Salbutamol up-regulates PDE4 activity and induces a heterologous desensitization of U937 cells to prostaglandin E2. Implications for the therapeutic use of beta-adrenoceptor agonists.

Previous studies with U937 cells, a human monocyte cell line, have shown that the activity of cyclic nucleotide phosphodiesterase 4 (PDE4) is increased by agents that elevate cyclic AMP content. The present experiments were conducted to determine 1) whether an increase in PDE4 steady-state message and/or protein accompanies the up-regulation of PDE4 activity and 2) whether the up-regulation changes the functional responses of U937 cells to activators of adenylyl cyclase. To up-regulate PDE4 activity, U937 cells were treated for 4 h with a combination of 1 μM salbutamol, a β-adrenoceptor agonist, and 30 μM rolipram, a PDE4 inhibitor. Cells were washed extensively to remove drugs and used immediately in various experimental protocols. Reverse transcriptase-polymerase chain reactions conducted with primers specific for the four PDE4 subtypes suggested that pretreatment with salbutamol and rolipram increased steady-state mRNA levels of PDE4A and PDE4B, but not PDE4C or PDE4D. Immunoblot analyses using two rabbit polyclonal antibodies, one directed against human recombinant PDE4A and PDE4D and a second directed against human recombinant PDE4B, revealed bands of immunoreactivity corresponding to ∼125 kDa (PDE4A) and ∼70 kDa (PDE4B), respectively, that increased in intensity after cells were treated with salbutamol and rolipram. As demonstrated in both time course and concentration-response studies with prostaglandin E2 (PGE2), an agent that activates adenylyl cyclase by a non-β-adrenoceptor-mediated mechanism, cAMP accumulation was substantially decreased in cells in which PDE4 activity had been up-regulated. The difference in PGE2-stimulated cAMP accumulation between control and PDE4 up-regulated cells was greatly reduced in the presence of rolipram, consistent with the notion that an increase in PDE4 activity was responsible for the heterologous desensitization. Functionally, up-regulation of PDE4 markedly decreased the ability of PGE2 to inhibit LTD4-induced Ca2+ mobilization in intact cells. A hypothetical implication of these results is that increasing PDE4 activity in vivo by administering β-adrenoceptor agonists could exacerbate inflammatory processes by decreasing the activity of endogenous anti-inflammatory agents such as PGE2.

Cyclic nucleotide phosphodiesterases (PDEs) 1 are a family of isozymes that catalyze the hydrolysis of the 3Ј-phosphoester bond on adenosine cyclic 3Ј,5Ј-monophosphate (cAMP) and guanosine cyclic 3Ј,5Ј-monophosphate to form the inactive 5Јmonophosphate products. Consequently, PDEs have a major role in regulating cellular cyclic nucleotide content. It is now recognized that PDEs represent a diverse family of isozymes, each with different kinetic and physical characteristics, tissue distribution, and sensitivity to endogenous regulators (Beavo, 1988). At least seven classes of PDE isozymes exist, some of which contain multiple subtypes Beavo et al., 1994). All of these isozymes as well as many of the subtypes are encoded by distinct genes (Beavo et al., 1994).
The cAMP-specific PDE, designated PDE4 (for nomenclature see Beavo et al., 1994), is the predominant cAMP hydrolyzing isozyme class found in most, if not all, immune and inflammatory cells (Torphy and Undem, 1991;Giembycz and Dent, 1992). This, coupled with the well defined role of cAMP as a second messenger mediating a generalized suppression of immune and inflammatory cell activity (Bourne et al., 1973(Bourne et al., , 1974Kammer 1988;), has led to the recognition that PDE4 plays a critical role in regulating the function of these cells (Torphy and Undem, 1991). Consequently, this isozyme has received considerable attention as a target for new antiinflammatory and immunomodulator drugs.
The PDE4 isozyme class is comprised of four subtypes, PDE4A through PDE4D (Beavo et al., 1994). The activity of certain subtypes can be increased in situ by either a short term regulatory process involving protein phosphorylation or by a long term regulatory process involving increased gene expression Sette et al., 1994aSette et al., , 1994b. Both of these regulatory mechanisms are cAMP-dependent and can be triggered by a variety of activators of adenylyl cyclase (Bourne et al., 1973;Conti et al., 1991;Torphy et al., 1992). This raises the possibility that PDE4 activity can be regulated in vivo by various hormones, drugs, and growth factors . Because of the predominance of PDE4 in immune and inflammatory cells, these regulatory pathways may have a substantial influence on the responsiveness of these cells to a variety of hormones, autacoids, and drugs.
