Integration of Lipidomics and Transcriptomics Reveals the Efficacy and Mechanism of Qige Decoction on NAFLD

The prevalence of nonalcoholic fatty liver disease (NAFLD) is increasing as obesity and diabetes become more common. There are no drugs approved for NAFLD yet. Qige decoction (QGD), a traditional Chinese medicine (TCM) formula, is used for NAFLD and hyperlipidemia treatment in TCM and has shown hypolipidemic and hepatoprotective effects. This study tried to interpret the pharmacology and molecular mechanisms of QGD in NAFLD rats. Firstly, the therapeutic effects of QGD on high-fat diet (HFD)-induced NAFLD rats were evaluated. Then, integration of lipidomics and transcriptomics was conducted to explore the possible pathways and targets of QGD against NAFLD. QGD at low dosage (QGL) administration reduced serum total cholesterol (TC), triglyceride (TG), and low-density lipoprotein cholesterol (LDL-C) (P < 0.05). Liver histopathology indicated that QGL could alleviate hepatic steatosis. The main differential lipids (DELs) affected by QGD were glycerolipids. KEGG enrichment analysis suggested that the main pathways by which QGD improved NAFLD may be cholesterol metabolism, glycerolipid metabolism, and insulin resistance. Transcriptome sequencing identified 179 upregulated and 194 downregulated mRNAs after QGD treatment. An interaction network based on DELs and differential genes (DEGs) suggested that QGD inhibited hepatic steatosis mainly by reducing hepatic insulin resistance and triglyceride biosynthesis via the PPP1R3C/SIK1/CRTC2 and PPP1R3C/SIK1/SREBP1 signal axis, respectively. These findings indicated that QGD could protect against NAFLD induced by HFD. The improvement of hepatic insulin resistance and the reduction of triglyceride biosynthesis might be the potential mechanisms.


