High-protein Diet More Effectively Reduces Hepatic Fat Than Low-protein Diet Despite Lower Autophagy and FGF21 Levels

Chenchen Xu; Mariya Markova; Nicole Seebeck; Anne Loft; Silke Hornemann; Thomas Gantert; Stefan Kabisch; Kathleen Herz; Jennifer Loske; Mario Ost; Verena Coleman; Frederick Klauschen; Anke Rosenthal; Volker Lange; Jürgen Machann; Susanne Klaus; Tilman Grune; Stephan Herzig; Olga Pivovarova-Ramich; Andreas F. H. Pfeiffer


Liver International. 2020;40(12):2982-2997. 

In This Article


The most important finding of our study is that, compared to LP intake, the 3-week HP diet more effectively reduced IHL in morbidly obese patients which was the primary outcome in this study. Both strategies were differentially accompanied by metabolic improvements of secondary outcomes - serum transaminases, lipid, uric acid, glucose and insulin levels. The HP strategy additionally showed a decrease in oxidative stress conditions. Furthermore, our study showed that different mechanisms are involved in metabolic improvements induced by HP and LP diets. Indeed, FGF21 and autophagy, which have been associated with metabolic improvements, increased only in the LP but not in the HP diets. Contrary with expectations, the mitochondrial activity and expression levels of genes involved in fatty acid β-oxidation were not increased in the HP compared to LP diets. Interestingly, however, genes related to lipid biosynthesis displayed lower expression levels in the condition of HP diets.

To our knowledge, this is the first human study investigating whether HP or LP diets are more effective for liver fat reduction in patients with morbid obesity and NAFLD indication undergoing bariatric surgery. Although all groups, LP, HP and RP, reduced weight and BMI because of the moderate caloric restriction, the HP group was more effective in eliminating the intrahepatocellular lipids than the LP group. Two participants in the LP group even slightly gained liver fat despite weight loss. Remarkably, there was no correlation between weight (or BMI) loss and reduction of liver fat. Moreover, although the LP and RP groups had similar baseline liver fat, the liver fat reduction was more effective in the RP group, suggesting that even a moderate increase of the protein intake up to 20–22 EN% has a stronger effect on liver fat reduction than LP intake. Similar results were published in other groups of subjects. Martens et al reported IHL reduction after isocaloric HP diet compared to LP diet.[31] However, the study was conducted in healthy subjects with BMI of 24 kg/m2 and IHL below 0.4% at baseline. Similarly, Drummen et al observed a significant decrease in IHL in obese participants after 8 weeks of hypocaloric HP diet (35–40 E% protein) and 20 weeks of hypocaloric HP diet.[32] Important result of our study is that we demonstrated the strong reduction in liver fat already after a short-term (3-weeks) HP diet. Taken together, we concluded that HP diet reduced liver fat more effectively than LP diet in the presence of moderate weight loss, regardless of the baseline liver fat content.

Hence, the effective reduction of liver fat by HP diets seems to be accomplished by pathways independent of overall fat mass reduction. We hypothesize that HP diet reduced liver fat by triggering mitochondrial activity and fatty acid β-oxidation. It was previously demonstrated that obesity, NAFLD, and insulin resistance are associated with mitochondrial dysfunction.[33–36] The lower mitochondrial activity and higher CS activity in NASH patients compared to subjects with normal liver or simple steatosis showed a similar pattern as previously reported by Koliaki et al.[37] In addition, CS activity was positively correlated with IHL via 1H-MRS and NAFLD stages diagnosed by SAF score indicating that CS activity is up-regulated by hepatic fat accumulation and may represent an attempt to counterbalance the increased lipids by increased oxidation. Interestingly, methionine-restriction was shown to increase mitochondrial mass, size and capacity in liver, muscle and adipose tissue in some rodent studies.[38–40] However, this was not the case in the LP group where methionine intake was very low because of the vegetarian protein source. On the other hand, Garcia-Caraballo et al[41] found an increased mitochondrial oxidative capacity, but no changes in CS activity by a HP diet, which corresponds to our findings.

