**2. Results**

#### *2.1. PLIN5 Was Up-Regulated in Liver Tissues of NAFLD Mice*

NAFLD is characterized by the accumulation of LDs and a raised ROS level. We induced NAFLD in mice by two classical methods, which were the methionine-choline-deficient diet (MCDD) treatment and high-fat diet (HFD) treatment, respectively. The liver tissues of mice fed with MCDD for 0 week, 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, and 8 weeks, and mice fed with HFD for 0 week and 10 weeks were collected. Then, the changes in hepatic *PLIN5* expression were investigated in these collected samples. The results showed that the mRNA level of *PLIN5* was up-regulated significantly in hepatic tissues of mice fed with MCDD for 4 weeks, 6 weeks, and 8 weeks, and mice fed with HFD for 10 weeks, compared to the corresponding control samples (fed with chow diet, CD; *p* < 0.05; Figure 1A–D). To validate this phenotype and further investigate the localization of *PLIN5*, immunohistochemistry was performed. The hepatic tissues of mice fed with MCDD for 0 week, 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, and 8 weeks were detected. The results showed that *PLIN5* protein was mainly localized surrounding the LDs, and additionally, in mice fed with MCDD, the expression of *PLIN5* was activated (Figure 1E). These results indicate that the expression levels of *PLIN5* were enhanced and *PLIN5* was mainly recruited on the surface of LDs during NAFLD development.

**Figure 1.** *Perilipin 5* (*PLIN5*) was up-regulated in liver tissues of non-alcoholic fatty liver disease (NAFLD) mice. (**A**) Six-week-old C57/bl male mice were fed with a high-fat diet (HFD) or chow diet (CD) for 10 weeks. The mRNA levels of *PLIN5* in liver tissues of mice fed with chow diet (CD) and HFD were detected by qRT-PCR. (**B**) The protein levels of *PLIN5* in liver tissues of mice fed with chow diet (CD) and HFD were detected by Western Blot. (**C**) Six-week-old C57/bl male mice were fed with methionine-choline-deficient diet (MCDD) for 1 week, 2 weeks, 3 weeks, 4 weeks, 6 weeks, and 8 weeks, respectively. The mRNA levels of *PLIN5* were detected by qRT-PCR. (**D**) The protein levels of *PLIN5* were detected by Western Blot. (**E**) Immunohistochemistry analysis of liver tissues of mice fed with MCDD (1–8 weeks) and control mice (0 week). Scale bar, 20 μm. These experiments were performed in triplicate. \* *p* < 0.05; and \*\* *p* < 0.01.

#### *2.2. Hydrogen Peroxide or LPS Treatment Enhanced Expression of PLIN5*

It was already well known that ROS levels were increased in hepatic tissues with NAFLD. Moreover, we did not observe any significant up-regulation of hepatic *PLIN5* in mice fed with MCDD for 1 week, 2 weeks, and 3 weeks, although many hepatic LDs had accumulated. Therefore, we assumed that the raised ROS levels activated *PLIN5* expression. ROS represents a variety of molecules and free radicals (chemical species with one unpaired electron) derived from molecular oxygen. Superoxide anion (O2−•), the product of a one-electron reduction of oxygen, is the precursor of most ROS and a mediator

in oxidative chain reactions [28], and then hydrogen peroxide arises from O2 −• [29,30]. Hydrogen peroxide and LPS are classical regents that induce cellular oxidation. Therefore, we investigated whether inducing cellular oxidative stress a ffected the expression level of *PLIN5*. The HepG2 cells were treated with 200 μM hydrogen peroxide for 12 h. qRT-PCR showed that *PLIN5* mRNA was up-regulated significantly by hydrogen peroxide treatment (*p* < 0.05; Figure 2A), and, additionally, the Western Blot indicated that *PLIN5* protein was also significantly up-regulated (*p* < 0.05; Figure 2B). Furthermore, we also investigated the changes in *PLIN5* expression levels with the lipopolysaccharide (LPS) treatment. The results showed that both mRNA and protein levels of *PLIN5* were up-regulated significantly (*p* < 0.05; Figure 2C,D). The oleic acid (OA) treatment can induce cells to form more and larger LDs, which is well applicable to observe the subcellular localization of *PLIN5*. Therefore, we subsequently detected the subcellular localization of *PLIN5* in OA treated cells by the PLIN5-EGFP expression vector. The results indicated that *PLIN5* was located on the surface of LDs (Figure 2E). To investigate whether hydrogen peroxide treatment changes the localization of *PLIN5*, we detected subcellular localization of *PLIN5* in cells with the 200 μM hydrogen peroxide treatment. We found that hydrogen peroxide treatment did not change the localization of *PLIN5* (Figure 2E). These results indicated that increased cellular ROS levels promoted the expression of *PLIN5*.

