`DOI 10.1007/s00125-003-1176-7
`
`Review
`
`Cyclic nucleotide phosphodiesterases in pancreatic islets
`N. J. Pyne, B. L. Furman
`
`Department of Physiology and Pharmacology, Strathclyde Institute for Biomedical Sciences, University of Strathclyde,
`Glasgow, Scotland
`
`Abstract
`
`Cyclic nucleotide phosphodiesterases (PDEs) com-
`prise a family of enzymes (PDE1-PDE11) which hy-
`drolyse cyclic AMP and cyclic GMP to their biologi-
`cally inactive 5′ derivatives. Cyclic AMP is an impor-
`tant physiological amplifier of glucose-induced insu-
`lin secretion. As PDEs are the only known mechanism
`for inactivating cyclic nucleotides, it is important to
`characterise the PDEs present in the pancreatic islet
`beta cells. Several studies have shown pancreatic is-
`lets or beta cells to contain PDE1C, PDE3B and
`PDE4, with some evidence for PDE10A. Most evi-
`dence suggests that PDE3B is the most important in
`relation to the regulation of insulin release, although
`PDE1C could have a role. PDE3-selective inhibitors
`augment glucose-induced insulin secretion. In con-
`trast, activation of beta-cell PDE3B could mediate the
`inhibitory effect of IGF-1 and leptin on insulin secre-
`tion. In vivo, although PDE3 inhibitors augment glu-
`
`cose-induced insulin secretion, concomitant inhibition
`of PDE3B in liver and adipose tissue induce insulin
`resistance and PDE3 inhibitors do not induce hypo-
`glycaemia. The development of PDE3 inhibitors as
`anti-diabetic agents would require differentiation be-
`tween PDE3B in the beta cell and that in hepatocytes
`and adipocytes. Through their effects in regulating be-
`ta-cell cyclic nucleotide concentrations, PDEs could
`modulate beta-cell growth, differentiation and surviv-
`al; some work has shown that selective inhibition of
`PDE4 prevents diabetes in NOD mice and that selec-
`tive PDE3 inhibition blocks cytokine-induced nitric
`oxide production in islet cells. Further work is re-
`quired to understand the mechanism of regulation and
`role of the various PDEs in islet-cell function and to
`validate them as targets for drugs to treat and prevent
`diabetes. [Diabetologia (2003) 46:1179–1189]
`
`Keywords Islet beta cell, phosphodiesterase, cyclic
`AMP, cyclic GMP, insulin secretion.
`
`Introduction
`
`Received: 28 February 2003 / Revised: 8 May 2003
`Published online: 7 August 2003
`© Springer-Verlag 2003
`
`Corresponding author: B. L. Furman, Department of Physiolo-
`gy and Pharmacology, Strathclyde Institute for Biomedical
`Sciences, University of Strathclyde, Taylor Street, Glasgow,
`G4 0NR Scotland
`E-mail: b.l.furman@strath.ac.uk
`Tel.: +44-141-5484707
`Abbreviations: PDE, phosphodiesterase; GIP, glucose-depen-
`dent insulinotropic peptide; GLP-1, glucagon like peptide 1;
`PDX-1, pancreatic duodenal homeobox-1; IBMX, isobutyl-
`methylxanthine.
`
`Cyclic nucleotide phosphodiesterases comprise a
`family of enzymes the function of which is the hy-
`drolysis of cyclic AMP and cyclic GMP to their bio-
`logically inactive 5′ derivatives. Currently there are
`11 known gene families of CN-PDEs consisting of
`more than 50 enzymes with differences in their sub-
`strate selectivities (cyclic AMP vs cyclic GMP), ki-
`netics, allosteric regulation, tissue distribution and
`susceptibility to pharmacological inhibition (see re-
`views [1, 2, 3]). The main properties of some of
`these enzymes are shown in Table 1 [3, 4, 5, 6, 7, 8,
`9, 10, 11, 12, 13, 14, 15, 16, 17]. In view of the clear
`
`MPI EXHIBIT 1076 PAGE 1
`
`MPI EXHIBIT 1076 PAGE 1
`
`
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`Tissues
`
`Reference
`
`Substrate Km µM Properties
`
`thyroid, testis, brain
`
`brain, lymphocytes
`blood vessels, testis
`
`[4]
`
`[5]
`[6]
`
`Brain, heart, platelets,
`liver, thymocytes
`
`[2] [7]
`
`[1] [2] [8] [9]
`
`Heart, blood
`vessels
`adipocyte
`hepatocyte
`lymphocyte
`
`cGMP
`cAMP
`
`cGMP
`
`cAMP
`cAMP
`
`3
`1–30
`
`10
`
`25
`0.5
`
`lung, inflammatory and
`immune cells, brain,
`blood vessels
`
`[10] [11]
`
`cAMP
`
`0.2–4
`
`Activated by Ca/
`calmodulin
`
`Activated by cGMP
`
`Inhibited by GMP
`Phosphorylation
`(PKA and insulin-sensitive
`kinase)
`
`Phosphorylation by PKA
`and p42/p44 MAPK
`(PDE4D3)
`
`1180
`
`Table 1.