Importantly, the expression levels of genes regulating fat uptake, de novo lipogenesis and lipid storage were down-regulated by the HP diet as confirmed by RNA-seq data for LPL, FABP4 and FABP5. LPL, which was down-regulated upon HP intake, can hydrolyse chylomicron triglycerides to NEFAs and promote hepatocellular uptake of chylomicron remnants and NEFAs.[42] FABP4 and FABP5 can bind and promote uptake of fatty acids as lipid transport proteins. Zhu et al[43] found that NASH patients had much higher expression of lipid transport proteins such as CD36 and FABP1 and concluded that an increase in fatty acid uptake contributes more to fat deposition than changes in de novo lipogenesis or fatty acid β-oxidation.

As the main source of de novo lipogenesis, dietary carbohydrates are widely accepted to up-regulate hepatic lipogenesis. Interestingly, previous studies revealed that dietary protein also played a role in the regulation of lipid biosynthesis. Expression levels of SREBP1c,[44,45] FASN,[18,46] ACC,[18,46] SCD1[45] and PPARγ,[18] involved in de novo lipogenesis and lipid storage, were down-regulated by HP intake in several rodent studies. The diminished lipogenesis could be related to HP diet stimulated glucagon secretion but lower glycemic action,[47] but also to the function of dietary ketogenic essential amino acids.[45] The relatively lower carbohydrate content in HP diets might have also partly contributed to the larger decrease in de novo lipogenesis and intrahepatic fat. Also, in contrast with lipogenesis genes, dietary protein intake did not change the expression of genes involved in lipid oxidation or the substrate oxidation estimated by indirect calorimetry in several rat studies.[44,46] In our study, no difference in gene expression levels involved in fatty acid β-oxidation between the LP and HP diets was observed. Stepien et al[46] attributed the unchanged lipid oxidation to a phenomenon observed by starvation, since the gene expression of enzymes involved in lipid oxidation changed more slowly than those involved in lipogenesis.[48] Additionally, Zhu et al[43] showed that the increased hepatic fat uptake and de novo lipogenesis might be the most common cause for the formation of steatosis in NASH patients. Then, HP diets might reverse steatosis through the suppression of these two processes.

Previous studies showed that LP diets promote autophagy by a GCN2-mediated stress response to the deficiency of amino acids. Also, 'lipophagy' as a potential player of autophagy in lipid metabolism has been described in several papers.[13,49] Originally, we hypothesized that autophagy could help to eliminate the intrahepatocellular lipids in LP diets by supporting lipophagy. From the dynamic analyses of autophagy flux after 3 weeks of intervention, we confirmed that the LP group displayed significantly elevated autophagy flux, while autophagy flux in the HP group was unchanged. We conclude that, although LP diets increased autophagy levels, the ability to eliminate liver fat did not increase correspondingly. Some papers reported that autophagy is involved in the formation and growth of lipid droplets.[14] How autophagy keeps a balance between 'lipophagy' or the genesis of lipid droplets and the final effect of autophagy on hepatic lipid metabolism is still not known. In order to gain further insight into the molecular mechanism associated with autophagy, correlation analyses were performed. Expression of autophagy-related genes was positively correlated with ER-stress-related genes, and autophagy flux was positively correlated with ATF4 gene expression, which is in line with the fact that the ATF4 pathway can promote stress-induced autophagy gene expression.[50–52] However, we found no significant correlation between autophagy flux and autophagy-related gene expression which might be explained by the fact that the dynamic levels of autophagy flux (LC3B II) are correlated with the function of autophagosomes rather than the number.[53] In addition, hepatic autophagy flux was positively correlated with circulating FGF21 and hepatic FGF21 gene expression levels. Zhu et al[11] proposed that FGF21, which is potently induced by ER-stress, may up-regulate autophagy.

Our data show a lower post-diet level of ER-stress in the HP group which may be related to the stronger reduction of liver fat. However, this is difficult to assess in our study cohort because liver fat in the HP group was lower at the onset of the study. Therefore, this group may have started out with lower levels of ER-stress. However, the correlation of ER-stress markers and hepatic triglycerides persisted in the HP group indicating that the protein induced reduction in liver fat also reduced hepatic ER-stress. This is also supported by the trend towards a reduction in serum FGF21 in the course of the study. Although the FGF21 decrease upon the HP diet did not achieve the statistical significance in the current study, possibly because of the relatively low subject number, it was clearly demonstrated in our previous study in 37 subjects upon diets rich in both animal or plant proteins.[54] The reduction of protein intake may have added the stress induced by amino acid deficiency on to the preexisting ER-stress induced by obesity and hepatic steatosis and thereby may have reduced the efficiency of liver fat reduction.

