GLP-1-mediated delivery of the PPAR 𝛼 / 𝛾 dual-agonist Tesaglitazar improves obesity and glucose metabolism in mice

Dual-agonists activating the peroxisome proliferator-activated receptors alpha and gamma (PPAR 𝛼 / 𝛾 ) have shown benecial effects on glucose and lipid metabolism in patients with type 2 diabetes, but their development was discontinued due to unfavorable cardiovascular and/or renal effects. Here we report the design and preclinical evaluation of a molecule that covalently links the PPAR 𝛼 / 𝛾 dual-agonist Tesaglitazar to GLP-1 to allow for the GLP-1 receptor-dependent delivery of Tesaglitazar. GLP-1/Tesaglitazar does not differ from matched GLP-1 in GLP-1R signaling, but shows GLP-1R-dependent PPAR 𝛾 -RXR heterodimerization with enhanced ecacy to improve body weight, food intake, and glucose metabolism relative to GLP-1 or Tesaglitazar in mice with diet- and genetically-induced obesity. The conjugate fails to affect body weight and glucose metabolism in GLP-1R knockout (ko) mice and shows preserved effects in DIO mice at doses subthreshold for GLP-1 and Tesaglitazar to improve metabolism. Consistent with the GLP-1R expression pattern, LC/MS-based proteomics identied a series of novel PPAR protein targets in the hypothalamus that are acutely upregulated by Tesaglitazar and by GLP-1/Tesaglitazar, but not by treatment with GLP-1. Collectively, our data show that GLP-1/Tesaglitazar improves energy and glucose metabolism with superior ecacy to GLP-1 or Tesaglitazar alone and suggest that this conjugate holds therapeutic value to treat hyperglycemia and insulin resistance. diet- and genetically-induced obesity. The ability of GLP-1/Tesaglitazar to decrease body weight and to improve glucose control is absent in GLP-1R knockout (ko) mice and is preserved in DIO mice even at doses subthreshold for GLP-1 and Tesaglitazar alone to improve metabolism. In line with previous reports indicating that PPAR 𝛾 may act on hypothalamic neurocircuitries 5,6 , we identied a series of novel PPAR protein targets in the hypothalamus that are acutely upregulated by Tesaglitazar and by GLP-1/Tesaglitazar, but not by treatment with GLP-1. Collectively, our data identify a series of novel Tesaglitazar targets in the hypothalamus and show that GLP-1/Tesaglitazar improves body weight, food intake, and glucose metabolism with superior ecacy relative to treatment with GLP-1 or Tesaglitazar alone. Our data suggest that GLP-1/Tesaglitazar might hold therapeutic value to treat conditions characterized by hyperglycemia and insulin resistance.


