Synthetic Prostacyclin Analogs Differentially Regulate Macrophage Function via Distinct Analog-Receptor Binding Specificities¹

Prostaglandin I₂ (prostacyclin; epoprostenol) is an oxygenated metabolite of arachidonic acid formed enzymatically by the sequential activities of cyclooxygenase and PGI synthase enzymes (1). It is produced constitutively by vascular endothelial and smooth muscle cells (2) and is induced under inflammatory conditions in vascular cells (3) and macrophages (4). PGI₂ is a potent vasodilator and antithrombotic agent (1) whose effects result from binding to a unique heptahelical G protein-coupled receptor termed the I prostanoid (IP)⁴ receptor (5). This receptor is G_αs-coupled and activates adenylate cyclase, resulting in an acute burst of intracellular cAMP.

In light of its potent vasodilatory properties, PGI₂ has been exploited as a pharmacological agent in the treatment of vasoconstrictive/ischemic diseases such as peripheral vascular arterial occlusive disease, cerebrovascular ischemia, and pulmonary arterial hypertension (PAH) (1). Although native PGI₂ is inherently unstable at room temperature with a t_1/2 of seconds to minutes (6), continuous infusions of a highly alkaline solution (pH 10.2 to 10.8) are approved by the U.S. Food and Drug Administration (FDA) in the treatment of PAH. To combat the technical difficulties associated with administering epoprostenol, more stable synthetic analogs of PGI₂ have been developed and approved for use in PAH. The compound treprostinil (UT-15, Uniprost), which has a half-life of several hours and is stable at room temperature, was approved by the FDA for continuous s.c. infusion. Treprostinil is a potent IP receptor agonist (7), although its specificity for this receptor is unknown. Iloprost (8) is an inhaled prostacyclin analog with a half-life between those of epoprostenol and treprostinil that was also approved by the FDA for use in PAH. Though iloprost is a potent IP receptor agonist that is commonly used in vitro, it also demonstrates significant binding to two of the four E prostanoid (EP) receptors, namely the G_αq-coupled EP1 and the G_αi-coupled EP3 subtypes (9).

Cyclic AMP is the quintessential intracellular second messenger that amplifies extracellular signals following the binding of G_αs-coupled receptors by appropriate ligands. Acute increases in cAMP have profound anti-inflammatory effects on a number of different cell types involved in both innate and acquired immunity (10). Our laboratory has studied the effects of cAMP-elevating compounds on the innate immune system, particularly in the lung (11, 12, 13). Recent work demonstrated that PGE₂ suppressed critical antimicrobial defense functions of alveolar macrophages (AMs), including phagocytosis, bacterial killing, and inflammatory mediator production, all in a cAMP-dependent manner, mediated via the G_αs-coupled EP2 and EP4 receptors (11, 12). Animal models have identified important roles for endogenously produced PGE₂ in regulating pulmonary host defense (14). Given this ability to stimulate cAMP production through the IP receptor, we hypothesized that PGI₂ analogs might inhibit pulmonary innate immunity in a manner analogous to PGE₂. This is particularly relevant given that: 1) PGI₂ analogs are prescribed for PAH; and 2) there are relatively few studies regarding the effects of PGI₂ on innate immunity.

To test our hypothesis in vitro, we isolated rat AMs and, for comparison, peritoneal macrophages (PMs) and examined the regulation of phagocytosis, bacterial killing, and inflammatory mediator production by stable PGI₂ analogs. For the first time, the present investigation identifies differences in the immunoregulatory properties of clinically administered PGI₂ analogs and their actions in AMs and PMs. These findings may predict important differences regarding the risk of infection for patients receiving such agents.

Pathogen-free 125- to 150-g female Wistar rats were obtained from Charles River Laboratories. Mice with a targeted disruption of the EP2 gene (15) backcrossed over 10 generations onto a C57BL/6 background (designated as EP2 knockout (KO) mice) were a gift from S. Narumiya (Kyoto University, Kyoto, Japan) and were obtained from ONO Pharmaceutical and bred in the University of Michigan Unit for Laboratory Animal Medicine (Ann Arbor, MI). Wild-type C57BL/6 mice were purchased from The Jackson Laboratory. Animals were treated according to National Institutes of Health guidelines for the use of experimental animals with the approval of the University of Michigan Committee for the Use and Care of Animals.

