Increased let-7d-5p in non-alcoholic fatty liver promotes insulin resistance and is a potential blood biomarker for diagnosis

2.1 Human subjects

This is a cross-sectional study. Individuals included in the study were subjected to a laparoscopic cholecystectomy in the Santa Cristina University Hospital (Madrid, Spain) and a liver biopsy was obtained with previous informed consent. None of the individuals included in the study were under treatment with ursodeoxycholic acid. The inclusion criteria were age between 18 and 75 years and absence of chronic liver disease different from NAFLD. Exclusion criteria included significant alcohol consumption (>20 g/day), iron overload (transferrin saturation index >55%), presence of autoantibodies and/or antibodies against hepatitis B or C viruses or human immunodeficiency virus and potentially hepatotoxic drug consumption. Liver biopsies were analysed by haematoxylin–eosin staining (Section 2.4) by an expert hepatopathologist to classify them in normal liver histology (NL, n = 21) and liver steatosis (NAFL, n = 30). Demographic, anthropometrical and plasma biochemical parameters were collected. All protocols were performed according to the directions established by the Santa Cristina University Hospital and Complutense University of Madrid Ethical Committees, as well as those established in the Declaration of Helsinki of 1975.

2.2 Mouse model of NAFLD

All animal studies were performed with the approval of the University of Texas at Austin Institutional Animal Care and Use Committee (IACUC) and in accordance with the NIH guidelines ‘Guide for Care and Use of Laboratory Animals’. Male and female apolipoprotein E deficient (ApoE^−/−) mice (B6.129P2-Apoe^tm1Unc/J) (Jackson Laboratories, Inc., Bar Harbour, ME, USA) were fed the standard formulation Clinton/Cybulsky high-fat rodent diet with regular casein and 1.25% added cholesterol (D12108C; Research Diets, Inc., New Brunswick, NJ, USA). At 13 weeks of high-fat diet (HFD), the mice were sacrificed (n = 10 per sex), and the blood and liver were harvested (ApoE^−/− HFD). As controls, we used male and female wild-type (WT) mice (C57BL/6J, n = 5 per sex) fed a standard diet (STD) for 13 weeks (WT STD). All mice were maintained on a 12-h light/dark cycle at room temperature.

2.3 Oil red O staining of liver tissue

Oil red O staining of liver sections was performed as described previously.¹⁶ The sections were mounted with aqueous mounting medium for imaging (Vector Laboratories, Burlingame, CA, USA). Images were acquired using an inverted Eclipse TE300 microscope coupled to a Digital Sight DS-U2 camera (Nikon, Tokyo, Japan).

2.4 NAFLD activity score assessment in liver tissue

To determine NAFLD severity, haematoxylin–eosin staining was performed on paraffin-embedded liver sections (4-μm thick) that were evaluated by a single-blinded clinical pathologist and the NAFLD activity score (NAS) was determined for each liver sample as described previously.¹⁷ Minimal criteria for NASH included the combined presence of grade 1 steatosis, hepatocellular ballooning and lobular inflammation.¹⁷

2.5 Cell culture

Huh7 cells were maintained in high-glucose Dulbecco's modified Eagle's medium (Cytiva, Marlborough, MA, USA) supplemented with 10% fetal bovine serum (FBS) (Gibco, Waltham, MA, USA), 100 U/mL penicillin–streptomycin (Gibco) and MycoZap (Lonza, Basel, Switzerland).

2.6 In vitro model of steatosis

Fatty acids (FA) were dissolved at a concentration of 200 mM in absolute ethanol and diluted to a final concentration of 5 mM of FA in Krebs–Ringer HEPES buffer, pH 7.4, containing 1% bovine serum albumin (BSA, ITW Reagents, Chicago, IL, USA). Cells (2 × 10⁵) were seeded in complete medium and, once attached, growth medium was replaced by stimulation medium containing 0.5% FBS, 1% BSA for 4 h. Then, cells were stimulated with either 300 μM oleic acid (OA, Sigma-Aldrich, St. Louis, MO, USA) and 200 μM palmitic acid (PA, Sigma-Aldrich)¹⁸ or 100 μM OA and 400 μM PA for 18 h.

