Liver Gene Expression Profiles in Sham, CLP, and CLPDCA Group

A total of 1208 DEGs were identified from 4 sham, 4 CLP, and 3 CLPDCA samples, including 800 up-regulated and 408 down-regulated genes between CLP and sham (Fig. 1a). A further 125 DEGs were identified between the CLPDCA and sham groups, including 67 up-regulated and 58 down-regulated genes. Inspection of Figure 1a shows that treatment with DCA had partially reversed the changes in gene expression found between the sham and CLP groups. Hierarchical cluster analysis of significantly different expression patterns in different groups of samples found DEG patterns to be more similar between CLPDCA and sham groups than between CLP and sham (Fig. 1b). Analysis of DEG overlap identified 48 co-up-regulated and 36 co-down-regulated DEGs in the CLP and CLPDCA treatment groups relative to the sham group. However, a comparison of sham and CLP groups revealed 750 up-and 372 down-regulated genes. The differences in numbers of up- and down-regulated genes support the view that treatment of CLP mice with DCA partially reversed abnormal gene expression induced by CLP (Fig. 1c).

Fig. 1

Gene expression profile of mouse liver tissue. a Significant DEGs in CLP and CLPDCA septic mice. b Expression heatmap of significant DEGs among CLP, CLPDCA, and sham. c Venn diagram showing overlap of DEGs in CLP and CLPDCA samples. d Bar plot of the most enriched GO biological processes of genes up-regulated in CLP. e Bar plot of the most enriched GO biological processes of genes down-regulated in CLP.

Functional analysis was performed on DEGs between CLP and sham. GO analysis showed co-up-regulated genes to be enriched in the biological processes of cell adhesion, inflammatory response, immune system processes, positive regulation of cell migration, negative regulation of external apoptotic signal pathways through death domain receptors, chemotaxis, and other functional pathways (Fig. 1d). Co-down-regulated DEGs were enriched in oxidation-reduction and lipid metabolic processes (Fig. 1e). Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis identified up-regulated genes as being involved in cytokine-cytokine receptor interaction and IL-17 signaling pathway and down-regulated genes in metabolic pathways. Genes that were up-regulated in the comparison of CLP with CLPDCA were enriched in FoxO signaling and glucagon signaling pathways (supplement1).

The involvement of lipid metabolism–related genes in the impact of sepsis on liver tissue was highlighted by the outcomes of the functional analyses described above. Inflammation/immune and cell apoptotic–related genes, including cytokine-cytokine receptor interaction and the IL-17 signaling pathways, also featured prominently in the differential expression results. It may be suggested that the actions of DCA on signaling of the transcription factor, FoxO, and hormone, glucagon, are involved in reversing abnormal gene expression. Lipid metabolism was also highlighted as a process worthy of further investigation as a potential therapeutic target.

Alternative Splicing Patterns in Sham, CLP, and CLPDCA Group

AS events at nine variable splicing sites, including A3SS, A5SS, and ES, were found from RNA-Seq data. Hierarchical clustering analysis showed that the 4 CLP samples formed one cluster, and the three sham and three CLPDCA samples formed a second cluster. Thus, the variable splicing patterns of genes in CLPDCA and sham were more similar to one another than either was to CLP (Fig. 2a). RASEs present in CLP and CLPDCA groups relative to sham were identified by t-test, and the most significant were A3SS and A5SS with the cassette exon and ES following (Fig. 2b). The implication of the findings above is that the pattern of A3SS and A5SS splicing may be associated with gene expression in the septic liver.

Fig. 2

Regulated alternative splicing events in septic mice. a PCA of RASE ratios with confidence ellipse for all groups. b Bar plot of RASEs in CPL versus sham liver and CLPDCA versus sham liver. c Scatter plot of enriched GO terms relating to ASEGs in CPL versus sham liver. d Scatter plot of enriched GO terms relating to ASEGs in CPLDCA versus sham liver.

