A loss of Pdxk model of Parkinson disease in Drosophila can be suppressed by Buffy

Background The identification of a DNA variant in pyridoxal kinase (Pdxk) associated with increased risk to Parkinson disease (PD) gene led us to study the inhibition of this gene in the Dopa decarboxylase (Ddc)-expressing neurons of the well-studied model organism Drosophila melanogaster. The multitude of biological functions attributable to the vitamers catalysed by this kinase reveal an overabundance of possible links to PD, that include dopamine synthesis, antioxidant activity and mitochondrial function. Drosophila possesses a single homologue of Pdxk and we used RNA interference to inhibit the activity of this kinase in the Ddc-Gal4-expressing neurons. We further investigated any association between this enhanced disease risk gene with the established PD model induced by expression of α-synuclein in the same neurons. We relied on the pro-survival functions of Buffy, an anti-apoptotic Bcl-2 homologue, to rescue the Pdxk-induced phenotypes. Results To drive the expression of Pdxk RNA interference in DA neurons of Drosophila, we used Ddc-Gal4 which drives expression in both dopaminergic and serotonergic neurons, to result in decreased longevity and compromised climbing ability, phenotypes that are strongly associated with Drosophila models of PD. The inhibition of Pdxk in the α-synuclein-induced Drosophila model of PD did not alter longevity and climbing ability of these flies. It has been previously shown that deficiency in vitamers lead to mitochondrial dysfunction and neuronal decay, therefore, co-expression of Pdxk-RNAi with the sole pro-survival Bcl-2 homologue Buffy in the Ddc-Gal4-expressing neurons, resulted in increased survival and a restored climbing ability. In a similar manner, when we inhibited Pdxk in the developing eye using GMR-Gal4, we found that there was a decrease in the number of ommatidia and the disruption of the ommatidial array was more pronounced. When Pdxk was inhibited with the α-synuclein-induced developmental eye defects, the eye phenotypes were unaltered. Interestingly co-expression with Buffy restored ommatidia number and decreased the severity of disruption of the ommatidial array. Conclusions Though Pdxk is not a confirmed Parkinson disease gene, the inhibition of this kinase recapitulated the PD-like symptoms of decreased lifespan and loss of locomotor function, possibly producing a new model of PD.

