HTML
-
DENV infection affects various host cellular pathways including interferon responses, apoptosis and endoplasmic reticulum stress. Knowledge of changes in host protein expression and phosphorylation caused by infection helps to understand the interaction between virus infection and host response. In this study, a dimethyl labelling-based quantitative proteomics strategy was employed to characterize the host cell proteome and phospho-proteome changes upon DENV infection in K562 cells (Fig. 1A). Briefly, K562 cells were infected with DENV-2 at a multiplicity of infection (MOI) of 10 or mock treated and were harvested and lysed at 48 h post infection (hpi). The DENV-2 infected K562 cells were analyzed by qRT-PCR (Fig. 1B). Extracted proteins were tryptic digested, and the digested peptides from mock-treated or DENV-infected cells were differentially labeled with light or medium dimethyl reagent, respectively. The labeled peptides were mixed in a ratio of 1:1, subjected to TiO2 phosphoenrichment, offlinely fractionated by hydrophilic interaction liquid chromatography (HILIC), and then analyzed by liquid chromatography-tandem mass spectrometry (LC– MS/MS) (Fig. 1A). Three independent biological replicates were performed.
Figure 1. Proteomics and phosphoproteomics profiling of DENV-2 infection in human K562 cells. A Workflow for quantitative proteomic analysis of DENV-2 infected K562 cells. B qRT-PCR analysis of DENV-2 levels in K562 cells. K562 cells were infected with DENV-2 strains TSV01 at a MOI of 10. The intracellular DENV-2 RNA was extracted and measured by qRT-PCR at different time points. Three independent biological replicates were performed. C An overview of all identified and regulated proteins, phosphoproteins and phosphosites. Orange box represented upregulated protein/ phosphosite, blue box represented downregulated protein/phosphosite, gray box represented unregulated protein/phosphosite.
MS data were submitted to Maxquant to perform peptide identification and quantitation. Using these criteria that only proteins with at least two quantified peptides with a confidence score of > 95 were quantified, a total of 2263 host proteins and 2440 phosphorylated sites on 799 host phosphoproteins were identified (Table 1 and Fig. 1C). The cutoff for differentially regulated proteins was set as described previously (Xin et al. 2017). Briefly, the Gaussian distribution of protein ratios was analyzed, and proteins with ratios deviating from the mean of the normally distributed data by 1.96 standard deviations (SDs) were considered differentially regulated (Supplementary Figure S1). Using this criterion, we found that 201 proteins were identified as being upregulated, while 120 were identified as being downregulated out of the 2263 quantified host proteins. And 103/85 phosphoproteins were identified as being upregulated/downregulated, while 301/200 phosphosites were identified as being upregulated/downregulated out of the 2440 quantified phosphosites among the 799 host phosphoprotein (Table 1) (detail information in Supplementary Table S1–S3). As indicated in Fig. 1C, 14.1% of all quantified proteins were significantly regulated, whereas higher percentages (23.4%/ 20.4%) of all quantified phospho-proteins/sites showed significant regulation, implying a bigger effect on host protein phosphorylation caused by DENV infection as compared to protein expression. Interestingly, relatively higher numbers of upregulation in expression and phosphorylation were observed for the same host proteins (Fig. 1C).
Biological replicate 1 Biological replicate 2 Biological replicate 3 Total Total protein Unregulated 1775 1845 1771 1942 Significantly upregulated 83 68 98 201 Significantly downregulated 57 72 60 120 Phosphoproteins Unregulated 549 622 617 611 Significantly upregulated 53 49 41 103 Significantly downregulated 28 38 55 85 Phosphosites Unregulated 1563 1840 1947 1939 Significantly upregulated 178 124 70 301 Significantly downregulated 107 79 64 200 In total, 2263 proteins, 799 phosphoproteins and 2440 phosphosites were quantified. Gaussian distribution (mean ± 1.96SD) was applied to filter out the significantly regulated proteins/sites. Table 1. An overview of proteins, phosphoproteins and phosphosites quantified in this study.
