New Insight into Drugs to Alleviate Atopic March via Network Pharmacology-Based Analysis

In the present study, a subject of atopic dermatitis (AD) is exposed progressively to allergic rhinitis (AR) and asthma (AS), which is defined as atopic march (AM). However, both the targets and compounds against AM are still largely unknown. Hence, we investigated the overlapping targets related directly to the occurrence and development of AD, AR, and AS through public databases (DisGeNET, and OMIM). The final overlapping targets were considered as key targets of AM, which were visualized by a Venn diagram. The protein–protein interaction (PPI) network was constructed using R package software. We retrieved the association between targets and ligands via scientific journals, and the ligands were filtered by physicochemical properties. Lastly, we performed a molecular docking test (MDT) to identify the significant ligand on each target. A total of 229 overlapping targets were considered as AM causal elements, and 210 out of them were interconnected with each other. We adopted 65 targets representing the top 30% highest in degree centrality among 210 targets. Then, we obtained 20 targets representing the top 30% greatest in betweenness centrality among 65 targets. The network analysis unveiled key targets against AM, and the MDT confirmed the affinity between significant compounds and targets. In this study, we described the significance of the eight uppermost targets (CCL2, CTLA4, CXCL8, ICAM1, IL10, IL17A, IL1B, and IL2) and eight ligands (Bindarit, CTLA-4 inhibitor, Danirixin, A-205804, AX-24 HCl, Y-320, T-5224, and Apilimod) against AM, providing a scientific basis for further experiments.


Introduction
Atopic lesions, including atopic dermatitis (AD), allergic rhinitis (AR), and allergic asthma (AA), have seen elevated incidence rates for some decades, impacting around 30% of the world population [1]. The atopic march (AM) occurs in a time-based sequence: from AD in infancy to AR and AA in childhood, which is developed from the skin to the respiratory tract [2]. Firstly, AD is a chronic autoimmune skin disorder caused by increases in serum Immunoglobulin E (IgE) levels and hypersensitivity to a diverse array of allergens, including food and bacteria [3,4]. AD is an initial step in the development of AM, and IgE is a key biomarker of allergic inflammation [5].
The first line drug of AD is topical corticosteroids such as Pimecrolimus and Tacrolimus, the mechanism of which is to block the T cell's activation to dampen immune responses [6,7].
Secondly, AR is an inflammatory disease of the nasal mucous membrane mediated by IgE [8]. The first line drug of AR is H1 antihistamine drugs such as cetirizine, levocetirizine, desloratadine, loratadine, and fexofenadine [9,10]. Thirdly, AA is a final step of AM, and the recommended drugs to alleviate AM are inhaled corticosteroids which suppress airway inflammation [11,12]. Collectively, a key symptom of AM is the inflammatory response, and the optimal strategy to relieve AM is to elucidate overlapping targets from AD, AR, and AA.
The analysis of protein-protein interaction (PPI) networks for drug repositioning is a significant approach, providing clues about causal relationships between targets that carry out certain functions [13]. Therefore, this research has focused on the topological properties of PPI networks to uncover the relatively important nodes among the overlapping targets from AD, AR, and AA.

Hypothesis
The targets on AD, AR, and AA were retrieved through DisGeNET, OMIM, and the literature, suggesting that the overlapping targets from the three diseases were considered as AM-related targets. We hypothesize that targets with the 30% greatest betweenness centrality (BC) from targets with the top 30% highest degree centrality (DC) values [14] in the overlapping targets are core targets to alleviate AM, thereby identifying the most stably binding ligands on them via molecular docking test (MDT).

Methods
The targets associated with AD, AR, and AA were acquired through DisGeNET (https://www.disgenet.org/) (accessed on 2 December 2021), and OMIM (https://www. omim.org/) (accessed on 3 December 2021), which are therapeutic relevant components. The overlapping targets from the three diseases (AD, AR, and AA) were considered as core targets of AM. The core targets were analyzed using the STRING database (https://string-db.org/) (accessed on 4 December 2021). We performed the topological analysis of AM on a PPI network. The workflow is as below.
Step 1: Identifying of targets (AD, AR, and AA) from DisGeNET and OMIM.
Step 2: Identifying of overlapping targets (AD, AR, and AA) via a Venn diagram.
Step 3: The overlapping targets were screened on the top 30% by degree centrality (DC), and the PPI network was constructed on RPackage software. Step 4: After constructing the PPI network of the top 30% DC, the core targets were identified, and the PPI network of the top 30% by betweenness centrality (BC) in the top 30% DC were constructed on RPackage software.
Step 5: Suggesting that targets with the highest DC in the top 30% BC targets were considered as the promising targets against AM.
Step 7: The first screening of ligands based on TPSA <140Å 2 or Lipinski's rule.
Step 8: The second screening of Step 7 ligands according to the docking score (<−6.0 kcal/mol) or the lowest binding energy (the highest negative value) of each target.
Step 9: The identified ligands were converted to .sdf format from PubChem into .pdb format utilizing Pymol, and the ligands were converted into .pdbqt format via AutoDock. Then, the PDB ID of targets were obtained via RCSB PDB (https://www.rcsb.org/) (accessed on 7 December 2021). The AutoDockTools-1.5.6 software was utilized to evaluate the affinity of the promising targets and ligands. The ligands were docked with targets using autodock4 by arranging 4 energy ranges and 8 exhaustiveness as default to identify 10 different poses of ligands [15].  [16]. The adopted ligands on key targets were regarded as key compounds against AM. The methodology workflow is represented in Figure 1.
Curr. Issues Mol. Biol. 2022, 2, FOR PEER REVIEW 3 were regarded as key compounds against AM. The methodology workflow is represented in Figure 1.
Based on the top 30% of the subnetwork by betweenness centrality (BC), 14 out of 20 targets had the same degree centrality (19) (Table 1), which were determined as crucial targets in the PPI network. Then, 8 out of 14 targets were reported with ligands, and 6 other targets had no reported ligands. A total of 8 targets were associated with the 34 ligands (Table 2). A total of 34 ligands related directly to the uppermost 8 targets were retrieved via chemical supplies websites and literature ( Figure 4A). The 34 ligands were filtered into 29 ligands with TPSA < 140 Å 2 or Lipinski's rule ( Figure 4B). MDT was performed on the 29 filtered ligands to select the most stable binding ligands for each target, as a consequence revealing that 8 ligands were critical compounds ( Figure 4C).  Based on the top 30% of the subnetwork by betweenness centrality (BC), 14 out of 20 targets had the same degree centrality (19) (Table 1), which were determined as crucial  Based on the top 30% of the subnetwork by betweenness centrality (BC), 14 out of 20 targets had the same degree centrality (19) (Table 1), which were determined as crucial  The MDT of eight ligands demonstrated that (1) Figure 5H). The MDT confirmed the stable binding between eight core ligands as well as eight targets (Table 3).   Figure 5H). The MDT confirmed the stable binding between eight core ligands as well as eight targets (Table 3).

