Differential regulation of TNF-α and IL-1β production from endotoxin stimulated human monocytes by phosphodiesterase inhibitors

The effect of selective PDE-I (vinpocetine), PDE-III (milrinone, CI-930), PDE-IV (rolipram, nitroquazone), and PDE-V (zaprinast) isozyme inhibitors on TNF-α and IL-1β production from LPS stimulated human monocytes was investigated. The PDE-IV inhibitors caused a concentration dependent inhibition of TNF-α production, but only partially inhibited IL-1β at high concentrations. High concentrations of the PDE-III inhibitors weakly inhibited TNF-α, but had no effect on IL-1β production. PDE-V inhibition was associated with an augmentation of cytokine secretion. Studies with combinations of PDE isozyme inhibitors indicated that PDE-III and PDE-V inhibitors modulate rolipram's suppression of TNF production in an additive manner. These data confirm that TNF-α and IL-1β production from LPS stimulated human monocytes are differentially regulated, and suggest that PDE-IV inhibitors have the potential to suppress TNF levels in man.


Introduction
Activation of monocytes results in the secretion of numerous cytokines, including 11-1 and tumour necrosis factor o (TNF-o), which play key roles in regulation of the immune system. 1'2 However, unregulated overproduction of these cytokines has been cited as a major factor in the development of septic shock, rheumatoid arthritis and inflammatory lesions. 1'2 Thus, the production of these cytokines must be tightly controlled by numerous mechanisms. Macrophage activation induces a rise in the rate of transcription for IL-1 and TNF-o mRNA and a concurrent increase in the stability 1'3 and translational efficiency of TNF-o mRNA. The precursors of the cytokines are enzymatically cleaved, and TNF-o and IL-lfl are released into the interstitial spaces or blood. 1'2 Typically, monocyte activation is associated with a 200--3000 fold and 150-1000 fold increase in TNF-o and IL-lfl production, respectively. Although endotoxin induces monocytes to secrete both IL-lfl and TNF-o, the production of these cytokines has been shown to be differentially regulated. [3][4][5] For example, IL-1, but not TNF-, production by endotoxin induced monocytes can be selectively decreased by a calmodulin dependent kinase inhibitor, H7. Cellular activation is often accompanied by changes in the intracellular levels of cyclic nucleotides, and phosphodiesterases, by metabolizing the intracellular cyclic nucleotides, play a major role in modulating their intracellular concentrations. 992 Rapid Communications of Oxford Ltd The nonspecific phosphodiesterase (PDE) inhibitors, theophylline, 3-isobutyl-l-methylxanthine (IBMX), and pentoxifylline, selectively decrease TNF-o mRNA steady state levels and TNFproduction from endotoxin-stimulated murine and human monocytes. ('-8 However, the effect of agents which augment intracellular levels of cAMP on IL-lfl production from LPS induced macrophages varies depending on the species, stimulus, and culture conditions; 2 results ranging from stimulation to no effect s to inhibition -2 have been reported. There are five distinct families of phosphodiesterase (PDE) isoenzymes which differ in their cyclic nucleotide substrate specificities and tissue distribution. 3'4 Selective inhibitors for PDE-I, PDE-III, PDE-IV and PDE-V have been identified, although the selective inhibitors may display some inhibitory effect on other PDE isoenzymes at higher concentrations. s PDE-IV has been isolated from human monocytes, 16 and PDE activity was also detected in peaks eluting in buffers for PDE-I and PDE-V isolation from routine macrophages. This study was undertaken to investigate the role of cAMP, cGMP and PDE isoenzymes in regulation of monocyte cytokine production. The effect of selective inhibitors of the PDE-I, PDE-III, PDE Dulbecco's phosphate buffer saline (PBS) (Gibco) to produce a stock containing 5 mg/ml, diluted to 1 mg/ml in PBS, aliquoted and then stored frozen at -20C. LPS was thawed and diluted with RPMI 1640-5% FCS as needed.
