An Adverse Outcome Pathway Linking Organohalogen Exposure to Mitochondrial Disease

Adverse outcome pathways (AOPs) are pragmatic tools in human health hazard characterization and risk assessment. As such, one of the main goals of AOP development is to provide a clear, progressive, and linear mechanistic representation of pertinent toxicological key events (KEs) occurring along the different levels of biological organization. Here, we present an AOP framework that depicts how exposure to organohalogens can lead to mitochondrial disease. Organohalogens are disinfectant by-products (DBPs) found in our drinking water. Chloroform, trichloroacetic acid, and trichlorophenol were selected to represent specific types of organohalogens for the development of this AOP. Although each of these compounds contains chlorine atoms, they differ in aromaticity and solubility, which have a significant impact on their potency. This AOP consists of two main pathways, both of which are triggered by the molecular initiating event (MIE) of excessive reactive oxygen species generation. Pathway 1 details the downstream consequences of oxidative stress, which include mitochondrial DNA damage, protein aggregation, and depolarization of the mitochondrial membrane. Pathway 2 shows the KEs that result from inadequate supply of glutathione, including calcium dysregulation and ATP depletion. Pathways 1 and 2 converge at a common KE: opening of the mitochondrial membrane transition pore (mPTP). This leads to the release of cytochrome c, caspase activation, apoptosis, and mitochondrial disease. This AOP was developed according to the Organisation for Economic Co-operation and Development guidance, including critical consideration of the Bradford Hill criteria for Weight of Evidence assessment and key questions for evaluating confidence. The presented AOP is expected to serve as the basis for designing new toxicological tests as well as the characterization of novel biomarkers for disinfectant by-product exposure and adverse health effects.


