Physiological mechanisms of sustained fumagillin-induced weight loss

Abstract

Current obesity interventions suffer from lack of durable effects and undesirable complications. Fumagillin, an inhibitor of methionine aminopeptidase-2, causes weight loss by reducing food intake, but with effects on weight that are superior to pair-feeding. Here, we show that feeding of rats on a high-fat diet supplemented with fumagillin (HF/FG) suppresses the aggressive feeding observed in pair-fed controls (HF/PF) and alters expression of circadian genes relative to the HF/PF group. Multiple indices of reduced energy expenditure are observed in HF/FG but not HF/PF rats. HF/FG rats also exhibit changes in gut hormones linked to food intake, increased energy harvest by gut microbiota, and caloric spilling in the urine. Studies in gnotobiotic mice reveal that effects of fumagillin on energy expenditure but not feeding behavior may be mediated by the gut microbiota. In sum, fumagillin engages weight loss-inducing behavioral and physiologic circuits distinct from those activated by simple caloric restriction.

Keywords: Behavior; Intermediary metabolism; Metabolism; Obesity; Therapeutics.

Figure 1. Fumagillin reduces caloric intake and body weight in obese rats.

(A) Body weight changes (% increase relative to starting weight) over the initial 12 weeks of feeding of HF or SC diets. Data are means of n = 86 for the HF group and n = 26 for the SC group. Weight gain was significantly different in the HF versus SC animals after 1 week of feeding (P < 0.0012). (B) Mean caloric intake over the 12-week feeding period in HF and SC diet groups, expressed as Kcal/rat/week. The difference between groups are significant at all time points (P < 0.0004). (C) After the initial 12-week feeding period, animals were divided into 4 intervention groups: (i) HF rats continued on HF diet ad libitum (HF); (ii) SC rats continued on SC diet ad libitum (SC); (iii) HF rats switched to HF/FG diet ad libitum (HF/FG); (iv) HF rats pair-fed on HF diet (HF/PF) to match food intake of the HF/FG group. Body weight change is expressed as percent change relative to the start of the intervention period. Beginning at day 5 of intervention, the HF/FG and HF/PF rats lost weight compared with HF or SC (P = 0.04–0.00001). Data represent n = 26 for HF, HF/PF, and SC groups at 2 weeks of diet intervention and an n = 13 for these groups between 2–8 weeks of intervention. For HF/FG, n = 34 at 2 weeks of diet intervention and n = 17 from 2–8 weeks. (D) Mean daily food intake in the 4 intervention groups described in C expressed as grams of food consumed/rat. (E) Daily caloric intake for the groups described in C, expressed as Kcal/gram body weight/day. The HF and SC groups and the HF/FG and HF/PF groups were not different at any time point. Consumption of HF and HF/FG diets were significantly different (P < 0.024) from day 2–38, except for days 13, 19, 20, 27, 32, 36, and 37. Consumption of HF food by the HF and HF/PF groups were significantly different (P < 0.015) from day 1–38, except for days 20, 27, 32, 36, and 37. (F) Mean daily caloric intake, Kcal/gram body weight/day, for groups described in C. Data are mean ± SD, with *P < 0.05 when compared with either HF or SC group. Two-tailed, unpaired t tests were performed in all panels. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 2. Fumagillin reduces white adipose tissue weights and improves insulin sensitivity in obese rats.

Diet-induced obesity was induced in Wistar rats by feeding a high-fat (HF) diet for 12 weeks. A control group was fed a standard chow (SC) diet for the same time period. After the feeding period, rats were then subjected to dietary intervention periods of 2 or 8 weeks as described in Figure 1C. (A) Epididymal white adipose tissue (EWAT) weights for the groups described in Figure 1C, fed the indicated diets for 2 or 8 weeks (2W, 8W, respectively). ⁺P < 0.05 compared with HF group after 2 weeks of diet intervention. ^@P < 0.05 when compared with HF and HF/FG groups after 2 weeks of diet intervention. ^#P < 0.05 when compared with the other 3 groups after 8 weeks of diet intervention. (B) Liver weights for the groups described in Figure 1C, fed the indicated diets for 2 or 8 weeks (2W, 8W, respectively). ⁺P < 0.05 when compared with HF group after 2 weeks of diet intervention. *P < 0.05 when compared with HF group after 8 weeks of diet intervention. In A and B, tissue weights were normalized to body weight. n = 6 for each group. Data are presented as mean ± SD. (C) Glucose infusion rate during hyperinsulinemic-euglycemic clamps. A cohort of animals treated as described in Figure 1C were used. Data are presented as mean ± SD for n = 5–7 rats per group. ⁺P < 0.05 when compared with other groups after 2 weeks of diet intervention. *P < 0.05 when compared with other groups after 8 weeks of diet interventions. For A–C, 2-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 3. Fumagillin feeding affects oxygen exchange rate and heat production.

