Certain ortho-hydroxylated brominated ethers are promiscuous kinase inhibitors that impair neuronal signaling and neurodevelopmental processes

The developing nervous system is remarkably sensitive to environmental signals, including disruptive toxins, such as polybrominated diphenyl ethers (PBDEs). PBDEs are an environmentally pervasive class of brominated flame retardants whose neurodevelopmental toxicity mechanisms remain largely unclear. Using dissociated cortical neurons from embryonic Rattus norvegicus, we found here that chronic exposure to 6-OH–BDE-47, one of the most prevalent hydroxylated PBDE metabolites, suppresses both spontaneous and evoked neuronal electrical activity. On the basis of our previous work on mitogen-activated protein kinase (MAPK)/extracellular signal-related kinase (ERK) (MEK) biology and our observation that 6-OH–BDE-47 is structurally similar to kinase inhibitors, we hypothesized that certain hydroxylated PBDEs mediate neurotoxicity, at least in part, by impairing the MEK–ERK axis of MAPK signal transduction. We tested this hypothesis on three experimental platforms: 1) in silico, where modeling ligand–protein docking suggested that 6-OH–BDE-47 is a promiscuous ATP-competitive kinase inhibitor; 2) in vitro in dissociated neurons, where 6-OH–BDE-47 and another specific hydroxylated BDE metabolite similarly impaired phosphorylation of MEK/ERK1/2 and activity-induced transcription of a neuronal immediate early gene; and 3) in vivo in Drosophila melanogaster, where developmental exposures to 6-OH–BDE-47 and a MAPK inhibitor resulted in offspring displaying similarly increased frequency of mushroom-body β–lobe midline crossing, a metric of axonal guidance. Taken together, our results support that certain ortho-hydroxylated PBDE metabolites are promiscuous kinase inhibitors and can cause disruptions of critical neurodevelopmental processes, including neuronal electrical activity, pre-synaptic functions, MEK–ERK signaling, and axonal guidance.

Humans around the world face exposure to a vast number of anthropogenic environmental contaminants that are largely untested for safety, many in the form of industrial chemicals. Use of one such class of environmentally-widespread flame retardants, polybrominated diphenyl ethers (PBDEs), 2 began in the 1970s in response to regulatory efforts concerning fire-related safety of consumer products. They are now known to be stable in the environment and highly lipophilic, enabling collection in household dust and persistent bioaccumulation. The primary route of human exposure is ingestion and inhalation of such dustborne PBDEs (1,2), with dietary intake from seafood and dairy products accounting for much of the remaining exposures (3). Exposure levels are known to be high in the United States in particular, with infants and toddlers carrying the highest body burdens due to elevated intake rates from household exposures (4).
PBDEs have also been identified as one of a number of environmental pollutants that are considered developmental neurotoxins (5). Concerningly, a substantial amount of research has established the relationship of PBDE exposure levels with behavioral deficits in humans, as well as in animal models. Recently, several large-scale and systematic reviews have concluded that PBDE exposures impact externalizing behaviors and IQ in children and that BDE-47, -99, and -209 affect learning in animal studies (6 -9). There is also concern for the relationship between environmental toxins like PBDEs and autism spectrum disorders, although the evidence is less conclusive, especially in human studies (6,(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20). Given the prevalence of these compounds and their association with behavioral deficits, it is imperative to understand the molecular and cellular mechanisms underlying PBDE neurotoxicity.
Exposure to PBDEs has been shown to have numerous effects on neural cells, which are thoroughly reviewed elsewhere (21).
Several of the major known mechanisms of PBDE toxicity include the following: 1) interference with endocrine signaling, likely caused by structural similarities to thyroid hormones and other hormone receptors' ligands (22)(23)(24)(25); 2) disruptions of calcium homeostasis/signaling as shown in one of the earliest mechanistic studies of PBDEs (26) and more recently demonstrated in several cell types, including human neuronal precursors (27,28); 3) toxicity produced by mitochondrial disruptions, including uncoupling of oxidative phosphorylation (29) and subsequent elevated production of reactive oxygen species that can lead to DNA damage and apoptosis; and 4) dysregulation of epigenetic mechanisms, including DNA methylation, chromatin dynamics, and noncoding RNAs-an emerging point of investigation that we have recently reviewed (30). It is possible that one or more of these molecular mechanisms contribute to exposure-related behavioral abnormalities; however, the extent of their involvement, if any, remains unclear.
The complex nature of the effects of these toxins is likely due in part to their diversity; there are 209 PBDE congeners, many found in commercial flame-retardant mixtures, and these compounds can be further endogenously metabolized in our cells. Such metabolism, mediated in humans by cytochrome P450 enzymes (31)(32)(33) and inherent in some other natural contexts (34 -37), leads to the production of hydroxylated and methoxylated forms of PBDEs. In addition to parent congeners, toxicity may also arise from these metabolic products of PBDEs, some of which (e.g. 6-OH-BDE-47) have been found to be more toxic than the parent compound (25,27). Given the numerous PBDEs to which humans are exposed, including these metabolized forms, understanding how the effects of exposures to various PBDEs impact neurodevelopmental processes has been a difficult process, and it is likely that unreported mechanisms are involved.
Previously, we have utilized the ability of neurons to induce gene transcription in response to elevated network activity as a tool for assessing developmental PBDE toxicity (38). Specifically, we assayed levels of the neuron-specific activity-induced immediate early gene (IEG) Arc after exposing embryonic cortical neurons across various stages of differentiation and maturation. Our rationale for employing this end point was that Arc induction relies on multiple levels of regulation, from electrical activity and synaptic function to intracellular signaling (particularly Ca 2ϩ -dependent mechanisms) and to nuclear regulation of gene expression, all of which are critical processes intertwined in regulating neurodevelopment (39). These processes are largely hardwired by genetic programs but are also modulated by environmental cues. Exposure to toxins like PBDEs represents an environmental challenge that may dysregulate these processes at multiple levels, the cumulative effects of which could be detected as altered transcriptional responses. Results from our previous study showed that chronic exposures to a hydroxylated metabolite of BDE-47, 6-OH-BDE-47 (6-OH), dysregulate basal and activity-dependent neuronal transcription. Among those findings were data that strongly indicated that the attenuated transcriptional response following 6-OH exposure is due to synaptic dysfunction and that 6-OH is potentially capable of inhibiting intracellular signaling. Either could explain the attenuated transcription of Arc, as its activity-induced transcription relies on synaptic activity as well as signaling via the MAPK cascade, one of the major signaling pathways that is important for neuronal plasticity (40) and is necessary for Arc induction (41).
In this study, we propose and test a novel mechanism of PBDE metabolite toxicity via impairment of MEK-ERK signaling. The notion that a PBDE metabolite could have the capacity to impair MAPK signaling was of particular interest to us for several reasons. The MAPK pathway is a deeply evolutionarilyconserved intracellular signaling mechanism that supports diverse processes in both prokaryotes and eukaryotes (42)(43)(44). In prokaryotes, PBDE-mediated inhibition of MAPK-homologous kinases could explain the natural biogenesis of these compounds as a defense mechanism, known to have antimicrobial activity and to be produced by bacteria in several marine contexts, including sponge-cyanobacterial symbiotic relationships (34 -37, 45-48). In eukaryotes, the most well-known role of MAPK is regulation of cell growth, division, and differentiation in many cell types, including stem cells in the developing nervous system (49 -51). It is also known to regulate axonal guidance and growth that have been demonstrated in multiple species (52)(53)(54)(55). Additionally, there are numerous neurodevelopmental disorders (NDDs) that are genetic syndromes involving the dysregulation of Ras and the MAPK pathway, thus known as "RASopathies" (56,57). In mature neurons-postmitotic nondividing cells-the pathway regulates synaptogenesis (58,59), synaptic plasticity (60), activity-induced gene expression (41,61), and generally communicates information to a large number of cellular substrates. These roles in neural cells make MAPK signaling a concerning target for environmental disruption. However, the MAPK pathway remains largely unexplored as a target for pollutants.
With these factors in mind, we sought to expand and clarify our previous findings with the studies reported here. Here, we show that chronic 6-OH exposure suppresses neuronal electrical activity, that 6-OH is a promiscuous kinase inhibitor, and that acute exposures can inhibit MEK-ERK signaling. We also report that in vivo exposures to 6-OH and a MEK inhibitor, PD0325901, similarly alter axonal guidance in the mushroom bodies of Drosophila melanogaster. These new findings corroborate our earlier work, provide evidence for the novel concept that certain hydroxylated PBDEs are promiscuous kinase inhibitors potentially capable of disrupting cellular processes via a range of intracellular signaling mechanisms, and further demonstrate the ability of PBDE metabolites to disrupt neurodevelopmentally-relevant neuronal processes.

