Mechanisms of Action of Agrochemicals Acting as Endocrine Disrupting Chemicals

Abstract

Agrochemicals represent a significant class of endocrine disrupting chemicals that humans and animals around the world are exposed to constantly. Agrochemicals can act as endocrine disrupting chemicals through a variety of mechanisms. Recent studies have shown that several mechanisms of action involve the ability of agrochemicals to mimic the interaction of endogenous hormones with nuclear receptors such as estrogen receptors, androgen receptors, peroxisome proliferator activated receptors, the aryl hydrocarbon receptor, and thyroid hormone receptors. Further, studies indicate that agrochemicals can exert toxicity through non-nuclear receptor-mediated mechanisms of action. Such non-genomic mechanisms of action include interference with peptide, steroid, or amino acid hormone response, synthesis and degradation as well as epigenetic changes (DNA methylation and histone modifications). This review summarizes the major mechanisms of action by which agrochemicals target the endocrine system.

Introduction

In Silent Spring in 1962, Rachel Carson recognized that chemicals in the environment, particularly the pesticide dichlorodiphenyltrichloroethane (DDT) and other agrochemicals, were impacting reproduction and development in wildlife and had the potential to harm humans (Carson, 1962). Although she did not know it at the time, Rachel Carson was describing the effects of endocrine disrupting chemicals (EDCs), compounds that interfere with the action of hormones in the body (Gore et al., 2015). Over the past half century, scientists have identified the mechanisms through which DDT causes the softening of eggshells and decreases in bird populations described in Silent Spring as well as numerous other mechanisms through which environmental chemicals can interfere with the complex operations of the endocrine system.

In this review, we summarize endocrine disruptor mechanisms of action with an emphasis on agrochemicals that act as EDCs. The endocrine system operates through complex and finely tuned hormone signaling pathways. As such, EDCs exhibit numerous modes of action. The most well-known mechanisms of EDC action involve mimicking the interaction of endogenous hormones with nuclear receptors, but evidence supporting non-nuclear receptor-mediated mechanisms of action will also be discussed herein.

The Endocrine System

The endocrine system controls reproduction, development, growth, metabolism, tissue and brain function, and other physiological functions in the body. Endocrine glands distributed throughout the body, including the brain, thyroid, mammary glands, cardiovascular system, and reproductive organs, produce and release hormones. Understanding the complexities of hormone signaling provides vital context to the wide range of EDC mechanisms.

The canonical pathway of hormone signaling involves the binding of a hormone to its corresponding nuclear receptor(s). Ligand binding induces structural changes in the receptor that lead to dimerization, exposure of co-factor binding sites, and DNA binding. Genomic binding may occur directly to response elements in the genome (Figure 1A) or through transcription factors (Figure 1B). Hormones may also function through non-genomic rapid signaling via secondary messengers (Figure 1C). Exogenous chemicals, including pharmaceuticals and EDCs, can mimic hormones as ligands via any of these three mechanisms as well as repress transcription by recruiting co-repressors (Figure 1D) (Heldring and Pike, 2007).

Figure 1: Model of the canonical hormone signaling via nuclear receptors (NR).

Hormones (grey triangles) act as ligands for nuclear receptors which can bind directly to DNA (A) or indirectly via transcription factors (TF) (B) to regulate gene expression. Hormones may also signal through non-genomic pathways by interacting with membrane nuclear receptors or cytoplasmic receptors to initiate signaling cascades via secondary messengers (SM) that lead to rapid physiological effects without altering gene expression (C). Hormone mimicking chemicals may operate through any of the mechanisms shown in A-C as well as recruit co-repressors to block gene transcription (D). Adapted from Helding and Pike, 2007.

Extremely low concentrations of hormones are present in the body, on the order of pg–ng/mL for estradiol, testosterone, and thyroid hormone (Vandenberg et al., 2012). Furthermore, only small percentages of many circulating hormones, including steroid and thyroid hormones, are bioavailable; the majority are bound to carrier proteins. The endocrine system is tuned to respond to minor changes in bioavailable hormone concentrations. Signaling at these extremely low concentrations is facilitated by the high affinity of hormones for their receptors as well as non-linear relationships between hormone concentration, receptor occupancy, and biological effect (Welshons et al., 2003). Thus, a change in hormone concentrations at low doses has a stronger effect than the same magnitude of change at a higher dose. For a full discussion of this, see Vandenberg et al., 2012. In the context of understanding endocrine disruptor mechanisms, this means that studying low doses of EDCs (below those typically used in toxicological testing) is required to identify mechanisms, and that the mechanisms may change at different doses. This is evident in non-monotonic dose response curves (NMDRCs) observed in toxicological studies of EDCs, in which the sign of the curve changes (i.e. U or inverted U) as the dose changes. NMDRCs are established for hormonal pharmaceuticals such as tamoxifen yet they continue to be challenged in the context of EDCs (Vandenberg et al., 2012).

EDC Mechanisms of Action

Endocrine disrupting chemicals can bind to receptors to mimic endogenous hormones, but they also act by altering hormone signaling in a variety of other ways. EDCs may interact with multiple receptors, including non-nuclear receptors, as agonists, in which they facilitate genomic interactions, or as antagonists, in which they cause a conformational change to the receptor to block action. They may also trigger non-genomic signaling that is independent of nuclear receptors. Importantly, EDCs can interfere with endogenous hormone synthesis and degradation to alter hormone levels. Recent studies have also identified how EDCs can follow an epigenetic mode of action by altering genomic methylation and histone modifications.

Nuclear Receptors

Nuclear receptors are ligand dependent transcription factors that bind hormones and exert long term control of their target cell phenotype. This is in contrast to membrane receptors which elicit faster, short term effects on their respective cells. Nuclear receptors play a crucial physiologic role in development, organ homeostasis, metabolism, immune function, and reproduction (Balaguer et al., 2019). Hormones are the endogenous ligands of most of the major classes of nuclear receptors, but EDCs may also bind nuclear receptors. Nuclear receptors can act as repressors or activators of gene transcription depending on ligand binding status, the identity of the ligand itself, and available coregulators.

Nuclear receptors are composed of several conserved structural domains, including a DNA binding domain and a ligand binding domain (Weikum et al., 2018). The DNA binding domain is responsible for interacting with the genome and typically interacts with specific sequences called DNA response elements, whereas the ligand binding domain contains the binding pocket for small molecules. Unliganded nuclear receptors can be found in the nucleus or cytoplasm and may bind to target genes, resulting in recruitment of corepressors. Following endogenous ligand binding, conformational changes to nuclear receptors disrupt corepressors leading to dimerization and interaction with DNA and coactivators. Upon binding to DNA or interacting with tethering co-factors, nuclear receptors further activate co-factors to facilitate transcription.

Endocrine disrupting chemicals can interact with nuclear receptors directly and elicit strong biological consequences (Gore et al., 2015). Agricultural chemicals, including herbicides, insecticides, rodenticides, and fungicides, are prime examples of synthetic endocrine disrupting molecules exogenous to normal eukaryotic biology that interact with nuclear receptors (Beischlag et al., 2008). Table 1 provides examples of the biological effects of select agrochemicals that act through nuclear receptors.