RPMI 1640 cell culture medium and a penicillin/streptomycin/amphotericin B solution were purchased from Invitrogen Life Technologies. Tryptic soy broth was supplied by Difco. LPS of Escherichia coli strain 055:B5, cytochalasin D, SDS, o-phenylenediamine dihydrochloride, saponin, MTT, and peroxidase-labeled monoclonal anti-rabbit IgG were purchased from Sigma-Aldrich. The nonspecific phosphodiesterase inhibitor 3-isobutyl-1-methylxantine (IBMX) was purchased from EMD Biosciences. PGE₂, carbaprostacyclin, iloprost, treprostinil, AH-6809 (EP2 receptor antagonist), and rabbit polyclonal anti-IP and anti-EP2 receptor Abs were from Cayman Chemicals. ONO-AE3-208 (EP4 receptor antagonist) was provided by ONO Pharmaceutical. Compounds requiring reconstitution were dissolved in DMSO. Required dilutions of all compounds were prepared immediately before use, and equivalent quantities of vehicle were added to the appropriate controls.

Resident AMs from rats and mice and PMs from rats were obtained via lung and peritoneal lavages. respectively, as previously described (16, 17) and resuspended in RPMI 1640 to a final concentration of 1–4 × 10⁶ cells/ml. Cells were allowed to adhere to tissue culture-treated slides or plates for 1 h (37°C with 5% CO₂) followed by two washes with warm RPMI 1640. Cells were cultured overnight in RPMI 1640 containing 10% FBS and 1% penicillin/streptomycin/amphotericin B before use. The following day cells were washed two times with warm medium to remove nonadherent cells.

The phagocytosis of IgG-opsonized, nonviable, FITC-labeled Escherichia coli BioParticles (Molecular Probes) was assessed as previously reported (11). Drugs of interest were added 5 min before phagocytosis was initiated. Results are expressed as a percentage of the control, to which only vehicle was added.

The ability of bacteria to survive within the macrophages was quantified using a tetrazolium dye reduction assay as described elsewhere (18, 19) and results are expressed as the percentage of survival of phagocytosed bacteria.

Rat or murine macrophages were harvested as described, and the ability of these cells to produce IL-6 and TNF-α was assessed by ELISA. Cells were treated according to a previously published protocol (13), including a 60-min preincubation (37°C with 5% CO₂) with compounds of interest in serum-free medium followed by incubation (37°C with 5% CO₂) for an additional 16 h in the presence of LPS 100 ng/ml (or vehicle) and 1% FCS. After this time, cell-free supernatants were collected and analyzed using commercially available ELISA kits for IL-6 (Assay Designs) or TNF-α (R&D Systems) per the manufacturer’s instructions.

Macrophages were cultured overnight in 6-well plates in RPMI 1640 at a concentration of 3 × 10⁶ cells/well. Cells were incubated for 30 min with the nonspecific phosphodiesterase inhibitor IBMX (250 μM) followed by 15 min in the presence or absence of compounds of interest. In some experiments the EP2 antagonist AH-6809 (100 μM), the EP4 antagonist ONO-AE3-208 (10 μM), or vehicle (DMSO) was added with IBMX for 30 min before adding PGI₂ analogs. The doses of AH-6809 and ONO-AE3-208 were based on previous reports (11, 20). Culture supernatants were aspirated and the cells were lysed by incubation for 20 min with 0.1 M HCl (22°C), followed by disruption using a cell scraper. Intracellular cAMP levels were determined by ELISA according to the manufacturer (Assay Designs).

Western blot analysis was performed as previously described (21). Briefly, whole cell protein extracts were obtained by lysing freshly harvested AMs and PMs in a buffer (50 mM Tris-HCl (pH 7.4), 25 mM KCl, 5 mM MgCl₂, and 0.2% Nonidet P-40) supplemented with protease inhibitors (Roche Diagnostics). Protein samples (40 μg) were resolved on 10% Tris-HCl polyacrylamide gels and subsequently transferred to a nitrocellulose membrane. Membranes were probed with commercially available rabbit polyclonal EP2 and IP receptor Abs (Cayman Chemicals) or a mouse monoclonal anti-β-actin Ab (Sigma-Aldrich) followed by HRP-conjugated anti-rabbit or anti-mouse secondary Abs and ECL Plus detection reagents (Amersham Biosciences). Relative band densities were determined by densitometric analysis using National Institutes of Health Image software and the ratios were calculated. The results were expressed as the ratio of IP to that of β-actin. In all instances, density values of bands were corrected by subtraction of the background values.

