Mutations in PNPLA6 are linked to photoreceptor degeneration and various forms of childhood blindness

Blindness due to retinal degeneration affects millions of people worldwide, but many disease-causing mutations remain unknown. PNPLA6 encodes the patatin-like phospholipase domain containing protein 6, also known as neuropathy target esterase (NTE), which is the target of toxic organophosphates that induce human paralysis due to severe axonopathy of large neurons. Mutations in PNPLA6 also cause human spastic paraplegia characterized by motor neuron degeneration. Here we identify PNPLA6 mutations in childhood blindness in seven families with retinal degeneration, including Leber congenital amaurosis and Oliver McFarlane syndrome. PNPLA6 localizes mostly at the inner segment plasma membrane in photoreceptors and mutations in Drosophila PNPLA6 lead to photoreceptor cell death. We also report that lysophosphatidylcholine and lysophosphatidic acid levels are elevated in mutant Drosophila. These findings show a role for PNPLA6 in photoreceptor survival and identify phospholipid metabolism as a potential therapeutic target for some forms of blindness. Blindness due to retinal degeneration affects millions of people worldwide, but many disease-causing mutations remain unknown. Here the authors link mutations in the gene PNPLA6 with childhood blindness in seven families with retinal degeneration and show that the gene plays a role in photoreceptor survival in Drosophila.

P NPLA6 (also previously referred to as neuropathy target esterase (NTE)) is a highly conserved lysophospholipase anchored to the cytoplasmic face of the endoplasmic reticulum (ER) [1][2][3] . It is the primary target of toxic organophosphorous compounds (found in agricultural pesticides and agents of chemical warfare) that cause axonal degeneration in large neurons. Human organophosphorous-induced delayed neuropathy (OPIDN) occurs when PNPLA6 becomes phosphorylated by these organophosphates, followed by dealkylation of the phosphoryl enzyme, inhibiting the catalytic domain 4,5 . Patients with mutations in the catalytic domain of PNPLA6 develop a Mendelian motor neuron disease called spastic paraplegia (SPG39), which is characterized by distal degeneration of motor axons surprisingly similar to OPIDN 6,7 . The exact mechanism of PNPLA6 function in both OPIDN and SPG39 remains unknown 8 .
PNPLA6 is expressed in the brain, where it deacetylates phosphatidylcholine (PC) to lysophosphatidylcholine (LPC) and LPC to glycerophosphocholine 1,9 . In Drosophila, the Swiss cheese protein (sws; the orthologue of vertebrate PNPLA6) is involved in the regulation of membrane lipid homeostasis and in the survival of both neurons and glia as well 10 . In mice, Pnpla6-deficient animals are embryonic lethal 11 but brain-specific mutants show neurodegeneration 12 , and there is much evidence that PNPLA6 is involved in maintaining nervous system integrity 13,14 . Interestingly, mice expressing a single copy of Pnpla6 are viable but show increased sensitivity to organophosphorous toxins 11 .
In this work, we identified biallelic PNLPA6 mutations in patients with childhood blindness due to severe photoreceptor death and clinical features of Leber congenital amaurosis (LCA) and, interestingly, also of the rare Oliver McFarlane syndrome (OMS; OMIM 275400). OMS patients have a complex phenotype characterized by blindness due to severe photoreceptor degeneration, dwarfism due to pituitary growth hormone deficiency, trichomegaly and progressive alopecia. Occasionally, OMS is associated with mental retardation, distal muscle weakness/ wasting and ataxia, due to axonal peripheral neuropathy. We document PNPLA6 expression in mouse and Drosophila retina, and test phenotypic and metabolic consequences (phospholipids) of PNLPA6 and SWS mutations in humans and Drosophila, respectively. We show that PNPLA6 has a role in photoreceptor survival and identify phospholipid metabolism as a potential therapeutic target for some forms of blindness.

