Towards Precision Ecotoxicology: Leveraging Evolutionary Conservation of Pharmaceutical and Personal Care Product Targets to Understand Adverse Outcomes Across Species and Life Stages

Abstract

Environmental science aims to protect biodiversity and ecosystem services, and our future ability to do that relies on a developing a precision ecotoxicology approach where we leverage the genetics and informatics of species to better understand and manage the risks of global pollution. A little over a decade ago, a workshop focusing on the risks of pharmaceuticals and personal care products (PPCPs) in the environment identified a priority research question, “What can be learned about the evolutionary conservation of PPCP targets across species and life stages in the context of potential adverse outcomes and effects?” Here we review the activities in this area over the past decade, consider prospects of more recent developments, and identify future research needs to develop next generation approaches for PPCPs and other global chemicals and waste challenges.

BACKGROUND AND NEED

Protection of biodiversity and ecosystem services are common goals during ecological risk assessment (ERA) and management, yet global biodiversity losses have been identified as occurring faster now than any other time in history (www.ipbes.net). Among the diverse factors driving such precipitous biodiversity decline, chemical impacts were recently highlighted during the United Nations Biodiversity Conference in Montreal (COP 15; www.unep.org/news-and-stories/story/cop15-ends-landmark-biodiversity-agreement), where several targets were selected as goals for 2030, including to “Reduce by half both excess nutrients and the overall risk posed by pesticides and highly hazardous chemicals” (www.cbd.int/article/cop15-cbd-press-release-final-19dec2022). Adoption of this landmark Global Biodiversity Framework agreement occurred in parallel with other efforts to establish an intergovernmental policy activity on chemicals and waste (Wang et al. 2021). For example, the United Nations Environment Assembly recently announced their intention to establish chemicals and waste as the third pillar of their approach to environmental protection (UNEP, 2022), supporting the existing international panels on biodiversity and ecosystem services (IPBES) and climate change (IPCC). Achieving such aspirational goals will be challenging, because for the majority of the over 350,000 chemicals and chemical mixtures registered for commercial use around the world (Wang et al. 2020), ecotoxicology information is limited, and mechanistic ecotoxicology data are decidedly lacking.

In contrast to the thousands of commercial chemicals without empirical toxicology information, pharmaceuticals, pesticides, and some ingredients in personal care products (PCPs) receive extensive study prior to introduction to commerce. Leveraging mammalian safety information from experience with pharmaceuticals has been advancing the science to identify problematic industrial chemicals in the environment and to identify and design less hazardous alternatives, while improving ecological risk assessment approaches for pharmaceuticals and personal care products (PPCP; Brooks 2014, 2018). Building from early recognition of opportunities to consider human safety information during environmental assessments of pharmaceuticals (Lange and Dietrich 2002, Seiler 2002), Huggett et al. (2003) proposed a fish plasma model, which anticipated evolutionary conservation of biological targets for PPCPs between humans and fish, to support prioritization and screening activities. Ankley et al. (2007) integrated this concept among species, along with other considerations for pharmaceuticals in the environment, within a risk-based framework, despite the paucity of genomic information for different species and limited mechanistic toxicology information for human pharmaceuticals in wildlife. For example, Owen and colleagues (2007) examined comparative physiology, pharmacology and toxicology relationships in beta-blockers, for which perhaps more mechanistic data was available for fish than any other group of pharmaceuticals at the time. Using available genome information, Gunnarsson et al. (2008) then explored evolutionary conservation of pharmaceutical targets and predicted orthologs among sixteen species for over 1300 drug targets. Brooks and colleagues (2009) built from this effort to consider reading mammalian information across to environmental data within tier-based effects analysis, along with Kostich and Lazorchak (2008), Berninger and Brooks (2010) and Fick et al (2010), who examined use of human therapeutic plasma values (C_max) and other pharmacology and toxicology information to potentially identify pharmaceutical hazards in the environment. Further, Winter et al. (2010) critically considered prospects for integrating diverse drug discovery and development information within ecological risk assessment activities.

