ABSTRACT

   Promoting thermogenesis in adipose tissue to enhance energy expenditure
   is widely regarded as a promising strategy for obesity treatment.
   However, the development of effective thermogenic drugs remains
   challenging. Our screenings identified the natural compound Akebia
   Saponin D (ASD) as a potent brown fat thermogenesis activator in mice,
   showing effects through mitochondrial brown fat uncoupling protein 1
   (UCP1)‐dependent pathways. ASD was found to significantly mitigate
   high‐fat diet‐induced obesity and enhance the mitochondrial quality of
   brown adipocytes to promote thermogenesis. Utilizing human protein
   microarrays, cellular thermal shift assay, and drug affinity responsive
   target stability, along with microscale thermophoresis and molecular
   docking analysis, we identified ubiquitin carboxyl‐terminal hydrolase 4
   (USP4) as a direct target of ASD. ASD interacts with USP4 and promotes
   the deubiquitination of peroxisome proliferator‐activated receptor
   gamma, thus inhibiting its proteasomal degradation and enhancing the
   transcriptional activation of UCP1 in brown adipocytes. Additionally,
   USP4 knockdown was shown to attenuate brown fat thermogenesis induced
   by ASD. In summary, our findings demonstrate that ASD promotes brown
   fat thermogenesis by targeting USP4, highlighting its potential as a
   promising natural small molecule for obesity treatment.

   Keywords: Akebia Saponin D, deubiquitination, mitochondria, obesity,
   ubiquitin carboxyl‐terminal hydrolase 4
     __________________________________________________________________

   ASD, a natural plant product screened from small molecule compound
   library, is an antiobesity agent in mice. ASD promotes BAT
   thermogenesis to inhibit body weight gain. USP4 was identified and
   validated as a direct molecular target of ASD. ASD‐induced
   thermogenesis is modulated by a USP4–PPARγ–UCP1 axis.

   graphic file with name MCO2-6-e70420-g005.jpg

1. Introduction

   Obesity refers to the pathological buildup of lipids resulting from an
   imbalance between caloric intake and expenditure over an extended
   period. This condition is often concomitant with various comorbidities
   and metabolic disorders [[50]1]. Over the past three decades, the
   global prevalence of obesity has more than doubled, making its
   treatment a critical public health priority [[51]2]. Thermogenesis in
   adipose tissue can regulate energy metabolism, offering a potential
   therapeutic strategy for obesity [[52]3, [53]4]. The classic
   thermogenesis is the futile cycle mediated by uncoupling protein 1
   (UCP1), which facilitates the release of energy produced by
   mitochondria as heat [[54]5, [55]6, [56]7]. Thermogenic brown adipose
   tissue (BAT) plays a pivotal role in the regulation of body temperature
   in response to cold, primarily through UCP1‐dependent mechanisms
   [[57]8, [58]9].

   Adult individuals possess functional BAT that is capable of
   contributing to thermogenesis; however, its activity is diminished in
   individuals with obesity [[59]10, [60]11, [61]12, [62]13]. In response
   to cold exposure or certain dietary stimuli, the sympathetic nervous
   system releases norepinephrine, which activates β‐3‐adrenergic
   receptors (β3‐AR) in adipose tissue [[63]14]. This activation triggers
   the cyclic adenosine monophosphate signaling pathway, leading to the
   nuclear activation of transcription factors such as cyclic
   AMP‐responsive element‐binding protein 1, histone‐lysine
   N‐methyltransferase PRDM16, cyclic AMP‐dependent transcription factor
   ATF‐2, and peroxisome proliferator‐activated receptor gamma (PPARγ),
   which in turn promote UCP1 transcription [[64]15, [65]16, [66]17,
   [67]18]. Stimulating the browning of white fat and activating BAT
   thermogenesis are considered potential treatments and interventions for
   obesity. However, most of the β3‐AR agonists have been found to have
   side effects, such as cardiovascular risks, due to off‐target effects
   of β3‐AR on other tissues [[68]1, [69]19]. In addition, studies have
   demonstrated that the activation of thermogenesis signals in human
   brown adipocytes is dependent on β‐2‐adrenergic receptors [[70]20].
   Consequently, developing antiobesity drugs that promote adipocyte
   browning and activate BAT thermogenesis remains a significant
   challenge.

   PPARγ, a member of the nuclear receptor superfamily of ligand‐induced
   transcription factors, is predominantly and highly expressed in adipose
   tissue [[71]21]. PPARγ can enhance the expression of particular genes
   within brown adipocytes, inhibit the expression of certain genes in
   white adipocytes, and induce the appearance of beige phenotype in white
   adipocytes [[72]22]. Previous studies have shown that various
   posttranslational modifications, including deacetylation,
   phosphorylation, sumoylation, and ubiquitination, are involved in the
   regulation of PPARγ activity [[73]23, [74]24, [75]25].

   Ubiquitination refers to the process in which ubiquitin (Ub) molecules
   are covalently attached to lysine residues of target proteins through
   the coordinated action of E1 Ub‐activating enzyme, E2 Ub‐conjugating
   enzyme, and E3 Ub ligase. This process facilitates the synthesis of Ub
   chains on target proteins, marking them for recognition and degradation
   by the proteasome [[76]26, [77]27]. Deubiquitinating enzymes (DUBs)
   remove the ubiquitination of proteins by reversing the synthesis of Ub
   chains [[78]28]. Ub carboxyl‐terminal hydrolase 4 (USP4), also known as
   ubiquitous nuclear protein, was initially identified by Gupta et al.
   [[79]29] as a mouse gene associated with the tre oncogene. Previous
   studies have found that USP4 plays a key inhibitory role in
   nonalcoholic fatty liver disease (NAFLD) and related metabolic
   disorders [[80]30].

   We identified a natural small molecule compound Akebia Saponin D (ASD),
   which targets USP4. ASD is a triterpenoid saponin derived from
   traditional Chinese herbal medicine, and can markedly enhance
   thermogenesis in brown adipocytes. In vivo, ASD targets USP4 to promote
   the deubiquitination of PPARγ, activate Ucp1 transcription, and mediate
   BAT thermogenesis. This intervention leads to a marked reduction in
   weight gain and notable alleviation of obesity in mice that are fed a
   high‐fat diet (HFD). Our research aims to explore the effect and
   mechanism of ASD in alleviating the occurrence and development of
   obesity. This study provides new ideas, drugs, and targets for obesity
   prevention and control, laying the foundation for the development of
   innovative antiobesity drugs.