`
`Enzyme
`
`PDE1A
`
`PDE1B
`PDE1C
`
`PDE2
`
`PDE3A
`
`PDE3B
`
`PDE4A
`
`PDE4B
`PDE4C
`PDE4D
`
`Inhibitors
`
`zaprinast
`
`vinpocetine
`SCH51866
`8-MM-IBMX
`EHNA
`
`milrinone
`Org 9935
`siguazodan
`cilostamide
`enoximone
`OPC3911
`rolipram
`
`RP73401
`zardaverine
`CP80633
`CDP840
`LAS31025
`SB207499
`zaprinast
`sildenafil
`dipyridamole
`SCH51866
`zaprinast
`dipyridamole
`
`dipyridamole
`SCH51866
`SCH51866
`dipyridamole
`zaprinast
`dipyridamole
`
`PDE5
`
`smooth muscle
`
`PDE6
`
`retinal photoreceptors
`
`PDE7A, PDE7B brain, lymphocytes,
`kidney
`PDE8A, PDE8B thyroid
`PDE9
`kidney
`PDE10A
`testis, brain
`
`PDE11A
`
`??
`
`[12]
`
`[13]
`
`[14]
`
`[15]
`[15]
`[3]
`
`[3]
`
`cGMP
`
`2
`
`Phosphorylation
`by PKA/PKG
`
`cGMP
`
`60–100
`
`cAMP
`
`cAMP
`cGMP
`cAMP
`cGMP
`cAMP
`cGMP
`
`0.2
`
`0.7
`0.17
`0.05
`3
`1
`0.5
`
`IBMX-insensitive
`IBMX-insensitive
`
`importance of cyclic nucleotides, especially cyclic
`AMP, in the pancreatic islet beta cell and the fact
`that phosphodiesterase-activity is the only known
`system for destroying these cyclic nucleotides, it is
`important to understand the nature and the roles of
`the phosphodiesterase enzymes expressed in islet be-
`ta cells. The aim of this short review is to examine
`two main questions: (i) what phosphodiesterase
`(PDE) isoenzymes are expressed in the pancreatic is-
`let beta cell? (ii) which PDE isoenzymes are func-
`tionally important in regulating cellular cyclic AMP
`concentrations in relation to insulin secretion?
`
`Roles of cyclic nucleotides in the pancreatic islet
`beta cell
`
`cAMP
`
`Insulin secretion
`
`Although glucose-induced insulin secretion does not
`seem to require increases in islet beta-cell cyclic AMP
`or the PKA-system [18, 19], intracellular concentra-
`tions of cyclic AMP in beta cells are increased by glu-
`cose [20, 21]. Moreover, this cyclic nucleotide is gen-
`erally accepted as an important amplifier of glucose-
`induced insulin release [22], particularly when its cel-
`lular concentrations are increased by various gut hor-
`mones implicated as incretins but also by glucose it-
`self [23]. Cyclic AMP augments glucose-induced in-
`sulin secretion through a number of mechanisms in-
`cluding increased opening of voltage-sensitive Ca2+
`
`MPI EXHIBIT 1076 PAGE 2
`
`MPI EXHIBIT 1076 PAGE 2
`
`
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`1181
`
`channels [24], calcium-induced Ca2+-release [25], ac-
`tivation of ryanodine receptors in the ER [26, 27],
`stimulation of beta-cell lipolysis [28] and direct ef-
`fects on exocytosis [23]. Most actions of cyclic AMP
`in the beta cell seem to be mediated through protein-
`kinase A (PKA)-catalysed phosphorylation events but
`direct effects of the cyclic nucleotide on exocytosis
`are partly PKA-independent [29]. PKA-independent
`effects on exocytosis can be mediated by the cyclic
`AMP-binding protein cAMP-GEFII, interacting with
`Rim2, a target of the small G-protein Rab3 [30]. Cy-
`clic AMP mediates the insulinotropic action of the in-
`cretin factors glucose-dependent insulinotropic pep-
`tide (GIP) and glucagon like peptide 1 (GLP-1) [31].