Table 1:

Agrochemical effects mediated through nuclear receptors

Chemical	Mode of Action	Effect	Reference
atrazine	AhR mediated	Nephrotoxicity	(Zhang et al., 2019)
dibutyltin (DBT)	PPARγ activation	Increased expression of adipogenic genes, promoted adipogenic differentiation, increased lipid accumulation, decreased glucose tolerance, increased circulating leptin levels in males	(Chamorro-García et al., 2018)
DDT	Androgenic antagonist: inhibits dihydrotestosterone (DHT) binding to AR	Decreased fertility and cryptorchidism. Reduction of testicular weight, decreased in number and motility of spermatozoa in epididymis, loss of gametes in lumen of seminiferous tubules, decreased testosterone production by testes, increased luteinizing hormone and follicle stimulating hormone	(Kelce et al., 1995; Sakly, 2001)
	ESR1 activation leading to altered expression of Pgr, Ccnd1, Cyp19a1	Hormonal carcinogenesis in uterus and ovaries	(Kalinina et al., 2017)
DDE	AR antagonist: inhibits DHT binding to AR	Decreased fertility and cryptorchidism	(Kelce et al., 1995)
dieldrin & aldrin	AR antagonist	Reproductive performance affected	(Danzo, 1997)
fenarimol	AR antagonist leading to decrease in prostate binding protein, ornithin decarboxylase, and insulin-like-growth factor 1 mRNA levels	Reduced weight of ventral prostate, seminal vesicles, and bulbourethral glands	(Vinggaard et al., 2005)
lindane (γ-HCH)	Androgenic antagonist: inhibits DHT binding to AR	Altered testis histology	(Danzo, 1997)
linuron	AR antagonist	Disruption of reproductive tract development, reduction of accessory gland weight, reduction of epidydimal weight, increased serum estradiol and luteinizing hormone	(Lambright, 2000)
methoxychlor	ESR1 and ESR2 agonist	Premature nuclear expression of ER gene in neonatal uterine epithelium	(Eroschenko et al., 1996)
methoxychlor procymidone	ESR1 and ESR2 agonist AR antagonist	Inhibition of folliculogenesis and stimulation of anti-Müllerian hormone production in the ovary	(Uzumcu et al., 2006)
		In males: shortened anogenital distance, hypospadias, reduced weight and altered histology of prostate, permanent nipples, ectopic undescended testes	(Ostby et al., 1999)
propiconazole & tebuconazole	Pregnane X receptor (PXR) activation	Hepatocellular hypertrophy and hepatocellular steatosis	(Knebel et al., 2019)
2,3,7,8-tetrachlorodibenzodioxin (TCDD)	AhR activation during in utero exposure	Cleft palate	(Abbott et al., 1996)
2,3,7,8-tetrachlorodibenzodioxin (TCDD) tributyltin (TBT)	AhR activation	Downregulation of uterine and hepatic ESR1 results in repressed estradiol function	(Romkes et al., 1987)
	AhR activation	UGT1 induction resulting in decreased serum T4 concentrations and compensatory excretion of thyroid stimulating hormone	(Nishimura et al., 2002)
	PPARγ and retinoid X receptor (RXR) activation	In utero exposure results in increase lipid accumulation in adipose tissue, liver, and testis	(Grün et al., 2006)
tributyltin (TBT) vinclozolin	PPARγ induction and Esr1 repression	Impaired metabolic functions on liver and pancreas via lipid accumulation in white adipose tissue and hepatic inflammation	(Bertuloso et al., 2015)
	AR antagonist	Hypospadias, undescended testes, delayed puberty, transgenerational prostate disease	(Monosson et al., 1999; Christiansen et al., 2008; Anway, 2015)

The structural similarities between EDCs and endogenous hormones form the basis for nuclear receptor-based mechanisms of endocrine disruption. Both classes of molecules are generally hydrophobic, lipid-soluble small molecules. In addition, many of the most worrisome EDCs have structural elements such as hydroxyl groups in the proper position to interact with amino acid residues in the ligand binding pocket in the same manner as endogenous hormones. For example, the methoxychlor metabolite 2,2-bis-(p-hydroxyphenyl)-1,1,1-trichloroethane (HPTE) contains both hydroxyl groups and hydrophobic aromatic regions that allow it to bind to estrogen receptors, whereas the pyrethroid deltamethrin shares a diphenyl ether structural moiety with thyroid hormones (Delfosse et al., 2015; Du et al., 2010) (Figure 2). Once EDCs bind a receptor, they either activate the hormone receptor, amplifying physiological hormonal activity, or antagonize endogenous hormone action to block activity. Due to the structural similarities between EDCs and endogenous hormones as exemplified in Figure 2, EDCs are typically investigated as competitive binders in the ligand binding domain of the receptor. In addition, EDCs which are not structurally similar to hormones can still directly affect nuclear receptors by stimulation or inhibition of receptor expression. This leads to imbalances in endocrine homeostasis through modification of hormone receptor turnover/availability.

Figure 2: Structural similarities between endogenous hormones and agrochemicals.

Top: Estradiol and the methoxychlor metabolite HPTE share phenolic structure (blue) that facilitates the binding of HPTE to human estrogen receptor 1 (ESR1) (Delfosse et al., 2015). The second hydroxyl group of HPTE does not interact in the ESR1 binding pocket in the same way as the second hydroxyl on estradiol. Bottom: The thyroid hormone triiodothyronine (T3) and pyrethroid deltamethrin share diphenyl ether functional groups (green) that facilitate binding to thyroid receptors (Du et al., 2010).

In vitro assays are valuable tools for investigating the ability of EDCs to directly interact with receptors and alter transcription processes. One commonly used assay is a recombinant receptor and reporter gene assay in which a luciferase reporter gene is associated with a nuclear receptor response element such that binding of a ligand to the receptor triggers transcription and luciferase activity (Legler et al., 1999). Although these assays are specific, responsive, and quick, they do not measure binding affinity nor identify which step of the nuclear receptor activation process is disrupted (Xiang et al., 2017; Judson et al., 2018). High throughput screening of nuclear receptor hormone agonists and antagonists is a top priority of regulatory agencies. In the US, the Tox21 program high throughput screening program utilizes multiple in vitro and in silico assays to assess ligand binding, dimerization, DNA binding, RNA transcription, and other steps in the activation process (Judson et al., 2018). However useful in vitro and silico models are for predicting EDC interaction with nuclear receptors, these models do not integrate multiple mechanisms of endocrine disruption that can occur in a whole animals. For this reason, in vivo testing is also necessary to understand mechanisms that may occur simultaneously in multiple organ systems to extrapolate to human and environmental health (Gore et al., 2015).

