Compounds disclosure from A. platensis C1 by multi-retrieving methods
A total of 95 metabolic pathways of A. platensis C1 (Additional file 1) were retrieved from the Spirulina-Proteome Repository (SpirPro) database [14]. The main metabolic pathways of this organism were related to carbohydrate, amino acid, and vitamin metabolism, which consisted of 15, 13, and 11 pathways, respectively. To find the compounds in A. platensis C1, enzymes that produced the metabolic products from substrates were firstly sought. Seventy-nine of these 95 pathways had enzyme-related metabolites, while the other 16 pathways did not. A total of 699 EC numbers (enzyme number in the KEGG database) were retrieved (Additional file 2). The syn00230 pathway associated with purine metabolism had the highest EC numbers of 41 enzymes. Many of these complex pathways contained only a few Spirulina-specific EC numbers. The 415 non-redundant enzymes were collected resulting a total of 968 products produced. After removing duplicated products, the remaining 369 non-redundant products were collected. These products were linked to 369 SIDs in the PubChem database. However, data on the chemical structures were only available for 302 CIDs. Literature review and text mining of published data were further done to find all known and potential chemical compounds in A. platensis C1. A genome-scale metabolic model of A. platensis C1 containing 692 genes, 837 metabolites, and 875 reactions has been reported [17], from which a total of 501 compounds were retrieved with their relevant CIDs. After combining the retrieved data, from SpirPro, Klanchui et al. [17], and literature review (Table 1), and excluding all redundancies, a total of 833 CIDs (Additional file 3) were obtained and used for further analysis.
Table 1. Compounds of A. platensis C1 retrieved from various sources.
|
SpirPro
|
Klanchui et al. study
|
Literature review
|
Metabolic pathways
|
95
|
NA
|
NA
|
Total enzymes with EC
|
699
|
619
|
NA
|
Non-redundant enzymes with EC
|
415
|
477
|
NA
|
Total products from
non-redundant enzymes
|
968
|
NA
|
NA
|
Non-redundant products from non-redundant enzymes
|
369
|
837
|
NA
|
Non-redundant products linked to CID in PubChem
|
302
|
501
|
264
|
Non-redundant products linked to CID in PubChem from all sources
|
833
|
EC; enzyme number in KEGG database, CID; compound identifier in PubChem database, NA; not available
Computational screening of immunosuppressive agents from various databases
Data on both the approved and potential immunosuppressive drugs for SLE treatment and their identity numbers were retrieved from various databases (DrugBank, KEGG, PubChem, TTD, DailyMed, and the ATC/DDD). These drugs and their identify numbers were matched with CID numbers in the PubChem database. Proteins and chemical compounds related to immunologic effects were also searched in PubChem database, using the terms “immunomodulator”, “immunosuppression”, “immunosuppressive”, and “immunosuppressive agents” as recommended by NCBI (Medical Subject Heading; MeSH). The corresponding identity (GI) numbers were also collected. After combined the data and excluding the redundancies, a total of 281 chemical compounds related to immunologic effects were obtained (Additional file 4). These were later matched with the compounds retrieved from A. platensis C1.
Structural similarity matching between compounds retrieved from A. platensis C1 and immunosuppressive agents
To find the algal potential bioactive compounds to be used as SLE drugs or immunosuppressive agents, matching between all potential compounds retrieved from A. platensis C1 and all immunosuppressive agents was done. Details of the 234,073 similarity matchings with their scores are shown in the Additional file 5. The structural similarity matching between the 833 compounds retrieved from this alga and 281 immunosuppressive agents resulting in 1637, 412, 203, 103, 28, and 5 matchings with Tanimoto scores of > 70%, > 80%, > 85%, > 90%, > 95%, and 100%, respectively (Additional file 6). For example (Table 2), there are 5 compounds, including cytidine, desthiobiotin, lomustine, agmatine, and anthranilic acid, matching of 100% with known Immunosuppressive agents. Whereas others, metrotrexate drug had a matching score of 87% with folic acid. Lobenzarit drug had a matching score of 80% with anthranillic acid, 72% with 4- aminobenzoic acid, and 71% with 1-(2-carboxyphenylamino)-1-deoxyribulose 5-phosphate.
Additionally, there were 103 matchings with Tanimoto scores of > 90%. The immunosuppressive agents, namely 8-aminoguanosine, 3868-32-4 (2,8-Diamino-9-[3,4-dihydroxy-5-(hydroxymethyl)-2-oxolanyl]-3H-purin-6-one), fludarabine phosphate, and cladribine had the most frequent matchings of 24 (23.3%), 24 (23.3%), 16 (15.5%), and 11 (10.7%), respectively. The compounds: cytidine, 5'-cytidylic acid, cytidine 5'-diphosphate, desthiobiotin, and cytidine triphosphate had the most frequent matchings of 8 (7.8%), 6 (5.8%), 6 (5.8%), 3 (2.9%), and 3 (2.9%), respectively.
Table 2. Structural similarity scores between the 833 compounds from A. platensis C1 and 281 immunosuppressive agents.
