Multiple Cyclic Nucleotide Phosphodiesterases in Human Trabecular Meshwork Cells

Abstract

purpose. To characterize cyclic nucleotide phosphodiesterase isozyme activities in human trabecular meshwork cells and primary cultures of porcine trabecular meshwork cells.

methods. Radioimmunoassay of acetylated acid extracts was used to determine changes in cyclic adenosine monophosphate (cAMP) and cyclic quanosine monophosphate (cGMP) in human trabecular meshwork cells treated with phosphodiesterase isoform selective inhibitors. Cyclic nucleotide phosphodiesterase activities were measured using the two-step radioisotope procedure (Thompson). Enzyme activities in the supernatant of human cells were fractionated using anion-exchange chromatography. Additionally, human and porcine trabecular meshwork cell transcripts of phosphodiesterase family–specific isoforms were studied by reverse transcription-polymerase chain reaction and nucleotide sequencing.

results. In intact human cells, selective inhibitors for phosphodiesterase 4 (rolipram) and 5 (E4021) gene families were effective in augmenting cyclic nucleotide accumulation in response to isoproterenol or sodium nitroprusside, respectively. cAMP and cGMP hydrolytic activities, resolved using Trisacryl M anion-exchange chromatography, showed a cAMP phosphodiesterase peak that was minimally sensitivity to cGMP but modestly inhibited by rolipram and a cGMP phosphodiesterase peak that was sensitive to inhibition by E4021. Further evaluation of the cGMP phosphodiesterase demonstrated Michaelis-Menten kinetics and competitive inhibition by E4021. Messenger RNA transcripts for phosphodiesterase 4, 5, and 7 isozymes were isolated in human trabecular meshwork cells. However, in porcine trabecular meshwork cells only isozymes for phosphodiesterase 4 and 5 isozymes were detected.

conclusions. Human trabecular meshwork cells express phosphodiesterase 4, 5, and 7 gene family isoforms and enzyme activities, suggesting that selective isoform inhibitors could be used to augment the actions of antiglaucoma drugs that use cyclic nucleotides as second messengers.

Cyclic nucleotide phosphodiesterases (CN PDE) are enzymes responsible for terminating cyclic nucleotide activity in intact cells. Thus far, nine different families of CN PDE have been defined in a variety of tissues. The complexity of these multiple forms of PDE has required the differential investigation of the multiple activities in various tissues and cells. This expression of different complements of isozymes in different tissues suggests that pharmacological manipulation can be achieved in a relatively selective manner. 

Normal function of trabecular meshwork (TM) cells in the outflow tract of the eye is essential in the regulation of intraocular pressure (IOP). The activities of these endothelial-like cells are modulated by various signaling substances produced locally or carried to the TM by the aqueous humor. Receptors for neurohormones are located on TM cell membranes, and their activation alters the level of second messengers such as cyclic nucleotides. Antiglaucoma drugs, such as epinephrine, elevate cyclic adenosine monophosphate (cAMP) in TM cells and are believed to lower IOP by enhancing outflow of aqueous humor. ¹ Moreover, when given directly into the anterior chamber of experimental animals, ^{2 3} cAMP has been shown to promote the efflux of aqueous humor. An increase in the level of cyclic guanosine monophosphate (cGMP) by the action of nitric oxide is postulated to enhance outflow of aqueous humor by inducing relaxation of TM cells. ⁴ Additional evidence suggests that PDE inhibitors cause ocular hypotensive effects when used alone and in combination with agents that stimulate cyclic nucleotide synthesis. ⁵  

Because CN PDEs act beyond cell surface receptors to regulate the tissue levels of cyclic nucleotides, these isozymes provide potential pharmacological targets for manipulating IOP at the level of signal transduction. Therefore, the characterization of CN PDE isozymes in TM cells should define more completely the role of cyclic nucleotides in regulating resistance to outflow of aqueous humor and provide a more rational approach for glaucoma therapy. The studies described here have identified PDE isozyme transcripts in human trabecular meshwork (HTM-3) and porcine trabecular meshwork (PTM) cells. Specifically, the results demonstrate specific PDE activities of gene families 4, 5, and 7 in HTM-3 cells and PDE4 and PDE5 in PTM cells. 

