Abstract

   Alisma orientale (Sam.) Juzep (A. orientale) is a traditional herb that
   is often used to treat disease including edema and hyperlipidemia.
   However, the molecular mechanism by which Alisma orientale (Sam.) Juzep
   exerts its hypolipidemic effects remains unclear. In this study, a
   diabetic rat model was established by feeding a high-fat and high-sugar
   diet combined with a low-dose streptozotocin injection (HFS). Then the
   rats were treated with an A. orientale water extract (AOW), an A.
   orientale ethanolic extract (AOE) or metform (MET). The gut microflora
   and liver transcriptome were analyzed by high-throughput
   next-generation sequencing. Ultra-performance liquid
   chromatography-triple quadrupole-mass spectrometry was employed to
   analyze the major compounds in the AOE. The results showed that the
   serum total cholesterol (TC) and low density lipoprotein cholesterol
   (LDL-C) levels in rats of the AOE group (2.10 g/kg/day, 14 days) were
   significantly lower than those in the HFS group (p<0.01). Moreover, AOE
   treatment altered the gut microecology, particularly modulating the
   relative abundance of gut microflora involved in lipid metabolism
   compared with the HFS group. Furthermore, compared with the HFS group,
   the mRNA expression levels of Fam13a, Mapk7, Mpp7, Chac1, Insig1,
   Mcpt10, Noct, Greb1l, Fabp12 and Hba-a3 were upregulated after the
   administration of AOE. In contrast, the mRNA expression levels of Lox,
   Mybl1, Arrdc3, Cyp4a2, Krt20, Vxn, Ggt1, Nr1d1 and S100a9 were
   downregulated. Moreover, AOE treatment for two weeks markedly promoted
   the relative abundance of Lachnospiraceae (p = 0.0013). The
   triterpenoids contents in AOE were alisol A, alisol A 24-acetate,
   alisol B, alisol B 23-acetate, alisol C 23-acetate, alisol F, alisol F
   24-acetate, and alisol G. Our findings above illustrated that the
   hypolipidemic effect of the triterpenoids of A. orientale is mediated
   mainly through alteration of the gut microecology and the regulation of
   genes involved in cholesterol metabolism, especially Insig1.

Introduction

   Diabetes mellitus (DM) is a metabolic disease associated with the
   inability to produce insulin and the inability to properly utilize
   glucose (T1D) or insulin resistance when interacting with its receptor
   (T2D) [[40]1]. Approximately 10% of the world’s population already has
   T2D or may develop it, and approximately 40% of adults are overweight
   or obese [[41]2]. Obesity is a major risk factor for metabolic syndrome
   and T2D, which share a potential mechanism of insulin resistance (IR)
   [[42]3, [43]4]. So far, many antihyperglycemic chemical or biological
   drugs have been developed, such as insulin secretagogues
   (Thiazolidinediones, Sulfonylureas, Biguanides), insulin sensitizer
   agents, α-glucosidase inhibitors, GLP-1 analogs, DPP-4 inhibitors,
   cannabinoid-1 receptor inhibitors, G protein-coupled receptor 40
   agonists and GPR119 agonists [[44]5]. However, some inevitable adverse
   effects, such as hypoglycemia, liver dysfunction and gastrointestinal
   reactions, have attracted considerable attention [[45]5].

   DM is known as Xiao-Ke disease in ancient China in terms of polydipsia,
   polyphagia, polyuria, and body weight loss. Traditional Chinese
   medicines were used to treat the disease for around 1, 500 years
   [[46]6]. In recent years, increasing amounts of researches have shown
   that some Chinese herbs or their extracts can effectively treat T2D
   and/or obesity [[47]7, [48]8]. Iridoid glycoside from Cornus
   officinalis effectively mitigated the general symptoms of DM rats
   including weight loss, polydipsia, polyphagia, polyuria, elevated blood
   glucose level and low serum insulin level [[49]9]. Purified
   anthraquinone-glycoside from Rheum palmatum L. could reduce oxidative
   stress in rats with DM by improving their blood lipid metabolism and
   enhancing their antioxidant capacity, thereby inhibiting beta-cell
   apoptosis and improving beta-cell function [[50]10, [51]11]. Quercetin
   extracted from the flowers of Edgeworthia gardneri exerted excellent
   properties in islet protection and amelioration [[52]12]. Lychee seed
   extract remarkably prevented neuronal injury and improved cognitive
   functions in T2DM rats [[53]13]. The two fractions of Morus alba fruit
   polysaccharides (MFP50 and MFP90) had markedly antihyperglycemic and
   antihyperlipidemic effects and could relieve diabetes symptoms in the
   T2D rat model [[54]14]. Hawthorn extracts decreased the levels of
   triglyceride, total cholesterol and blood glucose, whereas the level of
   plasma insulin released from the pancreas was increased [[55]15]. In
   addition, Zhang et al. showed that the ethyl acetate extract of
   Forsythia suspense (Thunb.) Vahl fruit could treat DM via modulation of
   oxidative stress, hepatic glucose metabolism and pancreatic insulin
   secretion [[56]16].

   Alisma orientale (Sam.) Juzep (A. orientale) is a well-known medical
   plant of the Alisma genus and Alismataceae family in China, and is also
   distributed in the former Soviet Union, Mongolia, Korea and Japan. A.
   orientale is sweet in taste and cold in properties. It is mainly used
   to treat wind, cold and dampness impediments and lactation
   difficulties, to eliminate water, to nourish the five Zang-organs, to
   increase energy and to strengthen the body. It is effective in
   promoting urination, eliminating dampness and resolving heat.
   Clinically, it is used to treat dysuria, edema, beriberi, diarrhea,
   stranguria and leucorrhea.

   The Liuwei Dihuang decoction is a well-known classic traditional
   Chinese medicine formula, which consists of six herbs including A.
   orientale. It can intervene in the insulin resistance of T2DM, in part
   through regulation of the canonical PI3K/Akt signaling pathway
   [[57]17]. Both A. orientale and gliclazide monotherapy are effective in
   lowering fasting blood glucose. As a single-target drug, gliclazide has
   a stronger effect in lowering fasting glucose. However, A. orientale
   can also control weight and improve glucose intolerance [[58]18].

   Kim et al. [[59]19] have reported metabolomic changes in the evaluation
   of the treatment effects of A. orientale and Artemisia capillaris on
   fatty livers in diabetic mice. The AMP/ATP ratio of cellular energy
   homeostasis in A. orientale-treated mice was increased relative to than
   that of the Artemisia capillaris-treated mice. A. orientale is a more
   effective natural product in the treatment of liver dysfunction than
   Artemisia capillaris. Unfortunately, they did not report the
   hypoglycemic effect of A. orientale and Artemisia capillaris in
   diabetic mice.

