2.1 Park7 is associated with liver regeneration after PHx

To investigate the possible involvement of Park7 in liver regeneration, we first examined the hepatic PARK7 levels in wild-type (WT) mice after PHx. Increased hepatic PARK7 expression was detected in mice 24–48 h after PHx using immunoblot analysis (Figure S1A,B). Next, we assessed human serum PARK7 levels in normal donors, who donated parts of healthy livers to their children receiving paediatric living donor liver transplants, at 24 and 72 h after left lateral lobectomy. Interestingly, the serum PARK7 levels significantly increased at both 24 and 72 h after PHx in healthy donors (Figure S1C). These results indicate that Park7 might act as a critical target in liver regeneration after PHx.

2.2 Delayed liver regeneration in Park7 knockout (Park7^−/−) mice

We performed 2/3 PHx in WT and Park7^−/− mice and evaluated their liver regeneration potential by examining the Ki67 (a nuclear antigen in proliferating cells) and PCNA (a nuclear protein necessary for DNA synthesis)/CCND1 (cyclin D1; a marker of G1/S) levels and the liver-to-body weight ratios. As shown in Figure 1A,B, immunohistochemistry staining showed that the hepatic Ki67 levels peaked at 36–48 h in WT mice but at 48–72 h in Park7^−/− mice after PHx. Consistently, western blot assays showed that compared with WT mice, Park7^−/− mice displayed delayed PCNA and CCND1 induction at 24 and 36 h after PHx (Figure 1C,D). Moreover, significantly decreased liver-to-body weight ratios were observed in Park7^−/− mice compared with WT mice at 24, 36 and 48 h after PHx (Figure 1E). Additionally, in agreement with the Ki67 results, lower and delayed cell mitotic counts were found at 36 h (Figure 1F). The induction of principal drivers of cell cycle progression was also delayed in Park7^−/− mice compared with WT mice after PHx (Figure S2A–I), which revealed deficient S/M phase progression. These findings indicate that Park7 deficiency significantly delays PHx-induced liver regeneration.

2.3 Enhanced TRAS in the regenerating liver of Park7^−/− mice after PHx

Park7 is a key regulator of inflammatory responses.¹ Given that inflammatory cytokines play an important role during the priming phase of liver regeneration,^{30, 31} we evaluated whether Park7 affects this period in liver regeneration. Both the mRNA and serum levels of IL-6 and TNFα were comparable in the priming phase between both genotypes (Figure S3A,B), suggesting that Park7 did not affect the priming phase in the process of liver regeneration. Correspondingly, there was no difference in the hepatic activation of STAT3 in the priming phase of liver regeneration between WT and Park7^−/− mice (Figure S3C,D). Interestingly, haematoxylin–eosin (HE) staining showed a greater number of fat droplets in the regenerating liver of Park7^−/− mice compared with WT mice at 36 and 48 h after PHx (Figure 2A). Meanwhile, the difference in hepatic lipid accumulation between WT and Park7^−/− mice was further explored by Oil Red O staining (Figure 2B,C). Additionally, ELISA assays showed that the transient hepatic accumulation of triglycerides (TG) and non-esterified fatty acids (NEFA) was elevated in the regenerating liver of Park7^−/− mice compared with WT mice (Figure 2D,E), along with liver cholesterol and cholesterol esters (Figure S4A,B). Interestingly, the serum NEFA level also increased in Park7^−/− mice (Figure S5), which suggests enhanced adipose tissue–associated lipolysis in Park7^−/− mice. Park7^−/− mice also displayed a downregulation of peroxisome β-oxidation genes, such as Acox1 and Crot, at 24–48 or 36 h (Figure S6A,B). Cd36, Fabp4, Fabp1 and Fatp4 expression was similar between WT and Park7^−/− mice, whereas Fabp2 expression was decreased in Park7^−/− mice during the regenerative phase, suggesting that Park7 has little impact on lipid import (Figure S6C–G). Additionally, comparable mRNA levels of Mttp and ApoB-100 showed similar VLDL assembly and excretion between the two groups (Figure S6H,I). These findings suggest that Park7 deficiency does not affect the priming phase of hepatic regeneration but results in enhanced TRAS in regenerating liver.