2.7 Transfection with let-7d-5p precursors and antagonists

Huh7 cells were seeded in complete medium (1 × 10⁵ cells/well). After 24 h, cells were transfected with a let-7d-5p precursor (AM17100, PM11778) or antagonist (AM17000, AM11778) (Ambion, Austin, TX, USA) using INTERFERin® (Polyplus Transfection, Strasbourg, France) as described previously.¹⁹ To evaluate let-7d-5p-induced insulin resistance, transfected cells were serum starved for 1 h and stimulated with 10 nM insulin (Sigma-Aldrich) for 5 min.

2.8 Extracellular vesicle isolation from plasma

EVs were isolated from the plasma of the patients according to the protocol described in the total exosome isolation kit (Invitrogen, Waltham, MA, USA).

2.9 Assessment of PNPLA3 rs738409 polymorphism

The analysis of the rs738409 polymorphism was performed in genomic DNA extracted from peripheral blood as described previously²⁰ and analysed by using the TaqMan Genotyping Master Mix (Applied Biosystems) and the corresponding TaqMan probe (Applied Biosystems) (Table S1).

2.10 RNA extraction

We isolated RNA from cultured cells and liver tissue with the miRVana miRNA isolation kit (Invitrogen) according to the manufacturer's protocol. RNA extraction from EVs and serum was carried out with the total exosome RNA and protein isolation kit (Invitrogen) according to the manufacturer's protocol.

2.11 MiRNA retrotranscription and quantitative PCR (RT-qPCR)

Reverse transcription (RT) was performed on 10 ng of miRNAs using the TaqMan™ advanced miRNA cDNA synthesis kit (Applied Biosystems, Waltham, MA, USA). MiRNA expression was determined by qPCR in a StepOnePlus™ Real-Time PCR System (Applied Biosystems) using the TaqMan Fast Advanced Master Mix (Applied Biosystems) and the corresponding TaqMan probe (Table S1). MiR-191-5p was used as an endogenous normalizer for liver and cell samples, and both miR-191-5p and miR-16-5p for plasma samples.^{21, 22} qPCR results were analysed as described previously.²³ For human liver samples, an exploratory screening of several miRNAs was performed to identify candidates with an altered expression in NAFLD. MiRNA candidates were selected after a search considering previous bibliography and gene expression datasets from GEO database (NCBI).

2.12 Total RNA RT-qPCR

Total RNA from liver biopsies was retrotranscribed using the high-capacity cDNA synthesis kit (Applied Biosystems) and analysed by qPCR using the PowerUp™ SYBR® Green Master Mix (Applied Biosystems) and the corresponding primers (Table S2). 36B4/RPLP0 was used as an endogenous normalizer. All experiments were carried out in a StepOnePlus™ Real-Time PCR System.

2.13 Protein extraction and western blotting

Protein extraction and Western Blotting were performed as described previously.¹⁹ Primary antibodies were diluted in 0.5% (v/v) Tween 20–TBS (TTBS) at the appropriate concentrations (Table S3). Primary antibodies were detected using horseradish peroxidase-coupled anti-rabbit IgG (NA931V, GE Healthcare, Chicago, IL, USA, 1:5000) or anti-mouse IgG (NA934, GE Healthcare, 1:5000). Protein expression was normalized using β-actin as a loading control.

2.14 Oil red O staining of cultured cells

Stimulated cells were fixed with 4% formalin for 1 h, washed with 60% isopropanol and stained for 10 min with a 0.21% oil red O solution. Images were acquired as described in Section 2.3. Oil red O staining quantification was performed using IP Win32 v4.5 software (Acromag, Wixom, MI, USA).

2.15 Statistical analyses

Normally distributed variables were expressed as mean ± SEM, and non-normally distributed variables were expressed as median (25th–75th percentile). Nominal variables were expressed as n (%). Statistically significant differences between two groups were assessed using two-tailed Student's t tests or two-tailed Mann–Whitney U tests depending on normality tests. Statistically significant differences between more than two groups were assessed using a one-way ANOVA followed by a Bonferroni post hoc test. qPCR results from patients were analysed using two-tailed Student's t tests to account for the influence of the two grouping variables (liver injury and body mass index). Correlation between variables was assessed by two-tailed Spearman's r correlation analyses. Receiver operating characteristic (ROC) analyses were performed to test the diagnostic accuracy of the evaluated miRNAs. All statistical procedures were performed using the GraphPad Prism v8.2.1 software (GraphPad Software, San Diego, CA, USA).

