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

Disclosures

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

In This Article

Results

Clinical Characteristics of Subjects

In cohort I, 19 subjects completed LP (age of 47 ± 9 years, BMI at baseline of 46.4 ± 3.4 kg/m2) or HP (age of 49 ± 9 years, BMI at baseline of 44.2 ± 1.3 kg/m2) dietary intervention. In cohort II, 10 subjects, who consumed a moderately high-protein diet and did not undergo bariatric surgery, was selected as a reference protein group (RP, age of 56 ± 2 years, BMI at baseline of 38.5 ± 1.7 kg/m2).

After 3 weeks of hypocaloric diets, BMI decreased significantly and similarly in all groups (LP: −1.7 kg/m2, P LP = .001; HP: −1.5 kg/m2, P HP = .001; RP: −1.9 kg/m2, P RP = 3.5 × 10−5) as did weight (LP: −5.3 kg, P LP = 4.43 × 10−4; HP: −4.7 kg, P HP = 2.36 × 10−4; RP: −5.3 kg, P RP = 3.6 × 10−5). The changes in BMI among the three groups showed no significant differences (P LPvs.HPvs.RP = .324) (Table 1). At the day of surgery, body weight remained decreased in comparison with baseline (LP: −4.2 kg, P LP = .004; HP: −4.4 kg, P HP = .001).

After 3 weeks of diets, serum γ-glutamyl transferase (GGT) decreased significantly in the LP and RP group, but increased significantly in the HP group (Table 1). The levels of serum urea, as a marker for dietary protein intake, decreased dramatically after the intervention in the LP group (P LP = 1.63 × 10−4), remained unchanged in the RP group and increased significantly in the HP group (P HP = 0.031), indicating a good compliance to the dietary intervention (Table 1).

After interventions, serum levels of total, HDL-, LDL-cholesterol and triglyceride decreased in the RP group as well as total and LDL-cholesterol in the HP group (P HP = .040 and P HP = .052 respectively) (Table 1). Fasting glucose decreased significantly only in the LP group (P LP = .002), while the mean fasting serum insulin and HOMA-IR, as a marker of insulin sensitivity, decreased by about 50% in all groups (insulin: P LP = 4.25 × 10−4; P HP = .042; P RP = .010; HOMA-IR: P LP = .004; P HP = .081; P RP = .028) (Table 1).

Intrahepatic Lipid and Hepatic Triglyceride Content

In order to determine IHL levels at the beginning of the intervention and 3 weeks later, 1H-MRS analyses were performed. 21 subjects (6 in the LP group, 5 in the HP group, and 10 in the RP group) completed 1H-MRS at baseline and after 3 weeks. IHL levels of all these 21 participants before the intervention were above the threshold for NAFLD definition of 5.56% by Szczepaniak et al.[29] Baseline levels of IHL in the LP group were higher than in the HP group (P LPvs.HP = .041), although both groups had similar baseline BMI (P LPvs.HP = .387).

There were strong reductions in IHL by 42.6% in the HP group (P HP = .014) and by 36.7% in the RP group (P RP = 2.45 × 10−4). By contrast, no significant reduction in IHL in the LP group (P LP = .144) was observed. The differences in the changes in IHL levels between the LP and HP groups, and between the LP and RP groups were significant (P LPvs.HP = .001; P LPvs.RP = .001) (Figure 1A). However, these 21 participants similarly and significantly reduced their BMI (LP: −4.4%, P LP = .028; HP: −4.6%, P HP = .001; RP: −4.9%, P RP = 3.5 × 10−5), without difference between the three groups (P LPvs.HPvs.RP = .895) (Figure 1B). In addition, there was no correlation between changes of IHL and changes of BMI (r = .212, P = .356), fat mass (r = .273, P = .417), fat-free mass (r=−.173, P = .612) and HOMA-IR (r = .033, P = .900). Figure 1C moreover shows the impressively greater reductions of IHL in virtually all patients in the HP group as compared to the LP group.

Figure 1.