Estrogen Receptors (ESR)

Estrogen receptor alpha and estrogen receptor beta (ESR2) have unique and overlapping physiological roles that are highly tissue and cell type dependent. Both receptors are expressed in the brain, lungs, uterus, ovaries, breast, heart, and intestines. ESR1 is predominantly expressed in hepatocytes of the liver and the hippocampus, whereas ESR2 is predominantly expressed in the prostate, vagina, and cerebellum (Taylor and Al-Azzawi, 2000). ESR1 and ESR2 have quite similar binding pockets with subtle differences in amino acids at the ligand binding domain that explain the ligand selectivity between the two types (Kuiper et al., 1998). Both ESR1 and ESR2 regulate gene expression in response to estrogen exposure via ligand dependent or ligand independent mechanisms, and each subtype can mediate unique responses to ligands.

Inappropriate ESR signaling can lead to increased risk of hormone-dependent cancer, impaired fertility, abnormal fetal growth and development, and altered metabolism in white adipose tissue (Zhao et al., 2008). Many EDCs display estrogenic activity and interfere with normal estrogen signaling mediated by ESR1 and ESR2. A notable characteristic of ESRs that makes them susceptible to direct interaction with EDCs is their large binding pockets and broad specificity for ligands. The binding pockets are much larger than the estrogen molecule itself (Pike et al., 1999; Brzozowski et al., 1997). Although ESR1 and ESR2 have similar affinities for estrogen, EDCs and other exogenous ligands may display higher affinity for one subtype over the other. For example, the o,p’ isomer of DDT, which is the most estrogenic isomer of DDT, has a higher relative binding affinity for ESR2 than ESR1 (Kuiper et al., 1998).

Organochlorine pesticides, including methoxychlor and DDT, are notorious estrogenic agrochemicals. They bind both ESR1 and ESR2 (Kuiper et al., 1998) and elicit ER binding to DNA both directly and through tethering as described above. DDT adversely affects the female reproductive tract by stimulating uterine proliferation and impairing normal follicle development (Tiemann, 2008). Although DDT was banned in the 1970s, it is persistent in the environment and accumulates in adipose tissue, making it a relevant threat to public health today. DDT and its dechlorination metabolite dichlorodiphenyldichloroethylene (DDE) have been detected in human adipose tissue around the world many years after its use was terminated (Turusov et al., 2002).

Methoxychlor is another potent endocrine disruptor which stimulates uterotrophic activity and impairs overall fertility in rat models (Cummings, 1997). Methoxychlor itself has low binding affinities for ER. However, its major metabolite HPTE is a well-studied agonist for both ESR1 and ESR2 and is likely the responsible agent for the endocrine disrupting effects of methoxychlor (Gaido et al., 2000). HPTE acts as an agonist for ESR1 and antagonist for ESR2 (Gaido et al., 2000). ESR1 knockout mice do not respond to HPTE treatment, indicating the effects of HPTE on gene regulation in the mouse uterus are dependent on ESR1. HPTE may also mediate effects through non-classical ER signaling mechanisms similar to estrogens. HPTE and estrogen treatment led to similar gene expression profiles in uteri from mice expressing an ESR1 mutant deficient in DNA binding, which limits ESR1 mediated gene regulation to the pathway in which ESR1 tethers to DNA through transcription factors like AP-1 and Sp1 (Hewitt and Korach, 2011).

Pyrethroids also exhibit estrogenic and antiestrogenic activity, with variable results observed in tests of different chemicals in vitro, suggesting that individual screening of members of a class of pesticides is necessary to identify endocrine disrupting activity (Saillenfait et al., 2016). In addition to in vitro screening, whole-organism bioassays have been used to detect the estrogenic activity of EDCs. In transgenic medaka expressing green fluorescent protein, the herbicide linuron elicited anti-estrogenic activity, whereas fenoxycarb, a less-studied insecticide, showed no effect on estrogen, androgen, or thyroid signaling (Spirhanzlova et al., 2017). Recent studies have also investigated the estrogenic properties of mixtures of pesticides. An in vitro analysis using multiple screening methods of individual pesticides and mixtures on ESR1 and ESR2 found additive effects of the mixtures (Seeger et al., 2016), suggesting the need for further experiments on mixtures, which are more representative of human exposure.

Androgen Receptor

The androgen receptor (AR) is a ligand dependent nuclear transcription factor responsible for male fetal development, secondary sex characteristics at puberty, and maintenance of spermatogenesis. Beyond sex differentiation, the AR plays a critical role in development and maintenance of the reproductive, musculoskeletal, cardiovascular, immune, neural, and haemopoietic systems (Rana et al., 2014). Consequences of dysfunctional AR include infertility (Nenonen et al., 2011), delayed puberty onset (Mouritsen et al., 2013), and cryptorchidism (Davis-Dao et al., 2012). AR signaling is also involved in the development of tumors in the prostate, bladder, liver, kidney and lung.

The fungicides vinclozolin (Gray et al., 1994), procymidone (Hosokawa et al., 1993), and prochloraz (Vinggaard et al., 2002) target the AR and their phenotypic consequences have been well documented. Five triazole fungicides, tebuconazole, uniconazole, hexaconazole, peneconazole, and bitertanol showed anti-androgenic activity toward human AR both before and after metabolism mediated by human liver microsomes (Lv et al., 2017). Many insecticides and their metabolites have also been shown to inhibit androgen receptor dependent transcriptional activity. These include multiple isomers of DDT, DDE, and dichlorodiphenyldichloroethane (DDD) as well as methoxychlor and HPTE (Maness et al., 1998). Urea based herbicides such as linuron have also been identified as endocrine disruptors via AR antagonism (Lambright, 2000; Spirhanzlova et al., 2017). Agrochemicals have also been studied for their effects against AR as mixtures. Various combinations of androgen antagonists have been shown to disrupt male reproductive development in a cumulative and additive matter, regardless of their individual modes of action, and differences in tissue selectivity (Wilson et al., 2008).

Peroxisome Proliferator Activated Receptor

The peroxisome proliferator activated receptor (PPAR) is a ligand activated nuclear receptor that participates in energy combustion via stimulation of lipid catabolism and energy storage via stimulation of adipogenesis (Neschen et al., 2007). Lipophilic hormones, monounsaturated fatty acids, polyunsaturated fatty acids, and eicosanoids are all endogenous ligands of PPAR (Ayisi et al., 2018). Three subtypes of PPAR, PPARα, PPARβ, and PPARγ exist, with differing tissue distribution, ligand specificity, physiologic role, and mechanism of action. Once bound to their endogenous ligands, PPARs can induce the expression of genes and enzymes involved in lipid metabolism through both genomic and non-genomic mechanisms.

In a study of the PPARα and PPARγ activity of 200 pesticides using in vitro reporter gene assays, only three pesticides, doclofop-methyl, pyrethrins, and imazalil, exhibited PPAR agonist activity, which was further confirmed in vitro (Takeuchi et al., 2006). Due to its primary physiologic role as a regulator of adipogenesis and lipolysis, PPARγ is a major target of agricultural chemicals that act as obesogens. Tributyltin (TBT), an antifouling agent, alters PPAR-mediated differentiation of adipocytes and promotes adipogenesis in liver and adipose tissue (Maradonna and Carnevali, 2018). Exposure to dibutyltin (DBT), the major metabolite of TBT in the body, binds PPARγ to accelerate adipogenesis in both human and mouse mesenchymal stem cells. Interestingly, human cells were shown to be significantly more responsive to DBT than mouse cells (Chamorro-García et al., 2018). The fungicide triflumizole has also been shown to activate PPARγ to promote adipogenesis, acting as an obesogen in vivo (Li et al., 2012).