Data are represented as mean ± SEM and were analyzed with the Prism 4.0 statistical program (GraphPad). Comparisons among three or more experimental groups were performed with ANOVA followed by the Bonferroni correction as indicated. Differences were considered significant if p ≤ 0.05. All experiments were performed on at least three separate occasions unless otherwise specified.

Iloprost and treprostinil are approved for use in the treatment of PAH. The former is a standard and well-characterized IP receptor agonist that also binds EP1 and EP3 receptors (9), whereas treprostinil has an extended half-life and chemical stability but its specificity for the IP receptor is unclear. We assessed the ability of iloprost and treprostinil to inhibit FcR-mediated phagocytosis, a process known to be inhibited by acute rises in intracellular cAMP (11). Native PGI₂ (epoprostenol) was not used due to its inherent instability (22). We compared the potencies of these synthetic PGI₂ analogs with that of PGE₂, which suppresses FcR phagocytosis through cAMP-mediated signaling (11). Rat AMs or PMs were pretreated for 5 min with iloprost, treprostinil, or PGE₂ before incubation with IgG-opsonized E. coli. A concentration of 1 μM of these compounds was used for interdrug comparisons because: 1) 1 μM PGE₂ predictably evokes a significant increase in cAMP and substantially inhibits FcR-mediated phagocytosis; and 2) 1 μM iloprost elicited near maximal cAMP response in rat macrophages (see Fig. 4).

Interestingly, iloprost inhibited the FcR-mediated phagocytosis of IgG-E. coli by 51.4 ± 6.5% in PMs but did not significantly impair phagocytosis in AMs (Fig. 1,A). Dose-response experiments revealed that iloprost significantly inhibited phagocytosis in PMs more potently than in AMs across all drug concentrations examined (p < 0.001, comparing suppression in PMs vs AMs for each dose; Fig. 1 B). Similar results were seen with carbaprostacyclin, another synthetic IP receptor agonist, which is not FDA approved for use in humans but is commonly used in vitro (not shown).

As we have previously shown in AMs (10), PGE₂ also inhibits FcR-mediated phagocytosis by PMs (Fig. 1,A). We anticipated that treprostinil would show a similar profile as iloprost and carbaprostacyclin; however, treprostinil (1 μM) inhibited FcR-mediated phagocytosis effectively in both AMs and PMs by 35.0 ± 5.6 and 58.5 ± 5.7%, respectively (Fig. 1 A), an effect parallel to that of PGE₂. Results similar to these were obtained using an alternative assay for FcR-mediated phagocytosis with IgG-coated sheep RBCs (data not shown).

Depending on the cell of interest and the experimental approach, acute increases in cAMP can be found to either enhance or suppress LPS-stimulated IL-6 production (23, 24). In rat AMs and PMs, however, cAMP augments LPS-stimulated IL-6 generation (13). We therefore hypothesized that PGI₂ analogs would increase IL-6 production following LPS treatment in both AMs and PMs. Although this was indeed the case for iloprost and carbaprostacyclin in PMs, as was also observed for PGE₂, neither iloprost nor carbaprostacyclin (1 μM) affected LPS-stimulated IL-6 release by AMs (Fig. 2,A). By contrast, treprostinil significantly up-regulated LPS-stimulated IL-6 in both AMs and PMs (Fig. 2 B). These data are in accord with our findings in phagocytosis and indicate that treprostinil behaves more like PGE₂ in AMs than either of the other two PGI₂ analogs.

