Identifcation of hub genes and pathways in psoriasis through bioinformatics and validation by RT-qPCR

Ling Ai Zou Huangshi Central Hospital, Afliated Hospital of Hubei Polytechnic University, Edong Health Care Group https://orcid.org/0000-0001-5357-9747 Qichao Jian (  [email protected] )

Research

Keywords: Psoriasis, Pathogenesis, Hub genes, Pathways, CXCL10

Posted Date: March 30th, 2021

DOI: https://doi.org/10.21203/rs.3.rs-357690/v1

License:   This work is licensed under a Creative Commons Attribution 4.0 International License. Read Full License

Page 1/23 Abstract Background

Although several studies have attempted to investigate the aetiology and mechanism of psoriasis, the precise molecular mechanism remains unclear. Our study aimed to identify the hub genes and associated pathways that promote its pathogenesis in psoriasis, which would be helpful for the discovery of diagnostic and therapeutic markers.

Methods

GSE30999, GSE34248, GSE41662, and GSE50790 datasets were extracted from the Gene Expression Omnibus (GEO) database. The GEO profles were integrated to obtain differentially expressed genes (DEGs) using the affy package in R software, with |logFC|> 1.5 and adjusted P < 0.05. The DEGs were utilised for Gene ontology (GO), Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway, and protein-protein interaction (PPI) network analyses. Hub genes were identifed using Cytoscape and enriched for analysis in www.bioinformatics.com.cn. These hub genes were validated in the four aforementioned datasets and M5-induced HaCaT cells using real-time quantitative polymerase chain reaction (RT-qPCR).

Results

A total of 359 DEGs were identifed, which were mostly associated with responses to bacterium, defence responses to other organism, and antimicrobial humoral response. These DEGs were mostly enriched in the steroid hormone biosynthesis pathway, NOD-like receptor signaling pathway, and cytokine-cytokine receptor interaction. PPI network analysis indicated seven genes (CXCL1, ISG15, CXCL10, STAT1, OASL, IFIT1, and IFIT3) as the probable hub genes of psoriasis; CXCL10 had a positive correlation with the other six hub genes. The chord plot results further supported the GO and KEGG analysis results of the 359 DEGs. Seven predicted hub genes were validated to be upregulated in four datasets and M5-induced HaCaT cells using RT-qPCR.

Conclusions

The pathogenesis of psoriasis may be associated with seven hub genes (CXCL1, ISG15, CXCL10, STAT1, OASL, IFIT1, and IFIT3) and pathways, such as the NOD-like receptor signaling pathway and cytokine- cytokine receptor interaction. These hub genes, especially CXCL10, can be used as potential biomarkers in psoriasis.

Background

Page 2/23 Psoriasis is a chronic immune-mediated infammatory skin disease with global prevalence of 2- 3% [1,2]. The occurrence of psoriasis is a complex interaction among genetics, immunology, and environment [1,3]. Although there are plenty of studies to investigate the etiologies and mechanisms, the precise molecular mechanism of psoriasis remains unclear [3]. Therefore, a thorough exploration of psoriasis’ pathogenesis is helpful for discovery of potential biomarkers and provides novel clues for diagnosis and treatment.

Up to now, microarray technology has been extensively utilized to analyze gene expression in many diseases, including psoriasis disease [4,5]. Gene Expression Omnibus (GEO) is a comprehensive public database, which contains various disease gene expression datasets [4]. Thus, bioinformatics analysis can mine the hub genes and associated pathways from GEO database, and provide a novel approach for exploring the molecular mechanism of psoriasis.

In the current study, we merged four independent gene expression datasets (GSE30999 [6], GSE34248 [7], GSE41662 [7] and GSE50790 [8]) from the GEO database using the affy package in R software, and identifed differentially expressed genes (DEGs) of the merged datasets. Then the DEGs were performed for a series of analyses, such as Gene ontology (GO) enrichment, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, protein-protein interaction (PPI) network analysis, and hub genes prediction.

In order to verify these predicted hub genes, we compared the expression levels of these genes between the lesional and non-lesional psoriatic skin in the four datasets, and performed RT-qPCR experiments in M5 induced HaCaT cells. Since some studies confrmed M5 (IL-1a, IL-17A, IL-22, TNF-a, and oncostatin M) can induce a better psoriasis cell model than other cytokines, we use M5 to mimic the hyperproliferative pathological state of psoriasis keratinocytes in our experiments [9-14]

Methods

Microarray data source

Fig. 1 shows the overall workfow of this study. Using the key words ‘psoriasis’, ‘lesion*’, ‘Tissue’, ‘Homo sapiens’ and ‘Expression profling by array’ were screened in the National Center of Biotechnology Information (NCBI) - GEO (https://www.ncbi.nlm.nih.gov/geo/). The raw profles of GSE30999 [6], GSE34248 [7], GSE41662 [7], and GSE50790 [8], were downloaded from the NCBI GEO database. The detailed information of four enrolled microarray datasets is shown in Table 1. In GSE30999, GSE34248, GSE41662, and GSE50790, only psoriatic lesional skin samples and their matched non-lesional skin samples were selected, which resulted in 127 pairs of skin samples from psoriasis patients for subsequent analysis.

