A Targeted, Low-Throughput Compound Screen in a Drosophila Model of Neurofibromatosis Type 1 Identifies Simvastatin and BMS-204352 as Potential Therapies for Autism Spectrum Disorder (ASD)

Abstract Autism spectrum disorder (ASD) is a common neurodevelopmental condition for which there are no pharmacological therapies that effectively target its core symptomatology. Animal models of syndromic forms of ASD, such as neurofibromatosis type 1, may be of use in screening for such treatments. Drosophila larvae lacking Nf1 expression exhibit tactile hypersensitivity following mechanical stimulation, proposed to mirror the sensory sensitivity issues comprising part of the ASD diagnostic criteria. Such behavior is associated with synaptic dysfunction at the neuromuscular junction (NMJ). Both phenotypes may thus provide tractable outputs with which to screen for potential ASD therapies. In this study, we demonstrate that, while loss of Nf1 expression within the embryo is sufficient to impair NMJ synaptic transmission in the larva, constitutive Nf1 knock-down is required to induce tactile hypersensitivity, suggesting that a compound must be administered throughout development to rescue this behavior. With such a feeding regime, we identify two compounds from a targeted, low-throughput screen that significantly and consistently reduce, but do not fully rescue, tactile hypersensitivity in Nf1P1 larvae. These are the HMG CoA-reductase inhibitor simvastatin, and the BKCa channel activator BMS-204352. At the NMJ, both compounds induce a significant reduction in the enhanced spontaneous transmission frequency of Nf1P1 larvae, though again not to the level of vehicle-treated controls. However, both compounds fully rescue the increased quantal size of Nf1P1 mutants, with simvastatin also fully rescuing their reduced quantal content. Thus, the further study of both compounds as potential ASD interventions is warranted.


Introduction
Autism spectrum disorder (ASD) is a common neurodevelopmental condition affecting 1-2% of children (Baird et al., 2006;Maenner et al., 2020). Clinically, it is characterized by impairments in social communication alongside repetitive and restricted behavior and interests, which include altered responses to sensory stimuli (American Psychiatric Association, 2013). Such sensory impairments may directly contribute to other ASD symptoms and correlate with worsening outcomes on several quality-of-life measures. Thus, they may provide an important target for therapeutic intervention (Lundqvist, 2015;Robertson and Baron-Cohen, 2017;Lin and Huang, 2019). The need for such interventions is urgent, given the lifelong impact on an individual's quality of life that ASD can impose (Lord et al., 2020), and the substantial economic burden arising from the need to support those affected (Buescher et al., 2014;Leigh and Du, 2015). Currently, the only two compounds approved by the Food and Drug Administration for the treatment of behavioral symptoms in ASD are the atypical antipsychotics aripiprazole and risperidone, both of which are prescribed for irritability and aggression (Lord et al., 2020). Additional medications commonly prescribed to individuals with ASD include stimulant drugs for the treatment of comorbid ADHD, selective serotonin-reuptake inhibitors for mood disturbances and anxiety, and melatonin for sleep disorders (Aishworiya et al., 2022). However, these are often associated with adverse side effects ranging from the mild (e.g., weight gain, appetite changes, fatigue, and headaches) to the severe (e.g., metabolic syndrome; Aishworiya et al., 2022). Importantly, none have been shown to consistently and effectively improve abnormal sensory sensitivity or indeed other aspects of core ASD symptomatology (Hyman et al., 2020).
While ;75% of ASD cases are idiopathic, ;4-5% occur in association with a monogenic neurodevelopmental syndrome (Fernandez and Scherer, 2017). Because syndromic forms of ASD have a known single causative mutation, they are comparatively simpler and offer a more tractable model from a biomedical research perspective (Sztainberg and Zoghbi, 2016). One such condition is neurofibromatosis type 1, an autosomal dominant disorder arising from loss of function of the NF1 gene on chromosome 17 (Gutmann et al., 2017). The prevalence of ASD among individuals with neurofibromatosis type 1 is estimated at 10-25%, with up to a further ;20% exhibiting some form of clinically relevant symptomatology (Garg et al., 2013;Walsh et al., 2013;Adviento et al., 2014;Plasschaert et al., 2015;Morris et al., 2016;Eijk et al., 2018). Furthermore, it is becoming increasingly apparent that ASD and neurofibromatosis type 1 share an overlapping pathophysiology. Thus, animal models of neurofibromatosis type 1 possess significant potential in screening for novel therapies for ASD (Molosh and Shekhar, 2018;Kaczorowski et al., 2020).
An effective drug screen foremost requires a suitable model of the disorder. Target-based, in vitro approaches have not proven suitable for conditions like ASD in which the underlying disease biochemistry is poorly understood and likely involves the dysfunction of multiple pathways. They also do not routinely permit behavioral outputs as measures of drug efficacy (Pandey and Nichols, 2011;Strange, 2016). More traditional, preclinical model organisms, such as mice, are equally unfeasible, given their high cost of maintenance, long generation time, and relatively small litters (Bell et al., 2009). By contrast, the fruit fly, Drosophila melanogaster, overcomes many of these limitations and provides a useful bridge between the two model systems (Bell et al., 2009;Pandey and Nichols, 2011;Strange, 2016). The potential of Drosophila in neurodevelopmental drug discovery was exemplified by a screen of 2000 small molecules for their ability to rescue glutamate-induced lethality in a fly model of ASD-associated Fragile X syndrome (Chang et al., 2008). Indeed, identification of GABA as a hit compound lent support for the use of GABA-promoting drugs in clinical trials (Braat and Kooy, 2015), although the success of these has been mixed (Berry-Kravis et al., 2012;Erickson et al., 2013Erickson et al., , 2014Ligsay et al., 2017;Veenstra-VanderWeele et al., 2017).
