C57BL/6J mice were purchased from Japan SLC (Shizuoka, Japan). PAI-1 KO mice (background strain C57BL/6J; strain name B6.129S2-Serpine1tm1Mlg/J; stock no. 002507) were obtained from Jackson Laboratory (Bar Harbor, ME, USA). Mice were housed under controlled temperature and humidity conditions on a 12-h light, 12-h dark cycle. The animals were housed individually and given standard laboratory diet (65% carbohydrate, 4% fat, 24% protein) and water ad libitum unless otherwise noted. All animal studies were conducted in accordance with the institutional guidelines for animal experiments at Tohoku University.

Two types of HFD were used in this study. A safflower oil-rich HFD (Yamada et al., 2006) (32% safflower oil, 33.1% casein, 17.6% sucrose, and 5.6% cellulose) was purchased from Oriental Yeast (Tokyo, Japan) and a lard-rich D12492 HFD (31.7% lard, 25.8% casein, 8.9% sucrose, and 6.5% cellulose) was purchased from Research Diet (New Brunswick, NJ, USA). Both of these HFD were purchased in powder form and then shaped into pellets in our laboratory.

The inhibitory activity and specificity of TM5441 were previously described in detail (Boe et al., 2013). TM5441 was mixed with HFD powder, and then shaped into pellets. Given that mice eat approximately 4 g of HFD per day (Figure 1E), 108 mg of TM5441 per 900 g of HFD were mixed together when the mice were approximately 30 g each. This enabled the mice to consume approximately 15 mg of TM5441 per kg body weight per day. The TM5441 dosage was set at 15 mg/kg of body weight because TM5441 treatment at a dose of 20 mg/kg of body weight has been shown to suppress HFD-induced body weight gain without exerting adverse effects (Piao et al., 2016). Control mice were fed HFD pellets without TM5441.

Blood samples, 30 µl per mouse, were collected by incision of the tail vein with a lance. Blood was collected into 0.5 ml Eppendorf tubes containing 0.2 μl of 0.5 M ethylenediaminetetraacetic acid (EDTA) and then centrifuged for 15 min at 8,000 rpm at 4°C to obtain plasma. Plasma leptin levels were determined with enzyme linked immunosorbent assay (ELISA) kits (Morinaga Institute of Biological Science, Yokohama, Japan) (Chiba et al., 2016). Plasma, 5 μl per mouse, was used for measuring leptin by ELISA. None of the samples required dilution for the assay. The detection range of the kit was 0.4–25.6 ng/ml.

Blood glucose levels were assayed using a Glutestmint kit (Sanwa Kagaku Kenkyusho, Nagoya, Japan) (Uno et al., 2006). Plasma insulin levels were determined with ELISA kits (Morinaga Institute of Biological Science, Yokohama, Japan) (Uno et al., 2006). Plasma, 5 μl per mouse, was used for measuring insulin by ELISA. None of the samples required dilution for the assay. The detection range of the kit was 0.156–10 ng/ml. Mice that had been fasted for 9 h during daylight hours were given glucose at a dose of 2 g/kg body weight (Yamada et al., 2006). Blood glucose and plasma insulin were measured both immediately before administration and then again 15, 30, 60, 90, and 120 min after glucose loading.

Spontaneous locomotor activities of mice were analyzed with an infrared activity monitor (Supermex; Muromachi Kikai, Tokyo, Japan) as previously reported (Tsukita et al., 2012).

Livers and white adipose tissues (WAT) were removed, fixed with 10% formalin and embedded in paraffin (Tsukita et al., 2012). Tissue sections were stained with hematoxylin eosin (HE). Adipocyte size was measured by the diameters of the adipocytes in light-microscopy images (20×) (n = 50 adipocytes per section, two sections per mouse) and analyzed using Keyence BZ-X710 Analyzer software. Experiments were carried out in a blinded manner by masking the slide number and then conducting the evaluation in a random order (Drareni et al., 2018).

Leptin loading tests were carried out as previously described (Tsukita et al., 2012) with slight modifications. To examine the effects of leptin treatment on body weight change, mice were given daily intraperitoneal administrations of leptin (5 mg/kg of body weight; R&D Systems, Minneapolis, MN, USA) or saline for 5 consecutive days. The leptin dosage was set at 5 mg/kg of body weight, based on a previous report (Kabra et al., 2016). To examine the effects of leptin treatment on hypothalamic neuropeptide expressions, 24-h fasted mice were given a single injection of either leptin (5 mg/kg of body weight) or saline, and were then sacrificed by decapitation 4 h after the injection. Brains were immediately frozen in isopentane with dry ice and stored at −80°C until RNA purification.