DEGs analysis and integration

After the raw data of four datasets were collated, the analyses were performed in the R programming environment (version4.0.2.). The affy package was used to preprocess the collated data by conducting

Page 3/23 background correction, normalization and expression calculation [15]. Then the ComBat function of sva package was applied for eliminating batch effect between datasets [16]. And the probes were annotated by matrix data combined with HG-U133_Plus_2-na36 annotation fle. Lastly we used the limma package of R language to identify the differential expression genes(DEGs). The signifcant differential expression cut-off was set as |logFC| >1.5 and adjusted P<0.05, which were demonstrated as Volcano plots. Furthermore, a heatmap was drawn to show the clustering of samples.

GO and KEGG pathway analysis of DEGs

Metascape (http://metascape.org) is a free, powerful and online analysis tool for gene annotation and analysis [17]. It is also an automated meta-analysis tool to annotate genes, conduct enrichment analysis and construct protein-protein interaction networks. In this study, Gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were enriched based on Metascape. GO terms included biological process, cellular component, and molecular function. The cutoff criterions were P-value < 0.05, minimum overlap of 3, and minimum enrichment of 1.5. The top 20 enrichment Go terms and KEGG pathways in our study were listed as fgures and tables respectively.

Protein-protein interaction(PPI) network construction and hub genes identifcation

The Search Tool for the Retrieval of Interacting Genes/Proteins (STRING, version 11.0, http://www.string- db.org/) was utilized to construct a PPI network for proteins encoded by DEGs, and the minimum required interaction score was set as 0.7 (referred as high confdence) [18]. In order to further analyze the results from STRING database, Cytoscape 3.8.0 was used as a tool to visualize the PPI network [19]. The modules in the PPI network were extracted using the Cluster Finding algorithm in MCODE with a max depth of 100, a node score cutoff of 0.2, a degree cutoff of 2, and a k-core of 2. The hub genes were identifed by CytoHubba, which was a useful plugin in Cytoscape. It can provide 12 topological analysis methods to explore top 10 genes. Specifcally, these methods are Maximal Clique Centrality (MCC), Density of Maximum Neighborhood Component (DMNC), Maximum Neighborhood Component (MNC), Degree, Edge Percolated Component (EPC), BottleNeck, EcCentricity, Closeness, Radiality, Betweenness, Stress, and ClusteringCoefcient. A hub gene can be identifed, if this gene was predicted as one of the top 10 genes in all the 12 methods. Additionally, interactions between hub genes were analyzed back in STRING database.

Functional enrichment analysis of hub genes

GO and KEGG analyses of 359 DEGs were performed by Metascape, including seven hub genes. Then the functional enrichment analysis of seven hub genes were explored by http: //www. bioinformatics. com.cn, an online platform for data analysis and visualization.

Validation of hub genes in datasets

Expression data of hub genes were extracted from GSE30999, GSE34248, GSE41662, and GSE50790. The correlations between CXCL10 and the other hub genes were confrmed in the four aforementioned Page 4/23 datasets. Then the differential expression of these hub genes was validated in these four datasets.

Cell culture and treatment

HaCaT cell line was purchased from Shanghai iCell Bioscience Inc. HaCaT cells were cultured in DMEM(Dulbecco's Modifed Eagle's Medium) with 10% fetal bovine serum and 1% penicillin and streptomycin, in a humidifed incubator at 37 °C and 5% CO2. Then the cells were treated with 10 ng/ml M5 (IL-1α [PeproTech, 200-01A], IL-17A [PeproTech, 200–17], IL-22 [PeproTech, 200–22], TNF-α [PeproTech, 300-01A], Oncostatin M [Peprotech, 300–10]) for 48 h, to establish the psoriasis cell model.

Real-time quantitative PCR (RT-qPCR)

Total RNA was extracted from M5 treated HaCaT cells by TRIpure Total RNA Extraction Reagen t(#EP013, ELK Biotechnology, China). Total RNA was reverse transcribed using EntiLink™ 1st Strand cDNA Synthesis Kit(#EQ003, ELK Biotechnology, China) according to the manufacturer's protocols. The expression levels of the seven predicted hub genes in Hacat cells(without M5 stimulation, control) and M5 treated HaCat cells (psoriasis cell model), were detected by RT‑qPCR using EnTurbo™ SYBR Green PCR SuperMix(#EQ001, ELK Biotechnology, China) on the StepOne™ Real-Time PCR System(Life technologies, USA). GAPDH was used as an internal control, and the mRNA levels were calculated using 2−△△CT method. The primer sequences are shown in Table 2.

Statistical analysis

The data extracted from GEO datasets were examined by the normality test and homogeneity of variance test. Spearman's correlation analysis was utilized to investigate the correlation between CXCL10 and the other hub genes. T-testing and analysis of variance (ANOVA) testing were used for comparisons between two groups and among three or more groups respectively. GraphPad Prism 8.0.1 was used to perform the above tests.

Results

Identifcation of differential expression genes (DEGs)

A total of 359 DEGs were screened between paired lesional skin and non-lesional skin groups with |logFC|>1.5 and adjusted P<0.05, among which 284 genes were upregulated and 65 genes were downregulated. The volcano plots of DEGs has been shown in Fig. 2, and the heatmap of DEGs has been shown in Fig. 3.