We previously reported that treatment of undifferentiated U937 cells, a human monocytic cell line, with salbutamol, a ␤-adrenoceptor agonist, or prostaglandin (PG) E 2 produces a 2-4-fold increase in the activity of PDE4, the major PDE in these cells (Torphy et al., 1992). This increase in activity develops after 2-4 h of agonist exposure and is: 1) preceded by an increase in cAMP content and cAMP-dependent protein kinase activity; 2) potentiated by cotreatment with rolipram, a PDE4 inhibitor; 3) mimicked by 8-bromo-cAMP; 4) marked by an increase in V max with no change in the cAMP K m ; 5) abolished by inhibitors of mRNA or protein synthesis; and 6) reversible within 3 h of agonist removal. In light of these results, the present experiments were conducted to determine whether treatment of U937 cells with ␤-adrenoceptor agonists increases the steady-state levels of PDE4 protein or transcript and, if so, whether this phenomenon changes the functional responsiveness of these cells to activators of adenylyl cyclase.

EXPERIMENTAL PROCEDURES
Cell Culture and Drug Pretreatments-U937 cells from the American Type Culture Collection were grown in plastic flasks (80 cm 2 ) in Roswell Park Memorial Institute (RPMI)-1640 medium supplemented with 10% (v/v) heat-inactivated fetal bovine serum at 37°C in a humidified atmosphere of 95% air, 5% CO 2 . Culture medium was changed every 2-3 days and always 24 h before harvest. Cells were seeded at densities of 0.075-0.15 ϫ 10 6 cells/ml and harvested at 0.75-1 ϫ 10 6 cells/ml. The cells were preincubated with vehicle (distilled water for salbutamol, 0.03% dimethyl sulfoxide for rolipram), or 1 M salbutamol plus 30 M rolipram for 4 h. Although treatment with salbutamol alone increases PDE4 activity, the magnitude is greater using the combination of an adenylyl cyclase activator and PDE4 inhibitor (Torphy et al., 1992). Consequently, the combination protocol was used in the present experiments to increase the opportunity to detect differences between control and PDE4-up-regulated cells.
To prepare human monocytes, heparinized whole blood was centrifuged at 350 ϫ g for 30 min over Ficoll-Hypaque to remove erythrocytes and granulocytes. The mononuclear cell layer was removed, washed twice with Ca 2ϩ -and Mg 2ϩ -free Hanks' balanced salt solution containing 1 mM EGTA, and resuspended in RPMI 1640 medium supplemented with 10% heat-inactivated fetal bovine serum, 2.5 mM HEPES, 20 mM L-glutamine, 100 U penicillin/ml, and 100 mg streptomycin/ml (RPMI). The suspension was underlaid with iso-osmotic Percoll adjusted to a density of 1.062 g/cm 3 with RPMI and centrifuged at 560 ϫ g for 30 min. Monocytes were harvested from the interface, washed twice with HBS, and resuspended in RPMI. Cells were incubated overnight at a density of 0.5-1 ϫ 10 6 cells/ml in 175-cm 2 flasks in a humidified, 37°C atmosphere of 95% air, 5% CO 2 .
When PGE 2 -induced cAMP accumulation was determined, control and "PDE4-induced" cells were centrifuged at 500 ϫ g for 5 min and washed two times with cold Krebs-Ringer-Henseleit buffer composed of 118 mM NaCl; 4.6 mM KCl, 24.9 mM NaHCO 3 , 1 mM KH 2 PO 4 , 11.1 mM D-glucose, 1 mM CaCl 2 , 1.1 mM MgCl 2 , and 5 mM HEPES, pH 7.4. Cells were then resuspended at a final concentration of 10 6 cells/ml in Krebs-Ringer-Henseleit buffer containing 0.1% bovine serum albumin (radioimmunoassay grade) and treated for the indicated times with different concentrations of PGE 2 in the absence or presence of 30 M rolipram.
Immunoblot Analysis-Following their exposure to various pharmacological agents, cells were harvested by centrifugation. Cells were washed twice (10 min, 400 ϫ g) with 20 mM phosphate-buffered saline (PBS), counted, and resuspended at 2 ϫ 10 8 cells/ml in 50 mM Tris homogenization buffer, pH 7.4, containing 10 mM EDTA, 2 mM EGTA, 2 mM benzamidine, 2.5 mM dithiothreitol, 2 g/ml soybean trypsin inhibitor, 100 M tosyl-L-lysine chloromethyl ketone, 200 M leupeptin, and 50 M phenylmethylsulfonyl fluoride. Suspensions were frozen in a methanol/dry ice bath and quickly thawed in a 37°C water bath for three cycles to lyse the cells. The lysed suspensions were centrifuged for 5 min at 14,000 revolutions/min in an Eppendorf microfuge to remove nuclei and cellular debris. Supernatants were quick-frozen and stored at Ϫ80°C.