#### *2.3. PLIN5 Regulated Cellular ROS Levels*

To investigate whether *PLIN5* was involved in the anti-oxidant process, we validated the e fficiency of *PLIN5* overexpression and knockdown by the Western Blot method first. The result showed that *PLIN5* was overexpressed and interfered successfully (Figure 3A,B). Subsequently, we knocked down and overexpressed *PLIN5* to detect ROS levels by the DCFH-DA (2,7-dichlorodihydrofluorescein diacetate) method, respectively. The results showed that *PLIN5* knockdown increased ROS levels, whereas *PLIN5* overexpression decreased ROS levels significantly (*p* < 0.05; Figure 3C). Subsequently, we used 200 μM hydrogen peroxide to treat cells with *PLIN5* knockdown and overexpression in order to investigate whether *PLIN5* expression a ffected ROS levels of cells in oxidative stress. The results indicated that *PLIN5* knockdown increased ROS levels, whereas *PLIN5* overexpression decreased ROS levels significantly in cells treated with 200 μM hydrogen peroxide (*p* < 0.05; Figure 3D). To validate the phenotype, we further used the DHE (dihydroethidium) method to detect the O2 −• levels in *PLIN5* knockdown, overexpression, and corresponding control cells, respectively. The microplate reader indicated that *PLIN5* knockdown increased ROS levels, whereas *PLIN5* overexpression decreased ROS levels (Figure 3E,F). The release level of cytochrome c from mitochondria to cell plasma is the gold standard to reflect the level of cellular oxidative stress. Therefore, we investigated whether *PLIN5* knockdown aggravated or whether *PLIN5* overexpression reduced hydrogen peroxide-induced cytochrome c release. Firstly, we treated the cells with 200 μM hydrogen peroxide and then isolated the cytosolic and mitochondrial fractions respectively to detect the levels of cytochrome c (Figure 3G). The result showed that hydrogen peroxide treatment increased cytosolic cytochrome c levels and decreased mitochondrial cytochrome c levels, which indicated that hydrogen peroxide treatment increased cytochrome c releasing from mitochondria to cytoplasm. Then we detected the e ffect of *PLIN5* overexpression on the hydrogen peroxide treatment-induced cytochrome c release. The cells were transfected with *PLIN5* expression vector or pcDNA3.1 (control) vector, and then treated with 200 μM hydrogen peroxide. The cytoplasm and mitochondria were isolated, and then cytosolic and mitochondrial cytochrome c levels were detected through Western Blot, respectively. The result showed that *PLIN5* overexpression decreased cytosolic cytochrome c levels but increased mitochondrial cytochrome c levels compared to the control in the presence of hydrogen peroxide (Figure 3H). Mitochondrial membrane potential is an important indicator for mitochondrial oxidative damage. Therefore, we detected the mitochondrial membrane potential in cells with *PLIN5* overexpression or control by the JC-1 (5,5,6,6-tetrachloro-1,1,3,3-tetraethyl-imidacarbocyanine) method. As expected, the mitochondrial membrane potential was significantly higher in cells with *PLIN5* overexpression than in the corresponding control cells (Figure 3I). Furthermore, we detected the mitochondrial membrane

potential in cells with *PLIN5* overexpression or control in the presence of hydrogen peroxide. The result showed that mitochondrial membrane potential of *PLIN5* overexpression group was higher than the control group (Figure 3I).

**Figure 2.** Hydrogen peroxide or lipopolysaccharide (LPS) treatment enhanced the expression of *PLIN5*. Hydrogen peroxide and LPS were used to induce cellular oxidative stress. (**A**) mRNA level of *PLIN5* in cells with H2O2 treatment or control (phosphate buffer saline, PBS). (**B**) Western Blot analysis of *PLIN5* protein levels in cells with H2O2 treatment or control (PBS). (**C**) mRNA level of *PLIN5* in cells with LPS treatment or control (PBS). (**D**) Western Blot analysis of *PLIN5* protein levels in cells with LPS treatment or control (PBS). (**E**) H2O2 treatment did not change the localization of *PLIN5*. *PLIN5* localization analysis of cells with 400 μM oleic acid medium or H2O2 treatment. Green, PLIN5-EGFP; red, lipid droplets; blue, nucleus. Bar, 10 μm. These experiments were performed in triplicate. \* *p* < 0.05; and \*\* *p* < 0.01.

