Neuronal overexpression of Alzheimer's disease and Down's syndrome associated DYRK1A/minibrain gene alters motor decline, neurodegeneration and synaptic plasticity in Drosophila

Down syndrome (DS) is characterised by abnormal cognitive and motor development, and later in life by progressive Alzheimer's disease (AD)-like dementia, neuropathology, declining motor function and shorter life expectancy. It is caused by trisomy of chromosome 21 (Hsa21), but how individual Hsa21 genes contribute to various aspects of the disorder is incompletely understood. Previous work has demonstrated a role for triplication of the Hsa21 gene DYRK1A in cognitive and motor deficits, as well as in altered neurogenesis and neurofibrillary degeneration in the DS brain, but its contribution to other DS phenotypes is unclear. Here we demonstrate that overexpression of minibrain (mnb), the Drosophila ortholog of DYRK1A, in the Drosophila nervous system accelerated age-dependent decline in motor performance and shortened lifespan. Overexpression of mnb in the eye was neurotoxic and overexpression in ellipsoid body neurons in the brain caused age-dependent neurodegeneration. At the larval neuromuscular junction, an established model for mammalian central glutamatergic synapses, neuronal mnb overexpression enhanced spontaneous vesicular transmitter release. It also slowed recovery from short-term depression of evoked transmitter release induced by high-frequency nerve stimulation and increased the number of boutons in one of the two glutamatergic motor neurons innervating the muscle. These results provide further insight into the roles of DYRK1A triplication in abnormal aging and synaptic dysfunction in DS.


Introduction
Down syndrome (DS, also known as Down's syndrome) or trisomy 21 is caused by the presence of three copies of chromosome 21 (Hsa21) instead of the usual two (Herault et al., 2017). It is characterised by cognitive impairment (Lott and Dierssen, 2010) and the delayed and incomplete acquisition of motor skills (Malak et al., 2015) as a result of abnormal development of the nervous system (Stagni et al., 2018). Individuals with DS almost invariably develop Alzheimer's disease (AD)-like symptoms (AD-DS). These include progressive dementia after 40 years of age, the onset of amyloid plaques, neurofibrillary tangles (NFTs) and neurodegeneration after 10-20 years (Wiseman et al., 2015;Zigman, 2013), faster age-dependent motor decline that is an early marker for the onset of cognitive decline and health deterioration (Anderson-Mooney et al., 2016;Buchman and Bennett, 2011), and a shorter mean life expectancy by approximately 28 years (O'Leary et al., 2018). Currently there is no treatment for DS or AD; our understanding of the mechanisms of the disorder is incomplete and this hampers the development of effective therapies.
One of the Hsa21 genes, DYRK1A (dual specificity tyrosine-phosphorylation-regulated kinase 1A), is a candidate causative gene for the structural and functional changes that occur in the DS brain, and for the associated cognitive and motor deficits (Herault et al., 2017;Stagni et al., 2018). DYRK1A/Dyrk1a mRNA and protein are expressed throughout the brain in humans and rodents, wherein DYRK1A controls aspects of neuronal development and function (Duchon and Herault, 2016;Kay et al., 2016;Stringer et al., 2017). DYRK1A/Dyrk1a mRNA and protein expression is increased in DS brain and in the brain of different mouse models of DS (Duchon and Herault, 2016;Garcia-Cerro et al., 2018;Kay et al., 2016;Stringer et al., 2017), whether the gene is triplicated as part of a genomic segment, as in Dp1Yey, Ts65Dn, Ts1Cje and Tc1 mice, or alone as in TgDyrk1a and TgDYRK1A mice (Herault et al., 2017).