Protein concentrations were determined using the Bio-Rad (modified Bradford) protein assay. Proteins (ϳ100 g/lane) were separated via electrophoresis (Bio-Rad) on SDS-8% polyacrylamide gels and electrophoretically transferred to nitrocellulose membrane (Amersham, Buckinghamshire, United Kingdom) using a tank electroblotter. Blots were briefly washed in PBS, 0.1% Tween-20 and then blocked overnight in PBS/Tween-20/5% nonfat dry milk. Blocked membranes were washed three times with PBS/Tween-20 before incubation with the primary antibody.
Blots were incubated for 1 h with a 1:2000 dilution of polyclonal serum produced in New Zealand White rabbits. One antibody was raised against a galK-hPDE-1 fusion protein containing a major fragment of PDE4A, which included the conserved PDE4 catalytic domain (Livi et al., 1990). Preliminary characterization of this antibody indicated cross-reactivity with PDE4D, but not PDE4B (recombinant PDE4C is not available). A second antibody was raised against a peptide representing the unique carboxyl terminus of PDE4B2 (CDIDIAT-EDKSPVDT). For blocked antibody experiments, the antibody was diluted 1:100 into a lysate of either the nontransfected PDE-deficient Saccharomyces cerevisiae strain GL62 (control) or the same strain engineered to express a 686-residue fragment of human recombinant PDE4A (hrPDE4A) containing the conserved catalytic domain of PDE4. The mixtures were incubated overnight at 4°C with gentle agitation. The antibody was used at a final dilution of 1:2000 in PBS/Tween-20, 1% nonfat dry milk.
After three washes with PBS/Tween-20 blots were then incubated for 1 h with horseradish peroxidase-linked anti-rabbit Ig whole antibody from donkey (Amersham) and washed five times with PBS/Tween-20. Immunoreactive proteins were detected by chemiluminescence (Amersham ECL reagents).
cAMP Assay-To stop the incubation, 900-l aliquots of incubation medium containing ϳ1 ϫ 10 6 cells were added to test tubes containing 100 l of 100% trichloroacetic acid and 4000 counts/min of [ 3 H]cAMP (added as a tracer). The samples were stored at Ϫ20°C before being assayed for cAMP content. When cAMP content was determined, the precipitated protein was separated from the soluble extract by centrifugation at 3000 ϫ g for 10 min. Trichloroacetic acid was removed with five successive extractions with water-saturated ether (Brooker et al. 1979). Cyclic AMP was measured following acetylation using commercially available radioimmunoassay kits (DuPont NEN). Cyclic AMP content was corrected for percentage of recovery (85-95%) and expressed as pmol of cAMP/10 6 cells.
Calcium Mobilization-After preincubating U937 cells with vehicle or 1 M salbutamol and 30 M rolipram for 3.25 h, the cells were centrifuged at 200 ϫ g for 5 min and resuspended to a concentration of 2 ϫ 10 6 cells/ml in Krebs-Ringer-Henseleit buffer (plus 0.1% BSA) containing vehicle (control cells) or1 M salbutamol plus 30 M rolipram (treated cells). Cells were then exposed to 2 M fura-2/AM, the pentacetoxymethyl ester of fura-2, at 32°C for 30 min. After the 30-min incubation period, cells were washed by centrifugation and resuspended in Krebs-Ringer-Henseleit buffer (plus 0.1% BSA) containing vehicle (control cells) or 1 M salbutamol and 30 M rolipram (treated cells) at 32°C for an additional 15 min to allow complete hydrolysis of the intracellular fura-2 ester. Using this protocol cells were exposed to vehicle or a combination of salbutamol and rolipram for a total of 4 h before Ca 2ϩ mobilization was assessed. Cells were centrifuged and washed two times with cold Krebs-Ringer-Henseleit buffer (plus 0.1% BSA) and kept at room temperature until used for fluorescence determination as described previously (Winkler et al., 1988). Cells were exposed to vehicle or rolipram (10 M) for 1 min before the addition of various concentrations of PGE 2 . One min after the addition of PGE 2 , cells were challenged with 3.3 nM leukotriene D 4 (LTD 4 ), and Ca 2ϩ mobilization was monitored and quantified via fura-2 fluorescence (Winkler et al., 1988).
Data Analysis and Statistical Evaluation-Data are expressed as the mean Ϯ standard error. Significant differences between two means were determined by an unpaired or paired Student's t test. Differences among the means of more than two groups were determined using analysis of variance and a Newman-Keuls multiple comparison test. Differences among means were accepted as significant at p Ͻ 0.05.