Aryl Hydrocarbon Receptor (AhR) Complex

The aryl hydrocarbon receptor complex includes the aryl hydrocarbon receptor (AhR) and the aryl hydrocarbon receptor nuclear translocator (ARNT). Although it is a basic region-helix/loop.helix (bHLH) protein and not a nuclear receptor, the AhR is a ligand dependent transcription factor with similar structure and function to nuclear receptors (Nebert, 2017). Following ligand binding, the AhR forms a heterodimer with ARNT that binds to DNA at xenobiotic response elements (Figure 3). For many years prior to the discovery of endogenous AhR ligands, the receptor’s most notable role was metabolism of exogenous, synthetic chemicals. The Ahr is activated by exogenous chemicals, and relays environmental signals to the cell (Elferinks et al., 1990). Thus, many mechanisms through which the AhR mediate toxic effects in the body are well studied. Much of what is known today regarding mechanisms of AhR mediated modulation of gene expression has been studied in the context of dioxin metabolism. Recently, endogenous ligands of AhR have been identified, whereas the receptor’s physiological roles in cell proliferation and differentiation, immune response, inflammation, and regulation of circadian rhythm are less understood (Nebert, 2017).

Figure 3: Mechanism of AhR signaling.

Synthetic chemicals such as dioxin (green triangles) activate AhR by binding to the receptor and translocating to the nucleus, where the complex dimerizes with the aryl hydrocarbon receptor nuclear translocator (ARNT) and controls transcription by interacting with xenobiotic response elements (XRE) in DNA. Transcriptional regulation by dioxin and other EDCs leads to an upregulation of xenobiotic metabolism to create toxic metabolites that impact the immune system, neurological function, reproduction, and carcinogenesis.

2,3,7,8-Tetrachlorodibenzodioxin (TCDD, also known as dioxin) is one of the most widely studied AhR ligands. The AhR has been shown to modulate the activity of other transcription factors including ESRs and ARs (Ohtake et al., 2011). In the presence of estrogens and androgens, the AhR-ARNT heterodimer downregulates transcriptional events that would otherwise be upregulated in the presence of these hormones. Dioxin bound AhR acts as a substrate specific adaptor component that targets estrogen and androgen bound receptors for degradation by the cullin 4B ubiquitin ligase complex (Ohtake et al., 2007). The opposite effect on transcription has been observed when AhR is activated in the absence of estrogen and androgen hormones. The AhR-ARNT complex physically associates with unliganded estrogen and androgen receptors bringing transcriptional co-activators to the promotors of each of these receptors (Ohtake et al., 2003). Hexachlorobenzene, a banned but persistent dioxin-like fungicide, also activates AhR to trigger both genomic and non-genomic effects (van Birgelen, 1998; Miret et al., 2019).

Reporter gene assays of 23 pesticides and insecticides in human and rat cell lines indicated AhR agonistic effects by iprodione, chlorpyrifos, and prochlorax, whereas five additional pesticides exhibited mixed activity in the two cell lines (Long et al., 2003). Another reporter gene assay study identified eleven pesticides that agonize AhR function, while two inhibited AhR activity (Ghisari et al., 2015). Pyrethroids have structural elements reminiscent of dioxins, suggesting potential AhR binding activity (Brander et al., 2016); cypermethrin has been shown to antagonize TCDD-induced AhR transactivation (Ghisari et al., 2015). The fungicide propiconazole also activates AhR (Knebel et al., 2018). This has been confirmed in silico, in vitro, and in vivo, wherein AhR was activated in a luciferase reporter gene assay in a human cell line and further increased mRNA and enzyme expression of genes controlled by AhR in the livers of rats treated with propiconazole via diet (Knebel et al., 2018).

Thyroid Receptors

Thyroid receptors (TR) are another class of ligand-dependent transcription factors. Studies show that four active isoforms of thyroid receptors (TRα1, TRβ1, TRβ2, and TRβ3) bind the endogenous thyroid hormones triiodothyronine (T3) and thyroxine (T4) (Zoeller, 2012). TRα and TRβ isoforms are differentially expressed in tissues throughout the body. Isoform distribution is particularly important during fetal brain development when thyroid hormone signaling is vital for normal development. Thyroid receptors are also important for physiological control of metabolism and the cardiovascular system (Zoeller, 2012).

Extensive evidence in humans and animals indicates that pesticides including chlorpyrifos, DDT, and methoxychlor can disrupt thyroid signaling and neurodevelopment (Boas et al., 2012). However, most evidence on pesticides implicates thyroid hormone synthesis, bioavailability, and metabolism for these effects (Ghassabian and Trasande, 2018). Pesticides generally do not have a high degree of structural similarity to thyroid hormones and thus have been less investigated as TR binders compared to environmental chemicals such as polychlorinated biphenyls (PCBs) and polybrominated diphenyl ethers (PBDEs) that have been shown bind directly to TR (Ghassabian and Trasande, 2018; Zoeller, 2007). However, in vitro reporter gene assays have indicated the ability of multiple pesticides to bind to TR. Ten pyrethroids or pyrethroid metabolites have demonstrated antagonistic activity against TR (Du et al., 2010). In an in vitro reporter assay analysis of the binding of 21 pesticides to human TR, 13 bound directly to TR, with 5 exhibiting agonistic activity, including procymidone, imidacloprid, mancozeb, and atrazine, whereas 11 were antagonistic and 4 exhibited both agonistic and antagonistic activity (Xiang et al., 2017).

Non-genomic Signaling

While genomic mechanisms of endocrine disruption involve activation of nuclear receptors that translocate to the nucleus to modulate gene expression, EDCs also exhibit non-genomic mechanisms of action. These actions are characterized as rapid effects that do not directly or initially influence gene expression (Figure 1C) (Lösel and Wehling, 2003). A non-genomic effect can be observed in a much shorter time frame (seconds to minutes of exposure to experimental chemical) than genomic effects that are constrained by transcription and translation. These effects are also typically unaffected by actinomycin D or cycloheximide treatment, which inhibit transcription and translation, respectively.

Many non-genomic actions are initiated at the level of the cell membrane through interactions with membrane-associated receptors. The fungicides prochloraz and vinclozolin have been shown to directly target the membrane androgen receptor ZIP9, a member of the solute carrier protein family, to block testosterone-driven zinc influx and apoptosis in PC3 prostate cancer cells (Thomas and Dong, 2019). Another study has demonstrated that the herbicide atrazine and pesticide trans-nonachlor target the epidermal growth factor receptor (EGFR) and block receptor activation and autophosphorylation (Hardesty et al., 2018). Further protein-ligand docking simulations suggest that trans-nonachlor acts as a competitive antagonist to the EGFR, whereas atrazine blocks the tyrosine kinase activity of the receptor (Hardesty et al., 2018). The insecticide cypermethrin can also inhibit EGFR activity and downstream MAPK activation by interfering with non-classical testosterone signaling in sertoli cells, leading to reduced cell viability and proliferation (Wang et al., 2019).