2.4 Hepatocyte-specific Park7 deficiency is sufficient to enhance TRAS and delay liver regeneration after PHx

To investigate whether hepatic Park7 deficiency is sufficient to delay liver regeneration, we generated hepatocyte-specific Park7 knockout (Park7^△hep) mice by crossing Park7^fl/fl mice with Alb-Cre⁺ mice. As shown in Figure S7A,B, hepatic Park7 deficiency was validated using a western blot assay. Compared with Park7^fl/fl controls, Park7^△hep mice showed severely impaired liver regeneration after PHx, as evidenced by decreased liver-to-body weight ratios (Figure 3A) and delayed Ki67 (Figure 3B,C) and PCNA/CCND1 induction (Figure 3D,E). Similar to the observations in Park7^−/− mice, HE staining, Oil Red O staining and ELISA assays of TG and NEFA showed enhanced TRAS in the regenerating liver of Park7^△hep mice compared with Park7^fl/fl controls (Figure 3F–J). These findings indicate that hepatic Park7 deficiency is sufficient to enhance TRAS and delay liver regeneration after PHx. Additionally, the blood glucose levels were elevated in Park7^△hep mice at 24-h post-PHx, but the levels were comparable at other time points (Figure S8A). Liver glycogen was also investigated using biochemical assays and periodic acid–Schiff staining. We found that hepatocyte-specific Park7 deficiency further impaired glycogen storage (Figure S8B,C), which might result from mitochondrial dysfunction in Park7^△hep mice after PHx.

2.5 Park7 deficiency affects the expression of key regulators of β-oxidation

It is generally acknowledged that TRAS is a result of the uptake of adipose-derived fat by the regenerating liver,³² which is a process unrelated to hepatic lipogenesis.³³ In-line with this notion, although the mRNA level of Acc1 increased at 24 h after PHx and that of Srebf1 decreased at 36 h, the mRNA levels of the main Srebf1-targeted lipogenesis genes, such as Fasn, Acc1 and Scd1, were comparable between Park7^fl/fl and Park7^△hep mice at 36 and 48-h post-PHx (Figure S9A). Similarly, despite an increasing level of ACC1 in Park7^△hep mice, we found no significant differences between the protein levels of FASN, ACLY and SCD1 at 36 h after PHx, implying the inconspicuous effect of hepatic lipogenesis (Figure S9B,C). TRAS in the regenerating liver could be affected by fatty acid β-oxidation. To investigate whether the enhanced TRAS in Park7^△hep mice resulted from impaired β-oxidation, we examined the level of Ppara, a key regulator of fatty acid β-oxidation in liver tissues.³⁴ Quantitative polymerase chain reaction (PCR) assays showed that hepatic Ppara was significantly downregulated in Park7^△hep mice compared with Park7^fl/fl controls at 36 and 48-h post-PHx (Figure 4A). Immunofluorescent (IF) staining and western blot assays of PPARa showed that either total PPARa levels or PPARa nuclear localization  was greatly reduced in Park7^△hep mice at 36 h after PHx compared with Park7^fl/fl controls (Figure 4B–D). As a result of decreased PPARa expression and nuclear localization, a decrease in the expression of Cpt1a, a downstream transcriptional target of PPARa and also a rate-limiting enzyme of β-oxidation, was found in Park7^△hep mice at 36 and 48-h post-PHx (Figure 4E–G). We then examined the PPARa and CPT1a expression before and after PHx in detail. The PPARa and CPT1a expression was relatively low before PHx compared with the expression at 36 h after PHx, and the PPARa and CPT1a expression was slightly increased before PHx in Park7^△hep mice (Figure S10A,B), suggesting that Park7 plays a different role before and after PHx. Considering that PARK7 disruption delayed but ultimately did not prevent PHx-induced hepatic regenerative recovery, we assumed that the drastic decrease in lipid accumulation after 48 h might stimulate liver regeneration. We therefore administered etomoxir, an inhibitor of CPT1a, every 12 h starting 36 h after PHx. The results showed that the liver-to-body weight ratios of Park7^△hep mice started to become comparable with Park7^fl/fl mice from 60 h after PHx in phosphate-buffered saline (PBS)-treated group, whereas the ratios of Park7^△hep mice failed to catch up with Park7^fl/fl mice in etomoxir-treated groups at 60–96 h after PHx. These results indicated that the FAO levels started to increase after 48 h to promote liver regeneration by supplying greater amounts of energy (Figure S11). To sum up, these findings suggest that hepatocyte-specific Park7 deficiency negatively affects fatty acid β-oxidation in the proliferation phase of liver regeneration,³⁵ which in turn results in enhanced TRAS and delayed liver regeneration.

