Endothelin receptor antagonism improves glucose handling, dyslipidemia, and 1 adipose tissue inflammation in obese mice

Endothelin-1 (ET-1) is elevated in patients with obesity; however, its contribution to the 35 pathophysiology related to obesity is not fully understood. We hypothesized that high ET- 36 1 levels cause dyslipidemia, inflammation, and insulin resistance within the adipose tissue 37 of obese mice. To test this hypothesis, male C57BL/6J mice were fed either normal diet 38 (NMD) or high fat diet (HFD) for 8 weeks followed by 2 weeks of treatment with either 39 vehicle, atrasentan (ET A receptor antagonist, 10mg/kg/day), or bosentan (ET A /ET B 40 receptor antagonist, 100mg/kg/day). Atrasentan and bosentan lowered circulating non- 41 esterified free fatty acids and triglycerides seen in HFD mice, while atrasentan-treated 42 mice had significantly lower liver triglycerides compared to non-treated HFD mice. ET-1 43 receptor blockade significantly improved insulin tolerance compared to insulin resistant 44 HFD mice and lowered expression of genes in epididymal white adipose tissue (eWAT) 45 associated with insulin resistance and inflammation. Flow cytometric analyses of eWAT 46 indicated that HFD mice had significantly higher percentages of both CD4 + and CD8 + T 47 cells compared to NMD mice, which was attenuated by treatment with atrasentan or 48 bosentan. Atrasentan treatment also abolished the decrease in eosinophils seen in HFD 49 mice. Taken together, these data indicate that ET A and ET A /ET B receptor blockade 50 improves peripheral glucose homeostasis, dyslipidemia, and liver triglycerides, and also 51 attenuates the proinflammatory immune profile in eWAT of mice fed a HFD. These data 52 suggest a potential use for ET A and ET A /ET B receptor blockers in the treatment of obesity- 53 associated dyslipidemia and insulin resistance.

two clinical trials suggest that high levels of ET-1 promote dyslipidemia in patients with 86 diabetic nephropathy 5, 6 ; however, the mechanisms are not yet understood. Therefore, 87 ET-1 appears to play a major role in pathophysiology of obesity, and ET-1 receptors may 88 be attractive targets to attenuate cardiovascular risk in patients with obesity. 89 Obesity causes a shift to a pro-inflammatory immune cell profile, 2 especially 90 observed in visceral adipose tissue. 7-9 In addition, visceral adipose tissue abundance is 91 highly correlated to cardiovascular disease risk, whereas other depots such as 92 subcutaneous adipose have little or no correlation. 10 In white adipose, obesity causes 93 tissue to become hypoxic and immune cell populations become dysregulated, leading to 94 an increase in macrophages, an increase in T lymphocytes and a decrease in 95 eosinophils, among others. 11 This shift in immune cell population promotes dysfunction 96 with the adipose tissue, contributing to pathophysiology related to obesity. This includes 97 insulin resistance at the level of the adipocyte as well as peripherally due to alterations in 98 circulating adipokines, including insulin sensitizing adipokines such as adiponectin and 99 adipsin. Given the importance of immune cells in modulating adipocyte function, more 100 studies are needed to identify methods to target and improve inflammation within the 101 adipose tissue. 102 A strong relationship between ET-1 and inflammation has been established in 103 several models of disease including chronic kidney disease and sickle cell disease. 12 ET-104 1 receptor antagonism has been shown to reduce pro-inflammatory cytokines such as 105 tumor necrosis factor alpha (TNF-a). Given the relationship between ET-1, obesity, and 106 inflammation, we hypothesize that upregulation of ET-1 within the eWAT of obese mice 107 promotes an inflammatory immune cell profile which contributes to dyslipidemia and 108 insulin resistance. The goal of the current study is to determine if ET-1 receptor 109 antagonism attenuates dyslipidemia, insulin resistance and adipose tissue inflammation that female mice are protected from HFD-induce metabolic syndrome compared to males. 120 Mice were habituated to the animal facility for one week upon arrival and fed ad libitum. 121 At 8 weeks of age, mice were randomized and individually housed into four groups; 122 normal diet-fed (NMD; 12.6% kcal fat, 30% kcal carbohydrate, Envigo, TD.05230) (n=7), 123 high fat diet-fed (HFD; 45% kcal fat, 42% kcal carbohydrate, Envigo TD.88137) (n=7), 124 HFD atrasentan (HFD Atr) (n=5), and HFD bosentan (HFD Bos) (n=6) for 8 weeks. Mice 125

