Exposure to bisphenol A differentially impacts neurodevelopment and behavior in Drosophila melanogaster from distinct genetic backgrounds

Abstract

Bisphenol A (BPA) is a ubiquitous environmental chemical that has been linked to behavioral differences in children and shown to impact critical neurodevelopmental processes in animal models. Though data is emerging, we still have an incomplete picture of how BPA disrupts neurodevelopment; in particular, how its impacts may vary across different genetic backgrounds. Given the genetic tractability of Drosophila melanogaster, they present a valuable model to address this question. Fruit flies are increasingly being used for assessment of neurotoxicants because of their relatively simple brain structure and variety of measurable behaviors. Here we investigated the neurodevelopmental impacts of BPA across two genetic strains of Drosophila—w¹¹¹⁸ (control) and the Fragile X Syndrome (FXS) model—by examining both behavioral and neuronal phenotypes. We show that BPA induces hyperactivity in larvae, increases repetitive grooming behavior in adults, reduces courtship behavior, impairs axon guidance in the mushroom body, and disrupts neural stem cell development in the w¹¹¹⁸ genetic strain. Remarkably, for every behavioral and neuronal phenotype examined, the impact of BPA in FXS flies was either insignificant or contrasted with the phenotypes observed in the w¹¹¹⁸ strain. This data indicates that the neurodevelopmental impacts of BPA can vary widely depending on genetic background and suggests BPA may elicit a gene-environment interaction with Drosophila fragile X mental retardation 1 (dFmr1)—the ortholog of human FMR1, which causes Fragile X Syndrome and is associated with autism spectrum disorder.

Graphical Abstract

1. INTRODUCTION

1. 1. Neurodevelopmental disorders

More than one in six children in the United States (U.S.) are diagnosed with a neurodevelopmental disorder (NDD) [1], which refers to a heterogeneous group of nervous system disorders with complex etiologies. The prevalence of NDDs has increased significantly over the past several decades—perhaps most notably, the incidence of autism spectrum disorder (ASD) has increased by almost 300 percent in the last 20 years [2, 3]. Many NDDs can result from both genetic susceptibilities and environmental factors [4, 5]. While genomic and genetic studies have identified thousands of genetic variants that either cause or confer risk of NDDs, we know relatively little about environmental risk factors. Identifying and characterizing environmental factors that confer risk of NDDs is crucial for enacting evidence-based preventative measures. We further contend that because both genetic and environmental susceptibilities are implicated—and because genetic risk factors can impact the physiological response to environmental factors—examining their combined effect is essential to advance our understanding of NDD etiology.

1. 2. Bisphenol A

One environmental chemical taking shape as a potential risk factor for NDDs is bisphenol A (BPA), a ubiquitous chemical used in the synthesis of polycarbonate plastic and epoxy resins. Structurally similar to estradiol, BPA can both agonize various estrogen receptor (ER) subtypes and antagonize androgen receptors [6, 7]. The ability of BPA to influence hormonal signaling was first documented in the 1930s [8, 9]. Despite understanding its endocrine disrupting capability, BPA was adapted for use in the synthesis of polycarbonate plastics in the 1950s and quickly became one of the most prevalent synthetic compounds in the world [10, 11]. An estimated 7.7 million metric tons of BPA was generated worldwide in 2015, and production is expected to rise to 10.6 million metric tons by 2022 [11]. BPA is found in a wide array of products used in everyday life, including plastic bags, plastic containers, thermal papers, food cans, and beverage cans [12–14]. Residual BPA can leach from these products due to incomplete polymerization during production or from depolymerization when exposed to high temperatures or extreme pH conditions, which speed up the hydrolysis of ester bonds that link BPA monomers [11]. Due to concerns surrounding the endocrine disrupting capabilities of BPA, the U.S. Food and Drug Administration eventually banned its use in all childhood products beginning in 2012. But BPA remains pervasive in our environments; of greatest concern, pregnant women are still persistently exposed to BPA in a variety of ways.

The lipophilic structure of BPA allows it to readily cross cell membranes, as well as both placental and blood-brain barriers, enabling its ability to potentially affect the neurodevelopmental program of a growing embryo or fetus [15–17]. Studies have linked prenatal and early childhood BPA exposure to behavioral problems in children. For example, positive correlations were found between BPA and ASD diagnoses by quantifying BPA in household products and in the urine of pregnant women exposed to those products, then tracking the developmental outcomes of their children [18]. Studies using model organisms ranging from fruit flies to rodents have shown that developmental exposure to BPA impacts both neuronal and behavioral phenotypes of offspring [19–21]. When exposed to BPA during embryonic and larval development, adult Drosophila exhibit increased locomotor activity [20], a phenotype thought to be the product of altered synaptogenesis. Hyperactivity has also been observed in mice and rats following pre- and perinatal BPA exposure, respectively [22, 23]. In addition, BPA has been shown to decrease the proliferative capacity of neural stem cells (NSCs) in rats and mice [24, 25]. A separate study using zebrafish found BPA increased, rather than decreased, NSC proliferation [21]. Such contradictory results may stem from the non-monotonic nature of BPA, a property shared by many endocrine disrupting chemicals (EDCs) [26, 27]—depending on the dose administered and duration of exposure, a higher dose of BPA can elicit a milder phenotype than a lower dose. BPA may also cause distinct responses depending on the developmental time point of exposure. Another explanation for such disparate phenotypes could be the variable genetic backgrounds and differences in the molecular networks that govern neurodevelopment within the organisms examined across different studies.