2.6 PTEN deficiency rescues delayed liver regeneration in Park7^△hep mice

Park7 is a negative regulator of PTEN. Although several previous studies have demonstrated that Park7 exclusively inhibits the phosphatase activity of PTEN instead of suppressing PTEN expression,^{28, 36} we detected remarkable PTEN induction in the liver tissues with Park7 deficiency (Figure 5A,B), suggesting that Park7 can negatively regulate PTEN expression. Given that PTEN negatively regulates liver regeneration, we hypothesized that the delayed liver regeneration in Park7^△hep mice is caused by enhanced PTEN expression. To this end, we generated hepatocyte-specific Park7-Pten double knockout (Park7-Pten^△hep-DKO) mice and confirmed PARK7 and PTEN deficiency in liver tissues using western blot assays (Figure S12A,B). Compared with Park7^fl/fl controls, significantly decreased liver-to-body weight ratios (Figure 5C) and hepatic PCNA/CCND1 and Ki67 levels (Figure 5D–G) were only observed in Park7^△hep mice but not in Park7-Pten^△hep-DKO mice at indicated time points post-PHx, suggesting that PTEN may be attributed to the delayed liver regeneration in Park7^△hep mice. An increase in the liver-to-body weight ratio at 24-h post-PHx was observed in Pten^△hep mice compared with Park7^fl/fl controls (Figure 5C), which is consistent with a previous finding revealing that PTEN deficiency enhanced liver regeneration.⁶ Meanwhile, the delayed induction of pro-cell cycle factors in Park7^△hep mice was restored in Park7-Pten^△hep-DKO mice after PHx (Figure S12C–H). Additionally, we estimated the liver-to-body weight ratios before PHx. We found no statistically significant differences between these four groups at baseline (Figure S13A). Moreover, comparable liver-to-body weight ratios were observed in Park7^fl/fl, Park7^△hep and Park7-Pten^△hep-DKO mice, whereas Pten^△hep mice displayed an increased liver-to-body weight ratio at 168 h after the completion of the regenerative response to PHx (Figure S13B). Collectively, these data indicate that PTEN is required for delayed liver regeneration in Park7^△hep mice and that PTEN deficiency can restore impaired liver regeneration in Park7^△hep mice.

2.7 Functional inhibition of PTEN is sufficient to rescue delayed liver regeneration in Park7^△hep mice

As a phosphatase, PTEN executes its function via dephosphorylation of downstream targets. To further examine the involvement of PTEN in the delayed liver regeneration in Park7^△hep mice, we treated Park7^fl/fl and Park7^△hep mice with VO-Ohpic,² a PTEN inhibitor, followed by a PHx-induced liver regeneration model. The treatment regimen is graphically illustrated in Figure 6A. The administration of VO-Ohpic diminished the differences in the liver-to-body weight ratios (Figure 6B) and hepatic Ki67 (Figure 6C,D) and PCNA/CCND1 (Figure 6E,F) levels between Park7^fl/fl and Park7^△hep mice. These findings suggest that inhibiting the phosphatase activity of PTEN is sufficient to rescue the delayed liver regeneration in Park7^△hep mice after PHx.