-
Gene ontology (GO) descriptions were referred to their annotations in the NCBI database. To obtain an overview of functional classification of all the differentially regulated proteins, the 201 upregulated and 120 downregulated proteins together with 103 upregulated and 85 downregulated phosphoproteins were submitted to Panther for GO analysis. The results showed that the regulated proteins appeared to be involved in a broad range of cellular biological processes and categorized into ten to twelve functional groups (Fig. 2A–2D) (detail information in Supplementary Table S4). Obviously, a broad spectrum of host proteins/phosphoproteins was involved in virus infection and host response with similar pattern of functional classification. The three largest groups of regulated proteins were identified involving in metabolic process, cellular process, and cellular component organization or biogenesis, composing high variety of proteins with vital cellular functions. In addition, a smaller quantity of the regulated proteins/phosphoproteins involved in response to stimulus and immune system process was found. Moreover, for both proteins and phosphoproteins, upregulated proteins exhibited different patterns as compared to downregulated proteins in most processes except for the three major processes. Especially, higher percentage of upregulated proteins was distributed in response to stimulus than downregulated proteins (Fig. 2A–2D, words highlighted in red). High similarity was shared between protein expression and phosphorylation in response to DENV infection.
Figure 2. Functional classification analysis. List of regulated total proteins and phosphoproteins were submitted to PANTHER to perform a gene classification analysis. A Upregulated total proteins, B downregulated total proteins, C upregulated phosphoproteins, D downregulated phosphoproteins. Different colors represent different biological processes; area of fans represent the number of included proteins. "response to stimulus" was highlighted in red.
To assess the statistical overrepresentation of specific categories of regulated proteins relative to all quantified proteins, a hypergeometric test in GO annotation was performed by using BinGo. Such statistical enrichment analysis can assess whether the numeric values of proteins in certain biological processes are nonrandomly distributed with respect to the numeric values of all quantified proteins. Categories with normalized P < 0.001 were considered overrepresented. As illustrated in Fig. 3A (detail information in Supplementary Table S5), the distributions of proteins in the categories—macromolecule biosynthetic process, RNA splicing, Goμgi vesicle transport and cellular membrane organization were identified as highly overrepresented in total proteins, suggesting that these biological processes were activated by DENV infection (Fig. 3A). In comparison, the distributions of phosphoproteins in the biological processes including negative regulation of nucleobase, nucleoside, nucleotide and nucleic acid metabolic process, negative regulation of gene expression, chromatin modification, macromolecular complex assembly, RNA splicing, cell cycle, mRNA Processing and regulation of transcription were significantly higher than the overall distribution, implying that these processes were highly regulated by reversible protein phosphorylation in virus and host interaction (Fig. 3B). Nucleic acid metabolic process was enriched for both regulated proteins and phosphoproteins emphasizing its role in response to DENV infection.
Figure 3. GO analysis of regulated protein and phosphoprotein. List of regulated total proteins and phosphoproteins were submitted to BINGO to perform a GO enrichment analysis. A Go analysis of regulated total proteins (P < 0.001). B GO analysis of regulated phosphoproteins (P < 0.001). Whole annotation is as reference. Area of circles represents included gene numbers. P < 0.001 defines overrepresent categories. Green line represents "Cellular macromolecule biosynthetic process". Purple line represents "RNA splicing". Blue line represents "mRNA processing". Black line represents "Regulation of transcription".
To better understand the host cellular response to DENV infection, annotations of all regulated proteins were examined manually. We found that the majority of regulated proteins were involved in RNA splicing and cellular macromolecule biosynthetic process. These two processes were found to be the two most overrepresented processes upon DENV infection with a low P values (highlighted in purple and green arrows, respectively in Fig. 3A). For protein phosphorylation, mRNA processing and transcriptional regulation were found to be the two most overrepresented processes in response to DENV infection (highlighted in blue and black arrows, respectively in Fig. 3B).
-
To explore potential protein network connections for the significantly enriched processes from GO analysis, the regulated proteins involved in the most enriched pathways were analyzed by using STRING in which the interactions were documented according to known and predicted protein–protein interactions. The resulting protein networks were visualized in Cytoscape (Fig. 4, detail information in Supplementary Table S6).
Figure 4. Protein–protein interaction network of regulated protein. List of "Cellular macromolecule biosynthetic process" and "RNA splicing" among total protein, "mRNA processing" and "Regulation of gene expression" among phosphoprotein were submitted to STRING to perform an interaction analysis and visualized by Cytoscape. A Network of "Cellular macromolecule biosynthetic process" and "RNA splicing" from regulated total proteins. The network includes 54 nodes (proteins) and 195 edges (interactions). Asterisk represents interested proteins. B Western blot validation of 5 interested genes. C Network of "mRNA processing" and "Regulation of gene expression" from regulated phosphoproteins. The network includes 48 nodes (proteins) and 133 edges (interactions). Black dot represents interacted predicted kinases. D Kinase prediction of regulated phosphoproteins.