Chronic 6-OH-BDE-47 exposure suppresses electrical activity in cortical neurons
We have previously shown that chronic exposure to nanomolar concentrations of 6-OH-BDE-47 (6-OH) impairs the ability of primary cortical neurons to induce transcription of the IEG Arc (38). Findings from that work indicated that the effects of the exposure were manifested at the level of synaptic functionality. To test this hypothesis directly, we utilized the same cell culture system but grew cells on microelectrode Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling arrays (MEAs). Cells were exposed to 500 nM 6-OH from the day of plating, and spontaneous electrical activity of the networks was recorded daily (Fig. 1A). There were no obvious morphological differences in cell growth. We also previously demonstrated that this exposure level does not significantly impact cell viability (38). Spontaneous activity was detected usually starting at 6 days in vitro (DIV), which is consistent with previous reports for cultured cortical neurons (62), and continued to the end of the experiment at DIV14. Networks exposed to 6-OH displayed significant suppression of spontaneous spiking across the recording period, which was quantified and is summarized in Fig. 1B. To further clarify the synaptic nature of the exposure effects, we isolated synaptosomes from 6-OHtreated cultures (DIV0 -10) and evaluated the levels of both pre-and post-synaptic protein markers by Western blotting. We found a significant decrease in the amount of detectable Synapsin1, but not several other pre-synaptic proteins or the post-synaptic proteins PSD-95 and gephyrin (Fig. 1C). This suggests that the observed activity deficits induced by chronic 6-OH exposure may be specifically mediated by pre-synaptic dysfunction involving synapsin, although effects unrelated to the markers assayed here cannot be ruled out. Together, these data support the hypothesis generated from our earlier work that chronic exposure to the BDE-47-hydroxylated metabolite, 6-OH, impairs synaptic function.
Next, we observed the effects of 6-OH on induced neuronal activity. This was done to partly explain one of our previous observations where activity-induced expression of the IEG Arc was significantly reduced in networks chronically exposed to 6-OH (38). To test whether chronic 6-OH exposure impaired firing patterns during induced activity, which could explain impaired Arc transcription, we evoked activity in unexposed and 6-OH-exposed networks using two different bicuculline (Bic) stimulation paradigms ( Fig. 2A). The first was continuous treatment with Bic, similar to the approach we previously used to induce Arc transcription (38). The other was a Bic stimulus followed by washout, a treatment paradigm known to produce long-lasting recurrent synchronous bursting that has been previously published as a model of in vitro plasticity (63). Both types of Bic treatments produced characteristic burst firing patterns (Fig. 2B) that were diminished in 6-OH-treated cells (Fig. 2, C-F). The observed reduction in activity after 15 min of continuous Bic treatment (Fig. 2, C and D) directly explains our earlier finding that chronic 6-OH exposure impairs activity-induced Arc transcription due to reduced firing of exposed neurons. The strong effect Primary rat cortical neurons were grown on MEAs in order to assess the effects of 6-OH exposure on electrical activity. A, example images of neurons at DIV10 with or without exposure to 500 nM 6-OH (left), and example traces of recorded spontaneous activity (right). B, detectable activity was measured daily with 3-min recordings and quantified through the first 2 weeks of growth in vitro, n ϭ 3-4. p values were generated with two-way ANOVA with post hoc FDR method. Time F(8,49) ϭ 2.09, p ϭ 0.054, treatment F(1,49) ϭ 36.04, p Ͻ 0.0001, interaction F(8,49) ϭ 1.503, p ϭ 0.1808. C, synaptosomes were isolated from DIV10 cultures. Pre-and post-synaptic markers were assessed by Western blotting (left) and densitometric quantification of blots (right) (note: the band between the Gephyrin and Synaptophysin1 is an additional band from the Gephyrin antibody), n ϭ 3-5. p values were generated with one-sample t tests using a hypothetical mean of 1.