Materials and Methods

Drugs/Chemicals

Cyclic-{8-³H}-AMP and cyclic-{8-³H}-GMP were purchased from Amersham (Amersham, Arlington Heights, IL). CGS9343B was purchased from CIBA Pharmaceutical (Summit, NJ). EHNA was purchased from RBI (Natick, MA). Rolipram was purchased from Berlex Laboratories (Cedar Knolls, NJ). E4021 was a generous gift from Eisai (Tokyo, Japan). Sodium nitroprusside (SNP), snake venom, cAMP, cGMP, and Trisacryl M anion-exchange chromatography materials were purchased from Sigma (St. Louis, MO). 

Cell Culture Conditions

HTM-3 cells (a generous gift from Iok–Hou Pang, Alcon Laboratories, Fort Worth, TX) were grown in Dulbecco’s minimal essential medium (GIBCO BRL, Grand Island, NY) supplemented with 10% (vol/vol) fetal calf serum (GIBCO BRL), 4 mM L-glutamine (GIBCO BRL), and 50 μg/ml gentamicin (GIBCO BRL) in a humidified incubator in the presence of 5% (vol/vol) CO₂ at 37°. Stock cultures were grown in T-flasks of 25-, 75-, or 225-cm² surface area (Fisher, Pittsburgh, PA), fed every 2 days, and subcultured every 5 to 7 days after treatment with 1× trypsin-EDTA (0.05% trypsin, 0.53 mM EDTA-4Na; GIBCO BRL). PTM cells were provided generously by Ramesh C. Tripathi (University of South Carolina School of Medicine, Columbus, SC). These cells were prepared and characterized as reported previously. ⁶  

CN PDE Inhibitor–Induced Cyclic Nucleotide Accumulation in Intact HTM-3 Cells

HTM-3 cells were grown to confluence in 6-well plates and preincubated for 30 minutes in serum-free Dulbecco’s minimal essential medium in the humidified incubator. Subsequently, the cells were treated with isoform-selective inhibitors cGS9343B (PDE1), EHNA (PDE2), indolidan (PDE3), and rolipram (PDE4) at concentrations ranging from 10⁻⁷ to 10⁻⁴ M at half-log increments and used along with 10⁻⁸ M isoproterenol in evaluating cAMP content of TM cells. Similarly, in studying the cGMP level in TM cells, HTM-3 cells were treated by using 10⁻² M SNP in the presence of cGS9343B, EHNA, and E4021 (PDE5) at concentrations covering full-log unit increments. Incubations were terminated after 15 minutes by adding 0.2 N HCl/50% methanol. After acid extracts were lyophilized and resuspended in radioimmunoassay (RIA) buffer, cAMP or cGMP concentrations were measured by RIA as described below. Triplicate determinations were made, and all experiments included vehicle controls. 

RIA of Cyclic Nucleotides

cAMP or cGMP levels of cells were determined by RIA using Amersham cAMP[¹²⁵I] or cGMP[¹²⁵I] assay systems. The system utilizes a high specific activity [¹²⁵I] 2′-0-succinyl-cAMP tyrosine methyl ester tracer, together with a highly specific and sensitive antiserum. Separation of the bound antibody from the free fraction was achieved with a second antibody Amerlex-M preparation and centrifugation. Standard curves ranging from 25 to 3200 fmol/per tube were obtained. Cell protein content was determined by Bradford assay (Bio-Rad), and levels of cAMP or cGMP were expressed as pmoles of cyclic nucleotide per milligram of protein. 

Cyclic Nucleotide Phosphodiesterase Assay, Kinetics, and Inhibitor Studies

HTM-3 cells were harvested mechanically with cell scraper and homogenized in 20 mM Tris–HCl/5 mM MgCl₂ (pH 7.4)/protease inhibitors buffer (TMPI, Tris–HCl/MgCl₂/protease inhibitors). The homogenate was centrifuged at 100,000g for 1 hour in a Beckman TLA-100.4F with a fixed-angle rotor at 4°C. CN PDE activities were determined in the homogenate, supernatant, and pellet using the modified two-step radioisotope procedure. ⁷ Enzyme assays were optimized to estimate the activity of each PDE isoform with the different substrate concentrations and various amounts of the enzymes. Incubation mixtures (0.4 ml) contained 40 mM Tris (pH 8.0), 10 mM magnesium acetate, 3.75 mM 2-mercaptoethanol, 30 μg of fatty acid–free bovine serum albumin, and 0.24 μM[ ³H]cAMP or 0.58 μM[ ³H]cGMP. Enzymatic reactions were incubated at 30°C for 20 minutes. Substrate concentrations of cAMP and cGMP were 0.24 and 0.58 μM, respectively, in the Trisacryl M column studies. The inhibition of fractionated PDE activity from a Triscryl M column was obtained using 10⁻⁴ M of E4021 or rolipram, respectively. The K _m for cGMP was determined from Lineweaver–Burk linear regression analysis using 3.0 × 10⁻⁶ to 1.0× 10⁻⁴ M cGMP. E4021 was used at concentration ranging from 10⁻¹⁰ to 10⁻⁴ M. Three E4021 concentrations (10⁻⁵, 10⁻⁴, and 10⁻³ M) and three cGMP concentrations (0.4 × 10⁻⁶, 1.2 × 10⁻⁶, and 2.0 × 10⁻⁶ M) were used to determine the K _i calculated from Dixon plots. 

