PPARγ agonists delay age‐associated metabolic disease and extend longevity

2.2. Rosiglitazone reduces inflammation, fibrosis, and atrophy in aging tissues

Given that the aging process is accompanied by elevated inflammation in tissues such as fat and liver and by increased sarcopenia and muscle atrophy (Dalle et al., 2017; Kalyani et al., 2014), we performed gross pathology and histological analysis to assess the effects of chronic Rosi treatment started when mice reach 14 months of age. Immuno‐histochemical analysis using Sirius Red staining and an F4/80 antibody of WAT revealed, respectively, diminished fibrosis and decreased macrophage infiltration in the Rosi‐treated group (Figure 2a,b). Furthermore, treated mice showed decreased hepatic inflammation (Figure ​(Figure2c)2c) and diminished islet degeneration (Figure ​(Figure2d2d).

Open in a separate window

FIGURE 2

Rosiglitazone treatment in aging mice reduces inflammation, increases oxidative fiber number, and mitochondrial area in muscle. (a–i) Analysis of 20‐month‐old mice after 6 months of treatment, either with control diet or diet supplemented with Rosi (1 mg/kg of mouse weight/day) (n = 6). Representative image of Sirius Red staining (a) and F4/80 staining (b) of eWAT and iWAT with quantification. (c) Representative H&E images and F4/80 staining of liver with quantification. (d) Representative H&E images of pancreas and quantification of islet area. (e) Cross‐sectional area (CSA) of fiber of gastrocnemius (Gas) muscle by Succinic dehydrogenase (SDH) staining and ratio of oxidative fibers. (f) Representative images of gastrocnemius muscle by transmission electron microscopy and quantification of mitochondrial area. Data are presented as mean ± SEM and Student t tests were performed to compare control and Rosi‐treated group. (*p < 0.05, **p < 0.01)

Given the known role of Rosi in regulating mitochondrial function in muscle (Rong et al., 2007), we compared muscle fiber size and number and mitochondrial morphology of muscle obtained from control and treated mice. As shown in Figure ​Figure2e,2e, SDH staining revealed increased oxidative fibers and electron microscopy analysis showed elevation in mitochondrial density in muscle of Rosi‐treated mice (Figure ​(Figure2f).2f). Overall, these results suggest that Rosi prevents age‐associated tissue inflammation, fibrosis, and atrophy in a number of metabolic tissues.

2.3. Rosiglitazone treatment relieves anxiety‐ and depression‐like symptoms and improves cognitive functionality

It has been previously reported that increased locomotor activity in mice may be a sign of anxiety‐ and depression‐like conditions (Seibenhener & Wooten, 2015). Given that Rosi‐treated mice showed reduced activity (Figure ​(Figure1i),1i), we assessed whether control and treated mice displayed differences in anxiety‐ and depression‐like behaviors. Behavioral studies revealed that 14‐month‐old mice receiving Rosi spent more time in the center area of the open field maze and in the open arm of the elevated zero maze compared to controls (Figure 3a,b), suggesting decreased anxiety‐like symptoms. Given the known association between stress and attenuated sucrose preference (Wang et al., 2018), we performed sucrose preference tests and demonstrated that Rosi‐treated mice consume more sucrose after exposure to stress than control mice (Figure ​(Figure3c),3c), suggesting a decrease in depression‐like symptoms in treated mice. The aging process is known to be accompanied by neurodegeneration and brain deterioration (Dillin et al., 2014); we therefore determined possible changes in learning abilities and memory in control and treated mice. Our data demonstrated that Rosi‐treated mice had improved cognitive function, as shown by their increased alternations in the T‐maze, and decreased escape latency in the water maze (Figure 3d,e).