Genes associated with alternative splicing events (ASEGs) were subjected to GO functional analysis. Those differentially expressed between the CLP and sham groups showed enrichment for the processes of oxidation-reduction, lipid metabolism, mitochondrial morphogenesis, and positive regulation of fatty acid biosynthesis (Fig. 2c). Those differentially expressed between CLPDCA and CLP were enriched for protein ubiquitination, lipid metabolism, redox balance, autophagy, DNA template–dependent negative and positive regulation of transcription (Fig. 2d). These results give further confirmation of the involvement of lipid metabolism and oxidation-reduction reactions in sepsis and also show that DCA treatment ameliorates the sepsis-dependent changes by acting on the same pathways.

Overlap analysis of DEGs and ASEGs between CLP and sham revealed 61 genes were differentially expressed between the two groups. Similar analysis of the CLPDCA and CLP groups showed only four genes with differential expression (Fig. 3a). GO analysis of the differences between CLP and sham showed enrichment for redox processes, lipid metabolism, and regulation of transcription by RNA polymerase II (Fig. 3b).

Fig. 3

Regulated alternative splicing events. a Venn diagram showing the overlap of ASEGs and DEGs. b Bar plot of enriched GO terms relating to the overlapping genes in a.

A comparison of AS event ratios of the CLP and sham groups revealed that 390 events were up-regulated and 396 down-regulated. However, a similar comparison of CLPDCA with sham revealed only 43 up-regulated and 66 down-regulated variable alternative splicing events. Again, it can be seen that the pattern observed for DEGs has been repeated. Abnormal splicing in the septic liver may cause tissue damage, and treatment with DCA reverses the effect by reversing the changes in abnormal splicing.

Up- and down-regulation of differential AS events in CLPDCA tissue were 244 and 249 compared with CLP, indicating that CLPDCA reversed the abnormal gene expression profile (Fig. 4a). Cluster analysis carried out by screening the covariant AS events with read number > 10 in at least 80% of the samples showed that the four CLP samples were clustered in one group and the 3 sham and 3 CLPDCA samples were clustered in a second group (Fig. 4b). Up-regulation of AS events with CLP treatment may be associated with septic liver damage, and up-regulation of AS events with DCA treatment reversed the CLP effect and may indicate a protective effect on the septic liver. GO analysis of the genes involved in the AS events above was performed.

Fig. 4

Differential AS events in the CLP- and CLPDCA-treated liver. a Venn diagram of overlapping AS events in CLP versus sham and CLPDCA versus sham liver. b Hierarchical clustering heat map of all significant ASEG ratios. The AS filter included all detectable splice junctions with at least 80% of samples having ≥ 10 splice junction supporting reads. c Bar plot of the most enriched GO biological processes of the CLP-specific ASEGs from b. d Bar plot of the most enriched GO biological processes of the CLPDCA-specific ASEGs from b.

The 10 most enriched pathways involved in biological processes relating to the AS genes identified above included triglyceride homeostasis, positive regulation of fatty acid biosynthesis, redox processes, negative regulation of mRNA splicing by spliceosomes, protein transport, lipid metabolism, negative regulation of adipocyte differentiation, lactation, and in utero embryonic development (Fig. 4c). GO analysis of DCA-specific AS events indicated that redox processes, actin cytoskeleton, phospholipid biosynthesis, RNA processing, protein ubiquitination, autophagy, immune system processes, mitochondrion organization, ubiquitin-dependent ERAD pathways, and positive transcriptional regulation of DNA templates (Fig. 4d).

