Developmental Toxicity Assessment of Piperonyl Butoxide Exposure Targeting Sonic Hedgehog Signaling and Forebrain and Face Morphogenesis in the Mouse: An in Vitro and in Vivo Study

Background: Piperonyl butoxide (PBO) is a pesticide synergist used in residential, commercial, and agricultural settings. PBO was recently found to inhibit Sonic hedgehog (Shh) signaling, a key developmental regulatory pathway. Disruption of Shh signaling is linked to birth defects, including holoprosencephaly (HPE), a malformation of the forebrain and face thought to result from complex gene–environment interactions. Objectives: The impact of PBO on Shh signaling in vitro and forebrain and face development in vivo was examined. Methods: The influence of PBO on Shh pathway transduction was assayed in mouse and human cell lines. To examine its teratogenic potential, a single dose of PBO (22–1,800mg/kg) was administered by oral gavage to C57BL/6J mice at gestational day 7.75, targeting the critical period for HPE. Gene–environment interactions were investigated using Shh+/− mice, which model human HPE-associated genetic mutations. Results: PBO attenuated Shh signaling in vitro through a mechanism similar to that of the known teratogen cyclopamine. In utero PBO exposure caused characteristic HPE facial dysmorphology including dose-dependent midface hypoplasia and hypotelorism, with a lowest observable effect level of 67mg/kg. Median forebrain deficiency characteristic of HPE was observed in severely affected animals, whereas all effective doses disrupted development of Shh-dependent transient forebrain structures that generate cortical interneurons. Normally silent heterozygous Shh null mutations exacerbated PBO teratogenicity at all doses tested, including 33mg/kg. Discussion: These findings demonstrate that prenatal PBO exposure can cause overt forebrain and face malformations or neurodevelopmental disruptions with subtle or no craniofacial dysmorphology in mice. By targeting Shh signaling as a sensitive mechanism of action and examining gene–environment interactions, this study defined a lowest observable effect level for PBO developmental toxicity in mice more than 30-fold lower than previously recognized. Human exposure to PBO and its potential contribution to etiologically complex birth defects should be rigorously examined. https://doi.org/10.1289/EHP5260


Introduction
Piperonyl butoxide (PBO) is a semisynthetic pesticide synergist that inhibits insect cytochrome p450 enzyme activity (U.S. EPA 2006;Scott 1996). Most commonly used in pyrethrin-and pyrethroidbased pesticide formulas, PBO is present in hundreds of products used in agricultural, commercial, and residential settings, including indoor and outdoor foggers and topical medications like lice shampoo. These products typically contain substantially higher concentrations of PBO than the active insecticidal ingredient (U.S. EPA 2006). PBO was one of the top 10 chemical contaminants found in indoor dust sampled from 119 homes in Cape Cod, Massachusetts (Rudel et al. 2003), and was detected in 75% of personal air samples of 230 pregnant women in New York City, New York, who self-identified as either African American or Dominican (Horton et al. 2011). Use of pyrethrin-and pyrethroid-based formulations containing PBO has increased with the decline of organophosphate pesticides, as reviewed in the U.S. Environmental Protection Agency (EPA) Registration Eligibility Decision for PBO (U.S. EPA 2006). Relative to active pesticidal components, the developmental toxicity of PBO has received little research attention.
PBO was recently discovered to inhibit the Sonic hedgehog (Shh) signaling pathway in a study using high-throughput in vitro screening and Shh-responsive cell-based and zebrafish assays (Wang et al. 2012). The Shh pathway is required for multiple aspects of embryonic development, including morphogenesis of the forebrain and midface, as illustrated by major malformations observed in Shh knockout mice (Chiang et al. 1996). Shh pathway inhibition is classically associated with holoprosencephaly (HPE), a congenital condition defined by median forebrain deficiency, typically occurring with characteristic facial dysmorphology, including hypotelorism and midface hypoplasia (Weiss et al. 2018). Mice with homozygous Shh null mutations exhibit severe HPE (Chiang et al. 1996), and SHH mutations have been identified as among the most common human HPE-associated gene mutations (Nanni et al. 1999;Roessler et al. 1996Roessler et al. , 2018Roessler and Muenke 2010). The Shh pathway also appears to be inherently sensitive to small molecule modulation (Chen et al. 2002b;Chen 2016). Well-characterized and structurally diverse pathway antagonists include the plant alkaloid cyclopamine, which was found to cause HPE in livestock, rabbits, and rodents (Keeler 1970(Keeler , 1975(Keeler , 1978, and the U.S. Federal Drug Administration (FDA)-approved drug vismodegib, which was shown to cause HPE in mice (Heyne et al. 2015a(Heyne et al. , 2016. Human HPE is an etiologically heterogeneous condition associated with significant morbidity. Although occurring in approximately 1 in 10,000 live births (Leoncini et al. 2008), HPE was estimated to have a prevalence of 1 in 250 conceptuses (Matsunaga and Shiota 1977). These observations suggest that HPE is one of the most common human developmental abnormalities, but the vast majority of affected embryos do not survive to birth. In surviving patients, HPE can cause severe intellectual disability and learning, behavior, and motor impairment (Levey et al. 2010;Weiss et al. 2018).
HPE is thought to arise from complex gene-environment interactions (Graham and Shaw 2005;Hong and Krauss 2018;Krauss and Hong 2016;Lovely et al. 2017;Petryk et al. 2015;Roessler et al. 2003Roessler et al. , 2018. Most single-gene mutations linked to HPE pathogenesis have been found to be heterozygous and associated with highly variable phenotypic outcomes (Roessler and Muenke 2010). Only 25% of patients with nonchromosomal HPE have been found to have mutations in one of the four most common HPE genes (Roessler and Muenke 2010), and documented examples of clear gene-gene interactions have been exceedingly rare in humans (Roessler et al. 2018). Therefore, the environmental contribution to HPE etiology is likely substantial. However, relative to genetic factors, examination of environmental influences in the complex etiology of HPE has been largely neglected. A logical focus of such investigation is common environmental chemicals that perturb the Shh signaling pathway.
The few existing studies of the potential developmental toxicity of PBO were conducted before its inhibitory action on the Shh pathway was known and were therefore untargeted (U.S. EPA 2006). Unlike previous investigations, the studies reported here were designed to specifically examine the impact of PBO on Shh signaling. First, in vitro assays were used to define the impact of PBO on the Shh pathway transduction. Then, the in vivo developmental toxicity of PBO was tested by targeting exposure to a critical period of forebrain and face development, incorporating a clinically relevant genetic cofactor, and rigorously examining Shh-dependent outcomes.