Open in a separate window

FIGURE 3

Rosiglitazone treatment is associated with decreased anxiety‐like behavior and depression‐like symptoms and improved cognitive function. (a–e) Analysis of 20‐month‐old mice after they were treated for 6 months either, with control or Rosi at 1 mg/kg/day (n = 6). (a) Open field analysis indicating time and distance. (b) Elevated O maze showing the length of time mice stay in the open arm and the distance. (c) Sucrose preference test. (d) Total alternation and alternation rate in the T‐maze test. (e) Water maze test. Data are presented as mean ± SEM. Student t tests were performed to compare control‐ and Rosi‐treated group (* p < 0.05, ** p < 0.01)

2.4. Rosiglitazone reprograms the expression profile of genes involved in inflammation and mitochondrial function in aging metabolic tissues

To better understand the molecular mechanisms underlying Rosi's action in metabolic tissues, we performed unbiased analysis of gene expression in adipose depots, liver, muscle, and brain. These analyses revealed that mice undergoing Rosi treatment have the highest percentage of gene expression changes in eWAT compared to the other tissues analyzed (Figure ​(Figure4a4a and Table S2). Analysis of average changes in gene expression demonstrated that factors involved in inflammatory responses, genomic stability, and mitochondrial functionality are the most differentially regulated genes in response to Rosi treatment in each of the tissues analyzed (Figure ​(Figure4b).4b). Specifically, Rosi‐treated mice showed decreased levels of inflammatory genes and increased expression of genes encoding for factors involved in mitochondrial functionality and genome stability. Furthermore, mitochondrial gene expression changes appeared to be WAT tissue‐selective, since minimal effects were observed in BAT, liver, and brain (Figure ​(Figure4c).4c). Validation of the differential expression of select genes confirmed the results obtained via the arrays (Figure ​(Figure4d).4d). Overall, our molecular analysis suggests that Rosi treatment is associated with decreased age‐associated dysfunction, possibly through Rosi effects in reducing inflammation in metabolic tissues and in preserving mitochondrial functionality.

Open in a separate window

FIGURE 4

Chronic treatment with low doses of rosiglitazone alters the levels of expression of genes involved in inflammatory responses and in mitochondrial functionality. (a) Percentage of differentially expressed genes (DEGs) in adipose tissues, liver, muscle and brain in 20‐month‐old mice after they were treated for 6 months. (b, c) Distribution of modulated genes in different pathways and tissues. (d) Expression levels of representative genes of each category in multiple organs of 20‐month‐old mice after 6 months of treatment. Data are presented as mean ± SEM and Student t tests were performed to compare control and Rosi‐treated group (*p < 0.05, **p < 0.01, n = 6 per group)

2.5. Effects of long‐term low‐dose rosiglitazone treatment on cardiac function and skeletal system

It has been previously reported that Rosi treatment of human subjects increases myocardial infarction, water retention, and bone fractures (Nissen & Wolski, 2007). To potentially reduce the incidence of issues arising from the long‐term use of Rosi, we exposed mice to lower doses than those previously used (Cellai et al., 2010). Comprehensive serum analysis revealed no significant differences except in glucose levels, which appeared to be significantly lower in Rosi‐treated mice (Table ​(Table1).1). Furthermore, as shown in Figure ​Figure5a,5a, no significant differences in water retention between treated mice and controls were revealed by Echo MRI analysis. In vivo echocardiogram showed no cardiac abnormalities (Table ​(Table2),2), and measurements of heart weight and WGA staining of heart biopsies (Figure 5b–g) revealed no differences in heart volume and in left ventricular ejection fraction between groups, although a minor trend to an increase in these parameters was noticed. To assess adverse effects of Rosi also on the bone, we performed micro‐computed tomography and demonstrated that Rosi‐treated mice have comparable amounts of trabecular and cortical bone to controls (Figure 5h–l). These results suggest that chronic treatment of Rosi at low doses has limited adverse effects.