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
seen under the Bic washout paradigm (Fig. 2, E and F) further demonstrates the compromised synaptic functionality of 6-OH-treated cells and suggests that 6-OH may specifically impair the molecular mechanisms regulating induction and maintenance of Bic-induced bursting.

6-OH-BDE-47 is a promiscuous kinase inhibitor
The report that previously published the Bic washout paradigm as a model of in vitro plasticity also demonstrated that the effect was regulated by MAPK signaling, and could be blocked with a pharmacological MEK inhibitor (63). We additionally found that chronically treating neurons with low concentrations of a similar MEK inhibitor, PD0325901 (PD), impairs spontaneous spiking as well (Fig. S1). Combined with our previous observation that 6-OH is capable of attenuating MAPKdriven gene transcription (38) and the observation that 6-OH shares key structural features with the same class of MEK inhibitors shown to regulate in vitro Bic-induced recurrent bursting (Fig. S2A), we hypothesized that 6-OH may be capable of directly inhibiting MEK1. To test this hypothesis and to gain insight into the plausibility of 6-OH acting as a small molecule kinase inhibitor, we conducted ligand-protein docking simulations using published crystal structures of MEK1 and 6-OH as well as a commercial inhibitor, PD (a type-III non-ATP competitive inhibitor), which are well-studied (64,65). Utilizing AutoDock Vina (66), which evaluated energetically-optimized binding poses for each ligand, we found that both 6-OH and PD were placed around the well-characterized allosteric binding pocket for this type of inhibitor, without providing prior information on the binding surface. Almost all of the generated top poses for both 6-OH and PD clustered around this surface when the simulation was initialized near the catalytic kinase domain (Fig. S2B). The poses for 6-OH and PD that are closest to the published binding modes of MEK1 inhibitors are superimposed in Fig. S2C. These results were encouraging and suggested that 6-OH is capable of realistically interacting with the type of binding surface known to be targeted by small molecules specifically designed to inhibit MEK1.
To our surprise, when we attempted to demonstrate direct MEK1-6-OH binding in a cell-free kinase-inhibition assay, there was no decrease in MEK1 activity across a range of 6-OH concentrations (Fig. 3B). However, when we screened a range of kinases, representing much of the human kinome, we found that 6-OH directly reduces the activity of a number of other kinases (Fig. 3A). In an attempt to identify shared features that enable the observed promiscuous inhibitory potential of 6-OH, further ligand-docking simulations were conducted with crystal structures of the top hits from the screen (Fig. 3C). Interestingly, the top 6-OH binding poses clustered largely in the wellknown ATP-binding pockets of each of the modeled kinases. This suggests that 6-OH may, in fact, be a promiscuous ATP-

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
competitive kinase inhibitor. Two of the top hits from the screen also suggest routes by which 6-OH may indirectly impair MEK-ERK signaling: Dyrk3, which is known to regulate sirtuin activation thereby potentiating ERK activation by insulin signaling (67-69), and intriguingly, CaMK1, which has been shown to regulate depolarization-induced MEK-ERK activity and ERK-regulated long-term potentiation (70 -73). Focusing on CaMK1, we found that acute 6-OH exposures completely attenuate phosphorylation levels of a well-known pre-synaptic CaMK1 target, synapsin1-serine 9 ( Fig. 3A) (74), providing in vitro confirmation of CaMK1 inhibition. We also reproduced an effect from a past report on CaMK1-MEK-ERK signaling, . Bars represent the mean percent activity of each kinase (screened in duplicate), and error bars represent standard deviation. The screening method utilizes a published radioactive filter-binding assay using [ 33 P]ATP (110,111) and was conducted by the International Centre for Kinase Profiling (ICKP) at the University of Dundee. Kinases with enlarged names are of particular interest for their relation to MEK-ERK signaling. B, IC 50 determinations were conducted by the ICKP (using the same methods as cited in A) for indicated compounds with either MEK1 or CaMK1. Error bars represent standard deviation of each data point, which were conducted in duplicate. No inhibition was observed for BDE-47 or 6-OH with MEK1, whereas the IC 50 for PD0325901 with MEK1 was detected here as 103 Ϯ 7 nM, and 6.82 Ϯ 0.02 M for 6-OH with CaMK1. It is worth noting that this IC 50 value for PD is around an order of magnitude higher than previously published, indicating that the estimations presented here may be affected by the sensitivity of this determination method. C, ligand-docking simulation results for the top five hits from the kinase screen. Crystal structures were obtained from the Protein Data Bank and are cited under "Materials and methods." Autodock-Vina was used to find energetically-optimal binding poses for 6-OH with each of the kinases, which are displayed in red in each panel. The last panel displays a structural alignment of the five kinases, with the docking results for CaMK1 superimposed. In each case, top binding poses clustered largely to the well-known ATP-binding pockets of the kinases. 6-OH acting as an ATP-competitive kinase inhibitor may help to explain the complex and diverse effects documented for PBDE exposures.

