Per- and poly-fluoroalkyl substances (PFAS), firstly defined by Buck et al. in 2011, are considered a highly fluorinated aliphatic family of compounds which are formed by one or more carbon atoms where all the hydrogen substituents have been replaced by atoms of fluorine. Structurally speaking and in general terms, the nomenclature can be represented as CnF2n+1 being n ≥ 1, with the molecular structure containing at least one CF3– group (Alderete et al., 2019; Scheringer et al., 2014; OECD/UNEP, 2018; http://www.pops.int/. 2004). Despite the current industrial value of PFAS, scientific evidence has raised concerns regarding their potential toxicity to wildlife and ecosystems, as well as to their potential role as inductive agents for physiological disorders on human metabolism, including diabetes, hypertension, and even various types of cancer (Bischel et al., 2018; Poothong et al., 2017; Brendel et al., 2018). A major characteristic of PFAS toxicity is their persistence in biological systems and the environment for extended periods in their unchanged forms. A recent investigation revealed that some PFAS not only accumulate but also are bio-transformed extensively into a wide range of structurally diverse chemicals capable of persisting for long time in the environment. (Tamara et al., 2021; Andersen et al., 2006) Furthermore, the continuous release of PFAS contaminants in the aqueous ecosystems can cause a secondary contamination of natural water leading to the spread of contamination in water levels of parts per trillion (Bischel et al., 2018; Wang et al., 2015; Krafft et al., 2015; Bowman et al., 2015; Butenhoff et al., 2012). In fact, according to the Stockholm Convention, PFAS have been considered as one of the most controversial among the category ‘persistent organic pollutant’ (POP) for the anthropologically manufactured worldwide environmental contaminants (http://www.pops.int/, 2004). As such, the need for sophisticated bio-remediation strategies in order to mitigate their environmental impact is urgent (Scheringer et al., 2014).
Indeed, proper knowledge regarding PFAS-bioaccumulation parameters alongside with appropriate environmental risk assessment models that addressed PFAS ecotoxicity (in particular when main food-chains are in question) are still amiss (Vestergren et al., 2009). In particular, this knowledge is absent for a wide variety of these main-food chains organisms, such as bacteria, algae, crustaceans, ciliates, fish, yeasts and nematodes. Nevertheless, research has evidenced a direct association between the bioaccumulation potential of PFAS and the length of their perfluoroalkyl chain, as PFAS analogs of longer chains, with at least six carbons, presented a higher bioconcentration potential (Bischel et al., 2018; Poothong et al., 2017; Brendel et al., 2018; Tamara et al., 2021; Andersen et al., 2006). These chemical attributes are related with increasing concentration of environmental pollution the PFAS cause through their exposure routes, which could involve different levels of trophic chains, including humans, and diverse environmental compartments, such as water, soil or air (Wang et al., 2015; Krafft et al., 2015; Butenhoff et al., 2012).
An interesting bio-target for a PFAS ecotoxicological predictive/assessment model in aqueous ecosystems is the zebrafish (Danio rerio), since the zebrafish model can be efficiently used to predict/study mechanisms of toxicity for both the macroscopic and cellular level of this organism regarding PFAS induced damages (Wang et al., 2015; Krafft et al., 2015; Bowman et al., 2015; Butenhoff et al., 2012; Vestergren et al., 2009; Hagenaars et al., 2013; O'Brien et al., 2004; Mashayekhi et al., 2015). Interestingly, among the most PFAS – induced macroscopic damages in the zebrafish models, reports include the endocrine disruption (thyroids), altered swimming, developmental and reproductive toxicity in sub-lethal doses. Furthermore, subcellular and biochemical level damages include cytotoxicity, perturbations in lipid metabolism, oxidative stress leading to genotoxicity, and mitochondrial dysfunction, such as ATP bioenergetics perturbations, and channel mitotoxicity.
A hypothesis that we propose in this paper, departing from the analysis of the published works (Wang et al., 2015; Krafft et al., 2015; Bowman et al., 2015; Butenhoff et al., 2012; Vestergren, et al., 2009; Hagenaars et al., 2013; O'Brien et al., 2004; Mashayekhi et al., 2015), is that PFAS could be bioaccumulated at mitochondrial level (i.e., at the mitochondrial matrix) after interacting with the zebrafish voltage-dependent anion channel 2 (zfVDAC2), or selectively induce significant perturbations in the native structure and physiological function of zfVDAC2, or selectively induce significant perturbations in the native structure and physiological function of zfVDAC2 by triggering mitotoxic events directly associated to the opening of the mitochondria permeability transition pore-based zfVDAC2 channel dysfunction (zfVDAC2-MPTP). The latter is directly involved in ROS induction, cardiolipin peroxidation, ADP/ATP transport perturbations, cytochrome c release, decreased ATP levels. From a phylogenetic point of view, and regardless of the animal species under study, it is considered that VDAC channel (zfVDAC2) is highly conserved and is involved in the mitochondrial bioenergetics and homeostasis regulation, as well as in the control of intracellular metabolism (Vestergren et al., 2009; Choi et al., 2017). With this accepted evidence as cornerstone, we explore a computational model for the mitochondrial channel dysfunction as an eco-toxicodynamic mechanism in zebrafish.
