Prenatal inflammation causes obesity and abnormal lipid metabolism via impaired energy expenditure in male offspring

Introduction Obesity has becoming a global health issue. Fetus exposed to adversity in the uterine are susceptible to unhealth stimulus in adulthood. Prenatal inflammation is related to poor neonatal outcomes like neurodevelopmental impairments and respiratory complications. Recent studies suggested prenatal lipopolysaccharide (LPS) exposure could result in metabolic disorders. Thus, we hypothesized that offspring exposed to prenatal inflammation could develop into metabolic disorder. Methods The pregnant C57BL/6J mice were intraperitoneally injected with 50 μg/kg LPS or saline only once at GD15. The male offspring were weighted weekly until sacrificed. Indirect calorimetry and body composition were both performed at 9 and 18 weeks old. At 20 weeks old, mice were fasted overnight before collecting blood samples and liver for metabolomics analysis and RNA sequencing, respectively. Differentially expressed genes were further verified by RT-qPCR and western blotting. Results Prenatal inflammation resulted in obesity with increased fat percentage and decreased energy expenditure in middle-age male offspring. Abnormal lipid accumulation, changes of gene expression profile and upregulation of multi-component mechanistic target of rapamycin complex 1 (mTOR)/Peroxisome proliferator-activated receptor-γ pathway was observed in liver, accompanied with decreased bile acids level, unsaturated fatty acids androgens and prostaglandins in serum. Indirect calorimetry showed increased respiratory exchange rate and deceased spontaneous activity at 9 weeks in LPS group. Impaired energy expenditure was also observed at 18 weeks in LPS group. Conclusion Prenatal LPS exposure led to obesity and abnormal lipid metabolism through impaired energy expenditure. Supplementary Information The online version contains supplementary material available at 10.1186/s12986-022-00642-y.


Introduction
Obesity affects a third of the world population [1,2], and it disturbs nearly all physiological functions of the body. Obesity increases the risk of developing multiple disease conditions, such as diabetes mellitus [3], cardiovascular disease [4], several types of cancers [5] and poor mental health [6]. Despite the risk factors as heredity and western diet [7], a clear understanding of the potential mechanisms originated from early life which might permanently program the metabolism is still lacking. According to the developmental origins of health and disease (DOHaD) theory, fetus exposed to adversity like toxins, malnutrition, infection and other none-infectious Ni et al. Nutrition & Metabolism (2022) 19:8 inflammation in the uterine are susceptible to stimulus in adulthood [8,9].
According to research, 63.6% of women reported at least one infection during pregnancy, with 49.6% reported a respiratory infection, 20.5% a fever and 17.1% a urinary tract infection [10]. Therefore, low levels exposure of LPS through Gram-negative bacteria infection [11] is a common condition in pregnant women [12]. Previous studies showed that prenatal LPS exposure could result in prematurity, abortion [13,14] and intrauterine growth restriction [15]. Furthermore, metabolic disorders such as glucose tolerance impairment, insulin resistance [16] and obesity were observed in offspring exposed to prenatal LPS exposure [17][18][19]. However, the other metabolic disorders and potential mechanism beneath it remain to be elucidated.
In the study, we aimed to investigate whether prenatal LPS exposure led to metabolic disorders in offspring and the potential mechanism might contribute to it.

