Metabolic advantage: how could it happen?
It is possible that metabolic efficiency may be decreased by oxidative uncoupling as described above. Polymorphisms connecting uncoupling proteins with obesity or propensity to gain weight have been identified in humans [11,12] although these are not firmly established and the effect of dietary intervention is unknown. Other mechanisms are better understood and are described below.
Substrate cycling and protein turnover
Substrate or "futile" cycles refer to the dynamic process that must accompany the thermodynamic steady state [13]. In particular, increased cycling of metabolic intermediates utilizes ATP and generates heat. The simplest examples are the numerous kinase-phosphatase pairs that regulate metabolism. In addition, although not generally considered in the category of substrate cycling, inefficiency results from the repeated breakdown and re-synthesis of proteins, lipids, and carbohydrates in cycles that use ATP for no apparent net gain. Such mechanisms, however, far from futile, allow for precision in the regulation of metabolism and constitute one of the uses of ATP. Protein turnover, in particular, provides for error correction or removal of "old" or damaged proteins. The effect of metabolic path on the energetics of oxidation is illustrated in Table 1 which summarizes the analysis from our earlier paper [2]. In this example, a mole of glucose directly oxidized to CO2 and water generates 38 moles of ATP with an overall efficiency of about 38.5%. On the other hand, if glucose is first incorporated into glycogen, followed by hydrolysis of the glucose and subsequent oxidation, 2 moles of ATP are lost per mole in this cycle with overall efficiency reduced to 35%. Similarly an amino acid from an "average" protein, when directly oxidized to CO2, produces ATP with an efficiency of about 33%. If the amino acid is first incorporated into a protein and later hydrolyzed and oxidized, four ATP's per molecule are used for synthesis of the peptide bond. This reduces the efficiency to 27%. Smaller degrees of inefficiency are seen for lipid cycles (Table 1) but multiple cycles may have a cumulative effect. It is estimated, for example, that half of depot fatty acids in triacylglycerol have been through at least one cycle [14]. It should be apparent that variation in efficiency is not a thermodynamic issue but an empiric question to be determined by the requirements of metabolism.
Table 1. Effect of Path on energetics of oxidation
Thyrotoxicosis
Thyroid hormone decreases efficiency possibly by mechanisms involving both uncoupling and cycling described above: oxidative uncoupling as well as increased futile cycling of intermediates [15]. It is observed in thyrotoxic mice that UCP1 decreases efficiency in brown fat at the mitochondrial level [8]. In humans, the role of UCP1 in thyrotoxicosis is less certain due to the relative paucity of brown fat. On the other hand, activation of the adrenergic system via phosphoenolpyruvate carboxykinase ultimately increases "futile" metabolic cycling of intermediates ([15]). Thyrotoxicosis is well known to result in weight loss, often with increased food intake and increased generation of heat, indicative of metabolic inefficiency. The use of thyroid hormone has even been suggested therapeutically to induce weight loss in obese individuals, although its toxicity has limited this application. Inefficiency in metabolic processes with weight loss and increased heat generation, therefore, is known to occur on clinical grounds. Even without a complete understanding of the relative importance of different underlying cellular mechanisms in humans, the potential for biochemical processes to reduce their efficiency must be considered established as a feature of mammalian metabolism.
Protein induced protein turnover
There is abundant evidence that dietary protein stimulates protein breakdown and re-synthesis. In particular, branched chain amino acids, and especially leucine, are documented to act as nutritional signals acting via both the insulin and mTOR signaling pathways [16-18]. On the macroscopic level, the energetic cost of protein turnover is demonstrable as excess heat generated during a high protein meal. Thermogenesis (thermogenic effect of feeding; old name: specific dynamic action) has been defined as the extra heat generated during a meal due to digestion or metabolism. Johnston et al [19] compared the energy expended during 9 hour intravenous feedings of a high protein meal, vs. an isocaloric high carbohydrate meal; both contrasted with a 9 hour fast. The protein meal, with 70% of its caloric value due to protein, had significantly greater thermogenesis than the high carbohydrate meal (70% of calories from carbohydrate). These data have been reproduced in numerous studies [19-22]. The overall energy costs of protein turnover and synthesis have been estimated in various animal species, including man, and tabulated by Vernon Young ([23]), based on data from other investigators [24-26]. Despite the substantial experimental difficulties involved, the cost of protein synthesis clusters at around 4–5 kcal/gram in 8 species of birds, marsupials and mammals, including man. The high energetic cost is understandable in view of the multiple ATP-requiring processes involved. The cost of protein turnover can reduce efficiency from 33% to 27%, merely in the formation and hydrolysis of a single peptide bond (requiring 4 ATP's per bond formed: Table 1). In addition, protein processes that are ATP-dependent include formation of the ribosomal initiation complex, translation and folding of the protein, and protein degradation (both ubiquitin-dependent and -independent pathways) [23]. The energy costs of protein turnover could therefore account for a metabolic advantage in high protein diets, independent of carbohydrate content. This mechanism may also contribute to inefficiency in low carbohydrate diets, often high in protein.
