Nutrient metabolism in the liver

Following digestion of dietary fat, proteins or carbohydrates in the gut, the end product nutrients released such as glucose, amino acids, NEFA, SCFA and odd-chain fatty acids (OCFA) such as propionic acid from fermentation by gut microbiota are taken up into the periportal blood supply that drains to the liver. Within the liver, nutrient homeostasis is maintained by metabolism to acetyl CoA and then to ATP through mitochondrial oxidative phosphorylation; acetyl CoA anabolism leading to fat accumulation or distribution of lipids in the systemic circulation through VLDL. The liver is the first major organ after the periportal vein to receive gut-derived nutrients. Therefore, controlled biogenesis of hepatic organelles and expression of uptake/transport proteins and metabolic enzymes are essential to minimise TAG accumulation, formation of reactive oxygen species, inflammation and development of NAFLD^{(Reference Ipsen, Lykkesfeldt and Tveden-Nyborg11)}; and PPARα, an indirect sensor of amino acid levels, plays a significant role in regulating lipid metabolism in the liver.

Hepatic fatty acid uptake and impairment during non-alcoholic fatty liver disease

Dietary fat contributes to NAFLD and fatty acid uptake mechanisms that are reviewed here to appreciate how low-protein diets may contribute to fatty acid uptake during fatty liver disease development.

The main plasma membrane transporters of NEFA into the liver postprandially and during obesity are fatty acid transporter proteins (FATP), caveolins, fatty acid translocase/cluster of differentiation 36(CD36) and fatty acid binding protein (FABP), whose expression is regulated by PPARα. Of the six member FATP family, only FATP2 and FATP5 are highly expressed in the liver^{(Reference Falcon, Doege and Fluitt12)}. Knockdown of FATP2 in mice decreases the uptake of NEFA and ameliorates hepatic steatosis induced by a high-fat diet^{(Reference Falcon, Doege and Fluitt12)}. Similarly, FATP5 knockout mice are resistant to diet-induced obesity and hepatic TAG storage accumulation^{(Reference Doege, Baillie and Ortegon13)}, highlighting the significant role played by FATP5 in liver lipid accumulation. The second family of lipid transporters, caveolins, have three members, namely caveolins 1, 2 and 3, and play an important role in the formation of lipid droplets^{(Reference Ipsen, Lykkesfeldt and Tveden-Nyborg11)}. Caveolin 1 is increased in the liver of mice with NAFLD, mainly in the centrilobular region, where the steatosis tends to be severe^{(Reference Qiu, Liu and Chen14)}. There are contradictory reports on the significance of caveolin 1 for NAFLD based on knockout studies^{(Reference Asterholm, Mundy and Weng15,Reference Li, Chen and Huang16)} ; this may reflect the different dietary induction protocols with caveolin 1 being important for NAFLD development in high-fat but not in fasting diets.

Fatty acid translocase/CD36 accelerates NEFA uptake via facilitated diffusion. Elevated hepatic expression of CD36 has been observed in NAFLD and is associated with enhanced uptake of NEFA^{(Reference Miquilena-Colina, Lima-Cabello and Sánchez-Campos17)}. High-fat-diet-fed mice develop hepatic steatosis in parallel with increased CD36 mRNA and protein expression^{(Reference Wilson, Tran and Erion18)}, promoting a positive feed-forward loop for lipid accumulation. After uptake, the cytosolic FABP facilitate intracellular transport of NEFA for metabolism and to lipid-sensitive transcription factors. Targeted deletion of the liver isoform, liver fatty acid binding protein (LFABP) in mice results in hepatic lipid accumulation in female mice^{(Reference Martin, Atshaves and Landrock19)} further illustrating the importance of transport protein expression to prevent fatty liver disease. Liver expression of FATP, caveolin, LFABP or CD36 is regulated by the common transcription factors FOXA1, CEBPα and PPARα, and there is evidence that CEBPα and PPARα are in turn regulated by the amino acid response to low dietary protein^{(Reference Haro, Marrero and Relat5)}. Together, this evidence suggests that dietary protein malnutrition will influence fatty acid uptake and storage in the liver, at least in part due to altered PPARα activity.

