IL-4 up-regulates cyclooxygenase-1 expression in macrophages

Macrophages use various cell-surface receptors to sense their environment and undergo polarized responses. The cytokines, interleukin (IL)-4 and IL-13, released from T-helper type 2 (Th2) cells, drive macrophage polarization toward an alternatively activated phenotype (M2). This phenotype is associated with the expression of potent pro-resolving mediators, such as the prostaglandin (PG) D2-derived cyclopentenone metabolite, 15d-PGJ2, produced by the cyclooxygenase (Ptgs; Cox) pathway. Interestingly, IL-4 treatment of bone marrow-derived macrophages (BMDMs) significantly down-regulates Cox-2 protein expression, whereas Cox-1 levels are significantly increased. This phenomenon not only challenges the dogma that Cox-1 is only developmentally regulated, but also demonstrates a novel mechanism in which IL-4-dependent regulation of Cox-1 involves the activation of the mechanistic target of rapamycin complex (mTORC). Using specific chemical inhibitors, we demonstrate here that IL-4-dependent Cox-1 up-regulation occurs at the post-transcriptional level via the Fes-Akt-mTORC axis. Activation of AMP-activated protein kinase (AMPK) by metformin, inhibition of mTORC by torin 1, or CRISPR/Cas9-mediated genetic knock-out of tuberous sclerosis complex-2 (Tsc2) blocked the IL-4-dependent expression of Cox-1 and the ability of macrophages to polarize to M2. However, use of 15d-PGJ2 partially rescued the effects of AMPK activation, suggesting the importance of Cox-1 in macrophage polarization as also observed in a model of gastrointestinal helminth clearance. In summary, these findings suggest a new paradigm where IL-4-dependent up-regulation of Cox-1 expression may play a key role in tissue homeostasis and wound healing during Th2-mediated immune responses, such as parasitic infections.

In addition to signaling through the Jak/Stat pathway, IL-4 receptor engagement also up-regulates protein translation through the recruitment of Fes (8). Fes activates phosphatidylinositide 3-kinase (PI3K) leading to the generation of phosphatidylinostol (3,4,5)-triphosphate. 3-Phosphoinositide-dependent protein kinase-1 (Pdk-1) is activated by membrane phospholipids and is a major regulator of protein kinase B/Akt signaling (9,10). Additionally, mTORC activity increases the phosphorylation of Akt. Akt inhibits tuberous sclerosis complex (TSC), activating mTORC. mTORC increases translation by activating ribosomal protein S6 kinase ␤-1 (p70S6K) and releasing elongation factor eIF4E from its inhibitor 4EBP. S6 kinase negatively feeds back to inhibit PI3K activity (11). Therefore Akt is both regulated and regulates mTORC signaling pathways. Finally, mTORC has been shown to play a critical role in M2 macrophage polarization in both peritoneal macrophages and BMDMs with IL-4 stimulation increasing the expression of Tsc1 and Tsc2 protein in stimulated macrophages (12).
Even though Cox-1 (Ptgs-1) and Cox-2 (Ptgs-2) have a high degree of sequence homology (ϳ63%), they differ in many respects from gene structure to enzymatic activity. Cox-1 has been generally considered as a constitutive enzyme, with some exceptions such as up-regulation of transcription in fibroblasts by TGF␤ and IL-1␤ and during cellular differentiation of monocytes with phorbol ester in THP-1 cells (27). Cox-2 is subjected to transcriptional and post-translational regulation by diverse stimuli (28 -31). Although Ptgs-2 promoter contains many regulatory elements present in "early immediate genes," the Ptgs-1 promoter is more reminiscent of "housekeeping" genes lacking TATA and CAAT boxes, high GC content, and multiple transcriptional start sites (TSS) (28 -31). Additionally, Ptgs-2 mRNA, as well as protein, has a short half-life of 2-7 h, whereas Ptgs-1 mRNA and protein is relatively stable (29). The instability of Cox-2 protein is due to a 27-amino acid cassette at the C terminus, not present in Cox-1, which targets it for degradation through the endoplasmic reticulum-associated protein degradation system followed by ubiquitination and proteolysis by 26S proteasome (29,32).
Here we describe for the first time the post-transcriptional regulation of Cox-1 expression by IL-4 in primary macrophages (both human and mouse) to be mediated through the Fes-Akt-mTORC axis rather than the Jak/Stat pathway. Activation of AMPK or direct inhibition of mTORC by torin 1 blocked the IL-4-dependent expression of Cox-1 and affected macrophage polarization toward the M2 phenotype. Use of 15d-PGJ 2 partially rescued the effects of AMPK activation in mice infected with N. brasiliensis, suggesting the importance of Cox-1 in the polarization of macrophages. In summary, these studies suggest a novel paradigm where IL-4-dependent post-transcriptional control of Cox-1 expression may play a key role in tissue homeostasis and wound healing during Th2-mediated immune responses, such as parasitic infections.

