Bioinformatics Study on the Response of Human Endothelial Cells to Different Strains of Staphylococcus Aureus

Background Staphylococcus aureus-induced bacteremia has an impact on human health due to its high mortality rate of 20–30%. To better study the invasion process of staphylococcus aureus, we conducted a study in human endothelial cells to try to nd a link between the infection process and bacteremia at the molecular level. In this study, the datasets GSE13736, GSE82036 were analyzed using R software to identify differentially expressed genes. Only the infection samples of four different strains had differential gene expression compared to the control samples. Then the GO analysis and KEGG analysis were conducted to construct a protein-protein interaction (PPI) network which shows the interaction and inuence relationship between these differential genes. Finally, the central gene of the selected CytoHubba plug-in was veried using GraphPad Prism 8. There were 421 differential genes in the Strain 6850, including 64 up-regulated and 357 down-regulated; There were 319 differential genes in the Strain 8325-4, including 14 up-regulated and 305 down-regulated. There were 90 differential genes in the Strain K70058396, including 12 up-regulated and 78 down-regulated. There were 876 differential genes in the Strain K1801/10, accompanied by 261 up-regulated and 615 down-regulated. An analysis of GO and KEGG revealed that these differentially expressed genes were signicantly enriched in pathways associated with immune response and cytokines; Verication of the hub gene can provide a molecular basis for studying the relationship between invasive endothelial infection and bacteremia. We found specic gene expression patterns in endothelial cells in response to infection with Strain K70058396, and these central genes and their expression products (RSAD2, DDX58, IFITT3, and IFIH1) play a key role in this process of infection.


Introduction
Staphylococcus aureus, characterized by staphylococcus or clusters, is considered one of the most common and infectious pathogens [1] . Staphylococcus aureus can colonize the human body and mucosa with a variety of pathogenic factors [2] , including immune escape surface factors, enzymes, α-toxin, and even causes invasive infection [3,4] .
Humans suffer from a series of diseases caused by Staphylococcus aureus, such as skin and soft tissue infections, infective endocarditis, pneumonia and S. aureus bacteremia (SAB), and so on [2,5,6] . Among these infections, SAB, causing 10-30 cases per 100000 people per year with the mortality rate is as high as 25% [7] , draws a great amount of medical workers' attention [6] . Patients with SAB exhibit a range of diseases and outcomes of varying severity. Studies have shown that some patients have eliminated pathogens in rst-line treatment, while others have not resolved the infection problem. Persistent bacteremia leads to an immune response imbalance in the host and is associated with a mortality rate of 20-30% after the illness [7] . In terms of treatment, the wild spread of antibiotic-resistant strains aggravates the disease and makes treatment more di cult.
Saureus bacteremia can be summarized as a systemic bacteria spreading disease, which may lead to systemic infection [8] . So, it's reasonable to consider that entering and crossing through the endothelium is the vital link of S. aureus bacteremia [4,8] . Previous studies show that Staphylococcus aureus can in uence endothelial barrier function through inducing an in ammatory response, apoptosis, microtubule destabilization, and so on [8,9] . However, it is well known that there are many different strains of Staphylococcus aureus causing staphylococcal bacteremia. Further studies, using proteomic methods [5] and bacteriological methods [10] to demonstrate the existence of differences in the interaction between S. aureus and the endothelium among different strains may facilitate an in-depth study of the molecular mechanisms underlying the infection process of the highly infectious strains.
This study intended to analyze the differences in gene expression between S. aureus strains with different endothelial invasion abilities. To understand the difference in gene expression between different strains of S. aureus, it is necessary to further understand how it infects the endothelial barrier function, to nd new multi-targets to block this progress, which may provide some molecular basis for the research on antibiotic resistance. In addition, the discovery of marker gene symbols may help determine the risk of S. aureus bacteremia in the event of local infection, which is of auxiliary signi cance for clinical diagnosis and treatment.

Microarray data
We searched the GEO database (https://www.ncbi.nlm.nih.gov/gds/) for the appropriate data set for our study, using the keywords "staphylococcus aureus" and "endothelial cells". Gene expression pro les of the GSE13736 and GSE82036 were downloaded from the GEO database for our study. Dataset GSE13736, based on the GPL570 platform [HG-U133_Plus_2] Affymetrix Human Genome U133 Plus 2.0 Array, contains samples of human umbilical vein endothelial cells platform Illumina HumanHT-12 V4.0 expression beadchip and contained 24 samples of HUVEC infected by different strains of S. aureus. Among the 24 samples, we selected 21 for our study (mock infection groups were excluded). The detailed information is listed in Table 1.

Identi cation of differentially expressed genes (DEGs)
We downloaded, calibrated and standardized the relative data les and then processed them with R package (R Foundation for Statistic Computing). The Fold Change (FC), P-Value and false discovery rate were calculated to screen DEGs between HUVEC infected by different strains of S. aureus and HUVEC went through monk infection. Finally, the cutoff point of DEGs was |log 2 fold change (FC)| > 1 and P-value < 0.05. Besides, hierarchical cluster analysis was employed to show the heat map ( Fig. 1) and volcano of DEGs (Fig. 2) identi ed above.

