Many tumors, including Hodgkin's lymphoma, are associated with decreased cellular immunity and elevated levels of prostaglandin E2 (PGE2), a known inhibitor of CD4+ T cell activation, suggested to be involved in immune deviation in cancer. To address the molecular mechanisms tumor-derived PGE2 might have on primary human CD4+ T cells, we used a whole genome-based transcriptional approach and show that PGE2 severely limited changes of gene expression induced by signaling through the T cell receptor and CD28. This data suggests an interference of PGE2 at an early step of T cell receptor signaling: indeed, PGE2 stimulation of T cells leads to inactivation of lck and reduced phosphorylation of ZAP70. Antiapoptotic genes escaped PGE2-induced inhibition resulting in partial protection from apoptosis in response to irradiation or Fas-mediated signaling. As a functional consequence, PGE2-treated CD4+ T cells are arrested in the cell cycle associated with up-regulation of the cyclin/cyclin-dependent kinase inhibitor p27kip1. Most importantly, CD4+ T cells in Hodgkin's lymphoma show similar regulation of genes that were altered in vitro by PGE2 in T cells from healthy individuals. These data strongly suggest that PGE2 is an important factor leading to CD4+ T cell impairment observed in Hodgkin's lymphoma. (Cancer Res 2006; 66(2): 1114-22)
Prostaglandins are lipid molecules generated from arachidonic acid by cyclooxygenases and cell-specific prostaglandin synthases regulating numerous processes including modulation of immune function (1–3). Prostaglandin E2 (PGE2) is produced by many different cell types, including malignant cells, thereby playing an essential role in the carcinogenesis of different tumors associated with chronic inflammatory responses (4–6). Deletion of the respective prostaglandin receptors leads to reduced carcinogenesis and enhanced antitumor immunity (7). Elevated levels of PGE2 have also been found in patients with Hodgkin's lymphoma, which is suggested to be partially responsible for decreased cellular immune function in these patients (8, 9).
It is well known that PGE2 has diverse effects on CD4+ T cells leading to inhibition of T cell activation (10). Less is known about the mechanism of PGE2-induced T cell inhibition. Four different PGE2 receptors termed EP1, EP2, EP3, and EP4 are differentially expressed on different cell types and even differently regulated by inflammatory agents or PGE2 itself (11, 12). These receptors couple through their intracellular sequences to specific G proteins with diverse second messenger signaling pathways. EP2 and EP4 receptors couple to a Gs-type G protein leading to stimulation of cyclic AMP (cAMP; refs. 13, 14), whereas the EP1 receptor leads to increases of intracellular calcium (15, 16). To date, there are seven EP3 splice variants identified in humans. Originally, the EP3 receptor was described to couple to a Gi-type G protein leading to inhibition of intracellular cAMP, but subsequently, it was shown that individual splice variants also lead to stimulation of cAMP and inositol 1,4,5-trisphosphate (IP3) generation (17, 18). There is still a limited understanding of how these potentially conflicting signals are integrated and coordinated within the cell and little is known which of the receptors is mediating the PGE2-induced inhibition of CD4+ T cell activation. Using specific agonists and antagonists for cAMP and its target protein kinase A (PKA) it was indirectly shown that the inhibitory effects of PGE2 could be mediated by an increase of intracellular cAMP (1), suggesting Gs-coupled EP receptors like EP2 and EP4 to be responsible for PGE2-mediated inhibition of T cell activation. cAMP itself has been recognized as an important second messenger regulating immune responses through inhibition of various T cell functions like cytokine production and proliferation. The best described substrate of cAMP is PKA (19), and it has been shown that antigen-specific T cell proliferation and cytokine production are also inhibited by PKA type I (20). Recently, it was shown that PKA-I in response to cAMP activates the COOH-terminal src kinase (Csk; ref. 21). Activated Csk subsequently phosphorylates the COOH-terminal inhibitory tyrosine residue in lck and thereby acts as a negative regulator of T cell receptor (TCR) signaling (22–24). It has also been shown that activated PKA regulates cell proliferation via alteration of the cyclin/cyclin-dependent kinase (CDK) complex (25, 26). This kinase complex is involved in the phosphorylation and inactivation of the retinoblastoma protein thereby allowing progression to enter the S phase of the cell cycle. Activated PKA reduces cyclin D3 expression and induces expression of the Cdk inhibitor p27kip1. However, a direct effect of PGE2 leading to inhibition of lck or elevation of p27kip1 has thus far not been shown in primary human CD4+ T cells.
