This study confirms that mice fed a high fat-normal protein diet gained significantly more weight, adipose tissue mass and liver fat compared to all other groups despite a similar energy intake 
. It further shows that in contrast to fat a high protein content in the diet induced a lower weight gain, a lower adipose tissue mass and a lower accumulation and storage of fat in the liver compared to all other groups. Interestingly the present study shows that the reducing effect of protein on adipose tissue mass and liver fat occurred both under high and low fat conditions. This suggests a specific effect of protein in liver independent of the fat and carbohydrate content of the diet. These results are in line with previous observations in which dietary protein reduced adiposity in rats 
and humans 
during short term interventions. Increasing protein intake was also reported to reduce liver fat accumulation in rats 
and mice 
as well as in the context of high fat diet in human 
. This was earlier interpreted as both a direct and indirect action of AA. However, the question why energy in the form of amino acids is less efficiently transferred to fat in the liver still remained unsolved 
The present results are in agreement with the idea that in order to handle the overflow of incoming AA under HP feeding, AA uptake capacity in the liver increases. Also hepatic genes involved in AA deamination and metabolism are induced whereas the use of AA for liver protein synthesis mainly remains unaffected. The results of GSEA revealed that HP feeding both in context of a high fat or a high carbohydrate diet strongly affects hepatic gene expression in mice. This effect includes induction of gene sets belonging to AA processing pathways such as AA transport and catabolism, transamination processes and nitrogen metabolism. In previous observations HP diet enhanced liver transporter activity and AA uptake in rats 
. Accordingly, the present study shows that liver expression of membrane transport proteins of the solute carrier (Slc) family was enhanced in mice adapted to the HP diet. It was also earlier observed in rats that the increased quantity of AA taken up by the liver was unable to further increase liver protein synthesis compared to a normal protein diet. The excess of AA was thus not used for protein synthesis but subjected to deamination with the release of amino acid-derived carbon skeleton 
. Extending these observations, the present study shows no modification of genes involved in liver protein synthetic pathways, whereas there is an induction of catabolic gene sets (). They contain a large number of amino transferases and transaminases (Table S1
, AA_catabolism) as well as gene sets involved in metabolism of single AA, amongst others histidine, glycine, arginine and phenylalanine.
In contrast to earlier findings, in the present study GSEA did not reveal any effects of the HP diet on de novo
. In our analysis we focused on the effect of the high protein diet independent from the background diet. De novo
lipogenesis was repeatedly demonstrated to be reduced during high fat feeding 
and is possibly a result of decreased carbohydrate intake. This effect should therefor occur in both HF groups and the possible additional effect of increasing dietary protein plays a less important role under these conditions.
It was also previously reported in rats that in response to HP feeding hepatic gluconeogenesis remains induced in the fed state 
. Our observations in mice reveal an induction of gene sets affecting carbohydrate metabolism and more specifically genes linked to gluconeogenesis (). However, it also has been reported that the amount of AA-derived carbon skeletons used for gluconeogenesis is very limited and this pathway cannot cope with the excess of incoming AA under HP feeding 
. The present study indicates that in order to handle the excess of AA taken up by the liver that are not used for protein synthesis or gluconeogenesis under HP feeding, hepatic capacity for AA oxidative deamination processes increases (). Indeed, biosynthesis of mitochondria and sets of genes encoding enzymes of the TCA cycle and oxidative phosphorylation were induced in the liver by a HP diet in mice. Via transamination reactions AA are metabolised into α-keto acids, then converted into acetyl-CoA, succinyl-CoA, fumarate, α-keto glutarate and oxaloacetate which subsequently enter the mitochondrial TCA cycle. The direct supply of dietary AA into the TCA cycle when fed a HP diet was described earlier 
. NADH is generated in the TCA cycle, which is a precursor required for oxidative phosphorylation and generation of ATP (Table S1
, TCA_cycle). Genes related to oxidative phosphorylation are encoding different sub-complexes of enzymes like cytochrome c oxidase, NADH dehydrogenase and ATP synthase (Table S1
, oxidative_phosphorylation). An energy restricted HP diet was previously described to induce a stronger increase in mitochondrial oxidation compared to a normal energy restricted diet 
. Moreover, in rats adapting to a HP diet a transient drop in carbohydrate oxidation and increase in protein oxidation was observed 
. In humans, both without and after adaptation to a HP diet, basal protein oxidation and post-prandial energy expenditure in response to a HP meal were higher compared to the control condition 
. Fatty acid oxidation was not detected to be altered in the present study. This is in line with previous data from rats which demonstrated that postprandial lipid oxidation increases on the first day of a HP diet, but returns to baseline during adaptation 
. In human post prandial lipid oxidation was even suggested to decrease after adaptation to a HP diet 
. Tremendous increases in acetyl-CoA, inhibiting β-oxidation may explain these findings.