Human RBP genes were intersected with the DEGs found above to be associated with CLP/sham and CLPDCA/CLP differences. A total of 37 RBP genes were up- or down-regulated in the CLP group compared with the sham (Fig. 5a). Treatment with DCA reversed the expression changes of the RBP genes brought about by CLP (Fig. 5a). Some of the RBP genes identified are likely to be involved in AS events. The 20 RBPs showing the greatest degree of up- or down-regulation in the CLP group were screened (Fig. 5b), and a co-expression network of hub RBPs and ASEGs was constructed. RBP S100A11 was found to be likely to regulate multiple differential AS events (Fig. 5c). GO and KEGG analyses of the 10 RBPs co-expressed most frequently with ASEGs were carried out. GO analysis showed enrichment in lipid metabolism, redox processes, drug responses, cell division, cell cycle, ion transport, proteolysis, positive and negative regulation of transcription by RNA polymerase II, and DNA-dependent negative regulation (Fig. 5d).

Fig. 5

Interaction network of RNA-binding proteins and alternative splicing-associated genes. a Venn diagram showing overlapping differentially expressed ASEGs and RBP genes in septic mice. b Heatmap of differentially expressed RBP genes in CLP samples. RBPs were filtered by expected fragments per kilobase of transcript per million fragments mapped (FPKM) ≥1 in 80% of samples. The color key from blue to red indicates z-score color range. c The scatter plot shows ASEGs by CLP versus sham co-expressed with differentially expressed RBP genes from b. d Enriched GO biological processes of ASEGs co-disturbed with the top 10 differentially expressed RBPs. e The co-deregulation in CLP-treated liver of AS network and RBPs (left; circle size, number of connections), AS events (middle left), ASEGs (center right), and enriched GO biological process terms of co-disturbed ASEGs and RBP genes (right).

In order to investigate the regulatory relationship between RASE and DERBPs, a co-expression analysis was conducted. Pearson’s correlation coefficients (PCCs) were calculated and used to categorize their association as positively correlated, negatively correlated, or non-correlated. Relationship pairs between DERBPs and RASE that exhibited a correlation coefficient value of at least 0.85 and a p-value of no greater than 0.01 were identified through screening. An interaction network of differentially expressed RBPs and ASEGs suggested that CLP-induced up- or down-regulation of liver RBPs could be reversed by DCA treatment, affecting splicing of downstream genes (Fig.5e). A KEGG analysis showed the ASEGs co-disturbed with differentially expressed RBPs enriched in metabolic pathways, drug metabolism, porphyrin and chlorophyll metabolism, steroid biosynthesis, ascorbic acid and uronate metabolism, cytochrome P450 metabolism, riboflavin metabolism, pentose and gluconate interconversion, and sulfur metabolism (Fig. 6).

Fig. 6

The most enriched KEGG pathways of ASEGs co-disturbed with the top 10 differentially expressed RBPs.

Co-expression Analysis Between Sepsis-Regulated RBPs and NIR

An interaction and co-expression network was constructed to encompass all connections among RBPs, AS events, and ASEGs. The dysregulation of RBP gene expression caused by CLP was found to be reversed by DCA treatment with a likely impact on AS events (Fig. 7a and b). Genes involved in metabolic pathways, in particular, lipid metabolism, which is subject to AS, were the focus of the investigation. The abnormally expressed RBP, S100A11, was found to affect AS of some genes encoding products involved in lipid metabolism, such as SREBF1 and CERS2 (Figs. 7c, d and 8).

Fig. 7

Abnormally expressed RBP genes with effects on lipid metabolism–associated genes. a Network showing deregulation of lipid metabolism by abnormal RBP expression. b Box plot showing expression of S100a11 in CLP, CLPDCA, and sham samples. c Box plot showing the splicing ratio profile of the Cers2 gene across 11 samples. d Visualization of junction read distribution of Cers2 gene in samples from different groups. Splice junctions are labeled with SJ read number.

Fig. 8

Gene expression and splicing regulation of lipid metabolism–associated genes. a Box plot showing the splicing ratio profile of the Serbf1 gene across 11 samples. b Visualization of junction reads distribution of Serbf1 gene in samples from different groups. Splice junctions are labeled with SJ read number.