IHL reduced effectively in the HP group after the dietary intervention; high correlation between hepatic triglycerides and histology. (A) IHL at the baseline and after 3-week dietary intervention in the LP, HP and RP groups by 1H-MRS. nLP = 6, nHP = 5, nRP = 10. (B) Body mass index at the baseline and after 3-week dietary intervention in the LP, HP and RP groups. nLP = 6, nHP = 5, nRP = 10. (C) ΔBMI change and ΔIHL change. nLP = 6, nHP = 5. (D) Liver triglyceride levels in the LP and HP groups (quantified in liver samples and normalized to protein concentration). nLP = 10, nHP = 9. (E) Hepatic triglyceride content in different stages of NAFLD by histology (scored by SAF). nLP = 10, nHP = 9. (F) Microscopy images of liver biospies of three patients as examples in different stages of NAFLD or hepatic triglyceride content after H&E stain (scored by SAF). Abbreviations: BMI, body mass index; HP, high-protein diet; IHL, intrahepatic lipid; LP, low-protein diet; NASH, non-alcoholic steatohepatitis; SAF, steatosis-activity-fibrosis; TG, triglyceride. *P < .05, **P < .01, ***P < .001. ns, not significant (P > .05). Data are presented as mean ± SEM

The hepatic triglyceride levels in liver biopsies, which are shown in Figure 1D, were approximately twofold higher in the LP than in the HP group (P LPvs.HP = 1.09 × 10−4) and highly correlated with the IHL contents detected via 1H-MRS after the 3-week intervention (r = .790, P = .004), indicating that both methods for determining liver fat content should be considered as correct and reliable (Table 2). Hepatic triglyceride levels were significantly different between "normal", "steatosis" and "NASH" groups divided according to the SAF score (P "normal"VS."steatosis"VS."NASH" = .002) being the lowest in the "normal" group, and highest in the "NASH" group (Figure 1F-E).

LC3B II Protein Levels in Autophagy Flux

As the autophagy was shown to essentially contribute to the regulation of hepatic lipid metabolism and intracellular lipid stores,[13,14] we investigated dynamic indicators of autophagy (LC3B II flux) upon HP and LP diets. For this, fresh liver tissue was obtained during bariatric surgery and incubated with or without lysosomal proteolytic inhibitors (bafilomycin and leupeptin) in ex vivo experiments. Subsequently, LC3B II protein levels were assessed via Western blotting.

Ratios of LC3B II levels with/without proteolytic inhibitors were used for the assessment of autophagy flux. Incubation with lysosomal proteolytic inhibitors induced a significant increase in LC3B II (the lipidated form of LC3) protein levels by 66.7% in the LP group compared to the incubation without the inhibitors (P LP = .038). The average ratio with/without inhibitors was 1.83 for the LP group and 0.97 for the HP group. However, in the HP group, incubation with lysosomal proteolytic inhibitors had no effect on LC3B II (−2.8%, P HP = .322), resulting in a marked difference between the LP and HP groups (P LPvs.HP = .029; P adj = .056 after the adjustment for the baseline IHL) (Figure 2 A, B).

Figure 2.

Autophagy flux and FGF21 increased significantly in the LP group after the dietary intervention. (A-B) Fold changes of LC3 II levels (in the absence and the presence of autophagy inhibitor) for dynamic autophagy flux. nLP = 7, nHP = 6. (C) Serum FGF21 at the baseline and 3-week dietary intervention in the LP and HP groups by ELISA. nLP = 10, nHP = 7. (D) Expression levels of FGF21 pathway (FGF21, FGFR1, FGFR2 and KLB) related genes via qRT-PCR (normalized to the geometric mean of HPRT and RPLP0). nLP = 10, nHP = 9. Abbreviations: Baf, bafilomycin A1; FGF21, fibroblast growth factor 21; FGFR1, fibroblast growth factor receptor 1; FGFR2, fibroblast growth factor receptor 2; HP, high-protein diet; KLB, klotho beta; Leu, leupeptin; LC3B, microtubule-associated protein 1B-light chain 3; LP, low-protein diet. *P < .05, **P < .01, ***P < .001. ns, not significant (P > .05). Data are presented as mean ± SEM

In order to gain further insight into the metabolic mechanism associated with autophagy, bivariate correlation analyses were performed. Autophagy flux was significantly correlated with IHL (r = .786, P = .036), hepatic triglycerides (r = .764. P = .002), serum FGF21 (r = .615, P = .025), hepatic gene expression of FGF21 (r = .555, P = .049) (Table 2) and ATF4 (r = .621, P = .024). However, there was no significant correlation between autophagy flux and expression of autophagy related genes (LC3A, LC3B, and ATG5).

Serum FGF21 and Expression Levels of Hepatic FGF21 Pathway Related Genes

FGF21 is a metabolic hormone which is strongly regulated by protein restriction and autophagy and also involved in endoplasmic reticulum (ER) stress upon liver fat accumulation.[15,16] To investigate the role of FGF21 in the effects of dietary protein on the liver fat, gene expression levels of FGF21, fibroblast growth factor receptor 1 (FGFR1), fibroblast growth factor receptor 2 (FGFR2) and klotho beta (KLB) were measured after the dietary intervention in the liver samples.