In addition, this departing concept is sustained by the zfVDAC2 being the most abundant and densely localized channel protein in the outer mitochondrial membrane of all cells in the zebrafish (Dreier et al., 2019), thus representing the main route of communication between the mitochondrial intermembrane space and the cytosol. It should be noted that the zfVDAC2 channel has an N-terminal α-helix segment located horizontally inside the pore, which is aligned nearly parallel to the membrane plane, causing a partial narrowing of the zfVDAC2 pore (Shimizu et al., 2015, Choi et al., 2017). In particular, the N-terminal α-helix segment can adopt different conformations within the voltage-gating channel depending on the internal/or external stimuli (Shimizu et al., 2015, Choi et al., 2017). Further, the importance of this channel covers many mitochondrial essential aspects, such as ideal equilibrium dynamics of the intracellular [Ca2+], or the ADP plus Pi/ATP influx/efflux transport which favor key biochemical functions as glycolysis and ATP-bioenergetics regulation (Shimizu et al., 2015, Choi et al., 2017, Zafeer et al., 2018, Liao et al., 2020, Girdhar et al., 2020).
Notwithstanding, the fact is that there is a data gap regarding the biological hazardousness factor per PFAS group and versus specific bio-targets (Cousins et al., 2020, Cousins, et al., 2022), something that is even more noticeable when considering mitochondria channel toxicity versus zfVDAC2 toxicity, as the evidence scenario, either experimental or in silico, is scarce if not inexistent. As such, within the in silico model proposed in this work regarding PFAs versus zfVDAC2 toxicity, validation versus experimental evidence can only be achieved when there is availability of such results. Therefore, the focus will be in the intrinsic in silico validation methodologies, that have been proven effective in other ecotoxicological approaches that were also validated versus experimental evidence.
The development of an in silico ecotoxicological predictive technique in PFAS environmental risk assessment, which is currently rather limited to our best knowledge, not only offers great advantages in time and money savings but is also aligned with the need of implementing effective predictive animal-free testing methodologies (Schredelseker et al., 2014; Cheng et al., 2019; Sima et al., 2021). Furthermore, in silico methods are strongly recommended in the very early stages of the predictive toxicological (ecotoxicological) investigation, only to be followed by in vitro and in vivo experimental validation after the computational and theoretical assessment.
As such, a synergy between classic computational approaches as structure-based virtual screening (SB-VS) and predictive Quantitative Structure-Activity Relationships (QSAR) modeling could be efficiently implemented to precisely assess the potential ecotoxicological effects of PFAS in aquatic organisms (Vestergren et al., 2009; Choi et al., 2017; Ankley et al., 2021; Schredelseker et al., 2014; Cheng et al., 2019). These in silico methodologies have demonstrated the ability to simultaneously evaluate the molecular mechanisms of interaction of a large number of compounds, including PFAS, while maintaining sensitivity, specificity, and accuracy to realistically reproduce the key interactions of traditional or emerging pollutants (Vestergren et al., 2009; Choi et al., 2017; Ankley et al., 2021; Schredelseker et al., 2014).
In fact, while molecular docking software allows for the rapid calculation of docking scores, the integration of the 3D-QSAR approach offers valuable additional insights and complements the docking analysis in our study, namely in two major vectors: the identification of relevant specific 2D molecular descriptors, and a wider range of prediction.
Regarding the first vector, while the identification of specific 2D molecular descriptors, i.e., structural features of molecules in a two-dimensional representation, allows to gain insight into the molecular properties that may contribute to the observed activity or properties of the compounds, the 3D-docked poses, i.e., the predicted spatial arrangements of the PFAS ligand molecules within the binding site of the target protein obtained through molecular docking simulations, represent potential binding configurations and help assess the binding affinity between the ligand and protein. Though the docking score obtained from the 3D-docked poses serves as an indicator of the binding affinity between the PFAS compounds and the zfVDAC2 target channel, it does not provide detailed information about the specific structural physico-chemical based molecular descriptors responsible for the PFAS binding activity on the zfVDAC2 channel. Therefore, the recourse to the QSAR methodology aims to elucidate the structure-mitotarget interactions activity relationship of the PFAS compounds on zfVDAC by identifying relevant specific 2D molecular descriptors derived from the 3D-docked poses.
Regarding the second vector, the proposed QSAR model will allow us to quantitatively predict the ecotoxicological properties of a larger set of PFAS compounds beyond those specifically docked in our study. This broader perspective enhances our understanding of the ecotoxic potential of various PFAS analogs, guiding future experimental efforts and providing insights for environmental risk assessment in the context of an ecotoxicological target as zfVDAC2. In addition, employing the QSAR model enables the analysis of the contributions of individual PFAS molecular descriptors and their respective weights, providing a more detailed understanding of the underlying molecular mechanisms governing PFAS ecotoxicity.
The present study proposes an effective methodology for assessing the molecular interactions between per- and poly-fluoroalkyl compounds and the zfVDAC2 channel protein. This is achieved through a synergistic approach combining structure-based virtual screening and 2D/3D-QSAR models, introducing a novel hybrid in silico risk assessment technique. This approach represents a qualitative advancement in ecotoxicological predictive techniques within the field of Computational Toxicology. Notwithstanding, the framework of the proposed approach is overall supported by well-recognized structural and thermodynamic techniques to delve into the potential mitochondrial toxicity response induced by PFAS compounds in the zebrafish. In addition, considering the importance of understanding mitochondrial dysfunctions in generating hypotheses across various biological levels, from molecular mechanisms to population health, in the context of Environmental Pollutants (Dreier et al., 2019), this work not only presents novelty but also holds significant relevance.