In this report, we used exclusively primary human CD4+ T cells to study the effects of PGE2 on T cell activation. Whole genome-based transcriptional analysis revealed an inhibition of a large number of TCR/CD28-regulated genes by PGE2, suggesting that PGE2 interferes with TCR/CD28 during an early step of signal transduction. We show that PGE2 leads to an inhibition of the src-kinase lck resulting in reduced phosphorylation of ZAP70. Additionally, we show that the inhibitory effects of PGE2 are in fact dependent on TCR signaling because CD4+ T cells activated with phorbol 12-myristate 13-acetate (PMA) and ionomycin, thereby circumventing the TCR, are not affected by PGE2.
By assessing genome-wide transcription profiles in CD4+ T cells derived from Hodgkin's lymphoma specimens, we found a similar transcriptional regulation of genes also altered in CD4+ T cells from healthy donors in response to PGE2 treatment in vitro. Thus, PGE2 might be an important mediator of impaired cellular immunity in patients with Hodgkin's lymphoma.
Materials and Methods
Isolation of CD4+ T cells and stimulation. Blood samples were collected from healthy blood donors at the Center for Transfusion Medicine (University of Cologne, Cologne, Germany) after informed written consent was obtained. CD4+ T cells were isolated by negative selection using RosetteSep CD4+ enrichment kit (StemCell Technologies, Meylan, France). The purity of the isolated CD4+ T cells was determined by flow cytometry and was routinely >80% CD3+ and CD4+, but was negative for CD8. Cells were cultured in RPMI 1640 supplemented with 10% FCS, glutamine, and antibiotics (Invitrogen Life Technologies, Karlsruhe, Germany) and stimulated at 37°C by mixing with artificial antigen-presenting cells (aAPC) at a ratio of 1:3 (cells/beads) comprised of magnetic beads (Dynal Biotech, Oslo, Norway) coated with the following antibodies: anti-CD3 (OKT3), anti-CD28 (9.3), and anti-MHC class I (W6/32). For most experiments, these aAPCs were coated with suboptimal anti-CD3 (5%), suboptimal anti-CD28 (14%), and anti-MHC-I constituting the remaining 81% of protein added to the bead, as previously described (27). In some experiments, CD4+ T cells were stimulated with 20 ng/mL PMA and 1 μmol/L ionomycin (Sigma-Aldrich, Munich, Germany).
CD4+ T cells from lymph node specimens were isolated by homogenizing the lymph node in RPMI. CD4+ T cells were purified using Miltenyi magnetic cell sorting columns (Miltenyi Biotech, Bergisch Gladbach, Germany) according to the manufacturer's instructions. All samples were taken after informed consent following Institutional Review Board approval.
RNA preparation, microarray hybridization, and microarray data processing. CD4+ T cells or subpopulations thereof (CD4+CD25− and CD4+CD25+) were either left unstimulated or stimulated with magnetic beads coated with CD3/MHC-I or CD3/CD28/MHC-I. PGE2-treated cells were also stimulated with CD3/CD28/MHC-I and PGE2 was added in a concentration of 1 μmol/L. After 8 hours, magnetic beads were removed and cells were lysed in TRIzol reagent (Invitrogen Life Technologies). RNA isolation and quantification was done as described previously (28). Biotin-labeled cRNA preparation was done using the Ambion Illumina RNA amplification kit (Ambion Europe, Huntington, Cambridgeshire, United Kingdom). cRNA (1.5 μg) was hybridized to Sentrix whole genome bead chips 6 × 2 (Illumina, San Diego, CA) and scanned on Illumina BeadStation 500×. For data collection and statistical analysis, we used Illumina BeadStudio software and dCHIP 1.3. The following filtering criteria were used for selection of differentially expressed genes: fold change ≥ 2, absolute difference in signal intensity between group means ≥ 50, P ≤ 0.05 (paired t test) and present call ≥ 25%. For gene ontology assessment and pathway visualization, GenMAPP, GeneSpring, and BioRag (http://www.biorag.org/) software was used.