2. Results

2.1. Small Molecule Compound Library Screening Brown Fat Thermogenic Agonists

   To screen out the natural active compounds that can activate the
   thermogenic effect of BAT, we selected 72 commercially available
   natural compounds related to antiobesity or anti‐inflammation from a
   natural product library containing 2542 compounds (Table [81]S1). We
   divided these 72 compounds into groups of 18 and conducted screening
   experiments until the most stable compound was identified for
   subsequent mouse experiments (Figure [82]1A). These compounds were
   tested using UCP1 luciferase reporter gene assays in brown adipocytes
   to assess their ability to stimulate UCP1 expression. Two‐dimensional
   plots were generated based on fold change (FC) and significance
   (Figure [83]1B). Based on this, we further evaluated the mRNA abundance
   of Ucp1 in brown adipocytes (Figure [84]1C). Brown fat organoids were
   generated and used to assess the effects of these compounds on
   mitochondrial function (Figures [85]1D,E and [86]S1A). The red dots in
   the first quadrant signify compounds potentially capable of stimulating
   BAT (p < 0.05, FC > 1.2). We selected five small molecules that
   exhibited the largest FC in a single experiment or were significantly
   altered in at least two experiments, which may be associated with
   activated BAT thermogenesis (Table [87]S2). These compounds were
   administered to brown adipocytes to evaluate their impact on UCP1
   protein abundance (Figure [88]1F). Among them, ASD exhibited the most
   stable effect. Furthermore, no prior studies have reported on the
   regulation of adipose tissue thermogenesis by ASD. Therefore, we chose
   ASD as a suitable compound to further investigate its potential ability
   to induce brown adipocytes thermogenesis and antiobesity.

FIGURE 1.

   FIGURE 1
   [89]Open in a new tab

   Small molecule compound library screening brown fat thermogenic
   agonists. (A) Schematic diagram of the screening process for natural
   small molecule compounds (created with BioRender.com). (B) Ucp1
   promoter activity of brown adipocytes treated with 18 different
   chemical compounds at the dose of 100 µM for 24 h (n = 3). (C) Ucp1
   mRNA levels of brown adipocytes treated with compounds at 100 µM for
   24 h (n = 3). (D) The isolated primary adipose precells were induced
   into globular‐like cells and differentiated into mature adipocytes,
   which were treated with drugs for 24 h (n = 4). (E) High‐content
   imaging scan images. See Table [90]S2 for the name of the compound
   corresponding to the number. Scale bar, 100 µm. (F) UCP1 protein
   concentration in brown adipocytes treated with compounds at 100 µM for
   48 h (n = 3). Statistical significance was defined as *p < 0.05,
   **p < 0.01, and ***p < 0.001. ns, not significant.

2.2. ASD Alleviates Obesity in Mice Induced by a HFD

   We used diet‐induced obese mice (DIO mice) to demonstrate whether ASD
   has an antiobesity effect. ASD showed a dose‐dependent reduction in
   weight gain in C57BL/6 mice, with the effect in the high‐dose group
   comparable to that observed in the metformin (MET) group
   (Figure [91]2A–C). Additionally, ASD treatment led to a reduction in
   the coefficients and volume of inguinal white adipose tissue (iWAT) and
   epididymal white adipose tissue (eWAT). Although there were no
   substantial alterations in BAT size or weight, its tissue color
   exhibited a more pronounced reddish hue following ASD treatment
   (Figure [92]2D,E).

FIGURE 2.

   FIGURE 2
   [93]Open in a new tab

   ASD treats obesity in mice caused by a high‐fat diet. (A) The chart
   shows the treatment of C57BL/6 mice on a high‐fat diet with ASD or MET
   (created with BioRender.com). (B) Weight changes of C57BL/6 mice during
   ASD treatment (n = 8). (C) Representative photo of a C57BL/6 mouse
   after ASD treatment. (D) Weights of fat tissue mass normalized by body
   weight (BW) (n = 8). (E) Representative photos of fat pad morphology.
   (F) Fasting lipid (n = 8). (G) Representative H&E staining of dissected
   tissues. Scale bar, 50 µm. (H) Representative MRI images (n = 3). (I)
   GTT and ITT of DIO mice (n = 8). Statistical significance was defined
   as *p < 0.05, **p < 0.01, and ***p < 0.001. The significance of the
   control group versus the model group was expressed as #p < 0.05,
   ##p < 0.01, and ###p < 0.001. ns, not significant. Significance markers
   match group colors in the line chart.

   Blood biochemical analyses revealed that ASD significantly reduced
   fasting blood glucose (GLU) and lipid levels, particularly total
   cholesterol, triglycerides (TGs), low‐density lipoprotein cholesterol,
   and nonesterified fatty acids. Moreover, ASD substantially decreased
   levels of aspartate aminotransferase and alanine aminotransferase,
   suggesting a protective effect on liver function. However, no
   significant changes were observed in high‐density lipoprotein
   cholesterol levels (Figures [94]2F and [95]S2A).

   Histological analysis showed a reduction in the size of lipid droplets
   within adipocytes and decreased inflammatory cell infiltration after
   ASD treatment (Figures [96]2G and [97]S2B). To assess the contribution
   of fat mass and lean mass to weight loss, we conducted small animal MRI
   scans to measure total body fat in mice. The results indicate that ASD
   leads to a reduction in fat mass, while having no impact on lean mass
   (Figures [98]2H and [99]S2C).

   Obesity and insulin resistance have a mutually causal relationship. In
   order to investigate this, the glucose tolerance test (GTT) and insulin
   tolerance test (ITT) were conducted. Our findings indicate that mice
   with ASD exhibited a dose‐dependent improvement in glucose tolerance
   and insulin sensitivity (Figures [100]2I and [101]S2D). In addition to
   attenuating pancreatic insulin secretion, the treatment for ASD also
   exhibited a decrease in fasting insulin levels and a notable reduction
   in the insulin resistance index (Figure [102]S2F,E). The findings of
   this study indicate that ASD effectively mitigates both obesity and
   insulin resistance.

