Compromised hepatic mitochondrial fatty acid oxidation and reduced markers of mitochondrial turnover in human NAFLD

Abstract

Background and Aims

Nonalcoholic fatty liver disease (NAFLD) and its more advanced form steatohepatitis (NASH) is associated with obesity and is an independent risk factor for cardiovascular, liver-related, and all-cause mortality. Available human data examining hepatic mitochondrial fatty acid oxidation and hepatic mitochondrial turnover in NAFLD and NASH are scant. To investigate this relationship, liver biopsies were obtained from patients with obesity undergoing bariatric surgery and data clustered into four groups based on hepatic histopathological classification: Control (no disease), NAFL (steatosis only), Borderline-NASH (steatosis with lobular inflammation or hepatocellular ballooning), and Definite-NASH (steatosis, lobular inflammation, and hepatocellular ballooning).

Results

Hepatic mitochondrial complete fatty acid oxidation to CO₂ and the rate limiting enzyme in β-oxidation (β-hydroxyacyl-CoA dehydrogenase activity) were reduced by ~40-50% with D-NASH compared with Control. This corresponded with increased hepatic mitochondrial reactive oxygen species production, as well as dramatic reductions in markers of mitochondrial biogenesis, autophagy, mitophagy, fission and fusion in NAFL and NASH.

Conclusions

These findings suggest that compromised hepatic fatty acid oxidation and mitochondrial turnover are intimately linked to increasing NAFLD severity in patients with obesity.

INTRODUCTION

Nonalcoholic fatty liver disease (NAFLD) and its progressive form of steatohepatitis (NASH) is the most prevalent liver disease in the U.S. (1) and is an independent risk factor for cardiovascular, liver-related, and all-cause mortality (2). Moreover, there are no established, validated pharmacological therapies for NAFLD and NASH. This is, in part, attributed to poorly understood pathological mechanisms of the disease process. An increasing body of evidence highlights hepatic mitochondrial dysfunction as an important pathological discovery in NAFLD/NASH in both rodent models (3-6) and humans (7-16). Koliaki et al. described the adaption of hepatic mitochondrial function to increase bioenergetic needs in the setting of obesity and NAFLD, which was then lost with NASH (15). Additionally, indirect measures of hepatic substrate oxidation have been shown to either be normal or increased in patients with NAFLD compared to no disease (7, 9-14, 17, 18). Whether perturbations in hepatic mitochondrial long-chain fatty acid oxidation occur in patients with increasing NASH severity is unknown. Moreover, despite being well-established in rodent literature (19-21), very little is known about alterations in markers of hepatic mitochondrial turnover, mitophagy, fission and fusion in humans with increasing NAFLD severity.

Here, we simultaneously assessed ex vivo fatty acid oxidation in isolated hepatic mitochondria and whole liver using radiolabeled [1-¹⁴C] palmitate, isolated mitochondria respiration and measures of mitochondrial biogenesis, mitophagy, and dynamics in patients with varying degrees of liver disease severity. Our findings from multiple, independent outcomes demonstrate a dramatic decline in hepatic fatty acid oxidation and markers of mitochondrial biogenesis, autophagy, mitophagy, fission and fusion with increasing liver disease. These findings provide mechanistic insights into the potential role of compromised mitochondrial function and turnover in the etiology of NAFLD/NASH.

EXPERIMENTAL METHODS

Study Participants

Liver samples were obtained from adults with clinical obesity undergoing elective bariatric surgery at the University of Missouri Hospital, Columbia MO. Before inclusion, all participants gave written informed consent to the protocol, which was approved by the Institutional Review Board (IRB) of University of Missouri (protocol #2008258) and conducted according to World’s Medical Association Declaration of Helsinki. This study is registered at ClinicalTrials.gov (Identifier: NCT03151798). Degree of NAFLD severity was determined by a blinded hepatopathologist, using validated guidelines of the hepatic histopathological classification (22). Patients were grouped based on their histological classification including degree of steatosis, lobular and portal inflammation, and hepatocellular ballooning. Patients were assigned into the following groups: Control (CTRL; no disease), NAFL (steatosis only), Borderline-NASH (B-NASH; (steatosis with lobular inflammation or hepatocellular ballooning), and Definite-NASH (D-NASH; steatosis, lobular inflammation, and hepatocellular ballooning). Inclusion criteria for NAFLD patients were based on an alcohol intake less than 20 g/day and histologically confirmed steatosis with/without necroinflammation and/or fibrosis. Other causes of liver disease were excluded based on history, laboratory data, and histological features.

Blood Metabolic Panels

Following an overnight fast, blood samples were drawn prior to anesthesia before metabolic surgery and were immediately processed for biochemical measurements by a CLIA-standardized laboratory (Quest Diagnostics, St. Louis, MO, Lic.#26D0652092), according to standard procedures. Lipid measurements [total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDLc), high-density lipoprotein cholesterol (HDLc)] were performed via auto-analyzer (Roche Cobas 8000 System, CV 0.6-0.9%, Indianapolis, IN) using electrochemiluminescent immunoassay. Liver enzymes [aspartate transaminase (AST), alanine aminotransferase (ALT), and alkaline phosphatase (ALP)] were measured using UV Absorbance (Roche Cobas 8000 System, CV 0.5-3.2% for AST and 0.5-3.1% for ALT, Indianapolis, IN).