GO and KEGG pathway analysis of DEGs

GO and KEGG pathway enrichment analysis were performed in Metascape. The top 20 GO enrichment items were classifed into two functional groups: biological process group (15 items), and molecular function group (5 items) (Fig. 4A, B and Table 3). The DEGs were mainly enriched in six clustering groups

Page 5/23 in the GO enrichment analysis, including response to bacterium, defense response to other organism, antimicrobial humoral response, favonoid glucuronidation, skin development, monocarboxylic acid metabolic process (Fig. 4A). The top 20 KEGG pathways enrichment items are shown in Fig. 4C, D and Table 4. The DEGs were mainly enriched in three clustering groups in the KEGG pathways enrichment analysis, including Steroid hormone biosynthesis, NOD-like receptor signaling pathway, Cytokine-cytokine receptor interaction (Fig. 4C).

PPI network construction and hub genes identifcation

The online database STRING was used to obtain PPI information on the 359 DEGs, was constructed with use of Cytoscape software (Fig. 5A). Then, the MCODE was used to identify functional modules. There were six modules, of which cluster 1 consisted of 28 nodes and 196 edges (Fig. 5A, B), including GAL, CXCL13, GBP1, IFIT3, RTP4, IFIT1, MX1, OASL, IFI44, ISG15, CCL27, OAS1, RSAD2, CXCR2, CXCL8, IFI44L, IRF7, STAT1, CXCL1, PTGER3, OAS2, HERC6, IFI6, IFI27, CCL20, CXCL9, CXCL10 and CXCL2(Table 5). Seven genes, namely CXCL1, ISG15, CXCL10, STAT1, OASL, IFIT1 and IFIT3, were identifed as the hub genes by the 12 analysis methods in CytoHubba. The detailed information of the seven hub genes was shown in Table 6.

Subsequently, the correlation of these seven hub genes were analyzed in STRING database (Fig. 5C). In Fig. 5C, red line means known interaction that has been determined experimentally, and skyblue line represents known interaction from curated databases, these genes including ISG15, STAT1, OASL, IFIT1 and IFIT3. Moreover, yellow green line means text mining evidence and black line represents co- expression, these genes including CXCL10, CXCL1, ISG15, STAT1, OASL, IFIT1 and IFIT3. Light blue line stands for protein homology, these genes including IFIT1 and IFIT3.

Functional enrichment analysis of seven hub genes

Seven hub genes were explored by Cytoscape (Table VI). GO and KEGG analyses of seven hub genes were performed by http: //www. bioinformatics. com.cn. The hub genes were enriched in the top fve of these terms, which based on their adjusted P-values were shown in chord plots (Fig 6, Table 7 and Table 8). For GO analysis，the top fve terms were response to bacterium, defense response to other organism, antimicrobial humoral response, favonoid glucuronidation, and chemokine receptor binding(Fig. 6A, Table 7). For KEGG pathway analysis, CXCL10, CXCL1, ISG15, STAT1, and IFIT1 were mostly enriched in NOD-like receptor signaling pathway, Cytokine-cytokine receptor interaction, and Amoebiasis, except for OASL and IFIT3(Fig. 6B). OASL and IFIT3 weren’t enriched in the top fve KEGG pathways (Table 8).

Correlation between CXCL10 and the other hub genes

Among the correlation of these seven hub genes, CXCL10 was in the central of the network (Fig. 5C), and had a higher rank (Table V), suggesting the correlation of CXCL10 with other hub genes. CXCL10 showed positive correlation with other six hub genes in GSE30999, GSE34248, GSE41662 and GSE50790 (Fig. 7).

Validation of the expression of hub genes in datasets Page 6/23 The seven hub genes underwent expression validation in GSE30999, GSE34248, GSE41662 and GSE50790. Except for CXCL1, STAT1, and OASL in GSE30999, all other hub genes were signifcantly up- regulated in psoriatic lesional skin samples from the four datasets (P < 0.05, 0.01, 0.001, or 0.0001; Fig. 8). This may be due to small samples in GSE50790(only 4 paired samples).

Validation of hub genes by qRT-PCR

The expression levels of seven predicted hub genes were validated by RT-qPCR(n=3). The results of qRT- PCT showed that the transcription levels of CXCL10, CXCL1, ISG15, STAT1, OASL, IFIT1 and ITIT3 were signifcantly up-regulated in 10 ng/ml M5 induced HaCaT cells (Fig. 9). The expression levels of these genes were consistent with the microarray results.

Discussion

An important fnding of this study was the identifcation of hub genes and associated pathways in psoriasis through bioinformatics and RT-qPCR. The predicted seven hub genes contained CXCL10, CXCL1, ISG15, STAT1, OASL, IFIT1, IFIT3, and the mostly enriched pathways were NOD-like receptor signaling pathway, and Cytokine-cytokine receptor interaction. These hue genes and pathways may be vital for the pathogenesis of psoriasis. Some studies reported several genes as hub genes in psoriasis, partly the same with ours [20-22].