DEAE Anion-Exchange Chromatography of HTM-3 Cell Supernatant

Ten confluent HTM-3 cell cultures (150-mm diameter dishes) were homogenized with 5 ml of TMPI buffer with a Teflon pestle grinder for 100 strokes. The 100,000g supernatant was applied to a Trisacryl M anion-exchange column of 15-ml bed volume at 1 ml/min loading rate. After the column was washed with 30 ml of homogenization buffer, PDE activities were eluted with two successive linear gradients of sodium acetate (NaOAC) (60 ml of 0–0.4 M and 60 ml of 0.4–1.6 M) also in homogenization buffer at a flow rate of 0.6 ml/min. The conductivity of each fraction was determined with a Conductivity Meter CDH-420 (Omega Engineering, Stanford, CT) to calculate the ionic strength along the elution profile. The pellet of the 100,000g centrifugation was resuspended with 5-ml TMPI for membrane preparation. 

CN PDE Isozyme–Specific Primer Design and Computer Analysis

CN PDE family–specific oligonucleotide primers were designed on the basis of comparison of all available sequences of cyclic nucleotide PDEs in GenBank. Those selected were analyzed with Oligo 5 primer design software for deficiencies. The nucleotide sequences of the sense and antisense oligonucleotides were as follows: PDE1, 5′ATCCACGACTATGAGCACACT3′ (sense), 5′TCCTTGTCACCCTGGCGGAAGAAT3′ (antisense); PDE2, 5′CCCAAAGTGGAGACTGTCTACACCTAC3′ (sense), 5′CTGGCCACAGTGCACCAAGATGA3′ (antisense); PDE3, 5′TCACCTCTCCAAGGGACTCCT3′(sense), 5′CAGCATGTAAAACATCAGTGGC3′ (antisense); PDE4, 5′ATGGT(TAG)GA(AG)AC(GC)AA(GA)AA(GA)GT3′ (sense), 5′AG(GA)TC(GCT)GCCCA(GAT)GT(GCT)TCCCA3′ (antisense); PDE5, 5′GTGAAAGATATTTCTAGTCACTTG3′ (sense), 5′ATACATGTAATTGATTCTGTTTGC3′ (antisense); PDE7, 5′GCTCTCTTCGGCTGCCCCAAT3′ (sense), 5′ACGAAGTTTCATCATATCTAA3′ (antisense). Programs (Fetch, Gap and Fasta) in the University of Wisconsin Genetics Computer Group (GCG) were used to analyze DNA sequences, and database searching occurred with a terminal linked to a Sun Workstation at the University of South Alabama. Complete nucleotide sequences were assembled with the GCG programs: Gelstart, Gelenter, and Gelassemble. 

Reverse Transcription–Polymerase Chain Reaction

Total RNA was isolated with RNAzol B method (Tel-Test, Friendswood, TX) from HTM-3, PTM, PC12 cells, and rat brain tissue. Poly(A) mRNAs were further extracted with Oligo(dT) column chromatography of Oligotex mRNA mini kit (QIAGEN, Chatsworth, CA). The integrity of the total RNA was examined with 1.2% agarose gel containing 1× MOPS and 1.1% formaldehyde showing clear 18S and 28S rRNA bands. The purity was determined by the ratio of OD_260/280. 

The poly(A) mRNAs were analyzed by reverse transcription–polymerase chain reaction (RT-PCR); the first-strand cDNAs were synthesized with AMV reverse transcriptase (Promega, Madison, WI), primed with oligo(dT)₁₅ primer at 42°C for 15 minutes, and further amplified with CN PDE family–specific primer sets. The PCR conditions were as follows: one cycle of 91°C for 4 minutes, 40 cycles of denaturing at 91°C for 1 minute, annealing for 1 minute at optimal temperature of each primer set (PDE1, 56°C; PDE2, 58°C; PDE3, 60°C; PDE4, 48°C; PDE5, 50°C, and PDE7, 49°C), and elongation at 72°C for 10 minutes. The reaction mixture contained 1.5 mM MgCl₂; 200 μM of dATP, dGTP, dTTP, and dCTP; 5% dimethyl sulfoxide; 50 ng each primer; 1× buffer and 1.25 U Taq polymerase (Fisher) in the total volume of 25 μl. 