Flow cytometric analyses of adipose tissue 171
Given the high correlation between visceral adipose tissue abundance, 172 inflammation and cardiovascular disease risk in obesity, 10 immune cell populations in 173 epididymal white adipose tissue (eWAT) were assessed. eWAT is the most abundant 174 visceral adipose tissue depot in mice. eWAT was homogenized in GentleMACS Octo 175 Dissociator (Miltenyi Biotec). Briefly, ~1 g of adipose tissue was minced into ~5 mm 176 pieces and placed in a GentleMACS C (Miltenyi Biotec) tube with 10 mL RPMI media 177 containing 200 U/mL DNase and 10 mg/mL collagenase IV. The sample was 178 homogenized using the manufacturer's program designed for adipose tissue. The 179 homogenate was then filtered through a 70 μm filter into a 50 mL tube and allowed to 180 settle for 10 min to allow the adipocytes to separate from the stromal vascular fraction 181 (SVF). The lower SVF was removed and transferred to a 15 mL tube and centrifuged for 182 10 min at 400 x g. The resulting cell pellet was resuspended in 3 mL of 1X PharmLyse 183 (BD Biosciences) and incubated for 5 min at room temperature to lyse erythrocytes. Ten 184 mL of 1X PBS, 2% FCS was added to the cells to wash followed by centrifugation at 350 185 x g for 5 min. Cells were then used for flow cytometry. 186 Briefly, cells were washed and resuspended in 1X PBS, 2% FCS, and 0.9% 187 sodium azide at a concentration of 1 x 10 7 cells/mL. 1X10 6 cells (100 μL) were aliquoted 188 into a flow cytometry tube and incubated with 0.25 μg of anti-mouse CD32/CD16 (FcR 189 block, BD Biosciences) for 5 min. on ice. Cells were then stained with either isotype 190 control antibodies or antibodies shown in Table 1

Randomisation and Statistics. 231
Although it is not possible to blind researchers from diet of mice, samples were 232 blinded for all assays. All data are expressed as mean ± SEM. Data were tested for 233 statistical significance by one-way ANOVA for one variable datasets or two-way repeated 234 measure ANOVA (ITT and OGTT). Tukey's post hoc test used to compare groups. P<0.05 235 was considered statistically significant. All graphs and statistical analyses were performed 236 using GraphPad Prism. 237

ET-1 receptor blockade does not significantly affect body weight or fat mass of 240
HFD-fed mice. 241 C57BL/6J mice were fed a NMD or HFD for 10 weeks. As expected, male mice fed 242 NMD had significantly lower body weight, fat mass, and lean mass compared to HFD fed 243 mice. Female mice on HFD gained less body weight and fat mass over the course of the 244 experiment compared to male mice (Supplemental Figure 2A and 2B). Further, there were 245 no significant differences in fasting blood glucose, glucose tolerance, or insulin tolerance 246 (Supplemental Figure 2D, 2E, and 2F) between NMD or HFD fed females; therefore, we 247 proceeded with only male mice. There were no differences in body weight, lean mass, or 248 fat mass between HFD, HFD+Atr and HFD+Bos mice throughout the experimental 249 protocol ( Fig. 1A-1C). In addition, there were no significant differences in total body water 250 during treatment between any group (Fig. 1D). Finally, there were no detectable 251 differences in caloric or water intake among all four treatment groups throughout the 252 duration of the treatment, indicating that the differences in any endpoints were 253 independent of caloric intake and hydration status ( Fig. 1E and 1F). 254