`GLP-1 is released during meals and markedly aug-
`ments the direct effect of glucose on insulin secretion
`via cyclic AMP-dependent mechanisms.
`
`Other roles in the beta cell
`
`In addition to its role in amplifying glucose-induced
`insulin secretion, cyclic AMP can mediate or modu-
`late other effects of glucose and insulinotropic hor-
`mones in the beta cell. Glucose-mediated increases
`in insulin synthesis involve the phosphorylation of
`the transcription factor pancreatic duodenal homeo-
`box-1 (PDX-1) and its translocation to the nucleus
`[32]. Whereas the effects of glucose on PDX-1 are
`not mediated by cyclic AMP, there is strong evidence
`for the importance of cyclic AMP in GLP-1-depen-
`dent stimulation of PDX-1 in the beta cell, as well as
`its translocation to the nucleus and its activation of
`the insulin gene promoter [33]. However, the role of
`cyclic AMP in regulating insulin synthesis is not
`clear because the adenylyl cyclase activator for-
`skolin, or the cyclic AMP analogue 8-bromo-cyclic
`AMP suppressed insulin transcription in INS-1 cells
`in a PKA-independent manner [34]. Cyclic AMP
`could mediate effects of glucose in stimulating the
`expression of immediate early response genes such
`as c-myc [35] and c-fos [36]. In this context, cyclic
`AMP can either activate or inhibit the p42/p44 mito-
`gen-activated protein kinase pathway depending on
`the cell type and conditions [37]. Glucose itself acti-
`vates the p42/p44 MAPK pathway [38, 39] and this
`effect is amplified, although not mediated by, cyclic
`AMP. Indeed, the activation of this pathway by GIP
`is cyclic AMP and PKA-dependent [40]. Regulation
`of gene expression by cyclic AMP could have rele-
`vance to effects on beta-cell growth, differentiation
`and apoptosis. For example, increased cyclic AMP
`concentrations protected rat islets against interleu-
`kin-1 beta mediated induction of nitric oxide syn-
`thase and nitric oxide production [41]. Excessive ni-
`tric oxide production seems to mediate subsequent
`apoptosis of beta cells. On the other hand, beta-cell
`lines were made more susceptible to apoptosis after
`
`exposure to dibutyryl cyclic AMP [42] or the cyclic
`AMP-increasing agent forskolin [43].
`
`Cyclic GMP
`
`Although increases in cyclic GMP have been linked
`with decreases in insulin secretion [44, 45], the role of
`this nucleotide in the islet beta cell remains unclear
`[46, 47]. More recent studies suggest that cyclic GMP
`could mediate the effects of low concentrations of ni-
`tric oxide in augmenting glucose-induced Ca2+ oscil-
`lations and insulin secretion in rat-isolated beta cells
`[48, 49, 50]. 8-bromoguanosine-cyclic GMP was
`found to augment glucose-induced Ca2+ oscillations in
`mouse islets and to inhibit KATP channel activity in
`mouse-intact beta cells [51]. Cyclic GMP was report-
`ed to mediate nitric oxide-induced apoptosis in the in-
`sulin-secreting cell line HIT-T15 [52].
`
`PDE isoforms present in the islets
`
`Relatively little is known about the PDE isoenzymes
`in the pancreatic islet beta cell. Most studies have
`used pancreatic islets, which contain four endocrine
`cell types; these comprise the insulin-secreting beta
`cells, the glucagon-secreting A cells, the somatostatin
`secreting D cells and the pancreatic polypeptide cells.
`Blood vessels also permeate the islets. Thus the pres-
`ence of a particular PDE in islets does not necessarily
`indicate its presence in the insulin-secreting cells.
`This technical problem has been circumvented by the
`use of insulin-secreting cell lines, although these have
`limitations as models for native beta cells.
`
`PDE1
`
`There are three distinct isoforms of the calcium-cal-
`modulin activated PDE1, termed PDE1A, B and C.
`These isoforms are encoded by distinct genes and
`show differential tissue expression. Earlier work
`showed islets to contain a cytosolic Ca-calmodulin ac-
`tivated PDE, with low and high Km components for
`both cAMP and cGMP and a particulate form, which
`appeared insensitive to Ca-calmodulin [53, 54, 55].
`We have recently shown a calcium-calmodulin acti-
`vated PDE activity in the pellet, but not the superna-
`tant, fraction of homogenates of an insulin-secreting
`cell line BRIN BD11 [56].