Drosophila express a highly conserved ortholog of NF1  with similar molecular, cellular, and behavioral functions to its mammalian counterpart (Guo et al., 2000;Walker et al., 2006). Accordingly, Nf1 À/À flies display phenotypes analogous to ASD in the clinic, including altered communication (Moscato et al., 2020), disrupted sleep (Bai and Sehgal, 2015), and repetitive behaviors in the form of excessive grooming (King et al., 2016(King et al., , 2020. More recently, it was shown that Drosophila larvae lacking Nf1 expression exhibit an increased likelihood of responding to a mechanical stimulus, thought to mirror the sensory sensitivity abnormalities comprising part of the ASD diagnostic criteria (Dyson et al., 2022). Here, we exploit this phenotype in a targeted, low-throughput screen to identify compounds that improve tactile hypersensitivity in Nf1 P1 larvae. Any such leads may therefore have potential in managing ASD symptomatology in affected individuals. We start by investigating the temporal requirements for Nf1, to determine when compounds must be administered for optimal activity. Then, using a protocol of feeding drugs during both embryonic and larval stages, we identify two compounds, simvastatin and BMS-204352, as capable of consistently improving, but not fully rescuing, the enhanced response to mechanical touch. Finally, we demonstrate that both compounds also reduce Nf1-associated synaptic transmission deficits at the neuromuscular junction (NMJ).

Fly lines and maintenance
The Nf1 P1 mutant and its parental K33 line used in this study  were initially obtained from Seth Tomchik (University of Iowa), where they were both backcrossed into the w CS10 background such that K33 acts as an isogenic control, as described previously (King et al., 2016;Dyson et al., 2022). For the temperature-dependent knock-down of Nf1 at different developmental stages, elav c155 -GAL4;tubP-GAL80 ts was crossed to either UAS-Nf1 RNAi ;UAS-Dicer2 (experiment) or UAS-GFP RNAi ;UAS-Dicer2 (control). These lines were generated by combining the following constructs together: elav c155 -GAL4 (Bloomington, #458; Giachello and Baines, 2015), tubP-GAL80 ts (McGuire et al., 2004), UAS-Nf1 RNAi (VDRC ID #109637; King et al., 2016), UAS-GFP RNAi (Bloomington #9331; Roignant et al., 2003), and UAS-Dicer2 (Dietzl et al., 2007). In these experiments, animals were maintained at either 30°C to permit GAL4dependent expression of UAS-Nf1 RNAi and thus knock-down of endogenous Nf1, or at 18°C to facilitate GAL80-dependent repression of GAL4, thereby allowing the expression of endogenous Nf1 (McGuire et al., 2004). Progenitor flies were set up to mate and left to deposit embryos on grape agar plates, which were collected at 4-h intervals before being transferred to food-containing vials. For the constitutive knock-down of Nf1, embryos and larvae were maintained constantly at 30°C until the experiment. For the embryonic knock-down of Nf1, embryos were maintained at 30°C for 17-21 h after egg laying (AEL), and then transferred to 18°C (Ashburner et al., 2005). For the larval knock-down of Nf1, embryos were maintained at 18°C for 42-to 46-h AEL, and then transferred to 30°C (Ashburner et al., 2005). For all other experiments, lines were maintained at 25°C. Fly food was standard cornmeal medium. All flies were kept in a 12/12 h light/dark cycle.

Compound administration
Compounds that were tested for their ability to improve tactile hypersensitivity in Nf1 P1 larvae are listed in Table 1. All compounds were made up as stock solutions in DMSO and stirred into molten fly food cooled to 60°C. Typically, progenitor flies were maintained on food containing either compound or vehicle (DMSO) for !3 d before being transferred onto fresh compound-containing or vehicle-containing food, from which larvae for the experiment were collected. However, for experiments examining embryonic exposure to compound, progenitor flies were maintained on food containing either compound or vehicle (DMSO) for !3 d before being transferred onto standard Compounds were selected based on either their ability to inhibit Ras/MAPK activity or to modulate ion channel function, mostly with the expectation of reducing neuronal excitability. As part of the "Ras/MAPK" targeting group, inhibitors of histone deacetylase (HDAC) and CDK4/6 were included since they have been demonstrated to improve viability in Drosophila models of other RASopathies (Das et al., 2021).
food (lacking compound), from which larvae for the experiment were collected. For experiments examining larval exposure to compound, progenitor flies were maintained on standard food before being transferred onto fresh compound-containing or vehicle-containing food, from which larvae for the experiment were collected. DMSO concentration was 0.05% for all experiments presented in this study.
Tactile sensitivity and compound screening To assess tactile sensitivity, a mechanical stimulus was applied briefly to the posterior end of each wall-climbing third instar larva (of either sex), as described previously (Dyson et al., 2022). Larvae were counted as "responding" only if they exhibited a full, 360°rolling motion. Video footage exemplifying the typical response of Nf1 P1 larvae (representative of larvae lacking Nf1 expression) compared with that of K33 controls has already been published (Dyson et al., 2022). All such experiments were conducted at room temperature by an experimenter blinded to genotype and/or treatment. For the initial screen, all compounds were tested at 50 mM on Nf1 P1 larvae and compared against Nf1 P1 larvae raised on an equivalent concentration of DMSO vehicle (n = 20 per treatment group). Data for each compound were accepted if vehicle-treated Nf1 P1 larvae exhibited the expected significant increase in likelihood of nocifensive response compared with vehicletreated K33 larvae (n = 20). For the compound screen, data were normalized with the equation: (no. of compound-treated responders/no. of vehicle-treated Nf1 P1 responders) Â 100, such that the number of Nf1 P1 larvae responding to the stimulus = 100%. Compounds found to exert a significant effect in the initial screen were then tested in four independent validation trials (n = 20 per treatment group per trial, thus n = 80 overall) at 50 mM. For compound validation, % responding larvae per trial [i.e., (no. of responders/20) Â 100] for each genotype was calculated.