Activation of membrane receptors triggers signal transduction cascades to regulate cellular responses like changes in gene expression, proliferation, and metabolism. The mitogen activated protein kinase pathway (MAPK) is a kinase cascasde activated downstream of membrane receptors for growth factors and cytokines that regulate cell growth, survival, and differentiation (Morrison, 2012). Several studies have shown MAPK pathway activation to be critical in mediating the effects of agricultural chemicals like nonylphenol, atrazine, and cypermethrin on Sertoli cell viability, testosterone production in Leydig cells, and gonadotropin production in cultured gonadotropes, respectively (Choi et al., 2014; Pogrmic-Majkic et al., 2016; F. Li et al., 2018). One study observed reduced membrane associated connexin43 expression and impaired inter- and intracellular signaling in liver progenitor cells treated with either methyoxylchlor or vinclonzin (Babica et al., 2016). This effect was dependent on p38 MAPK activity as well as activation of protein kinase C.

The phosphoinositide-3-kinase–protein kinase B/AKT (PI3K/AKT) pathway is another target of agricultural chemicals. The PI3K/AKT pathway is often triggered upstream by activation of membrane receptors targeted by growth factors and cytokines, and plays a critical role in cell proliferation, survival, and metabolism (Hemmings and Restuccia, 2012). Current studies have demonstrated the suppression of this signaling pathway by nonylphenol and ziram in the testes of exposed rats, resulting in the generation of reactive oxygen species and increased apoptosis (Huang et al., 2016; Xie et al., 2018).

Another common target of agricultural chemicals is the cyclic adenosine 3′,5′-monophosphate (cAMP) pathway. The second messenger cAMP accumulates in the cell through activation of adenylyl cyclase by membrane associated G-protein coupled receptors (Sassone-Corsi, 2012). In turn, cAMP activates protein kinase A (PKA) to regulate cell metabolism, ion transport, and gene expression (Sassone-Corsi, 2012). Atrazine has been shown to increase cAMP levels in Leydig cells by inhibiting the cAMP specific phosphodiesterase, an enzyme that catalyzes the hydrolysis of cAMP to the inactive 5-AMP (Karmaus and Zacharewski, 2015). Excessive accumulation of cAMP by atrazine increases PKA activity to modulate testosterone production in Leydig cells (Samardzija et al., 2016; Pogrmic-Majkic et al., 2016; Karmaus and Zacharewski, 2015). Other chemicals, like methyoxychlor and its metabolite HTPE, have been shown to inhibit follicle stimulating hormone (FSH) stimulated cAMP production, resulting in reduced estrogen production (Harvey et al., 2015). Another study demonstrated that exposure to a mixture containing organochlorine insecticides impaired cAMP signaling, leading to decreased cAMP mediated post-translational processing of the steroidogenic acute regulatory protein to its active form and reduced steroidogenesis in cultured Leydig cells (Enangue Njembele et al., 2014).

EDCs can also influence cell function by targeting membrane associated ion-channels to mediate rapid non-genomic responses. These channels open to form a pore, allowing the passive diffusion of specific ions into the cell that will mediate numerous cellular functions, including the control of electrical excitability and the exocytosis of secreted proteins (Jentsch et al., 2004). Insecticides cypermethrin and DDE as well as glyphosate-based herbicides have been shown to target membrane bound voltage gated calcium channels and/or intracellular ryanodine receptors and increase intracellular calcium concentrations (Tavares et al., 2013; De Liz Oliveira Cavalli et al., 2013; Ye et al., 2017; F. Li et al., 2018). Perturbation in calcium homeostasis by agricultural chemicals has been linked to several physiological effects, including reduced acrosomal integrity and mobility in sperm, increased oxidative damage and decreased viability in Leydig cells, and increased glucose-stimulated insulin secretion from pancreatic β-cells through generation of reactive oxygen species (Tavares et al., 2013, 2015; De Liz Oliveira Cavalli et al., 2013; Chen et al., 2017). Another study demonstrated that cypermethrin induced calcium influx triggered rapid activation of MAPK signaling, leading to increased gonadotropin secretion from cultured pituitary cells (F. Li et al., 2018). Cypermethrin has also been shown to target voltage gated sodium channels in the hypothalamus to increase gonadotropin-releasing hormone pulse frequency (Ye et al., 2017).

The non-genomic effects of endocrine disrupting chemicals are not fully understood, and even less is known about those driven by agricultural EDCs. This is partly due to the duration of treatment. Many studies using agrochemicals involve long term treatment periods spanning days to weeks, which make it difficult to discern the rapid non-genomic actions from the slower genomic actions. Use of shorter exposure periods in in vitro culture systems and the inclusion of actinomycin D and cycloheximide will greatly increase our understanding of the non-genomic mechanisms of agricultural EDCs.

Hormone Synthesis and Degradation

Hormones in the body can be characterized into three main groups depending on their major components: peptide/protein hormones, steroid hormones, and amino acid analogue hormones. Peptide/protein hormones are the most abundant in the body and include gonadotropin-releasing hormone, growth hormone, FSH, luteinizing hormone (LH), and thyroid stimulating hormone (TSH). Important steroid hormones include estrogens, progesterone, and testosterone. Lastly, amino acid analogues include iodothyronines such as thyroxine (T4) and 3,5,3’-triiodothyronine (T3) and amines such as dopamine and serotonin. The synthesis and degradation of peptide/protein, steroid, and thyroid hormones can be negatively affected by EDCs, including agrochemicals. Interfering with hormone action may cause negative downstream effects on reproductive and non-reproductive health.

Peptide/protein hormones

Peptide and protein hormones are composed of short and long amino acid chains, respectively. Synthesis occurs in the nucleus and cytoplasm of secretory cells via gene transcription, translation into the peptide chain, and finally post-translational modifications. After hormones are synthesized and packaged into secretory granules, neural and hormonal signals cause their secretion into extracellular space (Malandrino and Smith, 2018). Some peptide and protein hormones are secreted in pulsatile patterns that may have rhythmic changes that contribute to feedback mechanisms, such as the hypothalamic-pituitarygonadotropin (HPG) axis, to control other hormone production in the body. Most peptide/protein hormones are soluble in aqueous solvents and do not need carrier proteins for transport throughout the blood stream. Therefore, they are susceptible to rapid protease degradation, leading to a short half-life and duration of action (Malandrino and Smith, 2018). TSH, a glycoprotein hormone that stimulates the production of T4, is susceptible to EDC influence. Perinatal exposure in rats to a glyphosate-based herbicide resulted in decreased TSH levels, decreased gene expression of deiodinases and transporters, and altered other genes regulated by thyroid hormones or involved in thyroid hormone metabolism in male offspring (de Souza et al., 2017). Epidemiological studies of agricultural workers have shown associations between pesticide exposure and levels of FSH and LH (Recio et al., 2005; Cremonese et al., 2017). Extensive evidence in animal studies also indicates that chlorpyrifos treatment alters levels of FSH and LH, in addition to steroid hormones (Li et al., 2019).