RT–PCR products were cloned into the vector pCR2.1 (Invitrogen, San Diego, CA) and propagated in One Shot competent cells. Plasmid DNAs containing the RT–PCR fragments of HTM-3 and PTM cells were sequenced by the dideoxy chain termination method ¹¹ with AmpliTag DNA polymerase, fluorescent sequencing of ABI PRISM^T Dye Terminator Cycle Sequencing Core Kit on ABI 373A DNA Sequencer (Perkin Elmer Applied Biosystems, Foster City, CA). Positive control templates for PCR were rat brain tissue mRNA and BPDE1B1 (M94867) cDNA (kindly provided by J. A. Beavo, Department of Pharmacology, University of Washington, Seattle) for PDE1, PC12 cells (ATCC Cell Lines and Hybridomas, Rockville, MD), mRNA and RPDE2A1 (M94540) cDNA for PDE2, and HPDE3A1 (M91667) cDNA (generously provided by V. C. Manganiello, Pulmonary—Critical Care Medicine Branch, National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, MD) for PDE3. 

Statistical Analyses

The data were analyzed statistically using a Tukey–Kramer multiple comparison test. The difference between values were considered statistically significant when P < 0.05. 

Results

CN PDE Inhibitor–Induced Cyclic Nucleotide Accumulation in Intact HTM-3 Cells

HTM-3 cells used in this study have been demonstrated to have receptors and intracellular second messenger systems that respond to pharmacological agents. ⁸ The current studies were conducted to delineate the enzymes hydrolyzing cAMP and cGMP in HTM-3 cells. Therefore, the levels of cAMP and cGMP were measured in response to agonists that activate adenylyl and guanylyl cyclase activity in the presence of relatively selective phosphodiesterase isoform inhibitors. 

At 10⁻⁸ M isoproterenol, cAMP accumulation was determined in the absence and presence of cGS9343B (PDE1), EHNA (PDE2), indolidan (PDE3), and rolipram (PDE4) at 3 × 10⁻⁸ to 10⁻⁴ M. Although cGS9343B, EHNA, and indolidan failed to increase cAMP levels, the PDE4 inhibitor rolipram dose-dependently enhanced cAMP levels in response to the threshold dose of isoproterenol stimulation (Fig. 1) . The 50% of the maximum accumulation of cAMP content occurred at 30 ± 5 μM rolipram, which was the highest inhibitor concentration used in this particular intact cell system. 

At 10⁻⁴ M SNP, cGMP levels were determined in the presence of isoform selective inhibitors: cGS9343B (PDE1), EHNA (PDE2), or E4021 (PDE5) at 0.03 × 10⁻⁶ to 10^−-4 M (Fig. 2) . Note that SNP alone had no effect on cGMP levels in HTM-3 cells at concentrations of 10⁻⁶ to 10⁻² M. However, the PDE5 inhibitor E4021 enhanced the effects of SNP with an EC₅₀ of 5 ± 0.8 μM. In contrast, CGS9343B and EHNA did not augment cGMP levels in the presence of SNP. 

CN PDE Activities in HTM-3 Cells

Confluent HTM-3 cells were homogenized in TMPI buffer and further subjected to 100,000g ultracentrifugation as described in Materials and Methods. Seventy-two percent of the cAMP PDE activity was obtained from the supernatant; the remaining activity was located in the pellet. In contrast, 94% of cGMP PDE activity was detected in the supernatant; the remaining cGMP PDE activity was minimal and located in the pellet. The 100,000g supernatant was applied a Trisacryl M anion-exchange column. cGMP PDE was eluted with the linear 0 to 0.4 M NaOAC gradient at 0.19 M NaOAC in TMPI (pH 7.4) buffer, and cAMP PDE peak was eluted with 0.4 to 1.6 M NaOAC gradient at 0.40 M NaOAC (Fig. 3 A). The recoveries of activity from the columns for cGMP and cAMP PDEs were 71% and 59%, respectively. Figure 3A also shows the peak activities for cGMP and cAMP hydrolysis in the assay as determined in the presence of 100 μM E4021 or 100 μM rolipram, respectively. Although the majority of cGMP PDE activity was inhibited by E4021, only 26% cAMP PDE activity was sensitive to inhibition by rolipram. 