Introduction
In the early 1960s, researchers identified mitochondrial disease as a serious clinical condition and have since increased their efforts to identify its etiology [1]. Mitochondrial diseases are progressive, chronic, and irreversible illnesses that result from failure of mitochondrial organelles, which are specialized cellular compartments responsible for generating more than 90% of the energy required to sustain life. In general, mitochondrial failure leads to cell injury, closely followed by cellular demise [2]. When multiple cells expire through this pathway, the most common adverse outcome (AO) is organ failure [3][4][5][6].
With the exception of red blood cells, all human cells contain mitochondria. Therefore, mitochondrial dysfunction can occur in nearly any organ system of the human body and cause a variety of adverse health conditions, ranging from mild (i.e., nausea or mild cognitive impairment) to severe (heart failure or Parkinson's Disease) [7][8][9]. The most common symptoms of mitochondrial dysfunction include loss of muscle coordination, muscle weakness, developmental delays, learning disabilities, heart disease, diabetes, gastrointestinal disorders, liver disease, kidney disease, and neurological problems [10][11][12][13][14].
Mitochondrial dysfunction becomes mitochondrial disease as soon as mutations caused by xenobiotic exposure, genetics, or a combination thereof are identified. Mutation markers can be found in either mitochondrial or nuclear DNA [15][16][17]. While most cases of mitochondrial disease are linked to a genetic malfunction, substantial evidence published in the recent literature suggests that environmental triggers or exposure to certain xenobiotics may be responsible Toxicity LD 50 (rats,oal) 695 mg/kg [51][52][53][54] 425 mg/kg [55] 670 mg/kg [56][57][58][59] for the onset of some mitochondrial diseases [18,19]. Exposure to pesticides is one environmental trigger that is frequently cited in the literature [20][21][22]. Ingestion of disinfectant by-products (DBPs) from drinking water may also be a trigger of mitochondrial disease. Risk assessments performed on these environmental exposures provide evidence for addressing this growing public health concern [23][24][25][26][27]. Disinfectant products, such as chlorine or chloramine, are added to the drinking water supply to mitigate the onset of waterborne illnesses due to microorganisms. While this technology has greatly improved water quality around the world, the addition of chlorinated compounds to water has produced an unintended consequence: unnatural disinfectant by-products (DBPs) littering plumbing within the water distribution system (Table 1). Chlorine reacts readily with water constituents (such as metal ions and carbonaceous species) to produce DBPs, which have been associated with a variety of human health effects, including cancer [28][29][30][31][32][33]. A substantial amount of research has been performed investigating the chemical mechanisms of formation and subsequent compound identification [34][35][36][37]. However, fewer studies have reported the toxicological mechanisms of action after DBP exposure to humans [38][39][40]. The chemical mechanisms of DBP formation aid in exposure analyses, but, without adequate toxicological mechanisms of action reported, risk assessments are difficult to perform.
Toxicologists face many challenges when performing human risk assessments, such as incomplete dosimetry information, disparate in vitro and in vivo hazard results, and lack of human epidemiological data [41][42][43]. There is an ever-increasing number of substances (i.e. chemicals, particles, aerosols, pharmaceuticals, advanced materials, and by-products) that should be tested for toxicity, evaluated for exposure, and assessed for risk. Unfortunately, insufficient resources make these thorough analyses difficult to perform in a timely and cost-effective manner [44,45]. Additionally, there is pressure to reduce the use of animals used in toxicological analyses [46]. Therefore, it is increasingly necessary to use available data from the literature to design, collect, and interpret new toxicological studies that answer unresolved questions or gaps in data [47,48].
One promising pathway-based analysis tool is adverse outcome pathway (AOP) development ( Figure 1). An AOP describes the progression of adverse health effects from lower-level molecular reactions to higher-level disease onset [47,49]. AOP development begins with identifying a molecular initiating event (MIE) and ultimately concludes by recognizing the adverse outcome(s) (AO) of regulatory significance. The MIE and AO are related via a sequence of biologically plausible and scientifically supported key events (KEs) of increasing complexity. The relationships between the KEs are activated through structural and functional relationships coupled with weight of evidence criteria [50].
The purpose of this manuscript is to explain and support a developed AOP that relates a global environmental exposure (ingestion of organohalogens) to mitochondrial dysfunction (opening of the mitochondrial permeability transition pore, mPTP) using individual mechanistic events identified in the peer-reviewed literature ( Figure 2). This AOP has the potential to serve the scientific community as a basis for the development of new and targeted toxicological tests as well as the characterization of novel biomarkers of oxidative stressinduced organ system dysfunction.  AOP developers use different strategies to develop a single or network of pathways leading to adverse outcomes. The strategy most often utilized during the early stages of AOP development is data-mining [49]. This process includes analysis of relevant literature and database mining approaches to infer relationships between KEs (a.k.a, KERs). Mitochondrial disease AOPs are in their infancy; therefore, we used a combination of keyword searches. We conducted an extensive literature search using a combination of the following terms: "(mitochondrial * ) AND (AOP OR adverse outcome pathway)". The primary databases searched were PubMed, Web of Science, and Scopus. The search results were not restricted by a date range; however, most papers included in this analysis were published within the last 20 years (1998-2018). A total of 70 unique papers were returned with these keyword phrases. On the other hand, papers linking "(mitochondrial * ) AND (environment * )" are plentiful. A total of 15,221 unique papers were returned with these keywords. The next section describes the screening of these papers in an effort to narrow the developed AOP.

AOP Representation and Evaluation.
The search terms "mitochondrial" AND "environmental" produced too much data to mine AOP key events and their associated relationships. From this initial search, we screened abstracts using specific terms related to different phases with in the AOP:  Figure 3: Graphical representation of organohalogen exposure to mitochondrial disease AOP. Key events at the molecular and cellular levels lead to AOs at the tissue/organ and individual levels. Exposure to organohalogens cause increased ROS production (Molecular Initiating Event, MIE), triggering the key events (KEs) that result in the adverse outcome (AO), defined as the onset of mitochondrial disease and the associated symptoms of organ symptom dysfunction. This AOP depicts 2 distinct pathways from MIE to AO.
(1) First, we included "organohalogen" and eliminated other environmental factors to build the molecular level.
(3) Third, we included "mitochondrial permeability", "caspase activation", "apoptosis" to develop the organ level. Articles that focused on mitochondrial dysfunction and genetic causes were excluded.
Upon applying inclusionary and exclusionary criteria, we were left with a total of 189 unique eligible articles. This process ensured that the articles with the most relevance were included in the final AOP.
Appropriate KEs and their relationships were reported using a flow diagram showing the key event relationships (KERs) along the increasingly complicated levels of biological organization (i.e., molecular, cellular, tissue/organ, and individual levels) in a consecutive manner (Figure 1). Connections between events were decided based on a strengthbased weight-of-evidence (WoE) assessment of the MIE, KE, and AO linkages ( Figure 3). A final evaluation was conducted in two steps. First, Bradford Hill criteria were used to assess the WoE of the AOP to establish a causal link between the different blocks representing biological organization (Table 2). Second, we reported the confidence associated with each causal link by addressing OECD's proposed questions (Figures 4,5,6,and 7). This evaluation process ensures that the resultant AOP meets the minimal information requirements to establish plausibility of the proposed pathway.