During the dietary intervention, serum FGF21 levels were not significantly changed in the HP group (P HP = .220) and increased in the LP group by 42.2% (P LP = .047). The difference in the FGF21 changes between the LP and HP groups was significant (P LPvs.HP = .005) (Figure 2C). Post-intervention, hepatic gene expression levels of FGF21, FGFR1, FGFR2 and KLB were higher in the LP group compared to the HP group (FGF21: P LPvs.HP = .008; FGFR1: P LPvs.HP = .007; FGFR2: P LPvs.HP = .005; KLB: P LPvs.HP = .049) (Figure 2D). In addition, there was a positive correlation between serum FGF21 and hepatic FGF21 gene expression levels (r = .709, P = .001) (Table 2).

Hepatic Gene Expression Levels

We further investigated expression levels of genes involved in autophagy, ER-stress, fat uptake, fatty acid β-oxidation, de novo lipogenesis and lipid storage – processes involved in the regulation of liver fat accumulation - in the LP and HP groups. Expression levels of LC3A, ATG5, BiP, XBP1s, XBP1, DDIT3, LPL, ChREBP, FASN, SREBP1c, ACC1, ACC2, AMPKγ1 and SCD1 were higher in the LP group than those in the HP group (Figure 3 A, B, Table S5). Expression of genes involved in fatty acid β-oxidation was not different between groups (Figure 3 A, B).

Figure 3.

Expression of autophagy, ER stress, lipid biosynthesis and inflammatory genes were significantly higher in the LP group. (A) Expression levels of autophagy (LC3A, LC3B and Atg5) and ER stress (BiP, XBP1s, XBP1, ATF4 and DDIT3) related genes via qRT-PCR (normalized to HPRT). (B) Expression levels of fat uptake (LPL), fatty acid β-oxidation (IRS-1, MCAD, ACOX1, ACOX2, CPT1A and PPARα), de novo lipogenesis (ChREBP, FASN, SREBP1c, ACC1, ACC2 and AMPKγ1), and lipid storage (PPARγ and SCD1) related genes via qRT-PCR (normalized to the geometric mean of HPRT and RPLP0). (C) Expression of genes involved in inflammatory pathways (ITGAX, IL6, MCP1, TNFα, IL1B, IL10). nLP = 10, nHP = 9. Abbreviations: ACC1, acetyl-CoA carboxylase 1; ACC2, acetyl-CoA carboxylase2; ACOX1, acyl-CoA oxidase 1; ACOX2, acyl-CoA oxidase 2; AMPKγ1, AMP-activated protein kinase γ1; ATF4, activating transcription factor 4; Atg5, autophagy-related protein 5; BiP, binding of immunoglobulin protein; ChREBP, carbohydrate-responsive element-binding protein; CPT1A, carnitine palmitoyltransferase 1A; DDIT3, DNA damage-inducible transcript protein 3; ER, endoplasmic reticulum; FASN, fatty acid synthase; HP, high-protein diet; IL, interleukin; IRS1, insulin receptor substrate 1; ITGAX, Integrin subunit alpha x; LC3A, microtubule-associated protein 1A-light chain 3; LC3B, microtubule-associated protein 1B-light chain 3; LP, low-protein diet; LPL, lipoprotein lipase; MCAD, medium-chain specific acyl-CoA dehydrogenase; MCP1, monocyte chemoattractant protein 1; PPARα, peroxisome proliferator activated receptor α; PPARγ, peroxisome proliferator activated receptor γ; SCD1, stearoyl-CoA desaturase 1; SREBP1c, sterol regulatory element-binding protein 1c; TNFα, tumor necrosis factor alpha; XBP1, X-box binding protein 1; XBP1s, spliced X-box binding protein 1. *P < .05, **P < .01, ***P < .001. ns, not significant (P > .05). Data are presented as mean ± SEM

The expression of autophagy-related genes was positively correlated with that of ER-stress-related genes (Table 3). Furthermore, ATF4 gene expression was positively correlated with autophagy flux (r = .621, P = .024). Moreover, hepatic triglyceride content was significantly correlated with the gene expression levels of LC3A, ATG5, BiP, XBP1s, XBP1, ATF4, DDIT3 and FGF21 (Table 3).

