Abstract
Pyridox(am)ine 5’-phosphate oxidase (PNPO) is a rate-limiting enzyme in converting dietary vitamin B6 (VB6) to pyridoxal 5’-phosphate (PLP), the biologically active form of VB6, and involved in the synthesis of neuro-transmitters including GABA, dopamine, and serotonin. In humans, PNPO mutations have been increasingly identified in neonatal epileptic encephalopathy and more recently also in early-onset epilepsy. Till now, little is known about the neurobiological mechanisms underlying PNPO-deficiency-induced seizures due to the lack of animal models. Previously we identified a c.95 C > A missense mutation in sgll - the Drosophila homolog of human PNPO (hPNPO) and found mutant (sgll95) flies exhibiting a lethal phenotype on a diet devoid of VB6. Here we report the establishment of both sgll95 and ubiquitous sgll knockdown (KD) flies as valid animal models of PNPO-deficiency-induced epilepsy. Both sgll95 and sgll KD flies exhibit spontaneous seizures before they die. Electrophysiological recordings reveal that seizures caused by PNPO deficiency have characteristics similar to that in flies treated with GABA antagonist picrotoxin. Both seizures and lethality are associated with low PLP levels and can be rescued by ubiquitous expression of wild-type sgll or hPNPO, suggesting the functional conservation of the PNPO enzyme between humans and flies. Results from cell type-specific sgll KD further demonstrate that PNPO in the brain is necessary for seizure prevention and survival. Our establishment of the first animal model of PNPO deficiency will lead to better understanding of VB6 biology, the PNPO gene and its mutations discovered in patients, and can be a cost-effective system to test therapeutic strategies.
Introduction
Mutations in an autosomal gene encoding pyridox(am)ine 5’-phosphate oxidase (PNPO) cause neonatal epileptic encephalopathy (NEE, OMIM #610090), a severe neurological disease that leads to death if untreated [1]. Furthermore, PNPO mutations have more recently been identified also in patients with infantile spasms [2] and early-onset epilepsy [3, 4, 5]. While the causal relationship between PNPO deficiency and NEE has been established on the molecular level since 2005 [1], the neurobiological mechanisms of how PNPO deficiency leads to NEE and related epilepsy syndromes remain to be established.
PNPO is a rate-limiting enzyme in the synthesis of vitamin B6 (VB6), which comprises a group of six different forms including pyridoxine (PN), pyridoxal (PL), pyridoxamine (PM), pyridoxine 5’-phosphate (PNP), pyridoxal 5’-phosphate (PLP), and pyridoxamine 5’-phosphate (PMP) [6, 7]. PNPO converts PNP and PMP to PLP, the biologically active form of VB6, which is a co-factor for more than 140 different enzymes that are involved not only in the metabolism of amino acids, glycogen, and unsaturated fatty acids but also in the synthesis of various neurotransmitters including γ-aminobutyric acid (GABA), dopamine, and serotonin [8]. PNPO activity is essential to humans and other animals because unlike plants, fungi, and bacteria, they cannot synthesize VB6 de novo [9]. PNPO is highly expressed in liver, brain, and kidney in mammals [10] and previous studies have implicated the liver [11] as the primary site of conversion of dietary VB6 to PLP, which is further transported by circulation to different tissues/organs, including the brain [12, 13]. However, the contribution of brain-expressed PNPO to seizures and lethality, is unknown.
In recent years, PNPO mutations have been increasingly reported in human patients [1, 3, 5, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28]. Seizures in PNPO deficient patients usually do not respond to conventional anti-epileptic drugs but is responsive exclusively to PLP, or to PN in patients with partial PNPO deficiency [17, 20, 23]. Despite its effectiveness in seizure control, PLP or PN treatment has little effect on other neurological defects observed in PNPO deficiency patients, such as developmental delay or intellectual disability [20, 24, 26]. Moreover, long-term PLP treatment causes hepatic cirrhosis [29, 30]. There is a need for valid animal models to study the fundamental biology of PNPO and VB6, their role in neurodevelopment and neurotransmission, and to explore other treatment options for PNPO deficiency.
