Protocatechuic Acid as a Phenolic Intermediate to Ameliorate Non-alcoholic Fatty Liver Disease by Inhibiting Enterococcus Faecalis in Mice

Jijun Tan Hunan Agricultural University Ruizhi Hu Hunan Agricultural University Baizhen Li Hunan Agricultural University Yanli Li Hunan Agricultural University Xizi Yang Hunan Agricultural University Ziyu He Hunan Agricultural University Hongfu Zhang Chinese Academy of Agricultural Sciences De-Xing Hou Kagoshima University: Kagoshima Daigaku Jianhua He Hunan Agricultural University Shusong Wu (  wush688@hunau.edu.cn ) Hunan Agricultural University https://orcid.org/0000-0003-3532-1165


Introduction
The occurrence and development mechanism of non-alcoholic fatty liver disease (NAFLD) are complicated, although primarily based on the correlations between genetic and nutritional factors including lipids metabolism disorder, insulin resistance and intestinal microecological imbalance [1] . Since metabolic syndromes are popular in individuals with over-nutrition, gut microbiota are considered to play an important role in NAFLD, such as short-chain fatty acids (SCFAs)-producing bacteria [2,3] and microbiota belonging to Lactobacillales [4] . Recent studies have suggested that Enterococcus faecalis (E. faecalis), an opportunistic pathogen belonging to Lactobacillales, has the potential to induce liver steatosis since it can secrete cytolysin [5] , and has capability of metabolizing trimethyllysine into N,N,N-trimethyl-5-aminovaleric acid, which can reduce carnitine synthesis to decrease oxidation of fatty acids [6] . Considering that, changes of gut microbiota become the prime concern in understanding the pathogenesis and prevention of NAFLD.
Polyphenols offer a great potential as an alternative therapy for NAFLD due to their biological functions such as antioxidant [7] , anti-in ammatory [8] , and anti-bacterial [9] in recent years. Our previous study has suggested that cyanidin 3-glucoside (C3G), one of the most common anthocyanins in plant, has the potential to ameliorate in ammation and oxidative stress in NAFLD mice [10] , by modulating gut microbiota [11] . However, after ingestion, C3G can be rapidly degraded into phenolic acids, which may take the primary responsibility for the biological function [12] , and protocatechuic acid (PCA), phloroglucinaldehyde (PGA), vanillic acid (VA), ferulic acid (FA) and their derivates have been proved as the major phenolic metabolites of C3G in blood circulation [13,14] . Based on the antioxidant and anti-in ammatory activities, PCA, PGA, VA and FA are suggested to be the potential bioactive phenolic metabolites of C3G in our recent study [15] . Therefore, this study aimed to investigate the protective effects of C3G, PCA, PGA, VA and FA against NAFLD, and challenged to clarify the protective mechanisms of PCA, a typical bioactive metabolite, focusing on regulation of gut microbiota in a high fat diet-induced mouse model. and light (12 h light/day), and had free access to feed and water. In the rst experiment, twenty-eight C57BL/6J mice (SPF class, male, 6 weeks of age) were randomly allocated into 7 groups (n = 4) after acclimatization for 1 week. Mice were then fed a LFD, a HFD, or a HFD containing 0.4% (w/w) of C3G, PCA, PGA, VA, or FA for 12 weeks, respectively. The dosages of C3G and its phenolic metabolites were based on our previous study [10] . In the second experiment, twenty-four C57BL/6J mice (SPF class, male, 4 weeks of age) were randomly divided into 4 groups (n = 6) after acclimatization for 1 week, and then fed a LFD, a HFD, or a HFD containing 0.025% or 0.1% (w/w) of PCA for 12 weeks, respectively.

Bacteriostatic experiment
E. faecalis was cultured with brian heart infusion (BHI) medium and identi ed by comparing with sequence of E. faecalis ATCC 29212 on National Center for Biotechnology Information (NCBI) (Supplemental Fig. 2). The bacteriostatic effect of PCA (0.355-45.455 mM) was tested by detecting the optical density (OD) value under 600 nm after 24 h incubation. The calculation of Transmittance accords to the following formula: The minimum inhibitory concentration (MIC) depends on all value of Transmittance > 95 (only valid when negative controls were turbid) [16,17] . After that, "x" was converted to "log10(x)" to achieve tting curve between Concentration (x, mM) and Transmittance (y, %). And the theoretical MIC was calculated by assignment "y" is "95".