**Figure 3.** *PLIN5* regulated cellular reactive oxygen species (ROS) levels. (**A,B**) Western blot validation of *PLIN5* overexpression (**A**) and RNAi (**B**) efficiency. (**C**) HepG2 cells were transfected with *PLIN5* siRNAs or negative control siRNAs or *PLIN5* expression vector or pcDNA3.1 (control) vector, respectively. Then,

the cellular ROS levels were detected by DCFH-DA probes through a microplate reader (Ex = 488 nm, Em = 525 nm). ( **D**) The cells were transfected with *PLIN5* siRNAs or negative control siRNAs or *PLIN5* expression vector or pcDNA3.1 (control) vector, respectively, and then treated with 200 μM H2O2. Then, the cellular ROS levels were detected by DCFH-DA probes through a microplate reader (Ex = 488 nm, Em = 525 nm). (**E**) HepG2 cells were transfected with *PLIN5* siRNAs or negative control siRNAs or *PLIN5* expression vector or pcDNA3.1 (control) vector, respectively. Then, the cellular ROS levels were detected by DHE through a microplate reader (Ex = 535 nm, Em = 610 nm). (**F**) The cells were transfected with *PLIN5* siRNAs or negative control siRNAs or *PLIN5* expression vector or pcDNA3.1 (control) vector, respectively, and then treated with 200 μM H2O2. Then, the cellular ROS levels were detected by DCFH-DA probes through a microplate reader (Ex = 488 nm, Em = 525 nm). ( **G**) The cells were treated with 200 μM H2O2. The cytoplasm and mitochondria were isolated respectively, and then cytosolic and mitochondrial cytochrome c levels were detected respectively through Western Blot. ( **H**) The cells were transfected with *PLIN5* expression vector or pcDNA3.1 (control) vector, and then treated with 200 μM H2O2. The cytoplasm and mitochondria were isolated, and then cytosolic and mitochondrial cytochrome c levels were detected respectively through Western Blot. *GAPDH* was the reference protein of cytosolic component and the *Porin*/*VDAC1* was the reference protein of mitochondrial component. (**I**) The mitochondrial membrane potential (MMP) was detected by JC-1 probes using the flow cytometry method. These experiments were performed in triplicate. \* *p* < 0.05. Ex, excitation wavelength; Em, emission wavelength.

#### *2.4. PLIN5 Promoted LD Formation and Contact with Mitochondria*

We have shown that up-regulated *PLIN5* decreased cellular ROS levels, so we then investigated the regulatory mechanism. Our previous study has shown that the up-regulation of *PLIN2* promoted the formation of cellular LDs [12]. Therefore, *PLIN5* was overexpressed in HepG2 cells, and then the cellular LDs were labeled by BODIPY493/503 (4,4-difluoro-1,3,5,7,8-pentamethyl-4-bora-3a,4a-diaza-s-indacene) to reflect the e ffect of *PLIN5* up-regulation on LD content. As expected, the number of cellular LDs in cells with *PLIN5* overexpression was higher than the LDs in control cells (*p* < 0.05; Figure 4A,B). To validate the phenotype, the LD content was detected in cells transfected with siRNA oligos targeting *PLIN5* and control cells. The result indicated that *PLIN5* knockdown decreased the number of cellular LDs (*p* < 0.05; Figure 4C,D). A previous study showed that *PLIN5* promoted LD contact with mitochondria, and subsequently, we validated this phenotype in HepG2 cells. There was no doubt that the result indicated that the overexpression of *PLIN5* highly enhanced this contact in HepG2 cells (*p* < 0.05; Figure 4E,F). Moreover, we also investigated whether hydrogen peroxide treatment induced LD contact with mitochondria, since the high levels of ROS up-regulated *PLIN5*. The mitochondria were labeled by a mito-tracker, and LDs were marked by BODIPY. After the treatment with hydrogen peroxide, the contact events were increased significantly compared to corresponding cells (*p* < 0.05; Figure 4G,H). The results indicated that high ROS levels enhanced LD formation and promoted LD contact with mitochondria by up-regulating *PLIN5*.

