Mechanism of 1,25(OH)2D3 in the Treatment of Acute Lung Injury Based on UHPLC/Q-TOF MS-Based Metabolomics and Network Pharmacology

Background: Sepsis-induced acute lung injury (ALI), a high morbidity and mortality disease, still has no effective therapies. 1,25(OH) 2 D 3 is one of the indispensable nutrients in our body. The regulation mechanism of 1,25(OH) 2 D 3 in inammation has been recognized gradually. Network pharmacology was used wildly to broaden the understanding of diseases and advance drug discovery. In this study, we used network pharmacology and metabolomics to generate the potential mechanism of 1,25(OH) 2 D 3 on acute lung injury. Methods: We used metabolomics and network pharmacology to elucidate the therapeutic mechanism of 1,25(OH) 2 D 3 on acute lung injury. Serum samples, collected from mice with LPS-induced acute lung injury, were detected by UHPLC/Q-TOF MS to evaluate the differential metabolites from multiple metabolic pathways. Meanwhile, the H (cid:0) E staining, ELISA and QPCR were used to estimate the ecacy of 1,25(OH) 2 D 3 on acute lung injury. Results: The results of animal experiments showed that 1,25(OH) 2 D 3 could mitigate severe pulmonary edema and inammatory inltration caused by LPS, and the treatment of 1,25(OH) 2 D 3 reduced the levels of inammatory cytokines, interacted 25 related proteins and TNF signaling pathway, Toll-like receptor signaling pathway, PI3K-Akt signaling pathway. Conclusions: The integrated methods coupled with UHPLC/Q-TOF MS and network pharmacology provided a new way to study the potential mechanism of 1,25(OH) 2 D 3 on acute lung injury, which may provide a possible solution for patients with clinical acute lung injury.


Background:
Acute lung injury (ALI) and its severe form acute respiratory distress syndrome (ARDS) are two kinds of severe respiratory disorders clinically, which caused by many conditions, such as sepsis, severe trauma with shock, pancreatitis or inhalation of poisonous gases [1]. With a mortality over 40%, ALI and ARDS remain a leading cause of morbidity and mortality [2,3]. ALI is characterized with the disruption of the alveolar endothelial and epithelial barrier [4,5], which in turn leads to pulmonary alveolar and interstitial edema, impaired gas exchange and hypoxemia [6]. Although many efforts have been made, the mechanism of ALI is still unclear and there are no effective pharmacotherapies available that can improve survival of patients with ALI/ARDS. Therefore, an effective treatment of ALI must consider all these factors and the molecular networks behind them.
1,25-dihydroxyvitamin D (1,25(OH) 2 D 3 ) is a fat-soluble steroid hormone which regulates the calcium-phosphorus homeostasis. Growing evidences from epidemiological and basic researches reveal that 1,25(OH) 2 D 3 can modulate immune responses and many acute and chronic disorders [7]. 1,25(OH) 2 D 3 , the active form of vitamin D binds to vitamin D receptor (VDR) and retinoid X receptor (RXR) and the trimeric complex translocates into the nucleus, regulating the expression of more than 900 genes [8,9], regulating cell proliferation, differentiation and survival [10,11]. Recent studies have shown that 1,25(OH) 2 D 3 plays an important role in anti-infection and anti-in ammation [12]. In children and adults with asthma and chronic obstructive pulmonary disease, the levels of serum 25-hydroxyvitamin D are associated with impaired lung function, airway hyperresponsiveness and increased exacerbation frequency [13][14][15]. Besides, 1,25(OH) 2 D 3 de ciency has been found to be associated with lung injury, high-dose preoperative treatment with cholecalciferol reduced changes in postoperative pulmonary vascular permeability index effectively [16]. Despite the evidence linking 1,25(OH) 2 D 3 to lung diseases, there is little information concerning 1,25(OH) 2 D 3 and ALI and the potential targets are still unclear.
Because of the complex mechanisms of ALI, the traditional reductionist 'one drug, one target' cannot investigate all the mechanisms. With the rapid development of bioinformatics, metabolomics and network pharmacology have become effective methods to detect the molecular and pharmacological mechanisms of diseases and drugs, focusing on the multiple targets [17]. Network pharmacology can reveal the action mechanisms of the drug through the combination of computational biology, systems biology and "omics" related to target drugs [18]. Metabolomics, a new omics technique developed after genomics, transcriptomics, and proteomics, has developed rapidly. It focuses on the holistic investigation of multi-parametric metabolite responses of living systems and can be used to nd new biomarkers, gain a better understanding of metabolic pathways [19]. What's more, metabolomics combined with network pharmacology can be used to construct a network, containing drug-targets-pathways-disease, to detect the potential therapeutic mechanisms of drugs.
In this study, we sought to determine the potential therapeutic mechanisms of 1,25(OH) 2 D 3 on acute lung injury by metabolomics combining with network pharmacology. First, we identi ed the differential metabolites among mice in different groups and predicted the potential targets of these metabolites. Then we detected the GEO database for microarray data related to ALI. Since VDR is the receptor of 1,25(OH) 2 D 3 , we also detected the potential proteins related to VDR and examed the overlapping targets shared by above three target lists. Our work highlights the mechanisms of 1,25(OH) 2 D 3 treating ALI and facilitate the development of novel drugs.

