Specific localized expression of cGMP PDEs in Purkinje neurons and macrophages

Abstract

As cGMP hydrolyzing cyclic nucleotide phosphodiesterases (PDEs) have diverse regulatory and catalytic properties, the specific cGMP PDEs a cell expresses will determine the duration and intensity of a cGMP signal. This, in turn, results in different cellular responses between cell types and tissues. Therefore, identifying which cGMP PDEs are expressed in different tissues and cell types could increase our understanding of physiological and pathological processes. The brain is one area where large numbers of diverse cGMP PDEs are expressed in specific regions and cell types. A case in point is differential expression of cGMP PDEs in neuronal cells. For example, we have recently found that PDE5 is expressed in all Purkinje neurons while PDE1B is expressed in only a subset of these neurons. The expression of PDE2 has also been found to be selective for discrete populations of neurons. Another example of selective cGMP PDE expression is seen with cytokine-induced differentiation of monocytes to macrophages. We have recently discovered that monocyte differentiation with the cytokine macrophage colony-stimulating factor (M-CSF) causes an upregulation of PDE2 and a small increase in PDE1B while granulocyte-macrophage colony-stimulating factor (GM-CSF) causes a large increase in PDE1B but a decrease in PDE2. These same cytokines can influence the phenotype of microglial cells and are likely to affect their expression of cGMP PDEs. In this report, we present recent results from our laboratory and review earlier findings illustrating the concept of highly specific expression of cGMP PDEs and discuss how this may be important for understanding brain function and dysfunction.

Introduction

Cyclic nucleotide phosphodiesterases (PDEs) are responsible for degrading the second messengers cAMP and cGMP. Eleven different PDE families are currently known; most of which contain several different isoforms. Although the PDE families all have a homologous catalytic domain, they are distinguished by differing specificities for cAMP or cGMP, regulation by allosteric activators and inhibitors, enzymatic characteristics, and pharmacological inhibitor profiles. They also have different tissue as well as cellular and subcellular distributions. Given that these enzymes have different regulatory and kinetic properties, which isoforms are expressed in a cell will determine the intensity and duration of the cAMP or cGMP signal and will control the cellular response. This makes it crucial to identify the PDEs expressed in a cell to understand how they regulate cAMP/cGMP-mediated processes. Furthermore, as PDEs play roles in many physiological and pathological processes, identifying the PDEs involved in such processes leads to identification of potential therapeutic targets. This report will review recent findings from our laboratory concerning specific expression of cGMP PDEs in two different systems. Specific expression of cGMP PDE isoforms in Purkinje cells and cytokine-regulated expression of different cGMP PDE isoforms in macrophages will be presented as examples of how important the concept of targeted PDE expression is and how it can be important for understanding brain physiology and disease.

Nitric oxide (NO) is an important neurotransmitter in the brain and has been implicated to play roles in neuronal signaling, synaptic plasticity, learning, and perception of pain as well as in several pathological conditions (Esplugues, 2002). NO activates the soluble guanylate cyclase (sGC) leading to increased production of cGMP. The cGMP produced can then activate targets such as protein kinase G (PKG) or cyclic nucleotide gated channels. The duration and magnitude of a NO-induced cGMP signal will be determined by how it is terminated by the cGMP PDEs present. A good example of this has been presented in work that demonstrated disparate kinetics of cGMP accumulation and degradation between cerebellar astrocytes (Bellamy et al., 2000) and striatal neurons (Wykes et al., 2002) and the differences were in part due to the activities of the different PDE isoforms expressed in the two cell types.

The cGMP hydrolyzing PDE isoforms PDE1, PDE2, PDE3, PDE5, PDE9, PDE10, and PDE11 are all expressed in the brain but in different locations (Murashima et al., 1990, Repaske et al., 1993, Yan et al., 1994, Reinhardt and Bondy, 1996, Kotera et al., 1997, Juilfs et al., 1999, Andreeva et al., 2001, Yuasa et al., 2001). Fascinatingly, expression of the different isoforms of cGMP PDEs has been found to be highly specific to sets of neurons in different regions of the brain. One example of highly selective expression is PDE1. Three different PDE1 gene families (A, B, and C) have been identified and differ in their hydrolytic specificities for cAMP and cGMP as well as in several other modes of regulation such as phosphorylation by different kinases. By in situ hybridization, the different PDE1 isoforms were found to have distinct localizations in mouse brain (Yan et al., 1994). PDE1B1 was found in Purkinje cells and PDE1C1 was present in Purkinje and granule cells while PDE1C5 was expressed only in granule cells (Yan et al., 1994). Furthermore, another PDE1 variant, PDE1C2, was found to be specific for a set of olfactory neurons (Yan et al., 1995). In these neurons, PDE1C2 was expressed almost exclusively in the olfactory cilia and not in the cell bodies and axons suggesting that it is targeted in the cell to a specific location for a precise function. PDE1 activity is stimulated by calcium/calmodulin binding and is likely to regulate cGMP in response to cellular conditions that elevate Ca²⁺. For instance, PDE1 has been postulated to be involved in cGMP hydrolysis after NMDA activation of calcium dependent nNOS (Baltrons et al., 1997). Thus, PDE1 is likely to have a specialized role in neurons necessitating specific localization.