affected, including the locus coeruleus, the brain stem, raphe nucleus, dorsal motor nucleus of the vagus, basal nucleus of the Meynert, amygdala, and hippocampus [4]. PD is characterized by the presence of neuronal inclusions composed of abnormal α-synuclein and generally referred to as Lewy-related pathology [2,5]. This atypical protein accumulation is believed to lead to cellular toxicity and, eventually, the PD pathogenesis. A majority of PD cases are idiopathic but the emergence of familial cases led to the identification and study of genes that are highly associated with PD [6,7]. The understanding and exploitation of the genetic basis of PD has revealed over 20 genes that are implicated in PD pathogenesis [8], and highlighted the complexity of this neurodegenerative disease.
The link between vitamin B 6 and PD incidence has been explored for years, with some studies associating dietary vitamin B 6 with reduced effectiveness of Levodopa [9]. Other studies have shown the advantages of a higher dietary vitamin B 6 and the reduced risk of PD [10] or reported low dietary intake of vitamin B 6 with increased risk to PD [11], either via its antioxidant abilities or through dopamine biosynthesis. Vitamin B 6 is comprised of three pyridine derivatives or vitamers-which are chemical compounds that have a similar molecular structure and possess similar vitamin activity-known as pyridoxine (PN), pyridoxamine (PM), pyridoxal (PL) and their phosphorylated products pyridoxine-5′-phosphate (PNP), pyridoxamine-5′-phosphate (PMP) and pyridoxal-5′-phosphate (PLP) [12,13]. PLP is the most metabolically active form and responsible for more than 100 enzymatic reactions [12], predominantly in amino acid metabolism, and is implicated in nervous system function (neurotransmitter synthesis), red blood cell formation (heme biosynthesis), vitamin formation, one-carbon metabolism (nucleic acid synthesis) and as a potent antioxidant [14]. In neuronal function, PLP plays a key role in the metabolism of neurotransmitters, including dopamine, serotonin, glycine, GABA, glutamate, d-serine and histamine [12]. The deficiency of vitamin B 6 has been implicated in increased risk of cancer, neural decay and accelerated ageing. Mitochondrial oxidative decay is a major contributor to ageing [15,16]. Mitochondrial function is more dependent on PLP than any other organelle as PLP function as a coenzyme for transaminases that are involved in the catabolism of all amino acids by the urea cycle of the mitochondria [16]. PLP is involved in diverse biochemically important roles in the mitochondria including maintaining energy pathways, homocysteine and glutathione (an antioxidant) biosynthesis. The heme biosynthesis occurs predominantly in the mitochondria and depends on PLP as a coenzyme. The inadequate synthesis of heme can cause mitochondrial decay and oxidative DNA damage [15], whereas its inhibition can cause oxidant leakage, that increases cellular endogenous ROS formation. Vitamin B 6 has a direct antioxidant activity by preventing superoxide radical formation, glycated haemoglobin formation and erythrocyte lipid peroxidation [17]. The inter-conversion of the pyridines to the biologically active phosphate derivative PLP require the action of pyridoxal kinase, thus, the activation of vitamin B 6 to its active form, PLP, requires pyridoxal kinase.
Pyridoxal kinase (Pdxk) belongs to the phosphotransferase family of proteins that are involved in the phosphorylation of vitamin B 6 to pyridoxal-5-phosphate an important co-factor in intermediary metabolism [18,19]. They contain a ribokinase/pyridoxal domain and are highly conserved, being found in yeast, plants and animals. The association of the gene coding for Pdxk with Parkinson disease was through whole-genome expression profiling of human DA neurons, combined with association analysis in differentially regulated genes [20]. A DNA variant, single nucleotide polymorphism, in the Pdxk gene has been associated with an increased risk to PD [20], though other studies ruled out the association of the variant rs2010795 with PD in a cohort of patients [21]. The study did not rule out the existence of the Pdxk variants that may increase the risk for PD. The development of model systems to study potential therapies is important, and as such Drosophila melanogaster is a good model organism to study the pathophysiology of movement disorders [22]. The first Drosophila model of PD relied on the expression of human α-synucleinwhich has no known homologue in Drosophila-to induce the PD-like symptoms [23]. Its success anchors on the ability to recapitulate features of human PD such as (1) age-dependent loss in locomotor function (2) LBlike inclusions and (3) age-dependent loss of DA neurons; and is widely used in the study of α-synuclein-induced neurodegeneration [22][23][24][25][26][27]. The utilization of the UAS/ GAL4 spatio-temporal expression system [28], and the availability of a plethora of promoters or enhancers of which TH-Gal4, elav-Gal4 and Ddc-Gal4 are employed in modelling PD in flies, makes Drosophila a powerful model organism [22][23][24][25][26][27]. We have previously shown the pro-survival advantages of Buffy, the sole anti-apoptotic Bcl-2 homologue in Drosophila by its rescue of the α-synuclein-induced phenotypes [29], in the HtrA2 model of PD [30], we extended this study to investigate whether Buffy would rescue the loss of Pdxk-induced phenotypes in Drosophila.

Pdxk is evolutionarily conserved across diverse species
The bioinformatic analysis of protein sequences encoded by the human, mouse, mosquito and fruit fly Pdxk gene revealed a highly conserved ribokinase/pyridoxal kinase domain ( Fig. 1) as determined by NCBI Conserved Domain Database (CDD) [31] and Eukaryotic linear motif (ELM) resource [32]. Comparison of the human and the Drosophila homologues by NCBI's BLAST revealed a 46% identity and 64% similarity along the protein sequences. The alignment of protein sequences using Clustal Omega multiple sequence alignment [33] demonstrates the high level of conservation of the kinase domain. The predicted Drosophila protein appears to be localised in the cytoplasm as predicted by MultiLoc2 [34], and to possess active sites and motifs that include a weak nuclear export signal (NES) as determined by NetNES [35], a nuclear localization signal (NLS) that was detected by cNLS Mapper [36], and a transmembrane (TM) domain as predicted by TMpred [37]. Motifs identified by ELM that may contribute to the function of the Drosophila homologue are ubiquitination site, protein phosphatase 1 interacting motif, IAP binding motif, and an Src Homology 2 (SH2) domain binding motif.