`The presence of a Ca-calmodulin activated PDE
`was also observed in another beta-cell line β-TC3 and
`furthermore RT-PCR was used to show that this iso-
`form was PDE1C [4]. Thus islets, and possibly beta
`cells, express PDE1. This is supported to some extent
`by the observation that the PDE1/PDE5 inhibitor zap-
`rinast (10–5–10–4 mol/l) produced a 14 to 30% inhibi-
`
`MPI EXHIBIT 1076 PAGE 3
`
`MPI EXHIBIT 1076 PAGE 3
`
`
`
`1182
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`tion in both membrane and cytosolic fractions of rat
`islet homogenates [57] and 25% inhibition of cyclic
`AMP/cyclic GMP PDE activity in BRIN-BD11 cells
`(IC50~1–5 µmol/l, [56]. A study [4] also showed that
`the PDE was inhibited by zaprinast (IC50 4.5 µmol/l)
`or 8-MM-IBMX (IC50 7.5 µmol/l) but not by vinpoce-
`tine (IC50 >100 µmol/l for PDE1C vs 1.4 µmol/l and
`9.8 µmol/l for PDE1A and PDE1B) [5, 6]; these find-
`ings were compatible with the presence of PDE1C but
`not PDE1A or PDE1B. Although the data support the
`presence of a PDE1, other isoforms are clearly ex-
`pressed, as evidenced by the presence in islets [57] or
`beta-cell lines [4, 56] of zaprinast-insensitive PDE ac-
`tivity.
`
`activity. In addition, the PDE activity in anti-PDE3B-
`immunoprecipitates was completely inhibited by mil-
`rinone [59]. Again, recent work showed SK&F 94836
`and Org 9935 to inhibit cyclic AMP PDE in the BRIN
`BD11 insulin-secreting cell line [56]. Unlike in islets,
`we found that these drugs inhibited cyclic AMP hy-
`drolysis only in the membrane fraction of BRIN-
`BD11 cell homogenates. We proposed that the PDE3
`activity found in the soluble fraction of homogenates
`of rat islets might be PDE3A from blood vessels or
`other non-beta-cell tissue. PDE3A was shown exclu-
`sively in the cytosolic fractions of all tissues studied,
`whereas PDE3B was expressed in particulate fractions
`[64].
`
`PDE3
`
`PDE4
`
`Islet PDE activity was inhibited by 60 to 70% using
`10 µmol/l cyclic GMP [58]. The pellet fraction of
`homogenates of BRIN BD11 cells was also found to
`contain a cyclic AMP PDE activity that was potently
`(IC50–0.7 µmol/l) inhibited (up to 30–40%) by cyclic
`GMP [56]. Moreover, the membrane and cytosolic
`fractions of islet homogenates contained a low Km
`(1.4–2.2 µmol/l) cyclic AMP PDE activity [57]. These
`data are consistent with the presence of PDE3 in insu-
`lin-secreting cells. There are two isoforms of PDE3
`(PDE3A and PDE3B) and there is convincing evi-
`dence that only PDE3B is expressed in the beta cell.
`For example, the use of a polyclonal antibody against
`a GST-PD3B, showed a single protein band in western
`blots of extracts of rat pancreatic islets and the beta-
`cell line HIT-T15 that corresponded in size to the
`PDE3B protein found in extracts of rat epididymal ad-
`ipose tissue [59]. Furthermore, we applied RT-PCR to
`total RNA of BRIN-BD11 cells and amplified a prod-
`uct using PDE3B-specific primers. The product
`showed over 97% sequence homology with rat adi-
`pose tissue PDE3B [56]. No product was obtained us-
`ing PDE3A-specific primers. The presence of PDE3B
`in native beta cells was also supported strongly by im-
`munocytochemistry, showing that in rat islets PDE3B
`was expressed only in cells that were co-stained with
`anti-insulin antibodies [59].
`Pharmacological characterization of PDE3 has
`been greatly facilitated by the development of potent,
`highly selective inhibitors of this isoform, such as
`SK&F 94836 (siguazodan) [60] and Org 9935 [61].
`We found that SK&F 94836 and Org 9935 were very
`potent inhibitors of rat islet PDE activity, especially in
`membrane fractions [57] with up to 85% of mem-
`brane-bound PDE being inhibited with IC50 values of
`5.5 and 0.05 µmol/l respectively. Of interest, we have
`also made similar observations in human islets [62].