A second example of specific cGMP PDE localization in neurons is PDE2. The cAMP and cGMP hydrolytic activity of PDE2 is stimulated by cGMP binding to an allosteric GAF domain of the protein; consequently, it may function not only to degrade cGMP but also to provide cross-talk between a NO signal and cAMP. PDE2 has been shown to be localized with guanylate cyclase-d to a small set of olfactory neurons (Juilfs et al., 1997). These neurons are distinct from a separate set in olfactory tissue that expressed PDE1C2, PDE4A, and ACIII. PDE2 mRNA expression has also been found to be high in the habenula, olfactory cortices, hippocampus, basal ganglia, and cortex. By immunocytochemistry, it was shown that PDE2 expression is found in specific cells and the protein was found localized in cell bodies as well as in the axons and dendrites (Juilfs et al., 1999). These findings illustrate that cGMP PDEs are highly localized to specific regions and cell types in the brain and even to defined subcellular locations.

Our laboratory has recently characterized, in detail, the expression and regulation of PDE1 and PDE5 in mouse cerebellar Purkinje neurons (Shimizu-Albergine et al., 2003). By Mono Q chromatography, it was determined that PDE1B, PDE1C, and PDE5 constitute the majority of mouse cerebellar cGMP PDE activity. In previous work, it was shown by in situ hybridization that PDE1C is expressed in some Purkinje neurons (Yan et al., 1996). In our latest report, PDE1B and PDE5 were found by immunocytochemistry to be expressed in neurons of the Purkinje cell layer. However, unlike PDE5, which was expressed in all Purkinje neurons, PDE1B was expressed in only a subset of these cells (see Fig. 1). It is unclear exactly why PDE1B would be expressed in only a subset of Purkinje neurons, but it would give these cells precise control over cGMP degradation and allow individual Purkinje cells to regulate cGMP in a different way. In Purkinje neurons, NO and cGMP have been implicated in the induction of long-term depression (LTD) (Crepel and Jaillard, 1990, Daniel et al., 1993, Hartell, 1994, Lev-Ram et al., 1997). PDE5 and PDE1 in Purkinje neurons may play a role in limiting the cGMP-mediated LTD response.

Furthermore, it was demonstrated that PDE5 in these neurons could be phosphorylated and activated by PKG. An earlier report demonstrated that cGMP stimulates PKG-mediated phosphorylation of PDE5 in smooth muscle cells (Rybalkin et al., 2002) which serves to stimulate PDE5 activity (Corbin et al., 2000, Murthy, 2001). In this report, phosphorylated PDE5 could be detected in Purkinje neurons after microinjection of 8-Br-cGMP into the lateral ventricle. PDE5 phosphorylation could also be detected in Purkinje cell bodies and dendrites of cerebellar slices after treatment with the PDE5 selective inhibitor sildenafil or the non-selective PDE inhibitor IBMX and could be blocked by the guanylate cyclase inhibitor ODQ. Similar treatments of PKG I knockout mice did not demonstrate phosphorylated PDE5 by immunocytochemistry or Western blotting. Therefore, phosphorylation of PDE5 is likely to provide a feedback regulation of cGMP in Purkinje neurons and its phosphorylation could be used as an indicator of PKG activity in cells. Thus, in the cells expressing both PDE1B and PDE5, the amplitude and duration of the cGMP signal would be even smaller. These studies emphasize how the targeted expression of PDEs to specific tissues or individual cells can change their response to stimuli that induce cGMP elevation.