Introduction
Nonalcoholic fatty liver disease (NAFLD) comprises a wide disease spectrum ranging from steatosis to infammation to fbrosis [1]. NAFLD, regarded as a hepatic characteristic of the metabolic syndrome (MS) [2], has become more prevalent with the increased prevalence of obesity, diabetes, and MS. Te global prevalence of NAFLD is estimated at ∼25% [3], contributing to a growing economic burden [4]. NAFLD-related adverse hepatic outcomes may include cirrhosis, hepatocellular carcinoma, and liver failure. It is predicted to be the primary cause of liver transplants in ten years [5]. In addition, the main extrahepatic-related adverse outcomes are increased cardiovascular disease and malignancy [6].
Te liver is essential for glucose and lipid metabolism. Terefore, the onset and development of NAFLD are closely related to disorders of glucose and lipid metabolism, especially insulin resistance. Hepatic insulin resistance has been shown to promote hepatic de novo lipogenesis (DNL) and gluconeogenesis while decreasing glycogen synthesis, which was referred to as "selective insulin resistance" [7]. Hepatic insulin resistance was thought to direct glucose into lipogenesis pathways and further promote the development of NAFLD [8]. Meanwhile, increased intrahepatic ectopic lipids may promote hepatic insulin resistance [9]. A possible mechanism of lipid-induced hepatic insulin resistance is that hepatic diacylglycerol deposition inhibits IRTK activation while promoting PKC activation [10,11].
Although there has been steady progress in identifying therapeutic targets and promoting drug development. Improving metabolism is a potentially efective method for the treatment of NAFLD with approaches to improving insulin sensitivity, decreasing the delivery of metabolic substrates to the liver and hepatic DNL, or facilitating its safe disposal. For example, obeticholic acid (OCA), an activator of the farnesol X receptor, can improve insulin sensitivity and show antiinfammatory and antifbrotic activity in nonalcoholic steatohepatitis (NASH) patients [12]. Acyl-CoA carboxylase (ACC) inhibitors may potentially ameliorate NASH by reducing DNL [13]. PPARα/δ agonists possess dual activity on both PPARα and PPARδ receptors. Te PPARα component increases fatty acid β oxidation and the PPARδ component attenuates infammation [14]. However, these drugs undergoing research have some limitations. OCA could cause both pruritus and elevated low-density lipoprotein cholesterol (LDL-C) [15]. ACC inhibitors promote hypertriglyceridemia [13]. In short, there are great challenges in drug development, and no agents have been approved yet.
In China, traditional Chinese medicine (TCM) has a history of 5000 years and extensive use. Many TCMs have shown signifcant efcacy in the treatment of NAFLD, Silibinin capsules [16], Hugan Qingzhi tablet [17], etc. Qige decoction (QGD, a TCM formula) consists of a 6 : 2 : 1 ratio of Astragalus mongholicus Bunge (Huangqi), Pueraria montana var. Lobata (Willd.) Maesen and S. M. Almeida ex Sanjappa and Predeep (Gegen), and Citrus × aurantium L. (Chenpi). Te frst two are drug pairs that were frst recorded in "ZhengzhiHuibu" by Li Yongcui (1687 AD). Huangqi-Gegen drug pair possesses a wide use to alleviate NAFLD in clinical practice. Modern pharmaceutical research expressed that it could improve insulin resistance [18]. Chenpi, according to the theory of chief, deputy, assistant, and envoy that is highly respected by TCM formulation, is often used as an assistant medicine to strengthen the therapeutic efect of chief and deputy medicines. Moreover, our previous research showed that Chenpi could alleviate hyperlipidemia mainly through taurine and hypotaurine metabolism, fatty acid biosynthesis, and arginine and proline metabolism pathways [19]. QGD can strengthen the spleen, beneft qi, dispel dampness, and clear heat. It is suitable for NAFLD, which is believed to be due to spleen defciency leading to damp encumbrance in view of TCM pathophysiology. We have found that QGD could ameliorate CCl4-induced liver injury by activating the Nrf2 pathway as an antioxidant [20]. Herein, we tried to interpret the pharmacology of QGD on NAFLD and explore its potential mechanism via the integration of lipidomics and transcriptomics.
Lipidomics focuses on lipid metabolism and has been used to reveal the lipidome changes mediated by disease efects, especially metabolic diseases, such as NAFLD and MS [21]. Terefore, lipidomics was used to identify and quantify the hepatic lipidome to fnd out the changes in lipid metabolites during NAFLD and QGD administration. Transcriptomics can identify thousands of diferentially genes (DEGs) to show the overall gene expression profle during NAFLD and QGD administration. Te metabolite changes were tightly associated with gene transcription regulation. Integrating gene expression profles and metabolic profles was considered a novel approach to understanding the complicated regulatory network of genetic and metabolic pathways [22]. Hence, we also conducted a correlation analysis between lipid metabolites and gene expression to preliminarily explore the targets and pathways of QGD in the treatment of NAFLD.

Drugs Preparation.
Te plant compositions of QGD are shown in Table 1. All medicinal plants of QGD were simmered for 30 min three times after boiling (with distilled water, 1 : 20, 1 : 15, and 1 : 15 V/V, respectively). After fltration, the mixed solution was concentrated to 1 g raw herb/ mL.