Weight of Evidence
3.1.1. Empirical Support of the KERs. Based on well-established knowledge of mitochondrial function, the biological plausibility between increased ROS production and mitochondrial disease is strong. This is often associated with mitochondrial electron transport chain disruption or complex I inhibition [60][61][62]. Figure 2 contains the details for the biological plausibility of the KERs.
The degree of empirical support for the KERs ranges from insufficient to strong evidence. Organohalogens produce ROS species in water [63][64][65][66][67]. Many of these tri-chlorine containing compounds have a high affinity for disruption of mitochondrial electron transport chain (ETC) [63,65]. Table 1      Journal of Toxicology                 14 Journal of Toxicology      Increased ROS concentration caused by exposure to these organohalogens has been found to cause oxidative stress (A) and glutathione depletion (K).
Glutathione is an essential antioxidant and is one of the cell's main defenses against reactive oxygen species. When cells are exposed to a toxic amount of these organohalogens, the amount of ROS generated exceeds the amount of mitochondrial glutathione being produced, and glutathione stores become depleted. Glutathione depletion further increases cellular oxidative stress (L) because the cell lacks one of its main antioxidants. Glutathione depletion is a key event in the apoptotic cascade.
Weight of Evidence Oxidative stress causes strand breaks in mitochondrial DNA (mtDNA) and loss of its supercoiled structure, leading to genetic mutations (B). mtDNA is recognized as a major cause of cellular apoptosis, but the mechanisms of this are still not clear. Based on our research, we believe that the above-mentioned changes in mtDNA directly contribute to depolarization of the mitochondrial membrane (C). Unfortunately, there is not an abundant amount of literature on this, so it this KER is labeled as "weak". This mechanism should be studied further. Depolarization of the mitochondrial membrane causes opening of the mitochondrial permeability transition pore (mPTP) (D). mPTP opening allows rapid release of cytochrome c into the cell (E). Upon release from the mitochondria, cytochrome c activates caspases (F), which trigger apoptosis (G).    lists the physicochemical characteristics of three notable organohalogens (i.e., chloroform, trichloroacetic acid, and trichlorophenol) that are believed to inhibit the normal function of mitochondria in mammalian cells.

Weight of Evidence
There is extensive indirect evidence of chloroform acting as an inhibitor of the electron transport chain through the generation of ROS, induction of oxidative stress conditions, and depletion of glutathione and ATP [68,69]. Exposure to chloroform induces a dose-dependent cytotoxic response. It is classified as a Group 2B carcinogen by the IARC, however its propensity to induce tumors in humans is limited. LD 50 values assessed from oral exposure to chloroform vary in rodent studies; values range 446 mg/kg for 14-day-olds male rats to 2,180 mg/kg for adult rats [51,52]. But the mechanism of action at low-dose exposures is not yet established. Data suggest that chloroform metabolites produce DNA mutations [53,54]. Chloroform-induced cell death is observed through the biochemical hallmarks of apoptosis, cytochrome c release, and activation of caspases 3 and 9. Cytochrome c is released into the cytosol in a time-dependent manner. The longer cells are exposed to organohalogens, the more cytochrome c is released.
Chloroacetic acid also interferes with the mitochondria. Unlike chloroform, it is not classified as a carcinogen. The metabolites of chloroacetic acid, including glycolic acid and oxalate, do generate ROS within the cytosol and quickly deplete glutathione supplies. The LD 50 value of chloroacetic acid/trichloroacetic acid is reported to be 425 mg/kg (rats, oral) [55]. Chlorophenol acts in a similar manner but at lower doses (LD 50 of 670 mg/kg (rats, oral)), making it more cytotoxic than chloroform (695 mg/kg), but less toxic than chloroacetic acid (425 mg/kg) [56][57][58][59]. Chloroacetic acid may be less cytotoxic because of its high-water solubility (greater than 10,000 g/L at 20 ∘ C). It is able to metabolize quickly through Phase 1 and Phase 2 processes as compared to the less water-soluble compounds (trichloromethane (8.09 g/L) and trichlorophenol (20.0 g/L)). Trichlorophenol may be more cytotoxic because of its aromatic structure. Aromatic compounds are generally metabolized by P450 enzymes and form epoxides, which cause DNA damage (mitochondrial or nuclear).