Coronal cryostat sections (80 μm) of the hypothalamic arcuate nucleus (ARC) were placed on PEN-coated slides (Leica Microsystems, Wetzlar, Germany) (Tsukita et al., 2012). Laser microdissection was carried out on a Leica AS LMD (Leica Microsystems, Wetzlar, Germany) (Asai et al., 2017). Immediately after microdissection, total RNA was purified using an RNeasy Micro Kit (Qiagen, Valencia, CA, USA) (Tsukita et al., 2017).

Complementary DNA was synthesized from 1 μg of total RNA using a QuantiTect Reverse Transcription Kit (Qiagen) and was evaluated using a real-time PCR quantitative system (Light Cycler Quick System 350S; Roche Diagnostics, Mannheim, Germany) (Takahashi et al., 2013). mRNA expression levels were normalized against the levels of glyceraldehyde 3-phosphate dehydrogenase (gapdh) for the hypothalamic ARC and beta-actin for brown adipose tissue. The sequences of the forward and reverse primers used were as follows: agouti-related peptide (agrp), forward, 5′-CCAGAGTTCCCAGGTCTAAG-3′; agrp, reverse, 5′-CGGTTCTGTGGATCTAGCA-3′; neuropeptide y (npy), forward, 5′-GCTTGAAGACCCTTCCATTGGTG-3′; npy, reverse, 5′-GGCGGAGTCCAGCCTAGTGG-3′; proopiomelanocortin (Pomc), forward, 5′-TCACCACGGAGAGCAACCT-3′; Pomc, reverse, 5′-CAGCGGAAGTGACCCATGA-3′; gapdh, forward, 5′-TGAAGGTCGGTGTGAACG-3′; gapdh, reverse, 5′-CCATTCTCGGCCTTGACT-3′; uncoupling protein 1 (ucp1), forward, 5′-TACCAAGCTGTGCGATGT-3′; ucp1, reverse, 5′-AAGCCCAATGATGTTCAGT-3′; peroxisome proliferative activated receptor gamma coactivator 1 alpha (pgc-1a), forward, 5′-ATACCGCAAAGAGCACGAGAAG-3′; pgc-1a, reverse, 5′-CTCAAGAGCAGCGAAAGCGTCACAG-3′; type II iodothyronine deiodinase (dio2), forward, 5′-ACTCGGTCATTCTGCTCAA-3′; dio2, reverse, 5′- AAGCCCAATGATGTTCAGT-3′; cell death-inducing DNA fragmentation factor alpha subunit-like effector A (cidea), forward, 5′-ACTTCCTCGGCTGTCTCAATGT-3′; cidea, reverse, 5′- TATCGCCCAGTACTCGGAGCAT-3′; PR domain containing 16 (prdm16), forward, 5′-TGTACAGGCAGGCTAAGAA-3′; prdm16, reverse, 5′-CCTGTGATGTTCAAGGAA-3′; beta-actin, forward, 5′-GATGCCCTGAGGCTCTT-3′; and beta-actin, reverse, 5′-TGTGTTGGCATAGAGGTCTTTAC-3′.

All values are expressed as mean ± SE. Data were compared employing Student’s t test to examine the effect of one factor, two-way ANOVA followed by Tukey’s test to examine the effects of two factors, or two-way repeat measures ANOVA to determine variance with respect to time and mouse group. The level of significance was set at P < 0.05.

We first examined the effects of PAI-1 deficiency on body weight and food intake under normal chow feeding conditions (Experimental procedure 1). Eight-week-old PAI-1 KO mice (chow-KO mice) and wild type (WT) control mice (chow-WT mice) were fed the normal chow diet. Daily food intake from the 1st day (Figure 1A) of observation was similar in these two groups. Body weight of the two groups were similar throughout the observation period (Figure 1B), as expected from a previous report (Ma et al., 2004). Daily food intake from the 71st day (Figure 1C) of observation was also similar in these two groups. Therefore, PAI-1 deficiency itself appears to have almost no effects on either eating habits or body weight alterations under normal chow feeding conditions.

