T cell–intrinsic prostaglandin E2-EP2/EP4 signaling is critical in pathogenic TH17 cell–driven inflammation

Background IL-23 is the key cytokine for generation of pathogenic IL-17–producing helper T (TH17) cells, which contribute critically to autoimmune diseases. However, how IL-23 generates pathogenic TH17 cells remains to be elucidated. Objectives We sought to examine the involvement, molecular mechanisms, and clinical implications of prostaglandin (PG) E2–EP2/EP4 signaling in induction of IL-23–driven pathogenic TH17 cells. Methods The role of PGE2 in induction of pathogenic TH17 cells was investigated in mouse TH17 cells in culture in vitro and in an IL-23–induced psoriasis mouse model in vivo. Clinical relevance of the findings in mice was examined by using gene expression profiling of IL-23 and PGE2-EP2/EP4 signaling in psoriatic skin from patients. Results IL-23 induces Ptgs2, encoding COX2 in TH17 cells, and produces PGE2, which acts back on the PGE receptors EP2 and EP4 in these cells and enhances IL-23–induced expression of an IL-23 receptor subunit gene, Il23r, by activating signal transducer and activator of transcription (STAT) 3, cAMP-responsive element binding protein 1, and nuclear factor κ light chain enhancer of activated B cells (NF-κB) through cyclic AMP–protein kinase A signaling. This PGE2 signaling also induces expression of various inflammation-related genes, which possibly function in TH17 cell–mediated pathology. Combined deletion of EP2 and EP4 selectively in T cells suppressed accumulation of IL-17A+ and IL-17A+IFN-γ+ pathogenic Th17 cells and abolished skin inflammation in an IL-23–induced psoriasis mouse model. Analysis of human psoriatic skin biopsy specimens shows positive correlation between PGE2 signaling and the IL-23/TH17 pathway. Conclusions T cell–intrinsic EP2/EP4 signaling is critical in IL-23–driven generation of pathogenic TH17 cells and consequent pathogenesis in the skin.

these results suggest that exogenously added PGE 2 induces COX-2 and produces PGE 2 244 endogenously and continuously as we reported previously, 51 which makes more 245 contribution to Il23r induction, and that indomethacin and COX-2 inhibitor block this 246 process. Indeed, the addition of stable EP2 and EP4 agonists overcame the Il23r 247 suppression by indomethacin (Fig E1, B). Therefore, these data together suggest that IL-248 23 stimulates Th17 cells to produce PGE 2 , which acts back to EP2 and EP4 on these 249 cells to augment Il23r expression in a positive feedback manner. 250 251

Induction of Il23r expression by IL-23 and PGE 2 -cAMP signaling is mediated 252
through not only STAT3 but also CREB1 and NF-κ κ κ κB 253 We then investigated transcription factors responsible for induction of Il23r expression 254 in Th17 cells by IL-23 and PGE 2 -EP2/EP4 signaling. Because IL-23 activates STAT3 to 255 induce Il23r expression, 52 we first examined the effect of a STAT3 inhibitor. The 256 M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT addition of STAT3 inhibitor VII suppressed Il23r expression not only by IL-23, but also 257 by db-cAMP, and both (Fig. 3, A), indicating that the db-cAMP action was also 258 mediated by STAT3. Consistently, Y705 phosphorylation of STAT3 was increased by 259 db-cAMP at 5 and 30 min (Fig E2, A), which were ameliorated not only by the addition 260 of STAT3 inhibitor VII but also by H-89 (Fig. 3, B), indicating the involvement of PKA 261 in db-cAMP-mediated Y705 phosphorylation of STAT3. Intriguingly, the Y1007/1008 262 phosphorylation of JAK2, a kinase responsible for STAT3 Y705 phosphorylation in 263 Th17 cells, was enhanced by db-cAMP, and this enhancement was suppressed by Src 264 Kinase Inhibitor I ( Fig E2, B), indicating cAMP-PKA activates STAT3 through c-Src-265 JAK2 pathway. 266 Although the above findings demonstrated that IL-23 and PGE 2 -cAMP 267 signaling converge at STAT3 activation, it is well known that other STAT3 activators, 268 such as IL-6 and IL-21, cannot substitute for IL-23 in the expansion of Th17 269 population, 32 indicating that STAT3 is not the sole transcription factor regulating 270 expression of Il23r. Since PKA activates CREB1, 36  S933 phosphorylation of NF-κB p105 subunit, a precursor of p50, in response to db-281 cAMP alone and its combination with IL-23 in Th17 cells (Fig. 3, E). The latter is 282 consistent with our previous finding in dendritic cells that PGE 2 -cAMP signaling 283 activates the p50 subunit 54 and a report that phosphorylation of p105 S933 is PKA-284 dependent. 55 We therefore examined the involvement of NF-κB in Il23r induction by 285 using Nfkb1-deficient mice (p105 KO) or CTP-NBD, a NF-κB inhibitor. Interestingly, 286 both genetic deficiency and pharmacological inhibition of NF-κB suppressed Il23r 287 induction in response to db-cAMP, IL-23 and in combination (Fig. 3

