Abstract

   Build‐up of free cholesterol (FC) substantially contributes to the
   development and severity of non‐alcoholic fatty liver disease (NAFLD).
   Here, we investigate the specific mechanism by which FC induces liver
   injury in NAFLD and propose a novel therapeutic approach using
   dihydrotanshinone I (DhT). Rather than cholesterol ester (CE), we
   observed elevated levels of total cholesterol, FC, and alanine
   transaminase (ALT) in NAFLD patients and high‐cholesterol diet‐induced
   NAFLD mice compared to those in healthy controls. The FC level
   demonstrated a positive correlation with the ALT level in both patients
   and mice. Mechanistic studies revealed that FC elevated reactive oxygen
   species level, impaired the function of lysosomes, and disrupted
   lipophagy process, consequently inducing cell apoptosis. We then found
   that DhT protected mice on an HCD diet, independent of gut microbiota.
   DhT functioned as a potent ligand for peroxisome proliferator‐activated
   receptor α (PPARα), stimulating its transcriptional function and
   enhancing catalase expression to lower reactive oxygen species (ROS)
   level. Notably, the protective effect of DhT was nullified in mice with
   hepatic PPARα knockdown. Thus, these findings are the first to report
   the detrimental role of FC in NAFLD, which could lead to the
   development of new treatment strategies for NAFLD by leveraging the
   therapeutic potential of DhT and PPARα pathway.

   Keywords: NAFLD, Free cholesterol, ROS, Lysosome, Dihydrotanshinone I,
   PPARα
     __________________________________________________________________

   This study underscores the detrimental role of free cholesterol in
   non‐alcoholic fatty liver disease (NAFLD), demonstrating its ability to
   trigger a surge in reactive oxygen species (ROS) that leads to
   hepatocyte death. We also introduce dihydrotanshinone I (DhT), a small
   molecule that mitigates free cholesterol‐induced toxicity by activating
   peroxisome proliferator‐activated receptor α, thereby reducing ROS
   levels. This finding points to a promising new treatment strategy for
   NAFLD.

   graphic file with name ADVS-12-2406191-g006.jpg

1. Introduction

   Non‐alcoholic fatty liver disease (NAFLD) encompasses a wide range of
   conditions resulting from the accumulation of fat in the liver,
   including simple non‐alcoholic fatty liver (NAFL) and non‐alcoholic
   steatohepatitis (NASH), which may progress to fibrosis, cirrhosis, or
   even hepatocellular carcinoma.^[ [36]^1 ^] Currently, NAFLD is one of
   the major causes of morbidity and a healthcare burden worldwide, with a
   global prevalence higher than 30%.^[ [37]^2 ^] The current treatment
   mainly lies in lifestyle modifications, physical activity and
   pharmacological interventions; however, the benefits resulted from
   those treatment options are far from satisfaction due to low compliance
   and poor real‐world efficacies.^[ [38]^3 ^] The research and
   development of novel medication against NAFLD is slow‐moving, with the
   first Food and Drug Administration‐approved drug (Rezdiffra) only
   announced in March 2024, which is used to treat adults with
   noncirrhotic NASH with moderate to advanced fibrosis and to be used
   along with diet and exercise.^[ [39]^4 ^] Therefore, much more effort
   should be taken to develop novel therapies.

   NAFLD has traditionally been characterized by increased hepatic
   triglyceride accumulation.^[ [40]^5 ^] Recent studies have however
   revealed disturbance of cholesterol hemostasis as another crucial
   factor in the pathogenesis of NAFLD.^[ [41]^6 ^] Lipidomic analysis of
   NAFLD patients has demonstrated an increased accumulation of free
   cholesterol (FC), without a similar increase in cholesterol esters
   (CE), compared to healthy controls.^[ [42]^7 ^] FC has been well
   documented to be cytotoxic in several aspects. It is directly
   associated with inflammation and fibrosis in NAFLD,^[ [43]^8 ^]
   potentially through mechanisms that involve increasing Toll‐like
   receptor 4 signaling.^[ [44]^9 ^] FC also caused hepatocyte death by
   activating c‐Jun N‐Terminal Protein Kinase 1.^[ [45]^10 ^]
   Additionally, FC accumulation is observed in the development of liver
   cancer associated with fatty liver.^[ [46]^11 ^] Notably, recent
   studies have shown that hepatocyte death is one of the key triggers of
   NAFLD severity, leading to the subsequent development of inflammation,
   fibrosis, cirrhosis, and even hepatocellular carcinoma.^[ [47]^12 ^]
   However, the detailed mechanism by which cell death or injury responds
   to cholesterol accumulation (both FC and CE) in NAFLD remains yet to be
   fully understood.

   Given that cholesterol is closely involved in the pathology of NAFLD,
   cholesterol‐lowering therapy has been advocated and adopted in the
   management of this disease.^[ [48]^11 , [49]^13 ^] Statins no doubt
   have been the first choice.^[ [50]^14 ^] However, physicians are
   cautious about prescribing statins to patients with NAFLD, due to the
   concerns that statins might induce asymptomatic elevation in serum
   alanine transaminase (ALT),^[ [51]^15 ^] trigger autoimmune
   hepatitis,^[ [52]^16 ^] and rhabdomyolysis.^[ [53]^17 ^] Thus, statins
   are underused in NAFLD patients.^[ [54]^15 , [55]^18 ^] There is still
   an urgent need for seeking novel and effective pharmaceutical agents.

   Salvia miltiorrhiza Bge. (known as Danshen in Chinese) has been
   clinically shown to be effective in treating NAFLD, whether used alone
   or as part of an herbal formula.^[ [56]^19 ^] Salvia miltiorrhiza Bge.
   manages NAFLD involving the regulation of several key factors,
   including c‐Jun N‐terminal kinases, sterol regulatory element‐binding
   protein‐1c, carbohydrate response element binding proteins, peroxisome
   proliferator‐activated receptors (PPARs), and cytochrome P450.^[
   [57]^20 ^] One of the major tanshinones isolated from Salvia
   miltiorrhiza Bge, dihydrotanshinoneI(DhT), has been shown to have
   cardiovascular protective, anti‐inflammatory, anti‐oxidant, and
   hepaprotective actions.^[ [58]^21 ^] However, the therapeutic effect
   and mechanism of DhT in treating NAFLD are still unknown.

   In this study, we observed elevated levels of TC and FC in both NAFLD
   patients and high‐cholesterol diet (HCD)‐induced NAFLD mice compared to
   the healthy control, with a positive correlation with ALT levels.
   Rather than CE, FC was found to be detrimental to hepatocytes, as it
   increased the level of reactive oxygen species (ROS), impaired lysosome
   functions, and ultimately inhibited lipophagy, leading to cell
   apoptosis. Treatment of mice on an HCD with DhT alleviated FC‐induced
   cell apoptosis by activating PPAR𝛼 and promoting catalase expression,
   subsequently down‐regulating the FC‐triggered burst of ROS.

2. Results

2.1. The level of FC is significantly upregulated in NAFLD and positively
correlated with the ALT level

   NAFLD is accompanied by dyslipidemia and hepatic injury in clinic. To
   investigate the relationship between cholesterol (CHO) levels and
   hepatic injury, we collected plasma from 36 NAFLD patients and 30
   normal healthy controls to detect the levels of TC, FC, CE, and ALT,
   and to analyze their correlations. Figure [59]S1A‐C show the basic
   information of all individuals. Our findings revealed increased levels
   of TC, FC, and ALT in clinic plasma samples (Figure [60] 1A). Spearman
   correlation analysis showed a significant positive correlation between
   FC and ALT (Figure [61]1B), indicating the potential toxic role of FC
   in NAFLD.

Figure 1.

   Figure 1
   [62]Open in a new tab

   FC was positively correlated with hepatic injury in both NAFLD patients
   and mice. (A) Plasma levels of TC, FC, CE and ALT in human. (B)
   Correlation analysis between ALT value and TC, FC or CE levels in
   human. (C) Plasma levels of TC, FC, CE and ALT in mice. (D) Hepatic
   levels of TC, FC, CE and ALT in mice. (E‐F) Correlation analysis
   between ALT value and TC, FC or CE levels in mice plasma (E) and liver
   (F).

   Then, we established an NAFLD mouse model using a HCD.^[ [63]^22 ^] The
   successful model establishment was evidenced by increased body weight
   gain, higher liver index and NAFLD scores, and elevated levels of
   lipids and CHO compared to mice fed a normal diet (Figure [64]1C, D;
   [65]S1D‐I). The NAFLD mouse model produced similar outcomes to the
   human patients, demonstrating a positive correlation between FC and ALT
   in both the liver tissue and plasma (Figure [66]1E, F).