Open in a separate window

FIGURE 5

Limited adverse effects of long‐term rosiglitazone treatment at the low dose of 1 mg/kg/day on water retention, heart functionality, and bone density. (a–f) Analysis of 17‐month‐old mice after they were treated for 3 months with either control diet or Rosi‐containing diet (1 mg/kg/day) (n = 6). (a) Water content. (b, c) Heart weight and representative images of Wheat Germ Agglutinin (WGA) staining of heart. (d–g) Measurements of heart functionality include hemodynamics, left ventricular (LV) structure, LV function and LV outflow tract. (h) Representative images of femurs by X rays, (i) comparison of femur length, (j) measurements of X‐ray density, (k) micro‐CT reconstruction, and (l) parameters of trabecular and cortical bone. Data are presented as mean ± SEM

2.6. Long‐term treatment with rosiglitazone extends lifespan in mice

Given the effects of Rosi treatment on amelioration of the metabolic and cognitive functions we observed, we next determined whether exposure to Rosi may affect lifespan, even if the treatment is started in mice when they are 14 months of age and therefore have potentially already undergone some of the metabolic changes associated with aging. Our data demonstrate that chronic treatment with 1 mg/kg/day of Rosi started when the mice are 14 months old, extended median lifespan by 11% (Figure ​(Figure6a).6a). The median survival (864 days) was significantly increased in treated mice compared to the control group (777 days). Age is a major risk factor for the development of cancer; therefore, we evaluated the effects of Rosi treatment on the natural development of tumors. Rosi has been shown to have anti‐tumorigenic properties in lung, breast, and colon cancer through inhibition of cell growth and invasion and via induction of apoptosis in cancer cell lines (Fenner & Elstner, 2005). Our analysis of 50 pathology reports of mice in the longevity cohort determined by necropsy at the time of death revealed a trend in delayed development of spontaneous tumors, in Rosi‐treated mice (Figure ​(Figure6b6b and Figure S5). These data suggest that Rosi treatment is associated with increased lifespan and that the anti‐cancer properties of Rosi may not contribute to the effects of Rosi on longevity and healthspan.

Open in a separate window

FIGURE 6

Chronic treatment with low doses of rosiglitazone (1 mg/kg) of 14 months of age male mice is associated with increased lifespan with no alterations in tumor incidence and survival on pioglitazone in comparison to glimepiride. (a) Longevity curve, (b) tumor incidence at necropsy in control and Rosi‐treated mice. Curves were plotted with the Kaplan–Meier method and compared using the log‐rank test (control group, n = 31 and Rosi‐treated group, n = 34). (c) Survival curve of patients of the Veterans Health Administration on pioglitazone or glimepiride

4.2. Lifespan and cancer incidence analysis

Lifespan monitoring of male mice was performed as previously described (Ma et al., 2014). Briefly, 14‐month‐old male mice used to study longevity and cancer incidence were kept up to five per cage and were not used for any other metabolic or physiological analyses. The date of their death was recorded as the day that either the mice were found dead in the cage or when euthanasia was performed based on the veterinary technician recommendations due to severe discomfort. Survival curves were plotted with Kaplan–Meier methods, and statistical analysis was performed with log‐rank test by GraphPad. Pathological analysis of malignant tumors was performed by a veterinary pathologist of the Division of Veterinary Resources of the NIH on dissected tissues (see Section 4.4 below) and Fisher's exact test was used to compare control or Rosi‐treated mice.