Discussion
As a matter of fact, targets with a high betweenness centrality (BC) (the measurement of the shortest route between other targets) are the most significant element to be associated with vital multiple therapeutic utilization and then are a noteworthy value to be repositioned [13]. There is an observational result that the eight core targets (CCL2, CTLA4, CXCL8, ICAM1, IL10, IL17A, IL1B, and IL2) overlapped among AD, AR, and AA were the therapeutic targets with the most potential against AM. In addition, the compounds-targets relationship suggested that the therapeutic effect of AM was directly associated with eight ligands, including three benzenoids (Danirixin, Y-320, T-5224), two heterocyclic compounds (Bindarit, CTLA-4 inhibitor), one organosulfur (A-205804), one organic nitrogen compound (Apilimod), and one phenylpropanoid and polyketide (AX-024 HCl). It was reported that benzenoids isolated from Antrodia cinnamomea have a potent anti-inflammatory effects on AD, AR, and AA [17,18]. This implies that the benzenoids classification might be a good ligand to alleviate AM. Investigations into the treatment of heterocyclic compounds have revealed that the compounds are useful for the alleviation of inflammatory and allergic disorders including AD, AR, and AA [19]. Study indicated that organosulfur compounds are inflammatory mediators to dampen interleukin (IL)1B and IL17, particularly, AD, AR, and AA [20][21][22][23]. It was reported that nitrogen-containing compounds such as Apilimod are potent selective inhibitors for AD, AR, and AA [24][25][26]. A report indicated that plantderived phenylpropanoids (PPPs) are anti-inflammatory agents to modulate underlying inflammatory reactions in vitro or in vivo [27].
CCL2 (C-C motif chemokine ligand 2) is the initial chemokine stimulated by itching and pain symptoms on keratinocytes, fibroblasts, and endothelial cells' infiltered monocytes and lymphocytes [28]. The overexpression of CCL2 is a trigger to respond to allergies and asthma [29]. It could be thus speculated that the inhibition of CCL2 might be a good strategy for the treatment of AM. Study shows that miR-155 as a CTLA-4 inhibitor has inhibitory effects on AD that could potentially disturb the development of AM [30]. Thus, a CTLA-T inhibitor might be a promising target against AM. There is an observational report suggesting that AD, AR, and AA increased the expression level of CXCL8 in responding to innate immunity [31][32][33]. A report suggested that the intercellular adhesion molecule-1 (ICAM1) is upregulated in AD, AR, and AA, indicating that the ICAM1 is associated directly with AM [34][35][36]. IL10 is multiple-functional cytokine during allergic symptoms and a significant element in the progression of severe allergic diseases, including AD, AR, and AA [37][38][39]. IL17A plays a vital role in the occurrence and development of allergic diseases, particularly, AD, AR, and AA [40,41]. IL1B is a proinflammatory agent, mediated by a rise in the expression of ICAM-1, and thereby is related directly to the inflammatory responses such as AD, AR, and AA [42][43][44]. IL2 increased in severe AD, which secretes pruritus in almost all patients [45]. In addition, after treatment of the intranasal corticosteroid, the IL-2 were considerably diminished during 24 h, contained by significant enhancement in the upper airway function [43].
Collectively, this study indicates the potential ligands and critical targets to alleviate AM via orchestrating targets of AD, AR, and AA, providing theoretical clues for further experimental evaluation.

Conclusions
In summary, eight promising compounds (Bindarit, CTLA-4 inhibitor, Danirixin, A-205804, AX-24 HCl, Y-320, T-5224, and Apilimod) and eight targets (CCL2, CTLA4, CXCL8, ICAM1, IL10, IL17A, IL1B, and IL2) have been uncovered to alleviate AM with synergistic effects. Besides, the topological analysis on AM was orchestrated with a network-based concept. We conclude that multi-target agents are a potential strategy to treat comorbid AM, and network-based analysis especially provides critical clues to select targets on AM. Given the limitations of network pharmacology perspectives, the potential therapeutic ligands of the treatment of AM are predicted by topological analysis on PPI networks, and its efficacy in vivo was not validated, which needs to be further evaluated through therapeutic and clinical tests.