   Given that CHO can be transformed into oxysterols to maintain CHO
   hemostasis in the liver,^[ [67]^23 ^] gas chromatography–mass
   spectrometry was used to detect the major metabolites. Our findings
   revealed that an HCD notably increased the hepatic CHO levels, but
   there was no statistically significant increase in the levels of the
   five oxysterols (Figure [68]S2A). Additionally, we did not observe
   changes in CE levels (Figure [69]1C, D) or alterations in cholesterol
   esterase expression as a result of the HCD (Figure [70]S2B).

   These findings underscore the positive association between FC and liver
   injury in NAFLD, observed in both humans and mice, suggesting the
   potential toxic effects of FC.

2.2. FC triggered cell apoptosis

   To find out what was the impact of FC on the liver, we treated HepG2
   cell line and mouse primary hepatocytes with CHO, and different cell
   death inhibitors were applied. In Figure [71] 2A, CHO followed a
   dose‐dependent manner to lower the cell viability, which showed higher
   toxicity in comparison with the same doses of CE (Figure [72]S2C). At
   non‐toxic doses, Z‐VAD‐FMK, a pan‐caspase inhibitor of apoptosis, was
   observed to protect both HepG2 cells and primary hepatocytes from
   CHO‐induced toxicity (Figure [73]2B, E). Flow cytometry‐based apoptosis
   analysis revealed that CHO promoted cell apoptosis (Figure [74]2C, F)
   and resulted in an increase in the level of cleaved‐CASPASE3 (c‐CASP3)
   (Figure [75]2D, G). Furthermore, filipin staining of FC in mouse liver
   sections demonstrated that HCD administration led to a significant FC
   accumulation (Figure [76]S2D), which is consistent with the results
   obtained using a biochemical kit (Figure [77]1D). TUNEL staining of
   liver sections confirmed that HCD promoted cell apoptosis
   (Figure [78]S2E). Spearman correlation analysis indicated a strongly
   positive correlation between the FC level and the TUNEL positive area
   in mouse livers (Figure [79]2H). Immunoblotting of c‐CASP3 further
   supported the notion that FC increased cell apoptosis in NAFLD mouse
   livers (Figure [80]2I, [81]S3D). Therefore, we concluded that FC could
   trigger cell apoptosis in NAFLD mouse livers.

Figure 2.

   Figure 2
   [82]Open in a new tab

   FC triggered cell apoptosis. (A) Cell viability with CHO treatment.
   (B‐D) Cell viability with CHO and different cell death inhibitors
   treatment (B), flow analysis of apoptotic cells (C), and immunoblotting
   of c‐CASP3 (D) in HepG2 cells. (E‐G) Cell viability with CHO and
   different cell death inhibitors treatment (B), flow analysis of
   apoptotic cells (C), and immunoblotting of c‐CASP3 (D) in primary
   hepatocytes. (H‐I) Correlation analysis between TUNEL positive area and
   hepatic FC level (H) and immunoblotting of hepatic c‐CASP3 (I) in mice
   fed with ND or HCD.

2.3. FC induced lysosome dysfunction

   In addition to the apoptosis‐promoting effect of CHO, we also observed
   that the presence of chloroquine, an autophagy inhibitor, could
   exacerbate CHO toxicity on cell viability (Figure [83]2B, E).
   Considering the complex interactions between apoptosis and autophagy,^[
   [84]^24 ^] we sought to investigate whether autophagic flux played a
   role in the apoptotic death induced by CHO.

   Initially, we employed both an agonist (Rapamycin) and antagonists
   (Bafilomycin A1 and chloroquine) to co‐treat cells with CHO. The
   results indicated that both the agonist and the antagonists exacerbated
   the CHO‐induced toxicity, as evidenced by the lower cell viability
   compared to cells treated with CHO only (Figure [85]S2F, G).
   Subsequently, we delved into investigating how CHO impacted the
   autophagic flux.

   Subsequently, co‐immunostaining of microtubule‐associated proteins
   1A/1B light chain 3B (LC3B) and ubiquitin‐binding protein p62 (P62),
   two autophagic markers, revealed that CHO treatment not only increased
   the puncta and expression of LC3B and P62 but also promoted an increase
   in the co‐localization of these two markers (Figure [86] 3A; [87]S1A,
   BS3A, B). As autophagy in NAFLD is closely related to lipid
   degradation, commonly referred to as lipophagy, we proceeded to
   investigate the co‐localization of lipids and the lysosome. With CHO
   treatment, bodipy staining demonstrated elevated lipid accumulation in
   both HepG2 cells and primary hepatocytes, while lysosome puncta
   following lysotracker red staining showed reduced intensity, resulting
   in a decreased level of overlapped puncta (Figure [88]3B).

Figure 3.

   Figure 3
   [89]Open in a new tab

   FC induced lysosome dysfunctions. (A) Representative
   co‐immunofluorescence staining of DAPI, LC3B and P62 in HepG2 cells and
   primary hepatocytes without and with CHO treatment, scale bar: 50 µm.
   (B) Representative co‐staining of Hoechst 33342, lysotracker red and
   bodipy 505/515 in HepG2 cells and primary hepatocytes without and with
   CHO treatment, scale bar: 250 µm. (C) Representative fluorescence
   detection of HepG2 cells and primary culture transfected with a plasmid
   encoding GFP‐RFP‐LC3 without and with CHO treatment, scale bar: 250 µm.
   (D) Representative co‐immunofluorescence staining of DAPI, CTSB and
   LAMP1 in HepG2 cells and primary hepatocytes without and with CHO
   treatment, scale bar: 50 µm. (E) Transmission electron microscopy
   detection of autolysosomes (red arrow) in HepG2 cells without and with
   CHO treatment. (F) Immunoblotting of CTSB and CTSD in HepG2 cells and
   primary hepatocytes without and with CHO treatment. (G) Immunoblotting
   of hepatic P62, LC3B, CTSB and CTSD in mice fed with ND or HCD. (H)
   Hepatic ATP6V1A content detected using an Elisa kit in mice fed with ND
   or HCD.

   Next, the cells were transfected with a tandem mRFP‐GFP‐LC3 construct
   to investigate autolysosome formation. The assessment was based on the
   principle that the low pH inside the lysosome quenches the GFP
   fluorescent signal, while the mRFP fluorescence is more stable in an
   acidic environment.^[ [90]^25 ^] Therefore, normal autolysosome
   maturation would result in more RFP‐only puncta. However, if lysosome
   function is impaired, most of the puncta would display both red and
   green signals, appearing yellow. As illustrated in Figure [91]3C, cells
   challenged with CHO exhibited yellow signals. Additionally, we
   performed immunostaining of two proteins, cathepsin B (CTSB) and
   lysosomal‐associated membrane protein 1 (LAMP1), to confirm the
   impaired state of the lysosome, as CSTB is processed via proteases in
   lysosomes into a mature form, and LAMP1 represents a lysosomal membrane
   protein. As expected, CHO treatment resulted in the impairment of the
   lysosome, as evidenced by the significant downregulation of overlapped
   puncta of CTSB and LAMP1 (Figure [92]3D). Transmission electron
   microscopy was utilized to observe the lysosome and autolysosome, and
   the data demonstrated that autolysosomes accumulated after CHO
   treatment (Figure [93]3E), indicating the dysfunction of lysosomes.

   Then, we examined the expression of LC3B, P62, CTSB, and cathepsin D
   (CTSD) in both in vitro and in vivo samples. The results showed that
   CHO treatment increased P62 levels and the ratio of LC3BII to LC3BI
   (Figure [94]S3A, B, D; [95]3G). The reduced levels of mature CTSB and
   CTSD following CHO treatment indicated lysosome impairment
   (Figure [96]3F, G; [97]S3C, D). Additionally, we detected the
   expression of a key enzyme, V‐type proton ATPase catalytic subunit A
   (ATP6V1A), in liver homogenate as it is crucial for maintaining the
   acidic environment in lysosomes.^[ [98]^26 ^] Mice on an HCD showed
   decreased hepatic ATP6V1A activity, compared to the mice on an ND
   (Figure [99]3H).

   In summary, we could conclude that FC induced lysosome dysfunction.

2.4. FC triggered ROS burst to impair lysosome functions

   To find out how FC impaired lysosome, hepatic transcriptome study was
   performed. Differential gene expression between mice fed on a ND and
   those on a HCD was analyzed (Figure [100]S4A) and subjected to KEGG
   pathway enrichment analysis. Figure [101] 4A illustrates the most
   significant pathways, and the involvement of pathways related to NAFLD
   confirmed the successful establishment of the model. Furthermore, gene
   set enrichment analysis (GSEA) revealed the presence of autophagy,
   apoptosis, and inflammatory response pathways in HCD mice compared to
   ND mice (Figure [102]S4B‐D), consistent with our immunostaining and
   immunoblotting data (Figure [103]2, [104]3).

Figure 4.