Materials
Biologically active materials used for in vitro and/or in vivo studies included recombinant human SHH ligand (Catalog No. 1845-SH; R&D Systems), vismodegib (Catalog No. V-4050; LC Labs), cyclopamine (Catalog No. C-8700; LC Labs), the Smoothened (SMO) agonist (SAG; Catalog No. S7779; Selleckchem), and PBO (Catalog No. P490200; Toronto Research Chemicals). The supplier-stated purity of PBO was >96% and our own liquid chromatography-mass spectrometry analysis suggested a purity of 98.11%, with the presence of four minor impurities (see Figure S1).

Luciferase Assays
After media were removed, SHH LightII-3T3 MEF cells were washed once in PBS. Then lysates were collected in Passive Lysis Buffer (Catalog No. E1941; Promega) and luciferase activity was assayed using the Dual-Luciferase ® Reporter Assay System (Catalog No. E1910; Promega) according to manufacturer specifications. Relative light units (RLUs) for Firefly and Renilla luciferase were generated using a TD-20=20 luminometer (Catalog No. 2020-998;Turner Designs). The difference between SHH-and vehicle-treatment groups was used as a baseline to calculate the effect of drug exposure on relative GLI-dependent luciferase activity. Percentage reduction from SHH-treated values shown in Figure 1B were calculated for each experimental drug concentration treatment group (x) using the following equation: where SHH Ctl = relative luciferase of SHH + DMSO; Veh Ctl = relative luciferase of DMSO; and x = relative luciferase of SHH + drug at the given concentration. Calculations used SHH Ctl and Veh Ctl values generated within the same biological replicate experiment. Percentage reduction from SHH-treated values were used to establish concentration-response curves and approximate half maximal effective concentration (EC 50 ) values using Graphpad Prism (version 7; GraphPad Software Inc.).

Gene Expression Analysis of in Vitro Assays
For in vitro samples, cells were washed once in PBS before RNA isolation using GE Illustra RNAspin kits (Catalog No. 25-0500-72; GE Healthcare Life Sciences) according to manufacturer recommendations. Eluted RNA was stored at −80 C. Complementary DNA (cDNA) was synthesized from 250 ng of total RNA using GoScript reverse transcription reaction kits (Catalog No. A2791; Promega). Singleplex quantitative real-time polymerase chain reaction (qRT-PCR) was performed using a Bio-Rad CFX96 real-time PCR detection system (Bio-Rad Laboratories). The reaction mixture contained 0:75 lL cDNA, 6 lL SSoFast EvaGreen Supermix (Catalog No. 1725201; Bio-Rad Laboratories), 4:75 lL nuclease-free water, and 0:5 lL of 10 lM forward and reverse gene-specific primers designed using the IDT PrimerQuest tool (Integrated DNA Technologies). qRT-PCR primer sequences are listed in Table S1. Target gene specificity was confirmed using National Center for Biotechnology Information Primer-Basic Local Alignment Search Tool (NCBI Primer-BLAST). The gene glyceraldehyde-3-phosphate dehydrogenase (Gapdh) (mouse) or GAPDH (human) was used as the housekeeping gene, and analyses were conducted with the 2 -ddCt method. qRT-PCR conditions were as follows: one cycle at 95°C for 3 min, then 40 cycles × at 95°C for 10 s followed by 60°C for 30 s.