In April 2011, a workshop held at Niagara-on-the-Lake, Ontario, Canada, examined research needed to understand risks of PPCPs in environment. This unique and transparent exercise generally followed the methods of previous scoping workshops to identify research questions for conservation of global biodiversity (Sutherland et al. 2009). Following input from hundreds of scientists and engineers actively publishing on the subject, the Niagara-on-the-Lake synthesis workshop resulted in identification of 20 priority research questions, including (Boxall et al. 2012): “What can be learned about the evolutionary conservation of PPCP targets across species and life stages in the context of potential adverse outcomes and effects?” Here we reflect on research developments for this question over the past decade while providing current perspectives. We then conclude by recommending future research needs, including several research questions for the next decade.

WHAT IS OUR CURRENT UNDERSTANDING?

Adverse outcome pathways and taxonomic domains of applicability

Over the past decade, several advances in cross-species extrapolation for PPCPs and other contaminants have occurred in parallel with, and leveraged development of, adverse outcome pathways (AOP; Ankley et al. 2010). Within the context of the AOP framework, which aims to link molecular initiation events (MIEs) across levels of biological organization to adverse outcomes of relevance to risk assessment, structural and functional conservation of biological pathways are typically considered in understanding the taxonomic domain of applicability (tDOA). The published literature is now being mined to capture existing knowledge from toxicity studies to define key events and key event relationships along the AOP. The model organisms used in those studies provide the basis for defining the tDOA, with AOP developers then considering the biological plausibility of additional species/taxa that are likely to follow a similar trajectory in their biology. Since the initial guidance documents for AOP development were generated, there has been limited evidence collected for the tDOA for the majority of AOPs entered in the AOP-Wiki (aopwiki.org/), which is the central repository for AOPs. This is likely because toxicity and pathway data generated across species are typically relatively sparse and commonly only generated for a few model organisms. However, with advances in informatics, particularly bioinformatics, the possibilities to enhance understanding of conservation across species through extrapolation using comparative genomics, proteomics, and transcriptomics are being realized (Jensen et al., 2023).

Efforts are underway to harness the computational power of individual tools and databases that allow for cross-species comparison and extrapolation of biological pathway knowledge to better inform the tDOA. For most pharmaceuticals, relatively extensive knowledge is available describing how the drug enters the body and interacts with particular biomolecules (i.e., molecular initiating events) in model organisms and humans. Therefore, having an understanding of protein target conservation across species can provide a line of evidence that the pharmaceutical could interact with that protein in another species. As noted above, this concept of exploring gene/protein conservation for cross species extrapolation was detailed by Gunnarsson et al. (2008) and then explored by Kostich and Lazorchak (2008) for the prioritization of pharmaceuticals in the environment. Further, these concepts of gene/protein conservation, phylogeny, and ortholog detection for pharmaceuticals (LaLone et al. 2013; Rivetti et al., 2023) evolved to the development of tools for the translation of such comparative data to inform chemical safety decisions across species using bioinformatics. Examples of publicly available tools include the US EPA Sequence Alignment to Predict Across Species Susceptibility (SeqAPASS; seqapass.epa.gov/seqapass/) tool, which evaluates protein sequence and structural similarity across hundreds to thousands of species to understand pathway conservation and/or predict chemical susceptibility, and EcoDrug (www.ecodrug.org/), which contains information for >600 eukaryotes and allows users to identify human drug targets for >1000 pharmaceuticals and associated orthologue predictions. Additionally, there are efforts to make greater use of advances in other omics technologies, including high throughput transcriptomics, derivation of transcriptomic points of departure, and development of cross-species quantitative PCR arrays (e.g., EcoToxChips; Basu et al., 2019).

Such comparative approaches aim to understand the impacts of chemical stressors on diverse species representing different taxonomic groups by gathering knowledge regarding perturbation of biological pathways. Empirical data provides an understanding of chemical interactions that lead to perturbation of the biology and with mechanistic studies that incorporate high throughput toxicology and bioinformatics aiding in understanding which pathways are most likely to be impacted. This knowledge can inform the identification of tDOA for specific MIEs (e.g., ecdysone receptor involved in molting in invertebrate species) or early key events in an AOP. Beyond these approaches, there are more sophisticated computational molecular models that have been applied and advanced through drug discovery, specifically moving toward protein structural-based evaluations of chemical protein interactions across species (McRobb et al., 2014, Cheng et al., 2021; LaLone et al., 2023). The field of bioinformatics continues to evolve and is likely to play a prominent role in understanding and predicting the effects of pharmaceuticals to the diversity of species, particularly as global efforts are moving toward a non-animal testing regulatory agenda for chemical safety evaluations (LaLone et al., 2021; Langan et al., 2023).

