Abstract
We developed selective monoclonal antibodies and used them for Western and immunocytochemical analyses to determine the tissue and cellular distribution of the human cyclic GMP-stimulated phosphodiesterase (PDE2). Western analysis revealed PDE2A expression in a variety of tissue types, including cerebellum, neocortex, heart, kidney, lung, pulmonary artery, and skeletal muscle. Immunocytochemical analysis revealed PDE2A expression in a subset of tissue endothelial cells. PDE2A immunostaining was detected in venous and capillary endothelial cells in cardiac and renal tissue but not in arterial endothelial cells. These results were confirmed by in situ hybridization. PDE2A immunostaining was also absent from luminal endothelial cells of large vessels, such as aorta, pulmonary, and renal arteries, but was present in the endothelial cells of the vasa vasorum. PDE2A immunostaining was detected in the endothelial cells of a variety of microvessels, including those in renal and cardiac interstitial spaces, renal glomerulus, skin, brain, and liver. Although PDE2A was not readily detected in arterial endothelial cells by immunocytochemistry of intact tissue, it was detected at low levels in cultured arterial endothelial cells. These results suggest a possible role for PDE2A in modulating the effects of cyclic nucleotides on fluid and inflammatory cell transit through the endothelial cell barrier.
Keywords
T
The cGMP-stimulated phosphodiesterase (PDE2) is a homodimer of two 105-kD subunits and exists in particulate and soluble forms. PDE2 enzymatic activity was first described in rat liver extracts (Beavo et al. 1970) and was subsequently purified to apparent homogeneity from a variety of tissues, including bovine heart (Martins et al. 1982), adrenal gland (Miot et al. 1985), brain (Murashima et al. 1990), and liver (Yamamoto et al. 1983), rat liver (Pyne et al. 1986), and rabbit brain (Whalin et al. 1988). Although cGMP is the preferred substrate and effector molecule for this enzyme, PDE2 hydrolyzes both cGMP and cAMP with positively cooperative kinetics. At physiological concentrations of cyclic nucleotides, PDE2 responds to elevated cGMP with increased hydrolysis of cAMP (Manganiello et al. 1990). Therefore, this phosphodiesterase can provide crosstalk between the cAMP and cGMP signaling pathways.
To date, cDNAs encoding three variants of PDE2 have been described (PDE2A1, Sonnenburg et al. 1991; PDE2A2, Yang et al. 1994; PDE2A3, Rosman et al. 1997). These variants arise from a single gene by alternative splicing events at the 5′ end of the sequence with the potential of generating three almost identical proteins ranging in predicted molecular weight from 103 to 105 kD. Northern analyses using probes derived from these cDNA sequences reveal widespread expression of the PDE2 transcripts. As expected from the biochemical results, bovine PDE2A mRNA levels are most abundant in adrenal gland, heart, and brain (Sonnenburg et al. 1991). PDE2A is expressed in neuronal cells throughout the brain, as demonstrated by in situ hybridization analysis of rat brain, with the highest levels of PDE2A expression observed in the limbic system (Repaske et al. 1993). PDE2A has also been detected in a subset of rat olfactory neurons (Juilfs et al. 1997) and within the zona glomerulosa of bovine adrenal gland (MacFarland et al. 1991) using immunocytochemical approaches. Immunocytochemical analysis of PDE2A expression in tissue outside of the nervous system and the adrenal gland has not been reported.
In humans, PDE2A mRNA is readily detected in brain and heart, and at lower levels in a variety of human tissues including placenta, lung, liver, skeletal muscle, kidney, and pancreas (Rosman et al. 1997). To more fully characterize the distribution of PDE2A in human tissues and cells, we developed immunoreagents selective for PDE2A and used these immunoreagents for Western and immunocytochemical analyses. We demonstrate here that PDE2A is expressed in a variety of tissues and that this expression is greatest in capillary and venous endothelial cells.