Animal Husbandry
This study was conducted in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (NIH) (NRC 2011). The protocol was approved by the University of Wisconsin School of Veterinary Medicine Institutional Animal Care and Use Committee (Protocol No. 13-081.0). Mice harboring the Shh null allele as described by Chiang et al. (1996)   A two-tailed t-test with Holm-Sidak correction was used to compare the statistical significance of relative luciferase activity between SHHand vehicle-treatment groups. * p < 0:05. (B) The difference between SHH-and vehicle-treatment groups was used as a baseline to determine the effect of drug exposure, calculated as percentage reduction in luciferase activity relative to SHH treatment alone. These values were used to establish concentration-response curves and approximate half maximal effective concentration (EC 50 ) values for each drug listed in parentheses. Values represent the mean ± SE of four biological replicate experiments. (C) Human medulloblastoma (DAOY) cells were treated ± 0:4 lg=mL SHH ligand and indicated concentrations of PBO, and GLI1 relative to GAPDH expression was measured. Values represent the mean ± SE of four biological replicate experiments. A one-way ANOVA followed by twotailed t-tests with Holm-Sidak correction were used to compare the statistical significance of SHH ± PBO-treated groups against the control (DMSO) group. ** p < 0:01; *** p < 0:001. (D) Baseline Shh pathway activity in Ptch1 −= − MEFs and those overexpressing (OE) either SMO M2 or GLI1 was compared with that in WT MEFs by examining Gli1 relative to Gapdh expression. Values represent the mean ± SE of five biological replicate experiments. Two-tailed t-tests with Welch's correction were used to compare statistical significance of relative Gli1 expression in Ptch1 −= − MEFs, SMO M2 OE MEFs, and GLI1 OE MEFs against relative Gli1 expression in WT MEFs. * p < 0:05; ** p < 0:01; *** p < 0:001. (E) Ptch1 −= − , SMO M2 OE, or GLI1 OE cells were treated ± the indicated PBO concentrations, and Gli1 relative to Gapdh expression was measured. Values from individual biological replicate experiments were normalized to the vehicle control group to show comparable fold changes. Values represent the mean ± SE of six biological replicate experiments. A one-way ANOVA followed by two-tailed t-tests with Holm-Sidak correction were used to compare the statistical significance of PBO-treated groups against the control (DMSO) group. * p < 0:05; ** p < 0:01. (F) Ptch1 −= − MEFs were treated ± 2 lM PBO and ± 50 nM SMO agonist SAG. Values from individual biological replicates were normalized to the vehicle (DMSO) control group. Values represent the mean ± SE of four biological replicate experiments. One-way ANOVA followed by twotailed t-tests with Holm-Sidak correction were used to compare the statistical significance between the vehicle and PBO ± SAG treated groups. ** p < 0:01; **** p < 0:0001. Note: DMSO, dimethyl sulfoxide; MEFs, mouse embryonic fibroblasts; PBO, piperonyl butoxide; RLU, relative luciferase activity; SAG, Smoothened agonist; SE, standard error; SHH, Sonic hedgehog ligand; SMO, Smoothened; WT, wild type.

General Animal Study Design
One to two nulliparous female C57BL=6J wild-type (WT) mice were placed with a single C57BL=6J WT or Shh + = − male for 1-2 h and subsequently examined for the presence of copulation plugs. The beginning of the mating period was designated as gestational day (GD)0. Pregnancy was confirmed by assessing weight gain at GD7, as previously described (Heyne et al. 2015b). Timed-pregnant dams were administered PBO dissolved in reagent-grade olive oil (Catalog No. OL130; Spectrum Chemical), vehicle (olive oil alone), or vismodegib (as a positive control) by oral gavage at GD7:75 ± 20 min. Vismodegib was suspended in 0.5% methyl cellulose with 0.2% Tween-80 as previously described (Heyne et al. 2015a(Heyne et al. , 2016. Pregnant dams were euthanized by carbon dioxide inhalation followed by cervical dislocation at GD8:25 ± 1 h or GD11 ± 1 h for embryo collection or at GD17 ± 2 h for fetal collection. Animal husbandry, drug administration, and embryo/fetal collection were conducted by individuals not blinded to experimental condition.

Facial Morphometric Analysis
For quantitative assessment of fetal facial morphology, WT dams were administered vehicle (n = 11) or 22.22 (n = 3), 66.67 (n = 4), 200 (n = 6), 600 (n = 5), or 1,800 mg=kg PBO (n = 5). Following euthanasia at GD17, dam uteri were inspected for resorptions and live fetuses, and crown-rump length was recorded for each fetus. Population parameters for the WT mouse study arm are described in Table 1. To assess gene-environment interactions, WT female mice mated with Shh + = − male mice were administered vehicle (n = 7) or 33 (n = 6), 110 (n = 7), or 300 mg=kg PBO (n = 7). Following euthanasia at GD17, dam uteri were inspected for resorptions and live fetuses, and crownrump length was recorded for each fetus. Population parameters for the study arm using Shh + = + and Shh + = − mice are described in Table 2. Interdigitated doses of PBO were chosen for these separate study arms to maximize dose-response data and avoid experimental duplication. For both study arms, all viable fetuses collected from each litter were included in the facial morphometric analyses. Fetuses were fixed in 10% phosphate buffered formalin and imaged by light microscopy, with careful attention placed on consistent positioning of the head and face.
Light microscopic images of GD17 fetuses were captured using a MicroPublisher 5.0 camera (QImaging) connected to an SZX-10 stereomicroscope (Olympus). Interocular distance (IOD) and upper lip length (ULL) measurements were made from images using ImageJ (Schneider et al. 2012) by a single rater blinded to genotype and treatment group.

Analysis of Embryonic Morphology
For qualitative assessment of embryonic face and brain morphology, embryos exposed to vehicle or 1,800 mg=kg PBO were collected at GD11. For each treatment group, two to three litters were collected and fixed in 4% paraformaldehyde (PFA) overnight, then dehydrated into methanol and stored at −20 C. Following rehydration into PBS, representative embryos were imaged using a QImaging MicroPublisher 5.0 camera connected to an Olympus SZX-10 stereomicroscope.

Analysis of Fetal Face and Brain Morphology
For qualitative assessment of correlative face and brain morphology, litters exposed to vehicle; 200, 600, or 1,800 mg=kg PBO; or 40 mg=kg vismodegib were collected at GD17 and fixed in Bouin's solution. Single litters were generated for each treatment group. From each litter, fetuses with facial dysmorphology representative of the corresponding group in the larger WT population were selected for histological analysis. Following paraffin embedding, 10-lm sections were produced and stained with hematoxylin and eosin (H&E) by standard protocols. Light microscopic images of fetal faces (before or after Bouin's fixation) and histological sections were captured using a QImaging MicroPublisher 5.0 camera connected to an Olympus SZX-10 stereomicroscope.