Larval crawling
A 77-cm 2 arena formed of 2% agarose (depth ;2 mm) within the lid of a clear 96-well plate was placed into a DanioVision observation chamber. A 2-to 3-mm moat was maintained between the agar and the edge of the lid, which was filled with 5 M NaCl to deter larvae from crawling off the agar. Wall-climbing third instar larvae (of either sex) were rinsed in ddH 2 O, then placed onto the center of the arena, and left to acclimatize for 30 s. EthoVision XT Video Tracking Software (part of DanioVision) was used to measure total distance traveled over a 3-min period under white light. Experiments were conducted at room temperature by an experimenter blinded to genotype and/or treatment. Only larvae that remained on the agar for the entire recording period were included in our analyses.

Electrophysiology
Wall-climbing third instar larvae (of either sex) were fillet-dissected to permit the recording and analysis of excitatory junction potentials (EJPs) and miniature EJPs (mEJPs) in HL3 saline 1 1.5 mM CaCl 2 (Stewart et al., 1994), as described previously (Dyson et al., 2022). EJP amplitude and resting membrane potential were calculated using Clampfit 10.3 (Molecular Devices), while mEJP amplitude and frequency were calculated using MiniAnalysis (Synaptosoft Inc.). Amplitudes were corrected using the where v' refers to the corrected amplitude, v is the recorded amplitude, and E is the driving force, assumed to be equal to resting membrane potential if the reversal potential is 0 mV (Feeney et al., 1998). Quantal content was calculated as corrected EJP amplitude/corrected mEJP amplitude. All recordings were conducted blind to genotype or treatment.

Statistical analysis
All statistical analyses were conducted using GraphPad Prism (version 8). Pairs of quantitative datasets were compared via a two-tailed, unpaired Student's t test, while more than or equal to three quantitative datasets were analyzed via either one-way (ungrouped) or two-way (grouped) ANOVA, followed by Tukey's post hoc test. Fisher's exact test was used to compare two sets of categorical variables (e.g., vehicle-vs compound-treated Nf1 P1 larvae). All analysis was conducted on raw data before normalization. When analyzing compound validation experiments (see above, Tactile sensitivity and compound screening), comparisons were made between the percentage values of larvae responding per trial, such that sample size is the number of trials (i.e., n = 4). Statistically significant p values are given in the figures, while nonsignificant p values that are nevertheless relevant to interpreting the data are given in the figure legend.

Results
Constitutive knock-down of Nf1 is required to induce tactile hypersensitivity, while knock-down of Nf1 in the embryo alone is sufficient to impair synaptic transmission Because neurofibromatosis type 1 and ASD are primarily developmental disorders, it is probable that they arise, at least partially, from aberrations in brain development and/or function during early life stages that lead to permanent changes in postembryonic behavior (Yenkoyan et al., 2017;Courchesne et al., 2019). These aberrations may be irreversible, and, consequently, less susceptible to clinical intervention in later life (de la Torre-Ubieta et al., 2016). To account for this, we first investigated when the NF1 protein is required during the Drosophila life cycle to regulate larval tactile sensitivity. We also focused attention to synaptic transmission at the NMJ, since deficits in the latter may be associated with the hypersensitivity phenotype (Dyson et al., 2022). We used the UAS/GAL4/GAL80 ts system to restrict Nf1 expression to either embryonic or larval stages (McGuire et al., 2004). In these experiments, larvae in which Nf1 was knocked down via RNA interference were compared with a control line in which a dsRNA construct against GFP was expressed instead.
As expected, constitutive knock-down of Nf1 resulted in a significant increase in the likelihood of a larva exhibiting a nocifensive response following a brief, typically innocuous, mechanical stimulation (Fig. 1A). However, this was not observed when Nf1 knock-down was restricted solely to the embryonic period ( Fig. 1B) or, alternatively, to postembryonic larval stages (Fig. 1C). To completely rule out the possibility that some nonspecific effect of raising larvae at 18°C, independent of GAL80 ts expression, diminishes the nocifensive response, we also exposed Nf1 P1 mutant larvae and the isogenic K33 controls to the same shifts in temperature. In all such controls, the number of Nf1 P1 larvae displaying the nocifensive response was significantly greater than that of control larvae ( Fig. 1D-F). Thus, we conclude that the NF1 protein is involved in regulating normal tactile sensitivity during both embryogenesis and larval development.
Next, we examined the effect of identical shifts in Nf1 expression on synaptic transmission at the NMJ. In Nf1 À/À mutants, the frequency of spontaneous transmission events (mEJPs) is significantly increased, while evoked release (quantal content) is significantly reduced. Constitutive knock-down of Nf1 is required to induce tactile hypersensitivity in third instar larvae. Abbreviated genotypes for the lines tested are given in the figure panels. GFP RNAi (green) and Nf1 RNAi (orange) refer to lines in which GAL4 is expressed under the control of elav to drive expression of either UAS-GFP RNAi or UAS-Nf1 RNAi , respectively, and UAS-Dicer2, with GAL80 ts expressed under the control of the tubulin promoter. All transgenic constructs are hemizygous or heterozygous in the larvae tested. A, Constitutive knock-down of Nf1 expression (Nf1 RNAi ) throughout all life stages results in larval tactile hypersensitivity, as indicated by a significantly greater number of larvae responding to a mechanical stimulus compared with GFP RNAi . B, Knock-down of Nf1 only within the embryo has no significant impact on the number of responding larvae (p = 0.0580), nor does (C) knock-down of Nf1 within the larval stages (p . 0.9999). D-F, K33 and Nf1 P1 larvae were also subjected to the same shifts in temperature as those required for constitutive, embryonic, and larval knock-down of Nf1, respectively. Regardless of the temperature paradigm, Nf1 P1 larvae always demonstrated a significantly greater likelihood of displaying the nocifensive response following stimulation. Numbers within each bar represent the sample size for that group. For the ease of comparing groups in which different sample sizes were used, raw data have been presented as percentages within the figure. Statistical comparisons via Fisher's exact test were nonetheless conducted on raw data before normalization. This change is seemingly compensated for by an increase in postsynaptic input resistance, as evidenced via an enhanced quantal size (mEJP amplitude), rendering the amplitude of evoked events (EJPs) unchanged (Dyson et al., 2022). These changes are similarly observed following constitutive Nf1 knock-down ( Fig. 2A-F), as well as when knock-down of Nf1 is restricted to the embryonic period ( Fig. 2G-L). However, knock-down of Nf1 during larval stages is insufficient to disrupt synaptic transmission (Fig. 2M-R). Together, these data indicate an early developmental requirement for Nf1 in regulating synaptic transmission at the larval NMJ, but that the role of Nf1 in sensory sensitivity requires expression at both embryonic and larval stages.
