RNA-seq proles of LPS-induced transcriptional changes in LBP-decient rat and its possible implications for the liver dysregulation during sepsis

Sepsis is an organ dysfunction caused by the dysregulated inammatory response to infection. LBP binds to LPS, and modulates the inammatory response. Rare systematic study has been reported to detect the effect of LBP gene during the LPS-induced sepsis. Herein, we explored the RNA sequencing technology to prole the transcriptomic changes in liver tissue between LBP-decient rats and WT rats at multiple timepoints after LPS administration. We compared the serum ALT levels using the biochemistry analyzer and proceeded RNA sequencing of liver tissue to search differentially expressed genes and enriched biological processes and pathways between LBP-decient and WT groups at 0 h, 6 h, and 24 h. In total, 168, 284, and 307 differential expressed genes (DEGs) were identied at 0 h, 6 h, and 24 h respectively, including Lrp5, Cyp7a1, Nfkbiz, Sigmar1, Fabp7, and Hao1, which are related to the inammatory or lipid-related process. Functional enrichment analysis revealed that inammatory response to LPS mediated by Ifng, Cxcl10, Serpine1, and Lbp was enhanced at 6 h, while lipid-related metabolism associated with C5, Cyp4a1, and Eci1 was enriched at 24 h after LPS administration in the WT samples. The inammatory process was not found when the LBP gene was knocked out, lipid-related metabolic process and PPAR signaling pathway mediated by Dhrs7b and Tysnd1 were signicantly activated in LBP-decient samples.


Background
Sepsis is a life-threatening disorder accompanied with organ dysfunction [1], which remains the leading cause of mortality in critically ill patients [2]. Despite years of intensive study and advances in the pathogenesis and supportive care, sepsis is still an enigmatic disease and a horrendous nancial burden for the healthcare system [3].
LPS, a major constituent of the outer cell wall of gram-negative (GN) bacteria, is considered to be the most important activator in the pathogenesis of sepsis, of which minute amounts can initiate the molecular mechanisms of the innate immune response [5,6]. According to the previous study, the dysregulated in ammatory response initiated by the interaction between lipopolysaccharide-binding protein (LBP) and lipopolysaccharide (LPS) is closely related with the development of sepsis [4]. LBP is a class I acute-phase protein primarily synthesized by hepatocytes [7]. It rstly recognizes LPS released from infecting pathogens by forming a high-a nity complex. The LBP-LPS complex is transfered to cluster of differentiation 14 (CD14) and toll-like receptor (TLR) 4 to trigger the release of in ammatory cytokines [8]. Additionally, increasing evidence indicated that lipid metabolism is correlated with the host's pro-in ammatory status. The in ammatory response is promoted by both obesity and high-fat meals, which may alter the intestinal barrier via affecting the gut microbiota to translocate LPS into the bloodstream [9,10]. LBP acts catalytically to facilitate binding of LPS to lipoproteins such as very lowdensity lipoprotein, low-density lipoprotein, and high-density lipoprotein, which also represent an important mechanism in host defense to inactivate with LPS [11,12] To date, rare systematic study has been reported to detect the transcriptomic changes in liver tissue to investigate the role of LBP in LPS induced in ammatory response. To this end, we explored the RNA sequencing technology to compare gene expression pro ling between LBP-de cient groups and WT groups at 0 h, 6 h, and 24 h after LPS infection, and identi ed candidate genes, biological processes and signal pathways which functionally related to sepsis, providing new clues for clinical treatment.

Hepatocellular Damage in Normal and LBP-De cient Rats after LPS-Administration
Inferior vena cava blood was collected from 3 normal and 3 LBP-de cient rats after LPS-induced at 0 h, 6 h, and 24 h respectively, then we analyzed the serum levels of ALT. In the normal groups, the serum ALT level reached a peak at 6 h after LPS administration ( Fig. 1), suggesting the most severe liver injury. In contrast, less hepatocellular damage was observed in consistent with the obvious decreases in serum levels of ALT in LBP-de cient samples (Fig. 1).