4.3. Primary cells preparation and measurement of D‐[¹⁴C]‐glucose uptake

Mice were anesthetized via an intraperitoneal injection of pentobarbital sodium (65 mg/kg body weight), and muscle (gastrocnemius) and adipose tissues (inguinal and visceral) were excised. Skeletal muscle cells and adipocytes were isolated as previously described with some deifications. Briefly, excised tissues were washed twice in Hank's Buffered Salt Solution (HBSS) (Invitrogen) and then minced into a coarse slurry. The minced tissues were then digested with either a Ca‐free 0.25% trypsin solution containing EDTA (Gibco) for 20 min (for skeletal muscle cells) or with a Krebs‐Ringer bicarbonate HEPES (KRBH) buffer with 0.1% collagenase‐type I (Gibco) for 40 min (for adipocytes) in a 37°C water bath with gentle agitation. The digestion solution was filtered through a 100 μm nylon mesh (Falcon) and centrifuged for 5 min at 800 g. Cell pellets were then resuspended in Dulbecco's modified Eagle's medium (DMEM) (Invitrogen) and incubated for 30 min with gentle agitation. Glucose uptake assay was performed according to previously described protocols. Briefly, cells were washed twice in KRBH buffer containing 0.1% BSA and then resuspended in the wash buffer at 20% cytocrit. The cell suspension was equilibrated for 10 min at 37°C with gentle agitation and then stimulated with insulin (100 nM) for 15 min. Next, the ¹⁴C‐labeled glucose medium consisting of 5 mM D‐glucose (containing 2.5 μCi/ml D‐[14C] glucose (Perkin Elmer) was added and incubated for 5 min at 37°C. The uptake was terminated by adding cytochalasin B (Solarbio) and washed three times with KRBH buffer containing 0.1% BSA. Finally, the cell‐associated radioactivity was measured using the scintillation cocktail medium Ultima Gold XR (Perkin Elmer) in a Tri‐Carb 4910TR Liquid Scintillation Analyzer (Perkin Elmer). The non‐specific signal was determined in control samples containing cytochalasin B without¹⁴C‐labeled glucose medium. Glucose uptake was normalized for DNA content.

4.4. Histological analysis

Tissues dissected from control‐ or Rosi‐treated mice were fixed in 10% neutral‐buffered formalin (Sigma) and embedded in paraffin. Tissue sections of 5 μm thickness were stained with hematoxylin and eosin (Histoserv). Frozen sections of gastrocnemius muscle were stained with succinic dehydrogenase (SDH) for oxidative fiber assessment (Histoserv) and wheat germ agglutinin (WGA) (Histoserv) for fiber quantification, respectively. Slides were analyzed using a microscope (Olympus) at the indicated magnification and images were captured by a digital camera (Olympus). The sizes of adipocytes were determined with ImageJ software. Specifically, for quantification, five random fields of H&E stained eWAT and iWAT of 6 mice per group were quantified and analyzed. Necropsies were performed by the NIH DVR Department at the NIH by removing individual organs, following the pathology procedure described above.

4.5. Transmission Electron Microscopy (TEM) analysis

For TEM analysis, dissected gastrocnemius muscle was cut into pieces of 1 mm size and placed in fixing solution containing 0.1 M phosphate buffer, 2% paraformaldehyde, and 2.5% glutaraldehyde (EM facility, National Cancer Institute, NCI). Fixed muscle tissue was subsequently washed and stored in 0.1 M cacodylate buffer. TEM was imaged in a Hiachi 7650 TEM operating at 80 kV. Images were taken with a digital camera from Advanced Microscopy Techniques and quantification analysis was performed with Image J (version 1.44; NIH).

4.6. 3‐Dimensional Micro‐computed Tomography (CT) analysis

For Micro‐CT, the femur bone taken from male mice either control‐ or Rosi‐treated for 3 months was fixed in 10% neutral‐buffered formalin (Sigma) and scanned at a resolution of 9 µm pixels. The region analyzed was below the growth plate, from 720 μm (80 image slices) to 1620 μm (180 image slices). Trabecular and cortical bone parameters of the femur were calculated using CTan software (Bruker micro‐CT), and 3D reconstruction images were generated with software CTvol (Bruker micro‐CT).

