Anti-Obesity Activities of the Compounds from Perilla frutescens var. acuta and Chemical Profiling of the Extract

Perilla frutescens var. acuta (Lamiaceae) is widely used not only as an oil or a spice, but also as a traditional medicine to treat colds, coughs, fever, and indigestion. As an ongoing effort, luteolin-7-O-diglucuronide (1), apigenin-7-O-diglucuronide (2), and rosmarinic acid (3) isolated from P. frutescens var. acuta were investigated for their anti-adipogenic and thermogenic activities in 3T3-L1 cells. Compound 1 exhibited a strong inhibition against adipocyte differentiation by suppressing the expression of Pparg and Cebpa over 52.0% and 45.0%, respectively. Moreover, 2 inhibited the expression of those genes in a dose-dependent manner [Pparg: 41.7% (5 µM), 62.0% (10 µM), and 81.6% (50 µM); Cebpa: 13.8% (5 µM), 18.4% (10 µM), and 37.2% (50 µM)]. On the other hand, the P. frutescens var. acuta water extract showed moderate thermogenic activities. Compounds 1 and 3 also induced thermogenesis in a dose-dependent manner by stimulating the mRNA expressions of Ucp1, Pgc1a, and Prdm16. Moreover, an LC-MS/MS chromatogram of the extract was acquired using UHPLC-MS2 and it was analyzed by feature-based molecular networking (FBMN) and the Progenesis QI software (version 3.0). The chemical profiling of the extract demonstrated that flavonoids and their glycoside derivatives, including those isolated earlier as well as rosmarinic acid, are present in P. frutescens var. acuta.


Introduction
Obesity has arisen as a worldwide concern because it significantly increases the risk of diseases, including type 2 diabetes mellitus (T2-DM), dyslipidemia, hypertension, dementia, and cancers [1].Anti-obesity medications (AOMs), such as orlistat and phentermine, have long been used to treat patients, and recently, glucagon-like peptide 1 (GLP1) agonists have been introduced in the market for obesity treatment [2].Although these medications exhibit a potent efficacy to treat obesity, they still cause several side effects, such as cardiovascular and gastrointestinal symptoms.Therefore, efforts to find different AOM targets are necessary as patients with obesity need long-term treatments.
Testing anti-obesity effects can be divided into two approaches: anti-adipogenesis and adipose thermogenesis.Anti-adipogenic activity refers to the inhibition against the differentiation of preadipocytes into mature adipocytes and lipid accumulation [3].This process is mainly mediated by two transcription factors, peroxisome proliferator-activated receptor γ (PPAR-γ) and CCAAT/enhancer-binding protein α (C/EBP-α).On the other hand, adipose thermogenesis helps to increase energy expenditure by facilitating heat generation from the stored energy, and the transcription factors related to this process include PPAR-γ coactivator 1α (PGC-1α), PR domain-containing 16 (PRDM16), and uncoupling protein 1 (UCP1) [4].
The dried leaves of Perilla frutescens (L.) var.acuta (Thunb.)Kudô ("Perilla herb", "Ja So Yeop" in Korea, Lamiaceae) is used to treat colds, coughs, fever, and indigestion in Asian traditional medicine [5].It has been reported that it shows anti-inflammatory, antibacterial, antifungal, antidepressant, and cytotoxic activities [6].In phytochemical studies, it has been known to have terpenoids, anthocyanins, flavonoids, phenylpropanoids, and fatty acids [7].Thomas et al. showed the anti-adipogenic activity of a P. frutescens var.acuta extract in 3T3-L1 preadipocytes using a high-fat diet mouse model [8], and Choi et al. demonstrated that the P. frutescens var.acuta extract improves the pathology of metabolic syndrome and non-alcoholic fatty acid liver disease in a mouse model [9].However, finding biologically active entities in this plant has not been widely studied to develop novel AOMs.
We have recently reported the anti-inflammatory activities of luteolin-7-O-diglucuronide (1), apigenin-7-O-diglucuronide (2), and rosmarinic acid (3) isolated from P. frutescens var.acuta in LPS-treated RAW 264.7 cells through the NF-κB signaling pathway [10].As a follow-up to our previous study, a P. frutescens var.acuta water extract (PFW) and 1-3 were tested for their anti-adipogenic and thermogenic activities in 3T3-L1 cells, the most commonly used cell line for adipocyte research [11], in the current study.The anti-obesity activities were evaluated by measuring the gene expressions of Pparg, Cebpa, Ucp1, Pgc1a, and Prdm16 compared to that of 18s.We also aimed to systematically investigate the constituents of the PFW extract using UHPLC-MS/MS fragmentation and molecular networking (MN) approaches.