Macrophages are a heterogeneous cell type dispersed throughout all tissues in the body. Macrophages perform many functions as they are active in the innate defense and also serve as a link to the adaptive immune response (Wewers, 1997, Hart et al., 1998, Morrissette et al., 1999). However, the phenotype of resident macrophages varies between tissues. This probably reflects a need for specialized local physiological functions in different tissues. Microglia are the resident macrophages in the brain and are adapted to protect the neural environment. Microglia are normally in a resting state, but can be activated to mount an immune response through their ability to secrete cytokines, phagocytose invading pathogens, and present antigen to T cells (Aloisi, 2001, Hanisch, 2002). As with all macrophages, microglia are a double-edged sword and excessive or inappropriate activation can contribute to neural pathology. Through their inflammatory actions, microglia have been postulated to contribute to neural damage in disease states such as Alzheimer’s disease, AIDS dementia, Huntington’s disease, traumatic brain injury, and infectious diseases of the brain (Aloisi, 2001, Nakamura, 2002, Liu and Hong, 2003).

As microglia are involved in many pathological conditions, there is a great deal of interest in finding methods for altering their inflammatory functions. One possible target is the cGMP pathway. Several reports have shown that cGMP can inhibit microglial release of inflammatory mediators (Paris et al., 1999, Yoshikawa et al., 1999, Paris et al., 2000). One of the effective methods used to increase cGMP in these studies was through the use of cGMP PDE inhibitors. However, it has not been determined which cGMP PDE isoforms are expressed in microglia. Identifying which cGMP PDEs are expressed in microglia could yield selective targets for modifying various functions of microglia and inflammation in the brain.

As using an isoform-selective cGMP PDE inhibitor to decrease inflammatory actions would be preferential to using a non-selective inhibitor, we have determined which cGMP PDE isoforms are expressed in different monocyte-derived macrophage types (Bender et al., 2003). In our study, we used monocytes isolated from peripheral human blood and differentiated them in culture to macrophages with two different cytokines, macrophage colony-stimulating factor (M-CSF) or granulocyte-macrophage colony-stimulating factor (GM-CSF). These two cytokines are the most prominent promoters of monocyte differentiation and promote differentiation to macrophages with distinct phenotypes and unique functional abilities. GM-CSF and M-CSF have also been shown to affect the development of microglial cells and produce cells of different phenotypes (Fischer et al., 1993, Santambrogio et al., 2001). Several reports have documented how these cytokines can change microglia phenotype (Fischer et al., 1993, Fujita et al., 1996, John et al., 2003) and have identified changes in gene expression differentially induced by the two cytokines (Re et al., 2002). Two populations of microglia exist: one a permanent fixed resting population which are ramified throughout the brain parenchyma and a second population of perivascular microglia that is periodically replaced by infiltrating monocytes (Hickey and Kimura, 1988, Williams et al., 2001). It is likely that the phenotype of the perivascular population of microglia will be influenced by either M-CSF or GM-CSF-induced differentiation of monocytes.

In our study, we found that monocytes differentiated to macrophages with either M-CSF or GM-CSF have unique cGMP PDE expression profiles. PDE1B and PDE2A were found at low levels in monocytes but are the major cGMP PDEs expressed in macrophages (see Fig. 2). M-CSF differentiation triggers increased expression of PDE1B and PDE2A. In contrast, GM-CSF triggers differentiation to a macrophage type that does not express PDE2A but has PDE1B expression that is substantially higher than the M-CSF differentiated cells. Intriguingly, the different macrophage types also differed in their expression of guanylate cyclase isoforms. At this time, it is unknown what the functions of PDE1B and PDE2A are in these cells. The fact that they are upregulated during differentiation suggests that they may either be causal for the differentiation and determine the phenotype of the cell, or that they may regulate a functional ability that is gained when the cells become macrophages.

The different phenotypes of tissue macrophages likely arise from the varying concentrations of cytokines in the local environment. M-CSF differentiated monocytes resemble peritoneal macrophages (Andreesen et al., 1990, Hashimoto et al., 1997) while GM-CSF differentiated monocytes resemble alveolar macrophages (Andreesen et al., 1990, Akagawa, 2002). Additional cytokines in combination with either of these colony-stimulating factors can change the phenotype of the cells. For example, differentiation with GM-CSF in the presence of IL-4 leads to the formation of dendritic cells instead of macrophages (Sallusto and Lanzavecchia, 1994, Thurner et al., 1999). Dendritic cells have also been shown to have a PDE profile distinct from macrophages (Gantner et al., 1999). Therefore, cytokine-regulated expression of PDEs may be a general mechanism and may apply to many cell types such as microglia. Identification of unique cGMP PDEs may lead to targeted PDE inhibition and potential therapeutics for neuroinflammatory disorders.