Liver Biopsies

Wedge liver biopsies (200–700 mg tissue) were obtained by surgeons ~30 minutes after initiation of anesthesia according to standardized protocols. Approximately 200 mg was placed in ice-cold mitochondrial isolation buffer (220mM Mannitol, 70mM sucrose, 10 mM tris-base, 1 mM EDTA; pH 7.4) for high resolution respirometry and fatty acid oxidation, ~10 mg was fixed in 1% formaldehyde for histological examination, 1-2 mg was fixed in 2% paraformaldehyde, 2% glutaraldehyde in 100 mM sodium cacodylate buffer pH = 7.35 for transition electron microscopy, and the remaining tissue was snap-frozen in liquid nitrogen and stored at −80°C. According to standard techniques, liver histology was performed by an experienced pathologist using hematoxylin-eosin and Masson’s trichrome staining.

Hepatic Mitochondrial Isolation

Hepatic mitochondria were prepared by differential centrifugation as previously described (3, 23). The mitochondrial pellet was resuspended in MiPO₃ buffer (0.5 mM EGTA, 3 mM MgCl₂·6H₂0, 60 mM K-lactobionate, 20 mM Taurine, 10 mM KH₂P0₄, 20mM HEPES, 110 mM Sucrose, 1g/l BSA, 20 mM Histidine, 20 μM vitamin E succinate, 3 mM glutathione, 1 μM leupeptine, 2 mM glutamate, 2 mM malate, 2 mM Mg-ATP) for mitochondrial respiration. Protein concentration was determined via bicinchoninic acid reaction using Pierce bicinchoninic acid protein assay (no. 23225, ThermoFisher Scientific).

Hepatic Mitochondrial Respiration

Mitochondrial respiration was assessed using high-resolution respirometry (Oroboros Oxygraph-2k; Oroboros Instruments; Innsbruck, Austria) and values were corrected to total mitochondrial protein loaded, as previously described (24). Isolated mitochondria (100-150 μg protein) were placed in respiration chambers in respiration media (MiR05; sucrose, 100 mM; K-lactobionate, 60 mM; EGTA, 0.5 mM; MgCl₂, 3 mM; taurine, 20 mM; KH₂PO₄, 10 mM; HEPES, 20 mM; adjusted to pH 7.1 with KOH at 37C; and 1 g/L fatty acid free BSA) for assessment of basal respiration at 37°C. State 2, State 3, and maximal coupled and uncoupled O₂ flux rates were assessed upon sequential exposure to mitochondrial substrates and titrating ADP concentrations, using three separate protocols – malate-glutamate stimulated, malate-palmityolcarnitine stimulated, and malate-octanoylcarnitine stimulated. Due to limited sample availability, all protocols were not run in all samples. For details on substrate concentrations used and subject numbers see Supplementary Table 1.

Mitochondrial Content and Structure

Hepatic citrate synthase activity was determined as previously described (6). Briefly, liver homogenates were incubated in the presence of oxaloacetate, acetyl-CoA, and DTNB. Spectrophotometric detection of reduced DTNB at a wavelength of 412 nm served as an index of enzyme activity. Transmission electron microscopy (TEM) was used to visually assess mitochondrial content and ultrastructural differences. All tissue preparation and imaging were performed at the Electron Microscopy Core Facility, University of Missouri (Columbia, MO). See supplemental material for extended methods.

Fatty Acid Oxidation and β-hydroxyacyl-CoA dehydrogenase (β-HAD) activity

Fatty acid oxidation was measured with radiolabeled [1-¹⁴C] palmitate (American Radiochemicals) in fresh liver homogenate and isolated hepatic mitochondria, using methods from (6) and as previously described by (25, 26). Hepatic β-HAD activity was assessed as previously described by our group (6).

Hepatic Oxidative Stress

H₂O₂ emission in isolated hepatic mitochondria was measured via oxygen consumption and the addition of AmplexTM UltraRed reagent (#A36006, Thermo Fisher Scientific), reflecting reactive oxygen species production from complexes I and III at basal levels and following stimulation with palmitoyl-CoA (at increasing titrations 20 – 70 μM), using the Oroboros and as described previously (27). Due to smaller sample sizes data from Control and NAFL were pooled for H₂O₂ emission (Control + NAFL, n = 20; Borderline-NASH, n = 17; Definite-NASH, n = 35). Superoxide dismutase (SOD) activity was measured in hepatic tissue lysates according to manufacturer’s instructions (Cayman Chemical Company).

