*3.2. GCE Exerts Di*ff*erent E*ff*ects Compared to CAF on Hepatic TG Amount and Composition*

Blood cholesterol was unaffected by the diet or treatments, whereas blood TG levels were similarly increased in the HF-HFr, CAF, and GCE groups, compared to the control rats (Figure 2A,B).

**Figure 2.** Blood cholesterol (**A**), blood triglyceride (**B**), and liver triglyceride (**C**) levels. Results are the mean ± SD of values from 10–12 animals/group. (**D**) Bar plot representing the mean ± SD percentage of area of positive Oil Red O staining calculated as positive stained area per total area section in CT, HF-HFr, CAF, and GCE experimental groups (*n* = 10–12/group). Representative hematoxylin and eosin (**E**) and Oil Red O (**F**) stained liver sections from the four experimental groups. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.0001 vs. control. & *p* < 0.05 vs. CAF group.

The hepatic TG concentration was also increased in the HF-HFr group versus the CT group, and neither CAF or GCE attenuated this increase. Interestingly, hepatic TG levels were significantly lower in the GCE group than in the CAF group (Figure 2C). The same trend was observed in the liver sections stained with H&E and ORO, although the difference between the GCE and CAF groups was only marginally significant (*p* = 0.1) (Figure 2D–F).

We also aimed to determine the fatty acid profile of the hepatic TGs. As shown in Figure 3A, the amount of SFAs [palmitic acid (16:0) and stearic acid (18:0)] and MUFAs [palmitoleic acid (16:1 n-7) and oleic acid (18:1 n-9)] in the hepatic TG fraction was strikingly increased by the HF-HFr diet. The addition of CAF or GCE did not significantly affect the levels of these SFAs compared to the

HF-HFr group. Interestingly, the levels of both MUFAs were lower in the GCE group than in the CAF group, although the difference was significant only for palmitoleic acid. We also analyzed the levels of a less abundant MUFA in the TG fraction, 20:1 n-9, which showed also a significant increase in response to the HF-HFr diet and a decrease in the GCE group, compared to the CAF group (Figure 3A). Regarding PUFAs 20:4 n-6, 20:5 n-3, and 22:6 n-3, all showed lower levels in the GCE group than in the HF-HFr and the CT groups. The amount of linoleic acid (18:2 n-6) was not significantly altered by the diet or treatments (Figure 3B).

**Figure 3.** Fatty acid composition of hepatic triglycerides. (**A**) Saturated and monounsaturated fatty acids and (**B**) polyunsaturated fatty acids in the hepatic triglyceride fraction from CT, HF-HFr, CAF, and GCE experimental groups. Results are the mean ± SD of values from 9–10 animals/group. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001 vs. control. # *p* < 0.05; ## *p* < 0.01 vs. HF-HFr group. & *p* < 0.05; && *p* < 0.01 vs. CAF group.

### *3.3. Liver Lipidomic Signatures Induced by the HF-HFr Diet and E*ff*ects of CAF and GCE*

Analysis of hepatic DAGs showed a striking effect of the HF-HFr diet, which increased the amount of SFA-, and MUFA-containing DAGs (Figure 4A). The addition of GCE to the HF-HFr diet significantly attenuated the increase in DAG 18:0/18:0 whereas CAF supplementation had no effect on this species. By contrast, the HF-HFr diet did not significantly increase the levels of PUFA-containing DAG (Figure 4B) and caused a large reduction in DAG 18:2/18:2. GCE treatment reduced the amount of DAG 16:0/18:2 and DAG 18:0/20:4.

We also analyzed the effect of diet and treatments on the amount of hepatic Cer and HexCer (Figure 4C,D). The HF-HFr diet significantly reduced the levels of Cer 14:0 and Cer 16:0. By contrast, the amount of Cer 18:1 was increased by the HF-HFr diet, with CAF significantly attenuating this increase. Similarly, HexCer 18:0 and HexCer 20:0 levels were increased in the HF-HFr group, with GCE attenuating the increases. Moreover, GCE exerted specific effects on several species that were not modified by the HF-HFr diet, such as the reduction of Cer 20:0 and HexCer 16:0, 22:0, and 24:1 levels.