ET-1 receptor blockade improves dyslipidemia in HFD-fed mice. 255
Fasting plasma NEFA and triglyceride concentrations were increased in HFD-fed 256 mice compared to NMD-fed mice (p=0.001 and p<0.0001 respectively). The increase in 257 triglycerides and NEFA was attenuated in HFD-fed mice treated with atrasentan (p=0.005 258 and p=0.0003 respectively) or bosentan (p=0.06 and p=0.001 respectively; Fig. 2A and 259 2B). Similar to the circulating lipid profile, hepatic triglyceride content was significantly 260 increased in HFD-fed mice compared to NMD (p<0.001). Atrasentan treated mice had 261 lower hepatic triglyceride content compared to HFD-fed mice treated with vehicle 262 (p=0.009, Fig. 2C) while bosentan treatment had no significance difference in hepatic 263 triglycerides (p=0.09). Total cholesterol was increased in HFD-fed mice compared to 264 NMD-fed mice (p<0.0001), with no significant effect of treatment with either atrasentan or 265 bosentan (Fig. 2D). There were no differences in HDL levels between NMD-fed mice and 266 HFD-fed groups. There was, however, a significant increase in HDL in HFD+Bos (p=0.03) 267 treated mice compared to HFD vehicle, but no statistical difference between HFD and 268 HFD+Atr (p=0.08; Fig. 2E). LDL levels were higher all in all HFD-fed mice groups 269 compared to NMD-fed mice (Fig. 2F). 270

HFD-fed mice. 272
We next determined whether ET-1 receptor blockade improved glucose and insulin 273 tolerance in HFD-fed mice. After a 6 hour fast, HFD-fed mice exhibited significant 274 hyperglycemia compared to NMD-fed mice (p<0.0001), an effect that was attenuated in  Similarly, insulin tolerance was significantly impaired in HFD-fed mice compared to NMD 283 (p=0.001; Fig. 3E and 3F). Treatment with atrasentan or bosentan improved insulin 284 tolerance compared to HFD mice indicated by a significant increase in AUC (p<0.0001 285 and p=0.01 respectively; Fig. 3E and 3F). 286

ET-1 is upregulated in adipose tissue via Hif1α. 287
ET-1 expression has been shown to increase in hypoxic environments, which 288 occurs in the adipose tissue of patients with obesity and rodent models of obesity. In the 289 current study, Hif1α mRNA, a marker of hypoxia, 16 was significantly increased in eWAT 290 of HFD fed mice compared to NMD fed mice (p<0.0001). Interestingly the increase in 291 Hif1α mRNA was attenuated by 49% in mice treated with atrasentan (p=0.002; Fig. 4A). 292 In addition, ET-1 mRNA in eWAT was increased from 711 to 1086 copy counts/50 ng 293 RNA in response to diet induced obesity (p=0.04; Fig. 4B). Atrasentan exacerbated the 294 increase in ET-1 expression, although there was no detectable difference between 295 vehicle and bosentan treated HFD fed mice. (p=0.95; Fig. 4B). Surprisingly, protein 296 content had a negative correlation with mRNA expression, in that HFD fed mice had 297 significantly lower eWAT ET-1 protein content compared to NMD mice, and atrasentan 298 treatment reduced ET-1 content even more. This is likely due to increased ET-1 binding 299 to ETB receptors. ETA receptor expression was significantly reduced in HFD-fed and 300 atrasentan treated mice compared to NMD mice, while bosentan treated mice had no 301 detectable difference in ETA mRNA expression compared to NMD mice. (Fig. 4D). There 302 were no significant differences in ETB mRNA expression among all groups (Fig. 4E), 303 although eWAT had higher gene expression levels of ETB mRNA compared to ETA mRNA 304 in all groups. 305 To determine if ET-1 is elevated in response to hypoxia in adipocytes, 3T3-L1 306 fibroblast cells were differentiated into adipocytes and exposed to hypoxia. Hypoxia 307 induced a 3-fold increase in Hif1α mRNA (Fig. 4F) and this was associated with 4-fold 308 increase in ET-1 mRNA. The increase in ET-1 was attenuated in a dose-dependent 309 manner when cells were pretreated with the Hif1α inhibitor IDF-11774 (Fig. 4G) 310 suggesting that Hif1α drives the increase in ET-1 expression in the adipose of obese 311 mice. 312