Animals and Tissue Collection.
Fifty male Sprague-Dawley rats, VAH/SPF level, weighing 200 ± 20 g, were purchased from Guangdong Medical Laboratory Animal Centre (Guangzhou, China) and raised in an air-conditioned room at 21 ± 0.3°C with a regular 12 h light/dark photoperiod and 60 ± 5% relative humidity. Food and water were freely available. After 7 days of acclimation, all rats were randomly assigned to the control (CON) group, the model (MOD) group, the ATO group (5 mg/kg ATO), the QGD at low dosage (QGL) group (5 g crude drug/kg QGD, equivalent to 1 time the usual clinical therapeutic dose for QGD), and the QGD at high dosage (QGH) group (10 g crude drug/kg QGD, equivalent to 2 times the usual clinical therapeutic dose for QGD) (n � 10). CON rats were fed with a normal diet (Guangzhou, China, crude protein ≥1.8%, calcium 1%-1.8%, crude fber ≤5%, crude Ash ≤8%, total phosphorus 0.6%-1.2%, crude fat ≥4%, sodium chloride 0.5%-0.9%, lysine 0.82%, and water ≤100), while other groups were fed with a HFD (Guangzhou, China, 20% sucrose, 15% lard, 1.2% cholesterol, 0.2% sodium cholate, 10% casein, 0.6% calcium hydrogen phosphate, 0.4% stone powder, 0.4% premix, and 52.2% base feed) for 20 days. Body weight was recorded every day. All rats were anesthetized by intraperitoneal injection of sodium pentobarbital (30 mg/kg) on day 20. Blood was collected from the abdominal aorta and centrifuged at 3000 g for 10 min to get serum for subsequent biochemical analysis. Te livers were removed from each animal. Ten, 200 mg of samples excised from the left lobe of the liver were stored at −80°C for lipidomics (n � 6), while another 200 mg of samples excised from the left lobe were used for transcriptomics (n � 3). Te remaining liver lobes were subjected to hematoxylin-eosin (HE) and oil red O staining. All protocols and studies involving animals have been approved by the Animal Ethics Committee of the Guangzhou University of Chinese Medicine (No. TCMF1-2019011).

HE and Oil
Red O Staining. Liver histology was assessed by HE staining, while lipid accumulation in livers was observed by oil red O staining. Te liver tissue was fxed with 4% paraformaldehyde, then embedded in parafn, cut into slices of approximately 4 µm and stained with HE. Te frozen liver tissues were put in the optimal cutting temperature compound (OCT), cut into 8-10 μm thick, and stained using oil red O solution. Images were obtained under an Olympus BX54 inverted microscope (Tokyo, Japan) at 400× magnifcation.

Sample Preparation.
Since pharmacological results showed that the efects of the QGL group were superior to those of the QGH group, the QGL group was included in further lipidomics and transcriptomic analysis. Six rats in the CON, MOD, and QGL groups were randomly selected and included in the widely-targeted lipidomics analysis.
After thawing on ice, 20 g of liver tissues were homogenized with 1 mL lipid extraction reagent (methyl tertiary-butyl ether: methanol � 3 : 1, V/V, containing internal standard mixture). Samples were mixed on a vortexer for 15 min and 200 uL distilled water was added and then mixed for 1 min. After centrifugation at 12000 g for 10 min at 4°C, the supernatant was transferred and concentrated. Te mixture of 10% acetonitrile and 90% isopropanol with 0.1% formic acid and 10 mmol/L ammonium formate was added to the supernatant for the UPLC-MS/MS analysis. 2.7. Transcriptome Sequencing. Te total RNA of liver tissues was isolated with the RNAiso Plus Reagent (Takara Biotech, Kyoto, Japan). Te degradation and contamination of RNA were monitored by agarose gel electrophoresis. After the cDNA library was constructed, Qubit 2.0 (Life Technologies, CA, USA) was used for initial quantifcation and Agilent 2100 (Agilent Technologies, CA, USA) for quality testing.

UPLC-MS/MS
After quality testing, sequencing was performed on the Illumina HiSeq platform. Clean reads were obtained by fltering raw data, checking sequencing error ratio, and GC level distribution. Sequence alignment was performed with the designated reference genome using hisat2 2.1.0 to obtain mapped data. Te FPKM of each gene was calculated according to the length of the gene and the number of reads mapped to that gene. DEGs were analyzed with the DESeq2 R package (version 1.22.2). Te Kyoto Encyclopedia of Genes and Genomes (KEGG) was used for diferential gene annotation.

