Essential role for hematopoietic prostaglandin D2 synthase in the control of delayed type hypersensitivity

Abstract

Hematopoietic prostaglandin D₂ synthase (hPGD₂S) metabolizes cyclooxygenase-derived prostaglandin (PG) H₂ to PGD₂, which is dehydrated to cyclopentenone PGs, including 15-deoxy-Δ^12,14-PGJ₂ (15d-PGJ₂). PGD₂ acts through two receptors (DP1 and DP2/CRTH2), whereas 15d-PGJ₂ can activate peroxisome proliferator-activated receptors or inhibit a range of proinflammatory signaling pathways, including NF-κB. Despite eliciting asthmatic and allergic reactions through the generation of PGD₂, it is not known what role hPGD₂S plays in T helper (Th)1-driven adaptive immunity. To investigate this question, the severity and duration of a delayed type hypersensitivity reaction was examined in hPGD₂S knockout and transgenic mice. Compared with their respective controls, knockouts displayed a more severe inflammatory response that failed to resolve, characterized histologically as persistent acute inflammation, whereas transgenic mice had little detectable inflammation. Lymphocytes isolated from inguinal lymph nodes of hPGD₂S^−/− animals showed hyperproliferation and increased IL-2 synthesis effects that were rescued by 15d-PGJ₂, but not PGD₂, working through either of its receptors. Crucially, 15d-PGJ₂ exerted its suppressive effects through the inhibition of NF-κB activation and not through peroxisome proliferator-activated receptor signaling. In contrast, lymph node cultures from transgenics proliferated more slowly and synthesized significantly less IL-2 than controls. Therefore, contrary to its role in driving Th2-like responses, this report shows that hPGD₂S may act as an internal braking signal essential for bringing about the resolution of Th1-driven delayed type hypersensitivity reactions. Consequently, hPGD₂S-derived cyclopentenone PGs may protect against inflammatory diseases, where T lymphocytes play a pathogenic role, as in rheumatoid arthritis, atopic eczema, and chronic rejection.

Nonsteroidal antiinflammatory drugs have popularized the concept of inhibiting eicosanoids as a treatment for inflammation-driven diseases (1). However, it is now becoming clear that eicosanoids play both beneficial and detrimental roles during inflammation, depending on their site of action and the etiology of the inflammatory response (2, 3). Prostaglandin (PG) D₂ (PGD₂) and its cyclopentenone PG (cyPG) dehydration products are a good example of arachidonic acid metabolites that exert both protective and destructive effects, being beneficial for the resolution of innate immunity on the one hand (4) but instrumental in the pathogenesis of asthma and models of allergic inflammation on the other (5). PGD₂ is formed by the actions of two types of PGD₂ synthase isoforms (6). In general, one is present in the central nervous system, testis, and the human heart and is called lipocalin PGD₂ synthase. The second is present in the spleen and hematopoietic system (hPGD₂S), being widely distributed in antigen-presenting cells, T helper (Th)2 lymphocytes, mast cells, and megakaryocytes, where it metabolizes cyclooxygenase-derived PGH₂ to PGD₂. Although currently controversial, it is believed that PGD₂ is initially converted to the dehydration products PGJ₂ and 15d-PGD₂ in an albumin-independent manner. Thereafter, 15-deoxy-Δ^12,14-PGJ₂ (15d-PGJ₂) and Δ¹²-PGJ₂ are generated from PGJ₂ by albumin-independent and -dependent reactions, respectively (7).

PGD₂ is emerging as one of the predominant factors released in large amounts by mast cells during asthmatic attacks in humans (8). PGD₂ binds to and activates two G-protein-coupled receptors, DP1 and CRTH2 (chemoattractant receptor-homologous molecule expressed on Th type 2 cells), also known as DP2. Within the immune system, DP1 activation regulates human and murine dendritic cell migration (9), whereas CRTH2/DP2 is expressed on type 2 lymphocytes (10), eosinophils (11), and basophils (12), where it mediates eosinophil chemotaxis and is currently implicated as a central player in promoting Th2-related allergic inflammation (13). Although 15d-PGJ₂ was identified as a high-affinity natural ligand for peroxisome proliferator-activated receptor (PPAR)γ, it is now thought to exert its effects through PPARγ-dependent and -independent mechanisms, resulting in the suppression of various proinflammatory signaling pathways (2). These include pathways that operate through NF-κB, AP1, and signal transducers and activators of transcription as well as the suppression of inducible nitric oxide synthase and proinflammatory cytokine synthesis. Thus, although PGD₂ has been shown to drive allergic response, presumably synthesized by hPGD₂S, comparatively little is known about the role of this enzyme in the evolution and resolution of Th1-driven responses, such as rheumatoid arthritis, eczema, and chronic rejection.

