Volume 23, Issue 4, Pages 699-711 (April 2016) Sodium Intake Regulates Glucose Homeostasis through the PPARδ/Adiponectin- Mediated SGLT2 Pathway  Yu Zhao, Peng Gao, Fang Sun, Qiang Li, Jing Chen, Hao Yu, Li Li, Xing Wei, Hongbo He, Zongshi Lu, Xiao Wei, Bin Wang, Yuanting Cui, Shiqiang Xiong, Qianhui Shang, Aimin Xu, Yu Huang, Daoyan Liu, Zhiming Zhu  Cell Metabolism  Volume 23, Issue 4, Pages 699-711 (April 2016) DOI: 10.1016/j.cmet.2016.02.019 Copyright © 2016 Elsevier Inc. Terms and Conditions

Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 1 Adipose-Specific Knockout of PPARδ Ameliorates the Inhibitory Effect of an HSD on Natriuresis (A) Representative western blots of PPARα, PPARδ, PPARγ, and GAPDH in the perirenal fat of mice fed ND or HSD for 24 weeks. Quantitative results are shown on the right (n = 6). (B) Western blots of PPARα, PPARδ, PPARγ, and GAPDH in 3T3-L1 adipocytes treated by exogenous NaCl or mannitol for 24 hr. Quantitative results are shown below (n = 3). (C) The sodium content in the perirenal fat or renal cortex (dry weight) of wild-type mice fed an ND or HSD for 24 weeks (n = 6–9). (D) The fasting blood glucose of PPARδflox/flox and Fabp4-PPARδflox/flox mice fed an ND or HSD for 24 weeks (n = 6). (E) Intraperitoneal glucose tolerance test (IPGTT) results of indicated groups (n = 6). (F and G) The 24 hr urine volume (n = 9) (F) and urine sodium excretion (UNaV) (n = 8) (G) of indicated groups. The data in (A)–(G) are presented as the mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 for comparisons between the indicated groups in (A), (D), (F), and (G), or compared with ND group or control in (B), (C), and (E). See also Figures S1 and S2. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 2 SGLT2 Inhibition Blocks the Effect of Adipose PPARδ Knockout on Increasing Natriuresis and Glycosuria by High Sodium Intake (A) Western blots of PPARδ in perirenal fat and SGLT2 in renal cortex of PPARδflox/flox and Fabp4-PPARδflox/flox mice fed with ND or HSD for 24 weeks; quantitative results of renal SGLT2 are shown on the right (n = 6). (B and C) The blood glucose of PPARδflox/flox and Fabp4-PPARδflox/flox mice treated with placebo (vehicle) (B) or dapagliflozin (C) 0, 2, and 4 hr after oral glucose gavage (n = 8). (D) The 24 hr urinary sodium excretion of indicated mice treated with vehicle (n = 24) or dapagliflozin (n = 16). (E) The 24 hr urine glucose excretion of indicated mice treated with vehicle (n = 39) or dapagliflozin (n = 33). The data in (A)–(E) are expressed as the mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 compared with indicated group. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 3 Activation of Adipose PPARδ by a High-Sodium Diet Upregulates Adiponectin (A) Western blots of PPARδ, adiponectin, and GAPDH in the perirenal fat and renal cortex of PPARδflox/flox and Fabp4-PPARδflox/flox mice fed ND or HSD. (B) Quantitative analyses of the protein levels in (A) (n = 6). (C) Representative blots of the protein levels of adiponectin and GAPDH in 3T3-L1 adipocytes treated with exogenous NaCl or mannitol for 24 hr; quantitative results are shown on the right (n = 3). (D) The plasma adiponectin levels of PPARδflox/flox and Fabp4-PPARδflox/flox mice on ND or HSD (n = 8). (E) Western blots of PPARδ and adiponectin in primary cultured perirenal adipocytes from indicated mice and treated with GW501516 (GW) or GSK0660 (GSK). (F) Quantitative results of protein levels in (E) (n = 8). The data are expressed as the mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, compared between the indicated groups (B and D) or with control group (C and F). See also Figure S3. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 4 Adiponectin Transcriptionally Inhibits SGLT2 Expression Independent of PPARδ (A) Relative mRNA level of SGLT2 in HK-2 cells treated with control (vehicle) or 20 μg/ml Acrp30 for 24 hr (n = 3). (B) Representative western blots of SGLT2 and GAPDH in HK-2 cells treated with indicated dose of NaCl and/or 20 μg/ml Acrp30 for 24 hr; quantitative results are shown on the right (n = 3). (C) The protein expressions of adiponectin and SGLT2 in renal cortex tissues from C57BL/6 mice treated with control (vehicle) or 20 μg/ml Acrp30 for 24 hr in vitro in culture. Quantitative results are shown on the right (n = 6). (D) Western blots of adiponectin, SGLT2, and GAPDH in renal cortex of PPARδflox/flox and Fabp4-PPARδflox/flox mice treated with or without Acrp30 for 4 weeks; quantitative results are shown on the right (n = 8). (E) The binding sites of sp-1 and HNF-1α on SLC5A2 promoter are showed in the diagrammatic drawing, and the designed primers that covered these sites are shown below. (F) ChIP-qPCR results of the relative binding levels of IgG, sp-1, or HNF-1α to input detected by indicated primers. Data are shown as the mean ± SEM for three independent experiments. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, compared between the indicated groups (B, D, and F) or with control group (A, C, and D). See also Figure S4. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 5 Adiponectin Mediates the Inhibitory Effect of PPARδ Activation or High Salt Intake on SGLT2 Expression (A) Plasma adiponectin levels of Adn+/+ or Adn−/− mice treated with normal saline (Con) or a PPARδ agonist, GW501516, for 7 days (n = 8). (B and C) Western blots of PPARδ, adiponectin, SGLT2, and GAPDH in perirenal fat (B) and renal cortex (C) of Adn+/+ and Adn−/− mice treated with normal saline (Con) or GW501516; quantitative results are shown on the right (n = 8). (D) IPGTT results of Adn+/+ and Adn−/− mice treated with normal saline (Con) or GW501516 (n = 6). (E and F) The 24 hr urinary sodium excretion (n = 8) (E) and urinary glucose excretion (n = 16) (F) of Adn+/+ and Adn−/− mice treated with normal saline (Con) or GW501516. (G and H) Representative western blots of PPARδ, adiponectin, SGLT2, and GAPDH in the perirenal fat (G) or renal cortex (H) of Adn+/+ or Adn−/− mice fed with ND or HSD for 12 weeks; quantitative results are shown on the right (n = 4). (I) IPGTT results of Adn+/+ and Adn−/− mice on ND or HSD for 12 weeks (n = 4). The data are expressed as the mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001, compared between the indicated groups (C and H) or with ND or control group (A–I). See also Figure S5. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 6 Enhanced SGLT2 Activity Contributes to the Impaired Natriuretic Effect of an HSD in db/db Mice (A–D) The fasting blood glucose (n = 6–7) (A), urine glucose excretion (n = 16) (B), urine volume (n = 8–9) (C), and urinary sodium excretion (n = 8) (D) of db/m or db/db mice fed with ND or HSD for 24 weeks. (E) Relative mRNA level of SGLT2 in the renal cortex of db/m and db/db mice (n = 6). (F) Representative western blots of SGLT2 and GAPDH in renal cortex of db/m and db/db mice. Quantitative results are shown below (n = 6). (G and H) The blood glucose of ND- or HSD-fed db/m and db/db mice treated with vehicle (G) or dapagliflozin (H) 0, 2, and 4 hr after oral glucose gavage (n = 8). (I and J) The 24 hr urine sodium (I) and glucose excretion (J) of ND- or HSD-fed db/m and db/db mice treated with vehicle (n = 8) or dapagliflozin (n = 8). The data are expressed as the mean ± SEM. ∗p < 0.05, ∗∗p < 0.01, ∗∗∗p < 0.001 for comparisons between the indicated groups. See also Figure S6. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions

Figure 7 Urinary Sodium Excretion Is Negatively Correlated with the Blood Glucose Level and Positively Related with Plasma Adiponectin Level in Diabetic Patients (A and B) The scattered dots depict the relationship between urinary sodium excretion and blood glucose level (A) or HbA1c (B) in diabetic patients. (C and D) The 24 hr urinary sodium excretion of diabetic patients divided by 7.0 mM of fasting blood glucose (C) or 7.0% of HbA1c (D). The data in (C) and (D) are presented as the mean ± SEM. (E) The scattered dots depict the relationship between urinary sodium excretion and plasma adiponectin level in diabetic patients. Pearson’s correlation coefficient (r), statistical significance values (p) and number (n) are shown in (A), (B), and (E). (F) The working model of the study: high salt intake activates PPARδ in adipose tissue and stimulates adiponectin secretion, which reduces renal SGLT2, resulting in increased natriuresis. Hyperglycemia-increased SGLT2 activity contributes to the salt retention in diabetes. The green arrows, normal condition; the red arrows, hyperglycemia. See also Figure S7 and Tables S1 and S2. Cell Metabolism 2016 23, 699-711DOI: (10.1016/j.cmet.2016.02.019) Copyright © 2016 Elsevier Inc. Terms and Conditions