2.3. ASD Increases Mitochondrial Mass in BAT and Promotes Thermogenesis,
Rather Than Through Exercise

   To demonstrate that ASD can activate thermogenesis in mice, we measured
   24‐h oxygen consumption (VO[2]) and carbon dioxide production (VCO[2])
   using an energy metabolism monitoring system. Energy expenditure (EE)
   and respiratory exchange ratio (RER) were then calculated. In DIO mice,
   ASD administration elevated VO[2], VCO[2], and EE, with nocturnal VO[2]
   exceeding diurnal levels (Figures [103]3A–C and [104]S3A). There was no
   significant difference in RER, suggesting that the substrate the mice
   consumed was not biased toward carbohydrates or fat (Figure [105]S3B).
   The number of activities of the mice during this period was recorded by
   the light beam sensor, and the feed intake was also documented. Our
   results revealed that ASD was associated with increased calorie
   consumption, independent of exercise and diet factors (Figure
   [106]S3C,D). Generally speaking, mice, as homeothermic animals, do not
   show significant differences in body temperature under normal
   temperature conditions. However, in an environment of 4°C, the
   metabolically active mice were able to maintain their body temperature
   more effectively. Therefore, we subsequently exposed the mice to a
   temperature of 4°C in order to assess their ability to adapt to cold
   conditions. Mice administered with ASD exhibited enhanced thermogenic
   capacity and higher body surface temperatures compared with those in
   the HFD group (Figure [107]3D). We examined the mitochondrial mass of
   BAT mice through the use of transmission electron microscopy. The
   consumption of HFD led to a reduction in the quantity of mitochondrial
   cristae, while ASD helped to alleviate this damage (Figure [108]3E).
   Hormone‐sensitive lipase (HSL) is an enzyme that plays a crucial role
   in regulating the hydrolysis of TGs. ASD has the ability to increase
   the abundance of HSL protein depending on the dose‐dependent manner
   (Figure [109]3F). At the same time, the immunohistochemical staining
   intensity of UCP1 in BAT was enhanced (Figure [110]3G). Pathway
   enrichment analysis conducted by means of transcriptomic sequencing of
   BAT revealed that, in contrast to HFD mice, the ASD treat mice
   exhibited a greater number of alterations in genes associated with
   thermogenesis and oxidative phosphorylation (Figure [111]3H). ASD
   significantly upregulated thermogenesis genes in BAT, including Atp5a,
   Uqcrc2, Mtco1, Sdhb, and Ndufb8 (Figure [112]S3E). In addition, to
   further clarify the effect of ASD on brown adipocytes, we isolated
   brown adipose mesenchymal stem cells from BAT; then, induced
   differentiation into mature brown adipocytes. ASD had no obvious toxic
   effect on adipocytes (Figure [113]S3F). However, Ucp1 mRNA expression
   and protein abundance were increased in mature brown adipocytes in a
   dose‐dependent manner (Figures [114]3I and [115]S3G,H). Brown
   adipocytes exhibit smaller lipid droplets and greater mitochondrial
   activity, which is a sign that mitochondria are activated and promote
   fat consumption (Figures [116]3J,K and [117]S3I). We further examined
   the effect of ASD on the expression of genes related to fat synthesis,
   lipolysis and thermogenesis in brown adipocytes. The results showed
   that ASD promoted the mRNA expression of Atgl, Elovl3, Adrb3, Dio2,
   Prdm16, and Cpt1b in brown adipocytes (Figure [118]S3J,K). In short,
   the above results suggest that ASD increases mitochondrial mass in BAT
   and promotes thermogenesis in mice.

FIGURE 3.

   FIGURE 3
   [119]Open in a new tab

   ASD increases mitochondrial mass in BAT and promotes thermogenesis,
   rather than through exercise. (A) Whole‐body VO[2] normalized by body
   weight under basal conditions in DIO mice (n = 8). (B) Whole‐body
   energy exchange normalized by body weight under basal conditions in DIO
   mice (n = 8). (C) Mouse energy expenditure (kJ/h) plotted against body
   weight (kg), and the adjusted R ^2 = 0.622 (left). Adjusted means of
   energy expenditure for experimental groups (n = 8), analyzed using
   ANCOVA with body weight as covariate (right). Labels above the bars
   indicate p values relative to the HFD group. (D) Body surface
   temperature of DIO mice during low temperature exposure for 4 h
   (n = 5). (E) Representative BAT transmission electron microscope image.
   Scale, 2 µm. (F) Western blot analysis of BAT HSL in DIO mice. (G)
   Representative UCP1 immunohistochemical staining in BAT in DIO mice.
   Scale, 50 µm. (H) Enrichment analysis of Kyoto Encyclopedia of Genes
   and Genomes (KEGG) pathway of BAT RNA‐Seq differential genes (ASD 150
   vs. HFD). (I) Ucp1 mRNA relative expression in vehicle control and
   ASD‐treated primary BAT‐derived adipocytes (n = 3). (J) Representative
   phase‐contrast images (left) and oil red o staining (right) of vehicle
   control or 200 µM ASD‐treated primary BAT‐derived adipocytes. Scale,
   50 µm. (K) Fluorescence staining analysis of BODIPY, MitoTracker, and
   PLIN1 was performed in primary BAT‐derived adipocytes treated with
   either vehicle control or 200 µM ASD. Purple, PLIN1; red, MitoTracker;
   green, BODIPY; blue, DAPI for nuclei staining. Scale bar, 20 µm.
   Statistical significance was defined as *p < 0.05, **p < 0.01, and
   ***p < 0.001. The significance of the control group versus the model
   group was expressed as #p < 0.05, ##p < 0.01, and ###p < 0.001. ns, not
   significant. Significance markers match group colors in the line chart.

2.4. ASD Acts Directly on the Target USP4 in Brown Adipocytes

   Studies have reported that proinflammatory factors secreted by
   macrophages can inhibit adipocyte thermogenesis [[120]31]. It has been
   reported that ASD is also associated with anti‐inflammatory effects
   [[121]32, [122]33]. To confirm the anti‐inflammatory effect of ASD, we
   first assessed the cytotoxicity of ASD on RAW264.7 cells, observing no
   discernible toxicity (Figure [123]S4A). Using lipopolysaccharide (LPS)
   and interferon‐gamma (INFγ) to induce an inflammatory response in
   RAW264.7 cells, ASD treatment significantly reduced the expression of
   inflammatory markers Il1β, Il6, and Nos2 (Figure [124]S4B). At the same
   time, we collected the supernatant of RAW264.7 cells and analyzed the
   concentrations of IL6 and TNFα secreted by the cells. It was observed
   that ASD significantly decreased the concentrations of released
   inflammatory factors (Figure [125]S4C). Adipocyte hypoxia‐induced cell
   death occurs when adipose tissue expansion outpaces vascular oxygen
   supply. And the clearance of dead adipocytes depends on macrophages.
   Macrophages phagocytizing dead adipocytes will form characteristic
   crown‐like structures in the tissue [[126]3]. To observe this
   crown‐like structures, we detected CD68‐positive cells in BAT, eWAT,
   and iWAT by immunohistochemical staining. The results indicated that
   the infiltration of macrophages in BAT and eWAT was significantly
   reduced after ASD treatment. Interestingly, both eWAT and iWAT are
   white adipose tissues, but the crown‐like structures in iWAT of mice in
   the HFD group were far fewer than those in eWAT (Figure [127]S4D). Flow
   cytometry demonstrated that ASD diminished the recruitment of
   CD86‐positive cells within BAT (Figure [128]S4E). In a word, ASD
   treatment can alleviate the infiltration of inflammatory cells in the
   adipose tissue of mice.

   To ascertain whether adipocytes or macrophages are the primary target
   cells of ASD, LPS + INFγ‐induced conditioned medium (CMLPS) and
   LPS + INFγ + ASD‐induced conditioned medium (CMLPS + ASD) were
   collected from RAW264.7 cells. Mature brown adipocytes were cultured
   with ASD or the conditioned media, followed by a seahorse assay. The
   results revealed that ASD directly affects brown adipocytes rather than
   influencing thermogenesis through macrophages (Figure [129]4A).

FIGURE 4.