Based on the nutritional conditional lethal phenotype, we have recently identified a Drosophila melanogaster homolog gene of human PNPO (sgll) and identified a mutant with partial PNPO deficiency (sgll95) [31]. The sgll95 flies survive well on a normal diet, but due to their low PNPO activity, they exhibit lethality when reared on a sugar-only diet (4%sucrose in 1% agar, i.e., diet devoid of VB6). This nutritional conditional lethal phenotype can be rescued by the supplementation of either PN or PLP. Here we report that human and Drosophila PNPOs are functionally conserved and that PNPO deficiency in Drosophila causes seizures that are defined by both behavioral and electrophysiological parameters. These PNPO-deficiency-induced-phenotypes are associated with low levels of PLP and PNPO deficiency restricted to the neural tissue is sufficient to cause seizures and lethality.
Results
sgll95 flies exhibit seizure-like behavior before they die
Consistent with our previous report [31], we found that sgll95 homozygotes reared on a sugar-only diet, consisting of 4% sucrose in 1% agar, exhibited a striking lethality phenotype compared to w1118 counterparts. As shown in Figure 1A, nearly all sgll95 mutants had died after four days while no control flies had died. Before death (2 - 3 days on the sugar-only diet, green box in Figure 1A), sgll95 mutants displayed bouts of rapid ‘wing-buzzing’, and ‘body-rolling’, and later on, they often exhibited a ‘wings-up’ posture (Figure 1B). These seizure-like behavioral phenotypes occurred spontaneously and did not appear to be initiated by mechanical shock [32, 33, 34, 35] or temperature stress [36, 37, 38, 39, 40, 41].
To examine these behaviors, we recorded individual sgll95 or w1118 control flies walking in a 60 mm-wide open arena with a high-resolution video camera for 3 min (See Methods for details). Fly positions were then tracked with the IowaFLI Tracker [42], and the corresponding travel traces from each fly were plotted. We observed a smooth travel trace from each w1118 control fly (n = 22 on sugar; n = 19 on the normal diet) or from sgll95 fly on the normal diet (n = 14). However, travel traces of sgll95 flies reared on the sugar-only diet (n = 21) were heterogeneous with some traces similar to that of controls whereas others did not show smooth trajectories at all. Representative traces are shown in Figure 1C.
To quantify these travel traces, we plotted each trace on a scatter plot where the speeds of a fly during a frame t are plotted against the speeds in the next frame t+1 (Figure 1D) and then we calculated the speed correlation coefficients. The reasoning is that a smooth travel trace indicates consistent locomotion, i.e., similar speeds at two consecutive time points. On the scatter plot, all points from a smooth trace would land on the lower left to upper right diagonal line and give a high correlation coefficient of speed. On the opposite side, the less smooth travel traces will have their points spread out from the diagonal line and give lower correlation coefficients. Indeed, the average correlation coefficient is 0.654 ± 0.116 (Mean ± SD), 0.672 ± 0.120, 0.701 ±0.140, and 0.493 ± 0.260 for w1118 on the normal diet, w1118 on the sugar-only diet, sgll95 on the normal diet, and sgll95 on the sugar-only diet, respectively. Therefore, consistent with qualitative distinctions between sgll95 and w1118 flies, we found significantly reduced speed correlation coefficients in sgll95 when reared on the sugar-only diet (Figure 1E, P = 0.0047, gene by condition interaction P = 0.0065).
Ubiquitous RNAi-mediated knockdown of sgll leads to lethality and seizure-like behavior when reared on the sugar-only diet
We have previously shown that ubiquitous sgll knockdown (KD, i.e. actin-Gal4/UAS-sgll RNAi) flies also show the conditional lethal phenotype, which can be rescued by PLP but not PN [31], indicating severe PNPO deficiency in these flies compared to that in sgll95 flies. Indeed, sgll KD flies survived shorter than sgll95 flies (compared males to males); all sgll KD flies died within four days (Supplemental Figure S1A) while all sgll95 flies died within five days (Figure 1A). Prior to death, sgll KD flies displayed seizure-like behavior similar to that in sgll95 flies including the wing buzzes, wings-up posture and convoluted walking trajectories (Figure S1 B&C). The average speed correlation coefficient of sgll KD flies is 0.423 ± 0.240 (Mean ± SD), which is significantly lower than that from the parental controls (0.644 ± 0.139, and 0.645 ± 0.233 for actin-Gal4/+ and UAS-sgll RNAi/+, respectively. Figure S1 D; P = 5.947e-05).