**Figure 4.** *PLIN5* promoted LD formation and contact with mitochondria. (**A**) HepG2 cells were transfected with *PLIN5* expression vector or pcDNA3.1 (control) vector. Then, the cellular lipid droplets

were stained with BODIPY493/503 and observed by a confocal microscope. (**B**) The counts of cellular LDs in A. (**C**) HepG2 cells were transfected with *PLIN5* siRNAs or negative control siRNAs. Then, the cellular lipid droplets were stained with BODIPY493/503 and observed by a confocal microscope. (**D**) The counts of cellular LDs in C. (**E**) HepG2 cells were transfected with the mito-Dsred vector and *PLIN5* expression vector or pcDNA3.1 (control) vector. Then, the cellular lipid droplets were stained with BODIPY493/503 and observed by a confocal microscope. (**F**) HepG2 cells were transfected with mito-Dsred vector and *PLIN5* siRNAs or negative control siRNAs. Then, the cellular lipid droplets were stained with BODIPY493/503 and observed by a confocal microscope. (**G**) HepG2 cells were transfected with mito-Dsred vector and then treated with 200 μM H2O2. Then, the cellular lipid droplets were stained with BODIPY493/503 and observed by a confocal microscope. The fluorescence intensity along with the dotted line was performed to illustrate the contacts between LDs and mitochondria. (**H**) The ratio of contacts between LDs and mitochondria was analyzed. These experiments were performed in triplicate. \*\*\* *p* < 0.0001.

#### *2.5. PLIN5 Regulated the Expression Levels of Mitochondrial Function-Related Genes*

One of the functions of mitochondria is oxidative metabolism, which requires several mitochondrial respiratory chain oxidases, such as *COX* and *CS*. The expression levels of *COX* and *CS* are related to the mitochondria activity. Therefore, we further investigated whether PLIN5 expression affected the expression of *COX* and *CS*. The qPCR results showed that *COX2*, *COX4*, and *CS* were up-regulated in the cells with *PLIN5* overexpression (Figure 5A), whereas they were down-regulated significantly in cells with *PLIN5* knockdown (*p* < 0.05; Figure 5B). Subsequently, we investigated the effect of *PLIN5* expression on the expression levels of several cellular anti-oxidant genes including *GPX1, GPX2, SOD1, SOD2, TXNRD1, CAT,* and *PRDX3* through qPCR. The result showed that *GPX2* and *CAT* mRNA levels were increased significantly after *PLIN5* overexpression (*p* < 0.05), but *SOD2* mRNA level was decreased (Figure 5C). After *PLIN5* knockdown, the mRNA levels of *GPX1*, *SOD1*, and *TXNRD1* were decreased significantly (*p* < 0.05; Figure 5D).

#### *2.6. PLIN5 Reduced Apoptotic Rates of HepG2 Cells*

The mitochondrial activity is important for the cellular apoptosis process; therefore, we further investigated the effect of *PLIN5* on the regulation of cellular apoptosis. The rates of apoptosis of HepG2 cells were detected by the flow cytometry method. The results showed that *PLIN5* overexpression decreased the apoptotic rates significantly (*p* < 0.05; Figure 5E). Subsequently, a rescue experiment was carried out. *PLIN5* was overexpressed in the cells treated with hydrogen peroxide for 12 h, and then the apoptotic rates were detected. The result indicated that *PLIN5* overexpression rescued the enhancement of cellular apoptosis induced by hydrogen peroxide treatment (*p* < 0.05; Figure 5F).

**Figure 5.** *PLIN5* regulated the expression levels of mitochondrial function-related genes and apoptosis rate. (**A**) mRNA levels of *COX2, COX4,* and *CS* in cells transfected with *PLIN5* expression vector or control vector. (**B**) mRNA levels of *COX2, COX4,* and *CS* in cells transfected with *PLIN5* siRNAs or control siRNAs. (**C**) mRNA levels of *GPX1, GPX2, SOD1, SOD2, TXNRD1, CAT,* and *PRDX3* in cells transfected with *PLIN5* expression vector or control vector. (**D**) mRNA levels of *GPX1, GPX2, SOD1, SOD2, TXNRD1, CAT,* and *PRDX3* in cells transfected with *PLIN5* siRNAs or control siRNAs. (**E**) Apoptosis rate of cells transfected with *PLIN5* expression vector or control vector. (**F**) Cells transfected with *PLIN5* expression vector or control vector were treated with 200 μM H2O2 for 12 h. Then, the apoptosis rate was analyzed. These experiments were performed in triplicate. *GAPDH* was used as the reference gene. \* *p* < 0.05; and \*\* *p* < 0.01.