   Figure 4
   [105]Open in a new tab

   Oxidative stress involved in FC‐induced toxicity. (A) KEGG pathway
   enrichment of differential expressed genes from HCD mice versus ND
   mice. (B) GSEA analysis of reactive oxygen species pathway in HCD and
   ND mice. (C‐D) ROS levels in HepG2 cells challenged with CHO (C) and ND
   or HCD mouse liver (D). (E) HepG2 cell viability with CHO, NAC and
   mito‐TEMPO treatment for 24 hours. (F) Immunoblotting of P62, LC3B,
   c‐CASP3, CTSB and CTSD in HepG2 cells with CHO, NAC or mito‐TEMPO
   treatment. (G) HepG2 cells post‐treatment of CHO and NAC with
   co‐immunofluorescence staining of DAPI, CTSB and LAMP1 (upper left,
   scale bar: 50 µm); co‐staining of Hoechst 33342, lysotracker red and
   bodipy 505/515 (upper right, scale bar: 250 µm); co‐staining of LC3B,
   P62 and DAPI (lower left, scale bar: 50 µm); and fluorescence detection
   after transfected with a plasmid encoding GFP‐RFP‐LC3 (lower right,
   scale bar: 250 µm). (H) Mitochondrial membrane potential of HepG2 cells
   with CHO treatment (scale bar: 100 µm).

   Given that oxidative stress has been well documented in lysosome
   impairment,^[ [106]^27 ^] we paid close attention in the oxidation
   related pathways. GSEA demonstrated that HCD was positively related to
   the ROS pathway (Figure [107]4B), and CHO treatment upregulated the ROS
   level both in vitro and in vivo (Figure [108]4C, D). Scavenging ROS
   using N‐acetyl‐l‐cysteine (NAC) or Mito‐TEMPO exhibited protective
   effects on HepG2 cells challenged by CHO (Figure [109]4E). Furthermore,
   NAC treatment reduced CHO‐induced lysosome impairment, as evidenced by
   decreased expression and co‐localization of autophagic markers (P62 and
   LC3B), increased maturation of CTSB and CTSD and their overlap with
   LAMP1, as well as a reduction in merged puncta between lipids and
   lysosomes (Figure [110]4F, G; [111]S4E). As mitochondria are a major
   source of ROS and mitochondrial membrane potential is directly
   associated with ROS burst,^[ [112]^28 ^] we subsequently measured the
   mitochondrial membrane potential using tetramethyl rhodamine ethyl
   ester (TMRE). In Figure [113]4H, CHO treatment demonstrated a
   dose‐dependent manner in disrupting membrane potential.

   Therefore, we concluded that FC disrupted the membrane potential of
   mitochondria to induce ROS burst, subsequently impairing lysosome
   functions (Figure [114]4I).

2.5. DhT mitigated NAFLD in mice on an HCD, irrespective of gut microbiota

   DhT, extracted from Salvia miltiorrhiza Bunge, has been extensively
   researched for its ability to alleviate oxidative stress ^[ [115]^29
   ^]. However, it was previously unclear whether DhT is effective in
   mitigating NAFLD by suppressing HCD‐induced oxidative damage.

   In this study, in addition to the HCD, mice received low (5 mg/kg),
   medium (10 mg/kg), and high (20 mg/kg) doses of DhT (dissolved in 0.5%
   CMC‐Na), with the vehicle serving as the negative control (Figure [116]
   5A). Furthermore, atorvastatin (Ato, 20 mg/kg) was used as a positive
   control (Figure [117]5A). Following an eight‐week treatment period, it
   was observed that both DhT and Ato reduced mice weight gain
   (Figure [118]5B) and decreased the liver index (Figure [119]5C).
   Moreover, histological analysis revealed that DhT mitigated liver
   damage in mice challenged with HCD, as evidenced by fewer vacuoles,
   reduced cell infiltration, less lipid accumulation, and decreased
   collagen deposition compared to the mice administered with the vehicle
   (Figure [120]5D, E, F). Additionally, DhT downregulated levels of ALT
   and total bilirubin, further supporting the hepatic protective effect
   of DhT (Figure [121]5G, H). As expected, Ato treatment exhibited
   similar efficacy in this mouse model (Figure [122]5D, E, F, G, H).

Figure 5.

   Figure 5
   [123]Open in a new tab

   DhT mitigated NAFLD in HCD‐fed mice. (A) Structure of DhT and
   experimental groups in this study. (B‐C) Mouse weight (B) and liver
   index (C) on day 56. (D) H&E, oil red O and Sirius red staining of
   mouse liver sections. (E‐F) NAFLD score (E) and fibrosis area (F) in
   mouse liver. (G‐H) Liver function indices, ALT (G) and total bilirubin
   (H). (I‐L) Hepatic levels of TG (I), TC (J), FC (K) and CE (L). (M‐O)
   Hepatic ROS (M) and antioxidases (N, Catalase; O, SOD) level. (P)
   Hepatic inflammatory markers level.

   Furthermore, lipid and CHO (including TC, FC and CE) quantifications
   demonstrated the beneficial regulatory actions of both DhT and Ato.
   There was a significant decline in TG, TC, and FC levels after DhT or
   Ato treatment, except for CE (Figure [124]5I, J, K, L). Additionally,
   ROS triggered by HCD was notably downregulated, accompanied by a
   significant increase in the expression of antioxidant proteins
   including catalase and superoxide dismutase (SOD) (Figure [125]5M, N,
   O). Moreover, the anti‐inflammatory effect of DhT was evidenced by a
   decrease in the levels of tumor necrosis factor alpha (TNFα),
   Interleukin 6 (IL‐6), and IL‐1β (Figure [126]5P).

   As gut microbiota is actively involved in the pathological and
   therapeutic aspects of NAFLD,^[ [127]^30 ^] it is reasonable to
   hypothesize the possibility that DhT might follow a gut
   microbiota‐dependent manner to mitigate NAFLD. To explore this impact,
   we conducted an analysis of changes in gut microbiota by performing 16S
   rRNA sequencing of fecal samples from mice that were on a ND, an HFD,
   as well as two different doses of DhT (10mg/kg, DhT10; 20mg/kg, DhT20).
   No significant changes were observed in the Ace and Chao indices for
   alpha diversity following DhT administration (Figure [128]S5A, B). Beta
   diversity was only evident between the mice fed with the ND and the
   HCD, and DhT did not lead to adjustments in the microbial community
   (Figure [129]S5C). Further analysis of specific microbial taxa at the
   phylum and family levels yielded similar results, indicating that DhT
   was ineffective at reshaping the HCD‐disturbed gut microbiota
   (Figure [130]S5D, E). However, it is worth noting that while the HCD
   decreased the gut microbiota health index, DhT was found to increase
   this index (Figure [131]S5F, G).

   To further explore the role of gut microbiota in the effectiveness of
   DhT on NAFLD mice, we employed a pseudo germ‐free model to determine
   whether DhT provided protection for NAFLD mice with depleted gut
   microbiota. Establishing a pseudo germ‐free state involved the
   administration of antibiotics every other day, which was confirmed by
   qPCR analysis of fecal 16S rRNA levels (Figure [132]S6A, B). In this
   pseudo germ‐free state, DhT demonstrated beneficial effects for NAFLD
   mice, including reduced body and liver weight, decreased steatosis,
   inflammation, and fibrosis, preserved liver function, and enhanced
   antioxidant capacity (Figure [133]S6C‐N). These findings suggest that
   gut microbiota is not essential for DhT to exert its protective effects
   on NAFLD.

2.6. DhT protected lysosome from FC‐induced damage

   Although promoting CHO excretion may be a protective mechanism,^[
   [134]^23 ^] DhT administration did not increase the fecal TC level
   (Figure [135]S7A), suggesting a minimal role of DhT in cholesterol
   excretion. Therefore, the protective function of DhT on HCD‐fed mice
   might be attributed to interactions between DhT and FC.

   As shown in Figure [136]S7B, DhT protected HepG2 cells and primary
   hepatocytes from CHO‐induced toxicity. We previously confirmed that CHO
   induced cell apoptosis through impairing the functions of lysosomes.
   Here, DhT has been shown to reduce the expression and co‐localization
   of P62 and LC3B (Figure [137] 6A; [138]S7C, D). DhT treatment resulted
   in decreased lipid accumulation and reduced co‐localization of lipids
   with lysosomes. However, the overlap of CTSB and LAMP1 puncta was
   increased in both HepG2 cells and primary hepatocytes (Figure [139]6A;
   [140]S7C, D).

Figure 6.

   Figure 6
   [141]Open in a new tab

   DhT reversed FC‐ induced lipophagy blockage. (A) Co‐staining of DAPI,
   P62 and LC3B (upper, scale bar: 50 µm); Hoechst 33342, lysotracker red
   and bodipy 505/515 (middle, scale bar: 250 µm); DAPI, CTSB and LAMP1
   (lower, scale bar: 50 µm) in HepG2 cells and primary hepatocytes with
   CHO and DhT treatment. (B) Fluorescence detection after transfected
   with a plasmid encoding GFP‐RFP‐LC3 in HepG2 cells and primary
   hepatocytes with CHO and DhT treatment, scale bar: 250 µm. (C)
   Transmission electron microscopy detection of autolysosomes (red arrow)
   in HepG2 cells with CHO and DhT treatment. (D) Immunoblotting and
   quantification of hepatic P62, LC3B, CTSB and CTSD in HCD mice with DhT
   treatment. (E) Hepatic ATP6V1A content detected using an Elisa kit in
   HCD mice with DhT treatment.