Mapping and Annotation of RNA Sequencing Reads
The RNA sequencing technique integrated with bioinformatics analysis was used to characterize alteration in liver gene expression between WT and LBP-de cient samples triggered by LPS-induced systemic in ammation, and the analyze steps were sown in Fig. 2A. We obtained about 69.7 million (M) of 150 bp paired-end reads for each sample (ranging from 57.2 to 107.2 million reads) (Additional le 1).
To evaluate the segregation between WT and LBP-de cient samples during the different time after LPS administration, we conducted the neighbor-joining tree of samples based on the expression of all genes. As shown in Fig. 2C, a clear divergence between the time of LPS-treated (0 h, 6 h, and 24 h) was observed in this tree, and WT and LBP-de cient rats were also de ned their respective separate clades, suggesting high delity of our RNA-Seq data.

Systemic Administration of Bacterial LPS Induces Global Changes in the Liver Transcriptome
To further characterize the DEGs from our RNA-Seq data, an analysis was performed to screen DEGs with P-value less than 0.05 and log2(fold change) higher than 1.5 using DESeq2 R package [14] (Fig. 3). In total, we identi ed 168, 284 and 307 signi cantly alternative genes respectively during the time of 0 h, 6 h, and 24 h between the normal and LBP-de cient samples. Then we clustered these DEGs via hierarchical heatmap ( Fig. 3A-C) to depict the differential expression gene pro le between the normal and LBP −/− rats. The most signi cantly DEGs with P-value < 0.001 and log2(fold change) > 1.5 were labeled in the volcano plots ( Fig. 3D-F, Additional le 2). Among that, Lrp5 [15], Cyp7a1 [16], Nfkbiz [17], Sigmar1 [18], Fabp7 [19], and Hao1 [20] (Additional le 3) have been reported in in ammatory response and lipid metabolic process, suggesting these genes may play an important role in modulating sepsis-induced system in ammation in WT and LBP-de cient rats after LPS injection.

Gene Annotation and Gene Ontology Analyses of DEGs
To further study of signi cantly overrepresented gene ontology terms involving these DEGs during 0 h, 6 h, and 24 h after LPS administration, functional annotations were performed with the DAVID Bioinformatics Resources 6.7 (https://david-d.ncifcrf.gov/) respectively [21]. Selecting from the full enrichment data sets (Fig. 4A), we found ten representative terms with the exhibition of strong differential enrichment patterns were mainly related to in ammatory response, immune response and lipid metabolic processes (Fig. 4B). In further, to detect the most associated DEGs during those biological processes between healthy and LBP-de cient groups, we exhibited associated genes evolved in the representative terms and pathways (Fig. 5). Interestingly, we found that in the normal rats, LPS strongly upregulated genes involved in the processes of the in ammatory response and immunomodulation including Ifng [22], Cxcl10 [23], Serpine1 [24], and Lbp [25] (Additional le 4A-D) at 6 h after LPS injection, then proceed in lipid metabolic response including C5 [26], Cyp4a1 [27], and Eci1 [28] (Additional le 4E-G) at 24 h (Fig. 5A). And the enriched pathways were in accordance with the results of gene ontology (Fig. 5B), which revealed that in ammatory pathways containing the toll-like receptor signaling pathways and natural killer cell-mediated cytotoxicity were enhanced in the normal groups at 6 h and lipid-related metabolism of peroxisome proliferator activated receptor (PPAR) signaling pathway was enriched at 24 h after LPS administration. Conversely, the functional enrichment of DEGs in LBP-de cient groups predominantly activated the lipid metabolic response instead of in ammatory or immunological response during the rst two time points, enriching some up-regulated genes such as Dhrs7b [29] and Tysnd1 [30] (Additional le 4H-I). And the PPAR signaling pathway was signi cantly over-represented at the rst two time points, which suggesting modulating in ammation and bacterial killing after LPS challenge with the de cient of the LBP gene [31]. Interestingly, the DEGs both in healthy and LBP −/− groups were over-represented in the processes of lipid metabolic and repeatedly enriched genes of Eci1, Pnpla3 [32], Apoa5 [33], and Fabp1 [34] (Additional le 4J-L).