Animal Models and Sample Collection
24 Balb/C mice (22-24 g) were purchased from Cavens Animal Corp (Changzhou, Jiangsu, China), and housed in speci c pathogen-free conditions (24-26℃, 50 ± 5% humidity) with a 12 h light/dark cycle. Animals had free access to standard diet and water. All animals were randomly divided into three groups: the control group (n = 8), LPS group (n = 8) and LPS + VD group (n = 8). The LPS group was established by injecting lipolysaccharide (LPS, Sigma-Aldrich, St. Louis, MO, USA) into peritoneal cavity (10 mg/kg). The LPS + VD group was established by treating calcitriol (2ug/kg) at 48, 24 and 1 h before LPS injection [20]. 24 h after LPS injection, all animals were euthanized and the lung tissues were collected for further analysis. All studies were reviewed and approved by the Animal Experiment Administration Committee of the Shanghai Pulmonary Hospital (Shanghai, China).

Hematoxylin and Eosin (H&E) Staining
Lung tissues were xed with 4% paraformaldehyde for 24 hours, embedded in para n, stained with hematoxylin-eosin (H E) using the standard protocol. Then, images were captured with a digital camera (Nikon, Tokyo, Japan) by pathologists blinded to the groups.

RNA Extraction and Real-Time Quantitative PCR
Total RNA from the lung tissues were isolated by TRIzol reagent (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions and reverse-transcribed into cDNA at 37℃ for 15 minutes using a commercial kit from Takara according to the standard protocol. The expression levels of target genes were detected by the SYBR Premix Ex Taq™ II (RR820A, Takara, Japan) with an ABI 7500 PCR system (Life Technologies, USA). Primers for RT-qPCR analysis synthesized by Shanghai Sangon Biotech Co., Ltd (Shanghai, China) and listed in Supplementary Table 1. Expression levels were calculated using the 2 −ΔΔCt method and normalized with the level of β-actin gene in the same sample.

ELISA Analysis
The quanti cation of sera cytokines (IL-1β, IL-6 and TNF-α) were assessed with Mouse ELISA MAX™ Standard Sets (BioLegend, San Diego, CA) according to the manufacturer's instructions using serum samples.

Metabolomic Analysis
Serum samples were collected at room temperature, ice-cold methanol containing 25ug/ml 2-chloro-L-phenylalanine was added to 100 ul serum, prepared as the internal standard to precipitate the protein and extract the metabolites. After vortexing for 1 min, the serum samples were mixed with the internal standard for 5 minutes, tranquilized for 10 minutes and centrifuged at 13000 rpm, 4℃ for 15 minutes. The supernatant was transferred into a new vial. The QC sample was prepared from all samples collected in this study for monitoring the data acquisition performance during analysis.
UHPLC-Q-TOF/MS analysis was performed using an Agilent 1290 In nity LC system equipped with an Agilent 6538 Accurate-Mass Quadrupole Time-of-Flight (Q-TOF) mass spectrometer (Agilent, USA). Chromatographic separations were performed at 40℃ using an Acquity UPLC HSS T3 column (2.1 mm × 100 mm, 1.8 um; Waters, Milford, MA, USA). The ow rate was 0.4 mL/min and the injection volume was 3 µL. 0.1% formic acid (A) and ACN modi ed with 0.1% formic acid (B) were composed in the mobile phase. The total run time for equilibration and the optimized UHPLC elution conditions were set as follows: 5%B at 0-2 min, 5-95%B at 2-17 min, 95%B at 17-19 min. The ow rate was 0.4 mL/min and the injection volume was 3 µL.
Mass spectrometry was operated in both positive and negative ion modes of operation. The optimized conditions used set as follows: capillary voltage, 4 kV for the positive mode and 3.5 kV for the negative mode; drying gas ow, 11 L/min; gas temperature, 350℃; fragment or voltage, 120 V; nebulizer pressure, 45 psig; and skimmer voltage, 60 V. Data were collected in the pro le mode and the mass range was set at 50 to 1,100 m/z. The biomarkers were further analyzed by MS/MS, and the collision energy was set at 10 to 40 eV.
The raw data in the instrument speci c format (.d) were collected and converted into a common (mz.data) format using the Agilent Masshunter Qualitative Analysis B.04.00 software (Agilent Technologies, USA), in which the ltration threshold of the high of the absolute peak was set to 500, and the isotope interferences were excluded. We used the XCMS program [21] (https://xcmsonline.scripps.edu) to identify peak extraction, retention time correction, RT alignment, and integration, in order to generate a visualization matrix. After 80% based on the principle of selection, frequency of more than 80% of the ions present in each group retained samples [22], and to correct the MS response shift, all detected ions in each sample were normalized to total intensity. The sample names, RT, and m/z pairs, were imported to SIMCA-P software (version 13.0, Umetrics, Umea, Sweden) for principal component analysis (PCA) and partial least squares discriminate analysis (PLS-DA). Variable importance plot (VIP) was used to select metabolites with the threshold value of 1. The statistical signi cant differences were analyzed by SPSS 17.0. P < 0.05 was considered statistically signi cant.