**Figure 4.** Lipidomic analysis in rat liver homogenates. Levels of diacylglycerols (DAG) (**A**,**B**), ceramides (Cer) (**C**), and hexosylceramides (HexCer) (**D**) in CT, HF-HFr, CAF, and GCE experimental groups. Results are the mean ± SD of values from 9–10 animals/group. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001 vs. control. # *p* < 0.05; ## *p* < 0.01 vs. HF-HFr group. & *p* < 0.05; && *p* < 0.01; &&& *p* < 0.001 vs. CAF group.

As shown in Table 2, the ratio of HexCer 16:0, 18:0, 20:0, and 24:0 to the corresponding Cer was very low in the CT group and was increased by the HF-HFr diet. Again, we observed a differential effect of GCE, as this group showed lower HexCer/Cer ratios for 16:0 and 18:0 than the HF-HFr group, whereas CAF did not cause this effect.

**Table 2.** Ratio hexosylceramide/ceramide.


Values are expressed as mean <sup>±</sup> SD (*<sup>n</sup>* <sup>=</sup> 9–10). \* *<sup>p</sup>* <sup>&</sup>lt; 0.05; \*\* *<sup>p</sup>* <sup>&</sup>lt; 0.01; \*\*\* *<sup>p</sup>* <sup>&</sup>lt; 0.001 vs. CT. # *<sup>p</sup>* <sup>&</sup>lt; 0.05 vs. HF-HFr; && *<sup>p</sup>* <sup>&</sup>lt; 0.01 vs. CAF.

#### *3.4. E*ff*ects of the Diet and Treatments on the Fatty Acid Biosynthetic Pathway*

We determined the hepatic expression of sterol regulatory element-binding protein-1c (SREBP-1c), a transcription factor that controls the expression of enzymes involved in fatty acid synthesis. Both the precursor (125 kD) and the active form of SREBP-1c (68 kD) remained unaltered in the hepatic protein

samples of all groups (Figure 5A). By contrast, the hepatic protein level of fatty acid synthase (FAS), a lipogenic enzyme controlled by this transcription factor, was increased significantly by the HF-HFr diet, with GCE partially preventing this increase (Figure 5B). The mRNA levels of another lipogenic enzyme controlled by SREBP-1c, stearoyl-CoA desaturase (*Scd1*), followed the same pattern of an increase in the HF-HFr group (Figure 5C). Interestingly, CAF increased *Scd1* expression even more than the HF-HFr diet, whereas GCE did not.

**Figure 5.** Western Blot of precursor (125 kD) and mature (68 kD) SREBP-1c (**A**) FAS: fatty acid synthase (**B**) lipin-1 (**D**) and phospho-mTOR proteins (**E**) in liver samples. Bar plots represent the mean ± SD band intensity of the proteins obtained from three samples per group, each one pooled from two animals. Bands are shown in the upper part of the figures. (**C**) Bar plot representing the mean ± SD mRNA levels corresponding to liver *Scd1* from CT (*n* = 5), HF-HFr (*n* = 6), CAF (*n* = 6), and GCE (*n* = 6) experimental groups. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001 vs. CT. # *p* < 0.05; ## *p* < 0.01 vs. HF-HFr. & *p* < 0.05 vs. CAF.

The observed effects of the diet on FAS and SCD1 expression suggested increased SREBP-1c transcriptional activity despite no changes in the amount of the active form of the protein. As shown in Figure 5D, the HF-HFr group showed a significant decrease in hepatic nuclear levels of lipin-1, which could modulate the transcriptional activity of SREBP-1c. Accordingly, the expression of phosphorylated mammalian target of rapamycin (P-mTOR), which phosphorylates lipin-1 and causes its nuclear exclusion, was increased in the livers of the rats from the HF-HFr group (Figure 5E). Interestingly, GCE relieved the reduction in lipin-1 levels caused by the diet, increasing the amount of this protein in nuclear extracts above CT levels (Figure 5D). Moreover, the GCE group returned P-mTOR levels to the control values, showing a significant reduction compared to the HF-HFr and CAF groups, which was in accordance with the increase in nuclear lipin-1 levels (Figure 5E).