Steroid Hormones

Most steroid hormone synthesis takes place in the adrenal glands, ovary, testes, placenta, and adipose tissue. The synthesis of steroid hormones is known as steroidogenesis. Cholesterol is the pre-cursor for all steroid hormones, and it is transferred across the mitochondrial membrane by steroidogenic acute regulatory protein (StAR). Cholesterol is then converted to pregnenolone in a process that is common to all steroidogenic pathways (Miller and Auchus, 2011). After the production of pregnenolone, the pathway varies depending on the hormone being produced; pregnenolone may undergo various biotransformations to form aldosterone and cortisol in the adrenal gland or progesterone, estradiol, and testosterone in the gonads (Acconcia and Marino, 2018; Auchus, 2014) (Figure 4). Most steroids are hydrophobic in nature so hormones and their precursors can leave steroidogenic cells easily and are not stored (Acconcia and Marino, 2018). After a biological response occurs, hormone secretion ceases which contributes to feedback loops in the body, including the HPG axis and the hypothalamic-pituitary-adrenal (HPA) axis (Acconcia and Marino, 2018). Carrier proteins transport steroid hormones throughout the body and control the proportions of bound and unbound hormones (Leung and Farwell, 2018). Steroid hormones are eliminated through enzymatic reactions followed by transport across the cell membrane. Many of these reactions occur in the liver and include hydroxylation, conjugation, and reduction-oxidation (You, 2004). After metabolites reach circulation, they are readily excreted in both urine and bile as unconjugated or conjugated metabolites, but some may remain in circulation by binding to serum proteins (You, 2004).

Figure 4: Steroidogenesis.

Cholesterol is imported by steroidogenic acute regulatory protein (StAR) and transformed into pregnenolone, which is further converted to adrenal and gonadal hormones.

Agrochemical exposure causes detrimental effects on the process of steroidogenesis that occurs in the testis, ovary, and adrenal gland. These effects have been observed in both in vivo and in vitro experimental designs. Alteration of steroidogenesis is a widely studied mode of action of EDCs; thus, Table 2 provides examples of recent studies on the impacts of agrochemicals on steroid hormones. Considering the literature, some common targets of agrochemical EDCs emerge. Multiple chemicals have been shown to affect the synthesis of cholesterol, which is the pre-cursor for all steroid hormones. In the adrenal glands, agrochemicals negatively affect the hormones, genes, and proteins involved in the synthesis of steroid hormones. In the ovary, agrochemicals disrupt the production of estradiol by altering the levels of enzymes and pre-cursor hormones involved in steroidogenesis in both in vivo and in vitro studies. In the testes, agrochemicals impair the synthesis of testosterone, leading to detrimental effects on reproduction.

Table 2:

Agrochemical effects on steroid hormone synthesis, metabolism, and concentrations

Chemical	Mode of Action	Effect	Reference
2,4-D	Decreased serum testosterone levels, testis testosterone levels, Cyp17a1 levels, and total cholesterol levels in Leydig cells	Decreased pregnancy rate and number of pups	(Harada et al., 2016)
	Decreased serum testosterone levels	Decreased luminal spermatozoa	(Zhang et al., 2017)
atrazine	Decreased steroidogenic enzymes in Leydig cells, decreased serum levels of dihydrotestosterone and testosterone, decreased transport of cholesterol to mitochondria	Decreased testis, seminal vesicle, and prostate weight	(Pogrmic et al., 2009)
	Decreased testosterone levels and increased estradiol levels, decreased protein levels of 3β-hydroxysteroid dehydrogenase (3β-HSD) expression	Altered dilation of seminiferous tubules and decreased size of Leydig cells	(Victor-Costa et al., 2010)
	Increased estradiol and progesterone levels in granulosa cells, increased estradiol, estrone, and progesterone levels in adrenal cells	Altered steroidogenesis and aromatase activity	(Tinfo et al., 2011)
	Altered levels of steroidogenic enzymes	Decreased percent cell viability in Leydig cells	(Abarikwu et al., 2011)
	Decreased levels of estradiol, but increased levels of progesterone in granulosa cells	Impaired reproductive efficiency	(Basini et al., 2012)
	Altered levels of steroidogenic enzymes and increased levels of progesterone and testosterone in Leydig cells	Altered steroidogenesis	(Forgacs et al., 2013)
	Increased levels of estrogen and androgens in adrenal cells	Interference with androgen synthesis	(Háhn et al., 2016)
	Reduced levels of 3β-HSD expression	Decreased testis weight and heterogeneous testis morphology	(Martins-Santos et al., 2017)
	Reduced levels of steroidogenic enzymes	Decreased female fertility	(Pogrmic-Majkic et al., 2018)
cypermethrin	Altered levels of steroidogenic enzymes, increased levels of estrogen, cortisol, and aldosterone in adrenal cells	Altered the steroidogenic pathway	(Ji et al., 2019)
	Increased levels of steroidogenic enzymes, increased levels of luteinizing hormone, follicle-stimulating hormone, and testosterone	Accelerated onset of puberty in males	(Ye et al., 2017)
	Altered levels of steroidogenic enzymes, decreased levels of testosterone, increased levels of estradiol in vivo and in vitro	Altered testis weight over time, decreased germ cells, impaired spermatogenesis and steroidogenesis	(Huang and Li, 2014)
endosulfan	Decreased steroidogenic enzyme and testosterone levels	Decreased testis weight, sperm count, sperm motility, sperm viability, sperm production, and inhibited spermatogenesis	(Aly and Khafagy, 014)
fenvalerate	Decreased expression of steroidogenic enzymes and decreased serum and testicular testosterone	Impaired spermatogenesis and decreased sperm count	(Zhang et al., 2010)
glyphosate and glyphosate-based herbicides	Decreased levels of corticosterone, testosterone, cholesterol, but increased weight of adrenal glands	Disrupted function of HPA axis	(Pandey and Rudraiah, 2015)
	Altered levels of steroidogenic enzymes, decreased progesterone levels, but increased estradiol levels	Decreased ovary weight, altered ovarian histology, and sex ratio of pups	(Ren et al., 2018)
lambda-cyhalothrin	Increased levels of cholesterol but decreased levels of steroidogenic enzymes in the ovary, decreased levels of cholesterol, but increased levels of steroidogenic enzymes in the adrenal glands, and deceased serum levels of LH, FSH, estradiol, and progesterone	Altered ovarian structural degenerations and follicular maturation	(R. Ghosh et al., 2018)
	Decreased levels of steroidogenic enzymes and levels of testosterone, FSH, and LH	Delayed Leydig cell development	(H. Li et al., 2018)
mancozeb	Decreased levels of steroidogenic enzymes and levels of testosterone	Decreased testis, epididymis, seminal vesicle, vas deferens, and prostate weight, decreased sperm parameters	(Girish and Reddy, 2018)
	Decreased testosterone and FSH levels	Increased abnormal sperm morphology and sperm viability	(Elsharkawy et al., 2019)
methoxychlor	Altered levels of steroidogenic enzymes, decreased testosterone levels, increased estradiol levels	Decreased testis weight and disrupted spermatogenesis	(Du et al., 014)
	Decreased levels of steroidogenic enzymes and testosterone	Decreased testis weight, cauda sperm count, and sperm motility	(Aly and Azhar, 2013)
	Decreased levels of most steroidogenic enzymes and decreased levels of estradiol, testosterone, androstenedione, and progesterone	Decreased antral follicle growth	(Basavarajappa et al., 2012)
ziram	Decreased expression of steroidogenic enzymes, levels of testosterone, and levels of FSH	Decreased epididymis and testis weight	(Guo et al., 2017)
	Decreased steroidogenic enzyme, testosterone, and FSH levels	Disrupted Leydig cell development	(Xie et al., 2018)