The groups described in Figure 1C were studied in metabolic cages at 2 weeks or 8 weeks after the start of the dietary intervention period. (A) Oxygen exchange rate (VO₂) and (B) heat production (HEAT) are represented. For these measurements, upper panels represent data collected from the animals after 2 weeks and lower panels after 8 weeks of intervention. Left panels are raw data, and right panels summarize measurements as AUC. All groups received food once daily at ZT6:00 (1 pm), as shown by the downward-pointing arrows. Data for AUCs in the right panels are presented as mean ± SD for n = 7 for each group. *P < 0.05 when compared with the other 3 groups; ⁺P < 0.05 when compared with HF/PF group. Two-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 4. Fumagillin feeding affects ambulatory activity and has an acute effect on respiratory exchange rate (RER).

The groups described in Figure 1C were studied in metabolic cages at 2 weeks or 8 weeks after the start of the dietary intervention period. (A) Ambulatory activity (XAMB) and (B) respiratory exchange rate (RER) are represented. For these measurements, upper panels represent data collected from the animals after 2 weeks and lower panels after 8 weeks of intervention. Left panels are raw data, and right panels summarize measurements as AUC. All groups received food once daily at ZT6:00 (1 pm), as shown by the downward-pointing arrows. Data for AUCs in the right panels are presented as mean ± SD for n = 7 for each group. *P < 0.05 when compared with the other 3 groups; ⁺P < 0.05 when compared with HF/PF group. Two-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 5. Fumagillin feeding affects feeding rhythm and expression of circadian genes.

Wistar rats were fed a HF diet for 12 weeks and then subjected to an intervention period of 4 weeks, during which they consumed HF diet ad libitum (HF), HF diet with fumagillin (HF/FG), or an amount of HF food matched to the amount consumed by the HF/FG group (HF/PF). (A–C) Average food intake measured in these 3 groups of rats daily over 4 time periods after daily food provision at ZT6:00 (1:00 pm). The periods sampled were ZT6:00–10:00, ZT10:00–14:00, ZT14:00–2:00, and ZT2:00–6:00, equating to 1–5 pm, 5–9 pm, 9 pm–9 am, and 9 am–1 pm. n = 3, 5, and 4 for HF, HF/FG, and HF/PF groups, respectively. HF group compared with HF/FG or HF/PF groups for total food intake; P < 0.000009. HF group compared with HF/FG group for food intake during ZT6:00–2:00 (1 pm–9 am); P < 0.04. HF group compared with HF/PF group for food intake during all time periods except for ZT10:00–14:00 (5–9 pm); P < 0.04. HF/FG group compared with HF/PF group for food intake during all time periods except for ZT2:00–6:00 (9 am–1 pm); P < 0.002. (D) Expression levels of Bmal1 (upper panel), Clock (middle panel), and Per1 (lower panel) transcripts in liver, stomach, duodenum, jejunum, ileum, colon, cecum, epididymal white adipose tissue (EWAT), and mesenteric white adipose tissue (MWAT). Samples were collected after 10 days of diet intervention. Data represent means ± SD for n = 3, 5, and 4 for HF, HF/FG, and HF/PF groups, respectively. *P < 0.05 when compared with the other 2 groups. ^#P < 0.05 when compared with the HF group. ⁺P < 0.05 when compared with HF/FG group. For all panels, 2-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 6. Fumagillin treatment affects body temperature, expression of torpor-related genes, and urine catecholamines.