Specific ortho-hydroxylated PBDE metabolite exposures impair MEK-ERK signaling in vitro
In mature neurons, MEK-ERK signaling conveys information to many subcellular locations and substrates, including targets in synapses and the nucleus. To further test the prediction that 6-OH is capable of disrupting MEK-ERK signaling, we acutely exposed primary cortical neurons to BDE-47 and its hydroxylated metabolites and then stimulated them with a treatment of Bic and 4AP. The recurrent bursting driven by this treatment elevates intracellular calcium and induces the MEK-ERK-signaling pathway (driving gene expression and plasticityrelated events), which can be detected by measuring the phosphorylation status of both MEK and ERK. In whole-cell extracts, we found that activity-induced elevation of both pMEK and pERK was significantly reduced only in cells treated with 6-OH ( Fig. 4B), but not the parent compound or its other hydroxylated metabolites. The observed attenuation of pMEK induction is further evidence that 6-OH exerts effects somewhere above MEK in the signaling cascade, which includes CaMK1 as a possibility. As CaMK1 inhibition by 6-OH should only impact depolarization-induced MEK-ERK activation, we assessed the ability of 6-OH-exposed cells to induce pERK when the activation was driven synapse-independently. To do this, we also treated cells with TTX and PMA. This treatment, which is previously described (38) and depicted in Fig. 5A, silences propagation of neuronal activity while activating MEK-ERK signaling through protein kinase C (PKC) (which we knew was not inhibited by 6-OH from the kinase screen data). Again, we found that acute exposure only to 6-OH significantly attenuated pERK induction (Fig. 4C), indicating that 6-OH likely also impairs MEK-ERK signaling via synapseindependent mechanisms at high concentrations, although to a lesser extent than synapse-driven signaling. Inhibition of a synaptic signaling mechanism, such as CaMK1, could explain the stronger effect seen on synapse-dependent MEK-ERK induction.
Although strong inhibition of MEK-ERK induction was observed with higher 6-OH concentrations that are consistent with the IC 50 determination for CaMK1 (Fig. 3B), the lowest exposure level used (500 nM) was ineffective. The 500 nM dose was of interest to us as we have previously estimated it to be the approximate brain concentration reported from in vivo exposure studies of PBDEs in rodents (38) and is on the same order of magnitude as the Environmental Protection Agencyreported average for human exposure (4). As we initially did not see a significant decrease in pMEK/ERK levels with this environmentally-relevant concentration, we exposed cells to 500 nM 6-OH with increasing pre-treatment lengths to test the impact of providing more time for membrane penetrance and intracellular accumulation. This consistently yielded qualitatively mild but statistically insignificant inhibition of pERK in whole-cell extracts (Fig. S3A). Considering the subcellular location-specific functions of the MEK-ERK pathway, we next pre-pared nuclear extracts from neurons treated with 500 nM 6-OH to observe only the amount of pERK entering the nucleus (Fig.  S3B). Again, we observed mild reduction in pERK levels after acute exposure that was insignificant after quantification. At this point, it appeared that acute 500 nM 6-OH exposures are not sufficient to detectably inhibit high pERK levels induced by the strong Bic ϩ 4AP stimulus. To test the effect of stimulus strength, we activated neurons with 5 M Bic without 4AP. With this weaker stimulus, we found that acute exposure to 500 nM 6-OH significantly decreased nuclear levels of pERK (Fig. S3C).
Next, we measured pre-mRNA levels of the MEK-ERK-dependent and neuron-specific IEG Arc after acute PBDE exposure as a second approach to corroborate the above-mentioned pERK induction data. We acutely exposed neurons to BDE-47 and its hydroxylated metabolites as in Fig. 4B and found significant attenuation of Arc induction only after 6-OH exposure (Fig. 5B). We also found that acute exposure to 6-OH significantly attenuated Arc induction-driven synapse independently (Fig. 5C). Again, the inhibitory effect of 6-OH was stronger on synapse-dependent induction, with significant attenuation of Arc induction at all tested doses. With the synapse-independent treatment, the effect was again insignificant with a 500 nM exposure, as with the whole-cell pERK data (Fig. 4), so we again examined whether the strength of the stimulus impacts the extent of inhibition by titrating the concentration of PMA with which the cells were treated. As expected, we found that with a lighter stimulus the extent of 6-OH-induced inhibition was greater, and with Arc the pre-mRNA levels were significantly reduced when 500 nM 6-OH-exposed cells were stimulated with 0.01 M PMA (Fig. 5D). Taken together, these data indicate that acute 6-OH exposure impairs MEK-ERK signaling in neurons in dose-and stimulus strength-dependent manners and suggest that signaling in various subcellular locations may be differentially impacted.
To test the generality of our hypothesis regarding how hydroxylated PBDEs like 6-OH can impair MEK-ERK signaling, we acutely exposed cells to several other ortho-hydroxylated PBDE metabolites, whose possession or lack of the substituents hypothesized to mediate kinase inhibition are depicted in Fig. 6A. These include BDE-99, the second most environmentally-prevalent PBDE to which humans are exposed (4). It was surprising to find that only one of these ortho-hydroxylates metabolites, 6-OH-BDE-99, reduced pERK induction following Bic treatment (Fig. 6B). This metabolite produced nearly an identical effect on Arc induction as 6-OH-BDE-47 at various concentrations (Figs. 5B and 6C). This result suggests that for efficient MEK-ERK impairment, ortho-hydroxylated PBDE metabolites require a para-substituted halogen on the same ring as the hydroxyl group (Fig. 6A), potentially explaining why only certain ortho-hydroxylated metabolites impair elevation of pERK levels and subsequent Arc transcription.

Axonal guidance is dysregulated in vivo by both 6-OH-BDE-47 and PD0325901
Many developmental and functional processes in neurons are regulated by kinase activity, especially MEK-ERK signaling Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling (54,61,75). Specifically, it is known to be involved in axonal guidance, along with CaMK1 (73, 76,77), and that hydroxylated BDEs impact axonal growth in vitro (the evidence for which is covered in the discussion). We therefore hypothesized that developmental 6-OH exposures would lead to axonal guidance defects in vivo. To test this, we exposed adult flies (D. melanogaster) and their offspring to BDE-47, two of its hydroxylated metabolites, and PD. We then assessed axonal guidance in the brains of offspring by observing effects on mushroom-body ␤-lobe axonal midline crossing (see Fig. S4A for detailed experimental-timeline). The amino acid sequence of CaMK1 is highly conserved between Homo sapiens, Rattus norvegicus, and D. melanogaster, including key residues known to be involved in binding ATP-competitive kinase inhibitors (Fig.  7A) (78). In WT Drosophila, ␤-lobe axons are largely excluded from the midline of the mushroom body, but have been observed to have midline crossing defects at frequencies of ϳ20% (79). However, under conditions that disrupt normal axonal growth, increased frequency of midline crossing is observed for these axons (mild, moderate, and severe). The qualitative scheme for assessing the extent of midline crossing is depicted in Fig. 7B, and the frequencies of the different effects of severity are summarized in Fig. 7C. We found that 6-OH and PD exposure produced qualitatively similar increases in midline crossing, consistent with the involvements of both CaMK1 and MEK-ERK signaling in axonal growth. More mild effects were observed for BDE-47 and 5-OH-BDE-47 (5-OH) exposures. This became more evident when the distribution of crossing frequencies was rank-scored for each exposure and assessed quantitatively (Fig. 7D). Analysis of the rank score revealed that 6-OH and PD shift the median effect from "none" (as in the DMSO control) to "mild," although the parent compound and 5-OH did not produce a significant change. We also found that larval 6-OH exposure led to globally-reduced levels of ERK activation, measured as the fraction of pERK detectable in total ERK immunoprecipitates from larval tissue samples (Fig. S4B). These data indicate that 6-OH exposure can dysregulate axonal guidance in vivo and that the effect may be mediated via impairment of developmental CaMK1 and MEK-ERK signaling.