Evolutionary conservation of targets and new approach methodologies

Charles Darwin recognized the common Origin of Species, establishing evolutionary biology and our fundamental ability to consider species relatedness as we integrate biological read-across in ecotoxicology. Though conservation of genes and common physiological systems provide a necessary piece of the puzzle in understanding toxicity across species/taxa, biological complexity does not make it easy; for example, 70% of adversity-related genes in vertebrates may also be found across invertebrates (Colbourne et al., 2022). There remain significant challenges in extracting lines of evidence toward species relatedness that can be applied to understand the potential for chemical sensitivities among species and associated with species interactions. For example, in addition to evolutionary considerations are the needs to advance comparative pharmacokinetics, to define age and gender differences within and among species, and to deconvolute influences across environmental gradients and interactions with other species. In the last decade, we have innovated technologies from other fields, such as drug discovery and development, to improve environmental toxicology and ecotoxicology.

As introduced above, AOPs provide a framework for organizing existing biological knowledge that allows for comparative physiology across species/taxa (LaLone et al., 2017). Integration of quantitative information, when available, relative to AOP perturbation enables us to predict adverse outcomes from specific internal concentrations of chemicals in exposed wildlife (Margiotta-Casaluci et al., 2014, 2016; Conolly et al., 2017). Herein, pharmaceuticals as model compounds and pharmacology knowledge have been leveraged to better approach, understand, and predict effects of chemicals among non-mammalian model species with a known mode of action (Ankley et al., 2023). Most of these efforts have focused on fish. However, less comparative information is available for the majority of non-mammalian vertebrates, invertebrates, plants, and algal species.

Searching the genomes of wildlife species for pharmacological targets (Verbruggen et al., 2018) through simple online applications has become common place. Such comparative knowledge can then be used to inform the tDOA of AOPs (Jensen et al., 2023). We can also use >900 available databases that provide in silico chemical and drug safety information (Pawar et al., 2019), combined with in vitro studies such as those in the ToxCast database (comptox.epa.gov/dashboard/) and include published historical in vivo experimental data to gather broad knowledge of chemical pathway information across species using combined new approach methodologies (NAMs) for ecotoxicology (Brockmeier et al., 2017, Rivetti et al., 2020; Ankley et al, 2023). The term NAMs has been used as an umbrella term to capture all non-animal-based approaches for assessing chemical safety (e.g., in silico, in vitro, high-throughput). If we integrate this information transparently, predictive models for extrapolating toxicity knowledge to the diversity of species can be developed, not just for single compounds, but also for mixtures (Kidd et al., 2023). For example, Marmon and colleagues (2021) used an integrated NAM approach and created a pharmacology informed multiscale network model for all 25 non-steroidal anti-inflammatory drugs that examines measured concentrations in surface waters to anticipate toxicity. It is important to note that this approach was grounded in mechanistic evidence, including secondary targets of the drugs and potential interactions.

Such general approaches call for development of a new subdiscipline, precision ecotoxicology, which will inherently leverage developments in comparative biology, precision medicine and precision environmental health sciences, along with advances in ecology to understand perturbations at higher levels of biological organization (Brooks 2022). Though we are continuing to leverage lessons learned from the study of pharmaceuticals in the environment, we must build on these approaches to cover more modes of action, including personal PCPs, per- and polyfluoroalkyl substances (PFAS) and non-PPCPs. As we advance NAMs, we must also address the uncertainties; remain transparent in the limitation of approaches, and further define off-target effects that lead to variation in chemical susceptibilities of individuals, which will be critical to make precise predictions for populations.