1. 3. Fragile X mental retardation 1

We investigated the neurodevelopmental impacts of BPA exposure in w¹¹¹⁸ Drosophila (used as a control) and in the Fragile X syndrome (FXS) Drosophila model. In humans, FXS occurs when Fragile X Mental Retardation 1 (FMR1) is transcriptionally silenced [28]. Transcriptional repression of FMR1 occurs when a CGG trinucleotide repeat within the 5’-untranslated region (UTR) expands beyond 200 repeats, therein leading to hypermethylation of the promoter region [29–31]. FXS is regarded as the most common monogenic form of intellectual disability and ASD—approximately 25–52% of individuals with FXS are also diagnosed with ASD [32]. FMR1 encodes Fragile X Mental Retardation Protein (FMRP), an RNA-binding protein involved in post-transcriptional and translational regulation of RNA transcripts, most prominently within gonadal and neuronal tissues [33]. In the developing brain, FMRP regulates the expression of proteins and non-coding RNAs that participate in processes ranging from NSC proliferation, axon outgrowth and guidance, synaptogenesis, and synaptic plasticity [34–37]. The Drosophila ortholog of FMR1 (dFmr1) exhibits many functionally conserved neurodevelopmental roles, as elucidated through extensive analysis of the Drosophila FXS model [38–41]. The FXS strains used in this study—dFmr1³ and dFmr1^Delta113—are homozygous for dFmr1 null alleles, created by imprecise P-element excisions that removed large segments of the 5’UTR and open reading frame [41, 42]. Similar to vertebrate model organisms, dFmr1 mutant Drosophila exhibit a broad range of neuronal and behavioral phenotypes, including increased NSC proliferation [38], aberrant axon guidance [40], increased synapse formation [39], reduced courtship activity [43], excessive grooming [44], and increased locomotor activity [45, 46]. Expression of human FMR1 in FXS Drosophila can rescue neuronal defects [47], demonstrating the functional conservation of FMRP across species. We sought to determine how developmental exposure to BPA would impact behavioral and neuronal phenotypes already associated with loss of dFmr1. The w¹¹¹⁸ Drosophila strain was used as a control because the dFmr1 mutant lines were created in this genetic background. It is important to note that while the control w¹¹¹⁸ flies are wild-type at the dFmr1 locus, this strain carries a mutation in the white (w) gene; thus, w¹¹¹⁸ are not technically a wild-type strain. However, it was critical for the control strain to be w⁻ because dFmr1 mutant flies are w⁻.

1. 4. Drosophila as a model for developmental neurotoxicology

Although the FXS Drosophila model has been used for pharmacological analysis [48], it has not yet been taken advantage of for neurotoxicological assessment of environmental chemicals. Drosophila are increasingly being valued as a model for both NDDs and toxicology. Among Drosophila’s 14,000 protein-coding genes are orthologs to an estimated 61% of identified disease-related genes in humans, including orthologs to many NDD-associated genes like FMR1 [49–51]. Drosophila have advantageous life history traits that have been exploited to study neurodevelopment, neurodegeneration, and neurotoxicology [52–54]. Some of the measurable behaviors that can be used for such studies include courtship [43, 48], locomotor activity [45], repetitive grooming [44], circadian behavior [55], aggression [56], learning and memory [57], and open field exploration [58]. Given the relative simplicity and low cost of these experimental paradigms, Drosophila behavioral analyses are a pragmatic choice for initial assessment of xenobiotics to determine their potential impact on neurodevelopment. These studies can be followed by analysis of NDD-associated neuronal phenotypes through examination of the simple neural structures of fruit flies.

Our study indicates that BPA impacts both behavioral and neuronal phenotypes in Drosophila, but that the impact varies dramatically according to genetic background. In alignment with published analyses of BPA in Drosophila and other organisms, we predicted exposure would cause neurodevelopmental phenotypes in w¹¹¹⁸ flies. We also hypothesized that BPA would increase the severity of dFmr1 mutant phenotypes in an additive or synergistic manner. Instead, for almost every phenotype examined—larval locomotion, repetitive grooming, courtship behavior, axon guidance, and neural stem cell development—BPA exposure caused a significant response in w¹¹¹⁸ flies, but either elicited an insignificant or contrary response in dFmr1mutant flies.

2. MATERIALS AND METHODS

2. 1. Fly stocks and husbandry

Standard cornmeal-based fly food was used to rear all fly populations. Flies were cultured in K-resin plastic vials (a non-polycarbonate plastic) and housed in humidified incubators to maintain 25°C temperature and a 12-hour light/dark cycle. We used two dFmr1 null mutant strains—dFmr1³ and dFmr1^Delta113. Both include null alleles created via imprecise excision of a transposable element within the 5’ UTR of the dFmr1 locus from the w¹¹¹⁸ strain. The dFmr1³ line was a generous gift from Dr. Thomas Jongens (University of Pennsylvania, Philadelphia, Pennsylvania) [41]. The dFmr1^Delta113 mutant strain was obtained from the Bloomington Drosophila Stock Center (stock number 67403) [42].

2. 2. BPA exposure

Drosophila were exposed to BPA via oral administration through food. Two concentrations of BPA, 0.1mM and 1mM, and a negative control (water only) were used for all experiments with the exception of the axon guidance assay, which also included a 2mM exposure group. BPA was dissolved in water that was subsequently used to cook standard fly food. Freshly eclosed virgin female flies were collected and transferred to BPA treatment vials for a period of four days before introducing unexposed male flies. Exposing the parental (P1) generation females ensured the oocytes—and therefore embryos—were exposed to BPA. The P1 generation adults were removed after 7 days and their first filial (F1) larvae remained in the treatment vials to ensure BPA exposure during larval development.

2. 3. Courtship

The naïve courtship assay was modified from an established protocol [59]. F1 virgin males were collected and kept in separate food chambers on a 12-hour light/dark cycle for 5–7 days before introduction to a standard courtship chamber. Unexposed w¹¹¹⁸ virgin female flies aged 5–7 days were used in all courtship experiments. After introduction, flies were recorded for 10 minutes. Individuals blinded to genotype and exposure scored courtship behaviors—following/orientation, leg tapping, wing extension, licking, copulation attempts, and successful copulation. The courtship index (CI) was calculated as the total amount of time spent participating in courtship behaviors divided by the duration of the assay.

2.4. Larval locomotion

The larval locomotion assays were modified from two established protocols [45, 60]. Individual age-matched late third-instar larvae were transferred to a 15-cm petri dish containing a 2% agarose gel. After a one-minute acclimation period, the number of peristaltic contractions and orientation changes were recorded for one minute. Individuals blinded to genotype and exposure conditions scored the locomotor behaviors.

2. 5. Grooming

The grooming assay was adapted from an established protocol [61]. Male flies were collected and aged for five days in separate food chambers before being aspirated into an observation chamber. Flies were allowed to acclimate for one minute and were then recorded for five minutes. Individuals blinded to genotype and exposure analyzed the videos by recording the percentage of time flies spent grooming.