Microarrays of CD4+ T cells isolated from lymph node specimens were done on the Affymetrix platform (Affymetrix, Santa Clara, CA). RNA isolation, quantification and target preparation was done according to standard protocols for small samples and cRNA was hybridized to HG-U133A arrays. Cross-validation of data from the two different array platforms used revealed a valid comparability of both methods.3
Cell lysis and Western blotting. Primary human CD4+ T cells (1 × 107 per sample) were cultured in RPMI/FCS and stimulated with aAPC with and without the addition of PGE2. Cells were lysed for 30 minutes on ice with 50 μL lysis buffer [5 mL Triton 1% (Promega, Mannheim, Germany), 750 μL 150 mmol/L NaCl (Sigma-Aldrich), 250 μL 50 mmol/L Tris-HCl (Invitrogen), 50 μL phosphatase inhibitor cocktail 1 and 2 (Sigma-Aldrich), protease inhibitor (Roche, Basel, Switzerland), 10 μL PMSF 1 mol/L (Sigma-Aldrich)] and pelleted for 10 minutes at 14,000 rpm at 4°C. Lysates were resolved by SDS-PAGE and transferred to nitrocellulose. Blots were probed with the following antibodies: anti-Bcl-xL (sc-8392), anti-p27 (sc-1641; both from Santa Cruz Biotechnologies, Santa Cruz, CA), anti-lck (pY505), anti-p-ZAP70 (both from BD PharMingen, San Diego, CA) and anti-actin (Chemicon, Temecula, CA). As secondary antibodies, horseradish peroxidase–conjugated antimouse IgG and horseradish peroxidase–conjugated antirabbit IgG (both from DakoCytomation, Glostrup, Denmark) were used.
Cytometric bead array for cytokines. The concentration of tumor necrosis factor-α (TNF-α) and IFN-γ in cell culture supernatants was measured using the human Th1/Th2 cytokine kit II (BD PharMingen). In brief, capture beads were mixed with culture supernatants and phycoerythrin detection reagent and incubated for 3 hours at room temperature. The beads were then washed with wash buffer and analyzed according to the manufacturer's recommendation.
PGE2 leads to inhibition of CD4+ T cell proliferation and cytokine secretion. Although much effort has been undertaken to analyze the mechanism of PGE2-mediated inhibition of T cell activation, most of the previous work has been done using tumor cell lines deficient in key signaling enzymes or deficient in the complete expression of EP receptors that are expressed on the surface of primary human CD4+ T cells (29). In fact, when analyzing EP receptor expression on the human T cell line Jurkat in comparison to primary human T cells, we observed expression of all four EP receptors on primary human CD4+ T cells, whereas Jurkat cells lacked expression of EP1 (data not shown). Therefore, we chose to use only primary human CD4+ T cells for further analysis.
To analyze the effects of PGE2 on primary human CD4+ T cells, we established a system in which freshly isolated T cells were stimulated with aAPCs in the form of magnetic beads coated with antibodies against CD3 and CD28 to deliver TCR stimulation (signal 1) and costimulation via CD80/CD86 (signal 2). This system allows one to vary the amounts of anti-CD3 and anti-CD28 coated on the bead surface, thereby mimicking differential strength of TCR and costimulatory signaling. We first used optimal (albeit unphysiologically high) TCR/CD28 stimulation by preparing aAPC coated with only anti-CD3 and anti-CD28 antibodies at a ratio of 1:1 on the bead surface. To analyze T cell proliferation, CD4+ T cells were labeled with 5,6-carboxyfluorescin-diacetate-succinimidyl-ester (CFSE) and subsequently stimulated with these aAPC. After 4 days of culture, CFSE dilution was determined by flow cytometry. Under these conditions, we did not observe a significant inhibition of CD4+ T cell proliferation in the presence of up to 1 μmol/L PGE2 (Fig. 1A , top). To address the inhibitory effects of PGE2 on CD4+ T cells under more physiological conditions, the amount of CD3 and CD28 coated on the bead surface was subsequently reduced to 5% anti-CD3 and 14% anti-CD28 of the total protein added to the bead. Anti-MHC-I was used as a binding, but not signaling, antibody to ensure a consistent and suboptimal loading of anti-CD3 and anti-CD28 antibodies to the bead surface. Furthermore, beads coated with anti-MHC-I antibody have been shown not to alter the gene expression profile of resting human T cells (30).