3.1 Characteristics of the study group

The most relevant parameters of the individuals included in the study are displayed in Table 1. The NL and NAFL groups did not significantly differ in mean age, and both were composed of a higher percentage of female subjects. Circulating insulin levels were significantly higher in the NAFL group, as well as the insulin resistance indicator homeostasis model assessment of insulin resistance (HOMA-IR). In the NAFL group, TG levels were significantly higher and HDL-cholesterol (HDL-Ch) was significantly lower, indicating a trend towards dyslipidaemia. Serum alanine aminotransferase (ALT) activity was significantly higher in the NAFL group, whereas serum aspartate aminotransferase (AST) did not significantly differ between both groups. Concerning some co-morbidities commonly associated with NAFLD, only three individuals in the NAFL group and one in the NL group were diabetic, whereas only one individual in each group had evidence of cardiovascular alterations.

Regarding histological characterization (Table 1), 20 patients in the NAFL group (66.7%) exhibited grade 1 steatosis, 7 (23.3%) had grade 2 and only 3 (10%) displayed grade 3 steatosis. The majority of the NAFL patients (76.7%) showed no signs of lobular inflammation, whereas the rest (23.3%) were classified as grade 1 inflammation. None of the patients displayed hepatocyte ballooning, and thus none of the patients could be diagnosed with NASH. Therefore, 15 (50.0%) patients were assigned a NAS of 1, 11 (36.7%) had a NAS of 2, and 4 (13.3%) had a NAS of 3. Furthermore, we performed a genotyping analysis for the NAFLD risk variant rs738409 of PNPLA3,²⁴ which was present in 2 (16.7%) individuals in the NL group in heterozygosis, and in 14 (70.0%) in the NAFL group [11 (55.0%) heterozygous, 3 (15.0%) homozygous].

3.2 Let-7d-5p expression is up-regulated in the liver of NAFL patients and correlates with clinical and anthropometric variables

We analysed the expression of eight miRNAs in liver biopsies and found that let-7d-5p and miR-34a-5p are significantly up-regulated in the liver of our group of individuals with steatosis, whereas miR-26b-5p is significantly down-regulated (Figure 1A). Since we observed a strong positive correlation between the BMI and let-7d-5p expression in our study group (Figure 1B), we stratified the NL and NAFL groups according to the World Health Organization BMI classification criteria and found a clear overexpression of let-7d-5p in overweight and obese individuals, regardless of liver histology (Figure 1C). Moreover, correlation analyses showed that let-7d-5p levels positively and strongly correlated with the waist and hip circumference as well as circulating TG levels, whereas HDL-Ch negatively and strongly correlated to the hepatic expression of this miRNA (Figure 1D,E). No significant correlations were found between let-7d-5p expression and age, sex or ethnicity. Therefore, our results suggest a relationship between let-7d-5p dysregulation and metabolic dysfunction.

3.3 Overexpression of let-7d-5p decreases IGF-IR/IR signalling mediators in the liver of NAFL patients

Correlation analyses also showed a strong and positive correlation between let-7d-5p expression and the HOMA-IR (Figure 2A). MiRNA–target Interaction (MTI) databases showed that let-7d-5p might regulate genes involved in insulin/insulin-like growth factor I (IGF-I) signalling. Thus, we analysed the expression of these targets in the liver of NAFLD patients by qPCR. IGF1, IGF1R and INSR B were significantly down-regulated in all groups when compared to healthy, lean individuals. On the other hand, AKT expression was significantly lower in obese and overweight individuals in the NL and NAFL groups (Figure 2B). These expression profiles are consistent with the let-7d-5p up-regulation observed in NAFLD patients.

Moreover, the expression of the four studied genes showed negative correlations with let-7d-5p expression (Figure 2C). The strongest and most significant correlation was observed in the case of IGF1, though AKT and IGF1R also showed significant, moderate and negative correlations. INSR B mRNA expression also showed a trend of negative correlation with let-7d-5p levels. Furthermore, AKT levels were assessed in the liver biopsies of these patients by western blot (Figure 2D). In lean individuals with steatosis, AKT levels were slightly increased without reaching statistical significance. On the other hand, AKT protein expression was significantly lower in the liver of overweight and obese individuals who had liver steatosis.

3.4 Increased let-7d-5p and decreased insulin signalling protein expression in the liver of a NAFLD mouse model

To confirm the results obtained in patients, we used ApoE^−/− mice fed a HFD as a validated murine model of NAFLD.^{19, 25-27} We did not observe differences in plasma TG between the two groups; however, as expected, ApoE^−/− HFD mice exhibited significantly higher levels of cholesterol than WT STD mice (WT STD: 19.12 ± 2.42 mg/dL vs. ApoE^−/− HFD: 210.40 ± 12.28 mg/dL, p < 0.0001). Liver histology revealed that ApoE^−/− HFD mice developed NAFLD regardless of the sex (Figure 3A,B), displaying significantly higher grades of steatosis, inflammation and hepatocyte ballooning and increased NAS compared with WT STD mice (Figure 3C).