Fecal microbiota transplantation
Fecal microbiota transplantation (FMT) experiment was conducted based on a previous study [18] . Brie y, twelve C57BL/6J mice (SPF class, male, 4 weeks of age) were fed a LFD for 8 weeks and used as donor mice of normal gut microbiota. Meanwhile, another twelve C57BL/6J mice (SPF class, male, 4 weeks of age) were accommodated for 3 days, and then had free access to LFD and antibiotic cocktails containing vancomycin (0.5 g/L), neomycin sulfate (1 g/L), metronidazole (1 g/L) and ampicillin (1 g/L) in water for consecutive 3 weeks. Next the mice were randomly divided into 2 groups (n = 6), and administered with the fecal microbiota of donor mice or E. faecalis (verify information was shown in Supplemental Fig. 2), respectively. Feces from donor mice were collected sterilely every day and resuspended in PBS at 0.125 g/mL, followed by low-speed centrifugation (800 × g) for 10 min. A volume of 0.15 mL of the supernatant or the third generation of BHIcultured E. faecalis (1.75×10 11 CFU/mL) was administered to mice by oral gavage once a day for consecutive 7 days, and then fed with LFD for another 11 weeks.

Histomorphological analysis of fat and liver tissue
Intraperitoneal fat and liver tissue samples were xed in 4% paraformaldehyde or fat xative (G1119, Servicebio Technology Co., Ltd. Wuhan, Hubei, China), respectively for at least 24 h until slice production.

Measurement of indicators of oxidative stress in liver
Liver samples of mice were immediately collected into sterile tubes after weighing and frozen in liquid nitrogen before store at -80 ∘ C until use. The level of malondialdehyde (MDA) was measured with an assay kit (Nanjing Jiancheng Bioengineering Research Institute Co., Ltd., Nanjing, Jiangsu, China) according to the manufacturer's manual.

Analysis of gut microbiota
Feces were collected at the rst day and nal day of experiment, and recorded as Feces 0W and Feces 12W .
Cecal contents of mice were collected after sacri ce and recorded as Cecal contents 12W

Lipidomics
Liver samples (50 mg) was homogenized with 1mL lysis buffer including methanol, methyl tert-butyl ether (MTBE) and internal standard mixture (MTBE: methanol = 3:1) for 2 min. The homogenate was then mixed with 500 µL of pure water for 1 min before centrifuging at 10,000 g for 10 min at 4 ∘ C. The supernatant (500 µL) was concentrated and redissolved in 100 µL mobile phase B solution. The sample extracts were analyzed by using a LC-ESI-MS/MS system (UPLC, Shim-pack UFLC SHIMADZU CBM30A system; MS, QTRAP® 4500 System). The analytical condition was as follows, column, Thermo C30 (2.

Statistical analysis
Results are expressed as means ± SD. Signi cant differences between groups were determined using oneway analysis of variance (ANOVA) tests, followed by Fisher's least signi cant difference (LSD) and Duncan's Multiple Range test (SPSS21, IBM Corp., Armonk, NY, USA). Correlation of gut microbiota, serum indicators and hepatic metabolites was determined using Pearson's correlation analysis. A probability of P < 0.05 was considered signi cant.

Results
3.1 C3G, PCA and PGA showed protective effect against NAFLD in mice As shown in Figure 1A, HFD induced a signi cant increase in the body weight (BW) of mice (P < 0.05 vs LFD group). Dietary supplementation of C3G, PCA or PGA, but not VA or FA, decreased the BW of HFD-fed mice.
Histological analysis of liver showed that HFD led to diffuse bullous lipids accumulation around hepatic vein, which was attenuated by supplementation of C3G, PCA or PGA ( Figure 1B). Correspondingly, C3G decreased serum ALT level ( Figure 1C), while PCA and PGA reduced AST ( Figure 1D) and ALT levels in HFD-fed mice (P < 0.05). Further analysis on the serum indicators revealed that HFD signi cantly increased the level of glucose ( Figure 1E) and decreased the ratio of HDL-c to LDL-c (P < 0.05) ( Figure 1F), but were recovered by supplementation of C3G, PCA or PGA.
In ammatory cytokines were then measured in serum to evaluate the in ammation induced by HFD. As shown in Figure 1G showed that PCA signi cantly down-regulated the lipid metabolites in the pathways associated with insulin resistance (Supplemental Figure 1C).