Discussion
PBDEs are well-known environmentally-pervasive toxins that are concerning for human health, especially considering the growing body of evidence linking their exposure levels to behavioral abnormalities related to NDDs. Much work has been done characterizing the effects and mechanisms of toxicity of PBDEs, but there is still not a clear general understanding of how these compounds compromise the developing nervous system and how the effects are related to adverse phenotypic outcomes. In this study, we propose a novel mechanistic hypothesis regarding how certain hydroxylated metabolites of common PBDEs (including BDE-47 and BDE-99) exert effects in neurons. We corroborate earlier findings indicating that exposure to an ortho-hydroxylated BDE-47 metabolite, 6-OH, impacts synaptic functionality, and we extend those findings by showing that 6-OH exposure affects neurodevelopmental processes in vivo. We also, for the first time, demonstrate that an ortho-hydroxylated PBDE metabolite is a promiscuous kinase inhibitor and that specific metabolites can impair neuronal MEK-ERK signaling.
From our previously published work (38), we had made two predictions related to the effects of chronic 6-OH exposure that are relevant to the data presented here: 1) effects in developing neurons likely manifest at the level of synaptic functionality, and 2) 6-OH may be capable of inhibiting intracellular signaling at or downstream of membrane depolarization and PKC. Primary cortical neuronal cultures grown on MEAs (Figs. 1 and 2) allowed us to directly test the first hypothesis. While some recent studies have begun to employ MEAs to study PBDE developmental neurotoxicity as part of larger screens (80 -82), the data presented here are, to our knowledge, the first example of such evidence for hydroxylated PBDE metabolite-induced effects. They demonstrate that chronic nanomolar 6-OH exposures suppress both spontaneous and evoked electrical activity and suggest that the exposures may specifically impact presynapses, including synapsin-related signaling and function. Inhibition of CaMK1 may interfere with neurotransmitter vesicle release by preventing phosphorylation of a well-known presynaptic target, synapsin1-serine 9 (74). This phosphorylation event is known to promote release of sequestered vesicles for exocytosis at the pre-synaptic membrane. Subsequent ERK activation is also known to be involved in recycling of vesicles via further synapsin1 phosphorylation-dependent mechanisms (83,84). 6-OH-mediated CaMK1 inhibition and subsequent impairment of ERK induction could thus dysregulate pre-synaptic vesicle cycling, potentially explaining the observed effects of 6-OH on spontaneous electrical activity and synapsin protein levels (Fig. 1, B and C). These pre-synaptic defects and the pronounced effect of 6-OH on evoked activity under the Bic washout treatment paradigm, which has been previously published as a model of in vitro plasticity (63), may partially explain the reported relationship between learning deficits and BDE-47/99 exposure in animal studies (9), and it is known that their hydroxylated metabolites can be produced endogenously, primarily via CYP2B6, which is dynamically expressed in the brain (31)(32)(33)(85)(86)(87). A, top panel: synaptosomes were prepared from neurons acutely exposed (20 min pre-treatment) to 5 M 6-OH with and without Bic ϩ 4AP stimulus (10 min). Levels of the CaMK1 target pSynapsin1(serine 9) were assessed via Western blotting. n ϭ 2. Bottom panel: to reproduce a previously reported effect linking CaMK1 to MEK-ERK activity (70), neurons were acutely exposed (60 min pre-treatment) to the CaMKK inhibitor STO-609 prior to treatment with Bic ϩ 4AP for 10 min (indicated as ϩB10Ј). Inhibition of pERK induction was assessed by Western blotting and is quantified below. n ϭ 3, p value generated by unpaired one-tailed t test. B, cells acutely exposed (20 min pre-treatment) to BDE-47 or one of its hydroxylated metabolites were stimulated with Bic ϩ 4AP for 10 min before collecting whole-cell lysates. Activation of MEK-ERK signaling was assessed by Western blotting, measuring pERK and pMEK levels. Note, the lower band in the MEK panel is the ERK band visualized on the same blot. Quantification relative to total ERK and MEK is shown below representative blots. n ϭ 3-6, p values generated by one-way ANOVA with post hoc LSD. pERK quantification: treatment F(7,22) ϭ 9.822, p Ͻ 0.0001. pMEK quantification: treatment F(7,16) ϭ 7.805, p ϭ 0.0004. C, same experimental design as in B., but neurons were stimulated with TTX ϩ PMA to induce MEK-ERK signaling synapse-independently. n ϭ 3. pERK quantification: treatment F(7,16) ϭ 37.95, p Ͻ 0.0001. Investigation of the second prediction generated from our previous work was spurred on by several lines of evidence: our previous observation of disrupted MAPK-driven gene induction by 6-OH (38); the known role of MAPK signaling in the effects seen on Bic-induced neuronal activity (Fig. 2) (63); and the serendipitous observation that 6-OH shares key chemical features with a class of inhibitors designed for MEK1, a central kinase in the MAPK-signaling cascade (61,88). The structural similarities and results from ligand-protein docking simulations encouraged us to pursue 6-OH as a potential MEK1 inhibitor (Fig. S2A). Surprisingly, we came to find that 6-OH does not directly inhibit MEK1/2, but rather it appears to be a pro-miscuous ATP-competitive kinase inhibitor (Fig. 3). We believe this novel finding will be of use in generating future hypotheses and furthering the understanding of the complicated and diverse effects of the PBDE exposures documented to date. Here, we further pursued effects related to MEK-ERK signaling, particularly related to CaMK1, using multiple approaches to activate the pathway, mediated both synaptically and synapse-independently (Fig. 5A). We subsequently found that acute 6-OH exposures impair induction of MEK-ERK signaling and gene transcription (Figs. 4 and 5). Strong inhibition was found with 6-OH concentrations consistent with the data from the kinase screen (Fig. 3), whereas the lower, more environmen-

