A nutrient-hormone axis pivots an energy trade-off between resting thermogenesis and movement expenditure

Highlights

• •Chronic dietary protein restriction improves weight bearing exercise ability in male mice.
• •Chronic dietary protein restriction improves running economy and lowers daily physical activity energy expenditure.
• •These effects occur though lower body weight and adiposity driven by activation brown adipose tissue thermogenesis via liver-derived fibroblast growth factor 21.
• •Overall, the higher thermogenesis energy expenditure is offset by lower physical activity energy expenditure.
• •Our findings may provide a reason for the existence of diet-induced thermogenesis.

Abstract

Objective

The biological reason why mammals increase daily energy expenditure when consuming a protein-dilute diet is currently enigmatic. Recognizing that body weight is a chief determinant of movement economy, we hypothesized that there is an interaction between dietary protein dilution-induced resting energy expenditure and subsequent effects on body weight and energy expenditure during physical activity.

Methods

Using inbred and genetically modified mice and methods such as defined diet feeding, body composition analysis, indirect calorimetry combined with forced treadmill running or in-cage wheel running, tissue histological and molecular/biochemical analyses, we examined the interactions of diet and exercise biology.

Results

We found that mice chronically fed a protein-dilute diet exhibited enhanced weight-bearing exercise ability, an effect directly linked to the improved movement economy. Mechanistically, this improved movement economy and exercise ability resulted from a reduction in adiposity and body weight, which was facilitated by the increased liver-derived fibroblast growth factor 21 (Fgf21) that promotion of brown adipose tissue thermogenesis. Overall, dietary protein dilution-induced thermogenesis and resulting improved movement economy enabled a differential budgeting of energy expenditure between active and inactive states, resulting in balanced total energy expenditure.

Conclusion

The influence of liver-derived Fgf21 on brown adipose tissue thermogenesis and energy expenditure drives improved movement economy and weight-bearing exercise capability during chronic dietary protein dilution.

Graphical abstract

1. Introduction

Energy balance is critical for vitality and survival, primarily determined by the balance between food energy intake and somatic energy expenditure.^1,2 This balance between caloric intake and expenditure chronically shapes mammalian body composition, with body fat accumulating during periods of ample food supply to protect against periods of inadequate availability and defend against disease insults.³

An important consideration during times of limited food supply is the macronutrient composition, ensuring the appropriate balance of protein, carbohydrates, and fats according to the animal’s metabolic needs. Of these, protein is a vital nutrient because it contains essential amino acids, many of which must be sourced from diet as they are not sufficiently made, or not made at all, by the body. These amino acids are pivotal for protein synthesis and many other vital metabolic processes.⁴ Obligatory protein loss is an indispensable homeostatic process, and the amino acids lost must be replenished through dietary protein intake.^4,5 Thus, there are likely evolved physiological mechanisms linking periods of altered environmental protein availability to energy homeostasis through innate behaviors, such as food intake and physical activity. If such mechanisms did not exist, the consequences of inadequate protein balance are neurological, immune, and musculoskeletal dysfunction, as well as infertility.⁶ Recent studies have demonstrated that when an animal is in an environment with adequate calories but low protein abundance, it increases food energy consumption but remains lean because of a disproportionate increase in energy expenditure.7, 8, 9, 10, 11 While the increased food intake can be explained by the need to consume more protein,¹² the biological and mechanistic basis of why an animal would burn excess calories in the face of a diet with a limiting nutrient remains enigmatic.

Although often used interchangeably, the term movement economy differs from movement efficiency and specifically refers to the energetic cost associated with an animal maintaining a particular velocity, emphasizing the relationship between oxygen consumption and speed. In contrast, movement efficiency is the ratio of mechanical work performed to the energy expended. In simplified terms, movement economy is the energy required to move at a given speed, while movement efficiency measures how effectively energy is converted into mechanical work. As physical activity is a component of total daily energy expenditure,^1,13 and body weight is the chief determinant of physical activity energy expenditure and movement economy,14, 15, 16 we reasoned that there would be an interplay between dietary protein dilution (equicaloric low protein diet replaced by non-protein calories) driven resting energy expenditure and subsequent effects on body weight and physical activity energy expenditure. Indeed, here we demonstrate that dietary protein dilution pivots an energy trade-off between resting thermogenesis and movement energy expenditure through fibroblast growth factor 21 (Fgf21) driven brown adipose tissue (BAT) thermogenesis.

2. Methods

2.1. Animal ethics

Animal experiments were conducted according to regional, national, and continental ethical guidelines and protocols were approved by local regulatory authorities (Monash University Animal Ethics Committee; Project: 13263) and conformed to Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.

2.2. Diets

Diets were purchased from Specialty Feeds (Perth, WA, Australia; Supplementary Table 1) and all studies involved a comparison of diets SF17-180 (Specialty Feeds; 20% caloric energy from protein), SF17-175 (Specialty Feeds; 5% caloric energy from protein).

2.3. Recombinant viruses

A control (green fluorescent protein (GFP)), Cre-recombinase (CRE) cDNAs were cloned into adeno-associated viruses (AAV) genome plasmids for subsequent production of self-complimentary AAV using the 3-plasmid system (produced by the Lab of Oliver J. Müller¹⁷). The genome plasmid encoding either hepatocyte specific regulated (via liver promoter 1 (LP1)-promoter^18,19) GFP or CRE was co-transfected with the rAAV2 helper plasmid containing all genes required for rAAV2 production, but lacking the Cap gene (pDG∆VP) and the capsid plasmid p5E18VD2/8-mut6 (aa 589–592: QNTA to GNRQ; AAV8 to AAV2).²⁰ Purification and quantification of the final vector stocks was done by density gradient/filtration purification and quantitative real-time polymerase chain reaction (qPCR), respectively.¹⁷

2.4. Mouse experiments

Mice were assigned to groups based upon initial body mass for counterbalancing. Pre-established criteria for exclusion of mice from study groups were obvious infections/wounds which would impact on feeding behavior as well as metabolic profile. Where possible, analysis of samples was blinded. Unless stated otherwise, male mice aged 7 weeks upon arrival, were acclimated to the local housing facility (12–12 h light–dark cycle, 22°C–24°C) for 1 week prior to experimentation and were raised on standard rodent chow (102093; Barastoc, Melbourne, VIC, Australia). Mice used for experiments were male C57Bl/6 J (Monash Animal Research Platform, Clayton, VIC, Australia), male Fgf21+/+ and Fgf21−/−, and Fgf21fl/fl littermates, all on C57Bl/6 J background. Fgf21 flox mice (B6.129S6(SJL)- Fgf21tm1.2Djm/J) were crossed with human Cytomegalovirus (HCMV) immediate-early enhancer/promoter to drive strong, ubiquitous expression of Cre (CMV-Cre) mice (B6.C-Tg (CMV-cre)1Cgn/J) to obtain whole body Fgf21 knockout mice. CMV-Cre was bred out and homozygous Fgf21−/− (i.e., knockout (KO)) and Fgf21+/+ (i.e., wildtype (WT)) mice were established from the heterozygous mice. The mice were verified by genotyping. Mice are kept as WT and as homozygous KO for experiments, but inter-crossed regularly to minimize drifting apart of the 2 lines. The heterozygous mice are then used again to obtain new homozygous WT and KO mice.