Spontaneous seizure discharges in sgll95 and sgll KD mutants reared on the sugar-only diet
To monitor electrophysiological activity and identify aberrant patterns in sgll mutants, we undertook an analysis of flight muscle activity in a tethered fly preparation (Supplemental Video 1) [43, 44]. The isometric contractions of the dorsal longitudinal muscle (DLM) enable prolonged recordings of spiking activity with minimal damage to the muscle. Several studies have taken advantage of this stable recording preparation to monitor aberrant spike discharges associated with seizures that occur spontaneously in mutant flies [40, 41], seizures that are triggered by high-frequency electroconvulsive stimulation [43, 45, 46] or by proconvulsant administration [47]. Importantly, the bouts of DLM spiking activity during seizure discharges appears to be temporally correlated with extracellular synchronous activity across different body-axes [46].
We recorded spontaneous DLM spiking activity not associated with flight from sgll95, sgll KD flies and w1118 control flies, reared on either the normal diet or the sugar-only diet (Figure 2A). Flies reared on the normal diet displayed short bouts of a few spikes associated with wing depression events during grooming activity (Supplemental Video 1, also see reference [47]). Over the recording session (∼240 s) the average firing frequency (total # spikes / recording duration) was ∼0.3 Hz, for the three genotypes tested (Figure 2B). In a striking contrast, recordings from both sgll95 and sgll KD flies reared on the sugar-only diet showed high-frequency spike burst discharges qualitatively distinct from the grooming-related spiking seen in their counterparts reared on the normal diet (Figure 2A), and there was also significantly more firing (median firing rate of: 2.03 Hz.- vs. 0.32 Hz.- for sgll95, and 7.91 vs. 0.07 Hz for sgll KD flies, P = 0.0045 and P = 1.138e-6, for sgll95 and sgll KD, respectively. Figure 2B). These spike bursts occurred simultaneously in the left and right muscles (data not shown), suggesting that they were triggered by relatively widespread events across the nervous system. Notably, our sample included sgll95 and sgll KD flies with ‘wings-up’ as well as normal wing posture (Figure 1B). We found that between our recordings of sgll95 and sgll KD flies, all wings-up flies displayed spike bursts (6 / 6), and several individuals with normal wing posture did that as well (5 / 8 flies), suggesting that the burst discharge phenotype likely initiates prior to appearance of the ‘wings-up’ posture.
An important characteristic of the DLM spike trains recorded was the highly variable nature of inter-spike intervals (ISIs). Within a spike train, the instantaneous firing frequency between two spikes (defined as the inverse of the ISI, i.e. ISI−1) could range from less than 1 Hz to more than 100 Hz within a bout. To quantitatively delineate the grooming-related spike patterns, we employed a phase-space analysis of ISIs [47]. For the sequence of spikes within the recording, we plotted the ISI−1 of each spike against the spike’s instantaneous coefficient of variation, CV2, a measure of variability between successive ISI−1 values (see Methods for computational details). The resulting ISI−1 vs CV2 trajectories have been shown to clearly distinguish between spike patterns associated with flight, grooming, and electroconvulsive stimulation-induced seizure discharges in Drosophila [47].
The ISI−1 vs. CV2 trajectories of the burst discharges in sgll mutants on the sugar-only diet were readily distinguished from the spiking activities in counterparts on the normal diet and in w1118 flies (representative trajectories shown in Figure 2C). Specifically, trajectories of sgll95 flies on the sugar-only diet display ‘loops’ consisting of the ISI−1 of the trajectory with abrupt acceleration (often peaking between 50 and 100 Hz) followed by a gradual deceleration characterized by relatively low CV2 values. A few sgll KD flies displayed more extreme trajectories which were more compact, reflecting continuous high-frequency firing. In contrast, the ISI−1 vs. CV2 trajectories of sgll mutants on the normal diet as well as that of w1118 flies generally ambled within a limited region of higher CV2 values (> 0.5) reflecting a high degree of variability in successive spike intervals, consistent with observations of grooming spike patterns in other wild-type fly strains [47].
