Associated Lipidomics, Metabolomics and Gut Microbiota Changes in CDAA-induced NAFLD Mice After Polyene Phosphatidylcholine Treatment

Non-alcoholic fatty liver disease (NAFLD) is the most common chronic liver disease in most parts of the world. Currently, there is no drug approved for the treatment of NAFLD, and polyene phosphatidylcholine (PPC) is an important drug for clinical doctors to treat patients with NAFLD. Through the analysis of liver index, histopathological inspection and blood routine, it was obvious that PPC could signicantly improve Choline-Decient L-amino acid-dened (CDAA) diet induced NAFLD mice in the present work. We performed lipidomics and metabolomics analysis of 54 samples using ultraperformance liquid chromatography (UPLC) coupled to Thermo LTQ Orbitrap mass spectrometer to select differential metabolites associated with CDAA modeling and PPC treatment. A total of 19 differential metabolites including 5 polar metabolites and 14 lipids were obtained. We inferred that the protective therapeutic effect of PPC on liver was related to the supplement of phosphatidylcholine(cid:0)lysophosphatidylcholine and sphingomyelin(PC, LPC, SM) and acylcarnitine metabolism. In addition, we analyzed the gut microbiota of mice before and after modeling and treatment, signicant differences in the abundance of lactobacillus associated with NAFLD were found. This study provides more reference and data for exploring the pathogenesis of NAFLD and the therapeutic mechanism of PPC, and a methodological reference for the study of the mechanism.


Introduction
Nonalcoholic fatty liver disease (NAFLD) is a metabolic stress-related liver disease de ned as the hepatic accumulation of lipids, mainly triglyceride, in the absence of substantial alcohol consumption (< 20 g/day) or other secondary causes [1,2]. NAFLD has a high incidence of the disease worldwide, is one of the most common types of liver disease. An important characteristic of nonalcoholic steatosis is the accumulation of TG and TC in hepatocytes. Some patients with nonalcoholic fatty liver disease develop NASH and brosis, increasing the risk of cirrhosis and even hepatocellular carcinoma (HCC) [3].
Pathophysiology of NAFLD still has not been completely elucidated. Currently, the "two-hit hypothesis" proposed at the end of the 20th century is one of the most widely accepted model to explain the progression of NAFLD. The " rst hit" mainly refers to the excessive accumulation of fat in liver parenchymal cells. This process has been shown to be related to insulin resistance, which can lead to a dysfunction of intracellular triglyceride synthesis and transport. The "second hit" is an oxidative stress response, which is an in ammatory reaction that occurs in liver cells on the basis of the rst attack [4]. In addition, several recent studies have showed that the gut microbiota is an important factor that should be taken into account when studying NAFLD [5].
Currently, there is no approved drug for NAFLD, and polyene phosphatidylcholine (PPC) is an important drug for clinical doctors to treat patients with NAFLD [6]. The main component of PPC is extracted from soybean. The therapeutic and protective effects of PPC on liver have been reported in many studies [7,8], However, there have been few systematic studies on its mechanism of action, and this study aims to elaborate the mechanism of PPC on non-alcoholic fatty liver from the perspective of metabolomics, lipidomics, and gut microbiota. Choline-de cient, l-amino acid-de ned (CDAA) diet can interfere with fat metabolism in the liver of mice and fat transport from liver tissues to peripheral tissues, leading to excessive fat accumulation in the liver and formation of non-alcoholic fatty liver, which is similar to the pathological state of human NAFLD patients [9,10], Therefore, this model has been widely applied to study the therapeutic effect of drugs on NAFLD [11,12]. In this study, we show that PPC exhibits a good therapeutic effect on CDAA diet-induced NAFLD mice using liver histopathological inspection and serum biochemical indicators, and explored possible mechanisms for the positive effects of PPC to NAFLD mice using lipidomics, metabolomics and gut microbiota analyses.
All healthy male mice were allowed 1 week of acclimatization before onset of the experiments. All the mice were randomly divided into two groups: a group (n = 18) was fed with choline-su cient, l-amino acid-de ned CSAA (control feed, purchased from Nantong Trophic Feed Technology Co., LTD, article number TP-0100-G), another group (n = 38) was fed with CDAA (choline de cient feed, purchased from Nantong Trophic Feed Technology Co., LTD., article number TP-0100-C). After eight weeks of feeding, two mice was chosen from each group to execute and con rm the model of fatty liver has been established successfully by liver section. The group of mice fed with CDAA was divided into two groups (CDAA/model group and PPC group) according to the weight of mice to an average. The mice fed with CSAA (negative control group) and the model group were given 10 mL/kg of normal saline every day, and the PPC group was given the same amount of PPC (15 mg/mL). The mice were weighed once a week and treated with PPC for a total of four weeks. At the end of 8 weeks, mice were fasted for 12 hours and sacri ced by. Livers and intestine were rapidly excised and ash frozen in liquid nitrogen. Blood samples were collected and centrifuged at 14000 rpm for 10 min to obtain sera samples, and all the serum and tissues samples were stored at -80 ℃ until analysis. All animal studies were conducted with the approval and following the guidelines of the Institutional Animal Care and Use Committee of Qingdao.