Along with phagocytosis and the generation of inflammatory mediators, microbial killing is a critical defense function of the macrophage. Little is known about the direct effects of PGI₂ or its analogs on the regulation of leukocyte bactericidal activity (25). We therefore examined the effect of stable PGI₂ analogs on bacterial killing using the Gram-negative pathogen Klebsiella pneumoniae (26). As demonstrated in Fig. 3, both PGE₂ and treprostinil dramatically increased the ability of phagocytosed K. pneumoniae to survive within both AMs and PMs. By contrast, both iloprost and carbaprostacyclin were relatively weak at inhibiting killing in either cell type. The ability of carbaprostacyclin to block bactericidal mechanisms was minimal and only reached statistical significance in the PM (Fig. 3).

The IP receptor predominantly signals through a G_αs-coupled activation of adenylate cyclase. We have previously demonstrated that immunosuppression of macrophage functions by PGE₂ follows a similar G_αs-coupled receptor-based mechanism (11). As shown in Fig. 4,A, treating AMs for 15 min with either iloprost or carbaprostacyclin (1 μM) failed to enhance cAMP significantly in AMs, in stark contrast to both treprostinil and PGE₂. It is unlikely that we missed an early and transient increase in cAMP generated by iloprost or carbaprostacyclin, as these experiments were conducted in the presence of the PDE inhibitor IBMX, which sustains cAMP elevations. By contrast with these results in AMs, both iloprost (Fig. 4,B) and carbaprostacyclin (not shown) significantly increased cAMP in PMs. At a concentration of 1 μM iloprost, cAMP levels increased 23-fold in PMs but only 3.9-fold in AMs. Both treprostinil and PGE₂ (1 μM) increased cAMP to similar levels in AMs and PMs (Fig. 4 C).

The fact that iloprost and carbaprostacyclin increase cAMP and suppress the effector functions in PMs to a greater degree than in AMs suggested that AMs and PMs may differ in their levels of expression of the IP receptor. Protein preparations from lysates of freshly harvested, unstimulated rat AMs and PMs were subjected to Western blot analysis for the detection of IP receptor protein using a polyclonal Ab (Fig. 5). In view of the similar suppressive effect of PGE₂ in both cell populations, we also examined EP2 expression. We did not explore differences in the other G_αs-coupled EP receptor, EP4, as we have previously demonstrated that rat AMs express very little functional EP4 (11). Although the EP2 receptor was expressed at similar levels in both AMs and PMs, we consistently found that the IP receptor was present in greater abundance (∼2.3-fold) in the latter population (Fig. 5). Thus, differences in cAMP generation (and downstream immunoregulatory pathways) between AMs and PMs in response to iloprost and carbaprostacyclin correlate with differences in IP receptor expression.

The above data demonstrate that treprostinil is a more potent inhibitor of AM antimicrobial defenses than other PGI₂ analogs, a phenomenon reflected in its ability to increase cAMP in AMs more potently than either iloprost or carbaprostacyclin. This might reflect the ability of treprostinil to stimulate the relatively low-abundance IP receptor more potently than other agonists (7). However, as we were unable to find published data regarding the specificity of treprostinil, we hypothesized that it might bind not only the IP receptor but also the EP2 receptor, the latter being the key receptor mediating cAMP responses to PGE₂ in the rat AM (11). To test this hypothesis, we pretreated rat AMs for 30 min with antagonists for either the EP2 receptor (AH-6809; 100 μM) or the EP4 receptor (ONO-AE3-208; 10 μM) followed by a 15-min treatment with 1 μM treprostinil (Fig. 6,A). As illustrated, AH-6809 blocked ∼74% of the increase in cAMP stimulated by treprostinil, whereas the EP4 selective antagonist had no significant effect. We next examined the ability of treprostinil to stimulate cAMP in AMs harvested from EP2 KO mice and compared their responses to cells from wild-type animals (Fig. 6,B). Confirming our pharmacological data, the absence of the EP2 receptor reduced the treprostinil-stimulated increase in cAMP by ∼68%. To document that the lack of an EP2 receptor could reduce the potency of treprostinil in regulating inflammatory functions of the AM, we pretreated EP2-null or wild-type AMs for 1 h with either treprostinil or PGE₂ followed by an overnight incubation with LPS (100 ng/ml). As shown (Fig. 6,C), the ability of treprostinil to suppress TNF-α production was reduced in the absence of the EP2 receptor to the same extent as was the activity of PGE₂. It is notable that the lack of EP2 did not completely abrogate the ability of PGE₂ to decrease TNF-α production (Fig. 6 C). This most likely reflects the fact that AMs from EP2 KO animals express increased amounts of the G_αs-coupled EP4 receptor, perhaps to compensate for the lack of EP2 (14).

