*2.3. Biochemical Analysis and Real Time PCR*

Serum ALT, total cholesterol and glucose were measured on frozen sera using automated assays following the manufacturer's instructions (Reflotron Plus, Roche Diagnostic).

RNA extraction was performed from murine liver tissue (20 mg) of 23 samples: 8 SD, 8 HFD, 7 HFD + coffee (one sample was lost). Total RNA was extracted by using miRNeasy mini kit (Qiagen, Milan, Italy) according to the manufacturer's instructions [16,17]. The quantity and quality of total RNA were measured with *NanoDrop* (Thermo Fisher Scientific, Monza, Italy). Specific qPCR primers for NAFLD-associated lncRNAs and their mRNA targets were generated through Primer Blast [18] and are reported in Supplementary Table S2. Transcript expression was analyzed through real-time PCR assays by using *Power SYBR Green RNA-to-CT 1-Step kit* (Thermo Fisher Scientific) in *QuantStudio 5 Real-time PCR System* (Thermo Fisher Scientific). Gene expression fold changes (FC) were determined by applying the 2-ΔΔCt method and analyzing GAPDH as endogenous control [19,20].

Protein was extracted from 80 mg of liver tissue with RIPA lysis buffer and Western blot was made as previously reported [21]. All the immunoblot signals were detected using the Odyssey Fc System Infrared Scanner (LI-COR Biosciences, Lincoln, NE, USA) and densitometric analyses were performed by using Odyssey software Image Studio Lite Ver 5.2. We used the following antibodies: anti α-smooth muscle actin (Cell Signaling Technology, Danvers, MA, USA) and anti β-actin (Sigma Aldrich, St. Louis, MO, USA).

#### *2.4. Statistical Analysis*

Continuous variables are presented as mean ± SD or median IQR (interquartile range), based on data distribution assessed by D'Agostino and Pearson test. Statistical significance was evaluated applying the ordinary one-way ANOVA with Tukey's multiple comparisons test. Statistical significance was established at a two-tailed *p*-value < 0.05. GraphPad Prism 8 (GraphPad Software, Inc., San Diego, CA, USA) was employed for statistical analysis and graph-figure design.

#### **3. Results**

#### *3.1. Metabolic Parameters and Liver Histology*

At the beginning of the study, the three groups of mice had similar body weight (Table 1). At the end of the 12-week study period, mice of both HFD-fed groups, with or without coffee, had higher body weight compared to mice fed SD (Table 1).


All variables are presented as mean ± SD because of normal distribution assessed by D'Agostino and Pearson test. Statistical significance was assessed by ordinary one-way ANOVA with Tukey's multiple comparisons test. \* *p* < 0.5 vs. SD, † *p* < 0.5 vs. HFD.

> Mice fed HFD + coffee had lower body weight compared to HFD + vehicle despite similar food intake (Table 1). In agreement with body weight reduction, mice fed HFD + coffee displayed lower serum levels of total cholesterol and fasting glucose compared to mice fed HFD alone, whereas ALT were not significantly different among groups (Table 1).

Figure 1 shows representative pictures of liver hematoxylin-eosin staining in the three groups. All HFD animals showed some degree of steatosis, which was predominantly microvesicular in most cases (Supplementary Table S3). In coffee treated mice, macrovesicular steatosis disappeared and the degree of microvesicular steatosis was less severe, with most cases showing only Grade 1. Rare inflammatory foci were seen in four HFD mice and in none of the coffee treated animals.

**Figure 1.** Representative pictures of liver hematoxylin-eosin staining in the three groups. Normal liver histology in mice fed standard diet (Panel (**A**), original magnification 10×). Mice on HFD for 12 weeks showed severe mixed, micro- and macrovesicular steatosis (Panel (**B**), original magnification 10×). Two necro-inflammatory foci are visible in this field (Original magnification in the inserts 40×). Amelioration of liver histology in mice fed high fat diet + decaffeinated coffee for 12 weeks: absence of macrovesicular steatosis and inflammatory foci, reduction of microvescicular steatosis (Panel (**C**), original magnification 10×).

#### *3.2. Liver Expression of Long Non-Coding RNAs*

Based on literature data, we chose 14 specific lncRNAs involved in pathways related to NAFLD onset and progression including lipid metabolism, oxidative stress, inflammation, fibrosis, circadian rhythm regulation and apoptosis. For significantly (*p* < 0.05) deregulated lncRNAs with fold change values ≤ −2 or ≥2 in HFD + coffee versus HFD, qPCR analysis was extended also to their known validated direct or indirect targets.