#### *2.7. The Expression of PLIN5 Was Regulated by the JNK-p38-ATF Pathway*

We analyzed the promoter region to investigate the transcriptional regulation mechanism of *PLIN5* expression. The GeneHancer dataset showed two potential promoter/enhancer regions (GH19J004539 and GH19J004534) whose distances from TSS (transcription start site) were −5.3 kb and +0.1 kb. Then, the transcription factor binding sites in these two regions were analyzed. We found that these two regions contained *JNK, ATF1,* and *ATF4* binding sites. It is well known that JNK-p38 is an important signaling involved in stress response, which is activated by oxidative stress, DNA damage, and UV, and subsequently regulates the downstream targets' expression, such as ATFs and STATs [31–34]. We have shown that *PLIN5* was up-regulated by hydrogen peroxide treatment; therefore, we assumed that *PLIN5* expression was regulated by the JNK-p38-ATF pathway. The phosphorylation levels of *p38* and *JNK* were detected. The Western Blot results indicated that *p-p38* and *p-JNK* levels were significantly increased in cells with hydrogen peroxide treatment (*p* < 0.05; Figure 6A,B). Moreover, the downstream targets of JNK-p38, *ATF1*, and *ATF4* were also up-regulated (Figure 6A,B). To further investigate whether ATFs regulate *PLIN5* expression, we overexpressed ATF1 and ATF4 and detected the *PLIN5* expression levels of both mRNA and protein. The qPCR and WB results showed that both mRNA and protein levels of *PLIN5* were increased significantly by either *ATF1* or *ATF4* overexpression (*p* < 0.05; Figure 6C–E). Subsequently, we cloned the promoter/enhancer regions of *PLIN5* (−2 kb) into pGL3-basic reporter vector to confirm the regulatory role of ATFs in *PLIN5* expression. The dual luciferase reporter gene assay showed that the fluorescence intensity of cells with *ATF1* or *ATF4* overexpression was much higher than that of the control cells (Figure 6F). The results indicated that both *ATF1* and *ATF4* did promote the transcriptional activity of *PLIN5*. To further validate the effect of JNK-p38 pathway on the expression of *PLIN5*, we utilized the p38-JNK pathway inhibitor, GS-4997 (Selonsertib). GS-4997 could inhibit the activity of *ASK1* so that to suppress the phosphorylation of downstream targets, *JNK* and *p38*. We found that hydrogen peroxide treatment activated the JNK-p38 pathway, but the GS-4997 treatment suppressed the JNK-p38 pathway (Figure 6G,H). Furthermore, we also found that hydrogen peroxide treatment increased the expression levels of *PLIN5*, and whereas GS-4997 treatment blocked the upregulation of *PLIN5* induced by hydrogen peroxide treatment (Figure 6G,H).

#### *2.8. Low Expression of PLIN5 Is Associated with Poor Prognosis*

We have shown that *PLIN5* expression was enhanced by oxidative stress and *PLIN5* could alleviate cellular ROS levels. We then analyzed the expression level of *PLIN5* in different kinds of tumors via the GEPIA (gene expression profiling interactive analysis) database (http://gepia.cancer-pku.cn/). Interestingly, many kinds of tumor samples showed lower *PLIN5* expression compared to normal samples (Figure 7A). Furthermore, among these kinds of tumor samples, liver hepatocellular carcinoma (LIHC), ovarian serous cystadenocarcinoma (OA), pancreatic adenocarcinoma (PAAD), and stomach adenocarcinoma (STAD) showed the largest differences (Figure 7B). Subsequently, survival analysis showed that the prognosis of LIHC was poor with a low expression level of *PLIN5* (Figure 7C). Then, survival analysis was performed to predicate the prognosis of 31 kinds of tumors with lower expressions of *PLIN5* (including ACC, BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LAML, LGG, LICH, LUAD, LUSC, OV, PPAD, READ, SARC, SKCM, STAD, TGCT, THCA, THYM, UCEC, and UCS; the extension of tumor abbreviations can be referred to in GEPIA). The result indicated that a low expression level of *PLIN5* was associated with poor prognosis (Figure 7D). These results indicated that low expression levels of *PLIN5* were bad for the prognosis of tumors.