`This was later confirmed by others [63], who showed
`that rat and human islets contained a milrinone-sensi-
`tive PDE, accounting for up to 70% of total islet PDE
`
`Some experiments showed rolipram to be without ef-
`fect on cyclic AMP hydrolysis in rat islet homoge-
`nates. However, in other experiments homogenate cy-
`clic AMP-PDE activity was inhibited up to 20% by
`rolipram in concentrations of 10–5 to 10–4 mol/l [57],
`although this inhibitory effect is labile. These findings
`have been confirmed [63], while another study [4] re-
`ported a rolipram sensitive cyclic AMP PDE activity
`in βTC3 cells and showed the expression of PDE4A
`and PDE4D using RT-PCR. We reported that approxi-
`mately 25% of the BRIN-BD11 cell homogenate cy-
`clic AMP PDE activity in both cytosolic and mem-
`brane fractions were inhibited by rolipram (IC50
`0.04–0.1 µmol/l) [56]. Taken together, these results
`suggest that beta cells express functional PDE4.
`
`Other PDE isoforms
`
`A combination of maximal inhibitory concentrations
`of Org 9935 (PDE3), rolipram (PDE4) and zaprinast
`(PDE1/5) inhibited BRIN-BD11 cell cyclic AMP PDE
`at low substrate concentrations by around 90%. A
`similar degree of inhibition was obtained by maximal
`inhibitory concentrations of the non-selective PDE in-
`hibitor IBMX. Thus the total cyclic AMP PDE activi-
`ty could be largely accounted for by PDE1, PDE3 and
`PDE4. The ~10% activity not inhibited by IBMX
`could be attributable to the IBMX-insensitive, cyclic
`AMP specific enzyme PDE8, although there is no oth-
`er direct evidence for this. We have also found that
`Western blotting of islet homogenates showed a pro-
`tein band corresponding to PDE10A. The significance
`of this finding has yet to be determined. However, the
`very low Km for cyclic AMP of PDE10 suggests that
`these isoforms could function to control basal cyclic
`AMP under conditions of glucose fasting or in the ab-
`sence of circulating GLP-1.
`
`MPI EXHIBIT 1076 PAGE 4
`
`MPI EXHIBIT 1076 PAGE 4
`
`
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`1183
`
`Role of islet beta-cell PDE in modulating
`insulin secretion
`
`no effect [56, 57] or, in some conditions, inhibited
`[57] glucose-induced insulin secretion is not known.
`
`Modulation of glucose-induced insulin secretion
`by PDE inhibitors
`
`The non-selective PDE inhibitor IBMX has been
`shown by many authors to augment glucose-induced
`insulin secretion and has been used widely as a tool
`to investigate the role of cyclic AMP in the beta cell.
`However, apart from its lack of selectivity for differ-
`ent PDE isoforms, IBMX could have actions that are
`not related to PDE-inhibitory activity, for example
`activation of the ryanodine receptor [65] and antago-
`nism at adenosine receptors [66]. We found that sev-
`eral selective PDE3 inhibitors (Org 9935, siguazodan,
`SK&F 94120, ICI118233) augmented glucose-in-
`duced insulin secretion from rat and human islets,
`whereas selective inhibitors of PDE4 and PDE1/5 did
`not [57, 62]. Other selective PDE3 inhibitors were
`also found to augment insulin secretion. Thus, milri-
`none augmented glucose-induced insulin secretion in
`rat [4] and human [63] isolated islets and pimobendan
`augmented glucose-induced insulin secretion in rat is-
`lets [67]. Milrinone also augmented glucose-induced
`insulin secretion by monolayers of neonatal rat pan-
`creatic monolayer cells [59], excluding an indirect ef-
`fect by modulating the secretion of other islet hor-
`mones. Moreover, Org 9935 and siguazodan aug-
`mented glucose-induced insulin secretion in the insu-
`lin-secreting cell line BRIN-BD11 [56] and OPC3911
`produced a similar effect in INS-1(832/13) cells [23].