2. 6. Antibodies

Antibodies for neuroblast (NB) and intermediate neural progenitor (INP) detection were used at the following dilutions in 0.3% PBST (PBS + 10% Triton-X 100): Rabbit anti-phospho-Histone H3 (pH3; Cell Signaling) at 1:500, rat anti-Deadpan (Abcam) at 1:75, goat anti-rabbit Cy3 (Jackson Immunoresearch) at 1:300, and goat anti-rat Alexa 488 (Jackson ImmunoResearch) at 1:250. Anti-Deadpan marks all NBs and INPs and was used to define and exclude the optic lobe during analysis. Anti-pH3 marks all proliferating NBs and progenitor cells in M phase. Antibodies used for the detection of fasciculating axons in the mushroom body were used at the following dilutions in 0.1% PBST: Mouse anti-FasciclinII (Developmental Studies Hybridoma Bank) at 1:20 and goat anti-mouse Alexa 488 (Jackson ImmunoResearch) at 1:1000.

2. 7. Immunohistochemistry and confocal microscopy

For analysis of axon guidance in the mushroom body:

Using a previously established protocol [40], adult brains from 0–4 day old male flies were dissected in a 1% NGS PBS solution, and then fixed in 4% PFA for 25 minutes at RT. Samples were washed 3 × 15-minute at RT in 0.1% PBST, then blocked in 1% NGS, 0.1% BSA in PBST for 1 hour at RT. Samples were incubated with anti-FasII at 4°C for 16–24 hours. Samples were washed 3 × 20 minutes in 0.1% PBST at RT, then incubated with goat anti-mouse Alexa 488 at 4°C for 16–24 hours. Samples were then washed 6 × 20 minutes in 0.1% PBST at RT and mounted in ProLong Gold antifade mountant (Invitrogen). The degree of midline crossing was qualitatively assessed as described in Michel et al., 2004 using four categories: no crossing, mild crossing, moderate crossing, and severe crossing [40]. The individual that scored crossing severity was calibrated with another scorer and blinded to genotype and exposure.

For quantification of Dpn-positive or pH3-positive neuroblasts and progenitors:

Using a previously established protocol [38], age-matched late third-instar larvae were dissected in a 1% Normal Goat Serum (NGS) PBS solution and fixed in 4% Formaldehyde in PBS for 20 minutes at room temperature (RT), washed 3 × 20-minutes in 0.1% PBST (PBS + 10% Triton-X 100) at RT, then blocked in 1% NGS, 0.1% BSA in PBST for 1 hour at RT. Incubation in primary antibodies (anti-Dpn or anti-pH3) was performed at 4°C for 16–24 hours. A 5-minute wash followed by two 30-minute washes in 0.1% PBST were performed at RT, followed by secondary antibody (goat anti-rat Alexa 488 or goat anti-rabbit Cy3) incubation at 4°C for 16–24 hours. Samples were washed 3 × 30 minutes in 0.1% PBST and mounted in ProLong Gold antifade mountant (Invitrogen). Images which were quantified using IMARIS Bitplane software to determine the number of either Dpn+ or pH3+ cells in the central brain of each brain lobe (a single brain lobe is n = 1). The individual that scored positive cells was blinded to genotype and exposure.

An Olympus Fluoview FV10i confocal microscope was used to generate all images.

2. 8. Lethality assay

Exposed P1 fly populations were placed into embryo collection chambers containing grape juice agar plates. After 24 hours, the plates were replaced and embryos were collected and transferred in groups of 100 into fresh exposure vials. After 8–11 days, the number of F1 progeny that had reached the pharate adult stage in each vial was counted. To calculate survival for F1 dFmr1 homozygotes, Mendelian ratios were assumed after crossing P1 dFmr1 heterozygotes. For embryonic lethality experiments, the F1 vials were given 48 hours, then the vial was rinsed into a coffee filter placed in a Büchner funnel and affixed to a vacuum. The number of first and second instar larvae were counted for each vial. Surviving offspring counts for the w¹¹¹⁸ strain were calculated by dividing by the total number of embryos (100) to produce survival ratios. For dFmr1 homozygotes, survival offspring counts were divided by the total number of predicted homozygotes out of the 100 embryos (25).

2. 9. Statistical analysis

Prism 8 (GraphPad) was used to perform all statistical analyses, except for Bartlett’s test for homoscedasticity for which JMP software was used (heteroscedastic data was log transformed). Normality was tested using the Anderson-Darling test. Parametric data was analyzed using one-way ANOVA with Tukey’s multiple comparison test, or the two-tailed unpaired Student’s t-test. Non-parametric data was analyzed using the Kruskal-Wallis test with the Dunn’s multiple comparison test. Figures were prepared using Prism 8 and BioRender.com.

3. RESULTS

3. 1. BPA increases locomotor behavior of w¹¹¹⁸ larvae, but can reduce locomotor behavior of dFmr1 mutant larvae

Crawling is the mode of locomotion used by Drosophila larvae to forage for food and respond to environmental stimuli [62]. This navigational behavior is characterized by persistent forward movement driven by waves of peristaltic contractions from head to tail (Figure 1A), punctuated by random reorientation maneuvers that occur when the larva pauses and changes its direction (Figure 1B) [63]. Hyperactive locomotion of dFmr1 larvae was previously reported and is thought to be caused by an increase in synaptic boutons at the neuromuscular junction [46, 62]. Similarly, a prior study of BPA found exposure increased the mobility of adult wild-type Oregon-R flies [20].

Figure 1. BPA induced hyperactivity in w¹¹¹⁸ larvae, but partially reduced hyperactivity in dFmr1 mutant larvae.

