References 1. Palfreyman M. N. and Souness J. E. (1996) Prog. Med. Chem. 33, 1–52. 2. Prabhakar U., Lipscutz D., O‘Leary Bartus J., Slivjak M. J., Smith III E. F., Lee J. C. and Esser K. M. (1994) Int. J. Immunopharmac. 16, 805–816. 3. Schade F. U. and Schudt C. (1993) Eur. J. Pharmacol. 230, 9–14. 4. Tracey K. J. and Cerami A. (1993) Annu. Rev. Cell Biol. 9, 317–343. 5. Feuerstein G. Z., Liu T. and Barone F. C. (1994) Cerebrovasc. Brain Metab. Rev. 6, 341–360. 6. Fauci A. S., Schnittman S. M., Poli G., Koenig S. and Pantaleo G. (1991) Ann. Intern. Med. 114, 678–693. 7. Hinshaw L. B., Emerson T. E., Taylor F. B., Chang A. C. K., Duerr M., Peer G. T., Flournoy D. J., White G. L., Kosanske S. D., Murray C. K., Xu R., Pasey R. B. and Fournel M. A. (1992) J. Trauma 33, 568–573. 8. Sekut L., Yarnall D., Stimpson S. A., Noel, L. S., Batemanfite R., Clarke R. L., Brackeen M. F., Menius L. A. and Connolly K. M. (1995) Clin. Exp. Immunology 100, 126–132. 9. Genain C. P., Roberts T., Davis R. L., Nguyen M.-H., Uccelli A., Faulds D., Li Y., Hedgpeth J. and Hauser S. L. (1995) Proc. Natl. Acad. Sci. USA 92, 3601–3605. 10. Sommer N., Loschmann P.-A., Northoff G. H., Weller M., Steinbrecher A., Steinbach J. P., Lichtenfels R., Meyermann R., Riethmuller A., Fontana A., Dichgans J. and Martin R. (1995) Nature Med. 1, 244–248. 11. Badger A. M., Olivera D. L. and Esser K. M. (1994) Circ. Shock 44, 188–195. 12. Carpenter D. O., Briggs D. B. and Knox A. P. (1988) J. Neurophysiol. 59, 358–369. 13. Puurunen J., Lucke C. and Schwabe, U. (1978) NaunynSchmiedeberg’s Pharmacol. 304, 69–75. 14. Barnette M. S., Manning C. D., Cieslinski L. B., Burman M., Christensen S. B. and Torphy T. J. (1995) J. Pharmacol. Exp. Ther. 273, 674–679. 15. Barnette M. S., Grous M., Cieslinski L. B., Burman M., Christensen S. B. and Torphy T. J. (1995) J. Pharmacol. Exp. Ther. 273, 1396–1402. 16. Barnette M. S., Torphy T. J. and Christensen S. B. (1995a) WO Patent, 00139. 17. Barnette M. S., Christensen S. B., Essayan D. M., Esser K. M., Grous M., Huang S.-K., Manning C. D., Prabhaker U., Rush J. and Torphy T. J. (1994) Am. J. Resp. Crit. Care Med. 149, A209 [Abstract].