Amino Acid Hormones

Amino acid analogue hormones develop from amino acids. Specifically, the amines are derived from tyrosine and are secreted from both the thyroid and adrenal medulla (Koibuchi, 2018). The thyroid gland synthesizes the endogenous thyroid hormones T4 and T3, which require dietary iodine. Thyroid hormones are produced extracellularly in the follicular lumen (Koibuchi, 2018). The lumen is filled with the glycoprotein thyroglobulin, which is synthesized within the epithelial cells and secreted into the lumen (Koibuchi, 2018). Iodine is oxidized by the enzyme thyroperoxidase and then binds to tyrosine residues in the thyroglobulin, resulting in the formation of iodotyrosines (Koibuchi, 2018; Kopp et al., 2008). The iodotyrosines are coupled by thyroperoxidase to form T4, which is further converted into T3 in target tissues by deiodinases. Thyroid hormones are transported via carrier proteins including thyroid binding globulin and transthyretin. The levels of T4 and T3 are regulated by deiodinases in peripheral tissues and metabolism in the liver (Boas et al., 2012).

Although pesticides are generally poor structural mimics for thyroid hormones, they have been shown to exert significant effects on thyroid hormone synthesis, concentrations, and carrier proteins. Nonylphenol, a now-banned surfactant previously used in pesticide formulations, inhibits thioperoxides necessary for the synthesis of T4. (Schmutzler et al., 2004) The triazole herbicide amitrole has been shown to increase the transcription of thyroglobulin gene, while decreasing the uptake of iodide (Pan Hongmei et al., 2011). Further, DDT has been shown to down-regulate the iodine-accumulation function of follicular thyrocytes by suppressing sodium/iodide symporter synthesis and disrupting regular thyroid function in male rats (Yaglova and Yaglov, 2015). Multiple human epidemiological studies have indicated an association between disrupted thyroid hormone levels and pesticide use (Cremonese et al., 2017; Recio et al., 2005; Piccoli et al., 2016; Blanco-Muñoz et al., 2016; Lacasaña et al., 2010). Malathion, nonylphenol, ioxynil, and pentachlorophenol exposure decreased T3 binding to Japanese quail transthyretin (TTR), one of the carrier proteins, with no observed interaction with the ligand binding domain of thyroid receptors (Ishihara, Nishiyama, et al., 2003). Dicofol was shown to alter T3 binding to bullfrog TTR in a biphasic manner, with low doses of dicofol increasing binding and high doses inhibiting binding (Ishihara, Sawatsubashi, et al., 2003). Recently, enhanced binding of T4 to human TTR has been demonstrated in the presence of organophosphate triesters, which are typically used as flame retardants, but have similar structures to organophosphate pesticides (Hill et al., 2018).

Epigenetics

Epigenetics encompasses changes in gene expression without changing DNA sequences. The epigenome, which consists of all chemical and structural marks that control the accessibly of the genome, is highly sensitive to environmental conditions including nutrition and stress as well as chemical exposures. Changes to the epigenome can be heritable and can manifest in health impacts and disease long after the exposure has ended, even multiple generations after the exposure occurred (Rattan and Flaws, 2019). Observed transgenerational effects of environmental chemicals were one of the first indications that EDCs could alter the epigenome. For example, vinclozolin and methoxylchlor exposure during pregnancy have been shown to decrease fertility in male descendants for up to four generations (Anway et al., 2005). The effects of these chemicals on germ cell genomes make these outcomes heritable. Development is also a particularly sensitive window of exposure due to the extensive genetic programming that occurs in this time period. Exposures during periods of development can alter the epigenome to increase susceptibility to disease later in life, after the exposure has ended. This forms the mechanistic basis for the developmental origins of health and disease (DOHaD) (Heindel and Vandenberg, 2015).

In the past 15 years, mechanistic understanding of how EDCs can alter the epigenome has significantly improved and three mechanisms of epigenetic modification have been identified in recent EDC literature. DNA methylation and histone modification involve structural modifications to chromatin that control the accessibility of genes to transcription factors. Non-coding RNAs, particularly microRNAs (miRNAs), from the non-translated genome also regulate the epigenome and can be altered by EDCs. These mechanisms are illustrated in Figure 5.

Figure 5: Mechanisms of epigenetic modifications.

Chromosomes are composed of chromatin that is wrapped around proteins called histone. Modifications to histone tails and direct DNA methylation can control transcriptional access to DNA without modifying DNA sequence. Non-coding RNAs from the untranslated genome also regulate transcription.

DNA Methylation

Modification of DNA with methyl groups is the most well-studied epigenetic modification. Methylation of chromatin occurs at the 5-carbon position of cytosine to form 5-methylcytosine (5mC). The cytosine is typically adjacent to guanine in the 5′ → 3′ direction (CpG). CpG sites are less represented in the genome than expected by chance and appear in clusters known as CpG island in promoter regions of genes. Methylation is generally a silencing mark, which acts by blocking access of transcription factors to genes. DNA methyltransferases (DNMTs) are responsible for methylation and demethylation of CpG sites. DNMT1 performs maintenance of methyl marks during replication to preserve epigenetic marks through mitosis, whereas DNMT3a and DNMT3b perform de novo methylation.