ET-1 receptor blockade improves hypoadiponectinemia in HFD-fed mice. 313
Circulating plasma adiponectin was significantly decreased in HFD compared to 314  10 was only significantly attenuated by atrasentan treatment (Fig. 6I). There were no 345 significant differences in gene expression for Il-6 and Il-1r between treated and HFD-fed 346 mice (Supplemental Figure 2). CD8 + T cell percentages were significantly lower in mice that received bosentan, while 467 atrasentan caused a significant decrease CD8 + T cell percentages. CD4 + Th1 cells 468 secrete IFN-γ, which can disrupt insulin signaling in the adipose and could lead to or 469 exacerbate insulin resistance systematically. 53 TNF-α, which has been widely studied in 470 obesity, 54, 55 is also secreted by Th1 cells, among other cell types. Expression of TNF-α 471 was elevated in eWAT of mice fed a HFD, but was decreased in mice that received 472 atrasentan. Congruous with our results, previous studies have shown that blockade of the 473 ETA receptor lowered TNF-α concentrations in humans and rats, 56, 57 although the exact 474 mechanism has yet to be determined. Further studies are needed to characterize the Th 475 subset dynamics in adipose tissue in response to endothelin antagonism. CD8 + CTL 476 infiltrate the adipose in obesity and accumulate, along with macrophages, in crown-like 477 structures around dying adipocytes. These cells are important in the recruitment of 478 macrophages, and their depletion decreases pro-inflammatory cytokine production and 479 improves insulin sensitivity. 58 We also detected increased IL-12 and IL-10 mRNA 480 expression in eWAT of HFD-fed mice. IL-12 is a pro-inflammatory cytokine which is 481 important for the polarization of CTL to an activated Tc1 phenotype. 59 IL-10 is principally 482 known as an anti-inflammatory cytokine that decreases the activity and/or expression of 483 various proinflammatory cytokines and chemokines. 60 IL-10 is upregulated in white 484 adipose tissue during obesity, where it may have a deleterious role since knockdown of 485 IL-10 in mouse adipose tissue protects from obesity and promotes thermogenesis. 61 486 A seeming contradiction observed in the current study is the apparent negative 487 correlation between ET-1 mRNA and ET-1 peptide content in eWAT. Obesity increased 488 ET-1 mRNA in eWAT, and ETA blockade increased ET-1 mRNA further, while protein 489 content was proportionally lower. One potential reason is the increase in ET-1 receptor 490 expression, especially that of the ETB receptor. ETB receptor content in the lungs is high 491 compared to most other tissues and is responsible for clearing ET-1 from the circulation 492 by internalizing bound ET-1. This inverse relationship between ET-1 mRNA and protein 493 levels has been observed previously in obese mice. In fact, obese mice fed a high fat 494 diet had significantly lower peptide levels in the lung, even in mice that overexpress ET-495 1 in vascular endothelial cells. 62 This is thought to be from upregulation of ETB receptors 496 in the lungs leading to increased clearance. We speculate the same phenomenon in the 497 current study. (E) and water intake (ml/day) (F) was measured daily. A-D were analyzed by two-way 539 ANOVA, and panels E and F were analyzed by one-way ANOVA with post-hoc Tukey's 540 test between individual groups. Data are expressed as mean ± SEM * p<0.05, ** p<0.01, 541 *** p<0.001. (n=7), HFD (n=7), HFD+Atr (n=5), and HFD+Bos (n=6) mice after ten weeks on diets 546 and two weeks of treatment with atrasentan or bosentan. Liver triglycerides were 547 measured among all groups and normalized to liver weight (F). Data were analyzed by 548 one-way ANOVA with Tukey's post-hoc test for individual groups. E and F were 549 analyzed by two-way ANOVA. Data are expressed as mean ± SEM * p<0.05, ** p<0.01, 550 *** p<0.001, **** p<0.0001. 551 552 in HFD-fed mice. In vivo glucose homeostasis was determined by OGTT (A) and IPITT 554 (C) for of NMD (n=7), HFD (n=7), HFD+Atr (n=5), and HFD+Bos (n=6) mice after 10 555 weeks on diet and two weeks of treatment with atrasentan or bosentan. Area under the 556 curve for GTT (B) and GTT (D). Fasting blood glucose was measured after a 6-hour fast 557 (E). Plasma insulin was measured via ELISA (n=5 per group). Data in A, B, D, and F-H 558 were analyzed by one-way ANOVA with Tukey's post-hoc test for individual groups. 559 Statistical analysis for panels A, B D, F-H were done by one-way ANOVA with post-hoc 560 Tukey's test between individual groups, and panels C and E were analyzed by two-way 561 ANOVA with repeated measures. Data are expressed as mean ± SEM * p<0.05, 562 ** p<0.01, *** p<0.001, **** p<0.0001. 563 564