A Proposed Model of the Roles of NF-κB and PPAR Signaling Pathways in the WT and LBP-De cient Rats after LPS Challenge
Based on the biological functions of above-mentioned genes and previous studies of nuclear factor kappa B (NF-κB) and PPAR signaling pathways, we presented a proposed model for the development of sepsis in rats ( Fig. 6). At 6 h, the upregulation of Ifng, Cxcl10, Serpine1, and Lbp in WT rats trigger NF-κB signaling pathway induced in ammation response after LPS injection. And the activation of NF-κB signaling pathway is responsible for modulating the immune reaction via enhanced biosynthesis of large quantities of pro-in ammatory molecules, including cytokines, adhesion molecules, etc., which are frequently induce sepsis and cause tissue damage when their production is dysregulated and excessive [35,36].
At 24 h, PPAR signaling pathway was found and may function as bacterial clearance via the formation of NET by highlighted genes of C5, Cyp4a1, and Eci1 in SD rats, and enhanced Dhrs7b and Tysnd1 in the LBP −/− rats after LPS administration. Just as reports revealed, PPARs are a large superfamily of nuclear receptors and incorporate three isoforms (PPAR-α, PPAR-β, and PPAR-γ), which are broadly involved in the regulation of metabolism, especially associated with lipid and glucose homeostasis [37,38]. In the process of activating pathway, PPAR-γ negatively regulates the activity of the transcription factor to inhibit the expression of pro-in ammatory mediators such as tumor necrosis factor alpha (TNF-α), interleukin 12 (IL-12), and adhesion molecules which results in anti-in ammatory outcomes in the setting of sepsis induced by LPS [39].
Together, the proposed model re ects that invading LPS may interplay with LBP to activate NF-κB signaling pathway and trigger uncontrolled in ammatory response. However, when inhibiting the activity of NF-κB, lipid-related metabolism would make bacteria removal via the effect on PPAR signaling pathway in the absence of LBP gene.

Administration
In this study, we adopted RNA sequencing to con rm that LBP expression level was elevated after LPS treatment in vivo (Additional le 4D), following the previous research that LBP delivers LPS to CD14 and TLR4 and nally triggers a cascade of events including the translocation of NF-κB to the nucleus and the initiation of the production and release of in ammatory cytokines via the activation of TLR-4 signaling pathway [40].
PPAR-γ stimulated with correlative ligands performed anti-in ammation activity via down-regulating NF-κB actions and subsequently inhibiting the expression of in ammatory mediators, such as TNF-α, IL-12, and adhesion molecules [31]. It could be deemed that the diminished liver in ammation and injury in the normal groups at 24 h may be performed by modulating the in ammatory response through PPAR-γ signaling pathway. Additionally, combined with the enrichment results of functional annotations and pathways, we surmised that given at the time of resuscitation, LBP-de cient rats would reduce liver injury by enhancing bacterial clearance through PPAR-γ signaling pathway, as reported that the activation of PPAR-γ increased the formation of neutrophil extracellular traps (NET) containing neutrophil, histones, and granule proteins, which may potentially propose a protective mechanism of bacterial elimination in the LBP-de cient group [31].