Medial Ganglionic Eminence Morphometric Analysis
For quantitative assessment of medial ganglionic eminence (MGE) morphology, WT litters exposed to vehicle (n = 5) or 22.22 (n = 3), 66.67 (n = 3), 200 (n = 3), or 600 mg=kg PBO (n = 3) were collected at GD11. Embryos were fixed in 4% PFA overnight, then dehydrated into methanol and stored at −20 C. Embryos were then rehydrated into PBS, hemisected sagittally using a scalpel, and imaged by light microscopy using a Note: Timed-pregnant C57BL=6J mice were administered the indicated doses of piperonyl butoxide (PBO) at gestational day (GD)7.75 and inspected at GD17 for live fetuses and resorptions. One-way ANOVA was used to compare number of live fetuses, resorptions, and crown-rump lengths between vehicle-and PBO-treated groups. No statistically significant differences were detected. PBO, piperonyl butoxide; SD, standard deviation; WT, wild type.
a Includes two litters with early resorbing embryos that were difficult to individually distinguish.  Note: WT C57BL=6J female mice were mated with Shh + = − male mice and administered the indicated doses of piperonyl butoxide (PBO) at gestational day (GD)7.75. At GD17, dams were inspected for live fetuses and resorptions, and fetuses were genotyped. One-way ANOVA was used to compare number of live fetuses, resorptions, and crown-rump lengths between vehicle-and PBO-treated groups of Shh + = + and Shh + = − fetuses. No statistically significant differences were detected. ANOVA, analysis of variance; Shh, Sonic hedgehog; SD, standard deviation; WT, wild type.
QImaging MicroPublisher 5.0 camera connected to an Olympus SZX-10 stereomicroscope. Area measurements of the MGEs and head were made in ImageJ. Measurements were made by a single rater blinded to treatment group.

Embryonic Gene Expression Analysis
For gene expression analysis, vehicle-exposed (n = 5) and 1,800 mg=kg PBO-exposed litters (n = 4) were collected at GD8.25. Somites were counted (Theiler 1989) and from all embryos with 7-10 somites anterior neural folds were isolated by microdissection in cold PBS. Isolated neural fold tissue was pooled for each litter and frozen immediately on dry ice. RNA was isolated using GE Illustra RNAspin kits (Catalog No. 25-0500-72; GE Healthcare Life Sciences) and stored at −80 C. cDNA was synthesized from 250 ng of total RNA using GoScript reverse transcription reaction kits (Catalog No. A2791; Promega). Singleplex qRT-PCR was performed using a Bio-Rad CFX96 real-time PCR detection system (Bio-Rad Laboratories) as described above for gene expression analysis of in vitro samples.

In Situ Hybridization and Riboprobe Synthesis
Riboprobes were synthesized by PCR as previously described (Abler et al. 2011). Briefly, sequence-specific cDNA was first amplified and affixed with a 3 0 T7 RNA-polymerase recognition site using a Taq DNA-polymerase kit (Catalog No. 201203; Qiagen), GD12 whole-embryo cDNA as a template, and primers targeting Gli1 [forward-CCCTCCTCCTCTCATTCCAC and reverse-CGATGTTAATACGACTCACTATAGGGTCCAGCTGA-GTGTTGTCCAG (italicized nucleotides denote the T7-polymerase recognition site and a 5-bp spacer sequence)]. Gene-specific primers were designed using the IDT PrimerQuest tool, and sequence specificity was confirmed using NCBI Primer-BLAST. Following PCR, Gli1-amplified cDNA was recovered by gel extraction using an E. Z.N.A. Gel Extraction Kit (Catalog No. D2500-01; Omega Bio-Tek). Next, digoxygenen-11-uridine-5 0 -triphosphate (UTP)-labeled riboprobes were synthesized from the Gli1 cDNA template using T7 RNA-polymerase (

