Development of the microfluidic model of hepatic steatosis
In the present work, we first developed an in vitro model of hepatic steatosis, under 3D dynamic culture conditions, into a microfluidic sinusoid-like device (Liver-on-a-Chip device in Fig. 1) based on the model previously published (Gori et al., 2016). Hepatic steatosis was induced in HepG2 cells providing a fat overload with mixtures of two FFAs (PA and OA), added to the medium alone or in combination at 1:2 and 2:1 molar ratios, and at 1mM final concentration for 48h. Depending on the different proportions of saturated and unsaturated fatty acids used, may represent hepatic cellular models of steatosis that mimic benign chronic steatosis or a more severe and acute condition of steatosis [6]. Then, through the use of confocal microscopy-based HCA we evaluated in real-time the following parameters of steatosis: intrahepatic FFA accumulation in the form of TGs (Fig. 2a, b) and their cytotoxicity (Fig. 3a, b) as well as global levels of ROS/RNS and Superoxide species (Fig. 4a-c) generated as a result of the different FFA overloads.
[Figure 1]
Analysis of intrahepatic triglyceride accumulation, cytotoxicity and oxidative stress in our microfluidic model of NAFLD
After the treatments with the different mixtures of FFAs for 48h (Fig. 2a, and b in the representative confocal micrographs), only an overload of 1mM OA produced a statistically significant intracellular accumulation of triglycerides [a four-fold increase vs. control (ctrl), p<0.05]. The intracellular storage of triglycerides observed in the other treatments, was proportional to the prevalence of OA compared to PA. The analysis of FFA-induced cytotoxicity (Fig. 3a, and b in the representative confocal micrographs) showed a remarkable decrease in cell viability with 1mM PA (approximately 30% reductionvs. control, p<0.0001), followed by the treatments with PA:OA 2:1 (p<0.01) and PA:OA 1:2 (p<0.01), implying a clear dependence on the amount of PA in the FFA mixture. On the contrary, the 1mM administration of OA alone did not cause any significant reduction in cell viability compared to the ctrl. Lastly, the evaluation of the oxidative stress derived from exogenous FFA overload showed a statistically significant increase of both total ROS/RNS levels only in the PA and PA:OA 2:1 treatments (Fig. 4a and c in the representative confocal micrographs), and Superoxide levels only in the PA group (Fig. 4b and c in the representative confocal micrographs) compared to ctrl. Conversely, no significant rise was detected in any of the other treatments, thereby underlining a direct effect of PA in the generation of oxidative stress.
[Figure 2]
Overall, it turned out that while OA is more steatogenic than PA (Fig. 2a, b), the latter is more cytotoxic than OA (Fig. 3a, b), which is consistent with the literature (Gómez-Lechón et al., 2007b; Ricchi et al., 2009). Finally, in agreement with previous works (Gori et al., 2016; Ricchi et al., 2009), the overload of FFAs produced moderate levels of total ROS/RNS and Superoxide species (Fig. 4a-c) in every treatment except those with PA and PA:OA 2:1, which showed higher increases most likely responsible for the cytotoxic effects observed in Fig. 3 and that may then lead to inflammation and apoptosis (Assaily et al., 2011; Gori et al., 2014) correlated to disease progression and severity (Angulo, 2002; Feldstein, Canbay, Angulo, et al., 2003; Feldstein, Canbay, Guicciardi, et al., 2003).