Human monocyte assay: Human mononuclear cells were isolated by Ficoll hypaque density centrifugation (specific gravity 1.077) from Leukapaks obtained from healthy Caucasian males. The bully coat layer was washed three times in RPMI 1640 and enumerated. Human mononuclear cells were plated into 24-well plates (106 cells in l ml RPMI 1640-5% FCS/well) or flat-bottom microtitre plates (3.76 10 cells in 100/zl/ well) and incubated at 37C in 5% CO2 in a humidified atmosphere for 45 min. The wells were rinsed three times to remove non-adherent cells. The quantity of baseline and LPS stimulated cytokine production was measured from cells incubated in the absence or presence of 0.1 ng LPS (0111 :B4 subtype) per ml, respectively, for 16-24 h at 37C in 5% CO 2 in a humidified atmosphere. To assess the effect of drugs, varying concentrations of compounds or vehicle were added to the cells just prior to their incubation with 100 pg LPS/ml in RPMI 1640-5% FCS as above. All tests were performed in duplicate. To measure the concentration of secreted and cell associated IL-lfl and TNF-0, supernatants were harvested. To assess cell associated cytokine levels, RPMI 1640 (0.3 ml) was added to the cells. Samples were frozen at -80C and stored. Cell lysates were prepared by thawing plates, scraping cells from the well, adding 0.2 ml RPMI 1640, scraping remaining cells from the wells, and pooling samples followed by two additional freeze-thaw cycles. 17 The cell lysate was diluted two-fold with RPMI 1640 containing 10% FCS to yield a final volume of 1 ml. It should be noted that the effect of nitraquazone on monokine secretion was assessed in a set of experiments (n 3) in which samples were not processed for analysis of cell associated monokine levels.
The viability of the LPS stimulated cells in the presence or absence of a test compound was assessed in the microtitre plates following a 16-24 h incubation at 37C in 5% CO2 using the previously described assay based on the metabolic reduction of 3-(4,5-dimethyl-thiazol-2yl)-2,5 diphenyl tetrazolium bromide (MTT). 18 None of the PDE inhibitors tested at the same concentrations decreased the viability of the monocytes.
ILfl and TNFassays: The concentrations of IL-lfl and TNF-0 in the supernatants and cell lysates were measured in duplicate using ELISA kits (Cistron, Pine Brook, N J). The standards, supernatants and cell lysates were diluted with RPMI 1640-5% FCS just before use. After the addition of the samples and standards to the appropriate wells, the plates were incubated at 37C overnight in a humidified atmosphere containing 5% CO2. The plates were then processed according to the manufacturer's instructions. However, the analysis was performed as listed below. The standard curve was fit to the following equation: Where OD was set to the optical density, C the estimated minimum, D the estimated maximum and A and B are derived as parameters of intercept and slope, respectively. The concentration of cytokine was calculated by inverse prediction. Percent inhibition was calculated as follows: The ICs0 value, defined as 50% inhibition of the stimulated control response was determined by inverse prediction of the following equation: Concentration C 4-D Where D equals the cytokine concentration in the stimulated control less that of the baseline control.
Regulation of TNF-ot and IL-l fl by PDE inhibitors Statistics: To determine whether the individual PDE inhibitors significantly blocked TNF-0 and IL-lfl production and/or secretion, an analysis of variance for a randomized block design (using each donor as a block) was performed. 19 This included all treatments where a single PDE inhibitor was given. Similarly, to assess whether the PDE-I, PDE-1II and PDE-V inhibitors could modulate the inhibition of TNFby rolipram alone, a similar analysis was performed. A separate analysis was performed for each combination, and all treatment groups including either drug alone or in combination were included. A t-statistic taking into account the within-and between-donor variation was used to test whether the percent inhibition observed was significantly different from zero. Degrees of freedom were approximated by Satterwaite's formula. 2 Interaction between drugs was tested by estimating the expected inhibition for the combination based on the inhibition for the individual PDE inhibitors and testing whether this was significantly different from the observed inhibition for the combination. Pairwise comparisons using the least significant difference method were performed to test whether the combination was better than either drug alone.
Wasik and Beller 21 have observed that the inhibitory effect of compounds on cytokine production may be minimized by overstimulating the monocytes. As an initial step, the quantity of endotoxin was titrated to determine the concentration required to stimulate near maximal release of IL-lfl and TNFfrom human monocytes. As shown in Fig. 1, 0.1 ng/ml of lipopolysaccharide (LPS) induced near maximal release of both IL-lfl and TNF-, regardless of the LPS subtype. Therefore, 0.1 ng/ml LPS (subtype 0111 :BB4) was used to assess the effect of drugs on cytokine release.