Influence of PBO on Shh Signaling Pathway Transduction in Vitro
The influence of PBO on Shh signaling pathway transduction was examined with mammalian cell culture assays that have been used previously to discover and validate structurally diverse pathway inhibitors (Chen et al. 2002a(Chen et al. , 2002bLipinski and Bushman 2010). First, the concentration-response activity of PBO was tested relative to index pathway inhibitors cyclopamine and vismodegib. SHH LightII-3T3 cells were treated with recombinant human SHH ± graded concentrations of PBO, cyclopamine, vismodegib, or vehicle alone. Addition of SHH ligand resulted in significantly higher relative activity of GLI-responsive firefly luciferase compared with vehicle treatment ( Figure 1A). The difference between SHH-and vehicle-treatment groups was used as a baseline to determine the effect of drug exposure, calculated as percentage reduction in luciferase activity relative to SHH treatment alone. The addition of PBO, cyclopamine, and vismodegib each resulted in lower GLI-dependent luciferase activity in a manner that appeared concentration dependent ( Figure 1B). These values were used to establish concentration-response curves and approximate EC 50 values for each drug ( Figure 1B). Next, the effect of PBO on Shh signaling in human cells was examined in a Shh subgroup medulloblastoma (DAOY) cell line. DAOY cells were treated with SHH ligand ± graded concentrations of PBO or vehicle alone, and expression of the conserved Shh pathway gene GLI1 was measured relative to GAPDH. Addition of SHH ligand resulted in significantly higher GLI1 expression compared with the vehicle treatment ( Figure 1C). Compared with SHH treatment alone, addition of PBO resulted in lower GLI1 expression in a manner that appeared concentration dependent. No differences were observed in expression of the HPE-associated genes Bone morphogenetic protein 4 (BMP4) or SIX Homeobox 3 (SIX3) in response to SHH or PBO treatment (see Figure S2). Normally, SHH ligand binding to the Patched (PTCH1) receptor relieves its inhibition of SMO (Taipale et al. 2002), triggering downstream signal transduction that culminates in target gene regulation by the GLI transcription factors (Lipinski et al. 2006). Cyclopamine and vismodegib inhibit Shh signaling by directly binding to SMO (Chen et al. 2002a;Robarge et al. 2009). Cell lines with constitutive Shh pathway activity driven by genetic alterations at three unique signal transduction steps were utilized to examine the mechanism by which PBO impacts Shh signaling. Compared with WT cells, both Ptch1 null cells and WT cells expressing SMO M2 (a constitutively active form of SMO) or GLI1 exhibited higher expression levels of the conserved Shh pathway target gene Gli1 ( Figure 1D). In both Ptch1 null cells and WT cells expressing SMO M2 , PBO treatment resulted in lower Gli1 expression in a manner that appeared concentration dependent ( Figure 1E). In cells with GLI1 overexpression driving pathway activity downstream of SMO, no difference in Gli1 expression was observed between the PBO-and vehicletreatment groups. Finally, in Ptch1 null cells, we found that reduced Gli1 expression resulting from PBO exposure was partially rescued by the addition of SAG ( Figure 1F).

In Vivo Teratogenicity
Next, the teratogenic potential of PBO was examined by administering PBO acutely to timed-pregnant C57BL=6J mice by oral gavage at GD7.75 to target the critical period for induction of Environmental Health Perspectives 107006-5 127(10) October 2019 HPE (Heyne et al. 2015a;. Dams were euthanized at GD17, and uteri were carefully inspected. The number of resorptions and live fetuses in each litter, as well as mean crown-rump length were calculated (Table 1). No significant differences in these values were detected in any of the treatment groups. In 2 of 5 dams administered 1,800 mg=kg PBO, only resorbed embryos were observed, resulting in a modest and not statistically significant lower average number of live fetuses per litter in this treatment group. In all other PBO-dose groups (22:22-600 mg=kg), the average number of surviving fetuses per litter was similar to that of the vehicle-exposed group. Facial morphology was examined in GD17 fetuses by measuring ULL and IOD (Figure 2). These measurements have been used to assess midline morphogenesis and found to be useful predictors of underlying HPE-associated brain malformations (Heyne et al. 2016;Kietzman et al. 2014;Lipinski et al. 2012). Compared with the vehicle group, PBO exposure resulted in smaller IOD measurements. This appeared to be a dosedependent effect, with a lowest observable effect level (LOEL) of 66:67 mg=kg ( Figure 2G). The effect of PBO exposure on ULL was more variable, although significantly higher ULL measurements were observed in the 200 and 1,800 mg=kg PBO-exposure groups relative to the controls ( Figure 2H). Increased ULL has been shown to reflect median tissue deficiency and loss of the midline upper lip notch (Lipinski et al. 2012). No other gross structural malformations in PBO-exposed fetuses were noted upon visual inspection.

Correlative Face-Brain Dysmorphology
Next, qualitative assessment of correlative face and brain dysmorphology was conducted on GD17 fetuses exposed to vehicle; 200, 600, or 1,800 mg=kg PBO; or 40 mg=kg vismodegib (as a positive control) (Figure 3). In PBO-exposed fetuses, smaller IOD and larger ULL corresponded with increasingly severe median forebrain deficiency ( Figure 3B-E, 3B 0 -E 0 ). Relative to vehicle-exposed controls, PBO-exposed animals with subtle facial dysmorphology exhibited smaller forebrain septal region areas ( Figure 3A 0 -C 0 ), whereas those with severe facial dysmorphology exhibited overt HPE, illustrated by the absence of the septal region and a single forebrain ventricle ( Figure 3D 0 -E 0 ). Marked telencephalic hypoplasia and olfactory bulb aplasia were also observed in severely affected PBO-exposed animals (see Figure S3). As described previously (Heyne et al. 2015a(Heyne et al. , 2016, All individual data points are plotted along with bars representing the mean ± SD for each group. Filled circles in G and H indicate specific fetuses shown in A-F for the corresponding treatment group. One-way ANOVA with Tukey's post hoc test was used to compare the statistical significance of IOD and ULL measurements in PBO-exposed groups relative to the vehicle-exposed group. * p < 0:05; *** p < 0:001; **** p < 0:0001. Scale bars in A and A 0 : 1:0 mm. Note: ANOVA, analysis of variance; GD, gestational day; PBO, piperonyl butoxide; SD, standard deviation. fetuses exposed to 40 mg=kg vismodegib exhibited overt face and forebrain dysmorphology that was characteristic of HPE and similar to that observed in overtly affected fetuses exposed to 1,800 mg=kg PBO ( Figure 3F and 3F 0 ).
The pathogenesis of PBO-induced HPE was next examined by comparing GD11 vehicle-exposed control embryos to those from dams exposed to 1,800 mg=kg PBO at GD7.75. Overt malformations of the developing face and brain were evident at GD11, with affected PBO-exposed embryos exhibiting partially or completely fused telencephalic vesicles and closely spaced or fused nostrils ( Figure 3G-I). To examine the initial molecular pathogenesis of PBO-induced HPE, litters from dams exposed to vehicle or 1,800 mg=kg PBO at GD7.75 were harvested 12 h after exposure at GD8.25. Embryonic forebrain tissue was isolated by microdissection of the prosencephalic neural folds (dashed outline in Figure 3J) and harvested for mRNA. Compared with the vehicle-exposed group, PBO exposure resulted in significantly lower Gli1 expression, but no difference in expression of the HPE-associated genes Bmp4 and Six3 was detected ( Figure 3K).