   FIGURE 4
   [130]Open in a new tab

   ASD acts directly on the target USP4 on brown adipocytes. (A)
   LPS + INFγ or LPS + INFγ + ASD were treated for 12 h, and then the
   conditioned medium was changed and collected for 12 h. They were added
   24 h before seahorse testing, and the other two groups were separately
   added to 200 µM ASD or vehicle control (n = 3). (B) Target hook flow
   chart (created with BioRender.com). Scanning fluorescence analysis
   after binding of Bio‐ASD and protein array targets. Freshly prepared
   whole cell lysates were treated with or without ASD, and then digested
   by protease K or heated at 60°C and analyzed by mass spectrometry.
   Finally, obtain the intersection of targets. (C) Chemical structure of
   Bio‐ASD. (D) Representative image of protein array showing positive
   (red arrow) and negative control (blue arrow) spots, as well as spots
   for USP4 (green arrow). (E) Enlarged image of Bio‐ASD binding to USP4
   points on a protein array. The IMean Ratio is demonstrated in the
   figure. (The intergroup ratio between Bio‐ASD group and control group
   was calculated.) (F) The intersection of protein microarray, DARTS, and
   CETSA. DARTS (G) and CETSA (H) assays confirmed the binding of ASD
   (200 µM) to USP4 in BAT. (I) Microscale thermophoresis analysis of ASD
   binding to GFP–USP4 from 293T cell lysates or free GFP. GFP protein was
   the negative control. (J) Representative images of docking of USP4 and
   ASD (binding energy = −50.43 ± 1.66 kcal/mol). (K) Effect of ASD on the
   ability of USP4 to hydrolyze Ub‐Rh110. (L) OCR of primary brown
   adipocytes treated with vehicle control, 200 µM ASD and/ or PR‐619
   (n = 5). (M) RT‐qPCR analysis of gene expression in brown adipocytes
   treated with USP4 siRNA or/and 200 µM ASD (n = 6). Statistical
   significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001.
   ns, not significant.

   In order to discern the precise target of ASD, we employed protein
   microarray, cellular thermal shift assay (CETSA), and drug affinity
   responsive target stability (DARTS) experiments to identify the
   specific proteins that interact with ASD (Figure [131]4B). First,
   biotin was connected to ASD (Figure [132]4C). Next, HuProt proteome
   microarray was used for target screening (Figure [133]4D,E).
   Subsequently, adipose tissue and adipocytes were subjected to DARTS and
   CETSA assays, respectively, and protein identification was accomplished
   through liquid chromatography–tandem mass spectrometry (LC–MS/MS). The
   characteristic peptide of USP4 was identified in both DARTS and CETSA
   experiments, localized in the 545–559 region of amino acid residues
   within the ubiquitin‐like 2 (UBL2) domain (Figure [134]S4F). Finally,
   we combined the results from the three methods and found that USP4 was
   the only protein consistently identified by all approaches
   (Figure [135]4F and Table [136]S3). Therefore, we conclude that the
   direct target of ASD is USP4.

   We also demonstrated by DARTS and CETSA combined Western blot that the
   addition of ASD to cell lysates reduces the degradation of target USP4
   by proteases and temperature (Figure [137]4G,H). Microscale
   thermophoresis experiment showed that the affinity constant of ASD
   binding with USP4 was 37.8 ± 28.0 µM (Figure [138]4I). We used
   Discovery Studio to predict that ASD can combine with USP4 (PDB ID:
   2Y6E) (Figure [139]4J). Furthermore, USP4 activity was measured using
   Ub‐Rh110 as a substrate. Notably, ASD did not markedly influence the
   capacity of USP4 in catalyzing the degradation of Ub‐Rh110
   (Figure [140]4K). Oxygen consumption rate (OCR) is related to the
   activity of cellular mitochondria. To verify the effect of USP4 on
   brown adipocytes OCR, we executed seahorse experiments applying the DUB
   inhibitor PR‐619. The outcomes indicated that PR‐619 can counteract the
   effect of ASD in increasing OCR in brown adipocytes (Figure [141]4L).
   Usp4 small interfering RNA (siRNA) was used to interfere with Usp4
   expression. Initially, we validated the Usp4 siRNA with the best
   knock‐down capabilities on brown adipocytes (Figure [142]S4G). After
   interfering Usp4 expression with siRNA, Ucp1 mRNA expression was also
   downregulated (Figure [143]4M). There was no significant change in Usp4
   mRNA expression in brown adipocytes treated with ASD (Figure [144]S4H).
   However, USP4 protein abundance was increased in a dose‐dependent
   manner (Figure [145]S4I). Therefore, we hypothesized that ASD inhibits
   the degradation of USP4 protein and this was verified by using
   cycloheximide (CHX) (Figure [146]S4J). These results suggest that ASD
   directly acts on USP4, and there is a relationship between USP4 and
   UCP1‐dependent thermogenesis.

2.5. USP4 Enhances the UCP1‐Dependent Thermogenesis Pathway by Promoting
PPARγ Deubiquitination

   We observed that knockdown of USP4 resulted in inhibition of Ucp1 mRNA
   expression. Thus, it is plausible to suggest that USP4 may play a role
   in regulating the ubiquitination levels of upstream transcription
   factors of UCP1. Co‐immunoprecipitation (Co‐IP) with anti‐USP4 antibody
   was performed, followed by LC–MS/MS to identify possible transcription
   factors. In the results, we found the characteristic peptide of PPARγ
   (Figure [147]S5A). Subsequently, western blot was performed on the
   samples obtained after Co‐IP, and the results showed that USP4
   interacts with PPARγ (Figure [148]S5B). The interaction between USP4
   and PPARγ was further substantiated by immunofluorescence
   colocalization analysis of brown adipocytes (Figure [149]5A). Our study
   observed a positive correlation between the expression of USP4 and the
   levels of PPARγ and UCP1 in brown adipocytes treated with ASD.
   Additionally, USP4 levels were directly associated with the abundance
   of carnitine palmitoyltransferase 1B (CPT1B) protein, which facilitates
   the uptake of long‐chain fatty acids into mitochondria, without
   significantly affecting perilipin‐1 (PLIN1) levels. Despite unchanged
   HSL expression, the abundance of phosphorylated HSL increased in
   ASD‐treated brown adipocytes (Figures [150]5B and [151]S5C). PLIN1
   regulates lipolysis and TG levels to protect them from the breakdown of
   HSL [[152]34]. These findings indicate that ASD treatment promotes
   PPARγ expression, favoring UCP1‐dependent thermogenesis rather than
   adipocyte differentiation.

FIGURE 5.