IL-4 stimulation increases the expression of Cox-1
To determine the effects of IL-4 stimulation on Cox expression, murine BMDMs were treated with or without recombinant IL-4 for 20 h prior to collection. Cells were collected and analyzed by real-time PCR and Western immunoblot for Ptgs-1 (Cox-1) and Ptgs-2 (Cox-2) expression, and by flow cytometry to quantify M2 macrophages (CD206 ϩ Arg-1 ϩ F480 ϩ CD11b ϩ ). In addition, lipids were extracted from cell culture media supernatants to quantify the production of 15d-PGJ 2 . Ptgs-1 transcript levels were unaffected by IL-4 treatment (Fig. 1A). Cox-1 protein levels were increased 2.5-fold with IL-4 stimulation compared with untreated control BMDMs (Fig. 1B). Interestingly, the Ptgs-2 transcript was unaffected (data not shown) and Cox-2 protein was not detected in either untreated control or IL-4-treated BMDMs by Western immunoblot (Fig. 1C). Additionally, there was no increase in mPges-1 (microsomal PGE 2 synthase) or Txas-1 (thromboxane synthase) expression with IL-4 stimulation compared with untreated control BMDMs (supplemental Fig. S1, A and B). LC-MS/MS analysis of BMDM culture media supernatants showed a statistically significant increase in the production of 15d-PGJ 2 20 h post-IL-4 treatment as compared with untreated control BMDMs (Fig. 1D). IL-4 stimulation did not lead to an increase in the production of PGE 2 or TxB 2 compared with untreated control BMDMs (supplemental Fig. S1, C and D). BMDMs treated with the TxAS inhibitor, ozagrel, where able to polarize toward an M2 phenotype in response to IL-4 at similar levels to DMSO-treated BMDMs (supplemental Fig. S1E). Treatment with ozagrel lead to a decrease in the production of TxB 2 (supplemental Fig. 1F). Finally, IL-4 stimulation led to approximately a 10-fold increase in the expression of M2 macrophages that was previously shown to be sensitive to indomethacin (Fig. 1E) (17). Macrophages were first gated by their forward and side scatter characteristics followed by CD11b ϩ F480 ϩ cells, which consisted of ϳ98% of the macrophage population, and were further gated on CD206 ϩ and Arg-1 ϩ (supplemental Fig.  2A). Furthermore, human macrophages differentiated from frozen peripheral blood mononuclear cells (PBMCs) (33) also showed an increase in the expression of COX-1 protein upon treatment with IL-4 as with murine BMDMs. Differentiated human macrophages were confirmed by flow cytometry to be ϳ95% CD68 ϩ (supplemental Fig. 2B). PBMCs from subject 1 and subject 2 had a 2.5-and 2.1-fold increase in COX-1 after IL-4 stimulation, respectively (Fig. 1F). This data confirmed a previously reported increase in COX-1 expression after IL-4 stimulation (34).

Cox-1 is post-transcriptionally up-regulated by IL-4 stimulation
Based on the results above, to further examine whether regulation of Cox-1 expression was at the transcriptional or translational level, BMDMs were pre-treated with either actinomycin D or cycloheximide for 30 min or 4 h, respectively, prior to IL-4 stimulation. BMDMs treated with actinomycin D were collected 2 h post-IL-4 stimulation due to impact on cell viability. BMDMs treated with cycloheximide were collected 20 h post-IL-4 stimulation. Arg-1 expression, which is known to be transcriptionally regulated by IL-4 stimulation, was evaluated to determine the efficacy of the actinomycin D treatment. BMDMs treated with actinomycin D had a 4-fold decrease in Arg-1 expression compared with DMSO untreated control ( Fig.  2A). Cox-1 protein expression was not down-regulated after actinomycin D treatment (Fig. 2, B and C). Strikingly, Cox-1 protein was completely absent in both untreated control and IL-4-treated cells compared with DMSO controls after cycloheximide treatment (Fig. 2D). Because Cox-1 appears to be translationally regulated, we reasoned that Ptgs-1 likely has a 5Ј terminal oligopyrimidine (TOP)-like motif, which enables eIF4E to enhance translation. To this end, sequence analysis of PTGS-1 mRNA in both murine and human cells revealed the presence of TOP sequences (Fig. 2E). Interestingly, polysome profiling of murine BMDMs suggested that Ptgs-1 translation is modulated by mTORC signaling, based on a loss of Ptgs-1 mRNA from the heaviest polysome fractions (26% with IL-4 treatment) after torin 1 treatment (19% with IL-4 ϩ torin 1 treatment), but is not mTORC exclusive (Fig. 2F). As a comparison, Rpl-13a, a bona fide TOP mRNA (35), revealed a classical polysome profile for a typical mTORC-dependent mRNA, where all polysome fractions were sensitive to torin 1 treatment (30% with IL-4 treatment to 14% with IL-4 ϩ torin 1 treatment) (Fig. 2F).