Animals and treatments
The female C57BL/6J mice (6 weeks old, 16-18 g) were purchased from Shanghai Laboratory Animal Co. Ltd. (SLAC, Shanghai, China). Animals were maintained in specific pathogen free (SPF) environment with a 12-h cycle (light on, 7:00 a.m.; light off, 7:00 p.m.), a constant temperature (20-25 °C), and food and water were attained voluntarily. Females were mated with males (2: 1) at 5:00 p.m. overnight, and the presence of a vaginal plug the next morning was confirmed as gestational day (GD) 0. Pregnant mice were divided randomly into LPS or saline group. The dose-response effect of LPS (Sigma Chemical Co., St. Louis, MO, USA) was evaluated in pregnant mice (Additional file 1: Table S1) in a pilot experiment to optimize the concentration and time window of exposure. A dose of 50 µg/kg LPS at GD15 for a one-time exposure was confirmed with an acceptable survival rate of pups without significant change compared with the rate of saline group (Additional file 1: Figure S1, Table S2). After delivery, the pups were randomly discarded to keep the litter size as five pups per cage. At weaning, we combined pups from different litters of the same maternal treatment for further study. Only male pups were kept for analyses in following study, since obesity was not observed in female in the pilot study (not shown). Maternal and neonatal baseline characters were described (Additional file 1: Table S2).
The offspring were weighted weekly until sacrificed. Indirect calorimetry was performed at 9 and 18 weeks old, and body composition was measured at 9 and 18 weeks old. Intraperitoneal glucose tolerance tests (IPGTTs) and insulin tolerance tests (ITTs) were both performed at 9 and 18 weeks old. Mice, which were fasted overnight beforehand, were anesthetized using isoflurane and euthanized via cervical dislocation at 20 weeks old. Blood samples were collected, and then kept at 22-26 °C for 2 h before centrifugation at 4 °C and 5000 rpm for 10 min for the analysis. Liver, retroperitoneal and gonad fat were quickly weighed and collected in liquid nitrogen for further RNA and protein preparations, and then stored at − 80 °C (Fig. 1A).

Food intake
The offspring's daily caloric intake was measured once per week from 6 to 18 weeks of age during the experiments using the following equation:

Body composition analysis
The body composition of the offspring was assessed at 9 and 18 weeks of age using magnetic resonance imaging (EchoMRI 900, Houston, TX, USA), according to the manufacturer's instructions. Fat mass (g), lean mass (g) and free-water mass were measured and calculated as percentage.

Indirect calorimetry
At 9 and 18 weeks old, the offspring were measured for 24 h by compressive animal lab animal monitoring system (Oxymax-CLAMS, Columbus Instruments, Columbus, OH, USA). Gas mixture was calibrated before the experiments. The animals were placed individually in a single cage chamber and maintained for 12 h in advance to avoid anxiety. Oxygen consumption, carbon dioxide release, energy expenditure, and respiratory exchange ratio (RER) data were collected. The following equations were used for the calculation of corresponding parameters [20]: Total activity was evaluated by adding the spontaneous activity (X-Amb) and rearing activity (Z-Tot).

IPGTT and ITT
For the IPGTT, the mice were fasted overnight (from 5:00 p.m. to 9:00 a.m.). Each received an intraperitoneal injection of glucose (2 g/kg). The glucose concentration was measured from the tail at certain times (0, 15, 30, 60, 90, and 120 min) using a glucometer (Roche Accu-Chek Inform, Switzerland). For the ITT, the mice were fasted for 6 h (from 8:0 a.m. to 2:00 p.m.). Each received an intraperitoneal injection of insulin (0.75 U/kg), and the blood glucose was measured at certain time points described above. The tests were performed at 9 weeks old and 18 weeks old.

Histological analysis of liver and adipocytes
The dissected liver and adipose tissues were fixed with 4% paraformaldehyde in phosphate-buffered saline for 16 h at 4 °C, sequentially dehydrated and embedded in paraffin. The tissue samples were then sectioned at a 6 μm thickness for hematoxylin-eosin staining and quantification. For the Oil Red O staining, the sections were incubated in 60% isopropyl alcohol, and then stained with filtered Oil Red O solution (3 mg/mL) for 30 min and rinsed twice with distilled water. Images were obtained (Nikon Eclipse Ti microscope, Tokyo, Japan) and quantified (Image J software, USA). Five areas for calculation of the area under the curve (AUC) were randomly chosen at the interval of every two visual fields under 400X amplification.

Quantitative real-time polymerase chain reaction (qPCR)
The total RNA was extracted from liver and fat tissues with Trizol reagent (Takara, Shiga, Japan). Then, quantitative real-time reverse-transcription PCR was conducted using SYBR green real-time PCR master mix reagent (Toyobo, Osaka, Japan) on the platform of QuantStudio 3 Real-time PCR System (Life Technologies, Inc., Carlsbad, CA, USA). This method was performed for all samples in triplicates. The primers are listed in Additional file 1: Table S3.

