Abstract Metabolic syndrome (MetS) is associated with a group of conditions, which enhances the risk of diabetes, heart diseases and stroke in the affected individuals. Earlier reports from our lab have shown that Tumor necrosis factor-α (TNF-α) significantly modulates the expression of 56 genes at the alternative splicing level which are involved in various signaling and metabolic pathways (MetS genes) connected to MetS. These MetS genes were predicted to interact with various RNA-binding proteins (RBPs) when exposed to TNF-α, resulting changes in their alternative splicing patterns. Here we are presenting data of an RNA-Seq analysis, which identified 1218 unique, and significantly regulated genes by TNF-α, 15% of which are RBPs . Among the 1218 genes, 204 genes have been identified as MetS genes by the ingenuity pathway analysis, and 10% of the MetS genes are found as RBPs. Our results also show that TNF-α changes the phosphorylation status of certain RBPs such as SR proteins, crucial players in alternative splicing, possibly via changing the activation status of certain upstream signaling molecules which also act as upstream kinases for these proteins. Taken together, these findings suggest that TNF-α influences the regulation of the RBPs at the various levels for their expression, which may lead to the alteration of the splicing pattern of the MetS genes. MetS genes acting as RBPs and are modulated by TNF-α, predict the existence of highly interconnected mechanisms which require further analysis to understand their dual roles on the onset of these diseases. Introduction Metabolic syndromes (MetS) are associated with an assemblage of conditions including diabetes, obesity, insulin resistance, high triglyceride levels and cardiovascular risks. Inflammatory signals mostly mediated by the pro-inflammatory cytokines such as Tumor necrosis factor-α (TNF-α) and Interleukin-6 (IL-6), have stated to be the connecting link between these risk factors [[29]1,[30]2]. Infiltration of macrophages in hypoxic and enlarged adipocytes causes overproduction of inflammatory chemokines such as TNF-α [31][3], [32][4], [33][5], causing impairment of insulin signaling pathways and leading to insulin resistance and later stage of Type 2 diabetes (T2DM) [[34]6,[35]7]. Anti-TNF-α therapy has been proven efficient for various MetS [36][8]. Different signaling pathways modulated by TNF-α in MetS, are well documented [[37]5,[38]9,[39]10] TNF-α-mediated activation of stress related MAPKs (p38 MAPK and JNK) and various other Ser/Thr kinases (IKKβ, PKCθ etc.) have reported to contribute towards insulin resistance. TNF-α may trigger the activation of other cytokines-mediated pathways creating a second wave of signaling [40][11]. Other than these pathways, modulation of synthesis and activation of signaling proteins such as PPAR-γ and GLUT4 may also contribute towards metabolic disorders [[41]7,[42]12,[43]13]. Yet, the comprehensive role of TNF-α in MetS including its downstream effector targets have not been fully understood. Earlier report from our lab shows that TNF-α induces modulation of alternative splicing of genes (including CREB5, NFKB1, NFKB2, TP53 etc.) involved in various signaling and metabolic pathways (henceforth referred to as MetS genes), connected to MetS [44][14]. Alternative splicing is a posttranscriptional mechanism that creates diversity of proteins from a similar genetic background and is modulated by various RNA binding proteins (RBPs). The RBPs, once loaded, may decide the fate of the pre-mRNAs for their splicing, polyadenylation and capping [[45]15,[46]16]. The RBPs may have defined (canonical) or undefined (non-canonical) RNA binding domains (RBDs) [47][17]; the latter group uses intrinsically disordered regions (IDRs) for RNA binding [48][16]. The other classes of RBPs without a recognizable structural RBD or an assigned function in RNA biology are known as “enigmRBP” [49][17]. These unconventional RBPs include protein with various biological functions like cytoskeleton remodeling, protein folding, ATP-binding and enzymatic functions in classic metabolic pathways. Almost 9% of conserved core RBPs between human and yeast are shown to be metabolic enzymes. The classic enzymes belonging to this category includes oxidoreductases, isomerases, hydrolases, isomerases, lyases and ligases [50][18], [51][19], [52][20]. Metabolic enzymes were identified as RNA binding proteins by several groups [[53]18,[54][21], [55][22], [56][23], [57][24]]. A large-scale analysis revealed as many as 71 metabolic enzymes as RNA binding proteins in humans. Some of these enzymes are shown to bind to their own mRNAs, predicting a feedback loop (moonlighting activity), or contributors in modulation of enzyme activity [[58]25,[59]26]. These moonlighting enzymes include enzymes involved in glycolysis, tricarboxylic acid cycle, lipid metabolism and deoxynucleotide biosynthesis [60][25], [61][26], [62][27].Differential regulation of several RBP's such TTP, HuR, QKI, SRSF6 are reported in connection to various metabolic disorders [[63]28,[64]29] A global analysis of RBPs regulated by TNF-α at the transcriptional as well as at the post-transcriptional level is yet to be performed for a better understanding of the TNF-α associated MetS. A