The role of Fes/Akt/mTORC signaling pathway in IL-4dependent up-regulation of Cox-1 expression
The above data suggested that IL-4 treatment led to increased levels of Cox-1 protein. Therefore, we examined the underlying mechanisms downstream of the IL-4 receptor. representative Western immunoblot of Cox-2 and Gapdh expression in murine BMDMs stimulated with or without IL-4 or LPS for 20 h (n ϭ 1-2 biological replicates per group). D, 15d-PGJ 2 production 20 h post IL-4 stimulation. E, percentage of CD206 ϩ Arg-1 ϩ F480 ϩ CD11b ϩ murine BMDMs stimulated with or without IL-4 relative to untreated control. F, Western immunoblots of COX-1 and GAPDH from human PBMC-derived macrophages stimulated with or without human IL-4. Unpaired two-tailed t test. **, p Ͻ 0.01; ****, p Ͻ 0.0001. n ϭ 4 -7 biological replicates per group. All experimental data are expressed as mean Ϯ S.E.

IL-4 up-regulates Cox-1 expression
BMDMs were treated with or without Jak inhibitor I 4 h prior to IL-4 stimulation. Inhibitor efficacy was confirmed by the significant decrease in the expression of the Jak-dependent genes Arg-1 and Cd36 (supplemental Fig. S3, A and B). Surprisingly, Cox-1 was up-regulated upon IL-4 treatment regardless of the presence or absence of Jak inhibitor I (supplemental Fig. S3, C and D). However, M2 macrophages were decreased by ϳ25fold compared with untreated control BMDMs (supplemental Fig. S3E).
Because IL-4-mediated expression of Cox-1 was not dependent on the Jak/Stat signaling pathway, we explored if the proto-oncogene protein tyrosine kinase Fes was key in transducing the signal downstream of the IL-4R to effect increased Cox-1 translation. Western immunoblot analysis of BMDMs confirmed the expression of Fes (supplemental Fig. 4A). Incubation of BMDMs with the Fes inhibitor, herbimycin A, followed by treatment with IL-4 significantly decreased expression of the Cox-1 protein in BMDMs (Fig. 3A).
Fes has been shown to signal through the Akt/mTORC pathway (8 -10). To examine activation of the Akt/mTORC pathway, we first examined if PI3K was involved upstream of Akt. BMDMs were pretreated with the PI3K inhibitor, Ly294002, 4 h prior to IL-4 stimulation. Treatment of BMDMs with Ly294002 and IL-4 did not decrease Cox-1 protein (supplemental Fig. S4, B and C). BMDMs were next pre-treated with the Akt inhibitor VIII 4 h prior to IL-4 stimulation. Western immunoblot analysis clearly showed that Akt inhibitor VIII significantly decreased the IL-4-dependent expression of Cox-1 protein (Fig.  3B). Because polysome profiling suggested a loss of Ptgs-1 mRNA from the heaviest polysome fractions following torin 1 treatment of IL-4-stimulated BMDMs, we examined if torin 1 treatment also impacted Cox-1 expression at the protein level. BMDMs were pretreated with torin 1 4 h prior to IL-4 stimulation. Torin 1, completely inhibited Cox-1 protein expression in both untreated control and IL-4-treated BMDMs (Fig. 3C). Treatment with torin 1 also significantly inhibited the phosphorylation of ribosomal protein S6 kinase, which was slightly more phosphorylated in IL-4-treated BMDMs (supplemental Fig. 4D). Torin 1 significantly decreased the production of 15d-PGJ 2 compared with DMSO ϩ IL-4-treated cells (Fig. 3D). Torin 1 treatment also led to a significant decrease in IL-4-dependent skewing of BMDMs toward an M2 phenotype (Fig. 3E). Human PBMC-derived macrophages pre-treated with DMSO followed by IL-4 stimulation had a significant increase in COX-1, which was significantly decreased upon torin 1 pretreatment (Fig. 3F). IL-4 stimulation of Ntc BMDMs led to a 2.8-fold increase in Cox-1 expression as seen throughout this study (Fig. 3G). Knock-out of Tsc2 in Cas9 BMDMs led to a 2.4-fold decrease in Cox-1 expression compared with Ntc Cas9 BMDMs (Fig. 3G). Tsc2 genomic mutation was confirmed by PCR followed by a surveyor assay (data not shown). This further confirmed the role of mTORC signaling in Cox-1 protein expression through use of the CRISPR/Cas9 gene editing system. Fluorescent microscopy confirmed Ͼ70% population purity of the Ntc CRISPR gRNA and Tsc2 CRISPR gRNA (mCherry ϩ ) into Cas9 GFP ϩ bone marrow (supplemental Fig.  S5, A and B).

