Nonredundant Function of Phosphodiesterases 4D and 4B in Neutrophil Recruitment to the Site of Inflammation¹

Cyclic nucleotides play an important modulatory function in virtually all cell types involved in the pathogenesis of asthma and other chronic inflammatory diseases of the airway. Activation of cAMP signaling and protein kinase A (PKA)³ has a negative impact on T cell activation and proliferation (1, 2), production of cytokines and chemokines (3), and eosinophil recruitment to the site of inflammation (3). Furthermore, monocyte adhesion and migration are inhibited by high cAMP levels (4, 5, 6). Pharmacological activation of cAMP signaling suppresses several neutrophil responses, including degranulation (7, 8), superoxide anion generation (9, 10), release of IL-8 (11), and phagocytosis (10). In addition, an increase in cAMP impacts the expression of adhesion molecules (CD11b/CD18/L-selectin) and adhesion to other cells, and disrupts chemokine-induced chemotaxis (12, 13). Finally, smooth muscle contractility in the airway is also regulated by cAMP signaling (14).

Despite the overwhelming quantity of data in support of the importance of this pathway, it is still uncertain how cAMP signaling contributes to the orchestration of the inflammatory responses under physiological conditions as well as in pathological states, including those associated with the airway hyper-reactivity and airway remodeling found in asthmatic patients. Most of the observations pertinent to cAMP signaling have been based on in vitro pharmacological manipulations of inflammatory cell models, which sometimes produced conflicting results (3). In this study we investigated the role of cAMP signaling in proinflammatory cells in vivo using a genetic approach. By inactivating genes coding for three isoenzymes of type 4 phosphodiesterases (PDE4), the enzymes that degrade cAMP expressed in inflammatory cells, we have been able to test the effect of altered, but not interrupted, cAMP signaling on neutrophil function.

In a few instances it has been reported that signals that activate inflammatory cells also produce an increase in cAMP (15). However, the significance of the activation of cAMP signaling on the overall responses of inflammatory cells is largely unknown. Consistent with a negative role of cAMP in inflammatory cell activation, activation of T cells (3, 16), T cell lines (17), monocytes (18, 19, 20), and macrophages (18) is associated with an increase in PDE activity. This regulation is viewed as a positive feedback necessary to remove a cAMP negative constraint that prevents inflammatory cell activation. PDEs are a large superfamily of enzymes encoded by at least 25 genes subdivided into 11 families (21). The expression of PDE1, -3, -4, -5, -7, and -8 in inflammatory cells is inferred by mRNA detection, whereas the presence and function of corresponding proteins have been difficult to determine. PDE3, -4, -7, and -8 are regulated during T cell activation (16, 22). Of the four PDE4 genes present in the human genome, three are expressed ubiquitously in most inflammatory cells at least as mRNAs (3, 23, 24, 25). In some cases PDE4 activity has been characterized (26, 27), but few attempts have been made to determine the contribution of each of the isoenzymes to the overall PDE activity. This PDE4 family of enzymes has received attention because PDE4 inhibitors have important anti-inflammatory effects (3). However, the side effects of nonselective PDE4 inhibitors have hindered the development of useful drugs (28, 29). In addition, PDE4 target validation has been hampered by the fact that PDE4A, PDE4B, and PDE4D are expressed in most inflammatory cells, rendering it difficult to define the role of individual PDE4s in these and other cells.

Using a model of lung inflammation, this study investigated the function of each PDE4 gene in neutrophils in vivo by deleting individual PDE4 genes. We provide evidence that PDE4D and PDE4B, but not PDE4A, are necessary for neutrophil recruitment to the lung after exposure to endotoxin. Ablation of each of these genes produces impaired neutrophil function with altered chemotaxis and, possibly, adhesion. The functions of the two PDE4 genes overlap, but are not completely redundant, and one gene is unable to compensate for inactivation of the other.