Western Blotting

Western blot analysis was completed in whole liver homogenate. Western blot sample preparation and methods have been described previously by our group (6, 28, 29). Liver samples were homogenized using lysis buffer. Protein (10-20 ug) was loaded into an SDS-PAGE gel and probed with primary antibody. After washing, the membrane was incubated with horseradish peroxidase-conjugated secondary antibodies. Protein bands were quantified using a densitometer (Bio-Rad, Hercules, CA). To control for equal protein loading and transfer, the membranes were then stained with 0.1% amido-black (Sigma) and this total protein staining for each lane was quantified by densitometry and used to correct for any differences in protein loading or transfer of all band densities. The intensities of the bands and total protein staining were quantified using Image Lab Software. For a list of primary antibodies utilized see supplemental information.

Gene Expression

RNA and cDNA samples from whole liver tissue were prepared as previously described (24). Quantitative real-time PCR was completed with the ABI 7500 Fast Sequence Detection System (Applied Biosystems, Carlsbad, CA) using iTaq Universal SYBR Green Supermix (Bio-Rad).

GAPDH was selected as the housekeeping gene as it was the most stable transcript and did not differ significantly among the four groups (P = 0.536). Results were quantified using the ddC_T technique relative to the GAPDH housekeeping gene. Values were expressed relative to the CTRL group. All samples were run in triplicate on a 96 well plate. To reduce inter-plate variability, each plate contained a control sample that was used to adjust the CT thresholds of each plate to the same CT value. Primer pairs were obtained from Sigma (St. Louis, MO) and primer sequences are listed in Supplementary Table 2.

Ingenuity Pathway Analysis

For examination of the top up- and down-regulated proteins and the corresponding top pathway network, protein markers obtained from Western blot analysis that were significantly different between D-NASH vs CTRL following a one-way ANOVA and follow up post-hoc analysis were included in Ingenuity Pathway Analysis (IPA; Ingenuity Systems, Redwood, CA). Additionally, the Gene Ontology (GO) Consortium database was used to generate transcript lists of changes in relevant biological processes, cellular components, and molecular functions.

Statistical Analysis

Data was grouped as follows: Control, n = 13; NAFL, n = 34; Borderline-NASH, n = 27, and Definite-NASH, n = 56. To adjust for sex differences, comparisons were performed with one-way analysis of covariance (ANCOVA) with sex as a covariate, and significant main effects were followed up with Sidak post hoc analyses. Differences were considered statistically significant at P ≤ 0.05. Additionally, a secondary analysis was performed to evaluate the influence of diabetes status with a one-way ANCOVA with sex and HbA1c as covariates. Where there were significant differences when adjusting for sex and HbA1c, results are reported in the text. Correlations were analyzed using Pearson correlation. Data were also grouped based on histological fibrosis scores: F0-F1, n = 94; F2-F4, n = 12 and comparisons were performed using a one-way ANCOVA with sex as a covariate. Additional adjustments for multiple comparisons were performed for mRNA and western blot outcomes with Benjamini-Hochberg false discovery rates (FDR) at P < 0.10 as previously reported (30-32). All of the mRNA and western blot outcomes that were significant from the ANCOVA remained significant after FDR calculations. Data was graphed using GraphPad Prism 9.0 and analyses were performed using IBM SPSS Statistics Version 28.0.1.0. All data are presented as means ± standard error.

Contact for Reagent and Resource Sharing

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, R. Scott Rector (rectors@health.missouri.edu).

RESULTS

Group Characteristics

NAFLD severity was determined according to standard clinical definitions (22) based on the degree of histological hepatic steatosis, lobular inflammation, and hepatocellular ballooning present. Participants (n =130) were clustered into four groups based on NAS: No disease/control (CTRL, n = 13), Nonalcoholic fatty liver (NAFL, n = 34), Borderline-NASH (B-NASH, n = 27), and Definite-NASH (D-NASH, n = 56) disease (Fig. 1A & C). Individuals with definite NASH presented with significantly greater histological fibrosis compared with all other groups (Fig. 1B & D; P ≤ 0.05). Degree of NAFLD severity was also determined using the NAFLD activity score (NAS) (33) (Fig 1C). There was a significant main effect for HbA1c (P ≤ 0.05; Table 1) and a trend toward significance for plasma glucose concentrations (P = 0.10; Table 1). Similarly, there was a significant main effect across groups for aminotransferase enzymes, AST and ALT, markers of hepatic injury (P ≤ 0.05; Table 1). Groups did not differ in age, body mass, body mass index (BMI), or by fasting serum TG, LDL-cholesterol, or total cholesterol (Table 1; P > 0.05).

Figure 1.

Liver phenotype and measures of oxidative stress in humans. A) Representative liver H&E and B) trichrome staining, C) histological steatosis, inflammation, and ballooningNAFLD activity score, and D) histological fibrosis score. Values are presented as mean ± SE. CTRL = Control, NAFL = Nonalcoholic fatty liver, B-NASH = Borderline NASH, D-NASH = Definite-NASH. δ = main effect (P ≤ 0.05). *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Table 1.

