Protectin DX alleviates insulin resistance by activating a myokine-liver glucoregulatory axis

Transcription

1 Protectin DX alleviates insulin resistance by activating a myokine-liver glucoregulatory axis Phillip J White,, Philippe St-Pierre,, Alexandre Charbonneau,, Patricia L Mitchell,, Emmanuelle St-Amand,, Bruno Marcotte, & André Marette, We previously demonstrated that low biosynthesis of v fatty acid derived proresolution mediators, termed protectins, is associated with an impaired global resolution capacity, inflammation and insulin resistance in obese high-fat diet fed mice. These findings prompted a more direct study of the therapeutic potential of protectins for the treatment of metabolic disorders. Herein we show that protectin DX (PDX) exerts an unanticipated glucoregulatory activity that is distinct from its anti-inflammatory actions. We found that PDX selectively stimulated the release of the prototypic myokine interleukin- (IL-) from skeletal muscle and thereby initiated a myokine-liver signaling axis, which blunted hepatic glucose production via signal transducer and activator of transcription (STAT)-mediated transcriptional suppression of the gluconeogenic program. These effects of PDX were abrogated in Il-null mice. PDX also activated AMP-activated protein kinase (AMPK); however, it did so in an IL- independent manner. Notably, we demonstrated that administration of PDX to obese diabetic db/db mice raises skeletal muscle IL- levels and substantially improves their insulin sensitivity without any impact on adipose tissue inflammation. Our findings thus support the development of PDX-based selective muscle IL- secretagogues as a new class of therapy for the treatment of insulin resistance and type diabetes. PDX ((S,7S)-dihydroxydocosa-(Z,7Z,E,Z,E,9Z)- hexaenoic acid), is produced via sequential lipoxygenation of docosahexaenoic acid (DHA; : n ), and can be found together with the first identified protectin, protectin D (PD; (R,7S)- dihydroxydocosa-(z,7z,e,e,z,9z)-hexaenoic acid), in mouse inflammatory exudates. PDX may also be produced by human neutrophils exposed to DHA, albeit to a lesser extent than PD (ref. ). We evaluated the therapeutic potential of PDX for insulin resistance in the setting of lipid excess by employing a -h heparinized intralipid infusion paired to a hyperinsulinemic-euglycemic clamp in lean C7BL/ mice (Supplementary Fig. ). We administered equal volumes of ethanol vehicle or PDX ( µg intravenously (i.v.)) to lipidinfused mice immediately before and. h into the -h lipid infusion. We also clamped a saline-infused group that had been treated with vehicle to ascertain baseline insulin sensitivity. Preclamp glycemia was lower in PDX-treated mice compared to both saline-infused and lipidinfused animals treated with vehicle (Fig. a); however, preclamp insulinemia was not different among groups at this time point (saline,. ±.7 ng ml ; lipid,.7 ±.9 ng ml ; lipid + PDX,. ±. ng ml ; data are mean ± s.e.m., n = ). Thus, PDX appeared to possess an unanticipated insulin-independent glucoregulatory activity. PDX-treated mice were also protected from lipid-induced insulin resistance, as demonstrated by a higher glucose infusion rate (GIR) required to maintain euglycemia during the clamp in these mice compared to vehicle-treated, lipid-infused mice (Fig. a,b). The better insulin response in PDX-treated, lipid-infused animals involved partial protection of peripheral insulin action and markedly elevated hepatic insulin action (Fig. c). Phosphorylation of protein kinase B (Akt) on Ser7 confirmed the stronger insulin response in skeletal muscle and liver of PDX-treated mice compared to vehicle-treated, lipid-infused counterparts (Fig. d,e). In line with the classical role of protectins in the active resolution of inflammation, we found that PDX also prevented lipid-mediated induction of inducible nitric oxide synthase (inos) in both muscle and liver (Fig. d,e) and c-jun N-terminal kinase (JNK) activation in liver, as determined by phosphorylation on Thr and Tyr (Thr/Tyr) (Fig. e). PDX also strongly suppressed lipidinduced secretion of chemokine cc motif ligand (CCL) and CCL, as well as tumor necrosis factor-α (TNF-α), interferon-γ (IFN-γ), interleukin-β (IL-β), IL- and IL-7 (Fig. f). However, in contrast to all other chemokines and cytokines tested, IL- concentrations were approximately sevenfold higher in PDX-treated mice compared to in mice that received lipid infusion alone (Fig. f). Although IL- has been reported to underlie the insulin-sensitizing actions of adiponectin in liver, we found that adiponectin does not account for these higher concentrations of IL- in PDX-treated mice, as both vehicle- and PDX-treated lipid-infused mice displayed similarly low concentrations of adiponectin in plasma compared to saline-infused animals (Supplementary Fig. a). As macrophages are important contributors to global chemokine and cytokine production, we next examined whether PDX had the same influence on the lipid-induced inflammatory response exhibited by macrophages treated with palmitate in vitro. PDX effectively Department of Medicine, Québec Heart and Lung Institute, Laval University, Québec, Québec, Canada. Institute of Nutrition and Functional Foods, Laval University, Québec, Québec, Canada. Correspondence should be addressed to A.M. (andre.marette@criucpq.ulaval.ca). Received 9 January; accepted April; published online May ; doi:./nm.9 nature medicine advance online publication