Measurement of lipid in mice liver
The total cholesterol (TC) (A110-1-1) and triglyceride (TG) (A111-1-1) in the mice liver were extracted with absolute ethyl alcohol and measured by commercial kits, according to the manufacturer's instructions (Jiancheng Bioengineering Institute, Nanjing, China). The TC and TG levels were determined from the optical density of each well at 510 nm wavelength using a microplate reader (Synergy H1, Biotek, VT, USA).

Metabolomics analysis Sample preparation
All chemicals were analytical or high-performance liquid chromatography-grade with water, methanol, formic acid, acetonitrile from (CNW Technologies GmbH Düsseldorf, Germany) and L-2-chlorophenylalanine from (Hengchuang Bio-technology Co., Ltd., Shanghai, China). The samples stored at − 80 °C were thawed at 22-26 °C and prepared as previous researches described [21].

Data processing and statistical analysis
Progenesis QI software (Waters Corporation,Milford, CT, USA) was applied for the raw data of LC-MS. Parameters were set as followed with the precursor tolerance as 5 ppm, the fragment tolerance as 10 ppm and the retention time (RT) tolerance as 0.02 min. Isotopic peaks were determined according to peak RT alignment in internal standard detection. 15% of the base peak intensity was set to identify the minimum response. The parameters included m/z, peak RT, peak intensities while RT-m/z pairs were applied for identify the ion. After trimming samples with missing values (ion intensity = 0), less than 50% of the samples were finally included. The metabolites were mapped to the database (http:// www. hmdb. ca/; http:// www. lipid maps. org/; and self-built databases) Progenesis QI (Waters Corporation) was used for the data analysis. The data were processed by the R package 'ropls' . Mean centering (Ctr) and Pareto variance (Par) were applied for initial data processing to visualize the alterations by principal component analysis (PCA) and (orthogonal) partial least-squares-discriminant analysis (O) PLS-DA, respectively. The Hotelling's T 2 region was used to define the 95% confidence interval of the variation. Variable importance in the projection (VIP) > 1 are selected for discrimination. In this study, after oneseventh of the samples was excluding from the model in each round where the default seven-round crossvalidation was applied to avoid overfitting. Metabolites with VIP values > 1.0 and P < 0.05 were remained as differential metabolites. Hierarchical clustering and correlation analysis were performed on top 50 statistically different metabolites. Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment was performed to identify the pathways. Enrichment with q value < 0.05 was considered significant. For lipid analysis, differential metabolites were identified with VIP > 1 and P < 0.1. KEGG enrichment was performed using MetaboAnalyst 5.0 (https:// www. metab oanal yst. ca/).

RNA sequencing
The total RNA of liver was extracted with RNAiso Plus Total RNA extraction reagent (Cat#9109, TAKARA), as instructions mentioned. RNA integrity was validated referring RIN number through an Agilent Bioanalyzer 2100 (Agilent technologies, Santa Clara, CA, USA). Then, qualified total RNA was purified with RNA Clean XP Kit (Cat A63987, Beckman Coulter, Inc. Kraemer Boulevard Brea, CA, USA) and RNase-Free DNase Set (Cat#79254, QIAGEN, GmBH, Germany).The library was constructed with the TruSeq Stranded mRNA LT Sample Prep Kit (Illumina, CA, USA) with 150-bp paired-end on the platform (Illumina). The depth of the sequencing was larger than 6-Gb. The quality of the sequenced reads was estimated by the Q-value, then trimmed and mapped to the reference genome GRCm38 using Hisat2 (https:// github. com/ Daehw anKim Lab/ hisat2). The abundance of the annotated genes was estimated using Stringtie (version 1.3.0, Johns Hopkins University, USA), normalized with the trimmed mean of M values, and the Fragments Per Kilobase Million value was obtained on the perl script. The differentially expressed genes (DEGs) were identified with a threshold as q value < 0.05, and log-2FoldChange ≥ 1.0 or ≤ -1.0. Gene ontology and KEGG enrichment was performed using the R package 'cluster-profiler65. ' Enrichment with q value < 0.05 was considered significant. All DEGs were involved in the Gene Set Enrichment Analysis (GSEA) to investigate potential biological processes in priori defined gene sets. (http:// softw are. broad insti tute. org/ gsea/ index. jsp).