   In addition, the lipid accumulation and co‐localization between lipid
   and lysosome has been downregulated, whereas the overlapped puncta of
   CTSB and LAMP1 has been increased by DhT treatment in both HepG2 cells
   and primary hepatocytes (Figure [142]6A; [143]S7C, D). Plus, yellow
   signals after merging the GFP and RFP channels decreased after DhT
   treatment, suggesting the normal acidic environment of lysosome
   (Figure [144]6B). Transmission electron microscopy analysis
   demonstrated that DhT reduced autolysosomes accumulation induced by CHO
   treatment (Figure [145]6C).

   We next detected the level of autophagy markers (P62 and LC3B),
   apoptotic markers (c‐CASP3 and TUNEL staining), and lysosome function
   indicators (CTSB and CTSD) in NAFLD mouse liver samples. As expected,
   DhT reversed the detrimental effect of HCD following a dose‐dependent
   manner (Figure [146]6D, [147]S7C‐E). The level of ATP6V1A, a key factor
   to maintain the acidic environment in lysosomes, was promoted after DhT
   administration (Figure [148]6E), implying the recovered lysosomal
   functions.

2.7. Peroxisome proliferator‐activated receptor alpha (PPAR𝛼) is involved in
the effect of DhT

   Transcriptome analysis in mouse liver was applied to find out by which
   mechanism DhT was able to exert protection on NAFLD mice. First, we
   have observed that the disturbed genes by HCD could be reversed by DhT
   treatment (Figure [149] 7A), which indicated the therapeutical efficacy
   of DhT on mice challenged by HCD. KEGG pathway analysis of
   differentially expressed genes identified from DhT‐treated mice versus
   HCD mice revealed the involvement of the PPAR signaling pathway
   (Figure [150]7B), aligning with the concept that PPARs are potential
   therapeutic targets for NAFLD.^[ [151]^31 ^] Next, the expression of
   three isoforms of PPARs, including PPAR𝛼, PPAR𝛾 and PPAR𝛿, have been
   detected both in vitro and in vivo. With DhT treatment, PPAR𝛼 was found
   to be promoted, while the expression of PPAR𝛾 and PPAR𝛿 was not altered
   significantly (Figure [152]7C, [153]S8A).

Figure 7.

   Figure 7
   [154]Open in a new tab

   PPAR𝛼 is involved in the effect of DhT. (A) Differential expressed
   genes among ND, HCD and DhT mice. (B) KEGG pathway enrichment of
   differential expressed genes from HCD mice versus HCD+DhT mice. (C)
   Immunoblotting and quantification of PPARs in mouse liver. (D‐E) Cell
   viability (D left), ROS content (D right) and immunoblots of P62, LC3B,
   c‐CASP3, mature CTSB and mature CTSD (E) in siRNA‐transfected HepG2 and
   primary hepatocytes with CHO and DhT treatment. (F) Transmission
   electron microscopy detection of autolysosomes (red arrow) in
   siRNA‐transfected HepG2 cells with CHO and DhT treatment.

   Subsequently, siRNA‐mediated gene silencing was applied to elucidate
   the specific role of each PPAR isoform. The successful gene silencing
   was confirmed by the reduced expression at the translational level
   (Figure [155]SB). We assessed cell viability and ROS levels post‐siRNA
   transfection followed by CHO and DhT treatment. The protective effect
   of DhT on CHO‐challenged cells (HepG2 and primary hepatocytes) was
   diminished upon PPAR𝛼 silencing, with similar cell viability and ROS
   levels compared to cells treated with CHO only (Figure [156]7D).

   Next, autophagy markers (P62 and LC3B), apoptotic marker (c‐CASP3), and
   lysosome function indicators (CTSB and CTSD) were detected in cells
   with PPARs silencing. Similar to previous data, DhT was able to
   regulate those markers in cells with control siRNA transfection, and
   knockdown of PPAR𝛼, rather than PPAR𝛾 and PPAR𝛿, failed to exhibit the
   protective action of DhT (Figure [157]7E, [158]S8C). Additionally,
   autolysosomes displayed using transmission electron microscopy analysis
   gave another piece of evidence to elucidate the importance of PPAR𝛼
   (Figure [159]7F).

   Therefore, we could conclude that PPAR𝛼 is key to the effect of DhT on
   NAFLD.

2.8. DhT activated PPAR𝛼 to promote catalase expression

   PPAR𝛼 is a ligand‐activated transcription factor that governs the
   metabolic homeostasis. Upon activation, PPAR𝛼 translocates into the
   nucleus to exert its transcriptional regulation on target genes.^[
   [160]^32 ^] The distribution of PPAR𝛼 was examined using immunoblotting
   and immunostaining in HepG2 cells and mouse liver sections. Both
   techniques revealed that DhT promoted the nuclear distribution of PPAR𝛼
   (Figure [161] 8A‐D). In addition to the increased nuclear distribution,
   we observed the expression of the confirmed downstream target genes of
   PPAR𝛼, including fatty acid transport protein 1 (Fatp1), carnitine
   palmitoyl transferase 1 alpha (Cpt1𝛼) and acyl‐CoA oxidase 1 (Acox1),
   could be elevated by DhT (Figure [162]S8D). Thus, DhT has been
   confirmed to activate PPAR𝛼.

Figure 8.

   Figure 8
   [163]Open in a new tab

   DhT bound and activated PPAR𝛼 to promote catalase expression. (A‐B)
   PPAR𝛼 expression and localization in cytosol and nuclear fractions in
   HepG2 cells (A, C) and mouse liver (B, D) with CHO and DhT treatment.
   C: scale bar: 250 µm; D: scale bar: 100 µm. (E) Docking simulation of
   PPAR𝛼 LBD (PDB: 3vi8) and DhT. (F) Binding affinity between PPAR𝛼 LBD
   wild type and mutant using microscale thermophoresis. (G‐H) Catalase
   expression in siRNA‐transfected HepG2 with CHO and DhT treatment at the
   transcriptional (G) and translational (H) levels. (I) Level of catalase
   in mouse liver with HCD and DhT treatment. (J) Promoter activities of
   Ppar𝛼 and catalase in HepG2 cells with CHO and DhT treatment. (K)
   Graphical summary illustrating the protective mechanism of DhT.

   Subsequently, after conducting a simulation of the protein‐ligand
   binding interaction using Autodock 4.2, we observed that Threonine 283
   (T283) and Methionine 320 (M320) formed a binding pocket, anchoring DhT
   within the ligand binding domain (LBD) of PPAR𝛼 with a binding energy
   of ‐9.38 kcal/mol (Figure [164]8E). We next expressed, purified, and
   concentrated recombinant PPAR𝛼 LBD fused with an EGFP tag for the
   subsequent binding interaction analysis (Figure [165]S9). The binding
   affinity (Kd) between DhT and PPAR𝛼 LBD was measured using microscale
   thermophoresis. It was observed that EGFP did not interact with DhT,
   making the Kd calculation unfeasible (Figure [166]8F). Notably, the Kd
   between DhT and PPAR𝛼 LBD was determined to be 1.66 ± 0.5 μM
   (Figure [167]8F), indicating a possible interaction. Given DhT
   interacted with the main chain of M320 and the side chain of T283 in
   PPAR𝛼 we subsequently introduced a mutation where threonine was changed
   to alanine (T283A) (Figure [168]8E), and the Kd was re‐evaluated. The
   mutated protein exhibited a lower affinity, resulting in an increased
   Kd of 32.9 ± 6.1 μM (Figure [169]8F).

   Since DhT treatment led to the decrease in hepatic ROS content in NAFLD
   mice, we screened for potential antioxidants in wild‐type cells and
   PPAR𝛼‐silenced cells. Data showed that DhT elevated the levels of
   catalase, SOD1, SOD2, thioredoxin 1 (TRX1) in cells with CHO treatment
   (Figure [170]8G, H;  [171]S8E); whereas PPAR𝛼 knockdown abolished the
   action of DhT to promote catalase expression at the transcriptional and
   translational levels (Figure [172]8G, H;  [173]S8C). Additionally, the
   effect of DhT to promote catalase expression was observed in NAFLD mice
   following a dose‐dependent manner (Figure [174]8I).

   We further investigated the transcriptional regulation of PPAR𝛼 and
   catalase expression mediated by DhT‐activated PPARα using a luciferase
   assay. It is known that PPAR𝛼 can regulate its own expression by
   binding to its promoter upon activation by an agonist.^[ [175]^33 ^] In
   our study, DhT‐activated PPAR𝛼 increased the activity of the PPAR𝛼
   promoter (Figure [176]8J), leading to higher PPAR𝛼 expression.
   Additionally, we observed an increase in the luciferase signal of
   catalase promoter in the presence of DhT‐activated PPAR𝛼
   (Figure [177]8J). However, this effect was not observed when PPAR𝛼 was
   mutated (Figure [178]8J).