Seizures and lethality are correlated with low levels of internal PLP
PNPO converts intracellular PNP and PMP to PLP (Figure 3A), and PNPO deficiency is expected to result in a low level of PLP. However, both normal and decreased PLP levels have been reported in human PNPO deficient patients [26, 48]. Such inconsistency is conceivably due to confounding factors such as heterogeneous genetic backgrounds and different diets [3]. Since both genetic background and dietary conditions were well controlled in flies, we reasoned that by comparing the PLP level between sgll95 and w1118 control flies we should be able to characterize the impact of PNPO deficiency on VB6 metabolism. Furthermore, seizures in PNPO deficient patients cease with PLP or PN treatment [17, 20, 23, 24]. Our previous studies have also shown that PLP or PN supplementation can rescue the conditional lethality of sgll95 flies [31]. To further correlate the PLP level with seizures and lethality, we measured PLP in flies reared on the sugar-only diet as well as the sugar-only diet supplemented with PN.
As shown in Figure 3B, when reared on the sugar-only diet (S), sgll95 flies had significantly reduced PLP compared to w1118 flies (red bar vs. black bar, P = 4.020e-5), which was significantly improved by PN supplementation (red vs. pink bar, P = 0.0035). On the other hand, PN supplementation did not change the PLP level in w1118 flies (black vs. gray bar, P = 0.9126), suggesting that PLP is highly regulated in normal flies. The regulation of PLP in w1118 flies is most likely mediated through the conversion from PLP to PL as indicated by a significantly increased PL level after PN supplementation in w1118 flies compared to sgll95 flies (gene by treatment interaction: P = 0.0115; P = 0.0004 and P = 0.0988 the treatment effect in w1118 and sgll95 flies, respectively). In comparison, PMP and PA levels show no difference between w1118 and sgll95 flies before PN supplementation (P = 0.6914 and P = 0.5305 for PMP and PA, respectively). After PN supplementation, PA is significantly increased in w1118 flies compared to sgll95 flies (gene by treatment interaction: P = 0.0198; P = 0.0504 and P = 0.8017 for the treatment effect in w1118 and sgll95 flies, respectively), whereas PMP is similar between the two genotypes (gene by treatment interaction: P = 0.4062; P = 0.5159 and P = 0.9980 for the treatment effect in w1118 and sgll95 flies, respectively).
PNPO is functionally conserved between humans and flies
Amino acid sequence alignment analysis revealed that the protein product of sgll shares more than 75% similarity with human PNPO [31]. To further study if the PNPO enzyme is functionally conserved between humans and flies, we generated transgenic flies by sub-cloning cDNAs from wild-type (WT) human and Drosophila PNPO gene into a pUAST vector (UAS-hPNPO and UAS-sgll, respectively). We used the GAL4/UAS system [54] to drive the expression of transgenes. To examine the effect of the ubiquitous expression of each transgene, we first backcrossed each UAS line and an actin-Gal4 line (genotype: actin-Gal4/CyO) to the sgll95 background. We then crossed each backcrossed UAS line with the backcrossed actin-Gal4 line to generate flies in which the transgene was ubiquitously expressed on the sgll95 background. The conditional survival rates of these flies (genotypes: actin-Gal4/UAS-sgll; sgll95/sgll95 flies and actin-Gal4/UAS-hPNPO; sgll95/sgll95) and control flies (genotypes: +/UAS-hPNPO; sgll95/sgll95, +/UAS-sgll; sgll95/sgll95, and actin-Gal4/+; sgll95/sgll95 and sgll95/sgll95) were examined.