Activation of AMPK suppresses mTORC signaling and Cox-1 expression
We examined if metformin, a widely prescribed anti-diabetic drug that also activates AMPK, which inhibits mTORC, affected Cox-1 expression in BMDMs. We reasoned that if metformin affected Cox-1 expression, it would translate to adversely affecting the production of 15d-PGJ 2 and consequent polarization of macrophages. BMDMs were treated with or without metformin 4 h prior to IL-4 stimulation (36). BMDMs treated with IL-4 following metformin pre-treatment had a

IL-4 up-regulates Cox-1 expression
decrease in Cox-1 expression (Fig. 4A). LC-MS/MS analysis of the culture media supernatants showed a decrease in 15d-PGJ 2 production upon metformin pre-treatment and IL-4 stimulation (Fig. 4B). In addition, BMDMs treated with metformin and IL-4 had a slight but significantly decreased M2 macrophage phenotype compared with IL-4-stimulated BMDMs that were not treated with metformin (Fig. 4C).

In vivo AMPK activation exacerbates parasitic infection
To examine if inhibition of mTORC through activation of AMPK by metformin treatment of mice impacted their ability to clear N. brasiliensis and if Cox-1-derived 15d-PGJ 2 played any role, mice were administered 300 mg/kg of metformin water with or without 0.05 mg/kg/day 15d-PGJ 2 treatment 1 day prior to inoculation with infective N. brasiliensis thirdstage larvae (L3) and continued on treatment through day 8 of the infection. Control mice either received Milli-Q water with sterile PBS intraperitoneal (i.p.) injections or metformin water with sterile PBS i.p. injections. Immunoblot analysis of the jejunal extracts indicated a significant decrease in Cox-1 expression compared with the control jejunum (Fig. 5A). Interestingly, jejunal extracts of mice revealed no Cox-2 expression in any of the treatment groups suggesting that Cox-1 was the predominant enzyme during a Th2 cyokine (IL-4/IL-13) response . Cox-1 expression is controlled by the Fes/Akt/mTORC signaling pathway. A, representative Western immunoblot for Cox-1 and Gapdh expression in murine BMDMs stimulated with or without IL-4 and DMSO or herbimycin A. B, representative Western immunoblot for Cox-1 and Gapdh expression in murine BMDMs stimulated with or without IL-4 and DMSO or Akt inhibitor VIII. C, representative Western immunoblot for Cox-1 and Gapdh expression in murine BMDMs stimulated with or without IL-4 and DMSO or torin 1. D, production of 15d-PGJ 2 from murine BMDMs stimulated with or without IL-4 (20 h) and DMSO or torin 1. E, percentage CD206 ϩ Arg-1 ϩ F480 ϩ CD11b ϩ murine BMDMs stimulated with or without IL-4 or torin 1 relative to DMSO untreated control. F, Western immunoblots of COX-1 and GAPDH from human PBMC-derived macrophages stimulated with or without human IL-4 and DMSO or torin 1. G, representative Western immunoblot for Cox-1 and Gapdh expression in Cas9 murine BMDMs transduced with either Ntc or Tsc2-specific CRISPR gRNA and stimulated with or without IL-4 (n ϭ 1 per group). Two-way ANOVA with Fisher's LSD test. *, p Ͻ 0.05; **, p Ͻ 0.01; ****, p Ͻ 0.0001. n ϭ 3-6 biological replicates per group. All experimental data are expressed as mean Ϯ S.E.

IL-4 up-regulates Cox-1 expression
to infection with N. brasiliensis (supplemental Fig. S6A) (37). Mice receiving Milli-Q water with sterile PBS i.p. injections had the least parasitic burden as determined by number of adult N. brasiliensis worms in the small intestines (Fig. 5B). However, metformin administration and sterile PBS i.p. injections led to a 3-fold higher adult worm burden compared with mice on Milli-Q water and sterile PBS i.p. injections (Fig. 5B). Furthermore, mice on metformin water that were treated with exogenous 15d-PGJ 2 had a significant decrease in adult worm burden (Fig.  5B). Interestingly, there was a strong negative correlation between the amount of Cox-1 protein expression and number of adult worms present in mice (Fig. 5C). The expression of Arg-1, a M2 macrophage marker, was slightly decreased in the jejunum of mice treated with metformin; however, treatment of mice with metformin and 15d-PGJ 2 restored the jejunal expression of Arg-1 (Fig. 5D).