Materials and Methods
Antibody Preparation and Characterization
Full-length human recombinant PDE2A3 (Rosman et al. 1997) was expressed as a fusion protein with E. coli thioredoxin (LaVallie et al. 1993) under control of the tac promoter. Protein expression was induced in the E. coli strain XL-1 Blue (Stratagene; La Jolla, CA) with 1 mM isopropyl-β-D-thiogalactoside (IPTG) for either 16 hr at 22C or 1 hr at 37C. A 120-kD protein corresponding to the thioredoxin-PDE2A3 fusion protein was isolated from bacterial inclusion bodies by SDS-PAGE and electroelution. This material was used to generate mouse monoclonal antibodies (MAbs) 107B, 107E, and 107K using standard procedures (Harlow and Lane 1990).
Linear epitopes for MAbs 107B, 107E, and 107K were mapped by generating and expressing a series of 3′ deletions of the PDE2A3-thioredoxin fusion construct using the Promega Erase-a-Base system as described by the manufacturer (Promega; Madison, WI). Peptides corresponding to the linear epitopes AHPLFYRG (107B, 107E) and EEWSLQ (107K) were generated by Peptide Innovations (UCB-Bioproducts; Raleigh, NC). Blocking experiments were performed by preincubating the antibodies in a 50-fold molar excess of specific or nonspecific epitope peptide before immunostaining.
Western Analysis
Recombinant protein from human PDE isoforms 1–7 was electrophoresed on 10% SDS-PAGE and transferred to Immobilon P membranes (Millipore; Bedford, MA) for Western analysis to demonstrate the selectivity of the PDE2A antibodies. Samples of PDE1A and PDE1C (Loughney et al. 1998), PDE1B (Yu et al. 1997), PDE3A and PDE3B (Meacci et al. 1992; Miki et al. 1996) and PDE4A, PDE4B, PDE4C, and PDE4D (Bolger et al. 1993) were generously provided by Lothar Uher of ICOS Corporation. Samples of PDE2A3 (Rosman et al. 1997) and PDE5 (Loughney et al. 1998) were generously provided by Mike Gadau of ICOS Corporation. A sample of PDE7 (Michaeli et al. 1993) was generously provided by Cathy Farrell of ICOS Corporation. PDE1A, 1B, 2A3, 4B, 4C, 4D, 5, and 7 were at greater than 75% purity. Approximately 4 ng of each of these samples was loaded per lane, with the exception of 10 pg loaded per lane for PDE2A3. PDE1C was at ∼50% purity and therefore 8 ng of protein was loaded per lane. The PDE3A, 3B, and 4A samples were total cell extracts in which the expressed recombinant proteins represented 0.1–1% of the total cellular protein. Approximately 1 μg of each of these extracts was loaded per lane.
Extracts of human aorta, cerebellum, neocortex, heart, kidney, lung, pulmonary artery, and skeletal muscle were prepared from snap-frozen tissue obtained from either the National Disease Research Interchange (NDRI) or the Cooperative Human Tissue Network (CHTN). Briefly, the frozen tissue was pulverized in the presence of liquid nitrogen to a fine powder, using a mortar and pestle. Hot SDS sample buffer (125 mM Tris, pH 8.8, 2% SDS, 20% glycerol, 100 mM DTT, and 0.01% bromophenol blue) was immediately added to the frozen powder and the sample was boiled for 5 min. DNA in the extracts was sheared by repeated passage through a 24-gauge needle.
Approximately equal amounts of protein extract (as judged by Coomassie staining) were electrophoresed on 10% SDS-PAGE and transferred to Immobilon P membranes for Western analysis. Full-length recombinant PDE2A3 (in the absence of thioredoxin) was expressed in yeast as described by Rosman et al. (1997) and was used as a positive control for western analysis. MAb 107B was used at a concentration of 1 μg/ml in blocking Solution A (Tris-buffered saline, 5% powdered milk, 0.1% Tween). Horseradish peroxidase-conjugated, goat anti-mouse secondary antibody (BioRad; Richmond, CA) was used at a 1:10,000 dilution in blocking Solution A, followed by detection using the Renaissance Enhanced Luminal Western Blot Chemiluminescence Reagent (Dupont-NEN; Boston, MA).