Histological analysis.
The liver tissue of mouse was immobilized with 4% paraformaldehyde and embedded in para n. The liver was sectioned and stained with hematoxylin and eosin (H&E). To observe the degree of liver brosis, liver sections were stained with picric acid-Sirius red solution. To observe lipid precipitation, liver tissue was frozen in tissue-Tek OCT (Tissue-Tek, Sakura Finetek, USA) and the sections were stained with oil red O reagent. All the histological procedures were performed following the standard procedures as indicated in reagent speci cations. All the images were captured using an optical microscope (Nikon, ECLIPSE 80i).

Biochemical indexes analysis
Commercial kits were used to measure the contents of TG, TC, LDL-C, HDL-C, AST, ALT (Changchun Huili biotech co., Itd, Changchun, China) in mice serum according to the manufacturer's instructions.

Rat liver lipid analysis
A portion of the dissected liver tissues was ground with 9 times anhydrous ethanol. After centrifugation at 4000 g for 10 min, the supernatants were collected. Commercial kits were used to measure the contents of TG and TC (Nanjing Jiancheng Bioengineering Institute, Nanjing, China) in the liver homogenate of rats according to the manufacturer's instructions.
Both positive and negative ionization mode data were collected, and the mass range was 200-1600 m/z. MS and MS/MS were collected at a resolution of 70, 000 and 17, 500, respectively. The electrospray ionization (ESI) conditions were as follows: capillary voltages and capillary Temp were set at 35 V and 300 ℃ in the positive and negative modes for the analysis. Quality control (QC) samples prepared from pooled sera of mice were used to monitor the overall quality of the lipid extraction and mass spectrometry analyses. QC samples were included in batches of analytical samples during the study. The average coe cient of variation of major lipids detected in the QC samples was < 20%. The acquired MS and MS/MS spectral data were analyzed using MSDIAL software for lipid identi cation according to the instructions in the software tutorial [14,15]. The mass tolerance was set at 10 ppm.

Metabolomics analysis
A volume of 150 µL serum was mixed with 450 µL methanol and vortexed for 30 s. After centrifugation at 14000 rpm at 4 ℃ for 15 min, 500 µL of supernatant was added into a 1.5 mL centrifuge tube. The supernatant was dried under vacuum at 4 ℃ vacuum. The residue was dissolved in 100 µL acetonitrile/water (1:1, v/v) and vortexed for 30 s. The samples were centrifuged at 14,000 rpm at 4 ℃ for 15 min, and the supernatant was collected for analysis. The metabolomics analysis was performed using an Agilent 1290 In nity UPLC system coupled to a Thermo LTQ Orbitrap mass spectrometer equipped with a heated electrospray ion source (Thermo Scienti c, CA, USA). Metabolite extracts were separated on a Waters ACQUITY BEH C18 column (2.1 × 50 mm, 1.7 µm) with column temperature maintained at 40 ℃. The mobile phase was water (A) and methanol (B), both containing 0.1% formic acid, the sample was eluted with the following program: 0-1 minutes 2% B, 1-9 minutes 2% B to 98% B, 9-12 minutes 98% B, 12-12.1 minutes 98% B to 2% B, 12.1-15 minutes 2% B. The ow rate was 250 µL/min, the sample injection volume was 8 µL. The mass spectrometer was operated in positive ionization modes. The full scan was collected at a resolution of 60,000. The data were imported to Progenesis QI software for data processing and analysis. In this experiment, compounds with p value < 0.05 and fold change value > 1.5 were considered as differential metabolites. MS/MS spectra were compared with those from online databases (HMDB: http://www.hmdb.cal and METLIN: http://metin.scrippsed) for compound identi cation.