Network Pharmacology
The target proteins of the differential metabolites were searched for in the Search Tool for Interactions of Chemicals database (STITCH, http://stitch.embl.de). The gene expression data of LPS-induced ALI were searched from the GEO database (GSE1871), the differential gene was analyzed by GEO2R. P < 0.05, FC > 2 or FC<-2 were considered as signi cant differential genes. The proteins related to VDR were obtained from the Search Tool for the Retrieval of Interacting Genes database (STRING, https://string-db.org/). A VDR-metabolites-target-disease network was constructed using Cytoscape (http://cytoscape.org) software.

Protein-protein Interaction Data
The protein-protein interaction (PPI) data was obtained from STRING database (https://string-db.org/) [23]. STRING database provides information regarding the experimental and predicted interactions of these proteins.

Gene Ontology and Pathway Enrichment
Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis were performed with the STRING Database (string-db.org/). GO enrichment analysis and pathways with False Discovery Rate (FDR) < 0.01 were de ned as enriched terms and pathways. R software (R 3.6.0 for Windows) was used to chart the bubble plot in KEGG enrichment analysis.

Statistical analysis
Data were expressed as mean ± standard error of the mean (SEM) and were assessed by one-way ANOVA followed by a Tukey's post-test using SPSS 17.0 statistical software (SPSS Inc., Chicago, IL, USA). Values of P < 0.05 were considered statistically signi cant.

Scheme for precise investigation of calcitriol against acute lung injury
The pharmacologic effect of calcitriol against ALI was evaluated in vivo (Fig. 1). Firstly, the physiopathological changes of mice lung tissues and the changes of pro-and anti-in ammatory cytokines were measured. Then, the changes of the differential metabolites in mice were measured by metabolic pro ling analysis. Differential metabolites-related proteins, disease-related proteins from GSE1871 and vitamin D receptor-related proteins were obtained from databases. Finally, signatures biomarkers corresponding to the speci c e cacy of calcitriol can be obtained.

Effects of calcitriol on lung injury and production of pro-and antiin ammatory cytokines
The representative images of hematoxylin and eosin (H E) staining for each group are illustrated in Fig. 2A. In LPS group, treated with 10 mg/kg LPS, severe pulmonary edema and in ammatory in ltration were detected. It is noteworthy that when treated with calcitriol (2ug/kg), the lung injury was attenuated signi cantly.
The levels of pro-in ammatory cytokines in serum such as IL-1β, IL-6 and TNF-α were signi cantly higher than control group after LPS administration, while calcitriol treatment effectively decreased these pro-in ammatory cytokines in serum (Fig. 2B-D). The pro-in ammatory cytokines in the lung tissues were detected by RT-qPCR analysis, just as the trends in serum, the levels of pro-in ammatory cytokines were up-regulated and anti-in ammatory cytokine (IL-10) were down-regulated after LPS administration. Conversely, calcitriol pretreatment attenuated the changes in LPS group. Compared with LPS group, lower levels of pro-in ammatory cytokines and higher levels of anti-in ammatory cytokine were shown after calcitriol pretreatment ( Fig. 2E-H).

Metabolic pro ling analysis
The PLS-DA model was used to investigate the general interrelation among control group, LPS group and LPS + VD group to observe the clear separation in positive and negative ion models. In the PLS-DA model, R 2 Y and Q 2 [cumulative (cum)] were parameters used for evaluation of the models, indicating the prediction ability. R 2 Y and Q 2 (cum) in the PLS-DA model were 0.804 and 0.96 in the positive mode and 0.548 and 0.693 in the negative mode, respectively. The result of the permutation test is shown in Figure S1. As shown in Fig. 3A-F, S-plot and variable importance plots (VIPs) were used to select the potential biomarkers in control group, LPS group and LPS + VD group. The points farther away from the origin point were selected as potential biomarkers with higher possibility.