Western analysis of cultured human endothelial cells was performed using a commercially available immunoblot (EndoPanel 1) from Clonetics Products of BioWhittaker (San Diego, CA). This blot contained 75 μg of protein per lane from cultured human endothelial cells derived from the following tissue types: pulmonary artery (HPAEC), coronary artery (HCAEC), iliac artery (HIAEC), aorta (HAEC), lung microvasculature (HMVEC-L), umbilical vein (HUVEC), umbilical artery (HUAEC), and foreskin (dermal) microvasculature (HMVEC-d Neo). All of the cells used to generate this blot had been in culture through three or four passages at the time of extract preparation. The blot was treated with MAb 107B at 3 μg/ml and processed as described above. The blot was then treated (without stripping the PDE2 antibody) with an actin antibody (A-4700, AC-40; Sigma, St. Louis, MO) at a 1:1000 dilution and processed as described above. Both the PDE2A and the actin immunoreactivity were detected on a single piece of film.
Immunohistochemistry
Formalin-fixed, paraffin-embedded tissues were obtained from NDRI or CHTN. Tissue sections (6 μm) were deparaffinized sequentially through xylene, ethanol solutions (100%, 95%, 70%), PBS, and finally water. Sections were treated with Target Retrieval Solution for 30 min at 95C as described by the manufacturer (Dako; Carpinteria, CA). All subsequent steps were carried out at room temperature (RT). Endogenous peroxidase activity was inhibited by a 15-min incubation with 0.33% H2O2 in PBS. Sections were then incubated in blocking Solution B (20% normal human serum, 5% normal goat serum, and 1% BSA in PBS) for 30 min, followed by treatment with the Avidin/Biotin Blocking Kit as described by the manufacturer (Vector Laboratories; Burlingame, CA). Sections were subsequently incubated with either the PDE2-specific MAbs at 5-20 μg/ml or a mouse MAb against von Willebrand factor (Dako) at a 1:50 dilution in blocking Solution B for 2 hr. A mouse IgG1 antibody, MOPC-12 (Sigma), was used as an isotype-matched negative control at concentrations comparable to those used for the primary antibodies. The sections were washed with PBS, incubated with biotin-conjugated horse anti-mouse secondary antibody (Jackson Immunoresearch Laboratories; West Grove, PA) at a 1:200 dilution in blocking Solution B for 90 min, washed with PBS, and then incubated with the Vectastain Elite ABC Kit as described by the manufacturer (Vector Laboratories). The sections were then treated with 3,3′diamino-benzidine (DAB; Research Genetics, Huntsville, AL) for approximately 2 min until the color fully developed. The sections were then counterstained with Gill's hematoxylin #2, rinsed in PBS, dehydrated, and mounted.
In Situ Hybridization
An ∼400-
Sections of formalin-fixed, paraffin-embedded human kidney from NDRI were treated for in situ hybridization essentially as described by Yu et al. (1997), with several changes. After deparaffinization and before denaturation with 70% formamide, sections were digested with 6 μg/ml proteinase K in PBS for 10 min at RT, followed by heating to 95C for 2 min to inactivate the protease. Slides were dehydrated with ethanol and then prehybridized with hybridization buffer (50% formamide, 10% dextran sulfate, 1 X Denhardt's, 5 mM EDTA, 20 mM Tris-HCl, pH 8.0, 0.3 M NaCl, 100 mM DTT, and 300 μg/ml yeast tRNA) at 42C for 1 hr.
Hybridization was performed with 4 × 105 cpm of 35S-labeled riboprobe per sample in hybridization buffer at 50C for 16-20 hr. After hybridization, slides were washed in 4 × SSC, 10 mM DTT at RT for 1 hr, followed by a wash in 50% formamide, 1 × SSC, and 10 mM DTT at 60C for 40 min. Slides were washed at RT with 1 × SSC followed by 0.1 × SSC, each for 30 min. Slides were dehydrated through ethanol, air-dried, and dipped in undiluted Kodak NTB2 nuclear emulsion at 45C. Slides were air-dried for 2-3 hr and subsequently exposed in the dark at 4C with desiccant until development. Tissue samples were counterstained with hematoxylin and eosin.
