Non-alcoholic fatty liver disease (NAFLD) covers a broad range of liver disease from simple steatosis to steatohepatitis (NASH) to fibrosis or even cirrhosis. Today, the nature of the disease is supposed to be multifactorial including not only the accumulation of hepatic lipids but also accumulation of toxic lipid or reactive species, insulin resistance, or inflammation [1
]. In early stages, the accumulation of hepatic lipids is often associated with specific hepatic insulin resistance, while in aggravated lipid accumulation, systemic insulin resistance occurs. Accumulating lipids are most likely derived from increased hepatic lipid production, increased uptake from adipose tissue lipolysis, or diet [3
In physiology, the crosstalk between insulin-sensitive tissues defines a finely coordinated network to maintain systemic metabolism. Excess amounts of circulating lipids, as present in obesity, lead to adipose tissue dysfunction, which results in ectopic accumulation of excess lipids in peripheral organs like the liver [6
]. The increasing amounts of hepatic lipids cause lipotoxicity-mediated changes in hepatic metabolism, which consequently lead to hepatic or even systemic insulin resistance [6
]. Although the pathogenesis of NAFLD is extensively studied, it remains still unclear which mechanisms lead to the onset and progression of simple steatosis or more severe forms of the disease.
Recently, data-driven analysis clustered patients with diabetes to distinct heterogeneous subtypes to improve risk prediction for associated diseases in individual patients [8
]. One of the identified clusters showed a close relation between insulin resistance and the accumulation of hepatic lipids and, consequently, an increased risk for progression of fatty liver disease [8
]. In addition, several studies assign genetic predisposition to the pathogenesis of fatty liver disease and associated complications [10
]. Genetic studies clearly help to understand the pathogenesis of NAFLD and associated complications, but these association studies often lack translation to functional levels to assess physiological and pathological consequences.
In order to identify drivers of NAFLD pathogenesis, our study aimed to identify gene regulatory networks, which determine progression of hepatic lipid accumulation to the point of no return. Here, we used two mouse models with tissue-specific overexpression of the sterol regulatory element-binding protein (SREBP)-1c to discriminate genetic fatty liver mainly derived from increased hepatic de novo lipogenesis and metabolic fatty liver caused by excessive fatty acid influx and ectopic lipid accumulation [11
The transcription factor SREBP-1c is a central regulator of genes involved in lipid and cholesterol synthesis. SREBP-1c is the predominant isoform in lipid metabolism, especially in the activation of de novo lipogenesis (DNL) [13
]. The regulation of SREBP-1c proteins is complex [15
]. They are regulated on the transcriptional level, by a coordinated proteolytic release of the transcriptional active domain from a precursor molecule, and post-translational modification to regulate transcriptional activity and stability. Each step of this orchestrated regulation integrates information about the metabolic status of a cell into the transactivation of SREBP-1c. The tissue-specific overexpression of the N-terminal transcriptionally active domain of human SREBP-1c circumvents the complex regulation.
The liver-specific transgenic mouse model with SREBP-1c under control of the albumin promoter shows a mild fatty liver, hepatic insulin resistance with compensatory increased β-cell function, and massive obesity, but no signs of inflammation, metabolically healthy adipose tissue, and specific activation of hepatic DNL [11
The second mouse model with adipose tissue specific overexpression of the human transcription factor SREBP-1c included in the study displays a phenotype of fatty liver caused by dramatic increase of systemic lipid load due to the absence of adipose tissue. In these animals, the mechanism responsible for accumulation of lipids in the liver is certainly indirect, as the transgene is not expressed in liver tissue and systemic lipid overflow in the circulation is caused by absent adipose tissue [18
In sum, in our models the genetic phenotype displays a moderate accumulation of hepatic lipids with specific hepatic insulin resistance, while in the metabolic phenotype, aggravated fatty liver is present with systemic insulin resistance [11
In this study, we applied a transcriptome-wide differential gene expression analysis in liver tissue to identify regulator networks altered between moderate and aggravated fatty liver. Further, we independently validated our findings in primary hepatocytes, as the metabolically active unit of the organ from the different pathologies, and we further analyzed the functional impact on hepatic metabolism. Results derived from mouse studies were translated to men by measurement of the identified candidate, namely insulin-like growth factor binding protein (IGFBP) 2, in sera of obese patients with and without diabetes and biopsy-proven hepatic steatosis (NAFL) or steatohepatitis (NASH), and we further assessed whether intervention that mitigates hepatic steatosis reverses the observed relations.