2 a Saline Lipid Lipid + PDX b Preclamp e Saline Lipid pakt Total Akt inos pjnk Total JNK Lipid + PDX GIR (mg kg min ) Liver protein (AU) Saline Lipid Lipid + PDX c GIR (mg kg min ) Saline Lipid Lipid + PDX... pakt/akt ND inos pjnk/jnk f Plasma chemokines and cytokines (pg ml ) Fold increase in Rd Saline Lipid Lipid + PDX d Peripheral insulin action % suppression of HGP ND CCL TNF-α IL-7 Hepatic insulin action ND Plasma chemokines and cytokines (pg ml ) pakt Total Akt inos ND Saline Lipid Lipid + PDX. Lipid + Saline Lipid PDX. Saline Lipid Lipid + PDX Muscle protein (AU).. pakt/akt inos ND ND ND ND CCL IFN-γ IL-β IL- IL- Figure PDX prevents lipid-induced insulin resistance. (a) Preclamp glycemia (left) and clamp glucose excursion (right) in vehicle-treated, saline-infused (Saline), vehicle-treated, lipid-infused (Lipid) and PDX-treated, lipid-infused (Lipid + PDX) mice. (b) GIR (mg per kg body weight per minute) (left) and mean GIR for the final min of clamp (right) in Saline, Lipid and Lipid + PDX mice. (c) Peripheral insulin action expressed as fold increase in rate of disappearance (Rd) during the clamp (left) and hepatic insulin action expressed as percentage suppression of hepatic glucose production (HGP, right) in Saline, Lipid and Lipid + PDX mice. (d) Immunoblots for phosphorylated Akt (pakt) Ser7, total Akt and inos in gastrocnemius muscle of clamped Saline, Lipid and Lipid + PDX mice. Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses is shown to the right of the representative gels. (e) Immunoblots for pakt Ser7, total Akt, inos, pjnk Thr/Tyr and total JNK in liver of clamped Saline, Lipid and Lipid + PDX mice. Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses are shown to the right of the representative gels. AU, arbitrary units; ND, not detected. (f) Chemokines and cytokines in plasma from clamped Saline, Lipid and Lipid + PDX mice. Data are mean ± s.e.m., representative of mice per group. P <., P <., P <. calculated using analysis of variance (ANOVA). suppressed lipid-induced secretion of CCL, CCL, TNF-α, IL- and IL-, as well as inos and JNK activation (Supplementary Fig. b d). However, rather than stimulating IL- release, PDX repressed lipid-induced IL- production in macrophages (Supplementary Fig. b). These data suggest that the higher concentration of circulating IL- in PDX-treated mice was probably derived from an alternate cellular source. Given that IL- is well regarded as the prototypic myokine (muscle-derived cytokine),, we hypothesized that skeletal muscle could be the site of higher IL- release in PDX-treated animals. Notably, we found that the effect of PDX on IL- in muscle closely resembled that in plasma (Fig. a), but the livers of PDX-treated mice did not contain higher IL- concentrations than those from vehicletreated mice (Supplementary Fig. a). To confirm that PDX induces IL- expression and release from muscle in a cell-autonomous fashion, we next treated cultured CC myotubes with PDX. PDX stimulated a dose-dependent rise in IL- mrna expression and protein accumulation in medium within h of administration, with the greatest dose prompting more than twofold higher IL- secretion and mrna expression than vehicle (Fig. b). We also detected greater phosphorylation of AMPK at this time point in PDX-treated myotubes (Fig. c). Brown adipose tissue has recently been shown to regulate glucose homeostasis via secretion of IL- (ref. 7). However, although we observed IL- release from cultured T7i brown adipocytes in response to adrenergic stimuli ( µm norepinephrine), PDX did not stimulate IL- release from this cell type (Supplementary Fig. b). We also found that PDX neither stimulated nor repressed IL- release or mrna expression in macrophages that had not been treated with palmitate (Supplementary Fig. c,d). Despite its lack of effect on IL- release, PDX maintained its ability to promote AMPK phosphorylation (Supplementary Fig. e). Activation of AMPK represses inflammatory cytokine production in activated macrophages and thus may underlie the PDX-mediated inhibition of IL- release in our earlier macrophage studies with palmitate. To determine the structural specificity of PDX-mediated muscle IL- release, we next assessed the potential of three structurally related lipid mediators, PD, (S,S)-dihydroxyicosa-(Z,9E,Z,E)- tetraenoic acid ((S,S)-diHETE) and (7S,R,7S)-trihydroxydocosa- (Z,9E,E,Z,E,9Z)-hexaenoic acid (resolvin D), to stimulate IL- production from CC myotubes. We chose these lipids because they possess a broad range of common structural features with PDX (Supplementary Fig. f). However, only administration of PDX prompted IL- release from muscle cells (Supplementary Fig. f). These comparative structure-function analyses thus revealed that the ability of PDX to trigger IL- release was not solely dependent on the DHA backbone, or on the number, position and stereochemistry of the hydroxyl groups alone, and cannot simply be linked to the E,Z,E organization of the conjugated triene within the PDX molecule. Thus, skeletal muscle IL- release appears to be a unique consequence of stimulation by the PDX molecule rather than a shared feature of a new lipid class. Together, these data identify PDX as a structurally distinct selective skeletal muscle IL- secretagogue. IL- is thought to regulate hepatic glucose production in liver via STAT-mediated transcriptional suppression of the gluconeogenic genes peroxisome proliferator-activated receptor γ coactivator-α (Ppargca), phosphoenolpyruvate carboxykinase (Pck) and glucose- -phosphatase (Gpc) 9,. We observed that PDX promotes greater STAT phosphorylation in liver compared to vehicle-treated saline- or lipid-infused mice (Fig. d). The suppression of Ppargca and Pck in liver of PDX-treated mice was also greater than that observed in vehicletreated mice (Fig. e), and there was also improved suppression of advance online publication nature medicine