Development of quantitative adverse outcome pathways

To move towards precision ecotoxicology, it will be necessary to develop a quantitative understanding of comparative sensitivities to PPCPs and other contaminants among species that is informed by evidence at multiple levels of biological organization. As mentioned above, AOPs have been used to organize this information, with quantitative AOPs (qAOPs) describing response-response relationships leading to an adverse outcome. To date, multiple qAOPs have been developed, most notably a qAOP linking aromatase inhibition to decreased fecundity in fish (Conolly et al., 2017). The tDOA of this qAOP has been tested, with experimental evidence indicating that response-response relationships are conserved across three fish species (Doering et al., 2019). Further, this qAOP has been used to predict short-term in vivo responses in female fish from mammalian in vitro aromatase inhibition data (Villeneuve 2021). This application requires both structural and functional conservation of the molecular target across species, and a similar understanding has been applied in other case studies following the qAOP framework. For example, mammalian in vitro data for estrogen receptor agonism has been correlated with male fish vitellogenin production in vivo (Dreier et al., 2017), and aryl hydrocarbon receptor activation has been used to predict early life stage mortality in birds and fish (Doering et al., 2018). In contrast, when a molecular target is not conserved across species, there can be a divergence in sensitivity. For example, a lack of evidence for molecular target conservation in vertebrates was associated with higher acute toxicity values (i.e., reduced hazard) for an insecticidal target (LaLone et al., 2013a), and when a molecular target sequence is not conserved, such as the androgen receptor in invertebrates, it would be predicted that effects are limited from molecules that interact with that receptor (LaLone et al., 2013b). Therefore, a priori knowledge of chemical-biomolecule interaction defined the MIE(s) and associated AOP(s) offers insight towards characterizing species sensitivity and further predicting chemical susceptibility.

In the absence of formal qAOPs, other approaches have been used to support quantitative assessments of species sensitivity. At the molecular level, docking simulations combined with in vitro transactivation data have been used to understand the basis for differences in estrogen receptor subtype responses in various fish species (Tohyama et al., 2015). Such an approach, representing diverse targets across multiple species, may also be used to understand potential effects. For example, a library of chemicals has been screened in a multiplexed, multispecies assays to characterize activation of diverse nuclear receptors (Houck et al. 2021). This characterization at the molecular level offers a line of evidence to support species sensitivity assessments, but ultimately this information will need to be anchored to effects at the individual/population level for risk assessment (Margiotta-Casaluci et al., 2023).

Linkages to higher levels of biological organization

While the application of AOPs can serve as the basis to inform chemical effects on individuals, ecotoxicology requires those effects to be placed into the context of the environment (Chapman 2002). In doing so, the integration of exposure and effects is critical. Mechanistic ecological models (e.g., population models) integrate relevant exposure and effects components into various metrics of population vulnerability, sustainability, and potential risk. These models contain great flexibility to advance the ecotoxicology of PPCPs and other chemical contaminants both for broad applications as well as with higher levels of realism and precision when needed (Raimondo et al. 2018). While AOPs can assess the inherent susceptibility of a species to chemical exposure based on likelihood and magnitude of molecular responses, they must be nested within population structure, recruitment potential (e.g., intrinsic rate of increase), and the spatial and temporal co-occurrence of individual organisms and PPCPs in the environment. Survival, growth, and reproduction are highly variable across and sometimes even within species, and a molecular response shared by multiple species or unique within specific life stages or gender may result in significantly different manifestations at the population level (Forbes et al. 2008). Similarly, timing of critical life history events (e.g., development, rearing young) with influxes of exposure may result in considerably greater impact to some species, while populations of other species may not temporally or spatially overlap with compounds that exceed non-hazardous thresholds, resulting in lower realized impacts (Bennett and Etterson 2007). Mechanistic ecological effects models remain the best available tools to incorporate toxicological effects measured or predicted on individuals with life history, chemical fate, and co-occurrence of contaminants and individuals across a landscape (Forbes et al., 2017; Galic et al., 2018) to discern ecotoxicological impacts of PPCPs and other chemical stressors (Sumpter et al., 2023) within and among species.

In addition to considering the timing of critical life history events (e.g., development, rearing young), many more ecological or phenotypical attributes of organisms (e.g., size, feeding habits, life cycle duration) can influence the vulnerability of organisms to chemicals (Van den Brink et al., 2011). Trait-based approaches aim to understand such attributes and have proven useful by enhancing causal diagnosis and prediction (Rueda-Cediel et al., 2023). These trait-based approaches can be particularly useful at multiple levels of biological organization and can be used to describe three key aspects determining organismal vulnerability to chemical compounds: i) exposure, ii) intrinsic sensitivity, and iii) recovery potential. The traits (i.e., migration behavior, reproductive strategy) of a species can help determine the potential likelihood and magnitude of both external and internal exposure of an organism by describing where, when, and at what life stage exposure may occur in a contaminated system, where pH, intrinsic chemical properties and metabolism can largely determine internal exposure of ionizable PPCPs and other chemicals (Carter et al. 2023). Traits such as habitat choice, food choice, and life cycle duration can be useful for describing the likelihood and magnitude of external and internal exposure.