**Figure 6.** The expression of *PLIN5* was regulated by the JNK-p38-ATF pathway. (**A**) Protein levels of *ATF1, ATF4, p-p38, p38, p-JNK,* and *JNK* were detected by Western Blot. (**B**) The gray value analysis of A. (**C**) mRNA levels of *ATF1, ATF3,* and *ATF4* in cells with 200 μM H2O2 treatment were detected by qPCR. (**D**) The cells were transfected with *ATF1* expression vector or pcDNA3.1 vector. The protein levels of *ATF1* and *PLIN5* were detected through Western Blot. (**E**) The cells were transfected with *ATF4* expression vector or pcDNA3.1 vector. The protein levels of *ATF4* and *PLIN5* were detected through Western Blot. (**F**) The effects of ATFs' expression on *PLIN5* transcriptional activity were detected by dual-luciferase reporter assay. (**G**) Protein levels of *p-ASK1, Ask1, p-p38, p38, p-JNK, JNK,* and *PLIN5* were detected by Western Blot. (**H**) The gray value analysis of G. *GAPDH* was used as the reference protein. These experiments were performed in triplicate. \* *p* < 0.05; \*\* *p* < 0.01; and n. s., not significant. 307

**Figure 7.** Low expression of *PLIN5* is associated with poor prognosis. Gene expression analysis and survival analysis were performed using the GEPIA database (http://gepia.cancer-pku.cn/). (**A**) Gene

expression analysis of *PLIN5* by cancer type. The lower pattern is the histogram illustration of the upper pattern. (**B**) The expression levels of *PLIN5* in samples with LIHC, OV, PAAD, STAD, and corresponding controls, respectively. ( **C**) Survival analysis of *PLIN5* in LIHCtumor samples. ( **D**) Survival analysis of PLIN5 in different kinds of tumor samples including ACC, BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LAML, LGG, LICH, LUAD, LUSC, OV, PPAD, READ, SARC, SKCM, STAD, TGCT, THCA, THYM, UCEC and UCS; E, CESC, DLBC, HNSC, LIHC, LUSC, PPAD and THYM; F, ACC, BLCA, BRCA, CESC, CHOL, COAD, DLBC, ESCA, GBM, HNSC, KICH, KIRC, KIRP, LAML, LGG, LICH, LUAD, LUSC, OV, PPAD, READ, SARC, SKCM, STAD, TGCT, THCA, THYM, UCEC, and UCS. The extension of tumor abbreviations can be referred to in the GEPIA database. \* *p* < 0.05.

#### **3. Discussion and Conclusions**

NAFLD has become one of the most common liver metabolic diseases worldwide. One obvious characterization of NALFD is an accumulation of LDs. A high level of hepatic fat accumulation increased cellular free fatty acids in hepatocytes, which was induced by lipolysis. However, the overaccumulation of free acids is dangerous because of toxic metabolites generated by fatty acid breakdown. Moreover, high lipid content showed higher levels of ROS [35]. Indeed, in liver tissues with NAFLD, a high level of oxidative damage was observed [4]. The enhanced oxidative stress resulted in changes in mitochondrial permeability transition, which was able to decrease the mitochondrial membrane potential and subsequently induce cell apoptosis. It is well known that decreasing the number of hepatocytes impairs the hepatic function and promotes NAFLD/NASH development [36]. Therefore, reducing cellular ROS levels contributed to the alleviation of oxidative damage, which was good for NAFLD/NASH treatment. However, the expression levels of SOD and other antioxidant enzymes, the main scavengers of cellular ROS, were decreased in NAFLD/NASH tissues [4]. Therefore, we considered that a compensation mechanism could exist to respond to this case.

*PLIN5* is a conserved LD protein that belongs to the PAT family [13]. *PLIN5* is expressed mainly in tissues with high oxidative metabolism such as liver, skeletal muscle, cardiac muscle, and brown adipose tissues. It is interesting that *PLIN5* was reported to be the key factor regulating LDs contacting mitochondria [37]. 443-463aa is the key region that promotes LDs' recruitment to mitochondria [22]. The deletion of 443-463aa of *PLIN5* deprived the ability of *PLIN5* promoting LDs contacting mitochondria [22]. Additionally, 443-463aa region of *PLIN5* is highly conserved between di fferent species [22]. A previous study has showed that the overexpression of *PLIN5* promoted cellular LD accumulation, whereas knockdown *PLIN5* enhanced fatty acid oxidation metabolism in liver cells [38]. Moreover, an SNP (single nucleotide polymorphism; rs327694326, NC\_010444.4:g.74314701T>C) in Italy big white, Italy Duroc, and Peter ran pigs, which induced high expression levels of *PLIN5*, promoted lipid accumulation and decreased the levels of *HSL*, an important lipolysis. Ilan et al. overexpressed *PLIN5* in mouse brown adipocytes and found that more mitochondria surrounded LDs and lipid synthesis was enhanced to promote the expansion of LDs [22]. In our study, we found that the overexpression of *PLIN5* increased the number of cellular LDs, whereas *PLIN5* knockdown decreased the number of cellular LDs (Figure 4C,D). Moreover, we also found that *PLIN5* overexpression promoted LD contact with mitochondria.