   In summary, we concluded that DhT activated PPAR𝛼 to promote catalase
   expression, which could down‐regulate the ROS level induced by HCD
   (Figure [179]8K).

2.9. Adeno‐associated virus (AAV)‐mediated PPAR𝛼 knockdown reduced the
protective effect of DhT on NAFLD mice

   To further explore the importance of PPAR𝛼, we applied AAV‐mediated
   shRNA delivery to reduce the hepatic PPAR𝛼 level, followed by the NAFLD
   model establishment and DhT treatment (Figure [180] 9A). shRNA
   efficiencies were verified in murine cells lines and shRNA‐1 was
   selected for the in vivo experiment (Figure [181]S10).

Figure 9.

   Figure 9
   [182]Open in a new tab

   Hepatic PPAR𝛼 knockdown reduced the effect of DhT. (A) Experimental
   grouping in this study. (B–C) Mouse weight (B) and liver index (C) on
   day 56. (D) H&E, oil red O and Sirius red staining of mouse liver
   sections. (E–F) NAFLD score (E) and fibrosis area (F) in mouse liver.
   (G–H) Liver function indices, ALT (G) and total bilirubin (H). (I–L)
   Hepatic levels of TG (I), TC (J), FC (K) and CE (L). (M–N) Hepatic ROS
   (M) and catalase (N) level; (O–P) Hepatic inflammatory markers level.
   (R) Expressions of PPAR𝛼, catalase, P62, LC3B, c‐CASP3, mature CTSB and
   mature CTSD in AAV mouse liver.

   In mice receiving control shRNA (AAV‐Control), DhT was beneficial to
   reduce body weight gain (Figure [183]9B), liver index (Figure [184]9C),
   NAFLD score (Figure [185]9D, E), fibrosis area (Figure [186]9F) and
   level of TG (Figure [187]9I), TC (Figure [188]9J), FC (Figure [189]9K)
   and inflammatory markers (Figure [190]9O, P, Q). In addition, DhT
   maintained the liver function (Figure [191]9G, H) and increased
   antioxidant capacity (Figure [192]9M, N). Levels of PPAR𝛼, catalase,
   autophagy markers (P62 and LC3B), apoptotic marker (c‐CASP3), and
   lysosome function indicators (CTSB and CTSD) showed the similar trends
   (Figure [193]9R) after DhT treatment (Figure [194]6, [195]7
   and [196]8I).

   The successful hepatic PPAR𝛼 knockdown was confirmed by the reduced
   PPAR𝛼 expression detected by the immunoblotting (Figure [197]9R).
   Further, the action of DhT to promote PPAR𝛼 was prohibited
   (Figure [198]9R). As expected, the effect of DhT to ameliorate NAFLD
   was weakened as shown in Figure [199]9B‐R. Therefore, PPAR𝛼 is pivotal
   to the protective effect of DhT on NAFLD.

3. Discussion

   CHO is an important lipid molecule for animal cells, maintaining the
   barrier function between the environment and cells, modulating fluidity
   of biological membranes and forming lipid rafts that concentrate
   signaling molecules.^[ [200]^23 ^] Under the physiological condition,
   cellular CHO levels are determined by the interplay between de novo
   biosynthesis, uptake, export and storage to maintain the CHO
   hemostasis.^[ [201]^34 ^] The liver is a key organ involved in the
   regulation of CHO hemostasis, and hepatic CHO is found in the form of
   FC and CE, both of which are constituents of circulating
   lipoproteins.^[ [202]^34 ^] Besides the de novo synthesis, absorption
   from the diet is another major source of hepatic CHO. Thus, HCD
   containing 0.25 to 2% cholesterol has been widely used to establish
   NAFLD models.^[ [203]^35 ^]

   Here, despite the unchanged CE content, higher FC level was observed in
   NAFLD patients and HCD‐induced NAFLD mice, in consistent with the
   previous reports.^[ [204]^7 , [205]^36 ^] Though FC has been one of the
   lipotoxic chemicals, we are the first to report the positive
   correlation between the level of FC with that of ALT. As metabolizing
   CHO into oxysterols could detoxify the CHO‐induced toxicity, we thereby
   detected the levels of major oxysterols and found that, with HCD
   administration, only CHO content was promoted. Therefore, there might
   be a potential link between FC and NAFLD pathology.

   Despite the previously well‐documented mechanism of saturated fatty
   acid‐mediated lipotoxicity in the liver, recent case‐control studies
   and follow‐up research have shown that hypercholesterolemia or high
   consumption of CHO, rather than the intake of fat or saturated fatty
   acid, was associated with a higher risk of NAFLD.^[ [206]^37 ^] FC,
   rather than CE, has been widely confirmed to play a significant role in
   this context.^[ [207]^6 , [208]^7 , [209]^36 , [210]^38 ^] The toxicity
   of FC has been addressed in both parenchymal cells (hepatocytes) and
   non‐parenchymal cells (Kupffer cells and stellate cells) in the liver,
   and the injury of each cell type would aggravate FC‐induced damage to
   the other cells. The accumulation of FC, rather than triglycerides or
   fatty acids, increases the sensitivity of hepatocytes to inflammatory
   cytokines‐mediated steatohepatitis via decreasing the level of
   mitochondrial glutathione.^[ [211]^39 ^] CHO crystals, accumulated and
   precipitated in the damaged hepatocytes, contribute to triggering an
   inflammatory response by interacting with NLRP3 inflammasomes in
   Kupffer cells.^[ [212]^40 ^] Activation of hepatic stellate cells has
   been involved in FC accumulation and has been reported to regulate
   Toll‐like receptor 4 signaling^[9a] or increase the sensitivity
   transforming growth factor beta (TGFβ)‐induced activation in a vicious
   cycle,^[9b] leading to fibrosis in NASH. Additionally, as an endpoint
   outcome of FC‐induced toxicity, cell death has been described in this
   NAFLD. Being the predominant mode of cell death, the presence of
   apoptotic cell death has been observed in clinical samples as evident
   by the higher level of caspases or TUNEL positive signals,^[ [213]^41
   ^] which is possible due to subcellular dysfunctions including
   endoplasmic reticulum stress, mitochondrial dysfunction and lysosomal
   dysfunction.^[38b] Besides, necroptosis, pyroptosis and
   autophagy‐related death have been reported yet less understood than
   apoptosis in NAFLD.^[ [214]^42 ^] Thus, there lacks a detailed
   explanation to decipher how hepatocytes response to FC challenge, as
   well as a thorough mechanism by which FC kills or damages the normal
   cellular functions.

   As a defense mechanism in NAFLD, autophagy degrades accumulated lipids
   inside hepatocytes, a process termed lipophagy.^[ [215]^43 ^] However,
   lipophagy is functionally impaired in NAFLD patients, leading to lipid
   overaccumulation.^[ [216]^44 ^] In this study, we discovered that FC
   initially promotes autophagy induction but subsequently blocks the
   later autophagic degradation process. This finding is consistent with
   previous reports that hepatic autophagy is enhanced in the early stages
   of NAFLD but is blocked at later stages.^[ [217]^45 ^] Our study
   provides a molecular mechanism for this phenomenon: FC accumulation
   increases ROS production, damaging lysosomal function and thereby
   blocking the later stages of lipophagy. This insight offers a novel
   perspective on why lipophagy fails in NAFLD.

   Given that FC is more toxic than CE in HepG2 cells and primary
   hepatocytes, and that HCD administration increases TUNEL‐positive cells
   in liver sections, we investigated whether cell death was related to
   lipophagy blockage. By using various inhibitors targeting different
   cell death pathways (apoptosis, necroptosis, ferroptosis, and
   autophagy), we found that FC induces caspase‐dependent apoptosis.
   Additionally, in the presence of an autophagy agonist (rapamycin) or
   antagonists (chloroquine and bafilomycin A1), FC further enhances cell
   death. A myriad of evidence demonstrates crosstalk between autophagy
   (or lipophagy) and apoptosis.^[ [218]^46 ^] The interconnection between
   autophagy and apoptosis is complex, often occurring in sequence, with
   autophagy preceding apoptosis. In most cases, autophagy can remove
   toxic or obsolete proteins and organelles, thereby reducing the
   tendency of cells to undergo apoptosis.^[ [219]^47 ^] As lysosomes are
   responsible for autophagy‐related degradation, their dysfunction leads
   to autophagy blockage and debris accumulation, ultimately inducing
   apoptosis.^[ [220]^48 ^] The widely accepted mechanism explaining
   lysosome dysfunction‐related apoptosis involves mitochondrial
   participation. Exposed to stimuli such as growth factor deprivation,
   hypoxia, DNA damage, and oxidative stress, mitochondria directly
   trigger caspase activation or induce ROS bursts that rupture lysosomes,
   releasing cysteine proteases to augment apoptotic factors.^[48b] We
   propose that FC increases ROS levels, leading to lysosome dysfunction.
   Therefore, FC toxicity is attributed to ROS‐induced lysosomal damage.
   Our observation provides the first explanation for why FC accumulation
   is toxic in the liver.