When further considering trait-based approaches, it is useful to partition toxicokinetics (TK; uptake, biotransformation, and elimination) from toxicodynamics (TD; damage, internal recovery, and toxicity thresholds, Rubach et al., 2011) to define AOPs, as AOPs are intended to be chemical agnostic, and to understand sensitivities among species, both are required. Interspecific differences in these TKTD processes can disentangle differences in inherent sensitivity among species, and it is here where the combination of both traits and evolutionary conservation of specific targets and pathways play a complementary role. Studies describing species differences in TK parameters (e.g., Buchwalter et al., 2008, Rubach et al., 2012) found that traits like mode of respiration and body size are useful predictors of uptake rates, while elimination rates tend to have a very strong phylogenetic signal. Since TD parameters describe processes related to toxicity thresholds inside the organism, the evolutionary conservation of specific chemical receptors is likely to be a strong predictor of differences in the TD part of species sensitivity. Relatedness-based approaches (using metrics of taxonomic or phylogenetic relatedness between species as a proxy for differences in sensitivity) have the potential to describe aspects of both TK and TD, because relatedness acts as a proxy for the likelihood of sharing niche space and necessary traits, but also potentially for sharing similar toxicological targets and AOPs. Therefore, it is likely that optimally performing mechanistic models can be found by combining sensitivity-related, morphological traits with the conservation of the molecular targets of the chemical or MOA under study and using relatedness-based metrics to represent sensitivity related processes that are still unknown (Van den Berg et al., 2021).

It is also important to consider how the vulnerability of a species can be heavily influenced by population- or community-level processes that determine the potential impacts to and recovery of a species. Because a primary protection goal for many ERAs includes the population level of biological organization (Suter 2006; though the individual level represents the protection goal for threatened and endangered species), these processes should not be overlooked. Key traits that influence population growth rates and therefore the recovery potential of populations include life span, growth and development, differential survival across age and/or size cohorts, generation time, and the number of offspring per reproductive event. Key traits that influence the recolonization potential of an organism include dispersal capacity and dispersal mode. Herein, mechanistic knowledge guided by AOP and informed by trait-based approaches could be integrated to predict the critical apical endpoints at the individual level quantitatively, which could be further used to estimate the population dynamics across landscapes defined explicitly or implicitly (Etterson and Ankley, 2021). Thus, it appears possible to integrate taxonomic domains of qAOPs, population dynamics and food web models to then be used to simulate and define effects associated with a given exposure scenario on species assemblages. Furthermore, recent advances of genomic biomonitoring tools (Deiner et al., 2017), as an alternative approach for field-based surveillance, have provided a rapid and animal-free approach to examine populations and communities of aquatic species assemblages under real world exposure scenarios for multiple stressors, including PPCPs.

WHAT ARE THE PRIORITIES FOR FUTURE RESEARCH?

In this section, we consider future research opportunities and identify several timely research questions to advance precision ecotoxicology for PPCPs and other contaminants (Box 1). In developing these questions, we aimed to follow criteria employed during the original Boxall et al. (2012) and the Global Horizon Scanning project, through which a related question was identified: How can we extrapolate effects data across species using evolutionary conservation of biological pathways? (Fairbrother et al., 2019). Specifically, these key questions were intended to address important gaps in knowledge, have factual answers that do not depend on value judgments, be answerable through a realistic research design, cover a spatial and temporal scale that could realistically be addressed by a research team (e.g., 10M € over 5 years), not be answerable by “it all depends” or “yes” or “no,” and if a question was related to impact and interventions, it should have contained a subject, an intervention, and a measurable outcome (Boxall et al. 2012).

Box 1. Key research questions to advance precision ecotoxicology for pharmaceuticals, personal care products, other anthropogenic contaminants, and natural toxins.

• How can the quantity, taxonomic/species coverage, quality, and annotation of sequences improve to expand information available for comparative studies using bioinformatics and other omics approaches?

• What are the molecular targets of specific chemical classes, are these targets conserved across species used in and reflected by endpoints employed during regulatory ecotoxicology testing?