PGI₂ is a cyclooxygenase-derived lipid mediator well known for its regulatory effects on vascular endothelial and smooth muscle cells as well as platelets. Less appreciated, however, is the influence PGI₂ has on the immune system. PGI₂ has been postulated to play a key role in regulating both innate and acquired immunity and the effects are, for the most part, immunosuppressive or anti-inflammatory, resulting from IP-mediated increases in cAMP (reviewed in Ref. 27). Inhibitory effects on innate immunity include the suppression of adhesion molecule expression by endothelial cells and leukocytes (27), the reduction of reactive oxygen intermediate production by neutrophils (28), the impairment of inflammatory cytokine generation by pathogen-stimulated macrophages (29) and dendritic cells (R.S. Peebles, unpublished observation), and the inhibition of NK-cell mediated cytotoxicity (30). The overproduction of PGI₂ is believed to play a role in the pathogenesis of certain bacterial infections, perhaps by direct effects on innate immunity (31). However, data regarding the influence of PGI₂ on lung innate immunity are lacking. Such investigations are important given that stable, long-acting synthetic analogs of PGI₂ are in clinical use in patients with chronic pulmonary diseases (e.g., PAH) whose lower respiratory tract defense mechanisms may be impaired.

Our previous results have demonstrated that the suppressive effects of PGE₂ on AM functions follow its ability to acutely elevate cAMP via G_αs-coupled receptors (11, 12). We therefore postulated that PGI₂, which itself signals via the G_αs-coupled IP receptor, would suppress innate immunity in a manner similar to that of PGE₂. Surprisingly, our studies found that standard PGI₂ analogs (iloprost and carbaprostacyclin) only weakly regulated AM phagocytosis, inflammatory mediatory generation, and bacterial killing but were as potent as PGE₂ in regulating PMs. For these studies we examined FcR-mediated phagocytosis, a standard model of phagocytosis about which much is understood regarding its regulation by eicosanoids and cAMP (11, 12). We were unable to identify prior studies of PGI₂-regulation of FcR-mediated phagocytosis and now demonstrate that IP receptor activation inhibits the ingestion of IgG-opsonized targets in resident rat PMs to a much greater extent than in AMs. That the inhibitory effects of IP receptor activation might extend beyond FcR-mediated uptake is suggested by a report that PGI₂ was able to block the fibronectin-mediated ingestion of gelatinized sheep RBCs by casein-elicited rat PMs (32).

To assess whether the effect of iloprost and carbaprostacyclin was limited to phagocytosis, we turned our attention to the ability of these drugs to regulate LPS-stimulated IL-6. IL-6 is a cytokine with mixed pro and anti-inflammatory effects in the lung (33). We chose this cytokine for a number of reasons. First, IL-6 is well recognized to be regulated by cAMP and, in particular, both PGE₂ and PGI₂ have been shown to modulate its synthesis during LPS stimulation (34, 35). Second, in the rat model both AM and PM IL-6 synthesis are profoundly regulated by PGE₂ under conditions of LPS treatment (13). Lastly, as discussed below, IL-6 production by human AMs has been shown to be regulated by treprostinil during LPS exposure (29). Our data (Fig. 2) demonstrating that neither iloprost nor carbaprostacyclin stimulate IL-6 production in rat AMs but do in PMs suggest that there is a proximal signaling difference between AMs and PMs that explains the differing susceptibility of various functions of these cells to these PGI₂ analogs.

The killing of K. pneumoniae by AMs can be inhibited by PGE₂-induced increases in cAMP (12). Regulation of bacterial killing by PGI₂ has only been studied directly in neutrophils, where it was not found to have an effect (25). We find in both AMs and PMs (Fig. 3) that iloprost and carbaprostacyclin are weak inhibitors of bacterial killing when compared with PGE₂. The reason that these compounds are so much weaker than PGE₂ in suppressing bacterial killing by PMs is not entirely clear and requires further study. However, these experiments again revealed differential regulation of AMs and PMs by these PGI₂ mimics, as both compounds were significantly more potent at inhibiting bacterial killing in the latter cell. Whether such effects are relevant in vivo has yet to be determined.