   FIGURE 5
   [153]Open in a new tab

   USP4 enhances the UCP1‐dependent thermogenesis pathway by promoting
   PPARγ deubiquitination. (A) Immunofluorescence analysis shows PPARγ
   (green) interacting with USP4 (red) in brown adipocytes. Cells were
   treated with or without ASD (200 µM) for 48 h. Scale, 20 µm. (B)
   Immunoblots in control and ASD (200 µM)‐treated brown adipocytes. (C)
   Immunoblots of BAT in DIO mice after ASD treatment. (D) Immunoblots of
   BAT in C57BL/6 mice after cold exposure or room temperature. (E)
   Immunoblots of PR‐619 or/and ASD (200 µM)‐treated brown adipocytes. (F)
   Brown adipocytes were treated with or without ASD (200 µM)
   immunoblotting after transfection with USP4 siRNA. (G) Brown adipocytes
   were treated with 25 µg/mL CHX or with CHX and 50 µM MG132. The
   expression of USP4 was detected by Western blot. (H) 293T cells were
   transfected with USP4. After 48 h, the cells were treated with 25 µg/mL
   CHX for different times. The expression of USP4 was detected by Western
   blot. (I) After the 293T cells overexpressed USP4, PPARγ, and Ub for
   48 h, the cells were lysed and processed for immunoprecipitation (IP)
   with anti‐HA magnetic beads. 50 µM of MG132 was added 4 h before
   protein collection. (J) Schematic diagram of multipoint mutation of
   ubiquitin amino acid residues. (K) 293T cells transfected with WT or
   mutated ubiquitin plasmid to screened for PPARγ ubiquitination types.
   50 µM of MG132 was added 4 h before protein collection. Use anti‐HA
   magnetic beads for IP.

   In addition, we observed a positive correlation between USP4 and PPARγ
   in BAT of mice exposed to cold environment and treated with ASD mice
   (Figures [154]5C,D and [155]S5D,E). To verify that ASD regulates PPARγ
   expression through USP4, we used PR‐619 and USP4 siRNA to inhibit or
   knockdown USP4 in brown adipocytes, which decreased PPARγ abundance and
   reversed the upregulation of PPARγ induced by ASD (Figures [156]5E,F
   and [157]S5F,G). Notably, ASD did not affect PPARγ mRNA levels (Figure
   [158]S3K), suggesting that PPARγ regulation may occur via the
   proteasome pathway.

   To further confirm whether PPARγ is degraded via the proteasome
   pathway, we conducted experiments on brown adipocytes using the
   proteasome inhibitor MG132 and the protein synthesis inhibitor CHX.
   When protein synthesis is inhibited, MG132 significantly reduce the
   degradation of PPARγ, indicating that PPARγ degradation is mediated by
   the proteasome pathway (Figures [159]5G and [160]S5H). Additionally,
   overexpression of USP4 in 293T cells potently suppressed PPARγ
   degradation (Figures [161]5H and [162]S5I), suggesting that USP4, as
   the DUB for PPARγ, can reduce its degradation. Subsequent
   ubiquitination experiments further confirmed the ability of USP4 to
   deubiquitinate PPARγ (Figure [163]5I).

   To determine whether PPARγ undergoes ubiquitination and to identify the
   type of ubiquitination modification, we focused on seven lysine
   residues (K6, K11, K27, K29, K33, K48, and K63) involved in forming Ub
   chains. We maintained a specific lysine site on the Ub protein while
   mutating the others with arginine (Figure [164]5J). Our results
   revealed that PPARγ ubiquitination primarily occurs via K48‐linked Ub
   chains (Figure [165]5K). In summary, we found that the mechanism
   underlying ASD‐mediated thermogenesis involves USP4‐mediated
   deubiquitination of PPARγ.

2.6. USP4 is Associated with Obesity; the Effect of ASD is Dependent on USP4

   To investigate whether the antiobesity effect of ASD on mice is
   dependent on USP4, we injected lentivirus expressing USP4 short hairpin
   RNA (Lentivirus‐shUSP4) and then introduced it into the tail vein of
   UCP1‐cre mice to reduce the expression of Usp4 in BAT (Figure
   [166]S6A). The fluorescence signal of GFP was detectable in all four
   groups of BAT isolated at the termination of drug administration
   (Figure [167]S6B). Compared with the control Lentivirus‐GFP, injection
   of Lentivirus‐shUSP4 significantly suppressed USP4 expression in BAT
   (Figure [168]6A). ASD reduced weight gain in Lentivirus‐GFP mice, but
   this effect was completely abolished in Lentivirus‐shUSP4 mice
   (Figures [169]6B and [170]S6C). In line with this, the knockdown of
   USP4 significantly inhibited the reduction in BAT adipocyte area
   induced by ASD (Figures [171]6C and [172]S6D). Furthermore, the
   lipid‐lowering capacity of ASD was markedly impaired following USP4
   knockdown (Figure [173]6D).

FIGURE 6.

   FIGURE 6
   [174]Open in a new tab

   USP4 is associated with obesity, and ASD treatment (150 mg/kg) of
   diet‐induced obesity in mice is dependent on USP4. (A) Usp4 mRNA
   expression in BAT (n = 6). (B) Weight gain in mice during ASD treatment
   (n = 6). (C) Representative H&E images of BAT. (D) Biochemical
   detection of mouse blood (n = 6). (E) Immunoblotting of UCP1, PPARγ,
   and USP4 in BAT. (F) GTT and ITT after ASD treatment (n = 6). (G)
   Association of Usp4 mRNA expression in adipose tissue with
   mitochondrial copy number, body weight, visceral fat weight, and
   HOMA‐IR (male). Statistical significance was defined as *p < 0.05,
   **p < 0.01, and ***p < 0.001. ns, not significant.

   ASD administration significantly upregulates the abundance of Ucp1 and
   PPARγ in BAT of Lentivirus‐GFP mice, but not in BAT of
   Lentivirus‐shUSP4 mice (Figures [175]6E and [176]S6E). We previously
   noted that ASD alleviates insulin resistance in mice, while knockdown
   of USP4 exacerbates insulin resistance and diminishes the therapeutic
   efficacy of ASD in mice (Figure [177]6F). We utilized the Hybrid Mouse
   Diversity Panel (HMDP) database, developed by Professor Lusis, to
   perform correlation analysis [[178]35]. Results demonstrated that USP4
   expression in adipose tissue positively correlated with mitochondrial
   DNA copy number in high‐fat/high‐sugar DIO mice. Additionally, a
   negative correlation was observed between Usp4 expression and body
   weight, visceral fat weight, as well as homeostatic model assessment of
   insulin resistance (HOMA‐IR) levels (Figures [179]6G and [180]S6F).
   Together, these results confirm the important role of USP4 in
   ASD‐mediated thermogenic activation of BAT.

3. Discussion

   The high‐throughput screening of antitumor drugs using tumor organoids
   has become a widely employed approach [[181]36]. We employed the UCP1
   luciferase reporter gene assay and brown fat organoids to conduct a
   screening of a natural compound ASD, which can increase UCP1
   transcription and protein expression in brown adipocytes through a
   UCP1‐dependent pathway. In mice, ASD alleviates HFD‐induced obesity and
   insulin resistance without compromising their activity level. ASD
   effectively activated heat production in BAT and reduced lipid
   deposition in white adipose tissue. HFD is associated with
   mitochondrial damage in BAT [[182]37]. ASD protects mitochondrial
   function from HFD‐induced damage by increasing Ucp1 levels, enhancing
   mitochondrial mass, and upregulating oxidative phosphorylation and
   thermogenic genes in BAT. The above results provide further evidence
   for the thermogenic properties of small molecule ASD screened by brown
   fat organoids. There is still a great difference between the
   physiological and pathological environment of in vitro cultured cells
   and the body. However, the progress in stem cell and organoid
   technology has allowed for the development of three‐dimensional cell
   models that can replicate the histological, molecular, and
   physiological characteristics of certain human organs. This advancement
   will play a crucial role in the screening of potential therapeutic
   compounds for the treatment of various diseases.