Extensive reprogramming of methylation marks occurs during germ cell development and early embryonic development (Reik et al., 2001). In both cases, methylation marks are removed and remethylated, providing opportunities for disruption by environmental chemicals. Germ cell modifications are heritable, which can lead to the preservation of altered methylation marks in offspring and later generations, leading to transgenerational effects. For example, methoxychlor increased the incidence of kidney disease, ovarian disease, and obesity as well as decreased sperm concentrations in F3 mice following prenatal exposure by altering methylation patterns (Manikkam et al., 2014; Stouder and Paoloni-Giacobino, 2011). Gestational exposure of the F1 generation to the herbicide atrazine led to increased testis disease, lean phenotype, early puberty in females, and behavioral changes in the F3 generation (McBirney et al., 2017). Differently methylated regions in sperm were identified for each differently-exposed generation (F1 as embryo; F2 as germ cells; F3 no direct exposure), with consistently differently methylated regions in the F3 generation corresponding to observation of lean phenotype (McBirney et al., 2017). In sperm from F3 rats descended from F0 rats exposed gestationally to vinclozolin, differentially methylated regions were associated with testis disease, prostate disease, and kidney disease (Nilsson et al., 2018). In another transgenerational study of gestationally exposed rats, vinclozolin induced differential methylation in sperm and brain in F1 and F3 generations (Gillette et al., 2018). Adult exposure to methyl-parathion has been shown to decrease sperm quality in mice immediately following exposure due to increased methylation at promoter regions of DNA repair genes as well as global hypomethylation (Hernandez-Cortes et al., 2018). Although the heritability of these observed changes in male germ cells was not investigated, effects due to adult exposures can be inherited through the germ line. One study has examined methylation in sperm from humans exposed to TCDD, finding differentially methylated regions associated with peripubertal serum TCDD measurements (Pilsner et al., 2018).

Alternations in reprogramming during embryonic development can cause immediate toxicity or increase adult diseases. The insecticide fipronil is composed of an enantiomeric mixture, one of which is significantly more developmentally toxic to zebrafish. The toxic S-enantiomer hyper-methylated CpG regions during development, leading to disruption of signaling pathways including MAPK, Wnt, and hedgehog (Qian et al., 2017). Methoxychlor induced adult ovarian dysfunction in neonatally exposed rats. Methylation analysis showed hypermethylation in ERβ promoter regions and increased Dnmt3b expression (Zama and Uzumcu, 2009).

Histone Modification

DNA is organized by winding around proteins called histones to form chromatin that is compact enough to fit in the nucleus of a cell (Strahl and Allis, 2000). DNA must unwind from the histones to be accessible to transcription factors and thus gene expression is regulated by the structure of the histone. Like methylation of DNA, histones can be modified with chemical tags that limit the ability of DNA to be transcribed. Methylation and acetylation are two of the most common forms of post-transcriptional modifications on histone tails that can limit the accessibility of genes.

Atrazine treatment of mice leads to heritable transgenerational effects on sperm mediated through histone modifications. In male offspring following developmental exposure, trimethylation of the fourth lysine on histone 3 (H3K4me3) was altered in promoters of key pluripotency-associated genes in sperm (Hao et al., 2016). Furthermore, these marks were preserved in the F3 generation (Hao et al., 2016). These marks were compared to a previous study on vinclozolin exposed mice, showing overlap in differently methylated regions indicative of regions of the genome sensitive to epigenetic modifications by environmental chemicals (Guerrero-Bosagna et al., 2012). Another study on atrazine exposure in adult male mice found disruption of meiosis in sperm with alterations in H3K4me3 marks as well (Gely-Pernot et al., 2015).

Recent studies have examined changes in histone modification and other epigenetic modification as first steps in elucidating the mechanisms of action of less widely studied agrochemicals. The fungicide carbendazim has been shown to alter spermatogenesis and steroidogenesis in mouse testes. Epigenetic analysis has shown alterations in trimethylation at lysine 27 on H3 (H3K27me3) as well as changes in DNA methylation (Liu et al., 2019). The organophosphate pesticide chlorpyrifos has been shown to increase expression of histone deacetylase 1 (HDAC1) in mammary tissue of rats (Ventura et al., 2019). The estrogenic pesticide endosulfan has also been shown to increase histone deacetylase (HDAC) and DNMT expression in MCF-7 cells (K. Ghosh et al., 2018). Although each of these studies shows that EDCs can act through epigenetic pathways, the detailed mechanisms through which these chemicals alter expression of epigenetic regulatory proteins are largely unknown.

Non-Coding RNA Expression

Non-coding RNAs are nucleic acids transcribed from sections of DNA that do not code for proteins, once known as “junk” DNA. Non-coding RNAs, including small microRNAs (miRNAs), are important regulators of post-transcriptional gene expression by binding to untranslated regions of mRNAs to block protein translation. MiRNAs can target enzymes responsible for epigenetic marks such as DNMTs and HDACs. In addition, epigenetic marks control the expression of miRNAs (Yao et al., 2019). MiRNAs have been implicated in various human diseases, especially cancers (Tüfekci et al., 2014), but the role of miRNAs and other non-coding RNAs is less widely studied in relation to the endocrine system. However, several miRNAs that directly target ESR1 have been identified (Bhat-Nakshatri et al., 2009).

Developmental exposure to atrazine in zebrafish significantly altered both human and zebrafish miRNA expression. Further analysis revealed that the altered miRNAs are associated with epigenetic regulation of carcinogenesis, the cell cycle, and cell signaling (Wirbisky et al., 2016). In MCF-7 breast cancer cells, DDT induced a distinct pattern of miRNA expression compared to vehicle control, estradiol, or bisphenol A (BPA) (Tilghman et al., 2012). These preliminary studies show that EDCs can alter miRNA expression to epigenetically regulate biological processes, but further investigations are necessary for deeper understanding of the mechanism of action of EDCs on these miRNAs and identify miRNA targets.

Conclusions

As this review illustrates, environmental chemicals can act as EDCs through a variety of mechanisms. The diversity of pathways and precision of biological hormone actions in the endocrine system makes it particularly susceptible to disruption by exogenous agents. In addition, the wide range of possible phenotypes and endpoints makes integration of studies on EDCs to understand mechanisms a difficult task. However, the requirement in the European Union of evidence of a plausible mode of action for EDC regulation underscores the importance of mechanistic studies and analyses of these compounds (Solecki et al., 2017).

Agrochemicals represent a significant class of EDCs that humans and animals around the world are exposed to constantly. With such diversity of structures and uses, examples of agrochemicals that exhibit each of the major mechanisms of endocrine disruption have been identified. Much like other environmental contaminants, legacy pesticides continue to present a threat to human health, while newer replacement chemicals are less well understood in terms of their endocrine disrupting potential. Furthermore, newer chemicals with shorter half-lives and different routes of exposure pose challenges to understanding their mechanisms of action, much like other classes of replacement EDCs.

Given the diversity of mechanisms through which natural hormones and EDCs act, it is vital that 21^st century toxicology incorporates the principles of endocrinology into assessment of the safety of chemicals in our environment (Schug et al., 2016; Vandenberg et al., 2019, 2013, 2012). Future studies should recognize the prevalence of non-monotonic dose response curves and the importance of low dose studies. In addition, mechanistic studies are needed on newer chemicals on the market and suspected EDCs; legacy chemicals and controversial EDCs with lots of public interest have received most of the scientific attention to date. Improvements in assays and techniques to elucidate EDC mechanisms of action for computational, in vitro, and whole animal studies will facilitate interdisciplinary cooperation to identify additional unstudied mechanisms of action and help regulators and the public understand the various modes of action through which EDCs operate.