Identi cation of the potential biomarkers
Among these signi cant differential metabolite ions related to ALI, we identi ed 49 differential metabolites altered in LPS group, and the trends of 16 metabolites were reversed by calcitriol intervention. These reversed metabolites were regarded as the biomarkers related to the protective effect of calcitriol ( Table 1). As shown in Fig. 4A-B, all identi ed differential metabolites in three groups were presented in the heatmaps both in positive (A) and negative (B) modes. Moreover, metabolic pathway analysis of these metabolites were shown in Fig. 4C.

Network pharmacology analysis
Reversed differential metabolites were selected to do the further analysis. The protocol of the integrated systems pharmacology approach is listed in Fig. 1. The predicted proteins targeted by reversed differential metabolites was detected in STITCH database (http://stitch.embl.de) and 2202 proteins were obtained (Supplementary table 2). GEO database mining was used to search genes related to LPS-induced ALI. We screened the database of LPS-induced acute lung injury as the sample and GSE1871 were selected nally. We used the GEO2R to analysis the differential genes between vehicle group and LPS group and genes with signi cant difference (P < 0.05, FC > 2 or FC<-2) were compared with the metabolites-related proteins. Since calcitriol is the active ligand of vitamin D receptor (VDR) and the VD 3 -VDR axis has been shown to regulate both innate and adaptive immune responses, we used STRING database (https://string-db.org/) to detect the protein targets of VDR. In Fig. 5A, after combining these target proteins together and take their intersection, 25 proteins were obtained ( Fig. 5A and  Fig. 5B C and helpful to explore the protective effect of calcitriol on ALI by acting on these targets and their related pathways.

Functional enrichment analysis of the target proteins
To further understand the function and role of calcitriol in treating ALI, Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis were performed by STRING database (https://string-db.org/). For GO term, the top ten enriched biological process terms, molecular function terms and cellular component terms were shown in Fig. 6A-C. KEGG pathway analysis further showed that the proteins involved in TNF signaling pathway, Toll-like receptor signaling pathway, PI3K-Akt signaling pathway and so on. The top 20 pathways were shown in Fig. 6D in the form of bubble plot.