3 a IL- (pg g tissue) Saline Lipid + PDX Muscle Lipid b Fold vehicle Veh PDX nm IL- mrna IL- in medium PDX nm PDX µm c pampk AMPK CC protein (AU) PDX PDX PDX Veh nm nm µm pampk/ampk d pstat STAT Liver protein (AU) Saline Lipid + Lipid PDX pstat/stat Figure PDX stimulates skeletal muscle IL- expression. (a) Skeletal muscle IL- protein expression in gastrocnemius from clamped Saline, Lipid and Lipid + PDX mice. Data are mean ± s.e.m., representative of mice per group. Fold increase in IL- mrna expression over vehicle normalized to Gapdh and fold increase in IL- protein secretion in medium over vehicle (b) and immunoblots for pampk Thr7 and total AMPK (c) in CC myotubes treated with PDX for h compared to ethanol vehicle (Veh) control. Quantification of densitometry analyses for immunoblots is shown below the representative gels. Data are mean ± s.e.m. for three independent experiments. (d) Immunoblots for pstat Ser77 and total STAT in liver from clamped Saline, Lipid and Lipid + PDX mice (n = mice per group). Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses for immunoblots is shown below the representative gels. (e) Relative mrna expression of Ppargca, Pck and Gpc normalized to Gapdh in liver from clamped Saline, Lipid and Lipid + PDX mice (n = mice per group). (f) Immunoblots for pampk Thr7 and total AMPK in gastrocnemius muscle from clamped Saline (n = ), Lipid (n = ) and Lipid + PDX (n = ) mice. Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses for immunoblots is shown below the representative gels. Data are mean ± s.e.m. P <., P <. calculated using ANOVA. f Muscle protein (AU) e Relative expression pampk AMPK..... Saline Lipid + PDX Ppargca Pck Lipid Lipid + Saline Lipid PDX pampk/ampk Gpc Gpc in PDX-treated animals compared to vehicle-treated, salineinfused mice (Fig. e). In line with a reported role for AMPK in IL- mediated enhancement of skeletal muscle glucose metabolism, PDX-treated mice also displayed more phosphorylation of AMPK in muscle compared to their vehicle-treated counterparts (Fig. f). To confirm the involvement of IL- in the beneficial effects of PDX, we next performed a second round of paired lipid-infusion hyperinsulinemic-euglycemic clamp studies in Il-knockout mice alongside genetically matched wild-type (WT) nonlittermate counterparts. We added saline-infused mice treated with PDX to the study to ascertain whether PDX also improves glucose metabolism in insulin-sensitive animals (Supplementary Fig. ). As observed in the previous study, lipid-infused PDX-treated WT mice displayed lower preclamp glycemia compared to vehicle-treated counterparts (Fig. a) in the absence of any change in circulating insulin (data not shown). However, this effect of PDX was completely absent in Il-knockout mice (Fig. a). Lipid-infused Il-knockout mice were also refractory to the PDX-mediated enhancement of GIR (Fig. b), and although PDX improved both peripheral and hepatic insulin action in lipid-infused WT mice (Fig. c), these effects were substantially blunted in Il-knockout mice. As expected, PDX also failed to raise IL- concentrations in skeletal muscle of lipid-infused Il-knockout mice (Fig. d). Accordingly, the stimulatory effect of PDX on hepatic STAT phosphorylation (Fig. e) and its suppressive effect on hepatic Ppargca, Pck and Gpc (Fig. f) were completely absent in lipid-infused Il-knockout mice. Il-knockout mice also had higher Pck expression compared to vehicle-treated WT mice (Fig. f); however, this was not the case for Gpc, whose expression was significantly lower in Il-knockout mice compared to in vehicle-treated WT counterparts (Fig. f). This discrepant effect of IL- deficiency on Pck and Gpc expression may explain the lack of further deterioration of hepatic insulin action in lipid-infused Il-knockout mice. We observed similar effects of PDX on preclamp glycemia, GIR and skeletal muscle IL- concentration in insulin-sensitive salineinfused WT mice (Supplementary Fig. a,b,d). However, in contrast to the case in lipid-infused mice, the improved insulin sensitivity witnessed in the WT PDX-treated, saline-infused mice was entirely the result of superior hepatic insulin action (Supplementary Fig. c). Of note, the IL- dependent activation of STAT by PDX was not associated with further suppression of Ppargca in insulin-sensitive saline-infused mice (Supplementary Fig. e,f). However, PDX administration did potentiate the transcriptional repression of Pck downstream of Ppargca (Supplementary Fig. f). These effects of PDX on STAT and Pck were completely absent in saline-infused Il- knockout mice. PDX administration improved, though not in a statistically significant manner, the suppression of Gpc in saline-infused WT mice (P =.; Supplementary Fig. f), but this was not the case for Il-knockout counterparts. The expression of Pck and Gpc in PDX-treated, saline-infused Il-knockout mice was higher than in vehicle-treated WT mice (Supplementary Fig. f), which is consistent with the reduced hepatic insulin action in these mice (Supplementary Fig. c). It is noteworthy that PDX improved the inhibition of hepatic glucose output and suppression of gluconeogenic enzymes in these insulin-sensitive saline-infused mice without potentiating Akt phosphorylation (Supplementary Fig. a), which suggests that the hepatic STAT axis is the major pathway responsible for the glucoregulatory effects of PDX. We also found that PDX-treated Il-knockout mice exhibit comparable AMPK phosphorylation to PDX-treated WT mice (Fig. g and Supplementary Fig. g). In light of these data, we reexamined the early timeline of PDX-mediated IL- release and AMPK activation in CC myotubes. PDX promoted AMPK phosphorylation within min in these cells, but we did not detect IL- release into the medium at this time point (Supplementary Fig. b,c). We also found that PDX maintained its ability to potently suppress lipid-induced elevations in nature medicine advance online publication