Subject Characteristics (Mean ± SE)

	CTRL (n=13)	NAFL (n=34)	B-NASH (n=27)	D-NASH (n=56)	ANCOVA p-value
Age (years)	39±2	45±2	49±2	46±1	0.178
Sex (M/F)	0/13	5/29	7/20	9/47	-
Weight (kg)	129.5±7.7	138.2±5.6	134.8±4.2	138.2±3.6	0.787
BMI (kg/m2)	46.7±2.3	49.2±1.2	48.2±1.0	49.0±1.0	0.708
-	-	-	-	-	-
Glucose (mg/dL)	85±4.2	99.7±4.4	103.6±7.2	107.7±4.0	0.102
HbA1C (%)	5.5±0.2	6.0±0.2	6.0±0.2	6.5±0.2	0.038
AST (U/L)	23.2±1.5	27.1±2.1	31.1±2.0	41.7±5.4	0.043
ALT (U/L)	25.1±2.6	31.2±3.5	31.2±3.5	47.7±5.8	0.028
-	-	-	-	-	-
Cholesterol (mg/dL)	165.0±9.3	161.1±6.9	164.7±7.0	171.8±5.5	0.628
Triglycerides (mg/dL)	110.9±9.3	120.0±11.1	129.9±11.4	141.6±6.8	0.162
HDLc (mg/dL)	49.6±3.6	41.1±2.4	37.3±1.2	40.8±1.3	0.025
LDLc (mg/dL)	93.2±6.1	96.0±7.0	134.8±4.2	102.7±5.4	0.752

ANCOVA with sex as a covariate, followed with post hoc analysis: *p<0.05 vs CTRL. CTRL = Control, NAFL = Nonalcoholic fatty liver, B-NASH = Borderline NASH, D-NASH = Definite-NASH.

Reduced hepatic mitochondrial fatty acid oxidation and NAFLD

In rodent models, we have previously shown that NAFLD development and progression corresponds with impaired hepatic fatty acid oxidation (6). Isolated hepatic mitochondrial complete [1-¹⁴C] palmitate oxidation to CO₂ was significantly reduced with worsening NASH (D-NASH) compared to no disease (CTRL; Fig. 2D, P ≤ 0.05). In addition, there was a main effect for differences in complete [1-¹⁴C] palmitate oxidation in whole liver samples (Fig 2A, P<0.05), with post-hoc analyses revealing significant reductions in D-NASH versus NAFL (Fig. 2A; P ≤ 0.05), which remained significantly lower when adjusting for Sex and HbA1c as covariates (P ≤ 0.05). There was also a main effect for differences in incomplete and total [1-¹⁴C] palmitate oxidation in whole liver samples, with strong trends for differences in D-NASH vs CTRL (Fig. 2A). On the other hand, hepatic mitochondrial incomplete and total palmitate oxidation were significantly elevated in D-NASH relative to NAFL (Fig. 2D; P ≤ 0.05). It is also noted that complete, incomplete and total [1-¹⁴C] palmitate oxidation in whole liver samples were significantly lower (P<0.5) in B-NASH and D-NASH compared with CTRL when not adjusted for sex as a covariate (statistics not indicated in Fig 2A). In addition, the presence of lobular inflammation or hepatocellular ballooning was associated with lower complete, incomplete and total [1-¹⁴C] palmitate oxidation in whole liver samples compared with no histological inflammation or ballooning (P < 0.05, data not shown). In the context of histological fibrosis, increased hepatic fibrosis (F2-F4 vs F0-F1) was also accompanied by a significant reduction (~50%) in complete palmitate oxidation in isolated hepatic mitochondria (Fig. 2F, P ≤ 0.05). To compliment these findings, whole liver and isolated mitochondrial β-HAD activity, the rate limiting enzyme in fatty acid oxidation, was also reduced by ~50% across all levels of NAFL/NASH versus Control (Fig. 2B & E, P ≤ 0.05).

Figure 2. Whole liver and isolated hepatic mitochondria ex-vivo fatty acid oxidation in humans.

Whole liver A) complete oxidation to CO₂, incomplete, and total [1-¹⁴C] palmitate oxidation. B) Whole liver β-HAD activity and C) Whole liver complete palmitate oxidation across hepatic fibrosis scores. Isolated hepatic mitochondria D) complete oxidation to CO₂, incomplete, and total [1-¹⁴C] palmitate oxidation. E) Isolated hepatic mitochondria β-HAD activity and F) Isolated hepatic mitochondria complete palmitate oxidation across hepatic fibrosis scores. Values are presented as mean ± SE. CTRL = Control, NAFL = Nonalcoholic fatty liver, B-NASH = Borderline NASH, D-NASH = Definite-NASH. δ = main effect (P ≤ 0.05), *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