As shown in Fig. 1, stimulation of primary CD4+ T cells with CD3/CD28/MHC-I resulted in robust T cell expansion and cytokine secretion as previously described (31). The addition of PGE2 to the cultures inhibited T cell proliferation in a concentration-dependent manner (Fig. 1A,, bottom); 1 μmol/L PGE2 resulted in complete inhibition of T cell proliferation and this concentration was used in subsequent experiments. To analyze the effect of PGE2 on cytokine secretion, CD4+ T cells were stimulated as above and cytokines were measured in the collected supernatants. As shown in Fig. 1B, 1 μmol/L PGE2 resulted in strong inhibition of secretion of the cytokines IFN-γ and TNF-α.
To exclude the effect of regulatory CD4+ T cells on PGE2-mediated inhibition of T cell proliferation, conventional CD4+CD25− T cells were stimulated with CD3/CD28/MHC-I in the presence or absence of PGE2 and proliferation was assessed by CFSE dilution. Similar to the total CD4+ T cell population, CD4+CD25− T cell proliferation was inhibited by PGE2 (data not shown).
PGE2 induces overall leveling of transcriptional changes during T cell stimulation. To determine PGE2-mediated signaling events, we first did a whole genome-based transcriptional analysis. CD4+ T cells were either left unstimulated or stimulated with beads coated with CD3/MHC-I as a control or CD3/CD28/MHC-I with or without the addition of 1 μmol/L PGE2. Using stringent filter criteria, we detected no significant changes in the transcriptional profile after stimulation with CD3/MHC-I compared with resting cells (data not shown). However, 1,037 genes were significantly up-regulated compared with resting cells after stimulation with CD3/CD28/MHC-I. After the addition of PGE2, only 191 genes were detected to be up-regulated. Using Venn diagrams to visualize intersections, 177 of the 191 genes up-regulated after PGE2 treatment were also regulated after CD3/CD28/MHC-I stimulation, and only 14 genes were up-regulated in a CD3/CD28/MHC-I–independent fashion. Interestingly, PGE2 signaling targeted the genes that were down-regulated by CD3/CD28 less than those up-regulated. Five hundred and seventy-one genes were significantly down-regulated after stimulation with CD3/CD28/MHC-I beads, whereas 238 were down-regulated after additional PGE2 treatment. Two hundred and nineteen of the genes down-regulated after the addition of PGE2 were also found to be down-regulated after CD3/CD28/MHC-I stimulation, and 19 genes were down-regulated independently from CD3/CD28 (Fig. 2A). Using GeneMapp and BioRag software as tools to allocate regulated genes to different cellular pathways, we found a distinct up-regulation of genes belonging to metabolic and cellular signaling pathways after stimulating the cells with CD3/CD28/MHC-I beads, as expected. Additional treatment with PGE2 significantly reduced the number of genes associated with all pathways connected to CD3/CD28/MHC-I stimulation (Fig. 2B). This general leveling of CD3/CD28-regulated genes suggests an interference of PGE2-mediated signals during an early step of CD3/CD28-mediated signal transduction.
Genes belonging to antiapoptotic pathways escape PGE2-mediated inhibition of TCR signaling. After stimulation of T cells with CD3/CD28/MHC-I, different genes belonging to apoptotic pathways were transcriptionally regulated in comparison with unstimulated cells. These include genes encoding for TNF-α, TNF-β, Fas-L, and CD40-L with rather proapoptotic features but also for Bcl2, Bcl-xL, and Bcl2A1 with antiapoptotic attributes (Fig. 3, top). By analyzing unstimulated versus CD3/CD28/MHC-I–stimulated cells in the presence of PGE2 we found that PGE2 interfered with the majority of genes transcriptionally regulated after CD3/CD28/MHC-I stimulation, however, the antiapoptotic genes Bcl2, Bcl-xL, and Bcl2A1 escaped this PGE2-mediated inhibition. Bcl2 and Bcl2A1 showed the same or even stronger up-regulation compared with stimulation with CD3/CD28/MHC-I alone, Bcl-xL was found being strongly (14.91-fold) up-regulated after CD3/CD28/MHC-I and still 10.6-fold after additional treatment with PGE2. However, the up-regulation of TNF-α, TNF-β, and Fas-L was inhibited by PGE2 (Fig. 3, bottom). This data supports previous reports that PGE2 treatment leads to a blockade of Fas-L induction in T cells (32). The escape of antiapoptotic genes from PGE2-mediated inhibition may suggest that these genes are differently regulated by CD3/CD28 or might be less affected by PGE2-mediated signals.