Mitochondrial fatty acid metabolism and dysfunction in non-alcoholic fatty liver disease

Mitochondrial dysfunction has been associated with the pathogenesis of NAFLD^{(Reference Nassir and Ibdah20,Reference Begriche, Igoudjil and Pessayre21)} and hepatic steatosis^{(Reference Koliaki, Szendroedi and Kaul22–Reference Simões, Fontes and Pinton24)}. Mitochondrial-derived reactive oxygen species contribute to the progression of NAFLD^{(Reference Koliaki, Szendroedi and Kaul25)}. In conditions of dietary excess or obesity, oxidative phosphorylation promotes further reactive oxygen species production by mitochondria^{(Reference Koliaki, Szendroedi and Kaul25)}. This is likely to be exacerbated by increased conversion of accumulating palmitate to ceramide and inhibition of PPARα ^{(Reference Williams, Correnti and Oranu26)}, in turn inhibiting metabolic organelle biogenesis and activity.

Mitochondria play a central role in aerobic ATP production from fatty acid β-oxidation (acyl-chain length of ≤C20), ketogenesis and gluconeogenesis from pyruvate and tricarboxylic acid cycle intermediates.

After fatty acid uptake by the hepatocyte, CoA is added by fatty acyl-CoA synthase, forming long-chain acyl-CoA. Then, CPT1 catalyses conversion of the long-chain acyl-CoA to long-chain acylcarnitine to facilitate transport across the inner mitochondrial membrane. An inner mitochondrial membrane CPT2 then converts the long-chain acylcarnitine to acyl-CoA for β-oxidation. It has been reported that a low-protein diet (3 %) in weanling rodents impaired hepatic expression of CPT1 and CPT2 genes and promoted fatty liver development^{(Reference Kuwahata, Kubota and Amano27)}. Fatty liver induced by this 3 % protein diet was attenuated when the rat diets were supplemented with dietary medium-chain TAG^{(Reference Kuwahata, Kubota and Amano27)}. While the evidence from low protein dietary studies so far is limited, existing data suggest a role of PPARα regulation and mitochondria in mediating the effect of paternal protein malnutrition on offspring hepatic metabolism^{(Reference Contreras, Torres and Tovar28)}.

Mitochondrial amino acid metabolism

The branched-chain amino acids (BCAA) valine, leucine and isoleucine constitute between 15 and 25 % of total amino acids from dietary protein intake^{(Reference Holeček29)}. During BCAA catabolism, each BCAA is first converted into its respective α-keto acid through mitochondrial branched-chain aminotransferase in extrahepatic tissue. The α-keto acid is further metabolised by a rate-limiting step of BCAA catabolism via branched-chain keto-acid dehydrogenase (BCKDH) in liver, skeletal muscle, kidney and adipose tissue. Notably, a decrease in BCKDH activity is seen in several diseases such as maple syrup urine disease, type 2 diabetes and obesity, measured as increases in circulating BCAA levels^{(Reference She, Van Horn and Reid30)}. A recent study has identified higher valine and isoleucine concentrations in adolescents with NAFLD independently of the presence of diabetes. In addition, the BCAA metabolic signature was negatively correlated with insulin sensitivity^{(Reference Goffredo, Santoro and Trico31)}. There are at least two explanations for the presence of higher concentrations of circulating BCAA in blood during obesity: either down-regulation of BCAA catabolic enzymes^{(Reference Herman, She and Peroni32)} or more likely the inhibition of rate-limiting BCKDH in liver. The activity of BCKDH is regulated by inhibitory phosphorylation by BCKDH kinase, and preliminary studies have successfully targeted BCKDH kinase using a selective inhibitor (BT2) to increase BCKDH activity, reduce BCAA, inhibit lipogenesis and improve glycaemic control^{(Reference White, McGarrah and Grimsrud33)}.