Statistical method
The results were expressed as mean ± standard deviation (SD) or percentage (%) as needed. All the data were analyzed using Prism software version 8.0 (GraphPad software, San Diego, CA, USA) or R (version 3.6.3). The QQ-plots and Shapiro-Wilk test were applied for verifying the assumptions of normality before Student's t-test, while Mann-Whitney U test was used when data was abnormal distribution. A value of *P < 0.05, **P < 0.01, ***P < 0.001 were considered to be statistically significant.

Male offspring exposed to LPS developed obesity in middle age
Prenatal LPS exposure increased the body weight in the male offspring at weaning (P < 0.0001) (Fig. 1B). The weight of the LPS group remained higher slightly but not significantly until 14 weeks, while a catch-up growth was observed in the normal saline (NS) group. Interestingly, as the mice grew older, the LPS group gained more weight compared with that of the control group until being sacrificed. The food intake did not differ between the two groups through the life span, although fluctuations were observed at different weeks (Fig. 1C). The body composition measurement showed increased body fat rate and decreased lean weight in LPS-treated group at 18 weeks old (NS: (8.76 ± 0.00) × 100%; LPS: (10.88 ± 0.00) × 100%, P < 0.05) (Fig. 1E, F), yet not at 9 weeks (Fig. 1D). After sacrifice, the anatomical parameters were recorded. Body weight (LPS: 27.956 ± 0.788; NS: 26.468 ± 0.670; P = 0.0065), gonadal fat (LPS: 0.475 ± 0.128; NS: 0.364 ± 0.104; P = 0.0486) and perirenal and retroperitoneal fat (LPS: 0.140 ± 0.043; NS: 0.080 ± 0.046; P = 0.0153), especially liver (LPS: 0.965 ± 0.065, NS: 0.803 ± 0.075, P = 0.0231), were higher in the LPS group as compared to those in the control group (Table 1).

Lipid metabolism disorders in the serum of LPS-exposed male offspring
To further explore the metabolic disorders, a gas chromatography-mass spectrometer (GS-MS), based on untargeted metabolomics approach, was applied to the serum of mice. The primary bile acid synthesis (ID annotation: mmu00120) was enriched in Gene Ontology and KEGG analyses dramatically (Rich factor = 0.17, FDR = 4.24E−05) (Fig. 3F). The other metabolites enriched in the bile acid (BA) pathway are listed in Additional file 1: Table S4. Then, we investigated the pathway in detail and found that the serum levels of both unconjugated bile acid (UCBAs) and conjugated bile acids (CBAs) were decreased (Fig. 3G). A series of tauro-conjugated BAs consisting the majority of BAs in mice were dramatically declined, such as taurocholic acid (Fold Change (FC) = 0.0099, P = 0.04), taurochenodesoxycholic acid (FC = 0.0165, P = 0.04), taurodeoxycholic acid (FC = 0.0016, P = 0.04). To explore the potential reason for BAs' decrease, we measured gene expression of several key enzymes anticipated in primary bile acids synthesis in liver, including cytochrome P450 family 7 subfamily A member 1 (CYP7A1), cytochrome P450 family 27 subfamily A member 1 (CYP27A1) in the main stream and cytochrome P450 family 8 subfamily B member 1 (CYP8B1) in the bypass. However, all these enzymes were up-regulated in LPS group, which suggested negative feedback where BA's decrease was prior to the change in liver. Interestingly, the level of farnesoid X-activated receptor (FXR or Nr1h4), a nuclear transcription factor activated by BAs, was increased, too (Fig. 3H). Furthermore, citrate cycle (TCA cycle), arachidonic acid metabolism, and linoleic acid metabolism were enriched (Fig. 3F); the latter two unsaturated fatty acids play vital roles in many biochemical processes such as the balance of inflammation and anti-inflammation. To determine the change of the lipid composition and distribution, we isolated lipids to create a subset. The detected differentiated  lipid classes (VIP > 1, P < 0.1) were shown in Fig. 4A. We found 59 lipid species, consisting of 13 C24 BAs, seven amino acids, seven BAs, alcohols and their derivatives and other lipid classes. Functional enrichment analyses in KEGG pathways showed the levels of C24 BAs, lysophosphatidylcholine and amino acids were significantly enriched (Fig. 4B-D). Interestingly, most unsaturated fatty acids were decreased, especially eicosapentaenoic acid (EPA) (FC = 0.579), docosapentaenoic acid (DPA) (FC = 0.480) and α-linolentic acid (FC = 0.675), which are ω-3 essential fatty acids (Fig. 4E). Moreover, all androstane steroids were decreased in the LPS group, such as etiocholanolone (FC = 0.257), dihydrotestosterone (FC = 0.351) and androstanediol (FC = 0.271) (Fig. 4F). Prostaglandins (PGs), which are lipid signals derived from arachidonic acid that are produced by cyclooxygenase enzymes and are targeted by non-steroidal antiinflammatory drugs, were all decreased (Fig. 4G). All evidence indicated the occurrence of a lipid metabolism disorder, from triacylglycerol to cholesterol.