Results
Identification of PNPLA6 mutations in OMS. Our initial goal was to discover the causal genetic mutations for two sporadic cases (5,267 and 5,273) in two Canadian families and three affected siblings (166, 167 and 169), in a third OMS family from the Czech Republic ( Fig. 1). Although most OMS cases seem to be sporadic, segregation patterns of the phenotype support the recessive mode of inheritance. We therefore hypothesized that we could identify the unknown causal gene by excluding all genomic variants, except  Fig. 1).
those biallelic variants that we would find in the same gene in the five affected OMS patients (assuming one gene for OMS). Whole exome sequencing (WES) first excluded such mutations in the currently known 210 retinal disease genes. We then sifted through thousands of variants and identified that all five patients harboured and shared biallelic variants in PNPLA6 and not in any other gene (Fig. 1). We identified six missense mutations in these three OMS families in PNPLA6 (NM 001166111.1). With Sanger sequencing, we confirmed co-segregation in two out of the three families (one family was deceased) and excluded the variants from 100 normal controls. We then recruited five more OMS families from around the world (Fig. 1), and overall we identified ten novel PNPLA6 mutations ( Fig. 2b and Supplementary Fig. 1) in a total of seven families. Only one OMS family did not harbour PNPLA6 mutations, suggesting allelic heterogeneity and thus a second OMScausing gene. We then searched for PNPLA6 mutations in 200 cases with LCA that had undergone WES and found one patient with significant autism and two PNPLA6 mutations. We discovered a total of five missense, two nonsense, two frameshift and one splice site mutation(s) in PNPLA6 (Figs 1 and 2). One of the mutations, that is, p. Gly1129Arg was found three times, while another mutation, that is, p. Asp1125Asn was identified twice, affecting residues in the evolutionarily highly conserved patatin domain (down to Escherichia coli; Fig. 2c). In five out of seven patients, we found compound heterozygous missense mutations in both the regulatory (amino-terminal) and in the catalytic domain (carboxyterminal). The missense mutations are all in residues that are very well conserved (Fig. 2c). Targeted DNA screening revealed that the mutations found in the affected individuals are rare, as they were absent from 41,000 WES samples in our in-house databases of which at least 200 patients were genetically unsettled LCA.
Genotype/phenotype correlations in the cohort. All except one patient had OMS, as per standard clinical diagnostic criteria, with some patients exhibiting ataxia, mental retardation, or peripheral neuropathy. Detailed phenotypes are described in Table 1. All our OMS patients had significant retinal degeneration with photoreceptor death and blindness, long lashes and dwarfism. The retinal degeneration is severe and early onset, with photoreceptor death, but also with inner retinal changes and retinal thinning accompanied by lipofuscin metabolism abnormalities (Fig. 3). In proband 5,267 from family 1, we found two missense mutations (Figs 1 and 2b and Supplementary Fig. 1) One of the missense mutations resides in a residue in the regulatory region and the second in the catalytic region (patatin domain) (Fig. 2). Briefly, this affected French-Canadian child with OMS developed nyctalopia (significant nightblindness) at the age of 2 years. At age 9 years, she had 20/1,200 visual acuities from severe retinal and choroidal degeneration (Fig. 3a), with diffuse photoreceptor loss ( Fig. 3b), growth retardation and long lashes, but high intelligence. In proband 5,273 from family 2 with OMS, we found a nonsense mutation in the regulatory region and a missense mutation in the patatin domain of PNPLA6 (Figs 1, 2b and Supplementary Fig. 1). This 30-year-old Scottish male had light perception vision photoreceptor loss, dwarfism, peripheral neuropathy and ataxia.  ARTICLE In cases 166, 167 and 169 from the Czech OMS family 3 we found a splice site mutation in the regulatory region and a missense mutation in the patatin domain ( Fig. 2b and Supplementary Fig. 1). The three affected members had severe photoreceptor degeneration, dwarfism, long lashes, normal intelligence, ataxia and peripheral neuropathy. Proband 3,649 from family 4 with LCA was found to have a frameshift in the catalytic region and a missense mutation in the regulatory region of PNPLA6 ( Fig. 2b and Supplementary  Fig. 1). This Caucasian child had congenital visual loss and significant autism but no other systemic disease. Proband 5,477 from family 5 with OMS was found to have two missense mutations, both in the catalytic region, which co-segregate through this Brazilian family ( Fig. 2b and Supplementary Fig. 1). The affected child has severe visual loss (counting fingers vision) from photoreceptor degeneration, long lashes, short stature, mild mental handicap and neuropathy with ataxia. Proband 5,440 from family 6 with OMS from Scotland carried a frameshift mutation in the regulatory region and a missense mutation in the catalytic region of PNPLA6, which co-segregated in the family ( Fig. 2b and Supplementary Fig. 1). The affected child was noted to have congenital nystagmus from photoreceptor degeneration, normal mentation and lashes, but short stature and a motor neuropathy. Proband 5,748 from family 7 with OMS from Great Britain was found to harbour a nonsense mutation in the regulatory region of PNPLA6, shared by the father of the affected child, but we could not identify the maternal mutation ( Fig. 2b and Supplementary Fig. 1).