235 18. Duplantier A. J., Biggers M. S., Chambers R. J., Cheng J. B., Cooper K., Damon D. B., Eggler J. F., Kraus K. G., Marfat A., Masamune H., Pillar J. S., Shirley J. T., Umland J. P. and Watson J. W. (1996) J. Med. Chem. 39, 120–125. 19. Reeves M. L., Leigh B. K. and England P. J. (1987) Biochem. J. 241, 535–541. 20. Torphy T. J. and Cieslinski L. B. (1990) Mol. Pharmacol. 37, 206–214. 21. Souness J. E., Carter C. M., Diocee B. K., Hassall G. A., Wood L. J. and Turner N. C. (1991) Biochem. Pharmacol. 42, 937– 945. 22. Shakur Y., Pryde J. and Houslay M. D. (1993) Biochem. J. 292, 677–686. 23. Sette C., Vicini E. and Conti M. (1994) J. Biol. Chem. 269, 18271–18274. 24. Schwabe U., Miyake M., Ohaga Y. and Daly J. W. (1976) Mol. Pharmacol. 12, 900–910. 25. Souness J. E. and Scott L. C. (1993) Biochem. J. 291, 389– 395. 26. Souness J. E., Maslen C. and Scott L. C. (1992) FEBS Lett. 302, 181–184. 27. Souness J. E., Maslen C., Webber S., Foster M., Raeburn D., Palfreyman M. N., Ashton M. J. and Karlsson J-A. (1995) Br. J. Pharmacol. 114, 39–46. 28. Alvarez R., Sette C., Yang D., Eglen R. M., Wilhelm R., Shelton E. R. and Conti M. (1995) Mol. Pharmacol. 48, 616–622. 29. Engels P., Sullivan M., Muller T. and Lubbert H. (1995) FEBS Lett. 358, 305–310. 30. Bolger G., Michaeli T., Martins T., St John T., Steiner B., Rodgers L., Riggs M., Wigler M. and Ferguson K. (1993) Mol. Cell. Biol. 13, 6558–6571. 31. Hughes B., Howat D., Lisle H., Holbrook M., James T., Gozzard N., Blease K., Hughes P., Kingaby R., Warrellow G., Alexander R., Head J., Boyd E., Eaton M., Perry M., Wales M., Smith B., Owens R., Catterall C., Lumb S., Russell A., Allen R., Merriman M., Bloxham D. and Higgs G. (1996) Br. J. Pharmacol. 118, 1183–1191. 32. Livi G. P., Kmetz P., McHale M., Cieslinski L. B., Sathe G. M., Taylor D. J., Davis R. L., Torphy T. J. and Balcarek J. M. (1990) Mol. Cell. Biol. 10, 2678–2686. 33. Torphy, T. J., Stadel, J. M., Burman M., Cieslinski, L. B., McLaughlin, M. M., White, J. R. and Livi, G. P. (1992) J. Biol. Chem. 267, 1798–1804. 34. Harris A. L., Connell M. J., Ferguson E. W., Wallace A. M., Gordon R. J., Pagani E. D. and Silver P. J. (1989) J. Pharmacol. Exp. Ther. 251, 199–206. 35. Souness J. E., Griffin M., Maslen C., Ebsworth K., Scott L. C., Pollock K., Palfreyman M. N. and Karlsson J-A. (1996) Br. J. Pharmacol. 118, 649–658. 36. Barnette M. S., O’Leary Bartus J., Burman M., Christensen S. B., Cieslinski L. B., Esser K. B., Probhakar U. S., Rush J. A. and Torphy T. J. (1996) Biochem. Pharmacol. 51, 949–956. 37. Baecker P. A., Obernolte R., Bach C., Yee C. and Shelton E. R. (1994) Gene 138, 253–256. 38. Amegadzie B. Y., Hanning C. R., McLaughlin M. M., Burman M., Cieslinski L. B., Livi G. P. and Torphy T. J. (1995) Cell Biol. Int. 19, 477–484. 39. Schneider H. H., Schmiechen R., Brezinski M. and Seidler J. (1986) Eur. J. Pharmacol. 127, 105–115. 40. Schmiechen R., Schneider H. H. and Wachtel H. (1990) Psychopharmacology 102, 17–20. 41. Schultz J. E. and Schmidt B. (1986) Naunyn-Schmiedeberg’s Arch. Pharmacol. 333, 23–30. 42. Schneider H. H. (1984) Biochem. Pharmacol. 33, 1690–1693. 43. Schultz J. E. and Folkers G. (1988) Pharmacopsychiat. 21, 83–86. 44. McLaughlin M. M., Cieslinski L. B., Burman M., Torphy T. J. and Livi G. P. (1993) J. Biol. Chem. 268, 6470–6476.

