SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Objective:

Protein leverage plays a role in driving increased energy intakes that may promote weight gain. The influence of the protein to carbohydrate ratio (P:C) in diets of C57BL/6J mice on total energy intake, fat storage, and thermogenesis was investigated.

Design and Methods:

Male mice (9 weeks old) were provided ad libitum access to one of five isocaloric diets that differed in P:C. Food intake was recorded for 12 weeks. After 16 weeks, white adipose tissue (WAT) and brown adipose tissue (BAT) deposits were dissected, weighed, and the expression levels of key metabolic regulators were determined in BAT. In a separate cohort, body surface temperature was measured in response to 25 diets differing in protein, fat, and carbohydrate content.

Results:

Mice on low P:C diets (9:72 and 17:64) had greater total energy intake and increased WAT and BAT stores. Body surface temperature increased with total energy intake and with protein, fat, and carbohydrate, making similar contributions per kJ ingested. Expression of three key regulators of thermogenesis were downregulated in BAT in mice on the lowest P:C diet.

Conclusions:

Low-protein diets induced sustained hyperphagia and a generalized expansion of fat stores. Increased body surface temperature on low P:C diets was consistent with diet-induced thermogenesis (DIT) as a means to dissipate excess ingested energy on such diets, although this was not sufficient to prevent development of increased adiposity. Whether BAT was involved in DIT is not clear. Increased BAT mass on low P:C diets might suggest so, but patterns of thermogenic gene expression do not support a role for BAT in DIT, although they might reflect failure of thermogenic function with prolonged exposure to a low P:C diet.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

In a recent study on mice, Sørensen et al. (1) established that decreasing the dietary ratio of protein to carbohydrate (P:C) increased total energy intake and body adiposity. When mice were offered pair-wise choices of nutritionally complementary foods differing in P:C, they adjusted food intake to maintain intake of protein and carbohydrate at a common ratio (1:2.5 P:C), reflecting an intake target for both nutrients. The self-selected diet contained 23% of total energy as protein (1). Together, these results show that mice have separate regulatory systems for protein and carbohydrate intakes, and that protein regulation is somewhat stronger than that for carbohydrate when the two nutrients are provided in unbalanced ratios. The partial dominance of protein intake regulation is an example of what is termed “protein leverage” (2), which is also seen to be in varying degrees in several other omnivorous and herbivorous animals, from insects to humans (1, 3-12).

Protein leverage has been proposed to play a role in driving the increased energy intake that has accompanied the growing prevalence of human obesity in developed and developing countries over recent decades (2, 11–13). It is hypothesized that increased energy intake resulting from protein leverage will promote the development of weight gain and obesity unless excess ingested energy absorbed is voided by increased activity or a process termed diet-induced thermogenesis (DIT) (14). DIT arises from an upregulation of energy-dissipating biochemical pathways associated with futile cycles and/or uncoupling of mitochondrial electron transport from adenosine-5′-triphosphate (ATP) synthesis (32), both of which result in the dissipation of ingested energy as heat. These processes are separate from the thermic effects of processing ingested food.

That DIT occurs is well established, but the tissue sites where it occurs remain contentious; in particular the involvement of brown adipose tissue (BAT) (see review in Ref. 16). BAT is a specialized thermogenic tissue that has been implicated in DIT (17, 18). In mice, BAT is concentrated in the thorax, including the interscapular region, heart, and mediastinum (15), but it is present at lower levels in the abdomen, brain, and extremities.

If DIT is highly efficient, an animal could ingest excess energy to gain limiting protein without gaining in fat mass. Stock (14) suggested that DIT evolved “as a mechanism for enriching nutrient-poor diets by disposing of the excess non-essential energy.” Suggestively, the mechanisms for DIT are most highly developed in animals adapted to a habitually low-protein diet, such as nectar- and fruit-eating bats and marmosets (14), and in rodents hyperphagia on cafeteria diets, which are typically lower in protein, stimulates DIT (14, 19, 20).

In this study, first we sought to confirm the presence of protein leverage in mice by comparing total energy intakes on diets varying in P:C. Then we related energy intakes to differences in the mass of visceral white adipose tissue (WAT) pads and interscapular BAT. As a simple direct measure of thermogenic output that might be indicative of DIT, we investigated the impact of P:C ratio on body surface temperature. To establish whether BAT might be involved in DIT, we measured the impact of dietary P:C on BAT gene expression, with particular focus on genes that modulate energy expenditure via DIT.