[Figure 3]
Quercetin and Hydroxytyrosol individually reduce intrahepatic triglyceride accumulation induced by the FFAs
Next, we investigated the role of dietary concentrations of the two polyphenols in FFA-induced hepatic steatosis: i) Quercetin, a flavonoid widely distributed in nature in many foods, especially in vegetables, fruits and tea, at 10 μM final concentration, which is in the average circulating human plasma concentrations, and in line with previous reports (Radtke et al., 2002; Vidyashankar et al., 2013); ii) Hydroxytyrosol, produced by the hydrolysis of oleuropein, which is a polyphenol peculiar to olives and olive oil (Hur et al., 2012), at physiological concentration of 10 μM (Hur et al., 2012). These two natural compounds have been selected for their antioxidant, anti-inflammatory and hypolipidemic properties (Hur et al., 2012; Marcolin et al., 2012; Park, Choi, Um, Yoon, & Park, 2011; Peres et al., 2000; Priore et al., 2014; Vidyashankar et al., 2013) and, in turn, for the possibility to restrain or prevent the development of NAFLD in our microfluidic model (i.e., reducing lipid accumulation, and the related lipotoxicity and oxidative stress). Interestingly, when 10 μM Quercetin (Fig. 2c, and 2d in the representative confocal micrographs) and 10 μM Hydroxytyrosol (Fig. 2e, and 2f in the representative confocal micrographs) were separately added for 48h to the cells, along with the different mixtures of FFAs, intrahepatic triglyceride accumulation was remarkably reduced compared to FFAs alone (Fig. 2a, b), with a statistically significant difference in the OA treatment (Fig. 2c, e, p<0.05 and p<0.01 for Que and HT, respectively) that is the most steatogenic. Hence, such hypolipidemic effect of Que and HT showed a very similar trend in all treatments and thus a relatively comparable effect in the context of hepatic steatosis, which was further confirmed by the log2-fold change (log2 F.C. in Fig. 1Sa and b) analysis of the AdipoRed experiment. Indeed, the present analysis showed that the more the prevalence of OA in the FFA mix and the stronger the lipid-lowering effect of both Que and HT with the latter that, regardless of the treatment, presents more statistically significant fold changes compared to Que.
Quercetin and Hydroxytyrosol protect from the lipotoxicity of FFAs
Furthermore, a corresponding significant decrease in cytotoxicity was also observed in almost all the treatments with Quercetin (Fig. 3c and representative confocal micrographs in 3d) and Hydroxytyrosol (Fig. 3e and representative confocal micrographs in 3f) compared to the FFAs alone. Notably, regarding the strong lipotoxicity of PA, its effect was partially reduced by both polyphenols compared to the treatments w/o polyphenols, except for the PA:OA 1:2 condition in which the cell viability increase, both after Que and HT administration, became statistically significant (as shown in Fig. 3c, p<0.01, for Que, and in Fig. 3e, p<0.01, for HT). Nevertheless, when the combined treatments of PA with Que (Fig. 3c) and HT (Fig. 3e) were compared to that with PA alone (white bars), we detected a slight increase in cell viability that, however, was not statistically significant in either case.
Quercetin and Hydroxytyrosol lower oxidative stress generated by the FFAs
Finally, our data showed that also total ROS/RNS and Superoxide levels were dramatically lowered with the addition of Quercetin (Fig. 4d, e and f in the representative confocal micrographs) and Hydroxytyrosol (Fig. 4g, h and i in the representative confocal micrographs) compared to the FFAs alone (Fig. 4a, b and c in the representative confocal micrographs), as also proved by the log2-fold change analysis for both polyphenols in the (log2 F.C. in Fig. 1Sc-f). In particular, the most powerful effect of both polyphenolic compounds was observed against the PA-induced ROS/RNS and Superoxide species in which the downregulation was statistically significant (p<0.01 in Fig. 4d and e for Que, and p<0.05 in Fig. 4g and h for HT). In addition, in the PA:OA 2:1 treatment, the ROS/RNS and Superoxide production was significantly lowered by both Que and HT (Fig. 4d and e with p<0.001 for Que; Fig. 4g and h with p<0.01 for HT, respectively), highlighting a slightly stronger effect, in terms of statistical significance, of Quercetin against oxidative stress compared to Hydroxytyrosol at the chosen concentrations.
