2021 45(1): 65-77
DOI:10.32604/biocell.2021.014154
www.techscience.com/journal/biocell
| BIOCELL 2021 45(1): 65-77 DOI:10.32604/biocell.2021.014154 | www.techscience.com/journal/biocell |
In silico assessment of human health risks caused by cyanotoxins from cyanobacteria
1Department of Biomedical Science and Environmental Biology, College of Life Science, Kaohsiung Medical University, Kaohsiung, 807, Taiwan
2Univ. Lille, Laboratoire LASIRE, UMR CNRS 8516, Equipe Physico-Chimie de l’Environnement, Lille, F-59000, France
3Institute of Marine Biology, National Taiwan Ocean University, Keelung, 202, Taiwan
4Center of Excellence for the Oceans, National Taiwan Ocean University, Keelung, 202, Taiwan
5Center of Excellence for Ocean Engineering, National Taiwan Ocean University, Keelung, 202, Taiwan
6Research Center for Environmental Medicine, KMU-Kaohsiung Medical University, Kaohsiung, 807, Taiwan
7Department of Marine Biotechnology and Resources, National Sun Yat-sen University, Kaohsiung, 804, Taiwan
*Address correspondence to: Hans-Uwe Dahms, hudahms11@gmail.com
Received: 03 September 2020; Accepted: 12 October 2020
Abstract: Harmful algal blooms (HABs) that are formed by cyanobacteria have become a serious issue worldwide in recent years. Cyanobacteria can release a type of secondary metabolites called cyanotoxins into aquatic systems which may indirectly or directly provide health risks to the environment and humans. Cyanotoxins provide some of the most powerful natural poisons including potent neurotoxins, hepatotoxins, cytotoxins, and endotoxins that may result in environmental health risks, and long-term morbidity and mortality to animals and humans. In this research, we used the chemcomputational tool Molinspiration for molecular property predictions, Pred-hERG 4.2 web software for cardiac toxicity prediction, and Pred-Skin 2.0 web software for predicting skin sensitization. We are predicting some toxicological aspects of cyanobacteria here using chemcomputational tools with the hypothesis that cyanotoxins are providing a risk to human health. We are using the tool Pred-hERG 4.2 to predict hERG channel blocking potential and the Pred-skin tool to predict skin sensitization due to cyanotoxins. The potential of anatoxin, ambigol, the microcystin group, and lyngbyatoxin A, lyngbyatoxin B, nodularin-R, and saxitoxin were predicted to cause skin sensitization in the final results (consensus model). Anatoxin-a and lyngbyatoxin were predicted to allow GI absorption and blood–brain barrier penetration. Among the 20 predicted cyanotoxins only aeruginosin 103-A, ambigol A, and ambigol were predicted by Pred-hERG 4.2 according to the applicability domain results as potential cardiotoxins with weak or moderate potency. Lyngbyatoxin shows activity through the GPCR ligand and protease, kinase, and enzyme inhibitor.
Keywords: Cyanotoxins; Predictive model; Molinspiration; Bioactivity score; hERG blocker; Carcinogenicity
Cyanobacteria occur in aquatic environments such as freshwaters (rivers, lakes, ponds, etc.), brackish waters (Hamilton et al., 2014), and the oceans (Schaefer et al., 2020). They are primary producers and play important roles in aquatic ecosystems, converting nitrogen into organic forms that can be used as macronutrients by eukaryotes, and oxygenic photosynthesis (Gademann and Portmann, 2008; Harke et al., 2016; Wood et al., 2007). However, climate change and eutrophication can result in harmful algal blooms (HABs) that are caused by excessive growth of cyanobacteria, and the release of high concentrations of toxic secondary metabolites called cyanotoxins becomes a serious threat. This holds for other organisms as well as for the safety of drinking water, aquatic food sources (Agasild et al., 2019), and public health (Aráoz et al., 2010).