Methods and Procedures

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Animals and diet treatments

This study was approved by the Animal Ethics Committee of the University of Sydney (AEC Ref. No.: L04/12-2007/3/471350). Separate groups of male C57BL/6J mice were used to measure intake, growth, body composition, and thermogenic enzyme activity (group 1) and thermogenesis (group 2). For group 1, 50, eight-week-old mice were purchased from the Animal Resources Centre (Perth, WA). Mice were housed individually in environmentally controlled conditions (temperature 22°C, light cycle from 07:00 to 19:00 h and dark cycle from 19:00 to 07:00 h). Mice were allowed ad libitum access to food and water throughout the study. Tissue and carton rolls were provided for nest building and environmental enrichment.

Five custom diets were designed based on AIN-93M (21) and manufactured by a local stockfeed company (Gordon's Specialty Stockfeeds, Yanderra, NSW). All foods were presented as dry pellets (diameter 1.5 cm × length 5 cm) and were isocaloric in energy density, varying in protein to carbohydrate composition only (Supporting Information Table S1). Fifty mice were allocated among five groups (N = 10) and fed with standard laboratory chow for the first week after delivery to allow them to adapt to the new environment. From the second week, mice were fed one of the five experimental diets.

Accurate measures of food intake across 2-day periods were derived from the difference between the mass of food provided and the mass remaining after 2 days, including spillage. Food and spillage were weighed on an electronic balance (Mettler Toledo) to within 0.1 mg. Food spillage was recovered using a no-spill mouse cage insert (1). Food intake was measured until 12 weeks after entry into the experiment, by which time mice were 21 weeks old. From 12 to 16 weeks on experimental diets (21-25 weeks of age), food intake was no longer recorded due to mice having settled into a stable daily intake pattern. Body weight (BW) was recorded to within 0.1 mg throughout the 16 weeks, using a dynamic weighing module on an electronic balance (Mettler Toledo). At the end of the experiment (week 16, 25 weeks of age), all mice were fasted for 3 h and then administered a lethal dose of sodium pentobarbitone (120 mg/kg, ip). All mice were culled between 1,000 and 1,300 h to minimize circadian variation.

The effects of dietary P:C ratio on thermogenesis were measured using a larger group of 25-week-old mice (group 2 = 229 mice) confined to a wider range of diets (25 diets) than group 1 (see Supporting Information Table S2). This enabled comprehensive determination of the effects of diet on thermogenesis using response surface methodology (see below). These mice were part of a separate ongoing experiment and thus were not euthanized. Mice were housed in groups of three animals per cage and were maintained at 23.0 ± 0.2°C under a light–dark (LD) 12:12 photoregime with ad libitum access to water and one of the treatment diets that varied in the concentration and ratio of protein, digestible carbohydrate, and fat. Animals were subjected to one of the diet treatments from 3 weeks of age. Accurate food intake and BW measurements were taken weekly using the aforementioned same procedures for group 1.

BWs, body lengths, and fat pad weights

Final BWs were measured to within 1 mg. BW gain was calculated by difference between week 16 and week 0 in the experiment (ages 25 and 9 weeks, respectively). Body lengths (BLs) were derived by measuring from nose to anus of supine animals during the muscle relaxation stage in the euthanasia. After the animals had died, the epididymal (Epi) and retroperitoneal (Retro) white fat pads, and interscapular BAT pads were removed immediately and weighed to within 0.1 mg.

Reverse transcription real-time PCR

The mRNA levels of several key regulators in BAT were measured by real-time PCR. After the mice were culled, the interscapular BAT pads were immediately removed, weighed, wrapped in foil, and snap-frozen in liquid nitrogen, then stored at −80°C. We chose to measure the following thermogenic genes: uncoupling protein 1 (UCP1), a mitochondrial membrane protein that decreases the efficiency of oxidative phosphorylation and thereby promotes heat production; deiodinase 2 (DIO2), which converts thyroxine (T4) to its more potent analog 3,3′,5-triiodothyronine (T3); and peroxisome proliferator-activated receptors, γ isotype (PPARγ) coactivator 1 alpha (PGC1α), which is a transcriptional coactivator of PPARs involved in the induction of both UCP1 and DIO2, as well as mitochondrial biogenesis.