Therefore, we have tested the hypothesis that hPGD₂S has a role in the resolution of Th1-mediated inflammatory responses. To this end, we used methylated BSA (mBSA)-induced delayed type hypersensitivity (DTH) in hPGD₂S^−/− and transgenic mice. hPGD₂S was found to counterregulate the severity of a DTH response as well as its longevity. These protective effects were mediated through terminal 15d-PGJ₂ inhibiting lymphocyte NF-κB activation and not through either PPAR signaling or DP (DP1 or DP2/CRTH2) receptor activation. These findings advance our understanding of the role of hPGD₂S, PPARs, PGD₂, and the cyPGs in inflammation, provide new insights in the endogenous control of adaptive immunity, and underline the modulatory properties of eicosanoids in host defense and autoimmunity. Consequently, although hPGD₂S may be pathogenic in asthma, it may be protective in other chronic inflammatory diseases, and it is these diverse properties of eicosanoids that, more than ever, highlight the need to tailor inhibitors of eicosanoid biosynthetic pathways for specific disease processes.

Results

Inflammatory Response in hPGD₂S^−/− Mice Bearing a DTH.

Paw swelling in wild types peaked 24 h after challenge, remained elevated at 48 h, and began to resolve by 72 and 96 h. In contrast, swelling in hPGD₂S^−/− animals was immediately apparent at 1 h, remained significantly greater than wild-type controls at all subsequent time points, and failed to resolve by 96 h, after which time, experiments were terminated in accordance with United Kingdom Home Office regulations (Fig. 1A). By contrast, paw swelling in hPGD₂S transgenics was virtually undetectable compared with the wild-type controls (Fig. 1B). Thus, a T cell-driven immune response was more severe and failed to resolve in hPGD₂S^−/− mice but was blunted in transgenic mice, suggesting an important role for hPGD₂S in controlling the severity and duration of DTH reactions. Histologically, injected paws from wild-type and hPGD2S^−/− mice showed predominantly acute inflammation, as demonstrated by an infiltrate consisting mostly of neutrophils (see Fig. 6B, which is published as supporting information on the PNAS web site). Smaller numbers of mononuclear cells with morphological features of monocytes, macrophages, and lymphocytes were also present, with few detectable eosinophils. The extent and severity of the neutrophil infiltrate was greater in knockout mice than in wild types (Fig. 6). Neutrophils infiltrated the epidermis, skin adnexae, dermis, skeletal muscle, adipose tissue, and synovium. In parallel to the inflammatory cell infiltrate, there was edema of the dermis, spongiosis of the epidermis, keratinocyte mitoses (including suprabasal mitoses), and, interestingly, morphological features of keratinocyte apoptosis, all of which appeared more prominent in the knockouts. In the epidermis of hPGD2S^−/− mice, neutrophils accumulated superficially in foci of keratin crust and also formed small subcorneal microabscesses (Fig. 6B), whereas, in the dermis and skeletal muscle of some of the knockouts, there were dense accumulations of neutrophils forming small microabscesses (Fig. 6B), histological changes that were less apparent in wild types.

Fig. 1.

hPGD₂S controls the severity and duration of DTH responses. C57BL/6 mice lacking hPGD₂S (■) and wild-type controls (□) (A) along with FVB mice overexpressing this enzyme (♦) and their appropriate controls (□) (B) were sensitized at the base of the tail and challenged in the left paw 14 d later with mBSA, which induced a DTH reaction. Contralateral paws received saline. Paw swelling measurements were made by water-displacement plethysmometry at selected time points after challenge. n = 6–8 animals per group; ∗, P ≤ 0.05; ∗∗, P ≤ 0.01, as determined by ANOVA, followed by Bonferroni t test, with data expressed as mean ± SEM.