Discussion
Cyclooxygenase-1 has been generally regarded as a constitutive enzyme regulated developmentally and key in the production of prostanoids, which effect homeostasis in the kidney, stomach, and intestine. Previous studies have shown that various stimuli including TNF␣, IL-1␤, and LPS leads to the production of TxA 2 /TxB 2 in monocytes/macrophages as well as human umbilical vein cells (38,39). In unstimulated human umbilical vein cells, COX-1 is the predominant isoform leading to the production of TxA 2 , however, upon IL-1␤ stimulation, COX-2 is significantly increased with a corresponding increase in PGE 2 and PGI 2 (38). IL-4 treatment has been shown in other studies to decrease pro-inflammatory mediators such as TxB 2 and PGE 2 , which is also decreased in our study compared with LPS treatment. Additionally, we have shown that resting and IL-4-stimulated BMDMs produce baseline levels of TxB 2 and PGE 2 , which appear to not play a significant role in M2 macro-phage polarization. Our study describes a novel mechanism of up-regulation of Cox-1 expression by a Th2 cytokine, IL-4, in macrophages. Interestingly, Cox-2, which is a highly inducible gene regulated by diverse stimuli via mechanisms involving transcriptional and translational control of expression, was significantly down-regulated by IL-4, highlighting the importance of Cox-1 in a Th2 environment, such as during helminth infection.
Furthermore, the Akt-mTORC axis was involved in increased translation of Ptgs-1 mRNA upon IL-4 treatment. The presence of TOP sequences in the vicinity of the TSS in the 5Ј UTR of both murine and human PTGS-1 mRNA further supported the involvement of mTORC. Surprisingly, Ptgs-1 appears to not be exclusively regulated by mTORC in the same manner as traditional TOP mRNA sequences. The use of CRISPR/Cas9 to knock-out Tsc2 led to a decrease in Cox-1 expression. It is known that mTORC1 is negatively regulated by Tsc, whereas mTORC2 is positively regulated by the Tsc complex. Additionally, mTORC2 is not sensitive to rapamycin treatment (data not shown) but is torin 1 sensitive. Finally, because Cox-1 does not behave as a bona fide TOP sequence in regards to the polysome data, it is likely that mTORC2 may affect Cox-1 protein expression. However, the downstream targets of mTORC2 are unknown and further work is needed to delineate the functional differences between mTORC1 and mTORC2 in regulating Cox-1 protein expression. That said, mTORC signaling appears to play a role in the accumulation of Cox-1 protein over the course of IL-4 stimulation suggesting other mechanisms potentially involving protein stability.
There are a few studies that have examined a relationship between Cox-2 and HSP27 during myofibroblast migration (42,43). Additionally, a recent study reported that HSP90 inhibition decreased Cox-2 mRNA expression and protein level in a colorectal cancer cell line (44). In light of the fact that IL-4 treatment of macrophages increases the expression of these chaperones, whereas significantly down-regulating Cox-2 protein and increasing Cox-1 protein, it is plausible that Cox-1 interacts with HSPs, due to high structural similarities between Cox-1 and Cox-2, leading to the stabilization of Cox-1 protein. Further work is warranted to investigate this phenomenon in macrophages and its overall impact on the pathophysiology of IL-4/IL-13-dependent intestinal clearance of N. brasiliensis.
The canonical pathway of signaling downstream of IL-4R␣ typically involves Jak/Stat signaling and the PI3K/Akt/mTORC pathway (7,11). Although treatment of BMDMs with herbimycin A, Akt inhibitor VIII, or torin 1 blocked Cox-1 protein expression; treatment with Ly294002 (PI3K inhibitor), had no effect on Cox-1 expression implicating an alternate pathway of activation of Akt that involves Fes, as in T-cells and early myeloid progenitor cells (Fig. 6) (8 -10). Further work is warranted