DS-associated cognitive and motor deficits are replicated by overexpression of Dyr1ka in mice (Ahn et al., 2006;Altafaj et al., 2001;Cerro et al., 2018;Martínez de Lagrán et al. 2004;Ortiz-Abalia et al., 2008;Souchet et al., 2014;Watson-Scales et al., 2018). However, the contribution of DYRK1A overexpression to the shorter life expectancy, faster age-dependent decline in cognitive and motor function, and development of AD-like pathology in DS is unclear. It is predicted to play a role as it both phosphorylates tau and alters its splicing (Shi et al., 2008;Woods et al., 2001), promoting its self-aggregation (Liu et al., 2008) into NFTs. Dyrk1A is found physically associated with NFTs in the brain to a greater level in DS-AD than non-DS associated AD Wegiel et al., 2011). In the Ts65Dn and Ts1Cje mouse models of DS, Dyrk1a overexpression in the brain intensifies with age (Ahmed et al., 2017;Creau et al., 2016;Stringer et al., 2017;Watson-Scales et al., 2018), and this is associated with AD-DS-like histopathological changes in the aged Ts65Dn brain (García-Cerro et al., 2017;Wiseman et al., 2015). However, there is insufficient behavioural data from aged animals to directly assess the impact of DYRK1A overexpression in inducing DS-AD phenotypes.
Cognitive and motor dysfunction in individuals with DS and in mouse models of DS are associated with changes in synaptic plasticity and with changes in the number and structure of GABAergic and glutamatergic brain neurons and synapses (Battaglia et al., 2008;Contestabile et al., 2017;Duchon and Herault, 2016). Such modifications have been linked to Dyrk1a overexpression (Duchon and Herault, 2016;García-Cerro et al., 2017;Garcia-Cerro et al., 2018;Ruiz-Mejias et al., 2016), but the effects of Dyrk1a overexpression on the basic properties of synaptic function have rarely been explored. In one study, there was no change in the frequency of miniature excitatory synaptic currents (mEPSCs) or the probability of electrically-evoked glutamate release in the prefrontal cortex of TgDyrk1a mice (Thomazeau et al., 2014). Nevertheless, since Dyrk1a controls the activity of proteins that regulate endocytosis (Murakami et al., 2012) and DYRK1A overexpression slows endocytosis of transmitter vesicles in hippocampal presynaptic membranes from TgDYRK1A mice (Kim et al., 2010), modulation of transmitter release at other glutamatergic synapses is likely.
To investigate the contribution of DYRK1A overexpression in the nervous system to various aspects of DS, we overexpressed minibrain (mnb), the Drosophila ortholog of DYRK1A (Duchon and Herault, 2016), in the Drosophila nervous system and implemented well-established assays in larvae and adult flies (Bykhovskaia and Vasin, 2017;Lenz et al., 2013;McGurk et al., 2015). The assays monitored motor impairment and its development with age, lifespan, age-related neurodegeneration, and synaptic dysfunction. Due to their short lifecycle, Drosophila are one of the pre-eminent models for aging and neurodegeneration (Jones and Grotewiel, 2011), both aspects of DS that are more difficult to investigate in mice. The Drosophila larval neuromuscular junction (NMJ) is a well-established model for mammalian central glutamatergic synapses and is easily accessible to electrophysiology (Bykhovskaia and Vasin, 2017). Mnb is expressed presynaptically at larval NMJs and reducing its expression changes motor nerve terminal structure and impairs recycling of transmitter vesicles, while overexpression of one isoform, mnbeF, has no effect on basal transmission but ameliorates the effects of reduced mnb expression (Chen et al., 2014). Five mnb isoforms, E-I, have been identified, all of which share a highly conserved kinase domain (Gramates et al., 2017;Hong et al., 2012). Regions of DYRK1A outside the kinase domain also appear to play important roles, but which areas exactly and how they impact function is thus far incompletely understood (Jin et al., 2015;Kelly and Rahmani, 2005;von Groote-Bidlingmaier et al., 2003). We therefore utilised mnbeH, the isoform with the longest coding region (Gramates et al., 2017;Zerbino et al., 2018). Here we report the effects of neuronal overexpression of mnb-H on motor function, the rate of motor decline with age, lifespan, age-related neurodegeneration, presynaptic structure, spontaneous transmitter release and recovery from frequencydependent depression of electrically-evoked transmitter release.