A targeted screen identifies two compounds sufficient to improve tactile hypersensitivity in Nf1 P1 larvae To expedite the screening process, we opted to assay compounds for their ability to improve tactile hypersensitivity, rather than aberrant synaptic transmission. As knock-down of Nf1 in both the embryonic and larval periods is necessary to induce this behavioral phenotype, it is plausible that, for a compound to rescue such behavior, it must likewise be present throughout both stages. To achieve this, progenitor flies were provided compound-containing food for !3 d before being transferred onto fresh food containing the same compound on which eggs would be laid and larvae would develop. This was to ensure that the compound was present both during the embryonic and larval stages, as feeding compounds to mated females results in significant transfer to embryos (Marley and Baines, 2011). Tested compounds fell broadly into one of two classes of activity (Table 1; Das et al., 2021). First, compounds were selected for their ability to diminish Ras/MAPK signaling, since the NF1 protein acts as a negative regulator of Ras (Ratner and Miller, 2015), knock-down of either Ras85D or Ras64B fully rescues both tactile hypersensitivity and synaptic dysfunction in Nf1 P1 larvae (Dyson et al., 2022), and genetic variants associated with Ras/MAPK pathways are recurrent in idiopathic ASD (Luo et al., 2018). The second class comprised compounds known to modulate ion channel function, typically in favor of reducing neuronal excitability, as synaptic transmission deficits associated with tactile hypersensitivity in Nf1 P1 larvae are indicative of neuronal hyperexcitability (Dyson et al., 2022), and an imbalance in central excitatory/inhibitory activity has been proposed as a major pathophysiological mechanism underlying ASD (Rubenstein and Merzenich, 2003).
All drugs were tested at 50 mM (i.e., the concentration added to fly food), although the concentration reaching the CNS is unknown. While three of the compounds tested proved toxic at this concentration (Fluvastatin, PD0325901, and Trichostatin A), we identified two compounds that significantly reduced the number of Nf1 P1 larvae responding to mechanical stimulation (Fig. 3). These were the HMG CoAreductase inhibitor simvastatin, and the big conductance calcium-activated potassium channel (BK Ca ) activator BMS-204352. Four subsequent independent trials demonstrated that the effect of each compound was consistent, with a significantly lower mean percentage of Nf1 P1 responders in the compound-treated versus vehicle-treated groups (Fig. 4A,B). It is worth noting, however, that each compound exerted a significant effect in only three out of four trials when comparing treatment groups within the trial itself (Table 2).
In addition, we also tested our hypothesis that compound exposure is required in both the embryo and larval stage to improve sensory behavior. Accordingly, administering simvastatin solely during the embryo had no significant effect on the number of responding NF1 P1 larvae (Fig. 4C). By contrast, raising larvae on simvastatin, without embryonic exposure, significantly reduced the likelihood of a nocifensive response (Fig. 4D), but to a lesser degree than when exposing both embryo and the larva (Fig. 4A). These observations are consistent with our assumption that, based on Nf1 knock-down data (Fig. 1), a compound must be present during both embryonic and larval stages if it is to exert its optimal effect. However, in contrast to simvastatin, BMS-204352 administration during the embryo or larvae alone was sufficient to significantly reduce the mean percentage of responding larvae (Fig. 4E,F), although the effect was stronger with larval exposure. Indeed, administering BMS-204352 solely to larvae appears to be similarly effective as administering the drug throughout both stages (Fig. 4B), with changes from 91.3 6 2.4% to 68.8 6 4.3% responding (Fig. 4F), and 96.3 6 2.4% to 66.3 6 3.8% responding (Fig. 4A), respectively. Possibly, targeting a mechanism that may occur downstream of excessive Ras signaling, such as disrupted BK Ca activity, is a more robust way of improving behavior, such that pharmacological manipulation is still feasible in the larva.
While additional concentrations (10, 25, and 100 mM) of the two hit compounds were also tested, we found 50 mM the most consistently effective (data not shown). Furthermore, we also considered the possibility that one or both compounds reduce the likelihood of a nocifensive response in Nf1 P1 larvae via a nonspecific, sedative effect. To test this, the impact of each compound (50 mM) on larval mobility was measured over a 3-min period. We observed no significant difference in the mean distance traveled by Nf1 P1 and K33 larvae (Fig.  4G,H), nor was distance influenced in Nf1 P1 larvae by either simvastatin (Fig. 4G) or BMS-204352 (Fig. 4H). This supports the effect of these compounds being specific to tactile hypersensitivity, and not because of a global reduction in activity.