For protein expression, proteins involved in RNA splicing and cellular macromolecule biosynthetic process were of great interests and chosen for protein–protein interaction analysis. A highly interacting protein network was formed, consisting of 54 individual proteins and 195 distinct interactions (Fig. 4A). Proteins interact not only with proteins involved in the same process but also with proteins in the other process. RNA splicing and cellular macromolecule biosynthetic process were represented in rectangle and ellipse, respectively, with diamond representing the overlap. Up/down-regulation information was also included and highlighted in red/blue for clear comparison. As showed in Fig. 4A, more proteins were found to be upregulated than downregulated in both processes.
Those proteins which showed regulated expression by virus infection and appeared to actively interact with other proteins might be the key players in the antiviral response. Based on our MS results and the protein–protein interaction analysis as well as previous knowledge, several proteins drew our attention, which were EIF3E, BID, HDAC1, GRB2, and SAP18. They were all upregulated upon DENV infection revealed by mass spectrometric analysis and were highly interacting with proteins involved in cellular macromolecule biosynthetic process (Fig. 4A).
EIF3E is the component of the eukaryotic translation initiation factor 3 (eIF-3) complex, which is required for several steps in the initiation of protein synthesis (Masutani et al. 2007). EIF3E has been reported as a binding partner of classical swine fever virus (CSFV) NS5A protein. Overexpression of EIF3E markedly enhanced CSFV genomic RNA replication, viral protein expression and production of progeny virus (Liu et al. 2018). BID is a proapoptotic Bcl-2 protein containing only the BH3 domain (Wang et al. 1996). BID interacts with BAX, leading to the insertion of BAX into organelle membranes and finally induces apoptosis. Although apoptosis has been detected in vivo and in vitro in response to DENV infection, the mechanism is not completely understood (Thongtan et al. 2004). In our work, BID was 1.7 fold higher in abundance after infection. It may be involved in DENV-induced apoptosis and was worthy for further investigation. GRB2 (Growth factor receptor-bound protein 2) is an adapter protein that provides a critical link between cell surface growth factor receptors and the Ras signaling pathway. It has been reported that HCV NS5A can specifically interacts with GRB2 and inhibits the MAPK/ERK pathway (Huynh et al. 2016). SAP18 (Sin3a-associated protein of 18 kDa) is a component of the SIN3-HDAC1 complex which can direct the formation of a repressive complex to core histone proteins and induce transcriptional repression (Zhang et al. 1997). SAP18 was also found to be correlated with RNA splicing (Singh et al. 2010). One work has find that HIV-1 IN protein can specifically interact with SAP18 and recruit the Sin3a-HDAC1 complex into HIV-1 virion. The deacetylase activity is important for the subsequent stages of viral replication (Sorin et al. 2009).
To verify our MS data, Western blotting analysis of selected host proteins was performed. As shown in Fig. 4B, Western blotting data showed that the proteins EIF3E, BID, HDAC1, GRB and SAP18 were upregulated as a result of DENV infection (Fig. 4B), consistent with our MS data.
For protein phosphorylation, phosphoproteins which were significantly regulated upon infection and involved in mRNA processing and regulation of transcription were also further looked into for protein–protein interaction analysis. A network consisting of highly interacting 48 phosphoproteins was formed and 133 distinct interactions between phospho-proteins in the same process and in the other process were illustrated in Fig. 4C. Equal number of upregulated and downregulated phosphoproteins was found to be involved in both processes.
Our data set provides a source of phospho-peptide sequences for exploring the network between different kinase families and their substrates. NetworKIN was used to generate a list of predicted potential kinases for the regulated phosphorylation sites. Prediction was based on comparison of the phosphorylation sites with reported consensus sequences for different protein kinases. The number of distinct phosphorylation sites for each kinase family was illustrated in Fig. 4D (detailed information in Supplementary Table S7. The kinase prediction of identified phosphorylation sites exhibited a broad range of kinase families which might target various host proteins in the process of virus infection. It suggests major role for the CDK1, CK2alpha, PKCalpha, CLK1, GSK3alpha and PKBalpha kinase families, implicating the greatest proportion of the identified distinct phosphorylation sites are their potential substrates. As marked in Fig. 4C, several upregulated proteins involved in both mRNA processing and regulation of transcription, as well as more downregulated proteins involved in regulation of transcription were found to be the substrates of those major kinase families.