Introduction
The peroxisome proliferator-activated receptors alpha and gamma (PPAR / ) play important roles in the regulation of energy, lipid and glucose metabolism 1 . PPARs are nuclear transcription factors which upon ligand activation heterodimerize with the 9-cis retinoic acid receptor (RXR) to promote target gene expression via binding to DNA response elements 2 . While selective activation of PPAR by bric acid derivates ( brates) primarily improves hepatic lipid and cholesterol metabolism 3 , the selective activation of PPAR by thiazolidinediones (TZDs) enhances insulin sensitivity in peripheral tissues such as the adipose tissue, liver and skeletal muscle 4 . Although predominantly expressed in the adipose tissue 4 , PPAR mRNA and immunoreactivity has also been detected in hypothalamic nuclei governing energy balance 5 , and third ventricle adenoviral-mediated overexpression of PPAR decreases food intake in lean and diet-induced obese (DIO) mice 6 . Inhibition of food intake by hypothalamic PPAR activation is, however, not undisputed since also viral-mediated overexpression of PPAR in the hypothalamus, or administration of the PPAR agonist Rosiglitazar into the third ventricle, has been shown to increase food intake and body weight in rats 7 . Also systemic activation of PPAR increases food intake and body weight in rodents 8 and neuron-speci c deletion of PPAR decreases body weight and food intake in mice fed a high-fat diet (HFD) 9 . Fibrates and TZDs have both shown bene cial cardiovascular effects in clinical trials 3,10,11 , which together with their complementary action on glucose and lipid metabolism spurred the development of PPAR / dual-agonists (Glitazars) for the treatment of type 2 diabetes (T2D) and dyslipidemia 1 . These agents effectively improved glucose and lipid metabolism in clinical trials 12,13 , however, the development of many Glitazars was terminated due to unfavorable effects on the cardiovascular and/or renal system 14 . Adverse effects of Glitazars may include signs of myopathy and muscle catabolism, uid retention, renal damage, weight gain, peripheral edema and early indications of an increased cardiovascular risk 15 .
Tesaglitazar is a PPAR / dual-agonist, which in phase II and III clinical trials improved glucose and lipid metabolism with greater e cacy relative to selective PPAR agonism 13,16−20 . Unfavorable effects of Tesaglitazar are dose-dependent and may include undesired weight gain, increased serum levels of creatinine, and decreased glomerular ltration 13,16,18,19 . In 2006, the development of Tesaglitazar was terminated based on an estimated bene t/risk pro le that was not expected to be superior to existing therapies.
Here we report the preclinical pharmacological evaluation of a single molecule conjugate of Tesaglitazar covalently attached to the incretin hormone glucagon-like peptide-1 (GLP-1). We hypothesized that the insulinotropic effect of GLP-1 would synergize with the insulin-sensitizing effect of Tesaglitazar to optimize glucose handling, while the body weight lowering effect of GLP-1 would buffer against the obesogenic nature of PPAR agonism. We further hypothesized that GLP-1-mediated delivery of Tesaglitazar would provide bene cial glucometabolic effects at doses subthreshold for each monotherapy, thereby improving systemic metabolism at more tolerable doses. In HEK293 cells, we show that GLP-1/Tesaglitazar does not differ from the matched GLP-1 backbone in terms of ligand-induced GLP-1 receptor (GLP-1R) activation, internalization or degradation. GLP-1/Tesaglitazar shows GLP-1Rdependent PPAR -RXR heterodimerization and enhanced in vivo e cacy to reduce body weight and food intake, and to improve glucose metabolism relative to treatment with GLP-1 or Tesaglitazar alone in mice with diet-and genetically-induced obesity. The ability of GLP-1/Tesaglitazar to decrease body weight and to improve glucose control is absent in GLP-1R knockout (ko) mice and is preserved in DIO mice even at doses subthreshold for GLP-1 and Tesaglitazar alone to improve metabolism. In line with previous reports indicating that PPAR may act on hypothalamic neurocircuitries 5,6 , we identi ed a series of novel PPAR protein targets in the hypothalamus that are acutely upregulated by Tesaglitazar and by GLP-1/Tesaglitazar, but not by treatment with GLP-1. Collectively, our data identify a series of novel Tesaglitazar targets in the hypothalamus and show that GLP-1/Tesaglitazar improves body weight, food intake, and glucose metabolism with superior e cacy relative to treatment with GLP-1 or Tesaglitazar alone. Our data suggest that GLP-1/Tesaglitazar might hold therapeutic value to treat conditions characterized by hyperglycemia and insulin resistance.

Animals
C57BL/6J mice and Lepr db (db/db) mice were purchased from the Jackson Laboratory. Glp1r -/mice were kindly provided by Daniel Drucker (University of Toronto, CA). All wt and ko mice used in our studies were bred on a C57BL/6J background. Mice were double-housed and kept in a constant environment with the ambient temperature set to 22 ± 2°C with constant humidity (45 -65%) and a 12h/12h light/dark cycle. For studies in DIO mice, male C57BL/6J mice were fed with a high-fat diet (HFD) comprising 58% kcal from fat (D12331; Research Diets, New Brunswick, NJ). db/db mice were fed with a regular chow diet throughout the study. At the beginning of each experiment, mice were equally distributed into experimental groups according to their body weight and body composition. All animal studies were approved by the State of Bavaria, Germany, or the IACUC of the University of Cincinnati, OH, USA. Compounds were dissolved in PBS and were subcutaneously administered with a volume of 5 l/g body weight in the indicated doses.