The quantity of IL-lfl and TNF-0 secreted by LPS stimulated monocytes increased by 100 to 1000-fold over the baseline level of 4-13 pg/ml and 5-23 pg/ml, respectively (data not shown). The cell associated TNF-0 was less than 10% of the total produced, whereas the cell associated IL-lfl ranged from 33.8% to 61.8% of the total produced. The level of TNFand IL-lfl elicited by LPS stimulation of human monocytes varied from donor to donor, in agreement with prior reports. 22'23 Effect of PDE inhibitors on monokine release: TNF-0 secretion was significantly suppressed by the PDE-IV inhibitors, rolipram and nitraquazone, in a concentration dependent manner ( Fig. 2A) 2B). For example, IL-lfl inhibition at 10#M rolipram or nitraquazone ranged from -43% to 53% and 14% to 57%, respectively. Vinpocetine, CI 930 and milrinone did not have a significant eect on IL-lfl secretion. In contrast, zaprinast, a PDE-V inhibitor, significantly augmented the release of IL-lfi from LPS stimulated human monocytes. Zaprinast was ten times more potent in augmenting IL-lfl secretion than TNF-0 secretion. Dexamethasone suppressed IL-lfl production (ICs0 0.02 #M) as expected 8'23 (Fig. 3B).

Mediators of Inflammation. Vol
The PDE inhibitors tested at the above concentrations did not decrease the viability of the cells as determined by the MTT assay. Therefore, the effect on cytokine production was not due to cytotoxicity. ficantly augmented secreted TNF-z production (p 0.004), the apparent increase in cell associated TNF-0 by zaprinast was not significant (p 0.2).
Similar to zaprinast's effects on IL-lfi secretion, the level of cell associated IL-lfi was significantly augmented by zaprinast (Fig. 4B). Although vinpocetine, CI 930, and milrinone did not affect the intracellular levels of IL-1/, 10/M rolipram significantly increased the concentration of cell associated IL-lfl. This increase in cell associated IL-lfl occurred at rolipram concentrations similar to those which inhibited IL-1/ secretion. Dexamethasone significantly suppressed cell associated TNF-0 and IL-lfl (ICs0 =0.02 and 0.01 #M, respectively).

Discussion
Agents which regulate intracellular cyclic nucleotide levels such as dibutryl cAMP (dbcAMP), prostaglandin E 2 (PGE2) and phosphodiesterase inhibitors can modulate macrophage functions. Superoxide generation from guinea pig peritoneal macrophages was inhibited by rolipram and Ro-20-1724, two PDE-IV inhibitors, but not by SKF 94120, a PDE-III inhibitor. 24 Human alveolar macrophage activation was modestly suppressed by forskolin; Ro-20-1724 had no effect. 25 Agents which elevate cAMP levels, such as dbcAMP, isoproterenol, and nonselective PDE inhibitors (e.g. pentoxifylline, IBMX, and theophylline), suppressed the production of TNFfrom LPS induced monocytes and macrophages. -8'11 The present studies showed that selective PDE isoenzyme inhibitors can differentially effect the production of TNF-0 and IL-lfl from LPS induced human monocytes. Rolipram and nitraquazone, selective inhibitors of PDE-IV, suppressed production of secreted TNF-z with a maximal inhibition of approximately 80% (Fig. 3A). CI 930 and milrinone at concentrations of 3 and 10#M modestly depressed TNF-0 secretion. Although CI 930 and milrinone can inhibit PDE-IV enzymatic activity at concentrations greater than 100 #M and 10 #M, respectively, ls'26 the inhibition of TNF-0 production by CI 930 occurred at lower concentrations. Thus, the effect of at least CI 930 is most likely caused by suppression of inherent PDE-III activity in the monocytes. The observation that combinations of these PDE-III inhibitors and rolipram resulted in additive suppression of TNF-0 production are consistent with the expectation that these agents mediate their effect via the same mechanism (elevation of cAMP levels).