PCA alleviated NAFLD potentially by down-regulating the relative abundance of Enterococcus
Gut microbiota has been considered as the key factor that affects development of NAFLD. Thus, gut microbiota was then been characterized by 16S rDNA gene sequencing. At the phylum level (Table 1), HFD increased the relative abundance of Firmicutes both in feces and cecal contents (P < 0.05), but decreased the relative abundance of Bacteroidetes, which led to a signi cant increase in Firmicutes/Bacteroidetes ratio (P < 0.05). Supplementation of 0.1% PCA restored the ratio of Firmicutes to Bacteroidetes (P < 0.05). Further analysis at the genus level (Table 2) revealed that supplementation of PCA dose-dependently decreased the relative abundance of Roseburia, Intestinibacter and Enterococcus, which belong to Firmucutes.
Based on all data of each mouse, the correlation between cecal microbiota, serum indicators and hepatic lipid metabolites were analyzed by Pearson's correlation analysis. As shown in Figure 4A

PCA showed a direct inhibitory effect against E. faecalis in vitro
Considering PCA dose-dependently decreased the relative abundance of Enterococcus, the bacteriostasis effect of PCA against E. faecalis was further evaluated in vitro. The results indicated that a concentration of 2 M PCA showed obvious bacteriostatic circle agaisnt E. faecalis (Supplemental Figure 2E). MIC experiment showed that a concentration of 22.727 mM PCA inhibited the growth of E. faecalis with the transmittance > 95% ( Figure 4A). To access the accurate MIC of PCA against E. faecalis, a tting curve between Concentration (x, mM) and Transmittance (y, %) was designed by converting "x" to "log10(x)", and the theoretical MIC of PCA against E. faecalis is 12.457 mM based on the tting curve: Log10(x PCA )=0.8647-(Log10(97.942/(y-2.458)-1))/5.348 (R 2 =0.933) ( Figure 4B).

Transplantation of E. faecalis promoted NAFLD in mice
As Enterococcus showed signi cant correlation with 12 kinds of changed hepatic lipid metabolites in experiment 2. A FMT experiment was further designed to clarify the effect of E. faecalis in the pathogenesis of NAFLD ( Figure 4C). As shown in Figure 4, transplantation of E. faecalis (rE. faecalis) increased the body weight (E) and intraperitoneal fat deposition (D&F) as compared with control group (rLFD), with a P value of 0.072 and 0.092, respectively. Meanwhile, rE. Faecalis signi cantly increase the liver weight (G) and induced obvious diffuse bullous lipids accumulation (H).