Among the seven hub genes, CXCL10 was a higher rank gene and had positive correlations with other six hub genes, suggesting it might be the most valuable gene. Furthermore, the RT-qPCR result of CXCL10 confrmed the prediction. CXCL10, also named IFN-γinduced protein 10 (IP-10), is a member of the CXC family of chemokines, which is considered to play an important role in infammation via its T-cell chemotactic and adhesion properties [23]. Researches indicate CXCL10 is upregulated in the psoriatic skin lesions and serum [23,24]. It has been hypothesized that CXCL10 could be a good marker for psoriasis [25]. Although CXCL10 and CXCL1 belong to the chemokine “CXC” family [26], they play different roles in psoriasis. CXCL10 attracts T helper (Th) 1 cells, and CXCL1 attracts neutrophils [27]. It is reported that CXCL10 production by keratinocytes depends on STAT1 [28]. STAT1 is known as a member of the STAT family, involving in type I and type II interferon (IFN) signalling [29]. The expression of STAT1 is increased in lesional psoriatic skin, and it shows an important role in the pathogenesis of psoriasis [30].

The remaining four genes(ISG15, OASL, IFIT1, IFIT3) are antiviral genes, which may explain the relatively few skin viral infections found in the psoriasis patients. Ubiquitin-like protein ISG15 is an interferon- stimulated protein that has a critical role for the control of microbial infections [31,32]. Some studies reported that ISG15 was elevated in lesional psoriatic skin compared to atopic dermatitis skin and healthy skin[33]. IFN-inducible oligoadenylate synthetases-like (OASL) protein belongs to atypical oligoadenylate synthetase (OAS) family, possesses antiviral activity, and boosts the innate immunity[34,35]. Gao LJ et al. also identifed OASL as a hub gene [21], but it is rarely studied in psoriasis patients. Additionally, OASL wasn’t enriched in the top 5 pathways in our study. Although the expression Page 7/23 of OASL was up-regulated in M5 induced HaCat cells in the present study, further research is needed to confrm our results, including clinical specimens and animal experiment.

The interferon-induced proteins with tetratricopeptide repeats (IFITs) included IFIT1, IFIT2, IFIT3, and IFIT5 [36]. IFIT1 and IFIT3 are members of IFIT family, which regulate immune responses and function as essential anti-viral proteins [37]. The study indicated that IFIT3 binding to IFIT1 was vital for stabilizing IFIT1 expression, and was indispensable for inhibiting infection of viruses lacking 2’-O methylation [37]. In this study, IFIT1 was enriched in NOD-like receptor signaling pathway, but IFIT3 wasn’t. Currently there are few researches on IFITs in psoriasis, while IFIT1 and IFIT3 are overexpressed in oral squamous cell carcinoma that promote tumor growth, regional and distant metastasis [36]. Therefore, this new fnding warrants further study.

Nucleotide-binding and oligomerization domain (NOD)-like receptor is a kind of pattern-recognition receptor. It associates with various diseases related to infection and immunity [38]. Some studies show NOD-like receptor signaling pathway is highlighted in psoriatic epidermis [39]. Meanwhile, some researches have found cytokine-cytokine receptor interaction is related to the pathogenesis of psoriasis by combined transcriptomic analysis [40]. These results are consistent with ours.

However, there are some limitations in this study. For example, this study lacks clinical human specimens. Therefore, further experiments should be performed to confrm whether these hub genes can be used as biomarkers in psoriasis.

In conclusion, through an integrative bioinformatic analysis of the GEO data, this study identifed important hub genes and signaling pathways. These hub genes might be used as potential biomarkers in psoriasis.

Conclusions

The pathogenesis of psoriasis may associate with the seven hub genes (CXCL1, ISG15, CXCL10, STAT1, OASL, IFIT1, IFIT3) and pathways such as NOD-like receptor signaling pathway, and Cytokine-cytokine receptor interaction. These hub genes may be used as potential biomarkers in psoriasis, especially CXCL10.

Declarations

Acknowledgements

We would like to thank Editage (www.editage.cn) for English language editing.

Author Contributions

Ailing Zou designed the study, analyzed the data, prepared fgures and tables, authored or reviewed drafts of the paper, and approved the fnal draft.

Page 8/23 Qichao Jian authored or reviewed drafts of the paper, and approved the fnal draft.

Funding

Not applicable.

Availability of data and materials

The following information was supplied regarding data availability:

The data is available at NCBI GEO: GSE30999, GSE34248, GSE41667 and GSE50790.

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Abbreviations

RT-qPCR: real-time quantitative polymerase chain reaction; GEO: Gene Expression Omnibus; GO: gene ontology; KEGG: Kyoto Encyclopedia of Genes and Genomes; DEGs: diferentially expressed genes; DEGs: diferentially expressed genes; MCODE: Molecular Complex Detection; IP-10: IFN--induced protein 10; OAS: oligoadenylate synthetase; OASL: oligoadenylate synthetases-like; IFITs: IFN-induced proteins with tetratricopeptide repeats.

References

1. Perera GK, Di Meglio P, Nestle FO. Psoriasis. Annu Rev Pathol, 2012, 7: 385-422. 2. Michalek IM, Loring B, John SM. A systematic review of worldwide epidemiology of psoriasis. J Eur Acad Dermatol Venereol, 2017, 31(2): 205-212. 3. Boehncke WH, Schön MP. Psoriasis. Lancet, 2015, 386(9997): 983-994. 4. Chen D, Lu T, Tan J, et al. Identifcation of a transcription factor-microRNA network in esophageal adenocarcinoma through bioinformatics analysis and validation through qRT-PCR. Cancer Manag Res, 2019, 11: 3315-3326. 5. Delic D, Wolk K, Schmid R, et al. Integrated microRNA/mRNA expression profling of the skin of psoriasis patients. J Dermatol Sci, 2020, 97(1): 9-20.