RNA-seq-based Pathway Analysis

To gain further insight into the molecular mechanisms involved in dietary protein induced changes in liver, RNA-seq was performed (n = 10 in the LP group and n = 9 in the HP group). Differential expression analyses showed that 66 genes were higher and 70 genes were lower expressed in the HP group compared with the LP group from a total of 22 273 detected transcripts. GO analyses revealed a range of changed pathways (Table S6) including pathways linked to hepatic amino acid biosynthesis and amine metabolism which were enriched in the HP group (Table 4). Moreover, transcript coding argininosuccinate synthetase (ASS1), an important enzyme in the urea cycle, was expectedly up-regulated upon HP diet (Table S6). Furthermore, pathways involved in glucose import and glycogen biosynthesis in the liver were enriched in the HP group compared with the LP group (Table 4). In contrast, in the LP group, genes involved in triglyceride hydrolysis (LPL, FABP4, FABP5) were up-regulated compared with HP group (Table 4). Moreover, in the LP group, pathways involved in inflammation, apoptosis, chemotaxis, I − κB/NF − κB signalling, ERK1/ERK2 cascade showed higher activity and several markers of liver fibrosis (COL1A2, COL1A1) were expressed higher (Table 4). PCR analysis confirmed that the proinflammatory cytokine MCP1 and the M1 macrophage marker ITGAX (CD11c) were expressed higher upon LP diet (MCP1: P LPvs.HP = .015; ITGAX: P LPvs.HP = .043) (Figure 3C). Expression of fibrosis markers αSMA, COL1A1, COL3A1, COL6A1 and TGFβ1 showed no significant between-group differences (data not shown).

Hepatic Mitochondrial and Citrate Synthase Activity

We further hypothesized that HP diet reduced liver fat by triggering mitochondrial activity. In order to investigate the influence of dietary protein on the mitochondria of the liver, mitochondrial respiration analyses in fresh liver samples and citrate synthase (CS) activities were determined (n = 5 in the LP group and n = 5 in the HP group). Hepatic mitochondrial activity was normalized by CS activity as a marker of mitochondrial mass.[30]

There were significant differences of CS activity between "normal", "steatosis" and "NASH" groups (P "normal"VS."steatosis"VS."NASH" = .030). "NASH" group had the highest CS activity and was significantly different from the "steatosis" and the "normal" group (P "normal"vs."NASH" = .045; P "steatosis"vs."NASH" = .014). Additionally, the OXPHOS respiration and the maximal uncoupled respiration (ETSmax) were significantly lower in the "NASH" compared to the "normal" group (OXPHOS: P "normal"vs."NASH" = .050; ETSmax: P "normal"vs."NASH" = .040) (Figure 4 A-C).

Figure 4.

Liver mitochondrial and citrate synthase activity. (A) Representative trace of oxygen (O2) concentration (blue line) and O2 flux (red line) during substrate-uncoupler-inhibitor-titration (SUIT) protocol for mitochondrial respiratory capacity in human liver biopsies: Malate + Glutamate (Leak respiration), ADP (OXPHOS capacity), Cytochrome c (Integrity of outer mitochondrial-membrane), Pyruvate & Succinate (ETSCI&CII), FCCP (ETS maximum capacity), Rotenone + Antimycin A (less than 2% residual oxygen consumption, ROX). (B) Hepatic CS activity in different stages of NAFLD by histology (histology scored by SAF; normalized to protein concentration). nLP = 5, nHP = 5. (C) Hepatic mitochondrial activity in different stages of NAFLD by histology (histology scored by SAF; normalized to CS activity). nLP = 5, nHP = 5. (D) Hepatic CS activity in the LP and HP groups (normalized to protein concentration). nLP = 5, nHP = 5. (E) Hepatic mitochondrial activity in the LP and HP groups (normalized to protein concentration). nLP = 5, nHP = 5. Abbreviations: CS, citrate synthase; ETS, mitochondrial electron transport system; HP, high-protein diet; LP, low-protein diet; NASH, non-alcoholic steatohepatitis; Oxphos, oxidative phosphorylation; SAF, steatosis-activity-fibrosis; TG, triglyceride. *P < .05, **P < .01, ***P < .001. ns, not significant (P > .05). Data are presented as mean ± SEM

The CS activity, however, was not influenced by protein intake (P LPvs.HP = .735). Mitochondrial activity in the HP group was not significantly different from that in the LP group although it showed a tendency to increase in all the respiration processes in the HP group (Figure 4 D, E). However, there were strong correlations between hepatic CS activity and IHL content (r = .963, P = .002) (Table 2), and between ETSmax respiration and IHL content (r=−.906, P = .013).

processing....