   In this study, in order to explore the hypoglycemic and hypolipidemic
   effects of A. orientale and its potential molecular mechanism, a DM rat
   model was established. Subsequently, the water extract (AOW) and
   ethanolic extract (AOE) of A. orientale were separately administered
   for hypoglycemic and hypolipidemic treatment. The liver is an important
   organ that plays a central role in the regulation of systemic glucose
   and lipid metabolism [[60]19]. What’s more, the microbiota also affects
   many metabolic functions, including the regulation of glucose and lipid
   homeostasis. There is increasing evidence that any modification of the
   microbiota composition may lead to a variety of diseases, including
   metabolic diseases, such as diabetes, obesity, and cardiovascular
   diseases [[61]20–[62]22]. In view of the above factors, the serum
   biochemical indicators, liver histopathology observation, liver
   transcriptome analysis and gut microflora were evaluated. The gut
   microflora and liver transcriptome were analyzed by high-throughput
   next-generation sequencing. This is the first report of the regulatory
   efficacy of A. orientale on gut microecology and the liver
   transcriptome profile in vivo.

Materials and methods

Ethics statement

   The study was approved by the ethics review board of Fujian Academy of
   Medical Sciences (Fuzhou, Fujian Province, China). All experimental
   procedures were performed in accordance with the China legislation
   regarding Laboratory animal—Guideline for ethical review of animal
   welfare (GB/T 35892–2018) published by the General Administration of
   Quality Supervision, Inspection and Quarantine of the People’s Republic
   of China (02-06-2018 issued, 09-01-2018 implemented). All surgery was
   performed under anesthesia, and all efforts were made to alleviate
   suffering.

Preparation of the Alisma orientalis extract

   The dried samples of A. orientale were provided by Zhuqing Lv at the
   Station for Popularizing Agricultural Technique of Nanping City, Fujian
   Province, China. 252 g of A. orientale was boiled in 5040 ml (1:20) of
   80% ethanol or water twice, 1.5 hours each time, using the method of
   reflux extraction. The extracts were then combined, filtered and
   concentrated to a density of 0.21 g crude herb/ml by a rotary
   evaporator, and stored at 4°C until further use.

Establishment of diabetic rat model and experimental design

   Forty-eight specific pathogen-free male Sprague Dawley (SD) rats (eight
   weeks old, 180–220 g) used in this experiment were purchased from the
   Shanghai Laboratory Animal Company (SLAC, Shanghai, China). All rats
   were housed in sanitized polypropylene cages in a hygienic room with a
   natural day-night light cycle with a regulated temperature (23–26°C)
   and 55–65% relative humidity. Standard water and chow diet were
   provided ad libitum.

   After a week of adaptive feeding, the rats were randomly divided into a
   normal diet group (n = 16) and high fat diet group (n = 32). Rats in
   the normal diet group were fed with standard rodent chow, while those
   in the high fat diet group were fed with a high-fat diet containing 60%
   energy from fat for four weeks (Trophic Animal Feed High-Tech Co.,
   Ltd., Nantong, China). Their food intake was detected daily and the
   body weight was measured once a week.

   On day 29, we fast all rats for 6 hours (from 7:00 a.m. until 1:00
   p.m.) prior to STZ treatment. Water was provided as normal. The
   experimental rats were injected with STZ intraperitoneally at 40 mg/kg
   in 0.1 M citrate buffer (pH 4.5). The control rats were given an equal
   dose of citrate buffer [[63]23]. Three days after STZ administration,
   we measured their blood glucose level in a tail-vein blood sample using
   a One Touch Basic blood glucose monitoring system. Rats with a fasting
   blood glucose range of 11–14 mmol/L were considered T2DM and
   subsequently used for the study [[64]24].

   The rats fed a high-fat diet with STZ intraperitoneally injected were
   then randomly divided into four groups: the first group was orally
   administered saline as a model control (HFS group, n = 8), while the
   second group was orally administered metformin as a positive control at
   250mg/ (kg·d) (MET group, n = 8). The third group was orally
   administered the water extract of A. orientale at 2.1g/ (kg·d) (AOW
   group, n = 8), while the fourth group was orally administered the
   ethanol extract of A. orientale at 2.1g/ (kg·d) (AOE group, n = 8). The
   dose of 2.1g/(kg·d) was decided basing on twice the usual clinical
   dose. For the low fat diet group, the eight rats intraperitoneally
   injected with STZ only were classified as the STZ group, while the
   remaining eight rats intraperitoneally injected with citrate buffer
   were used as a control group (CON group).

Biochemical measurements of the serum

   After two weeks of treatment, all rats were kept on a fast for 12
   hours. The rats were then sacrificed under anesthesia with
   pentobarbital sodium at a dose of 40 mg/kg, and blood was obtained from
   the abdominal aorta. The Serum was separated by centrifugation at
   3000rpm for 15 min at 4°C. The serum aspartate transaminase (AST),
   alanine aminotransferase (ALT), triglyceride (TG), total cholesterol
   (TC), low-density-lipoprotein cholesterol (LDL-C), and
   high-density-lipoprotein cholesterol (HDL-C) levels were measured by an
   automatic biochemical analyzer (BS-400 MINDRAY Chemistry Analyzer,
   China).

Liver and cecal content samples collection

   Right after the blood collection, the livers were removed and weighed,
   then washed with normal saline. The partial liver tissues of the right
   lobe were excised and fixed in 10% formaldehyde solution for paraffin
   embedded sections. The remaining parts of the liver tissues were
   immediately cut into small pieces and frozen in liquid nitrogen for
   approximately 15 min then stored at −80 °C until liver transcriptome
   and Q-PCR determination. Simultaneously, the tissues of the cecum were
   cut and the cecal content samples were collected into a cryotube and
   frozen in liquid nitrogen for approximately 15 min then stored at −80
   °C for throughput sequencing analysis of the gut microbiota.

Liver histopathological observation

   For histopathological observation, the collected liver samples fixed in
   formaldehyde solution were treated with a series of different
   concentrations of ethanol solutions. The liver tissues samples were
   embedded in paraffin and cut into thick sections by a pathological
   slicer, then stained with hematoxylin-eosin (H&E) and oil red O, and
   liver histopathological observation were performed under an optical
   microscope.