For liver/hepatocyte specific Fgf21 silencing experiments, 7 weeks old Fgf21fl/fl mice were administered a total of 1.0 × 10¹¹ virus particles per mouse via the tail vein. Mice either received Cre-AAV for Fgf21 KO or Gfp-AAV as negative control (NC). One week following this time, the dietary intervention was initiated.

For all experiments, 8 weeks old mice were randomly distributed to cages (3–4 per cage) and then weighed and counterbalanced to diet study groups. In 1 experiment, mice were single housed for accurate measurement of food intake per mouse. For chronic diet experiments, mice were fed the diets for ≥12 weeks after which functional phenotyping was conducted. Graphical depictions of experimental designs are shown in Supplementary Fig.1.

2.5. Biometric and metabolic phenotyping

In all experiments, mice were carefully weighed on a precise scale. In some experiments, 2 days prior to sacrifice, body composition was assessed at the end of the experiment by ECHO-magnetic resonance imaging (EchoMRI^TM 3-in-1 Body Composition Analyzer; Echo Medical Systems, Houston, TX, USA) and immediately before the initiation of diet feeding.

For assessment of in-cage indirect calorimetry, mice were individually housed in metabolic phenotyping system cages (Promethion-M High Definition Multiplexed Respirometry System, Sable Systems International, North Las Vegas, NV, USA) and assessed essentially according to Lark et al.²¹ At the end of the 12 weeks feeding period, mice were allocated to cages and allowed to acclimate for 3–4 days and maintained on the prior diets fed throughout the experiment. After the acclimation, a 10-day measurement was conducted. Running wheels were present in the cages the entire time but remained locked for the initial 4 days of the 10 days period. O₂ and CO₂ measurements and voluntary wheel running (VWR) measurements were collected at a second-by-second resolution. Energy expenditure (W) was estimated using the Weir equation²² from oxygen consumption (VO₂) and carbon dioxide production (VCO₂) rate measurements: energy expenditure (EE) (W or J/s) = ((1.44 × (3.94 × VO₂ (mL/h) + 1.11 × VCO₂ (mL/h))/1000 × 4.196) × 0.28. For VWR calculations, the data from the final 3 days were assessed and averaged.

For forced treadmill exercise indirect calorimetry, we employed the metabolic treadmill (0195-8001; Modular Enclosed Metabolic Treadmill, Columbus Instruments, Columbus, OH, USA), and tests were conducted essentially according to Lee-Young, et al.²³ Two days prior to the exercise stress test, all mice were acclimatized to treadmill running by performing 10 min of exercise at a speed of 10 m/min (0% incline). To determine VO₂, VCO₂, respiratory exchange ratio (RER), and energy expenditure, mice were placed in an enclosed single lane treadmill connected to O₂ and CO₂ sensors (1000-4207; Oxymax High Speed Zirconia O₂ and Non-dispersive infrared (NDIR) CO₂ indirect calorimeter, Columbus Instruments, Columbus, OH, USA). Following a 10-min basal period, mice commenced running at 3 m/min on a 5% incline. Running speed was increased by 3 m/min every 3 min until 24 m/min (total test time of 24 min). VO₂ and VCO₂ rates were assessed at 30 s intervals throughout the exercise periods. Energy expenditure (W) was estimated using the Weir equation²² from VO₂ and VCO₂ measurements: EE (W or J/s) = ((1.44 × (3.94 × VO₂ (mL/h) + 1.11 × VCO₂ (mL/h))/1000 × 4.196) × 0.28. RER was calculated as the ratio of VCO₂ to VO₂. Power output was calculated based upon the mouse weight and its vertical displacement per unit time.

2.6. Wire hang test

The mouse was placed on the cage top, which was then inverted and suspended above a cage, with the latency to when the animal falls recorded. This test was performed over 3 consecutive days with 3 trials per session. The average performance for each session is presented as the average of the 3 trials.

2.7. Treadmill exercise to exhaustion

Mice were exercised on a multilane treadmill (0208-8101, Exer 3/6 Treadmill; Columbus Instruments) at 0% slope. Mice were acclimated to treadmill running for 3 days before the initiation of the experiments. For acclimation, mice were placed on the multilane treadmill for 10 min and run for 5 min at 10 m/min and 1 min at 15 m/min at 0% slope. All animals were randomized before the initiation of exercise tests. The endurance test was performed with a warm-up period of 5 min at 10 m/min, followed by 20 m/min until fatigue was reached. The time until fatigue at 20 m/min was considered the representative measure of endurance capacity. Fatigue was determined as the time point where the mice could not be prompted to continue running for at least 5 s.

2.8. Brown fat thermosensing telemetry

To obtain real-time BAT temperature measurements, a temperature sensitive transponder (UID Temperature Microchip, UCT-2112; Unified Information Devices, Kenosha, WI, USA) was surgically implanted between the 2 lobes of the interscapular BAT. Continuous temperature measurements were made using the mouse matrix home cage monitoring system (UID Mouse Matrix, United Information Devices) that allows the simultaneous measurement of BAT temperature and movement using radio-frequency identification (RFID) technology.

2.9. In vivo glucose uptake

Experiments were conducted essentially as previously described.²⁴ In short, mice were fasted for 4.5 h and then administered 12 µCi of 2-[1,2-³H(N)]-Deoxy-D-glucose (NET549A250UC; Revvity, Scoresby, VIC, Australia), USA). Blood (25 uL) was sampled (16.443; Microvette tube, Sarstedt, Nümbrecht, Germany) from the tail beforehand and at 15 min, 30 min, and 45 min. Mice were killed by cervical dislocation and tissues were rapidly collected and frozen. Blood and tissue samples were proceeded accordingly and tissue glucose uptake rates were calculated.⁷

Blood serum metabolites & hormones. Serum Fgf21 (MF2100; R&D Systems, Minneapolis, MN, USA) and urea (Z5030016; Biochain, Newark, CA, USA) were measured from the blood serum samples. All samples were loaded in order to fit within the assay range of the reagents supplied.