Materials-hrPDE4A, hrPDE4B, and hrPDE4D for use as standards were kindly prepared and provided by G. Livi and M. McLaughlin (SmithKline Beecham Pharmaceuticals, King of Prussia, PA). Racemic salbutamol (albuterol), (Ϯ)-rolipram, and LTD 4 were obtained from SmithKline Beecham Pharmaceuticals. Dithiothreitol was obtained from Boehringer Mannheim. PGE 2 and all other reagents were obtained from Sigma. Cyclic AMP radioimmunoassay kits were obtained from DuPont NEN. PGE 2 and (Ϯ)-rolipram were dissolved in dimethyl sulfoxide as stock solutions (10 and 100 mM, respectively). The final concentration of dimethyl sulfoxide never exceeded 0.1%. All other drugs were prepared as aqueous solutions.

RESULTS
PDE4 Subtype Gene Expression-PDE4 subtype-specific RT-PCR products in control and PDE4-up-regulated cells are shown in Fig. 1. Transcripts for PDE4B and PDE4D were detected in untreated cells, whereas only a faint band corresponding to PDE4A was apparent. After treating cells for 4 h with a combination of salbutamol (1 M) and rolipram (30 M), conditions previously shown to produce a 2-4-fold increase in PDE4 catalytic activity (Torphy et al., 1992), the intensity of the bands corresponding to the PCR products for PDE4A and PDE4B increased. Interestingly, the PCR product derived from PDE4D message appeared to decrease. Transcript for PDE4C was not detected using this method. Equivalent intensities of the ␤-actin-specific bands were seen in control versus treated cells, confirming successful normalization of the mRNA. A PCR product for ␤-actin was not detected in the absence of reverse transcriptase, indicating that the RNA preparation was not contaminated with intact DNA.
PDE4 Immunoreactivity-Western analyses of PDE4 immunoreactivity were carried out using a polyclonal antibody raised against a purified GalK-hrPDE4A fusion protein or against a synthetic peptide unique to PDE4B. An immunblot represent-ative of results from four experiments using the GalK-hrPDE4A antiserum is shown in Fig. 2. In none of the experiments was significant immunoreactivity detected that corresponded to a large molecular mass protein (ϳ60 -120 kDa) in untreated U937 cells ( Fig. 2A). However, after 4 h of treatment with salbutamol and rolipram a band of immunoreactivity was consistently observed. The band corresponded to a molecular mass of approximately 125-kDa, identical in size to a band of immunoreactivity detected in human blood monocytes. As expected, the antibody also bound to a truncated form of hrPDE4A that lacks 264 N-terminal amino acids. To demonstrate that the 125-kDa protein was indeed PDE4, studies were conducted in which the antibody was saturated with the truncated hrPDE4A before it was exposed to the immunoblot (Fig.  2B). Preabsorbing the antibody in this manner virtually abolished its ability to detect either hrPDE4A or U937 cell PDE4. Note that the preabsorption procedure also reduced or eliminated the immunoreactivity directed against several other minor bands in U937 cells, suggesting that these proteins represented fragments of PDE4.
The antibody raised against the GalK-hrPDE4 also detected hrPDE4D, but not hrPDE4B (Fig. 3A). Although a faint band of immunoreactivity in U937 cells was observed that corresponded to the appropriate molecular mass for PDE4D (ϳ92 kDa), its intensity was not increased in PDE4-up-regulated cells (Fig. 3A).
An immunoblot produced with the antiserum raised against the PDE4B peptide is shown in Fig. 3B. Two bands of immunoreactivity were detected with hrPDE4B, a major band at ϳ70 kDa and a minor band at ϳ52 kDa. Several bands were detected in U937 cells, but only those corresponding to 70 and 52 kDa were increased in intensity after treating cells with a combination of rolipram and salbutamol. Preabsorbing the antiserum with PDE4B eliminated these two bands, but not the others.
Effect of Up-regulating PDE4 Activity on PGE 2 -stimulated FIG. 1. RT-PCR of transcripts for PDE4 subtypes in U937 cells. U937 cells were untreated or exposed to 1 M salbutamol and 30 M rolipram for 4 h. After the 4-h treatment period, RNA was prepared, reverse transcribed into cDNA, and amplified using primer sets specific for the four subtypes of PDE4. The PCR products were electrophoresed on 3% agarose gels and visualized with ethidium bromide. RNA normalization for control and treated samples was confirmed by conducting RT-PCR reactions using primers for ␤-actin RNA (ϩ). Lack of DNA contamination was confirmed by conducting the ␤-actin reaction in the absence of RT (Ϫ). DNA molecular mass standards appear in the lane on the far left. The results are representative of three experiments using material from three different cell preparations.