Results
Specificity of PDE2A antibodies and Western analysis of PDE2A distribution in human tissue. (
Full-length recombinant human PDE2A3 linked to thioredoxin was used as an immunogen to generate mouse MAbs selective for PDE2A. Three antibodies, labeled 107B, 107E, and 107K, were characterized by Western analysis and epitope mapping. These three antibodies were selective by Western analysis for PDE2A and lacked immunoreactivity with recombinant representatives of human PDE1A, 1B, 1C, 3A, 3B, 4A, 4B, 4C, 4D, 5, and 7. An example of this analysis with 107B is shown in Figure 1A. Similar results were obtained with MAbs 107E and 107K (not shown). The linear epitope of each of these immunoreagents was determined using a 3′-exonuclease deletion series of the thioredoxin-PDE2A3 fusion protein in combination with western analysis. MAbs 107B and 107E both mapped to amino acids 483-490 of PDE2A3, represented by the sequence AHPLFYRG. MAb 107K mapped to amino acids 173-178 of PDE2A3, represented by the sequence EEWSLQ. The linear epitopes recognized by MAbs 107B, 107E, and 107K fall within regions of PDE2A3 common to all of the reported PDE2A splice variants. These sequences are unique to PDE2A relative to the other known PDE family members. These MAbs were subsequently used for Western and immunocytochemical analyses.
Western analysis was done to investigate the expression of PDE2A in protein extracts from a variety of human tissue types. An example of this analysis with MAb 107B is shown in Figure 1B. Similar results were observed with MAbs 107E and 107K (not shown). An immunoreactive band of approximately 100 kD was detected in all of the tissues examined. This immunoreactive band was similar in size to that of recombinant PDE2A3 (Figure 1B, Lane 1). The strongest immunoreactive signal was observed in the sample of neocortex. Lower levels of signal were detected in cerebellum, heart, kidney, lung, pulmonary artery, and skeletal muscle. An immunoreactive band at 100 kD was also detected in aorta after a longer exposure.
MAbs 107B, 107E, and 107K were subsequently used for immunocytochemical analyses to investigate in more detail the cellular distribution of PDE2A in human tissues. Initial studies focused on the distribution of PDE2A in renal tissue and demonstrated PDE2A-associated immunostaining in association with renal blood vessels and the glomerulus. PDE2A immunostaining was readily detected in venous endothelial cells in the kidney (Figure 2A, white arrows) but was absent from endothelial cells in an adjacent artery (Figure 2A, black arrows). In contrast, an antibody against von Willebrand factor, a marker of endothelial cells, was positive in both arterial and venous endothelial cells (Figure 2B. PDE2A immunostaining was also detected in endothelial cells associated with interstitial capillaries interspersed among the distal and proximal tubules in the renal cortex (Figure 2C. Lastly, PDE2A immunostaining was observed in a subset of cells within the glomerulus (Figure 2D. At higher magnification (Figure 2E, the immunostaining was clearly associated with endothelial cells of the glomerular capillary bed. In both the interstitial and glomerular endothelial cells, the PDE2A immunostaining appeared to co-localize with the hematoxylin-stained nuclei, whereas the staining in the venous endothelial cells appeared to be cytoplasmic. The PDE2A immunostaining pattern observed in renal tissue was observed with all three of the MAbs and was completely blocked when the immunocytochemistry was performed in the presence of peptides spanning the linear epitope sequences (not shown).
The results from the immunocytochemical analysis were further supported by results from in situ hybridization. A radiolabeled antisense riboprobe derived from the human PDE2A3 cDNA sequence was used to detect PDE2A mRNA in human renal tissue (Figure 3). In the darkfield example shown (Figure 3B, signal associated with PDE2A mRNA was readily detected in endothelial cells of a large vein but was not detected in endothelial cells in an adjacent artery. A complimentary sense riboprobe gave no signal (Figure 3C. Under brightfield microscopy, black silver grains associated with the antisense probe were clearly detected in the venous endothelial cells but were absent from the endothelial cells of the adjacent artery (Figure 3D. Positive signal was also detected in the interstitial endothelial cells in the renal cortex (Figure 3E and in a subset of cells within the glomerulus (Figure 3F.