Simvastatin and BMS-204352 improve synaptic transmission deficits at the NF1 P1 larval NMJ Tactile hypersensitivity in Nf1 À/À larvae is associated with impaired synaptic transmission at the NMJ, with both phenotypes arising in a Ras-dependent manner (Dyson et al., 2022). Therefore, we next investigated the effect of simvastatin (Fig. 5A-F) and BMS-204352 (Fig. 5G-L) exposure throughout embryonic and larval stages on activity at this peripheral glutamatergic synapse. Exposure to simvastatin (50 mM) led to a significant reduction in the typically enhanced mEJP frequency of Nf1 P1 larvae, although rescue  Figure 1. A-F, Raising Nf1 RNAi larvae at 30°C, sufficient to ensure constitutive knock-down of Nf1, mimics the Nf1 À/À larval phenotype (Dyson et al., 2022), in that it does not affect EJP amplitude (p = 0.5334) but increases mEJP frequency and amplitude, and reduces quantal content. G-L, Knock-down of Nf1 within the embryonic period has a similar effect on synaptic transmission as constitutive knock-down on all parameters examined, including no significant impact on (G) EJP amplitude (p = 0.8967). M-R, In contrast, knock-down of Nf1 only during larval stages has no effect on (M) EJP amplitude (p = 0.3010), (N) mEJP frequency (p = 0.3230), (O) mEJP amplitude (p = 0.3694), or (P) quantal content (p = 0.9998). Data in each panel were analyzed via unpaired, two-tailed Student's t test. For each experiment, n = 13. Data are presented as mean 6 SEM. was only partial since this was still significantly greater than that of vehicle-treated K33 controls (Fig. 5B). Conversely, the same treatment led to full rescue of the increased quantal size (mEJP amplitude) and reduced quantal content (Fig.  5C,D). There was no effect on EJP amplitude, which is also unaltered by the Nf1 P1 mutation (Fig. 5A), and simvastatin treatment had no effect on any parameter measured in K33 larvae. Similar effects on EJP amplitude (Fig. 5G), mEJP frequency (Fig. 5H), and mEJP amplitude (Fig. 5I) were also observed following BMS-204352 (50 mM) treatment, with the exception that the increase in quantal content in Nf1 P1 larvae treated with BMS-204352, compared with those treated with DMSO, was not significant (Fig. 5J). However, the quantal content of Nf1 P1 larvae treated with BMS-204352 was also not significantly different from that of vehicle-treated K33 larvae. Again, there was no effect of BMS-204352 on any parameter measured in K33 larvae.

Discussion
Clinical trials have thus far failed to identify treatments that effectively and consistently help to manage core ASD symptomatology, including abnormal responses to sensory stimuli. To address this, we examined whether compounds that target either Ras/MAPK signaling or ion channel activity can improve tactile hypersensitivity in Nf1 P1 larvae. We also investigated the temporal requirements for Nf1 in regulating the two ASD-associated phenotypes employed in this study. This is because the onset of certain ASD symptoms may occur as a result of specific developmental disruption; consequently, the time during which treatment is administered may be of equal importance as the molecular mechanism targeted. For example, in a mouse model of the syndromic ASD Tuberous Sclerosis, a four-week regimen of rapamycin treatment beginning at postnatal day 7 prevented the occurrence of impaired social behavior, even four weeks after the end of treatment (Gibson et al., 2022). Conversely, rapamycin administration starting at 10 weeks of age was unable to rescue the deficit (Tsai et al., 2018).
Here, we find that Nf1 likely regulates NMJ synaptic transmission via a developmental role, since its early, embryonic, downregulation is sufficient to induce persistent deficits in the third instar larva. An early, Nf1-dependent developmental window has also been identified in the regulation of Drosophila motor activity, with Nf1 expression required within the pupa, but not the adult, to regulate grooming behavior in the latter (King et al., 2020). In addition, the loss of function of other ASD-associated genes required for NMJ development, such as the adaptor protein ank2 (Koch et al., 2008;Stevens and Rasband, 2021) and the presynaptic adhesion molecular neurexin-1 (J. Li et al., 2007;Hu et al., 2019), has also been shown to disrupt third instar larval NMJ synaptic transmission. Thus, that the NF1 protein regulates synaptic transmission in a developmental manner is not necessarily surprising, although the mechanism via which it does so remains to be investigated. One possibility is via the regulation of reactive oxygen species (ROS) generation. ROS generation is an important contributor to the regulation of NMJ development, structure, and plasticity (Milton et al., 2011;Oswald et al., 2018), and excessive mitochondrial ROS levels are observed in Nf1 À/À flies, giving rise to a reduced lifespan and poor stress tolerance (Tong et al., 2007). In Nf1 P1 larvae, early abnormalities in ROS production may lead to abnormal NMJ development, which in turn may give rise to persistent changes in synaptic transmission.
In GABAergic hippocampal neurons of mice, loss of Nf1 causes enhanced inhibitory transmission via increased Ras/MAPK-dependent phosphorylation of synapsin I, a synaptic vesicle protein that, on phosphorylation, dissociates from vesicles to facilitate their recruitment to the readily-releasable pool (Cui et al., 2008). Thus, the mechanism of Nf1-dependent transmission here appears to be physiological, rather than developmental. Yet, differences in how the loss of Nf1 impacts neurotransmission in the mouse hippocampus relative to that at the Drosophila NMJ have already been discussed (Dyson et al., 2022), suggesting that Nf1 may regulate this process via multiple mechanisms, and/or in a cell type-dependent manner. It should also be noted that, given the temporal resolution of the UAS/GAL4/GAL80 ts system (McGuire et al., 2004), we cannot entirely rule out an additional requirement for Nf1 in the early first instar larval stage.