• How can we better take advantage of combined approaches considering molecular target sequence similarity and protein structural knowledge to inform functional predictions for species extrapolation in chemical safety assessments?

• Which molecular targets have a quantitative relationship between evolutionary conservation and species sensitivity?

• What vulnerable species co-occur with contaminants in specific locations at concentrations above adverse effect thresholds that are anticipated based on cross-species extrapolation approaches?

• How can advances in eco-exposomics and precision exposure science accelerate development of precision ecotoxicology and its applications to improve the practice of environmental science and technology?

• What are the criteria that would drive the adoption of alternative methods, model systems and bioinformatics approaches for regulatory decision-making regarding cross species extrapolation?

Advancing –omics to understand molecular initiation events across species

Although the grouping of PPCPs originated from Daughton and Ternes (1998), these are diverse groups of chemicals that have been considered together during discussion of anthropogenic impacts of chemicals in the environment. However, the existing knowledge of specific chemical mechanisms of action and impacts on the environment differ substantially within and among groups of PPCPs. For example, PPCPs, which can also include PFAS (Whitehead et al., 2021), are not regulated similar to pharmaceuticals in their premarket approval, and require toxicity testing prior to production. Further, sales data and environmental introduction amounts are not equivalent among these diverse groups of chemicals used by humans and in veterinary applications. Additional research is needed to identify molecular targets for the numerous chemical categories encompassed by PPCPs. This is necessary to have the most basic biological understanding to leverage any approaches that have been described above for evaluating conservation of biological pathways across species/taxa.

To take advantage of omics and bioinformatics approaches for extrapolation of pathway knowledge, the breadth and quality of sequence information must continue to expand and incorporate broader species coverage. Advances have been made in more rapidly and cost effectively generating sequences from whole genomes, and hundreds of genomes have been annotated. In fact, there are many international efforts (e.g., www.darwintreeoflife.org) to generate sequence data for a broad range of species including non-mammalian species, invertebrates, and plants and algae. Expanding research to generate more sequence data and advancing curation and methods for annotation are essential steps for generating better predictions of chemical susceptibility across species.

Leveraging advances among disciplines to develop quantitative adverse outcome pathways

The greatest potential impact on advancing an area of science is through interaction among scientists from diverse fields of study and areas of expertise. An example of this includes capitalizing on the biomedical and drug development/discovery fields, where the evaluation of chemical-biomolecule interactions is paramount to understand which particular structures, out of hundreds, can bind to a given protein target to identify candidates to move forward in drug development. In principle, these same automated virtual pipelines could be applied in support of chemical safety assessments considering interactions across species (Margiotta-Casaluci et al., 2023; LaLone et al., 2023). In fact, advances in precision medicine and precision environmental health science provide robust foundations from which precision ecotoxicology can advance (Brooks 2022).

The AOP-Wiki currently contains numerous AOPs, though relevance of AOPs across species needs to advance. Using a mechanistic approach, we can progress towards describing, understanding, and then predicting toxicity for a large number of chemicals. For some classes of chemicals, we have advanced our understanding about how they induce toxicity and can identify an AOP relevant for these stressors, yet numerous data gaps remain and we know very little about many classes of chemical stressors. We also have a challenge in the enormous number of chemicals currently used, many of which we have limited to no mechanistic knowledge (Wang et al., 2020). Clearly all chemicals cannot be tested and assessed individually. However, we could use the tools described above and others at our disposal to group chemicals by structure, mechanism, and pathway to prioritize which taxa are more susceptible to specific MIEs and thus chemicals initiating AOPs through evolutionary conserved targets (Margiotta-Casaluci et al., 2023). Capitalizing on mechanistic approaches to inform chemical grouping has been suggested as a gold standard for grouping chemicals for human hazard assessment, and when mechanistic information is missing then a common AOP is considered a pragmatic approach (EFSA, 2021). Could we extend this approach from human data or other model organism information to the diversity of species in the environment? If so, a key question in addition to those in Textbox 1 is, how many AOPs or modes and mechanisms of action are the most critical to inform biodiversity and ecosystem services protection goals, and how many will need to be developed with changes in the chemical universe and subsequent environmental exposure scenarios?