It is well known that organelle contacts usually induce the exchange of proteins in the outer membrane. LD is a highly dynamic organelle, which contacts other organelles frequently, such as endoplasmic reticulum (ER), mitochondria, peroxisome, and autolysosome. Many studies have reported that LD–ER contacts resulted in ER proteins, such as lipid synthetases (*DGAT2*, *GPAT4*) transferring to LDs [39–42]. LD contact with ER is important to LDs' expansion and cellular lipid homeostasis. A previous study showed an interesting phenotype involving LDs contacting mitochondria and clearing harmful proteins from the outer mitochondrial membrane [26]. A high level of cellular ROS induced the damage, and when accumulated damage exceeded a certain threshold, the cells would undergo the apoptosis process. During this process, some specific proteins were translocated to mitochondria such as pro- and anti-apoptotic proteins, for example, *BAX, BCL-XS, BIK, BAK, BCL-2,* *BCL-XL,* and *CED* [43,44]. Among these proteins, *BAX* played an important role in leading to a permeabilization of the outer mitochondrial membrane, which was able to subsequently induce the release of cytochrome c and apoptosis [45,46]. Interestingly, *BAX* and *BCL-XL* contained a protein domain consisting of two α-helices, which allowed them to localize to LDs. When LDs were in contact with mitochondria, *BAX* and *BCL-XL* were translocated to LDs from mitochondria [26]. Therefore, we considered that enhancing LD contact with mitochondria promoted the translocation process, which would subsequently modulate the stress response. In the present study, *PLIN5* overexpression enhanced the contacts between LDs and mitochondria, and the cellular ROS levels were significantly decreased (*p* < 0.05; Figures 3 and 4). Moreover, we also detected the influence of *PLIN5* expression on the expression of several anti-oxidant genes. The results showed that *PLIN5* did affect several anti-oxidant enzymes. We considered that there was a little effect, because not so many anti-oxidant enzymes such as some isoforms of *SOD, CAT,* and *GPX* can be influenced by the change of *PLIN5* expression. Therefore, the *PLIN5*-mediated LD contact with mitochondria could be an important mechanism for cells to respond to oxidative stress. For further study, the proteins of mitochondria and LDs in cells with *PLIN5* overexpression and control could be isolated, respectively, to investigate whether *PLIN5* overexpression promotes the proteins' translocation between these two organelles and to analyze the terms of proteins translocated through the mass spectrum method.

We investigated the regulatory pathway of *PLIN5* during the oxidative stress process. It is well known that the JNK-p38 signaling pathway plays an important role in the stress response. When the ROS levels (such as cellular hydrogen peroxide) were elevated, apoptosis signal-regulating kinase 1 (*ASK1*) was activated and subsequently sustained the activation of *JNK* and *p38 MAPK* signaling [47]. The activated *JNK* and *p38 MAPK* signaling would further activate ATFs' expression [31–34]. In our study, JNK-p38 MAPK signaling was activated by hydrogen peroxide treatment, and then *ATF1* and *ATF4* expression levels were significantly increased (*p* < 0.05; Figure 6A), which corresponds to the previous studies. Through bioinformatic analysis, we found that the promoter region of *PLIN5* contained the binding sites of ATFs. Therefore, we considered that the expression of ATFs could affect the expression levels of *PLIN5*. Overexpression of *ATF1* or *ATF4* indeed up-regulated *PLIN5* (Figure 6B,C). Furthermore, we also validated the regulatory role of ATFs on *PLIN5* expression experimentally, through dual luciferase reporter gene assay. The results confirmed that ATFs did indeed enhance the transcriptional activity of *PLIN5*. Therefore, we demonstrated that the ROS-JNK-p38-ATFs regulatory axis modulated the expression of *PLIN5* so that it regulated the cellular stress response process. Moreover, many studies have reported that *ASK1* signaling played an important role in NAFLD/NASH processes by promoting the inhibition of lipid and glucose metabolism [48–50] and by driving a strong inflammatory response [51]. Currently, *ASK1* has become a key therapeutic target for NAFLD/NASH. For example, Selonsertib (GS-4997) is a highly selective and potent *ASK1* inhibitor with potential anti-inflammatory, anti-tumor, and anti-fibrotic activities [52]. ASK1-JNK-p38 signaling was activated in NAFLD/NASH; therefore, *PLIN5* expression levels were supported to increase also. Our results showed that *PLIN5* expression was indeed up-regulated in liver tissues of mice fed with MCDD, which supported our hypothesis. We considered that *PLIN5* up-regulation could be a rescue mechanism during NAFLD/NASH processes. Increased *PLIN5* expression promoted LDs contacting mitochondria, enhanced the expression of mitochondrial functional genes and subsequently alleviated the cellular oxidative stress. We found that many kinds of tumors cells showed low expression levels of *PLIN5* (Figure 7A). Previous studies showed that NAFLD and NASH were well-known risk factors of hepatocellular carcinoma (HCC) [53,54], whereas and HCC was a lipid-rich tumor. Patients with obesity and NAFLD/NASH show an increased intake of dietary fatty acids (FAs). Meanwhile, insulin resistance enhances lipolysis of adipose tissue, which causes an increased exogenous FA supply and results in the development of a "lipid-rich" environment for hepatocytes. As we all know, more FAs would promote cells to generate more ROS through β-oxidation process. High level of cellular ROS often induced cellular stress and promoted cell apoptosis. However, we found that *PLIN5* could reduce cellular ROS levels and reduce cell apoptosis in the present study. Moreover, we also found