Poor Effects of Anti-in ammatory Therapies during Sepsis
Our ndings support that in ammatory response is closely associated with liver injury, which can further demonstrate that dysregulated in ammatory response exerts a crucial part in the development of sepsis [4]. However, during the last decades, the effects of many clinical trials testing anti-in ammatory approaches on patients with sepsis were rather disappointing. Gordon et al. [41] suggested that AZD9773, a polyclonal fragment antibody which has the effect of decreasing the concentration of TNF-α in circulation, was short of clinical bene t. Steven et al. [42] demonstrated that Eritoran did not improve survival among patients with sepsis shock as the antagonist of the MD2-TLR4 receptor for treatment. The administration of a high dose of corticosteroids, anti-in ammatory agents that globally depress the activity of the immune system and reduce the damage from cytokines and neutrophils, also failed to bring about improving outcomes for patients with sepsis [43]. Thus, it can be possible to conclude that the development of sepsis in humans is not merely the modulation of the in ammatory response, more comprehensive exploration concerning complicated molecular mechanisms requires undertaking for the more effective clinical treatment.

The SNPs in LBP and the Potential of LBP as a Biomarker in Clinical Application
The mechanisms mediated by LBP is a crucial player in the production of sepsis and related metabolic disorders, which makes it rational to suppose that single nucleotide polymorphisms (SNPs) within LBP gene might be determinants for interindividual susceptibility. Eckert et al. [44] previously found the rs2232613 polymorphism, leading to the substitution of proline with leucine at position 333 of LBP protein, was associated with a reduced ability to bind LPS or induce cytokines in vitro. The phenotype of individuals carrying the rs2232618 (Phe436Leu) had signi cant relevance with the higher incidence of sepsis and multiple organ dysfunction [45]. Another study also showed that susceptibility to severe sepsis was strongly correlative with a common haplotype from the 5'-anking region of the LBP gene [46]. Additionally, a foregoing study reported the rs2232592 polymorphism, located in the intron of LBP, were signi cantly related to type 2 diabetes [47]. Brie y, polymorphisms within the LBP gene might have an intensive association with sepsis and metabolic risk, which emphasize the immense potential of LBP in clinical application.
It has been performed that the up-regulation of LBP was widely observed in patients with severe infectious diseases [48] and increased circulating LBP-levels are correlative with the severity of sepsis [49], suggesting LBP may serve as a valuable biological marker for diagnosis and prognosis of patients with sepsis. However, previous reports showed LBP provided little clinical favorable information. Compared to other traditional biomarkers, such as procalcitonin and C-reactive protein, LBP has a moderate degree of diagnostic accuracy for sepsis [50]. Similarly, Sakr et al. [48] demonstrated that LBP moderately discriminated patients without infection from patients with severe sepsis.
What worth noting is that although LBP concentrations may weakly correlate with the severity and outcome of sepsis, circulating LBP was elevated when it came to obesity, metabolic syndrome (MetS), and type 2 diabetes in apparently healthy Chinese [51]. According to the investigation concerning the association between LBP levels and 6-year incident MetS, Liu et al. [52] suggested that LBP was positively correlatively with the increased 6-year risks of MetS among middle-aged and older Chinese. Besides, higher LPS or LBP concentrations could be observed than in diabetic subjects in healthy controls [53]. In short, LBP might be a promising biomarker of metabolic endotoxemia, but future prospective studies are still recommended for elucidating the potential biological mechanisms.

Conclusions
Taken together, to the best of our knowledge, we present here the rst comprehensive pro le of gene expression between the healthy and LBP-de cient rats after LPS-induced at multiple time points using RNA sequencing technology. With all these data surrounding the in uence of sepsis evoked by acute administration of LPS, we reported a list of genes that tremendously altered in the liver tissue. Most importantly, we emphasized the modulation of uncontrolled in ammatory response triggered by the NF-κB signaling pathway and bacterial elimination via lipid-related metabolism with the effect of the PPAR signaling pathway, which may as the potential reason for the alleviated in ammatory response and the attenuated liver damage and mortality of rats. And we also exhibited the proposal model to explain the genetic mechanisms in LBP −/− rats after LPS challenge, which may have more biological and clinical implications. However, further and ongoing in vivo studies are still required to con rm the proposed model and the candidate genes to ultimately validate the functional role of these ndings.