Figure 3B shows CN PDE activities as modified by E4021 and rolipram in the HTM-3 membrane fraction (100,000g pellet). Although 97% of cGMP PDE activities were inhibited by 100 μM E4021, only 48% of cAMP PDE activities were sensitive to inhibition by 100 μM rolipram. Inhibition by both PDE inhibitors was statistically significant. Therefore, these results aid in localizing the cGMP PDE activity and suggest that the isozyme present is PDE5 based on the specificity of E4021. On the other hand, the cAMP PDE activities in the membrane fraction were only partially due to PDE4 activity, suggesting that the rest of cAMP PDE activities were due to rolipram-insensitive cAMP PDE isoform. 

Subsequently, cAMP hydrolysis, measured in the presence of 5 μM cGMP, or cGMP hydrolysis, measured in the presence of 5 μM cAMP, showed no statistically significant inhibition or activation throughout the fractions of Trisacryl M anion-exchange chromatography and membrane preparation (data not shown). Therefore, these results suggest that cAMP hydrolysis in HTM cells is contributed to by isozymes from PDE4 and seven families and that cGMP hydrolysis was due to PDE5 isoforms. There were no detectable activities of PDE2 and PDE3. If present, they must be in minimal amounts. 

Kinetic Study of cGMP PDE Peak from Trisacryl M Column

cGMP hydrolysis in the fractions from the DEAE column showed a symmetrical peak whose kinetic behavior was characterized further. Initial velocity studies showed that cGMP PDE activity displayed Michaelis–Menten kinetic behavior. The apparent K _m of this enzyme for cGMP was 5.3 ± 1.0 μM as determined from a double-reciprocal Lineweaver–Burk plot. Subsequently, representative cGMP PDE inhibitors were used to study the pharmacological properties of the cGMP PDE peak. CGS9343B and EHNA (up to100 μM) showed minimal inhibition of cGMP hydrolysis. In contrast, E4021 showed dose-dependent inhibition with an IC₅₀ of 0.4 ± 0.1 μM (Fig. 4 , left); competitive inhibition as determined with a Dixon plot showed an apparent K _i = 0.6 ± 0.1 μM cGMP (Fig. 4 , right). 

RT–PCR–Mediated Identification of PDE Isozymes in HTM-3 Cells

PCR fragments were amplified from HTM-3 first-strand cDNAs with PDE isozyme–specific primers and showed PDE4, PDE5, and PDE7 products. No products were observed with PDE1, PDE2, or PDE3 primers. The relative size of the former bands in 1.2% agarose were estimated as 300, 600, 800, and 500 bp, respectively. The four bands for PDE4, PDE5, and PDE7 were reamplified for further analysis using the same primer sets from which they were originally generated. After PCR products were cloned and sequenced, the 600-bp product derived with PDE4 primers was shown to be an artifact by nucleotide sequencing. The other three respective fragments, which were generated by PDE4, PDE5, and PDE7 primers, contained: 305, 769, and 486 bp, respectively (Fig. 5) . Database homology searching with the Fasta program in GCG for the HTM-3 305-bp clone showed the following: 98% identity with human monocyte HSPDE4A4 (m37744), including a 305-bp overlap; 98% identity with human T-cell PDE4A (s75213), including 305-bp overlap; and 89.2% identity with rat brain PDE4 (M26715), including a 305-bp overlap. The HTM-3 769-bp clone showed 93% identity with bovine lung PDE5 (l16545), including 769-bp overlap. The HTM-3 486-bp clone showed 99% identity with human glioblastoma PDE7 (l12052), including a 486-bp overlap. 

RT–PCR–Facilitated Identification of PDE Isozymes in PTM Cells and Comparison with HTM-3 Cells

PTM cells were investigated by RT–PCR with PDE family–specific primers. Two products were identified: One was amplified with PDE4 primers, and the other one was amplified with PDE5 primers. The nucleotide sequences obtained were 303 bp and 769 bp in length and showed similarity (98%, 98%) with HTM-3 PDE4 and PDE5, respectively (data not shown). In contrast to HTM-3 results, there was no product observed with PDE7 primers (Table 1) . 