Graphical Representation and Plausibility of the KERs.
Many of the individual KERs within this developed AOP are strong. Figure 3 presents the overall graphical representation of organohalogen exposure to mitochondrial disease with each pathway's biological plausibility detailed in subsequent figures. Refer to Table 2 for the specific references from the scientific literature that support each of these KERs.
The electron transport chain, located on the inner membrane of the mitochondria, is the site of oxidative phosphorylation. When mammalian cells are exposed to organohalogens, their electron transport chain produces an increased amount of reactive oxygen species (ROS), inducing oxidative stress conditions within the cell (A). In an attempt to mediate oxidative stress, glutathione, a powerful antioxidant, is put to work. Eventually, ROS is produced faster than the cell produces antioxidants and glutathione stores are depleted (K), allowing the oxidative stress conditions to persist (L).
Because mitochondrial DNA is housed so close to the electron transport chain, it is exposed to high amounts of ROS, which causes strand breakage and loss of its tertiary structure (B). These changes affect mtDNA translation, eventually leading to depolarization of the mitochondrial membrane (C). Depolarization of the mitochondrial membrane contributes to opening of the mitochondrial permeability transition pore, or mPTP (D). When the mitochondrial membrane opens, cytochrome c is released into the cytosol of the cell (E). Cytochrome c activates caspases (F), which trigger the apoptotic cascade (G).
In addition to mitochondrial DNA damage, oxidative stress conditions also inhibit the function of the ubiquitin proteasome system (H), or UPS, which functions to clear out aggregated proteins within the cell. UPS dysfunction leads to a buildup of aggregated proteins within the cell (I). These aggregated proteins also contribute to depolarization of the mitochondrial membrane (J), which then triggers opening of the mPTP, cytochrome c release, caspase activation, and apoptosis.
Glutathione depletion caused by increased ROS production contributes to sustained oxidative stress conditions (L) and inhibits the production of ATP (O) in complex I of the mitochondrial electron transport chain. Both glutathione depletion and ATP depletion cause an interruption of calcium homeostasis within the cell (M, P). These conditions cause a large influx of calcium into the cell, which is then transported into the mitochondria in an attempt to regulate cellular calcium levels. This causes an overload of calcium in the mitochondria, which triggers opening of the calcium-mediated mPTP (N). Opening of the mPTP triggers cytochrome c release, caspase activation, and apoptosis, as discussed above.
Caspase activation causes two notable adverse biochemical reactions. The first is apoptosis, which occurs at the cellular level (G). The second is mitochondrial disease, which occurs at the tissue level. Apoptosis in cells with high mitochondrial content also contributes to the onset of mitochondrial disease. Mitochondria are responsible for providing cells with the energy needed to function normally. When mitochondrial disease is present, the cells are unable to function as efficiently, and therefore the organs are unable to function efficiently. This leads to symptoms of organ system failure or dysfunction. Apoptosis in cells with high mitochondria content can also lead to symptoms of organ dysfunction. Symptoms vary, depending on which organ(s) are affected by the mitochondrial dysfunction as well as which part of the organ is affected. For example, mitochondrial disease affecting the nigrostriatal pathway in the brain can cause motor deficits as manifested by Parkinson's disease [70]. Meanwhile, mitochondrial disease of the heart muscle can cause edema, shortness of breath, and arrhythmias, as manifested by mitochondrial cardiomyopathy [71].
It is important to note that many of the KEs presented in these pathways contain positive feedback loops. For example, calcium dysregulation, glutathione depletion, and ATP depletion all contribute to a further increase in ROS generation. These feedback loops were omitted from the figures for the sake of simplicity, but it is an important to recognize that these synergistic relationships do exist.