Loss of Pdxk decreases lifespan and locomotor ability
The inhibition of Pdxk in the Ddc-Gal4-expressing neurons by RNA interference results in flies with a decreased lifespan and impaired locomotor function. The median lifespan for Pdxk-RNAi flies was determined to be 50 days compared to 70 days for the control flies that express the lacZ transgene (Fig. 2a). When Pdxk was suppressed Fig. 1 The ribokinase/pyridoxal domain is evolutionarily conserved. A Clustal Omega multiple sequence alignment [33] of Drosophila melanogaster Pdxk with that of mammalian and insect homologues shows an evolutionarily conserved kinase domain (H. sapiens is Homo sapiens NP_003672.1, M. musculus is Mus musculus NP_742146.1, A. gambiae is Anopheles gambiae XP_315959.4, and D. melanogaster is Drosophila melanogaster NP_996031.1). The Drosophila transcript shows presence of motifs for NLS, a weak NES, MTS, and TM domains. Domains were identified using the NCBI Conserved Domain Database (CDD) [31] and the Eukaryotic Linear Motif resource [32]. Asterisk indicate the residues that are identical, colon indicate the conserved substitutions, dot indicate the semi-conserved substitutions. Colours show the chemical nature of amino acids. Red is small hydrophobic (including aromatic), blue is acidic, magenta is basic, and green is basic with hydroxyl or amine groups in the Ddc-Gal4-expressing neurons the flies showed impaired locomotor ability as determined by the nonlinear fitting of the climbing curves (Fig. 2b). The 95% confidence intervals (CI) were 0.038-0.049 compared to 0.051-0.076 for the controls. Taken together, these results indicate an important role for this kinase in the Ddc-Gal4-expressing neurons of Drosophila as interference with its activity phenocopies some of the well-established phenotypes observed in other Drosophila models of PD.

Inhibiting Pdxk does not suppress α-synuclein-induced phenotypes
The co-expression of Pdxk-RNAi with the expression of α-synuclein does not alter the diminished longevity nor does it change the observed loss of climbing ability over time. The median lifespan was 58 days for Pdxk flies compared to 60 days for control flies (Fig. 3a). A comparison of the climbing curves at 95% CI for Pdxk-RNAi flies was 0.033-0.045 compared to 0.036-0.048 for the controls. This comparison showed that they were not significantly different from each other (Fig. 3b). This implies that the inhibition of Pdxk in the Ddc-Gal4expressing neurons does not influence the neurotoxic effects of α-synuclein.

The pro-survival Buffy suppresses the loss of Pdxk phenotypes
The co-expression of the pro-survival Bcl-2 homologue Buffy with Pdxk-RNAi in Ddc-Gal4-expressing neurons results in a slightly increased lifespan and an improved climbing ability. The median survival of Pdxk-RNAi flies with co-expression of Buffy was 68 days when compared to 74 days for the controls that expressed the lacZ transgene and Buffy, as determined by Log-rank at a P < 0.0001 (Fig. 4a). The climbing ability was significantly improved as determined by a nonlinear fitting of the climbing curves and compared at 95% CI (Fig. 4b). These results suggest a pro-survival role for Buffy as exhibited by significant increases in both survival and locomotor function when Pdxk is inhibited in the Ddc-Gal4expressing neurons.