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
tally-relevant doses had milder effects, although still pronounced in the case of Arc transcription (Fig. 5). Additionally, we found that time-of-exposure does not appear to increase the extent of inhibition by these lower concentrations of 6-OH (Fig.  S3A) and that the strength of the stimulus driving MEK-ERK activation affects the extent of 6-OH-mediated impairment, with lighter stimuli leading to stronger impairment for both the synaptic and PMA-driven MEK activation (Fig. S3C and Fig.  5D). Overall, it is likely that acute exposures to environmentally-relevant levels of MEK-inhibiting ortho-hydroxylated metabolites are not sufficient to widely suppress MEK-ERK signaling throughout the cell (especially given the extensive feedback opposing perturbations), but they may chronically and regionally accumulate to high enough concentrations to inhibit various kinases regulating MEK-ERK signaling in specific subcellular locations, an interesting possibility that is further discussed below.
To generalize our findings, we screened several other orthohydroxylated metabolites of BDEs-68/99/123 and found that the metabolite of BDE-99 also impairs MEK-ERK signaling following acute exposure (Fig. 6). While initially surprising, these data indicate that in order for an ortho-hydroxylated PBDE metabolite to be effective at impairing MEK-ERK signaling, it also specifically requires an ortho-substituted halogen that presumably enables binding to susceptible kinases. Taken together, these several lines of molecular evidence demonstrate Figure 7. In vivo exposure to 6-OH-BDE-47 and the MEK inhibitor PD0325901 disrupt axonal guidance. Canton-S flies (D. melanogaster) were exposed to BDE-47, 5-OH, 6-OH, or PD via feeding. The mushroom bodies of offspring were then assessed for aberrant midline crossing. A, alignment of the amino acid sequences of human CaMK1 and its rat and fly orthologs. The proteins are highly conserved, including the key residues thought to mediate ATP-competitive inhibitor binding (highlighted in yellow). B, example images of the effect categories used in the qualitative scoring scheme. C, crossing frequency by severity level for each of the treatment groups, n ϭ 12-16. D, crossing severity was numerically ranked scored as follows. No crossing, 4; mild crossing, 3; moderate crossing, 2; severe crossing and/or lobe malformation, 1. Bars represent the median score of each treatment group that was generated from the values of the crossing data from C, after application of the numerical scoring scheme (the score for each fly is additionally plotted as gray circles). Error bars represent 95% confidence intervals. p values were generated using exact one-tailed Mann-Whitney tests to compare the median of each group with that of the DMSO control, n ϭ 12-16.

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
that specific ortho-hydroxylated PBDE metabolites can inhibit MEK-ERK signaling in vitro. This potential for acute MEK-ERK impairment, combined with chronic exposure-induced deficits in synaptic activity and composition, corroborates and further explains our previous findings concerning the ability of cortical neurons to induce Arc expression following chronic 6-OH exposure (38).
To validate our in silico and in vitro findings and extend them in vivo, we compared the exposure effects of 6-OH and PD using a neurodevelopmentally-relevant readout in fruit flies (D. melanogaster). Drosophila was chosen due to its amenability as an animal model for screening axonal guidance. Importantly, the Drosophila ortholog of CaMK1 is highly conserved, including the key amino acid residues that mediate binding of ATP-competitive inhibitors (Fig. 7A). Furthermore, it is known that CaMK1 and MEK-ERK signaling plays a role in developmental axon growth (89). One study reported that a let-60c mutation (a Caenorhabditis elegans ortholog of KRAS, a membrane GTPase upstream of MEK) led to axonal guidance defects manifested as aberrant ventral midline crossing (52). Another confirmed that MEK-ERK signaling regulates netrin-1-dependent axonal branching in cultured hamster sensorimotor cortical neurons (53). In our studies, exposure was accomplished via oral administration to the P1 generation-to ensure embryonic exposure-and also to F1 larva to cover the neurodevelopmental period of the mushroom-body formation. The brains of offspring were examined post-eclosion to assess the extent of mushroom-body ␤-lobe axonal midline crossing (see Fig. 4A for detailed timeline). Here, we found that 6-OH and PD exposures produced qualitatively similar increases in midline crossing, with BDE-47 and 5-OH also increasing crossing frequency, but to a lesser extent (Fig. 7C). After applying a numerical scoring scheme and quantifying the effects, we found that 6-OH and PD shift the median effect to "mild," whereas the parent compound and 5-OH did not produce a significant change. These data represent an in vivo confirmation of the effects of 6-OH exposure on axonal guidance (previously reported in primary hippocampal cultures (90)) and indicate that the effect may be mediated by disruptions to CaMK1 and MEK-ERK signaling. These findings align well with previous reports in C. elegans, where mutants of let-c, an ortholog of an upstream MAPKsignaling component, KRAS, exhibit aberrant axonal crossing of the ventral midline (52).
Although the evidence we present here demonstrates the promiscuous inhibitory potential of ortho-hydroxylated metabolites toward ATP-dependent kinases and MEK-ERK signaling, we note that such effects are highly nuanced and will require further investigation. MEK-ERK signaling is activated by a variety of stimuli in neurons and is a very dynamic process. Here alone we have identified CaMK1 and DYRK3 disruption as potential avenues of MEK-ERK impairment. Known modulation of Ca 2ϩ homeostasis by 6-OH further complicates interpretation of acute exposures and potentially explains the inhibition of pERK levels by 6-OH following the synapseindependent stimulus paradigm (Fig. 4C). Further examples of nuanced effects include cultures chronically exposed to 500 nM 6-OH were able to synaptically induce cytosolic pERK to the same extent as untreated controls, while significantly less pERK was available in the nucleus (Fig. S3, B and C). This was unexpected, especially given the compromised electrical activity observed for chronically-treated cells, but could be explained by the extensive homeostatic feedback mechanisms known to regulate MEK-ERK signaling (91). Such feedback mechanisms may also explain the discrepancy between our findings and those of another recent report that showed that exposures to 6-OH selectively impaired ERK5 phosphorylation, but not ERK1/2, in adult neural stem cells (aNSCs) (92). In this study, the authors incubated aNSCs overnight with 5 M 6-OH, followed by a stimulus with epidermal growth factor/basic fibroblast growth factor to induced MAPK, and they reported no decrease in the amount of pERK1/2. In addition to the data from this study showing that after prolonged treatment with 6-OH the cortical neurons induce the same level of pERK as untreated controls, it has been shown in mouse embryonic stem cells that acute PD treatment reduces pERK levels, but between 12 and 24 h later, the pERK signal strongly returns (93). Therefore, it is possible that either feedback within the MAPK pathways or differences in signaling dynamics in aNSCs, particularly those that activate MEK-ERK signaling, account for this discrepancy between our observation and that by Li et al. (92).
Interestingly, chronic 6-OH treatment-induced differences in cytosolic and nuclear pERK inhibition also suggest another nuanced aspect of PBDE toxicity: their subcellular localization and localized effects. Because of the highly-lipophilic nature of PBDEs, they are likely to be more concentrated within close proximity of lipid membranes. Supporting this notion, it has been demonstrated that several PBDEs localize in higher amounts in mitochondrial and microsomal fractions collected from cerebral granule neurons (94). Accordingly, it will be interesting and necessary to evaluate the potential of orthohydroxylated PBDE metabolites to inhibit kinases in specific subcellular locations, as well as MEK-ERK signaling events, which are also known to be localized in different cellular compartments, including the cytosol, mitochondria, outgrowing neurites, and synapses (95,96). Considering the effects on such compartmentalized signaling, inhibition may help to generally explain some of the various documented effects of exposure to certain PBDEs, especially well-documented mitochondrial toxicity (21). Furthermore, mitochondrial toxicity may link PBDE exposure to NDD etiology given the known comorbidity of mitochondrial deficits and autism (12). Pre-synaptic accumulation could also be involved in the strong effect seen on synapsin (Figs. 1C and 4A), as it and other synaptic proteins are known targets of CaMK1 and synapse-localized MAPKs (97,98). Both CaMK-and MAPK-dependent synapsin phosphorylation are also known to play a role in pre-synaptic plasticity (59,74,84,99). Local inhibition of CaMK1 and MAPK signaling and subsequent impairment of pre-synaptic vesicle regulation may explain the suppressed electrical activity caused by chronic 6-OH exposure (Figs. 1 and 2), although more work is needed for confirmation.
Taken together, the findings presented here provide new evidence on the effects of exposure to hydroxylated PBDE metabolites and establish novel insight into the underlying molecular mechanisms that mediate their toxicity. The indication that