FIG. 2. Western blot analysis of PDE4 immunoreactivity in U937 cells and human monocytes using antibody raised against GalK-hrPDE4A. Panel A, U937 cells were untreated or exposed to 1 M salbutamol and 30 M rolipram for 4 h. After the treatment period cells were washed extensively and lysed. Supernatant fractions were then prepared and identical amounts of protein (ϳ100 g) were subjected to SDS-polyacrylamide gel electrophoresis. Also run in these studies were untreated human monocyte supernatants and lysates from yeast engineered to express a truncated form of hrPDE4A that lacked 265 amino acids on the N terminus. The data are representative of results from four (U937 cells) or three (monocytes) experiments. Panel B, additional experiments were conducted to confirm that the antibody was detecting PDE4 protein in U937 cells. In these studies, the antibody was diluted 1:100 into a lysate of either nontransfected yeast (control) or yeast that had been engineered to produce the truncated form of hrPDE4A (blocked). The following day, hrPDE4A (panel B, left) and treated (1 M salbutamol plus 30 M rolipram) U937 cell supernatants (panel B, right) were run on SDS-polyacrylamide gels which were then transferred to nitrocellulose membranes and probed with the control or preadsorbed antibody.
cAMP Accumulation-To determine whether up-regulating PDE4 activity had an impact on agonist-induced cAMP accumulation, U937 cells were treated for 4 h with vehicle or a combination of salbutamol (1 M) and rolipram (30 M). At the end of the 4-h pretreatment period, cells were washed extensively to remove drugs before being resuspended in drug-free Krebs-Ringer-Henseleit buffer. The cells were then exposed to 1 M of PGE 2 and cAMP was measured at various times over a 15-min period (Fig. 4). In vehicle-pretreated cells, basal cAMP content (1.7 Ϯ 0.2 pmol/10 6 cells) was increased by 25-fold (41.5 Ϯ 7.8 pmol/10 6 cells) within 2 min of being exposed to PGE 2 . In cells in which PDE4 activity had been up-regulated, basal cAMP content (2.4 Ϯ 0.3 pmol/10 6 cells) was not significantly different from control cells. However, PGE 2 -stimulated cAMP accumulation was substantially less, reaching a maximum of only 23.3 Ϯ 7.3 pmol/10 6 cells within 2 min. A pattern virtually identical to this was observed in separate studies using a larger concentration (10 M) of PGE 2 (data not shown).
PGE 2 -induced cAMP accumulation in control and PDE4 upregulated cells was also assessed in the presence of 30 M rolipram (Fig. 4). We reasoned that if the decrease in PGE 2stimulated cAMP accumulation observed in salbutamol-pretreated cells was due, at least in part, to an up-regulation of PDE4 activity, then inhibiting PDE4 in these cells with rolipram would tend to normalize their responsiveness to PGE 2 . Indeed, in the presence of rolipram, PGE 2 -stimulated cAMP accumulation over the 15-min time course was virtually identical in control versus PDE4-induced cells (Fig. 4). Only after 15 min of exposure to PGE 2 was cAMP content slightly less in PDE4 up-regulated cells (68.2 Ϯ 11.2 pmol/10 6 cells) than in control cells (89.4 Ϯ 9.6 pmol/10 6 cells).
The ability of a range of PGE 2 concentrations (1 nM-10 M) to elevate cAMP content in control and PDE4-up-regulated U937 cells is shown in Fig. 5. The conditions of these experiments were identical to those of the time course studies, except that cells were treated with various concentrations of PGE 2 for a single fixed time (15 min) before cAMP content was determined. In control cells PGE 2 produced a large, concentrationrelated increase in cAMP accumulation. For example, basal cAMP content (0.99 Ϯ 0.14 pmol/10 6 cells) was increased to 52.9 Ϯ 2.9 pmol/10 6 cells by 10 M PGE 2 , greater than 50-fold over the basal level. In contrast, PGE 2 had much less effect on cAMP content in cells in which PDE4 activity had been upregulated. In fact, cAMP accumulation stimulated by 10 M PGE 2 was only 5.4 Ϯ 0.5 pmol/10 6 cells, nearly 10-fold less than in control cells and only 5-fold above basal cAMP content (0.97 Ϯ 0.12 pmol/10 6 cells).