Gut microbiota total DNA extraction and high throughput sequencing

   Three samples from the gut microbiota were collected randomly from each
   of the groups. The total DNA was extracted from the cecal contents of
   the rats using the QIAamp Fast DNA Stool mini kit (Cat ID: 51604,
   Qiagen, Germany) according to the manufacturer’s instructions. The
   V3-V4 hypervariable regions of 16S rRNA from the gut microbiota were
   amplified using standard primers (forward primer
   5′-ACTCCTACGGGAGGCAGCA-3′ and reverse primer 5′-
   GGACTACHVGGGTWTCTAAT-3′). Sequencing libraries were prepared using
   Illumina’s TruSeq Nano DNA LT Library Prep Kit. The quality of the
   libraries was tested on the Agilent Bioanalyzer using the Agilent High
   Sensitivity DNA Kit (Agilent, USA). The libraries were quantified on a
   Promega QuantiFluor fluorescence quantification system using the
   Quant-iT PicoGreen dsDNA Assay Kit (Thermo Fisher Scientific, USA). The
   sequencing was performed on the Illumina MiSeq instrument with a 2 ×
   300 paired-end configuration using the MiSeq Reagent Kit V3 (Illumina,
   USA).

Microbiota diversity analysis

   To explore the diversity of the microbiota species composition, the
   paired-end sequences were paired according to overlapping bases using
   the FLASH software (v1.2.7, [65]http://ccb.jhu.edu/software/FLASH/)
   [[66]25], and the connected paired sequences were assigned to the
   corresponding samples according to the barcode sequences, thereby
   obtaining an effective sequence of each sample. Operational taxonomic
   units (OTUs) were generated by clustering at a similarity of 97% using
   the QIIME software and the UCLUST sequence alignment tool [[67]26]
   against the Greengenes database [[68]27], RDP (Ribosomal Database
   Project) database [[69]28] and the Silva database [[70]29]. The alpha
   diversity of each sample was calculated using the Chao1 estimator
   [[71]30], the ACE estimator [[72]31], the Shannon diversity index
   [[73]32], and the Simpson index [[74]33]. The composition and abundance
   distribution of each sample at the five classification levels of
   phylum, class, order, family and genus were obtained using QIIME
   software. Beta diversity analysis was performed by the principal
   component analysis (PCA), multidimensional scaling (MDS) and clustering
   analysis methods. The Spearman rank correlation coefficient between the
   dominant genera was calculated using Mothur software. The associated
   network was constructed and visualized in Cytoscape
   ([75]http://www.Cytoscape.org/) [[76]34]. The PICRUSt (phylogenetic
   investigation of communities by reconstruction of unobserved states)
   was used to predict the bacterial metabolic function [[77]35].

Total RNA extraction and RNA-seq analysis

   Five liver samples were collected randomly from each of the CON, HFS,
   MET and AOE groups. Total RNA was extracted using TRIzol reagent
   (Invitrogen Life Technologies, USA) according to the manufacturer’s
   protocol. The same RNA samples were used for qRT-PCR analysis. The
   concentration, quality and integrity were determined using a BioDrop
   spectrophotometer (BioDrop Technologies, UK) and an Agilent bioanalyzer
   2100 (Agilent, USA). Sequencing libraries were constructed using the
   TruSeq RNA sample preparation kit (Illumina, USA). Briefly, mRNA was
   purified from total RNA using poly-T oligo-attached magnetic beads.
   Fragmentation was carried out using divalent cations under an elevated
   temperature in an Illumina proprietary fragmentation buffer. First
   strand cDNA was synthesized using random oligonucleotides and
   SuperScript II. Second strand cDNA synthesis was subsequently performed
   using DNA polymerase I and RNase H. The remaining overhangs were
   converted into blunt ends via exonuclease/polymerase activity and the
   enzymes were removed. After adenylation of the 3′ ends of the DNA
   fragments, Illumina PE adapter oligonucleotides were ligated to prepare
   for hybridization. To select cDNA fragments of the preferred 200 bp
   length, the library fragments were purified using the AMPure XP system
   (Beckman Coulter, Beverly, CA, USA). DNA fragments with ligated adaptor
   molecules on both ends were selectively enriched using an Illumina PCR
   primer cocktail in a 15-cycle PCR. Products were purified (AMPure XP
   system) and quantified using an Agilent high sensitivity DNA assay on a
   Bioanalyzer 2100 system (Agilent). The sequencing library was then
   sequenced on a Hiseq platform (Illumina) by Shanghai Personal
   Biotechnology Co. Ltd.

Bioinformatic analysis

   After the raw data collation, filtering and quality assessment, the
   filtered reads were mapped to the reference genome using HISAT2
   software ([78]http://ccb.jhu.edu/software/hisat2/index.shtml). The gene
   expression was normalized using FPKM, which is the number of Fragments
   Per Kilo bases per Million fragments. Difference analysis of gene
   expression was performed by DESeq, and the differentially expressed
   genes (DEGs) were screened with |log2 Fold Change| > 1, p <0.05.
   GO-enrichment analysis was performed using topGO, and the P-value was
   calculated by the hypergeometric distribution test. Pathway-enrichment
   analysis was performed using the Kyoto Encyclopedia of Genes and
   Genomes (KEGG) database annotation. The results of the differentially
   expressed genes, and the significance value of the differential gene
   enrichment was p<0.05. Protein and protein interacting (PPI) analysis
   was performed using STRING (Search3 Tool for the Retrieval of
   Interacting Genes/Proteins). The correlations heatmap was drawn using
   the OmicShare ([79]http://www.omicshare.com/tools/Home/Soft/senior)
   online tool.

qRT-PCR validation

   To validate the Illumina sequencing data, Insig1 was selected for
   qRT-PCR analysis, using a SYBR Green PCR Kit (TaKaRa, Dalian, China)
   with 100 ng of cDNA as a template. After mixing, the PCR was performed
   using a LightCycler 480 instrument. The Gapdh gene was used as an
   internal control, and the relative gene expression level was analyzed
   using the 2^-△△CT method. The primers were synthesized by Biosune
   Biotech (Shanghai) Co., Ltd. Primer sequences will be provided upon
   request.

Determination of the contents of triterpenoids in A. orientale

   AOW and AOE samples were added to an equal amount of acetonitrile,
   ultrasonically mixed, then filtered through a 0.45μm filter membrane
   for content determination. Alisol A, alisol A 24-acetate, alisol B,
   alisol B 23-acetate, alisol C 23-acetate, alisol F, alisol F
   24-acetate, and alisol G (MANSITE Biotechnology Co. Ltd., Chendu,
   China.) were selected as reference substances. Quantitative analysis of
   eight triterpenoids was performed on the Waters TQ TQS system of ultra
   high-performance liquid chromatography configured with a triple
   quadruple mass spectrometer (UPLC-QQQ-MS, Waters, UK), according to the
   methods used in previous research [[80]36].