Page 9/23 6. Suárez-Fariñas M, Li K Fau - Fuentes-Duculan J, Fuentes-Duculan J Fau - Hayden K, et al. Expanding the psoriasis disease profle: interrogation of the skin and serum of patients with moderate-to-severe psoriasis. J Invest Dermatol, 2012, 135(11): 2552-2564. 7. Bigler J, Rand Ha Fau - Kerkof K, Kerkof K Fau - Timour M, et al. Cross-study homogeneity of psoriasis gene expression in skin across a large expression range. PLoS One, 2013, 8(1): e52242. 8. Swindell WR, Xing X Fau - Stuart PE, Stuart Pe Fau - Chen CS, et al. Heterogeneity of infammatory and cytokine networks in chronic plaque psoriasis. PLoS One, 2012, 7(3): e34594. 9. Guilloteau K, Paris I, Pedretti N, et al. Skin Infammation Induced by the Synergistic Action of IL-17A, IL-22, Oncostatin M, IL-1{alpha}, and TNF-{alpha} Recapitulates Some Features of Psoriasis. J Immunol, 2010, 184(9): 5263-5270. 10. Mee JB, Johnson Cm Fau - Morar N, Morar N Fau - Burslem F, et al. The psoriatic transcriptome closely resembles that induced by interleukin-1 in cultured keratinocytes: dominance of innate immune responses in psoriasis. Am J Pathol, 2007, 171(1): 32-42. 11. Nograles KE, Zaba Lc Fau - Guttman-Yassky E, Guttman-Yassky E Fau - Fuentes-Duculan J, et al. Th17 cytokines interleukin (IL)-17 and IL-22 modulate distinct infammatory and keratinocyte- response pathways. Br J Dermatol., 2008, 159(5): 1092-1102. 12. Ma HL, Liang S Fau - Li J, Li J Fau - Napierata L, et al. IL-22 is required for Th17 cell-mediated pathology in a mouse model of psoriasis-like skin infammation. J Clin Invest, 2008, 118(2): 597-607. 13. Banno T, Gazel A Fau - Blumenberg M, Blumenberg M. Effects of tumor necrosis factor-alpha (TNF alpha) in epidermal keratinocytes revealed using global transcriptional profling. J Biol Chem, 2004, 279(31): 32633-32642. 14. Boniface K, Diveu C Fau - Morel F, Morel F Fau - Pedretti N, et al. Oncostatin M secreted by skin infltrating T lymphocytes is a potent keratinocyte activator involved in skin infammation. J Immunol, 2007, 178(7): 4615-4622. 15. Gautier L, Cope L Fau - Bolstad BM, Bolstad Bm Fau - Irizarry RA, et al. affy--analysis of Affymetrix GeneChip data at the probe level. Bioinformatics., 2004, 20(3): 307-315. 16. Leek JT, Johnson We Fau - Parker HS, Parker Hs Fau - Jaffe AE, et al. The sva package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics. , 2012, 28(6): 882-883. 17. Zhou Y, Zhou B, Pache L, et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun., 2019, 10(1): 1523. 18. Szklarczyk D, Gable AL, Lyon D, et al. STRING v11: protein-protein association networks with increased coverage, supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res., 2019, 47(D1): D607-D613. 19. Shannon P, Markiel A Fau - Ozier O, Ozier O Fau - Baliga NS, et al. Cytoscape: a software environment for integrated models of biomolecular interaction networks. Genome Res. , 2003, 13(11): 2498-2504. 20. Choudhary S, Pradhan D, Khan NS, et al. Decoding Psoriasis: Integrated Bioinformatics Approach to Understand Hub Genes and Involved Pathways. Curr Pharm Des., 2020, 26(29): 3619-3630. Page 10/23 21. Gao LJ, Shen J, Ren YN, et al. Discovering Novel Hub Genes and Pathways Associated with the Pathogenesis of Psoriasis. Dermatol Ther, 2020: e13993. 22. Luo Y, Luo Y, Chang J, et al. Identifcation of candidate biomarkers and pathways associated with psoriasis using bioinformatics analysis. Hereditas, 2020, 157(1): 30. 23. Boorsma DM FJ, Sampat S, Ottevanger C, de Haan P, Hooft L, Willemze R, Tensen CP, Stoof TJ. Chemokine IP-10 expression in cultured human keratinocytes. Arch Dermatol Res, 1998, 290(6): 335- 341. 24. Antonelli A, Fallahi P Fau - Delle Sedie A, Delle Sedie A Fau - Ferrari SM, et al. High values of Th1 (CXCL10) and Th2 (CCL2) chemokines in patients with psoriatic arthtritis. Clin Exp Rheumatol, 2009, 27(1): 22-27. 25. Ferrari SM, Ruflli I, Colaci M, et al. CXCL10 in psoriasis. Adv Med Sci, 2015, 60(2): 349-354. 26. Pohl D AC, Borská L, Fiala Z, Hamáková K, Ettler K, Krejsek J. CC and CXC chemokines patterns in psoriasis determined by protein array method were infuenced by Goeckerman's therapy Acta Medica (Hradec Kralove), 2009, 52(1): 9-13. 27. Méhul B, Laffet G, Séraïdaris A, et al. Noninvasive proteome analysis of psoriatic stratum corneum refects pathophysiological pathways and is useful for drug profling. Br J Dermatol., 2017, 177(2): 470-488. 28. Kanda N, Watanabe S. Prolactin enhances interferon-gamma-induced production of CXC ligand 9 (CXCL9), CXCL10, and CXCL11 in human keratinocytes. Endocrinology, 2007, 148(5): 2317-2325. 29. Bai L, Fang H, Xia S, et al. STAT1 activation represses IL-22 gene expression and psoriasis pathogenesis. Biochem Biophys Res Commun. , 2018, 501(2): 563-569. 30. Hald A, Andrés Rm Fau - Salskov-Iversen ML, Salskov-Iversen Ml Fau - Kjellerup RB, et al. STAT1 expression and activation is increased in lesional psoriatic skin. Br J Dermatol., 2013, 168(2): 302- 310. 31. Freitas BT, Scholte FEM, Bergeron É, et al. How ISG15 combats viral infection. Virus Res, 2020, 286: 198036. 32. Perng YC, Lenschow DJ. ISG15 in antiviral immunity and beyond. Nat Rev Microbiol, 2018, 16(7): 423-439. 33. Raposo RA, Gupta R, Abdel-Mohsen M, et al. Antiviral gene expression in psoriasis. J Eur Acad Dermatol Venereol, 2015, 29(10): 1951-1957. 34. Zhu J, Ghosh A, Sarkar SN. OASL—a new player in controlling antiviral innate immunity. Current Opinion in Virology, 2015, 12: 15-19. 35. Zhao C, Zheng S, Zhu D, et al. Identifcation of a novel porcine OASL variant exhibiting antiviral activity. Virus Res., 2018, 244: 199-207. 36. Pidugu VK, Wu MM, Yen AH, et al. IFIT1 and IFIT3 promote oral squamous cell carcinoma metastasis and contribute to the anti-tumor effect of geftinib via enhancing p-EGFR recycling. Oncogene, 2019, 38(17): 3232-3247.