Metabolism of BCAA adds to the lipogenic acetyl-CoA and propionyl CoA pools. Using isotopomer spectral analysis^{(Reference Crown, Marze and Antoniewicz34)}, BCAA were shown to be involved in the synthesis of OCFA and even-chain fatty acids in adipocytes. Propionyl CoA derived solely from isoleucine and valine catabolism is a primer for fatty acid synthase; thus, C5 : 0 may then be extended in the synthesis of longer chain OCFA (Fig. 2). This pathway is also used in the metabolism of methionine, although its contribution to the total OCFA pool is unknown. The OCFA C15 : 0 and C17 : 0 are both associated with good metabolic health, possibly mediated through PPARα activation^{(Reference Jenkins, West and Koulman35)}. It has been shown that plasma circulating OCFA correlate inversely with a range of metabolic conditions including fatty liver disease and insulin resistance^{(Reference Jenkins, West and Koulman35)}. It remains to be determined whether, in conditions where BCAA metabolism is inhibited or when dietary protein intake is low, OCFA synthesis is also lower. These findings suggest a novel pathway for investigating the protective effect of dietary protein against hepatic lipid accumulation, through enhanced BCAA metabolism and the production of OCFA.

Fig. 2. Pathways for synthesis of C15 : 0 and C17 : 0 in the liver. PHYH, phytanoyl-CoA α-hydroxylase; HACL1, 2-hydroxyphytanoyl-CoA lyase; PDH, pristanal dehydrogenase; BCAT, branched-chain amino acid aminotransferase; BCKD, branched-chain ketoacid dehydrogenase; VLCFA, very long-chain fatty acids.

Mitochondrial metabolism of fatty acids and BCAA is impaired in obesity and fatty liver disease but whether these changes are causal or a consequence of disease is unknown. Nevertheless, metabolic inhibition in liver mitochondria correlates with lower circulating OCFA and elevated BCAA. As OCFA are protective, the potential for targeting BCAA metabolism to increase OCFA in fatty liver disease merits consideration.

Peroxisomes in fatty acid α- and β-oxidation

The peroxisome, a single membrane-bound organelle, plays important roles in plasmalogen and bile acid synthesis, β-oxidation of very long-chain fatty acids, α-oxidation of branched-chain fatty acids and BCAA, and in maintaining cellular redox homeostasis^{(Reference Wanders36)}. Peroxisomes work in concert with mitochondria to provide substrates for oxidative metabolism by regulating early metabolism of long-chain fatty acids, branched-chain fatty acids and possibly also BCAA^{(Reference Smith and Aitchison37)}. They are also enriched in catalase which supports the detoxification of the reactive oxygen species hydrogen peroxide produced during metabolism thereby mitigating the risk for oxidative damage^{(Reference Walker, Pomatto and Tripathi38)}.

Very long-chain fatty acids, including C22 : 0, C24 : 0 and C26 : 0, undergo β-oxidation in peroxisomes because very long-chain fatty acids are not substrates for CPT1, and it is the only acylcarnitine fatty acid product of CPT1 that can be transported into mitochondria. Peroxisomes also lack a citric acid cycle and respiratory chain; hence, the end products of β-oxidation in peroxisomes including NADH, acetyl-CoA, propionyl-CoA and other longer chain acyl-CoA are shuttled from peroxisomes to mitochondria for complete oxidation^{(Reference Wanders36)}. The rate-limiting enzyme in peroxisomal fatty acid β-oxidation, acyl-coA oxidase 1 which produces hydrogen peroxide as a byproduct, regulates metabolism, spontaneous hepatic steatosis and hepatocellular damage over time^{(Reference Wanders36)}.