This study identifies IGFBP2 as the most consistent effector network of differential gene expression in fatty liver, dependent on the degree of fatty liver. Physiologically this is accompanied by reduced fatty acid oxidation, increased methyltransferase and sirtuin activity, hypermethylation of the Igfbp2 promoter, and concomitant decreases in Igfbp2 mRNA and protein abundance. Furthermore, IGFBP2 secretion from primary mouse hepatocytes directly correlated with circulating IGFBP2 plasma levels, which show a reduction depending on the degree of fatty liver. Ex vivo, exposure to the saturated fatty acid palmitate lowers IGFBP2 protein secretion by primary hepatocytes even in metabolically healthy mice. We further show that IGFBP2 blunts the stimulation of de novo lipogenesis by IGF1 in hepatocytes from healthy mice or mouse models with fatty liver disease. Consequently, reductions in IGFBP2 levels by lipids may aggravate the development of fatty liver disease. Finally, we show that the observations obtained in the mouse models are clinically relevant, as the circulating levels of IGFBP2 are lower in patients with NAFL and NASH, and IGFBP2 levels are restored after weight loss following bariatric surgery along with reductions in hepatic fat content.
The role of IGFBP2 in health and disease is still not fully understood. A negative correlation of IGFBP2 with body composition, BMI [26
], metabolic syndrome [29
], type 2 diabetes mellitus [28
], or NAFLD [20
], and a positive correlation to insulin sensitivity independent to BMI [31
] were described. In contrast, IGFBP2 concentrations are high in patients with anorexia nervosa [32
Studies in mice support these findings. Overexpression of IGFBP2 reduced the predisposition to obesity and improved insulin resistance under normal and high-fat diet in transgenic mouse models such as the ob/ob mouse or diet-induced obesity models [26
], supporting a rather systemic impact of IGFBP2 in metabolic diseases including obesity and NAFLD. Here, we show that changes in hepatic IGFBP2 secretion were directly reflected in circulating IGFBP2 plasma levels and metabolic effects. In the fatty liver mouse models, an increase in the degree of hepatocyte lipid accumulation resulted in a decline of IGFBP2 secretion. In parallel, fatty acid oxidation is reduced, but hepatocellular SIRT activity is gradually elevated with increasing lipid content. SIRT1 activity depends on the energy level of a cell, and thereof on alterations in energy and lipid metabolism. Metabolic pathologies interfere with SIRT/NAD+
] and SIRT activity, as it is tightly regulated by the cellular availability of NAD+
]. Therefore, SIRT1 is not only a sensor but also a potential regulator of metabolism [36
]. SIRT1 has been shown to regulate DNA methyltransferase 1 [38
], whereas the methyltransferase activity itself is increased in a kind of feedback mechanism to interfere with regulation of rate-limiting enzymes of NAD+
metabolism on the transcriptional level [38
]. This may be important to maintain liver functionality and insulin sensitivity in response to altered metabolic pressure [39
]. In line with this hypothesis, our data show changes in energy metabolism accompanied with increased activity of SIRT1 and methyltransferase activity along with the gradual increase of hepatic lipid content of the mouse models.
The increased activity of methyltransferases further interferes with gene regulation due to alterations in DNA and histone methylation. Igfbp2 promoter methylation in whole blood cells was shown to correlate with an increased type 2 diabetes risk in patients without obesity [40
]. As IGFBP2 abundance was already low before the onset of type 2 diabetes in that study [40
], it may probably require further physiological alterations. This is supported by a study using a mouse model of diet-induced obesity, where hypermethylation of the Igfbp2 promoter and reduced Igfbp2
gene expression in early life correlated with the development of fatty liver and impaired glucose metabolism in adolescence [21
]. In our study, we demonstrated that hypermethylation of the specific murine Igfbp2 promoter region is not simply dependent on diabetes state, as the alb-SREBP-1c mice already display specific hepatic insulin resistance but no changes in IGFBP2 secretion. Our data indicate regulation of IGFBP2 to be more dependent on the lipid status of the liver, as shown in the aP2-SREBP-1c mice with systemic insulin resistance and aggravated fatty liver.
The liver metabolism of the alb-SREBP-1c mice, driven by the constitutive active transcription factor hSREBP-1c, permanently builds up triglycerides [12
], which could be exported to the adipose tissue for storage. In contrast, aP2-SREBP-1c mice display a lipodystrophic phenotype and do not have this deposit to compensate hepatic lipid overflow. Our ex vivo studies indicate that IGFBP2 secretion was significantly altered when hepatocytes were exposed to the saturated fatty acid palmitate as surrogate for saturated free fatty acids, whereas the unsaturated fatty acid oleate does not interfere with IGFBP2 secretion. Palmitate is known to induce ER stress in primary hepatocytes, which contributes to lipotoxicity [41
] and, at least in part, to a decrease in IGFBP2 secretion.