2.8 PTEN deficiency or inhibition restores PPARa and CPT1a expression and hepatic lipid utilization in the regenerating liver of Park7^△hep mice

To investigate whether PTEN deficiency or inhibition restores the β-oxidation impairment in the regenerating liver of Park7^△hep mice, we first compared the changes in hepatic expression of PPARa and CPT1a between Park7^△hep and Park7-Pten^△hep-DKO mice after PHx. Compared with Park7^fl/fl controls, western blot assays showed a decreased hepatic expression of PPARa and CPT1a in Park7^△hep mice and restored the expression of these two genes in Park7-Pten^△hep-DKO mice at 36 h after PHx (Figure 7A,B). IF staining showed reduced PPARa nuclear localization in the regenerating liver of Park7^△hep mice and increased nuclear PPARa signals in the regenerating liver of Park7-Pten^△hep-DKO mice at 36-h post-PHx (Figure 7C). The mRNA levels of PPARa target genes, such as Acox1, Crot and Hmgcs2, and other β-oxidation-associated genes, such as Acadm and Slc25a20, were also examined and validated the changes in PPARa (Figure S14A–E). To further confirm whether Park7 affects the rate of β-oxidation, we examined the NAD⁺/NADH ratio levels. Supporting our conclusion, Park7 deletion decreased the NAD⁺/NADH ratio 24–36 h after PHx, which was restored in Park7-Pten^△hep-DKO mice (Figure 7D). Additionally, compared with Park7^fl/fl controls, Park7^△hep mice displayed decreased serum β-hydroxybutyrate levels (Figure S15A). We next measured the ketogenesis in vivo. Park7^△hep mice displayed severe impairment in their ability to produce β-hydroxybutyrate compared with Park7^fl/fl controls, which was also restored in Park7-Pten^△hep-DKO mice (Figure S15B). We then found that hepatic malonyl CoA levels increased in Park7^△hep mice and became comparable between Pten^△hep and Park7-Pten^△hep-DKO mice (Figure S15C). The previous findings suggest that PTEN deficiency restores the impaired β-oxidation in Park7^△hep mice. Consequently, the regenerating liver of Park7-Pten^△hep-DKO mice produced significantly high levels of ATP compared with that of Park7^△hep mice (Figure 7E). As a result of the enhanced β-oxidation, Oil Red O staining of hepatic lipids and ELISA assays of hepatic TG and NEFA levels showed reduced lipid accumulation in the regenerating liver of Park7-Pten^△hep-DKO mice at 36 h after PHx (Figure 7F–I). Consistent with the results of genomic Pten deficiency, the inhibition of the phosphatase activity of PTEN by VO-Ohpic significantly restored the expression of PPARa and CPT1a in Park7^△hep mice at 36-h post-PHx (Figure 7J,K). Meanwhile, the results of Oil Red O staining and assays of hepatic lipids showed that the difference in TRAS after PHx between Park7^fl/fl and Park7^△hep mice was diminished upon VO-Ohpic treatment (Figure 7L–O). Therefore, these findings suggest that PTEN deficiency or inhibition reverses the β-oxidation impairment in Park7^△hep mice and results in accelerated utilization of hepatic lipids derived from TRAS.

2.9 Hepatocyte growth factor (HGF)/epidermal growth factor (EGF)-activated NRF2 plays a key role in regulating Park7 expression after PHx