Branched-chain fatty acids, which cannot undergo β-oxidation directly due to the location of the methyl group at position 3, undergo α-oxidation in peroxisomes by removing the terminal carbon to generate a 2-methyl fatty acid before entering into the β-oxidation pathway^{(Reference Wanders36)}. For instance, phytanic acid which is a 3-methyl branched-chain fatty acid undergoes α-oxidation beginning with the formation of phytanoyl-CoA, followed by hydroxylation to produce 2-hydroxyphytanoyl-CoA, a reaction catalysed by the enzyme phytanoyl-CoA 2-hydroxylase. Subsequently, 2-hydroxyphytanoyl-CoA is lysed by the enzyme 2-hydroxyacyl-CoA lyase to pristanal and formyl-CoA and then hydrolysed into formic acid and CoASH. Pristanal is oxidised to pristanic acid (2, 6, 10, 14-tetramethylpentadecanoic acid) by a yet undefined peroxisomal aldehyde dehydrogenase. Finally after activation to its CoA-ester, pristanoyl-CoA undergoes three cycles of β-oxidation in peroxisomes prior to transport of end products to mitochondria for complete oxidation^{(Reference Wanders36)}. Both propionyl CoA and pristanal produced in peroxisomes by phytanoyl-CoA 2-hydroxy lyase activity are important precursors for biosynthesis of OCFA^{(Reference Jenkins, de Schryver and Van Veldhoven39)} (Fig. 2). It remains to be determined whether impaired peroxisomal function reduces levels of circulating protective OCFA. Nevertheless, deficits in peroxisomal biogenesis have been associated with liver disease development^{(Reference van Zutphen, Ciapaite and Bloks40)}. In later sections, we explore the relationship between protein malnutrition and deficits in peroxisomal function and number.

Odd-chain fatty acids and non-alcoholic fatty liver disease

Two key odd-chain SFA that have become increasingly important biomarkers for predicting metabolic diseases including fatty liver disease are pentadecanoic acid (C15 : 0) and heptadecanoic acid (C17 : 0)^{(Reference Jenkins, West and Koulman35)}. Lower levels of OCFA are associated with reduced insulin sensitivity and increased fatty liver. Similarly, serum levels of C15 : 0 and C17 : 0 are inversely correlated with NAFLD activity scores and hepatocyte ballooning scores in humans^{(Reference Yoo, Gjuka and Stevenson41)}. OCFA originating from microbial fermentation have been used as biomarkers for dairy fat intake in humans^{(Reference Pfeuffer and Jaudszus42)}. As described previously, OCFA biosynthesis is catalysed by fatty acid synthase from propionic acid^{(Reference Al-Lahham, Peppelenbosch and Roelofsen43)} and in humans, the SCFA propionic acid can be generated via BCAA degradation, peroxisomal oxidation of cholesterol side-chain during bile acid formation and peroxisomal α-decarboxylation, followed by successive β-oxidative degradation of the branched-chain fatty acid, phytanic acid (Fig. 2)^{(Reference Jenkins, West and Koulman35)}. Propionic acid produced via gut microbiotic fermentation of dietary fibre may then absorbed and metabolised by the liver to produce further OCFA^{(Reference Pfeuffer and Jaudszus42)}. The relative contribution of propionic acid produced during metabolism of BCAA to the serum levels of C15 : 0 and C17 : 0 remains unknown. Nevertheless, it has been suggested that a high-protein diet can minimise diet-induced development of fatty liver and even reverse pre-existing steatosis state^{(Reference Pfeuffer and Jaudszus42)}. In order to investigate whether the beneficial effect of high-protein diet is mediated via production of OCFA, the anaplerotic lipid C7 : 0 was investigated in a high-fat diet-induced fatty liver disease mouse model. Despite beneficial effects of dietary C7 : 0 on expression of hepatic PPARα that regulates peroxisomal and mitochondrial biogenesis, increased acyl-CoA oxidase expression (rate-limiting step in peroxisomal metabolism of fatty acids), CD36 expression (fatty acid uptake) and a decline in plasma acetyl-carnitines, there was no change in liver fat accumulation^{(Reference Comhair, Garcia Caraballo and Dejong44)}. A complex picture is emerging where adequate dietary protein contributes to the synthesis of protective OCFA, but in models of liver disease that are induced by high-fat diet, no protective effect of the OCFA C7 : 0 was observed. The interpretation of C7 : 0 supplementation study is further complicated by the observation that a high-fat diet, which was used to induce liver fat accumulation, reduces C15 : 0^{(Reference Jenkins, Aoun and Feillet-Coudray45)}. C7 : 0 supplementation to prevent NAFLD during a high-fat diet is not likely to be effective; however, where low protein is a causative factor for fatty liver disease, further investigation of the effect of C7 : 0 supplementation is merited.