Analysis the Genome-Wide AS in Lipid Metabolism and Oxidation-Reduction–Related Genes

In the CLP group, the top ten alternative splicing event genes were Cers2, Tm7sf2, Crot, Pcyt2, Mgll, Sc5d, Elovl5, Akr1c6, Srebf1, and Gm44805 when compared with the Sham group. These genes were mainly enriched in metabolic lipid processes in the GO enrichment analysis of biological processes. Additionally, Cyp2d10, Tm7sf2, Paox, Uox, Sc5d, P4ha1, Haao, Akr1c6, Cp, and Bdh2 were the alternative splicing event genes that were mainly enriched in the oxidation-reduction process in the GO enrichment analysis of biological processes. The alternative splicing ratio profile of lipid metabolism and oxidation-reduction–related genes is shown in Table 2 and Fig. 9.

Table 2 Lipid Metabolic and Oxidation-reduction–related RASGFig. 9

Top ten node lipid metabolic and oxidation-reduction–related alternative splicing genes and relative alternative splicing events ratio in Sham, CLP, and CLPDCA group.

Furthermore, in the CLP group, the splicing ratio of Lyplal, Ppfibp1, Pnpla2, Lrps3, Pnpla6, Pnpla8, Mgll, Apoc3, and Lipc exhibits significant differences when compared to the Sham group. Additionally, the splicing ratio of P4ha1, Akr1c6, Cyp2d10, Tm7sf2, Bdh2, Gm44805, and Sc5d decreased in the CLP group compared to the Sham group. On the other hand, the Pcyt2, Uox, Paox, and Elovl5 genes showed an increase in their splicing ratio in the CLP group. The CLP group showed a significant reduction in the reads of ES in Lyplal1 compared to the Sham group. The reads of A3SS were increased in Mgll and Apoc3, and there was a significant reduction in the reads of cassette exon in Ppfibp1 and Pnpla2. as illustrated in Fig. 10. These findings suggest that the alternative splicing of genes related to lipid metabolism and oxidation might be involved in the development of liver damage during sepsis.

Fig. 10

Splicing ratio profile of specific lipid metabolism–related genes in Sham, CLP, and CLPDCA.

Validation of Co-expressed RBPs and ASEs

We observed abnormal alternative splicing of genes related to lipid metabolism in the liver tissues of septic animal model. These genes include Cers2, Tm7sf2, Crot, Pcyt2, Mgll, Sc5d, Elovl5, Akr1c6, Srebf1, Gm44805, Lylal1, Pnpla, Lyplal1, Ppfibp1, Pnpla2, Lrp3, Pnpla6, Apoc3, Pnpla8, Mgll, Apoc3, and Lipc. Dysregulated expression of cers2 and srebf1 has been reported in sepsis, and they are considered hub genes of lipid metabolism. However, there is a lack of evidence on the expression and alternative splicing of other lipid metabolism–related genes in sepsis. To address this, Considering the evidence for the genes and alternative splicing linked to lipid metabolism in sepsis, we validated only cers2 and srebf1 in a septic animal model. S100a11 expression levels were significantly greater in the CLP group than in the sham group, although DCA eliminated this difference. Furthermore, the alternative splicing ratio of Srebf1 and Cers2 was reduced compared with the sham group and increased after DCA treatment. These results suggested that S100a11 is relevant to the pathogenesis of sepsis-induced liver damage. Its mechanism may be related to Srebf1 and Cers2 alternative splicing, regulated by S100a11 (Fig. 11).

Fig. 11

Validation of RASEs and RBP. a Box plot showing expression status of S100a11 in CLP, CLPDCA, and Sham samples by qPCR validation. b Box plot showing splicing ratio profile of the Cers2 splicing event by qPCR validation. c Box plot showing the splicing ratio profile of the Serbf1 splicing event by qPCR validation. Data are shown as mean ± SD, *p < 0.05, **p < 0.01, ****p < 0.0001.