2.10. Adipose tissue RNA extraction and analysis

RNA was extracted from tissues using QIAzol RNAzol (R4533; Merck, Bayswater, VIC, Australia) and cDNA was synthesized using the Quantitect Reverse Transcription Kit (205314; Qiagen, Germantown, MD, USA). qPCR was conducted using Quantitect Sybr Green qPCR (204143; Qiagen, Germantown, MD, USA) with the following primers: mouse Tbp (QT00198443 (Quantitect, Qiagen)) as a housekeeping gene, Ucp1 (GCTTTGCCTCACTCAGGATTGG and CCAATGAACACTGCCACACCTC), Ppargc1 (GAATCAAGCCACTACAGACACCG and CATCCCTCTTGAGCCTTTCGTG), Cidea (GGTGGACACAGAGGAGTTCTTTC and CGAAGGTGACTCTGGCTATTCC), Cox8b (AGCCAAAACTCCCACTTCC and TCTCAGGGATGTGCAACTTC), Cox7a1 (AAACCGTGTGGCAGAGAAGCAG and CCCAAGCAGTATAAGCAGTAGGC). Relative mRNA expression was calculated using the comparative C(T) method.²⁵

2.11. Tissue histology

Intrascapular BAT and subcutaneous white adipose tissue (scWAT) samples were harvested, fixed overnight in 10% formalin (formulated and provided by the Monash Histology Platform), paraffin embedded and sectioned (6 mM) for immunohistochemistry or hematoxylin & eosin (H&E) staining (Monash Histology Platform). For Uncoupling protein 1 (UCP1), sections were incubated in primary antibody (ab10983, 1:1000 dilution; Abcam, Cambridge, UK) and localized with a diaminobenzidine-generated brown reaction product.

For skeletal muscle histology, samples were snap frozen in liquid nitrogen cooled isopentane and archived at –80°C until required. Gastrocnemius cryosections (10 mM thick) were sectioned from the midpoint of the muscle using a CM1850 cryostat (Lecia BioSystems, Mt Waverley, VIC, Australia) and mounted onto Superfrost Plus Adhesion Microscope Slides (EPBRSF41296SP; Epredia, Portsmouth, NJ, USA). For immunostaining, sections were fixed in ice-cold acetone for 5 min and washed three times with phosphate buffered saline (PBS) (5 min each). Sections were then permeabilized with 0.1% TX-100 in PBS for 5 min, followed by washing via immersion in PBS for 5 min. Blocking was performed for 1 h at room temperature in 5% bovine serum albumin (BSA; A7030; Merck), 10% donkey (D9663; Merck) and 10% goat (G9023; Merck) serums in PBS. Sections were co-incubated with primary antibodies diluted in blocking buffer overnight at 4°C including dystrophin (ab15277; 1:100; Abcam, Cambridge, UK) and fiber-type specific antibodies Type I (MYH7, BA-F8), Type IIa (MYH2, SC-71), Type IIb (MYH4, BF-F3), and Type IIx (MYH1, 6H1). All fiber-typing monoclonal antibodies were diluted 1:100 and were developed by S. Schiaffino (Universita degli Studi di Padova, Padova, Italy) or C. Lucas (University of Sydney, Sydney, NSW, Australia) and obtained from the Developmental Studies Hybridoma Bank, created by the National Institute of Child Health and Human Development (NICHD) of the National Institutes of Health (NIH) USA and maintained at the Department of Biology, The University of Iowa, Iowa City, IA, USA. Following primary antibody incubation, sections were washed extensively with PBS and incubated for 1 h at room temperature with Alexa Fluor-conjugated secondary antibodies (diluted 1:500 in blocking buffer; A21141, A21044, A31573; ThermoFisher Scientific, Waltham, MA, USA). Sections were again washed in PBS and coverslips mounted using Fluoromount G (00-4958-02; ThermoFisher Scientific). Slides were scanned using an Olympus VS200 universal whole slide imaging scanner (VS200; Evodent Scintific, Nagano, Japan). ImageJ 1.54e (Java, NIH, Bethesda, MD, USA) with MuscleJ2 plugin²⁶ installed was used for automated analysis of minimal “Feret’s diameter” for muscle fibers, total number of muscle fibers and fiber type composition.

2.12. Tissue citrate synthase activity assays

Frozen gastrocnemius skeletal muscle and heart tissues were pulverized in LN₂ then homogenized with a pestle in lysis buffer (50 mM Tris pH 7.5, 1 mM ethylenediaminetetraacetic acid (EDTA), 10% v/v glycerol, 1% v/v TritonX-100, 50 mM NaF, 5 mM Na₄P₂O₇, 1 mM phenylmethylsulfonyl fluoride (PMSF) and 5 μL/mL Protease Inhibitor Cocktail (P8340; Merck, Bayswater)) at 10 μL/mg muscle tissue or 20 μL/mg heart tissue. Lysed samples were incubated on ice for 20 min, freeze-thawed twice then centrifuged at 10,000 × g for 10 min. Supernatant was collected and a bicinchoninic acid (BCA) protein assay (23225; ThermoFischer Scientific) performed to dilute samples to 1.5 mg/mL (gastrocnemius) or 0.5 mg/mL (heart). Sample (5 μL gastrocnemius, 3 μL heart) was added to a 250 μL reaction containing 80 mM TRIS buffer pH 8.3, 0.1 mM 5,5′-dithiobis (2-nitrobenzoic acid) (DTNB), 0.48 mM acetyl CoA, and 0.6 mM oxaloacetate. Citrate synthase activity was measured in a microplate reader at 25°C via change in absorbance at 412 nm based on the molar extinction coefficient for 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) (e = 13,600/M).²⁷

2.13. Quantification and statistical analyses

2.13.1. Calculations

Body, lean and fat energy gain was calculated from ECHO-magnetic resonance imaging values over the experimental time from standard calorific values: lean = 7.53 kJ/g, fat = 39.33 kJ/g.²⁸ Total body energy gain was the sum of fat and lean energy gain. Cumulative food intake was calculated from the total food weight eaten over the experimental time period multiplied by the digestible energy density of the diets. Metabolic efficiency was then calculated as the quotient of total body energy gained and the food energy eaten over the experimental period.