Dietary protein effects on liver peroxisome and mitochondrial function

Peroxisomal biogenesis is regulated by PPARα and coordinated with mitochondrial biogenesis through PGC1α activity^{(Reference Wu, Puigserver and Andersson46,Reference Mandard, Müller and Kersten47)} .

van Zutphen et al. showed diminished hepatic peroxisome content and impaired peroxisomal function in rats following a 4 weeks low-protein diet (5 % energy content as protein) and attributed the cause of hepatic steatosis to loss or dysfunction of the peroxisomes. Using electron microscopy, peroxisomes in the periportal area of the liver were rarely observed in low-protein diet animals compared with controls as indicated by decreased immunofluorescence staining of the peroxisomal membrane protein PEX14. Also, protein levels of peroxisomal membrane protein 70 and the antioxidant enzyme catalase were decreased after 4 weeks of a low-protein diet^{(Reference van Zutphen, Ciapaite and Bloks40)}, increasing the risk for oxidative damage and lipid accumulation. In the same study, in the mitochondria, there was a reduction in both complex I and complex IV activity. Lower expression of medium-chain specific acyl-CoA dehydrogenase, medium and short-chain l-3-hydroxyacyl-coA dehydrogenase and enoyl-CoA hydratase were observed, suggesting that peroxisomes and mitochondria are affected by conditions of low dietary protein availability^{(Reference van Zutphen, Ciapaite and Bloks40)}. More studies are required for further understanding of the impact of protein restriction on the biogenesis of peroxisomes and on mitochondrial function.

Dietary protein restriction and hepatic lipid accumulation

Both human and animal studies have shown that a low-protein diet (below 9 % of total energy) causes an increase in hepatic lipid accumulation. In human studies, a diet generally low in protein and high in carbohydrates, which may been seen in the elderly with significant co-morbidity and is present at all ages in developing countries, has been shown to cause liver lipid accumulation^{(Reference Ghosh, Gupta and Bhattacharya48)}. Moreover, children in India that presented with severe malnutrition were found to show several metabolic disturbances including increased oxidative stress and hepatic steatosis^{(Reference Ghosh, Gupta and Bhattacharya48)}.

In animal models, feeding a 5 % of low-protein diet to Wistar rats for 4 weeks resulted in increased hepatic TAG accumulation^{(Reference Kuwahata, Kubota and Amano27)}. Similarly, feeding Sprague–Dawley male rats with a low-protein diet for 4 weeks also led to hepatic TAG accumulation which was associated with down-regulation of hepatic microsomal TAG transfer protein and increased expression of acetyl CoA carboxylase^{(Reference Kang, Lee and Baik49)}. Similarly, Kwon et al. reported that an 8 % low-protein diet resulted in steatohepatitis with severe steatosis in lactating female Sprague–Dawley rats^{(Reference Kwon, Kang and Nam50)}. Despite the emerging evidence from animal models, there is a potential of confounding in the interpretation of the data; for example, a low-protein diet in mice for 16 weeks led to increased body weight, adiposity and fatty liver^{(Reference Huang, Hancock and Gosby51)} but the lack of dietary protein was substituted with a high carbohydrate content in the diet. It has been suggested that decreasing protein and replacing energy loss with a higher proportion of sucrose may be the driving force of the emerging high incidence of fatty liver disease in developing countries. In the studies we have explored here, protein was substituted by sucrose in two studies^{(Reference Kuwahata, Kubota and Amano27,Reference Kwon, Kang and Nam50)} and by a variety of starches in four other studies^{(Reference Mayneris-Perxachs, Bolick and Leng9,Reference van Zutphen, Ciapaite and Bloks40,Reference Kang, Lee and Baik49,Reference Lyall, Cartier and Richards55)} demonstrating that replacing protein with a specific source of simple sugars was not solely responsible for the effects of low-protein diets observed. Is the development of fatty liver in this model due to low protein or high carbohydrate or a combination of both? To address this, some studies have explored the effect of supplementation with protein extract or amino acids on low-protein-induced fatty liver disease.