We next looked at PDE2A immunostaining in human cardiac tissue. In a cross-section of cardiac tissue, PDE2A immunostaining was again detected in endothelial cells lining veins (Figure 4A, white arrow) but was absent from endothelial cells in adjacent arteries (Figure 4A, black arrow). As in the kidney, PDE2A immunostaining was observed in the endothelial cells of the interstitial capillaries and small venules (Figure 4B, white arrows) and was consistently absent from the endothelial cells of adjacent arterioles (Figure 4B, black arrow). PDE2A immunostaining was also detected in the cardiac myocytes (Figure 4B. In the example shown here, the brown PDE2A immunostaining appears to circumscribe the cross-sectioned myocytes. This immunostaining pattern became more evident when a fluorescent detection method was used and disappeared when the antibodies were preincubated with epitope-specific peptides (not shown).
PDE2A expression was next assessed in human liver, cerebellum, skin, and lung. In the liver, PDE2A immunostaining was readily detected in venous endothelial cells of the portal veins (Figure 4C, white arrow) and of the central vein (not shown) but was not detected in adjacent arterial endothelial cells (Figure 4C, black arrow). PDE2A immunostaining was also detected in endothelial cells in the sinusoidal capillaries (Figure 4D, arrows). PDE2A immunostaining was also detected in the microvasculature of the CNS. In the example shown, PDE2A immunostaining was detected in the endothelial cells of a capillary running through the granular layer of the cerebellum (Figure 4E. Lastly, PDE2A immunostaining was detected in microvessel endothelial cells in human foreskin (Figure 4F. Despite being detected in a variety of microvessel endothelial cells, PDE2A immunostaining was not observed in the endothelial cells comprising the pulmonary capillary bed of the alveoli (Figure 4G. In contrast, PDE2A immunostaining was detected in lung endothelial cells of small vessels adjacent to the bronchial smooth muscle cells (Figure 4G. In all of the examples described above, the PDE2A immunostaining appeared to be cytoplasmic or perinuclear.
Immunolocalization of PDE2A in human renal tissue. Sections of formalin-fixed human renal tissue were treated with the PDE2A-specific MAbs (
Localization of PDE2A mRNA in human renal tissue by in situ hybridization. An ∼400-bp cDNA fragment of human PDE2A3 was labeled with [35S]-UTP and used for in situ hybridization of formalin-fixed sections of human kidney as described in Materials and Methods. (
We then looked at PDE2A expression in the endothelial cells of large arterial vessels. Figure 5 shows a comparison between PDE2A immunostaining and von Willebrand factor immunostaining through a cross-section of human aorta. Whereas von Willebrand factor was clearly detected in the luminal endothelial cells (Figure 5D and 5E), PDE2A immunostaining was absent (Figures 5A and 5B). In contrast, immunostaining for both PDE2A and von Willebrand factor was detected in endothelial cells of the small vessels of the vasa vasorum (Figures 5A and 5C and 5D and 5F, respectively). PDE2A immunostaining was also present in the vasa vasorum of the carotid and pulmonary arteries but was absent from the luminal endothelial cells (not shown). These results, along with those from the tissues described previously, demonstrate that PDE2A immunostaining is readily detected in venous and capillary endothelial cells but is not detected in arterial endothelial cells.
Immunolocalization of PDE2A in venous and capillary endothelial cells in heart, lung, liver, and skin. Sections of formalin-fixed heart, liver, cerebellum, skin, and lung were treated with the PDE2A-specific MAbs (
Immunolocalization of PDE2A in human aorta. Sections of formalin-fixed human aorta were stained with the PDE2A-specific MAb 107E (
Lastly, we investigated the expression of PDE2A in cultured endothelial cells. A commercially available immunoblot was treated with MAb 107B as described in the Materials and Methods (Figure 6). An immunoreactive band of 42 kD representing actin was detected at equal levels in all of the cell extracts tested. In contrast, the immunoreactive band at approximately 100 kD, representing PDE2A, was detected at different levels in the cell extracts. PDE2A levels were low in arterial endothelial cells derived from the pulmonary artery (HPAEC), coronary artery (HCAEC), iliac artery (HIAEC), aorta (HAEC), and umbilical artery (HUAEC). As expected from the results described above, PDE2A levels were low in the microvascular endothelial cells from the lung (HMVEC-L). Surprisingly, PDE2A levels were also low in the umbilical vein-derived endothelial cells (HUVECs), possibly reflecting the arterial character of the umbilical vein. Finally, PDE2A was detected at the highest levels in microvascular endothelial cells derived from foreskin (HMVEC-d Neo), which is consistent with the immunocytochemical observations.