For quantification of the steady-state RNA levels, total RNA was extracted from homogenates of BAT samples according to the manufacturer's instructions (Trizol® Reagent, Invitrogen). The ratio of absorbance at 260 and 280 nm was used to assess the purity of RNA (Nanodrop®). The integrity of RNA samples was verified by denatured gel electrophoresis. Thirty high-quality RNA samples (six biological samples from each group), were selected and treated with DNase I (RQ1 RNase-Free DNase, Promega). cDNA first strands were prepared by reverse transcriptase according to the manufacturer's instructions (SuperScript® VILO™ cDNA Synthesis Kit, Invitrogen). For quantitative real-time PCR, primers were designed for AKT2, CEBPα, DIO2, GLUT4, HSL, IRS1, PGC1α, PIK3R1, PPARα, PPARγ, PRDM16, and UCP1 using Pimer3 (http://frodo.wi.mit.edu/primer3/). Primer sequences are listed in Supporting Information Table S3. Quantitative Ct values were determined using a 7500 Fast Real-Time PCR System (Applied Biosystems) and Sybr Green detection. Mouse ribosomal 18S RNA was used as an endogenous control. Reverse transcriptase (−) was checked for genomic DNA contamination. Melting curve analysis was done for each reaction. For experimental results, ΔCt values were normalized to 18S and expressed as copy number per 1,000,000 18S.

Measurement of body surface temperature

At 25 weeks of age, a thermographic technique (22) was used to determine body surface temperatures of unshaved mice on each of the 25 diets. Each mouse was placed unrestrained in a plastic box (10 cm long × 10 cm wide × 20 cm high), which was well ventilated and the top was open. An infrared CCD camera (ThermaCam S65 PAL, FLIR Systems) was placed 20 cm above the box and a single picture was taken for each mouse twice during the 24-h cycle—once in the middle of the light phase and the other in mid-dark phase (i.e., 12:00 and 14:00; 24:00-02:00), while mice had ad libitum access to food. The box was wiped after delivering a 70% ethanol spray between mice. The surface temperatures of three body regions, that is, head (excluding the eyes) together with the central regions of the upper and lower back were determined by image analysis (ThermaCAM™ Researcher—version 2.9, FLIR Systems). The temperatures of the head and back regions were combined as trunk temperature (Ttrunk).

To visualize temperature differences in relation to diet composition, nonparametric thin-plate splines were fitted and response surfaces were visualized using the fields package in R (version 2.5.1) (23). Trunk temperatures for the three mice within each cage were averaged and these mean values were used to construct temperature-response surfaces. Surfaces were fitted onto nutrient intake arrays for protein, fat, and carbohydrate eaten per mouse per cage. These intake values were derived from weekly measurements of food intake recorded at the cage level and averaged across the period 12-24 weeks after commencement of the study. Food intake remained stable on each diet across this period.

Data analysis

Cumulative nutrient intakes for five diets were presented as bi-coordinate plots and analyzed according to the geometric framework (GF) (7, 8). Body mass growth and fat pad weights were compared using ANOVA, followed by post hoc pair-wise comparison (Tukey HSD test). Gene expression data were analyzed using ANOVA. For all tests, a significance level of 5% was adopted. The effects of nutrient intakes (ingested protein, carbohydrate, and fat and their interactions) on trunk temperature (visualized using thin-plate spline procedures in R) were tested using Repeated Measures General Linear Model (GLM) in PASW (formerly SPSS) Statistics, v. 18, with day and night temperatures providing the within-subject factor.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

The effect of dietary P:C ratio on energy intake and BW

Male C57BL/6J mice responded to changes in dietary P:C ratio in a manner closely similar to that reported in the previous study on NMRI mice (1), prioritizing intake of protein over carbohydrates when confined to imbalanced diets. Accordingly, energy intake was negatively correlated with dietary protein content, although protein leveraging was not complete, as was also found previously (1). Thus, mice restricted to low P:C diets exhibited excess energy intake when compared when mice restricted to medium or high P:C diets (Figure 1).

thumbnail image

Figure 1. (A) Nutrient intakes and protein leverage effects in mice maintained for 12 weeks (ages from 9 to 21 weeks) with ad libitum access to one of the five isocaloric diets differing in P:C. Mice were continued for a further 4 weeks without intake being measured before being dissected. The black dots on the dashed radials (intake rails) represent the cumulative nutrient intakes (g); the dotted lines with inclination of −1 represent isocaloric lines and the shadowed area indicates increased energy intake in mice fed low P:C diets, relative to the 23:57 diet—i.e., the protein leverage effect. (B) Differences in total energy intake per day over 12 weeks.