We next examined the effects of PAI-1 deficiency on body weight and food intake under HFD-fed conditions. Safflower oil-rich HFD was given to 8-week-old PAI-1 KO mice (HFD-KO mice) and WT mice (HFD-WT mice) (Experimental procedure 2). The HFD-KO mice gained less weight than the HFD-WT mice (Two-way repeat measures ANOVA; F = 4.53, P < 0.001, HFD-WT mice 40.3 ± 1.68 g vs HFD-KO mice 34.6 ± 1.84 g, P = 0.039), and the difference between the two groups became statistically significant after 106 days (Figure 1D). Strikingly, HFD-KO mice ate significantly less than HFD-WT mice before their body weights showed significant differences (Two-way repeat measures ANOVA; F = 35.89, P < 0.001) (Figure 1E) and the average food intake amounts during the experimental period (21 days) were also significantly smaller in HFD-KO mice than in HFD-WT mice (HFD-WT mice 3.98 ± 0.08 g/day vs HFD-KO mice 3.73 ± 0.07 g/day, P = 0.021) (Figure 1F). Since reduced food intake preceded the evident differences in body weight, reduced energy intake is likely to be a direct cause, rather than a result, of suppressed weight gain. Thus, PAI-1 deficiency may suppress HFD-induced weight gain, while exerting minimal effects on weight regulation under normal chow-fed conditions.

We next performed oral glucose tolerance tests on HFD-KO mice and HFD-WT mice after 22 days of HFD feeding. Blood glucose levels of HFD-KO mice tended to be decreased as compared to those of HFD-WT mice (Figure 1G). Fasting plasma insulin levels were significantly lower in HFD-KO mice despite fasting blood glucose levels being similar (Two-way repeat measures ANOVA; F = 5.17, P = 0.036, HFD-WT mice 0.69 ± 0.03 ng/ml vs HFD-KO mice 0.52 ± 0.04 ng/ml, P = 0.005) (Figure 1H), indicating that these mice had better insulin sensitivity. In addition, plasma insulin levels 15 min after glucose loading were significantly higher in HFD-KO mice than in HFD-WT mice (HFD-WT mice 1.33 ± 0.09 ng/ml vs HFD-KO mice 1.65 ± 0.11 ng/ml, P = 0.048). Taken together, these results suggest that suppression of body weight gain in PAI-1 KO mice improves both insulin sensitivity and glucose-stimulated secretion of insulin.

We then examined the effects of TM5441, a selective PAI-1 inhibitor, on body weight in HFD-fed mice. In the following experiments, we switched from safflower oil-rich HFD to lard-rich HFD (Experimental procedure 3), because mice fed the latter gained more weight than those given the former (Two-way repeat measures ANOVA; F = 12.38, P < 0.001, lard-rich HFD mice 41.5 ± 1.21 g vs safflower oil-rich HFD mice 34.1 ± 0.98 g on day 49, P = 0.003) (Figure 2A). Ten-week-old WT mice that had been pre-fed a lard-rich HFD for 2 weeks were divided into two groups. The first group was then fed HFD mixed with TM5441 (HFD-WT-TM5441 mice), while the second group continued to eat the HFD without TM5441 (HFD-WT-Control mice) (Experimental procedure 5). No death, hair loss, signs of respiratory infections such as open-mouth breathing, or other adverse effects were observed in any of the TM5441 treated mice. Body weight gain was slower in mice that ate HFD mixed with TM5441, and these reductions reached statistical significance 35 days after the initiation of inhibitor treatment (Two-way repeat measures ANOVA; F = 5.24, P < 0.001, HFD-WT-Control mice 37.6 ± 1.07 g vs HFD-WT-TM5441 mice 33.8 ± 0.97 g on day 35, P = 0.017) (Figure 2B). Liver weights on day 7 were similar in the two groups (Figure 2C), and there was no evidence of either hepatocyte ballooning or lobular inflammation in the two groups (Figure 2D). On the other hand, epididymal white adipose tissue (WAT) weight on day 7 was significantly lower in HFD-WT-TM5441 mice than in HFD-WT-Control mice (HFD-WT-Control mice 0.98 ± 0.08 g vs HFD-WT-TM5441 mice 0.63 ± 0.06 g, P = 0.003) (Figure 2E). On average, adipocytes were markedly smaller in HFD-WT-TM5441 mice (HFD-WT-Control mice 71.05 ± 1.49 μm vs HFD-WT-TM5441 mice 60.35 ± 2.89 μm, P = 0.003) (Figures 2F, G), indicating decreased accumulation of lipid per adipocyte in HFD-WT-TM5441 mice. Thus, TM5441 treatment alleviates the weight gain induced by HFD, after reducing WAT weight. If loss of lean muscle mass had occurred, it might also have contributed to reduced body weight gain. Although direct measurement of lean muscle mass will be required to examine this possibility, spontaneous locomotor activities on day 7, during both light hours and dark hours, were similar in these groups of mice (light hours: HFD-WT-Control mice 25.68 ± 1.81 counts/min vs HFD-WT-TM5441 mice 25.56 ± 2.76 counts/min, P = 0.972, dark hours: HFD-WT-Control mice 96.88 ± 2.93 counts/min vs HFD-WT-TM5441 mice 90.06 ± 10.53 counts/min, P = 0.544) (Figure 2H) (Experimental procedure 6), which might suggest that severe loss of muscle mass would have been unlikely to have occurred in HFD-WT-TM5441 mice.