Change in the transcriptome in LPS-exposed male offspring
Abnormal lipid accumulation in the liver, a core organ of nutrient metabolism, was observed in previous results. Therefore, we performed RNA sequencing of the liver. When all the genes were involved, GSEA indicated the changes in adipogenesis, fatty acid metabolism and bile acid metabolism ( Fig. 5D-F, Additional file 1: Table S5). KEGG analysis with a threshold of adjusted P value < 0.05 and fold change > 2 showed enrichment in the digestive system as pancreatic secretion was highlighted. Moreover, fat digestion and absorption were also increased in the LPS group (Fig. 5H, I). In the DEGs identified by RNAseq, the expression of uncoupling protein-1 (Ucp1), a gene encoding a thermogenesis factor, was decreased (FC = 0.10 ± 0.04, P < 0.05); while those of carboxyl ester lipase (Cel) (FC = 8.15 ± 3.91, P < 0.05) and carboxylesterase 1G (Ceslg) (FC = 4.46 ± 1.79, P < 0.05), which are genes encoding lipid and protein digestive enzymes, were increased (Fig. 6A, B). We also explored the key enzymes of glycolipid metabolism. The increase in fatty acid synthase (Fasn) and glucose 6 phosphatase (G-6-p) levels was confirmed by RT-PCR (Fig. 6C, D). As the pathway analysis indicated, PPAR-γ, a member of the PPAR signing pathway and an upstream regulative factor of lipid metabolism (lipogenesis and cholesterol synthesis), which functioned as core metabolic regulative factor, was enriched slightly (Fig. 5H). We also examined the levels of mTOR, up-steam regulator of PPAR-γ, and found that the phosphorylation level was increased in the LPS group (Fig. 6E, F).

Discussion
In the present study, we confirmed that prenatal LPS exposure resulted in obesity, increased body fat percentage and impaired energy expenditure in middle-aged male offspring. Furthermore, abnormal lipid accumulation in the liver and adipocyte tissues was observed, accompanied with metabolites changed in serum such as unsaturated fatty acids, sex hormones and PGs. In addition, fat digestion and absorption were enriched in the LPS group in KEGG analysis. Moreover, the Ucp1 level was decreased, while the level of Cel and Ceslg was increased in the liver. We further investigated the key enzymes of glucolipid metabolism and found that Fasn and G-6-P were upregulated. Furthermore, we found mTOR/PPAR-γ pathway, an upstream regulatory pathway of both lipid metabolism  F). Spontaneous activity (G, H), Values were presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. RER respiratory exchange rate, EE energy expenditure and primary BA synthesis was upregulated in the liver. The schematic diagram of the study was shown (Fig. 9).