PNPLA6 is expressed in mouse and fly photoreceptors. All our subjects with PNPLA6 mutations have photoreceptor degeneration resulting in blindness. However, PNPLA6 mutations associated with SPG39 and OPIDN diseases are not known to cause retinal degeneration or blindness. As it was previously not known whether PNPLA6 is expressed in retinal tissue, we investigated the spatiotemporal retinal expression pattern of the PNPLA6 orthologues in Drosophila (SWS) and in mouse retina (Fig. 4a,b). Using an anti-serum against fly SWS, we found expression around photoreceptor nuclei and along the entire photoreceptor (Fig. 4c). Performing co-stainings with the plasma membrane marker antihorseradish peroxidase, we found that some of the SWS is associated with the plasma membrane, but that it can also be found in the cytoplasm ( Supplementary Fig. 2). This is in agreement with our previous results in central nervous system neurons, which showed that SWS can be detected in the ER 10 . In the mouse, we found PNPLA6-expressing cells in both developing and adult retinas ( Fig. 4d-f). Co-staining of PNPLA6 with cell-type-specific markers showed restricted expression in horizontal, amacrine and photoreceptor cells (Fig. 4g-o), but not in ganglion cells (not shown). To determine the subcellular localization of PNPLA6 in photoreceptors, we co-stained adult mouse retinal sections with PNPLA6 and the inner segment (IS) plasma membrane marker Na þ /K þ ATPase. Three-dimensional confocal imaging revealed clear co-labelling ( Fig. 4p-r), indicating primarily IS plasma membrane localization in adult photoreceptors, although some intracellular staining was also observed (Fig. 4q). Interestingly, PNPLA6 was not found in the outer segments (OS). Peanut agglutinin co-staining in adult mice showed that PNPLA6 is also expressed in cones ( Fig. 4s-u).
PNPLA6 is required for photoreceptor survival in Drosophila.
To study the consequences of PNPLA6 loss, we analysed the recessive sws 1 -null mutant in Drosophila, which shows severe agedependent neurodegeneration 15 starting around day five of adulthood, but no overt eye phenotype. Electron microscopic studies from sws 1 flies shortly before eclosion revealed that the eye developed normally with all photoreceptors present (Fig. 5b). Whereas 14-day-old flies still had mostly intact ommatidia (occasionally one photoreceptor was lost, arrow, Fig. 5c), at 23 days the photoreceptors were either lost (arrows, Fig. 5d) or showed degenerative signs such as condensed cell bodies (arrowheads, Fig. 5d,f), intracellular membranous bodies (arrow, Fig. 5g), or vacuoles (arrowhead, Fig. 5h). Thirty-dayold wild-type flies did not reveal any of these phenotypes (Fig. 5a,e). To exclude that the photoreceptor degeneration is a secondary effect of the brain degeneration in sws 1 , we also analysed an eye-specific knockdown of SWS using GMR-GAL4. Again, the photoreceptors developed normally (Fig. 5i,m) and still appeared unaffected in 7-day-old flies ( Fig. 5j,n), but when 30 days old flies showed obvious signs of degeneration (Fig. 5k,o, arrows) and some photoreceptors were lost (Fig. 5k, arrowhead).
To determine whether the degeneration was activity dependent, we raised the flies in the dark. As seen in Fig. 5l,p, this did not prevent the degeneration, showing that neuronal activity is not affecting the degeneration. To further assess a connection between degeneration and photoreceptor activity, we directly measured activity using electroretinograms (ERGs). Again, we could not detect any defects neither in young flies (1-3 days old) nor in aged ones (28-30 days old, Supplementary Fig. 3), which strongly suggests that phototransduction is not affected by the loss of SWS.