Functional enrichment analysis of DEGs
Gene ontology (GO) databases can help annotate genes and thus learn about relevant gene function. It contains three sections: biological progress (BP), cellular component (CC) and molecular function (MF). Kyoto Encyclopedia of Genes and Genomes (KEGG) is another commonly used database that helps users learn about gene functions. In our study, we make use of the GO and KEGG databases to further investigate the function of DEGs. We performed GO analysis and KEGG analysis on strains 6850, 8325-4, K70058396, and K1801/10, respectively, and the GO analysis chart and KEGG analysis chart for each strain were listed as one chart (Fig. 3, Fig. 4, Fig. 5, Fig. 6).

PPI network construction and key module identi cation
The PPI network of these DEGs was constructed using the Search Tool for the Retrieval of Interacting Genes/Proteins (STRING), an online database that helps assess and integrate protein-protein interaction (PPI) information. Besides, interaction is admitted when the interaction score is > 0.4. The protein interaction network was also adjusted and visualized by Cytoscape. Subsequently, the CytoHubba plug-in was also utilized to lter hub genes and the top 10 hub. The genes with the highest degree of connectivity within the PPI network were gured out and shown in the interaction network (Fig. 7).

Validation of gene expression levels
We found and used the data set GSE65088 in the geodatabase, and selected three samples from the control group and three samples from the infection group to verify the central genes screened by the CytoHubba plug-in, and used GraphPad Prism 8 software to produce veri cation maps related to the infection of four groups of S. aureus strains (Fig. 8).

Identi cation of DEGs
Two datasets (GSE13736, GSE82036) were obtained from the GEO database and analyzed in the R language. A total of 7 groups of DEGs analysis were  (Table 2). Volcano plot and heat map of each group of DEGs were also generated respectively (Figure 1

GO functional enrichment analysis and KEGG pathway analysis of DEGs
With the help of the GO and KEGG database, we analyzed the enrichment of each group of DEGs respectively. The most signi cant biological processes for each group of DEGs enrichment are response and defense response to the virus, response to interferon−gamma and they are shown in Figure 3,4,5,6. As for cellular components (CC), organelle outer membrane, outer membrane, proteasome core complex, beta−subunit complex are thought to be important in Strain 6850. Cytoplasmic ribonucleoprotein granule, ribonucleoprotein granule and proteasome core complex, beta−subunit complex are clearly enriched in strain Strain 8325-4. The aggregation of cellular components (CC) of Strain K70058396 is embodied in particles, such as cytoplasmic ribonucleoprotein granule, speci c granule and ribonucleoprotein granule. The cell component enrichment of the Strain K1801/10 is mainly re ected in membrane raft, membrane microdomain and membrane region ( Figure. 6). In addition, in terms of molecular function, for strain 6850, DEGs is mainly enriched and expressed in terms of receptor-ligand activity, signaling receptor activator activity and cytokine receptor binding; For Strain 8325-4, DEGs was mainly and signi cantly expressed in cytokine receptor binding, cytokine activity, and double−stranded RNA binding. For Strain K70058396, the differential genes were mainly enriched in receptorligand activity, signal receptor activation activity and cytokine activity. Finally, for Strain K1801/10, DEGs were signi cantly enriched in DNA binding transcription activator activity, RNA polymerase II speci c, and DNA binding transcription activator activity. The last but not the least, the KEGG pathway analysis showed that the overlapping DEGs of each group were mainly enriched in In uenza A, NOD−like receptor signaling pathway, TNF signaling pathway, and Epstein−Barr virus infection, respectively. Through these functions enrich the analysis, we further study endothelial cells reactions toward different strains of S. aureus. What's more, we calculated the Degree using CytoHubba plug-in and found out 10 Hub genes in each PPI network and showed their interaction.

Validation of gene expression levels
We performed validation experiments on the genes screened by the CytoHubba plug-in and found that Strain K1801/10 and Strain K70058396 were signi cantly expressed in S. aureus strains that equip themselves with invasiveness to HUVEC compared to strains with no invasiveness. These signi cantly expressed genes may play an important role in the invasion of endothelial cells by S. aureus. In addition, the central genes screened after the differential expression of S. aureus strain invading endothelial cells did not exactly match the hub genes that were signi cantly differentially expressed in the validation data set, which may further suggest that the bacteremia caused by the endothelial invasion and whole blood infection was not identical.