Gluconeogenesis-stimulated protein turnover in carbohydrate restriction
The following hypothesis is suggested from classic studies of starvation done in chronically fasted obese individuals [27,28]. The brain's metabolism requires 100 grams of glucose per day. In the early phase of starvation, glycogen stores are rapidly reduced, so the requirement for glucose, is met by gluconeogenesis. Approximately 15–20 grams are available from glycerol production due to lipolysis, but fatty acid oxidation generally cannot be used to produce glucose. Therefore, protein breakdown must supply the rest of substrate for conversion to glucose in the early phases of starvation. By 6 weeks of starvation, ketone bodies plus glycerol can replace 85% of the brain's metabolic needs, the remainder still arising from gluconeogenesis due to protein. It should be mentioned that, since the fundamental role of ketones is to spare protein, it might be expected that the reliance on protein would actually decrease with time, perhaps relating to the anecdotal observation of "hitting the wall" on weight loss diets.
Very low carbohydrate diets, in their early phases, also must supply substantial glucose to the brain from gluconeogenesis. For example, the early phase of the popular Atkins or Protein Power diet restricts dieters to about 20–30 grams of carbohydrate per day, leaving 60–65 grams to be made up from protein-originated gluconeogenesis. One hundred grams of an "average" protein can supply about 57 grams of glucose so 110 grams protein would be needed to provide 60–65 grams glucose. Increased gluconeogenesis has been directly confirmed using tracer studies on day 11 of a very low carbohydrate diet (approx 8 grams/day) [29]. If indeed, 110 grams of endogenous protein is broken down for gluconeogenesis and re-synthesized, the energy cost, at 4–5 kcal/gram could amount to as much as 400–600 kcal/day. This is a sizable metabolic advantage. Of course, the source of protein for gluconeogenesis may be dietary rather than endogenous. Whereas endogenous protein breakdown is likely to evoke energetically costly re-synthesis in an organism in homeostasis, dietary protein may conserve energy. The source of protein for the observed gluconeogenesis [29] remains an open question, but there is no a priori reason to exclude endogenous rather than dietary sources. This is therefore a hypothesis that would need to be tested. The extent to which the protein for gluconeogenesis is supplied by endogenous protein would explain very high-energy costs. It should be noted, however, that even if limited to breakdown of dietary protein sources, there would be some energy cost associated with gluconeogenesis.
Metabolic advantage: does it happen?
Having established that there is no theoretical barrier to metabolic advantage and that there are plausible mechanisms that could account for such an effect, we must ask whether it can be demonstrated experimentally, that is, whether the proposed effects are of sufficient magnitude to be a practical feature of weight reduction strategies, in particular very low carbohydrate diets. If so there will be increased weight loss for the same caloric intake, or metabolic advantage. A recent animal model provides support for greater metabolic inefficiency in rats fed carbohydrate restricted diets compared with higher carbohydrate, leading to excess weight loss [30]. Human data in Table 2 illustrates 10 clinical trials of isocaloric diets with a lower versus higher carbohydrate arm in each trial [31-40]. It can be seen that the lower carbohydrate arm in 9 of 10 studies demonstrates increased weight reduction in comparison with the higher carbohydrate arm. Three of the studies show statistical significance (p < 0.05 or better). Even without statistical significance of individual studies, however, the likelihood that the lower carbohydrate arm would have an advantage in 9 of 10 studies is equivalent to the likelihood of 9 coin toss experiments havin