`These data strongly support a role for PDE3, rather
`than other isoforms, in regulating the cyclic AMP
`pool that modulates insulin secretion. However, this
`is complicated by the observations that glucose-in-
`duced insulin secretion is augmented in beta-cell
`lines, as distinct from intact islets, by rolipram, a se-
`lective PDE4 inhibitor [4, 56]. There is good evi-
`dence in other, non-beta-cell systems for compart-
`mentalization of PDEs [68, 69, 70]. Therefore com-
`partmentalization of PDE4 isoforms might be differ-
`ent between beta-cell lines and native beta cells. In
`cell lines, PDE4 can interact with relevant cyclic
`AMP pools regulating insulin secretion due to its re-
`localisation into compartments containing the insulin
`secreting machinery. A further and more intriguing
`complication arises from the observation that 8-MM-
`IBMX, purported to be a selective inhibitor of the Ca-
`calmodulin activated PDE1C isoform, augmented
`glucose-induced insulin secretion in both β-TC3 cells
`and in islets [4]. Zaprinast, a PDE1/5 inhibitor also
`augmented glucose-induced insulin secretion in β-
`TC3 cells. The selectivity of 8-MM-IBMX for
`PDE1C at the concentrations (150 and 500 µmol/l)
`used to modulate insulin secretion is not clear. More-
`over, the reason for the discrepancy between these
`observations and those in which zaprinast either had
`
`Mechanisms underlying the augmentation of glucose-
`induced insulin secretion by PDE3 inhibitors. Caution
`must be exercised in interpreting the effects of all “se-
`lective” PDE inhibitors because it has not been estab-
`lished that their effects on insulin secretion are medi-
`ated solely through inhibition of cyclic AMP hydroly-
`sis. For example, effects of the selective PDE3 inhibi-
`tor pimobendan could involve a calcium-sensitising
`effect unrelated to PDE inhibition [67]. Effects of se-
`lective PDE3 inhibitors on glucose-induced insulin se-
`cretion can be shown in the absence of a measured in-
`crease in cyclic AMP concentrations [57, 67] although
`the agents clearly increased cyclic AMP when adenyl-
`yl cyclase was activated by forskolin [57]. In βTC3
`cells rolipram (PDE4) and 8-MM-IBMX (PDE1C) but
`not milrinone (PDE3) produced modest increases in
`cyclic AMP [4]. The failure consistently to demon-
`strate global increases in islet cyclic AMP concentra-
`tions in response to the selective PDE3 inhibitors
`could reflect compartmentalization of the PDEs as has
`been shown in a number of other systems, including
`adipocytes and cardiovascular tissues [70, 71, 72].
`Moreover, cyclic AMP accumulated as a result of se-
`lective inhibition of PDE3 can be destroyed by other
`PDE isoforms as it diffuses from its site of action. Al-
`though detailed studies on the compartmentalization
`of PDE3B in the islet beta cell have not been carried
`out, it is clear that the enzyme is membrane bound
`[23, 56, 57]. Nevertheless, effects of PDE3 inhibitors
`in the islets are compatible with effects mediated
`through cyclic AMP. Thus, the selective PDE3 inhibi-
`tor Org 9935 increased [Ca2+]i in rat intact, isolated is-
`lets of Langerhans [73]. The selective PDE3 inhibitor
`OPC3911 augmented Ca2+-induced exocytosis from
`INS-1(832/13) cells in the presence of normal intra-
`cellular concentrations of cyclic AMP but not when
`cyclic AMP was replaced by the PDE-resistant cyclic
`AMP analogue Sp-cAMPS [23]. OPC3911 also aug-
`mented depolarization-evoked exocytosis when a sub-
`maximal concentration of forskolin (500 nmol/l) was
`present to increase basal cyclic AMP concentrations
`[23].
`
`Effect of over-expression of PDE3B. Strong evidence
`supporting a role for PDE3B in the beta cell has been
`obtained using alternative approaches (thereby vali-
`dating studies based on pharmacological inhibition) in
`which adenoviral plasmid constructs were used to
`over-express PDE3B in insulinoma cells and native rat
`islets [23]. This resulted in a six- to eightfold over-ex-
`pression in membrane-associated PDE3B activity. In-
`sulin secretion in response to glucose and GLP-1-me-
`diated augmentation of glucose-induced insulin secre-
`tion were markedly reduced, consistent with an oppos-
`ing action of PDE3B on insulin release. Moreover, cy-
`
`MPI EXHIBIT 1076 PAGE 5
`
`MPI EXHIBIT 1076 PAGE 5
`
`
`
`1184
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`Fig. 1. Hypothetical regulation of pancreatic islet beta-cell
`phosphodiesterases. Some sites of action ofcAMP in amplify-
`ing glucose-induced insulin secretion are indicated by the star
`symbol (see also the text)
`
`clic AMP augmentation of Ca2+-induced exocytosis
`measured by capacitance changes in patch-clamped in
`single beta cells was also considerably diminished in
`cells over-expressing PDE3B, whereas the response to
`a PDE-resistant cAMP analogue (Sp-CAMPS) was
`not modified.
`The involvement of PDE3 in regulating the cyclic
`AMP pool relevant to insulin secretion could, in part,
`explain the conflicting findings concerning the role of
`cyclic GMP in beta-cell function. Increases in cyclic
`GMP could inhibit beta-cell PDE3, leading to increas-
`es in cyclic AMP. Thus any direct effects of cyclic
`GMP could be confounded indirectly by the actions of
`cyclic AMP. This cross-talk has not yet been found in
`the islet beta cell but does occur in other cell types, in-
`cluding endocrine cells. For example, the increase in
`renin secretion from renal juxtaglomerular cells in re-
`sponse to nitric oxide could be mediated by a rise in
`cyclic AMP as a result of cyclic GMP-mediated inhi-
`bition of PDE3 [74]. This could explain the increased
`glucose-induced insulin release in response to the
`PDE1/5 inhibitor zaprinast reported by [4], although
`we found no such effect either in islets or in a beta-
`cell line [56].