(A) Peristaltic contractions measured in late third-instar larvae. (B) Compared the w¹¹¹⁸ control group, exposure to 0.1mM and 1mM BPA significantly increased the number of peristaltic contractions in w¹¹¹⁸ larvae. BPA exposure did not significantly affect the number of peristaltic contractions in dFmr1^Delta113 mutant larvae. Bars represent mean ± SEM; **** = P < 0.0001; ns = not significant. Sample sizes for w¹¹¹⁸: Unexposed, n = 40; 0.1mM, n = 42; 1mM, n = 41. Sample sizes for dFmr1^Delta113 mutants: Unexposed, n = 46; 0.1mM, n = 46; 1mM, n = 41. (One-way ANOVA and Tukey’s multiple comparison test.) (C) Orientation behavior measured in late third-instar larvae. (D) Compared the w¹¹¹⁸ control, exposure to 0.1mM and 1mM BPA significantly increased the number of orientation changes in w¹¹¹⁸ larvae. In contrast, exposure to 0.1mM BPA significantly reduced the number of orientation events in dFmr1^Delta113 mutant larvae, though exposure to 1mM BPA did not have a significant impact. Bars represent mean ± SEM; *** = P < 0.001; **** = P < 0.0001; ns = not significant. Sample sizes for w¹¹¹⁸: Unexposed, n = 40; 0.1mM, n = 43; 1mM, n = 41. Sample sizes for dFmr1^Delta113 mutants: Unexposed, n = 48; 0.1mM, n = 45; 1mM, n = 40. (Kruskal-Wallis test and Dunn’s multiple comparison test.)

Consistent with these studies, we found a significant increase in the number of peristaltic contractions of unexposed dFmr1 larvae (20.74 ±0.91) compared to unexposed w¹¹¹⁸ larvae (13.73 ±0.72; Figure 1C). We also found that BPA increased peristaltic contractions in w¹¹¹⁸ larvae; exposure to both 0.1mM and 1mM BPA significantly increased the number of peristaltic contractions to 23.41 ±0.82 and 22.41 ±1.23, respectively, compared to 13.73 ±0.73 performed by unexposed w¹¹¹⁸ larvae. In contrast, BPA did not have a significant impact on the number of peristaltic contractions in dFmr1 larvae at 0.1mM (18.80 ±1.00) or 1mM BPA (19.85 ±1.01) compared to unexposed dFmr1 larvae (20.74 ±0.92).

We also observed a significant increase in the number of reorientation events in unexposed dFmr1 larvae (10.52 ±0.63) compared to unexposed w¹¹¹⁸ larvae (4.20 ±0.25; Figure 1D). Exposure to 0.1mM and 1mM BPA significantly increased orientation changes in w¹¹¹⁸ larvae, as the mean number of reorientation events increased from 4.200 ±0.25 (unexposed) to 6.35 ±0.51 (0.1mM BPA) and 7.20 ±0.38 (1mM BPA). However, BPA had the opposite effect in dFmr1 larvae by rescuing the increase in reorientation caused by loss of dFmr1. Exposure to 0.1mM BPA significantly reduced the reorientation events in dFmr1 larvae (7.20 ±0.39) compared to unexposed dFmr1 larvae (10.52 ±0.63). There was also a decrease in reorientation in response to 1mM BPA in dFmr1 larvae (8.75 ±0.53), although this trend was not statistically significant. These data indicate that BPA increases hyperactive locomotion in w¹¹¹⁸ larvae, but can lead to a reduction in hyperactivity in dFmr1 larvae.

3.2. BPA increases repetitive grooming activity of w¹¹¹⁸ flies, but reduces grooming activity of dFmr1 mutant flies

Grooming is a common animal behavior defined as cleaning activities directed at the external surface of the body (Figure 2A) [61]. dFmr1 mutant Drosophila were previously reported to exhibit more repetitive grooming compared to wild-type controls [44], a phenotype suggested as being analogous to the repetitive behaviors observed in individuals with FXS and ASD. We also observed a significant increase in repetitive grooming caused by loss of dFmr1 (Figure 2B). The mean amount of time spent on repetitive grooming in unexposed w¹¹¹⁸ flies was 130.5 ±9.70 seconds, while the mean grooming time of dFmr1 flies was 187.8 ±14.9 seconds. In w¹¹¹⁸ flies, exposure to 1mM BPA significantly increased the amount of grooming time from 130.5 ±9.70 to 174.8 ±14.9 seconds. Time spent grooming following exposure to 0.1mM BPA was 142.1 ±6.07, which represented an insignificant increase. However, for dFmr1 flies exposure to both 0.1mM and 1mM BPA significantly reduced grooming time to 96.44 ±11.5 and 124.5 ±22.0 seconds, respectively. BPA exposure again elicited a contrasting phenotype in dFmr1 flies compared to w¹¹¹⁸ flies and appeared to rescue the excessive grooming phenotype associated with loss of dFmr1. We also noticed that this observed rescue effect of BPA in dFmr1 flies was more robust at 0.1mM than 1mM BPA, a trend also observed during analysis of dFmr1 mutant larval reorientation.

Figure 2. BPA increased grooming time in w¹¹¹⁸ flies, but reduced grooming time in dFmr1 mutant flies.

(A) Example of repetitive grooming behavior in adult flies, which includes rubbing prothoracic or metathoracic legs against each other or other body parts. (B) Compared to the unexposed w¹¹¹⁸ control, exposure to 1mM BPA, but not 0.1mM BPA, significantly increased the time spent grooming by w¹¹¹⁸ flies. In dFmr1^Delta113 mutant flies, exposure to both 0.1mM and 1mM BPA resulted in a significant decrease in grooming time compared to the unexposed dFmr1^Delta113 control group. * = P < 0.05; ** = P < 0.01; ns = not significant. Sample sizes for w¹¹¹⁸: Unexposed, n = 13; 0.1mM, n = 12; 1mM, n = 15. Sample sizes for dFmr1^Delta113 mutants: Unexposed, n = 9; 0.1mM, n = 9; 1mM, n = 8. (One-way ANOVA and Tukey’s multiple comparison test of log transformed data.)