Because the IP receptor is G_αs coupled, we were able to measure intracellular cAMP responses as a surrogate marker of IP activation in the presence of iloprost and carbaprostacyclin (Fig. 4). These data not only confirmed differences in sensitivity of AMs and PMs to these drugs but importantly suggested a mechanism underlying such sensitivities: namely, that the IP receptor itself might be differentially expressed by these cells. Indeed, by Western blot analysis we found greater IP receptor protein expression in rat PMs than in AMs. Whether such discrepant receptor expression is the sole explanation for differences in cAMP synthesis between the two macrophage cell types requires further study. It remains possible that the coupling between the IP receptor and adenylate cyclases or the types and amounts of adenylate cyclase isoforms expressed in the two cells may differ. It is unlikely that differences in phosphodiesterase isoform expression explains our results, as we inhibited all cAMP phosphodiesterases with IBMX.

A quite unexpected result was the finding that treprostinil did not share the same regulatory phenotype as the other PGI₂ analogs. Treprostinil is clearly a potent IP receptor agonist (7), but its specificity for that receptor has not been published. We have recently found that treprostinil modulates T lymphocyte and dendritic cell functions in an IP receptor-independent fashion (R.S. Peebles, unpublished observation). Such data suggest that this drug may bind other G_αs-coupled receptors and we hypothesized that the EP2 receptor might be one such unintended target. Although treprostinil has been suggested to also be a ligand for the peroxisome proliferator-activated receptor β (PPARβ) receptor (36), we did not consider this to be a likely mechanism to explain our findings. Our treatments were brief (minutes) in the studies of cAMP elevation, FcR-mediated phagocytosis, and bacterial killing, whereas PPARβ is a transcription factor whose effects would be expected to require a number of hours. Though it remains possible that the effect of treprostinil on cytokine synthesis involved the PPARβ system because those experiments were conducted in the presence of treprostinil for >16 h, we have previously shown that cAMP signaling mechanisms are sufficient to explain such results (13). The finding that treprostinil is more promiscuous in binding G protein-coupled receptors than previously thought warns us against assuming that the pharmacological effects of this drug in vivo result from IP-mediated signaling exclusively.

Whereas we were unable to identify previous studies of either iloprost or carbaprostacyclin in regulating AMs, we identified one previous investigation of treprostinil involving human AMs (29). That study showed that treprostinil could dose-dependently regulate IL-6 production by LPS-treated AMs in a manner associated with inhibition of NF-κB activation. However, a requirement for the IP receptor (or an involvement of EP receptors) was not investigated.

In the rat, the AM appears to be less vulnerable to immunomodulation by IP receptor stimulation than the PM. Although teleological reasons for this are unclear, it has been noted before that AMs are relatively resistant to regulation by PGI₂ analogs (37). For example, Beusenberg et al. (37) noted that the stable PGI₂ analog DC-PGI₂ only weakly enhanced cAMP levels in guinea pig AMs when compared with PGE₂. Although direct comparisons of human AMs and PMs are lacking, the same investigative group separately published results of cAMP modulation by DC-PGI₂ in either human AMs (38) or PMs (obtained through peritoneal dialysis catheters) (39). Maximal cAMP responses to DC-PGI₂ (10 μM) were lower in AMs (3.31-fold) compared with PMs (6.34-fold).

In summary, we have shown that rat macrophages are regulated differently by stable PGI₂ analogs depending upon their anatomic source and the receptor-binding properties of the analog. We have also identified differences in IP receptor expression between AMs and PMs, which correlate with differences in cAMP responsiveness between the two cell types. Although the in vivo relevance of these findings has yet to be determined, the novel findings in this report bring to light important differences in the behavior of synthetic PGI₂ analogs that are already in clinical use. Our results suggest a novel hypothesis: important differences exist in the risks to the innate immune system associated with individual PGI₂ analogs. This hypothesis requires testing in the clinical setting. In this context, it will be important to consider additional medication- or disease-related factors that might influence the risks posed by PGI₂ analogs in particular subpopulations of patients.

We have no financial conflict of interest.