In the presence of 10 M rolipram, PGE 2 -induced cAMP accumulation was greater in both control cells and, even more impressively, cells in which PDE4 activity had been up-regulated. Overall, inhibiting PDE4 activity with rolipram increased PGE 2 -stimulated cAMP accumulation by 2-fold in control cells and 12-fold in PDE4 up-regulated cells. Thus, in the presence of rolipram, maximal PGE 2 -stimulated cAMP accu-  Fig. 2A, U937 cells were untreated or exposed to 1 M salbutamol and 30 M rolipram for 4 h. After the treatment period cells were washed extensively and lysed. Supernatant fractions were then prepared and identical amounts of protein (ϳ100 g) were subjected to SDS-polyacrylamide gel electrophoresis. Immunoreactivity was determined using antibody raised against GalK-hrPDE4A. Also run in these studies were lysates from yeast engineered to express hrPDE4A (full-length), PDE4B, or PDE4D. The data are representative of results from four (U937 cell lysates) or two (hrPDE4 subtypes) experiments. Panel B, additional experiments were conducted to detect PDE4B immunoreactivity in control U937 cells and cells treated for 4 h with salbutamol (1 M) and rolipram (30 M). For reference purposes, hrPDE4B was also run. The antibody used in these studies was raised against a unique carboxyl-terminal PDE4B peptide. The data are representative of two experiments.

FIG. 5. Effect of up-regulating PDE4 activity on concentrationresponse curves for PGE 2 -stimulated cAMP accumulation in U937 cells. Control cells (solid symbols) and cells treated for 4 h with 1 M salbutamol and 30 (Ϯ)
M rolipram (open symbols) were washed extensively and treated with PGE 2 in the absence (circles) or presence (squares) of 10 M (Ϯ) rolipram. When used, rolipram was reintroduced into the incubation medium 1 min before the addition of PGE 2 . The cells were then exposed to various concentrations of PGE 2 (1-10,000 nM). Cyclic AMP content was determined 15 min after the addition of PGE 2 . Basal cAMP content was 0.99 Ϯ 0.14 pmol/10 6 cells in control cells and 0.97 Ϯ 0.12 pmol/10 6 cells in PDE4 up-regulated cells. The values represent the mean Ϯ S.E. of five experiments. *Significantly less than value in cells not pretreated with rolipram and salbutamol (p Ͻ 0.05). mulation in cells in which PDE4 activity had been up-regulated was only 2-fold less than in control cells. This contrasts with the 10-fold difference detected in the absence of rolipram.
Effect of PDE4 Up-regulation on Agonist-stimulated Ca 2ϩ Mobilization-Consistent with the results of other studies (Saussy et al. 1989;Winkler et al., 1990), LTD 4 produced a concentration-dependent Ca 2ϩ mobilization with an EC 50 ϭ 3-5 nM. Note that up-regulating PDE4 activity had no effect on the LTD 4 concentration-response curve (Fig. 6).
Although up-regulation of PDE4 activity had no direct effect on LTD 4 -stimulated Ca 2ϩ mobilization, it had a substantial effect on the ability of PGE 2 to inhibit this response (Fig. 7). In control cells PGE 2 suppressed maximal Ca 2ϩ mobilization induced by 3.3 nM LTD 4 with an IC 50 ϭ 30 nM and a maximal inhibitory effect of 70 Ϯ 1% (Fig. 7A). Prior exposure of U937 cells to salbutamol (1 M) and rolipram (30 M) for 4 h to induce PDE4 resulted in a substantial reduction in the inhibitory effect of PGE 2 . In these cells, PGE 2 suppressed Ca 2ϩ mobilization with an IC 50 ϭ 150 nM, 5-fold greater than in untreated cells, and had a maximal inhibitory effect of only 27 Ϯ 2%.
The heterologous desensitization to the inhibitory effect of PGE 2 in PDE4 up-regulated cells was largely reversed in the presence of 10 M rolipram (Fig. 7B). For example, although the maximal inhibitory effect of PGE 2 on Ca 2ϩ mobilization was statistically less in PDE4 up-regulated cells even in the presence of 10 M rolipram, the difference from control cells was extremely small (70 Ϯ 3%, control versus 65 Ϯ 1%, upregulated). Qualitatively, similar results were obtained in studies in which PGE 2 -stimulated cAMP accumulation was assessed in the presence of 100 M rolipram, although a significant diminution in the ability of PGE 2 to inhibit Ca 2ϩ mobilization in PDE4 up-regulated cells under these conditions was observed only at one concentration of PGE 2 (10 nM PGE 2 ; data not shown). No statistically significant difference was observed with any of the other concentrations of PGE 2 . Thus, the ability of PGE 2 to inhibit Ca 2ϩ mobilization in cells that have increased PDE4 activity is largely recovered in the presence of a PDE4 inhibitor.