Additional research at the molecular level will be necessary to support quantitative assessments across species. These efforts should emphasize the MIE, leveraging both computational and empirical data to ascertain relevant interactions. Fortunately, there have been recent and significant advances in structural biology, with deep learning methods yielding protein structure predictions from various tools such as AlphaFold (Jumper et al., 2021), RoseTTAFold (Baek et al., 2021), and RGN2 (Chowdhury et al. 2021). These structures can be used to support molecular docking experiments for a variety of protein targets expressed across diverse taxa, therefore offering a quantitative approach to accurately characterize MIEs beyond molecular target sequence conservation (LaLone et al., 2023). Results from molecular docking experiments can be further validated by generating empirical data, such as those from receptor binding assays, as well as other chemical proteomics approaches to identify molecular targets (Ross and Peng, 2020). Molecular responses downstream of the MIE can also be used to infer species sensitivity. For example, transcriptomics has been used to identify upstream drug (Pabon et al., 2018) and chemical (Harrill et al., 2021) targets in human cells, and this approach could be applied in additional species to determine whether specific responses are functionally conserved across taxa.

While molecular data offer an important line of evidence to characterize species sensitivity, it remains crucial to establish whether individual or population effects are expected from exposure to PPCPs. In a movement towards precision ecotoxicology, it may first be necessary to generalize effects within PPCP categories (e.g., defined by MIEs, or other characteristics) to subsequently identify future basic research and translational risk assessment needs. For example, large databases of ecotoxicology data are available (Connors et al., 2019; Olker et al., 2022), and this information could be mined to generate unique probabilistic chemical toxicity distributions (Dobbins et al., 2008, Williams et al., 2011; Connors et al., 2014) for specific species with conserved MIEs leading to adverse outcomes and categories of PPCPs. Subsequently, hazardous concentration for five percent of the species (HC5) estimates for susceptible species that are identified with common AOPs can be combined to form a unique species sensitivity distribution, therefore identifying potentially susceptible taxa, as well as defining potential ecological thresholds of toxicological concern (Belanger et al., 2021) by species and for specific MIEs. These data-driven approaches will be necessary to harness our current knowledge and offer quantitative estimates of hazard, thereby enhancing our general understanding and highlighting priorities.

Developing linkages across biological organization from the laboratory to the field

Improving the precision of ecotoxicology and translation of the scientific advances within risk-based evaluations involves improving our understanding of where, when, and how much PPCPs and other contaminants are being consumed and entering the environment. It further involves understanding what species are most likely to be exposed and/or most vulnerable at the population level, as noted by another key question identified by Boxall et al., (2012) and further explored elsewhere in this special issue (Wilkinson et al., accepted with revision). The primary exposure route of PPCPs introduction to the environment has historically been attributed to wastewater treatment effluent and biosolids (Xia et al. 2005). More recently, the importance of landfill leachate and recreational swimming/snorkeling have emerged as additional point sources for PPCPs (Chung et al., 2018; Mitchelmore et al. 2019, Yu et al. 2020); however, the contribution of these sources to overall PPCP concentrations in environment matrices remains unknown. For example, PPCP concentrations vary widely within and among lentic, lotic, and terrestrial systems (Meyers et al. 2019; Wilkinson et al., 2022), resulting in disproportionate exposure of species inhabiting these different systems (Bouzas-Monroy et al., 2022).

It is important to note that vulnerable populations such as threatened and endangered species, amphibians, and many coral reef species may experience more dramatic impacts at lower PPCP concentrations than more resilient species exposed to higher chemical concentrations. For example, emerging concerns over potential impacts of ultraviolet filters used in sunscreen on coral reef species (Downs et al. 2014) may represent a very small proportion of PPCPs in the environment, but if impacts of lower incident compounds act on and affect keystone species, then they could have more extensive ecological impacts than other compounds present in higher, more wide-spread concentrations (Bean et al., 2023). Future research is needed to concurrently characterize exposure to PPCPs and leverage cross-species extrapolation approaches to identify the most at-risk species/taxa based on overlapping distributions of compounds and diverse species/taxa. Evaluation and characterization of the ecological receptors most at risk from PPCPs, other anthropogenic contaminants and natural toxins must include both inherent susceptibilities detected at molecular levels and species life history, distribution, and population status (e.g., threatened, endangered, species of conservation concern). Herein, advances in eco-exposomics and precision exposure science in environmental matrices will facilitate acceleration of basic science in precision ecotoxicology and its subsequent translation to practice, particularly for susceptible species and biodiversity hotspots.