Discussion

Trabecular meshwork cells are located in the primary site of aqueous outflow from the anterior segment of the eye. Dysfunction of this filtration apparatus underlies the etiology of certain forms of open-angle glaucoma. ⁹ Therefore, rectifying the pathogenic process within the TM has been recognized as a potential therapeutic approach for ocular hypertension and glaucoma. Indeed, some ocular hypotensive agents, such as epinephrine, modulate aqueous outflow by improving the facility of aqueous outflow through the TM. ¹ However, the exact cellular mechanism responsible for the regulation of IOP by the TM is not well understood. For example, the quantity, and the exact chemical components of the extracellular matrix produced by these cells, may affect the resistance to fluid flow through the trabecular structure. ^{10 11 12} Moreover, the phagocytic actions of TM cells may contribute to the clearance of debris that, if not removed, can block the meshwork. ^{13 14 15 16}  

Several functions of the TM cells are hypothesized to be involved in the action of outflow-enhancing drugs. Recently, the TM has been discovered to have contractile elements. ^{17 18 19} Endothelin-1 and muscarinic and adrenergic agonists induce contraction of the TM, ^{20 21} suggesting that mechanical tension of cells within the outflow tract may be another means of regulating aqueous humor hydrodynamics. Therefore, understanding the physiologic and biochemical actions of agents within these TM cells will provide crucial insights into the control of IOP, and ultimately, the treatment of glaucoma. 

The small amount of TM tissue that can be obtained from a single eye impedes the study of its biology at the cellular level. However, the establishment and propagation of cultured TM cells ^{22 23 24} have aided in solving some of these difficulties, but the relatively slow growth rate of these cells has limited biochemical and pharmacological experimentation. HTM-3 cells were derived by transforming TM cells of a male glaucoma patient with an origin defective mutant of SV40 virus. ⁸ Pharmacological characterization of transformed TM cells has shown expression of several components of critical signal transduction systems. The results of this study demonstrate that HTM-3 cells show multiple PDE isozymes activities, namely PDE4, 5, and 7 gene families, thereby providing the first characterization of CN PDE isozymes in TM cells. PDE4 has been found in many cells studied, but the finding of a rolipram-insensitive high affinity cAMP PDE (PDE7) in the human eye was an interesting finding. Moreover, the expression of cGMP PDE (PDE5) in cells composing the outflow tract of the eye may prove to be pivotal in defining cyclic nucleotide–mediated signal transduction in TM cells because this isoform has a more limited tissue distribution. 

The isolation and verification of PDE4 and PDE5 from primary porcine TM cells make a significant contribution to the data derived from HTM-3 cells in the sense of demonstrating similarities between two species (human, porcine) and between primary cells versus a virally transformed cell line. High homology of PDE4 and PDE5 nucleotide sequences between HTM-3 and PTM cells suggests that there are few differences between these species within the same isozyme family and within the selected PDE primer regions. These results are consistent with the initial idea for the primer design. Therefore, representing the TM’s primary machinery for cAMP and cGMP hydrolysis. In addition, the detection of PDE7 in HTM-3, but not in PTM cells, could be due to: (a) PDE7 gene expression unique to human trabecular meshwork; (b) a product of the disease stage; (c) a lack of cross-reactivity of human PDE7 primers with the porcine isoform. Because HTM-3 cells are thought to be representative of primary TM tissue in terms of biochemical and pharmacological characteristics, it is most likely that PDE7 is an isozyme uniquely transcribed in either the normal or a diseased stage of human trabecular meshwork. 

In intact HTM-3 cells, cAMP and cGMP accumulation in response to phosphodiesterase inhibitors and agonists that activate cyclases suggested that PDE4 and 5 isoforms might serve as intracellular detectable receptors for cAMP, cGMP, or both. In general, PDE inhibitors are at least 30-fold selective for the PDE against which they are targeted. Most are substrate-site–directed competitive inhibitors, but a few act at allosteric sites. Selective inhibitors of PDE7 have yet to be identified. In fact this enzyme is resistant to all standard PDE inhibitors, including nonselective compounds such as isobutylmethylxanthine (IBMX). ²⁵ In these studies, selective inhibition of PDE4 by rolipram caused a more robust cAMP accumulation in the presence of isoproterenol. No enhancement of cAMP accumulation was observed in this system in response to cGS9343B, EHNA, or indolidan, inhibitors of PDE1, 2, and 3 gene families, respectively. These results suggest a functional coupling between β-adrenergic receptors and cAMP hydrolysis and demonstrate that PDE4 participates the regulation of intracellular cAMP homeostasis in HTM-3 cells. Because no PDE7 inhibitors are available, the role PDE7 may play in cAMP metabolism remains to be determined. In addition, 100 μM rolipram may target other cAMP PDE isozymes, for instance, PDE7. ²⁶ In our previous studies, the nonselective inhibitor IBMX had a modest effect on cAMP accumulation; therefore, the cAMP hydrolysis in HTM-3 cells may be due in part to PDE8 activity because PDE8, a new cAMP-specific phosphodiesterase, is insensitive to IBMX. ^{27 28 29}  