Obesity and energy expenditure
Prenatal LPS exposure could lead to metabolic disorders. LPS is an endotoxin produced by Gram-negative bacteria. When challenged by LPS, the type 1 immune response occurs, including secretion of inflammatory cytokines, infiltration of immune cells and subsequent adaptive response. Prenatal LPS exposure could result in abortion and prenatal morality [22,23], and litter size  F). Spontaneous activity (G, H), Values are presented as mean ± SEM. *P < 0.05; **P < 0.01; ***P < 0.001; **** P < 0.0001 might affect adult metabolism in rodents [24]. To minimize the effect of litter size, we randomly discarded pups to keep five pups in each litter. In line with our study, the exposure pregnant rats to LPS during mid-gestation led to increased body weight and adiposity in the offspring at 11 weeks old [19,25]. In Qin's study, the results showed the glucose level of offspring was impaired [26], while in Nilsson's work the glucose level remained intact [25]. In general, these studies indicate that prenatal LPS exposure could result in lipid and/or glucose metabolism disorders and sometimes obesity. The discrepancies among groups may be due to differences in the dosage, time and frequency of LPS exposure. Impaired energy expenditure might participate in the development of obesity. We decomposed the factors that might contribute to obesity as energy balance equation that energy input (EI) > energy output (EO). Then, the surplus energy is stored (ES) and converted to tissue mass (termed Ec). For two externally-modifiable components, the EI was represented by food intake. The EO, involving basal metabolic rate (BMR) and the energy spend on physical activity, was represented as an activity factor (PAF) multiplied by (BMR), as EO = PAF × BMR. Therefore, the equation for obesity becomes EI − (PAF × BMR + Ec) = dES/dt [27]. At 9 weeks old, LPS group manifested an increased respiratory exchange rate which indicated a preference for glucose as substrate for energy supply. At the same time, LPS group showed decreased spontaneous activity at night. However, the body weight, body fat percentage and energy expenditure were not affected. As grown old, no difference was found in the food intake and activity between the two groups at 18 weeks old instead of energy expenditure alternation in LPS-exposed male offspring. It was also accompanied with the decreased lipid oxidation and glucose oxidation, especially at night when rodent is physiological arousal. Interestingly, it also suggested that obesity in middle age might originate from metabolic disorders in early life. [28,29].

Metabolites change
Obesity in middle age of LPS exposed male offspring was companied with several metabolite change in serum such as bile acid, unsaturated fatty acid, sex hormone and prostanoid.
Bile acids are amphipathic steroid molecules and synthesized in the hepatocytes from cholesterol [30]. The rate-limiting enzymes in the bile acids synthesis includes CYP7A1, which produces the majority BA pool. Bile acid synthesis is negatively regulated via the FXR by induction of small heterodimer protein in the liver. In our study, all kinds of BAs in the serum were dramatically decreased in LPS-exposed male offspring when fasted while the expression of CYP7A1, CYP8A1 and CYP8B1 was upregulated in liver. The results indicated a secondary change from negative feedback. Bile acid has a close relationship with obesity. Researches have shown that low serum BAs are associated with fasting-induced hepatic steatosis in mice [31]. Some chemicals like theabrownin could elevate conjugated BA levels in the serum and ileum to alleviate the high-fat diet (HFD)-induced obesity [32]. However, in human, the total BA concentrations were found to be increased in obese, type II diabetic and nonalcoholic steatohepatitis patients [33]. Altogether, the discrepancy in the BA spectrum might due to the different models applied and the stage of the disease.
Omega-3 polyunsaturated fatty acids (PUFAs) have been known to be important in reducing the incidence of infection and inflammation [34][35][36]. Recent researches have confirmed the value of omega-3 PUFAs in reducing cardiovascular diseases and associated inflammation [37]. Interestingly, most unsaturated fatty acids like EPA, DPA, docosahexaenoic acid (DHA) and α-linolentic acid were decreased, while stearic acid, a saturated fatty acid was significantly increased in the LPS group, which indicated an inflammatory state accompanied with obesity.
PGs are derived from arachidonic acid and modulate diverse biological processes such as vascular permeability, hyperalgesia [38] and fever [39,40]. And it is rapidly degraded and inactivated by the lung and liver after releasing and functioning locally within 1-2 mins. It is generally believed that the PG is a tissue hormone rather than a circulating hormone. Besides, PGs are affected by body type [41] and eating state [42]. In our study, decrease of all PGs might result from the reduction of arachidonic acid levels which was the precursor of all PGs, since mice were starved for 24 h before euthanasia and all unsaturated fatty acids as EPA, DHA were decreased.
Previous researches showed sex hormones may play a role as sex-dimorphic etiology in the process of obesity [43]. In our pilot experiments, obesity was not observed in female offspring exposed to prenatal LPS exposure. In addition, all androstane steroid levels were decreased in the LPS group, such as etiocholanolone, dihydrotestosterone and androstanediol, while estrogen levels were slightly increased in the LPS group. Although we did not confirm the role of sex hormone in our study due to the limitation of sample size, several researches have filled the gap. Sex hormones are related to the obesity through energy metabolism [44], body fat distribution [45] and regulation of proteins in adipose tissues [46]. A report indicated that the total testosterone and sex hormonebinding globulin were negatively associated with diabetes in male, while the result from estrogen was inverse [47]. In Xu's work, prenatal LPS exposure led to an upregulated aromatase expression, reduced the androstanediol/ estradiol ratio and altered sex hormone receptor activity in glucose metabolism disorders in middle-aged male offspring [16]. Nonetheless, how sex hormones participate in the development of obesity needs further study.