To advance such integrative efforts within precision ecotoxicology, we need to identify the most important processes in the three vulnerability aspects identified earlier (external and internal exposure, intrinsic sensitivity, and recovery potential), and determine which measures can be used to quantify these processes, particularly for sensitive species. This will consist of parallel research lines, combining measures from diverse scientific knowledge (e.g., traits, evolutionary conservation, relatedness). We need to advance the science to integrate these different vulnerability aspects, ideally resulting in common vulnerability measures at the individual and population level of biological organization, which represent environmental protection goals (Suter 2006). Some promising frameworks are being developed in this space, among which the integration of energy budgets with AOPs show promise (e.g., Murphy et al., 2018, Goodchild et al., 2019). Specifically, Dynamic Energy Budget (DEB) models incorporate bioenergetics to characterize sublethal effects of chemicals and provide quantitative measures of energy costs related to chemical exposure (Kooijman et al., 2009). Directly linking AOPs to DEB parameters allows the connection of sub-organismal endpoints to the level of the individual or even the population (Martin et al. 2013). However, these frameworks need to be developed further, particularly among MIEs resulting from PPCP exposures to ensure they consider the most important sensitivity-related processes, which could be informed by evolutionary conservation of targets within and among species.

Especially since the dawn of the Anthropocene, human activities have impacted public health and the environment. Many consider recent changes in climate, land use, and nutrient loading to be top factors leading to biodiversity decline in aquatic ecosystems. However, a large body of evidence demonstrates that PPCPs and other chemical contaminants can impact biodiversity and ecosystem services, especially at the local and landscape scales (Bernhardt et al., 2017). Landscapes not only drive the production, use, and emission of man-made chemicals, but also provide habitat for species assemblages (Li et al., 2021). Both the presence of chemicals and the occurrence of species can be measured by new analytical methodologies, such as high-resolution mass spectrometry and environmental DNA technologies (Zhang, 2019). The integration of satellite remote sensing data and in-situ monitoring, in addition to environmental exposure modeling and machine learning, is most likely to inform the precise ecological risks of PPCPs, other anthropogenic contaminants and natural toxins in the future.

FUTURE PERSPECTIVES

The Global Horizon Scanning Project recently identified that step changes are needed in the ways we assess and manage chemical risks to public health and the environment (van den Brink et al., 2018; Fairbrother et al., 2019). Over the past decade, advances in –omics, AOPs, and population and bioenergetic modelling efforts in ecotoxicology have afforded unprecedented opportunities to begin to answer, “What can be learned about the evolutionary conservation of PPCP targets across species and life stages in the context of potential adverse outcomes and effects?” Acceleration within these efforts, along with developments in precision medicine and precision environmental health science, has provided a foundation from which precision ecotoxicology can move forward. But much more basic and translational work is needed across scales of biological organization and environmental gradients for PPCPs within the context of other stressors (Sumpter et al., 2023; Mo et al., 2023). As with other basic science advances within the ecotoxicology discipline, translation to the practice of risk assessment will be crucial to advance environmental management (Oldenkamp et al., 2023). To achieve paradigm shifts in chemical risk assessment and management activities, it will be important to advance NAMs for PPCPs and other chemicals within the risk assessment paradigm (Ankley et al., 2023) and to realize more rapid integration of scientific advances by regulatory agencies (Oldenkamp et al, 2023).

An understanding of the multinational institutions that are currently in place, together with how consumer patterns, policy developments and technological advancements are going to change the risk assessment landscape, will ensure sufficient preparedness of individuals and institutions to lead effective and more sustainable environmental management efforts. Herein, it is important to reduce, and pragmatically replace, the use of animals in toxicology. Therefore, we need to advance NAMs (Langan et al., 2023) to reduce testing with laboratory animals as we aim to advance green and sustainable chemistry, better protect the environment, and restore ecological damages. Building from a developing understanding of evolutionary conservation of PPCP targets and potential linkages with adverse outcomes, it will be important to engage opportunities and limit challenges of the applicability of mechanistic approaches for biological read-across based on evolutionary conservation that incorporates complexity when extending observations from laboratory to the field. Such considerations will be key to realizing a future where animal studies are unnecessary and ecosystem level AOPs are available as we sustainably protect biodiversity and ecosystem services from chemicals and waste.

Data availability

No data are associated with this Perspectives Article.