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
PBDE metabolites are capable of impairing MEK-ERK signaling sets up several potential avenues for further investigation. One takes into account another aspect of our previously published work (38), i.e. 6-OH exposure dysregulates expression of BAF chromatin remodeling subunits. This evidence, along with a recent preprint report concerning ERK1/2 regulation of neurodevelopmental polycomb-repressive complex chromatin remodeling that further speculates about additional regulation of BAF complex composition (100), suggests that 6-OH exposure may interact with chromatin remodeling via modulation of MAPK signaling. More closely related to this study, it will be critical to investigate the role of ortho-hydroxylated PBDEs as promiscuous kinase inhibitors, their links to impairment of MEK-ERK signaling, and the many established routes of PBDE toxicity. This will be especially interesting for effects related to mitochondrial function, calcium homeostasis, and regulation of synaptic function. Overall, such future studies may help to clarify the growing evidence linking human PBDE exposures to NDDs and perhaps provide useful insight into these challenging disorders.

Primary neuronal culture and cell treatment
Timed-pregnant Sprague-Dawley rats were obtained from Charles River Laboratories. Dams were consistently delivered 2 days prior to dissection and housed individually on a 12-h light/ dark cycle with access to food and water ad libitum. On the day of dissection, dams were anesthetized with Euthasol euthanasia solution (pentobarbital 390 mg/ml, phenytoin 50 mg/ml) in order to minimize distress. Loss of responsiveness was ensured with the toe pinch method, confirming a lack of reflex prior to decapitation via guillotine. Primary cultures of cortical neurons were then prepared from embryonic day 18 pups (IACUC approval, AUP 16-0004, University of California, Merced). Cortical hemispheres were isolated and pooled from male and female embryos in magnesium-and calcium-containing HBSS (Gibco, catalog no. 14025092). They were then mechanically dissociated by pipetting with a fire-polished Pasteur pipette following a 7-min digestion with StemPro Accutase (Life Technologies, Inc., catalog no. A1110501). The enzyme was then deactivated by diluting the triturated cell suspension in HBSS lacking calcium and magnesium (Gibco, catalog no. 14175095). The dissociated cortical neurons were then pelleted by centrifugation, resuspended in plating media, counted using trypan blue and a TC20 TM automated cell counter (Bio-Rad, catalog no. 1450102), and subsequently plated by addition to dishes containing pre-warmed Neurobasal medium (Gibco, catalog no. 21103049) supplemented with 25 M glutamate (Sigma-Aldrich, catalog no. 1446600), 0.5 mM L-glutamine (Sigma-Aldrich, catalog no. G8540), and 2% NS21 supplement. Cells were maintained at 37°C in a humidified incubator with 5% CO 2 . They were grown in the medium described above without glutamate, replacing half the media every 3-4 days. NS21 was prepared in the laboratory as described previously (101). Neurons were used for various assays between 10 and 14 days in vitro. No blinding was performed for experiments from cultured cells. To induce MEK-ERK signaling and gene transcription using syn-aptic activity, neurons were co-treated with 50 or 5 M bicuculline (Sigma-Aldrich, catalog no. 14340) with or without 75 M 4-aminopyridine (Acros Organics, catalog no. 104571000). To induce gene transcription synapse independently, membrane activity was blocked with 1 M TTX (Calbiochem, catalog no. 554412), and the MAPK pathways were activated via PKC with 1 M to 1 nM phorbol 12-myristate 13-acetate (PMA; Sigma-Aldrich, catalog no. P1585). BDE-47 and hydroxylated BDE metabolites used to treat cultures were obtained from AccuStandard (BDE-047N, HBDE-4003N, HBDE-4004N,  HBDE-4005N, HBDE-4006N, HBDE-5006-N, and HBDE-5011N) and initially dissolved in dimethyl sulfoxide (DMSO) at 2 and 5 mM stock concentrations. Final treatment concentrations were produced at the time of treatment by dilution in culture medium. PD0325901 was obtained from AdipoGen (SYN-1059), dissolved in DMSO at 5 mM, and similarly diluted at the time of treatment.