4 a WT-LV WT-LP KO-LV KO-LP b WT-LV WT-LP KO-LV KO-LP c WT-LV WT-LP KO-LV KO-LP d Preclamp GIR (mg kg min ) GIR (mg kg min ) Fold increase in Rd..... Peripheral insulin action % suppression of HGP Hepatic insulin action IL- (pg g tissue) 7 WT-LV WT-LP KO-LV KO-LP Muscle Figure IL- is required for the insulin-sensitizing e WT WT KO KO f g WT WT KO KO actions of PDX. (a) Preclamp glycemia (left) and LV LP LV LP WT-LV WT-LP KO-LV KO-LP LV LP LV LP clamp glucose excursion (right) in WT or Il-knockout pstat pampk (KO) lipid-infused (L) mice treated with PDX (P) or vehicle (V). (b) GIR (left) and mean GIR for the STAT AMPK final min of clamp (right) in WT or Il-knockout lipid-infused mice treated with PDX or vehicle. (c) Peripheral insulin action expressed as fold increase in Rd during the clamp (left) and hepatic insulin action expressed as percentage suppression of hepatic glucose production (right) in WT or Il-knockout lipid-infused mice treated with PDX or vehicle. (d f) Skeletal muscle IL- content (d), pstat/stat pampk/ampk immunoblots for pstat Ser77 and total STAT (e) and relative mrna expression of Ppargca, Pck and Gpc normalized to Gapdh (f) in liver from clamped WT or Il-knockout lipid-infused mice treated with PDX or vehicle. Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses for immunoblots is shown below the representative gels. (g) Immunoblots for pampk Thr7 and total AMPK in gastrocnemius muscle from clamped WT or Il-knockout lipid-infused mice treated with PDX or vehicle. Dashed lines indicate that lanes were run on the same gel but were noncontiguous. Quantification of densitometry analyses for immunoblots is shown below the representative gels. WT-LV, n = mice; WT-LP, n = mice; KO-LV, n = mice; KO-LP, n = mice. Data are mean ± s.e.m. P <., P <., P <. calculated using ANOVA. Liver protein (AU) Relative expression Ppargca Pck Gpc Muscle protein (AU) circulating TNF-α in Il-knockout mice (Supplementary Fig. d). Thus, our data suggest that the major glucoregulatory mechanisms of PDX are distinct from its anti-inflammatory actions and its ability to activate AMPK. We further explored the therapeutic efficacy of PDX in genetically obese db/db mice, a well-established model of type diabetes. In these experiments, we treated db/db mice with vehicle or PDX ( µg i.v.) and 9 min before the initiation of a hyperinsulinemic-isoglycemic clamp. As observed previously, PDX tended to lower preclamp glycemia (Fig. a), but this effect did not reach statistical significance in these severely diabetic mice. Nonetheless, PDX treatment substantially improved insulin sensitivity, as revealed by a.-fold higher GIR in PDX-treated mice compared to vehicle-treated animals (Fig. b). It is noteworthy that in these obese diabetic mice, the impact of PDX on GIR began to dissipate around min into the clamp. This aligns well with a lack of enhanced STAT signal in liver at the completion of the clamp (Supplementary Fig. e) and a recent report showing that obese mice express threefold higher levels of -hydroxyprostaglandin dehydrogenase, which is responsible for the enzymatic conversion of docosanoid resolution mediators to less active oxo-derivatives. Nevertheless, we still observed greater suppression of Ppargca, Pck and Gpc in liver of PDX-treated db/db mice compared to vehicletreated counterparts (Fig. c). Of note, these effects occurred in the absence of any significant anti-inflammatory impact of PDX on adipose tissue chemokines or cytokines (Fig. d). However, as observed in earlier studies, PDX-treated mice exhibited higher IL- concentrations in skeletal muscle and plasma (Fig. e). In order to determine whether the beneficial effects of PDX could be sustained or improved with a prolonged treatment regimen, we next administered µg of PDX or vehicle to 7-week-old db/db mice twice daily for d leading up to the clamp experiment. As in the acute study, PDX administration tended to lower glycemia without affecting circulating insulin (Supplementary Fig. a,b); however, this change in glycemia did not reach significance in these severely obese diabetic mice. Nonetheless, the improved insulin sensitivity observed in db/db mice treated acutely with PDX was conserved in those mice treated with PDX for d (Supplementary Fig. c) and was once again associated with higher concentrations of skeletal muscle IL- and a trend toward higher plasma IL- levels (Supplementary Fig. d). Thus, the effectiveness of PDX does not appear to dissipate with extended treatment. As the lengthening of the PDX treatment regimen to d was not sufficient to resolve inflammatory cytokine production in adipose tissue or plasma of these very obese diabetic mice (Supplementary Fig. e,f), it appears that the glucoregulatory axis of PDX action is entirely responsible for the improved glucose utilization observed in these obese diabetic mice. To the best of our knowledge, this is the first report of an agent other than contraction or exercise that selectively promotes skeletal muscle IL- expression and release. The notable insulin-sensitizing effect of PDX observed in our animal models of insulin resistance and type diabetes suggests that skeletal muscle IL- secretagogues could become a new class of agents for diabetes therapy. Further study of the mechanisms by which PDX promotes muscle IL- release is thus warranted. This is also, to our knowledge, the first report where Il-knockout mice have been studied using the hyperinsulinemic-euglycemic clamp in conditions of lipid excess. Although lipid infusion and palmitate treatment do increase systemic and macrophage IL- production, advance online publication nature medicine