In contrast to sgll95 homozygotes that died within five days on the sugar-only diet (Figure 4A, red curve), sgll95 flies with ubiquitous expression of WT sgll or hPNPO (i.e. rescue flies) displayed no mortality (blue or green circle versus red point-up triangle; P < 0.001 for both genotypes compared to sgll95 homozygotes) and did not show any seizure-like behavior. To confirm that, we performed behavioral characterization on rescue flies reared on the sugar-only diet for two or five days. As shown in Figure 4B-D, rescue flies behaved normally and had similar average speed correlation coefficients to sgll95 flies on the normal diet. The average speed correlation coefficient on Day 2 is 0.601 ±0.141 (Mean ± SD) and 0.702 ± 0.106 for actin-Gal4/UAS-sgll; sgll95/sgll95 (n = 14) and actin-Gal4/UAS-hPNPO; sgll95/sgll95 (n = 11), respec-tively. The average speed correlation coefficient on Day 5 is 0.722 ± 0.107 and 0.675 ± 0.160 for actin-Gal4/UAS-sgll; sgll95/sgll95 (n = 10) and actin-Gal4/UAS-hPNPO; sgll95/sgll95 (n = 10), respectively. These coefficients are not significantly different from those in sgll95 flies on the normal fly diet (P = 0.0766, P = 0.9866, P = 0.6873, and P = 0.6852, unpaired t-test). Furthermore, recordings of DLM activities from actin-Gal4/ UAS-sgll; sgll95/sgll95 and actin-Gal4/UAS-hPNPO; sgll95/sgll95 flies did not show the spiking burst discharges characteristic of sgll95 flies (Figure 4E, n = 7 for each genotype). Together, these data indicate that ubiquitous expression of hPNPO completely rescued the conditional seizures and lethality of sgll95 homozygotes and demonstrate the functional conservation of the PNPO enzyme in humans and flies.
PNPO in the brain is necessary for normal brain function
The brain obtains VB6 from the circulation [12, 13]. Previous studies have shown that PLP and PL are the primary forms of VB6 in the circulation [55, 56], raising the question whether PNPO in the brain is necessary for normal brain functions.
To answer this question, we generated neural-specific sgll KD flies using an elav-Gal4 driver [57]. KD and corresponding control flies were then tested on the sugar-only diet. We observed both lethality (Figure 5A, P < 0.001) and seizures (Figure 5B) in sgll KD flies, both of which were absent from controls. However, compared to sgll95 or ubiquitous sgll KD flies (Figure 1A and Figure S1 A), neural-specific sgll KD flies were not as severely impaired. One possible explanation is that neural-specific sgll KD was not efficient. Since it was not feasible to quantify cell type-specific sgll KD efficiency precisely, we generated neural-specific sgll knockout (KO) flies using guide RNA stocks generated by the Transgenic RNAi project (https://fgr.hms.harvard.edu) and tested them on the sugar-only diet. As shown in Figure S2, neural-specific sgll KO flies also showed lethality. The survival rates of neural-specific sgll KD flies and neural-specific sgll KO flies are comparable, suggesting that the knockdown efficiency was high. It’s worth noting that in the ubiquitous sgll KD flies, the knockdown efficiency was over 90% as measured by the mRNA level [31]. Therefore, expression of PNPO in neurons is essential for both seizure prevention and survival but PNPO expression in other cell types also contributes. Since PLP can be converted to PL, cell membranes are permeable to PL, and PL can be converted to PLP independent of PNPO, it is conceivable that PNPO functions are only partially cell-autonomous.
Discussion
The identification of a hypomorphic PNPO allele, sgll95, in Drosophila by our earlier work [31] has opened up opportunities to study PNPO functions in an animal model for the first time. Here we report that PNPO deficiency causes seizures in Drosophila, similar to human conditions associated with PNPO deficiency. The seizure and lethality phenotypes are associated with low PLP levels and are completely rescued by VB6 supplementation and by ubiquitous expression of either WT sgll or WT human PNPO (hPNPO). Cell type-specific KD further suggests that PNPO in the brain is necessary for seizure prevention and survival.
The amino acid sequence of PNPO is evolutionarily conserved [31, 58]. The fact that PNPO deficiency induces seizures in flies as it does in humans (Figures 1 & 2) demonstrates that the biological function of PNPO is also conserved. Complete rescue of the conditional lethality and seizures of sgll95 flies by ubiquitous expression of hPNPO further confirmed this notion (Figure 4). All these provide a solid foundation for studying the fundamental biology of PNPO (Figure 5) and for future characterizations of human-disease-associated hPNPO mutations in Drosophila models.