IL-4 up-regulates Cox-1 expression
to characterize the under-appreciated interaction between IL-4R␣, Fes, and Akt in the context of mTORC activation. To extend the analysis of mTORC signaling and Cox-1 in macrophages, we showed that the production of 15d-PGJ 2 was significantly increased with IL-4 stimulation, which was inhibited upon loss of Cox-1 expression by torin 1 as well as metformin. Our data suggested that Cox-1 functionally couples with H-PGDS to likely provide the ligand (15d-PGJ 2 ) to PPAR␥, which acts in concert with Stat6 to activate transcription of PPAR␥ target genes to aid in macrophage (M2) polarization (15)(16)(17)45). More importantly, inhibition of mTORC by metformin, which negatively impacted M2 macrophage polarization through decrease in Cox-1 expression, was rescued by exogenous treatment with 15d-PGJ 2 suggesting the importance of Cox-1 activity in macrophage function.
We utilized a helminth model to induce Th2-type inflammation to test the role of Cox-1 and metformin in helminth clearance. Mice treated with metformin had a significantly higher adult parasite burden compared with control mice, which was partially rescued with exogenous administration of 15d-PGJ 2 .
As in the helminth model, Cox-1 could also play a critical role in asthma and airway function. Cox-1 Ϫ/Ϫ mice are hyper-responsive to bronchial constrictors and non-selective Cox inhibitors increased bronchial constriction in wild type mice; whereas rofecoxib (a Cox-2 selective inhibitor) had no effect on airway hyper-responsiveness despite the expression of Cox-2 in epithelial tissue (46,47). This clearly suggested Cox-1, and not Cox-2, activity regulated airway function. However, metformin administration in asthmatics has proven to be beneficial (13) perhaps given its ability to impact a large set of mTORC targets, in addition to COX-1. In light of our studies demonstrating the regulation of Cox-1 by IL-4 in macrophage polarization, it is necessary to revisit the role of Cox-1 in the context of Th2 diseases in an effort to reconcile such apparent differences. Additionally, Langenbach et al. (27) have shown that Cox-1 Ϫ/Ϫ mice exhibit slightly reduced resolution and macrophage recruitment in addition to delayed parturition, which all have varying extents of involvement of M2 macrophages. Thus, we can only speculate that there could be an impact on M2 polarization in Cox-1 Ϫ/Ϫ mice, which will need to be further tested.
In conclusion, our studies provide evidence for a novel mechanism involving post-transcriptional regulation of Cox-1

IL-4 up-regulates Cox-1 expression
expression upon IL-4 stimulation in human and murine macrophages (Fig. 6). This is relevant because Cox activity is required to generate bioactive prostanoids, particularly PGJ 2 metabolites, that serve as endogenous ligands for PPAR␥ and drive macrophage polarization (15)(16)(17)48). As shown here, only Cox-1, and not Cox-2, was expressed both in vitro and in vivo during a Th2-mediated response emphasizing the need to better understand the role of Cox-1 to enable development of new and more specific therapeutic regimens in addition to understanding how current drugs, such as metformin, may have unintentional effects on Cox regulation. Specifically, because inhibition of Cox-1 with indomethacin (17) or metformin was rescued by exogenous treatment with 15d-PGJ 2 to significantly alleviate nematode burden, it is clear that further studies are necessary to unravel the role of Cox-1 in promoting resolution in Th2-mediated inflammation.

Materials and methods
Mice C57BL/6 male age-matched mice were purchased from Taconic Biosciences, Inc. (Hudson, NY) and maintained on an AIN-76-based semi-purified diet from Harlan-Teklad (Madison, WI). All studies were pre-approved by the Institutional Animal Care and Use Committee and the Institutional Biosafety Committee at Penn State University.

Treatments
Various compounds were used at the following concentrations: 7.5 M herbimycin A (Santa Cruz), 10 M Ly294002 (Cayman), 10 M Akt inhibitor VIII (Cayman), 250 nM torin 1 (Selleck Chemicals), 80 nM actinomycin D (Sigma), 10 M cycloheximide (gift from Dr. Gary Perdew, Penn State University), 50 nM Jak inhibitor I (Santa Cruz), 20 M metformin (Sigma), and 30 nM ozagrel (Cayman). Metformin was reconstituted in cell culture grade water, whereas all other compounds were reconstituted in cell culture grade DMSO (Sigma) and added to cells at concentrations not exceeding 0.1% (v/v). The above mentioned compounds were added to BMDM cultures 4 h prior to IL-4 stimulation, except for actinomycin D, which was added 30 min prior to IL-4 stimulation due to impact on cell viability. IL-4 (10 ng/ml; R&D Systems) was added for 20 h, unless described otherwise in text, and cells were collected for analysis as described below. Cell viability was confirmed by trypan blue staining.

RNA extraction and real-time PCR
Cell pellets were stored in 1 ml of Tri-reagent (Thermo) at Ϫ80°C until RNA extraction. RNA extraction was performed according to the manufacturer's protocol as described previously (21). RNA was reverse transcribed to cDNA using a high