`
`Regulation of PDEs in the pancreatic islet beta cell
`
`Relatively little is known about the long- or short-term
`regulation of PDEs in the islet beta cell. In other tis-
`sues PDE1, by definition, is activated by calcium-cal-
`modulin and, as described above, there is evidence for
`this in the pancreatic islet beta cell, although the role
`of this regulation in modulating insulin secretion or
`other islet beta cell functions has not been established.
`In a thyroid cell line, PDE4 is rapidly activated by cy-
`clic AMP through a PKA-mediated phosphorylation
`[75], whereas in smooth muscle both PDE3 and PDE4
`are activated in a similar fashion [76]. The role of
`PKA-catalysed activation of these enzymes has not
`yet been investigated in the pancreatic islet beta cell.
`The PDE3 activity of the pancreatic islet beta cell
`seems to be regulated by insulin, insulin-like growth
`factor-1 (IGF-1), leptin and by glucose itself. The reg-
`ulation of PDEs in the pancreatic islet beta cell is
`summarized in Fig. 1 and is further described below.
`
`Insulin and insulin-like growth factor-1 (IGF-1). It is
`well known that insulin activates PDE3B in adipo-
`cytes and that this contributes importantly to the anti-
`lipolytic effects of insulin [71]. Insulin receptors have
`been observed on the beta cell [77] and insulin was
`shown in some studies to inhibit its own release [77,
`78, 79]. Thus this inhibition could be achieved by in-
`sulin-mediated activation of beta-cell PDE3 and con-
`sequent reduction in beta-cell cyclic AMP concentra-
`tions. However, we were unable to show any effect of
`insulin on PDE activity in freshly isolated rat islets
`[57]. In contrast IGF-1 potently activated PDE activity
`in neonatal rat pancreatic monolayer cell cultures [59]
`
`MPI EXHIBIT 1076 PAGE 6
`
`MPI EXHIBIT 1076 PAGE 6
`
`
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`1185
`
`and in BRIN-BD11 cells [43, 56], through a PI-3 ki-
`nase-dependent pathway [43]. Another study [59] also
`showed that IGF-1 reduced cyclic AMP concentra-
`tions. Moreover, IGF-1 attenuated the augmentation of
`glucose-induced insulin release produced by the
`PDE3B-hydrolyzable cyclic AMP analogue 8-Br- cy-
`clic AMP but not that produced by the non-hydrolyz-
`able N6-benzoyl-cAMP, further suggesting that its ef-
`fects were due to activation of PDE. Moreover, these
`effects of IGF-1 were prevented by inhibition of
`PDE3 using milrinone. Thus insulin at high concentra-
`tions acting through low-affinity binding to IGF-1 re-
`ceptors, or at physiological concentrations through in-
`sulin receptors, might modulate its own release via
`PDE3B activation. The finding that inhibitors of PI-3
`kinase augmented insulin secretion is compatible with
`this hypothesis [80].
`
`Leptin. Leptin markedly inhibited GLP-1-augmenta-
`tion of glucose-induced insulin release in a neonatal
`rat pancreatic culture [81]. This effect was accompa-
`nied by a reduction in cyclic AMP concentrations and
`an activation of PDE. Activation of PDE3B probably
`occurs via a cytokine-dependent regulation of IRS-1/2
`and associated binding of PI3K. Indeed, PKB is a pos-
`sible candidate for phosphorylation of PDE3B in re-
`sponse to both leptin and IGF-1. Leptin attenuated the
`augmentation of glucose-induced insulin release pro-
`duced by the PDE3B-hydrolyzable cyclic AMP ana-
`logue 8-Br-cAMP but not that produced by the non-
`hydrolyzable N6-benzoyl-cAMP, further suggesting
`that its effects were due to activation of PDE [81].
`However, the role of PDE activation in leptin-induced
`inhibition of insulin secretion in vivo was questioned
`in view of the failure of the PDE3 inhibitor milrinone
`to abrogate this inhibition [82].
`
`Glucose. PDE1 activity was increased in the insulin-se-
`creting cell line βTC3 when the cells were cultured in a
`high glucose medium and was reduced in glucose-de-
`prived cells [4]. On the other hand, a short-term increase
`of glucose concentrations (to 11.1 mmol/l) activated
`PDE3B activity in another insulin secreting cell line
`(HIT-T15) [58]. The mechanisms underlying the effects
`of glucose in these cell lines has not been investigated.