Statistical analysis

   All of the experimental data are expressed as the mean ± standard
   deviations. Statistically significant differences between the groups
   were analyzed by one-way analysis of variance (ANOVA) followed by post
   hoc multiple comparison Fisher’s LSD t-tests using SPSS 18.0 software
   (IBM, USA). Statistically significant differences were defined as p<
   0.05.

Results

Effects of A. orientale on serum lipid and FBG levels in diabetic rats

   After treatment with metformin or A. orientale for two weeks, the serum
   level of LDL-C in the AOE group was markedly reduced compared with the
   HFS group. Compared to the CON group, the level of TC was increased in
   response to the HFS, which was then attenuated by AOE feeding. However,
   the levels of HDL-C, TG and FBG were not ultimately altered by the
   administration of metformin or A. orientale ([81]Fig 1).

Fig 1. Effects of A. orientale on the levels of serum biochemical parameters
and blood glucose in diabetic rats.

   [82]Fig 1
   [83]Open in a new tab

   (A) The serum level of TC. (B) The serum level of TG. (C) The serum
   level of LDL-C. (D) The serum level of HDL-C. (E) The level of blood
   glucose. Data represent the mean±SEM (n = 8). * p < 0.05 vs. CON group,
   ^# p < 0.05 vs. HFS group. TC, total cholesterol; TG, total
   triglyceride; LDL-C, low density lipoprotein cholesterol; HDL-C,
   high-density-lipoprotein cholesterol.

Effect of A. orientale on liver weigh, liver function and histopathology in
diabetic rats

   There were no significant differences in the weight of livers between
   the HFS group and the CON group. The rats in the HFS group had
   dramatically higher levels of ALT compared with those in the CON group,
   which indicated that rats in the HFS group had severe liver damage.
   However, after treatment with metformin or A. orientale, no significant
   improvements were found. Moreover, no significant changes in the level
   of AST were observed among the different groups ([84]Fig 2A).

Fig 2. Effect of A. orientale on the ratio of liver weight to body weight,
liver function and histopathology in diabetic rats.

   [85]Fig 2
   [86]Open in a new tab

   (A) The ratio of liver weight to body weight and the serum levels of
   ALT and AST. (B) The gross appearance of the rat livers. (C)
   Histopathological observations of the rat livers. H&E stained sections
   were photographed at 200×magnification, n = 3. (D) The Oil Red O
   staining of the rat livers. Data represent the mean±SEM (n = 8). * p <
   0.05 vs. CON group, ^# p < 0.05 vs. HFS group. AST, aspartate
   transaminase; ALT, alanine aminotransferase.

   The liver tissues in the CON group were normal in appearance, soft in
   texture, and dark red in color ([87]Fig 2B). H&E staining showed that
   the liver of this group was observed to be complete, and the structure
   of the liver lobules was normal ([88]Fig 2C). The results of Oil Red O
   staining revealed that the liver cells were not affected by steatosis
   ([89]Fig 2D). The gross appearance of the liver in the HFS group was
   different from that of the CON group. The liver volume of the HFS group
   was significantly increased, the texture was brittle, and the color was
   greasy yellow. Correspondingly, HE staining showed that the hepatic
   cord structure was disordered and irregularly arranged. Most
   hepatocytes were swollen with severe steatosis. Moreover, an uneven
   size and quantity of lipid droplets were visible in the cytoplasm after
   of Oil Red O staining. However, after treatment with metformin or A.
   orientale, all of these symptoms were obviously ameliorated. In
   conclusion, these results indicated that A. orientale might ameliorate
   HFS-induced hepatic steatosis in rats.

Effect of A. orientale on gut microecology in diabetic rats

   To investigate whether the treatment of A. orientale in diabetic rats
   is related to the changes in their gut microecology, 16S rRNA of 18
   intestinal content samples were detected by high-throughput sequencing.
   A total of 968,160 effective reads were obtained from the samples of
   the six groups rats (n = 3). After obtaining the OTU abundance matrix,
   alpha diversity analysis was performed to calculate the diversity of
   each sample population. Rarefaction curves of the Chao1 and Shannon
   index indicated that the current sequencing depth of each sample was
   sufficient to reflect the microbial diversity contained in the
   community. The Chao1 index, ACE index, Shannon index and the Simpson
   index are commonly used metrics for alpha diversity analysis. The Chao1
   index and ACE index focus on community richness, while the Shannon
   index and Simpson index take into account community uniformity. There
   was no obvious difference among the six groups ([90]Fig 3).

Fig 3. Effect of A. orientale on gut microecology.

   [91]Fig 3
   [92]Open in a new tab

   (A) Rarefaction curve of the Chao1 index. (B) Rarefaction curve of the
   Shannon index. (C) Alpha diversity analysis. (D) PCA analysis: each
   point represents a sample, drawn by the first principal component on
   the X-axis and the second principal component on the Y-axis, with one
   color per group. n = 3.

   The OTU was divided and the classification status identification
   results were statistically analyzed. R-software was used to draw the
   histogram to visually compare the difference between the OTU number and
   the classification status of different groups of samples. The phylum,
   class, and order were the top three classifications with the largest
   OTU number, and the number of OTU classified to the species was the
   least ([93]Fig 4A). The relative abundance of bacterial phylum types
   was compared among the six groups to further investigate the
   composition alterations in the gut microecology. Firmicutes,
   Bacteroidetes, Proteobacteria, Verrucomicrobia were the four most
   abundant bacterial phyla in the six groups. There was no obvious
   difference in the four phyla bacteria among the six groups, but the AOW
   group had the highest ratio of Firmicutes and Bacteroidetes compared
   with other groups ([94]Fig 4B). At the class level, the STZ group
   displayed an increased relative abundance of bacilli in gut bacteria
   when compared with the CON group ([95]Fig 4C). However, after treatment
   with A. orientale, the bacilli abundance was decreased notably. At the
   order level, gut bacteria from the STZ group displayed the same
   increasing trend of Lactobacillales abundance when compared with the
   other groups as the class level ([96]Fig 4D). It’s a pity that there
   was no statistical significance between the CON_vs_HFS group and
   HFS_vs_AOE group in the above two levels.

Fig 4. Taxonomic diversity of A. orientale on gut microecology in diabetic
rats.