Gene-Environment Interactions in HPE Genesis
The influence of normally silent single-allele mutations in the Shh gene was examined by administering PBO to timed-pregnant WT C57BL=6J female mice mated to Shh + = − male mice. This experimental paradigm was chosen to specifically isolate the variable of Shh heterozygosity. Mice harboring the Shh null allele were backcrossed to the C57BL=6J strain for more than 15 generations, and the mating strategy generated litters with both Shh + = + and Shh + = − embryos at equal proportion. Litters were harvested at GD17, and the number of live fetuses and resorptions, as well as mean crown-rump measurements were calculated (Table 2). No significant differences in these values were observed between dose or treatment groups. Facial morphology was examined as previously in WT trials ( Figure 4A-H). No difference was observed in ULL or IOD between vehicle-exposed Shh + = + and Shh + = − fetuses ( Figure 4I-J). However, at all three PBO doses (33.3, 100, or 300 mg=kg), Shh + = − fetuses exhibited more severe facial dysmorphology (lower IOD and higher ULL measurements) than WT fetuses ( Figure 4I-J).

Subteratogenic Impact on Forebrain Development
Finally, the impact of acute prenatal PBO exposure on morphogenesis of the MGEs was examined because development of these transient forebrain structures has been shown to be Shh dependent (Kohtz et al. 1998;Xu et al. 2005). MGE morphology was examined in embryos exposed to vehicle or PBO by measuring the area of MGEs relative to total head size ( Figure 5A,B). PBO exposure resulted in smaller normalized MGE areas, the size of which appeared to be PBO dose dependent with an LOEL of 66:67 mg=kg ( Figure 5B-G 0 ). Severe disruption of MGE development in the 600 mg=kg PBO-exposure group was comparable to that observed in embryos exposed to 15 mg=kg vismodegib (see Figure S4).