Proliferation and Cytokine Analysis.

Inguinal lymph nodes from both wild types and knockouts were excised 24, 48, and 96 h after mBSA challenge and incubated for 72 h in the presence of the specific antigen (mBSA) or the polyclonal T cell activator concanavalin A (Con A). In response to these stimuli, proliferation of lymph-node-derived cells from wild-type animals was maximal when extracted at 48 h, whereas proliferation of those extracted 96 h after challenge was lower, with proliferation being expectedly higher in the Con A-stimulated group (Fig. 2A). However, the proliferative capacity of lymph node cultures from knockouts was significantly greater than controls at all time points in both mBSA- and Con A-treated cultures (Fig. 2A), mirroring the exaggerated inflammatory response exhibited in the paws of hPGD₂S^−/− mice, (Fig. 1A). IL-2 was then measured in the supernatants of these cultures at 24 h (peak of IL-2 synthesis in this system) and found to be significantly higher in knockouts compared with controls (Fig. 2B), which may explain the heightened proliferation of ex vivo lymph node cultures in hPGD₂S^−/− mice. In contrast, proliferation of lymph node cultures from hPGD₂S transgenics was significantly lower than their appropriate controls (Fig. 2C) and had correspondingly lower IL-2 synthesis (Fig. 2D), reflecting the reduced inflammatory response exhibited by these animals (Fig. 1B). These data suggest that hPGD₂S may act as an internal braking signal on T cell proliferation, possibly by controlling IL-2 synthesis.

Fig. 2.

hPGD₂S acts as a braking signal for lymphocyte proliferation and IL-2 synthesis. Inguinal lymph nodes from hPGD₂S^−/− mice previously sensitized to mBSA to trigger a DTH reaction were incubated with the specific antigen (mBSA) for 72 h (A), revealing significantly increased proliferation of these cells, and for 24 h (B), showing enhanced IL-2 synthesis. Similar experiments were carried out on mice overexpressing hPGD₂S, where proliferation (C) and IL-2 synthesis (D) were found to be significantly reduced compared with controls. n = 5 animals per group; ∗, P ≤ 0.05; ∗∗, P ≤ 0.01, as determined by t test, with data expressed as mean ± SEM.

PG Synthesis in Lymph Nodes and Inflamed Paws.

Given that mBSA-elicited DTH is T cell-driven, we then determined the temporal profile of PG synthesis in the lymph nodes of inflamed animals from onset to resolution. Inguinal lymph nodes were excised from wild-type and hPGD₂S^−/− mice 8, 24, 48, and 96 h after mBSA challenge and cultured for 24 h to allow for accumulation of PGs within the culture medium. PGD₂ was synthesized within lymph node cultures of sensitized but not challenged animals (0 h), with synthesis decreasing 3-fold as the lesion progressed up to 48 h. From 48 h onward, however, PGD₂ synthesis increased dramatically up to 96 h, in line with resolution, Fig. 3A. Total lymph-node levels of PGD₂ followed a similar profile: 150 ± 20 pg per node at 0 h, reducing to 100 ± 25 pg per node at 24 h, and rising to 450 ± 55 pg per node at 96 h. Similarly, 15d-PGJ₂ synthesis was comparatively high at the early-onset phase of the response, decreased as the lesion progressed up to 48 h and increased again up to 96 h, as the inflammatory response began to resolve (Fig. 3B). Levels of PGE₂ were also measured in the lymph node cultures of wild-type animals and found to change only slightly from onset to resolution, with no significant difference in PGE₂ levels detectable between wild-type and hPGD₂S^−/− animals (Fig. 3C). Importantly, PG generation was antigen-specific, because mice that were sensitized with complete Freund’s adjuvant (not including mBSA) and challenged 2 weeks later with mBSA displayed PG levels that were 50–80% less across the time course than mice that were sensitized and challenged with mBSA. Inflamed-paw levels of PGD₂ were also determined in this DTH model. Similar to lymph nodes, PGD₂ was comparatively high in naïve, uninflamed paws (1,980 ± 210 pg per paw), waned as the inflammation progressed, up to 24 h (1,255 ± 160 pg per paw), but increased again as inflammation resolved at 96 h (2,650 ± 250 pg per paw). Thus, hPGD₂S-derivd PGs were detectable in the naïve tissues but were elevated dramatically during the resolving phase in both lymph node cultures and inflamed tissues of a DTH.