For VWR calculations, the final 3 days of data of the 6 days period with unlocked running wheels was assessed and averaged. Binning (if/then arguments) were made in Microsoft excel from second-to-second measurements. Time budgets and energy expenditure for velocity ranges (5 m/min bins) were determined and total meters per velocity range and for the entire 24 h period was calculated. The metabolic cost of exercise of VWR was calculated by dividing the total number of kJ expended at a given velocity range by the total distance traveled at that velocity. The average energy cost of VWR across all speeds was used to calculate the metabolic cost of exercise per meter (i.e., running economy) for each individual mouse.

2.13.2. Statistical analyses

Statistical analyses were performed using t tests (two-sided), or two-way analysis of variance (ANOVA) with or without repeated measures, where appropriate, with Holm-Sidak-adjusted post-tests. Nonparametric tests were conducted when data were not normally distributed. All analyses were carried out with SigmaPlot v.13 software (Grafiti, Palo Alto, CA, USA). Statistical details can be found within the Figure legends. Differences between groups were considered significant when p < 0.05.

3. Results

3.1. Chronic dietary protein dilution (DPD) improves weight-bearing exercise ability and movement economy

Initially, we tested whether chronic exposure of mice to a protein dilute diet (i.e., DPD) would affect their exercise ability. We fed the mice with a control (20% energy from protein) diet or a protein dilute (5% energy from protein) diet for 12 weeks which resulted in the expected,^7,29 ∼50% lower body weight gain associated with low dietary protein intake (Fig. 1A), and marked reductions on lower adipose tissue weights (Supplementary Fig. 2A). During a treadmill exercise test, chronic DPD mice had a ∼40% increase in endurance exercise ability (Fig. 1B). Similarly, DPD mice had a 25% higher ability to hold themselves on a wire mesh before falling (Fig. 1C). DPD mice had equal (weight-supported) grip strength compared to controls (Supplementary Fig. 2B), revealing a higher grip strength when expressed relative to body weight (Fig. 1D). In summary, chronic DPD improves weight-bearing exercise ability without affecting basic neuromuscular function.

Fig. 1.

Chronic dietary protein dilution improves weight-bearing exercise ability and movement economy of male mice. (A) Body weight of male C57Bl/6 J mice after 12 weeks of feeding a normal (20%P), or protein-dilute (5%P) diet. (B) Time to fatigue during steady-state treadmill exercise of mice. (C) Time to fall during a wire hang test of mice. (D) Four limb grip strength of mice. (E) Power output during treadmill running of male C57Bl/6 J mice after 12 weeks of feeding a normal (20%P), or protein-dilute (5%P) diet. (F) Running economy of mice. (G) Running efficiency of mice. (H) Respiratory exchange ratio of mice. Data are presented as mean ± SEM, n = 9/group. Statistical tests: (A–D): Students t tests (two-sided); (E–H): two-way repeated measures ANOVA with Holm-Sidak post hoc tests. * p < 0.05, ** p < 0.01, *** p < 0.001. ANOVA = analysis of variance; EE = energy expenditure; g = gram; J = joule; mg = milligram; N = newton; P = protein; SEM = standard error of the mean; VCO₂ = carbon dioxide production; VO₂ = oxygen consumption.

As exercise performance correlates strongly with economy of movement,¹⁴ we then had mice perform a graded treadmill exercise test with a 5% slope with simultaneous indirect calorimetry measurement to assess energy expenditure and RER. The power output during exercise at each velocity was lower for the lighter DPD mice (Fig. 1E). This coincided with a ∼15% lower energy expenditure of the DPD mice at each velocity (Fig. 1F), meaning that the mice had better economy of movement. In contrast, there was no difference in running efficiency (i.e., the energy expenditure needed per power output) at any velocity (Fig. 1G). As a higher RER at a given velocity, and the shape of the curve during graded exercise, indicates exercise exertion level,³⁰ we assessed this (Fig. 1H). Congruent with the effects on movement economy (Fig. 1F), DPD mice had a lower RER at higher velocities, indicating that they were performing less stressful exercise at each velocity (Fig. 1H).

We then wanted to assess whether chronic DPD also has effects on movement economy in a more “free-living” situation rather than forced treadmill exercise during light hours as in Fig. 1. Hence, we adopted protocols established by Lark et al.,²¹ where mice were placed in cages to assess indirect calorimetry during VWR throughout the day. Importantly, the control diet fed mice had the expected VWR patterns of running distance (Fig. 2A) and percentage of time spent at each speed (Supplementary Fig. 3A). Chronic DPD-adapted mice spent less time at lower velocities but more time at higher velocities (Fig. 2A and Supplementary Fig. 3A), with a similar total running distance performed over a day (Fig. 2B). Similar to the effects seen with forced treadmill exercise, DPD mice had a lower energy expenditure at each velocity (Fig. 2C and Supplementary Fig. 3B), again meaning that they had improved movement economy. When summated, the total metabolic cost at lower running velocities was substantially lower (Supplementary Fig. 3C); a product of both the lower amount of time spent at those speeds and a lower energy expenditure at those speeds. In contrast, the higher amount of time of DPD mice running at higher speeds meant that they spent more of their daily energy running at those higher speeds (Supplementary Fig. 3C). Nonetheless, this did not offset the effects of the lower energy expenditure at low speeds, resulting in a lower total daily energy expenditure due to running in DPD mice (Supplementary Fig. 3D). Thus, since the total amount of meters ran was the same (Fig. 2B), coupled with the lower overall energy expenditure with VWR (Supplementary Fig. 3D), meant that overall the DPD mice had a 32% improvement in running economy (Fig. 2D).

Fig. 2.

Chronic dietary protein dilution improves movement economy during voluntary wheel running and alters energy expenditure budgeting. (A) Daily VWR distance at certain speed ranges during in-cage wheel running of male C57Bl/6 J mice after 12 weeks of feeding a normal (20%P), or protein-dilute (5%P) diet (n = 9/group). (B) Total daily running distance of mice as in A. (C) Energy expenditure at certain speed ranges of mice as in A. (D) Daily VWR economy of mice as in A. (E) Body weight of male C57Bl/6 J mice at Week 2 as well as Week 12 after initiation of feeding a normal (20%P), or protein-dilute (5%P) diet (n = 8/group). (F) Mice in E were housed in metabolic cages with free access to running wheels, and energy expenditure was measured using indirect calorimetry. Daily energy expenditure was measured from a summated time when the mouse was off the running wheel. (G) Daily energy expenditure was measured from a summated time when the mouse was on the running wheel as in F. (H) The total energy expenditure from mice as in F. Data are presented as mean ± SEM. Statistical tests: (A, C, and E–H): two-way repeated measures ANOVA with Holm-Sidak post hoc tests; (B and D): Students t tests (two-sided). Diet group difference: * p < 0.05, ** p < 0.01. Time difference: ^##p < 0.01. ANOVA = analysis of variance; g = gram; J = joule; kJ = kilojoule; P = protein; SEM = Standard error of the mean; VWR = voluntary wheel running.