Mutations in PNPLA6 alters phospholipid levels. Loss of SWS/ PNPLA6 results in increased PC levels in flies and mice 10,13 . We therefore predicted that sws 1 Drosophila as well as our OMS patients may show abnormalities in phospholipid levels due to lower PNPLA6 enzymatic activity. If true, this may then provide a tool for clinical diagnoses and potentially provide a therapeutic target. Based on the phospholipid metabolism showed in Fig. 6a, we predicted that mutant PNPLA6 (NTE) would result in elevated PC, LPC and LPA levels. This was confirmed using mass spectrometry analysis of these phospholipids in extracts from whole wild-type and mutant Drosophila (Fig. 6b). We also measured phospholipid levels in serum from family 1 (5,267) and family 3 (166-171) members. We found on average 10-15% increased levels of PC, LPA and LCA in patients and mutation carriers compared with controls (data not shown). However, as levels of these phospholipids are found to vary about two-to sixfold in normal human plasma/serum samples, with limited sample size and no opportunity to replicate these measurements in additional patients, the detected changes did not reach statistical significance.

Discussion
Our results suggest that PNPLA6 is essential for vision and photoreceptor biology, and that mutations in PNPLA6 cause photoreceptor neuron death and blindness in children with OMS  Why do the mutations in PNPLA6 described here cause photoreceptor cell death, while patients with SPG39 and OPIDN do not? One explanation is that the PNPLA6 mutations in five out of seven of our families were compound heterozygotes, with one mutation in the regulatory and one mutation in the catalytic domain ( Fig. 2b and Supplementary Fig. 1). SPG39 patients have homozygous mutations in the catalytic (patatin) domain 12,16 . In OPIDN patients, the organophosphates bind to and inhibit only the catalytic domain and the N-terminal functions may not be affected. Having mutations in both domains may cause a more severe overall impairment of PNPLA6 function, leading to photoreceptor death. Alternatively, the mutations in the N-terminal regulatory domain might induce specific phenotypes. It is also possible that the harmful effects of organophosphates are not strong enough to affect retinal photoreceptors but only large neurons, such as motor neurons. It will be possible to test this hypothesis in future experiments. PNPLA6 has been known for over 45 years and despite intense investigations (600 þ publications), its various functions have not been fully elucidated 13 . Previously, PNPLA6 was found to be expressed in the brain, most prominently in large (motor) neurons, but not in the retinal tissue 2,17 . We now show (Fig. 4) that PNPLA6 is expressed in the developing and mature mouse retina in horizontal, amacrine and photoreceptors cells, and in Drosophila photoreceptors. Unlike mouse neurons, where PNPLA6 localizes to the ER 12,18 , we show here that PNPLA6 is mostly localized to the IS plasma membrane of mouse photoreceptors, suggesting a possible role for PNPLA6 in the formation of distinct plasma membrane domains in the IS and in the OS, or in the development/renewal of OS disc membranes, which are essential for maintaining the integrity of the photoreceptors and vision 16,19,20 . Previous studies indeed support a role for PNPLA6 in membrane growth, as knockdowns of Pnpla6 in zebrafish caused axonal truncation and abnormal branching in addition to small eyes 17 . Furthermore, it was proposed that altered membrane trafficking in neuronal-specific Pnpla6 knockout mice causes axonal degeneration in spinal neurons 13 .
Similar to vertebrates, SWS is expressed in photoreceptor cells in flies. Although the photoreceptors develop normally in mutants, the eye-specific knockdown of SWS shows severe photoreceptor degeneration when aged, revealing that SWS is crucial for photoreceptor maintenance but not development. In addition, the degeneration is not dependent on light input, showing that the degeneration is not activity dependent. This suggests that the degeneration is caused by a failure to maintain the integrity of the photoreceptor cells and not by deleterious effects due to alterations in photoreception. This is also in agreement with the fact that the loss of SWS in other neurons also causes a progressive degeneration.
In agreement with the role of PNPLA6 as a phospholipase, we found alterations in phospholipid composition (LPC, LPA and PC) in Drosophila. We carefully examined the wild type, sws mutants and heterozygotes. Interestingly and importantly, we show a significant accumulation of all, LPA, LPC and LC, which may explain some of the clinical features of OMS. Indeed, elevated concentrations of LPA are known to be cytotoxic to neuronal cells, including photoreceptors 21 , and defects in the LPC acyltransferase-1, an enzyme catalysing the conversion of LPC to PC, cause photoreceptor degeneration in mice 22 . LPA also promotes hair growth 23 and defects either in LPA production 24,25 or LPA signalling 26,27 are associated with hair growth abnormalities, as observed in our OMS patients. It is therefore possible that elevations in LPA lead to the trichomegaly, wooly hair and progressive alopecia in our OMS patients. Altered LPA signalling may also be responsible for the abnormalities in bone and tooth development 28 , and growth hormone deficiency 25 , also seen in our OMS patients (Table 1). Interestingly, transient juvenile trichomegaly and progressive alopecia, as well as a strong family history of different early-onset cancers 29 were also found in some of our families (data not shown). However, this link needs to be further explored.