NAFLD, hepatic mitochondrial respiration, and increased oxidative stress

Previous studies have shown reduced hepatic mitochondrial respiration and OXPHOS complex activity with NASH in humans (15, 34). Here we demonstrate that malate + glutamate, malate + palmitoylcarnitine, and malate + octanoylcarnitine-stimulated hepatic mitochondrial respiration did not differ across groups (Fig. 3A, P > 0.05) when controlling for sex or sex+HbA1c as covariates. However, when examining the presence or absence of individual components of NASH (steatosis, inflammation, or ballooning), patients with hepatocellular ballooning exhibited elevated malate + glutamate and malate + octonylcarnitine-stimulated state 3-complex I+II and FCCP-stimulated maximal uncoupled respiration vs no ballooning (P < 0.05 from ANCOVA, data not shown). There were no differences in carnitine palmitoyltransferase 1A (CPT1A) mRNA expression or protein content in whole liver lysates (data not shown). Hepatic mitochondrial H₂O₂ emission was significantly elevated (>30%) with D-NASH relative to B-NASH, NAFL, and CTRL (Fig. 3B, P ≤ 0.05). Hepatic mitochondrial H₂O₂ emission was also significantly elevated with increases in hepatocellular inflammation and ballooning when adjusted for sex and sex + HbA1c covariates (Fig. 3B, P ≤ 0.05), but not with elevated histological steatosis (P > 0.05, data not shown). Elevated mitochondrial H₂O₂ emission was not related to differences in superoxide dismutase (SOD) activity or markers of oxidative stress [nuclear factor erythroid 2-related factor 2 (NRF2), catalase, glutathione peroxidase (GPx)-1, GPx-4, SOD1, and SOD2] in whole liver (Supplementary Fig. 1A-C, P > 0.05).

Figure 3. Measures of hepatic mitochondrial function, content and dynamics in humans with NAFL, B-NASH and D-NASH.

O₂ consumption in isolated hepatic mitochondria A) malate + glutamate-stimulated, malate + palmitoylcarnitine-stimulated, and malate + octanoylcarnitine-stimulated; followed by the addition of adenosine diphosphate (ADP), succinate, and carbonyl cyanide 4-trifluoromethoxy-phenylhydrazone (FCCP). B) Palmitoyl-CoA stimulated H₂O₂ emission in isolated hepatic mitochondria across NAFLD and comparisons across histological inflammation and histological ballooning. C) Whole liver citrate synthase activity. D) Mitochondrial OXPHOS complexes protein content across hepatic fibrosis scores measured in whole liver, and representative Western blot images. E) Mitochondrial biogenesis markers mRNA expression in whole liver. F) Pgc1α mRNA expression across fibrosis scores. All Western blots were run on continuous gels. Values are presented as mean ± SE. CTRL = Control, NAFL = Nonalcoholic fatty liver, B-NASH = Borderline NASH, D-NASH = Definite-NASH. δ = main effect (P ≤ 0.05), P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Mechanisms underlying reduced hepatic fatty acid oxidation with NASH were tested through investigation of markers of hepatic mitochondrial content and biogenesis. Hepatic citrate synthase activity, a surrogate measure of mitochondrial mass/content, was not significantly different across groups (Fig. 3C, P > 0.05). Similarly, protein content of hepatic oxidative phosphorylation (OXPHOS) complexes did differ not across groups (Supplementary Fig. 1D, P > 0.05). However, increasing histological fibrosis did correspond with a loss in the OXPHOS complexes CII, CIII, and CIV (F0-F1 versus F2-F4; Fig. 3D, P ≤ 0.05). These findings are in line with previous work demonstrating no significant differences in OXPHOS complexes protein content in patients with NASH compared to lean controls (15). Whereas, Pérez-Carreras and colleagues demonstrated a significant reduction in OXPHOS enzyme activity in patients with NASH compared to controls (34).

Assessment of mitochondrial biogenesis markers revealed a dramatic attenuation in peroxisome proliferative activated receptor gamma coactivator 1α (PGC1α) mRNA expression in B-NASH and D-NASH compared with CTRL and NAFL (Fig. 3E, P ≤ 0.05) and remained significant when adjusting for sex and HbA1c. In addition, PGC1α mRNA expression was significantly reduced with elevated liver fibrosis (F0-F1 versus F2-F4; Fig. 3F, P < 0.05). Similarly, sirtuin 1 (SIRT1), 5'-AMP-activated protein kinase α (AMPKα) mRNA expression was significantly lower by 20% in D-NASH versus CTRL (Fig. 3E, P ≤ 0.05). Further, AMPKα mRNA expression remained significantly lower when adjusting for sex and HbA1c (P ≤ 0.05).

Worsening NAFLD is linked to abnormalities in mitochondrial structure and morphology

Mitochondrial morphology was determined via TEM and revealed that patients with B-NASH and D-NASH presented with more round, swollen, hypodense mitochondria with loss of cristae, compared to CTRL (Fig. 4). These findings correspond with previous analyses of biopsied liver from patients with NASH showing abnormal mitochondrial ultrastructure compared to controls (35-38).

Figure 4. Mitochondrial structure and morphology with increasing NAFLD/NASH.

Representative transmission electron microscopy images of whole liver in patients CTRL, B-NASH, and D-NASH at A) 5,000x and B) 10,000x magnification.