Fig. 3.

Synthesis of hPGD₂S-derived PGD₂ and 15d-PGJ₂ increases during inflammatory resolution. Inguinal lymph nodes were excised from groups of mBSA-sensitized wild-type and hPGD₂S^−/− mice 8, 24, 48, and 96 h after mBSA challenge and cultured for 24 h to allow for accumulation of PGs within the culture medium. Synthesis of PGD₂ (A) and its metabolite 15d-PGJ₂ (B) was found to increase during resolution in wild-type (□) animals, with levels barely detectable in hPGD₂S^−/− (■) samples. (C) PGE₂ was also measured in culture supernatants and found not to change substantially through the time course of the reaction in wild types (□), and no significant difference was detectable in PGE₂ levels between wild types and knockouts (■). n = 3–5 animals per group. Data are expressed as mean ± SEM.

The cyPG 15d-PGJ₂, but Not PGD_2, Rescues Lymphocyte Hyperproliferation in hPGD₂S^−/− Mice.

Because PGD₂ is metabolized to 15d-PGJ₂, we next sought to determine the relative roles of PGD₂ and 15d-PGJ₂ in the control of DTH reactions. Therefore, inguinal lymph nodes were excised from mBSA-sensitized and -challenged animals and lymph-node-derived cells incubated with BW245C (DP1 agonist), 15R-PGD₂ (DP2 agonist), or 15d-PGJ₂ for 24 h, followed by a determination of lymphocyte proliferation and IL-2 production. The concentrations of BW245C used were based on those shown to activate the DP1 receptor (0.9–2.5 nM) (14, 15), whereas those for 15R-PGD₂ were of a concentration shown to activate the DP2 receptor (1 μM) (16). The 15d-PGJ₂ was used at levels known to alter various proinflammatory signaling pathways and protein expression (17). DP receptor agonists were without effect on proliferation of lymph node cultures (Fig. 4A) and culture media levels of IL-2 (Fig. 4B) obtained from hPGD₂S^−/− mice. This surprising lack of effect of DP2 receptor activation on proliferation and cytokine synthesis was confirmed by using 15(R)-15 methyl PGD₂, again, at concentrations that selectively activate DP2 (1.25, 2.5, and 5.0 nM, data not shown) (18). In contrast, the hyperproliferation characteristic of hPGD₂S^−/− mice was reversed by 15d-PGJ₂, whereas levels of IL-2 were also significantly reduced (Fig. 4 A and B) at a concentration of the cyPG that did not alter lymph node culture apoptosis as determined by annexin V labeling and morphology (data not shown). Taken together, these data suggest that hPGD₂S controls DTH reactions, not through PGD_2, but, at least in part, by its dehydration cyPG product 15d-PGJ₂. Importantly, the results of these experiments support a role for 15d-PGJ₂ in the pathophysiology of the inflammatory response.

Fig. 4.

The 15d-PGJ_2, but not PGD₂, reverses lymphocyte hyperproliferation and IL-2 synthesis in hPGD₂S^−/− mice. Inguinal lymph nodes were taken from mBSA-sensitized animals 24 h after in vivo challenge and incubated ex vivo for 24 h with BW245C (DP1 receptor), 15R-PGD₂ (DP2 receptor agonist), and 15d-PGJ₂. Only 15d-PGJ₂ was found to reverse the hyperproliferative phenotype (A) and IL-2 synthesis (B) of cells deficient in hPGD₂S, whereas DP1 and DP2 agonists were without effect. In the experiments shown in B, BW245C and 15R-PGD₂ were also used at 30 nM and 2 μM, respectively, without detectable effect (data not shown). Experiments with a second DP2 receptor agonist, 15(R)-15 methyl PGD₂ at 1.25, 2.5, and 5.0 nM confirmed original findings (data not shown). Importantly, the neutralizing property of 15d-PGJ₂ was effective at doses that did not cause lymphocyte apoptosis. n = 3–5 animals per group; ∗, P ≤ 0.05; ∗∗, P ≤ 0.01, as determined by ANOVA, followed by post hoc Dunnett’s test, with data expressed as mean ± SEM

The 15d-PGJ₂ Signals Through NF-κB and Not PPARs in Controlling Antigen-Induced Inflammation.