5 a Preclamp Veh PDX b Veh PDX GIR (mg kg min ) GIR (mg kg min ) c Relative expression Ppargca Pck Figure PDX therapy improves insulin sensitivity in diabetic mice. (a) Preclamp glycemia (left) and clamp glucose excursion (right) in genetically obese diabetic db/db mice treated with vehicle (Veh) or treated acutely with PDX (PDX). (b) GIR (left) and mean GIR for the final min of clamp (right) in vehicle- and PDX-treated db/db mice. (c) Relative mrna expression of Ppargca, Pck and Gpc normalized to Gapdh in liver from clamped vehicleand PDX-treated db/db mice. (d) Chemokines and cytokines in epididymal white adipose tissue (ewat) clamped vehicle- and PDX-treated db/db mice. (e) Skeletal muscle (left) and plasma (right) IL- content from of clamped vehicle- and PDX-treated db/db mice. Data are mean ± s.e.m. Veh, n = ; PDX n =. P <., P <. calculated using Student s t-test. respectively, our findings do not support a role for IL- in the development of lipid-induced insulin resistance,, as we found that lack of IL- does not prevent insulin resistance in lipid-infused mice. To the contrary, we found that insulin-sensitive saline-infused Il-knockout mice display a slight defect in hepatic insulin action that is associated with altered regulation of hepatic Pck and Gpc. Our data thus join a growing body of work,,7 9 that argues for a positive role of IL- in the regulation of glucose metabolism. We found that in addition to potentiating insulin action, PDX administration also induced a characteristic lowering of basal glycemia that was IL- dependent and preceded insulin administration in both saline- and lipid-infused mice. In additional studies performed in nonclamped mice, we observed that this PDX-dependent glucoselowering effect lasts for approximately. h in lean mice (data not shown). This suggests that PDX and IL- might also represent promising therapeutic targets as insulin-independent glucose-lowering agents. This finding is in agreement with work demonstrating that exposure of mouse soleus to IL- and soluble IL- receptor increases glucose transport ex vivo and with a study showing that the hypoglycemic response to endotoxemia is absent in Il-knockout mice. Notably, our data suggest that the glucose-lowering effect of PDX is dependent on IL- mediated activation of hepatic STAT, which suppresses the expression of Ppargca, Pck and Gpc. In addition to showing the insulin-sensitizing and glucoregulatory actions of PDX, this is also, to our knowledge, the first report demonstrating that PDX can suppress lipid-induced inflammation. In light of a recent report 9 that IL- promotes alternate activation of macrophages in inflammatory conditions such as obesity and endotoxemia by stimulating higher expression of the IL- receptor, it seems possible that IL- could underlie most of the anti-inflammatory activity of PDX. However, herein we observed PDX-dependent anti-inflammatory activity in two experimental paradigms in which IL- secretion was either repressed by PDX treatment (Supplementary Fig. ) or not present due to genetic deletion of Il (Supplementary Fig. d). Thus, our data suggest that the intrinsic anti-inflammatory activity of PDX does not hinge upon skeletal muscle IL- release. Alternatively, it is plausible that IL- independent activation of AMPK might underlie part of the anti-inflammatory activity of PDX,. Nonetheless, our Veh PDX Gpc d Tissue chemokines and cytokines (pg mg ewat) studies in Il-knockout mice revealed that PDX-dependent AMPK activation and TNF-α repression in the absence of IL- is not sufficient to entirely prevent the onset of lipid-induced insulin resistance. Although we cannot rule out the possibility that higher AMPK activation is required in concert with release of IL- to elicit the glucoregulatory effects of PDX, we did find that PDX treatment improves insulin sensitivity in obese db/db mice without affecting the inflammatory phenotype of these animals. Further studies are therefore warranted in animals lacking AMPK to determine what role AMPK might have in the glucoregulatory and anti-inflammatory activities of PDX. It may also be possible that AMPK activation is necessary for PDX-mediated IL- release. In conclusion, our studies identify the docosanoid resolution mediator, PDX, as an agent that carries a previously uncharacterized potent therapeutic potential for lipid-induced and obesity-linked insulin resistance and type diabetes. PDX enhances both hepatic and peripheral glucose metabolism in vivo by promoting release of the prototypic myokine, IL-. Notably, PDX stimulates skeletal muscle IL- release in a cell-specific manner that is very selective to the chemical structure of PDX. These findings may lead to the development of PDX-based muscle IL- secretagogues as a new class of drugs for the treatment of insulin resistance and type diabetes. Methods Methods and any associated references are available in the online version of the paper. Note: Any Supplementary Information and Source Data files are available in the online version of the paper. Acknowledgments We thank C. Dion for surgical preparation of the mice and S. Pelletier for assistance with the palmitate macrophage studies. We also thank C. Serhan (Brigham and Women s Hospital, Harvard University) and N. Flamand (Heart and Lung Institute, Laval University) for providing PD and (S,S)-diHETE, respectively, and M. Schwab and K. Bellmann for their help with the PCR analyses. T7i fibroblasts were a gift from M. Lombès, INSERM U7. This work was supported by grants e IL- (pg per µg protein) Veh CCL IL-β IL- Veh Muscle PDX Tissue chemokines and cytokines (pg mg ewat) IL- (pg ml ) CCL PDX 7 TNF-α Plasma nature medicine advance online publication