Gut microbiota
With sterile scalpel, the entire intestine was taken out in a sterile state, the outer surface of the intestine was cleaned with sterile water, and the contents of the intestinal segment up to 3-4 cm of cecum were cut

Results And Discussion
3.1 PPC has a good therapeutic effect on CDAA -modeled fatty liver in mice As can be seen from the CDAA diet for up to two months, the weight of mice was signi cantly reduced compared to CSAA group, but the weight of mice was not regained by PPC administration ( Figure 1A).
Liver index (liver weight/body weight) is an important indicator of fatty liver. As shown in Figure 1B, we can clearly nd that compared with the control group, the liver index of the CDAA group is signi cantly increased, and the liver index of the CSAA group is also signi cantly smaller than that of the CDAA group after PPC treatment. The degree of fatty liver lesions can be most intuitively seen through histological sections ( Figure 2).
Accumulation of TC and TG in the liver cells is one of the most important features of NAFLD. After a long time of modeling process, the contents of TG and TC in liver homogenates in the CDAA group were signi cantly higher than those in the negative control group, and were signi cantly reduced after PPC treatment ( Figure 3). Serum transaminase concentration (especially ALT and AST) is also an important indicator of liver injury. Many liver diseases lead to substantial liver damage and abnormal increase of transaminase concentration. We found that the concentration of transaminase in the serum of mice signi cantly increased in CDAA modeling group, indicating a relatively serious injury in the liver of mice, and signi cantly reduced in the PPC treatment group ( Figure 4A and B). HDL and LDL levels also decreased signi cantly after modeling and recovered after treatment ( Figure 4C and D). Choline de ciency diet may lead to defects in lipoprotein secretion [16], and PPC treatment signi cantly reverses these defects. Interestingly, we found the serum TG and TC levels of the model group mice were signi cantly decreased compared with those of the negative control (CSAA) group, and returned in some degree after PPC treatment ( Figure 4E and F). This may be caused by the modeling principle of choline de ciency feeding: when choline is de cient, the content of PC decreases, and the synthesis and secretion of VLDL in the liver slows down [17], resulting in reduced rate of lipid transportation to blood and accumulation of lipids in liver cells, and nally leads to the decrease of serum lipid level. The same result was also shown in the study by Tomokatsu Miyaki et al [18], while the mechanism needs to be con rmed by further experiments.
Numerous experiments have shown that the NAFLD model caused by insu cient choline intake includes a series of processes including steatosis, brosis and cirrhosis, which is very similar to the development process of human NAFLD, and is suitable for human NAFLD study [19]. Histological examination is still the most accurate method for fatty liver diagnosis. Serum biochemical indicators such as (ALT) are also the most commonly used biochemical markers to assist liver function evaluation. According to the results of histological sections and biochemical indexes, after two months of CDAA diet induction, all mice in the model group developed severe fatty liver disease, and some mice also developed obvious brotic lesions.
After one month of PPC administration (150 mg/kg), the mice in the treatment group showed signi cant improvement compared to the model group. These results demonstrated that this dose of PPC has a good therapeutic effect on fatty liver disease in mice. According to previous reports in the literature, PPC mainly achieves the therapeutic effect from three aspects: rstly, it provides the liver with a large amount of high-energy phospholipids, which penetrate into the liver cell membrane in the form of intact molecules, leading to increased cell membrane uidity and liver cell regeneration [20]; secondly, it converts cholesterol into a mobile form, reducing the degeneration and necrosis of liver cells to achieve the effect of fat removal; thirdly, it enhances the uidity and the ability of cell membrane to absorb metabolic cholesterol of HDL, and improves the lipid metabolism of blood and liver [21]. In order to further explore its treatment mechanism, we have conducted research from the perspectives of lipidomics, metabolomics, and gut microbiota.