Genes regulating cell cycle progression are significantly regulated by PGE2. Regulation of cell cycle and cell proliferation constitute other pathways of genes significantly regulated after additional PGE2 treatment. In a direct comparison of regulated genes after stimulation with CD3/CD28/MHC-I in the absence or presence of PGE2, we found genes acting as negative regulators of cell proliferation (e.g., FTH1; fold change, 2.67) or cell cycle (e.g., RGS2; fold change, 2.78) up-regulated after PGE2 treatment. Recently it was shown that FTH1 secreted from melanoma cells may suppress immune responses by preferential activation of regulatory CD4+ CD25+ T cells producing IL-10 and leading to changes in antigen-presenting cells (33, 34). RGS2 affects cell cycle progression by interaction with Gaq and stimulates its GTPase activity (35). The cell cycle inhibitor GADD45α (36, 37) was also up-regulated (2.06-fold) after PGE2, but this did not reach statistical significance between the different samples.
Taken together, the genome-wide analysis of transcriptional changes induced by PGE2 in human CD4+ T cells revealed at least three major aspects: (a) PGE2 interfered with the majority of genes regulated after TCR/CD28 stimulation rather than having a unique transcriptional profile suggesting an impairment of TCR/CD28 signaling after PGE2, (b) the escape of antiapoptotic genes from PGE2-mediated inhibition suggests that PGE2 might protect T cells from apoptosis, and (c) genes belonging to pathways negatively regulating cell cycle progression and cell proliferation were significantly regulated by PGE2, suggesting that the cells are alive but not capable of proliferating. Next, we were interested to investigate the functional consequences of the transcriptional changes induced by PGE2.
PGE2 induces partial protection from apoptosis in human CD4+ T cells. First, we aimed to evaluate the transcriptional up-regulation of Bcl-xL observed in the microarray experiments on protein level using Western blot. As shown in Fig. 4A, stimulation with CD3/CD28/MHC-I induced a strong up-regulation of Bcl-xL, which was sustained after additional treatment with PGE2.
To study the functional aspects of sustained expression of antiapoptotic proteins, CD4+ T cells were left unstimulated or stimulated with CD3/CD28/MHC-I with or without PGE2. After 3 days of incubation, samples were irradiated with 26 Gy and incubated for another 24 hours before staining the cells with Annexin V and 7-AAD to determine the percentage of apoptotic and necrotic cells. As shown in Fig. 4B (left), the addition of PGE2 during stimulation with CD3/CD28/MHC-I reduced the percentage of apoptotic cells after irradiation nearly to levels observed in resting cells. In a different set of experiments, cells were stimulated as above and incubated with Fas-L or control IgM for an additional 24 hours. As shown in Fig. 4B (right), PGE2 treatment reduced the sensitivity of the cells for Fas-L-induced apoptosis. Taken together, these functional results support the data obtained in microarray experiments.
PGE2 inhibits cell cycle entry of CD4+ T cells associated with elevated levels of p27kip1. Two hallmarks of T cell unresponsiveness are partial protection from apoptosis and significantly reduced transcriptional changes following encounters with antigens (38) similar to what we observed in T cells stimulated by TCR/CD28 in the presence of PGE2. CD4+ T cell unresponsiveness has been described to be associated with elevated levels of the cell cycle inhibitor p27kip1. Intracellular levels of p27kip1 are regulated by transcription, posttranslational modification, and ubiquitin-dependent degradation. We detected a slight but not significant transcriptional up-regulation of p27kip1 (1.67-fold) in transcriptome experiments when comparing T cell stimulation in the absence or presence of PGE2. This observation was confirmed on the protein level. Figure 4C shows that p27kip1 is down-regulated in CD4+ T cells after stimulation with CD3/CD28/MHC-I. The addition of PGE2 restored the amount of p27kip1 to levels observed in resting cells.