Most of the identified mutations in OMS are in highly conserved residues, supporting their functional importance and potential use for rapid clinical testing of OMS and LCA patients. Our studies have opened a new avenue in investigating functions of PNPLA6, including whether single PNPLA6 mutations increase photoreceptor susceptibility to environmental toxins. In the near future, it will be interesting to generate and characterize a photoreceptor-specific Pnpla6 mouse knockout to study the fundamental mechanism of action of PNPLA6 and evaluate the potential therapeutic benefits of lowering LPC and LPA.

Methods
Patient information. Patients were phenotyped at several international centres, after informed consents were obtained, approved by the ethical and scientific review boards. Detailed ophthalmological evaluations were followed by medical and clinical genetics consultations. All patients underwent a detailed history and pedigree analysis and detailed eye examinations, including best-corrected visual acuities by projected Snellen charts, near vision, slitlamp biomicroscopy and dilated retinal exams. In vivo retinal imaging was performed by the Heidelberg OCT (Heidelberg Engineering), followed by fundus autofluorescence. Kinetic fields were measured by Goldmann perimetry. In selected patients, we performed International Society for Clinical Electrophysiology of Vision (ISCEV) standard ERGs and multifocal ERGs (Diagnosys, Boston, USA). Peripheral blood was collected in lavender top (EDTA) tubes for DNA extraction. Performing sections from the retina of a sws 1 mutant fly shortly before eclosion (pharate adult) shows that photoreceptors develop normally (b) and persist during the first 2 weeks of adulthood (c, 14 days), although occasionally ommatidia missing a single photoreceptor can be found (arrow). However, after 23 days of ageing (which, due to the reduced live span of sws 1 , is shortly before these flies die), photoreceptor degeneration is evident in all ommatidia by the loss of photoreceptors (arrows in d), and in the condensed and darkly stained cell somata of the photoreceptors that are still present (arrowheads). In addition, vacuoles can be found in the somata (arrowheads in f,h) and the rhabdomeres are shrunken. In some photoreceptors, membranous structures can be found (arrow in g).
Knocking down SWS specifically in the eye using an RNA interference (RNAi) construct expressed by GMR-GAL4 showed that newly enclosed flies look normal with all the seven photoreceptors R1-7 present (i,m). At 7 days of age the ommatidia still look wild type (j,n), but at 4 weeks the photoreceptors clearly show degeneration with vacuoles forming in the photoreceptor somata (arrows in k,o) and loss of some photoreceptors (arrowhead in k). Raising the flies in the dark resulted in a comparable photoreceptor degeneration in 30-day-old flies (l,p). Scale bar, 4 mm (a), 2 mm (e).
NATURE COMMUNICATIONS | DOI: 10.1038/ncomms6614 ARTICLE Animals. All animal work was carried out in accordance with the Canadian Council on Animal Care guidelines and was approved by the Institut de recherches cliniques de Montréal Animal Care Committee. CD1 and C57/Bl6 mice of either sex were used in this study.
WES analysis. DNA was extracted from whole blood using the FlexiGene kit or the QIAamp DNA blood kit according to the manufacturer's protocol. DNA quantity and quality were verified by using NanoDrop. The exomes of two sporadic cases (5,273 and 5,267) and all individuals from the Czech family 3 (166-171) with childhood blindness and OMS were captured by the Sure Select Human Exome Kit version 4 (Agilent Technologies, Inc., Santa Clara, CA), using 3 mg of genomic DNA of each. Exon-enriched DNA libraries were then sequenced either on Illumine HiSeq2000 that generate 100 bp paired-end reads or on SOLiD 4 that generate 50 bp reads. The details of bioinformatics analysis have been reported previously [30][31][32][33] . Briefly, the high-quality trimmed reads from Illumina HiSeq2000 were aligned to the positive strand of human genome (hg19) by BWA version 0.5.9 (ref. 29) and a mean coverage of 128 Â (5,273) and 131 Â (5,267) was obtained for the targeted coding regions (consensus coding sequence). Reads from SOLiD 4 were aligned using Novoalign version 2.08.03. Quality score recalibration was performed during alignment and reads that aligned to multiple locations were discarded. Following alignment PCR, duplicate reads were removed using Picard tools version 1.84. Variant calling and annotation were performed by SAMtools version 0.1.7 (ref. 34) and ANNOVAR 35 , respectively. Filtering steps were carried out with public (dbsnp135, 1,000 genome project and NHLBI exomes) and our in-house exome (contains more than 800 samples) databases. Subsequently, variants that were predicted to be protein-altering (that is, missense, nonsense, frameshift, or canonical splice site changes) were prioritized and were manually examined in IGV 36 . Variants identified in individuals from Czech family 3 were also used for homozygosity mapping, linkage analysis and detection of copynumber variants.