In follow-up experiments, we aimed to examine the relative energy budgets for active and resting energy expenditure over time in response to dietary protein dilution, and whether this was affected by body weight differences. To this end, we examined the effects of dietary protein dilution on daily energy expenditure budgeting through indirect calorimetry following relatively short-term (i.e., 2 weeks) or long-term (12 weeks) diet feeding. Importantly, mice had voluntary access to a running wheel and we could then calculate the cumulative energy expenditure both when the mouse was off and on the running wheel.²¹ Although there was no substantial difference in body weight after 2 weeks of feeding, by Week 12, there was a ∼10 g difference in body weight (Fig. 2E) due to substantial weight gain of the mice on the control diet in contrast to the mice subjected to DPD. In Week 2, DPD mice exhibited a ∼10% higher off-wheel energy expenditure with no difference in on-wheel energy expenditure (Fig. 2F and 2G). Cumulatively, DPD mice showed a higher total energy expenditure at Week 2 (Fig. 2H). By Week 12, the control diet-fed mice had a higher off-wheel, on-wheel, and total energy expenditure (Fig. 2F–2H), likely due to their higher body weight (Fig. 2E) compared to the 2-week time point. In contrast, while DPD mice continued to show higher off-wheel energy expenditure, they did not increase their on-wheel energy expenditure, leading to no change in total energy expenditure over time and a similar total energy expenditure as the control diet mice at Week 12 (Fig. 2F–2H). At Week 12, DPD mice had a 12%:88% on-wheel to off-wheel energy budget ratio compared to a 14%:86% ratio for control mice (p < 0.001). Taken together, this demonstrates that chronic DPD-induced reductions in body weight gain result in a relatively lower budgeting of total energy expenditure to movement, even with a higher resting energy expenditure.

We next wanted to test whether the effects on physical ability were related to the diet itself or rather the effects of the chronic adaptations to the protein dilute diet. To this end, we conducted a study in mice after 10 days of diet feeding, resulting in no difference on body weight (Supplementary Fig. 4A). Accordingly, there we no differences in weight-bearing exercise ability (Supplementary Fig. 4B and 4C), despite clear adaptations to DPD⁷ such as lower (∼30%) serum urea (Supplementary Fig. 4D) and higher (∼8-fold) levels of serum Fgf21 (Supplementary Fig. 4E). Thus, the effects on weight-bearing exercise ability and movement economy appear to be directly related to differences in body weight. We and others have previously reported that DPD induces metabolic inefficiency, leading to less gain in body weight.^7,8,29,31, 32, 33 Again, here we could observe that there was a relative metabolic inefficiency with chronic dietary protein dilution as assessed by body energy gain, which was determined by changes in body fat and lean mass, and cumulative food energy intake (Supplementary Fig. 5A–5H). The absence of a body weight difference in the acute study enabled us to assess the effects on energy expenditure without the confounder of altered body weight,³⁴ and indeed energy expenditure was substantially (∼15%) higher following acute DPD (Supplementary Figs. 4F and 5I).

3.2. Sex-dependent differences in physical ability with DPD are primarily related to body weight

As phenotypic responses to DPD can depend on biological sex,^11,35 we sought to determine whether the effects of DPD on body weight and related physical capability were different in male vs. female mice. While there was a slightly lower (∼10%) food intake rate in females vs. males, the expected¹² increase (5%–7%) in food intake with DPD occurred in both male and female mice (Fig. 3A; Supplementary Fig. 6A and 6B). Despite this slight difference in food energy intake, there were large differences in body weight and composition between sexes after chronic DPD. Indeed, male mice gained a substantial amount of weight on the control diet versus females, and, DPD male mice gained far less weight (∼25%) than their control diet counterparts (Fig. 3B; Supplementary Fig. 6C and 6D). This difference in body weight in response to DPD in male mice was most related to reduced adiposity and not lean tissue (Fig. 3C and 3D; Supplementary Fig. 6E–6J), and this was thus reflected in changes in body energy amounts (Fig. 3E). To gauge chronic changes in energy expenditure in these mice, feed efficiency was calculated, and feed efficiency was clearly lower (∼33%) in female compared with male mice on the control diet mice and not influenced substantially by DPD (Fig. 3F). In contrast, male mice were clearly more feed efficient when on the control diet, and this gain was substantially lower (∼33%) during DPD (Fig. 3F). The measures of physical capability were performed at the end of the experimental period and female mice on the control diet were clearly more capable in terms of treadmill running (∼50%), wire hang (∼50%) and relative grip strength (∼25%) (Fig. 3G–3J), probably owing to their lower body mass. Males had worse physical capability when on the control diet, which was improved with DPD (Fig. 3G–3J), again closely reflecting the differences in body weight. Taken together, these results further exemplify that body weight is a clear determinant of weight-bearing exercise ability.

Fig. 3.

Effect of chronic dietary protein dilution on behavioral, biometric and exercise ability parameters in male and female mice. (A) Cumulative food energy intake over a 20-week period of feeding a normal (20%P) or protein-dilute (5%P) in male and female C57Bl/6 J mice. (B) The change in total body weight of the mice. (C) The change in fat weight of the mice. (D) The change in fat-free weight of the mice. (E) The change in body energy amount of the mice. (F) Feed efficiency of the mice. (G) Time to fatigue during steady-state treadmill exercise at Week 19 of the mice. (H) Time to fall during a wire hang test at Week 19 of the mice. (I) Absolute grip strength at Week 19 of the mice. (J) Relative (body weight) grip strength at Week 19 of mice. Data are presented as mean ± SEM, n = 7–8/group. Statistical tests: (A–J): two-way ANOVA with Holm-Sidak post hoc tests. Diet group difference: ** p < 0.01, *** p < 0.001. Sex difference: ^# p < 0.05, ^## p < 0.01, ^### p < 0.001. ANOVA = analysis of variance; g = gram; kJ = kilojoule; mg = milligram; MJ = megajoule; N = newton; P = protein; SEM = standard error of the mean.