Thermogenesis of the Extract and 1-3
Activating adipose thermogenesis exerts beneficial effects on the body fat content and metabolic health; therefore, the induction of thermogenesis by drugs can be a start to investigate its potential as an obesity treatment.To evaluate thermogenesis induction by the PFW extract and 1-3, fully differentiated 3T3-L1 adipocytes were treated with the PFW extract and 1-3 in addition to CL316,243 (CL).CL is a β3-adrenoreceptor agonist that induces adipose thermogenesis, and we tried to determine the additional induction of thermogenesis by the PFW extract and 1-3 (Figure 2 and Table S3).The extract itself upregulated Ucp1, Pgc1a, and Prdm16 in a dose-dependent manner.Compounds 1 and 2 showed a modest induction on the expression of Ucp1 and Pgc1a.Three concentrations (5, 10, and 50 µM) of 3 showed a dose-response relationship in the stimulation of the gene

Thermogenesis of the Extract and 1-3
Activating adipose thermogenesis exerts beneficial effects on the body fat content and metabolic health; therefore, the induction of thermogenesis by drugs can be a start to investigate its potential as an obesity treatment.To evaluate thermogenesis induction by the PFW extract and 1-3, fully differentiated 3T3-L1 adipocytes were treated with the PFW extract and 1-3 in addition to CL316,243 (CL).CL is a β 3 -adrenoreceptor agonist that induces adipose thermogenesis, and we tried to determine the additional induction of thermogenesis by the PFW extract and 1-3 (Figure 2 and Table S3).The extract itself upregulated Ucp1, Pgc1a, and Prdm16 in a dose-dependent manner.Compounds 1 and 2 showed a modest induction on the expression of Ucp1 and Pgc1a.Three concentrations (5, 10, and 50 µM) of 3 showed a dose-response relationship in the stimulation of the gene expressions (Ucp1, Pgc1a, and Prdm16).In general, CL stimulated the action of the components on the thermogenic gene expressions.expressions (Ucp1, Pgc1a, and Prdm16).In general, CL stimulated the action of the components on the thermogenic gene expressions.

Feature-Based Molecular Networking of the Extract
As demonstrated in Figure 3 and Table 1, 315 mass ions extracted from the Progensis QI (ProQI) software (version 3.0) were organized into a feature-based molecular networking (FBMN), which included 34 single nodes and 10 clusters (nodes ≥2).A total of seven compounds (3 and P1-P6) from the PFW extract were characterized based on the global natural products social molecular networking (GNPS) library, and Clusters 1, 4, 8, and 9