Discussion
Identification of environmental agents that contribute to birth defect etiology but can be avoided during pregnancy through risk communication offers a remarkably direct path to prevention. To the best of our knowledge, this report is the first to mechanistically investigate the developmental toxicity of PBO by specifically targeting the Shh signaling pathway during the critical period for HPE induction. We found that acute PBO exposure at Figure 3. Assessment of corresponding face and brain morphology and Gli1 expression in PBO-exposed embryos and fetuses. (A-F) Midfacial morphology is shown in representative GD17 fetuses exposed to vehicle (olive oil), the indicated doses of PBO, or 40 mg=kg vismodegib. Images taken before (D-F) or after fixation (A-C) show facial dysmorphology. (A 0 -F 0 ) Coronal sections from mice in A-F show forebrain abnormalities co-occurring with facial dysmorphology involving the forebrain septal region(s) and lateral ventricles. (G-I) Light microscopy images of unstained representative GD11 embryos exposed to vehicle or 1,800 mg=kg PBO show abnormalities involving the telencephalic vesicles (tv) and nostrils (arrows). (J) A GD8.25 embryo stained by ISH to visualize Gli1 expression is shown to illustrate the anterior neural fold region (dashed outline) that was microdissected from GD8.25 embryos exposed to vehicle or 1,800 mg=kg PBO. (K) Gene expression analysis by qRT-PCR of Gli1, Bmp4, and Six3 relative to Gapdh. Values represent the mean ± SE of n = 5 vehicletreated litters and n = 4 PBO-treated litters. Two-tailed t-tests with Holm-Sidak correction were used to compare the statistical significance of gene expression in the PBO-treatment group to the vehicle group. ** p < 0:01. Scale bars in A and A 0 : 1 mm, and in G and J: 0:5 mm. Note: GD, gestational day; ISH, in situ hybridization; PBO, piperonyl butoxide; qRT-PCR, quantitative real-time polymerase chain reaction; SE, standard error. GD7.75 in the mouse, equivalent to 3 weeks postconception in human development, disrupted development of the forebrain and face. The severity of PBO-induced outcomes was dose dependent and ranged from subtle to overt facial dysmorphology, corresponding with forebrain abnormalities that ranged from mild deficiency to the severe midline loss of full-blown HPE. Highlighting the potential for gene-environment interactions, we observed that the teratogenic effects of PBO were exacerbated by normally silent, single-allele Shh mutations that model common human HPE-associated mutations.
Unlike pyrethrin insecticides themselves, the potential developmental toxicity of PBO has received little attention. Previous examinations of PBO's teratogenic potential were untargeted and produced inconsistent results. A study of PBO exposure in pregnant rats published in 1952 reported that chronic dietary PBO administration adversely affected reproduction (Sarles and Vandegrift 1952). Two subsequent teratogenicity studies in rats were apparently negative, but high doses were reported to cause embryonic loss (Kennedy et al. 1977;Khera et al. 1979). In these studies, daily acute PBO exposure encompassed the critical period for HPE, but whether the phenotypic analysis was adequate to detect HPE-like phenotypes in surviving fetuses or offspring is unclear, and neither images nor craniofacial morphometric analyses were presented. A more recent examination of PBO teratogenicity in CD-1 mice found that a single maternal oral dose of PBO (as high as 1,065 or 1,800 mg=kg) administered at approximately GD9 resulted in forelimb digit reduction (Tanaka et al. 1994). Shh signaling is known to regulate limb development and digit patterning (Riddle et al. 1993), and the limb malformations observed following PBO exposure mimic outcomes we previously described in mice exposed to the potent Shh pathway inhibitor vismodegib at the same critical period of forelimb development (Heyne et al. 2015a). Although the study by Tanaka et al. (1994) did not examine the critical period for HPE (Heyne et al. 2015a), the finding that PBO resulted in phenotypically similar developmental limb abnormalities supports the hypothesis that in utero PBO exposure can cause structural malformations by inhibiting Shh signaling.
PBO was discovered as a Shh pathway inhibitor in a small molecule screening assay of over 1,400 toxicants (Wang et al. 2012). This original report found that PBO inhibited Shh signaling in vitro and was capable of competitively displacing radiolabeled cyclopamine, which inhibits the pathway by binding to the signal transducing protein SMO (Chen et al. 2002a). Here, we provide evidence suggesting that PBO inhibits Shh signaling in vitro in both mouse and human cells ( Figure 1A-C). Consistent with acting directly on SMO, the results from our in vitro experiments suggest that PBO acts downstream of PTCH1 and upstream of GLI1 ( Figure 1D-E) and that inhibition of Shh signaling transduction by PBO can be rescued by the addition of a SMO agonist ( Figure 1F). Concentration-response comparisons suggested that PBO was approximately 4-fold less potent than cyclopamine, a plant alkaloid that was discovered as a teratogen causing HPE in the offspring of grazing sheep (Keeler 1978). PBO was found to be approximately 50-fold less potent than the highly potent and specific synthetic Shh pathway inhibitor vismodegib. We also found that exposure to vismodegib or PBO at the same time in development causes Figure 4. Evaluation of facial dysmorphology in PBO-exposed Shh + = + and Shh + = − fetuses. (A-H) Shh + = + and Shh + = − GD17 fetuses exposed to vehicle (olive oil) or indicated doses of PBO at GD7.75. (I,J) Interocular distance (IOD) and upper lip length (ULL) measurement for vehicle-exposed Shh + = + (n = 26) and Shh + = − (n = 22) fetuses, 33:33 mg=kg PBO-exposed Shh + = + (n = 21) and Shh + = − (n = 18), 100 mg=kg PBO-exposed Shh + = + (n = 22) and Shh + = − (n = 20), and 300 mg=kg PBO-exposed Shh + = + (n = 24) and Shh + = − (n = 24) fetuses. All individual data points are plotted along with bars representing the mean ± SD for each group. Filled circles in I and J indicate specific fetuses shown in A-H for the corresponding treatment group and genotype. One-way ANOVA with Tukey's post hoc test was used to compare the statistical significance of IOD and ULL measurements between genotype groups for each dose group. * p < 0:05. Scale bar: 1:0 mm. Note: ANOVA, analysis of variance GD, gestational day; PBO, piperonyl butoxide; SD, standard deviation; Shh, Sonic hedgehog. comparable malformations of the brain and face (Figure 3; see also Figure S4).
Although gene-environment interactions are widely invoked to explain the complex etiology of human HPE, research efforts have primarily focused on the genetic side of this equation. SHH is among the most common genes associated with HPE (Nanni et al. 1999) but phenotypic outcomes are highly variable, such that even within a single family pedigree, some mutation carriers can be severely affected and others apparently normal (Solomon et al. 2012). In mice, homozygous Shh null mutations have been described to cause severe HPE (Chiang et al. 1996) while heterozygous Shh null animals appeared grossly indistinguishable from their WT siblings (Kietzman et al. 2014). Here, we found that Shh heterozygosity, although not resulting in detectable differences alone, exacerbated the teratogenic impact of PBO exposure (Figure 4). The implications of this specific gene-environment interaction may extend even more broadly, as most other mutations linked to HPE have been reported to be heterozygous and associated with wide phenotypic variability (Roessler and Muenke 2010). Efforts to elucidate the multifactorial etiology of HPE should also consider interactions between PBO and other HPE-associated environmental factors such as maternal diabetes (Johnson and Rasmussen 2010), alcohol consumption (Hong and Krauss 2017), and exposure to other environmental and dietary Shh signaling inhibitors (Elamin et al. 2010;Lipinski et al. 2007;Lipinski and Bushman 2010).