4.7. Gene expression analysis

Total RNA was extracted with TRIzol (Invitrogen), according to procedures described previously (Verma et al., 2020), from iWAT, eWAT, and BAT, liver, brain, and muscle tissues of 17‐month‐old mice, after three months of treatment with either a control diet or a diet containing Rosi. RNA was reverse transcribed into cDNA (First Strand cDNA Synthesis Kit, Roche), according to the manufacturers’ instructions. For unbiased gene expression analysis of tissues obtained from control‐ and Rosi‐treated mice, we used Mouse RT² Profiler™ PCR Array for Aging (Qiagen, PAMM‐178Z) for ABI PRISM 7900HT sequence detection system (ABI), using SYBR green (Roche). For validation of array studies and targeted analysis, we used the following primers: S100a8, Forward: AAATCACCATGCCCTCTACAAG; Reverse: CCCACTTTTATCACCATCGCAA; Tfb1m, Forward: CGGGAGATC ATTAAGTTGTTCGG; Reverse: GCCCAGGA CCCACTTCATAAA; Txnip, Forward: TCTTTTGAGGTGGTCTTCAACG; Reverse: GCTTTGACTC GGG TAACTTCACA; Terf1, Forward: CAGCAGTCTACAGA AACTGAACC; Reverse: ACTGAAATCTGATGGAGCACG; UCP1, Forward: GGCCCTT GTAAACAACAAAATAC; Reverse: GGCAACAAGAGCTGACAG TAAAT; Dio2, Forward AATTATGCCTCGGAGAAGACCG; Reverse: GGCAG TTGCC TAGTGAAAGGT.

4.8. Triglycerides analysis

100 mg of tissue was homogenized in phosphate‐buffered saline and lipids were extracted using a chloroform:methanol mix (2:1) and 0.1% sulfuric acid. The aliquot of the organic phase was collected, dried, and resuspended. Triglycerides content was then determined using commercially available kits (Sigma), following the manufacturer's instructions and data normalized to the respective tissue weight.

4.9. Retrospective survival analysis

We performed a retrospective survival analysis based on de‐identified electronic medical record data from the United States Veterans Health Administration (VHA) corporate data warehouse made available to us as part of the VA Obesity Medication Use Study (MIRB #1733), approved by the Research & Development committee and the Institutional Review Board at the Manhattan campus of the VHA New York Harbor Healthcare System. We focused on a comparison between subjects receiving Pio and the Sulfonylurea glimepiride, using an active comparator/new user approach. Of the TZD drugs used to treat insulin resistance, we chose to analyze the effects of Pio, given that this compound is the only PPARγ agonist that has been used without special restrictions over the course of the last 20 years in the United States. We selected glimepiride as a comparator because is an insulin secretagogue with a distinct mode of action from PPARγ agonists and because, similarly to pioglitazone, it is used almost exclusively for the treatment of type 2 diabetes. We queried the database to identify prescriptions for the medications of interest documented in the pharmacy records as released to the patients and not returned to stock. For each patient, we estimated the treatment start date to be the earliest day of release of the medication by the pharmacy. We queried the same database for dates of death. When not documented, we right‐censored survival status on the last date of vital sign documentation, therefore considering patients as lost to follow‐up after that date. In total, we identified 190590 subjects who were prescribed Pio and 47987 who were prescribed glimepiride (Table S3). We used non‐parametric approaches to estimate and compare survival after medication start date, without attempting to adjust for other factors, by means of Kaplan–Meier statistics and log‐rank testing, using R, a language and environment for statistical computing, version 3.6.1.

The studies were supported by the NIDDK Intramural Research Program of the NIH (to E.M.), by funding from the National Natural Science Foundation of China (31770840 to X.M and 31800989 to L.X), National Key Research and Development Program of China (2018YFC1313803 to X.M and 2019YFA0904500 to L.X.), by an NIH NHLBI institutional training grant (T32HL98129 to J.P.), by AHA 17‐SFRN33520045, NIH K08‐DK117064 and the Doris Duke Foundation (to JOA), by an NIH/NIAMS and NIH/NIA grant (K08AR069099, R01AG056169 to P.L), and by start‐up funds of New York University (to E.M.). We acknowledge that retrospective de‐identified data collection and analysis through the VA Informatics and Computation Infrastructure is approved by the VA NY Harbor Healthcare System IRB under protocol 07133. The authors have declared no conflict of interest.