Gene expression change
Gene Ontology and KEGG analysis of the DEGs in liver showed that pancreatic secretion was significantly enriched. The enzyme CEL, also known as bile saltdependent or -stimulated lipase, expresses mainly in pancreatic acinar cells and performs functions of hydrolyzing fat, cholesteryl esters and fat-soluble vitamins in the duodenum [48]. When inactivated, CEL can significantly reduce susceptibility to high fat/high cholesterol diet-induced obesity [49]. Ces1g encodes a kind of carboxylesterase responsible for the hydrolysis of many xenobiotics and endogenous substrates involved in fatty acyl and cholesterol ester metabolism. The global loss of Ces1/Ces1g avoids the development of atherosclerosis by inhibiting intestinal cholesterol and triglyceride absorption [50]. However, other researches showed contradicting results, wherein Ces1/Es-x knockout mice resulted in increased hepatic lipogenesis and hyperlipidemia, accompanied by fat deposition in peripheral tissues [51]. Furthermore, Ucp1 expression was decreased in the LPS group. UCP1 is a mitochondrial protein responsible for thermogenic respiration specialized in brown adipose tissue that participates in thermogenesis, temperature regulation and the regulation of energy balance. Altogether, the digestive enzymes were increased in the liver and might affect the absorption rate. At the same time, decreased expression of Ucp1 might partly explain the impaired energy expenditure. PPAR-γ is an important nuclear receptor that participates in lipid and cholesterol metabolism [52]. mTORC1, a core target that coordinates cell growth with the availability of nutrients, energy and growth factors participates in the PPAR-γ pathway [53,54]. However, the function of mTOR/PPAR-γ depends on the characteristic of model. In mice fed with HFD, rosiglitazone could reduce hepatic lipid accumulation via inducing PPAR-γ [55]. Furthermore, irbesartan can ameliorate hyperlipidemia and liver steatosis by upregulating the expression of PPAR-γ, activating the AMP-activated protein kinase/protein kinase B/mTOR signaling pathway and inducing liver autophagy [56]. On the contrary, studies found resveratrol and genistein attenuate hepatic steatosis in HFD-fed mice by repressing PPAR-γ [57,58]. In addition, ghrelin can promote lipogenesis via mTOR/PPAR-γ signaling pathway in hepatic cells [59]. In our study, PPAR-γ and p-mTOR/ mTOR ratio was upregulated in LPS-exposed offspring, which might lead to the abnormal lipogenesis and accumulation in liver. Since obesity can cause and exacerbate male-factor infertility by several means [60], the change of mTOR pathway in our study might affect male reproductive system [61], which needs further research.

Limitation and perspective
There are some limitations of present study. First, we couldn't fully determine the role of mTOR/PPAR-γ pathway in the development of obesity, as an initiator or a responsor. Thus, future research on mice pre-treated with mTOR/PPAR-γ inhibitor is necessary. Second, endocrine profiles should be verified in the future to determine which one contributes to the obesity dominantly. Third, several metabolites have changed in our study, but few has been studied in the context of obesity. It would be interesting to explore the relationship between metabolites and obesity.
In conclusion, our study demonstrated that prenatal LPS exposure led to obesity and abnormal lipid metabolism through impaired energy expenditure.