#### *3.3. Coffee Inhibits De Novo Lipogenesis via lncRNA Gm16551/Srebf1 Pathway*

Figure 2 shows expression levels of Gm16551, a liver-specific lncRNA that regulates de novo lipogenesis through its interaction with the transcription factor Sterol regulatory element-binding protein isoform 1c (SREBP-1c) [22] (UniProtKB-Q9WTN3, encoded by the gene Srebf1). HFD caused a 2-fold downregulation of Gm16551 while administration of decaffeinated coffee solution determined a 3-fold upregulation of Gm16551 compared to HFD alone, restoring its expression to levels similar to SD condition (Figure 2, Panel A).

Surprisingly, mRNA for Srebf1 displayed an increasing trend of expression from SD towards HFD to HFD + coffee conditions (Figure 2, Panel B). This may depend on the fact that we evaluated the transcript for Srebf1 instead of measuring this factor at the translational level. However, although Srebf1 mRNA was upregulated by coffee, mRNA expression of its downstream targets acetyl coenzyme A carboxylase 1 (UniProtKB-Q5SWU9 encoded by Acaca) and stearoyl coenzyme A desaturase 1 (UniProtKB-P13516 encoded by Scd1) was downregulated. In detail, the administration of coffee in HFD mice induced a 3-fold down-regulation of mRNA for Acaca in comparison both to mice fed SD and HFD + vehicle (Figure 2, Panel C). mRNA for Scd1 had a similar expression trend, with a six-fold downregulation in HFD + coffee vs. HFD alone and respect to SD (Figure 2, Panel D).

**Figure 2.** Dot plots of the hepatic expression of *Gm16551* lncRNA (Panel **A**), *sterol regulatory element-binding protein factor 1* (Srebf1) mRNA (Panel **B**), *acetyl coenzyme A carboxylase alpha* (Acaca) mRNA (Panel **C**), *stearoyl coenzyme A desaturase 1* (Scd1) mRNA (Panel **D**), *cardiac mesoderm enhancer-associated* (CARMN) lncRNA (Panel **E**) and *steroid receptor RNA activator* (SRA) lncRNA (Panel **F**), analyzed through qPCR, in mice fed standard diet (SD), high fat diet (HFD) and HFD plus decaffeinated coffee; *n* = 23: 8 SD, 8 HFD, 7 HFD + coffee. Transcript statistical significance of DE transcripts was evaluated with one-way ANOVA with Tukey post-hoc test for multiple comparisons (two-tailed *p*-value < 0.05); FC = fold change.

Two other lncRNAs, also involved in lipid metabolism, were slightly modified by HFD and coffee intake, the lncRNA *cardiac mesoderm enhancer-associated* (CARMN) [23] and the *steroid receptor RNA activator* (SRA) [24]. HFD induced a slight increase in the expression of CARMN and SRA versus SD, while coffee supplementation significantly decreased their expression and restored them to levels of mice fed SD (Figure 2, Panel E,F).

#### *3.4. Coffee Inhibits Expression of the Fibrosis-Associated lncRNA H19*

Figure 3 shows the expression levels of *H19*, a lncRNA that is involved in liver fibrogenesis [25]. We found a 2.6 up-regulation of H19 in mice fed HFD compared to SD, whereas decaffeinated coffee reduced the expression of this lncRNA to levels lower than those observed in mice fed HFD alone and even SD (Figure 3, Panel A). We observed that mRNA for *Collagen alpha-1(I) chain* (UniProtKB-P11087 encoded by Col1a1) was downregulated in HFD + coffee in comparison to HFD and SD (Figure 3, Panel B). Although HFD is a model of early NAFLD without histological fibrosis, we also found an up-regulation of α-SMA protein expression evaluated by Western blot analysis (Figure 3, Panel C) suggesting the activation of hepatic stellate cells. In agreement with H19 down-regulation by coffee, the expression of α-SMA was restored by coffee intake to levels observed in mice fed SD (Figure 3, Panel D).

**Figure 3.** Hepatic expression of *H19* lncRNA (Panel **A**) and *Collagen alpha-1(I) chain* (Col1a1) mRNA (Panel **B**), analyzed through qPCR, in mice fed standard diet (SD), high fat diet (HFD) and HFD plus decaffeinated coffee. Transcript statistical significance was evaluated with one-way ANOVA with Tukey post-hoc test for multiple comparisons (two-tailed *p*-value < 0.05); FC = fold change. Liver expression of *alpha-smooth muscle actin* (α-SMA) protein, analyzed by Western blot (Panel **C**), and relative densitometry normalized for the housekeeping β-actin (Panel **D**). *n* = 23: 8 SD, 8 HFD, 7 HFD+ coffee.