Feature-Based Molecular Networking of the Extract
As demonstrated in Figure 3 and Table 1, 315 mass ions extracted from the Progensis QI (ProQI) software (version 3.0) were organized into a feature-based molecular networking (FBMN), which included 34 single nodes and 10 clusters (nodes ≥2).A total of seven compounds (3 and P1-P6) from the PFW extract were characterized based on the global natural products social molecular networking (GNPS) library, and Clusters 1, 4, 8, and 9 corresponded to flavonoid glycosides, phenylpropanoid glycosides, polyphenols, and flavonoids, respectively.Although P5 (apigenin) and P6 (luteolin) were connected to each other in Cluster 9, they did not show correlations with the nodes in Cluster 1, which represents a flavonoid-related cluster.More details can be found on the GNPS website https:// gnps.ucsd.edu/ProteoSAFe/status.jsp?task=105dc5b8472b4fe39b8b2fccbf4ca55b (accessed on 10 October 2023).The compounds in the orange boxes (3, P2, P5, and P6) were found simultaneously using the FBMN and ProQI approaches.corresponded to flavonoid glycosides, phenylpropanoid glycosides, polyphenols, and flavonoids, respectively.Although P5 (apigenin) and P6 (luteolin) were connected to each other in Cluster 9, they did not show correlations with the nodes in Cluster 1, which represents a flavonoid-related cluster.More details can be found on the GNPS website h ps://gnps.ucsd.edu/ProteoSAFe/status.jsp?task=105dc5b8472b4fe39b8b2fccbf4ca55b (accessed on 10 October 2023).The compounds in the orange boxes (3, P2, P5, and P6) were found simultaneously using the FBMN and ProQI approaches.The identification of potential chemical markers in the PFW extract was also carried out with the retention behavior and mass assignment of the components using ChemSpider and Element Composition in the ProQI software ver.3.0.The candidate compounds The identification of potential chemical markers in the PFW extract was also carried out with the retention behavior and mass assignment of the components using ChemSpider and Element Composition in the ProQI software ver.3.0.The candidate compounds were selected on the basis of matching score (≥40), fragmentation score (30), mass error (<5 ppm), and isotope similarity (≥90), and finally five chemical markers (1, 3, P2, P5, and P6) were identified (Table S4).Although the node "637.104" of Cluster 1 in Figure 3 was connected with the other flavonoid nodes, the FBMN analysis did not identify any known compound to this node.Later, it was identified as compound 1 (luteolin-7-O-diglucuronide/luteolin-7-O-[β-D-glucuronosyl-(1→2)-β-D-glucuronide]) with the assistance of ProQI by considering the fragmented ions "m/z 351.0555" and "m/z 285.0392" at the retention time of 6.07 min (Figure 4).were selected on the basis of matching score (≥40), fragmentation score (30), mass error (<5 ppm), and isotope similarity (≥90), and finally five chemical markers (1, 3, P2, P5, and P6) were identified (Table S4).Although the node "637.104" of Cluster 1 in Figure 3 was connected with the other flavonoid nodes, the FBMN analysis did not identify any known compound to this node.Later, it was identified as compound 1 (luteolin-7-O-diglucuwith the assistance of ProQI by considering the fragmented ions "m/z 351.0555" and "m/z 285.0392" at the retention time of 6.07 min (Figure 4).

Discussion and Conclusions
Phenolic compounds, widely used as food colorants, have received a ention since researchers have shown the relationship of phenolic-rich food intake and the decrease in the incidence of several diseases, such as cardiovascular diseases and cancers [12].Moreover, several clinical studies have been conducted for the anti-obesity activity of phenolic compounds [13].For example, epigallocatechin gallate (EGCG) in green tea and anthocyanins in Hibiscus sabdariffa have shown positive effects on the reduction in body weight, body mass index (BMI), and body fat.
To the best of our knowledge, this study reports for the first time that 1 and 2 suppressed adipocyte differentiation by reducing the mRNA expressions of Pparg and Cebpa.The differentiation of preadipocytes into mature adipocytes plays a key role in the development of obesity [14].PPAR-γ has been known to be indispensable in the differentiation of adipocytes [15,16], and C/EBP-α is also very important in the late stage of adipocyte differentiation.The target genes of PPAR-γ and C/EBP-α are upregulated in this process, and lipids are accumulated in adipocytes [17].As 1 and 2 did not show activity as PPARγ agonists in the previous study [10], we investigated the effects of 1 and 2 on the mRNA expression level of Pparg.Although 3 (50, 100, and 200 µM) suppressed the protein activities of PPAR-γ and C/EBP-α [18], it did not downregulate the mRNA expressions of those transcription factors at concentrations of 5, 10, and 50 µM in this study.While Thomas et al. reported that the extract of P. frutescens var.acuta suppressed the mRNA expressions of Pparg and Cebpa at concentrations of 0.1, 0.2, and 0.4 mg/mL [8], the inhibition of those mRNAs was not observed at concentrations of 10, 50, and 100 µg/mL in this study.
Although P. frutescens (L.) Bri on has been reported to promote thermogenesis through the p38 mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3kinase (PI3K)/Akt pathways, the biologically active components in the extract have never been investigated.The compounds 1 and 3 showed strong thermogenic induction by