Experimental Animals and Tissue Collection
A total of 18 male SD and LBP −/− rats (body weight 230 ± 20 g) were used in this study. SD rats were originally provided by Beijing Vital River Laboratory Animal Technology Co., Ltd., and LBP −/− rats were purchased from Nanjing Biomedical Research Institute of Nanjing University, which had the same genetic background as SD rats. All animals were housed under standard animal care conditions and had free to access to water and rat chow ad libitum. Thick corn cob padding and nest material was used for enrichment of housing environment. Animals were acclimatized for 7 days before treatment. All procedures were carried out according to the Animal Welfare legislation of China. Animal experiments were approved by the ethics committee of Anhui Medical University. All the treatment were performed under inhalation anesthesia using vaporized iso urane (Raymain, Shanghai, China). The anesthesia was induced in a chamber and maintained using a face mask with a 0.5L/min oxygen ow mixed with 3% iso urane. The injection, and operation started when the rat had no more pain re exes, e.g. no response to clamping the skin using surgical forceps.
SD rats and LBP −/− rats were divided into the control and treated groups respectively (n = 9 per group) and anesthetized. Rats were challenged with a sub-lethal LPS injection (2 mg/kg, intravenous injection, E. coli serotype O55:B05 type, Sigma-Aldrich, St. Louis, USA). Meloxicam (0.2 mg/kg, subcutaneous injection, targetmol, ) was administered to achieve the postoperative analgesia. Penicillin was not applied considering no open wound and low possibility of infection within 24 h after LPS administration. At 0 h (n = 3), 6 h (n = 3), and 24 h (n = 3) after LPS administration, rats were sacri ced under 5% iso urane (Raymain, Shanghai, China). Blood was taken from the inferior vena cava, and the liver tissues were collected, and used for succedent transcriptome sequencing and data analysis. Subsequently, all the rats were euthanatized with 5% iso urane (Raymain, Shanghai, China).
Phenotypic values were presented as mean ± standard deviation (M ± SD). Statistical comparisons of phenotypic values between the experimental and normal groups were conducted by the Student t-test. The statistical difference was considered as signi cant at P < 0.05 and highly signi cant at P < 0.01.

Liver Enzyme
To investigate the hepatocellular injury in normal and LBP-De cient rats after LPS-induced, we took venous blood from the cavity and measured the levels of serum alanine transaminase (ALT) using an Automated Chemical Analyzer (Bayer Advia 1650; Leverkusen, Germany).

RNA Extraction and Sequencing
Total RNA was extracted from 100 mg of adipose tissue from three LBP-de cient experimental individuals and three normal individuals using the RiboPure kit (Ambion, Austin, USA) according to the manufacturer's protocol. The RNA integrity was assessed by an Agilent Bioanalyser 2100 and RNA Nano 6000 Lab chip kit (Agilent Technologies, USA). Sequencing libraries were generated using the NEBNext UltraTM Directional RNA Library Prep Kit (Illumina, USA) following the manufacturer's recommendations, and index codes were added to attribute sequences to each sample. Then the paired-end sequencing of the libraries was constructed on a Hi-Seq 4000 platform (Illumina, USA) via Novogene (Novogene, USA). The resultant data will be deposited at NCBI Sequence Read Archive (SRA) database upon acceptance.
First, the RNA-seq reads were discriminated based on the indexing adaptors. Low-quality reads and those containing ploy-N were then removed from raw data using FastQC v0.11.7 (http://www.bioinformatics.bbsrc.ac.uk/projects/fastqc). Next, the ltered reads were mapped against the chicken reference genome Gallus_gallus-5.0 (Ensembl) using STAR-2.5.3a [13], a fast splice junction mapper for short and long RNA-seq reads to a reference genome using uncompressed su x arrays.
Parameters of STAR were set to only allow unique alignment to the reference genome. Transcripts were assembled and quanti ed by Stringtie-1.3.3b [54]. In addition, we explored S-MART (http://urgi.versailles.inra.fr/Tools/S-MART) to calculate the distribution of reads mapped to exons, introns and 1 kb upstream/downstream of the annotated genes. To count the number of reads that uniquely mapped to an exon, featureCounts was used with 'gene' as feature and not strand-speci c [55].
Since low expressed genes are more vulnerable to measurement errors, we removed low expressed genes whose counts were lower than 2 in 90% samples. And then FPKM (expected number of Fragments Per Kilobase of transcript sequence per Millions base pairs sequenced) of each gene was calculated based on the length of the gene and read count mapped to this gene. FPKM considers the effect of sequencing depth and gene length for the read counts at the same time and is currently the most commonly used method for estimating gene expression levels from RNA-seq data [14].

