**1. Introduction**

There are a large number of patients suffering from non-alcoholic fatty liver disease (NAFLD) all over the world. This disease increases the risk of non-alcoholic hepatitis (NASH), chronic interstitial hepatitis, hepatic failure, and even hepatocellular carcinoma [1–3]. One of the characterizations of NAFLD is increased levels of reactive oxygen species (ROS) [4]. Some studies have shown that high levels of ROS promote the development of NAFLD/NASH and hepatocellular carcinoma by inducing ER stress [5] or regulating the AMPK signaling pathway [6]. There are several antioxidant enzymes such as superoxide dismutase (*SOD*), catalase (*CAT*), and glutathione peroxidase (*GPX*) responsible for scavenging cellular ROS. Meanwhile, the expression levels of *SOD* and other antioxidant enzymes are decreased in NAFLD/NASH [4]. Interestingly, lipid droplets (LDs) have been shown to be involved in the cellular stress response process. Bailey et al. demonstrated that lipid droplets can act as antioxidant organelles that protect *Drosophila* neural stem cells from hypoxia-triggered ROS [7], by allowing neuronal stem cells to keep proliferating under hypoxic conditions, and protection likely involves sequestering vulnerable membrane lipids away from ROS [8]. Furthermore, LDs also respond to starvation-induced stress by increasing their contact with mitochondria and lysosomes, which could

consist in the role of these contacts in transferring fatty acids from LDs to mitochondria or lysosomes for energy supply [9]. Moreover, the formation of nuclear LDs is related to the stress induced by phospholipid shortages [10,11]. Our previous study has shown that hydrogen peroxide promoted the formation of cellular LDs [12]. However, whether the increased cellular LDs play a role as anti-oxidants is largely unknown.

*Perilipin 5* (*PLIN5*) is one of the conserved LD proteins, which belongs to the PAT (*perilipin, adipophilin,* and *TIP47*) protein family [13]. Oxidative tissues such as skeletal muscle, liver, and brown fat have high expression levels of *PLIN5*, indicating that *PLIN5* plays an important role in lipid storage and LD function [14–16]. Previous studies identified that *PLIN5* regulated triglyceride contents in hepatocytes [17] and skeletal muscle [18]. Overexpression of *PLIN5* in skeletal muscle promotes oxidative gene expression and lipid content [19]. Recently, *PLIN5* was reported to be the key factor that regulated LD contacting mitochondria [20,21]. The N-terminal (1-188aa) of *PLIN5* is the conserved PAT domain, and 189-391aa is the domain contacting with patatin like phospholipase domain containing 2 (*PNPLA2, ATGL*). The C-terminal of *PLIN5* (443-463aa) is the key sequence related to mitochondrial recruiting [22]. LD–mitochondria contact is important for the energy supply during starvation stress, which promotes lipid β-oxidation [9,23], and the transfer process of fatty acids from LDs to mitochondria was also observed by probe imaging [24,25]. Recently, a study also found that LD–mitochondria contact contributed to lipid synthesis and LD expansion [22]. Furthermore, LDs are able to protect against cellular apoptosis by clearing harmful proteins from outer mitochondrial membranes [26]. Moreover, *PLIN5* has been shown to limit fatty acid toxicity [27]. These studies suggested that *PLIN5* was involved in the process of cellular anti-oxidation.

In the present study, we found that hydrogen peroxide- or lipopolysaccharide (LPS)-induced oxidative stress up-regulated both mRNA and protein levels of *PLIN5*. The overexpression of *PLIN5* increased the cellular LD content, promoted LD–mitochondria contact, reduced cellular ROS level, and enhanced mitochondrial function-related gene expression, whereas knockdown *PLIN5* indicated opposite phenotypes. Moreover, we identified that the promoter region of *PLIN5* contained the binding sites of *JUN, ATF1, ATF3,* and *ATF4*, and therefore *PLIN5* expression was activated by the JNK-p38-ATF pathway. By bioinformatic analysis, it has been found that *PLIN5* has a high expression in liver hepatocellular carcinoma (LIHC), and additionally, low expression of *PLIN5* is correlated with poor prognosis in LIHC. Therefore, *PLIN5* can be a potential therapeutic target in NAFLD and NAFLD-induced LIHC.