#### *3.5. CAF or GCE Does Not A*ff*ect Lipid Catabolic Pathways*

To explore other mechanisms potentially involved in the observed effects on hepatic TGs, we determined the β-oxidation activity in liver samples. The results showed a significant decrease in response to the HF-HFr diet, with CAF or GCE addition having no effect on this decrease (Figure 6A). The mRNA levels of peroxisome proliferator-activated receptor α (*Ppar*α), and the PPARα target genes acyl-CoA oxidase (*Aco*) and very-low density lipoprotein receptor (*Vldlr*) were not modified by the diet or treatments (Figure 6B–D). However, the protein levels of VLDLR despite not being increased by the HF-HFr diet were significantly lower in the CAF and GCE groups, and GCE even lowered the amount of this protein compared to the CAF group (Figure 6E).

The autophagy of lipid droplets was described as another form of lipid catabolism. As shown in Figure 6F, the ratio of the microtubule-associated protein 1A/1B-light chain 3 (LC3) B-II/I was significantly reduced in the HF-HFr group, with CAF or GCE treatment not reversing this decrease. However, neither diet nor treatments reduced the levels of the autophagy substrate p62 (Figure 6G), while beclin-1 levels showed a small but significant increase in the CAF group (Figure 6H).

#### *3.6. Endoplasmic Reticulum Stress, Inflammation, and Oxidative Stress Markers*

We also explored other cell signaling pathways that could modulate hepatic lipid levels, such as endoplasmic reticulum (ER) stress. The HF-HFr diet significantly increased inositol-requiring enzyme-1α (IRE1α) phosphorylation, with neither CAF nor GCE reversing this increase (Figure 7A). However, levels of the active/spliced form of X-box-binding protein 1 (XBP-1s) protein in nuclear extracts were not significantly modified by any treatment, and mRNA levels of the XBP-1s target gene ER degradation-enhancing α-mannosidase-like 1 (*Edem1*) were not altered by HF-HFr diet and showed reduced expression in the CAF group (Figure 7B,C). Levels of the precursor (90 kD) and mature form (50 kD) of activating transcription factor 6 (ATF6) and phosphorylation of protein kinase RNA-like ER kinase (PERK) were not altered in any group (Figure 7D,E).

Finally, we assessed the expression of several inflammation and oxidative stress markers. The experimental diet used did not induce an inflammatory response in the liver. In fact, the mRNA expression of several inflammation-related genes was reduced, with the treatments showing negligible effect (Figure 8A). In line with these results, the plasma levels of the inflammation marker ALT were not increased by the diet or treatments (Table 1). Similarly, the HF-HFr diet did not induce hepatic oxidative stress, and even reduced glutathione peroxidase 1 (*Gpx1*) expression. The GCE group showed lower mRNA levels of superoxide dismutase 2 (*Sod2*) compared to the CAF group (Figure 8B).

**Figure 6.** (**A**) β-oxidation activity in liver samples. Bars represent the mean ± SD of 10–12 samples per group. Bar plots representing the mean ± SD mRNA levels corresponding to liver *Ppar*α (**B**), *Aco* (**C**), and *Vldlr* (**D**) genes from CT (*n* = 5), HF-HFr (*n* = 6), CAF (*n* = 6), and GCE (*n* = 6) experimental groups. Western Blot of VLDLR (**E**), LCII/I ratio (**F**), p62 (**G**), and beclin-1 (**H**) proteins, in liver samples obtained from CT, HF-HFr, CAF, and GCE experimental groups. Bar plots represent the mean ± SD band intensity of the proteins obtained from three samples per group, each one pooled from two animals. Bands are shown in the upper part of the figures. \* *p* < 0.05; \*\* *p*< 0.01 vs. CT. # *p* < 0.05 vs. HF-HFr group.

**Figure 7.** Western Blot of phosphorylated and total IRE1 (**A**), nuclear XBP1S (**B**), precursor (90 kD) and mature (50 kD) ATF6 (**D**), and phosphorylated and total PERK (**E**) in liver samples. Bar plots represent the mean ± SD band intensity of the proteins obtained from three samples per group, each one pooled from two animals. Bands are shown in the upper part of the figures. (**C**) mRNA levels of *Edem1* in the livers from CT (*n* = 5), HF-HFr (*n* = 6), CAF (*n* = 6), and GCE (*n* = 6) experimental groups. \* *p* <0.05; \*\*\* *p* < 0.001 vs. CT.