Flow cytometry
Cells were enumerated by trypan blue staining using a hemocytometer. 100,000 -200,000 cells were resuspended in 100 l of flow buffer (phosphate-buffered saline (Life Technologies) containing 100 IU/ml of penicillin (Corning), 100 g/ml of streptomycin (Corning), and 2% (v/v) fetal bovine serum (Atlanta Biologicals)). Extracellular staining was performed by incubating cells with 0.25 g of F c block (BD Biosciences) for 10 min followed 30 min at 4°C with the following antibodies: 0.25 g of FITC-conjugated CD206 (Biolegend), 0.10 g of PE-Cy7conjugated CD11b (BD Biosciences), and 0.15 g of APC-conjugated F4/80 (Miltenyi). Cells were then fixed by adding 2% formaldehyde solution (Sigma), and incubated for 20 min at room temperature. To perform intracellular staining, cells were resuspended in a permeability buffer (48.5 ml of phosphatebuffered saline containing 1.5 ml of fetal bovine serum (Atlanta Biologicals), 50 mg of saponin (Sigma), and 45 mg of sodium azide (J.T. Baker)) and incubated at room temperature for 15 min. Cells were resuspended in 100 l of permeability buffer and 0.125 g of PE-conjugated arginase-1 (R&D Systems) antibody or 2.5 l of Alexa Fluor 647-conjugated CD68 (BD Biosciences) and incubated for 30 min at room temperature. Cells were collected on the C6 Accuri Flow Cytometer (BD Biosciences) in addition to the following controls: unstained, OneComp eBeads (eBioscience) single stained, and fluores-cence minus one (FMO). Data were analyzed using FlowJo. Gating was based on fluorescence minus one staining controls.

Lipid extraction and liquid chromatography-mass spectrometry
Endogenous lipids were extracted from cell culture media samples using C-18 Sep-Pak cartridges (Waters, Milford, MA). Briefly, cell-free culture supernatants were acidified with 6 N HCl and loaded on to the Sep-Pak cartridge, bound lipids were eluted with methanol, evaporated, and stored in ethyl acetate, until analyzed at Ϫ80°C. Lipids were suspended in 70% methanol and analyzed by LC-MS/MS on an API 2000 LC-MS/MS system with Turbo V source and an electrospray ionization probe in negative ion mode (15). Quantitative analysis of 15d-PGJ 2 was performed using a calibration curve generated by respective standards in multiple reaction monitoring mode using three transitions for 15d-PGJ 2

Polysome profile assay
Polysome profiling and mRNA distribution analysis was performed as described previously (50). Briefly, BMDMs were seeded in 150-mm Petri plates and grown up to 70% confluence (ϳ1.5 ϫ 10 7 cells). Following DMSO or IL-4 treatment for 20 h, cycloheximide (CHX) was added to cells in 100 g/ml for 10 min at 37°C. Cells were collected at 1,500 ϫ g for 10 min and washed twice with cold PBS containing 100 g/ml of CHX. The cell pellets were suspended in 300 l of lysis buffer (10 mM HEPES-KOH (pH 7.5), 2.5 mM MgCl 2 , 100 mM KCl, 0.25% Nonidet P-40, 100 g/ml of CHX, 1 mM DTT, 200 units/ml of RNase inhibitor (RNaseOUT, Invitrogen) and EDTA-free protease inhibitor (Roche)), kept on ice for 20 min, and then passed 15 times through a 23-gauge needle. Lysates were cleared by centrifugation at 16,000 ϫ g for 20 min and supernatants (cytosolic cell extracts) were collected and absorbance was measured at 260 nm. Approximately 7 absorbance units (at 260 nm) of lysates were layered over 15-54% cold sucrose gradients in buffer (10 mM HEPES-KOH (pH 7.5), 2.5 mM MgCl 2 , and 100 mM KCl). Gradients were centrifuged at 121,000 ϫ g in a Beckman SW28 rotor for 4 h at 4°C. After centrifugation, 12 equal size fractions (1.2 ml/fraction) were collected. RNA from each fraction was isolated using TRIzol LS reagent (Invitrogen) and an equal volume of RNA from each fraction was used for cDNA synthesis with SuperScript III First-Strand Synthesis Supermix (Invitrogen). The relative quantities of specific mRNAs were measured by quantitative PCR using the VeriQuest SYBR Green qPCR master Mix (Affymetrix) on a StepOnePlus Real-Time PCR system (Applied Biosystem).

Infection with N. brasiliensis
Mice were pretreated with sterile PBS or 0.05 mg/kg/day of 15d-PGJ 2 (formulated in sterile PBS) intraperitoneally and

IL-4 up-regulates Cox-1 expression
administered metformin (300 mg/kg in Milli-Q water) or Milli-Q water alone starting 1 day prior to subcutaneous inoculation with 500 third stage N. brasiliensis larvae (L3) (17,51). Metformin water was made fresh and changed every other day throughout infection. Mice received i.p. injections daily. Treatment was continued throughout the infection until sacrifice on day 8 post-inoculation when adult worms in the jejunum were enumerated (52).