To further investigate the upstream regulatory factor of Park7, we first detected the mRNA levels of hepatocyte growth factor (HGF) and heparin-binding epidermal growth factor-like growth factor (HB-EGF) in the liver at indicated time points after PHx (Figure 8A). We found that the mRNA levels of HGF peaked at 24–48 h after PHx and those of HB-EGF peaked at 12 h and were maintained at a relatively high level at 12–48 h after PHx. To further confirm whether HGF/EGF upregulates Park7 expression, AML12 cells were treated with HGF or EGF (50 ng/ml) for 12 and 24 h. PARK7 was upregulated in AML12 cells after HGF and EGF treatment (Figure 8B,C), suggesting that HGF/EGF induces Park7 expression. Previously, it has been reported that HGF and EGF can induce the activation of NRF2, an important transcription factor associated with oxidative stress.³⁷ Next, we found that the NRF2 level in the nucleus was markedly elevated at 36 and 48-h post-PHx (Figure 8D,E). Using PROMO, a virtual laboratory for the identification of putative transcription factor binding sites in DNA sequences from a species or groups of species of interest, we found that the promoter region of Park7 carries a putative NRF2 binding site. Next, a ChIP–PCR assay was performed using NRF2 antibodies in HGF or EGF (50 ng/ml)-stimulated AML12 cells. The ChIP assay showed increased NRF2 recruitment to the promoter region of Park7 upon HGF or EGF treatment (Figure 8F,G). To further confirm this, we performed an Nrf2 knockdown experiment using Nrf2 siRNA. The results displayed that Nrf2 knockdown decreased both mRNA and protein levels of PARK7 in AML12 cells treated with HGF or EGF (Figure 8H–J). These results reveal that Park7 is a target gene of NRF2 in the regenerating liver.

2.10 oxPARK7 is responsible for PARK7-mediated regulation of liver regeneration and hepatic β-oxidation

We and other groups have shown that PARK7 is involved in the pathogenesis of several liver diseases based on its redox function. As reported previously,³⁸ the Cys106 of PARK7 is oxidized under oxidative stress and is critical for the biological function of PARK7. We detected the levels of liver malondialdehyde (MDA) and 2′,7′-dichlorofluorescin diacetate to analyse the oxidative status. The MDA levels were upregulated in Park7^fl/fl mice at 12–48 h after PHx, as reported previously.³⁹ Additionally, compared with Park7^fl/fl mice, the MDA and ROS levels were elevated in Park7^△hep mice at 36 and 48-h post-PHx (Figure S16A,B), which suggests that PARK7 also functions as an antioxidant response regulator in our study. We then examined the oxPARK7 expression in WT mice after PHx. Interestingly, elevated hepatic oxPARK7 expression was detected at 24–48 h after PHx (Figure S17A,B).

To investigate whether the role of PARK7 in PHx relies on its oxidized form, we established a hepatic overexpression of PARK7^WT or PARK7^C106A in Park7^△hep mice by the administration of AAV-PARK7^WT or AAV-PARK7^C106A by tail vein injection. After AAV administration, the mice were fed with a normal diet for another 6 weeks and were then subjected to 2/3 PHx and sacrificed after 36 h. By examining the PCNA/CCND1 and Ki67 levels at 36 h after PHx, we found that the hepatic administration of PARK7^WT, but not PARK7^C106A, could rescue the impaired liver regeneration in Park7^△hep mice (Figure S17C–F). Consistently, Oil Red O staining and TG measurement revealed that the hepatic expression of PARK7^WT, but not PARK7^C106A, could restore hepatic lipid utilization compared with Park7^△hep mice after PHx (Figure S17G–I). The hepatic expression of PARK7^WT, but not PARK7^C106A, restored the hepatic expression of PPARa and CPT1a at 36 h after PHx (Figure S17J,K). Next, we examined serum ketone and liver ATP levels. Hepatic PARK7^WT overexpression resulted in increased serum β-hydroxybutyrate and liver ATP levels compared with hepatic PARK7^C106A expression at 36-h post-PHx (Figure S17L,M). These results indicate that the PARK7 function in liver regeneration and hepatic β-oxidation was dependent on its antioxidant capacity.