Effect of amino acid supplementation on low-protein-induced fatty liver disease

A previous study in an animal model of paternal inheritance has shown that supplementing a low-protein diet with specific amino acids, for example, methionine, moderated fat accumulation in major metabolic tissues such as the liver, muscle and subcutaneous adipose tissue of the adult offspring^{(Reference Watkins and Sinclair52)}. This may be due to changes in gene expression through altered promotor methylation.

The amino acid derivative, betaine (trimethylglycine), occurs naturally in most living organisms. It is synthesised during the oxidation of choline^{(Reference Ueland53)}. Betaine and choline are methylating agents and are important for regulating DNA methylation and downstream gene expression. Choline is directed to maintain the S‐adenosyl methionine cycle; both choline-deficient diets and methionine- and choline-deficient dietary models may be useful for understanding the role of dietary amino acids for human NAFLD partly due to their histological similarity with these diseases^{(Reference Lyall, Cartier and Richards55,Reference Kulinski, Vance and Vance54)} .

A study by Madeira et al. has demonstrated that betaine and arginine supplementation, either individually or combined, into reduced protein diets (160 v. 130 g/kg of crude protein) decreased plasma total lipids and total cholesterol in lean pigs^{(Reference Madeira, Rolo and Lopes56)}. In the presence of either betaine or arginine supplementation, except for fatty acid synthase which was down‐regulated by protein reduction, no effects on hepatic fatty acid composition and gene expression levels of lipid‐sensitive factors were induced by the protein-restricted diet^{(Reference Madeira, Rolo and Lopes56)}. Another study showed that supplementation with fish-derived protein was hepatoprotective during a low-protein diet^{(Reference Bjørndal, Berge and Ramsvik8)}. Together these studies indicate that in animals receiving low-protein diets, the associated development of fatty liver may be prevented by simple amino acid supplementation. The beneficial effects of supplementing with methyl donors that are absent during protein malnutrition may be explained, at least in part, by epigenetic regulation of hepatic metabolic gene expression. Conversely, it has been demonstrated in rat pups from dams previously fed low-protein diets during lactation; hypomethylation of PGC1-α was observed in the offspring, decreased metabolic gene expression and subsequently impaired mitochondrial fatty acid oxidation^{(Reference Pooya, Blaise and Moreno Garcia57)}. It is not known whether a similar pattern of gene promotor hypomethylation may be induced by protein malnutrition in human liver.

Effect of protein variation in the diet on gut microbiota

In humans, dietary patterns shape the gut microbiota^{(Reference David, Maurice and Carmody58–Reference Wu, Chen and Hoffmann61)}. Previous studies have demonstrated that malnutrition in children can change the structure and function of the bacterial populations that are found in the gut with subsequent changes in the nutrient flow, for example, SCFA to the host^{(Reference Ghosh, Gupta and Bhattacharya48,Reference Smith, Yatsunenko and Manary62)} . Notably, fat absorption is impacted by the microbiota and has been linked to the development of steatosis^{(Reference Kolodziejczyk, Zheng and Shibolet63)}.