Download figure to PowerPoint

Mice on the near-balanced diet [P:C = 23:57; see (1)] had the greatest increases in body mass, whereas mice on the high P:C diet (48:33) had the lowest weight gains (Figure 2). Mice on the lowest P:C diets (9:72) exhibited excess weight gains after week 4 and the effects were most marked at week 16. Mice on the 23:57 diet were maximal in BL at 16 weeks (Supporting Information Table S4).

thumbnail image

Figure 2. BW gain at week 4, week 8, week 12, and week 16 (ages 13, 21, and 25 weeks) in mice maintained for 16 weeks with ad libitum access to one of the five isocaloric diets differing in P:C. At week 4, the 17:64 diet group had the greatest BW gain (significantly different from the 48:33 group), while the 9:72 and 48:33 had similarly low growth, representing a reverse U-shaped response to dietary P:C. The mice fed the 23:57 diet had gained the most weight at 12 and 16 weeks. The mice fed the 48:33 diet had gained the least weight at 8 weeks in comparison to mice fed 17:64 and 23:57 and at 12 weeks in comparison to the mice fed the 23:57 diet, the mice fed the lowest P:C ratio (9:72) gained more weight by 8, 12, and 16 weeks in comparison to the mice fed the 48:33 diet. *P < 0.05 for post-hoc pairwise tests following significant ANOVA (see text).

Download figure to PowerPoint

The effect of P:C ratio on the masses of various fat pads

To determine whether the observed increases in body mass on the lowest P:C diet relative to the highest P:C diet were linked to changes in fat storage, selected fat pads were weighed immediately after euthanasia (Supporting Information Table S4). Body fat stores were negatively correlated with dietary protein content, such that mice on the lowest P:C diet had the greatest depots of visceral fat pads, including epididymal white fat (F4,41 = 2.79, P = 0.04), retroperitoneal white fat (F4,41 = 4.06, P = 0.007), and interscapular BAT (F4,38 = 4.81, P = 0.003). The masses of the fat depots in mice on the lowest P:C ratio (9:72) at all these sites were nearly double those observed in mice on the highest P:C ratio (48:33) (Figure 3).

thumbnail image

Figure 3. Increased fat storage in the mice fed low P:C diets. Epi, Retro, and interscapular BAT were collected from mice fed ad libitum on one of the five isocaloric diets differing in P:C for 16 weeks. *P < 0.05; versus the 9:72 for post-hoc pairwise tests following significant ANOVA (see text).

Download figure to PowerPoint

The effect of dietary P:C on the expression of key regulators of metabolism in BAT

Reverse transcription real-time PCR was used to evaluate changes in gene expression in the BAT of fasted mice following the five dietary treatments. Three key regulators of thermogenesis were upregulated in mice fed the highest P:C diet and downregulated in mice fed the lowest P:C diet. Thus, compared to the mid-point 23:57 P:C diet, UCP1 and DIO2 mRNA expression decreased by around 30% and 50%, respectively, on the lowest P:C (9:72) diet and increased by around 1.7-fold and 3.5-fold, respectively, on the highest P:C diet (Table 1). Similarly, there were substantial progressive increases in PGC1α and PPARα expression by around nine-fold and 4.5-fold, respectively, as dietary P:C ratio increased from the lowest (9:72) to the highest (48:33) ratio (Table 1). The impacts of P:C ratio on the expression of several metabolism-regulating genes that are unrelated to thermogenesis were also assessed, including IRS1, AKT2, GLUT4, HSL, PPARγ, and CEBPα (Table 1). With respect to insulin signaling, IRS-1 mRNA levels increased modestly and GLUT4 mRNA levels decreased modestly as P:C ratio increased. The expressions of other genes tested were unaffected by changes in dietary P:C ratio (Table 1).