Generation of PDE4B- and PDE4D-deficient mice has been described previously (18, 30). Following a similar strategy, generation of PDE4A knockout mice will be reported elsewhere. In all cases the effect of the null PDE4 allele was tested on mice of a mixed background (C57BL/6 × 129/Ola) using littermates as controls. Colonies were maintained by out-breeding heterozygous pairs and genotyping the offspring. All mice used in these experiments were >2 mo old. To generate the compound heterozygous PDE4D^+/−/PDE4B^+/−, homozygous null PDE4D and PDE4B mice were mated. The compound heterozygous mice thus obtained were mated again to produce PDE4D^+/−/PDE4B^−/− or PDE4D^−/−/PDE4B^+/− mice. All procedures were approved by the administrative panel on laboratory animal care at Stanford University.

Mice were exposed to saline or 100 μg/ml LPS aerosols in PBS for 1 h in a container connected to a SPAG-2 system (ICN Pharmaceuticals, Costa Mesa, CA). At 1, 2, 4, or 24 h after LPS exposure, mice were overdosed with an i.p. injection of 6% ketamine and 2% xylazine (v/v) in PBS, and a partial tracheotomy was performed. Lungs were flushed with PBS six times through a cannula, and a total of 1.6–1.7 ml of lavage fluid was recovered. Total cells in the BAL fluid were counted using a hemocytometer. Cells present in the BAL (1.6 × 10⁴ cells) also were prepared on slides by cytospin centrifugation (Thermo Shandon, Pittsburgh, PA) and stained by May-Grünwald and Giemsa stain (Sigma-Aldrich, St. Louis, MO). After staining, percentages of neutrophils, macrophages, and lymphocytes were determined by counting at least 200 cells. The levels of TNF-α, KC, and MIP-2 in the supernatant of BAL fluid were measured using a mouse TNF-α immunoassay kit (BD Pharmingen, San Diego, CA), a mouse KC immunoassay, and a mouse MIP-2 immunoassay (R&D Systems, Minneapolis, MN).

Mice were injected i.p. with either polyethylene glycol (PEG; m.w., 200) or rolipram (3 mg/kg mouse) dissolved in PEG 45 min before the exposure to LPS. Four hours later, BAL fluid was obtained, and total cells and neutrophils were determined as described above.

Mice were treated with LPS for 1 h, and BAL fluid was collected 4 h later as described above. Cells from the BAL fluid were recovered by centrifugation, and pellets were lysed in a buffer containing 1% Nonidet P-40, 50 mM Tris-HCl (pH 7.5), 250 mM NaCl, 5% glycerol, 1 mM EDTA, 0.2 mM EGTA, 10 mM NaF, 10 mM sodium pyrophosphate, 1 mM Na₃VO₄, 1 mM 4-(2-aminoethyl)benzenesulfonyl fluoride (Roche, Mannheim, Germany), and 1 tablet/10 ml of protease inhibitor mixture (Roche). After sonication, aliquots of the cell extract were assayed for total PDE activity and rolipram-insensitive PDE activity in the presence of 10 μM rolipram. The PDE assay was performed according to the method of Thompson and Appleman (31) as detailed previously (30). The rolipram-sensitive activity (PDE4 activity) was derived by subtracting the rolipram-insensitive activity from the total activity. The protein concentration was determined by the Bradford method.

Mice were treated with LPS for 1 h, and BAL fluid was collected after 24 h as described above. Neutrophils were isolated using a Percoll gradient (Amersham Biosciences, Piscataway, NJ) as described previously (32). Approximately 95% of the recovered cells were neutrophils, as judged by morphology and Gr-1 expression. Crude extracts were prepared from these neutrophils and assayed for PDE activity as described above. For immunoprecipitation, cell extracts were clarified by centrifugation at 20,000 × g for 15 min, and supernatants were incubated with rabbit preimmune serum, PDE4D-specific (M3S1), or PDE4B-specific Abs (K118) that were preincubated with protein G-Sepharose. After 1-h incubation with Abs, samples were centrifuged, and pellets were washed, resuspended in PBS, and assayed for PDE activity.