PDE5 plays an important role in maintaining cGMP homeostasis in HTM-3 cells. E4021, a selective PDE5 inhibitor, increased cGMP content in response to SNP in a dose-dependent manner in our early studies. Sodium nitroprusside promotes formation of NO, which activates cytosolic soluble guanylyl cyclase generating cGMP from GTP. In HTM-3 cells, no increase in cGMP content was found up to 10⁻² M SNP, suggesting either limited activity of guanylyl cyclase or high cGMP PDE activities in these cells. The initial rate study of the cGMP PDE peak obeyed Michaelis–Menten kinetics. E4021 showed dose-dependent inhibition with an IC₅₀ of 0.4 ± 0.1 μM compared to 1.3 ± 0.3 μM of the unpurified 100,000g supernatant study. The kinetics of E4021 inhibition at different cGMP concentrations converged at one point in the left upper quadrant, indicating competitive interaction between E4021 and cGMP for the binding sites within PDE5 enzyme molecule. The K _i for E4021 resolved from these studies is approximately 100-fold greater than reported for PDE5 in the literature; ³⁰ the explanation could be that the cell preparation for enzymatic assay has been only partially purified with anion-exchange chromatography, or during this preparation, the enzyme was not at optimal conformation stage. Because the majority cGMP PDE activities (97%) were sensitive to the inhibition of E4021, cGMP PDE activities in HTM-3 cells were attributable to PDE5 based on E4021’s specificity. However, at a high enough concentration, E4021 may crossover and inhibit other isozymes ³⁰ ; therefore, there is the possibility of other cGMP PDEs participating in cGMP hydrolysis in HTM-3 cells as well, such as PDE9. ^{31 32} With the elevation of cGMP content through PDE5 inhibition, cGMP-mediated events such as modulation of protein kinase, ³³ regulation of calcium channel activities, ³⁴ inhibition of PDE3/activation of PDE2, ³⁵ or alteration of cGMP-gated Na channels ³⁶ can be studied. 

The findings described above were corroborated by the isolation of PDE4, PDE5, and PDE7 cDNAs from HTM-3 cells by RT–PCR and molecular cloning techniques. These isolates showed a high percentage of homology with the authentic PDE families at the nucleotide level. It was verified that the clones from HTM-3 cells contained sequences of RNA for PDE4, PDE5, and PDE7. PDE family–specific primers for PDE1, PDE2, and PDE3 were negative for the respective PDE isozymes in HTM-3 cells. These primers were able to identify positive signals in the appropriate templates, including BPDE1B1 and rat brain mRNA, RPDE2A plasmid and PC12 mRNA for PDE2, and HSPDE3 for PDE3. In the case of PDE4, RT–PCR products showed two bands recovered with the degenerate, family-specific primers. However, only the 300-bp isolate proved to be the PDE4 gene transcript because the 600-bp product proved to be satellite DNA artifact by nucleotide sequencing verification. 

Computer alignments of the fragments showed that the HTM-3 PDE4 fragment was located with the conserved catalytic domain in PDE4 enzyme molecule. This region of PDE4 most likely accounts for rolipram binding and catalytic activity as reported by Torphy et al ³⁷ from human monocyte PDE4 (HSPDE4A4, m37744). The HTM-3 PDE5 fragment is located in the noncatalytic regulatory region of the PDE5 molecule. The noncatalytic cGMP binding to PDE2 and PDE5 has been extensively studied using cGMP analogues. ³⁸ A general noncatalytic PDE-binding motif for cGMP appears to include constraint about the 8-position and positive binding interactions at the 6-oxo, N1-nitrogen, and 2-amino moieties and with the 2′ hydroxyl group of the ribose cyclic phosphate. The PDE7 fragment of HTM-3 cells was located in the putative regulatory domain in the N-terminal of the human glioblastoma PDE7 molecule. 

These data provide the first biochemical and molecular evidence of the existence of multiple isozymes of CN PDEs in HTM-3 and PTM cells. The results indicate the presence of rolipram-insensitive cAMP PDE 7 and rolipram-sensitive cAMP PDE4 along with E4021-sensitive PDE5 in HTM-3 cells. PTM cells contained PDE4 and PDE5. These findings can be used to better understand the signal transduction pathways in trabecular meshwork cells under normal and pathophysiological conditions. The potential application includes a rationale for directing therapy of glaucoma to the molecular level. 