Table 1. mRNA expression of thermogenic and insulin-signaling genes in BAT
#copies per 106 18S9:72, mean ± SEM17:64, mean ± SEM23:57, mean ± SEM31:50, mean ± SEM48:33, mean ± SEMANOVA, F4,25Sig.
AKT240.26 ± 1.9936.60 ± 4.2839.61 ± 3.5942.53 ± 3.9935.17 ± 3.100.710.59
CEBPA33.28 ± 4.1138.34 ± 3.5441.43 ± 2.0748.12 ± 2.9441.54 ± 4.742.250.09
DIO22.17 ± 0.233.18 ± 0.643.53 ± 0.669.03 ± 1.2712.40 ± 1.4920.140.00
GLUT488.07 ± 7.4281.67 ± 10.3553.05 ± 4.8646.02 ± 4.1248.04 ± 4.498.890.00
HSL526.17 ± 28.05490.62 ± 60.90481.15 ± 25.88468.95 ± 21.37388.48 ± 26.942.040.12
IRS15.45 ± 0.957.43 ± 0.547.45 ± 0.638.31 ± 0.788.76 ± 0.722.940.04
PGC1A19.01 ± 1.7720.41 ± 3.0282.91 ± 34.60197.57 ± 39.80172.60 ± 19.7510.980.00
PIK3R114.37 ± 1.5413.29 ± 1.8612.84 ± 1.7918.91 ± 3.1018.09 ± 3.501.280.30
PPARA72.73 ± 5.1376.84 ± 7.02188.06 ± 52.57244.02 ± 27.54331.30 ± 27.4614.120.00
PPARG52.12 ± 2.3941.41 ± 4.3849.01 ± 5.9154.66 ± 5.5149.70 ± 2.841.260.31
PRDM169.85 ± 0.718.59 ± 1.097.96 ± 1.078.97 ± 0.717.25 ± 0.631.300.30
UCP11,536.91 ± 201.881,497.49 ± 265.821,869.70 ± 473.002,853.06 ± 323.193,064.74 ± 239.955.560.00

The impact of macronutrient intake on body surface temperature

Pooled across all diets, surface temperatures in mice were lower during the light phase (30.4 ± 0.07°C; n = 227) than the dark phase (31.1 ± 0.06°C; n = 222), consistent with the diet pattern in core body temperature, which in ad libitum feeding mice switches from 36 to 37°C in the light phase to 37-38°C in the dark phase in association with increases in activity and oxygen consumption (24, 25). In addition, there was a significant between-group effect of total energy intake on surface temperature (F1,78 = 7.061; P < 0.01) such that surface temperature rose as a linear function of total energy intake. We next parsed total energy into protein and nonprotein energy (carbohydrate + fat) intakes. Neither protein (F1,77 = 0.062; P = 0.804) nor nonprotein intake (F1,77 = 0.004; P = 0.953) interacted with time of day in their associations with surface temperature. Nevertheless, both protein intake (F1,77 = 6.631; P = 0.012) and nonprotein intake (F1,77 = 13.753; P < 0.0005) were positively correlated with surface temperature, but the intake of nonprotein energy had a stronger association. Including an interaction between protein and nonprotein intake as an additional term in the repeated measures GLM yielded no significant improvement (F1,76 = 0.004; P = 0.948), suggesting that the contributions of protein and nonprotein energy intake on body surface temperature were additive.

We further parsed nutrient intake by separating non-protein energy into carbohydrate and fat eaten. When the model included protein, carbohydrate, and fat intake separately, none of the nutrients interacted significantly with time of day in affecting surface temperature (F1,76 = 0.124, P = 0.725; F1,76 = 0.322, P = 0.572; and F1,76 = 0.209, P = 0.649, respectively). The intakes of protein, carbohydrate, and fat all significantly influenced trunk surface temperature, with carbohydrate having the most significant association (protein: F1,76 = 7.171, P = 0.009; carbohydrate: F1,76 = 13.600, P < 0.0005; and fat: F1,76 = 6.645, P = 0.012).