NEE caused by severe PNPO deficiency is lethal if un-treated. However, favorable outcomes are still possible if patients could be diagnosed and treated timely [59]. A sensitive and specific biomarker will undoubtedly help in the diagnosis. Given that PNPO determines the synthesis of PLP, the PLP level along with other B6 vitamers has been scrutinized in both PNPO deficient patients and individuals without PNPO deficiency. However, results have not been consistent for PLP level measurements [26], which can be attributed to the heterogeneous genetic background, different diets, as well as other uncontrolled factors. Recently, it has been shown that PNPO deficient patients have high plasma levels of PM [49], but the sample size is relatively small (n = 6) and majority of them (5 / 6) were on either PN or PN+PLP treatment. More samples are needed to validate this finding. In our fly models with well-controlled genetic background and environmental conditions, it is clear that PLP level is the best predictor of symptoms among B6 vitamers (Figure 3). It is worth noting that PMP instead of PM was detected in flies under the current conditions and the PMP level is not changed by PN supplementation and is not altered in sgll95 flies compared to w1118 controls.
Little is known about the neurobiological mechanisms underlying PNPO-deficiency-induced seizures. In humans, different metabolites have been measured in PNPO deficient patients, including those involved in the synthesis of dopamine, serotonin, GABA, and other amino acids such as threonine and glycine. A potential contributor to the seizure phenotype in PNPO deficiency flies is an altered inhibitory central circuit. Notably, our spike pattern analysis suggests that seizures in PNPO deficient flies (Figure 2) share characteristics with spike discharges in wild-type flies administered with picro-toxin, a GABAA antagonist [47], supporting the notion of a potential role of dysfunctional GABAergic neurons as a consequence of PNPO deficiency. This is not surprising as PLP is required for GABA synthesis and GABA deficiency is known to be involved in the development of epilepsy.
This interpretation is also in agreement with results from animal models of other gene mutations in the VB6 pathway. Other than PNPO, several genes have been identified in VB6-dependent epilepsy patients, including ALDH7A1 [60], ALPL [61], and PROSC [62, 63], which encodes aldehyde dehydrogenase 7A1, tissue non-specific alkaline phosphatase (TNSALP), and proline synthetase co-transcribed homolog, respectively. These gene products affect the availability of PLP through different mechanisms. Knockout animal models have been generated to study the function of ALPL [64] and ALDH7A1 genes [65]. Those studies suggest that reduced intracellular PLP and in consequence, the GABA deficiency are likely the main contributors to their seizure phenotype.
Seizures have long been studied in Drosophila [32]. Fly models based on mutations in a number of genes have been established and well characterized, including genes encoding sodium, potassium, and calcium channels [66, 67, 39], as well as genes encoding proteins that can affect the functions of ion transporters [68, 69, 70]. The seizure activity in sgll is distinct from two major classes of hyperexcitable Drosophila mutants, bang-sensitive mutants (e.g. bss, eas [33]) and temperature-sensitive mutants (e.g. sei [38], cac [39, 40], Shu [41]), in which wing-buzzing and DLM spiking can be triggered by mechanical shock or high temperature exposure. In sgll mutants, wing-buzzing and DLM spike discharges appear spontaneously. Notably, the spontaneous DLM spike discharges in sgll mutants are reminiscent to spike patterns recruited by injection of the GABAA antagonist picrotoxin in WT flies [47] (Figure 2C). However, additional work is required to demonstrate the role of GABAergic circuitry in driving seizure discharges in sgll mutants.
Characterization of PNPO mutants makes it possible to examine genetic/functional interactions between PNPO and other proteins implicated in epileptogenesis. For example, VB6 has been shown to control seizures in human patients who carry mutations in KCNQ2 [71, 72] or CACNA1A [73], both of which encode voltage-gated ion channels. Detailed analysis of the VB6 metabolism pathway in known epilepsy models could reveal a better understanding of how seizures arise in these seizure-prone patients and potentially provide new therapeutic avenues to suppress seizure activity.
Epilepsy affects more than 70 million people in the world with one-third of them not well-controlled by drug treatments [74, 75]. Now with the advent of next-generation sequencing, treatments may be designed based on genetic information. The power of such an approach is especially exciting for de-fects in genes encoding metabolic enzymes as treatments may include simple dietary changes. Thus, valid animal models are valuable tools for testing treatment options, and help to elucidate the fundamental biology of these causal genes.