We next assessed whether these changes in expression of p27kip1 are associated with differential cell cycle regulation (Fig. 4D). After 3 days of stimulation, cells were fixed with paraformaldehyde and stained with propidium iodide to analyze the DNA content of individual cells. After stimulation with CD3/CD28/MHC-I, 42.4% of the cells were in G1, 42.6% in S phase, and 14.5% were in G2 phase of the cell cycle. However, after additional treatment with PGE2, 80.8% of the total cell population remained in G1, whereas only 14.7% entered S phase, respectively, 5.5% were in the G2 phase of the cell cycle. Thus, after PGE2 treatment, the majority of cells remain in G1 phase and do not proceed through the cell cycle.
PGE2 interferes with early TCR signaling events by phosphorylation of lck505. We next wanted to evaluate how PGE2 leads to the general leveling of gene transcription induced by TCR/CD28 stimulation (see Fig. 2). Increasing evidence relates the inhibitory effects of PGE2 to an elevation of intracytoplasmatic cAMP and further to activation of the cAMP-dependent target PKA. This hypothesis is due to (indirect) experiments with specific agonists and antagonists of the adenylate cyclase or PKA (20, 29). Indeed, we were able to corroborate these findings when using aAPC and primary human CD4+ T cells as well as the respective cAMP and PKA agonists (data not shown). However, a direct approach to confirm the hypothesis that PGE2 is acting on CD4+ T cells via cAMP and PKA would be to analyze directly the effects of PGE2 on possible downstream targets of PKA. As mentioned above, PKA activation ultimately leads to phosphorylation and thereby inactivation of lck. p-lck505 could therefore be used to monitor PKA-dependent signaling.
As shown in Fig. 5A (left), lck is phosphorylated at position 505 in resting CD4+ T cells. Activation of the cells for 5 minutes with CD3/CD28/MHC-I reduced the amounts of p-lck505, however, additional treatment with PGE2 restored the levels of p-lck505 to levels observed in resting cells. To confirm that elevated levels of p-lck505 in fact lead to impaired TCR signaling, we studied downstream TCR signals. As shown in Fig. 5A (right), stimulation with CD3/CD28/MHC-I for 5 minutes resulted in increased phosphorylation of ZAP70, which was reduced after additional treatment with PGE2.
If interference with early events in TCR signaling via phosphorylation of lck505 is the mechanism by which PGE2 acts to inhibit CD4+ T cell activation, bypassing the TCR by using the mitogens, PMA and ionomycin, should not result in the inhibition of T cell activation by PGE2. We therefore stimulated freshly isolated CD4+ T cells with CD3/CD28/MHC-I as a control or PMA/ionomycin with and without increasing concentrations of PGE2. As shown in Fig. 5B, T cell proliferation was inhibited by PGE2 in a concentration-dependent manner after stimulation with CD3/CD28/MHC-I. However, stimulation with PMA/ionomycin rendered the cells unresponsive for PGE2-mediated inhibition. Thus, PGE2-mediated proliferation arrest is dependent on stimulation of T cells via the TCR. This data further suggests that the mechanism of PGE2-mediated T cell inhibition acts via impairment of TCR signaling.
CD4+ T cells in Hodgkin's lymphoma show transcriptional changes associated with PGE2 signaling. As the potentially best tumor model studying chronic inflammatory responses in close proximity to tumor cells, we assessed T cells derived from lymph node biopsies of patients with Hodgkin's lymphoma. Elevated levels of PGE2 have been found in patients with Hodgkin's lymphoma (9). Monocytes, which have been described as being the source of higher PGE2 concentrations, and impaired cellular immune functions described in Hodgkin's lymphoma have been attributed in part to PGE2 (8). Therefore, we were interested in comparing the transcriptional profile of PGE2-treated CD4+ T cells from healthy donors with the transcriptional profile of CD4+ T cells from patients with Hodgkin's lymphoma. By analyzing CD4+ T cells from healthy donors and comparing unstimulated cells versus cells stimulated with CD3/CD28/MHC-I in the absence or presence of PGE2, we found a number of genes independently regulated after treatment with PGE2 (Fig. 2A; Supplemental Table S1). We used these genes as a filter prior to comparing the transcriptional profiles of CD4+ T cells isolated from lymph nodes of either five patients with Hodgkin's lymphoma or five patients with reactive lymph nodes. On the basis of this list of genes, CD4+ T cells from patients with Hodgkin's lymphoma present with a distinct transcriptional profile and cluster separately from CD4+ T cells isolated from reactive lymph nodes, as shown in Fig. 6A. The transcriptional changes between the two groups mirror significantly transcriptional changes after PGE2 treatment of healthy donor cells because many genes independently up-regulated after PGE2 treatment in vitro were also found to be up-regulated in patients with Hodgkin's lymphoma and vice versa (Supplemental Table S2).