Discussion
Staphylococcus aureus, as the most common pathogen in human pyogenic infection, can cause local pyogenic infection or severe systemic infection. In this paper, we analyzed the gene expression of infections caused by the invasion of different strains of Staphylococcus aureus in human endothelium from the perspective of bioinformatics, which opened up a new idea for the molecular mechanism of disease infection and the early diagnosis and treatment in clinical practice.
In this study, we analyzed two datasets in the GEO database: GSE13736 and GSE82036. Samples with differential expression were selected for R analysis, and GO analysis and KEGG analysis were performed. We found that the cellular components, molecular functions and biological processes signi cantly expressed in GO analysis were related to the expression of KEGG pathways. At the same time, the central genes screened from PPI networks were signi cantly enriched in the related pathways and affected the expression of molecular functions and biological processes. Strain K70058396: The different genes of Strain K70058396 mainly included RSAD2, DDX58, IFIT3, IFIH1 and so on. Among them, RSAD2 (eristostatin) is an evolutionally-conserved and interferon-induced protein, participating in the innate immune responses of cells to a variety of viruses in a variety of ways, such as regulating cellular signaling process. Especially, Epstein−Barr virus infection was signi cantly enriched in this study. The proteins get encoded by the Epstein−Barr virus, involved in viral replication and expression of virus particles packing and adjust the immune response process of the host cell. This may be related to the pathway through which S. aureus invades the endothelium to cause signi cant expression and functional changes of cell particles [11] .
Additionally, DDX58 is a defense reactive protein that is associated with constitutive upregtion of type I interferon [12,13] . Studies have shown that type I interferon increases the expression of functional tumor necrosis factor-associated apoptotic ligand (TRAIL), which may be related to the viral defense response.
IFIT3 is an IFN-induced antiviral protein, and the PPI network shows its signi cant expression after the invasion of human endothelial tissues [14] . At the same time, the In uenza A, Epstein−Barr virus infection pathways enriched in KEGG may be related to the signi cant expression of viral genes caused by the interaction between cytokines. The enrichment of ribonucleoprotein granule in the cell fraction may be related to the cosmids of IFIT3 and ribosomes from lysates of infected cells [15] .
In addition, IFIH1 heightens antiviral response in vitro and enhanced virus control in vivo [16] . The enrichment pathway obtained by KEGG analysis con rmed that S. aureus could be the main pathogen causing the spread of the in uenza virus, with a possible pro-in ammatory effect [17] . According to GO analysis, after S. aureus invades human endothelial cells, the body will initiate the biological response process to the virus and carry out selfdefense. In addition, the enrichment of the tumor necrosis factor signaling pathway in the KEGG assay showed that it was active as a common response to bacterial infection, with signi cant upregulation of both the protein-encoding genes, IFIT3 and ISG15.
Studies have shown that IFIT2, belonging to the same family as IFITT3, can enhance the expression of factors that selectively target and inhibit viral mRNA transcription cells to invade proteins and nucleic acids into pathogens. Therefore, we speculate that the RNA bound by IFITT3 is important to enhance viral gene expression [15] .
As an interferon-induced protein, ISG15 inhibits virus replication, regulates host injury, repair response, host immunity and other signal pathways. ISG15 can be expressed under the stimulation of numerous interferons, thus encoding anti-viral mediators and establishing a cellular anti-viral state in which there is a process of signal receptor activation and interaction between cytokines. The expression is con rmed in the molecular function of GO analysis [18] .
In Strain K1801/10, MX1 is an intracellular antiviral protein that activates the type I and type III interferon signals and can inhibit viral replication by blocking the transcription of viral RNA [19] . In viral replication, the MX1 protein interacts with viral nucleoproteins (NP) and PB2 to affect polymerase activity and provide interspeci c limitations [20] . In addition, the high expression of MX1 protein is also associated with tumor invasion [21] .
However, this study has certain limitations: for example, the highly infectious strain described in this study, Strain K70058396, was obtained by cell experiment and does not represent the strain that actually cause Staphylococcus aureus bacteremia in clinical practice. The sample size of each strain is small; No further wet experimental veri cation was performed. Central gene recognition facilitates the selection of biomarkers and targets genes for S. aureus invasion of endothelial cells, which may provide new ideas and directions for the early detection, control and treatment of infection in patients clinically.
Based on the above analysis, we speculate that there may be overlapping pathways between S. aureus and EB virus and the in uenza virus in the process of infection that invade endothelial cells. Their co-infection aggravates the damage to endothelial cells, which may be related to the fact that highly invasive strains are more likely to cause bacteremia after infecting the endothelium.

Conclusion
In this study, we found that endothelial cells have differential expression in response to strains with different invasion abilities. In response to the Strain K70058396 strain infection, the differential gene expression has a certain speci city. At the same time, RSAD2, DDX58, IFITT3, and IFIH1 play a key role in the endothelial cell response. Further studies on this basis may be bene cial for early recognition of Staphylococcus aureus infections and selection of action targets for treatment.

Declarations
Editorial policies Not applicable.

Consent for publication
All of the authors have read and approved the content, and agree to submit the whole article in your journal.

Availability of data and materials
The GSE13736 and GSE82036 datasets used during the current study are available from GEO database(https://www.ncbi.nlm.nih.gov/gds/).