PDE2A expression in cultured human endothelial cells. Western analysis was performed with MAb 107B and an anti-actin antibody using a commercially available blot containing extracts from various human endothelial cells as described in Materials and Methods. HPAEC, human pulmonary artery endothelial cells; HCAEC, human coronary artery endothelial cells; HIAEC, human iliac artery endothelial cells; HAEC, human aortic artery endothelial cells; HMVEC-L, human microvascular endothelial cells, lung; HUVEC, human umbilical vein endothelial cells; HUAEC, human umbilical artery endothelial cells; and HMVEC-d Neo, human microvascular endothelial cells, dermal. The migration of prestained molecular weight markers (in kD) is shown at far left.
Discussion
In this report we describe the generation and use of selective MAbs to study the distribution of the cGMP-stimulated phosphodiesterase (PDE2) in a variety of human tissue types. MAbs were developed against recombinant human PDE2A and were shown by Western analysis and epitope mapping to be specific for PDE2A relative to other members of the PDE gene family. These three antibodies generated similar results in both Western and immunocytochemical analyses. PDE2A expression was detected at low levels by Western analysis in a variety of human tissues, including aorta, cerebellum, heart, kidney, lung, pulmonary artery, and skeletal muscle, and at high levels in the neocortex. A detailed analysis of human renal tissue using immunocytochemical approaches revealed selective expression of PDE2A in capillary and venous endothelial cells. These results were confirmed by in situ hybridization. PDE2A immunostaining was also detected in capillary and venous endothelial cells in human cardiac and hepatic tissues and in microvessel endothelial cells in skin, brain, and aorta. In contrast, PDE2A immunostaining was not detected in arterial endothelial cells in any of the intact tissues studied.
PDE2A was detected by Western analysis at low levels in cultured human arterial endothelial cells. This is consistent with reports of PDE2 enzymatic activity in a variety of cultured arterial endothelial cells, including bovine aortic endothelial cells (Lugnier and Schini 1990; Kishi et al. 1992) and porcine pulmonary artery (Suttorp et al. 1996a,b) and aortic endothelial cells (Souness et al. 1990). However, PDE2A immunostaining was not detected in arterial endothelial cells in any of the intact human tissues examined. It is possible that PDE2A is expressed in arterial endothelial cells in intact tissue at a level undetectable by the antibodies and immunocytochemical approach described here. Alternatively, the inability to detect PDE2A in intact human arterial endothelial cells may reflect differences in metabolic state and protein expression in cultured vs intact endothelial cells (Liaw and Schwartz 1993; Cines et al. 1998). In any event, it is clear from the immunocytochemical, in situ hybridization, and Western analyses that PDE2A is expressed at a much higher level in venous and capillary endothelial cells relative to arterial endothelial cells.
The subcellular distribution of PDE2A in endothelial cells remains to be clarified. In the majority of the tissues examined, the PDE2A immunostaining in endothelial cells was cytoplasmic or perinuclear. In contrast, the immunostaining in the renal interstitial and glomerular endothelial cells appeared to co-distribute with the hematoxylin-stained nuclei. Additional experiments are needed to clarify whether the apparent nuclear staining in these cells is really inside the nucleus or is closely associated with the outside of the nucleus. Preliminary experiments using immunofluorescence techniques and confocal microscopy suggest that the PDE2 immunostaining in glomerular endothelial cells is distributed at the periphery of the nucleus (unpublished observations). Additional experiments are needed to confirm this observation.
The role of PDE2A in endothelial cells is unclear. Although PDE2 is capable of hydrolyzing both cAMP and cGMP, it does so at relatively high substrate concentrations. Unless PDE2 is responsible for a very defined intracellular pool of cAMP, it is reasonable to assume that if other cAMP hydrolyzing phosphodiesterases, such as PDE3 or PDE4, are present, these latter PDEs, with their higher affinity for substrate, will take prime responsibility for hydrolyzing cAMP. In the absence of the cGMP-specific phosphodiesterase (PDE5), PDE2 may be primarily responsible for cGMP hydrolysis. In addition, PDE2 can respond to elevated cGMP by increasing its hydrolysis of cAMP, enabling crosstalk between the cAMP and cGMP signaling pathways. Examples of this phenomenon have been observed in a variety of cell types, including adrenal zona glomerulosa cells (MacFarland et al. 1991), frog ventricular cells (Meyer and Huxley 1992), and PC12 cells (Whalin et al. 1988). In bovine aortic endothelial cells, atrial natriuretic peptide (ANP), which activates guanylate cyclase and increases intracellular cGMP, was shown to reduce cAMP, presumably through activation of PDE2 (Kishi et al. 1994). Therefore, PDE2 may play a role in modulating both cAMP and cGMP levels in endothelial cells.