We sought to determine through which intracellular signaling pathways (PPARs or NF-κB) 15d-PGJ₂ exerts its suppressive effect on lymphocyte functioning. Lymph node cultures were incubated with rosiglitazone (PPARγ agonist) or 15d-PGJ₂, with or without a preincubation period with GW9662 (PPARγ antagonist) at concentrations of rosiglitazone (19) and GW9662 (20) that are known to alter PPARγ signaling (0.1–10 μM). Rosiglitazone had no effect on lymph node culture proliferation, whereas antagonizing PPARγ by prior incubation for 1 h with GW9662 had no effect on the antiproliferative effects of 15d-PGJ₂ (Fig. 5A). Because 15d-PGJ₂ can also activate PPARα (21) and β/δ [David Bishop-Bailey (William Harvey Research Institute, London), personal communication] in some cell systems, lymph node cultures from wild-type animals were incubated with 0.1–3.0 μM PPARα agonist GW7647 or 0.01–1 μM PPAR β/δ agonist L-165041, with no detectable effect (data not shown). NF-κB is also required for the proliferation and production of cytokines in response to T cell activation and is a well described target for cyPGs (22). As a result, we determined NF-κB DNA binding in lymph node extracts from hPGD₂S^−/− as well as transgenic mice and found that NF-κB binding was higher in both antigen- and Con A-stimulated lymph node cultures of hPGD₂S^−/− mice compared with wild-type controls (Fig. 5B). Conversely, binding of NF-κB in mice overexpressing hPGD₂S was lower than corresponding controls (Fig. 5C). These experiments suggest that the suppressive effects of hPGD₂S-derived 15d-PGJ₂ on lymphocyte function may be independent of PPAR signaling but, at least in part, depends on the inhibition of NF-κB activity.

Fig. 5.

The inhibitory property of 15d-PGJ₂ on lymphocyte proliferation is not mediated through PPAR signaling but, at least in part, through inhibition of NF-κB. (A) Inguinal lymph nodes were taken from mBSA-sensitized hPGD₂S^−/− and wild-type animals 24 h after in vivo challenge and incubated for 24 h with a PPARγ agonist (rosiglitazone) or antagonist (GW9662) as well as 15d-PGJ₂ alone or with GW9662, where GW9662 was preincubated for 1 h before adding the cyPG. As before (Fig. 4), only 15d-PGJ₂ altered proliferation, whereas PPARγ agonists and antagonists were without effect. Even prior blocking of PPARγ did not alter the rescuing effects of 15d-PGJ₂ on knockout cells when compared with corresponding controls. A PPARα agonist (GW7647, 0.1 and 1.0 μg/ml) as well as a PPARβ/δ agonist (L-165041, 0.01–1 μM) were also without effect on proliferation (data not shown). In contrast, NF-κB binding was found to be greatly enhanced in lymph node extract from hPGD₂S^−/− animals (B), including freshly isolated cells from sensitized mice (labeled C) as well as those stimulated for 24 h with either the specific antigen (mBSA, labeled M) or Con A (nonspecific, labeled A). The converse also held true for hPGD₂S transgenic mice (C), where NF-κB binding was reduced in all treatment groups. n = 3–5 animals per group; ∗∗, P ≤ 0.01, as determined by ANOVA, followed by post hoc Dunnett’s test, with data expressed as mean ± SEM.