, F and G). 288
These results together suggest that the PGE 2  Since pathogenic Th17 cells should express various molecules in addition to IL-23R to 295 exert their pathogenicity, we next examined how PGE 2 -EP2/EP4-cAMP signaling 296 contributes to expression of such pathogenic genes in Th17 cells. CD4 + T cells were 297 cultured under the Th17-skewing conditions with IL-6 and TGF-β1 for 3 days, then 298 incubated with db-cAMP alone, IL-23 alone or both for 24 h, and subjected to 299 microarray analysis. The numbers of genes up/down-regulated more than 2-folds by 300 each stimulation were examined by Venn-diagrams (Fig. 4, A), and the genes expressed 301 in each cluster (Table E, [1][2][3][4][5][6][7][8] were subjected to heat-map analysis (Fig E3, A) and gene 302 ontology analysis (Fig E3, B; Table E, 9-11). Expression of representative genes in each evidences suggest that Th17 cells become pathogenic via the IL-23-IL23  328   receptor axis and play crucial roles in development of various autoimmune diseases  329 including psoriasis. 8,56,57 However, how these Th17 cells acquired the pathogenicity in 330 vivo and to what extent the microenvironment of diseases contributes to this process 331 remain to be defined. In the IL-23-induced psoriasis mouse model, gene expression of 332 enzymes involved in PGE 2 biosynthesis including Ptgs2 encoding COX-2, Ptges 333 encoding membrane-associated PGE synthase, mPGES1, and Ptges2 encoding 334 membrane-associated PGE synthase-2, mPGES2, were all up-regulated by IL-23 335 administration into the skin (Fig E4, A), which is consistent with clinical observation 336 that local PGE 2 levels are elevated in blister fluids from human psoriatic skin. 58 We 337 therefore hypothesized that IL-23 possibly activates PGE 2 -EP2/EP4 signaling, which 338 may contribute to psoriasis pathogenesis. 339 To test this hypothesis, we injected IL-23 into the skin of WT C57BL/6N mice 340 as well as EP2 knockout (KO) mice 46 with or without administration with a EP4 341 antagonist, AS1954813, 49 and assessed skin inflammation by ear thickness and 342 histology. The EP2 deficiency or the EP4 antagonism alone reduced IL-23-induced ear 343 swelling by half and attenuated edema and cell infiltration, and, when combined, led to 344 nearly complete suppression of IL-23-dependent skin inflammation (Fig. 5, A and B). 345 Blockade of EP2 and/or EP4 caused no alteration in PBS-injected control ear (Fig E3,  346 B). To examine at which step of inflammation EP2 deficiency and EP4 antagonism 347 exert their effects and whether it is related to generation of pathogenic Th17 cells, we 348 digested ear tissues and analyzed CD4 + T cell populations in the skin by flow cytometry. of Il23r by IL-23 injection (Fig. 5 D, right). These findings together indicate that the 363 EP2/EP4 signaling is indeed involved in the generation of pathogenic Th17 cells and 364 elicitation of inflammation in this model. We then asked whether T-cell intrinsic 365 EP2/EP4 signaling is responsible for these IL-23-induced phenotypes. To this end, we 366 used EP2 flox/flox mice 47 and EP4 flox/flox mice 48 and generated EP2 flox/flox EP4 flox/flox Lck-367 Cre + mice. EP2 flox/flox EP4 flox/flox Lck-Cre + mice showed no significant differences in the 368