The RNA-seq analysis of liver samples revealed a range of pathways which are differently regulated upon HP and LP intake and could affect NAFLD pathogenesis. As expected, GO analyses revealed that pathways linked to hepatic amino acid biosynthesis and amine metabolism were enriched in the HP group. Moreover, transcript coding ASS1, an important enzyme in the urea cycle, was expectedly up-regulated upon HP diet. We also found a range of additional pathways differently regulated upon HP and LP intake. Particularly, we found that inflammatory pathways (involved in leukocyte migration, chemotaxis, I − κB/NF − κB signaling, and ERK1/ERK2 cascade) showed lower activity upon HP diet which was confirmed by qPCR data of the expression of proinflammatory markers MCP1 and ITGAX in the liver. These data agree with our previous observation that HP diet moderately reduces proinflammatory cytokine levels in the circulation of subjects with type 2 diabetes.[3] Interestingly, inflammasome NLRP6, which expression was also affected by the protein intake, is regulated by microbial-derived metabolites and represents a promising cross-point between innate immunity and metabolic control.[55] Further studies are required to elucidate detailed molecular mechanisms of the regulation of liver inflammation by dietary protein intake.

There are some limitations of the study that should be mentioned. 19 subjects were divided into two different dietary groups prior to the intervention matched by age, gender and BMI. However, as it was mentioned above, individuals in the LP group had higher liver fat content and higher serum FGF21 and GGT levels at baseline. Considering the correlation of autophagy flux with IHL, serum FGF21, hepatic gene expression of FGF21 and ATF4 levels, the unequal liver fat content might impede the interpretation of the effect of dietary intervention. Furthermore, a higher fat accumulation in the LP group may signal a depot more resistant to mobilization upon the dietary intervention. Nevertheless, adjustment of the data on the autophagy flux and gene expression to the baseline hepatic fat content showed a similar tendency as unadjusted data. Four genes remained significantly lower in the HP group and many others showed a trend (0.1 > P>.05) to be lower expressed in the HP group (Table S5). Thus, our analysis confirmed the effect of the dietary protein intake on various mechanisms of the liver fat reduction independent from the baseline hepatic fat content.

The next limitation is the relatively small number of study subjects because of the inclusion criteria prior to a bariatric surgery. Nevertheless, according to the power calculation, our study was powered sufficiently to detect dietary-induced differences of liver fat content.

The next limitation is that protein intake during the intervention was not checked directly (eg by food protocol), so one cannot be sure about the consumed levels. However, study participants obtained food plans given as 10-d rotating menus which were easy to follow. Furthermore, we measured serum urea levels, a well-established marker of protein intake, and found a significant increase in the HP group and a significant decrease in the LP group, from where we could conclude a good compliance to the diets, although we cannot define the exact protein intake achieved. Although a 3–5 days gap was between the dietary intervention and the surgery, patients were asked to continue following the diets and their body weight remained decreased in comparison with baseline without marked regain. Data from the RP group demonstrate that even a moderate increase in protein intake leads to more effective decrease in liver fat compared to the LP group, supporting our findings and interpretation. Future studies in rodents and cell culture are needed to elucidate molecular mechanisms underlying the observed effects of dietary protein intake and the role of FGF21, inflammatory factors and other biomarkers.

In summary, HP diet had stronger effects on liver fat reduction than LP diet, even though LP diet seemed to increase glucose tolerance at least as effectively as HP diet. Although LP diet could dramatically elevate autophagy flux and FGF21 levels in liver and circulation compared to HP diets, there was no evidence that autophagy could help to eliminate the intrahepatocellular lipids. A triggered mitochondrial activity and fatty acid β-oxidation upon HP diet was also one of our mechanistic hypotheses to explain the effective liver fat reduction. However, hepatic mitochondrial activity and fatty acid β-oxidation gene expression did not change upon HP diet. Diet induced changes of other metabolic pathways, eg suppression of hepatic fat uptake, de novo lipogenesis and lipid storage upon HP diet might be explanatory for the marked liver fat reduction.