Discussion

We found that hPGD₂S^−/− mice bearing a DTH reaction display an exaggerated inflammatory response that fails to resolve. Although hPGD₂S-derived PGD₂ and the cyPGs possess potent but diverse biological roles in host defense, the suppressive effects of hPGD₂S on T lymphocyte functioning appear to be mediated by 15d-PGJ₂ and its inhibition of NF-κB DNA binding, with no contribution from PGD₂-activated DP1 or DP2/CRTH2 receptors. These findings suggest an important role for hPGD₂S as a checkpoint controller in the progression from acute to resolving inflammation, underpinned by the changing profile of PGD₂ and 15d-PGJ₂ synthesis throughout the time course of this DTH, possibly resulting from a change in T cell subsets that populate the inflamed lymph node during a DTH with Th1 cells at onset giving way to Th2 cell at resolution; Th2, but not Th1, cells express hPGD₂S (23). Levels of both eicosanoids are comparatively high in the lymph nodes at onset but drop as animals are challenged, suggesting that hPGD₂S might play an important role in controlling the early-onset phase of the response, possibly acting as an endogenous brake on lymph node cell proliferation. The reappearance of hPGD₂S-derived PGs during resolution would support an additional role in limiting the duration of DTH reactions, findings that are reminiscent of studies showing a role for cyPGs in the resolution of innate inflammation (4). Whether the absence of hPGD₂S predisposes to chronic inflammation or autoimmunity has yet to be determined. Nonetheless, the clear lack of inflammation in animals that overexpressed hPGD₂S further reinforces the critical role this downstream PGH₂-metabolizing enzyme plays in the etiology of T lymphocyte-driven immune responses.

DP2/CRTH2 receptors are not found on Th1 cells but are present on Th2 and on a small percentage of CD8⁺ T cells, where the receptors are coupled with Gαi-type G protein to induce cell migration in eosinophils, basophils, and Th2 cells (24). In contrast, DP1 is not present on T cells (5) but is expressed on dendritic cells, where it regulates their maturation and migration (9, 25). Despite this understanding of DP receptor activation on T cell signaling, the lack of effect of DP1 and DP2/CRTH2 agonists on proliferation and cytokine production by lymph node cultures obtained from both hPGD₂S knockouts and wild types was surprising. An equally unexpected observation was that, whereas 15d-PGJ₂ rescued the hyperproliferative phenotype of hPGD₂S^−/−-derived lymph node cultures, activation of PPARγ was without effect. Moreover, although 15d-PGJ₂ was also shown to activate PPARα (21) and PPARβ/δ (David Bishop-Bailey, personal communication), we found that selective activation of these PPARs was also without effect on lymph node cell proliferation, suggesting PPAR-independent protective effects of 15d-PGJ₂. However, the role of 15d-PGJ₂ and PPAR signaling in T lymphocyte functioning is controversial and depends highly on the source of the T cell (primary or cell lines) as well as the concentrations of PPAR agonists and cyPGs used (26–29). Thus, when investigating the role of 15d-PGJ₂ and PPARs in T cells, it is important to use an appropriate lymphocyte model system to interpret results in the context of in vivo adaptive immune responses. In this study, we used inguinal lymph node cultures, obtained from sensitized animals, maintained in an ex vivo environment that best represents that which occurs in the intact sensitized lymph node in vivo. These cultures comprised lymphocytes as well as antigen-presenting cells stimulated with doses of 15d-PGJ₂ and PPAR agonists that do not cause apoptosis but at concentrations known to selectively activate their respective receptors. Notwithstanding, although our finding shows a role for cyPGs only in this setting, it must be pointed out that DP receptors and PPARs may control other facets of the adaptive immune response that would not otherwise be observed in the ex vivo culture system used in this report, including dendritic cell migration/T cell interaction, T cell receptor activation, and antigen presentation.

One of the most well described PPAR-independent properties of 15d-PGJ₂ is the inhibition of NF-κB activation. Through Michael addition reactions, the highly reactive electrophilic carbon atom in the unsaturated carbonyl group of the cyclopentenone ring can react with nucleophiles, such as the free sulfydryl groups of glutathione and cysteine residues that form disulphide bonds in proteins (for detailed review see ref. 2). In this report, we found that NF-κB activation within mBSA-sensitized lymph nodes of hPGD₂S^−/− animals was increased, whereas, in mice that overexpress hPDG₂S, NF-κB DNA binding was decreased. In T lymphocytes, NF-κB is most strikingly required for proliferation and the production of cytokines in response to T cell activation (22). In addition, 15d-PGJ₂ is also known to inhibit activation of AP-1 (17), which is an intracellular signaling protein that is also involved in lymphocyte proliferation and cytokine synthesis (30). Interestingly, DNA binding of this signaling factor was enhanced in the lymph node cultures of hPGD₂S^−/− mice, including basal and antigen-stimulated (mBSA) as well as nonspecifically by Con A (data not shown). Thus, from these data, it seems that 15d-PGJ₂ exerts its protective effect in adaptive immunity, not through PPAR signaling directly, but through the inhibition of proinflammatory signaling pathways central to T lymphocyte functioning.