   CHO‐lowering medications, including but not limited to atorvastatin,
   are beneficial to ameliorate NAFLD from this study as well as others’
   research.^[ [221]^14 , [222]^15 ^] Patients receiving statins treatment
   not only exhibited lower lipid and CHO levels but also showed decreased
   c‐CASP3 content and increased catalase levels (Figure [223]S1J, K).
   Moreover, we have introduced DhT as a new potential candidate for not
   only lowering CHO levels but also attenuating FC‐induced toxicity,
   which, to our best of knowledge, it is the first to report this
   efficacy. PPARs are ligand‐activated transcription factors of nuclear
   hormone receptor superfamily comprising of the following three subtypes
   including PPAR𝛼, PPAR𝛾 and PPARδ. Each PPAR consists of an LBD and a
   DNA binding domain. After interacting with ligands, PPARs are
   translocated to the nucleus, where they activate their transcriptional
   function by binding to specific DNA regions of target genes known as
   peroxisome proliferator hormone response elements.^[ [224]^31 ^] PPARs
   have emerged as integrators of energy homeostasis and metabolic
   function, and their pivotal roles in disease management have well been
   documented in metabolic disorders, cancers and liver diseases.^[
   [225]^49 ^] Due to crucial metabolic regulatory roles and excellent
   draggability, PPARs have been widely discussed clinically in the
   treatment of metabolic disorders, including NAFLD. For example, PPAR𝛼
   agonists, such as Fenofibrate and Gemfibrozil, have demonstrated
   reliable improvement of liver function and lipid profiles. However,
   they have only exhibited minimal effects on histopathology in NAFLD.^[
   [226]^50 ^] Pioglitazone, a PPAR𝛾 selective agonist, has been confirmed
   effective in treating patients with NAFLD and type 2 diabetes. However,
   the efficacy to treat NAFLD patients without type 2 diabetes is
   limited.^[ [227]^51 ^] The research progress of PPAR𝛿 agonists has not
   been as promising as others. For example, Seladelpar's clinical trial
   has been discontinued due to atypical histological findings.^[ [228]^52
   ^] What is more, PPAR dual/pan agonists, including Lanifibranor and
   Saroglitazar, have been developed for treating NAFLD.^[ [229]^52 ^]
   Apart from the beneficial effects of these agonists, adverse events
   such as impaired kidney function, rhabdomyolysis, weight gain, and
   cardiovascular dysfunctions have also been reported.^[ [230]^52 ,
   [231]^53 ^] Therefore, addressing the resolution of NAFLD remains a
   challenging task, requiring further concerted efforts.

   Our data confirm that DhT is a promising candidate for treating NAFLD.
   This study is the first to report that DhT is a novel and potent ligand
   for PPAR𝛼. As a ligand‐inducible transcription factor, DhT binds to and
   activates PPAR𝛼, thereby triggering its transcriptional regulatory
   actions. We observed that DhT‐activated PPAR𝛼 increased the activity of
   the Ppar𝛼 promoter, indicating a self‐regulatory effect, which is a
   typical characteristic of PPARs.^[ [232]^33 , [233]^49 , [234]^50 ,
   [235]^51 , [236]^52 , [237]^53 , [238]^54 ^] ^[33,49b‐d,54] Given that
   oxidative stress is a key contributor to the detrimental effects
   induced by FC accumulation, we investigated whether the beneficial
   effects of PPAR𝛼 in NAFLD were related to its antioxidant action.
   Knockdown of PPAR𝛼 abolished the antioxidant effect of DhT. Next, we
   screened for antioxidant genes regulated by DhT‐activated PPAR𝛼 and
   identified catalase as a potential downstream target. Similar to Ppar𝛼,
   catalase expression was upregulated following DhT treatment, which was
   achieved through increased recruitment of the catalase promoter to
   DhT‐activated PPAR𝛼.

   Although the PPAR𝛼‐catalase pathway has been reported in the treatment
   of alcoholic liver disease,^[ [239]^55 ^] hepatic fibrosis,^[ [240]^56
   ^] and skin aging,^[ [241]^57 ^] this study is the first to elucidate
   its role in mitigating FC‐induced oxidative stress in NAFLD through DhT
   treatment. Furthermore, while the antioxidant effects of DhT have been
   previously reported in stroke^[ [242]^58 ^] and atherosclerosis,^[
   [243]^59 ^] this is the first study to demonstrate that antioxidant
   action of DhT is effective in treating NAFLD, highlighting its
   potential as a therapeutic agent. To more comprehensively evaluate the
   anti‐NAFLD effects of DhT, we suggest that additional NAFLD models
   beyond the HCD‐induced model should be employed, which will be a focus
   of our future research efforts.

   In summary, our work delineate that FC is toxic in the pathology of
   NAFLD, owing to the detrimental effect of FC to induce lysosome
   dysfunctions to provoke hepatocyte apoptosis. In addition, we propose a
   novel PPAR𝛼 agonist, DhT, to treat NAFLD, through a mechanism by which
   DhT activates and binds to PPAR𝛼 agonist to down‐regulate FC‐induced
   ROS burst, therefore showing protective efficacy.

4. Experimental Section

Human samples

   This study was approved by the Ethics Committee of the Second
   Affiliated Hospital, Zhejiang University School of Medicine (Approval
   No. 2019–420). The study was non‐interventional and did not involve
   additional treatment, operations, or examinations on the subjects, and
   did not pose any adverse reaction or risk. Thirty normal healthy
   controls, thirty‐six NAFLD patients without treatment (clinically
   confirmed by physicians), and thirty NAFLD patients after statin
   treatment were enrolled from the outpatient records after obtaining
   written consent. Plasma samples were collected to determine lipid,
   cholesterol (CHO) and ALT levels.

Animal study

   The animal studies described in this research were approved and carried
   out in accordance with the guidelines of the University Animal
   Experimentation Ethics Committee at The Chinese University of Hong Kong
   (21‐210‐MIS). Male four‐week‐old C57BL/6J mice, provided by Laboratory
   Animal Services Centre at The Chinese University of Hong Kong, were
   used to establish the NAFLD model.

Animal study—High‐cholesterol diet (HCD) induced NAFLD mouse model

   HCD was prepared by adding 1% CHO (sc‐202539A, Santa Cruz, Dallas, TX,
   USA) to an HFD consisting of 60% fat.^[49b] The NAFLD model was induced
   by administering HCD for eight weeks, while the normal chow diet (ND)
   was used as the control. Plasma and liver samples were collected after
   the eight‐week administration.

Animal study—Drug treatment on HCD‐induced NAFLD mice

   Dihydrotanshinone I (DhT, SD8290, Solarbio) was administered to mice by
   gavage every other day during the HCD‐fed period at doses of 5, 10, and
   20 mg/kg, dissolved in 0.5% CMC‐Na solution (HY‐Y1889, MedChemExpress,
   Monmouth Junction, NJ, USA). Atorvastatin (Ato, A6800, Solarbio) was
   used as a positive control. The vehicle (0.5% CMC‐Na solution) was the
   negative control for both ND‐ or HCD‐fed mice. The grouping information
   is presented in Figure [244]5A.

Animal study—Pseudo‐germ‐free mouse model establishment and DhT treatment

   A pseudo‐germ‐free model was created by administering a mixture of
   antibiotics, including cefadroxil (50 mg/kg; IC0160, Solarbio),
   terramycin (50 mg/kg; IO0130, Solarbio), and erythromycin (50 mg/kg;
   E8100, Solarbio) every other day.^[ [245]^60 ^] Fecal 16S rDNA levels
   were assessed using qPCR analysis every two weeks to confirm the
   pseudo‐germ‐free state. The NAFLD model was established as above at
   4.2.2 during the antibiotics administration period, and DhT (20 mg/kg)
   was orally administered to these mice every other day. The grouping
   information is presented in Figure [246]S6A.

Animal study—Adeno‐associated virus (AAV)‐mediated hepatic PPAR𝛼 knockdown
and DhT treatment

   The AAV containing PPAR𝛼 shRNA was prepared following our previous
   work.^[ [247]^49c ^] In vivo knockdown of PPAR𝛼 was achieved by
   tail‐vein injection of AAV (1 × 10^12 gc/mouse in 50 μl of normal
   saline), followed by HCD administration to establish the NAFLD state.
   Additionally, DhT (20 mg/kg) was administered orally to these mice
   every other day. The grouping information is presented in
   Figure [248]9A.

Primary hepatocytes isolation

   Primary hepatocytes were isolated from mice (three mice at a time)
   following a published protocol.^[ [249]^61 ^] After anesthesia, the
   liver was perfused, hepatocytes were dissociated using collagenase,
   separated from other cells, and cultured for subsequent experiments.