   ASD is the main active ingredient of Dipsacus asper Wall, and the
   extract of Dipsacus asper Wall is mainly used in the treatment of
   osteoporosis, waist and leg weakness, fractures, and other joint
   diseases [[183]38, [184]39]. ASD has been reported to ameliorate
   metabolic syndrome, relieve skeletal muscle insulin resistance, and
   treat hepatic steatosis [[185]40, [186]41, [187]42]. However, the
   effects of ASD on adipose tissue thermogenesis have not been previously
   reported. Our study suggests that the metabolic benefits of ASD in
   high‐fat dietary patterns are at least partially mediated by
   ASD‐activated thermogenesis.

   Numerous studies have demonstrated the anti‐inflammatory activity of
   ASD in vitro [[188]33, [189]43]. In RAW264.7 cells, our research also
   revealed that ASD could inhibit the inflammatory responses triggered by
   LPS and INFγ. Although inflammatory cytokine secretion levels were
   significantly reduced in the ASD‐treated group compared with the model
   group, the magnitude of reduction was less pronounced than the
   corresponding mRNA changes. This might be because the process of mRNA
   translating into proteins is related to the efficiency of the promoter,
   the availability of the translation initiation complex, and the
   structure of the mRNA [[190]44]. In addition, the secretory factors in
   the cell supernatant will increase over time, but the mRNA level can
   only reflect the gene expression at this moment. The underlying
   mechanism behind this difference requires further research to be
   confirmed.

   We also noted that ASD could alleviate the inflammatory response in
   adipose tissue, but the infiltration degree of inflammatory cells in
   eWAT was much higher than that in iWAT. According to the literature
   reports, among the two types of white adipose tissues, iWAT belongs to
   subcutaneous adipose tissue while eWAT belongs to visceral adipose
   tissue. The number of inflammatory cells in visceral adipose tissue is
   more abundant than that in subcutaneous adipose tissue, and the
   infiltration of inflammatory cells in visceral adipose tissue is more
   severe under the conditions of obesity or insulin resistance [[191]45].
   Although some studies have explored the involvement of drugs in
   adipocyte thermogenesis via macrophages [[192]46], we have not observed
   such an effect with ASD.

   Targeting methods based on affinity of small molecules and proteins,
   such as chemical crosslinking, photo‐crosslinking, click chemistry,
   biotin labeling, and other technologies, have been widely used
   successfully. However, linking small molecules with probes may alter
   their biological activity or binding specificity [[193]47]. Classical
   small‐molecule nonlabeling techniques, including DARTS and CETSA, have
   low sensitivity and low‐abundance target protein signals tend to be
   masked. This limitation can be optimized in combination with the use of
   LC–MS/MS. The target of our screening is to obtain the target protein
   with stable effect and high repeatability. Therefore, we conducted
   simultaneous biotin‐labeled binding protein microarrays for ASD and
   target screening for DARTS and CETSA binding LC–MS/MS, the only common
   target identified across these methods was USP4.

   Previous studies have found that PPARγ acts as a transcription factor
   capable of binding to UCP1 and other thermogenic gene promoters and
   activating transcription [[194]48, [195]49, [196]50]. We found that ASD
   is functionally regulated primarily by activating the PPARγ–UCP1 axis
   by targeting USP4. In adipocytes, PPARγ is involved in the regulation
   of browning of white fat, adaptive thermogenesis production, and
   adipocyte differentiation [[197]22, [198]51]. The USP4–PPARγ–UCP1 axis
   may be the key signaling pathway for brown adipocytes to exert their
   thermogenic effect. ASD had no significant effect on maker protein
   PLIN1 in adipocyte differentiation. This may be due to the fact that
   PPARγ upregulated by ASD is mainly involved in regulating the output of
   downstream UCP1.

   USP4 is a DUB enzyme whose downregulation is associated with insulin
   resistance, NAFLD, inflammation, and acute injury [[199]30, [200]52,
   [201]53]. In addition, we also found in the HMDP database correlation
   analysis that USP4 expression decreased in adipose tissue of obese
   mice. HMDP database has been successfully used to identify
   metabolism‐related genes [[202]54, [203]55]. We observed that ASD binds
   to the UBL2 region of USP4, which interacts with the DUSP and
   ubiquitin‐like 1 domains and is required for USP4 activation [[204]53].
   Although ASD did not affect the catalytic activity of USP4, the protein
   degradation pathway was inhibited after ASD acted on USP4. The specific
   mechanism needs to be further studied.

   Overall, our study found that ASD‐mediated brown adipose thermogenesis
   is dependent on upregulation of USP4. Cold exposure also upregulates
   USP4, which may be a potential target for obesity treatment.
   Additionally, ASD demonstrates promise for treating insulin resistance
   and obesity. Structural optimization of ASD as a lead compound could
   enhance its biological activity while reducing off‐target effects in
   both animal models and humans. However, our study has several
   limitations. First, we only focused on the potential therapeutic effect
   of USP4 in antiobesity through thermogenesis in brown adipocytes.
   Whether the browning of white adipose tissue also plays a role in this
   process requires further investigation. Second, existing literature
   suggests ASD activates the PPARγ pathway in microglia, promoting
   downstream neurotrophic factor BDNF precursor form (BDNF) expression
   [[205]56]. Another study indicates that BDNF can activate the UCP1
   thermogenic pathway by increasing the noradrenaline content in BAT
   [[206]57]. This suggests ASD might regulate UCP1 expression in BAT
   through neural signaling, though the precise mechanism warrants further
   study. Third, when ASD targets USP4, it does not affect the enzyme
   activity but is related to its expression. To further explore the
   effects of ASD targeting USP4, it is necessary to identify the types of
   epigenetic modifications that USP4 undergoes and generate site
   mutations or protein segments of USP4 for research. Fourth, it is
   necessary to determine whether there is a direct interaction between
   USP4 and PPARγ and then conduct an in vitro ubiquitination reaction to
   test the inhibitory effect of USP4 on the formation process of multiple
   Ub chains of PPARγ. Fifth, drug adverse reactions often correlate with
   off‐target effects. The potential toxicity and side effects of chronic
   ASD administration require evaluation. Furthermore, ASD may engage
   additional targets relevant to this study, which merit attention in
   future research.