An alternative possibility to analyze the transcriptional changes after treatment with PGE2 is to directly compare CD3/CD28/MHC-I–stimulated CD4+ T cells in the presence or absence of PGE2. This analysis results in a distinct list of genes regulated by PGE2. When using these genes as a filter, transcriptional profiles of CD4+ T cells from patients with Hodgkin's lymphoma also cluster separately from CD4+ T cells derived from reactive lymph nodes (Fig. 6B). By the use of other filter criteria (e.g., a list of cluster of CD markers or nuclear factor κB target genes) samples from patients with Hodgkin's lymphoma were not separated from patients with reactive lymph nodes (data not shown). To exclude the possibility that different ratios of T cell subpopulations from the different samples were responsible for the observed differences, we isolated CD4+CD25+ regulatory and CD4+CD25− conventional T cells from healthy donors and obtained transcriptional profiles. We used regulatory T cells as an example because they are particularly enriched in Hodgkin's lymphoma (39). We then extracted the data for genes specifically regulated by PGE2 (Supplemental Table S1) and used these genes to cluster regulatory and conventional T cells. Using several different clustering approaches, these genes never did segregate the two different T cell subpopulations (data not shown). Taken together, this data strongly suggests that CD4+ T cells in Hodgkin's lymphoma lymph nodes are under the influence of PGE2 leading to the transcriptional changes observed in human CD4+ T cells in the presence of PGE2 in vitro.
PGE2 has been implicated as a potential inhibitor of T cell function in the context of malignant disease (40, 41). Albeit the outcome of PGE2 signaling is well established, the molecular mechanisms involved are still not completely understood. The present study was designed to determine the potential mechanisms of PGE2 leading to inhibition of CD4+ T cell activation. We exclusively used primary human CD4+ T cells to achieve more physiologic conditions compared with cell line models defective in key enzymes of T cell signaling (42, 43). Analysis of the transcriptional profile after PGE2 treatment revealed that PGE2 leads to a general interference with the great majority of genes regulated by signals 1 and 2 rather than to a generation of a unique PGE2-associated transcriptional profile. This interference affected both genes up-regulated and down-regulated after CD3/CD28 stimulation suggesting an inhibition of early TCR-mediated signaling events. We showed that PGE2 leads to phosphorylation of lck505, thereby inactivating the most proximal event of TCR signaling. Moreover, we were able to show that PGE2-mediated inhibition of T cell activation is in fact dependent on activation of the cells via the T cell receptor because activation of T cells bypassing the TCR renders the cells unresponsive to PGE2-mediated inhibition. Alternatively, when providing unphysiologically high TCR/CD28 signals, the effect of PGE2 could also be overcome.