Effect of PPC on serum lipidome
To understand the effect of CDAA feeding on serum lipidome, lipid extracts of mice sera were analyzed using UPLC-Orbitrap mass spectrometer. Spectral data was analyzed using MSDIAL software, and eight important lipid subclasses were identi ed [15]. A principal component analysis (PCA) was performed for sample clustering, using the Metaboanalyst 4.0 web portal (www.metaboanalyst.ca). As shown in Figure  5, there were clear differences between CSAA and CDAA diet-fed rats serum samples, indicating considerable variation in the serum lipid composition, and PPC treated samples were well separated from CDAA group. Further analysis of the lipid abundance of each lipid class revealed that all lipid classes except phosphatidylethanolamines (PE) and ceramide (Cer) were signi cantly reduced after CDAA diet feeding ( Figure 6A). Interestingly, the level of these lipids in the PPC treated group showed opposite result, all lipids except PE and Cer were signi cantly increased after PPC treatment, compared to CDAA group ( Figure 6B). The alteration in TG level correspond to the results of serum biochemical analysis showing the total triglyceride content in serum of mice was reduced after long-term CDAA diet, meanwhile, the contents of PC and LPC in serum were signi cantly reduced. It is reported that long-term CDAA feeding leads to the decrease of PC and LPC, and the lack of PC, LPC and SM leads to the decrease in lipoprotein synthesis and secretion, resulting in decreased TG transportation from the liver to serum, accumulation of TG in the liver and a decrease of TG in the serum [22], and PPC therapy reverses this effect by supplementing PC. Table 1 and 2 show identi ed differential lipids (absolute value of fold change > 1.5 and p value < 0.05) between CSAA and CDAA group, and between CDAA and PPC group), respectively. Figure 6C and D show lipid species with signi cant differences before and after modeling and treatment. Except for PC 19:2_19:2, other lipids decreased signi cantly after modeling and recovered after treatment ( Figure 6C and D). The CSAA Group is the reference group. "+" sign refers to abundance increase in CDAA Group, while "−" sign refers to abundance decrease in CDAA Group. *The CDAA Group is the reference group. "+" sign refers to abundance increase in PPC Group, while "−" sign refers to abundance decrease in PPC Group. Table 3 and 4 show the differential metabolites absolute value of fold change > 1.5 and p value < 0.05 between CDAA and CSAA group, and between PPC and CDAA Group. It is worth noting that the levels of hexanoylcarnitine, octadecenoylcarnitine and L-carnitine were signi cantly decreased in the CDAA group compared to CSAA group. And after treatment with PPC, the abundances of two acylcarnitines increased signi cantly, while L-carnitine decreased further. Carnitine is known to be an important biological factor in fatty acid oxidation. It is essential for transporting long-chain fatty acids from the cytoplasm to the mitochondria. Acylcarnitines derived from long chain fatty acids and carnitine are transported by ester linkage to the mitochondria, where they are converted into acyl CoA on the inner mitochondrial membrane and serve as a substrate for β-oxidation [23]. L-carnitine is reported to have adjuvant therapeutic effects on fatty liver disease and insulin resistance [24][25][26]. PPC has been proved in vitro to improve the oxidative substrates of mitochondria, restore the respiratory chain activity stimulated by ADP, and improve the activity of mitochondrial cytochrome oxidase [27]. Katz et al. reported PPC can prevent oxidative phosphorylation of mitochondria and changes in mitochondrial skeleton and loss of mitochondrial cristae, and inhibit the activities of caspase-3 and caspase-9, thereby inhibiting mitochondrial apoptosis [28]. We hypothesize that PPC might alleviate the appearance of insulin resistance through the protection of mitochondria, and it might reduce the degree of fatty liver disease. According to the results, PPC might improve the absorption and utilization of L-carnitine by cells. In the PPC treatment group, L-carnitine level was further decreased but acylcarnitine levels increased, compared to untreated group. PPC and L-carnitine might have a synergistic effect, which needs to be veri ed by further experiments.