3.3. BPA affects courtship behavior in w¹¹¹⁸, but not dFmr1 mutant Drosophila

The sequential behaviors performed by male fruit flies during courtship (Figure 3A) are genetically programmed and involve the integration of multiple sensory modalities; thus, the naïve courtship paradigm can reveal a variety of neurodevelopmental impairments. The principal quantitative output of the courtship assay is the courtship index (CI), which reflects the percent time male flies participate in courtship behaviors over the duration of the assay. Loss of dFmr1 was previously shown to result in a significantly lower CI compared to wild-type controls [43]. Here, we found a similar result—unexposed dFmr1 mutant flies failed to actively participate in courtship, giving rise to a significantly lower CI (0.185 ±0.05) compared to unexposed w¹¹¹⁸ flies (0.681 ±0.03; Figure 3B). BPA was also previously found to impair the courtship behavior of wild-type Oregon R flies [64]. Similarly, we found that exposure of w¹¹¹⁸ flies to 1mM BPA led to a significant decrease in CI (0.484 ±0.03) compared to unexposed w¹¹¹⁸ flies (0.681 ±0.03). However, there was no significant difference in the CI values of dFmr1 flies upon exposure to BPA—the mean CI of unexposed flies compared to those exposed to 0.1mM and 1mM BPA were 0.185 ±0.05, 0.178 ±0.05, and 0.187 ±0.03, respectively. This data indicates that BPA impairs courtship behavior in w¹¹¹⁸ flies, but does not impact the courtship index of dFmr1 mutant flies at the concentrations examined.

Figure 3. BPA reduced the courtship index in w¹¹¹⁸ flies, but did not impact the courtship index in dFmr1 mutant flies.

(A) Standard courtship behaviors—orientation/following, leg tapping, wing extension, genital licking, attempted copulation and successful copulation—displayed by male flies were quantified in this experimental paradigm. The courtship index is calculated by dividing the time spent participating in these behaviors by the total duration of the assay. (B) Exposure to 1mM BPA, but not 0.1mM BPA, significantly reduced the courtship index of w¹¹¹⁸ flies when compared to the unexposed w¹¹¹⁸ control group. Exposure to BPA had no impact on the courtship index of dFmr1^Delta113 mutant flies. Bars represent mean ± SEM; *** = P < 0.001; ns = not significant. Sample sizes for w¹¹¹⁸: n = 36 for all conditions. Sample sizes for dFmr1^Delta113 mutants: unexposed, n = 9; 0.1mM, n = 10; 1mM, n = 10. (One-way ANOVA and Tukey’s multiple comparison test.)

3. 4. BPA exposure increases axon guidance defects of w¹¹¹⁸ flies, but rescues axon guidance defects of dFmr1 mutant flies

Abnormal axon guidance is a cellular phenotype associated with a number of NDDs [65, 66]. In Drosophila, axon guidance can be examined in the adult mushroom body, a midbrain structure required for learning and memory. The mushroom body is a bilaterally symmetric structure formed by a pair of neuropils that include axonal trajectories in α, β, and γ lobes (Figure 4A). When axon guidance is disrupted—as in dFmr1 mutants—one consequent mutant phenotype is midline crossing of β-lobe axons. This phenotype can be scored based on both severity (Figure 4B) and frequency [40].

Figure 4. BPA exposure increased the frequency of β-lobe midline crossing defects in w¹¹¹⁸ mushroom bodies, but reduced the frequency of β-lobe midline crossing in dFmr1 flies.

(A) Diagram of the Drosophila mushroom body, including the alpha (α), beta (β) and gamma lobes (γ). (B) Representative images of anti-FasciclinII (anti-FasII, green) stained mushroom bodies highlighting the different severities of β-lobe midline crossing phenotypes, which range from no crossing to mild, moderate, or severe crossing. White arrowheads indicate the midline of each brain, where severity is assessed. (C) In w¹¹¹⁸ flies, the frequency of midline crossing increased with exposure to 0.1mM and 1mM BPA, but then decreased with exposure to 2mM BPA. (D) In dFmr1³ flies, the frequency of midline crossing was reduced by exposure to 0.1mM, 1mM and 2mM BPA. Sample sizes for w¹¹¹⁸: Unexposed, n = 20; 0.1mM, n = 14; 1mM, n = 17; 2mM, n = 32. Sample sizes for dFmr1³ mutants: Unexposed, n = 14; 0.1mM, n = 16; 1mM, n = 14; 2mM, n = 15.

dFmr1 mutant flies exhibit increased frequency and severity of midline crossing by β-lobe axons relative to wild-type flies [40]. Our results were consistent, as demonstrated by our unexposed control groups for each genotype (Figures 4C–D). The midline crossing frequency was 15% and 79% in unexposed w¹¹¹⁸ and dFmr1 flies, respectively. The frequency of midline crosses among w¹¹¹⁸ males increased 26% between unexposed and 1mM BPA exposure then decreased by 19% upon exposure to 2mM BPA. This finding is reflective of the non-monotonicity previously observed with BPA [26]. Conversely, in dFmr1 mutant flies, the frequency of midline crossing decreased steadily with increasing BPA exposure, from 79% for unexposed dFmr1 mutants down to 47% at 2mM BPA. This finding again suggests that BPA exposure can rescue some dFmr1 mutant phenotypes.

3. 5. BPA increases lethality of w¹¹¹⁸, but not dFmr1 mutant flies

One explanation for the reduction of midline crossing defects in dFmr1 mutant flies following exposure is that BPA is eliciting a neuroprotective effect. Another possibility is that BPA exposure is causing lethality during embryonic-to-pupal stages. To distinguish between these two possibilities, we examined lethality following embryonic and larval exposure to 1mM BPA across the two genotypes by counting how many flies survived to the pharate adult stage. While a significant decrease in survival (i.e. increase in lethality) was observed in w¹¹¹⁸ flies exposed to 1mM BPA, no significant difference in survival was observed in dFmr1 mutant flies exposed to BPA (Figure 5A). The mean ratio of dFmr1 offspring that survived to the pharate adult stage decreased from 0.69 to 0.58 (an insignificant difference), while the mean ratio of w¹¹¹⁸ offspring that survived to the pharate adult stage significantly decreased from 0.65 to 0.33 upon exposure to 1mM BPA. When w¹¹¹⁸ flies were examined for lethality immediately post-embryonically—at the first to second instar larval stages (Figure 5B)—the results persisted. The mean ratio of w¹¹¹⁸ flies surviving to the early larval stages decreased from 0.61 to 0.42 upon exposure to 1mM BPA (Figure 5C). This indicates that the combination of BPA and homozygous loss of dFmr1 again rescue a mutant phenotype—this time, BPA-induced lethality.

Figure 5. BPA increased lethality in w¹¹¹⁸ flies, but not dFmr1 mutant flies.