Neuronal overexpression of mnb produced motor deficits in larvae, accelerated age-dependent motor decline in adult flies and shortened adult lifespan
The effect of mnb overexpression in the nervous system on motor function (specifically the mnb-H splice variant) was tested using two assays of fly larval locomotion. Elav > mnb larvae, overexpressing mnb throughout the nervous system under the control of the Elav-Gal4 driver (Robinow and White, 1991), did not move as far as control larvae (Elav/ +) in a free movement assay (Fig. 1A), which measures the ability of larvae to perform rhythmic muscle contractions necessary for gross locomotion (Kohsaka et al., 2017). They also took longer to complete a self-righting assay (Fig. 1B), which is a more complex motor task requiring larvae to enact a co-ordinated sequence of movements to right themselves after being rolled onto their backs (Picao-Osorio et al., 2015). To assess the impact of neuronal mnb overexpression on agerelated decline in locomotor function, the performance of the same cohorts of adult flies was assessed in a negative geotaxis assay at different ages (Jones and Grotewiel, 2011). This showed acceleration in Elav > mnb flies of the usual age-related decline in performance (Elav/ +). There was also evident shortening of the lifespan of Elav > mnb flies, so that the median lifespan was reduced by almost 50% (Elav/+, (A) Elav > mnb larvae crossed fewer lines of a 0.5 cm grid in 60 s than control larvae (Elav/+, 14.7 ± 0.44, n = 15; Elav > mnb, 12.1 ± 0.86, n = 15; mean ± SEM, *P = .014, Student's t-test), and with greater variance (F (14,14) = 3.768, P = .0184, F-test). (B) Elav > mnb larvae took longer than controls to perform a self-righting task (Elav/+, 5.5 ± 1.16 s, n = 15; Elav > mnb, 12 ± 2.56 s; n = 15; mean ± SEM, *P = .029, Student's t-test) and with greater variance (F (14,14) = 4.802, P = .0059, F-test). Each point in the plots (A, B) is from a different animal; horizontal lines indicate mean values. (C) The age-dependent decline in climbing ability in a negative geotaxis assay was steeper and showed greater variance for Elav > mnb adult flies than for controls (F (3,84) = 13.8; ***P < .0001, repeated measures two-way ANOVA, n = 15 groups of flies); at 1 day old there was no difference in the percentage of flies that climbed successfully (Elav/+, 90.59 ± 1.82%, n = 15; Elav > mnb ± 85.94 ± 1.46%, n = 15; mean ± SEM, P = .8262, repeated measures two-way ANOVA and Sidak's multiple comparison). Values plotted are mean ± SEM, n = 15 groups of 10 flies for each genotype. (D) Elav > mnb flies had a shorter lifespan relative to controls (n = 100 animals per genotype at day 0, ***P < .0001, log-rank (Mantel-Cox) test). 73 days; Elav > mnb, 38 days; Fig. 1D). These results indicate that neuronal overexpression of mnb alone produced a motor deficit and abnormal aging characterised by accelerated age-related locomotor impairment and a shorter lifespan.

Overexpression of mnb caused neurodegeneration in adult flies
As DYRK1A triplication has been linked to degeneration of brain neurons in AD-DS and in Ts65Dn mice (García-Cerro et al., 2017;Wegiel et al., 2008), we tested the possibility that neuronal overexpression of mnb is sufficient to cause neurotoxicity and age-related neurodegeneration using two established assays of neurodegeneration in adult flies (Lenz et al., 2013;McGurk et al., 2015). In the first, mnb was overexpressed in the eye through development and adulthood using the Glass multimer reporter driver (GMR-Gal4) (Ellis et al., 1993). The GMR > mnb flies, but not control flies (GMR/+), had a reduced eye surface area and a visible "rough eye" phenotype ( Fig. 2A), both of which indicate neural death and the resultant breakdown of the regularly spaced array of ommatidia making up the retina. In a second assay, the EB1 driver (EB1-Gal4) was used to overexpress mnb in the ellipsoid body (EB), a subpopulation of neurons within the central complex of the brain implicated in locomotor control ( Fig. 2B) (Diaper et al., 2013). The EB cells also expressed membrane-bound GFP which enabled their visualisation. At 1 day old, there was no difference in the number of GFP-positive EB neurons between control (EB1/+) and EB1 > mnb flies, whereas at day 40 the number of EB neurons was significantly reduced in EB1 > mnb flies but not in control flies ( Fig. 2C). Therefore, neurotoxicity caused by mnb overexpression promoted age-related neurodegeneration in a central neuron population.