Discussion and Conclusions
Phenolic compounds, widely used as food colorants, have received attention since researchers have shown the relationship of phenolic-rich food intake and the decrease in the incidence of several diseases, such as cardiovascular diseases and cancers [12].Moreover, several clinical studies have been conducted for the anti-obesity activity of phenolic compounds [13].For example, epigallocatechin gallate (EGCG) in green tea and anthocyanins in Hibiscus sabdariffa have shown positive effects on the reduction in body weight, body mass index (BMI), and body fat.
To the best of our knowledge, this study reports for the first time that 1 and 2 suppressed adipocyte differentiation by reducing the mRNA expressions of Pparg and Cebpa.The differentiation of preadipocytes into mature adipocytes plays a key role in the development of obesity [14].PPAR-γ has been known to be indispensable in the differentiation of adipocytes [15,16], and C/EBP-α is also very important in the late stage of adipocyte differentiation.The target genes of PPAR-γ and C/EBP-α are upregulated in this process, and lipids are accumulated in adipocytes [17].As 1 and 2 did not show activity as PPAR-γ agonists in the previous study [10], we investigated the effects of 1 and 2 on the mRNA expression level of Pparg.Although 3 (50, 100, and 200 µM) suppressed the protein activities of PPAR-γ and C/EBP-α [18], it did not downregulate the mRNA expressions of those transcription factors at concentrations of 5, 10, and 50 µM in this study.While Thomas et al. reported that the extract of P. frutescens var.acuta suppressed the mRNA expressions of Pparg and Cebpa at concentrations of 0.1, 0.2, and 0.4 mg/mL [8], the inhibition of those mRNAs was not observed at concentrations of 10, 50, and 100 µg/mL in this study.
Although P. frutescens (L.) Britton has been reported to promote thermogenesis through the p38 mitogen-activated protein kinase (MAPK) and phosphatidylinositol 3-kinase (PI3K)/Akt pathways, the biologically active components in the extract have never been investigated.The compounds 1 and 3 showed strong thermogenic induction by upregulating the mRNA expressions of Ucp1, Pgc1a, and Prdm16, while 2 mildly stimulated the thermogenesis of 3T3-L1 cells.The thermogenic activities of 1-3 are reported for the first time in this study.
In addition to the isolation of pure compounds (1-3) from the PFW extract, the chemical profiling of the extract was performed in this study.Tandem mass (MS/MS) spectrometrybased molecular networking (MN) is a tool to identify ingredients in extracts or mixtures.In FBMN, the feature detection and alignment of spectra are applied to the acquired LC-MS/MS data, and then the results are produced as a feature quantification table and MS 2 spectral summary [19].These results are uploaded into the GNPS web platform for MN and compound identification.Compared to the classical GNPS-MN, FBMN discriminates isomers with similar MS/MS data, which are distinguished by the retention time or ion mobility separation.In this study, the FBMN analysis of the PFW extract resulted in the identification of seven compounds, which were mainly apigenin and luteolin derivatives.Although the FBMN results did not assign the accurate structure to each node in the network, the analysis helped to improve spectral annotation and identify the clusters of specific compound types in the plant material.The same LC-MS 2 data were analyzed again using ChemSpider and Elemental Composition in ProQI according to the accurate mass, isotope distribution, and fragmentation pattern.The compounds 3, P2, P5, and P6 were simultaneously identified by the FBMN and ProQI approaches.However, 2 was not found in either analysis.
Several studies confirmed the correlation of obesity with inflammation [20].In an obese status, macrophages accumulated in the white adipose tissue release pro-inflammatory cytokines, including interleukins IL-1β, IL-6, IL-8, and IL-29, and tumor necrosis factor α (TNF-α), thus causing obesity-related inflammation [21].Obesity also changes adipokine secretion.The plasma level of leptin with pro-inflammatory properties tends to increase, and that of adiponectin with anti-inflammatory functions decreases in obese individuals [22,23].In our previous study, we demonstrated that 1-3 have anti-inflammatory activities via the suppression of NF-κB activity and the downregulation of the mRNA expressions of NF-κB target genes, including Il6, Mcp1, and Tnfa [10].In this paper, we also found that compounds 1-3 isolated from P. frutescens var.acuta have anti-obesity effects, indicating that 1-3 would be good candidates for drug development for obesity treatments.However, the underlying mechanisms for their anti-obesity activities need to be elucidated in further studies.