recent study from our lab has reported a genome wide map of 13,395 unique RNA-RBP interactions (RPIs) upon TNF-α treatment on the 228 significantly regulated alternatively spliced genes. Among which, 22% of the total interactions were observed on the 56 MetS genes with 32 unique RBPs (predicted by RBPDP database) [65][30].This result demonstrated the potential involvement of the RBPs in the modulation of the MetS genes at the alternative splicing level. Expression and functional aspects of RBPs are mostly determined by various signaling pathways [[66]31,[67]32] modulated by extracellular cues including cytokines [68][33]. Apart from the transcriptional regulation, the functional aspects as well as their shuttling back and forth between the nucleo-cytoplasmic pools of some of the RBPs, especially true for the SR-proteins, depend on their phosphorylation status [[69]34,[70]35]. Reported kinases responsible for phosphorylation of these RBPs include SRPK1/2, CLK and PRP4 [71][36]. Other kinases involved in crucial cellular signaling cascades including PI3K, P38MAPK, PKB, mTORC etc., are also known to modulate the phosphorylation status of these RBPs, either directly or by recruiting mediator proteins, such as Sam68 and HuR [72][37]. In the present study, we have identified the regulation of RBPs at the mRNA and post-translational level by TNF-α. Our data obtained from an RNA-Seq analysis shows that 10% of the MetS genes, regulated by TNF-α, are RBPs. We also obtained results showing that the phosphorylation status of some of the SR-proteins, which are the crucial players in alternative splicing, are regulated by TNF-α treatment. The change in the phosphorylation status of these SR-proteins correlated with the phosphorylation and thus the activation status of two crucial Ser-Thr kinases; p38 MAPK and mTOR. Taken together, this study substantiates our prediction that regulation of RBPs by TNF-α, at various levels, potentially involving upstream signaling pathways, may contribute towards differential regulation of MetS genes and in turn, influences the onset of MetS. Methodology Mammalian cell culture: Human Embryonic kidney (HEK 293) cells (NCCS, Pune) and Human skeletal myoblasts (A12555, Thermo Scientific), were maintained in low glucose (HEK 293; 1g/liter) or in normal glucose (4.5 g/l) DMEM media (HiMedia) supplemented with 10% FBS (Gibco, South American origin) at 37 °C in 5% CO[2]. A12555 cells were incubated for 48 h to induce differentiation as per the manufacture's recommendation (Thermoscientific). The cells were treated with TNF-α (Sigma) (10 ng /ml) for different periods in serum-deprived media. Most of the experiments were performed after 6 h of TNF-α treatment at the concentration of 10ng/ml in vitro. RNA Isolation: Total RNA was extracted from the cells using RNA Xpress reagent (HiMedia) according to manufacturer's instruction. The isolated RNA was quantified using Nanodrop spectrophotometer (Thermofischer scientific). The quantified total RNA (1 µg) was used for cDNA synthesis following a protocol of High Capacity Reverse Transcription kit (Applied Biosystems, Thermoscientific). Quantitative real time PCR: 2.5 ng of cDNA were used to perform qPCR. The reactions were performed using Brilliant Ш Ultra-Fast SYBR green qPCR Master Mix in AriaMx PCR System (Agilent Technologies) with the following cycling conditions: 95 °C for 3 min (initial denaturation), followed by 40 cycles of 95 °C for 30 s, 58 °C for 30 s and 72 °C for 60 s. 18srRNA served as the internal control. The primers used have been enlisted in in the Supplementary Table 1. Transcriptome sequencing: A reference based transcriptome sequencing (for Homo sapiens) was performed from the total RNA isolated from one set of TNF-α treated and one set of untreated cells after the respective cDNA library preparation. No replicates of the RNA-Seq was performed for the data analysis. The protocol for RNA extraction, cDNA synthesis and NGS analysis is the same as was mentioned in our previous publication [73][14]. Cufflinks-2.1.1 was used for differential regulation analysis from RNA-seq samples. RPKM (Reads per kilobase per million mapped reads) value was calculated using the formula, RPKM = 10^9 x C/N*L, Where C = number of mapping reads, N = Total number of reads, L = length of the transcript for each gene. A minimal read count of 5 reads was used to filter out the genes not expressed. To determine the differentially expressed genes, Cuffdiff was used to compare the two categories (Treated vs Control). A two-fold change and p-value <0.05 cut-off was used to report the DEGs. Owing to the absence of replicates, no FDR correction was used. Gene Ontology Analysis: WEB-based GEne SeT AnaLysis Toolkit (WebGestalt): Over-Representation Analysis (ORA) with the 1218 short-listed genes was performed using the WEB-based GEne SeT AnaLysis Toolkit (WebGestalt; doi: 10.1093/nar/gkz401), considering all the unique genes reported by the RNA-Seq analysis (12,715) as the background. Out of the 1218 genes, 1162 were mapped against the 11,464 unique genes (out of 12,715 genes) from the reference list. For significance level, FDR<0.05 was chosen. Pathway analysis Ingenuity pathway analysis (IPA): The enriched pathways from the data set and their ratios and