that expression of *PLIN5* could increase cellular lipid content. Previous studies reported that the lipid-rich environment is considered to promote the proliferation and metastasis of tumor cells [55–57]. Therefore, we considered that both functions of *PLIN5*, regulating cellular ROS levels and regulating cellular lipid content and lipolysis, could influence the tumor development process. Consequently, the down-regulation of *PLIN5* could be a predisposition for tumors' occurrence. *PLIN5* can be a good therapeutic target for NAFLD due to its ability to protect against oxidative stress and enhance mitochondrial function.

In the present study, we found that increase of cellular ROS induced by hydrogen peroxide or LPS treatment could up-regulate *PLIN5* expression. We then identified that ROS regulates the expression levels of *PLIN5* through JNK-p38-ATF signaling. Furthermore, we found that *PLIN5* could regulate the expression levels of mitochondrial cytochrome c oxidases (COXs) such as *COX2*, *COX4* and *CS*. Therefore, *PLIN5* could decrease cellular ROS levels through reducing the generation of ROS products by mitochondria, because up-regulation of COXs could reduce ROS products. Above is the novelty of this study. However, there are also several limitations in this study. The regulatory mechanism of *PLIN5* modulating the expression of COXs need further study. For example, studying the mechanism of protein exchange between LD and mitochondria during these two organelles contact. The LDs in cells with *PLIN5* overexpression could be isolated and the LD-related proteins on LD surface could be analyzed by mass spectrometry. Subsequently, whether *PLIN5* could promote protein exchange between LD and mitochondria can be investigated, by detecting the levels of mitochondrial-derived proteins on LD surface. Moreover, we noted that *PLIN5* could influence the expression levels of cellular anti-oxidative enzymes, such as *SOD1, SOD2, GPX1, GPX2,* and *CAT*. Although the e ffect of *PLIN5* on the expression of these enzymes was very mild, the mechanism is worth further study.

In conclusion, ROS-mediated activation of JNK-p38-ATF signaling up-regulated expression levels of *PLIN5*, and, then, increased *PLIN5* levels enhanced lipid synthesis and promoted LD contact with mitochondria, which helped cells to modulate stress response (Figure 8). Moreover, our study suggests that *PLIN5* could be a therapeutic target for NAFLD.

**Figure 8.** Diagrammatic sketch of this study. *PLIN5* was up-regulated by cellular stress induced by H2O2 or LPS treatment. Then, the increased *PLIN5* levels promoted cellular LD formation and expansion, expression levels of COXs and LDs contacting with mitochondria. Subsequently, LD formation and expansion reduced the levels of cellular fatty acids, which promoted the alleviation of stress. COXs' up-regulation reduced the release of cytochrome c from mitochondria to cytoplasm and reduced the mitochondrial damage. The contacts between LDs and mitochondria helped the transfer of potential harmful proteins from mitochondria to LDs. Therefore, cellular stress was alleviated.

#### **4. Materials and Methods**