Rats were fed the HF diet for 12 weeks and then subjected to an intervention period of 2 or 4 weeks, during which time they consumed HF diet ad libitum (HF), HF diet with fumagillin (HF/FG), or an amount of HF food matched to the amount consumed by the HF/FG group (HF/PF). (A) Following 4 weeks of diet intervention, rectal temperatures were measured in rats at 6 time points over a 20-hour period that covered light and dark cycles. Data are presented as mean ± SD for n = 7 per group. *P < 0.05 when compared with the other 2 groups. (B) Urine catecholamine levels, normalized to creatinine (CREA), in samples collected after 4 weeks of diet intervention. Data are presented as mean ± SD for n = 8 per group. *P < 0.05 when compared with the other 2 groups. (C and D) Hepatic expression of torpor-related transcripts after 2 weeks of diet interventions. Data are presented as mean ± SD. n = 6 for each group. ⁺P = 0.057 when compared with HF group. Two-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 7. Fumagillin treatment affects glycogen levels, but hepatic glycogen is not the mediator of fumagillin-induced changes in feeding behavior.

First 3 panels: rats were fed HF or SC diet for 12 weeks and then subjected to dietary intervention periods of 2 or 8 weeks, as described in Figure 1C. Liver and muscle samples were collected for measurement of glycogen levels or phospho-GSK3β levels at ZT6:00 (1 pm) at the 2-week or 8-week time points. Last 3 panels: rats were fed on HF diet for 12 weeks and then subjected to an intervention period of 4 weeks, during which rats were fed HF diet with fumagillin ad libitum (HF/FG) or an amount of HF food matched to the amount consumed by the HF/FG group (HF/PF), in the presence or absence of methionine restriction (MR). Liver samples were collected at ZT6:00 (1 pm), immediately prior to the normal once-daily provision of food. (A) Hepatic glycogen levels. Data are mean ± SD for n = 6 for each group. *P < 0.05 when compared with other groups at each time point. (B) Glycogen levels in gastrocnemius muscle. Data are mean ± SD for n = 6 for each group. *P < 0.05 when compared with other groups at each time point. (C) Ratio of phosphorylated GSK3β compared with total GSK3β in liver. Data are mean ± SD for n = 6 for each group. *P < 0.05 when compared with other groups at each time point. (D) Immunoblot analysis of liver samples obtained from rats treated with adenoviruses expressing either a Flag-tagged C-terminal deleted version of the muscle isoform of glycogen-targeting subunit of protein phosphatase 1 (GmΔC-Flag) (21) or β-galactosidase (βGal) and continuously fed the indicated diets for 1 week. (E) Hepatic glycogen levels in rats treated with the indicated adenoviruses. Data are mean ± SD for n = 5–7 animals per group. *P < 0.02 when compared with the other groups. (F) Changes in RER in response to daily provision of food at ZT6:00 (1 pm) (downward arrow) in rats fed the indicated diets and treated with the indicated adenoviruses. All diets in this experiment contained normal methionine levels. n = 5–7 for each treatment group. For all panels, 2-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 8. Feeding rhythms and hepatic torpor and Clock gene expression of gnotobiotic mice receiving cecal microbiota transplants from HF/FG or HF/PF rats.

All of the experiments in A–F, are from n = 5 mice/group. (A) Daily accumulated food intake of mice colonized with the cecal microbiota of a HF/FG rat. HF diet was provided ad libitum. (B) Daily accumulated food intake of mice colonized with the cecal microbiota of a HF/PF rat. HF diet was provided ad libitum. (C) Daily accumulated food intake of mice colonized with the cecal microbiota from a HF/FG rat. HF diet was provided ad libitum from experimental day 1–5 and then restricted by 10% from day 6–10. (D) Daily accumulated food intake of mice that were the recipients of a cecal microbiota transplant from a HF/PF rat. HF diet was provided ad libitum from day 1–5 and then restricted by 10% from day 6–10. (E) Expression of hepatic torpor genes in mice from the studies shown in A–D. ⁺P = 0.06 when compared with HF/FG ad libitum/CR group. (F) Expression of hepatic circadian genes in mice from the studies shown in A–D. *P < 0.05 when compared with the corresponding calorically restricted groups. In E and F, data are presented as mean ± SD, and 2-tailed, unpaired t tests were performed. P < 0.05 with a Bonferroni correction was used to define statistical significance among groups.

Figure 9. Cumulative change in body weight in germ-free mice after fumagillin administration.

Germ-free mice were provided with HF diet ad libitum from day –4 to –1. Thereafter, mice received either HF or HF/FG diets ad libitum from day 0–8. The 2 diets were then switched from day 9–18. Two-tailed, unpaired t tests were performed. P < 0.05 was used to define statistical significance among groups.*P < 0.05 between the 2 groups. n = 6/group.