Overexpression of mnb in motor neurons altered basal synaptic transmission at the larval NMJ
As neuronal overexpression of mnb increased the number of 1b boutons at the larval NMJ, and because previous studies have implicated Dyrk1a/mnb in the control of the recycling of neurotransmitter vesicles (Chen et al., 2014;Kim et al., 2010;Murakami et al., 2012), we investigated if spontaneous glutamate release was altered by recording spontaneously occurring miniature excitatory junction potentials (mEJPs) with intracellular microelectrodes. Since the muscle 6/7 NMJ is innervated by 1b and 1 s boutons, mEJPs usually result from the release of neurotransmitter vesicles from both types of boutons (Newman et al., 2017). Our recordings revealed that mEJPs occurred more frequently but were smaller in OK371 > mnb larvae (Fig. 4A-B). The decrease in amplitude was not secondary to a change in the electrical properties of the muscle as there was no difference in input resistance (OK371/+, 3.37 ± 0.69 MΩ, n = 8; OK371 > mnb = 3.99 ± 0.99 MΩ, n = 8, P = .613) or resting potential (OK371/+, −69.6 ± 1.48 mV, n = 8; OK371 > mnb, −67.6 ± 1.55 mV, n = 8, P = .365). Although we did not directly investigate the source of the more frequent smaller mEJPs, the notion that they are due to the observed selective increase in the number of 1b boutons is suggested by the fact that 1b-dependent mEJPs are smaller than 1 s-dependent mEJPs (Newman et al., 2017). In parallel with the changes in spontaneous synaptic events, there was a small (~11%) decrease in the mean amplitude of electrically-evoked excitatory junction potentials (EJPs) caused by single stimuli applied to the nerve at a low frequency (0.1 Hz) (Fig. 4C). There was no difference between EJPs in mean rise time (OK371/+, 2.67 ± 0.178 ms, n = 8; OK371 > mnb, 3.23 ± 0.33 ms, n = 8, P = .728) or mean time constant of decay (OK371/+, 44.9 ± 3.1 ms, n = 8; OK371 > mnb, 36.6 ± 4.6 ms, n = 8, P = .154). The relatively small fall in EJP amplitude is likely to reflect the smaller size of the 1b-dependent component of the EJP relative to that of the 1 s-dependent component (Newman et al., 2017).

Overexpression of mnb in motor neurons slowed recovery from frequency-dependent depression at the larval NMJ
To investigate the effects of neuronal mnb overexpression on recycling of synaptic vesicles during electrically-evoked transmitter release, EJPs were evoked with pairs of electrical stimuli separated by intervals of varying duration (10 ms -10 s) or with repeated trains of 10 stimuli applied at a high frequency (10 Hz, a frequency 100 times higher than that at which the single EJPs were evoked) (Kauwe and Isacoff, 2013). At control NMJs, paired pulses separated by intervals shorter than 200 ms caused depression of the amplitude of the second EJP relative to that of the first and the depression was stronger for shorter inter-stimulus intervals (Fig. 5A). The dependence of pairedpulse depression on interval duration was unaltered in OK371 > mnb larvae (Fig. 5A), indicating that mnb overexpression did not alter release from a readily releasable pool of vesicles (Regehr, 2012). When transmitter release was evoked at control NMJs with a train of 10 stimuli at 10 Hz, there was rapid depression of the EJP amplitude bỹ 20% within the first 3 events (Fig. 5B). In the one-minute interval before the next train, the EJP amplitude recovered fully so that the amplitude of the first EJP in the second train was the same as in the first train (Fig. 5B). This ability to recover did not wane during the recording; the amplitude of the first EJP in each train did not differ between 8 trains (Fig. 5B). These effects are consistent with previous studies (Kauwe and Isacoff, 2013) and confirm rapid depletion and replenishment of the readily releasable pool of vesicles (Regehr, 2012). However, the same pattern of nerve stimulation produced different effects at OK371 > mnb NMJs (Fig. 