**Figure 8.** Bar plots showing the mean ± SD of specific mRNAs of pro-inflammatory molecules *iNos*, *Mcp1*, *Nlrp3*, *Tlr4*, *Tnf*α, *Il-6*, and *F4*/*80* (**A**) and oxidative stress genes *Gpx1*, *Sod2*, *Nrf2*, and *Cat* (**B**) in the livers from CT (*n* = 5), HF-HFr (*n* = 6), CAF (*n* = 6), and GCE (*n* = 6) experimental groups. \* *p* < 0.05; \*\* *p* < 0.01; \*\*\* *p* < 0.001 vs. control. & *p* < 0.05 vs. CAF group.

#### **4. Discussion**

Although nearly all rodent models on a high-fat diet rich in saturated fatty acids are characterized by obesity and insulin resistance [20], it is increasingly being recognized that a substantial proportion of individuals present NAFLD without obesity [2]. To obtain a model of NAFLD in its initial phase of simple hepatic steatosis, we fed female Sprague-Dawley rats a high-fat diet, which provides an exogenous source of fatty acids, and added liquid fructose (10% w/v) to their drinking water to promote de novo lipogenesis (DNL) [21]. To avoid the dietary intake of cholesterol, which is thought to activate Kupffer cells and stellate cells, and induce inflammation and fibrosis characteristic of NASH [22], we used cocoa butter instead of milk, as the source of saturated fatty acids in the high-fat diet.

Administration of the HF-HFr diet for three months caused hypertriglyceridemia and hepatic lipid deposition in the female Sprague-Dawley rats, but not inflammation, ER stress, or oxidative stress. Moreover, the rats fed the HF-HFr diet did not show an increase in body weight and adiposity, despite receiving around 1.8-times more calories than the control rats, which could not be ascribed to increased energy expenditure through spontaneous locomotor activity. Furthermore, although the rats on the HF-HFr diet responded to a glucose challenge with a higher insulin secretion, the increased insulin levels successfully controlled blood glucose levels and reduced the expression of hepatic gluconeogenic genes. This suggests that despite a decrease in the ISI, the hepatic glucose output was reduced, whereas in a typical situation of hepatic insulin resistance it would be increased [23].

The lipidomic analysis of liver samples from the rats offered some clues to explain these features of the HF-HFr diet. One of the most important bioactive lipids are ceramides, a class of sphingolipids involved in insulin resistance, inflammation, oxidative stress, and NAFLD development [24]. It was suggested that saturated fat derived from DNL or from the diet induces ceramide synthesis and insulin resistance [25]. However, we found that the hepatic levels of most ceramide species were not increased by the diet, which could be attributed to the absence of inflammation, as liver ceramides were reported to be increased in NASH, but not in simple steatosis in humans [26]. Remarkably, mice deficient in ceramide synthase 5 (CerS5), which exhibit lower hepatic levels of Cer 16:0, were protected from developing obesity and insulin resistance when fed a high-fat diet [27]. Therefore, the 50% reduction in hepatic Cer 16:0 levels in the HF-HFr group might help explain the absence of liver insulin resistance and obesity in rats fed this diet.

Having obtained a model of lean NAFLD with simple hepatic steatosis, we aimed to determine whether a moderate dose of caffeine, or a green coffee extract providing the same dose of caffeine, was effective in reducing the liver lipid burden. One of the major sources of caffeine in the human diet is coffee, which was reported to have beneficial effects on liver health [9]. However, a positive effect of coffee on NAFLD was not clearly established in human studies. Thus, a lower prevalence of NAFLD was associated with higher coffee intake in the NHANES study [28] and in some meta-analyses [10], but this association was not confirmed in other studies [29,30]. However, studies in animal models of diet-induced steatosis showed that several components of coffee, including caffeine, might be effective in reducing liver fat deposition [31]. The different outcomes from the human and animal studies might be due to the high doses of caffeine administered to laboratory animals, which in some studies were equivalent to 6 cups of coffee a day, much higher than the usual consumption in humans [32]. To provide a more realistic scenario, we treated rats with a low dose of caffeine (5 mg/kg/day), which after conversion based on body surface area was equivalent to 66 mg of caffeine for a 70-kg human being [15]. This amount roughly corresponded to 1 cup (20–25 mL) of espresso coffee per day, which was reported to contain 2.4 to 4.5 mg/mL of caffeine [33].