 Reprint requests: Ling Zhou, Department of Pharmacology and Toxicology, Morehouse School of Medicine, 720 Westview Drive SW, Atlanta, GA 30310.

 Submitted as partial fulfillment for a doctoral degree (LZ).

 Supported by Grants EY06338 from the National Eye Institute, G12RR03034 from the National Institute of General Medicinal Sciences, (DEP) and Grant HL46494 from the National Heart, Lung and Blood Institute (WJT).

 Submitted for publication May 20, 1998; revised November 30, 1998; accepted January 11, 1999.

 Proprietary interest category: N.

 

Figure 1.

 

cAMP response to isoproterenol in the presence of isoform-selective PDE inhibitors in intact HTM-3 cells. cAMP contents in HTM-3 cells were measured by RIA in the presence of rolipram (closed circle), cGS9342B (closed square), EHNA (closed triangle), and indolidan (closed inverted triangle). Three replicates were conducted in each experiment, and the study has been repeated three times. The data were analyzed statistically with Student’s t-test.

Figure 2.

 

Sodium nitroprusside response in intact HTM-3 cells in the presence of isoform-selective PDE inhibitors. cGMP contents in HTM-3 cells were measured by RIA in the presence of E4021 (closed circle), cGS 9343B (closed diamond), and EHNA (closed triangle). Three replicates were conducted in each experiment, and the study was repeated three times. The data were analyzed statistically with Student’s t-test.

Figure 3.

 

(A) Trisacryl M anion-exchange fractionation of HTM-3 CN PDE activities. CN PDE activities were measured at 0.58 μM cGMP as substrate (closed circle), or presence of 100 μM E4021 (open circle); 0.24 μM cAMP as substrate (closed triangles), or presence of 100 μM rolipram (open triangles). The dashed line with no symbols shows NaOAC gradients from 0 to 1.6 M. (B) Sensitivities of CN PDEs in membrane fraction to E4021 or rolipram. HTM-3 membranes were prepared as described in Methods. The hatched bars represent cGMP or cAMP PDE activities with 0.58 μM cGMP or 0.24 μM cAMP as substrates, respectively. The closed gray bar represents the cGMP PDE activity in the presence of 100 μM E4021. The closed black bar represents cAMP PDE activities in the presence of 100 μM rolipram. Three replicates were conducted in each experiment, and the study was repeated three times. *P < 0.05.

Figure 4.

 

Inhibition of Trisacryl M HTM-3 cGMP PDE pool by selective PDE inhibitors. Activity analysis of the cGMP PDE pool was performed as described in the Methods section in the presence of cGS9343B, EHNA, and E4021 (left panel). IC₅₀ values were determined using power series in GraphPad Prism. Also shown on the right panel, Dixon plot of E4021 inhibition using cGMP concentrations of 0.4, 1.2, and 2 μM K _i = 0.6 ± 0.1 μM with Michaelis–Menten behavior.

Figure 5.

 

RT–PCR of HTM-3 cell mRNA using PDE family–specific primers. Lane 1, 100-bp molecular marker; lane 2, PDE1B1 was positive cDNA control for PDE1; lane 3, rat brain tissue served as positive tissue for PDE1; lane 4, HTM-3 detected by PDE1 primers; lane 5, PDE2A was positive cDNA control for PDE2; lane 6, PC12 cells were positive tissue control for PDE2; lane 7, HTM-3 detected by PDE2 primers; lane 8, human cardiac (HCAR) was positive cDNA control for PDE3; lane 9, HTM-3 detected by PDE3 primers; lane 10, HTM-3 detected by PDE4 primers; lane 11, HTM-3 detected by PDE5 primers; and lane 12, HTM-3 detected by PDE7 primers.

Table 1.

 

Table 1.

 

Summary of PDE mRNA in Trabecular Meshwork

	HTM-3	PTM	Pair Identity*	Identity with Authentic PDE Isoform
PDE families	PDE4 (305 bp)	PDE4 (303 bp)	98%	97% to 98% human monocyte PDE4A (m37744)
	PDE5 (769 bp)	PDE5 (769 bp)	98%	93% bovine lung PDE5 (116545)
	PDE7 (486 bp)	ND, #		99% human glioblastoma PDE7 (112052)

*  Pair identity was defined as the comparison of the same PDE isozyme between species.

#  Not detectable.