   [97]Fig 4
   [98]Open in a new tab

   (A) The number of OTUs and the classification status of different
   groups of samples. R-software was used to draw the histogram to
   visually compare the difference. (B) The ratio of Firmicutes to
   Bacteroidetes at the phylum level. (C) The relative abundance of
   bacilli in gut bacteria at the class level. (D) The relative abundance
   of Lactobacillales at the order level. (E) The relative abundance of
   Ruminococcaceae, Lactobacillale and Lachnospiraceae at the family
   level. (F) The relative abundance of Ruminococcaceae, Lactobacillale
   and Lachnospiraceae at the genus level. ^a p < 0.05 vs. the CON group,
   ^bp < 0.05 vs. the STZ group, ^cp < 0.05 vs. the HFS group. n = 3.

   At the family level, as shown in [99]Fig 4E, the STZ group displayed an
   increased number of Lactobacillaceae in gut bacteria when compared with
   the CON group, and A. orientale administration had a notable decreasing
   effect on Lactobacillaceae abundance. The HFS and MET groups displayed
   a decreased relative abundance of Ruminococcaceae in the gut bacteria
   when compared with the CON group or STZ group. The finding most worth
   mentioning is the Lachnospiraceae relative abundance, which in the HFS,
   MET and AOW groups displayed an increased effect, while AOE treatment
   had a notable decreasing effect. In addition, there was no difference
   in Lachnospiraceae relative abundance among the AOE, CON and STZ
   groups. At the genus level, the STZ group displayed an increased
   relative abundance of Lactobacillus in the gut bacteria when compared
   with the CON group, but after treatment with A. orientale or MET, it
   was notably decreased ([100]Fig 4F). The HFS group displayed a
   decreased relative abundance of Ruminococcus in the gut bacteria when
   compared with the CON group or STZ group, but there were no significant
   differences after treatment with A. orientale or MET ([101]Fig 4F).
   Interestingly, there was no difference in the relative abundance of
   Lachnospira among the CON, STZ, HFS, MET and AOE groups.

   In addition to the composition and structural analysis of the gut
   bacteria, the metabolic function of the bacteria is also a focus of gut
   microecology. In this study, the KEGG level 2 PICRUSt results showed
   that carbohydrate metabolism and amino acid metabolism were the two
   highest relative abundance metabolism pathways of the gut bacteria
   ([102]Fig 5A). In the carbohydrate metabolism pathway, the relative
   abundance of gut bacteria in the HFS group increased when compared with
   the CON group, but after treatment with AOE, the relative abundance of
   gut bacteria significantly decreased ([103]Fig 5B). In the amino acid
   metabolism pathway, the relative abundance of gut bacteria in the MET
   group, AOW group and AOE group had increased when compared with the STZ
   group ([104]Fig 5C).

Fig 5. KEGG second level distribution predicted PICRUSt.

   [105]Fig 5
   [106]Open in a new tab

   Carbohydrate metabolism and amino acid metabolism were the two highest
   relative abundance metabolism pathways of the gut bacteria. (A) The
   KEGG level 2 PICRUSt. (B) Carbohydrate metabolism. (C) Amino acid
   metabolism. ^a p < 0.05 vs. CON group, ^bp < 0.05 vs. STZ group, ^cp <
   0.05 vs. HFS group. n = 3.

Effect of A. orientale on the liver transcriptome in diabetic rats

   To obtain the transcriptome dynamic profile of the hypolipidemic effect
   of AOE, four mRNA libraries were constructed of five liver samples from
   the CON, HFS, MET and AOE groups. RNA sequencing was performing on the
   Illumina HiSeq (TM) 2500 system. The Q30 values in all of the libraries
   were more than 94%. From the libraries of the CON, HFS, MET and AOE
   group rats, 214,205,422, 210,281,068, 230,221,008 and 218,810,100
   high-quality reads were obtained respectively. The uniquely mapped
   rates in four of the libraries were all more than 91%. The results of
   the alignment of the gene regions showed that the ratio of reads mapped
   to exons were all over 90%, indicating that the prepared RNA libraries
   were of good quality and that the RNA sequencing data analysis results
   are reliable.

   The analysis results showed that there were 3,321 DEGs in the six
   groups (CON_vs_HFS, CON_vs_MET, CON_vs_AOE, HFS_vs_AOE, HFS_vs_MET and
   MET_vs_AOE). The total number of DEGs of CON_vs_HFS and HFS_vs_AOE were
   872 and 126, of which 467 and 56 were upregulated, whereas 405 and 70
   were downregulated, respectively ([107]Fig 6A). Venn diagrams were
   plotted to obtain the intersection and union among the four groups
   (CON_vs_HFS, HFS_vs_AOE, HFS_vs_MET and MET_vs_AOE) using the OmicShare
   tools, a free online platform for data analysis
   ([108]www.omicshare.com/tools). The results showed that there were 789
   specific DEGs in CON_vs_HFS, 62 specific genes in HFS_vs_AOE, 76
   specific genes in HFS_vs_MET and 135 specific genes in MET_vs_AOE.
   There were 30 common DEGs between CON_vs_HFS and HFS_vs_AOE, 29 common
   DEGs between CON_vs_HFS and HFS_vs_MET, and 3 common DEGs among
   CON_vs_HFS, HFS_vs_AOE and HFS_vs_MET. There were no common DEGs among
   the four groups ([109]Fig 6B).

Fig 6. Statistical analysis of differentially expressed genes.

   [110]Fig 6
   [111]Open in a new tab

   (A) The number of significantly upregulated and downregulated genes in
   CON_vs_HFS, CON_vs_MET, CON_vs_AOE, HFS_vs_AOE, HFS_vs_MET and
   MET_vs_AOE. (B) Venn diagram of the counts of DEGs in CON_vs_HFS,
   HFS_vs_AOE, HFS_vs_MET and MET_vs_AOE (n = 5).