Next, HFD-WT-TM5441 mice and HFD-WT-Control mice were subjected to oral glucose tolerance tests on day 35 (Experimental procedure 7). Blood glucose levels of HFD-WT-TM5441 mice tended to be lower than those of HFD-WT-Control mice (Figure 2I). Plasma insulin levels were significantly decreased in HFD-WT-TM5441 mice as compared to HFD-WT-Control mice (Two-way repeat measures ANOVA; F = 5.17, P = 0.036, 0 min: 1.43 ± 0.14 ng/ml vs 0.98 ± 0.09 ng/ml, P = 0.017, 30 min: 2.35 ± 0.28 ng/ml vs 1.40 ± 0.19 ng/ml, P = 0.015, 60 min: 1.34 ± 0.12 ng/ml vs 0.85 ± 0.11 ng/ml, P = 0.007, 90 min: 1.47 ± 0.14 ng/ml vs 1.01 ± 0.10 ng/ml, P = 0.021, HFD-WT-Control mice vs HFD-WT-TM5441 mice, respectively) (Figure 2J). These results suggest that insulin sensitivity of HFD-WT-TM5441 mice is increased as compared to that of HFD-WT-Control mice.

Taken together, these results indicate that pharmacological inhibition of PAI-1 activity after initiation of HFD feeding attenuates body weight gain and improves insulin sensitivity.

To rule out the possibility that body weight reduction seen in HFD-WT-TM5441 mice (Figure 2B) was a consequence of non-specific effects of TM5441, i.e. effects other than PAI-1 inhibition, such as an unpleasant taste, we treated HFD-fed PAI-1 KO mice with TM5441 (Experimental procedure 4). Ten-week-old PAI-1 KO mice that had been pre-fed HFD for 2 weeks were divided into two groups. The first group was then fed HFD mixed with TM5441 (HFD-KO-TM5441 mice), while the second group continued to eat HFD without TM5441 (HFD-KO-Control mice). As expected, weekly body weights of HFD-KO-TM5441 mice after TM5441 loading were almost equal to those in HFD-KO-Control mice (Figure 3A). In addition, not only liver but also epididymal WAT weights on day 35 were similar in the two groups (Figures 3B, C). Thus, anti-obesity effects of TM5441 treatment in HFD-fed mice were observed only in the presence of intrinsic PAI-1 (Figure 2B). Thus, TM5441 was confirmed to exert its anti-obesity effects via inhibition of PAI-1 activity. In addition, these findings clearly indicate that treatment with a PAI-1 inhibitor after obesity development is also effective for suppressing weight gain in mice that are not depleted of PAI-1.