BAL fluids were collected from the mice 4 h after LPS inhalation. Cells recovered from the BAL fluid were washed once with a FACS buffer (2% FBS and 0.01% sodium azide in PBS) and then incubated with mouse Fc block (BD Pharmingen) on ice for 15 min before staining with FITC anti-mouse Gr-1 Ab plus either PE anti-mouse CD18 or PE anti-mouse CD11b Ab (eBioscience, San Diego, CA) for 40 min on ice. For the detection of CD11a, PE Ly-6G Ab plus either FITC anti-mouse CD11a Ab or isotype control FITC rat IgG2a,κ was used. As an isotype control for anti-CD18 Ab, PE-rat IgG2a,κ (eBioscience) was used. PE-rat IgG2b,κ (eBioscience) was used as a control for the anti-CD11b Ab. The adhesion molecule expression on neutrophils was then quantitated using a FACScan flow cytometer (BD Biosciences, Mountain View, CA).

Bone marrow cells collected from the femurs of wild-type PDE4B^−/− and PDE4D^−/− mice were incubated for 5 days at a density of 0.7–1 × 10⁶ cells/ml in RPMI 1640 medium containing 10% FBS, 100 U/ml penicillin, and 100 μg/ml streptomycin. The culture was supplemented with 20 ng/ml G-CSF to induce neutrophil differentiation. For phagocytosis assay, aliquots of 1 × 10⁶ cells in 1 ml of medium were incubated for 1 h in the absence or the presence of 10 μl of FITC-latex bead suspension (Sigma-Aldrich). As a negative control, 1 × 10⁶ splenocytes were prepared from wild-type mouse spleen and incubated with FITC-latex beads for 1 h. To terminate phagocytosis, the cells were pelleted and washed once with FACS buffer, followed by incubation with anti-CD16/CD32 mAb (2.4G2; BD Pharmingen) on ice for 15 min. A PE anti-mouse Gr1 mAb (BD Pharmingen) then was used to stain the cells on ice for 40 min. After washing, cells were resuspended in FACS buffer and subsequently added to an equal volume of FACS buffer containing 1 μg/ml 7-aminoactinomycin D (Sigma-Aldrich). Green fluorescence in neutrophils was quantitated using a FACScan flow cytometer as described above.

All migration assays were performed in 24-well plates with 3-μm pore size polycarbonate filters (Corning, Corning, NY) using RPMI 1640 medium supplemented with 10% FBS (Life Technologies, Auckland, New Zealand). Unfractionated splenocytes (8 × 10⁵ cells) in 100 μl of medium were placed in the upper chamber, and increasing concentrations of recombinant murine KC (PeproTech, Rocky Hill, NJ) or recombinant murine MIP-2 (R&D Systems) were added in the lower chamber. In the total splenocytes derived from wild-type, PDE4B-deficient, and PDE4D-deficient mice, the percentages of neutrophils were 2.4 ± 0.4, 2.1 ± 1.1, and 2.2 ± 0.1, respectively. After 2-h incubation at 37°C, upper chambers were removed, 20 μl of polystyrene beads (Polysciences, Warrington, PA) were added to each well, and the suspension of cells/beads was transferred to tubes. Wells were rinsed with 5 mM PBS containing 5 mM EDTA to detach remaining cells, and the washes were combined. The migrated cells were counted according to the method of Campbell et al. (33). Briefly, cells were stained with FITC-anti-mouse Gr-1 Ab, and the stained cell/bead suspension was loaded onto a FACScan flow cytometer. Because beads could be distinguished from cells in the plot of side and forward scatter, the ratio of cells to beads was calculated to give the number of migrated Gr-1-positive cells. To assess the effect of acute PDE inhibition on chemotaxis, cells were pretreated with vehicle (DMSO) or rolipram (10 μM) for 1 h, then transferred to the upper chamber. Either vehicle (DMSO) or rolipram (10 μM) was added to the lower chamber containing increasing concentrations of KC (0 or 100 ng/ml).