(A) The mean ratio of w¹¹¹⁸ F1 progeny surviving to the pharate adult (late pupal) stage decreased upon exposure to 1mM BPA. The mean ratio of dFmr1³ F1 progeny surviving to the pharate adult stage was not significantly changed by exposure to 1mM BPA. Bars represent mean ± SEM; ** = P < 0.005. For each genotype and exposure, n = 4 replicates with 100 embryos each. (One-way ANOVA and Tukey’s multiple comparison test.) (B) The Drosophila lifecycle. Adults mate and lay embryos that hatch and progress through first instar, second instar, and third instar larval stages before pupation and successive emergence of adult flies. The pharate adult stage is the final stage before eclosion, at which point the pupa is considered to be an ensheathed adult. (C) The mean ratio of w¹¹¹⁸ F1 progeny, surviving to the first and second instar larval stages decreased upon exposure to 1mM BPA. Bars represent mean ± SEM; ** = P < 0.005. For each exposure, n = 4 replicates with 100 embryos each. (Student’s t test.)

3. 6. BPA differentially impacts the number and proliferative capacity of neuroblast and progenitor cells

Impaired neural stem cell (NSC) development is a common theme in NDDs and is often regarded as a phenotypic point of convergence [67–69]. In the Drosophila brain, NSCs are referred to as neuroblasts (NBs) and they give rise to two distinct progenitor cell populations. We quantified NBs and progenitor cells, as well as mitotically active cells in the central brain of third instar larvae because dFmr1 mutant flies were previously shown to exhibit increased proliferative capacity in this region of the larval brain [38].

First, we examined the total number of NBs and intermediate neural progenitor (INP) cells using Deadpan (Dpn), a transcription factor specifically expressed in these cell populations [70, 71]. We fluorescently labeled Dpn⁺ cells in whole mount larval brains (Figure 6A), and used IMARIS Bitplane software to quantify Dpn⁺ cells within the central brain. Using this approach we found a significantly higher number of total combined Dpn⁺ cells in the central brain of unexposed dFmr1 mutant larvae (138.3 ±7.72 Dpn⁺ cells) compared to unexposed w¹¹¹⁸ larvae (114.6 ±6.77 Dpn⁺ cells; Figure 6A–B). We did not detect a significant difference between Dpn⁺ cells in w¹¹¹⁸ larval central brains following exposure to 0.1mM BPA or 1mM BPA. In contrast, dFmr1 mutant larvae exposed to 0.1mM and 1mM BPA exhibited a significantly reduced mean number of Dpn⁺ cells in the central brain from 138.3±7.72 (unexposed) to 90.08±4.05 (0.1mM BPA) and 100.5±5.18 (1mM BPA). Thus, BPA exposure rescued the increase in Dpn⁺ cells caused by loss of dFmr1. These data indicate that BPA exposure does not affect the total combined number of NBs and INPs in w¹¹¹⁸ larval brains, but significantly reduces this number in dFmr1 larvae—once again exhibiting a different impact on the w¹¹¹⁸ and dFmr1 mutant strains.

Figure 6. BPA differentially affected neural stem cell development in w¹¹¹⁸ and dFmr1 flies.

(A) Representative images from each genotype and exposure of late third-instar larval brains stained with anti-Deadpan (Dpn) to mark neuroblasts (NBs) and mature intermediate neural progenitors (INPs). (B) In w¹¹¹⁸ larvae, exposure to BPA did not impact the mean number of NBs and mature INPs in the larval central brain when compared to the unexposed w¹¹¹⁸ control. In dFmr1^Delta113 mutant larvae, exposure to both 0.1mM and 1mM BPA significantly reduced the number of NBs and mature INPs when compared to the unexposed dFmr1^Delta113 control group. Bars represent mean ± SEM; *** = P < 0.001; **** = P < 0.0001; ns = not significant. Sample sizes for w¹¹¹⁸: Unexposed, n = 10; 0.1mM, n = 9; 1mM, n = 13. Sample sizes for dFmr1^Delta113 mutants: Unexposed, n = 10; 0.1mM, n = 12; 1mM, n = 10. (One-way ANOVA and Tukey’s multiple comparison test.) (C) Representative images from each genotype and exposure of late third-instar larval brains stained with anti-phospho-Histone H3 (pH3) to mark mitotically active NBs and progenitors. (D) In w¹¹¹⁸ larvae, exposure to 1mM BPA significantly reduced the number of mitotically active cells in the central brain compared to unexposed w¹¹¹⁸ controls. BPA exposure did not significantly impact the number of mitotically active cells in the central brains of dFmr1³ mutant larvae. Bars represent mean ± SEM; * = P < 0.05; ns = not significant. Sample sizes for w¹¹¹⁸: Unexposed, n = 8; 1mM, n = 9. Sample sizes for dFmr1³ mutants: Unexposed, n = 11; 1mM, n = 17. (Student’s t test.)

Next, we measured the number of mitotically active cells in the larval central brain. This population includes NBs, INPs, as well as ganglion mother cell (GMC) progenitor cells. Using an antibody for phospho-Histone H3 (pH3) to mark cells in M-phase (Figure 6C), and IMARIS Bitplane software for quantification, we found that w¹¹¹⁸ larvae have a mean of 145.8 ±17.2 pH3⁺ cells, while dFmr1 mutant larvae have a higher mean of 169.3 ±12 pH3⁺ cells (Figure 6D). In response to 1mM of BPA, the mean number of mitotically active cells in the w¹¹¹⁸ central brain decreased significantly from 145.8 ±17.2 to 101.6 ±9.7 pH3+ cells (Figure 6D). In our dFmr1 mutant larvae, the mean number of mitotically active cells in the central brain decreased from 169.3±12 to 149.4±13.6 in response to 1mM BPA exposure, but this trend was not statistically significant. Our results indicate that 1mM BPA may reduce the total number of proliferating NBs and progenitor cells in the central brains of w¹¹¹⁸, but not dFmr1 larvae—yet again displaying the ability of BPA to exert differential impacts on two distinct genotypes.