The rescuing of T cell proliferation and cytokine synthesis by 15d-PGJ₂ only, in hPGD₂S^−/− mice, raises the controversial issue of cyPGs in pathophysiology. It was shown that, although 15d-PGJ₂ is detectable in human urine, levels are low and are unaltered in response to lipopolysaccharide (31). Moreover, 15d-PGJ₂ biosynthesis is not augmented in the joint fluid of patients with arthritis nor is its urinary excretion increased in patients with diabetes or obesity, circumstances where PPARγ ligand are used therapeutically (31). As a result, Bell-Parikh et al. questioned the role of 15d-PGJ₂ as an endogenous ligand for PPARγ, in particular, and in the pathophysiology of inflammatory diseases, in general. In this study, we found that, in an animal system where synthesis of 15d-PGJ₂ is depleted, parameters of inflammation (proliferation and cytokine synthesis) are exaggerated and that this hyperresponsiveness is rescued by exogenous addition of 15d-PGJ₂. The degree of rescue was not total in its effect, suggesting that perhaps some other intermediate cyPGs (PGJ₂ or Δ¹²-PGJ₂, for instance) may also play a role in controlling lymphocyte functioning in DTH responses. Nonetheless, these experiments not only support a role for 15d-PGJ₂ as an endogenous controller of T cell functioning in vivo but that cyPGs may, indeed, be expressed during and instrumental in immune responses. However, given its short half-life and propensity to bind avidly to sulfydryl groups, one possible reason why 15d-PGJ₂ is not readily detectable in biological fluids is because it rapidly reacts with proinflammatory signaling proteins including NF-κB. Perhaps a more appropriate strategy is to look, for example, for intracellular 15d-PGJ₂/protein adducts.

In summary, we have shown that hPGD₂S negatively regulates the severity and duration of DTH responses. In doing so, we have shown the role of this enzyme in adaptive immunity and that it exerts these protective effects mainly through its terminal cyPGs, underlining a central role for these eicosanoids in the etiology of T cell-mediated events. In reporting these data, we are also highlighting the diverse roles hPGD₂S plays in inflammation and caution against targeting hPGD₂S as a general antiinflammatory therapy, because although its metabolites are detrimental to allergic reactions, they protect against immune responses, whereas Th1 cells play a pathogenic role. In doing so, these data highlight the diverse role eicosanoids play in inflammatory diseases and underscore the need to tailor inhibitors of their biosynthetic pathways appropriately for diseases where they are pathogenic.

Materials and Methods

Animal Maintenance, Generation of hPGD₂S-Altered Mice, and Induction of Antigen-Induced Inflammation.

hPGD₂S knockout and transgenic mice were generated as described in detail in Fig. 7 A–C, which is published as supporting information on the PNAS web site. Animals were bred under standard conditions and maintained in a 12-h/12-h light/dark cycle at 22 ± 1°C and given food and tap water ad libitum in accordance with United Kingdom Home Office regulations. Mice were sensitized at the base of the tail with a 50-μl injection of mBSA in Freund’s complete adjuvant (20 mg/ml solution of mBSA in saline emulsified with an equal volume Freund’s complete adjuvant containing 4 mg/ml Mycobacterium tuberculosis H37 RA; Difco). Inflammation was induced 14 d later by a subplantar challenge with 50 μl of mBSA (10 mg/ml) in saline, with the contralateral paw receiving saline only and serving as control. Inflammation is reported as the difference in paw swelling as determined by water-displacement plethysmometry (Ugo Basile) between left and right paws. All experiments in this article were carried out with the stated number of animals per group mentioned in the figure legends and repeated on three separate occasions to test for consistency.

Histology.