Compound
|
Immunosuppressive agent
|
Similarity score (%)
|
Cytidine
|
Beta-arabinosylcytosine
|
100
|
Cytarabine
|
100
|
Arabinofuranosylcytosine (Iretin)
|
100
|
Desthiobiotin
|
Desthiobiotin
|
100
|
D-dethiobiotin
|
100
|
Lomustine
|
Lomustine
|
100
|
Agmatine
|
Agmatine
|
100
|
Anthranilic acid
|
Anthranilic acid
|
100
|
Lobenzarit
|
80
|
Folic acid
|
Metrotrexate
|
87
|
4-Aminobenzoic acid
|
Lobenzarit
|
72
|
1-(2-carboxyphenylamino)-1-deoxyribulose 5-phosphate
|
Lobenzarit
|
71
|
Bioassay matching between compound bioassay of A. platensis C1 and bioassay associated with SLE
PubChem was searched to find all proteins and bioassays associated with SLE. A total of 660 bioassays and their relevant assay identifier (AID) numbers were also retrieved (Additional file 7). The five compounds and their related bioassays were matched with the 660 bioassays associated SLE. AIDs and targets (proteins or nucleotides) of these matchings, between bioassays associated with SLE and bioassays associated with immunosuppressive compounds, were as follows (Additional file 8); SLE and cytidine: AID504734 protein target - inhibitor of human Toll-like receptor 9; SLE and lomustine: AID625175 protein target - protease, cathepsin G enzyme inhibition, and AID625176 protein target - peptidase, ELA2 (Neutrophil Elastase 2) enzyme inhibition; SLE and anthranilic acid: AID651758 protein target - interleukin 8 stimulation. Unfortunately, all of the SLE matching bioassays above are non-active bioassays. However, cytidine, a nucleoside molecule, has 7 active bioassays with activities against Plasmodium falciparum, antibabesial activity, and Ewings sarcoma. Whereas, desthiobiotin has only 1 active bioassay for in vivo anticancer drug screening for tumor model Friend Virus Leukemia in mice. Though, cytidine and desthiobiotin were respectively identified in the pyrimidine and biotin metabolic pathways in A. platensis, the latter desthiobiotin has been defined as an immunosuppressive agent in MESH term. Lastly, Lomustine, an alkylating antineoplastic agent, has 39 active bioassays in vivo anticancer drug tests. As most of the approved SLE treatment drugs and the potential drugs that can be used are categorized in the immunosuppressant and antineoplastic groups, so that these 3 compounds which have anticancer activities can be proposed the potential to be used for SLE treatment.
Disease-gene-drug-compound network for SLE
The disease-gene-drug network was created using VisANT 4.0. The network consisted of 471 genes/proteins associated with SLE, 219 antineoplastic drugs, and 54 immunosuppressive drugs (Figure 2). From the matchings between 833 compounds retrieved from A. platensis C1 and 281 immunosuppressive agents with Tanimoto scores of > 90%; two immunosuppressive agents, namely fludarabine phosphate [18] and cladribine [19], were included in this SLE-gene-drug network. Only fludarabine phosphate was found to be associated with the SLE-related genes, as shown in the figure.
The disease-gene-drug-compound network was created by mapping 5 bioactive compounds from A. platensis C1 with the greatest potential, associated-immunosuppressive agents, and 471 genes/proteins associated with SLE onto the interaction network, after which the network was visualized using Cytoscape (Figure 3).
Integration of these data in the network offered insights into the potential immunologic effect of the compounds and their target genes/proteins. The target genes for cytarabine were MDM2, TP53, and JAK2, and targets for d-dethiobiotin were PPARG and IL1B. All of these 5 genes associated with SLE. Cytidine and desthiobiotin, from A. platensis C1, had a structural similarity score of 100% with cytarabine and d-dethiobiotin, respectively. Lomustine, agmatine, and anthranilic acid, from A. platensis C1, did not have target genes associated with SLE directly. These three compounds are defined as immunosuppressive agents, and their targets are shown in the network (Figure 3).
Molecular docking and molecular dynamics of the potential bioactive compounds
Cytidine, desthiobiotin, lomustine, agmatine, and anthralinic acid, were identified as the bioactive compounds with potential for SLE treatment from the above analyses. Only the first two bioactive compounds, cytidine and desthiobiotin (Figure 4), were not annotated as the immunosuppressive drugs by themselves. To determine their potential drug effect, molecular docking was used to check the binding affinity between cytarabine drug (ligand), cytidine bioactive compound (ligand) and the following receptors: MDM2, TP53, and JAK2; and the binding affinity between d-dethiobiotin drug (ligand), destiobiotin bioactive compound (ligand) and the following receptors: PPARG and IL1B. Both cytidine and desthiobiotin had good docking energy scores with their corresponding receptor proteins and similar to cytarabine and d-dethibiotin (Table 3). Additionally, MD simulation was performed to confirm the binding affinity of the compound and receptor protein. The RMSD plot for backbone, complex, and ligand of cytidine, desthiobiotin and their corresponding receptor proteins were shown (Figure 5). The RMSD values of all complexes were rather stable and maintained within the fluctuation of only 1 Å. The RMSD values of all complexes showed very small fluctuation and reached the equilibrium within 5 ns. The RMSD of backbone displayed the same fluctuation pattern with the complex.
Table 3. Binding affinity scores between the ligands including drugs and bioactive compounds, with their corresponding receptors by molecular docking
Receptor protein
|
Docking energy (kcal/mol); mean (SD)
|
Drug
|
Bioactive compound
|
|
Cytarabine
|
Cytidine
|
MDM2
|
-4.34 (0.40)
|
-4.59 (0.36)
|
TP53
|
-5.96 (0.21)
|
-6.12 (0.26)
|
JAK2
|
-6.01 (0.21)
|
-6.27 (0.32)
|
|
D-dethiobiotin
|
Desthiobiotin
|
PPARG
|
-5.72 (0.36)
|
-5.4 (0.45)
|
IL1B
|
-4.93 (0.20)
|
-5.04 (0.24)
|