4. Materials and Methods

4.1. Mouse Models

   Animal experiments are conducted in accordance with the provisions and
   general recommendations of the China Laboratory Animal Management Law
   and approved by the Ministry of Science and Technology. The C57BL/6J
   (SPF grade) was purchased from Beijing Vital River Company. They were
   kept in room temperature (22–26°C) and humidity controlled room with
   light/dark cycle for 12 h. Standard chow diet and sterilized water are
   given at will. All mice were fed adaptively for 1 week after entering
   the facility. Male mice aged 8–9 weeks were used in all experiments. In
   the study of a specific diet, 8–9‐week‐old C57BL/6J was given a
   high‐fat (60% fat, Research Diet; Cat# D12492) diet for 12 weeks before
   performing other experiments to produce HFD mice. At the same time as
   the HFD was started, the HFD mice were treated with ASD by gavage once
   a day (37.5, 75, and 150 mg/kg), the control and HFD mice were given
   normal saline. Metformin (200 mg/kg) was given in the MET group.

4.2. Brown Adipose Mesenchymal Stem Cells were Isolated and Cultured

   C57BL/6J mice aged 5–6 weeks were taken, after anesthesia, BAT was
   removed from the shoulder area of the back, cut and digested with type
   1 collagenase in 37°C water bath for 60 min. Brown adipose mesenchymal
   stem cells resuspended in medium with DMEM/F12. The cells were filtered
   with 70 µM cell sieve and transferred to a 37°C, 5% CO[2] constant
   temperature cell incubator.

4.3. Brown Adipose Mesenchymal Stem Cells Differentiation

   Brown adipose mesenchymal stem cells were cultured in DMEM/F12
   supplemented with 10% FBS, 5 µg/mL insulin, and 1 nM triiodothyronine
   (T3) (differentiation medium, DM). Cells were cultured in induction
   medium with 1 µM dexamethasone, 125 µM indomethacin, 0.5 mM
   isobutylmethylxanthine, and 100 nM T3 in DM for 2 days. The cells were
   then cultured in DM until day 8, till they exhibited a fully
   differentiated phenotype.

4.4. Spheroid Formation

   Follow the sphere formation method described in the previous article
   [[207]58]. A total of 50,000 cells were inoculated in 100 µL
   endothelial growth medium‐2 (PromoCell; Cat# C‐22011) on ultra‐low
   adherence 96‐well round plate (Corning Life Sciences, USA). The cells
   were maintained in low serum medium for 2 days until spheroid
   formation, transferred to low adsorption 24‐well plate (Thermo
   Scientific, USA), then added to DM and changed every 10 days until
   detection.

4.5. In Vitro Compounds Treatment

   A total of 72 compounds related to antiobesity or anti‐inflammation
   were provided by Selleck Company. Differentiated mature adipocytes were
   treated with candidate compounds (100 µM) for 24–48 h. For siRNA
   transfection, Lipofectamine RNAiMAX (Invitrogen; Cat# 13778100) was
   mixed with siRNA (GenePharma, China) and added to mature fat cells
   according to the instructions. The sequence of siRNA refers to Table
   [208]S4. After 24 h, the adherent cells were treated with ASD (200 µM)
   or vehicle control (0.1% DMSO) for 24–48 h. The inhibitor was
   preincubated with PR‐619 (5 µM) for 24 h and then treated with ASD
   (200 µM) for 48 h. 293T cells were transfected with USP4 plasmid and
   proteins were collected 48 h after transfection with Lipofectamine 3000
   (Invitrogen; Cat# L3000001).

4.6. RT‐qPCR

   Total RNA was extracted from mature adipocytes or mouse BAT using
   TRIzol reagent (Takara, Japan). Hiscript II reverse transcriptase
   (Vazyme; Cat# R323) was used. Real‐time fluorescence quantitative PCR
   system (Thermo Fisher ABI, QuantStudio 5) and SYBRgreen PCR mix
   (Vazyme; Cat# Q712) were used to analyze the expression levels of
   specific genes. 18S, tubulin, or β‐actin were selected as reference
   genes. RT‐qPCR primers are shown in Table [209]S5.

4.7. Western Blot

   Whole cell lysate was prepared using RIPA lysate buffer containing
   complete protease inhibitor cocktail. The protein concentration of cell
   lysate was determined with BCA protein assay reagent. The cell lysate
   was incubated in SDS‐PAGE sample loading buffer at 98°C for 10 min,
   separated by 10% SDS‐PAGE, transferred to polyvinylidene difluoride
   membrane, and incubated with target protein‐specific primary antibody
   and horseradish peroxidase‐coupled secondary antibody. Proteins were
   detected using a fully automated GEL imaging analysis system (Bio‐Rad,
   USA). The sources of the antibodies are shown in Table [210]S6.

4.8. Human Protein Microarray

   The target protein identification of ASD was performed using a human
   protein chip at room temperature. Commercial HuProt proteome chips (Bc
   Bio) were enclosed in PBS containing 3% BSA at room temperature for
   1 h. Incubate the protein chip with 10 µM ASD‐linked biotin (Bio‐ASD)
   in the reaction buffer (PBS containing 1% BSA) at room temperature for
   1 h. The microarray was washed three times with PBS and incubated with
   Cy3‐streptavidin diluted at 1:1000 for 1 h. After rotary drying, the
   interaction between Bio‐ASD and protein was detected at 635 nm by
   GenePix 4000B microarray scanner, and biotin was used as a control.
   GenePixTM Pro v6.0 software was used for data analysis.

4.9. BAT USP4 Knockdown Mice

   Construct a lentivirus‐based conditional, Cre–Lox‐regulated RNAi vector
   for Cre‐dependent termination of shRNA expression [[211]59]. UCP1–Cre
   mice were injected with lentivirus in a tail vein at a dose of
   6 × 10^11 PFU per mouse. After 3 days, the lentivirus‐injected mice
   were given either normal saline or ASD (150 mg/kg) once a day for 90
   days. The lentivirus was supplemented with a dose of 6 × 10^11 PFU at
   45 days. The mice were killed by fasting for 12 h after the last
   injection of ASD or saline (n = 6 per group).

4.10. Statistics Analysis

   Continuous variables are expressed as mean ± SEM. The normality of the
   results distribution was evaluated by the or Kolmogorov–Smirnov test.
   Mann–Whitney (Two groups) or Kruskal–Wallis (multiple groups)
   nonparametric test was used to analyze correlations for samples that
   did not satisfy the normal distribution. For results that satisfy a
   normal distribution; in the case of two groups, the analysis was
   performed using the Student's t‐test. In the case of multiple groups,
   one‐way analysis of variance was performed using the LSD post hoc test.
   Each immunoblot and immunofluorescence was performed at least three
   times. All analyses were performed using SPSS 21.0 software.
   Visualizations were performed using GraphPad Prism 9 software.