Integrating previous data, we would suggest the following model of PGE2-mediated T cell inhibition. Increasing evidence correlates PGE2-mediated events to elevated levels of intracytoplasmic cAMP and subsequent activation of PKA (20, 29). This data results from experiments using specific agonists and antagonists of cAMP and the cAMP-dependent substrates, PKA and EPAC. The fact that specific agonists of cAMP and PKA, but not EPAC, mirrored PGE2-mediated inhibition of T cell activation as well as that this inhibition was abrogated under the influence of specific antagonists of the adenylate cyclase or PKA provided indirect evidence of the cAMP/PKA pathway linked to the PGE2 receptors EP2 and EP4 to be responsible for PGE2-mediated signaling. Increase of cAMP and stimulation of PKA by respective chemical agonists can regulate the expression of p27kip1 and cyclin D3 to suppress proliferation of leukemic T cell lines and human peripheral blood lymphocytes (25, 26). Additionally, PKA type I, in response to cAMP, activates Csk in Jurkat cells and peripheral T cells (44). Activated Csk was shown to phosphorylate the COOH-terminal inhibitory tyrosine residue in lck (22–24), thereby acting as a negative regulator of TCR signaling. Altogether, these data suggested that PGE2 might be able to inhibit T cell activation via up-regulation of cAMP, activation of PKA, and subsequent activation of Csk inducing phosphorylation of lck505, thereby inhibiting TCR-mediated T cell activation. Here, we directly show that PGE2-mediated T cell inhibition is indeed mediated by inactivation of lck. This is associated with reduced ZAP70 phosphorylation and followed by a general leveling of the TCR-mediated transcriptional response.
Interestingly, transcripts for the antiapoptotic proteins Bcl2, Bcl-xL, and Bcl2A1 were not affected or were significantly less affected by PGE2 treatment, and PGE2-treated cells were in part protected from apoptosis induced by irradiation or Fas-L. Our observations are in line with a previous report by Seddon et al. (45). In this report, the peripheral T cell pool of p56lck (lck)-deficient mice was reconstituted by the expression of an inducible lck transgene. After switching off the transgene, a continued survival of peripheral naïve T cells was observed, confirming that T cell survival was independent of lck. However, the essential role of lck for homeostatic T cell proliferation became obvious because lck-deficient T cells did not proliferate in a T cell–deficient host environment in the absence of lck. These data further suggest that survival and TCR-driven homeostatic expansion possibly depend on different signals.
Cyclooxygenase-2 (COX-2) overexpression and PGE2 secretion is described to be associated with reduced T cell and dendritic cell functions in different solid tumors, especially lung (40), breast (41), and colon carcinoma (46). Selective COX-2 inhibitors are tested in clinical trials for prevention and tumor therapy (47–49) and preclinical data suggests that COX-2 inhibitors might act additively or synergistically with the specific chemotherapeutic agents used in the treatment of these tumors.
It is known that Hodgkin's lymphoma is associated with decreased cellular immunity (50, 51) and that lymphocyte function influences the prognosis (52). Furthermore, it is known that monocytes produce increased amounts of PGE2 in these patients (53, 54). We show that the transcriptional profile of CD4+ T cells in patients with Hodgkin's lymphoma has significant similarities to CD4+ T cells of healthy donors treated with PGE2. Moreover, on the basis of genes independently regulated after PGE2 treatment of CD4+ T cells from healthy donors, T cells originating from the lymph nodes of patients with Hodgkin's lymphoma could be separated from reactive lymph nodes by cluster analysis. This data further suggests the influence of PGE2 on CD4+ T cells in Hodgkin's lymphoma and supports the hypothesis that PGE2 might contribute to impaired T cell function in these patients.
In conclusion, our data provide direct evidence that PGE2 affects TCR/CD28-driven signals by inactivation of lck resulting in impaired activation of downstream signals as shown by decreased phosphorylation of ZAP70. The functional consequence is a state of CD4+ T cell unresponsiveness despite sufficient CD28 costimulation. Based on our observations, we postulate PGE2 as an important factor responsible for impaired CD4+ T cell function in Hodgkin's lymphoma, and potentially also in other tumors associated with increased levels of PGE2. Current and future work will elucidate whether T cells in other PGE2-associated tumors also show hallmarks of impaired T cell function. Moreover, the identification of these central mechanisms leading to T cell inhibition might open new avenues towards active cancer immunotherapy by reversing PGE2-mediated signaling events.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
J.M. Chemnitz and J. Driesen contributed equally to this work.
Grant support: Alexander von Humboldt Foundation via a Sofia-Kovalevskaja Award to J.L. Schultze. J.M. Chemnitz was supported by a fellowship from the Mildred Scheel Foundation from the Deutsche Krebshilfe, T. Zander was supported by the Frauke-Weiskam-Christel Ruranski foundation. J.L. Schultze is a member of the National Genome Research Network.
We thank our colleagues at the Division of Transfusion Medicine for their technical support.