Inhibiting Pdxk in the developing eye results in disruption of the ommatidial array that can be suppressed by Buffy
The inhibition of Pdxk in the developing eye directed by the GMR-Gal4 transgene results in eyes with a significantly lower mean number of ommatidia, at 686.4 for Pdxk-RNAi flies compared to 706.9 for the lacZ control flies (Fig. 5a, b) as determined by an unpaired T test with a P value of 0.0007. More disruption of the ommatidial array was observed in Pdxk-RNAi flies, at 37.3% when compared to the lacZ control flies at 25.5% (Fig. 5a, b) as determined by an unpaired T test with a P value of 0.0001. The inhibition of Pdxk along with α-synuclein expression did not result in marked differences in either the number of ommatidia or the disruption of the eye between the flies that express Pdxk-RNAi along with α-synuclein, when compared to control flies that expressed lacZ along with α-synuclein (Fig. 5a, c). The mean number of ommatidia for Pdxk-RNAi/α-synuclein flies was 643.8 compared to 675.6 for the lacZ/α-synuclein flies, a not significant difference as determined by an unpaired T test (P = 0.0616). Additionally, the disruption of the ommatidial array was not significantly different, at 45.8 and 49.8% respectively as determined by an unpaired T test (P = 0.3117). However, a one-way analysis of variance between Pdxk-RNAi/α-synuclein and Pdxk-RNAi flies revealed that the expression of α-synuclein along with Pdxk-RNAi had a significantly severe phenotype in the eye compared to the Pdxk-RNAi alone flies (Fig. 5a, c), marked by decreased mean ommatidia number and higher disruption of the ommatidial array. It appears that the inhibition of Pdxk in the developing eye along with α-synuclein expression enhances the Pdxk-RNAi-induced eye phenotype. When we coexpressed Pdxk-RNAi with Buffy, the number of ommatidia for Pdxk-RNAi/Buffy flies was 704.3 compared to 711.9 for the flies that expressed lacZ along with Buffy overexpression, this restored the Pdxk-RNAi-induced eye phenotypes to control levels. Similarly, the disruption of the ommatidial array was down to 8.4% for Pdxk-RNAi/Buffy flies compared to 6.4% for lacZ/Buffy flies. These comparisons were not significant as indicated by an unpaired T test with P values of 0.469 and 0.4115 respectively (Fig. 5a, d). However, a one-way analysis of variance between Pdxk-RNAi/Buffy and Pdxk-RNAi alone flies (Fig. 5a, b) showed a remarkable recovery of the Pdxk-RNAi-induced phenotypes. Taken together, these results suggest that Buffy suppresses the developmental eye defects resulting from the inhibition of Pdxk in the developing eye.

Discussion
The directed inhibition of Pdxk in the Ddc-Gal4-expressing neurons of D. melanogaster result in decreased lifespan and an age-dependent loss in climbing ability, phenotypes strongly associated with models of PD. Pyridoxal kinase is involved in the conversion of pyridoxal into pyridoxal-5′-phosphate (PLP), an important enzyme cofactor in intermediary metabolism [14]. These B 6 vitamers seem to be involved in all vital cellular functions, from glucose metabolism to nucleic acid synthesis to being potent antioxidants. Our inhibition of this protein kinase in the Ddc-Gal4-expressing neurons resulted in shortened lifespan and impaired climbing ability, while in supportive experiments, its inhibition in the developing eye resulted in reduced ommatidia number and a high degree of ommatidial disarray. The involvement of PLP in the conversion of dopa to dopamine is an important step in its synthesis, it seems therefore, that decreased neuronal PLP can mimic the effects of decreased dopamine, a key step in PD development, and possibly of major impact in dopamine-containing cells. It is not surprising therefore, that decrease in the availability of such an important coenzyme by inhibition of the kinase function involved in its conversion results in defective neuronal function that may lead to neurodegeneration. The expression of α-synuclein in Ddc-Gal4-expressing neurons can result in PD-like characteristics in Drosophila [23]. We investigated the possible existence of a link between α-synuclein-induced phenotypes and the inhibition of Pdxk in the Ddc-Gal4-expressing neurons of Drosophila. Our results indicate that loss of Pdxk function in Ddc-Gal4-expressing neurons that were additionally expressing α-synuclein did not enhance the α-synuclein-induced phenotypes of decreased lifespan and impaired climbing ability. Similarly, when we conducted similar comparisons in the neuron rich Drosophila eye, we observed that inhibition of Pdxk did not alter the consequences of α-synuclein expression in the developing eye. Therefore, the toxic effects of α-synuclein are sufficient to generate phenotypes but inhibition of Pdxk activity does not confer any additional disadvantage. Alternatively, loss-of-Pdxk-induced toxicity may precede the effects of α-synuclein accumulation with the expression of α-synuclein enhancing the Pdxk-RNAi-induced phenotypes and specifically in the developing eye. This observation indicates a cell specific mechanism, where the expression of α-synuclein in the Pdxk-RNAi background in the developing eye and not in the DA neurons enhances the phenotypes. Additionally, the multitude of biological functions that are dependent on PLP from a dysfunctional antioxidant activity to mitochondrial dysfunction may induce neurotoxicity in these sensitive neurons.
The directed expression of the pro-survival Bcl-2 homologue Buffy with Pdxk RNA interference resulted in the suppression of the loss of Pdxk-induced phenotypes of decreased survival and impaired locomotor ability. Buffy like many other pro-survival Bcl-2 proteins, is thought to be a "guardian" of the mitochondria and confers survival advantages by restricting death-promoting molecules [38,39]. Buffy suppressed the Pdxk-induced phenotypes of reduced lifespan and locomotor ability and further restored ommatidia number and decreased the disruption of the ommatidial array. Not only do PLP deficiencies accelerate mitochondrial decay, but they increase ROS radicals [15,16], a property that make neurons more vulnerable. It is possible that Buffy restores this balance by initiating or participation in pro-survival signals.