In contrast to NMJ phenotypes, constitutive loss of Nf1 is required to induce larval tactile hypersensitivity. One possible explanation for this difference is that, while abnormal synaptic transmission may contribute to tactile hypersensitivity, other pathophysiological mechanisms, occurring within the larval stage, are also necessary, such that the former is insufficient to induce the behavioral phenotype alone. Alternatively, while Nf1 may regulate Figure 3. A targeted pharmacological screen identifies simvastatin and BMS-204352 to improve tactile hypersensitivity in Nf1 P1 larvae. Twenty compounds were screened for their ability to reduce the number of Nf1 P1 larvae responding to a mechanical stimulus, with three proving toxic at the concentration (50 mM) tested. Only administration of simvastatin (50% of vehicletreated Nf1 P1 ) and BMS-204352 (61.1% of vehicle-treated Nf1 P1 ) resulted in a significant decrease in the number of responding larvae. Data are presented as a percentage of the number of Nf1 P1 larvae raised on an equivalent concentration of DMSO (vehicle) that responded to the stimulus (dotted line = 100%). Statistical comparisons were conducted on raw data using Fisher's exact test between compound-treated and vehicle-treated Nf1 P1 larvae.
Research Article: New Research Figure 4. Simvastatin and BMS-204352 consistently improve, but do not fully rescue, tactile hypersensitivity in Nf1 P1 larvae, while having no effect on overall activity. A, Across four independent trials, 50 mM Simvastatin significantly reduces the mean percentage responding larvae, as does (B) 50 mM BMS-204352. C, The presence of simvastatin solely during the embryo has no impact on the percentage of Nf1 P1 larvae responding to stimulation (p = 0.3026). D, Conversely, exposing only larval progeny to simvastatin results in a significant reduction in the mean percentage of responding Nf1 P1 larvae. E, The presence of BMS-204352 solely during the embryo results in a significant reduction in the percentage of Nf1 P1 larvae responding to stimulation. F, The percentage of responding Nf1 P1 larvae is also significantly reduced when only larval progeny are exposed to BMS-204352. G, K33 and Nf1 P1 larvae do not significantly differ in their total distance traveled over a 3-min period at room temperature, nor is this impacted by treatment with 50 mM simvastatin (p = 0.9518). H, 50 mM BMS-204352 treatment also does not alter distance traveled in Nf1 P1 larvae, which again show no significant difference in crawling behavior compared with K33 controls (p = 0.1881). In panels A-F, each data point represents the percentage responding larvae from a single trial, with n = 20 per trial, such that n = 80 larvae overall. Data in these panels were analyzed via unpaired, two-tailed Student's t test. Comparisons were made between individual trials, such that n = 4 per group. Data in panels G, H were analyzed via one-way ANOVA followed by Tukey's post hoc test. All data are presented as mean 6 SEM. The data in the table refer to that depicted in Figure 4A,B. For each trial, the number of vehicle-treated and compound-treated Nf1 P1 larvae responding to the stimulus is given, both as a fraction of total larvae tested (n = 20 per trial) and as a percentage. The outcome (p value) of the statistical comparison between these two groups in each trial is also stated. Both compounds reduced the number of responding larvae in all four validation trials, with this reduction reaching significance (p , 0.05) in three out of four trials for each. For trials in which the reduction was not significant, p values were nevertheless approaching 0.05. Statistical analysis (via Fisher's exact text) was conducted on raw data rather than percentage values. ns = not significant. Figure 5. Simvastatin and BMS-204352 improve synaptic transmission deficits at the Nf1 P1 larval NMJ, and have no impact on normal transmission in K33 larvae. A, Simvastatin (50 mM) treatment has no effect on EJP amplitude in either Nf1 P1 or K33 larvae (p = 0.2793). B, The increased mEJP frequency of Nf1 P1 larvae is reduced following simvastatin treatment, although this is still significantly greater than that of vehicle-treated K33 larvae. There is no significant difference between vehicle-treated and simvastatintreated K33 larvae (p = 0.9784). C, Simvastatin rescues the enhanced mEJP amplitude of Nf1 P1 larvae to values indistinguishable from those of vehicle-treated K33 larvae (p = 0.8097), which do not show a significant difference compared with simvastatin-treated K33 larvae (p . 0.9999). D, Simvastatin rescues the reduced quantal content of Nf1 P1 larvae to values indistinguishable from those of vehicle-treated K33 larvae (p = 0.4934), which also do not show a significant difference compared with simvastatin-treated K33 larvae (p = 0.9989). E, F, Representative traces of data presented in panels A-D. G, BMS-204352 (50 mM) treatment has no effect on EJP amplitude in either Nf1 P1 or K33 larvae (p = 0.5878). H, The increase in mEJP frequency of Nf1 P1 larvae is reduced following BMS-204352 treatment, although this is still significantly greater than that of vehicle-treated K33 larvae. There is no significant difference between vehicle-treated and simvastatin-treated K33 larvae (p = 0.9098). I, BMS-204352 rescues the enhanced mEJP amplitude of Nf1 P1 larvae to values indistinguishable from those of vehicle-treated K33 larvae (p = 0.9211), which do not show a significant difference compared with BMS-204352-treated K33 larvae (p = 0.9996). J, The increase in quantal content in Nf1 P1 larvae following BMS-204352 treatment is not significant (p = 0.1075); however, the quantal content of BMS-204352 treated Nf1 P1 larvae also does not differ from that of vehicle-treated K33 larvae (p = 0.8849). BMS-204352 does not significantly alter quantal content in K33 larvae either (p = 0.9991). All statistical comparisons were made via two-way ANOVA followed by Tukey's post hoc test, in which each genotype 1 treatment group was compared with all others. n = 13 for each group. All data are presented as mean 6 SEM. Although not explicitly stated in the figure or legend, in panels B-D and H-J, mEJP frequency and amplitude were both significantly increased, and quantal content significantly reduced, in vehicle-treated Nf1 P1 larvae relative to vehicle-treated K33 controls, as would be expected in larvae lacking Nf1 expression (Dyson et al., 2022). synaptic transmission at the NMJ in a developmental manner, it may not function similarly at central synapses, as it was previously shown that tactile hypersensitivity in Nf1 P1 larvae likely arises from a central, cholinergic deficit (Dyson et al., 2022). It is also conceivable that Nf1 may be required for some later compensatory mechanism that does not directly correct synaptic dysfunction but, nevertheless, ensures that it does not lead to changes in mechanosensory behavior. This may mean that, in the Nf1 P1 mutant, no compensation occurs following early changes to synaptic transmission because the Nf1 gene is permanently deleted; conversely, in the UAS/GAL4/GAL80 ts paradigm, re-expression of Nf1 in later larval stages following embryonic knock-down permits compensatory changes to prevent deficits in behavior. Future work is required to investigate these, and other, possibilities, and to also narrow down the Nf1-dependent critical period of synaptic transmission.