Several animal models recapitulate the pathological changes observed in chronic obstructive pulmonary disease, nonallergic asthma, and other chronic inflammatory disorders, including airway neutrophilia and enhanced cytokine production (29). One in vivo mouse model developed to assess neutrophil function is the airway infiltration that follows LPS administration by aerosol. To investigate the effect of PDE4 ablation on neutrophil recruitment, wild-type and PDE4A-, PDE4B-, and PDE4D-deficient mice were exposed to LPS for 1 h, and the number of cells recruited in BAL was scored 24 h from the end of the treatment. The neutrophil number recovered in BAL was reduced by 31 and 48% in PDE4B- and PDE4D-deficient mice, respectively (Table I). Conversely, no significant differences were observed between wild-type and PDE4A-deficient mice (PDE4A^+/+, 15.85 ± 2.73 × 10⁵ neutrophils/mouse (n = 4); PDE4A^−/−, 16.65 ± 5.41 × 10⁵ neutrophils/mouse (n = 4)). In addition, the number of cells recruited to the lung after LPS was not different in all genotypes (Table I). Consistent with the decreased neutrophil cell counts, myeloperoxidase activity was significantly decreased in the lungs of PDE4-deficient mice after LPS inhalation (data not shown). A more detailed time-course study demonstrated that the decreased neutrophil recruitment in the lungs after LPS inhalation was evident 4 h after treatment and was maintained for at least 20 h (Fig. 1). The reduced number of neutrophils in the BAL fluid of PDE4-null mice was not due to a decreased production of cells by bone marrow, because the number of circulating neutrophils was not different among the three genotypes (wild-type, 0.27 ± 0.02 × 10³ cells/μl blood (n = 8); PDE4D-deficient mice, 0.29 ± 0.04 × 10³ cells/μl blood (n = 5); PDE4B-deficient mice, 0.22 ± 0.03 × 10³ cells/μl blood (n = 5)).

To investigate the cause of reduction in neutrophil recruitment to the airway, cytokine and chemokine accumulation was measured at different times after LPS inhalation in wild-type and PDE4-deficient mice. TNF-α accumulation in BAL fluid was not significantly different between wild-type and PDE4-deficient mice 1 h after LPS inhalation (Fig. 2); however, the levels of TNF-α were significantly reduced in PDE4D-deficient mice compared with wild-type mice at 2 h and remained lower for up to 4 h. There was a trend toward a similar decrease in TNF-α accumulation in PDE4B-deficient mice, although it did not reach statistical significance. Conversely, no significant difference in either KC or MIP-2 accumulation was observed in the three genotypes at any time point investigated. Thus, the decreased neutrophil recruitment observed in the two mouse lines is not associated with major changes in chemokine production.

To investigate the levels of PDE4 expression in the neutrophils, PDE4 activity was assessed in cells recovered from the BAL of wild-type and PDE4-deficient mice. A major decrease in PDE4 activity was observed in PDE4D-deficient neutrophils (Fig. 3,A). Similar results were obtained with neutrophils purified from BAL of PDE4D-deficient mice (data not shown). Furthermore, a small, but significant, decrease in PDE4 activity was observed in the PDE4B-deficient neutrophils (Fig. 3,B). To confirm this pattern of expression, PDE4 activity in wild-type BAL cells after LPS inhalation was measured after immunoprecipitation of the cell extracts with PDE4D- and PDE4B-specific Abs. This alternative approach confirmed that PDE4D is expressed at much higher levels than PDE4B in BAL neutrophils (Fig. 3 C). These data demonstrate that both PDE4s are expressed in these cells, with PDE4D being the predominant form. No significant compensatory increase in rolipram-insensitive PDE activity was observed in any cell preparation.