4. DISCUSSION

4.1. BPA elicits disparate neurodevelopmental phenotypes in w1118 and dFmr1 mutant fly strains

Contrary to our initial hypothesis, BPA exposure caused distinct phenotypes in w¹¹¹⁸ and dFmr1 mutant fly strains, an observation that held true across multiple behavioral and neuronal phenotypes (Table 1). To comprehensively understand the molecular mechanisms by which BPA elicits such contrasting phenotypes in these fly strains, transcriptomic and translatomic analysis is likely required. BPA has been shown to impact expression of mRNAs [72–74], as well as non-coding RNAs, like long non-coding RNAs and microRNAs [75–77]; thus, similar to loss of dFmr1 expression, the molecular consequences of BPA exposure are vast. Metabolic profiling could also be useful given that the toxicokinetics of BPA in Drosophila are unknown, although neither dFmr1 nor its mammalian orthologs have been implicated in xenobiotic metabolism.

Table 1. Summary of BPA-induced phenotypes in w¹¹¹⁸ and dFmr1 mutant fly strains.

Each solid arrow indicates a significant difference, either increasing (up arrow) or decreasing (down arrow) the phenotype examined. Light grey arrows indicate significance at 0.1mM BPA and black arrows indicate significance at 1mM BPA. Grey arrows with dashed lines represent the qualitative differences in axon guidance. (ns = not significant.)

	W1118	dFmr1
Peristaltic Contractions		ns
Reorientation Events
Grooming Time
Courtship Index		ns
Lethality		ns
β-Lobe Axon Guidance Defects
Total # of NBs & INPs (Dpn+ Cells)	ns
Mitotic Activity (pH3+ Cells)		ns

In some cases, targeted analysis of proteins may shed incremental light on the differential impact of BPA. Loss of dFmr1 leads to the translational upregulation of many critical neuronal transcripts. For example, FMRP is thought to control locomotion through the translational downregulation of Protein Pickpocket1 (PPK1) [45], an Epithelial Sodium Channel family member that elicits sensory neuron activity in response to external stimuli [78]. Loss of FMRP increases PPK1 expression, thereby disrupting the neural circuits that drive larval locomotion and causing the hyperactivity of dFmr1 mutants [45]. Another example is the increased expression of Vesicular Monoamine Transporter (VMAT) in dFmr1 mutant flies, which is thought to contribute to the increased grooming activity [44]. Given that BPA rescues reorientation and grooming phenotypes in dFmr1 mutants, it is tempting to reason that BPA must decrease PPK and VMAT expression. But BPA causes increased orientation changes and grooming in w¹¹¹⁸ flies, so if PPK and VMAT are indeed molecular targets of BPA, they must be impacted differentially according to genetic background. This scenario is plausible—because of its critical role in translational regulation, loss of FMRP creates an immensely different molecular environment. We speculate that among the proteins upregulated in dFmr1 mutants, there must be alternative receptors or proteins that modulate BPA reception and response.

In addition to specific proteins, analyzing signaling pathways and epigenetic changes is also called for. Dopamine signaling has been previously connected to both BPA and dFmr1—BPA exposure can decrease dopamine production in Drosophila [79], while loss of dFmr1 increases dopamine levels [80]. Dopamine signaling is critical for a number of Drosophila behaviors. For example, elevated dopamine levels have been associated with increased adult grooming behavior [81] and adult locomotor activity [79]. These findings provide a potentially simple explanation for the BPA-exposed dFmr1 mutants that spent less time grooming; BPA exposure could have reduced the elevated dopamine levels caused by loss of dFmr1. However, if BPA depresses dopamine production, it is unclear why we observed an increase in grooming activity by w¹¹¹⁸ flies exposed to 1mM BPA. One explanation could be epigenetic changes in neurons. Studies centered on behavioral epigenetics have demonstrated that social environment influences the epigenetic landscape of dopaminergic neurons in both mice and fruit flies [82, 83]. Social isolation of male fruit flies, in particular, was shown to affect epigenetic changes in these neurons [82]. We isolated post-eclosion males prior to measuring their behaviors; thus, if social isolation epigenetically alters dopaminergic neurons, it is possible that such changes modified the molecular reception or response to BPA in our w¹¹¹⁸ population. It is equally plausible that dopamine-related impacts were confounded by other BPA-related effects in w¹¹¹⁸ flies. On the topic of epigenetics, BPA itself has been shown to influence DNA methylation in both humans and rodent models [84–86]. While our study focused solely on neurodevelopmental and behavioral outcomes of F1 populations, we began exposure with P1 females to ensure embryonic exposure to BPA. Future work should be done to examine how P1 exposure affects epigenetic modification of DNA in oocytes of dFmr1 mutant and w¹¹¹⁸ females and how those potential changes may impact development of their offspring.

Parsing the data surrounding the impact of BPA on neuroblast development is particularly interesting. Neuroblasts are specified during embryogenesis when they delaminate from the neuroectoderm [87]. By late embryonic stages, NBs are classified as either type I, type II or mushroom body NBs. During larval stages, NBs undergo asymmetric divisions, which serve to maintain their populations via self-renewal and also give rise to different progenitor populations [87]. The two major types of progenitor cells are INPs and GMCs [87, 88]. Each NB lineage is maintained in specific quantities throughout this developmental program; when the program is perturbed, different numbers of cells can be observed in the larval brain, providing a quantifiable measure of genetic and environmental influences on neurodevelopment. A variety of processes can be disrupted during NSC development to impact cell numbers, including specification, proliferation, and differentiation. Flies lacking dFmr1 have an increased number of Dpn+ cells (which marks the total number of NBs and mature INPs), and BPA exposure significantly reduced that number. But it is important to note that the total number of NBs and INPs does not necessarily reflect a change in their proliferative capacity during larval stages—during this developmental stage NBs and INPs undergo asymmetric divisions that maintain their populations. Differences in Dpn⁺ cells could indicate: 1) a change in the number of NBs specified during embryogenesis, 2) a disruption in the homeostatic mechanisms that regulate asymmetric versus symmetric division, or 3) a change in NB proliferative capacity—though the latter could only be determined if we were to compare the ratio of NBs to INPs, which our study did not distinguish between. However, our data showing that BPA reduced pH3⁺ cells in w¹¹¹⁸ larval brains, but had no impact on pH3⁺ cells in dFmr1 mutant brains suggests the observed changes in Dpn⁺ cells are not reflective of a change in proliferative capacity. Finally, an important caveat to our experimental approach is that, unlike mammalian models, there is no pan-mitotic marker for Drosophila; our labeling of pH3+ cells limited our assessment to mitotically active cells in M phase.