Leptin functions as an afferent signal in a negative feedback loop of energy metabolism leading to reduced feeding and promotion of energy expenditure (Friedman, 2016). However, hypothalamic leptin sensitivity is known to be impaired despite elevated circulating leptin concentrations under conditions of obesity (Cui et al., 2017), a phenomenon regarded as leptin resistance. Leptin resistance is considered to further promote body weight gain through a feed-forward system (Yamada et al., 2013). Hence, we measured circulating levels of leptin in HFD-WT-TM5441 mice and HFD-WT-Control mice 1 week after TM5441 loading (Experimental procedure 5). At this time-point, although body weights were similar in the two groups (day 7 of Figure 2B), interestingly, plasma leptin levels in HFD-WT-TM5441 mice had already decreased to less than half of those in HFD-WT-Control mice (HFD-WT-Control mice 8.38 ± 1.13 ng/ml vs HFD-WT-TM5441 mice 3.37 ± 0.68 ng/ml, P = 0.002) (Figure 4A). In addition, during the following weeks, HFD-WT-TM5441 mice gained less body weight than HFD-WT-Control mice. Therefore, we hypothesized that TM5441 treatment alleviated leptin resistance under HFD-fed conditions. To examine this possibility, we administered exogenous leptin to these mice, followed by monitoring body weights (Experimental procedure 8). After 1 week of HFD with or without TM5441, all mice were given a daily intraperitoneal injection of either leptin (HFD-WT-TM5441-Leptin mice and HFD-WT-Control-Leptin mice) or saline (HFD-WT-TM5441-Saline mice and HFD-WT-Control-Saline mice) for 5 consecutive days. There were no differences in body weights between the mice group before the first day of leptin or saline injection (Figure 4B). Under these settings, no differences in body weight change after leptin or saline injection were seen between HFD-WT-Control-Saline mice and HFD-WT-Control-Leptin mice (Figure 4C), suggesting that these HFD-fed mice have severe leptin resistance. Interestingly, administration of leptin after TM5441 treatment resulted in a remarkable suppression of body weight gain; HFD-WT-TM5441-Leptin mice gained markedly less weight than any of the other groups (HFD-WT-Control-Saline mice 1.70 ± 0.34 g vs HFD-WT-TM5441-Leptin mice 0.06 ± 0.38 g, P = 0.013, HFD-WT-Control-Leptin mice 1.55 ± 0.70 g vs HFD-WT-TM5441-Leptin mice 0.06 ± 0.38 g, P = 0.030, HFD-WT-TM5441-Saline mice 1.32 ± 0.31 g vs HFD-WT-TM5441-Leptin mice 0.06 ± 0.38 g, P = 0.049) (Figure 4C). Because body weights of all four groups of mice were similar on the first day of leptin or saline injection (Figure 4B), improvement of leptin sensitivity is suggested to be a direct effect of PAI-1 inhibition rather than a secondary effect reflecting suppressed weight gain. Thus, TM5441 treatment primarily improves leptin sensitivity in HFD-fed mice, thereby suppressing weight gain.

We next investigated the effect of TM5441 on leptin receptor signaling in the hypothalamus in HFD-fed mice (Experimental procedure 9). Leptin binds to the leptin receptor on proopiomelanocortin (POMC) neurons in the arcuate nucleus (ARC) to directly activate and induce POMC expression, while exerting an inhibitory effect on agouti-related peptide (AgRP)/neuropeptide Y (NPY) neurons (Timper and Bruning, 2017). We therefore studied ARC of HFD-WT-TM5441-Leptin, HFD-WT-Control-Leptin, HFD-WT-TM5441-Saline, and HFD-WT-Control-Saline mice fed HFD for 1 week, which had similar body weight (Figure 4D). Expression levels of these neuropeptides were examined in the ARC of mice 4 h after leptin administration. Although mRNA expression levels of agrp and npy showed no alterations in these four groups, Pomc expression was significantly increased in HFD-WT-TM5441-Leptin mice as compared to WT-Control-Leptin mice (HFD-WT-Control-Leptin mice 1.01 ± 0.12 vs HFD-WT-TM5441-Leptin mice 1.47 ± 0.23, P = 0.031) (Figure 4E). We further studied the effect of TM5441 on expressions of thermogenesis-related genes, such as ucp1, dio2, cidea, and prdm16, in brown adipose tissue (BAT) (Experimental procedure 10), because enhanced leptin signaling in the hypothalamic ARC increases BAT thermogenesis (Rezai-Zadeh and Munzberg, 2013). After 1 week of HFD with or without TM5441, all of the thermogenesis-related genes, including ucp 1 (HFD-WT-Control mice 1.00 ± 0.07 vs HFD-WT-TM5441 mice 1.32 ± 0.05, P = 0.002), were significantly up-regulated by TM5441 treatment (Figure 4F). These findings, taken together, suggest that administering TM5441 improves leptin sensitivity in neurons comprising the hypothalamic ARC, thereby activating BAT. Notably, this improvement was also observed, before suppression of weight gain became evident. Thus, an underlying mechanism whereby TM5441 treatment suppresses weight gain is alleviation of hypothalamic leptin resistance induced by HFD.