DISCUSSION
PDE4 is the predominant cAMP-metabolizing enzyme family in inflammatory cells and has been identified as an important new molecular target for novel antiasthmatic and anti-inflammatory drugs (Torphy and Undem, 1991;Giembycz and Dent, 1992). Based upon the results of recent studies in which Sertoli cells were used as a model system (Sette et al., 1994a(Sette et al., , 1994b, considerable attention has been focused on two general mechanisms by which the activity of PDE4 is regulated by hormones, particularly those that stimulate adenylyl cyclase activity. One regulatory mechanism, designated "short term activation," involves a protein kinase A-mediated phosphorylation of a specific splice variant PDE4D. This phosphorylation results in an increase in catalytic activity, perhaps by allosteric modification of the catalytic domain (Sette et al., 1994b). A second regulatory mechanism, designated "long term activation," occurs with two other splice variants of PDE4D (Swinnen et al., 1989. Activation of protein kinase A in intact cells increases the expression of these latter forms by enhancing mRNA synthesis or increasing mRNA stability. Although indirect evidence suggests that activators of adenylyl cyclase can regulate PDE4 activity in immune and inflammatory cells (Chan et al., 1982;Holden et al., 1987, Bourne et al., 1973Torphy et al., 1992), definitive information on the precise nature of this phenomenon in these cells is not available. Moreover, despite the growing body of evidence suggesting that the activity of PDE4 can be up-regulated by hormonal stimulation, little is known about the biological importance of this regulation. We have begun to address these deficiencies by examining the nature and functional consequences of PDE4 up-regulation in U937 cells, a human monocytic cell line. As previously reported (Torphy et al., 1992), activation of the protein kinase A cascade in these cells by ␤-adrenoceptor agonists increases PDE4 catalytic activity. The magnitude of this up-regulation is enhanced if rolipram is included in the incubation medium, presumably because inclusion of a PDE inhibitor both heightens and prolongs the increase in cAMP content produced by ␤-adrenoceptor agonists. The increase is prevented by actinomycin D or cycloheximide, indicating that the up-regulation of PDE4 activity depends upon the synthesis of both mRNA and protein. The results of the present experiments are consistent with this conclusion. Treatment of U937 cells for 4 h with a combination of salbutamol and rolipram increased the amount of immunoreactive PDE4A detected in cell supernatants. The results were particularly striking in that PDE4A was virtually undetectable in untreated cells but rolipram. When rolipram was used in combination with PGE 2 in the Ca 2ϩ mobilization experiments, it was reintroduced into the incubation medium 1 min before the addition of PGE 2 . Cells were challenged with LTD 4 (3.3 nM) 5 min after the addition of PGE 2 , and Ca 2ϩ mobilization was monitored via fura-2 fluorescence. The data represent the mean Ϯ S.E. of four experiments and reflect maximal cytosolic free Ca 2ϩ concentrations achieved in response to LTD 4 . *Significantly less in  clearly evident in the induced cells. An increase in PDE4B immunoreactivity also occurred in response to 4-h exposure to salbutamol and rolipram. In contrast, this treatment regimen had no effect on PDE4D3 levels. These studies do not, however, eliminate the possibility that PDE4D activity can be regulated directly by a protein kinase A-mediated phosphorylation pathway. Moreover, since our focus was on PDE4D3, we cannot exclude the possibility that ␤-adrenoceptor agonists regulate the expression of other mRNA splice variants (i.e., PDE4D1 and PDE4D2). mRNA transcripts encoding PDE4 subtypes were identified through RT-PCR methodology using subtype-specific oligonucleotide primers. In untreated U937 cells, the only PCR products detected were those corresponding to PDE4B and PDE4D. Although the technique utilized was not designed to be quantitative, distinct changes in the pattern of PCR products were detected after treatment with rolipram and salbutamol. Specifically, PCR product for PDE4A transcript was barely detectable in untreated cells, but a PCR product of the appropriate length was observed clearly and consistently in stimulated cells. In addition, the amount of PCR product corresponding to PDE4B appeared to increase, whereas that corresponding to PDE4D appeared to decrease. The functional consequences of these apparent changes in steady-state transcript levels as they relate to PDE4B and PDE4D protein expression are unknown. Consistent with the results of this study, Engels and colleagues (1994) also detected an increase in message for PDE4A and PDE4B in response to 18-h treatment of U937 cells with 0.5 mM dibutyryl cAMP.