Regardless of the precise timings of Nf1 function, we have identified two compounds that are able to improve tactile hypersensitivity and impaired synaptic transmission in Nf1 P1 larvae. These are the HMG CoA-reductase inhibitor simvastatin, and the BK Ca channel activator BMS-204352. Importantly, neither drug influenced synaptic transmission in the K33 control line. Yet, it should be noted that feeding is increased in Nf1 P1 adult flies (Botero et al., 2021); if the same is true in Nf1 P1 larvae, this raises the possibility that they ingest more drug than K33 controls. Consequently, the lack of effect of simvastatin and BMS-204352 on synaptic transmission in K33 larvae may, in part, be attributed to a comparatively lower amount of compound being present, rather than indicating the efficacy of the compounds in managing Nf1-dependent pathophysiological mechanisms. Conversely, increased feeding in Nf1 P1 flies is suggested to act as a homeostatic response to an increased metabolic rate (Botero et al., 2021), which would suggest that the amount of drug active in both lines should be similar.
Simvastatin, currently prescribed as a cholesterol-lowering medication, has been examined in clinical trials for its efficacy in treating behavioral symptoms of neurofibromatosis type 1, with mixed success. It did not improve cognition in two large randomized controlled trials of children and adolescents (Krab et al., 2008;van der Vaart et al., 2013), possibly because the intervention occurred too late, with it being suggested that simvastatin treatment may have been beneficial in younger children (van der Vaart et al., 2013). Our data support this, since the improvement in tactile hypersensitivity in Nf1 P1 larvae was strongest when administration of the compound throughout the larval stage was combined with embryonic exposure. Indeed, some improvement in ASD symptomatology following simvastatin treatment was demonstrated in a more recent, pilot study of younger children with neurofibromatosis type 1 (Stivaros et al., 2018). The same study also concluded that simvastatin is well-tolerated in such children (Stivaros et al., 2018), in contrast to many therapies currently prescribed for co-morbid behavioral symptoms in ASD (Aishworiya et al., 2022). However, if a drug must indeed be present during early (e.g., embryonic) as well as later development to elicit an effect, as our data suggest, this may entail considerable practical and ethical implications in the clinic.
BMS-204352 currently has no clinical application, having originally been developed as a therapy for acute ischemic stroke but displaying no benefit over placebo in Phase III trials (Jensen, 2002). Since nanomolar concentrations of BMS-204352 are sufficient to activate the BK Ca channel in vitro (Jensen, 2002), it most likely exerts its effect on sensory sensitivity via this mechanism, thereby implicating the reduced expression, or impaired activity, of BK Ca channels in tactile hypersensitivity and impaired transmission in Nf1 P1 larvae. This is consistent with a role for BK Ca channels at the plasma membrane in regulating neuronal excitability (N'Gouemo, 2014). However, BK Ca channels are also expressed in mitochondria, where their activity is both important for the regulation of ROS, and susceptible to manipulation by oxidizing agents (Balderas et al., 2015;Hermann et al., 2015). Given the possible involvement of ROS in Nf1-dependent NMJ development and activity suggested above, the modulation of ROS may comprise an alternative mechanism by which BMS-204352 influences synaptic transmission (and behavior) via BK Ca .
Roles for Nf1 in regulating potassium currents have been demonstrated previously, as the postsynaptic K 1 current typically elicited by application of PACAP38 is diminished at the Nf1 P1 NMJ (Guo et al., 1997), and treatment with apamin, an inhibitor of the SK Ca channel, rescues spatial learning deficits in Nf1 1/À mice (Kallarackal et al., 2013). Furthermore, haploinsufficiency of KCNMA1 (encoding BK Ca ) has been previously identified in an ASD individual, resulting in a diminished BK Ca current in patientderived lymphoblastoid cells that was enhanced via application of BMS-204352 (Laumonnier et al., 2006). BMS-204352 treatment has also been shown to rescue sensory hypersensitivity (Zhang et al., 2014), altered social preference (Hébert et al., 2014), and hyperactivity in an unfamiliar environment (Carreno-Munoz et al., 2018) in mouse models of the syndromic ASD Fragile X. In addition, BMS-204352 and an alternative BK Ca activator, BMS-191001, were found to rescue structural deficits (Hébert et al., 2014) and hyperexcitability (Zhang et al., 2014), respectively, in dendrites. Thus, our finding may hint at a shared mechanism underlying sensory hypersensitivity, and potentially other symptoms, in different forms of syndromic ASD. To further explore this possibility, one should seek to confirm that the improvement elicited by BMS-204352 is indeed mediated via BK Ca , and not via other channels that the compound is also able to activate, such as those of the KCNQ family (Jensen, 2002). To do so, BMS-204352 may be administered alongside a blocker of the BK Ca channel, such as paxilline or iberiotoxin, to determine whether this abrogates the effect (Balderas et al., 2015). In addition, the genetic upregulation of slowpoke (encoding the Drosophila BK Ca channel a subunit) in Nf1 P1 larvae would, if leading to successful phenotypic rescue, strengthen the association of BK Ca dysfunction with Nf1 P1 -dependent tactile hypersensitivity and impaired synaptic transmission.