Inhibition of PDE2A in endothelial cells would be expected to increase intracellular levels of cAMP and/or cGMP. Elevation of cyclic nucleotide levels in arterial endothelial cells has been shown to improve barrier function (Suttorp et al. 1993, 1996a,b; Westendorp et al. 1994), to modulate expression of adhesion molecules (Pober et al. 1993; Ghersa et al. 1994; Morandini et al. 1996), and to modulate cell proliferation (D'Angelo et al. 1997). The role of cyclic nucleotides in modulating venous or capillary endothelial cell function is more ambiguous. Several studies suggest that cGMP analogues and NO donors can improve barrier function (Kurose et al. 1993) and modulate the expression of cell adhesion molecules (De Caterina et al. 1995) in venous and microvessel endothelial cells. For example, 8-bromo-cGMP attenuated L-NAME induced increases in leukocyte adhesion, vascular albumin leakage, and the permeability index in rat mesenteric venules (Kurose et al. 1993). However, in frog mesenteric capillaries, luminal exposure to sodium nitroprusside (SNP) or ANP had the opposite effect, increasing capillary hydraulic conductivity by several fold (Meyer and Huxley 1992). In summary, the role of PDE2A and cyclic nucleotides in endothelial cells remains to be clarified.
The expression of PDE2A is not limited to endothelial cells. As noted earlier, PDE2A is highly expressed in various parts of the nervous system (Repaske et al. 1993; Juilfs et al. 1997) and in the adrenal gland (MacFarland et al. 1991). In the studies reported here, PDE2A immunostaining was also detected in cardiac myocytes, specifically at the periphery of the cell. This immunostaining pattern became more apparent upon using fluorescent detection methods (unpublished observations) and was absent when the antibodies were preincubated with epitope specific peptides. Because the staining in the myocytes was not the focus of this manuscript, we chose not to include much detail regarding this observation. Although additional studies are needed to determine the significance of this observed cellular distribution, these results are consistent with the proposed role of PDE2 in modulating the L-type calcium channels of myocytes (Rivet-Bastide et al. 1997).
The differential expression of PDE2A in human endothelial cells clearly warrants further investigation. This necessitates potent inhibitors and appropriate model systems. At present, EHNA is the most selective PDE2 inhibitor available (Podzuweit et al. 1995). However, this compound is also a very potent inhibitor of adenosine deaminase, complicating its use. Therefore, development of potent and selective inhibitors of PDE2 would greatly facilitate studies aimed at understanding the functional role of PDE2 in endothelial cells. The observation that PDE2A is not readily detected in intact human arterial endothelial cells and is expressed at relatively low levels in cultured arterial endothelial cells suggests that studies using intact veins or microvessels might provide the most biologically relevant information regarding the role of this phosphodiesterase in endothelial cell function.
Footnotes
Acknowledgements
We thank Cathy Farrell for gel purification of the thioredoxin-PDE2A3 fusion protein used for antibody generation and for the sample of purified recombinant PDE7, Janelle Taylor from the ICOS Hybridoma facility for generation of the PDE2 monoclonal antibodies, Mike Gadau for purified recombinant PDE2A3 used as a positive control for Western analyses, and for purified recombinant PDE5, Lothar Uher for purified recombinant PDE1A-C, PDE4B-D, PDE7, and cell extracts of PDE3A, 3B, and 4A, James Esselstyn for Western analysis with the recombinant PDEs and cultured endothelial cells, and Janine Harrison for help with generating figures. We would also like to thank Drs Kate Loughney, Christopher Irons, and Ken Ferguson for critical review of the manuscript.