Our data indicate that a combination of impaired cellular pathways, regulation of gene expression by hypermethylation, and ER stress lead to altered secretion of IGFBP2. Regulated secretion of IGFBP2 might, therefore, indicate cellular dysfunction of hepatocytes rather than severe cellular damage as indicated by commonly used liver markers, for instance transaminases.
The mechanistic observations drawn from the experimental mouse models might also point towards a function of IGFBP2 as hepatokine. An autocrine loop might interfere with the activation and activity of metabolic sensors like SIRT and methyltransferases to modulate gene expression, including the direct regulation of IGFBP2 expression and consequently the IGF1 effect on metabolic pathways, including DNL.
The translational approach of our investigation supports this hypothesis. In humans, we found lower IGFBP2 serum concentration in obese men with NAFL or NASH compared to control men without obesity. In men with class III obesity, IGFBP2 levels correlated with the grade of hepatic steatosis and disease progression staged by the NAFLD activity score. In addition, recent data published by Wittenbecher et al. showed circulating IGFBP2 serum concentration to be associated with FLI, serum triglyceride, ALT, and γGT levels, as well as a high risk for the development of metabolic complications in patients without obesity [40
], further indicating a potential link to IGFBP2 serum concentration and liver function. Even in patients without obesity, a negative association between FLI and IGFBP2 serum concentration is accompanied by an increase of BMI and waist circumference [40
] indicating a gradual decrease of IGFBP2 with the degree of fatty liver and adipose tissue function. Furthermore, in the intervention study presented here, in men with obesity the significant reduction of BMI two years after surgery was accompanied with the reduction of fatty liver and serum IGFBP2 concentration. Taken together, the metabolic improvement of adipose tissue by invasive reduction also interfered with the metabolic health of the adipose tissue and improved liver function.
NAFLD often associates with impaired insulin sensitivity and type 2 diabetes, as in the studies presented here. Recently, Ahlqvist et al. found that the heterogeneity of type 2 diabetes can be clustered according to distinct clinical variables to take the individual nature of the disease into consideration [8
]. Among these, patients grouped to the severe insulin-resistant diabetes (SIRD) cluster showed low whole-body insulin resistance and the highest hepatic lipid content when newly diagnosed with type 2 diabetes, and in the 5-year follow up analysis they were shown to develop more severe NAFLD [8
]. In our study, IGFBP2 showed no changes in its serum concentration in terms of specific hepatic insulin resistance, but a significant reduction in serum was related to aggravate liver fat when whole-body insulin resistance is present. In respect to this, IGFBP2 might serve as an additional variable to better classify hepatic status in patients assigned to the SIRD cluster as well as a risk predictor for the development of more severe forms of NAFLD.
The study groups investigated here have some limitations. The data obtained in animal and human data sets are related to male subjects only. Mouse liver tissue in aP2-SREBP-1c showed portal and lobular inflammation without hepatocellular ballooning or fibrosis in histology-based assessment of hepatic pathological changes [11
], limiting direct comparison to human steatohepatitis pathology. Liver biopsies from the HepObster control group were not subjected to histological assessment of fatty liver disease state. The absence of liver disease in non-obese control patients was based on overall good health without medication and with normal liver function tests [23
]. Human serum analysis was derived from a relatively small number of patients and needs further investigation in larger-scaled studies. Liver biopsy material to verify hepatocyte physiological function, as performed in the animal studies, was not available for use in this study. Furthermore, obesity, type 2 diabetes, and NAFLD share common causative metabolic impairments that affect whole-body metabolism. This implicates caution to differentiate whether the observed association of IGFBP2 with the degree of liver fat is an observation restricted to fatty liver disease. According to study design, FLI was calculated for both human cohorts to strengthen comparability of the results for the degree of fatty liver.
In conclusion, an RNA screening approach for regulatory network genes between mouse models with different degrees of fatty liver identified IGFBP2, a molecule integrated in the IGF system, as the central mediator between moderate and aggravated fatty liver. These results were verified on the mRNA and protein levels, and secretion from primary hepatocytes depends on the degree of fatty liver in mice. Exposure to the saturated fatty acid palmitate is sufficient to lower IGFBP2 secretion in metabolically healthy mice. Mechanistically, IGFBP2 inhibits IGF1 activation on DNL, and the reduction of IGFBP2 may, therefore, aggravate the development of fatty liver disease. In a translational approach, circulating levels of IGFBP2 were lower in obese men with NAFL as well as in those with NASH and were restored after weight loss intervention along with reductions in hepatic fat content. Our results imply IGFBP2 as non-invasive biomarker for the degree of hepatic lipid accumulation when hepatic fat content exceeds a certain level towards disease progression. Further, this study provides IGFBP2 as a variable which might be added to improve reliability of the type 2 diabetes cluster with increased risk to develop severe NAFLD.