Page 11/23 37. Johnson B, VanBlargan LA, Xu W, et al. Human IFIT3 Modulates IFIT1 RNA Binding Specifcity and Protein Stability. Immunity., 2018, 48(3): 487-499.e485. 38. Kim YK, Shin JS, Nahm MH. NOD-Like Receptors in Infection, Immunity, and Diseases. Yonsei Med J., 2016, 57(1): 5-14. 39. Tervaniemi MH, Katayama S, Skoog T, et al. NOD-like receptor signaling and infammasome-related pathways are highlighted in psoriatic epidermis. Sci Rep, 2016, 6: 22745. 40. Gao Y, Yi X, Ding Y. Combined Transcriptomic Analysis Revealed AKR1B10 Played an Important Role in Psoriasis through the Dysregulated Lipid Pathway and Overproliferation of Keratinocyte. Biomed Res Int, 2017, 2017: 8717369.

Tables

Table 1 Information of five enrolled microarray datasets EO GSE30999 GSE34248 GSE41662 GSE50790 ccession rganism Homo sapiens Homo sapiens Homo sapiens Homo sapiens issue Skin Skin Skin Skin latform GPL570 GPL570 GPL570 GPL570 ample Paired LS/NL Paired LS/NL Paired LS/NL Paired LS/NL air no. 85 14 24 4 ype Plaque psoriasis Plaque Plaque Plaque psoriasis psoriasis psoriasis itation Suárez-Fariñas, et al. Bigler, et al. Bigler, et al. Swindell, et al. (6) (7) (7) (8)

LS, lesional skin; NL, non-lesional skin

Table 2 PCR primer for Quantitative Real-Time PCR confirmation of hub genes Gene Forward primer(5− 3) Reverse primer(5− 3) GAPDH CATCATCCCTGCCTCTACTGG GTGGGTGTCGCTGTTGAAGTC CXCL10 AAGTGGCATTCAAGGAGTACCTC TGGATTCAGACATCTCTTCTCACC CXCL1 AACCGAAGTCATAGCCACACTC CTTCTCCTAAGCGATGCTCAAA ISG15 GGACAAATGCGACGAACCTC GCTCACTTGCTGCTTCAGGTG STAT1 ACTTTCCCTGACATCATTCGC TCTACAGAGCCCACTATCCGAG OASL AGCAAGGGTGTCTCCAAGAAG GGAGGCTGGTGTGCTATACAGT IFIT1 TCCACTATGGTCGGTTTCAGG TTGTAGACGAACCCAAGGAGG IFIT3 GAATCCTCTGAATGCATACTCCG AAACCCTCTAAACCATGTTGCAC