Discussion
The diseases of acute lung injury (ALI) and acute respiratory distress syndrome (ARDS) still keep a poor prognosis, which leads to high mortality every year worldwide. It is important to gure out the potential mechanisms of ALI and nd better treatments for patients. LPS can regulate the lung injury involving biomacromolecule synthesis, hepatocyte apoptosis stimulation, free radical generation, lipid peroxidation and other various mechanisms. In this study, high levels of proin ammatory cytokines were found in serum and lung tissues from animal model of acute lung injury induced by LPS. When treated with calcitriol, the expression of these pro-in ammatory cytokines were decreased and anti-in ammatory cytokines were increased signi cantly.
The role of calcitriol on ALI has not been clear till now. It is generally accepted that vitamin D has an effective therapeutic role in sepsis and calcitriol, the active form of vitamin D, has been reported to induce host defense [24]. Since ALI is characterized with the disruption of the alveolar endothelial and epithelial barrier [4,5], vitamin D has been reported to attenuate lung edema via blocking the Ang-2-Tie-2-MLC kinase cascade and the renin-angiotensin system to repair alveolar endothelial and epithelial barrier [25]. Meanwhile, several researches have demonstrated that 1,25(OH) 2 D 3 , the hormonal form of vitamin D, has been shown to reduce vascular permeability and ameliorate pulmonary edema [25][26][27]. Nie et al. showed that vitamin D augments transalveolar uid clearance via unregulating the expression of a-ENaC [28]. Moreover, vitamin D has been reported to alleviate seawater aspiration-induced lung injury via inhibiting nuclear translocation of NF-κB and membrane translocation of RhoA and protecting alveolar epithelial and pulmonary microvascular endothelial barrier [27].
In the current study, metabolomics was used to detect the changes of differential metabolites in the three groups. A total of 49 metabolites were identi ed as the differential metabolites and the trends of 16 metabolites were opposite in LPS group and LPS + VD group. These metabolites were selected to detect their target proteins via STITCH database. We integrated these target proteins with differential genes in GSE1871 and target proteins of VDR. Finally, a total of 25 proteins were selected and considered to be closely related to the mechanism of calcitriol ameliorates ALI. STRING database was used to demonstrate the GO analysis and KEGG pathway analysis.
From the KEGG pathways, TNF signaling pathway, PI3K-Akt signaling pathway and Toll-like receptor signaling pathway are closely related to the mechanism of calcitriol ameliorates ALI.
TNF signaling pathway is one of the critical pathways in the development and maintenance of in ammation [29] and it has been found to provoke the release of in ammatory cytokines in ALI and mediate the following innate immune response and in ammatory process [30]. Previous study demonstrated that TNF is initiated by TNF-α receptor 1 (TNFR1), its activation induces the increased expression of nuclear factor kappa B (NF-kB) [31]. Zhang et. al. has indicated that 1,25(OH) 2 D 3 downregulated TNF signaling pathway to alleviate osteoarthritis [32].
Toll-like receptor signaling pathway functions as molecular patterns in adaptive or innate immunity and the activation of tolllike receptor signaling pathway leads to the production of costimulatory factors and cytokines, which is one important reasons contributing to ALI [33]. The normal activation of TLRs is critical to defense molecules in the recognition of pathogens, including M. tuberculosis [34][35][36]. In HIV-seropositive individuals, vitamin D rescues impaired Mycobacterium tuberculosismediated tumor necrosis factor release in macrophages by enhancing toll-like receptor signaling pathway [37]. However, uncontrolled excessive TLR activation may lead to serious infection [38]. Multiple lines of evidence showed that ARDS is associated with the vigorous activation of TLR2, TLR3, TLR4 and TLR9 [39] and noninfectious ALI is associated with the activation of TLR2 and TLR4 [40]. In addition, the survival rates of ARDS patients were correlated with downregulation of TLR1, TLR4 and TLR5 [41]. One pharmacologic strategy to manage the excessive in ammation in ALI is to regulate over activation of Toll-like receptors (TLRs) [42].
PI3K-Akt signaling pathway is expressed in eukaryotes widely, which plays critical roles in growth, differentiation, proliferation, and survival [43,44]. activates PI3K-Akt signaling pathway to protect podocytes from apoptosis [47]. Studies showed that the phosphorylation of Akt downregulates the downstream pro-apoptotic factor Bax and upregulates the anti-apoptotic factor Bcl-2, which inhibits cell apoptosis [48]. LPS-induced ALI is highly associated with cell apoptosis [49] and thus in the treatment of calcitriol, PI3K-Akt signaling pathway plays a critical role.
In conclusion, in this study, the pharmacologic effect of calcitriol against LPS-induced ALI was evaluated in vivo and with the combination of metabolomics, GEO database and network pharmacology, we found out 25 related proteins which were associated to the protective effects of calcitriol. GO and KEGG analysis were used to classify these identi ed proteins and we selected TNF signaling pathway, PI3K-Akt signaling pathway and Toll-like receptor signaling pathway to be the potential pathway which are closely related to the mechanism of calcitriol ameliorates ALI. Since the network pharmacology strategy produces many possible targets, these differential targets should be thoroughly tested in our further studies.

Conclusions:
In this study, the pharmacologic effect of 1,25(OH) 2 D 3 against ALI was evaluated in vivo. Then we performed metabolomics to nd out 16 differential metabolites which were associated with the protective effects of 1,25(OH) 2 D 3 against ALI.
Meanwhile, STRING database and GEO database was used to nd out the differential metabolites-related proteins and disease-related proteins. GO and KEGG analysis were used to classify these identi ed proteins. Further studies will be performed about the differential proteins to understand the mechanism of 1,25(OH) 2 D 3 against ALI and the potential clinical applications.

Declarations
Ethics approval and consent to participate: All studies were reviewed and approved by the Animal Experiment Administration Committee of the Shanghai Pulmonary Hospital (Shanghai, China).

Figure 1
Scheme for signature-oriented precise investigation of calcitriol against sepsis-induced ALI. E cacy evaluation of calcitriol was measured by H E and in ammatory cytokines. Signatures of sepsis-induced ALI and calcitriol pretreatment were measured by metabolomics and network pharmacology. Shared proteins, targets and pathways were obtained by combining these target proteins together.   Topological network schematic of differential metabolites-related proteins, disease-related proteins and VDR-related proteins. (A) A venn diagram of proteins in these three groups and 25 proteins were shown in all three groups. (B) Protein-protein interactions networks of proteins targeted by differential metabolites, genes associated with acute lung injury and genes associated with VDR. (C) Metabolites-targets-disease network. The hexagon outside represent the differential metabolites; the nodes in the middle part represent the proteins overlapping proteins. Nodes with deeper color means they have more interactions with other nodes.