We observed that neither CAF nor GCE alleviated the hypertriglyceridemia and hepatic steatosis caused by the HF-HFr diet. However, rats treated with GCE exhibited lower levels of hepatic TG than those treated with CAF. When we analyzed the fatty acid composition of these TGs, we found that the amounts of palmitoleic acid and 20:1 *n*-9 were increased by the HF-HFr diet but were lower in the GCE group than in the CAF group. Palmitoleic acid is generated from palmitic acid through SCD1, which, together with FAS, are lipogenic enzymes regulated by SREBP-1c. The HF-HFr diet, despite not affecting the SREBP-1c levels, increased mTOR phosphorylation, which is known to phosphorylate and exclude lipin-1 from the nucleus [34]. This might lead to increased SREBP-1 transcriptional activity and, consequently, to FAS and SCD1 induction. Interestingly, the livers of the HF-HFr rats showed reduced nuclear levels of lipin-1 together with increased FAS and *Scd1* expression. These changes were not reversed neither by CAF nor by GCE. In fact, the expression of *Scd1* was higher in the CAF group than in the HF-HFr group, suggesting that CAF could further increase hepatic lipogenesis and worsen hepatic lipid deposition. However, neither the amount of hepatic TG nor lipin-1 or p-mTOR protein levels were different between the CAF and HF-HFr groups. In contrast, the GCE-treated rats showed lower mTOR phosphorylation and higher nuclear levels of lipin-1 than those of the rats from the HF-HFr group, suggesting lower SREBP-1 transcriptional activity. This might explain why FAS and SCD1 expression were induced to a lesser extent by GCE than CAF, and was in accordance with the lower levels of palmitoleic acid and total TGs observed in the livers of the GCE-treated rats.

The different effects of GCE compared to CAF were also observed with several DAG species, namely 18:0/18:0, 16:0/18:2, and 18:0/20:4, whose levels were reduced by GCE treatment compared to the HF-HFr group, but not by CAF treatment. Although there is a paucity of information about the effects of specific DAG species, it is generally assumed that DAGs play a role not only in insulin resistance but also in hepatic steatosis [35]. Therefore, the reduction of at least some of the DAGs accumulated in the liver might be regarded as a positive effect of other compounds contained in the GCE, given that caffeine alone did not cause such a reduction.

The hepatic levels of HexCer, which are formed from Cer by the enzyme glucosylceramide synthase (GCS), were reduced in the GCE group, as well as the 16\_0 and 18\_0 HexCer/Cer ratio. The HexCer/Cer ratio was considered to be an indicator of GCS activity. Interestingly, treatment of

ob/ob mice with an inhibitor of GCS was reported to reduce TG accumulation in the liver [36]. Thus, the lower HexCer/Cer ratios observed in the GCE group might also be associated with reduced GCS activity and lower levels of liver triglycerides in this group, suggesting beneficial effects of GCE on hepatic steatosis.

To gain more insight into the mechanisms involved in the regulation of hepatic TG accumulation, we also examined several pathways linked to fatty acid catabolism. The reduced hepatic activity of β-oxidation could contribute to increased liver TG accumulation in the HF-HFr group. However, none of the treatments reversed this decrease, suggesting that reduced catabolism of fatty acids also occurred in the CAF and GCE groups. By contrast, the hepatic protein levels of VLDLR, which were reported to increase in animal models and humans with hepatic steatosis [37], were significantly reduced by GCE, compared to the CAF group, although they were not significantly modified by the HF-HFr diet. Reduced VLDLR levels could contribute, at least partly, to the lower hepatic TG accumulation observed in the GCE group.