NAFLD and NADH are linked to a loss in markers of hepatic mitochondrial turnover and dynamics

Given the dramatic reduction in markers of hepatic mitochondrial biogenesis, we next examined the potential influence of increasing liver disease severity on markers of autophagy and mitophagy. Hepatic markers of macro-autophagy, including Unc-51 like autophagy activating kinase 1 (ULK1), serine phosphorylated ULK1 (pULK1^s555), Beclin1, autophagy related protein (ATG) 5, ATG12:5, and P62, were significantly lower in D-NASH compared to CTRL (Fig. 5A, P ≤ 0.05). In addition, increasing fibrosis (F0-F1 versus F2-F4) was associated with a loss in P62 protein content (Fig. 5C, P ≤ 0.05). On the other hand, the protein content of ATG7 and microtubule associated protein 1 light chain 3 A/B (LC3) I and II were not significantly different with D-NASH compared with CTRL (Fig. 5A, P ≤ 0.05). Further, increasing disease severity was not associated with significant changes in LC3II/LC3I ratio, a marker of LC3 activation (Fig. 5A, P > 0.05); however, LC3II/LC3I ratio was significantly upregulated with increased hepatic fibrosis (F0-F1 versus F2-F4; Fig. 5B, P ≤ 0.05). Further, ULK1, ATG5, ATG12:5, and P62 remained significantly lower in D-NASH compared with CTRL after further adjustment for sex + HbA1c (P ≤ 0.05).

Figure 5. Measures of autophagy and mitophagy in whole liver.

A) Protein content of autophagy markers in whole liver and representative Western blot images. B) LC3II/LC3I ratio across fibrosis scores and C) P62 protein content across fibrosis scores. D) BNIP3, PARKIN, and PINK1 protein content in whole liver and representative Western blot images. E) Bnip3 mRNA expression across fibrosis scores. F) Markers of mitochondrial fission and fusion protein content in whole liver and representative Western blot images. All Western blots were run on continuous gels. Values are presented as mean ± SE. CTRL = Control, NAFL = Nonalcoholic fatty liver, B-NASH = Borderline NASH, D-NASH = Definite-NASH. δ = main effect (P ≤ 0.05), *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

In rodent models, reduced mitophagic processes correspond with NAFLD development and progression (39), but to our knowledge these processes have not been previously explored in human NASH. Remarkably, hepatic BCL2 interacting protein 3 (BNIP3) protein content was 35-40% lower across all stages of NAFLD compared with CTRL when adjusted for sex and sex + HbA1C (Fig. 5D, P ≤ 0.05). Similar reductions were exhibited with increasing fibrosis score (Fig. 5E, P ≤ 0.05). Interestingly, hepatic Parkin and PTEN induced kinase 1 (PINK1) protein content did not differ across groups (Fig. 5D & G, P > 0.05). Markers of mitochondrial dynamics revealed significantly lower hepatic mitochondrial fission [Dynamin-related protein 1 (DRP1) and serine 616 phosphorylated DRP1 (pDRP^S616)] and fusion [OPA1 Mitochondrial Dynamin Like GTPase (OPA1), mitofusin (MFN) 2] with increased disease when adjusted for sex and sex + HbA1C (Fig. 5F, P ≤ 0.05).

Mitochondrial functional measures are correlated with markers of mitochondrial quality control

The important role of mitochondrial quality control in the maintenance of hepatic mitochondrial metabolism was evident in the data (Fig. 6). Of note, lower markers of autophagy, mitophagy, mitochondrial biogenesis, fission, and fusion correlated with worsening ex vivo complete, incomplete and total FAO in whole liver (P ≤ 0.05). Further, lower mitochondrial respiration was significantly related to lower markers of autophagy, mitochondrial biogenesis, and fission, and significantly negatively related to markers of oxidative stress (P ≤ 0.05). Finally, higher palmitoyl-CoA stimulated H₂O₂ emission correlated significantly with lower markers of mitochondrial health (oxidative stress, autophagy, and mitochondrial fission) (P ≤ 0.05).

Figure 6. Correlational Analysis.

Correlation Heat map - showing Pearson’s correlation coefficients of functional ex vivo measures of hepatic fatty acid oxidation, O₂ respiration, and ROS with protein content of markers related to mitochondrial content, quality control and oxidative stress. Protein content was measured via Western blot. Significant correlations are indicated on heat map with r value and significance level. Red indicates a positive correlation; blue indicates a negative correlation between measures. *P ≤ 0.05, **P ≤ 0.01, ***P ≤ 0.001, ****P ≤ 0.0001.