3.3. Chronic DPD does not substantially affect skeletal muscle

It is well established that muscle mitochondrial volume relates to exercise ability.^36,37 Hence, we assessed this by measuring citrate synthase activity, a marker of mitochondrial density.³⁸ Interestingly, while there was no difference in heart citrate synthase activity, gastrocnemius skeletal muscle citrate synthase activity was ∼15% lower in mice adapted to DPD (Fig. 4A and 4B). Given the additional difference in gastrocnemius weight (Supplementary Fig. 2A), we then chose to examine this muscle histologically using whole-mount cross-sectional imaging of fiber size and types (Supplementary Fig. 7A and 7B). While there was no difference in the total number of muscle fibers (Supplementary Fig. 7C and 7D), the mean fiber size was slightly smaller with DPD (Fig. 4C). Fiber type analysis revealed no differences (Fig. 4D), indicating that the reduced volume of muscle mitochondria and fiber size was independent of muscle fiber type switching. In any case, there was no significant difference in skeletal muscle histology and biochemistry that might then explain the improved physical capacity with DPD consistent with no absolute changes in grip strength ability (Supplementary Fig. 2B).

Fig. 4.

Dietary protein dilution does not substantially affect skeletal muscle but promotes brown fat thermogenesis. (A) Heart citrate synthase activity of male C57Bl/6 J mice after 12 weeks of feeding a normal (20%P), or protein-dilute (5%P) diet. Data are presented as mean ± SEM. n = 6/group. (B) GCM citrate synthase activity of mice as in A. (C) Mean muscle fiber size per cross-sectional are as determined from images Supplementary Fig. 7A. (D) Fiber type ratios per cross-sectional are as determined from images in Supplementary Fig. 7A and 7B. (E) Violin plots of BAT mRNA expression of select brown fat markers of mice adapted for 3 weeks to a protein-dilute (5%P) or matched control (20%P) diet. n = 8/group. (F) Violin plots of scWAT mRNA expression of select brown fat markers of mice as in A. (G) BAT histology of mice as in A. Stains: H&E and UCP1. Shown are representative images from 5 individual mice. Scale bar = 200 µm. (H) Subcutaneous white adipose tissue histology of mice as in C. (I) Time-course of telemetry-based BAT temperature of mice fed a protein-dilute (5%P) or matched control (20%P) diet. (J) Hourly circadian BAT temperature averaged from Days 6–10 of mice as in E. Statistical tests: (A–F): Students t tests; (E–F): two-way ANOVA with Holm-Sidak post hoc tests. Diet group difference: * p < 0.05, ** p < 0.01. ANOVA = analysis of variance; AU = arbitrary units; BAT= brown adipose tissue; Cidea = Cell death-inducing DNA fragmentation factor alpha; Cox7a1 = cytochrome c oxidase subunit 7a1; Cox8b = cytochrome c oxidase subunit 8b; g = gram; GCM = Gastrocnemius complex skeletal muscle; H&E = hematoxylin and eosin; P = protein; Ppargc1a = peroxisome proliferator receptor gamma coactivator 1 alpha; scWAT = subcutaneous white adipose tissue; SEM = standard error of the mean; UCP1 = uncoupling protein 1; ZT = zeitgeber time in hours.

3.4. DPD induces BAT activity independent of changes in adiposity

We questioned what mechanisms might promote the improved movement economy, reduced body weight, and increased basal energy expenditure. Since the activation of adipose tissue “browning” occurs with DPD,^7,31,33,39 and BAT is critical for adaptive non-shivering thermogenesis,²⁴ we re-assessed this using a combination of techniques. We could demonstrate an increase (∼1-fold) in Uncoupling protein 1 (Ucp1) mRNA levels in intra-scapular BAT of mice adapted to DPD (Fig. 4E). In addition, we observed a profound increase in brown fat markers (∼60-fold higher Ucp1) in subcutaneous white adipose tissue (scWAT) of DPD mice (Fig. 4F). Furthermore, UCP1 expression was upregulated in both BAT and scWAT as assessed histologically (Fig. 4G and 4H). Lastly, to confirm the higher BAT activity, we performed studies using simultaneous thermosensing and movement telemetry in cage housing conditions. Importantly, we could show a typical circadian rhythm in BAT temperature (Fig. 4I and 4J) and activity (Supplementary Fig. 8). With a time-course assessment, BAT temperature was higher 1.5 days after the onset of exposure to the DPD and remained ∼0.3°C elevated for the 14 days period of study (Fig. 4I). The higher BAT temperature with DPD was evident in both the light and dark cycles (Fig. 4J) and was independent of physical activity, which was not different (Supplementary Fig. 8).

Since a reduction in thermal insulation by fat may lead to an increase in adaptive thermogenesis,⁴⁰ we aimed to test whether DPD-induced thermogenesis was linked to reduced adiposity. To achieve this, we conducted time-course experiments to examine body/adipose weight, BAT activation and energy expenditure following the initiation of the DPD. In particular, in a continuous experiment, we observed that increases in both BAT temperature (Supplementary Fig. 9A), and energy expenditure (Supplementary Fig. 9B) occurred at about 3–4 days after the start of DPD, well before any detectable difference in body weight (Supplementary Fig. 9C). In a separate cohort study design, there was higher BAT Ucp1 and Cox8b mRNA expression (Supplementary Fig. 9D and 9E), as well as subcutaneous white adipose tissue Ucp1 and Cox8b mRNA expression (Supplementary Fig. 9F and 9G), 3.5 days after the diet began, at a time when there was no change in body weight (Supplementary Fig. 9H) or adiposity (Supplementary Fig. 9I). Taken together, these findings demonstrate that DPD induction of adipose tissue thermogenesis occurs independently of any changes in adiposity. On the contrary, the reduced adiposity is likely a result of the induction of adipose thermogenesis and energy expenditure with DPD.

3.5. Hepatocyte-derived Fgf21 connects heightened BAT activity and improved movement economy with chronic DPD

We then investigated the potential mechanisms underlying the increased BAT activity during DPD. Others have reported that the elevated levels of the peptide hormone Fgf21 affect the “browning” of adipose tissue^7,33,39 and the heightened energy expenditure^7,8,33 with DPD. Indeed, when assessing mRNA levels of browning markers in BAT and scWAT, the increased expression in response to DPD was absent in the in whole-body Fgf21 KO mice (Fig. 5A–5C). When examining BAT activity through thermosensing in vivo, the whole-body Fgf21 knockout mice were refractory to the effects of DPD for increasing BAT temperature (Fig. 5D), with no differences in physical activity (Supplementary Fig. 10A and 10B). Furthermore, as a measure of enhanced metabolic activity, we assessed in vivo glucose uptake in hepatocyte-specific Fgf21 KO mice and demonstrated that the higher BAT and scWAT glucose uptake with DPD was completely abolished in the KO mice, with no effects on muscle glucose uptake (Supplementary Fig. 10C–10E).