In further analyses, we systematically explored the effects of the two- and three-way interaction terms between protein, fat, and carbohydrate intakes. In none of the model variants did any of these interaction terms achieve significance, either within or between subjects (P > 0.149 in all cases), indicating that the most parsimonious model includes the additive effects of protein, carbohydrate, and fat intake on surface temperature. The parameter estimates from this GLM are presented in Table 2. The coefficients for the three macronutrients are statistically similar, indicating a similar thermogenic effect per kJ eaten for protein, fat, and carbohydrate.

Table 2. Parameter estimates from most parsimonious GLM testing the additive effects of protein, carbohydrate, and fat intakes on surface temperature
Dependent variableParameterBSEMtSig.95% Confidence interval
Lower boundUpper bound
TLIntercept29.2540.41271.0830.00028.43430.073
P0.0410.0192.2160.0300.0040.079
C0.0400.0133.0950.0030.0140.066
F0.0220.0131.7140.091−0.0040.048
TDIntercept30.0400.34586.9700.00029.35230.728
P0.0350.0162.1980.0310.0030.066
C0.0330.0112.9760.0040.0110.054
F0.0280.0112.6160.0110.0070.050

Hence, we can conclude that increased intakes of protein, carbohydrate, and fat are all associated with an increase in body surface temperature. In consequence, when surface temperature response profiles were mapped onto nutrient intakes (Figure 4), there was a general trend for temperature to increase with increased energy intake, with the regions of highest surface temperature being those associated with low percent dietary protein [i.e., toward the top left of the panels in which protein intake is plotted against either carbohydrate or fat (Figure 4) or where protein is plotted against carbohydrate + fat intake (Figure 5)]. These are the diets on which energy intake is the highest because of the effects of protein leverage (Figure 1).

thumbnail image

Figure 4. Thin-plate spline response surfaces for average trunk temperature in the light phase and the dark phase plotted onto nutrient intakes (kJ/d) for protein, carbohydrate, or fat. Mice were fed one of the 25 diets differing in protein, carbohydrate, and fat ad libitum for 25 weeks.

Download figure to PowerPoint

thumbnail image

Figure 5. Thin-plate spline response surfaces for the trunk surface temperature plotted onto protein versus nonprotein (C+F) energy (kJ/d) intakes. Mice were fed one of the 25 diets differing in protein, carbohydrate, and fat ad libitum for 25 weeks.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Male inbred C57BL/6J mice constrained to diets with low P:C ratios (9:72 or 17:64) exhibited increased energy intake and heavier depots of visceral fat pads when compared with mice constrained to diets with mid-range (23:57) or high P:C ratios (31:50 or 48:33), consistent with the protein leverage effect (PL). However, BW and BL were maximal in mice constrained to mid-range P:C ratio diets (17:64 and 23:57), reflecting the fact that both protein and carbohydrate intake are required for maximal skeletal growth and the masses of major structural tissues including muscle and bone. The findings closely mirror those reported for outbred NMRI mice (1) and support the notion that mice prioritize a dietary protein target, most likely due to its significance for growth and development (12). Although total body fat was not measured in this study, the diet-dependent patterns of visceral WAT accorded closely with measures of whole body lipid measurements in the study on NMRI mice (1).

In addition to the positive effects of low P:C diets on energy intake and WAT stores, C57BL/6J mice on low P:C diets had larger interscapular BAT depots and increased body surface temperatures. Given the link between low P:C ratio and energy intake, this result is consistent with the idea that BAT serves a regulatory function in the dispersal of excess energy ingested as a consequence of protein leverage. However, the downregulation of BAT thermogenic genes including DIO2, UCP1, and PGC-1α at the conclusion of the study in mice constrained to low P:C diets does not support this interpretation. It is possible that BAT thermogenic systems respond initially when an animal experiences a low P:C diet but fail in the face of chronic overingestion of energy (see below), leading to BAT hyperplasia as a compensatory response. We cannot, however, exclude the alternative possibility that the increase in the size of BAT deposits on low P:C diets resulted from infiltration of excess lipid, resulting in a phenotype more typical of WAT. Nor did we establish whether there was any evidence of the browning of WAT (26).