Body composition analysis:
Fat and lean tissue mass were measured via Nuclear Magnetic Resonance (NMR) technology (EchoMRI, Houston, TX, USA).

Glucose-and insulin tolerance tests (GTT/ITT):
Glucose tolerance was assessed in 6 h fasted mice after intraperitoneal (i.p) injection of 1.75 g glucose per kg body weight (DIO mice). Insulin tolerance was assessed in 6 h fasted mice after i.p. injection of either 0.75 U Insulin per gram body weight (DIO mice) or 1 U Insulin per gram body weight (db/db mice) (Humalog, Eli Lilly, Indianapolis, IN, USA). Tail vein blood glucose was subsequently measured using a handheld glucometer (TheraSense Freestyle) at baseline and after 15, 30, 60 and 120 min.

Plasma creatinine
Plasma creatinine was measured by a uorometric creatinine assay kit (Cat#: ab65340, Abcam, Cambridge, United Kingdom) based on the manufacturer's instruction.

Glucose-stimulated insulin secretion (GSIS) in primary murine islets
Isolation of islets was performed by collagenase P digestion of the wt adult pancreas. 3 mL of collagenase P (1 mg/mL) was injected into the bile duct, and the perfused pancreas was consequently dissected and placed into 3 mL collagenase P for 15 min at 37 °C. Then, 10 mL of G-solution (HBSS + 1% BSA) was added, followed by 2 min centrifugation at 1600 rpm at 4 °C. Pellets were re-suspended in 5.5 mL of gradient preparation (5 mL 10% RPMI + 3 mL 40% Optiprep/ per sample) and placed on top of 2.5 mL of the same solution. To form a 3-layers gradient, 6 mL of G-solution was added on the top. Samples were then subjected to 10 min centrifugation at 1700 rpm. Finally, the interphase between the upper and the middle layers of the gradient was harvested and ltered through a 70 μm Nylon lter. The eluent was washed with G-solution, and the islets were handpicked under the microscope. For GSIS analysis, the isolated islets were cultured overnight before being transferred to a 96-well plate containing modi ed Krebs Ringer phosphate Hepes (KRPH) buffer with 1 mM glucose for 1 h. Afterwards, islets were exposed to equal concentrations of the compounds (3 nM) at the same time when undergoing glucose concentrations of 16.8 mM (2 h each). Insulin was quanti ed from the supernatant. Finally, the islets were lysed for DNA extraction. Insulin concentrations were measured using an ultrasensitive insulin ELISA kit (Cristal Chem). The analysis was performed using a standard curve and the data were normalized to total content of DNA. . BRET measurements were taken every 30 sec -2 min using a PHERAstar FS multi-mode microplate reader. Baseline measurements were taken after an initial 5 min incubation with coelenterazine-h or NanoGlo-containing PBS after which cells were then treated with either vehicle (PBS) or respective ligands. The resulting ligand-speci c ratiometric BRET signals were normalized to vehicle producing the "ligand-induced BRET ratio" 21 , followed by an additional normalization step to well-speci c baseline readings. Ligand-induced measurements on the temporal scale is represented as the subsequent measurement after time point zero. Positive or negative incremental area under the curves (+iAUC/-iAUC) were calculated where noted. Each experiment was independently performed at least three times, each with at least three technical replicates for each group. Research, Nedlands, WA, Australia). Intracellular cAMP was measured using the YFP-Epac-Rluc CAMYEL sensor 23 . hGLP-1R-RLUC8 internalization was quanti ed using the intracellular plasma membrane marker Venus-KRAS 24 . Venus-KRAS was a kind gift from Kevin P eger. GLP-1R-Rluc8 lysosomal co-localization was measured using Lamp1-mNeonGreen. Lamp1-mNeonGreen was a gift from Dorus Gadella (Addgene plasmid # 98882). Time-dependent RXR/PPARg heterodimerization was measured using RXR-Rluc8 and PPARg2-YFP, both of which were kind gifts from Professor Vincent Ollendorf 25 .