Integration of Lipidomics and Transcriptomic Analysis.
Te correlation analysis of the DEGs and diferential lipids (DELs) was carried out via calculating Pearson's correlation coefcient (PCC) using R 1.22.2. Ten the DEGs and DELs with PCC >0.8 were screened. Te correlation coefcient cluster heat map was conducted with R 1.22.2, and the Cytoscape was adopted to show the relationship between DEGs and DELs. Te DEGs and DELs were mapped to the KEGG pathway map. Enrichment analysis was performed to obtain the pathway which a p value < 0.05 and draw a histogram.

Statistical Analysis.
Te results of blood lipid and body weight were analyzed by the SPSS 25.0 software (IBM Corporation, Armonk, NY, USA) and expressed as the mean ± standard deviation. A one-way variance analysis was adopted when both normality and homogeneity were met; otherwise, a nonparametric test was used. Te statistical signifcance was defned as P < 0.05, and P < 0.01 was considered highly signifcant.

QGD Protected Rats against HFD-Induced Hepatic
Steatosis. Body weight showed no signifcant diferences among the groups (P > 0.05) (Figure 1(a)). Serum TC, TG, and LDL-C levels were signifcantly increased in the MOD group after 20 days of HFD administration, while HDL-C was signifcantly decreased (P < 0.01) (Figures 1(b)-1(e)). Compared to the MOD group, both QGL and ATO could lower serum TC, TG, and LDL-C (P < 0.05), while the reductions of these indexes were not signifcant in the QGH group (P > 0.05). Tere were no signifcant diferences in HDL-C levels among the MOD, ATO, QGL, and QGH groups (P > 0.05). HE staining and oil red O staining demonstrated obvious lipid deposition in the livers of the MOD group, while ATO and QGL could signifcantly decrease the size and number of hepatic lipid droplets. However, QGL improved lipid deposition more signifcantly than QGH (Figures 1(f ) and 1(g)).

QGD Modulated Hepatic Lipid Profles.
To analyze the efects of QGD on hepatic lipid profles of NAFLD rats, PCA and OPLS-DA models were adopted to compare the liver samples obtained from the QGL and MOD groups because of the more signifcant improvement of liver histopathology and biochemical parameters in the QGL group than the QGH group. Although PCA analysis presented a partial overlap of the MOD and QGL groups (Figure 2(a)). Te OPLS-DA score plot (Figure 2(b)) showed that the two groups were separated into diferent clusters, which suggested that the hepatic metabolic phenotypes between the MOD and QGL groups were diferent. 16 DELs identifed between the MOD and QGL groups, which suggested QGD mainly afected glycerolipids (13 in total: triglycerides) and fatty acyls (2 in total: acylcarnitine), are shown in Figure 2(c) and details are shown in Table 2. KEGG enrichment (Figure 2(d)) suggested cholesterol metabolism, fat digestion and absorption, glycerolipid metabolism, insulin resistance, regulation of lipolysis in adipocytes, thermogenesis and vitamin digestion and absorption were the main pathways.
Among these pathways, cholesterol metabolism, glycerolipid metabolism, and insulin resistance are closely related to NAFLD.

Hepatic Transcriptomic Expression Alteration.
In transcriptome sequencing, the DEGs were determined with the screening criteria of |log2FC| > 1 and P < 0.05. 373 DEGs (179 upregulated and 194 downregulated) were identifed between the CON and MOD groups. After QGL administration, 262 genes in the MOD group were signifcantly regulated (134 upregulated and 128 downregulated) (Figure 3(a)). A cluster heatmap of 262 DEGs between the MOD and QGL groups is shown in Figure 3(b). An analysis of KEGG gene annotation classifcation revealed that 23 (16.55%) DEGs concentrated on the functional categories were related to metabolic pathway (Figure 3(c)).