[Figure 4]
Discussion
Drug discovery is currently hindered by the inability of conventional 2D cell culture models as well as animal experiments to accurately predict human responses. Liver-on-a-chip platforms may revolutionize this scenario by reproducing the natural 3D microenvironment of the cells and recapitulating some functionality of the hepatic tissue, allowing us to imitate liver pathophysiology more closely to its in vivocounterpart. Hence, we leveraged on such disease-on-a-chip technology to model the condition of hepatic steatosis, and to investigate at the cell level the protective effects of dietary concentrations of two natural polyphenols against some important features of the disease, in particular the intrahepatic fat accumulation and its related lipotoxicity and oxidative stress. Regarding Quercetin, its therapeutic potential and hepatoprotective effect has been thus far attributed to its antioxidant, anti-inflammatory and hypolipidemic activity (Marcolin et al., 2012; Peres et al., 2000; Vidyashankar et al., 2013). Instead, the beneficial effect in human health of extra virgin olive oil has been long ascribed to its high content of oleic acid (Carluccio, Massaro, Scoditti, & De Caterina, 2007; María-Isabel Covas, Konstantinidou, & Fitó, 2009; Esposito & Giugliano, 2010). However, more recently, also the important role played by the phenolic components (such as the oleuropein-derivative Hydroxytyrosol) has been increasingly emerging, not only for their known anti-oxidant and anti-inflammatory properties, but also for their lipid-lowering ability (Bendini et al., 2007; Carluccio et al., 2003; M.-I. Covas, 2008; Gordon, Paiva-Martins, & Almeida, 2001; Hur et al., 2012; Jemai, Fki, et al., 2008; Jemai, Bouaziz, Fki, El Feki, & Sayadi, 2008; Park et al., 2011; Pérez-Jiménez, Ruano, Perez-Martinez, Lopez-Segura, & Lopez-Miranda, 2007; Priore et al., 2014). Indeed, in this work we observed a stronger effect of HT, compared to Que, in mitigating the steatogenic effect of OA as well as the different mixtures of FFAs (see Fig. 2 and Fig. 1S). Actually, our in vitro results on the effects of Que and HT in the framework of NAFLD confirmed in vitro as well as in vivo data from the literature (Hur et al., 2012; LI et al., 2013; Pirozzi et al., 2016; Porras et al., 2017; Priore et al., 2014; Valenzuela et al., 2017; Vidyashankar et al., 2013) but in a more realistic scenario that is closer to the in vivo situation. In fact, our microfluidic model of NAFLD, providing us a more physiological setting than conventional static 2D culture systems, may represent a more suitable platform for simulating the chronicity of the disease and, as such, getting closer to the animal model, despite its intrinsic limitations that have yet to be completely overcome and improved.
These findings very importantly show a protective and comparable effect of Quercetin and Hydroxytyrosol against FFA-induced hepatic steatosis (confirming their lipid-lowering activity), lipotoxicity and oxidative stress, in which they are able to scavenge free radicals. Therefore, their role in counteracting excessive ROS/Superoxide generation and boosting the antioxidant defenses of hepatic cells, along with the reduction of excessive fat accumulation, seems to promote cell viability, and may represent an appealing strategy for the treatment of NAFLD. To date, no regulatory agency- approved cure for NAFLD has been found yet. As a matter of fact, the results reported herein together with the many beneficial pharmacological effects of Quercetin and Hydroxytyrosol on liver damage may promote their future clinical application as safe and effective therapeutic agents (Cao et al., 2014; Echeverría et al., 2018; Marcolin et al., 2012; Peres et al., 2000; Pirozzi et al., 2016; Tang et al., 2016; Valenzuela et al., 2017). Future experiments will also include: i) the combined administration of the two polyphenols to steatotic cells, in order to investigate a possible synergistic effect of the two compounds; ii) the use of a more complex liver microarchitecture that will involve different cell types, including other parenchymal (i.e., primary human hepatocytes or iPSC-derived hepatocytes) and non-parenchymal liver cells (e.g., endothelial cells, Kupffer cells and hepatic stellate cells) to enable also the analysis of the expression levels of inflammatory and fibrogenic cytokines (e.g., IL-6, IL-8, IL-1β, TNF-α, TGF-β1, CTGF), involved in the development of NAFLD, from the supernatants of such a multicellular hepatic microenvironment. In conclusion, our NAFLD-on-a-chip approach may also pave the way to the technological advancement of drug research, providing a promising tool to face the challenges of drug screening with the final goal, in the next future, to connect different tissues or even organs into a complex model system for the study of human development and disease.