Gene expression analysis
For assessment of acute drug effects, animals were treated with the respective compounds 4 hrs prior to tissue harvesting. Dissected tissues were frozen immediately on dry ice, and RNA was isolated using RNeasy Mini Kits (Qiagen). mRNA levels were determined using TaqMan probes in custom-made low density array cards (ThermoFisher) or in single assays on a QuantStudio 7 Real-Time PCR system. Target gene expression was normalized to HPRT and fold change was calculated relative to vehicle treated controls.

LC-MS/MS analysis
Single-shot measurements were performed with 500 ng of puri ed peptides, determined by absorbance at 280 nm on a NanoDrop 2000. Peptides were loaded onto a 50-cm column, packed in-house with 1.9 µm C18 ReproSil particles (Dr. Maisch GmbH) with an EASY-nLC 1200 system (Thermo Fisher Scienti c). Column temperature was kept at 60°C using a column oven. Peptides were eluted over 60 min using a binary buffer system consisting of buffer A (0.1% formic acid) and buffer B (80% ACN, 0.1% formic acid). In brief, the gradient started with 5% buffer B and increased stepwise to 45% over 45min, followed by a wash-out at 95% buffer B, all at a owrate of 300 nl/min. Peptides were then transferred to the gas phase using electrospray ionization (ESI), pre-ltered by a FAIMS device (CV -50 V) before entering the Orbitrap Exploris 480 (Thermo Fisher Scienti c) mass spectrometer. A data-independent (DIA) acquisition method was used, in which one full scan (300-1650 m/z, max. ion ll time of 45ms, normalized AGC target = 300%, R= 120.000 at 200 m/z) was followed by 66 tMS2 fragment scans of unequally-spaced windows with 1 Th overlap, covering the same m/z range ( ll time = 22 ms, normalized AGC target = 1000%, normalized HCD collision energy = 30%, R= 15.000).
GLP-1/ Tesaglitazar improves body weight, food intake, and glucose metabolism in DIO mice In DIO mice, 14-day treatment with GLP-1/Tesaglitazar (50 nmol/kg/day) decreased body weight and food intake with superior potency relative to treatment with equimolar doses of GLP-1 or Tesaglitazar (Figure 2a,b). The decreased body weight in GLP-1/Tesaglitazar-treated mice was associated with a decrease in body fat mass, with only a slight but signi cant decrease in lean tissue mass (Figure 2c,d). GLP-1/Tesaglitazar, but not GLP-1 or Tesaglitazar alone, decreased levels of fasted blood glucose ( Figure  2e). This was paralleled by a more signi cant decrease in levels of fasted insulin (Figure 2f), and improved insulin sensitivity relative to vehicle treated controls, as indicated by HOMA-IR ( Figure 2g) and by direct assessment of insulin tolerance (Figure 2h,i). In isolated murine islets, GLP-1/Tesaglitazar showed preserved but slightly decreased ability to stimulate insulin secretion under conditions of high glucose relative to GLP-1 (Figure 2j). These data align with the demonstration that GLP-1/Tesaglitazar is not superior to GLP-1 in either GLP-1R activation, internalization or degradation (Figure 1b- Tesaglitazar monotherapies in reducing body weight and food intake, and to improve glycemic control, and this is not accompanied by chances in markers indicative of liver function, (ALT, AST), kidney function (creatinine) or heart hypertrophy.

Single low-dose treatment with GLP-1/Tesaglitazar acutely improves glucose tolerance in DIO mice
Based on the insulin-sensitizing effect of GLP-1/Tesaglitazar in DIO mice (Figure 2h,i), we next assessed whether the acute glycemic bene ts prevail even at doses subthreshold for GLP-1 to improve glucose metabolism. Bolus peripheral treatment of DIO mice with Tesaglitazar at doses of 10 and 100 nmol/kg acutely worsened intraperitoneal glucose tolerance (Figure 3a,b). When given at a dose of 100 nmol/kg, GLP-1 and GLP-1/Tesaglitazar both improved glucose tolerance with only slightly enhanced e cacy of the conjugate relative to GLP-1 (Figure 3c