Cyanotoxins provide toxic compounds that may affect the environment, animals, and humans by exposure through, e.g., seafood and aquaculture consumption, water contact during recreation, and drinking water (Meriluoto et al., 2017; Bukaveckas et al., 2018). Cyanotoxins can cause serious public health issues after long-term exposure through contaminated drinking water and recreational contact in fresh- and saltwater systems. Seafood consumption is another concern (Stone and Bress, 2007). Exposure to cyanobacteria can cause health effects in humans. These may include vomiting, flu-like symptoms, rashes, nausea, fever, gastroenteritis, blistered mouth, skin, eye, and ear irritation, visual disturbances, abdominal pain, and systemic effects such as hepatic failure, neurological damage, and death (Brown et al., 2018; Codd et al., 2005; Kubickova et al., 2019). Skin irritation can result from contact with toxic cyanobacteria (Pilotto et al., 2004). Cyanobacteria can also cause allergic reactions with symptoms like hives, conjunctivitis, and asthma (Farrer et al., 2015).
Cyanobacterial toxigenic compounds provide major waterborne health risks globally (Schaefer et al., 2020). Several toxins appear to be confined to specific cyanobacteria, but some cyanobacteria are known to produce a variety of toxic compounds (Haque et al., 2017). Cyanotoxins contain a variety of chemical compounds that differ by their chemical structure. Toxicological endpoints can be separated by their effects on liver toxins, neurotoxins, dermatotoxins, and endotoxins (Brown et al., 2018; Mankiewicz et al., 2003; Haque et al., 2017) (Tab. 1). Different approaches try to classify cyanotoxins, such as their respective biological sources, toxicity, molecular mass, and structural characteristics (Johnson et al., 2011). Reproductive and developmental toxicity, as well as carcinogenicity, are as yet not conclusively connected to cyanotoxins. This holds for marine systems but also for freshwater ones, where cyanotoxins represent an emerging health concern (IJCH Report, 2017).
Table 1: Health risks provided by toxins from cyanobacteria
Cyanotoxins can cause alimentary poisoning in humans and animals with different syndromes such as: Diarrheic shellfish poisoning (DSP), amnesic shellfish poisoning (ASP), paralytic shellfish poisoning (PSP), neurotoxic shellfish poisoning (NSP), and azaspiracid shellfish poisoning (ASP) (Bigalke and Rummel, 2005). Several toxic compounds were found in cyanobacteria worldwide from oceanic environments and reservoirs and freshwater lakes worldwide (Huang and Zimba, 2019). Most threatening are neurotoxins affecting the nervous system (e.g., tetrodotoxin and saxitoxin), providing considerable potential as toxins even for military use (Pitschmann and Hon, 2016). For example, saxitoxin (STX) binds to the voltage-gated Na+ channel, subsequently blocking neuronal transmissions (Arnich and Thebault, 2018). It is the most studied and the first known compound to cause paralytic shellfish poisoning (PSP). The aerosol of the unstable saxitoxin degrades per minute at a rate of 17% (Jansson and Åstot, 2015). Saxitoxin is able to overcome the dermal barrier (similar to brevetoxin, anatoxin, and tetrodotoxin). However, it does not reach the efficacy of organophosphate as a chemical warfare agent (CWA) with another example of cyanobacterial neurotoxin, β-N-methylamino-l-alanine (BMAA), which is a cyanobacterial neurotoxin (Kubo et al., 2008). The toxins of several cyanobacterial genera also showed antineoblastic potential in human cell lines and promising results with respect to human adenocarcinomas (Zanchett and Oliveira-Filho, 2013).
We are applying chemoinformatics here as a search for chemical information that transfers chemical data to simulations. In silico approaches in toxicological studies make use of informational techniques that allow the predictions of toxic effects of cyanotoxins. We are using chemcomputational tools here following the hypothesis that cyanotoxins are providing risks to human health.
We accessed 21 cyanotoxins through the database of PubChem (https://pubchem.ncbi.nlm.nih.gov/) and a literature report. Isomeric SMILES notations of cyanotoxins collected from PubChem are presented in Tab. 2, and structures sketched by software ChemDraw 18.1 (https://www.perkinelmer.com/au/category/chemdraw) are presented in Fig. 1. Linear kernels like SMILES were used for our toxicity predictions since their similarity functions are computationally more efficient (Frenzel et al., 2017).