The functions of the two PDE4s in neutrophil recruitment were further assessed by comparing the effects of rolipram on wild-type and PDE4-deficient mice. Rolipram should have no additional effect on neutrophil recruitment in the PDE4-deficient mice if the functions of the two genes are redundant. If, instead, PDE4D and PDE4B have distinct functions, rolipram by itself would have a more pronounced effect than the genetic inactivation of each individual PDE4 and should further decrease neutrophil recruitment when tested in the PDE4-deficient mice. Rolipram at 3 mg/kg administered 45 min before LPS inhalation produced a decrease in the neutrophil recruitment in wild-type mice more pronounced than that obtained with inactivation of either PDE4D or PDE4B (Fig. 4). In addition, rolipram pretreatment significantly decreased neutrophil recruitment in PDE4B-deficient mice and, to a lesser extent, in PDE4D-deficient mice. A combination of rolipram plus inactivation of one of the PDE4 genes decreased neutrophil recruitment to levels similar to those observed with rolipram in wild-type mice.

A similar conclusion about the additive effects of PDE4 inactivation was reached using an alternative genetic approach, i.e., by measuring neutrophil recruitment in mice heterozygous for one PDE4 and homozygous null for the other. Mice homozygous null for both PDE4B and PDE4D could not be used in this study because they do not survive after birth. In PDE4B^−/−/PDE4D^+/− and PDE4B^+/−/PDE4D^−/− mice, neutrophil recruitment was decreased more than when a single PDE4 was inactivated (Fig. 5). These experiments suggest that PDE4D and PDE4B have complementary roles in the control of neutrophil recruitment.

The above data suggest that dysfunction of the neutrophils themselves rather than a disruption of chemotactic cues is the primary cause of the reduced recruitment of neutrophils. Neutrophil recruitment is dependent on expression of β₂ integrins upon activation. To investigate whether the expression of integrin subunits is one of the causes of decreased recruitment of neutrophils, the expression of the two α subunits, CD11a and CD11b, and the β₂ subunit, CD18, was assessed in neutrophils recovered from BAL after LPS treatment. Whereas no major differences were detected in the expression of either CD11a or CD11b (data not shown), the expression of CD18 was significantly reduced in PDE4B- and PDE4D-deficient neutrophils compared with that in wild-type cells (Fig. 6). It should be pointed out that the decrease in integrin expression observed in neutrophils should be considered a minimal estimate, because the neutrophils used in the assay have successfully migrated across the epithelium. We inferred that a more dramatic reduction in integrin expression would occur in cells that failed to migrate.

To further assess the possibility that neutrophils derived from PDE4-deficient mice have intrinsic dysfunction in migration, neutrophils were obtained from the spleen of wild-type and PDE4-deficient mice and used in a Boyden two-chamber chemotactic assay. In the presence of KC, wild-type neutrophils exhibited a concentration-dependent transmigration with an EC₅₀ of 10–15 ng/ml, values comparable to those previously reported (34). Conversely, the responses of both PDE4D- and PDE4B-deficient neutrophils were significantly reduced (Fig. 7,A). Similar results were obtained with MIP-2 (data not shown). When the wild-type cells were exposed to either 10 or 50 μM rolipram, a significant reduction in KC-induced neutrophil migration was observed (Fig. 7 B), and the decrease was comparable to that observed in the two PDE4-deficient neutrophils. This latter finding is in agreement with the hypothesis that PDE4 activity is necessary for efficient chemotaxis in response to chemokines.

To exclude the possibility that the disruption of chemotaxis is due to nonspecific generalized loss of cell function, neutrophils from wild-type and PDE4-deficient mice were tested for their phagocytic activity. Fig. 8 shows that phagocytosis of FITC-latex beads was identical in the three genotypes. This finding confirms that inactivation of PDE4 does not compromise all functions of neutrophils, and only chemokine-stimulated migration is affected.