Author Contributions

   Lili Gong, Lihong Liu, and Lang Chen conceived and designed the
   project. Lang Chen, Donghai Liu, and Yuxi Li performed the experiments.
   Song Yang, Weihua Jia, Xingbo Wang, and Bing Hu contributed to the
   experiments. Liang Peng, Honglin Liu, Yuchen Wang, Calvin Pan, and
   Aldons J. Lusis provided valuable suggestions to improve the project.
   Lang Chen, Donghai Liu, Lihong Liu, and Lili Gong drafted and/or
   revised the manuscript. Lili Gong and Lihong Liu supervised the
   project. All authors have read and approved the final manuscript.

Ethics Statement

   All animal study protocols were approved by the Ethics Committee of
   China‐Japan Friendship Hospital (no. ZRDWLL230108).

Conflicts of Interest

   The authors declare no conflicts of interest.

Supporting information

   Figure S1: Identification of brown fat cell spheres. (A) Fluorescent
   staining detection of BODIPY and UCP1 in brown fat organoids. Red,
   UCP1; green, BODIPY; blue, DAPI for nuclei staining. Immunofluorescence
   image scale bar, 20 µm. H&E staining scale bar, 50 µm.

   Figure S2: ASD can alleviate insulin resistance in DIO mice. (A)
   Fasting blood biochemical indices after ASD treatment (n = 8). (B)
   Quantification of eWAT and iWAT and BAT lipid droplet area. (C) Fat
   mass and lean mass of C57BL/6 mice by MRI scans (n = 3). (D) Area under
   the curve for GTT and ITT (n = 8). (E) Fasting insulin and HOMA‐IR
   (n = 5). (F) Representative pancreatic insulin staining and
   quantitative. Scale bar, 50 µM. Statistical significance was defined as
   *p < 0.05, **p < 0.01, and ***p < 0.001. The significance of the
   control group versus the model group was expressed as #p < 0.05,
   ##p < 0.01, and ###p < 0.001. ns, not significant.

   Figure S3: ASD activation of thermogenesis was independent of activity
   levels. (A) Whole‐body VCO[2] normalized by body weight under basal
   conditions in DIO mice (n = 8). (B) RER in DIO mice (n = 8). (C) DIO
   mice 24 h activity times (n = 3). (D) The average food intake of each
   cage of DIO mice accumulated to the 10th week (n = 2). (E) Upregulation
   of thermogenic genes in BAT RNA‐Seq ASD 150 group compared with HFD
   group. (F) CCK8 assay of cell viability in ASD‐treated primary brown
   adipocytes (n = 4). (G) Immunoblots of UCP1 in vehicle control and
   ASD‐treated primary brown adipocytes. (H) Quantification of UCP1 in
   vehicle control and ASD‐treated primary brown adipocytes. (I)
   MitoTracker fluorescence intensity in vehicle control and ASD‐treated
   primary brown adipocytes (n = 6). (J) Lipolysis gene mRNA relative
   expression in vehicle control and 200 µM ASD‐treated primary brown
   adipocytes (n = 3). (K) Thermogenic gene mRNA relative expression in
   vehicle control and 200 µM ASD‐treated primary brown adipocytes
   (n = 3). Statistical significance was defined as *p < 0.05, **p < 0.01,
   and ***p < 0.001. The significance of the control group versus the
   model group was expressed as #p < 0.05, ##p < 0.01, and ###p < 0.001.
   ns, not significant.

   Figure S4: ASD reduces inflammatory macrophage infiltration in adipose
   tissue. (A) CCK8 assay of cell viability in ASD‐treated RAW264.7 cells
   (n = 4). (B) Relative expression of inflammatory gene mRNA in RAW264.7
   cells. LPS and INFγ were treated for 12 h, and ASD (200 µM) was added
   to the other group simultaneously (n = 6). (C) The content of
   inflammatory factors secreted into the medium supernatant of RAW264.7
   cells treated in different groups (n = 4). (D) Adipose tissue CD68
   immunohistochemical staining. Scale, 50 µm. (E) The proportion of CD68
   and CD86 positive cells in BAT single cell suspension was detected by
   flow cytometry. (F) Mass spectrum of USP4 characteristic peptide
   detected by DARTS and CETSA. (G) Different siRNA interferes with the
   efficiency of USP4 mRNA expression (n = 3). (H) USP4 mRNA expression in
   brown adipocytes treated with ASD (n = 3). (I) Western blot of USP4 in
   ASD‐treated brown adipocytes. (J) Brown adipocytes were treated with
   25 µg/mL CHX or with CHX and 200 µM ASD. The expression of USP4 was
   detected by Western blot. Statistical significance was defined as
   *p < 0.05, **p < 0.01, and ***p < 0.001. The significance of the
   control group versus the model group was expressed as #p < 0.05,
   ##p < 0.01, and ###p < 0.001. ns, not significant.

   Figure S5: USP4 can interact with PPARγ. (A) After
   co‐immunoprecipitation with USP4 antibody, mass spectrum of PPARγ
   characteristic peptide was detected by LC–MS/MS. (B) The target protein
   was detected by IP with USP4 antibody. (C) Western blot quantification
   of ASD in brown adipocytes with or without drug administration. (D)
   Western blot quantification of BAT protein abundance in DIO mice. (E)
   Western blot quantification of BAT in 8‐week‐old male C57BL/6J mice
   exposed to 25°C or 4°C for 16 h. (F) Western blot quantification of
   PR‐619 or/and ASD‐treated brown adipocytes. (G) Western blot
   quantification of Brown adipocytes were treated with or without ASD
   after transfection with USP4 siRNA. (H) PPARγ protein degradation
   curve. (I) Degradation curve of PPARγ protein during overexpression of
   USP4. Statistical significance was defined as *p < 0.05, **p < 0.01,
   and ***p < 0.001. The significance of the control group versus the
   model group was expressed as #p < 0.05, ##p < 0.01, and ###p < 0.001.

   Figure S6: Knock down USP4 in mice BAT tissue and induce obesity model.
   (A) Workflow of ASD administration in BAT‐specific USP4 knockdown mice
   (created with BioRender.com). (B) Fluorescence imaging was used to
   analyze the fluorescence intensity of BAT in mice. (C) Representative
   photos of mice after USP4 knockdown and ASD treatment. (D)
   Quantification of BAT lipid droplet area. (E)Western blot quantitative
   analysis of mouse BAT after USP4 knockdown. (F) Association of Usp4
   mRNA expression in adipose tissue with mitochondrial copy number, body
   weight, visceral fat weight, and HOMA‐IR (female). Statistical
   significance was defined as *p < 0.05, **p < 0.01, and ***p < 0.001.
   ns, not significant.

   Table S1: 72 compounds related to antiobesity or anti‐inflammation.

   Table S2: Candidate compounds for evaluation of thermogenic potential,
   related to Figure 1.

   Table S3: ASD target candidate proteins from protein microarray, CETSA
   and DARTS intersection relationship analysis, related to Figure 5F.

   Table S4: Sequences list for siRNA.

   Table S5: Primer sequences list for qRT‐PCR.

   Table S6: Antibody resource.
   [212]MCO2-6-e70420-s001.docx^ (15MB, docx)

Acknowledgments