5B). The percentage decrease in EJP amplitude during each train was the same as at control NMJs, but the depression was not fully reversed during the intervals between trains, so that the first EJP in each train was smaller than the first EJP in the preceding train. The depression in amplitude accumulated over the 8 trains, resulting in an overall fall of 10%. To confirm that the changes in EJP amplitude were due to presynaptic changes in transmitter release and were not postsynaptically mediated by a decrease in the unitary depolarisations comprising each EJP, we measured the amplitudes of 200 mEJPs immediately before and 200 mEJPs immediately after the series of trains at each NMJ. At both control and OK371 > mnb NMJs, the cumulative distribution of mEJP amplitudes before and after a series of trains was similar; although they were not identical, the observed slight increase in the number of larger mEJPs cannot explain the decline in EJP amplitude (Fig. 5C). These results show that mnb overexpression slowed replenishment of the readily releasable pool of vesicles, an effect consistent with the reported slowing of endocytosis of transmitter vesicles by DYRK1A overexpression (Kim et al., 2010).

Discussion
This study demonstrated that neuronal overexpression of mnb, the Drosophila ortholog of DYRK1A, is sufficient to induce motor impairment, accelerate age-related decline in motor performance, shorten lifespan and cause age-dependent neurodegeneration. This study also found that neuronal mnb overexpression at a glutamatergic synapse alters presynaptic structure, modifies basal synaptic transmission and delays recovery from short-term synaptic depression. This provides useful information about the gene's function and the pathological effects of increased expression in a model system. However, it is important to note that this does not represent a high-fidelity recapitulation of DS or the complexity of the human DYRK1A locus, because Gal4mediated overexpression of a specific isoform does not accurately replicate the expression level or pattern caused by triplication of a whole genomic region of human chromosome 21. People with DS have impaired motor skills which are evident from childhood and are caused by abnormal development of the nervous system (Malak et al., 2015;Stagni et al., 2018). Later, in middle age, they undergo faster age-dependent motor decline, which is an early marker of future dementia, comorbidities and mortality, and is likely caused by histopathological changes in the brain (Anderson-Mooney et al., 2016; Buchman and Bennett, 2011). The life expectancy of people with DS is about 28 years shorter than the general population (O'Leary et al., 2018). By taking advantage of the relatively short life cycle of Drosophila and transgenic overexpression of mnb in neurons, we have demonstrated a potential role for neuronal DYRK1A overexpression in the accelerated age-dependent decline of motor function and shortening of life expectancy in DS. The genetic basis of these aspects of DS are more difficult and costly to explore in mouse models of DS, due to the time required to study aged mice. Our finding that mnb overexpression causes age-related neurodegeneration confirms previous studies inferring a link between DYRK1A overexpression and degeneration and loss of neurons (Duchon and Herault, 2016;García-Cerro et al., 2017;Watson-Scales et al., 2018;Wegiel et al., 2008), which is associated with faster age-related decline in motor and cognitive function in DS and AD-DS. Our results also reinforce the conclusion from earlier studies with adult mice overexpressing DYRK1A or Dyrk1a, alone or as part of a chromosomal segment, that triplication of DYRK1A is likely to contribute to motor deficits in DS (Altafaj et al., 2001;Arque et al., 2013;Garcia-Cerro et al., 2018;Martínez de Lagrán et al., 2004;Ortiz-Abalia et al., 2008;Souchet et al., 2014;Watson-Scales et al., 2018).