In both animal models (Heyne et al. 2016;Kietzman et al. 2014) and clinical cases (DeMyer et al. 1964;DeMyer 1971;El-Hawrani et al. 2006) of HPE, the degree of facial dysmorphology has been shown to predict the severity of co-occurring brain abnormalities. True HPE, defined by incomplete division of the cerebral cortices, causes profound intellectual disability and motor impairment (Weiss et al. 2018). We found that acute PBO exposure resulted in phenotypes consistent with true HPE, marked by severe median forebrain deficiency, hypotelorism, and midface hypoplasia, as illustrated by mice exposed to 1,800 mg=kg at GD7.75 (Figure 3). We also observed that lower doses of PBO (in addition to higher doses) caused more subtle disruptions of forebrain development, including abnormal morphogenesis of Shh-dependent MGEs ( Figure 5). MGE-derived inhibitory gamma-aminobutyic acid-producing (GABAergic) cortical interneurons regulate multiple aspects of cognitive processing including emotion, self-control, learning, and memory (Le Magueresse and Monyer 2013). Dysfunction in these neurobehavioral domains has been linked to cortical interneuron deficiency in mouse genetic models and cortical interneuron imbalance is widely implicated in human neuropsychiatric disorders (Marín 2012). We have shown previously that perturbation of MGE development by transient Shh pathway Figure 5. Evaluation of MGE morphology in PBO-exposed embryos. (A-A 0 ) GD11 embryos exposed to vehicle (olive oil) or the indicated doses of PBO were hemisected at the sagittal midline (dashed line) to visualize the inside of the developing brain, including the MGE (dashed black outline in A 0 ). (B) Quantification of MGE area normalized to head area. Measurements of all embryos collected from litters exposed to vehicle (n = 5) or 22.22 (n = 3), 66.67 (n = 3), 200 (n = 3), or 600 mg=kg PBO (n = 3) are plotted along with bars representing the mean ± SD for each group. One-way ANOVA with Tukey's post hoc test was used to compare the statistical significance of normalized MGE measurements in PBO-exposed groups relative to the vehicle-exposed group. **** p < 0:0001. Filled circles indicate the specific embryos shown in C-G and C 0 -G 0 . (C-G) Frontal images of GD11 embryos exposed to vehicle or the indicated dose of PBO at GD7.75. (C 0 -G 0 ) Magnified images of hemisected embryos from C-G with the MGE indicated with the dashed black outline. Scale bars in C and C 0 : 0:25 mm. Note: ANOVA, analysis of variance; MGE, medial ganglionic eminence; PBO, piperonyl butoxide; GD, gestational day; SD, standard deviation. inhibition leads to sustained disruption of cortical interneuron migration and specification in the mouse (Ansen-Wilson et al. 2018). Patients with true HPE were found to exhibit a depletion of GABAergic cortical interneurons (Fertuzinhos et al. 2009), although the contribution of this particular deficit to the intellectual disability observed in such patient cohorts is unclear (Fertuzinhos et al. 2009).
The finding that PBO caused subtle forebrain abnormalities in the absence of overt facial dysmorphology may shed new light on human and animal model studies linking prenatal PBO exposure to neurobehavioral outcomes. An epidemiologic study found a significant inverse correlation between prenatal exposure to PBO and neurodevelopment in 36-month-old children (Horton et al. 2011). Interestingly, no correlation was found for exposure of the active pesticidal ingredient, suggesting an independent effect of PBO itself. Developmental exposure to PBO through maternal diet has also been shown to cause neurobehavioral deficits in mice. In these studies, the offspring of mice maintained on chow containing 0.01-0.09% PBO during gestation and lactation exhibited deficits in several tasks including exploratory and spontaneous behavior Inomata 2015, 2016). To the best of our knowledge, neither human (Horton et al. 2011) nor animal Inomata 2015, 2016) studies linking prenatal PBO exposure to neurobehavior deficits have examined a causative mechanism. The results reported here implicate disruption of Shh signaling as a plausible mechanism underlying the neurodevelopmental deficits linked to prenatal PBO exposure. Together, these findings support the premise that subteratogenic PBO exposure may result in subtle forebrain abnormalities in the absence of overt, easily recognizable facial dysmorphology.
Relative to active pesticidal components, the availability of empirical data informing our understanding of human exposure to PBO is limited. As a synergist, PBO is not included in the suite of pesticides examined in the National Health and Nutritional Examination Survey, and its developmental toxicity has not been assessed within the National Toxicology Program. In the present study, PBO was administered acutely to specifically target Shh signaling during the defined, narrow critical period for HPE. The environmental relevance of the doses that were found to impact brain and face development in mice is unclear because of the limited understanding of human PBO exposure and its pharmacokinetics in mice and humans. Administration by oral gavage was chosen because ingestion is likely a main exposure route of PBO. In addition to indoor dust, PBO consumption also occurs as food residue. PBO is used in food production and processing, including pre-and postharvest application on agricultural crops, treatment of livestock animals and premises, and treatment of commercial facilities where raw and processed food commodities are stored, processed, and prepared (U.S. EPA 2006). The recently emerging Zika health crisis portends increased PBO exposure to women of childbearing age given that insecticides containing PBO are among mosquito control products that may be used to prevent transmission of the Zika virus (U.S. EPA 2006). Because many PBO-containing products are applied as aerosol sprays or directly to the skin, human exposure to PBO likely also occurs through inhalation and dermal absorption. With reported logP values between 4.60 and 4.95 (U.S. EPA 2006; Tomlin and British Crop Protection Council 2000) PBO is lipophilic and may have the potential for bioaccumulation. However, to the best of our knowledge, the pharmacokinetics of PBO and its concentration range in human serum have not been rigorously examined.
In the present study, we provide evidence suggesting that the environmental toxicant PBO inhibits the Shh signaling pathway in vitro in mouse and human cells and that its targeted prenatal exposure in mice causes face and brain dysmorphology that is consistent with Shh pathway disruption and exacerbated by normally silent Shh mutations. By targeting a sensitive mechanism of action, this study defined an LOEL for PBO-induced developmental toxicity that is more than 30-fold lower than was previously recognized (U.S. EPA 2006; Tanaka et al. 1994). PBO exposure should be considered alongside other nongenetic HPEassociated factors, particularly in genetically sensitive populations. More broadly, these findings should prompt efforts to address major knowledge gaps in our understanding of human exposure to PBO and its potential role in the complex etiology of Shh signaling-related human birth defects and neurodevelopmental abnormalities.