Discussion
Excessive energy intake [20] and energy imbalance [21] are considered as the primary factor causing overweight and obesity. Insulin resistance and abnormal lipids metabolism play an important role in the onset of NAFLD, typically characterized as high serum triglyceride, high LDL-c and low HDL-c levels [22] . As the rst-hit to induce NAFLD, insulin resistance in uenced by diets results in decrease of sugar utilization for tissues, together with the decreased sensitivity to insulin, which conversely aggravates the utilization of lipids for cells and a large amount of insulin accumulated as hyperinsulinemia [23] . Lipid accumulation in the liver is the main cause of cytotoxicity to hepatocytes, which induces hepatocyte injury and secondary in ammation/ brosis during the disease [24] . In this study, HFD caused lipid metabolism disorder and insulin resistance (HOMA-IR) as re ected by the lower HDL-c/LDL-c ratio and higher levels of both glucose and insulin, accompanied by hepatic lipid accumulation and in ammation. Supplementation of C3G and its main phenolic metabolites PCA and PGA showed protective effect against NAFLD in mice, in particular, PCA dosedependently ameliorated hepatic steatohepatitis and in ammation by attenuating insulin resistance, as re ected by lower levels of AST, ALT, MDA, TNF and HOMA-IR index, but higher HDL-c/LDL-c ratio.
Increased Firmicutes/Bacteroidetes ratio was reported to have a close correlation with obesity [25] , and our results indicated that the Firmicutes/Bacteroidetes ratio was increased by HFD but decreased by PCA. At the genus level, PCA decreased the relative abundance of Roseburia, Intestinibacter and Enterococcus that belonging to Firmicutes, and the correlation analysis revealed that Roseburia, Intestinibacter and Enterococcus had a positive correlation with serum indicators of hepatic damage (AST & ALT). Intestinibacter was reported to be involved in mucus degradation and metabolic disorder in type-2 diabetes [26] , and Enterococcus was considered as one of the major causative agents of infection in HFD-induced NAFLD [27] .
Enterococcus can secret cytolysin to cause hepatocyte injury [5] , and metabolize trimethyllysine into N,N,Ntrimethyl-5-aminovaleric acid so that reducing oxidation of fatty acids [6] . Roseburia is one of the short-chain fatty acid (SCFA)-producing bacteria, and closely associated with metabolic syndrome [28,29] , which suggesting that dysbiosis of gut microbiota may cause disorder of SCFA production in NAFLD [30] . Other studies have also indicated that NAFLD patients have higher SCFA levels along with SCFA-producing bacteria [3] . Therefore, PCA might decrease the relative abundance of harmful bacteria in eutrophic state, such as Intestinibacter and Enterococcus, to maintain microecological balance and energy balance, and decrease the relative abundances of speci c SCFA-producing bacteria like Roseburia to reduce the production of SCFAs to suppress lipogenesis and gluconeogenesis [30] .
The metabolism of hepatic neutral lipid, mainly triacylglycerol, is critical in the development of NAFLD [31] . In this study, lipidomics analysis revealed that HFD signi cantly up-regulated 120 kinds of lipids metabolites, while 41 kinds of lipids metabolites were down-regulated (Supplemental Figure 1B), which suggesting that HFD-induced disorder of lipid metabolism involved in both up-regulation and down-regulation of lipids. Most of the HFD up-regulated lipid metabolites belong to triradylglycerols (data are not shown), which can explain the lipid accumulation in liver. Most of the lipid metabolites down-regulated by HFD, such as phosphatidylcholine (PC), phosphatidylethanolamine (PE) and phosphatidylglycerol (PG), which were reported to be correlated with the composition of intestinal microvillus membrane [32] , and possess antiin ammatory activity to protect intestinal epithelial cells [33][34][35] . Thus, HFD might potentially increase cell membrane permeability and weakened the barrier function to promote the translocation of bacteria from gut to liver [36] . On the other hand, most of the lipid metabolites down-regulated by PCA are belonging to triradylglycerols (Supplemental Table 2), and KEGG analysis indicated that all these triradylglycerols are involved in insulin resistance-related pathways. The correlation analysis between gut microbiota and lipid metabolites indicated that Roseburia and Enterococcus are positively correlated with the production of Cer (m18:1(4E)/24:0), which has been proved to induce insulin resistance [37] and promote the development of NAFLD [38] . Our results indicated that PCA showed a direct inhibitory effect on E. faecalis, a representative strain of Enterococcus, and FMT results revealed that E. faecalis aggravate the progress of NAFLD. Therefore, the protective effect of PCA on NAFLD was largely dependent on the attenuated insulin resistance by down-regulating the relative abundance of Enterococcus.

Not applicable
Availability of data and materials The datasets during and/or analysed during the current study available from the corresponding author on reasonable request.   (K) HOMA-IR index based on levels of glucose and insulin in serum. Data represent as mean ± SD (n=6). Bars with different letters differ signi cantly (P < 0.05). ALT, alanine aminotransferase; AST, aspartate aminotransferase; HDL-c, high density lipoprotein cholesterol; HFD, high-fat diet; HOMA-IR, homeostasis model assessment-estimated insulin resistance; LDL-c, low density lipoprotein cholesterol; LFD, low-fat diet; MDA, malondialdehyde; PCA, protocatechuic acid; TNF, tumor necrosis factor.