4.2. Gene-environment interaction in FXS Drosophila?

While our data indicates BPA can indeed have distinct impacts on different genetic strains, this study is only suggestive of a specific gene-environment interaction between dFmr1 and BPA. To definitively make this claim, the w¹¹¹⁸ and dFmr1 mutant lines should be isogenic outside of the dFmr1 locus. While the dFmr1 lines used here were originally created twenty years ago via P-element excision from the w¹¹¹⁸ line, our w¹¹¹⁸ strain and dFmr1 strains are not isogenic. We did observe similar phenomena across two distinct dFmr1 null mutant lines—dFmr1³ was used for assessment of axon guidance, lethality, and NB and progenitor cell mitotic activity, while dFmr1^Delta113 was used for all other analyses. We observed disparate BPA-related phenotypes in both dFmr1 mutant lines relative to w¹¹¹⁸ flies—lending credence to the idea that loss of dFmr1 expression is responsible for the distinct BPA-related phenotypes. Though, we cannot exclude the potential involvement of other genetic loci. In addition, a broader concentration range, including lower concentrations, would aid in the delineation of gene-environment interactions. While our data provides a compelling starting point, follow on studies should include an expansion of concentrations, as well as more dFmr1 mutant and control strains.

Despite the observation that BPA partially rescued some dFmr1 mutant phenotypes, we would reject any suggestion that BPA could serve as a viable treatment option for any human disorders given the multitude of ways BPA impairs development and adult physiology. However, if this putative gene-environment interaction is validated in mammalian model organisms, understanding the molecular mechanism by which BPA potentially interferes with dFmr1-induced neurodevelopmental impairments could provide new pharmacological targets for FXS.

4. 3. Endocrine disrupting chemicals and gender

As an EDC, BPA is known to have gender-specific impacts in a variety of vertebrate organisms [6, 10]. It is important to note that our analysis was not designed to determine gender-specificity in Drosophila. In the courtship assay, we assessed the impacts of BPA on males because this experimental paradigm is traditionally centered on male-specific courtship patterns. The female flies used in the courtship assay were all unexposed w¹¹¹⁸ flies. For consistency, we chose to use male flies in other experiments involving adult flies (grooming and axon guidance). But because larvae do not have external genitalia or other morphological markers that enable reliable identification, larvae are typically not separated by gender. As such, we did not segregate genders in experiments involving larvae (locomotion, lethality, and NSC analyses). Thus, if BPA does have gender-specific responses in Drosophila, our mixed population of larvae in these assays may have confounded the results. While there is currently no evidence to suggest sexually dimorphic expression of a putative BPA receptor in fruit flies, that is largely because no study has addressed this question. An interesting future direction would be to investigate if BPA has gender specific impacts on Drosophila phenotypes.

In vertebrates, the endocrine-disrupting capabilities of BPA lie in its structural resemblance to estradiol, which facilitates binding to multiple estrogen receptor (ER) subtypes [6, 10, 89, 90]. Drosophila, however, do not have ERs encoded in their genome. Instead, they possess an estrogen related receptor (dERR) that belongs to the same nuclear receptor superfamily as ERs, but does not bind estrogen [91, 92]. However, ERRs can bind other ER ligands, thereby demonstrating some overlapping binding specificity [92]. In Drosophila, dERR activity is required for metabolism, mitochondrial function, testicular development, and testicular function [91, 93]. A study using Chironomus riparius, the harlequin fly, showed that BPA can induce the expression of genes linked to ERR activity [94]; although, no binding assay has directly assessed the capacity of BPA to directly bind ERRs in any organism. Receptor binding assays and other molecular analyses are warranted to determine if BPA-dERR binding confers any of the observed neurodevelopmental phenotypes.

Many EDCs, BPA included, exhibit non-monotonic dose response curves [27, 95]. A major limitation of this study was that our analysis was largely limited to two concentrations of BPA, which is not sufficient for delineating a dose-response curve. The only exception was our examination of axon guidance in the mushroom body where we included a 2mM exposure condition. In this case, relative to control w¹¹¹⁸ flies, β-lobe midline crossing defects increased by 6% with 0.1mM BPA and 26% with 1mM BPA. Exposure to 2mM BPA exposure only caused a 6.9% increase in midline crossing defects above the control condition—a 19.1% decrease from the 1mM BPA group. Although reminiscent of a non-monotonic response, the qualitative nature of this experimental paradigm prevents us from drawing that conclusion. In addition, our observed axon guidance frequency at 2mM may not reflect the true severity of the crossing phenotype, which was scored in 0- to 4-day old adult brains. We found that 1mM BPA increased embryonic-to-pupal lethality of w¹¹¹⁸ flies. While our lethality study should be broadened to include more concentrations, the fact that exposure to 1mM BPA affects lethality indicates we cannot rule out the possibility that 2mM BPA may have caused more severe midline crossing phenotypes in w¹¹¹⁸ flies that failed to be observed because of early lethality—particularly given that we did observe an increase in the severity of midline crossing at this concentration. Finally, other studies have found that BPA exposure impacts relevant adult phenotypes in wild-type flies, including causing increased mortality [79] and reduced fecundity [96]. Given that we administered BPA to the P1 generation in order to ensure embryonic exposure of the F1 generation, we cannot rule out the possibility that this approach may have culled P1 flies that could have produced an F1 generation that responded differently to BPA. We would also need to compare mortality in the parental dFmr1 mutant flies to discern the relative impact among the two genotypes.

4. 4. Conclusion

Although several studies of NDDs, most notably ASD, have linked BPA to neurodevelopmental and behavioral problems, this is the first attempt to investigate the effects of BPA in combination with a specific genetic risk factor. Our findings are consistent with prior examinations of dFmr1 mutant strains for all phenotypes examined and support previous data indicating BPA impacts hyperactive locomotor responses, repetitive grooming behaviors, and courtship behaviors of Drosophila. Our results add to this story by indicating BPA can impact NSC development and axon guidance in Drosophila. This study also highlights the complexity—and, in this case, unpredictable nature—of genotype-environment interactions and the need for more expansive investigations of environmental chemicals in different genetic strains to adequately assess their potential contribution to NDD pathophysiology. The genetic tractability and economy of fruit flies makes them a valuable in vivo model for such efforts.