Our data from trunk surface temperature suggest that, whatever the role of BAT, DIT arose as a response to the ingestion of excess energy on low P:C diets. Trunk surface temperature is largely determined by skin blood flow and BAT-dependent thermogenesis (22, 27), and, thus, we used it as a simple surrogate for thermogenesis in this study. Surface temperature might also be affected by differences in subcutaneous fat. However, such differences do not explain the observed patterns of surface temperature since adiposity and hence thermal insulation was highest on diets that gave the highest surface temperature readings. It is also notable that surface temperatures reflected the expected diurnal rhythm in metabolic rate, being significantly greater during the dark phase than the light phase.

The difference in strength of association between nutrient intakes and surface temperature observed (see Results section) could result either from differences in the thermogenic effects of the different nutrients (per kJ ingested), with carbohydrate having greatest effect, or to differences in the contributions of the nutrients to total energy intake across experimental subjects. It is apparent from estimates for the linear coefficients (Table 2) that protein and carbohydrate had similar thermogenic effects per kJ ingested (light phase: B = 0.041 ± 0.019 and 0.040 ± 0.013; dark phase: B = 0.035 ± 0.016 and 0.032 ± 0.011, respectively), whereas fat had a somewhat lesser effect per kJ ingested (B = 0.021 ± 0.013 and 0.029 ± 0.011, for light and dark phases, respectively). However, the 95% confidence limits overlapped, leaving no statistical grounds for concluding that the thermogenic effects of the three nutrients differed. It is interesting to note that the equivalent coefficients from the simpler model of protein vs. nonprotein energy intake were also similar (light phase: B = 0.039 ± 0.019 and 0.031 ± 0.011; dark phase: B = 0.034 ± 0.016 and 0.030 ± 0.009, respectively). The observed difference in average daily energy intake between mice on the lowest and highest P:C diets in group 1 (7 kJ/d; Figure 1B) corresponds to a trunk surface temperature difference of 0.3°C.

It has been reported previously that rats on both high-carbohydrate and high-fat cafeteria diets, which are typically commensurately low in protein concentration, exhibit increased BAT masses, increased BAT blood flow and DIT (14, 20). Other reports indicate that high-protein diets reduce interscapular BAT mass and thermogenic capacity (28). Our demonstration that dietary P:C ratio negatively modulates total energy intake is consistent with these reports. Contrary reports of high-protein diet preload studies in humans increasing rather than reducing thermogenesis have all come from short-term studies (29).

The finding that body fat mass increased substantially on the lowest P:C diet in this study indicates that, despite the evidence of increased DIT from body surface temperature data, DIT was insufficient to dissipate all excess ingested energy. Thus, the apparent efficiency of DIT fell in parallel with a general increase in body fat stores. It is possible that DIT and associated increases in thermogenic gene expression (e.g., in BAT) are better suited to short term rather than prolonged increases in energy intake. Insulin resistance provides an alternative explanation for a reduced efficiency of DIT with an increase in the mass of body fat stores. We did not measure insulin sensitivity but IRS1 mRNA expression in BAT was modestly reduced in mice constrained to lower P:C diets, suggestive of insulin resistance. Conversely, GLUT4 expression was enhanced in mice constrained to low P:C diets, possibly secondary to changes in the activity of the AMPK pathway (30).

These results may have implications for the development of obesity in humans on prolonged exposure to low-protein, high-energy diets. The protein leverage effect has been demonstrated recently in lean humans and, therefore, a low-P:C diet may drive overconsumption and, if maintained, could induce obesity (11). The role of BAT, which regresses after birth but persists in the neck and mediastinum in humans throughout adulthood (31) or the browning of WAT, in modulating these processes remains to be determined but may contribute to differences in genetic susceptibility to weight gain and obesity (18).

In conclusion, our results suggest that ad libitum access of mice to diets with low P:C ratios induces increased energy intake, WAT mass, BAT mass, and surface temperature but is associated with decreased expression of thermogenic genes in BAT. Furthermore, maximal body growth and lean body mass is associated with intermediate (17:64 and 23:57) dietary P:C ratios, confirming the importance of dietary macronutrient balance for development as well as the prevention of excessive adiposity in adulthood.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

We thank Professor Leonard Storlien for his help and for providing valuable comments on the manuscript. We thank Chris Maloney, Victoria Cogger and Jenny Phuyal for helping in the dissection of the project.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods and Procedures
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  9. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
OBY_20007_sm_SuppTabS1.doc37KSupporting Table S1
OBY_20007_sm_SuppTabS2.doc34KSupporting Table S2

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.