Many gut microbiota also digest complex carbohydrates, such as dietary fibre, to release the OCFA and SCFA such as acetic, propionic, butyric and valeric acids^{(Reference Hoverstad and Midtvedt64)}. These SCFA may act on receptors in the gut or be taken up directly through monocarboxylate transporters into the periportal vein for delivery to the liver and entry into metabolic pathways. Whether gut microbiota-derived OCFA are a source of circulating C15 : 0 or C17 : 0 is unclear^{(Reference Kimura, Ozawa and Inoue65)}. Nevertheless, it has been shown in mice fed a high-fat diet that butyrate generated by probiotics can promote peroxisomal biogenesis in a PPARα-dependent manner, thereby facilitating further lipid metabolism^{(Reference Weng, Endo and Li66)}.

The protective effect of gut-derived SCFA against development of fatty liver disease is, at least in part, due to their activation of the G-protein coupled receptor, GPR43, present in liver and adipose tissue^{(Reference Maslowski, Vieira and Ng67)}. In germ-free mice without gut microbiota or associated production of SCFA and in a GPR43 knockout mouse model, increases in pro‐inflammatory cytokine production and immune cell recruitment to the liver were observed^{(Reference Maslowski, Vieira and Ng67)}. As inflammation plays a major role in the progression of early stage fatty liver to fibrotic liver disease, dietary patterns that influence gut microbiota may exert distant effects on hepatic metabolism via their metabolites, for example, SCFA^{(Reference Jenkins, Seyssel and Chiu68)}.

To date, few studies in humans and animals have investigated any association between dietary proteins, liver metabolism and gut microbiota in relation to metabolic diseases including fatty liver disease; several studies have focused on a high-protein/low-carbohydrate diet rather than low-protein/high-carbohydrate diet. The limited numbers of protein-varied diet studies available have shown that duration and extent of protein malnutrition influence the gut microbiota and their function including secretion of SCFA and proinflammatory mediators (Table 1)^{(Reference Mayneris-Perxachs, Bolick and Leng9,Reference Li, Lauber and Czarnecki-Maulden70,Reference Zhou, Fang and Sun71)} . On the one hand, a higher-protein diet (>20 % of energy as protein) that led to weight loss in obese subjects was suggested to pose a risk for metabolic health through altering the gut microbiota and reducing the levels of protective SCFA production in the gut^{(Reference Russell, Gratz and Duncan69)}. On the other hand, specific populations at risk for protein malnutrition, including children in developing countries (Malawi and India) and older adults merit further study^{(Reference Ghosh, Gupta and Bhattacharya48,Reference Smith, Yatsunenko and Manary62,Reference Claesson, Jeffery and Conde72)} . They may benefit from increased dietary protein through effects on gut microbiota producing OCFA/SCFA and modulation of lipid accumulation in the liver.

Table 1. Dietary protein alters gut microbiota; resulting inflammatory effect associated with non-alcoholic fatty liver disease

Associative evidence between low dietary protein and fatty liver disease from population studies

Epidemiological studies have reported an association between poor outcomes and high intake of animal but not plant protein and commonly describe these effects in individuals with existing metabolic disease^{(Reference Budhathoki, Sawada and Iwasaki73,Reference Alferink, Kiefte-de Jong and Erler74)} . These studies highlight the need for assessing individual protein requirement, since an increase in dietary protein may have adverse outcomes for some populations, for example, younger people and those with chronic kidney disease, but potentially beneficial effects in those over 65 years of age or who have dietary protein insufficiency. In contrast, a low-protein diet is used in the management of organic acidaemias; however, to the best of our knowledge, no study has reported the outcome of fatty liver associated with protein restriction after the long-term management. Any analysis of fatty liver in these patients is in part confounded by the use of liver transplantation to address the metabolic deficit in patients and because conservative management is achieved not only by dietary protein restriction and metronidazole but also by l-carnitine supplements. A systematic review of the general population concluded that higher protein diets probably improve adiposity, blood pressure and TAG levels, but that these effects are small and need to be weighed against potential for adverse effects^{(Reference Santesso, Akl and Bianchi75)}. Nevertheless, our focus here has been to understand the effects of improving dietary protein intake on fatty liver disease risk in people who are malnourished and who have low protein intake. In Europe, the most recent analyses of National Dietary Surveys suggest that protein intake is adequate over all age ranges examined^{(Reference Rippin, Hutchinson and Jewell76)}, although there is some evidence that in older adults with frailty syndrome, protein intake may be reduced. Dietary surveys are limited in developing countries; however, a review of dietary surveys conducted on the adult South African population from 2000 to 2015 revealed that out of the total energy intake of men and women, the % energy from protein ranged from 10·9 to 18·3 %; fat from 17 to 37·1 %; and carbohydrate from 47·0 to 69 %, highlighting the variation that exists in country^{(Reference Mchiza, Steyn and Hill77)}.