Integrated Analysis of Lipidomics and Transcriptomics.
PCC was applied to represent the correlations between DEGs and DELs. Te coexpression network analysis of DEGs and DELs (Figures 4(a) and 4(b)) and the correlation coefcient cluster heat map (Figure 4(c)) based on Pearson's correlation are shown. Te DEGs and DELs between the QGL and MOD groups, such as TG, SIK1, PPP1R3C, CRTC2, PGC-1α, G6PC, PPP1R3G, and SREBP1 were found to be relevant to insulin resistance. While TG, SIK1, PPP1R3C, and SREBP1 were mainly enriched in triglyceride biosynthesis. DEGs involved in insulin resistance and triglyceride biosynthesis pathways are summarized in Table 3. Moreover, according to the KEGG database, DEGs

Discussion
Based on our current knowledge, the primary pathogenic driver of NAFLD is the overwhelmed ability of the liver to process excess carbohydrates and fatty acids, resulting in the   [23]. Tis process is closely related to abnormal lipid metabolism and insulin resistance [24]. It is reported that 69% of hyperlipidemia patients [25] and over more than three-quarters of diabetics sufer from NAFLD [26]. Huangqi-Gegen drug pair and Chenpi, the components of QGD, were found to improve insulin resistance and hyperlipidemia, respectively [19,20]. However, the efects of QGD on NAFLD have not been reported. Herein, we conducted pharmacodynamics to evaluate the efcacy of QGD on NAFLD. Ten, a multiomics analysis via integrating lipidomics and transcriptomics to elucidate the molecular mechanisms. Te results showed that QGD could alleviate hyperlipidemia and hepatic lipid accumulation by improving lipid metabolism and insulin resistance in the early stage of hepatic steatosis. Tus, QGD is of positive signifcance in preventing the progress of NAFLD. We found that QGD ameliorated lipid metabolism disorders at the transcriptional and metabolic levels mainly through insulin resistance and the triglyceride biosynthesis pathway (Figure 4(e)). Insulin modulates hepatic gluconeogenesis-related genes, thus controlling glucose production to ultimately maintain normoglycemia [27]. In the case of insulin resistance, insulin fails to inhibit postprandial hepatic gluconeogenesis, resulting in an increase in  Evidence-Based Complementary and Alternative Medicine 9 blood glucose. Terefore, impaired suppression of gluconeogenesis is widely considered to be a sign of compromised hepatic insulin signaling [28]. Our research indicated that QGD may improve insulin resistance via regulating the expression of CRTC2, SIK1, PPP1R3C, PGC-1α, G6PC, PPP1R3G, and SREBP1, which are involved in gluconeogenesis. CREB-regulated transcription coactivator 2 (CRTC2) is critical in fasting-induced hepatic gluconeogenesis. Te binding of dephosphorylated CRTC2 to phosphorylated cAMP response element-binding protein (CREB) leads to an enhancement of gluconeogenic gene expression, thereby promoting gluconeogenesis [29]. Saltinduced kinase 1 (SIK1) is one of the AMP-activated kinases. PPP1R3C is one of seven genes encoding PP1 regulatory  subunits. PP1's regulatory subunit and catalytic subunit (PP1C) constitute protein phosphatase 1 (PP1), which has a critical function in glucose metabolism. By now, SIK1 and PPP1R3C have emerged as regulators of glucose and lipid metabolism. Overexpression of SIK1 increases the phosphorylation level of CRTC2 at the Ser171 site [30,31]. Silencing PPP1R3C blocked the dephosphorylation of CRTC2 and limited CRTC2 nuclear entry [32], which suppressed its binding to CREB, thereby suppressing downstream genes expression involved in gluconeogenesis, mainly including peroxisome proliferator-activated receptor ccoactivator-1α (PGC-1α), phosphoenolpyruvate carboxykinase (Pck1), and glucose-6-phosphatase (G6PC) [33,34]. PGC-1α, which is a direct target for CREB-CRTC2 complex regulation in vivo, promotes Pck1 and G6PC induction [34]. Pck1 