numbers of total cells, B cells, T cells, CD4 T cells, CD8 T cells, Th1 cells, Th17 cells 369
and Treg cells in the thymus, spleen, lymph node and peripheral blood compared to 370 control WT Lck-Cre + mice (Fig E5, A). However, deficiency of both EP2 and EP4 371 selectively in T cells prevented accumulation of Th17 cells in the ear and almost 372 completely attenuated IL-23-induced skin inflammation (Fig. 5, E and F). These results 373 together therefore suggest that the T cell-intrinsic PGE 2 -EP2/EP4 signaling is critical 374 for the generation of pathogenic Th17 cells in psoriasis model. We also performed M A N U S C R I P T A C C E P T E D ACCEPTED MANUSCRIPT imiquimod (IMQ)-induced psoriasis model 8 , in which we applied IMQ to the ear of WT 376 or EP2 KO mice with or without EP4 antagonist for 6 days (Fig E6, A). We found that 377 ear swelling was also significantly reduced by EP2 deficiency and EP4 antagonism and 378 additively in combination similar to the results in IL-23-induced psoriasis model. 379 Given the above findings, we next examined the effects of COX inhibitors on 380 skin inflammation in IL-23-induced psoriasis model (Fig E6, B  Finally, to extrapolate our findings in mice to humans, we analyzed a public microarray 389 dataset on gene expression profiles in skin biopsies from psoriasis patients and healthy 390 control individuals, 61 with a particular interest in correlation of PGE 2 signaling and the 391 IL-23/Th17 pathway. As expected, psoriatic lesional skin overexpressed Th17 signature 392 genes (including IL23A, IL12B, IL23R, IL17A, IL17F, and IL22), STAT3 and NFKB1 393 (encoding NF-κB p105) (Fig. 6, A). Moreover, psoriatic lesional skin overexpressed 394 enzymes in PGE 2 biosynthesis, e.g., PTGS2, PTGES and PTGES2, and the EP4 receptor 395 (PTGER4) but under-expressed the PGE 2 degrading enzyme, 15-PGDH (encoded by 396 HPGD) (Fig. 6, A). Interestingly, expression of Th17 signature genes positively 397 correlated with those involved in PGE 2 biosynthesis (e.g. PTGES and PTGES2) and 398 receptor (e.g. PTGER4) but negatively correlated with HPGD (Fig. 6, B). In addition, the clinically effective anti-IL-23 therapy 62 down-regulated gene expression of not only  400 the IL-23/IL-17 pathway (e.g., IL23A, IL23R, IL17A) but also those in PGE synthesis 401 like PTGES (Fig. 6, C and D). These findings support a potential crosstalk between the 402 PGE 2 and IL-23/IL-17 pathways also in human psoriatic skin inflammation. 403 The IL-23-IL-23 receptor signaling plays a critical role in generation of pathogenic 405 Th17 cells in autoimmunity. 5 We have first found that PGE 2 synergizes with IL-23 and enhances Il23r 413 expression through EP2 and EP4, a finding consistent with the findings in human Th17 414 cells. 44 We have then found that IL-23 stimulation induces PGE 2 production in Th17 415 cells and the IL-23-induced Il23r expression was attenuated by the treatment of cells 416 with indomethacin or EP2/EP4 antagonists. These results thus suggest a previously 417 unsuspected intrinsic amplification mechanism mediated by the PGE 2 -EP2/EP4 418 signaling in Th17 cells that helps trigger the initial IL-23 responses in premature Th17 419

cells. 420
We have further analyzed the transcriptional mechanisms underlying the 421 synergistic action of IL-23 and PGE 2 , and found that this action is mediated by not only 422 STAT3 but also CREB1 and NF-κB. The involvement of CREB1 is analogous to that in 423 the PGE 2 -EP2/EP4-mediated Il12rb2 induction during Th1 cell differentiation, 42 and 424 may be consistent with the findings by Hernandez et al. 63 showing that the 425 CREB1/CRTC2 pathway regulates expression of IL-17A and IL-17F and that Th17 426 differentiation is defective in CRTC2 mutant mice. IL-23R and IL-12Rβ2 make a pair M A N U S C R I P T

A C C E P T E D ACCEPTED MANUSCRIPT
with the same molecule, IL-12Rβ1, to form IL-23 receptor and IL-12 receptor, 428 respectively. It is interesting that the same pathway regulates expression of these two 429 genes. We have also used T cells from p105 NF-κB1-deficeint mice and CTP-NBD and 430 unraveled the involvement of NF-κB in the IL-23/cAMP-induced Il23r expression in 431 Th17 cells. Consistent with these findings, we previously found that PGE 2 through EP2 432 or EP4 activates NF-κB1-containing NF-κB in various types of cells including 433 macrophages and dendritic cells, and induces expression of inflammation-related genes 434 including COX-2, which then produces PGE 2 and amplifies this process. 47 Cluster 4U by our microarray analysis (Fig. 4 B, left). However, we assume that this 441 mechanism does not critically operate in our experiment, because the addition of anti-442 IL-1β antibody to the medium did not reduce the Il23r induction ( Fig E7). 443 In addition to Il23r, our microarray analysis has revealed that stimulation of the 444 EP2/EP4 signaling together with IL-23 facilitates expression of a variety of pathogenic 445 Th17 signature genes (i.e. Il17a, Il17f, Il18r1 and Tgfb3). Interestingly, PGE 2 -EP2/EP4 446 signaling also up-regulated the expression of various genes related to chemotaxis and 447 migration such as S1pr1, Ccr2, Cxcl3, Cx3cr1, Cxcr4, Sema4f, Sell, Sema3c and 448 Sema6a (Fig. 4 B, left). These results suggest that PGE 2 -EP2/EP4 signaling may 449 contribute to migration, infiltration and accumulation of Th17 cells into inflammation Another topic to be discussed on PGE 2 in psoriasis is its facilitative action in 503 ultraviolet (UV) irradiation therapy, which at a glance contradicts our present findings 504 on the facilitative action of PGE 2 on Th17 pathogenicity. UVB-irradiation is an 505 effective therapeutic treatment of psoriasis by inducing immunosuppression. 72 We 506 previously showed that UVB induces PGE 2     (10 ng/ml) and PGE 2 (100 nM) or IL-23 (10 ng/ml), EP2 agonist (100 nM) and EP4 6 agonist (100 nM) in the absence or presence of indomethacin (100 µM) for 3 days, and 7 then harvested to analyze for Il23r expression by qRT-PCR.