significant scores (using the right-tailed Fisher exact test) were calculated using IPA. Calculation of the ratio and the p-values is the same as mentioned in our previous publication [74][14]. A threshold score of p-value 0.05 and 0.01 corresponding to a significance score of 1.3 and 2 respectively, was used to report the significantly enriched pathways. KEGG pathway analysis: The enriched pathways and the number of genes mapped against each pathway are represented by KEGG. The pathways related to metabolic disorders such as type 2 diabetes were extracted. The distributions of mapped genes related to metabolic disorders were represented in the pie chart. Network analysis STRING database analysis: STRING database represents a network detailing the interaction between the genes or proteins used as input. Isolated nodes represent the genes without any interaction. The genes/proteins are represented as nodes and the edges represent the interactions among them. The output from the STRING DB showing the connections among the genes (nodes) (Supplementary Datasheet 3) was used to assess the pairwise connections among the genes. Genes having ten or more connections were considered as highly connected nodes. IPA network analysis: IPA network analysis was performed as mentioned earlier [75][14]. Enrichment score was given to the focus molecules (genes from the user's list). EuRBPDB analysis: The differentially regulated genes by TNF-α treatment were used as inputs for the EuRBPDB (Eukaryotic RNA binding protein database; http://eurbpdb.syshospital.org/search_rbp.php) database to retrieve the RBPs [76][17]. EuRBPDB updated version, v1.2 was used as the search engine with the specific chosen parameter “RBP type”. Data from our RNA-seq analysis (in terms of gene names), was fed into the EuRBPDB database using the “requests” python library. The search pages were analyzed to identify RBP type for the genes that are fed into it. TRRUST analysis: The differentially regulated MetS genes by TNF-α treatment were used as inputs for the analysis into TRRUST (Transcriptionally regulatory relationships unraveled by sentence-based text mining) version 2 database (https://www.grnpedia.org/trrust/) to identify the transcription factors. Western Blot analysis: Cells were exposed TNF-α for different time period (30’ to 6 h) and cell lysate was prepared using a lysis buffer (4x Lammeli buffer supplemented with 1X mammalian protease phosphatase inhibitor cocktail) and stored at -20 °C. After SDS-PAGE, wet transfer was carried out using PVDF membrane. After incubation with the primary and the secondary antibodies, the blots were developed using ECL plus western blot developing system (GE healthcare) on X-ray films. β-Actin served as the internal control for all the experiments. Antibodies used are as follows; p-SR (1:1000, MABE50, Merck), p-P38MAPK (1:1000, cell signaling Technology), p-mTOR (1:1000, CST), β-Actin (1:1000, R&D), Anti-Mouse HRP (1:2000, Sigma) and Anti-Rabbit HRP (1:2000, Cell signaling Technology). Statistical analysis: Statistical evaluation for the p-values was done by paired, two-tailed t-test in each data set. The symbols *, ** and *** represent p < 0.05, < 0.01 and < 0.001, respectively. Results Genome-wide transcriptome analysis revealed the identity of differentially regulated genes by TNF- α: Identification of the differentially regulated genes upon stimulation by TNF- α was carried out by a genome wide transcriptome analysis, with the total RNA extracted from the TNF-α treated and untreated HEK293 cells. Considering a twofold change cut-off, a total number of 6056 and 5495 transcripts were found to be up and down regulated, respectively (Supplementary Datasheet 1). On excluding the transcripts with no gene names, the number got reduced to 4566 and 4465 for the up and down regulated transcripts, respectively. These transcripts corresponded to 3645 up-regulated and 3502 down-regulated genes ([77]Fig. 1a). From this gene pool, considering a p-value cut-off <0.05, 1218 unique genes were sorted out, combining the up and down regulated ones ([78]Fig. 1b and Supplementary Datasheet 1). Further analysis has been carried out with this gene pool. Fig. 1. [79]Fig 1 [80]Open in a new tab Quantitative representation of the data obtained from the NextSeq analysis. (a) The bar diagrams depict the no. of transcripts and genes got up or down regulated with a two-fold cut-off, upon TNF-α treatment compared to the control (b) Representation of the no. of genes up and down regulated with p-value <0.05 and the total no. of unique genes. Validation of RNA-seq data Among the differentially expressed genes identified by RNA-seq analysis, a few candidate genes, which were already reported in the context of metabolic disorders, were short-listed for validation. We have previously reported the modulation of MetS genes at the alternative splicing level by TNF-α. Here we chose a few MetS genes whose expression is modulated at the mRNA level ([81]Table 1). Table 1. Genes validated for regulation by TNF-α at the mRNA level and their association with MetS. Gene Name Chromosome Transcript Start (bp) Transcript End (bp) FPKM control FPKM TNF-Alpha log2 (fold_change) P_value Regulation Function/association with T2DM/metabolic disorders References