6 to A.M. from the Canadian Diabetes Association and from the Canadian Institutes of Health Research (CIHR). A.M. is partially funded by a CIHR/Pfizer Chair in the pathogenesis of insulin resistance and cardiovascular diseases. P.J.W. is the recipient of a PhD studentship award from the Fonds de la Recherche en Santé du Quebec. AUTHOR CONTRIBUTIONS P.J.W. and A.M. conceived of the study and wrote the manuscript. P.J.W., P.S.-P., A.C., P.L.M. and E.S.-A. performed mouse experiments. P.J.W., P.L.M., B.M. and E.S.-A. performed cell culture experiments. P.J.W., P.L.M., E.S.-A. and B.M. conducted ELISA and multiplex analyses. P.J.W. and P.L.M. carried out western blotting and PCR. All authors analyzed and discussed data and commented on the manuscript. COMPETING FINANCIAL INTERESTS The authors declare competing financial interests: details are available in the online version of the paper. Reprints and permissions information is available online at reprints/index.html.. White, P.J., Arita, M., Taguchi, R., Kang, J.X. & Marette, A. Transgenic restoration of long-chain n fatty acids in insulin target tissues improves resolution capacity and alleviates obesity-linked inflammation and insulin resistance in high-fat fed mice. Diabetes 9, 7 ().. Serhan, C.N. et al. Anti-inflammatory actions of neuroprotectin D/protectin D and its natural stereoisomers: assignments of dihydroxy-containing docosatrienes. J. Immunol. 7, 9 ().. Chen, P. et al. Full characterization of PDX, a neuroprotectin/protectin D isomer, which inhibits blood platelet aggregation. FEBS Lett., 7 (9).. Awazawa, M. et al. Adiponectin enhances insulin sensitivity by increasing hepatic IRS- expression via a macrophage-derived IL- dependent pathway. Cell Metab., ().. Pedersen, B.K. & Febbraio, M.A. Muscle as an endocrine organ: focus on musclederived interleukin-. Physiol. Rev., 79 ().. Pedersen, B.K. & Febbraio, M.A. Muscles, exercise and obesity: skeletal muscle as a secretory organ. Nat. Rev. Endocrinol., 7 (). 7. Stanford, K.I. et al. Brown adipose tissue regulates glucose homeostasis and insulin sensitivity. J. Clin. Invest., ().. Sag, D., Carling, D., Stout, R.D. & Suttles, J. Adenosine -monophosphate activated protein kinase promotes macrophage polarization to an anti-inflammatory functional phenotype. J. Immunol., (). 9. Inoue, H. et al. Role of hepatic STAT in brain-insulin action on hepatic glucose production. Cell Metab., 7 7 ().. Inoue, H. et al. Role of STAT- in regulation of hepatic gluconeogenic genes and carbohydrate metabolism in vivo. Nat. Med., 7 ().. Carey, A.L. et al. Interleukin- increases insulin-stimulated glucose disposal in humans and glucose uptake and fatty acid oxidation in vitro via AMP-activated protein kinase. Diabetes, 97 ().. Kelly, M., Gauthier, M.S., Saha, A.K. & Ruderman, N.B. Activation of AMP-activated protein kinase by interleukin- in rat skeletal muscle: association with changes in camp, energy state, and endogenous fuel mobilization. Diabetes, 9 9 (9).. Kelly, M. et al. AMPK activity is diminished in tissues of IL- knockout mice: the effect of exercise. Biochem. Biophys. Res. Commun., 9 ().. Clària, J., Dalli, J., Yacoubian, S., Gao, F. & Serhan, C.N. Resolvin D and resolvin D govern local inflammatory tone in obese fat. J. Immunol. 9, 97 ().. Jové, M., Planavila, A., Laguna, J.C. & Vázquez-Carrera, M. Palmitate-induced interleukin production is mediated by protein kinase C and nuclear-factor κb activation and leads to glucose transporter down-regulation in skeletal muscle cells. Endocrinology, 7 9 ().. Shi, H. et al. TLR links innate immunity and fatty acid-induced insulin resistance. J. Clin. Invest., (). 7. Wunderlich, F.T. et al. Interleukin- signaling in liver-parenchymal cells suppresses hepatic inflammation and improves systemic insulin action. Cell Metab., 7 9 ().. Pedersen, B.K. & Febbraio, M.A. Point: interleukin- does have a beneficial role in insulin sensitivity and glucose homeostasis. J. Appl. Physiol., (7). 9. Mauer, J. et al. Signaling by IL- promotes alternative activation of macrophages to limit endotoxemia and obesity-associated resistance to insulin. Nat. Immunol., ().. Gray, S.R., Ratkevicius, A., Wackerhage, H., Coats, P. & Nimmo, M.A. The effect of interleukin- and the interleukin- receptor on glucose transport in mouse skeletal muscle. Exp. Physiol. 9, 99 9 (9).. Tweedell, A. et al. Metabolic response to endotoxin in vivo in the conscious mouse: role of interleukin-. Metabolism, 9 9 ().. Pilon, G., Dallaire, P. & Marette, A. Inhibition of inducible nitric-oxide synthase by activators of AMP-activated protein kinase: a new mechanism of action of insulinsensitizing drugs. J. Biol. Chem. 79, ().. Centeno-Baez, C., Dallaire, P. & Marette, A. Resveratrol inhibition of inducible nitric oxide synthase in skeletal muscle involves AMPK but not SIRT. Am. J. Physiol. Endocrinol. Metab., E9 E9 (). advance online publication nature medicine