In addition to the smaller brain size and fewer brain neurons in DS and mouse models of DS, there are alterations in the structure of brain synapses that are predicted to modify synaptic function (Contestabile et al., 2017;Dierssen, 2012;Stagni et al., 2018). A previous study of paired-pulse ratio (EJP 2 /EJP 1 , mean ± SEM, n = 8 for each genotype) against inter-pulse interval reveal no difference in synaptic depression (F (9, 126) = 0.1343; n = 8; P = .9987, repeated measures two-way ANOVA). (B) High-frequency stimulation protocol. 10 EJPs were evoked at 0.1 Hz to establish a mean baseline amplitude followed by 8 trains of 10 EJPs at 10 Hz, at one minute intervals. (Left) Representative traces of train 1 (upper) and train 8 (lower) recorded from OK371/+ and OK371 > mnb NMJs. Dashed lines compare EJP 1 in each train. (Right) Plots of EJP amplitude during trains 1 and 8, expressed as a fraction of baseline amplitude (mean ± SEM, n = 8 for each genotype). During the first train, the decline in EJP amplitude was unaffected by genotype (F (1, 14) = 0.22, P = .6486, repeated measures two-way ANOVA). In the eighth train, EJPs were smaller at OK371 > mnb NMJs (F (1,14) = 4.98, P = .0426) but the rate of decline during the 8th train was not different (F (9, 126) = 0.99, P = .4540, repeated measures two-way ANOVA). (C) Cumulative frequency distributions of mEJP amplitudes (n = 1600 from 8 NMJ recordings) from immediately before (solid line) and immediately after (dashed line) the trains, for OK371/+ (left) and OK371 > mnb (right) NMJs; after the trains, a minority of mEJPs were larger in both OK371/+ (*P = .0117) and OK371 > mnb (*P = .018) NMJs (log-rank (Mantel-Cox) test). Lowe et al. Neurobiology of Disease 125 (2019) 107-114 showed that DYRK1A overexpression in mice changes postsynaptic morphology in the cortex and in cultured cortical neurons by reducing the number and length of dendrites and by reducing the number of dendritic spines but elongating their shape (Martinez de Lagran et al., 2012). It also decreased the number of synapses formed. Our study shows that mnb overexpression changes presynaptic structure and that this happens in a neuron-specific manner; mnb overexpression in the two glutamatergic motoneurons innervating the larval NMJ increased the number of 1b boutons without changing the number of 1 s boutons. These data are consistent with a previous study which demonstrated that reduced levels of mnb caused a decrease, and increased levels of the mnb-F transcript an increase, in the number of boutons at the NMJ (Chen et al., 2014), but did not differentiate between 1b and 1 s boutons.

S.A.
The cognitive and motor deficits in DS arise from aberrant information processing in the brain that is likely due, in part, to changes in synaptic transmission or synaptic plasticity. Individuals with DS have impaired synaptic plasticity in the motor cortex (Battaglia et al., 2008). Our finding that mnb overexpression slows replenishment of the readily releasable pool of vesicles, and also modifies basal synaptic transmission, confirms a previous suggestion that DYRK1A overexpression contributes to synaptic dysfunction and cognitive deficits associated with DS, made on the basis of the observed slowing of endocytosis of transmitter vesicles in cultured mouse hippocampal neurons overexpressing human DYRK1A (Kim et al., 2010). The effects of DYRK1A on synaptic function may be splice variant specific as we found that overexpression of the mnb-H transcript caused a decrease in mEJP and EJP amplitude, whereas overexpression of mnb-F in a previous study did not alter mEJP or EJP amplitude at the larval NMJ (Chen et al., 2014).
The effects of neuronal mnb overexpression on larval NMJ function replicate some, but not all, the documented changes in glutamatergic synaptic transmission in the brain of mouse models of DS. These include a decrease in the amplitude of spontaneous excitatory postsynaptic currents (sEPSCs) in neocortical neurons of Ts65Dn mice (Cramer et al., 2015), compromised glutamate release in response to stimuli trains at hippocampal CA1 synapses of Ts1Cje mice (Siarey et al., 2005) and a decrease in EPSC amplitude in hippocampal CA3 neurons of Ts65Dn mice (Hanson et al., 2007). However, in contrast to the increase in mEJP frequency caused by mnb overexpression at the larval NMJ, electrophysiological studies have found a decrease in the frequency of mEPSCs in hippocampal CA3 neurons of Ts65Dn mice, sEPSCs in neocortical neurons of Ts65Dn mice and sEPSCs in neurons derived from trisomy 21 induced pluripotent stem cells, or no change in mEPSC frequency in the prefrontal cortex of TgDyrk1a mice or mossy fibre-CA3 synapses in Tc1 mice (Contestabile et al., 2017).