Low-dose GLP-1/Tesaglitazar chronically improves body weight and glucose handling in DIO mice
We next assessed whether daily low-dose chronic treatment with GLP-1/Tesaglitazar also affects energy and glucose tolerance in DIO mice. When given at a daily dose of 5 nmol/kg, 7-day treatment with GLP-1/Tesaglitazar decreased body weight and food intake in DIO mice, while mice treated with GLP-1 showed no difference in either body weight or food intake relative to vehicle controls (Figure 4a (Figure 6a). Consistent with the greater decrease in body weight, GLP-1/Tesaglitazar signi cantly reduced food intake relative to treatment with GLP-1 or Tesaglitazar alone ( Figure 6b) and this was paralleled by a greater decrease in fasted blood glucose (Figure 6c,d) and further improvement of insulin sensitivity (Figure 6e,f). Collectively, these data show that GLP-1/Tesaglitazar outperforms GLP-1 and Tesaglitazar to decrease body weight and improve glucose metabolism in mice with genetically-induced obesity and glucose intolerance. The observation that the metabolic effects of GLP-1/Tesaglitazar depend on functional GLP-1R in vivo align with our in vitro demonstration that GLP-1/Tesaglitazar fails to induce PPARγ/RXR heterodimerization in the absence of GLP-1R (Figure 1i-l).
LC/MS-based proteomics reveals novel PPAR-regulated targets engaged by GLP-1/Tesaglitazar in the hypothalamus Based on recent data indicating that PPAR might affect systemic energy metabolism via hypothalamic neurocircuitries 5,6 , and since the glycemic bene ts of GLP-1/Tesaglitazar are most pronounced after acute treatment (Figure 3g,h), we next assessed the response of the hypothalamic proteome of DIO mice after single peripheral drug treatment using LC/MS. In single-shot 60 min DIA analyses, we quanti ed more than 6,500 hypothalamic proteins (Figure 7a). High reproducibility was indicated by a median  (Figure 7c), which aligns with our data in isolated islets showing somewhat decreased insulinotropic action of GLP-1/Tesaglitazar relative to GLP-1 (Fig. 2j). However, we observed stronger Tesaglitazar-selective changes in the proteome by treatment with GLP-1/Tesaglitazar relative to Tesaglitazar alone (Figure 7c). These data con rm that GLP-1/Tesaglitazar engages PPAR targets in the hypothalamus and suggest that the known high expression of GLP-1R in the hypothalamus allows for a favorable delivery of GLP-1/Tesaglitazar into this tissue. The enriched amplitude of PPAR-induced signaling observed with the GLP-1/Tesaglitazar conjugate is potentially due to selective biodistribution of the Tesaglitazar to GLP-1R positive tissues, as opposed to systemic biodistribution achieved with unconjugated Tesaglitazar treatment that would otherwise dilute the signaling amplitude.