Cell culture

   HepG2 cells (ATCC, Manassas, VA, USA) and primary hepatocytes were
   cultured in Dulbecco's Modified Eagle's Medium (DMEM, 11965092, Thermo
   Fisher Scientific, Waltham, MA, USA) supplemented with 10% fetal bovine
   serum (A5256701, Thermo Fisher Scientific) and 1%
   penicillin/streptomycin (15140122, Thermo Fisher Scientific).

MTT assay

   3‐(4,5‐dimethylthiazol‐2‐yl)‐2,5‐diphenyltetrazolium (MTT) assay
   (M6494, Thermo Fisher Scientific) was used to assess cell viability.
   Cells were seeded onto 96‐well plates and treated for 24 hours with the
   following compounds: CHO (sc‐202539A, Santa Cruz, Dallas, TX, USA; 1,
   10, and 100 μM), cholesterol esters (cholesterol palmitate, sc‐214704,
   Santa Cruz: 10 and 100 μM; cholesterol stearate, sc‐214708, Santa Cruz
   10 and 100 μM), DhT (0.1, 0.5, and 1 µg/mL), N‐acetyl‐l‐cysteine (NAC,
   A7250, Sigma‐Aldrich, Saint Louis, MO, USA: 0.1, 0.2, and 2.0 mM),
   Mito‐TEMPO (SML0737, Sigma‐Aldrich: 1, 5, and 10 μM), chloroquine
   (14194, Cayman chemical, Ann Arbor, MI, USA: 10 and 50 μM), Z‐VAD‐FMK
   (C1202, Beyotime, Beijing, China: 5 and 10 μM), necrostatin‐1 (SC4359,
   Beyotime, Beijing, China: 5 and 10 μM), ferrostatin‐1 (17729, Cayman
   chemical: 5 and 10 μM), bafilomycin A1 (S1413, Selleckchem, Houston,
   TX, USA: 25 nM), or rapamycin (sc‐3504, Santa Cruz: 500 nM).

Flow cytometer analysis (ROS and Apoptosis)

   Cellular and hepatic levels of reactive oxygen species (ROS) were
   measured using a Total ROS Assay kit (88‐5930‐74, Thermo Fisher
   Scientific). The probe from the kit was added to the cells (after CHO
   or/and DhT treatment) or liver homogenate and incubated for 30 minutes,
   followed by flow cytometric analysis.

   Cell apoptosis was assessed using an Annexin V‐FITC cell apoptosis
   detection kit (C1062S, Beyotime). Flow cytometry was employed to detect
   apoptosis in both HepG2 cells and primary hepatocytes challenged with
   CHO or/and DhT.

Biochemical assays

   Levels of lipids, total cholesterol (TC), free cholesterol (FC),
   alanine transaminase (ALT), triglycerides (TG), high density
   lipoprotein cholesterol (HDL‐c), low density lipoprotein cholesterol
   (LDL‐c), cleaved CASPASE3 (c‐CASP3), catalase, V‐type proton ATPase
   catalytic subunit A (ATP6V1A), superoxide dismutase (SOD), tumor
   necrosis factor alpha (TNFα), Interleukin 6 (IL‐6) and IL‐1β were
   detected using the following kits: TC content assay kit (BS1985,
   Solarbio); FC content assay kit (BS1895, Solarbio); ALT activity assay
   Kit (ab105134, Abcam, Cambridge, UK), Stanbio™ Triglyceride Liquid
   Reagent (2100430, Fisher Scientific, Pittsburgh, PA, USA), HDL‐c
   (BC5325, Solarbio), LDL‐c (BC5335, Solarbio), human cleaved caspase‐3
   (Asp175) ELISA kit (ab220655, Abcam), catalase activity assay kit
   (ab83464, Abcam; BC0205, Solarbio), mouse ATP6V1A ELISA kit (RK13939,
   ABclonal, Wuhan, China), SOD activity kit (BC5165, Solarbio), mouse
   TNFα (SEKM‐0034, Solarbio), mouse IL‐6 (SEKM‐0007, Solarbio) and mouse
   IL‐1β (SEKM‐0002, Solarbio).

siRNA transfection

   Transfection of control siRNA (sc‐37007, Santa Cruz), Pparα siRNA
   (human: sc‐36307; mouse: sc‐36308; Santa Cruz), Pparγ siRNA (human:
   sc‐29455; mouse: sc‐29456; Santa Cruz) or Pparδ siRNA (human: sc‐36305;
   mouse: sc‐36306) was carried out using Lipofectamine 2000 (11668027,
   Thermo Fisher Scientific) as recommended by the manufacturer. Then,
   cells with siRNA transfection were subject to CHO and DhT treatment as
   above.

Transmission Electron Microscopy

   HepG2 cells treated with CHO and DhT were collected and washed three
   times with PBS. The cells were then fixed with a mixture of 2%
   formaldehyde and 2.5% glutaraldehyde in 0.1 M Sodium Cacodylate buffer
   (pH 7.4) at 4 °C. After three times of washing with 0.1 M Sodium
   Cacodylate buffer, the cells were post‐fixed with 1% OsO[4] for 1 hour
   at room temperature. Then cells were undergoing dehydration using a
   graded acetone series (10%‐100%). Subsequently, samples were embedded
   in Embed‐812 resin (14120, Electron Microscopy Sciences, Hatfield, PA).
   Thin sections (100 nm thick) from the sample blocks were examined using
   a Hitachi 7400 TEM (Hitachi‐High Technologies, Tokyo, Japan) operated
   at 80 kV.

Immunoblot

   Total protein from cells or tissues was extracted using RIPA lysis and
   extraction buffer (89900, Thermo Fisher Scientific) and quantified with
   the Pierce™ BCA Protein Assay Kit (23227, Thermo Fisher Scientific).
   Nuclear and cytosol fractions were isolated using NE‐PER™ Nuclear and
   Cytoplasmic Extraction Reagents (78833, Thermo Fisher Scientific). The
   following antibodies were used for immunoblotting: PPARα (sc‐398394,
   Santa Cruz), PPARγ (2435T, Cell Signaling Technology, Danvers, MA,
   USA), PPARδ (GTX113250, GeneTex, Irvine, CA, USA), cleaved Caspase‐3
   (c‐CASP3, 9664T, Cell Signaling Technology), LC3B (3868T, Cell
   Signaling Technology), SOAT1 (sc‐137013, Santa Cruz), SOAT2 (sc‐69837,
   Santa Cruz), cathepsin D (CTSD, sc‐377299, Santa Cruz), cathepsin B
   (CTSB, sc‐365558, Santa Cruz), LC3B (3868T, Cell Signaling Technology),
   P62 (sc‐48402, Santa Cruz), catalase (A11220, ABclonal), Histone H3
   (9715S, Cell Signaling Technology), and β‐Actin (sc‐47778, Santa Cruz).

AAV preparation

   AAV‐ short hairpin RNA (shRNA)‐ctrl was a gift from Hongjun Song
   (Addgene plasmid # 85741; [250]http://n2t.net/addgene:85741; RRID:
   Addgene 85741). pAAV2/8 and pAdDeltaF6 plasmids were gifts from James
   M. Wilson (Addgene plasmid # 112864;
   [251]http://n2t.net/addgene:112864; RRID: Addgene 112864; Addgene
   plasmid # 112867; [252]http://n2t.net/addgene:112867; RRID:
   Addgene_112867). The shRNA sequence targeting PPAR𝛼 (Accession:
   [253]NM_001113418.1) was designed using an online tool‐
   [254]https://biosettia.com/support/shrna‐designer/ (Table [255]S1). An
   AAV‐shRNA‐ctrl (AAV‐Control) plasmid was constructed to incorporate the
   PPAR𝛼 shRNA (AAV‐ PPAR𝛼). AAV packaging was performed according to our
   previous publication^[49c] AAV titer was calculated using qPCR
   (Table [256]S1)

RNA extraction, real‐time PCR analysis and RNA Sequencing

   RNA from cells and tissues was extracted from cells and tissues with a
   RNAeasy™ Animal Long RNA Isolation Kit with Spin Column (R0027,
   Beyotime) and quantified using NanoDrop™ One/OneC Microvolume UV‐Vis
   Spectrophotometer (Thermo Fisher Scientific). cDNA conversion from mRNA
   was conducted using a BeyoRT™ III First Strand cDNA Synthesis Kit
   (D7178M, Beyotime). mRNA levels were analyzed with Luna Universal qPCR
   Master Mix (M3003L, New England Biolabs, Ipswich, MA, USA). Primers are
   listed in Table [257]S2.