Due to its role in lipid droplet degradation, autophagy is another mechanism that can lead to liver fat removal [38]. Our group previously showed that liquid fructose supplementation in female rats inhibit liver autophagy, as shown by the lower LC3II/I ratio, which leads to increased liver TG levels [39]. In the current study, we also observed a reduced LC3II/I ratio and TG accumulation in the livers of rats receiving the HF-HFr diet, suggesting inhibition of hepatic autophagy. CAF-treated rats showed the lowest LC3II/I ratio and the highest TG levels in the liver, which despite a slight increase in the beclin-1 protein levels indicated that CAF did not activate autophagy in our model. Other studies suggest that CAF induced autophagy in the liver [11], but they used higher doses of CAF (30 mg/kg/day compared to 5 mg/kg/day in our study).

In conclusion, a moderate dose of caffeine, equivalent to 1 cup of coffee a day in humans, did not alleviate liver lipid deposition in a model of diet-induced hepatic steatosis, without obesity and inflammation. One limitation of our study was that we did not treat rats fed a control diet, so we cannot rule out that caffeine could have exerted some effects in rats not exposed to HF-HFr. However, our goal was to investigate whether caffeine could reverse the hepatic steatosis induced by the HF-HFr diet. The lack of effect of caffeine in our study could be attributed to the duration of treatment, to the fact that treatment was initiated two months after the introduction of the HF-HFr diet or to the low dose used. However, when the same dose of caffeine was administered through a coffee extract, despite not normalizing the hepatic TG levels, these were lower than when the caffeine was administered alone. The coffee extract was rich in other compounds such as polyphenols, which might be responsible for the different effects observed. Vitaglione et al. showed that decaffeinated coffee reduced lipid droplet accumulation in hepatocytes, in a model of NASH, suggesting that caffeine was not essential for the anti-steatotic effect of coffee [40]. However, few studies compared the effects of caffeine with other coffee compounds on hepatic steatosis. A study conducted in mice concluded that only treatment with chlorogenic acid significantly reduced hepatic TG levels, whereas administration of pure caffeine did not [41]. Similarly, female mice treated with catechines or with catechines combined with caffeine, reduced liver TG levels, whereas caffeine alone did not [42]. In mice fed an HFD, administration of chlorogenic acid or caffeine alone did not reduce hepatic TG, but a combination of both compounds was effective [43]. Although the molecular mechanisms involved are not clearly established, this study suggest a synergistic effect on several pathways controlling fatty acid metabolism, including SREBP1c and lipogenic enzymes, such as SCD1 and FAS. Along the same lines, our results suggest that GCE components, either independently or in combination with CAF, might lead to: (i) less lipogenesis due to lower mTOR phosphorylation and higher nuclear levels of lipin-1, affecting FAS and SCD1 expression; (ii) a reduced amount of several DAG species; (iii) a lower HexCer/Cer ratio, which is a marker of GCS activity; and (iv) reduced expression of hepatic VLDLR. Although these changes are subtle, their combination might contribute to the different effects of the extract when compared to caffeine alone.

**Supplementary Materials:** The following are available online at http://www.mdpi.com/2072-6643/12/11/3240/s1, Supplemental Methods: Measurement of fatty acid methyl esters in liver TGs, Lipidomic analysis in rat liver homogenates. Table S1: Composition of the diets used in the study. Table S2: Primers used for RT-PCR. Table S3: GC-MS parameters for detection of fatty acid methyl esters using single ion monitoring. Table S4: Retention times and transitions for detection of lipids using LC-MS/MS.

**Author Contributions:** A.M.V. and N.R. were in charge of all experiments and prepared the figures; R.B. and M.G. contributed to the PCR/Western blot experiments; A.S.-V. and I.L. performed FAME analysis; J.R.-M. Performed lipidomic analysis; R.M.S. helped in data interpretation and reviewed the manuscript; J.C.L. and M.A. designed the experiments, analyzed the data and wrote the manuscript. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research was funded by the Spanish Ministry of Economy, Industry, and Competitiveness (SAF2017-82369-R), Generalitat de Catalunya (2017 SGR 38), and European Commission FEDER funds. A.M.V. is a predoctoral fellow, BECAL grant program BCAL04-327, from the Government of Paraguay. A.S.-V. is recipient of the Instituto de Salud Carlos III Miguel Servet fellowship (grant CP II 17/00029).

**Conflicts of Interest:** The authors declare no conflict of interest.