Discussion
We here report the development and molecular characterization of a GLP-1 peptide biochemically modi ed for GLP-1R-dependent delivery of the PPAR / dual-agonist Tesaglitazar. We demonstrate that GLP-1/Tesaglitazar is comparably e cacious as GLP-1 to promote GLP-1R-mediated Gα s recruitment, cAMP production, as well as internalization and degradation of the GLP-1 receptor in vitro, but is superior in decreasing body weight and improving glucose metabolism in vivo. Consistent with the in vitro data showing that GLP-1/Tesaglitazar is not simply a GLP-1 analog of enhanced potency, we see comparable, yet not superior, stimulation of insulin secretion by GLP-1/Tesaglitazar in isolated murine islets relative to GLP-1. Rather, GLP-1/Tesaglitazar demonstrates unique pharmacodynamic properties by inducing PPAR -RXR heterodimerization only in the presence of GLP-1R. This indicates that GLP-1/GLP1R interaction acts as a viable vector for Tesaglitazar intracellular delivery. This results in comparable e cacy relative to Tesaglitazar, yet with a notable delayed onset. This delayed onset of PPAR -RXR heterodimerization by GLP-1/Tesaglitazar relative to Tesaglitazar is consistent with the observed timedependent kinetics of ligand-induced GLP-1R internalization, and suggests that the rate of GLP-1Rmediated internalization of GLP-1/Tesaglitazar is critical to allow for an adequate induction of PPAR -RXR heterodimerization and hence to promote PPAR target effects. Consistent with this, we show lack of GLP-1/Tesaglitazar effects in the GLP-1R negative tissues in vivo, notably liver and skeletal muscle, as well as in GLP-1R ko mice. Consistent with a differential pharmacodynamic signature, treatment with GLP-1/Tesaglitazar led to greater reduction in body weight and food intake in DIO and db/db mice compared to treatment with GLP-1 or Tesaglitazar, whereas the glucoregulatory and weight loss actions were eliminated in GLP-1R de cient mice. Interestingly, GLP-1/Tesaglitazar was exceptionally potent to acutely improve glucose tolerance, particularly at very low doses subthreshold for either GLP-1 or Tesaglitazar alone. Our data in isolated islets suggest that this enhanced effectiveness is not mediated by enhanced insulin secretion.
Although this will require further testing, it is possible that the enhanced glycemic effects of the conjugate may be mediated by glucoregulatory organs that receive feedback from the CNS, and speci cally the hypothalamus. The aforementioned possibility would be in accordance with the differential hypothalamic proteomic signatures observed with GLP-1/Tesaglitazar relative to GLP-1 alone. By using LC/MS proteomic analysis, we identi ed the hypothalamus as a target of GLP-1/Tesaglitazar, consistent with the known distribution pattern of the GLP-1 receptor in mice. This is consistent with the observation that the hypothalamus and the hindbrain are primary targets of uorescently-labeled GLP-1R agonists 28,29 and aligns with the demonstration of high GLP-1R abundance in these brain regions 30,31 . PPARy is expressed in various nuclei of the hypothalamus but not in the nucleus tractus solitarius (NTS) of the brainstem 5 and targeted overexpression of PPARy in the hypothalamus decreases food intake in lean and DIO mice 6 .
While the mechanistic underpinnings underlying food intake regulation by PPARy remain largely unknown, in a hypothalamic cell line, stimulation of POMC expression by bisphenol A (BPA) is blocked by pretreatment with the PPARy antagonist T0070907 32 . In summary, there is accumulating evidence indicating that PPARy signaling engages hypothalamic neurocircuitries to control food intake, yet much more work is required to delineate precise mechanisms. he discussed potential projects with and has signed/signs contracts for his institute(s) and for the staff for research funding and/or collaborations with industry and academia, worldwide, including but not limited to pharmaceutical corporations like Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Medigene, Arbormed, BioSyngen, and others. In this role, he was/is further responsible for commercial technology transfer activities of his institute(s), including diabetes related patent portfolios of Helmholtz Zentrum München as, e.g., WO/2016/188932 A2 or WO/2017/194499 A1. MHT con rms that to the best of his knowledge none of the above funding sources were involved in the preparation of this paper. TDM and K.S. receive research funding by Novo Nordisk. DJD has received speaking and consulting fees from Eli     GLP-1/Tesaglitazar fails to affect body weight and glucose metabolism in obese GLP-1R ko mice. Body weight (a,g), change in fat and lean tissue mass (b,c, h,i), cumulative food intake (d,j), blood glucose (e,k) and intraperitoneal glucose tolerance (f,l) in 36-wk old male diet-induced obese wildtype (a-f) or GLP-1R ko mice (g-l) treated for 6-days with either Vehicle or 50 nmol/kg/day of GLP-1/Tesaglitazar. N=7-8 mice each group. Data in panel a, d, f, g, j, and l have been analyzed by 2-way ANOVA with Bonferroni post-hoc comparison for individual time points. Data in panel b, c, e, h, i and k have been analyzed by 1-way ANOVA using the Bonferroni's multiple comparison test. Data represent means ± SEM; asterisks indicate * p<0.05, ** p<0.01, *** p<0.001. Food intake (d,j) was assessed per cage in double-housed mice.