Footpads were fixed in formal saline, decalcified in 10% formic acid, dehydrated, and embedded in paraffin. Blocks were sectioned longitudinally at 10 μm and stained with hematoxylin and eosin.

Ex Vivo Cell Culture.

Inguinal lymph nodes were dissected and placed in RPMI medium 1640 supplemented with 5% FCS, 2 mM sodium pyruvate, 25 μM mercaptoethanol, and 100 μg/ml gentamycin (Sigma). Lymph nodes were dispersed through a 50-mesh tissue sieve and filtered through a 70-μm sterile cell sieve (Becton Dickinson) to yield a single-cell suspension, counted by hemocytometer, and cultured in round-bottom 96-well plates at 2 × 10⁶ cells per ml in the above medium. Cultures were left unstimulated to serve as controls or stimulated with 25 μg/ml of the specific antigen mBSA or 5 μg/ml of the nonspecific mitogen Con A. Cells were cultured at 37°C in a humidified atmosphere with 5% CO₂ for 24–72 h. Cell viability and apoptosis were determined by measurement of supernatant levels of lactate dehydrogenase (Sigma) and annexin V labeling (Sigma), respectively, as described in ref. 32.

In Vitro Stimulation with 15d-PGJ₂ or DP and PPAR Agonists.

Lymph node cells were cultured as above and treated with 25 μg/ml mBSA and one of 15d-PGJ₂, BW245C (DP1 agonist), or 15(R)-PGD₂ as well as 15(R)-15 methyl PGD₂ (DP2 agonists), rosiglitazone (PPARγ, Cayman Chemical, Ann Arbor, MI), and the PPARα and PPARβ/δ agonists GW7647 and L165041 (Calbiochem). Drug vehicle (methyl acetate- or DMSO-) treated cells were found not to differ significantly from unstimulated controls, with the former serving as the control for all experiments. The effects of antagonizing PPARγ were investigated by preincubating lymph node cells with the PPARγ antagonist GW9662 (Cayman Chemical) for 1 h, followed by costimulation with mBSA and 15d-PGJ₂, rosiglitazone, or vehicle. All experiments were carried out in triplicate and repeated on three separate occasions to confirm consistency.

Proliferation Assays.

After 60 h of culture, lymph node cells were pulsed with methyl [³H]thymidine [1.0 μCi (1 Ci = 37 GBq), Amersham Pharmacia] in a 10-μl volume of RPMI medium 1640. Twelve to 18 h later, cells were harvested onto filter paper by using a semiautomatic cell harvester (Skatron). Filter paper was left to dry for at least 2 h at 60°C, after which time [³H]thymidine incorporation was assessed by liquid-scintillation counting for 1 min. All experiments were performed in triplicate and repeated on three separate occasions. Proliferative responses were calculated as stimulation indices (SI) (SI = cpm in stimulated cultures − cpm in unstimulated cultures/cpm in unstimulated cultures).

PG, IL-2, and EMSA for Measurement of NF-κB Activity.

PGs in lymph node culture supernatants or inflamed paws (pulverized with a nitrogen bomb) were acidified and extracted by using C-18 Sep-Pak cartridge columns (Waters). PGD₂, PGE₂, and 15d-PGJ₂ were measured by using competitive enzyme immunoassays (PGD₂ and PGE₂, Cayman Chemical; 15d-PGJ₂ R & D Systems). IL-2 production was measured by using commercially available ELISA (Pharmingen), whereas NF-κB activity in lymph node extracts was determined by EMSA as described in ref. 33.

Statistics.

Statistical analysis was performed by using ANOVA, followed by a post hoc Bonferroni test when repeated readings taken in a continuous time course were used. Where multiple treatments of differing action or dose were compared with a single control, data were assessed by ANOVA, followed by a post hoc Dunnett’s test. All other comparisons between two variables were assessed by using a paired t test.

Abbreviations

Con A

concanavalin A

DTH

delayed type hypersensitivity

hPGD₂S

hematopoietic prostaglandin D₂ synthase

PG

prostaglandin

15d-PGJ₂

15-deoxy-Δ^12,14-PGJ₂

PPAR

peroxisome proliferator-activated receptor

cyPG

cyclopentenone PG

mBSA

methylated BSA

DP

PGD₂ receptor

Th

T helper.