Microelectrode array recording and data processing
Neurons from the preparations described above were plated on poly-L-lysine/laminin-coated MEAs (60MEA200/30-Ti, MultiChannel Systems (MCS), Reutlingen, Germany) in 500 l of the NS21-supplemented Neurobasal plating medium described above. When cells were fed, again every 3-4 days, approximately half the media was replaced with NS21-supplemented BrainPhys feeding media (StemCell). This was done to promote optimal conditions for neuronal firing, which has been shown to be enhanced in BrainPhys media (102). Recordings were made with an MEA2100-lite system that interfaces with MCS provided MultiChannel Experimenter software. Sampling was conducted at 10 -20 kHz in 3-min sessions at room temperature (arrays were covered to prevent contamination). Recording were post-processed in a MultiChannel Analyzer with a high-pass 1st order Butterworth filter with 100 Hz cutoff prior to generation of spike time-stamps. Spikes were detected by an automatic threshold estimator set to 6 -8 standard deviations from the baseline signal. To quantify burst properties, the MultiChannel Analyzer burst detection tool was used with the following settings: maximum interval to start burst, 25 ms; maximum interval to end burst, 250 ms; minimum interval between bursts, 1500 ms; minimum duration of burst, 50 ms; minimum number of spikes in burst, 5. Raster images from example recordings were generated in NeuroExplorer (RRID: SCR_001818, NexTechnologies, Herndon, VA).

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling
The simulations were run with standard parameters for a rigid receptor and ligands with rotatable bonds (108). The binding structures were evaluated visually based on comparison with the well-characterized binding modes of allosteric MEK inhibitors, the first example of which was published in 2004 (109), as well as with various ATP-competitive kinase inhibitors crystallized with some the of the other kinases.

RNA extraction and gene transcription quantitation
Total RNA was isolated from cultured neurons using the Illustra RNAspin mini kit (GE Healthcare, catalog no. 25050072). Specific pre-mRNAs from these total RNA samples were initially amplified by cDNA synthesis (14 cycles) using Arc and Gapdh primers overlapping an intron-exon junction and a OneStep RT-PCR kit (Qiagen, catalog no. 210212). Quantitative real-time PCR was then performed from this cDNA to quantify levels of specific transcripts using PerfeCTa SYBR Green FastMix (QuantaBio, catalog no. 95072-012) and the Bio-Rad CFX Connect real-time PCR detection system. Foldchange from control was estimated using the ⌬⌬Ct method.

Drosophila rearing and chemical exposure
WT Canton-S Drosophila were maintained at 25°C in a humidified incubator on a standard cornmeal diet (Bloomington Drosophila Stock Center standard cornmeal medium). No randomization was performed to allocate individuals in the various treatment groups. For oral administration of PD0325901, BDE-47, 5-OH, and 6-OH, compounds were dissolved in dimethyl sulfoxide (DMSO) and then added directly to warm food, prior to solidification. The concentration of DMSO was maintained at 0.02% for all exposures. Drosophila were exposed to PD (n ϭ 16) at a concentration of 100 nM. For all other chemicals (BDE-47 (n ϭ 12), 5-OH (n ϭ 14), and 6-OH (n ϭ 14)), Drosophila were exposed to a concentration of 1 M. Control flies were exposed to DMSO alone (n ϭ 14).

Fluorescence labeling and microscopy of Drosophila whole-mount brain preparations
Whole-mount immunostaining of adult male brains (0 -2 days old) was performed as described previously (79), except that brains were fixed for 25 min in 4% paraformaldehyde. Anti-Fasciclin II (RRID:AB_528235) was used at a 1:20 dilution. Alexa Fluor 488 -conjugated goat anti-mouse secondary (RRID:AB_2338840) was used at a 1:1000 dilution. Images were captured using an Olympus laser-scanning confocal microscope. Separate individuals, who were blinded to the exposure group and file-naming system, assessed the images for phenotypic scoring.

Immunoprecipitations (IPs) from Drosophila larval tissue
IPs were conducted by pooling the heads from 10 chemicallyexposed larvae. Tissue was lysed in ice-cold 1ϫ RIPA buffer (25 mM Tris-HCl, pH 7.5, 150 mM NaCl, 1% Na-deoxycholate, 0.1% SDS, 0.1% NP-40) supplemented with 1:100 protease/phosphatase inhibitor mixture (Thermo Fisher Scientific, catalog no. 78442). Lysate was added to a total volume of 1 ml in IP buffer (0.5% Triton X-100, 2 mM EDTA, 0.02 M Tris, pH 7.75, 150 mM NaCl, and 10% glycerol) with 4 g of total ERK antibody (CST, catalog no. 4696, RRID:AB_390780) and rotated for 2 h at room temperature. Protein A/G magnetic beads (Thermo Fisher Scientific, catalog no. 88803) were then added to the IPs, followed by rotation at room temperature for 1 h. IPs were then washed three times on a magnetic rack with 1 ml of wash buffer (0.5% Triton X-100, 2 mM EDTA, 0.02 M Tris, pH 7.75, 150 mM NaCl, 10% SDS), resuspending the magnetic beads between each wash. After the final wash, the isolated proteins were denatured at 95°C for 5 min prior to gel electrophoresis and Western blotting by the procedure described under "Western blotting, imaging, and antibodies."

Certain hydroxylated PBDEs impair neuronal MEK-ERK signaling Study design and statistical analysis
This study was not pre-registered. Data presented were generated from both hypothesis-confirming and exploratory experiments. For rat primary culture experiments, no sample size calculation was performed. For Drosophila experiments, sample size was estimated based on previously published studies that have examined the mushroom-body phenotype (79). There were no pre-determined exclusion criteria for animal work. Statistical analyses were conducted using GraphPad Prism 7 (RRID:SCR_002798, GraphPad software, San Diego, CA). Error bars represent standard error of the mean throughout, except in Fig. 7D where they represent 95% confidence intervals. Data were assessed for normality with the Shapiro-Wilk normality test. No tests for outliers were conducted; therefore all data points were included. Effects were determined by t test or one-/two-way ANOVA with appropriate post hoc tests for generation of specific p values; details are indicated in the figure legends. The only exception is Fig. 7D where the nonparametric Mann-Whitney U test was used to compare the medians of effect score distributions for the various treatment groups. Specific p values are indicated on the figures: * indicates p value Ͻ0.05 and ** indicates p value Ͻ0.01 throughout. Biological replicates are indicated throughout as N in corresponding figure legends. Biological replicates constitute cell culture preparations from the pooled cortices of embryos from independent litters or brains of individual flies.

Data availability
All data are included in the text. Raw datasets are available upon request.