and G6PC are the rate-limiting enzymes of the gluconeogenesis pathway. Tey catalyze the transformations of oxaloacetate and glucose-6-phosphate into phosphoenolpyruvate and glucose, respectively [27]. QGD increased the expression of SIK1 while inhibiting PPP1R3C, resulting in a reduction of the CRTC2 dephosphorylated level. Te transcriptions of PGC-1α and G6PC were subsequently downregulated. As a result, gluconeogenesis and insulin resistance were improved by QGD administration. Terefore, the PPP1R3C/ SIK1/CRTC2 signal axis is supposed to be the target of QGD. Moreover, we found that PPP1R3G, which is directly phosphorylated by AKT to control glycogen synthesis and glucose homeostasis in the insulin signaling pathway, was obviously elevated in the QGD group. Te upregulation of PPP1R3G may be another mechanism to improve insulin sensitivity [35].
Although insulin promotes lipogenesis by increasing sterol-regulatoryelement-binding transcription factor 1 (SREBP1) expression under physiological conditions, hepatic lipogenesis is not inhibited but increased in an insulin resistance state [36]. Tis may be the result of mTORC1 inducing SREBP1 expression under insulin resistance [37]. SREBP1, which promotes the expression of genes involved in the biosynthesis of fatty acids and triglycerides, particularly ACC, fatty acid synthase (FAS), glycerol-3-phosphate acyltransferase (GPAT3), and stearoyl-CoA desaturase (SCD1), is a major regulator of the lipogenic pathway [13,38]. Overexpression of SREBP1 led to increased hepatic lipid accumulation, indicating that SREBP1 played a key role in lipogenic gene activation [39,40]. Compared with the MOD group, QGD signifcantly decreased the expressions of SREBP1 and its downstream target gene GPAT3 in HFD rats. Te improvement of insulin resistance could be the main reason. Another possible mechanism is the transcriptional regulation of SIK1 and PPP1R3C by QGD. In addition to modulating gluconeogenesis and insulin resistance and thus indirectly infuencing SREBP1 activity, both SIK1 and PPP1R3C can directly regulate SREBP1 expression. SIK1 mediates the inhibition of SREBP1 activity. On the contrary, PPP1R3C increases SREBP1 activity [40,41]. QGD could upregulate SIK1 while downregulating PPP1R3C. Tese changes inSIK1, PPP1R3C, and SREBP1 expression levels in HFD rats after QGD intervention contributed to the reduction of hepatic triglyceride biosynthesis.
Moreover, lipidomics and transcriptomics analysis showed that QGD could signifcantly reduce the content of carnitine and increase carnitine palmitoyltransferase 2 (CPT2) activity, compared with the MOD group. Carnitine, which is crucial for the long-chain fatty acids to transfer across the inner mitochondrial membrane, is a critical substance for fatty acid β oxidation. CPT2 converts acylcarnitine esters back to acyl CoA esters for subsequent β oxidation. However, the changes in other genes involved in this pathway were not obvious. Further study is needed to illuminate the efects of QGD on fatty acid oxidation.

Conclusion
In conclusion, the lipidomics results of this research proved QGD inhibited hepatic steatosis mainly by reducing hepatic insulin resistance and triglyceride biosynthesis. Hepatic transcriptomic analysis further revealed that QGD signifcantly lowered the transcription of genes involved in insulin resistance and triglyceride biosynthesis pathways. Te targets of QGD were supposed to be the PPP1R3C/SIK1/ CRTC2 and PPP1R3C/SIK1/SREBP1 signal axis, which still need confrmation and support from further studies. In short, our fndings ofer a therapeutic prospect of QGD against NAFLD with a possible mechanism to improve glycolipid metabolism.

Data Availability
Te data used to support the fndings of this study are included within the article.

Conflicts of Interest
Te authors declare that there are no conficts of interest.