On the other hand, there is some question over the feasibility of targeting BK Ca via BMS-204352 in a clinical context. Other studies have considered the utility of BK Ca as a target in the treatment of epilepsy, a neurologic disorder characterized by an excess of excitatory activity (N'Gouemo, 2011). However, both loss-of-function and gain-of-function mutations within the channel can give rise to neuronal hyperexcitability, and different anti-epileptic drugs have opposing effects on BK Ca activity. Moreover, pharmacological BK Ca activation may either protect against or, conversely, induce seizures in different contexts. As such, the effective modulation of BK Ca channels as a treatment for neurodevelopmental disorders may be challenging (N'Gouemo, 2011(N'Gouemo, , 2014. Although Phase I and II trials have demonstrated that BMS-204352 is safe and well-tolerated in both healthy adults and acute stroke patients (Jensen, 2002), whether this will also be the case in young children, and over an extended dosing period, is unknown.
Despite demonstrating the consistent efficacy of simvastatin and BMS-204352 in improving Nf1-dependent tactile hypersensitivity in four subsequent trials, we were unable to fully rescue the phenotype, since K33 control larvae typically show a 0-20% likelihood of responding in the mechanoreception assay (Dyson et al., 2022). One explanation is that we did not test compounds at optimum concentrations. Alternatively, it may be because the target of the drug is not the only mechanism concerned. For instance, while BK Ca dysfunction may contribute to tactile hypersensitivity, the dysfunction of other ion channels, for example, may also be involved, which must also be pharmacologically corrected to completely rescue the phenotype. However, it is difficult to apply this reasoning to simvastatin: although we have not shown biochemically that simvastatin functions in our assay by targeting Ras (e.g., via Western blot of larval CNS extracts to examine phospho-MAPK levels following treatment), the most parsimonious explanation for its effect is that, by inhibiting HMG CoA-reductase, it prevents the farnesylation of Ras and its subsequent anchoring to the plasma membrane (W. Li et al., 2005). Yet, genetically attenuating Ras protein expression via RNA interference fully rescues the hypersensitivity phenotype (Dyson et al., 2022).
We did not examine the possibility of a combinatorial effect of the compounds on tactile hypersensitivity, such that administering both simvastatin and BMS-204352 simultaneously results in a stronger phenotypic rescue. Yet, if the target of BMS-204352 (assumed to be altered BK Ca activity) arises directly downstream of elevated Ras/ MAPK signaling, the utility of combinatorial therapy here may be limited. Alternatively, combining simvastatin or BMS-204352 with compounds targeting distinct molecular pathways may provide a more effective approach. Such compounds may include antioxidant therapies, like N-acetylcysteine, which was found to improve irritability and stereotypy in children with ASD (Hardan et al., 2012), and may function in part by ameliorating an excitation/inhibition imbalance via a reduction in glutamatergic transmission (Dean et al., 2011;Hardan et al., 2012). Oxidative stress has previously been implicated in the pathophysiology of idiopathic ASD (Bjørklund et al., 2020), and the possibility of abnormal ROS underlying defective transmission in Nf1 À/À larvae has been considered above. Screening antioxidant compounds in such animals may shed light on this proposed mechanism.
A subsequent screen may also involve activators of the cAMP/PKA pathway. The NF1 protein acts as a positive regulator of cAMP/PKA signaling by stimulating adenylyl cyclase activity (Guo et al., 2000;Tong et al., 2002), and genetic or pharmacological stimulation of this pathway rescues anatomic and behavioral phenotypes in both fly and mouse models of neurofibromatosis type 1 (Guo et al., 2000;Tsai et al., 2012;Diggs-Andrews et al., 2013;Sutton et al., 2019). While knock-down of Ras expression is sufficient to rescue tactile hypersensitivity and abnormal synaptic transmission in Nf1 P1 larvae (Dyson et al., 2022), this does not necessarily rule out impaired cAMP/ PKA signaling in the mechanism, since the NF1 protein is capable of regulating both Ras and cAMP even within the same pathway(s) (Anastasaki and Gutmann, 2014;Walker and Bernards, 2014).
To translate our findings into the clinic, it will be important to test these compounds in mammalian models of neurofibromatosis type 1 and/or other syndromic forms of ASD. Unlike Drosophila larvae, rodent models of neurofibromatosis type 1 do not consistently and strongly exhibit enhanced sensitivity to a mechanical stimulus, and while such models may display thermal hyperalgesia, this is considered a proxy for chronic pain in neurofibromatosis type 1 and arises because of sensory impairment (Moutal et al., 2017;Bellampalli and Khanna, 2019). Conversely, tactile hypersensitivity in Nf1 À/À larvae likely originates in the CNS, and may therefore have different mechanistic underpinnings (Dyson et al., 2022). Nevertheless, Nf1 1/À mice do display behaviors reflecting other aspects of ASD symptomatology, such as altered social behavior and communication (Molosh et al., 2014;Maloney et al., 2018). Since a single compound may not target multiple ASD symptom domains, before preclinical testing, it would be useful to test simvastatin and BMS-204352 on phenotypes in the fly that are also analogous to social impairment, such as defective courtship (Moscato et al., 2020). Additionally, determination of the efficacy of these compounds on their purported mechanistic targets, e.g., MAPK phosphorylation and BK Ca currents, respectively, in Nf1 1/À cell lines derived from human patients may prove fruitful.
In summary, we have shed light on the temporal requirements for the NF1 protein in regulating two previously identified larval phenotypes that arguably parallel ASD symptomatology and etiology. Subsequently, we identified two compounds that improve, but do not necessarily fully rescue, the same phenotypes. Despite the previous failure of simvastatin to improve cognitive symptoms in children with neurofibromatosis type 1 (Krab et al., 2008;van der Vaart et al., 2013), our findings suggest that the further investigation of this compound, and of BMS-204352, specifically in the context of ASD may be warranted.