Methods
Drosophila strains
Electrophysiology experiments were performed on flies bred on Frankel & Brosseau’s media [76] at the University of Iowa. For all other experiments, flies were bred on the standard cornmeal-yeast-molasses medium from Fly Kitchen at the University of Chicago. Flies used in all experiments were raised and tested at room temperature (∼23 °C).The sgll95 line and its genetic control wild-type line, w1118, were described before [31]. The UAS-sgll RNAi transgenic line (VDRC #105941), as well as its control, was obtained from the Vienna Drosophila Resource Center (VDRC). The actin-Gal4 (BDSC #4414), elav-Gal4 (#25750), elav-Gal4; UAS-Cas9/CyO (#67073), and sgll gRNA (#79746) were obtained from the Bloomington Drosophila Stock Center (BDSC) at Indiana.
Generation of transgenic flies
To generate transgenic flies, we constructed UAS-sgll and UAS-hPNPO by sub-cloning the cDNAs of Drosophila wild-type sgll and human wild-type PNPO, respectively, into pUAST and injected into w1118 commercially (Rainbow Transgenic Flies, Inc). Specifically, fly WT sgll cDNA (RA form, see reference [31]) were amplified from w1118 cDNAs with primers: 5’-CGA ATT CGC CAC CAT GAA GTT GCT GCA AAC AAT TCG AAG G-3’ and 5’-AGA GCT CCT AAG GAG CCA GCC GTT CGT ACA CCC A-3’; and human WT hP-NPO cDNA was amplified from human brain cDNA library (TaKaRa, Cat #637242) with primers: 5’-CGA ATT CGC CAC CAT GAC GTG CTG GCT GCG GGG CGT CA-3’ and 5’-GGA GCT CTT AAG GTG CAA GTC TCT CAT AGA GCC AGT CTT-3’. The sequences of constructs were confirmed by Sanger sequencing at the DNA Sequencing Facility at the University of Chicago.
Survival study
We followed our previous method [31]. In brief, male flies, 1-3-day-old, were anesthetized briefly and transferred into a culture vial filled with 4% sucrose in 1% agar. Fifteen to twenty flies were grouped into a vial. Survival was checked daily.
Behavioral recording and data analysis
For recordings in sgll95 or rescue flies, 1-2-day-old male flies were anesthetized briefly and maintained on 4% sucrose in 1% agar for two or five days as indicated in the text. For recordings in ubiquitous sgll KD flies, we used female flies because male flies died sooner and gave a shorter time window to do recordings (Figure S1). Therefore, 1-2-day-old female sgll KD flies were anesthetized briefly and maintained on 4% sucrose in 1% agar for two days. After treatment, each individual fly was transferred without anesthetization to a 60-mm petri dish (Corning, Cat #430166) for recording. The dish was pre-filled with 1% agar to maintain humidity. Each fly was recorded for 3 min at 20 frames per second using a Flea3 video camera (FLIR Integrated Imaging Solutions, Inc). Videos were saved to the computer with FlyCapture (FLIR Integrated Imaging Solutions, Inc) and tracked with the IowaFLI Tracker [42]. After tracking, the travel traces and scatter plots were plotted in Matlab (R2018b, Mathworks, Natick, MA). For clarity, only the first five hundred frames were plotted in the scatter plots shown in Figures 1, Figure S1, and Figure 4.
Electrophysiological recording of DLM flight muscle activity
We followed the protocol from our previous publications [43, 46]. In brief, a single fly, either male or female, was shortly anesthetized on ice and glued to a tungsten wire between neck and thorax. After a 30-minute recovery period, two sharpened tungsten electrodes were inserted into the left and right thorax, one on each side, targeting the top-most dorsal longitudinal muscle (DLMa, [77]). A reference electrode was inserted into the abdomen of the fly. Electrical activity in each muscle were picked up with an AC amplifier (AM Systems Model 1800, Carlsbourg, WA) and digitized by data acquisition card (USB 6210 National Instruments, Austin TX) controlled by LabVIEW8.6 (National Instruments). Spike detection was done using a custom-written Matlab script. Following the conventions in our previous publication [47], the instantaneous firing frequency for an individual spike was defined as the reciprocal of the inter-spike interval between the current spike and the succeeding spike (ISI−1). The instantaneous coefficient of variation, CV2 [78] for a pair of ISI−1 corresponding to adjacent interval i and I +1 was shown as:
Lower CV2 values indicate ISI−1 values with little variability, and higher CV2 values correspond to irregular ISI−1 values [47].