GO functional classification of DEGs

   GO classification can be divided into three functional categories:
   Biological Process (BP), Cellular Component (CC) and Molecular Function
   (MF). After obtaining the DEGs, the top 10 GO categories enrichment
   comparing the three groups (CON_vs_HFS and HFS_vs_AOE) were analyzed
   ([112]Fig 7). In the BP categories, the most significant of the DEGs in
   the CON_vs_HFS group were assigned to small molecule metabolic process
   (142 genes, approximately 10% of total DEGs), fatty acid metabolic
   process (49 genes, approximately 19% of total DEGs), and lipid
   metabolic process single cell processes (106 genes, approximately 11%
   of DEG). In the CC categories, the most significant of the DEGs were
   classified into hemoglobin complex (8 genes, approximately 61% of the
   DEGs), cytoplasm (407 genes, approximately 5% of the DEGs), integral
   component of plasma membrane (48 genes, approximately 8% of the DEGs).
   In MF categories, the most significant of the DEGs were classified into
   catalytic activity (328 genes, approximately 6% of the
   DEGs),monooxygenase activity (24 genes, approximately 20% of the DEGs),
   heme binding (28 genes, approximately 18% of the DEGs); In the BP
   categories, the most significant of the DEGs in the HFS_vs_AOE group
   were assigned to response to external stimulus (25 genes, approximately
   1% of total the DEGs), gut absorption (3 genes, approximately 12% of
   the total DEGs), glutathione catabolic process (2 genes, approximately
   33% of the total DEGs); in the CC categories, the most significant of
   the DEGs were classified into apical part of cell (8 genes,
   approximately 3% of the DEGs), MHC class II protein complex (2 genes,
   approximately 15% of the DEGs), MHC protein complex (2 genes,
   approximately 8% of the DEGs); and in the MF categories, the most
   significant of the DEGs were classified into icosanoid binding (2
   genes, approximately 33% of the DEGs), arachidonic acid binding (2
   genes, approximately 33% of the DEGs), icosatetraenoic acid binding (2
   genes, approximately 28% of the DEGs).

Fig 7. GO enrichment analysis histogram.

   [113]Fig 7
   [114]Open in a new tab

   (A) GO term of co-expressed DEGs in CON_vs_HFS group. (B) GO term of
   co-expressed DEGs in HFS_vs_AOE group.

KEGG classification of DEGs

   To identify the pathogenesis of diabetes and the biological pathway
   activated during AOE treatment, the three groups of DEGs were mapped
   into the five categories (Cellular Processes, Environmental Information
   Processing, Human Diseases, Metabolism, Organismal Systems) of the KEGG
   database. The pathways with the lowest p-value and the top 20 pathways
   with the lowest False Discovery Rate (FDR) are shown in [115]Fig 8.
   There were 60 known KEGG pathways in the CON_vs_HFS group. The top 10
   pathways were the PPAR signaling pathway (19 DEGs), Fatty acid
   degradation (11 DEGs), Malaria (11 DEGs), AMPK signaling pathway (17
   DEGs), Biosynthesis of unsaturated fatty acids (8 DEGs), and Fatty acid
   elongation (8 DEGs), Steroid hormone biosynthesis (11 DEGs), ABC
   transporters (9 DEGs), African trypanosomiasis (8 DEGs), and Fatty acid
   biosynthesis (5 DEGs). In addition, the statistically significant lipid
   metabolism pathways were Arachidonic acid metabolism, Linoleic acid
   metabolism, Glycerolipid metabolism, Glycerophospholipid metabolism,
   and Primary bile acid biosynthesis. DEGs in the HFS_vs_AOE group were
   only involved in 21 known KEGG pathways, of which many DEGs were
   involved in human diseases, such as Type I diabetes mellitus, Fluid
   shear stress and atherosclerosis, Asthma, Rheumatoid arthritis,
   Systemic lupus erythematosus, Graft-versus-host disease, Allograft
   rejection, Autoimmune thyroid disease, Inflammatory bowel disease
   (IBD), Tuberculosis, Staphylococcus aureus infection, Toxoplasmosis,
   and Malaria; followed by the Organismal Systems of Hematopoietic cell
   lineage, Th1 and Th2 cell differentiation, Th17 cell differentiation,
   Gut immune network for IgA production, and Nitrogen metabolism,
   Arachidonic acid metabolism at the metabolism level. For the HFS_vs_MET
   group, the DEGs were involved in a total of 29 known KEGG pathways, of
   which 14 pathways were classified into human diseases, such as Viral
   myocarditis, Type I diabetes mellitus, Primary immunodeficiency,
   Graft-versus-host disease, Allograft rejection, Autoimmune thyroid
   disease, Asthma, Rheumatoid arthritis, Systemic lupus erythematosus,
   Tuberculosis, Staphylococcus aureus infection, Toxoplasmosis,
   Epstein-Barr virus infection, and HTLV-I infection. This was followed
   by10 pathways of Organismal Systems in Vascular smooth muscle
   contraction, Pancreatic secretion, Insulin secretion, Ovarian
   steroidogenesis Gut immune network for IgA production, Antigen
   processing and presentation,Th1 and Th2 cell differentiation,Th17 cell
   differentiation, Hematopoietic cell lineage, and the IL-17 signaling
   pathway. The KEGG classification of DEGs showed that the pathways of
   AOE treatment were different from those of MET treatment.

Fig 8. KEGG enrichment analysis histogram.

   [116]Fig 8
   [117]Open in a new tab

   (A) Pathway functional enrichment of co-expressed DEGs in CON_vs_HFS
   group. (B) Pathway functional enrichment of co-expressed DEGs in
   HFS_vs_AOE group.

   To explore the interactive features between the DEGs of the liver
   transcriptome and the different gut microbiota (DGM) after the
   treatment with AOE, 19 DEGs between the CON_vs_HFS group and the
   HFS_vs_AOE group were selected to analyze the possible connection
   between the DGM and the DEGs ([118]Table 1). The Pearson correlation
   analysis showed that several DEGs were significantly correlated with
   the DGM ([119]Fig 9). Specifically, Arrdc3, Cyp4a2, Ggt1, Krt20, Lox,
   Mybl1, Nr1d1 and Vxn were positively correlated with
   Firmicutes/Bacteroidetes, Bacilli, Lactobacillales, Lachnospiraceae,
   Lactobacillaceae and Lactobacillus but negatively correlated with
   Ruminococcaceae and Ruminococcus; conversely, Fabp12, Fam13a, Greb1l,
   Mpp7 and Noct were negatively correlated with Firmicutes/Bacteroidetes,
   Bacilli, Lactobacillales, Lachnospiraceae, Lactobacillaceae and
   Lactobacillus but positively correlated with Ruminococcaceae and
   Ruminococcus; Insig1 and Chac1 were positively correlated with
   Ruminococcaceae and Ruminococcus; Mapk7 was negatively correlated with
   Lactobacillaceae and Lactobacillus; Hba-a3 was negatively correlated
   with Lachnospiraceae, and S100a9 was positively correlated with
   Lachnospiraceae; there was no obvious correlation between Mcpt10 and
   DGM, and there was no obvious correlation between Lachnospira and the
   19 DEGs. Therefore, there is a significant correlation between the DEGs
   and DGM, and the treatment with AOE not only contributed to the DEGs of
   the liver transcriptome but also influenced the composition of the gut
   microbiota.