Interestingly, the HP diet increased plasma β-hydroxybutyrate (BHB) levels, suggesting the induction of a fasting-like situation and a channelling of the excess of acetyl-CoA to BHB synthesis. It is likely that during HP feeding the increased catabolism of AA acids induces an overflow of acetyl-CoA exceeding the cellular energy requirements. In the liver it is not converted into fat but into BHB that is excreted and may act as energy source for extrahepatic organs. The HP diet increased postprandial plasma TG which suggests a larger excretion of hepatic lipids via VLDL which at the same time decreases lipid storage. Indeed VLDL production rate decreases with a HF-NP diet but this effect is restored under HF-HP conditions. In humans increased postprandial TG levels in response to a HP adaptation were likewise observed 
. Although an increased formation of apolipoproteins is not detected on a transcriptional level, this does not exclude the possibility that due to the high influx amino acids lipoprotein synthesis will increase. This will stimulate hepatic formation and excretion of VLDL particles. Other processes enriched under HP conditions are related to oxidative stress response. The high rate of mitochondrial oxidation confronts the liver with large amount of reactive electrophiles and changes the redox state of hepatocytes leading to an activation of Nrf2 
. Part of the Nrf2 target genes belong to the glutathione (GSH)-dependent detoxification system 
, encompassing amongst others glutathione synthase and glutathione peroxidases for neutralising oxidative stress (Table S1
, oxidative_stress_response). Moreover, several genes of the glutathione S-transferase (GST) family are overexpressed under HP feeding. GST catalyses the conjugation of reduced GSH to electrophiles, e.g. lipid peroxides, thereby protecting against peroxidation 
. GSH is regenerated by glutathione reductase (GSR). The hepatic synthesis of GSH uses AA (cysteine, glutamate and glycine) provided by the HP diet 
. These processes require energy in form of ATP, which is not available for other metabolic functions or storage as lipid.
Taken together, the present data suggest that at a similar energy intake, mice ingesting HP as compared to NP diets possess a lower body weight, adiposity and hepatic fat due to a combination of modulated processes (). Firstly, the hepatic capacity to store AA as protein is limited and the excess AA is deaminated possibly to prevent a rise in cellular and blood AA concentration. Secondly, as the transfer of AA-derived carbon to glucose also is limited and de novo
lipogenesis seems absent, these carbon skeletons are metabolized and transferred to acetyl-CoA. Thirdly, as result of an overload of acetyl-CoA and limited capacities to further metabolise it, energy is excreted from the liver in form of BHB and can act as a substrate to extrahepatic organs. In addition, more energy has to be used for detoxification of electrophiles produced during protein catabolism. Thus, handling of large amounts of amino acids leads to an increased energy utilisation in the liver. At the same time hepatic lipid storage is reduced. Additional effects contributing to the leaner phenotype of HP fed mice might be a slight, non-significant reduction in food intake, modifications in intestinal microbiota resulting in reduced nutrient availability as well as increased thermogenesis and therefore easier maintenance of body temperature at 20°C. This, however, has to be subject to further investigation. At last, HP in comparison with NP feeding promoted down regulation of both adaptive and innate immunity. Fatty liver is often associated with inflammation 
. The diminished hepatic lipid content in HP-fed animals prevented an increase in inflammation markers and a response of the immune system, even explicit inflammation was not yet detected in NP fed mice.
Schematic fate of dietary amino acid utilisation in the liver.
In conclusion, the liver plays a buffering role in preventing the extrahepatic tissues from an overload of AA. Moreover, a shift in energy from the liver to peripheral tissues and its use for example in muscle are suggested and should be further investigated also on a longer term.