In contrast, zaprinast, a PDE-V inhibitor that is expected to regulate cGMP levels, modestly but significantly increased secreted TNF-0 at 10 #M, in agreement with reports of other agents which elevate intracellular cGMP levels (sodium nitroprusside and exogenous cGMP) and which stimulate TNF-0 production. 6 The observation that the combination of zaprinast with rolipram decreased the TNF-0 inhibition observed with rolipram alone in an additive manner is consistent with opposing roles for PDE-V/cGMP and PDE-IV/cAMP in TNF-0 secretion by endotoxin stimulated monocytes.
TNF-0 production from activated macrophages is increased 100 to 5000-fold by a combination of elevated transcription rate, increased mRNA stability and improved translational ediciency.
Since cell associated as well as secreted TNF-0 levels were inhibited by rolipram and CI 930, these drugs probably do not inhibit secretion by blocking precursor processing. The nonspecific PDE inhibitors, pentoxifylline and IBMX, have been shown to reduce steady state levels of mRNA and suppress TNF-0 production by LPS stimulated monocytes and macrophages. Elevation of intracellular cAMP levels in LPS stimulated murine macrophages by dbcAMP is also reported to suppress TNF-o production, lower the transcription rate and decrease steady state levels of mRNA. 8'11 Thus, one likely mechanism by which PDE-I11 and PDE-IV inhibitors may regulate TNF-o is by decreasing the steady state level of TNF-o mRNA in the cell.
The effect of cAMP modulating agents on IL-1// production from LPS activated monocytes varies with the culture conditions, type of stimulus and species. 2 Kassis et al. 9 observed that agents which elevate cAMP stimulate IL-lfl production. Endres et al. 8 have observed no effect on total IL-lfl protein or mRNA levels in LPS stimulated monocytes or macrophages in the presence of agents which elevated cAMP levels including nonspecific PDE inhibitors. In contrast, Brandwein, 27 Hurme, 12 and Knudsen and colleagues 1 reported that concurrent treatment of LPS induced macrophages with cAMP elevating agents result in a modest decrease in secretion of IL-lfl. No effect on IL-I mRNA steady state levels is observed with treatment with agents which elevate cAMP levels. [10][11][12] In the present studies, rolipram and nitraquazone at high concentrations did significantly decrease IL-1// secretion, in agreement with the studies of Knudsen, Hurme and Brandwein. 1'12'27 Since rolipram increased the quantity of cell associated IL-1/ (nitraquazone was not tested), while it decreased secreted IL-lfl, this data supports the interesting suggestion that PDE-IV may play a role in modulating IL-lfl release from the cell. 1 This action by rolipram may help to account for some of the differences between previous reports which assessed the effects on secreted vs. total IL-lfl production. Secondly, variability of sensitivity of IL-lfl production to PDE-IV inhibitors between donor monocyte preparations may also account for some differences.
Disparate results are observed with cGMP elevating agents on IL-lfl secretion where SIN-1 treatment inhibits IL-lfl, dibutryl cGMP has no effect 12 and in the present study, zaprinast increased IL-1/ secretion. These disparities may have arisen due to different levels of stimulation and/or culture conditions. However, under the same conditions, the observations that PDE-IV and PDE-V inhibitors exerted opposite effects on IL-1// and TNF-o secretion raise the possibility that cAMP and cGMP may affect endotoxin induced cytokine secretion in an opposing manner.
The suppressive effect of selective PDE-IV inhibitors on TNF-o suggests that these and similar compounds may be useful in the prophylactic and/or therapeutic treatment of pathologies involving overproduction of TNF-o. Since anti-TNFreagents have been reported to decrease symptoms in animal models of septic shock, experimental allergic encephalitis, and collagen induced arthritis, PDE-IV inhibitors may have a therapeutic benefit in diseases such as arthritis, septic shock, and multiple sclerosis. PDE-IV inhibitors have been suggested also to be of potential benefit in the treatment of asthma, based upon their significant bronchodilatory and anti-inflammatory activities. 14'28 Although the role of TNF-o in asthma has not been clearly established, a recent report suggested that increases in bronchoalveolar lavage TNF-o levels are associated with exacerbations of asthma. 29 Furthermore, TNFwas reported to stimulate eosinophil oxidant production and toxicity towards human endothelium. 3 Thus, it may also be suggested that the anti-TNF<z activities of PDE-IV inhibitors will contribute to their anti-asthma profiles. Further experiments will be needed to explore this possibility.