Hierarchical Clustering
After quality control, we investigated sample heterogeneity between wild-and LBP-de cient liver transcriptome data by performing unsupervised hierarchical cluster analysis. Raw z-scores were rstly calculated from counts of wild-and LBP-de cient samples and then subjected to agglomerative hierarchical clustering analysis based on Ward's method and Euclidean distance. Bioinformatics analysis was performed in R version 3.5.1 and heat map was generated by pheatmap package from CRAN Rproject (https://CRAN.R-project.org/package=pheatmap).

Differential Gene Expression Analyses
Differential expression analyses of the LBP −/− experimental and normal groups were performed using the DESeq2 R package [14]. It provides statistical routines for determining DEGs from digital gene expression data using a model based on the negative binomial distribution. The resulting P-values were adjusted using the Benjamini and Hochberg's approach for controlling the false discovery rate. Genes with adjusted P-value less than 0.05 and log2(fold change) greater than 1.5 were assigned as DEGs.
Gene Ontology and Pathway Enrichment Analyses DAVID (https://david-d.ncifcrf.gov/) and PANTHER (http://www.pantherdb.org/) were executed to identify over-represented gene ontology (GO) terms and pathways of the DEGs. GO terms with corrected P-value less than 0.05 were considered signi cantly enriched by DEGs. KEGG is a database resource for understanding high-level functions and utilities of the biological system, such as the cell, the organism and the ecosystem, from molecular-level information, especially large-scale molecular datasets generated by genome sequencing and other high-through put experimental technologies (http://www.genome.jp/kegg/). We used the KOBAS software (http://kobas.cbi.pku.edu.cn) to test the statistical enrichment of DEGs in KEGG pathways.  (7):923-30. Figure 1 Levels of liver enzymes in wild type and LBP-de cient groups after LPS injection. ALT levels in serum samples collected at 0h, 6h, and 24h after LPS challenge. Serum ALT levels were analyzed as a measure of hepatocellular injury. Data are shown as means and standard deviations (n=3 per group at each time point). *p < 0.05, **p < 0.005, ***p < 0.0005, signi cantly different from the wild type groups. Neighbor joining tree of normal and LBP-de cient samples that treated with LPS for the times indicated (0h, 6h, 24h). Each condition has 3 replicates. Logarithm transformed counts from RNA-Seq dataset were computed for sample correlation by Pearson's correlation. CTR: normal rat; LBP: LBP-de cient rat.

Figure 3
Distinct transcriptional signature between WT and LBP-de cient rats. (A-C) Transcription pro les of signi cantly differentially expressed genes (DEGs) with log2(Fold Change) larger than 1.5 at P-value < 0.01 at 0h, 6h, 24h respectively. The labeling condition and DEGs were adapted as previous panels. (D-F) The volcano plot of LPS-induced transcriptional changes between normal and LBP-de cient rats with the time of 0h, 6h, 24h respectively. Differential expression genes with log2(Fold Change) larger than 1.5 at Pvalue < 0.05 were colored with blue (down-regulated) and red (up-regulated).