Identification and validation of targets of swertiamarin on idiopathic pulmonary fibrosis through bioinformatics and molecular docking-based approach

IPF is a chronic progressive disorder and shares many common clinical, pathological, and immune characteristics with pulmonary fibrosis [1], one of the post-sequelae of COVID-19 [2]. IPF can significantly lower patients’ life quality and life expectancy [3] because of the irreversible decline in lung functions. Some environmental factors, genetic susceptibilities [4], and oxidative stress [5] were reported to increase the risk of getting IPF. Oxidative stress is believed to play an important role in the initiation of inflammation and the damage to DNA [6], which first occurs in the development of IPF. Various signaling pathways, e.g., hedgehog [7], TGF-β/Smad [8], and Wnt/β-catenin (Wingless and Int-1/beta catenin) [9], etc., are involved in the initiation and development of the IPF. Marker proteins of these signaling pathways, for instance, TGFBRI (transforming-beta type I receptor) [10] and CTGF [11], etc., were reported to be the targets of anti-IPF drugs. Our previous study of screening novel anti-IPF drugs with machine learning from traditional Chinese medicines found that swertiamarin, a secoiridoid glycoside with high anti-oxidation and anti-inflammatory effects [12], has a strong effect on arresting the development of IPF [13], and identified that α-SMA and E-cadherin, two marker proteins of TGF-β/Smad signaling pathway, were significantly downregulated and upregulated, respectively, by swertiamarin. However, the targets still need to be investigated. For this new anti-IPF lead, it is tough and time-and-cost-consuming work to identify its targets with wet-lab experiments. Fortunately, bioinformatics provides us with powerful tools to exploit genomic, transcriptomic, and proteomic data [14] to gain insights into the anti-IPF mechanisms and to identify targets for further validation. Molecular docking (MD) is another effective in silico approach to accurately predict the stability of a ligand-receptor complex and understand the activity of the ligand. MD showed remarkable advantages [15] over traditional experiment paradigms in avoiding large amounts of intensive experiments. In this study, bioinformatics was first used to analyze the microarray datasets from the GEO database to obtain hub genes. MD was then applied to decide whether these hub genes were the targets of swertiamarin by predicting the interactions between swertiamarin and the corresponding proteins of these genes. Finally, the screened targets were experimentally validated.

Target identification is of central importance to the understanding of the anti-IPF mechanism of swertiamarin. However, solving this problem using web-lab experiments usually means expensive and slow processes, whereas computation-aided approaches provide efficient complements. Here we used the bioinformatics and MD-based approach to screen the anti-IPF targets of swertiamarin with the public database. In this study, GSE10667, GSE110147, and GSE71351 datasets were used to analyze the DEGs. It is important to clarify here that the GSE10667 dataset contains samples with interstitial pneumonia and it made the results more reliable but also more difficult to obtain the overlapping DEGs with high log|FC| values because of considering the gene expressions in the early stage of IPF. Even with lower log|FC| values, the hub genes were still successfully screened out. The KEGG analysis suggested that the selected DEGs were related to tissue and cancer development. This result was consistent with the fact that IPF shares several pathogenetic similarities, e.g., DNA methylation [27], with lung cancer [28] and that patients with IPF are at high risk of getting lung cancer [29].

Three hub genes (LOX, COL5A2, and CTGF) were screened out through the above analysis. The LOX is a cuproenzyme and is also known as protein-lysine 6-oxidase encoded by the human LOX gene [30]. LOX catalyzes the conversion of lysine into highly reactive aldehydes that form the cross-linking collagen and elastin in ECM proteins [31], contributing to the ECM’s stiffness. The stiffness of the ECM would increase fibroblast proliferation and contraction [32]. Aberrant expression and activity of LOX were associated with IPF [33] and led to the development of the IPF microenvironment. Therefore, LOX is a key participant in ECM remodeling [34]. In this study, the PPI analysis of LOX showed that the LOX has strong interactions with TGF-β1, TGFBR1, TGFBR2 (TGF-β receptor 2), and SMAD2/3 (suppressor of mothers against decapentaplegic family member 2 and 3) (Supplement 5), which are the main marker proteins of TGF-β/Smad signaling pathway. Therefore, LOX is the potential target of IPF [35] and inhibition of its activity would alleviate the IPF [33], which was identified by the MD model. The interactions between LOX and swertiamarin indicated a stable ligand-receptor complex (Fig. 4A), which is consistent with the western blot results showing the downregulation of the LOX expression.

CTGF, also known as CCN2 (cellular communication network factor 2), is a matricellular protein of the CCN family of ECM-associated heparin-binding proteins [36]. CTGF was reported to be associated with wound healing (the key initial process of IPF) and all fibrotic pathology [37]. In IPF tissues, CTGF expression is upregulated by TGF-βs [38], SMAD2 [39], and other physiological and pathological factors. The upregulation of CTGF expression would further exacerbate the ECM accumulation and aggravate the development of IPF [37]. The CTGF-involved PPI analysis (Supplement 6) showed that CTGF has interactions with TGF-β1, TGF-β3, and TGFBR2 (one of the key targets of exogenous factors that promote the expression of TGF-βs). The MD model also suggested that the swertiamarin-CTGF complex is highly stable and CTGF should be the target of swertiamarin on IPF. Western blot results showed that swertiamarin slightly but not significantly downregulated the expression of CTGF. Therefore, we concluded that swertiamarin can only inhibit the activity of CTGF rather than downregulate the expression of CTGF.

COL5A2 is a protein encoded by the COL5A2 gene [40] and is responsible for the formation of other collagen fibrils in tissues of the body. COL5A2 was reported to be involved in the development of pathological scarring [41]. COL5A2 is modulated by TGF-βs [42] and is highly related to human systemic sclerosis [43] and IPF [44]. In this study, the COL5A2 protein has a strong interaction with ITGB1 (integrin beta-1, a cell surface receptor in human and function as collagen receptors and modulate the migration across basement membranes in human [45]) and ADAMTS14 (ADAM metallopeptidase with thrombospondin type 1 motif 14, an enzyme cleaves the amino-propeptide of fibrillar collagens) (Supplement 7). The DEG analysis showed that the COL5A2 gene was significantly upregulated in the IPF patients. The MD model suggested that swertiamarin could modulate the activity of COL5A2 and the western blot showed that the expression of COL5A2 was downregulated by swertiamarin.

Our bioinformatics analysis identified 103 DEGs that co-exist in GSE10667, GSE110147, and GSE71351 datasets. The GO and KEGG functional analyses showed that these DEGs were mainly related to the cell process that related to cancer and tissue development. The MD models and experimental results proved that these three genes/proteins are the targets of swertiamarin. The experiment data didn’t show significant downregulation of CTGF. It is very important to further experimentally investigate the effects of swertiamarin on the activities of LOX, CTGF, and COL5A2, and especially to validate its effect on the expression of CTGF.