We were using Swiss ADME online software (http://www.swissadme.ch) in our toxicity discovery to predict parameters related to “absorption, distribution, metabolism, and excretion” (ADME) such as medicinal chemistry, druglike nature, friendliness of one or multiple small molecules as well as to compute parameters related to their physicochemistry. Swiss ADME enables assessment of ADME parameters of drug candidates and small molecules and provides information that allows early risk assessment in the drug development process. Notably, Swiss ADME provides a platform to assess the drug-likeness of oral bioavailability through Lipinski’s rule of five (Jayaseelan et al., 2012; Lipinski, 2000). Swiss ADME is a more recent and comprehensive site run by the Swiss Institute of Bioinformatics (SIB), which provides bioinformatics services and resources for scientists worldwide. SIB has over 800 scientists and 65 bioinformatic teams from major Swiss research and higher education institutes (Daina et al., 2017).
One form of the Long QT syndrome (LQTS) with a lack of repolarization of the heart after a heartbeat was disturbed, is related to a defect of the hERG protein that affects the K+ channel functioning during cardiac electric excitability. Identification of a hydrophobic area, which represents the putative interface and tight binding region with the interface of the K+ channel, became possible by the screening for mutagenesis of the domain surface. Once this hydrophobic domain of the channel is present, the rate of deactivation is slowed down (Cabral et al., 1998). The first three-dimensional model of a eukaryotic domain called Per-Arnt-Sim (PAS) is structurally similar to a yellow protein functioning as a bacterial light sensor.
In order to confirm or reject the sensitization effect of cyanobacterial compounds evaluated here, we used Pred-Skin 3.0 web online software (http://labmol.com.br/predskin/), available as an online software program (Braga et al., 2017).
The bioactivity potential of cyanotoxins was predicted by calculating the activity score for the ion channel modulator, the GPCR ligand, the nuclear receptor ligand, and the inhibitors of enzymes, proteases, and kinases (Jabeen and Ranganathan, 2019). All the above parameters were tested by applying the drug-likeness score of Molinspiration software (www.molinspiration.com).
Figure 1: Chemical structure of cyanotoxins.
Table 2: Isomeric SMILES of cyanobacterial toxins
Anatoxin-a and Lyngbyatoxin were predicted to have GI absorption and blood–brain barrier (BBB) penetration (Tab. 3).
Table 3: ADME prediction results of 20 cyanotoxins
Among the 20 predicted cyanotoxins only aeruginosin 103-A, ambigol A, and ambigol B have cardiotoxic potential and weak or moderate potency as predicted by Pred-hERG 4.2 according to the applicability domain results (Tab. 4).
Anatoxin, ambigol, the microcystin group, lyngbyatoxin A, lyngbyatoxin B, nodularin-R, and saxitoxin were predicted to have the potential to cause skin sensitization in the final results (consensus model) (Tab. 5).
The calculated values of the isolated compound endpoints and the drug-likeness scores are summarized in Tab. 6. Lyngbyatoxin, as an enzyme inhibitor, shows inhibition activity against proteases and kinases through the GPCR ligand and the inhibitors of enzyme, kinase, and protease.
Anatoxin-a and Lyngbyatoxin were predicted to have GI absorption and BBB penetration (Tab. 3). The BBB maintains homeostasis of the Central Nervous System (CNS), which is provided by the permeability BBB (Gao et al., 2017).
The inability of most compounds to cross the BBB is a major limitation to effectively treat diseases in the central nervous system (Alexander et al., 2019). The BBB is a highly evolved microvascular system comprised of brain endothelial cells (ECs) lining the vascular lumen, pericytes in the basal lamina, and associating astrocytic end-feet, microglia, and neurons. This cellular architecture forms functional neurovascular units that regulate molecular trafficking between the blood and the brain (Zhang et al., 2019).
The extent of binding can greatly influence The ADME (adsorption, distribution, metabolism, and excretion) profile that can greatly be affected by toxin binding. The evaluation of biological substance effects very much depends on gastrointestinal absorption (GI absorption). In the evaluation of the biological effects of substances, this is a key criterion (Diukendjieva et al., 2019). Cost-effective and reliable and cost-effective bioavailability studies early in the drug discovery process can lead to an improvement in the success rate for compounds entering clinical trials (Fabini and Danielson, 2017).