Table 3 The top 20 GO function enrichment items of DEGs

Page 12/23 O Category Description Count % Log10(P) Log10(q) O:0009617 BP response to bacterium 53 14.25 -22.54 -18.21 O:0098542 BP defense response to other 48 12.9 -22.25 -18.21 organism O:0019730 BP antimicrobial humoral response 23 6.18 -18.1 -14.22 O:0052696 BP flavonoid glucuronidation 9 2.42 -16.93 -13.18 O:0043588 BP skin development 31 8.33 -13.69 -10.48 O:0032787 BP monocarboxylic acid metabolic 33 8.87 -9.98 -7.24 process O:0001906 BP cell killing 17 4.57 -9.62 -6.91 O:0032103 BP positive regulation of response to 20 5.38 -7.45 -5.01 external stimulus O:0044282 BP small molecule catabolic process 23 6.18 -7.24 -4.82 O:0009611 BP response to wounding 29 7.8 -6.99 -4.58 O:0008202 BP steroid metabolic process 19 5.11 -6.9 -4.51 O:0016042 BP lipid catabolic process 19 5.11 -6.69 -4.3 O:0006766 BP vitamin metabolic process 12 3.23 -6.53 -4.16 O:0036230 BP granulocyte activation 23 6.18 -6.38 -4.02 O:0002831 BP regulation of response to biotic 12 3.23 -6.17 -3.82 stimulus O:0042379 MF chemokine receptor binding 16 4.3 -15.05 -11.66 O:0016491 MF oxidoreductase activity 31 8.33 -7.43 -5 O:0045236 MF CXCR chemokine receptor 6 1.61 -7.07 -4.65 binding O:0004252 MF serine-type endopeptidase activity 13 3.49 -6.27 -3.92 O:0004867 MF serine-type endopeptidase 10 2.69 -6.12 -3.79 inhibitor activity

GO, gene ontology; BP, biological process; MF, molecular function.

Table 4 The top 20 KEGG pathways of DEGs

Page 13/23 GO Category Description Count % Log10(P) Log10(q) ko00140 KEGG Steroid hormone 13 3.49 -12.22 -9.75 Pathway biosynthesis hsa04621 KEGG NOD-like receptor signaling 19 5.11 -11.11 -8.9 Pathway pathway hsa04060 KEGG Cytokine-cytokine receptor 24 6.45 -10.74 -8.68 Pathway interaction ko05146 KEGG Amoebiasis 9 2.42 -5.26 -3.91 Pathway ko03320 KEGG PPAR signaling pathway 7 1.88 -4.31 -3.01 Pathway hsa05219 KEGG Bladder cancer 5 1.34 -3.54 -2.29 Pathway ko04110 KEGG Cell cycle 7 1.88 -2.86 -1.65 Pathway hsa00630 KEGG Glyoxylate and 4 1.08 -2.84 -1.65 Pathway dicarboxylate metabolism ko04623 KEGG Cytosolic DNA-sensing 5 1.34 -2.82 -1.64 Pathway pathway ko05321 KEGG Inflammatory bowel disease 5 1.34 -2.76 -1.59 Pathway (IBD) ko00100 KEGG Steroid biosynthesis 3 0.81 -2.65 -1.51 Pathway hsa00240 KEGG Pyrimidine metabolism 6 1.61 -2.47 -1.37 Pathway hsa00380 KEGG Tryptophan metabolism 4 1.08 -2.36 -1.28 Pathway hsa04978 KEGG Mineral absorption 4 1.08 -2.27 -1.2 Pathway hsa05150 KEGG Staphylococcus aureus 4 1.08 -1.66 -0.68 Pathway infection hsa05205 KEGG Proteoglycans in cancer 7 1.88 -1.57 -0.61 Pathway hsa00565 KEGG Ether lipid metabolism 3 0.81 -1.56 -0.61 Pathway ko04972 KEGG Pancreatic secretion 4 1.08 -1.41 -0.48 Pathway ko04310 KEGG Wnt signaling pathway 5 1.34 -1.37 -0.45 Pathway ko00561 KEGG Glycerolipid metabolism 3 0.81 -1.36 -0.44 Pathway

KEGG, Kyoto Encyclopedia of Genes and Genomes; GEDs, differentially expressed genes.

Table 5 The detailed information of cluster networks in MCODE.

Page 14/23 Cluster Score Nodes Edges Node IDs (Density*#Nodes) 1 14.519 28 196 GAL, CXCL13, GBP1, IFIT3, RTP4, IFIT1, MX1, OASL, IFI44, ISG15, CCL27, OAS1, RSAD2, CXCR2, CXCL8, IFI44L, IRF7, STAT1, CXCL1, PTGER3, OAS2, HERC6, IFI6, IFI27, CCL20, CXCL9, CXCL10, CXCL2 2 12.167 13 73 DLGAP5, KIAA0101, KIF20A, CCNA2, CDC6, BUB1, CENPE, RRM2, CDK1, CCNB1, MCM10, TTK, SPC25 3 9.8 11 49 LCE3D, SPRR2B, IVL, SPRR3, SPRR1A, PKP1, DSG3, CDSN, DSC2, TGM1, PI3 4 6 9 24 TCN1, IL1B, IL17A, RAB27A, LRG1, HPSE, ARG1, LCN2, LTF 5 4 4 6 KRT16, KRT6A, KRT77, KRT6B 6 3 3 3 S100A12, S100A9, S100A8

Table 6 The detailed information of seven hub genes Occurrences in 12 Statistical methods Rank Gene Change LogFC Adj.P.Val by cytoHubba Full name(human) Symbol

1 CXCL1 up 3.4145 9.27E-42 8 C-X-C motif chemokine ligand 1 1 ISG15 up 1.9624 2.62E-38 8 ISG15 ubiquitin like modifier 2 CXCL10 up 2.3289 2.19E-27 7 C-X-C motif chemokine ligand 10 3 STAT1 up 1.7999 1.36E-47 6 signal transducer and activator of transcription 1 4 OASL up 3.7520 1.47E-66 5 2'-5'-oligoadenylate synthetase like interferon induced 4 IFIT1 up 1.7003 3.18E-33 5 protein with tetratricopeptide repeats 1 4 IFIT3 up 1.6149 4.02E-34 5 interferon induced protein with tetratricopeptide repeats 3

Table 7 The enrichment of seven hub genes in the top 5 GO terms.