Ingenuity Pathway Analysis (IPA) of protein markers related to mitochondrial biogenesis, autophagy/mitophagy and mitochondrial dynamics

To determine how NASH is related to hepatic mitochondrial health, differentially expressed proteins were entered into IPA software. The significant downregulation in hepatic protein content of autophagy (ULK1, SQSTM1, ATG5), mitophagy (BNIP3), and mitochondrial dynamics (MFN2, OPA1) markers reveal a reduction in processes related to ‘cellular homeostasis, formation of autophagosomes, macroautophagy of cells, autophagy of mitochondria, and inflammation of organ’ with increasing NAFLD/NASH (Fig. 7; D-NASH versus Control). Although not presented here, similar outcomes were observed for B-NASH versus CTRL and NAFL versus CTRL (data not shown), thus highlighting the intricate role these pathways play in early NAFLD and its more advanced form NASH.

Figure 7. Ingenuity Pathway Analysis network revealing changes in the regulation of disease and biological functions for D-NASH versus CTRL.

Protein content of markers related to mitochondrial content, quality control and oxidative stress were measured via Western blot. Significant differences were determined using a one-way ANCOVA, with significance level set a P ≤ 0.05, and followed up with a Tukey post-hoc analysis. Values shown in Figure 7 are fold change and p-value of the individual proteins. CTRL = Control, D-NASH = Definite-NASH.

DISCUSSION

Human data investigating the role of hepatic mitochondrial fatty acid oxidation and markers of hepatic mitochondrial turnover in liver disease progression are limited. We provide novel evidence that not only is hepatic fatty acid oxidation reduced in NASH, but also multiple markers of mitochondrial biogenesis, autophagy, mitophagy, fission and fusion are reduced with increasing NAFLD severity and hepatic fibrosis in humans with obesity.

Whole liver tissue and hepatic mitochondria long chain complete fatty acid oxidation and β-HAD activity were dramatically reduced in humans with worsening NASH, findings supported by our previous work in rodent models (6). On the other hand, malate + glutamate, malate + palmitoylcarnitine, and malate + octanoylcarnitine stimulated mitochondrial respiration did not differ across NAFLD/NASH groups. However, the presence of hepatocellular ballooning was associated with elevated malate + glutamate and malate + octanoylcarnitine stimulated respiration. These findings somewhat differ from work by Roden and colleagues (15) who found a compensatory upregulation in malate + octanoylcarnitine-stimulated hepatic mitochondrial respiration in patients with NAFLD and obesity that was largely lost in patients with NASH. In addition, our data also differ from some (14, 17, 40, 41), but not all (12) in vivo studies showing elevated hepatic TCA cycle flux or 13C-octanoate oxidation in humans with increased intrahepatic triglyceride content compared with lean controls. Differences compared to previous reports likely relate to a lack of lean control comparison in our cohort. Furthermore, Roden et al (15) reported ~25% higher hepatic mitochondrial mass (assessed by citrate synthase activity) in individuals with NASH compared to individuals with obesity, NAFL, and lean controls; findings not witnessed in the current study. Collectively, there is evidence for altered hepatic mitochondrial function in the setting of obesity (15, 42), insulin resistance (18, 43) and NAFLD/NASH (14, 15, 34, 38). However, our data collected simultaneously in the same patient population with similar BMIs, suggests that long chain fatty acid oxidation and the rate limiting step in β-oxidation are downregulated with increasing disease severity in the setting of NAFL/NASH. These data highlight the possibility of mitochondrial function and quality control playing a potentially causal role in advancement of liver disease and warrants future investigation.

Reduced hepatic mitochondrial fatty acid oxidation with D-NASH was not accompanied by prominent differences in assessed markers of hepatic mitochondrial content. Existing literature has either reported no change or an increase in hepatic mitochondrial content with NAFLD/NASH (15, 38), thus further supporting the notion that worsening NAFLD is not likely explained simply by a reduction in mitochondrial content/mass but rather is associated with impaired mitochondrial function and mitochondrial quality control. In support, mitochondrial ultrastructural abnormalities were noted with B-NASH and D-NASH in the current study, supporting existing literature of deterioration in hepatic mitochondria ultrastructural integrity with increasing NAFLD severity (35-38). Further, worsening fibrosis corresponded with reductions in hepatic mitochondrial OXPHOS protein content. Deteriorating mitochondrial morphology may be indicative of increasing bioenergetic or oxidative stress, and perturbations in machinery that regulate mitochondrial biogenesis, mitophagy, fission, and fusion (44). Indeed, the perturbations in mitochondrial function shown here likely explain the higher mitochondrial H₂O₂ emission observed with D-NASH and increasing histological lobular inflammation and hepatocellular ballooning. Taken together, these data highlight that NAFLD and its more advanced form NASH, are linked to increasing loss of hepatic mitochondrial function, reduced mitochondrial health, and elevations in hepatic oxidative stress.

Mitochondrial health and content are governed by the production of new healthy mitochondria (mitochondrial biogenesis) and the degradation of poorly functioning mitochondria (mitophagy) (39). Multiple markers of hepatic mitochondrial biogenesis (PGC1α, SIRT1, and AMPK) were downregulated with elevated NAS, indicating an impairment in these processes with worsening disease. PGC-1α is considered the master transcriptional regulator of mitochondrial biogenesis and is reduced in patients with obesity and NASH (15, 45). Furthermore, we have previously demonstrated that liver-specific PGC-1α overexpression increases hepatic mitochondrial function and reduces hepatic lipid accumulation in vivo and in vitro in rats (20, 46). The present observations point to the importance of hepatic mitochondrial biogenesis in the regulation of a healthy mitochondrial pool and in susceptibility to NAFLD/NASH in humans.