Human macrophage culture
Frozen human PBMC were purchased from AllCells (Alameda, CA) from donors. Upon arrival, cells were thawed in a 37°C water and enumerated with trypan blue staining using a hemocytometer to determine cell viability. Cells were transferred to a 50-ml tube and 20 ml of RPMI 1640 (Corning) containing 2 mM L-glutamine (Corning), 10% (v/v) fetal bovine serum (Atlanta Biologicals), 100 IU of penicillin (Corning), and 100 g/ml of streptomycin (Corning) was added dropwise. Cells were centrifuged at 200 ϫ g for 15 min at room temperature. This was repeated and PBMCs were finally resuspended in 2 ml of media and enumerated by trypan blue using a hemocytometer. PBMCs were brought to a concentration of 3.0 ϫ 10 6 cells/ml and plated in a 6-well culture plate. Cells were cultured with 10 ng/ml of recombinant human macrophage colonystimulating factor (GoldBio) for 24 h, then at 25 ng/ml for remaining culturing period. Media was changed every 3 days for 8 days adding fresh macrophage colony-stimulating factor with each change (33). Adherent cells (comprising of 95% macrophages) were pre-treated with either DMSO or 250 nM torin 1 followed by 20 h with or without recombinant human IL-4 (R&D Systems) stimulation. Cells were harvested and analyzed by flow cytometry and Western immunoblot analysis as described above.

Transfection and transduction
HEK293TN cells (System Biosciences; 350,000 cells/2 ml) were plated in DMEM (Life Technologies) with 1 mM sodium pyruvate (Life Technologies), 2 mM L-glutamine (Corning), 100 IU/ml of penicillin (Corning), 100 g/ml of streptomycin (Corning), and 10% (v/v) fetal bovine serum (Atlanta Biologicals) in a 6-well plate and allowed to adhere overnight. Cells were then transfected three times, once every 24 h with 2.45 g of pMD2.G (Addgene), 4.9 g of psPAX-2 (Addgene), 4.9 g of lentiviral CRISPR gRNA plasmid (Amp r /Pur r ) expressing mCherry (Vector Builder), 26.6 l of TransIT 293 transfection reagent (Mirus), and 700 l of serum-free DMEM (Life Technologies). The gRNA sequences for non-targeting control and Tsc2 were GCACTACCAGAGCTAACTCA and CAGGAG-GACCTGCGCGCGAA, respectively. Viruses were harvested and passed through a 0.45-m filter. Bone marrow was harvested from Cas9 GFP ϩ mice and incubated with 1 part viral supernatant to 1 part BMDM. After 48 h of viral transduction, bone marrow cells were centrifuged at 100 ϫ g for 5 min at 4°C and resuspended in DMEM containing 2 g/ml of puromycin (GoldBio) and continued for 4 days with puromycin, changing media every 2 days. Cells were imaged by fluorescent microscopy to detect GFP ϩ and mCherry ϩ . After 4 days of puromycin selection, cells were cultured in media without puromycin for an additional 2 days. BMDMs were stimulated with or without 10 ng/ml of recombinant murine IL-4 (R&D Systems) for 20 h and analyzed for Tsc2 knock-out and Cox-1 expression by surveyor assay and Western immunoblot analysis, respectively.

Bioinformatics and statistics
The TSS of murine and human PTGS-1 were analyzed with Promoter 2.0 (cbs.dtu.dk) and DBTSS (dbtss.hgc.jp) that predict TSS of vertebrate PolII promoters in DNA sequences (35,53). The most common TSS in PTGS-1 that was predicted by Promoter 2.0 corroborated with consensus sequences defined by earlier studies (GenBank TM AF440204.1 and AH015269.2). All experimental data are expressed as mean Ϯ S.E. Outliers were removed using ROUT method (Q ϭ 5.0%), where appropriate (Fig. 5). Either an unpaired one-tailed t test, an unpaired two-tailed t test, a one-way ANOVA with Fisher's LSD test, or a two-way ANOVA with Fisher's LSD test was used, where appropriate, to compare the mean of each treatment group with the mean of each other treatment group using GraphPad Prism (GraphPad Software). A linear regression analysis was also performed, where appropriate. p values Յ 0.05 were considered statistically significant.
Author contributions-A. E. S. and K. S. P. conceived and coordinated the study as well as wrote the manuscript and created the schematic. A. E. S. designed, performed, and analyzed the data shown in Figs. 1-6 and supplemental Figs. S1-S6. B. T. D. performed and analyzed the data shown in Figs. 1C, 3D, 4B, supplemental Figs. S1, D, E, and G, and 4D and assisted with manuscript preparation. B. J. G. performed and analyzed the data shown in Fig. 2F and assisted with manuscript preparation. V. N. cultured, stimulated, and harvested the BMDMs required for the polysome assay in Fig. 2F and assisted with manuscript preparation. J. F. U. provided L3 stage N. brasiliensis larvae and assisted with manuscript preparation. All authors reviewed the results and approved the final version of the manuscript.