Table 1. The possible connection between the DEGs of the liver transcriptome
and different gut microbiota after the treatment with AOE.

   gene_id Name CON_vs_HFS _ log2fc CON_vs_HFS_pval HFS_vs_AOE_log2fc
   HFS_vs_AOE_pval
   ENSRNOG00000007947 Fam13a -1.20127584 0.00182383 1.08665777 0.0003144
   ENSRNOG00000002412 Mapk7 -Inf 0.03477149 Inf 0.0019114
   ENSRNOG00000018760 Mpp7 -1.54415823 0.033892265 1.446105017 0.0184865
   ENSRNOG00000014387 Chac1 -3.32515557 4.53425E-07 1.076356803 0.0436458
   ENSRNOG00000006859 Insig1 -3.90168675 7.81859E-19 1.082190399 0.0132363
   ENSRNOG00000049991 Mcpt10 -1.99047147 0.028280076 1.859241805 0.024636
   ENSRNOG00000010799 Noct -1.6026617 0.000285117 1.509174242 0.0017221
   ENSRNOG00000023492 Greb1l -1.42721832 2.61438E-05 1.361789787 0.0004628
   ENSRNOG00000038825 Fabp12 -1.99693729 1.81747E-07 1.454191507 0.0072356
   ENSRNOG00000045989 Hba-a3 -3.79286311 0.001462438 3.458842797 0.0002627
   ENSRNOG00000014426 Lox 1.273966701 0.021787572 -1.68047127 0.0048201
   ENSRNOG00000021669 Mybl1 2.128045034 0.014143272 -1.61791255 0.0406767
   ENSRNOG00000045649 Arrdc3 1.596898139 1.12023E-07 -1.18187643 0.0001116
   ENSRNOG00000030154 Cyp4a2 2.420294518 1.07669E-05 -1.06099037 0.0247949
   ENSRNOG00000027139 Krt20 1.632926348 0.003394672 -1.64311171 0.0009927
   ENSRNOG00000021663 Vxn 1.649218031 0.016859805 -1.44041241 0.0253653
   ENSRNOG00000047697 Ggt1 1.967731427 0.00867844 -1.4849093 0.0222422
   ENSRNOG00000009329 Nr1d1 1.737400351 3.25719E-08 -1.16116203 0.0002606
   ENSRNOG00000011483 S100a9 1.931674927 0.001516338 -1.10779264 0.0323183
   [120]Open in a new tab

Fig 9. The Pearson correlation heatmap.

   [121]Fig 9
   [122]Open in a new tab

   (A) Correlation analysis between biochemical parameters and the
   different gut microbiota at the genus level. (B) Correlation analysis
   among biochemical parameters, the different gut microbiota and the 19
   selected DEGs. A positive correlation is represented by red while a
   negative correlation is shown in green in the heat map.

Transcriptional level verification of Insig1

   Recent studies have indicated that Insig1 plays an important role in
   the regulation of fatty acid and cholesterol metabolism [[123]37]. To
   verify the accuracy of the Illumina sequencing, we selected Insig1 to
   verify the differences obtained from transcriptome sequencing by
   qRT-PCR. The results indicate that the gene expression level of Insig1
   verified by qRT-PCR revealed similar trend when compared with the
   RNA-Seq samples ([124]Fig 10).

Fig 10. The expression levels of the Insig1 were verified by qRT-PCR (n = 3).

   [125]Fig 10
   [126]Open in a new tab

   * p < 0.05 vs. CON group, ^# p < 0.05 vs. HFS group.

Characterization of major active compounds in AOE

   The contents of the eight compounds of alisol A, alisol A 24-acetate,
   alisol B, alisol B 23-acetate, alisol C 23-acetate, alisol F, alisol F
   24-acetate, and alisol G determinate by UPLC-QQQ-MS are shown in
   [127]Fig 11. The results indicated that all eight compounds could be
   detected in the AOE. Alisol B 23-acetate, alisol B and alisol C
   23-acetate were the major compounds, which may be the active compounds
   of the hyperlipidemic effect of AOE.

Fig 11. The characterization of major active compounds in AOE.

   Fig 11
   [128]Open in a new tab

Discussion

   In this study, the HFS group rats were given a high fat feed and
   exhibited persistent hyperglycemia during the experiment compared with
   the CON and STZ group rats. After two weeks of treatment, the serum
   LDL-C level decreased significantly in the AOE group compared with the
   HFS group. The macroscopic and pathological analysis of the livers
   showed that the liver morphology was ameliorated after administration
   of AOE. These results indicate that AOE could promote the recovery of
   fatty liver in HFS rats. Previous studies have reported significant
   reduction in serum TC and LDL-C atherogenic indexes in high-fat diet
   model mice or rats following administration of the triterpene fraction
   of A. orientale [[129]38, [130]39]. Therefore, our results are
   basically consistent with the reduction in serum LDL-C level as
   previously reported with other high fat model rodents.

   The key roles of gut microbiota in the etiology and development of
   various physiological diseases, and changes in the microbiota are
   related to a numbers of human diseases, such as allergic diseases,
   autoimmune encephalitis, atherosclerosis, colorectal cancer, obesity,
   and diabetes [[131]40, [132]41]. Ruminococcaceae and Lachnospiraceae
   are dominant families in energy metabolism, and they are strongly
   correlated with adipokine levels. There is a positive correlation
   between the Ruminococcaceae and insulin, C-peptide, and leptin levels
   [[133]42]. Lactobacillus is a kind of probiotic in the human
   gastrointestinal tract for preventing or treating T2DM [[134]43].

   In this study, there were significant differences in the abundance of
   Lactobacillaceae/Lactobacillus, Ruminococcaceae/Ruminococcus and
   Lachnospiraceae/Lachnospira among the six groups. A. orientale was
   found to normalize the high-fat diet and streptozotocin-induced gut
   microbiota dysbiosis. Particularly, A. orientale ameliorated gut
   microecology changes in the abundance of the three taxa back to normal
   status (e.g., AOE increased Ruminococcus, AOW enhanced Lachnospira, but
   both of them decreased Lactobacillus, which was significantly increased
   in the STZ group). In this study, FBG and serum TG level were
   significantly negative correlated with the abundance of
   Ruminococcaceae/Ruminococcus, and the serum TC level exhibited
   pronounced positive correlations with the abundance of
   Lactobacillaceae/Lactobacillus. Based on these findings, we believed
   that the hypolipidemic effect of A. orientale in diabetic rats is
   associated with an alteration of gut microbiota composition.