Next, when compared to sex-matched controls, let-7d-5p expression significantly increased in ApoE^−/− HFD mice (Figure 3D), unrelated to vascular damage. Western blot analysis of the down-regulated let-7d-5p targets in patient samples showed that the expression of AKT and the insulin receptor (IR) significantly decreased in the liver of ApoE^−/− HFD mice in a sex-independent manner (Figure 3E). Moreover, this decrease negatively correlated with the increase in let-7d-5p expression, consistent with our previous findings in NAFL clinical samples. To evaluate the subsequent insulin resistance in ApoE^−/− HFD mice, we assessed the phosphorylation status of glycogen synthase kinase 3 (GSK3), which was significantly reduced in ApoE^−/− HFD mice (Figure 3F). Therefore, AKT activity and insulin signalling were impaired in ApoE^−/− HFD mice.

3.5 Overexpression of let-7d-5p in Huh7 cells results in insulin resistance

Since our data indicated that steatosis could induce let-7d-5p, we stimulated Huh7 cells with OA and PA and analysed its expression. We found that a combination of 100 μM OA and 400 μM PA induced a significant accumulation of lipid droplets (Figure 4A) as well as a significant increase in let-7d-5p expression (Figure 4B). These results demonstrate that in an in vitro model of lipid overload, the expression of let-7d-5p is also induced.

To determine whether the induction of let-7d-5p is involved in hepatic insulin resistance during NAFLD, we transfected Huh7 cells with a let-7d-5p precursor and an inhibitor. A remarkable increase in let-7d-5p expression was observed 96 h after transfection in cells transfected with the let-7d-5p precursor. Conversely, no decrease in let-7d-5p expression was observed in cells transfected with the let-7d-5p inhibitor (Figure 4C). Next, we analysed the expression of let-7d-5p targets involved in insulin signalling by western blot and found that all of the predicted targets were down-regulated in cells transfected with the let-7d-5p precursor compared to the control. As expected, we observed at least the same protein expression level in cells transfected with the miRNA inhibitor as the control. Moreover, IGF-I receptor (IGF-IR) and IR protein expression significantly increased in cells transfected with the let-7d-5p inhibitor when compared to untransfected cells (Figure 4D,E).

To investigate whether this let-7d-5p-induced down-regulation of insulin/IGF-I signalling pathway proteins could induce insulin resistance, we assessed the insulin-induced AKT phosphorylation. In non-transfected cells, we observed a statistically significant increase in AKT phosphorylation after stimulation with insulin compared to non-treated control cells. However, after let-7d-5p precursor transfection, we did not observe any increase in AKT phosphorylation following insulin treatment (Figure 4F,G), demonstrating that let-7d-5p overexpression is enough to induce insulin resistance in vitro.

3.6 Decreased circulating let-7d-5p levels as a potential biomarker in NAFLD

Finally, to evaluate the role of novel miRNAs as possible non-invasive diagnostic biomarkers for NAFLD, we isolated miRNAs from plasma EVs of individuals from the same study group. To validate that the isolated EVs from plasma could be exosomes, we determined the presence of the exosome markers CD63 antigen (CD63) and tumour susceptibility gene 101 protein (TSG101) and the absence of the Golgi vesicle marker Golgin subfamily A member 2 (GM130). We did not detect GM130 in our isolated EVs when compared to a positive control (Huh7 cell lysate), but observed CD63 and TSG101 expression in our samples (Figure 5A), suggesting that the isolated EVs could be enriched in exosomes.

Regarding circulating miRNA expression, the only statistically significant change that we observed was a reduction of let-7d-5p in the NAFL group (Figure 5B). To evaluate its putative role as a diagnostic biomarker in NAFLD, we performed ROC analyses (Figure 5C). For this miRNA, the area under the curve was 0.8571 (p = 0.0206). The optimal cut-off value for NAFL diagnosis was 0.6461 or lower, with a sensitivity of 87.50% and a specificity of 85.71%. Moreover, circulating let-7d-5p expression negatively correlated with the increase of the NAS (r = −0.6211, p = 0.0158).