Our study further elucidates the effect of DYRK1A overexpression in a model system, giving insight into the contribution of increased dosage to various DS phenotypes. It supports the future development of pharmacological inhibitors of DYRK1A as treatments for multiple aspects of DS and DS-AD (Duchon and Herault, 2016;Stringer et al., 2017). Further work is necessary to fully understand interactions between DYRK1A and other triplicated Hsa21 genes in DS, in specific cell types and during defined periods of development and aging.

Behaviour
mnb expression was driven throughout the nervous system using Elav-Gal4 (Robinow and White, 1991) for experiments investigating behaviour of wandering third instar larvae, the number of boutons at the larval NMJ, and synaptic transmission at the larval NMJ. All behavioural experiments took place at 25°C. Larval locomotor experiments were conducted on a 9.5 cm petri dish containing 1.6% agarose. A single third instar wandering larva was selected, washed in a drop of distilled H 2 O, transferred to the agarose and allowed 30 s to acclimatise. To analyse free movement, the dish was placed over a 0.5 cm grid and the number of lines the larva crawled across in one minute was counted by eye. The self-righting assay was conducted as described elsewhere (Lowe et al., 2018;Picao-Osorio et al., 2015); the larva was gently rolled onto its back on the agarose using a fine moistened paintbrush, held for one second and released, and the time for it to right itself was recorded.
The negative geotaxis assay was performed as described previously (Ali et al., 2011). A cohort of 10 flies was transferred without anaesthesia to an empty 9.5 cm tube with a line drawn 2 cm from the top. After 1 min acclimatisation, the vial was sharply tapped 3 times to knock the flies to bottom. The number of flies to climb past the line within 10 s was recorded. 15 cohorts of 10 flies were tested for each genotype. Age-dependent changes in climbing were assessed by repeating the negative geotaxis assay at 10, 20 and 30 days post-eclosure (Gargano et al., 2005). For the survival assay, 10 cohorts of 10 oncemated females were transferred to a vial of fresh food twice weekly and the number of surviving flies recorded at each transfer.

Neurotoxicity
Overexpression of mnb was driven in the eye using the Glass multimer reporter (GMR-Gal4). Images of the whole head of 1-2 day old flies were taken via a Zeiss AxioCam MRm camera attached to a stereomicroscope (Zeiss SteREO Discovery.V8, up to 8× magnification), and the surface area of the eye was calculated by manually outlining the eye in ImageJ (rsb.info.nih.gov/ij/). Overexpression of mnb in GFP-tagged ellipsoid body (EB) ring neurons was achieved by crossing EB1-Gal4; UAS-mCD8-GFP flies with UAS-mnb flies. Following published methods (Williamson and Hiesinger, 2010), adult brains were dissected, fixed for S.A. Lowe et al. Neurobiology of Disease 125 (2019) [107][108][109][110][111][112][113][114] 30 min in 4% paraformaldehyde and mounted on a coverslip in Vectashield (Vector Laboratories). Slides were imaged on a Leica SP5-II confocal laser scanning microscope using an oil immersion 40 × objective. A Z-stack of 25 images at 1 μm increments was captured and combined into a 3-D projection using ImageJ (rsb.info.nih.gov/ij/); analysis was performed by scrolling through all 25 images and counting the number of cells in one brain hemisphere.