Fig. 5.

Fgf21 is required for brown adipose tissue activation and thermogenesis during dietary protein dilution. (A) Violin plots of brown adipose tissue mRNA expression of Ucp1 in Fgf21+/+ and Fgf21−/− male mice fed a protein-dilute (5%P) diet or matched control (20%P) diet. (B) Violin plots of scWAT mRNA expression of Ucp1 and cell death inducing DFFA like effector a (Cidea) (n = 6/group). (C) BAT histology of Fgf21+/+ and Fgf21−/− male mice fed a protein-dilute (5%P) diet or matched control (20%P) diet for 3 weeks. Stains: H&E and UCP1. Shown are representative stain images from 2 individual mice. Scale bar = 100 µm. (D) Hourly circadian telemetry-based BAT temperature of Fgf21+/+ and Fgf21−/− male mice fed a protein-dilute (5%P) diet or matched control (20%P) diet. Data are those from time points averaged over Days 6–10 post diet switching (n = 6/group). Statistical tests: (A and B): two-way ANOVA with Holm-Sidak post hoc tests; (D): two-way repeated measures ANOVA with Holm-Sidak post hoc tests. Diet group difference: * p < 0.05. Genotype difference: ^# p < 0.05. ANOVA = analysis of variance; AU = arbitrary units; BAT = brown adipose tissue; DFFA = DNA fragmentation factor alpha; Fgf21 = fibroblast growth factor 21; g = gram; H&E = hematoxylin and eosin; P = protein; scWAT = subcutaneous white adipose tissue; Ucp1 (UCP1) = uncoupling protein 1; ZT = zeitgeber time in hours.

Given that Fgf21 simultaneously affects energy expenditure^7,8 and browning (Figs. 4 and 5) with DPD, this then provided an opportunity to assess whether DPD induced BAT thermogenesis and metabolic inefficiency would affect exercise ability and movement economy. To this end, we employed both whole-body Fgf21 KO mice and hepatocyte-specific Fgf21 KO mice and chronically adapted them to DPD prior to testing. Importantly, the induction of serum levels of Fgf21, basally and in response to DPD, was completely abolished in both mouse models (Supplementary Fig. 11A and 11B). In both models, the effect of chronic DPD on lower body weight was abolished (Fig. 6A and 6B; Supplementary Fig. 11C and 6D), and this was mostly attributable to effects on adiposity (Supplementary Fig. 11E and 11F). Concordant with the effects on body weight, both models of Fgf21 loss-of-function demonstrated that they were refractory to the effects of DPD to improve treadmill running endurance (Fig. 6C and 6D) and wire hang ability (Fig. 6E and 6F). These effects coincided with Fgf21 knockout mice having a lack of improved running economy with DPD (Fig. 6G and 6H).

Fig. 6.

Liver-derived Fgf21 is required for the improved exercise ability and economy with chronic dietary protein dilution. (A) Body weight of male in Fgf21+/+ and Fgf21−/− mice after 12 weeks of feeding a protein-dilute (5%P) diet or matched control (20%P) diet (n = 6/group). (B) Body weight of male Fgf21fl/fl mice pretreated with either AAV8-GFP or AAV8-CRE and then fed a protein-dilute (5%P) diet or matched control (20%P) diet for 12 weeks (n = 8/group). (C) Time to fatigue during a treadmill exercise test of mice as in A. (D) Time to fatigue during a treadmill exercise test of mice as in B. (E) Time to fall during a wire hang test of mice as in A. (F) Time to fall during a wire hang test of mice as in B. (G) Running economy during a fixed treadmill exercise test of mice as in A. (H) Running economy during a fixed treadmill exercise test of mice as in B. Data are presented as mean ± SEM. Statistical tests: (A–H): two-way ANOVA with Holm-Sidak post hoc tests. Diet group difference: *** p < 0.001. Genotype difference: ^### p < 0.001. AAV = adeno-associated viruses; ANOVA = analysis of variance; Fgf21 = fibroblast growth factor 21; g = gram; J = joule; P = protein; SEM = standard error of the mean; W = watt.

In summary, liver-derived Fgf21 effects on BAT thermogenesis to lower adiposity drive improved movement economy and weight-bearing exercise ability during chronic dietary protein dilution (Supplementary Fig. 12).

4. Discussion

Here we demonstrate that the effects of liver-derived Fgf21 on BAT thermogenesis and energy expenditure enhance weight-bearing exercise ability and movement economy following chronic DPD (Figs. 1 and 2). This reinforces the role of Fgf21 in the controversial^41,42 phenomenon of diet-induced thermogenesis, as previously indicated,^7,31 and provides an ecological rationale for the inherent existence of diet-induced thermogenesis (Supplementary Fig. 12).

One can envision that the reduction of body weight and subsequent improved movement economy of physical activity confers advantages from an ecological perspective. Our findings show that mice adapted to DPD exhibited improved weight-bearing exercise capability potentially enhancing their aptitude to avoid predation. This adaptation also leads to less of the total energy expenditure “budget” being allocated to physical activity (Fig. 2), which would be particularly important in protein-deplete environments where the animal may need to search extensively to find a protein food source.

Importantly, DPD-induced thermogenesis appears to be present in humans as well,³² albeit with a wide variability.^43,44 Understanding the reason for this individual variability is crucial, as susceptibility to weight gain is important for understanding human susceptibility to obesity 44, 45, 46 and correlates with the ability to activate Fgf21 and energy expenditure through DPD.⁴³ This becomes particularly evident when comparing sex-differences in responses to DPD, as female mice have been shown to exhibit a weaker response to DPD compared to males.^9,11,35 Additionally, we could also demonstrate that female mice gain far less weight than their male counterparts when fed the control diet with consequent effects on their weight-bearing exercise ability (Fig. 3). When male mice were chronically exposed to DPD, they exhibited similar characteristics to female mice, such as lower feeding efficiency, reduced body weight gain, and consequently enhanced weight-bearing physical ability, again highlighting the close relationship between body weight and physical capability. Another way to explore the source of such variability would be the use of mouse strain panels,⁴⁷ especially as caloric restriction effects vary widely among different strains of mice.^48,49 Nonetheless, it is already established that DPD inconsistently induces energy expenditure and reduces body weight and in male mice across various strains,^10,11 yet whether the effects on movement economy differ according to body weight changes between strains remains unknown. Such studies, when combined with genomic, transcriptomic or proteomic analyses, could potentially identify inherent factors contributing to such variable responses.