Applicability of the AOP.
In terms of taxonomic applicability, this AOP describes the toxic mechanisms of organohalogens on mitochondrial electron transport chain and is therefore relevant to any organism that utilizes mitochondria for energy production. Mitochondrial disease can be caused by mitochondria respiratory deficiencies, which are often present at birth or in early childhood and typically result in progressive muscular and neurological degenerative disorders. Adult onset of mitochondria respiratory defects is increasingly common, thus increasing the applicability of organohalogens-to-mitochondrial disease AOP to multiple stages of life. Mitochondria serve several roles in maintaining homeostasis and these roles evolve from fertilized egg to old age. Although this AOP focuses on organohalogen exposure, this pathway can be adapted to fit a wide range of environmental triggers of mitochondrial disease.

Discussion
An AOP was created to best characterize the pathway-based analysis of organohalogen exposure that results in dysfunction of the mitochondrial electron transport chain in humans ( Figure 1). For the purposes of this AOP, more emphasis was placed on the front-end of the AOP (MIE and the KEs) than the back-end (the AO). Mitochondrial disease can present itself in various physiological or toxicological manifestations, which makes it difficult to pinpoint one specific disease of interest. Instead, we provide evidence of the MIE responsible for initiating irreversible damage to mitochondria, which causes cell death and progression towards a diagnosable mitochondrial disease.
There is a suite of exogenous substances that have be shown to trigger changes in mitochondrial flux. While organohalogen compounds are highly cited as generators of ROS in cell and tissue systems, other materials, such as exhaust fumes, pesticides, tobacco products, smoke, drugs, or metals have been shown to increase oxidative stress in biological tests systems [72][73][74]. Furthermore, intangible stressors, such as radiation, heat, or ultraviolet light exposure, induce similar oxidative stress endpoints [75][76][77][78]. Most recently, exposure to engineered nanomaterials has been consistently linked to ROS generation in both cell-based and cell-free test systems which has been implicated as the main source of inflammatory responses in rodent studies [79][80][81].
Relating individual components of this AOP to other chemicals, particles, and fibers is possible. Inter-relationships between (or the ability to read-across) multiple test substances are dependent upon the level of biological organization. For example, the properties among classes of test substances vary greatly; i.e., chemicals are generally in a liquid form where particles and fibers are in a solid form. Liquids and solids have different physicochemical characteristics and hence the molecular initiating event is dependent upon the immediate chemical reaction between the substance of interest and subcellular entities; the physical (shape or size) and chemical (solubility or composition) characteristics often dictate the biochemical effects. Therefore, there are limitations in establishing interrelationships among different materials. However, as the AOP moves downstream from MIE to higher levels of biological organization, the dependency on physicochemical properties of the test system becomes less pronounced. Key events, key event relationships, and adverse outcomes at the tissue, organ, and individual levels are increasingly independent of material properties. Therefore, the ability to define interrelationships among classes of materials is possible.
The MIE is considered the "first domino" of the pathway, meaning that once the MIE occurs, reversible or irreversible damage will continuously cause KEs until the AO is achieved. Without the MIE, the progression of the AOP is halted and the KEs will not occur [47,49,50]. With regards to this AOP, the necessary MIE is the increase in ROS species. We have identified two main pathways that contribute to the development of mitochondrial disease from organohalogen exposure. These pathways diverge from the MIE and converge onto a single KE: opening of the mitochondrial permeability transition pore (mPTP).
Pathway 1 follows the direct effects of oxidative stress conditions caused by increased ROS generation. There are two main effects of oxidative stress identified in this pathway. The first is mitochondrial DNA (mtDNA) damage (Pathway 1A). Because ROS damages protein structures, the free radicals cause mtDNA strand breaks, loss of supercoiled structure, and interrupt mtDNA translation [82][83][84]. The second is dysfunction of the ubiquitin-proteasome system, which causes protein aggregation (Pathway 1B) [85,86]. Both mtDNA damage and protein aggregation contribute to depolarization of the mitochondrial membrane, which in turn causes opening of the mPTP [87,88].