B6 vitamer measurements
For measuring levels of B6 vitamers, a group of 20 male sgll95 or w1118 flies, 1-3-day-old, were picked into vials supplied with 4% sucrose in 1% agar, which was supplemented with or without 500 ng/ml of PN. The concentration of 500 ng/ml was selected based on our previous studies ([31]) show-ing that all flies survive under this condition. Forty-eight hours later, whole flies in each vial were homogenized in cold trichloroacetic acid. After centrifugation, the supernatant in each vial was collected into new Eppendorf tubes, which were frozen at −80 °C until B6 vitamer measurements. The concentration of each B6 vitamer in fly lysates was quantified by Ultra-Performance-Liquid-Chromatography-Tandem-Mass-Spectrometry (UPLC-MS/MS) [79].
Survival and seizures in neural-specific sgll KD and control flies
For survival, 1-3-day-old male flies, were anesthetized briefly and transferred into a culture vial filled with 4% sucrose in 1% agar. Fifteen to twenty flies were grouped into a vial. Survival was checked daily. We also recorded the number of flies that showed seizure phenotypes during a 2-minute-time window each day. Seizure phenotypes include wing-buzzing, body-rolling, and wings-up, all of which were described in Figure 1.
Statistical analysis
Statistical analysis was performed using R (version 3.3.2). Details on statistical analyses, including sample sizes, tests performed, exact P values, and multiple tests correction if necessary, are provided within the figure legends or in the text describing each figure.
Author contributions
W.C. and X.Z. conceived the project. W.C. wrote the first draft of the manuscript. All authors edited the manuscript. W.C. designed and conducted rescue experiments, behavioral characterization of seizures, and knockdown experiments. C.-F.W. and A.I. designed the electrophysiological studies. A.I. and W.C. conducted electrophysiological characterization of seizures and A.I. performed data analysis. M.A., M.B., and N.V.D. conducted VB6 measurements. W.C. and A.I. con-structed figures and performed statistical analyses.
Conflict of Interest Statement
The authors declare no competing conflict of interest.
Acknowledgments
We thank Dr. Edwin Ferguson (Department of Molecular Ge-netics and Cell Biology) for commenting on the manuscript. We thank Dr. Wei Du (Ben May Department for Cancer Research) for providing pUAST plasmids. We thank Qian Yang, Susie A. Turkson, Chris Wreden, Hanyu Li, Wenhan Chang, Wei Liu, and Wenqin Fu for technical support. We thank the Bloomington Drosophila Stock Center and the Vienna Drosophila Resource Center for Drosophila stocks. We thank Rainbow Transgenic Flies, Inc. for generating transgenic flies. This work was supported by the National Institutes of Health [T32MH020065 to W.C., R01GM100768 to X.Z., AG051513 to C.F.W]; the Donald F. Steiner Scholarship Fund (to W.C.); and Iowa Neuroscience Institute Fellowship (to A.I.).
Footnotes
Figure 5 updated.
References
- [1].↵
- [2].↵
- [3].↵
- [4].↵
- [5].↵
- [6].↵
- [7].↵
- [8].↵
- [9].↵
- [10].↵
- [11].↵
- [12].↵
- [13].↵
- [14].↵
- [15].↵
- [16].↵
- [17].↵
- [18].↵
- [19].↵
- [20].↵
- [21].↵
- [22].↵
- [23].↵
- [24].↵
- [25].↵
- [26].↵
- [27].↵
- [28].↵
- [29].↵
- [30].↵
- [31].↵
- [32].↵
- [33].↵
- [34].↵
- [35].↵
- [36].↵
- [37].↵
- [38].↵
- [39].↵
- [40].↵
- [41].↵
- [42].↵
- [43].↵
- [44].↵
- [45].↵
- [46].↵
- [47].↵
- [48].↵
- [49].↵
- [50].↵
- [51].↵
- [52].↵
- [53].↵
- [54].↵
- [55].↵
- [56].↵
- [57].↵
- [58].↵
- [59].↵
- [60].↵
- [61].↵
- [62].↵
- [63].↵
- [64].↵
- [65].↵
- [66].↵
- [67].↵
- [68].↵
- [69].↵
- [70].↵
- [71].↵
- [72].↵
- [73].↵
- [74].↵
- [75].↵
- [76].↵
- [77].↵
- [78].↵
- [79].↵