236 45. Christensen S. B., DeWolf W. E. Jr., Ryan M. D. and Torphy T. J. (1996) In: Dent G., Rabe K. and Schudt C., eds, Handbook of Immunopharmacol: Phosphodiesterase Inhibitors, Academic Press, London: 185–207. 46. Jacobitz S., McLaughlin M. M., Livi G. P., Ryan M. D. and Torphy T. J. (1994) FASEB J. 8, A371 [Abstract]. 47. Jin S.-L., Swinnen J. V. and Conti M. (1992) J. Biol. Chem. 267, 18929–18939. 48. Francis S. H., Colbran J. L., McAllister-Lucas L. M. and Corbin J. D. (1994) J. Biol. Chem. 269, 22477–22480. 49. Kelly J. J., Barnes P. J. and Giembycz M. A. (1996) Biochem. J. 318, 425–436. 50. Souness J. E., Villamil M. E., Scott L. C., Tomkinson A., Giembycz M. A. and Raeburn D. (1994) Br. J. Pharmacol. 111, 1081–1088. 51. Giembycz M. A. and Souness J. E. (1994) In: Postma D, Gerritsen J. eds. Bronchitis V. Assen: Van Gorcum & Co, 319–332. 52. Souness J. E. (1996) In: Dent G., Rabe K. and Schudt C., eds, Handbook of Immunopharmacol: Phosphodiesterase Inhibitors. Academic Press, London: 173–184. 53. Conti M. and Swinnen J. V. (1990) In: Houslay, M. D., Beavo, J. A., eds, Molecular Pharmacology of Cell Regulation: Cyclic Nucleotide Phosphodiesterase Structure, Regulation and Drug Action. Wiley, New York: 243–266. 54. Bolger, G. (1994) Cell. Signalling 6, 851–859. 55. Engels P., Fichtel K. and Lubbert H. (1994) FEBS Lett. 350, 291–295. 56. Monaco L., Vicini E. and Conti M. (1994) J. Biol. Chem. 269, 347–357.

J. E. Souness and S. Rao 57. Sette C., Iona S. and Conti M. (1994) J. Biol. Chem. 269, 9245–9252. 58. Manganiello V. C., Taira M., Degerman E. and Belfrage P. (1995) Cell Signalling 7, 445–455. 59. Lowe III J. A., Archer R. L., Chapin D. S., Cheng J. B., Helwig D., Johnson J. L., Koe B. K., Lebel L. A., Moore P. F., Nielsen J. A., Russo L. L. and Shirley J. T. (1991) J. Med. Chem. 34, 624–628. 60. Lenhard J. M., Kassel D. B., Rocque W. J., Hamacher L., Holmes W. D., Patel I., Hoffman C. and Luther M. (1996) Biochem. J. 316, 751–758. 61. Huston E., Pooley L., Julien P., Scotland G., McPhee I., Sullivan M., Bolger G. and Houslay M. (1996) J. Biol. Chem. 271, 31334–31344. 62. Torphy T. J., Zheng C., Peterson S. M., Fiscus R. R., Rinard G. A. and Mayer S. E. (1985) J. Pharmacol. Exp. Ther. 233, 409–417. 63. Masamune H., Cheng J. B., Cooper K., Eggler J. F., Marfat A., Marshall S. C., Shirley J. T., Tickner J. E., Umland J. P. and Vazquez E. (1995) Bioorganic Med. Chem. Lett. 5, 1965–1968. 64. Cheng J. B., Cooper K., Duplantier A. J., Eggler J. F., Kraus K. G., Marshall S. C., Marfat A., Masamune H., Shirley J. T., Tickner J. E. and Umland J. P. (1995) Bioorganic Med. Chem. Lett. 5, 1969–1972. 65. Bolger G., Rodgers L. K. and Riggs M. (1994) Gene 149, 237– 244. 66. Kemp B. E. and Pearson R. B. (1990) Trends Biochem. Sci. 15, 342–346.