Effect of PPC on gut microbiota
Abnormal changes of gut microbiota are closely related to NAFLD [5] . Therefore, in order to better explore the therapeutic mechanism of PPC on NAFLD, we further conducted gut microbiota analysis in mice. Figure 7 shows the overall difference in gut microbiota of mice in the CSAA group, CDAA group and PPC group. Studies have found that a signi cant feature of NAFLD patients on a high-fat diet was an increase in Firmicutes and a decrease in Bacteroidetes, this could be due to different energy residues in the faeces [29,30]. Another reason is that the abnormal bile acid level caused by high-fat diet changes the intestinal pH, and Firmicutes and Bacteroidetes have different adaptability to the environmental pH [31]. Different from the high-fat diet, we chose choline-de cient diet to build the NAFLD model. The weight change of mice was opposite to NAFLD patients on a high-fat diet. It is preliminarily speculated that Firmicutes/Bacteroidetes are closely related to the energy in feces. Top ten genus that have signi cant differences among CSAA group, CDAA group and PPC group are shown in Figure 8. The most notable genus is Lactobacillus. Studies have shown that Lactobacillus is an important probiotic, which is very bene cial for maintaining health [32,33]. Jinchi Jiang et al. isolated two Lactobacillus species from Chinese super-long-lived populations, and found they could regulate lipid metabolism in hypercholesterolemic rat models [34]. Experiment indicated that 20 adult patients with histologyproven NASH were randomly allocated to receive a probiotic formula containing Lactobacillus plantarum, Lactobacillus delbrueckii, Lactobacillus acido philus, Lactobacillus rhamnosus and Bi dobacterium bi dum, the authors found patients who had received this formula had reduced intrahepatic triglyceride content [35]. The development of NAFLD is also associated with the production of alcohol by some intestinal bacteria. A study showed that the blood and respiratory levels of ethanol in NAFLD mice were signi cantly higher than those in normal mice, and the activation of AMPK by Lactobacillus rhamnoides GG strain attenuated the accumulation of fat in the liver caused by alcohol [36]. In conclusion, studies have proved that Lactobacillus abundance can improve the intestinal barrier, reduce LPS levels in portal venous blood, attenuate in ammation, and inhibit fatty acid accumulation in the liver. In our study, The amount of Lactobacillus in CDAA group was signi cantly reduced compared with CSAA group, and after PPC treatment, Lactobacillus was signi cantly improved. Therefore, we speculate that PPC improves NAFLD possibly by restoring the abundance of Lactobacilli in NAFLD mice.

Conclusion
Non-alcoholic fatty liver disease (NAFLD) is a complex disease arising from both genetic and environmental factors. Our study showed that the choline de cient diet could induce mice to develop severe NAFLD and even NASH. Lipidomics, metabolomics and gut microbiota analyses combined with histopathological examination and blood routine examination had been employed to study the protective effect of PPC against CDAA-induced NAFLD mice and its possible mechanism. The content of major lipids in CDAA-induced NAFLD mice signi cantly changed compared with that in normal mice, and PPC treatment improved these lipid abnormalities to a certain extent, especially the lipids such as PC, LPC and SM that are associated with the synthesis of VLDL. Five metabolites were identi ed to have signi cant changes before and after modeling and treatment. The therapeutic effect of PPC on NAFLD might be related to acylcarnitine metabolism. In addition, the gut microbiota of the three groups of mice also showed signi cant differences. Further study is needed to elucidate the mechanism of PPC treatment on NAFLD. Our work studied the effect of PPC on NAFLD treatment in vivo from the perspective of lipidomics, metabolomics and gut microbiota, and provided experimental evidence for the study of PPC mechanism.

Declarations
The body weight of mice at week 12 (A), the liver index (liver wet weight/body weight ratio) at the end of week 12 (B), *p < 0.05, ** p < 0.01 compared with negative control (CSAA diet) group; #p < 0.05, ##p < 0.01 compared with model (CDAA diet) group.

Figure 2
Section with H&E staining (A), section with oil red O staining (B) and section with Sirius red staining (C) of mouse liver tissue.     Fold changes of serum lipid classes before and after CDAA diet feeding (A); fold changes of serum lipid classes before and after treatment of PPC (B); fold changes of serum lipid species before and after CDAA diet feeding (C); fold changes of serum lipid species before and after treatment of PPC (D).

Figure 7
Average phylum distribution of gut microbiomes of CDAA group, CSAA group and PPC group. Figure 8