Up-regulation of PDE4 activity in U937 cells had a substantial impact on their responsiveness to PGE 2 , an activator of adenylyl cyclase. As demonstrated in both time course and concentration-response studies, pretreatment of cells with a combination of salbutamol and rolipram substantially decreased the ability of PGE 2 to elevate cAMP content. This loss of activity was mirrored by a decrease in the ability of PGE 2 to inhibit LTD 4 -induced Ca 2ϩ mobilization. We reasoned that if an increase in PDE4 activity had a role in producing the heterologous desensitization in pretreated U937 cells, then inhibiting PDE4 activity in these cells would tend to normalize their sensitivity to PGE 2 . This was indeed the case. Whereas there was a substantial difference in the ability of PGE 2 to stimulate cAMP accumulation and suppress Ca 2ϩ mobilization in desensitized versus control U937 cells, the difference was markedly reduced or, in some instances, virtually abolished when functional studies were carried in the presence of rolipram (10 M).
It is not yet known whether regulation of PDE4 activity represents a general homeostatic mechanism by which all target cells, particularly inflammatory cells, modulate their responsiveness to hormones and autacoids that activate adenylyl cyclase. However, in support of the broad applicability of this phenomenon, we recently have demonstrated that PDE4 activity and steady-state protein levels in human monocytes is upregulated by ␤-adrenoceptor agonists in a manner similar to that seen in U937 cells. 2 The results of these studies have implications regarding the use of ␤-adrenoceptor agonists as bronchodilators in the treatment of asthma. Normally, endogenous activators of adenylyl cyclase such as epinephrine, PGE 2 , and prostacyclin act as natural anti-inflammatory and bronchodilator agents (Barnes, 1986;Kuehl et al., 1987), presumably by elevating cAMP content in the appropriate target tissues. In theory, the beneficial actions of these agents would be compromised if chronic treatment of asthmatic individuals with ␤-adrenoceptor agonists resulted in an up-regulation of PDE4 activity in inflammatory cells and airway smooth muscle. This would allow inflammatory processes and bronchoconstriction to proceed unchecked. Indeed, chronic use of inhaled salbutamol increases airway responsiveness to allergen and causes tolerance to the protective effect of the ␤-adrenoceptor agonist against allergen challenge (Cockcroft et al., 1993). Presumably, this occurs as a result of a diminished ability of ␤-adrenoceptor agonists to inhibit mast cell mediator release (Cockcroft et al., 1993).
In the long term, a generalized induction of tolerance in inflammatory cells could lead to a worsening of disease status. Regarding this proposal, several clinical reports have linked a deterioration of disease status and increased mortality with the excessive use of ␤-adrenoceptor agonists (Barnes and Chung, 1992;Sears et al., 1990;Spitzer et al., 1992). The results of these studies, the interpretation(s) of which are controversial and often conflicting, have led to the proposal of several hypotheses to explain the apparent detrimental effect of sympathomimetic bronchodilators. These include the possibility that the overuse of ␤-adrenoceptor agonists: 1) down-regulate ␤-adrenoceptors, 2) mask a worsening of disease status, 3) increase antigen load to the distal airway, or 4) compromise the "protective" role of lung mast cells (Barnes and Chung, 1992;Nelson et al., 1991;Page, 1991). It is tempting to speculate that an up-regulation of PDE4 activity represents an additional factor that contributes to the "bronchodilator paradox." An issue that remains to be addressed is whether the concentrations of salbutamol that increase the expression of PDE4 in U937 cells are similar to those required to produce bronchodilation in the clinic. In a previous study (Torphy et al., 1992), we demonstrated that PDE4 activity is up-regulated by salbutamol, used alone or in combination with rolipram, over a concentration range of 10 nM to 10 M. This range is identical to that required to relax human airway smooth muscle in vitro (Goldie et al. 1986;Nally et al., 1994). While it is virtually impossible to determine the local concentrations of inhaled salbutamol within the airway, this information strongly suggests that salbutamol up-regulates PDE4 activity at clinically relevant concentrations.
In conclusion, treatment of U937 cells with agents that activate the cAMP/protein kinase A cascade results in an increase in PDE4 activity. Coincident with the elevation in total cellular PDE4 catalytic activity is an increase in steady-state message and protein for PDE4A. From a functional standpoint, the up-regulation of PDE4 activity results in a heterologous desensitization of U937 cells to the actions of PGE 2 and, presumably, other adenylyl cyclase activators. Conceivably, this regulatory pathway could compromise the long term anti-asthmatic efficacy of ␤-adrenoceptor agonists, the most commonly used class of bronchodilators.
Note Added in Proof-Cyclic AMP-elevating agents produce a similar, but not identical pattern of PDE4 subtype up-regulation in human monocytes and Mono Mac 6 cells, a human monocytic cell line (Verghese et al., 1995).