Using a combination of pharmacological and genetic strategies, we provide evidence that the two PDE4B and PDE4D genes play a nonredundant function in polymorphonuclear neutrophil recruitment to the lung. Ablation of each of the two genes has a significant impact on the neutrophil ability for transepithelial migration in vivo. Because cytokine and chemokine accumulation is not greatly affected in the two mouse strains after LPS exposure, we surmise that this loss of function is intrinsic to neutrophils. Indeed, this reduction in recruitment to the lung in vivo is associated with an altered neutrophil expression of adhesion molecules and reduced chemotaxis measured in isolated cells in vitro. Both genetic and pharmacological evidence indicates that the functions of the two PDE4 genes are nonredundant and complementary, even though PDE4B has an impact on neutrophil function greater than that expected from its level of expression in these cells. These findings underscore the important role of cAMP homeostasis in neutrophil activation and open new avenues for pharmacological intervention. Thus, PDE4 subtype-selective inhibitors may be useful in the treatment of chronic inflammatory diseases of the lung.

Both immunological data and activity measurements in PDE4-deficient cells show that PDE4D and PDE4B are expressed in neutrophils, although at different levels. This finding is consistent with several reports on expression of PDE4 mRNAs in these cells from rodents and humans (19, 26, 35). PDE4 inhibitors also have been used to indirectly assess the presence of PDE4 in neutrophils (3, 26, 36). PDE4A may be expressed in these cells, albeit at very low levels, and its ablation does not produce significant effects on neutrophil recruitment. Despite the complex pattern of PDE expression in neutrophils, we have been able to dissect the function of each individual PDE4 expressed with selective ablation of PDE4 genes. Although we have measured major differences in the expression of PDE4D and PDE4B in neutrophils, the phenotypes associated with the ablation of the two genes are comparable. There are several possibilities that should be considered for the interpretation of these findings. First, PDE activity was measured in activated neutrophils that had emigrated from the vasculature and were recovered in the BAL. It is possible that before LPS stimulation, the proportion of PDE4D/PDE4B is different from that in activated cells. Indeed, PDE4B expression is induced by LPS exposure in monocytes/macrophages (18) even though it has been reported that the expression of PDE4B is constitutive in human neutrophils (19). Additional studies are required to address the possibility of changes in the pattern of PDE expression in neutrophils during their recruitment. In addition, that the level of PDE expression may not be a predictor of its impact on cell function should be considered. In several instances we observed that minimal changes in PDE activity have profound effects on cell function with no detected changes in overall cAMP levels (S.-L. C. Jin, L. Lan, M. Zoudilova, and M. Conti, manuscript in preparation). To explain these findings, we hypothesize that PDE4D and PDE4B control two different pools of cAMP in neutrophils, and disruption of either of these pools affects their ability to migrate, possibly by independent mechanisms. Because of the scarcity of cells available with the murine models, we have not been able to define whether the loss of neutrophil function in PDE4D- and PDE4B-deficient mice is indeed due to separate or overlapping mechanisms. Certainly, ablation of the activity of PDE4 is required for these effects because acute inhibition of PDE4 with rolipram produces a loss of function comparable to that produced by PDE4 genetic ablation. These findings also indicate that rolipram pharmacological effects on neutrophil recruitment are due to inhibition of both PDE4D and PDE4B, but not PDE4A.

From numerous studies of neutrophils as well as other inflammatory cells, it is generally agreed that activation of the cAMP pathway has a negative impact on activation of these cells. However, the mechanistic details of cAMP inhibition are unclear (37). Although we were not able to measure cAMP levels in neutrophils deficient in the two PDE4 genes, it is unlikely that major changes in cAMP levels would be observed. This view is supported by observations made in macrophages, where no detectable changes in cAMP were observed after PDE4 ablation (S.-L. C. Jin, L. Lan, M. Zoudilova, and M. Conti, manuscript in preparation). PDE4 inactivation most likely has subtle and no generalized disruptive effects on cAMP signaling. It has been shown that cAMP has a biphasic effect on permeabilized neutrophil migration because it is stimulatory at low concentrations, but inhibitory at high concentrations (15, 38). Thus, it is possible that ablation of PDE4 disrupts the fine balance required for chemotaxis that involves movement of signaling molecules between compartments (39). In human neutrophils, a cAMP analog abrogated the response to chemoattractants and caused inhibition of integrin-mediated adhesion through down-regulation of Rho activity (40). Moreover, integrin-mediated adhesion is inhibited by activation of PKA and requires A-kinase anchor protein binding (41). In addition, PDE4D and PKA are found in complex with AKAPs (42). It is then likely that the effect of PDE4 ablation leads to activation of PKA and interference with integrin expression and signaling. The formation of complexes of the chemokine receptor with β-arrestin-2 has been observed in response to a chemoattractant (43, 44). Fong et al. (45) showed that in mouse neutrophils in which β-arrestin-2 has been depleted, the chemotactic response to KC is defective. It is possible, then, that the function of β-arrestin-2 is also affected by PDE4 ablation in view of the observation that β-arrestins are in complex with PDE4 (46, 47).