   Meanwhile, RNA samples were sent to BGI Tech Solutions Co., Ltd. (Hong
   Kong) for RNA sequencing. After assessing RNA integrity and
   quantification, the downstream sequencing process involved removing
   ribosomal RNA before library construction, followed by mRNA
   fragmentation, cDNA synthesis, and fragment enrichment. The libraries
   were then loaded onto the Illumina HiSeq4000 platform for sequencing,
   and the data were analyzed using the Standard Dr. Tom analysis. The
   transcriptome datasets of the current study are available in the
   Sequence Read Archive Database in National Center for Biotechnology
   Information with the accession number (PRJNA1118074).

Histological examination and immunostaining

   Buffered‐formalin‐fixed, paraffin‐embedded liver sections were stained
   with hematoxylin and eosin (H&E) or Sirius red to assess
   histopathological severity and fibrosis. Frozen liver sections were
   subject to Oil Red O staining to evaluate lipid accumulation.

   If necessary, paraffin‐embedded and frozen liver sections underwent
   antigen retrieval using an Abcam heat‐induced epitope retrieval
   protocol. Then, liver sections or cells were stained with DAPI (62248,
   Thermo Fisher Scientific), Hoechst 33342 (62249, Thermo Fisher
   Scientific), Filipin (sc‐205323, Santa Cruz), TUNEL Assay Kit – HRP‐DAB
   (ab206386, Abcam), LC3B (3868T, Cell Signaling Technology), P62
   (sc‐48402, Santa Cruz), Lysotracker Red (C1046, Beyotime), BODIPY™
   493/503 (D3922, Thermo Fisher Scientific), CTSB (sc‐365558, Santa
   Cruz), LAMP1 (sc‐17768, Santa Cruz), Mitochondrial Membrane Potential
   Assay Kit with TMRE (C2001S, Beyotime), PPARα (sc‐398394, Santa Cruz).

   All the images of stained sections or cells were captured using Carl
   Zeiss PALM Inverted Microscope (Carl Zeiss, Oberkochen, Germany) and
   Leica TCS SP8 Confocal Microscope System (Leica Microsystems, Wetzlar,
   Germany).

Fecal genomic DNA extraction and 16SrRNA sequencing

   Feces were collected from mice following DhT treatment and promptly
   stored at −80 °C prior to analysis. Fecal genomic DNA extraction and
   16S rRNA sequencing were conducted in accordance with our previous
   publication.^[49c]

Protein expression and purification

   The coding sequence for the human PPAR𝛼‐ligand binding domain (LBD)
   (GenBank accession number [258]NM_005036.6, containing amino acids
   200–468) was cloned into the expression vector pET28a to produce a
   recombinant protein fused with a N‐terminal HIS6‐EGFP tag. Mutated
   PPAR𝛼 LBD was obtained PCR‐aided site directed mutation (T283A).
   Escherichia coli BL21(DE3) containing the constructed plasmids
   (wild‐type or mutated construct) was cultured in 4 L of luria broth
   medium at 37 °C until reaching an OD[600] value of 0.4. The culture was
   then cooled to 10 °C and induced with 0.1 mM IPTG for 30 hours. Cells
   were harvested by centrifugation at 8000 rpm, resuspended in lysis
   buffer (20 mM Tris pH 8.0, 150 mM NaCl, 5 mM imidazole, 10% glycerol,
   and 10 mM β‐mercaptoethanol), and lysed with a JN‐Mini Low Temperature
   Ultra‐High Pressure Cell Disrupter (JNBio, China). The resulting
   supernatant was collected after centrifugation at 20,000 rpm for 60
   minutes, and then loaded onto a 2 mL Ni2+ column. The column was washed
   with 50 mL of wash buffer 1 (20 mM Tris pH 8.0, 150 mM NaCl, 10 mM
   imidazole, 10% glycerol), 50 mL of wash buffer 2 (20 mM Tris pH 8.0,
   200 mM NaCl, 15 mM imidazole, 10% glycerol), 50 mL of wash buffer 3 (20
   mM Tris pH 8.0, 200 mM NaCl, 15 mM imidazole, 10% glycerol), and 50 mL
   of wash buffer 4 (20 mM Tris pH 8.0, 150 mM NaCl, 20 mM imidazole, 10%
   glycerol), followed by elution of bound protein with buffer 5 (20 mM
   Tris pH 8.0, 150 mM NaCl, 25–500 mM imidazole, 10% glycerol, 10 mM
   β‐mercaptoethanol). The eluted protein was concentrated with an Amicon®
   Stirred Cell (MERCK, Rahway, NJ, USA) and the buffer was changed to
   buffer 5 (20 mM Tris pH 8.0, 150 mM NaCl, 10% glycerol, and 10 mM
   β‐mercaptoethanol). The protein was concentrated to 2 mL and loaded
   onto a Superdex 75 size exclusion column with buffer 6 (20 mM Tris pH
   8.0, 150 mM NaCl, 10% glycerol, and 10 mM DTT). The eluted protein was
   collected, pooled, and concentrated to a final concentration of 10
   mg/mL.

Binding assay

   The binding affinity was evaluated using a Microscale Thermophoresis
   system, Monolith NT.115 (NanoTemper, Watertown, MA, USA). The
   experiments involved a 1:1 mixture of recombinant PPAR𝛼 and DhT
   (dissolved in DMSO) in standard label grade capillaries. A 250 nM
   protein concentration was titrated against a serial dilution of DhT
   (ranging from 0 to 500 μM) to generate 16 samples (using a two‐fold
   dilution) with a total volume of 10 µL. Measurements were conducted at
   25 °C with 20% MST power. The thermophoresis data were analyzed using
   the Kd model assuming 1:1 binding, utilizing NanoTemper's MO Affinity
   V2.3 software.

Luciferase reporter assay

   The promoter regions of PPAR𝛼 (1500 bp from ATG) and catalase (2000 bp
   from ATG) were synthesized by TWSIT Bioscience (South San Francisco,
   CA, USA) and subsequently ligated into the pGL3 Basic Vector (Promega,
   Madison, WI, USA). The wild type full‐length PPAR𝛼 sequence (GenBank
   accession number [259]NM_005036.6) was synthesized by Integrated DNA
   Technologies (San Diego, CA, USA), and Mutated full length PPAR𝛼 was
   obtained PCR‐aided site directed mutation (T283A). Both wild type and
   mutated PPAR𝛼 sequences were inserted into the pcDNA™3.1(+) vector
   (V79020, Thermo Fisher Scientific). Transfection of each of the
   plasmids, along with a control pEGFP‐C1 plasmid (V012024, NovoPro
   Bioscience, Shanghai, China), into HepG2 cells was accomplished using
   Lipofectamine 2000 (11668027, Thermo Fisher Scientific). After
   overnight transfection, the medium was replaced with fresh medium
   containing CHO (10 μM) and/or DhT (1.0 µg/mL). Sample preparation and
   signal detection were conducted following the methodology outlined in a
   previous publication.^[49c]

Determination of CHO and major oxysterols using Gas Chromatography‐Mass
Spectrometry

   Liver tissues collected from ND and HCD mice were homogenized and
   5α‐cholestane (internal standard, C8003, Sigma‐Aldrich) was added. The
   mixture was then extracted three times with isopropanol. The combined
   extraction was subject to N, O‐bis‐trimethylsilyl‐trifluoroacetamide
   with 1% trimethylchlorosilane (BSTFA‐TMCS, T6381, Sigma‐Aldrich)
   derivatization (Valverde‐Som et al., 2018). GC‐MS analysis was
   performed using an Agilent GC‐MS system (7890B, Agilent Technologies,
   Santa Clara, CA, USA) following the procedure outlined in a previous
   publication (ref. [260]62). Quantification was carried out with
   reference to standard chemicals, including Cholesterol (sc‐202539A,
   Santa Cruz); 27‐Hydroxycholesterol (sc‐358756A, Santa Cruz),
   25‐Hydroxycholesterol (sc‐214091, Santa Cruz), 24‐Hydroxycholesterol
   (sc‐489589, Santa Cruz), 7‐Ketocholesterol (sc‐210630, Santa Cruz), and
   7β‐Hydroxy Cholesterol (sc‐210655, Santa Cruz).

Data analysis

   Data from a minimum of three independent experiments were analyzed
   using one‐way ANOVA and Student's t‐test with GraphPad Prism Version
   10.1.0 (264) (GraphPad Software, Boston, MA, USA). The data were
   presented as means ± SD. P values less than 0.05 were considered
   statistically significant.

Conflict of Interest

   The authors declare no conflict of interest.

Author Contributions

   Jia‐Wen Shou: conceptualization, investigation, methodology,
   validation, formal analysis, data curation and writing – original
   draft, and writing – review & editing. Juncai Ma and Byung‐Ho Kang:
   transmission electron microscopy analysis. Xiao‐Xiao Li: Biochemical
   parameter analysis. Shu‐Cheng Chen: conceptualization and discussion.
   Pang‐Chui Shaw: conceptualization, supervision, writing – original
   draft, and writing – review & editing.

Supporting information

   Supporting Information
   [261]ADVS-12-2406191-s001.pdf^ (8.4MB, pdf)

Acknowledgements