Page 15/23 GO:0009617 GO:0098542 GO:0019730 GO:0052696 GO:0042379 logFC response to defense antimicrobial flavonoid chemokine bacterium response to humoral glucuronidation receptor other response binding organism 10 1.0 1.0 1.0 0.0 1.0 2.33 1 1.0 0.0 1.0 0.0 1.0 3.41 1.0 1.0 1.0 0.0 0.0 1.96 1 0.0 1.0 0.0 1.0 1.0 1.80 0.0 1.0 0.0 0.0 0.0 3.75 0.0 1.0 0.0 0.0 0.0 1.70 0.0 1.0 0.0 0.0 0.0 1.61

1 indicates the genes were present in GO terms, 0 indicates the genes were absent in GO terms.

Table 8 The enrichment of seven hub genes in the top 5 KEGG pathways. ko00140 hsa04621 hsa04060 ko05146 ko03320 logFC Steroid NOD-like Cytokine- Amoebiasis PPAR hormone receptor cytokine signaling biosynthesis signaling receptor pathway pathway interaction 10 0.0 1.0 1.0 0.0 0.0 2.33 1 0.0 1.0 1.0 1.0 0.0 3.41 0.0 1.0 0.0 0.0 0.0 1.96 1 0.0 1.0 1.0 0.0 0.0 1.80 0.0 0.0 0.0 0.0 0.0 3.75 0.0 1.0 0.0 0.0 0.0 1.70 0.0 0.0 0.0 0.0 0.0 1.61

1 indicates the genes were present in KEGG pathways, 0 indicates the genes were absent in KEGG pathways.

Figures

Page 16/23 Figure 1

Flow chart of this study. KEGG, Gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes; GO, Gene Ontology; PPI, Protein-protein interaction

Page 17/23 Figure 2

Volcano plot of differential expression genes(DEGs) between paired lesional skin and non-lesional skin. Red and green colors indicate upregulated and downregulated genes, respectively. And black color represent genes with no signifcant difference.There are a total of 359 DEGs based on | log2 FC | > 1.5 and adjusted p < 0.05.

Figure 3

Heatmap of DEGs. The orange and blue colors represent upregulated and downregulated genes, respectively.

Page 18/23 Figure 4

GO and KEGG pathway analysis of DEGs in Metascape. (A) Heatmap of Gene Ontology (GO) enriched terms colored by p-values. (B) Network of GO enriched terms colored by p-value, where terms containing more genes tend to have a more signifcant p-value. (C) Heatmap of Kyoto Encyclopedia of Genes and Genomes (KEGG) enriched terms colored by p-values. (D) Network of KEGG enriched terms colored by p- value, where terms containing more genes tend to have a more signifcant p-value

Page 19/23 Figure 5

PPI network and hub genes. (A) The PPI network was constructed in Cytoscape, with upregulated genes revealed in red ellipses, downregulated genes in blue ellipses and cluster 1 genes in yellow ellipses. (B) Cluster 1 network was constructed with upregulated genes revealed in red ellipses and downregulated genes in blue ellipses. (C) The connection function between seven hub genes in STRING database.

Figure 6

Page 20/23 Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) analyses of hub genes. (A) GO enrichment analysis. (B) KEGG enrichment analysis.

Figure 7

Spearman's correlation analysis confrming the signifcant correlation between CXCL10 and the other hub genes. (A) GSE30999: CXCL1(R=0.72), ISG15(R=0.67), STAT1(R=0.77), OASL(R=0.71), IFIT1(R=0.73), IFIT3(R=0.79). (B) GSE34248: CXCL1(R=0.70), ISG15(R=0.75), STAT1(R=0.85), OASL(R=0.78), IFIT1(R=0.67), IFIT3(R=0.78). (C) GSE41662: CXCL1(R=0.62), ISG15(R=0.58), STAT1(R=0.78), OASL(R=0.74), IFIT1(R=0.53), IFIT3(R=0.68). (D) GSE50790: CXCL1(R=0.76), ISG15(R=0.81), STAT1(R=0.90), OASL(R=0.92), IFIT1(R=0.81), IFIT3(R=0.76).

Page 21/23 Figure 8

Expression of seven hub genes in four datasets. The green bar indicates non-lesional skin groups, and the red bar indicates lesional skin groups. Paired T-testing was performed to compare the means of two groups. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. NL, non-lesional skin; LS, lesional skin; ns, no signifcance.

Page 22/23 Figure 9

Results of RT-qPCR. The transcription levels of CXCL10, CXCL1, ISG15, STAT1, OASL, IFIT1 and ITIT3 were signifcantly up-regulated in M5 group. T-testing was performed to compare the means of two groups. **P < 0.01; ***P < 0.001; ****P < 0.0001. M5, HaCat cells were treated with 10 ng/ml M5(IL-1a, IL- 17, IL-22, TNF-α and Oncostatin M); Control, HaCat cells without M5 stimulation.

Page 23/23