   The pathogenesis of obesity is a complex process involving the
   regulation of many genes. To identify the process of obesity and the
   DEGs after AOE treatment, the total RNAs of the liver from CON rats,
   HFS rats, MET and AOE-treated group rats were extracted and
   RNA-sequenced. According to the DEGs analysis in the liver
   transcriptome of the four groups of rats, there were 19 significant
   DEGs among the groups. The results showed that ten coexpressed DEGs
   were significantly upregulated and nine coexpressed DEGs were
   significantly downregulated in AOE_vs_HFS and HFS_vs_CON. Among the 19
   DEGs, there were five genes (Insig1, CYP4A2, Fabp12, nocturnin and
   Nr1d1) involved in the pathways of lipid metabolism. Insulin-induced
   gene 1 protein (Insig1) mediates feedback control of cholesterol
   synthesis by controlling SCAP and HMGCR [[135]44–[136]47]. Cytochrome
   P450 4A2 (CYP4A2) catalyzes the omega- and (omega-1)-hydroxylation of
   various fatty acids [[137]48, [138]49]. Fatty acid-binding protein 12
   (Fabp12) may play a role in lipid transport [[139]50, [140]51].
   Nocturnin is a circadian deadenylase, which plays an important role in
   posttranscriptional regulation of metabolic genes under circadian
   control and exerts a rhythmic posttranscriptional control of genes
   necessary for metabolic functions [[141]52–[142]56]. Moreover, Nuclear
   receptor subfamily 1 group D member 1 (Nr1d1) is a transcriptional
   repressor, which coordinates the circadian rhythm and metabolic
   pathways in a heme-dependent manner [[143]57]. It regulates genes
   involved in metabolic functions, including lipid and bile acid
   metabolism [[144]58, [145]59], adipogenesis [[146]60], glucose
   homeostasis [[147]61] and the macrophage inflammatory response
   [[148]62, [149]63]. These differentially expressed genes provide new
   target proteins for the molecular mechanism of the hypolipidemic effect
   of A. orientale.

   In current study, the FBG level were significantly negative
   correlations with the expression level of insig1, and the body weight
   exhibited strongly positive correlations with the expression level of
   insig1, The serum TC, HDL-C and LDH-C level were significantly negative
   correlations with the expression level of nocturnin, and the serum TC,
   TG, HDL-C, LDH-C level, and the body weight were positively correlated
   with the expression level of Nr1d1. Based on these findings, we believe
   that changes in the expression levels of these DEGs may be the
   molecular basis for the reduction of blood lipid levels in diabetic
   rats after AOE treatment.

   Cholesterol is an extremely important biological molecule that has
   roles in membrane structure as well as being a precursor molecule for
   the synthesis of the steroid hormones, the bile acids, and vitamin D
   [[150]64]. In the pathway of cholesterol biosynthesis, the rate
   limiting enzyme is 3-hydroxy-3-methylglutaryl-CoA (HMG-CoA) reducatase
   (HMGCR) [[151]65]. In the pathway of cholesterol metabolism, CYP7A1 is
   the rate limiting enzyme in the primary pathway of [152]bile acid
   synthesis pathway [[153]66, [154]67]. Our results revealed that oral
   administration of AOE did not affect the mRNA levels of HMGCR and
   CYP7A1, but increase the mRNA and protein expression levels of insig1.
   Insig1 is one of the insig encoding genes, the other one is insig2.
   Insig1 gene expression is highest in human liver while Insig2 gene
   expression is ubiquitous. Both insig proteins can cause ER retention of
   the SREBP/SCAP complex, which results in decrease of the intracellular
   cholesterol content. Expression of Insig1 has also been shown to be
   regulated by several members of the nuclear receptor family including
   PPARδ, PXR and CAR. In addition to their role in regulating
   sterol-dependent gene regulation, both Insig proteins activate
   sterol-dependent proteasomal degradation of HMGCR. In present study,
   there were no changes in the expression levels of insig2, Scap, Srebf1,
   Srebf2 and Hmgcr between the HFS group and AOE group. Based on these
   findings, it suggested that the hypolipidemic effect of A. orientale in
   diabetic rats was attributed to increasing levels of Insig1
   transcription.

   There were certain limitations to this study. First, two weeks of
   treatments with metformin or A. orientale might not be long enough to
   change the liver function, nor to make significant changes in gut
   microecology. Further study should be carried out to investigate what
   would happen to these indicators if the administration time was
   prolonged. In addition, further research was needed to determine which
   active compound in AOE plays the leading role in hypolipidemic effect,
   and the corresponding hypolipidemic pathways about Insig1.

Conclusion

   Based on the results of RNA-Seq transcriptomics, we identified some
   DEGs involved in the pathogenesis of diabetes and elucidated the
   potential molecular mechanism, by which AOE exerted its protective
   effects in diabetic rats. The active ingredients in AOE exerted
   hypolipidemic effect potentially by acting on Insig1. Finally, we
   presented a hypothetical model for how AOE played its role of
   hypolipidemic effects that will help us to understand more about the
   molecular ecological mechanisms.

Supporting information

   S1 Data

   (XLSX)
   [155]Click here for additional data file.^ (3MB, xlsx)
   S1 Fig

   (TIF)
   [156]Click here for additional data file.^ (1.3MB, tif)

Data Availability

   All data are contained within the article. Raw data are available at
   [157]https://dataview.ncbi.nlm.nih.gov/object/PRJNA632724?reviewer=g6oa
   hiaevhnv3q75r6pei3foeq and
   [158]https://dataview.ncbi.nlm.nih.gov/object/PRJNA632739?reviewer=eg4r
   sqb9k8r7blmntnj7j23hnb.

Funding Statement

   This work was supported by the Basic Research Project of Fujian
   Provincial Public Welfare Research Institute, China (2017R1031-2,
   2018R1031-3 and 2019R1011-5), Medical Innovation Project of Fujian
   Provincial Health and Family Planning Commission, China (2018-CX-15),
   and High-Level Hospital grants from Fujian Provincial Hospital, Fujian
   province, China (2017GL-001). The funders had no additional role in
   study design, data collection and analysis, decision to publish, or
   preparation of the manuscript. The author Wei Lin received a salary
   from the funder of Fujian Provincial Hospital, Fujian province, China.
   The authors received no specific funding for this work.

References