Two population-based studies have explored the effects of increasing dietary protein after weight loss in people with diabetes^{(Reference van Baak, Larsen and Jebb79,Reference Sluik, Brouwer-Brolsma and Berendsen78)} . In the PREVIEW study, the prospective association between increased plant protein intake and improvement in diabetes indicators and fatty liver was reported to be independent of BMI^{(Reference Sluik, Brouwer-Brolsma and Berendsen78,Reference Drummen, Dorenbos and Vreugdenhil80)} . In a secondary analysis of data from the randomised controlled DIOGenes trial, which investigated dietary protein composition and glycaemic index on weight loss maintenance and cardiometabolic risk factors in obesity, plant protein offered the greatest health benefits, while a higher plant protein intake from non-cereal products instead of cereal products (matched by a decrease in other sources of protein intake) benefited by body weight maintenance and blood pressure^{(Reference van Baak, Larsen and Jebb79)}. In a 5-year follow-up study, it was reported that substitution with 10 g carbohydrate with plant (but not animal) protein did not lead to any weight change and all-cause mortality was reduced^{(Reference Campmans-Kuijpers, Sluijs and Nöthlings81)}.

Another study explored liver fat deposition and showed that a high-protein diet compared with an isoenergetic high-carbohydrate diet, without any change in weight, reduced intrahepatic lipid storage, consistent with our suggestion that improving protein intake in low-income countries may reduce risk for NAFLD^{(Reference Martens, Gatta-Cherifi and Gonnissen82)}. To date, there is a lack of detailed biochemistry to describe the underpinning association between low dietary protein, lipid deposition and whether this relates impaired OCFA production.

Conclusion

The importance of dietary composition on lipid storage in metabolic health and diseases cannot be over-emphasised. Altered lipid metabolism is strongly associated with NAFLD, obesity, diabetes, the metabolic syndrome and other cardiometabolic diseases, and diet plays a critical role in the development of these diseases. Several human and animal studies have shown that high-fat and high-carbohydrate diets are associated with the development of NAFLD. These diets affect lipid metabolism resulting in accumulation of TAG in the liver. Here, we suggest that the effects of insufficient dietary protein for fatty liver disease risk are underexplored in specific populations that have a protein-restricted diet, either for medical reasons or because of limited availability of food resources in developing countries.

We have gathered evidence that suggests mitochondria, peroxisomes and altered gut microbiota may each play a role in the development of NAFLD in response to protein restriction and possibly mediated through epigenetic effects and changes to circulating SCFA and OCFA.

The limited existing data suggest that targeting specific sources of protein and amino acid intake provides an opportunity for reducing risk of fatty liver disease, particularly in vulnerable groups^{(Reference Santesso, Akl and Bianchi75)} and those with restricted protein diets in developing countries. Such studies would also need to address the type, source and quality of amino acid/protein used. For example, plant protein but not animal protein has been reported to reduce mortality risk in people with diabetes^{(Reference Campmans-Kuijpers, Sluijs and Nöthlings81)}. The feasibility of improving dietary protein as an approach to NAFLD is further supported by the recent report that an increase in daily protein intake to 25 % by weight was sufficient to reduce intrahepatic lipid accumulation in people with diabetes following a weight-reduction diet^{(Reference Drummen, Dorenbos and Vreugdenhil80)}. This beneficial effect on fat storage in the liver was observed during a 2-year follow-up period of weight maintenance and was independent of BMI.