Plant Materials and Sample Preparation
The dried leaves of P. frutescens var.acuta were provided by Megabiosoop in Seoul, Republic of Korea, in April 2019 and a voucher specimen (No.EA387) was stored in the Natural Product Chemistry Laboratory, College of Pharmacy, Ewha Womans University, Seoul, Republic of Korea.
The plant material (2 kg) was extracted with water (20 L) for 15 h at room temperature, and then the solvent was evaporated in vacuo at 40 • C to obtain a concentrated water extract (352.78 g).Then, 5 mg of the PFW extract was transferred to a 1.5 mL vial and dissolved in 1 mL of 70% of LC-MS-grade methanol.

Statistical Analysis
A one-way or two-way analysis of variance (ANOVA) with Duncan's post hoc test was performed to determine significant differences among groups.To determine the antiadipogenic effects of the PFW extract and 1-3, a one-way ANOVA was conducted.To determine the thermogenic effects, a two-way ANOVA was conducted for the effects of the PFW extract or 1-3, CL, and the interaction.The data are presented as the means ± SEM and were analyzed using the SPSS software (Version 26, IBM, Armonk, NY, USA).4.6.UPLC-MS/MS Analysis 4.6.1.UHPLC and MS/MS Conditions LC-MS/MS data acquisition was conducted using a Waters ACQUITY UPLC system (Waters Corporation, Milford, MA, USA) connected to a SYNAPT XS quadrupole time-offlight (Q-TOF) mass spectrometer (Waters Corporation).The separation of the components was performed on a BEH C18 chromatography column (2.1 × 100 mm, 1.7 µM, Waters Corporation).Mobile phase was (A) water with 0.1% formic acid and (B) acetonitrile with 0.1% formic acid, and the gradient elution was as follows: 0-2.5 min, 5% B; 2.5-21 min, 5-100% B; 21-26 min, 100% B; and 26.1-30 min, 5% B. The column oven was set to 40 • C and the autosampler temperature to 15 • C. The flow rate was 0.4 mL/min and injection volume was 1 µL.Comprehensive mass spectra information was acquired using MS E in the negative mode.The parameters of the MS E mode were set as follows: mass range, 100-1200 Da; capillary voltage, 2.2 kV; sampling cone voltage, 50 eV; source offset voltage, 30 eV; source temperature, 120 • C; desolvation temperature, 450 • C; cone gas flow rate, 50 L/h; and desolvation gas flow rate, 800 L/h.Nitrogen and argon were applied as cone and collision gases, respectively.The collision energy was 20-40 eV for high-energy function, and the scan time was 0.5 s.The data were calibrated in real time using a leucine enkephalin solution (m/z 554.2615 [M-H] − ) as an external reference (LockSpray™) at a

Figure 4 .
Figure 4. Example of the component prediction of the extract using ProQI.Compound 1 was identified as luteolin-7-O-diglucuronide by analyzing the fragmentation pa erns of the peak "m/z 637.1031".This peak was divided into two substructures: a flavonoid (m/z 285.0392) and a diglucuronide (m/z 351.0555).

Figure 4 .
Figure 4. Example of the component prediction of the extract using ProQI.Compound 1 was identified as luteolin-7-O-diglucuronide by analyzing the fragmentation patterns of the peak "m/z 637.1031".This peak was divided into two substructures: a flavonoid (m/z 285.0392) and a diglucuronide (m/z 351.0555).

Table 1 .
Compounds identified by the FBMN analysis.

Table 1 .
Compounds identified by the FBMN analysis.