Notably, while our work enables invasive and genetic experimentation, a limitation is that it was conducted exclusively in mice. Whether this interplay between diet and exercise energy balancing exists in other species is currently unclear. A promising avenue for future research would involve mammals that engage in inefficient forms of activity, such as flight. For example, the fruit bat consumes a very high carbohydrate but low-protein diet, and interestingly, it possesses BAT despite living in tropical climates, suggesting that it probably never requires this tissue for maintaining core body temperature in colder environments but rather may be beneficial for exercise/flight economy.³² In future, diet–exercise interplay studies should also be extended to humans, as body weight is a clear determinant of weight-bearing human movement economy,^14,15,50,51 and the increase in Fgf21^7,8,43,52 and energy expenditure^43,52 with DPD is conserved in humans. While the therapeutic effects of Fgf21 on metabolic health are being explored in humans,⁵³ whether dietary protein dilution and Fgf21 affects body weight and consequent exercise ability has not been studied, and our observations from mouse studies here should spur further research into this direction.

Our studies clearly indicate that liver-derived Fgf21 is required for DPD’s effects on BAT thermogenesis, adiposity, body weight, and movement economy (Figs. 5 and 6). This aligns with the finding that core body temperature increases with DPD,⁷ and highlights that UCP1, a protein almost exclusively expressed in brown and beige fat cells and required for uncoupled respiration,⁵⁴ is essential for DPD’s promotion of energy expenditure.³³ While a number of studies have consistently shown that Fgf21 is responsible for DPD’s effects on adiposity reduction,^{7,11,39,55,56} our findings contradict Hill et al.³³ who concluded that Fgf21 is not responsible for thermodynamic effects associated with DPD. However, their study focused on core body temperature in responses to cold,³³ rather than the circadian BAT temperature measured in our study, which may explain this discrepancy. Fgf21 can signal to a multitude of tissues that express both Fgf receptor 1 and the co-receptor β-klotho.⁵³ Yet, studies using neural-specific β-klotho silencing in mice have demonstrated the necessity of brain-neuron Fgf21 signaling for DPD’s effects on energy expenditure and its associated impacts on body weight and adiposity effects.^31,55 Indeed, we have shown that Fgf21 is responsible for inducing both BAT glucose (Supplementary Fig. 11) and fatty acid²⁹ metabolism with DPD, as well as enhanced browning of white adipose tissue (Fig. 5). This brain-adipose pathway is probably mediated by the sympathetic nervous system (SNS), as brain Fgf21 action activates the SNS,^57,58 and SNS inactivation by a broad spectrum β-adrenoreceptor inhibitor blunts the increase in energy expenditure with DPD.⁵⁹ Taken together, these studies demonstrate that a liver-brain-adipose tissue axis is involved in the metabolic adaptations to DPD, ultimately affecting body weight and exercise ability. Whether BAT thermogenesis fully accounts for the body weight, and downstream movement economy, differences with dietary protein dilution remains unknown and is a clear direction for future studies.

Our results also highlight that diet-induced thermogenesis is distinct from meal/postprandial thermogenesis. We show here, using carefully conducted time-course experiments, that diet-induced thermogenesis increases after persistent feeding (>48 h) which aligns with other findings regarding the induction of energy expenditure happening within 3–5 days after the onset of feeding a protein-dilute diet^7,8,33 (Supplementary Fig. 9). This contrasts with meal-induced thermogenesis which occurs within minutes after feeding, and is likely a result of the combined thermal effects of food (i.e., heat generated from chemical digestion) as well as heightened metabolic processes within the body, whether they are neural-controlled or otherwise.⁶⁰

5. Conclusion

We show that dietary protein dilution pivots an energy trade-off between resting thermogenesis and movement energy expenditure, via Fgf21 driven BAT thermogenesis.

Authors’ contributions

PMR and OHA performed mouse and wet lab experiments, performed data analysis and visualization, and wrote and edited the manuscript; AYC, YW, MASC, MBL, and EG performed mouse experiments; GTD performed mouse experiments and edited the manuscript; AS performed mouse experiments, data analysis, provided resources and edited the manuscript; AJT, SG, AF, and SAF performed wet lab experiments; SH and OJM provided resources; MJM provided resources and supervision and edited the manuscript; AJR conceived and managed the study, performed mouse experiments, performed data analysis and visualization, provided resources and supervision, and wrote and edited the manuscript. All authors have read and approved the final version of the manuscript, and agree with the order of presentation of the authors.

Declaration of competing interest

AJR and PMR have received research funding from Boehringer-Ingelheim for studies unrelated to the work shown here. GTD is the Founder and Chief Scientific Officer of Gallant Bio, a biotechnology company developing targeted therapies for obesity and type 2 diabetes. GTD is the Founder and Chief Scientific Officer of NewCo Bio, which focuses on developing G-protein coupled receptor (GPCR) therapeutics for metabolic disorders. Additionally, GTD is an inventor on patents related to neurofibrosis-targeting agents for restoring metabolic control, extracellular matrix (ECM)-targeted co-therapies to enhance glucagon like peptide 1 (GLP-1) receptor agonist efficacy, and a novel peptide therapeutic for an uncharacterized GPCR linked to energy balance. GTD has received funding and translational support from Novo Nordisk, Eli Lilly, and Gallant Bio for work unrelated to the work shown here. All the support had no involvement in the study design and writing of the manuscript or the decision to submit it for publication. The other authors declare no conflicts of interest.

Data availability

The data that support the findings of this study are within the main or supplementary figures. All correspondence and requests for materials should be made to the corresponding author AJR.

Acknowledgments

The authors thank Dr. Robert Lee (Monash Metabolic Phenotyping Platform) for discussions around initial planning and conduct of these studies. This work was supported by the Monash Animal Research Platform, the Monash Metabolic Phenotyping Platform, and the Monash Histology Platform. These studies were supported by Monash Biomedicine Discovery Institute laboratory start-up funds to AJR. The authors dedicate this paper to the memory of Prof. David H. Wasserman, a pioneer on the study of liver and exercise biology.