Conclusions
Although Pdxk has not been confirmed to be a PDcausative gene, it has been associated to increased risk of the disease. The inhibition of this gene activity by the directed expression of an RNAi transgene in the Ddc-Gal4-expressing neurons phenocopies PD-like symptoms in Drosophila, and therefore may represent a novel model of PD. This appear to corroborate other studies that show several mechanisms involved in the aetiology of PD. More studies are required to chart out a pathway for Pdxk activity in Drosophila, and importantly to show at a molecular level the changes associated with the lossof-function of this kinase in the development and function of dopaminergic neurons.

Drosophila media and culture
Stocks and crosses were maintained on a standard medium prepared from cornmeal, molasses, yeast, agar, water and treated with propionic acid and methylparaben to inhibit fungal growth. Stocks were maintained at 23 ± 2 °C while crosses and experiments were carried out at 25 and 29 °C. [38] was generously supplied by Dr. Leonie Quinn (University of Melbourne), UAS-α-synuclein [23] by Dr. Mel Feany of Harvard Medical School, and Ddc-Gal4 [41] by Dr. Jay Hirsch (University of Virginia). The UAS-Pdxk-RNAi (P{KK108240}VIE-260B) [42] flies were obtained from Vienna Drosophila Resource Center while GMR-Gal4 [43] and UAS-lacZ flies were obtained from the Bloomington Drosophila Stock Center at Indiana University. The UAS-α-synuclein/CyO; Ddc-Gal4/TM3; UAS-α-synuclein/CyO; GMR-Gal4; UAS-Buffy/CyO; Ddc-Gal4; and UAS-Buffy/CyO; GMR-Gal4 complex lines were generated following a standard protocol as previously described [44,45]. They expressed α-synuclein or overexpressed Buffy either in the Ddc-Gal4-expressing neurons using the Ddc-Gal4 transgene or in the developing eye using the GMR-Gal4 transgene. Analysis of complex lines was confirmed via PCR reactions and gel electrophoresis.

Ageing assay
Multiple matings of each genotype were carried out and male flies collected upon eclosion and analysed following a protocol that has been described [29,46]. At least 200 flies were aged per genotype and scored every 2 days for the presence of deceased adults [47]. Longevity data was analysed using the GraphPad Prism version 5.04 and survival curves were compared using the Log-rank (Mantel-Cox) test with significance determined with a 95% confidence interval, at a P ≤ 0.05 with Bonferroni correction.

Climbing assay
A batch of male flies was analysed for their ability to climb according to a standard protocol [48]. In brief, every 7 days after eclosion, 50 or fewer males of each genotype were assayed for their ability to climb. Climbing indices were computed and analysed with GraphPad Prism version 5.04. Climbing curves were fitted using non-linear regression and compared using 95% confidence interval with a 0.05 P value.

Scanning electron microscopy of the Drosophila eye
Male flies were prepared for scanning electron microscopy using a standard protocol as previously described [29]. Ten different eye images for each genotype were analysed using the National Institutes of Health (NIH) ImageJ software [49] and biometric analysis performed using GraphPad Prism version 5.04. The percentage disruption of the eye was calculated as previously described [50]. Statistical analysis comprised of unpaired student T test and one-way analyses of variance with Newman-Keuls multiple comparison post-test and significance determined at P values equal or less than 0.05.