To our knowledge, we demonstrate for the first time that NAFL (steatosis only) and more advanced NASH are associated with a loss in hepatic protein content of several markers related to autophagy (ULK1, pULK1, Beclin1, ATG5, ATG12:5 and P62) compared to CTRL in patients with obesity. Reductions in autophagy and mitophagy processes are associated with an increasing pool of dysfunctional mitochondria in the liver (47). Rodent models of diet-induced NAFLD/NASH exhibit reduced autophagy (19, 48), with direct deletion of autophagy related proteins causing NAFLD and NASH (48, 49). Moreover, our findings also highlight that onset of NAFLD, more advanced NASH, and fibrosis corresponded with a loss in hepatic BNIP3 protein content, a key regulator of mitophagy. BNIP3 is known to play a functional role in the regulation of lipid metabolism and fatty liver disease in rodent models (27, 50). Further, the adaptor proteins, BNIP3 and P62 - which play a role in LC3 facilitated mitophagy, were downregulated with increasing fibrosis, despite no changes in LC3II/I ratio. Together, these data highlight that autophagy/mitophagy is interrupted in the early onset of NAFLD and even preceded losses in hepatic mitochondrial fatty acid oxidation and may also play a role in fibrogenic processes with more advanced disease. To further evaluate this point, the downregulation in several of these proteins correlated with reduced functional outcomes of mitochondrial fatty acid oxidation and respiration (Fig. 6), thus supporting the importance of these processes in maintaining hepatic mitochondrial health and metabolism.

Mitochondria are dynamic organelles that constantly undergo fusion and fission processes in order to maintain a healthy mitochondrial pool, and changes in mitochondrial dynamics are associated with adaptations to cellular apoptotic, autophagic and bioenergetic processes (51, 52). Here we show that pDRP^S616 protein content, a marker of mitochondrial fission activation, was significantly lower with NAFL onset and D-NASH. DRP1 is reduced in rodent models of diet-induced NASH (53), and liver-specific deletion causes mitochondrial enlargement and loss of fatty acid oxidation (21). Similarly, markers of hepatic mitochondrial fusion (OPA1 and MFN2) were significantly reduced with NAFL onset and worsening NASH, in line with previous studies in rodents with NAFLD (54, 55). Additionally, the inhibition of repeated fission/fusion cycles results in mitophagic arrest and the subsequent accumulation of damaged mitochondria and decreased function (56, 57). These findings collectively highlight that a loss in mitochondrial quality control likely contributes to NAFLD onset and a worsening NASH phenotype, and future mechanistic studies in human NAFLD pathogenesis are warranted.

Despite the cross-sectional nature of this study, we demonstrate through multiple, independent lines of evidence that increasing NAFLD severity corresponded with a loss in hepatic fatty acid oxidation and an impairment in mitochondrial biogenesis, mitophagy, and dynamics, likely resulting in a stagnant dysfunctional hepatic mitochondria pool. Lower markers of mitochondrial turnover were present prior to reduced hepatic fatty acid oxidation and increased ROS production, suggesting that these reductions could trigger further deterioration of mitochondria and subsequent NASH. These data collectively highlight a critical and novel role for fatty acid oxidation and mitochondrial turnover in the liver and indicate potential future targets for prevention and treatment of NAFLD/NASH in humans.

List of Abbreviations:

ALP

alkaline phosphatase

ALT

alanine aminotransferase

AMPKα

5'-AMP-activated protein kinase α

AST

aspartate transaminase

ATG

autophagy related protein

β-HAD

β-hydroxyacyl-CoA dehydrogenase

BMI

body mass index

BNIP3

BCL2 interacting protein 3

CPT1A

carnitine palmitoyltransferase 1A

DRP1

Dynamin-related protein 1

FCCP

Carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone

GO

Gene Ontology

GPx

glutathione peroxidase

HDLc

high-density lipoprotein cholesterol

IPA

Ingenuity Pathway Analysis

IRB

institutional review board

LC3

microtubule associated protein 1 light chain 3 A/B

LDLc

low-density lipoprotein cholesterol

MFN

Mitofusin

NAFLD

nonalcoholic fatty liver disease

NAS

NAFLD activity score

NASH

nonalcoholic steatohepatitis

NRF2

nuclear factor erythroid 2-related factor 2

OXPHOS

oxidative phosphorylation

PGC1α

peroxisome proliferative activated receptor gamma coactivator 1α

PINK1

PTEN induced kinase 1

PPARα

peroxisome proliferative activated receptor α

SIRT

sirtuin

SOD

superoxide dismutase

TC

total cholesterol

TEM

Transmission electron microscopy

TG

triglycerides

ULK1

Unc-51 like autophagy activating kinase 1