Pred-hERG 4.2 server (http://labmol.com.br/predherg) predicts the probability maps where compounds block the K+ ion channel coded by a human Ether-à-go-go-Related Gene (hERG) (Tab. 4). The results show that not all tested marine cyanobacterial toxins are hERG blockers in a multiclass prediction. All tested cyanobacterial compounds are hERG blockers as predicted by the Pred-hERG binary prediction, except anatoxin-A, BMAA, and hypoxanthine.
Anatoxin, ambigol, and the microcystin group, lyngbyatoxin A, lyngbyatoxin B, nodularin-R, and saxitoxin were predicted to cause skin sensitization (Tab. 5). In case a susceptible individual is exposed to a contact allergen, upregulation and clonal expansion of allergen-responsive T-cells occurs (Gilmour et al., 2019).
The computed values of different parameters of the isolated compounds are summarized in Tab. 6. We found that Lyngbyatoxin shows activity through the protease inhibitor, the G-protein-coupled receptor (GPCR ligand), the GPCR ligand, and the kinase inhibitor. In clinical medicine, GPCR ligands became the most successful molecular drug targets. Antagonists, as well as agonists of the GPCR class, were applied in the treatment of every major organ system, including the respiratory, metabolic, urogenital, and cardiovascular systems. Considering their widespread expression and important regulatory and mechanistic characteristics, GPCRs have manifold functions in living beings (Insel et al., 2007; Campbell and Smrcka, 2018). Our study predicted anatoxin-a and lyngbyatoxin to affect GI absorption and BBB penetration. Factors ensuring overall drug absorption include compound solubility and permeability as well as compound dissolution and gastrointestinal conditions, drug- and formulation- related parameters, drug product dissolution, as well as active transport, passive diffusion, and metabolism of toxicants of drugs (Freerks et al., 2019).
Table 4: hERG-predictions of 20 cyanotoxins from Pred-hERG 4.2
Table 5: Skin sensitization predictions of cyanotoxins
Table 6: Bioactivity potential results as predicted by Molinspiration
Chemcomputation adds to the instrumentation applied in ecotoxicology, such as the prediction of the physiological effects of cyanotoxins. In silico tools add objectivity combines several approaches in repeatable and intelligent ways. Economic demands and fast technological progress are in favor of chemcomputational tools. Cheminformatics allows for higher throughput and constant optimization. Cheminformatic approaches have a higher reproducibility, are less time consuming, and are less expensive. Computational approaches can also prioritize chemicals in order to reduce the amount of costly in vivo and in vitro toxicological screening and provide early alerts for unexplored, newly discovered, or newly developed substances. Through their replacement, they reduce the use of experimental efforts.
A lack of transparency and quality of the training set of experimental data provide some limitations. For example, carcinogenicity prediction can only be applied with non-genotoxic compounds. Neurotoxicity, hepatotoxicity, and developmental toxicity cannot be accurately predicted by in silico methods. These are complex phenomena with multiple endpoints, unlike issues with well-understood mechanisms like mutagenicity, sensitization, and aquatic toxicity – as exemplified in this study by cyanotoxins.
Acknowledgement: We are thankful to Mr. Charli Deepak and Cheng-Han Liu, Mrs. Sravya Kosuru, and Revathi Gurunandam for their assistance in this computational study.
Availability of Data and Materials: The datasets analyzed during the current study are available from the corresponding author on reasonable request.
Funding Statement: This research was funded by the Ministry of Science and Technology of Taiwan, R.O.C. to HUD, grant number MOST 107-2621-M-037-001, MOST108-2621-M-037-001, and MOST 109-2621-M-037-001 provided for Tan, Han-Shih. This work was financially supported by the Research Center for Environmental Medicine, Kaohsiung Medical University, Kaohsiung, Taiwan from The Featured Areas Research Center Program within the framework of the Higher Education Sprout Project by the Ministry of Education (MOE) in Taiwan. An NSYSU/KMU collaboration is acknowledged (108-PO25).
Conflicts of Interest: The authors declare that they have no conflicts of interest to report regarding the present study.
 | This work is licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. |