`
`PDEs in other islet cell types
`
`There are no studies concerning the PDE enzymes
`present in the other cell types of the islets. However,
`protein kinase A mediates the effect of adrenaline on
`glucagon secretion [83] and glucagon synthesis and
`secretion are stimulated by a combination of forskolin
`and IBMX [84, 85]. Thus, the presence of the adenyl-
`yl cyclase-cyclic AMP-PKA system strongly indicates
`that phosphodiesterases are likely to be present and
`functionally active in the islet A-cell. It is important to
`
`determine their nature and how they modulate gluca-
`gon secretion.
`
`Is there potential for the therapeutic application
`of PDE inhibitors in treating diabetes mellitus?
`
`Type 2 diabetes mellitus is characterised by impaired
`insulin secretion and peripheral insensitivity to the hor-
`mone [86]. Treatment of Type 2 diabetes is currently
`unsatisfactory and new agents are needed. One ap-
`proach is to develop non-sulphonylurea drugs that will
`augment insulin secretion through mechanisms other
`than blocking KATP channels. Agents increasing islet
`beta-cell cyclic AMP have potential as therapeutic
`agents and GLP-1 and its derivatives have been shown
`to normalise insulin responses to glucose and almost to
`normalize overnight and daytime glucose concentra-
`tions [87, 88]. However, GLP-1 has the disadvantages
`associated with peptides, namely rapid degradation and
`inactivity by the oral route. Selective inhibition of
`PDE3 in the islet beta cell will augment meal-related
`insulin secretion, because of the amplification of the
`effect of incretin factors, particularly GLP-1. Thus
`PDE3 offers a target for developing drugs for the treat-
`ment of Type 2 diabetes mellitus. Development of
`PDE3 inhibitors for this purpose will require their se-
`lectivity for islet beta cell PDE3, as PDE3 seems also
`to be the important isoenzyme in the liver and adipose
`tissue [89], where its activation mediates some of the
`effects of insulin. Indeed, our in vivo work showed
`that a potent PDE3 inhibitor, Org 9935, considerably
`augmented glucose-induced increases in plasma insu-
`lin concentrations, but did not modify plasma glucose
`concentrations, possibly due to concomitant inhibition
`of hepatic and adipose tissue PDE3 [90]. Milrinone
`improved glucose tolerance in the normal mouse but
`was hyperglycaemic in the ob/ob mouse, despite in-
`creases in plasma insulin [91]. This group also showed
`milrinone to antagonise insulin-mediated inhibition of
`lipolysis in isolated adipocytes and insulin-mediated
`inhibition of glucose production in isolated hepa-
`tocytes. Of interest, the non-selective PDE inhibitor
`caffeine was shown recently to augment glucose-in-
`duced increases in serum insulin concentrations, while
`tending to increase blood glucose during OGTT in
`young, healthy male volunteers, again suggesting insu-
`lin resistance [92]. These observations could explain
`why metabolic effects such as hypoglycaemia were not
`reported in the major clinical trial of the PDE3 inhibi-
`tors milrinone for heart failure [93]. Although highly
`speculative, hyperinsulinaemia might have contributed
`to the higher mortality in milrinone-treated patients in
`this trial, as hyperinsulinaemia could activate PAI-1
`[94], thereby impairing fibrinolysis.
`Another intriguing possibility lies in the potential
`of PDE inhibitors to prevent beta-cell loss in both
`Type 1 and Type 2 diabetes. The non-selective PDE
`
`MPI EXHIBIT 1076 PAGE 7
`
`MPI EXHIBIT 1076 PAGE 7
`
`
`
`1186
`
`N. J. Pyne et al.: Cyclic nucleotide phosphodiesterases in pancreatic islets
`
`inhibitor pentoxifylline and the PDE4-selective agent
`rolipram were shown to reduce insulitis and prevent
`diabetes in non-obese diabetic (NOD) mice [95].
`While this effect can be explained by inhibition of cy-
`tokine production by pro-inflammatory cells such as
`macrophages there could also be a contribution from
`an inhibition of nitric oxide production by islet cells in
`response to cytokines. Thus pentoxifylline and the se-
`lective PDE3 inhibitor cilostamide blocked nitric ox-
`ide production by mouse islet cells and a beta-cell line
`(NIT-1) in response to stimulation by lipopolysaccha-
`ride and inflammatory cytokines [96] and prevented
`the expression of inducible nitric oxide synthase both
`in vitro and in vivo in the NOD