Pathway 2 follows the direct effects of glutathione depletion caused by increased ROS concentration [89]. We have identified three effects of glutathione depletion. The first is that, because glutathione is one of the cell's main antioxidants, its depletion contributes to oxidative stress conditions (Pathway 2A) [90]. Glutathione depletion also causes a dysregulation in calcium homeostasis (Pathway 2B) [91]. When glutathione stores are depleted, a large amount of calcium is sent into the cell, which is then sequestered into the mitochondria and causes calcium overload [92]. This calcium overload within the mitochondria triggers opening of the Ca 2+ -mediated mPTP [93]. ROS generation and glutathione depletion perturb the mitochondrial electron transport chain [94]. As a result, the enzyme is unable to produce the proton gradient needed for ATP synthase to convert ADP to ATP, and ATP levels are depleted (Pathway 2C) [95]. ATP depletion contributes to calcium dysregulation in the same manner as glutathione depletion, as described above [96].
Pathways 1 and 2 converge on a common Key Event: opening of the mPTP. This is a crucial step in the apoptotic pathway, as it allows release of cytochrome c into the cytosol. Cytochrome c then activates caspases, which are responsible for programmed cell death.
The plausibility table included in Figure 5 outlines the research needed to strengthen the weight of evidence for key event C, i.e., mitochondrial DNA damage to depolarization of mitochondrial membrane. A few noteworthy studies have found evidence for this key event relationship [87,97,98]. Specifically, however, more evidence is needed to strengthen the KER by demonstrating that oxidative stress causes strand breaks in mitochondrial DNA (mtDNA) and subsequent loss of its supercoiled structure. Loss of structure is believed to lead to genetic mutations. Mitochondrial DNA (mtDNA) is recognized as a major cause of cellular apoptosis, but the exact mechanisms of toxic action are still not clear. Based on our research, we believe that the abovementioned changes in mtDNA directly contribute to depolarization of the mitochondrial membrane. Unfortunately, there is not an abundant amount of literature on event C so it this KER is labeled as "weak". This mechanism should be studied further.
The Adverse Outcomes (AOs) identified in this AOP are mitochondrial disease (at the tissue/organ level) and symptoms of organ system dysfunction (at the individual level). The onset of mitochondrial disease occurs when enough cells in an organ exhibit significant caspase activation and/or induced apoptosis [13,99]. Caspases cause programmed cell death via apoptosis [100]. After cell death, the tissue and organ to which the deceased cells belong become nonfunctional or nonviable, resulting in tissue and organ system failure. This systemic failure can lead to the typical clinical symptoms associated with mitochondrial disease, such as neurodegeneration, impaired growth, muscle weakness, and developmental delays, before evolving into a diagnosable mitochondrial disease (AO) [101,102]. In other words, the inhibiting ability of organohalogen compounds has multilevel detrimental effects, including the macromolecular level, cellular/tissue level, organ level, organism level, and population or ecosystem level.

Conclusion
Mitochondrial function is an evolving field of study. Its role was previously described in simple terms, such as "mitochondria is the powerhouse of the cell"; however, new research is emerging that enables our understanding of mitochondria-to-disease pathways. With the motivation to reduce the number of animals in experimental designs and to create unprecedented value in the enormous amount of published studies on mitochondrial dysfunction, AOP development has the unique opportunity to serve as a framework for understanding mitochondrial disease. The effect of organohalogen exposure on intracellular respiration and oxidative stress conditions is one example in a long list of possible environmental exposures causing mitochondrial dysfunction. Overall, the biological plausibility of the key events and their relationships is strong and exists in the form of in vivo and in vitro studies. Although the strength of the biological plausibility of our AOP is strong, the strength of empirical evidence ranges from weak to strong, with the least amount of evidence existing to support the AO.

Data Availability
The data used to support the findings of this study are available from the corresponding author upon request.

Conflicts of Interest
The authors declare that they have no conflicts of interest.