7 ONLINE METHODS Animal studies. -week-old male C7BL/ mice from Jackson Labs were used for the first paired lipid infusion hyperinsulinemic-euglycemic clamp study. These mice were placed on a standard laboratory chow diet with free access to food and water and kept in a -h light-dark cycle at the Laval University Hospital Research Center Animal Facility. Mice were randomly assigned to saline, lipid or lipid + PDX groups. Five days before the experiment, mice were anesthetized, and PE- catheters (Harvard Apparatus, QC, Canada) were inserted into the left common carotid artery and the right jugular vein for blood sampling and infusions, respectively. Mice were fasted for h leading up to the protocol. Immediately before the start of the lipid infusion, PDX ( µg) or an equal volume of vehicle was administered via the jugular catheter to each group. Mice were then infused for h with saline ( ml per kg body weight per hour) or lipid (% intralipid emulsion (Baxter, ON, Canada) ml per kg body weight per hour with IU ml heparin (LEO Pharma, ON, Canada)).. h into the infusion, PDX ( µg) or vehicle was again administered to the appropriate groups, and the hyperinsulinemic-euglycemic clamp was initiated as previously described,. The clamp protocol consisted of a 9-min tracer equilibration period followed by a -min experimental period. A -µci bolus of [- H]glucose was given at the start of the tracer equilibration period, followed by an infusion of. µci per minute for 9 min. Blood samples were drawn for the assessment of glycemia, plasma insulin and glucose turnover. Following the 9-min tracer equilibration period, the clamp began with a primed-continuous infusion of human insulin ( mu per kg body weight bolus followed by mu per kg body weight per min, Humulin R, Eli Lilly, Indianapolis, IN). The [- H]glucose infusion was increased to. µci min for the remainder of the experiment. Euglycemia (. 7. mm) was maintained during clamps by infusing % dextrose as necessary. Blood samples were taken continuously to determine glucosespecific activity as well as insulin concentrations. Mice received saline-washed erythrocytes from donor mice throughout the experimental period ( µl min ) to prevent a fall of % hematocrit. Hepatic glucose production and Rd were determined using Mari s non steady-state equations for a twocompartmental model. For the second paired lipid infusion hyperinsulinemic-euglycemic clamp study, -week-old male B.9S-Il tmkopf (Il-knockout) and genetically matched nonlittermate control (WT) mice from Jackson Labs were used. Mice from each genetic background were randomly assigned to saline-vehicle, saline- PDX, lipid-vehicle and lipid-pdx groups. The lipid-infusion clamp study was performed as described above, except the clamp was performed with an insulin infusion of. mu per kg body weight per minute 7 to ensure detection of potential differences between insulin-sensitive saline-infused animals. The mu per kg body weight hyperinsulinemic-isoglycemic clamp study was performed in -week-old db/db mice from the BKS.Cg-Dock7 m +/+ Lepr db strain at Jackson Labs. Mice were randomly assigned to vehicle or PDX treatment groups. In preparation for the clamps, mice were catheterized as described for the lipid infusion study. Mice were administered µg of PDX or an equal volume of vehicle i.v. at h and also 9 min before the initiation of the insulin pump. Preclamp glycemia was taken immediately before the second injection of PDX. The clamp began 9 min later with a primed-continuous infusion of human insulin ( mu per kg body weight bolus followed by mu per kg body weight per minute, Humulin R, Eli Lilly, Indianapolis, IN). Glycemia was maintained as close as possible to individual fasting values by infusing % dextrose as necessary. Blood samples were taken continuously to determine glycemia as well as insulin concentrations. Mice received saline-washed erythrocytes from donor mice throughout the experimental period ( µl min ) to prevent a fall of % hematocrit. The chronic PDX experiments were performed as described for the acute study, except PDX was administered twice daily at a dose of µg i.v. and for five consecutive days until the day of the clamp. On clamp day, the protocol was performed precisely as described above for the acute study. All animal procedures were approved and carried out in accordance with directions of the Laval University Council for Animal Care and the Canadian Council for Animal Care. CC myotubes. CC myoblasts were maintained in DMEM containing % FBS. Differentiation to myotubes was initiated by addition of DMEM containing % horse serum. The experiment was conducted d after the addition of the differentiation medium. Immediately before the commencement of experiments, fresh medium was added to the cells. Next, vehicle or PDX ( nm, nm and µg) was added to the appropriate wells. and min after the addition of PDX, medium was collected and frozen for IL- quantification, and cells were washed in ice-cold PBS. For extraction of mrna, cells were then lysed and scraped in µl of RLT buffer (QIAGEN). For examination of AMPK activity, cells were lysed and scraped in µl of ice-cold lysis buffer containing mm HEPES, ph 7., mm NaCl, mm EGTA, mm β-glycerophosphate, % NP, mm NaF, mm Na VO, protease inhibitor cocktail (Sigma). For structure-function studies, we compared the activity of PDX ( nm) with nm of PD, resolvin D (Cayman Chemical) or (S,S)-diHETE (Cayman Chemical). After h of incubation with these lipid mediators, medium was collected and IL- determined as described above. Macrophages. J77A. mouse macrophages (ATCC) were maintained in DMEM (% FBS) until % confluence. For palmitate-stimulation studies, a mm palmitate solution or methanol vehicle in Alpha MEM (% BSA) was added to fresh DMEM (% FBS) to give a final concentration of µm palmitate. Concomitantly, PDX ( or nm, Cayman Chemical) or vehicle was added to the medium. After h, the medium was collected, and cells were lysed as described for the studies in CC myotubes. For IL- release studies, fresh DMEM (% FBS) was added to the cells, and they were stimulated with PDX (Cayman Chemical), vehicle or LPS ( ng ml ). The medium was collected, and cells were lysed as described above. Brown adipocytes. T7i fibroblasts were cultured in DMEM/HamF (Gibco- Invitrogen) supplemented with % FBS, IU ml penicillin, g ml streptomycin and mm HEPES and grown at 7 C in a humidified atmosphere with % CO. Differentiation into brown adipocytes was achieved under standard conditions by incubation of subconfluent T7i cells with nm triiodothyronine and nm insulin for at least 7 d. Experiments were performed 7 d after differentiation. On the day of the experiment, medium was replaced and cells were stimulated with nm PDX (Cayman Chemical), vehicle or norepinephrine (Sigma). The medium was collected and IL- determined as described above. Western blotting. Snap-frozen gastrocnemius muscle and liver from mice were pulverized in liquid nitrogen and then lysed overnight at C in the lysis buffer described for the CC myotube experiments. Immunoblotting of myotube, macrophage, liver and muscle lysates was then performed as previously described. Briefly, µg of protein was loaded onto a 7.% acrylamide gel, subjected to SDS-PAGE and then transferred onto nitrocellulose membranes. Membranes were cut into regions of specific molecular weights containing the protein of interest (Akt 7 kda; JNK and AMPK 7 kda; STAT 7 kda; inos kda) and then blocked and probed with the appropriate antibodies. All primary antibodies were used at a concentration of :,. Secondary antibodies were diluted :,. Antibodies for pakt Ser7 (CST97),, pjnk Thr/Tyr (CST9),9, JNK (CST9),9, pampk Thr7 (CST),, AMPK (CST),, pstat Ser77 (CST9), and STAT (CST9), were obtained from Cell Signaling Technology (MA, USA). Antibodies for total Akt (sc-), and inos (), were from Santa Cruz Biotechnology (CA, USA) and BD Transduction Laboratories (Canada), respectively. Real-time RT-PCR. RNA was extracted from CC myotubes using an RNeasy mini kit from QIAGEN. RNA from homogenized liver tissue was extracted using an RNeasy Fibrous Tissue Mini Kit from QIAGEN. RNA was then reverse transcribed to cdna using the high-capacity cdna Reverse Transcription Kit from Applied Biosystems. Real-time PCR for Ppargca, Pck, Gpc and Gapdh was then performed using TaqMan assay on demand probes and primers from Applied Biosystems in a CFX9 real-time system from BIO-RAD. The relative expression of genes of interest was then determined by normalization to the reference gene Gapdh using the comparative C T method for relative gene expression. doi:./nm.9 nature medicine

8 Analytical methods. Chemokines and cytokines were quantified in macrophage medium or mouse plasma using the MILLIPLEX MAP Mouse Cytokine/ Chemokine Kit (Millipore Corporation, MA, USA). Nitrite accumulation in medium was determined by Greiss assay as previously described. Plasma insulin levels were assessed by RIA (Linco, MI, USA). IL- was quantified in mouse skeletal muscle and liver as well as in CC, T7i and nonstimulated macrophages using the mouse IL- ELISA kit from R&D systems. Total plasma adiponectin was determined using the ELISA from ALPCO. Statistical analyses. An unpaired Student s t-test was used to identify statistically significant differences between two groups. For comparison among more than two groups, a one-way ANOVA was used followed by Newman-Keuls multiple comparison test. P values of <. were considered significant.. Charbonneau, A. & Marette, A. Inducible nitric oxide synthase induction underlies lipid-induced hepatic insulin resistance in mice: potential role of tyrosine nitration of insulin signaling proteins. Diabetes 9, 7 ().. Xu, E. et al. Targeted disruption of carcinoembryonic antigen-related cell adhesion molecule promotes diet-induced hepatic steatosis and insulin resistance. Endocrinology, (9).. Mari, A. Estimation of the rate of appearance in the non-steady state with a twocompartment model. Am. J. Physiol., E E (99). 7. Mulvihill, E.E. et al. Nobiletin attenuates VLDL overproduction, dyslipidemia, and atherosclerosis in mice with diet-induced insulin resistance. Diabetes, 7 ().. Xu, E. et al. Hepatocyte-specific Ptpn deletion protects from obesity-linked hepatic insulin resistance. Diabetes, 99 9 (). 9. Chung, J. et al. HSP7 protects against obesity-induced insulin resistance. Proc. Natl. Acad. Sci. USA, 79 7 ().. Jenkins, Y. et al. AMPK activation through mitochondrial regulation results in increased substrate oxidation and improved metabolic parameters in models of diabetes. PLoS ONE, e7 ().. Moh, A. et al. STAT sensitizes insulin signaling by negatively regulating glycogen synthase kinase- β. Diabetes 7, 7 ().. Río, A. et al. Reduced liver injury in the interleukin- knockout mice by chronic carbon tetrachloride administration. Eur. J. Clin. Invest., ().. Schmittgen, T.D. & Livak, K.J. Analyzing real-time PCR data by the comparative C T method. Nat. Protoc., (). nature medicine doi:./nm.9