Both PDE4B- and PDE4D-deficient neutrophils display a reduced expression of CD18. As discussed above, we believe that the reduction detected is more profound in neutrophils that failed to transmigrate. Whether this reduction in integrin is the major cause of the inability to transmigrate remains to be determined. Conflicting results have been reported on the effect of CD18 ablation on neutrophil recruitment to the lung (48). It is also unclear whether deficits in the vasculature or endothelial cells contribute to the phenotype. This is a possibility that should be investigated; however, in other paradigms of recruitment of inflammatory cells to the lung, no major deficits in eosinophil recruitment were observed in PDE4-null mice, arguing against a defect in the vasculature (49).

The effects of inactivation of the two PDE genes do not completely overlap with the effects of acute inhibition of PDE4 with rolipram. This discrepancy is most prominent for cytokine and chemokine production. With the caveat that considerable variation among these mice was observed, possibly due to the mixed background and outbreeding of the colony, there was no major difference among the three groups in TNF-α accumulation in the BAL fluid 1 h after LPS. A trend toward a decrease was observed at 2 and 4 h with both knockouts, even though statistical significance was reached in only one case. On the contrary, PDE4 inhibitors have been shown to markedly decrease LPS-induced TNF-α production in BAL (8, 50, 51, 52, 53). There are several explanations that could reconcile these differences. The first possibility is that nonselective PDE4 inhibitors have an effect because they block all PDE4s, whereas ablation of a single PDE4 is not sufficient to affect TNF-α accumulation. However, we have shown that ablation of PDE4B in leukocytes is sufficient to markedly suppress TNF-α production (18). Thus, it appears more likely that although inhibition of a single PDE4 may be sufficient to block TNF-α production in a given cell, a variety of cell types contributes to overall cytokine production in vivo. In the airway, TNF-α derives from monocytes/macrophages, airway epithelial cells, and perhaps neutrophils (54, 55, 56, 57). In our experimental conditions, it is possible that the major producers of TNF-α are perhaps epithelial cells or macrophages, and inactivation of a single PDE is not sufficient.

From a pharmacological standpoint, we propose that inhibitors with increased selectivity toward one PDE4 may have advantages over nonselective PDE4 inhibitors because they avoid potential side effects. Robichaud et al. (58) indicate that PDE4D, but not PDE4B, may be responsible for the emetic side effects often observed with PDE4 inhibitors. Thus, a compound that inhibits PDE4B preferentially should retain many of the beneficial properties of PDE4 inhibition (blockade of neutrophil adhesion and chemotaxis); however, it could probably be used at higher concentrations because of the decreased emetic effects.

In conclusion, we determined that PDE4B and PDE4D are both important for neutrophil recruitment during lung inflammation caused by LPS. The two PDEs have complementary roles, and their ablation affects both the expression of adhesion molecules and chemotaxis. Additional studies are required to determine the signaling pathways affected by the two PDEs. Understanding their exact function would certainly provide a rational basis for more selective pharmacological strategies.

We thank Dr. Stephen J. Galli and his colleagues for advice and support and for sharing the FACScan equipment, Drs. Brent Johnston and Eugene Butcher for advice on the chemotaxis assay, and Maria Zoudilova for the genotyping of the mice.