Electrophysiology
Wandering third instar larvae were dissected as for antibody staining. The motor nerves were severed just below the ventral ganglion and the brain was removed. CaCl 2 (1 mM) was added to the bath solution for intracellular recording from muscle 6 of abdominal segments 2-4. Most recordings were made in the presence of thapsigargin (2 μM), which minimises contraction and hence facilitates intracellular recording, without affecting amplitudes of mEJPs or EJPs (Guerrero et al., 2005;Newman et al., 2017). Sharp microelectrodes (thick-walled borosilicate capillaries, pulled on a Sutter Flaming/Brown P-97 micropipette puller) were filled with 3 M KCl and had resistances of 20-30 MΩ. For recording of stimulus evoked excitatory junction potentials (EJPs), severed nerves were drawn into a thin-walled glass-stimulating pipette and stimulated with square-wave voltage pulses (0.1 ms, 10 V, A-M Systems Model 2100 Isolated Pulse Simulator), 10 times at 0.1 Hz. EJPs and spontaneously-occurring miniature EJPs (mEJPs) were recorded at a controlled room temperature of 22-25°C with a Geneclamp 500 amplifier (Axon Instruments) and were further amplified with a LHBF-48× amplifier (NPI Electronic). The membrane potential was allowed to stabilise for one minute, the initial value was recorded, and then set to −70 mV by injecting current with the Geneclamp 500 amplifier. The muscle input resistance was measured by injecting current using the Axon Geneclamp 500, to bring the membrane potential to −100, −80, −60 and − 40 mV, and subtracting the electrode resistance from the slope of the resulting voltage/current graph. Voltage signals were lowpass filtered at 1.67 kHz (10 kHz 4 pole Bessel on Geneclamp 500, 1.7 kHz 8-pole Bessel on LHBF-48×) and digitised at 25 kHz by a CED-1401 plus A/D interface (Cambridge Electronic Design, UK) using Spike2 software (v. 5.13) (CED, Cambridge, UK). Recordings were discarded if the initial resting membrane potential was more positive than −60 mV or varied by > 10% throughout the recording. Synaptic potentials were analysed off line using Strathclyde Electrophysiology Software WinEDR (v3.5.2) and GraphPad Prism (v.6). All synaptic events were verified manually.
Amplitudes and intervals of mEJPs were compared by creating a cumulative distribution for each genotype of 1600 measurements across 8 animals, with each animal contributing 200 values. To analyse the mEJP waveform, a mean mEJP was constructed for each recording from events showing a single clear peak and a smooth decay, so as to prevent distortion of the waveform by closely occurring mEJPs. A single exponential was fitted to the decay of the mean mEJP and the 10-90% rise-time was measured. Time zero for the exponential fit was set to the time at the peak of the mEJP. EJPs were analysed by forming a mean of 10 events, measuring the 10-90% rise-time of the mean event, and fitting the decay with the sum of three exponentials (time zero was set at the time of the peak). A mean weighted time constant of decay was calculated as A 1 .τ 1 + A 2 .τ 2 + A 3 .τ 3 , where A 1 , A 2 and A 3 are the fractional amplitudes of the three components, and τ 1 , τ 2 and τ 3 are their time constants.
For paired pulse analysis, two EJPs were evoked, separated with intervals varying in duration between 0.01 and 10 s. The amplitude of the second event was calculated as a fraction of the amplitude of the first. Pairs of stimuli were separated by 30 s. For high-frequency stimulation, trains of 10 events evoked at 10 Hz were repeated 8 times at 1 min intervals. The amplitude of each event was expressed as a fraction of a baseline value, defined as the mean amplitude of 10 single EJPs evoked at 0.1 Hz. To compare mEJP amplitude before and after the trains, 200 mEJP amplitudes were measured immediately before the first train, and another 200 immediately after the eighth train. The measured values were pooled from 8 NMJs for each genotype and cumulative amplitude distributions compared.

Statistical analysis
Statistical analysis was conducted in GraphPad Prism (v. 6, La Jolla, CA). Data were tested for normality using the Kolmogorov-Smirnov test; where appropriate means were compared using Student's unpaired t-test, or medians were compared using a Mann-Whitney U test. EJPs evoked by pairs or trains of stimuli were compared using repeated measures 2-way ANOVA. Cumulative distributions were compared with a Kolmogorov-Smirnov test. Survival curves were compared with a Mantel-Cox test. Data are given as median or mean ± SEM. n is given per genotype. An α level of P < .05 was considered significant.