Sanger sequencing and co-segregation. The selected variants were confirmed by Sanger sequencing in the patients and co-segregation was tested in their family members on an ABI 3730xl capillary sequencers. The Sanger sequencing results were analysed with the Sequencher software. Mutations were excluded from normal controls (at least 100 normal controls) and co-segregation in the family was studied. All mutations were studied by in silico analysis (Pmut, Blossum62, SIFT and Polyphen). All patients had two mutations, either homozygous or compound heterozygous. To ensure the completeness and quality of the sequences, primers were designed from intron sequences so that the entire exon and at least 50 bp of intron DNA on each side of each exon were amplified. Primers for the PCRs were designed by primer3 (http://frodo.wi.mit.edu/primer3/).
Drosophila PNPLA6 localization and knockout experiments. The 50-mm vibratome head sections from white-eyed flies (w 1118 , to avoid interference by the autofluorescence from the eye pigment) were performed as described in Bolkan et al. 37 and stained with our anti-SWS serum at a dilution of 1:100. Detection was done with a biotinylated secondary antibody (1:200; Vector Laboratories) and a Cy5-conjugated streptavidin (1:100; Jackson ImmunoResearch). Sections were counterstained with DAPI (4 0 ,6-diamidino-2-phenylindole; Invitrogen). Optical images (0.5 mm thick) were taken on an Olympus FW1000 confocal microscope. Semi-thin and ultra-thin Epon plastic sections were prepared as described in ref.
15. Semi-thin sections (1 mm) were stained with 1% toluidine blue and 1% borax, and analysed by light microscopy. Ultra-thin sections were analysed with a FEI Morgagni 268 electron microscope. sws 1 is described in Kretzschmar et al. 15   Extraction of Drosophila phospoholipids for flow injection-electrospray ionization-tandem mass spectrometry analysis. Whole flies were homogenized in MilliQ water to acquired concentration 80 mg ml À 1 . Homogenization was performed with Branson Digital Sonifier 450 working as cup horn set to pulse mode (1 s pulse/1 s pause) with amplitude maximum at 100watt each sample was total homogenization time was 2 Â 30 s. The 250 ml of fly homogenates was extracted by modified method of Natomi 38 . Fly homogenates (250 ml) corresponding to 20 mg was extracted by 2 ml of mixture of chloroform:methanol:water (20:10:1 v/v/v). Extraction mixture was repeatedly vortexed in 15-min steps during 2 h and then left at À 20°C overnight. Supernatant was filtered through glass filter to a glass tube. Sediment was re-extracted with 2 ml of mixture of chloroform:methanol:water (10:10:1 v/v/v). Extraction mixture was vortexed for 1 h and then filtered through glass filter and combined with the first extract. Final re-extraction step was with 2 ml mixture of chloroform:methanol:water (10:20:1 v/v/v). The mixture was vortexed for 1 h and then filtered through glass filter and combined with previous extracts. The combined supernatants were dried under a stream of nitrogen. Aliquot corresponding to 1 mg of flies was mixed with 50 pmol of internal standards: , and scan times for each transition pair was 500 ms. The ESI condition were as follows: curtain gas (N 2 ), 14 psi; collision gas (N 2 ), 12 psi; source gas 1 (N 2 ), 5 psi; source gas 2 (N 2 ), 25 psi; nebulizing gas (N 2 ) temperature, 200°C; the capillary spray voltage was À 4.5 kV, interface heater, ON. The ion optics settings for LPA measurement were: À 118 V for the declustering potential and À 10.7 V for the entrance potential. The collision energy was set to À 35 V with a collision cell exit potential of À 5 V.
PC, LPC and LPA concentrations were calculated using known amount (50 pmol) of internal standard added to samples. Final concentrations were expressed as pmol of lipids per mg of Drosophila. The Analyst 1.5 software was used to operate the instruments and process the data.