4.1 Human samples

Serum samples of seven patients were collected before and after PHx at 24 and 72 h from the left lateral lobectomy of normal donors who donated parts of the healthy livers to their children receiving paediatric living donor liver transplant at the Department of Liver Surgery, Renji Hospital, School of Medicine, Shanghai Jiao Tong University. Blood samples were incubated for 2 h at 37°C, centrifuged at 3000 rpm for 15 min at 4°C, and the supernatant was then collected. Matched pairs of serum and plasma were collected under the approval of the Institutional Ethic Committee of Renji Hospital of Shanghai Jiao Tong University. Written informed consent was obtained from each patient based on the policies of the committee. All research was performed in accordance with the government policies and the Declaration of Helsinki.

4.2 Animals

Park7^−/− mice were obtained from the Jackson Laboratory (Bar Harbor, Maine, US), and C57BL/6 WT mice were purchased from SLAC Laboratory (Shanghai, China). The experimental Park7^−/− and WT littermates were generated by crossing Park7-heterozygous mice. The floxed Park7 (Park7^fl/fl) mice (Shanghai Biomodel Organism, Shanghai, China) and the mice expressing Cre recombinase in the control of the mouse albumin promoter (Alb-Cre; Shanghai Biomodel Organism, Shanghai, China) were applied to generate hepatocyte-specific Park7 knockout (Park7^△hep) mice. We used two steps to generate Park7^△hep mice. First, a homozygous loxP-flanked Park7 mouse was mated with a homozygous Alb-Cre mouse to generate the F1 mice that were heterozygous for a loxP-flanked Park7 allele and heterozygous for the Alb-Cre allele. Second, we backcrossed these F1 mice to the homozygous loxP-flanked Park7 mice to generate the Park7^△hep, which were heterozygous for the Alb-Cre allele and homozygous for the loxP-flanked Park7 allele. The homologous recombination of Park7^fl/fl and Pten^fl/fl mice is shown in Figure S18. The mice had a C57BL/6J background. The following primers were used for genotyping: CTCCTCTACTCCATTCTTCC and ACTCCCACCAATGAACAAAC (Pten), AGAACTCCACCTGCCTCTG and TGCCTCTGTAACAACCATCTG (Park7), CAGCATTGCTGTCACTTGGTC and ATTTGC CTGCATTACCGGTCG (Alb-Cre).

4.3 Animal treatment and surgery

Male mice, 8–10 weeks of age, were subjected to PHx as previously described.²⁵ Briefly, we first removed the liver from the ligaments after making a midline incision. Second, the pedicle of the left lobe was ligated (silk, 4/0) and resected. Finally, the middle lobe was ligated (silk 4/0) and resected. VO-Ohpic (10 μg/kg, MedChemExpress, HY-13074), the inhibitor of Pten, was administrated (.1 ml/10 g) to the mice by intra-peritoneal as shown in Figure 6A for Pten inhibiting experiments. Dimethyl sulfoxide was used as the vehicle. Etomoxir (10 mg/kg, MedChemExpress, HY-50202), the inhibitor of CPT1a, was administrated (.05 ml/10 g) to the mice by tail vein. PBS was used as the vehicle. AAV-PARK7^WT, AAV-PARK7^C106A or AAV-control were injected via the tail vein with 150 μl of PBS containing 1 × 10¹² VG. Mice were fed normally for another 6 weeks and subjected to 2/3 PHx. At each time point after PHx, blood was collected by heart puncture in anaesthetized mice before cervical dislocation. Liver samples were then harvested immediately for an examination. In all experiments, Park7^fl/fl littermates were used as the control. This study was performed in strict adherence to the recommendations in the Guide for the Care and Use of Laboratory Animals published by the National Institutes of Health. All animal studies were approved by the Institutional Animal Care and Use Committees of Renji Hospital and Shanghai Jiao Tong University. All authors have reviewed and approved the final manuscript and have access to the study data.

4.4 Statistical analysis

Data are presented as the mean ± SEM and analysed by permutation t-test and Pearson correlation. Two-sided p values of less than .05 were considered statistically significant. Multiple group comparisons were performed using one-way ANOVA followed by Bonferroni's post hoc test. We also performed Welch's ANOVA for multiple group comparisons when groups had unequal variances.