A Saposin deficiency model in Drosophila: lysosomal storage, progressive neurodegeneration, sensory physiological decline and defective calcium homeostasis

Saposin deficiency is a childhood neurodegenerative lysosomal storage disorder (LSD) that can cause premature death within three months of life. Saposins are activator proteins that promote the function of lysosomal hydrolases in the degradation of sphingolipids. There are four saposin proteins in humans, which are encoded by the prosaposin gene. Mutations causing an absence of individual saposins or the whole prosaposin gene lead to distinct LSDs due to the storage of different classes of sphingolipids. The pathological events leading to neuronal dysfunction induced by lysosomal storage of sphingolipids are as yet poorly defined. We have generated and characterised a Drosophila model of saposin deficiency that shows striking similarities to the human diseases. Drosophila saposin-related (dSap-r) mutants show a reduced longevity, progressive neurodegeneration, lysosomal storage, dramatic swelling of neuronal soma, perturbations in sphingolipid catabolism, and sensory physiological deterioration. We have also revealed a genetic interaction with a calcium exchanger (CalX), suggesting that calcium homeostasis may be altered in saposin deficiency. Together these findings support the use of dSap-r mutants in advancing our understanding of the cellular pathology implicated in saposin deficiency and related LSDs.


Introduction
Saposin deficiency is an autosomal recessive lysosomal storage disorder (LSD) that is typically associated with severe, age-dependent neurodegeneration and premature death during early childhood. In humans there are four saposins (saposins A -D), which are encoded by the prosaposin gene (1,2,3). The mature, active saposins are produced by cleavage of the prosaposin precursor during passage through the endosomes to the lysosomes; this function is primarily performed by Cathepsin D (4).
Once in the acidic lysosome environment, saposins act as activator proteins and promote the function of hydrolases involved in sphingolipid degradation (5)(6)(7)(8)(9). Mutations in prosaposin therefore cause a primary accumulation of sphingolipid species in the lysosomes. The location and severity of the prosaposin mutation dictates the number of saposins that are affected and hence the degree of sphingolipid accumulation and age of lethality. Mutations abolishing the prosaposin start codon result in an absence of prosaposin and therefore all 4 saposins; this causes the most severe pathology and individuals present with severe neurodegeneration at birth and die within 4 months (10)(11)(12). Of the single saposin disorders, saposin A deficiency is the most severe and results in death at 8 months old (13), whereas the mildest of the saposin C mutations cause non-neuronopathic disorders with relatively mild symptoms until the fourth decade of life (14). No single saposin D deficiencies have been reported in humans.
Because each saposin generally promotes the function of a specific sphingolipid hydrolase, the single saposin deficiencies resemble the pathology caused by mutations in their cognate hydrolase (e.g. 15; reviewed in 16); total prosaposin deficiency encapsulates many aspects of the single saposin deficiencies but to a more severe degree (10)(11)(12).
The sphingolipdoses form the largest group of LSDs, yet to date only one sphingolipidosis (Niemann-Pick Type C (NPC)) has been modelled in Drosophila (17)(18)(19)(20). To broaden our understanding of the sphingolipidoses, and to help identify pathological events subsequent to sphingolipid storage, we generated a Drosophila model of saposin deficiency. The Drosophila Saposin-related (dSap-r) locus encodes a protein predicted to contain eight saposin-like domains, each containing the classic six-cysteine arrangement found in all mammalian saposins. Characterisation of dSap-r mutants revealed pathology similar to that of the human disorders, including reduced longevity, progressive neurodegeneration, aberrant sphingolipid levels, and physiological deterioration; all hallmark signs of lysosomal storage. Our analysis reveals a genetic interaction with the Na+/Ca+ exchanger, CalX, and suggest a deficit in calcium regulation in the Drosophila model of saposin deficiency.

Identification of Drosophila Sap-r
A blastp search (NCBI; www.ncbi.nlm.nih.gov) was performed to identify the Drosophila melanogaster prosaposin (PSAP) homologue. The entire Homo sapiens PSAP protein sequence (CAG33027) was used to search the D. melanogaster protein database. Standard blastp assumptions were applied. A reciprocal search was performed, using the D. melanogaster d-Sap-rPA sequence to blast the H. sapiens protein database, to ensure the correct homologue was identified.

Identification of Drosophila Sap-r monosaposins
To identify the putative monosaposins encoded by the dSap-r gene, each human monosaposin sequence (RCSB Protein Data Bank entries 2DOB, 1N69, 2GTG, 2RB3) was aligned against the full-length dSap-rPA sequence using the bl2seq tool at NCBI. The H. sapiens and D. melanogaster monosaposins were aligned using default settings in ClustalX.
Drosophila stocks. All experimental crosses were grown on maize-meal fly food at 25°C. Newly eclosed flies were transferred to standard yeast-sucroseagar fly food. The wild type (+/+) control for all experiments was w 1118 crossed to Canton-S. The dSap-r PBac and Df(3R)tll-e (subsequently referred to as Df) stocks were from the Bloomington stock centre and the dSap-r NP7456 stock was from the Kyoto stock centre. The dSap-r C27 deletion was generated by mobilising the P-element from the dSap-r NP7456 parent line. The UAS-dSap-r transgenic stock was generated for this study. Briefly, pUAST-dSap-r was generated by excision of dSap-r cDNA from the pOT2 vector (clone GH08312, BDGP Gold collection), using XhoI and EcoRI followed by ligation into the pUAST vector. pUAST-dSap-r was microinjected into w 1118 embryos with helper plasmid ∆2-3. UAS-mCD8GFP and 1407-GAL4 stocks were kindly Immunohistochemistry. For the dSap-r expression pattern, third instar larvae (n>5) were dissected and fixed in 3.7% formaldehyde/PBS for 10 min, followed by 3 x 5 min washes in 0.1% PBST (PBS with 0.1% Triton X-100).
To label lysosomes, aged adult brains were dissected in 4% paraformaldehyde/PBS and transferred to fresh fixative for 20 min (n>8). Sections were stained with lead citrate for 10 min in the presence of sodium hydroxide pellets, followed by washing in dH 2 O. Images were captured using analysis software on a TECNAI G 2 (Version 2.18) transmission electron microscope (120 kV).
Neuronal soma area quantification. Neuronal soma area was quantified using ImageJ software. The soma and nuclear boundary of each neuron were demarcated and the area calculated by first inputting the number of pixels per micron then using the ImageJ area measurement tool. Neuronal soma area were normalised to nuclear area (n=3 flies). Pseudocolour images were produced using Adobe Illustrator CS4. Student's t-tests were performed in Microsoft Excel to determine statistical significance.

Lipid extraction and lipidomics analyses.
Female adults of the required genotypes were collected on emergence and aged for 5 days. Three brains/biological replicate were dissected in PBS, flash-frozen in 20% methanol using liquid nitrogen, and stored at -80°C until lipid extraction was performed. Brain samples were homogenized and extracted according to the methyl-tert-butyl ether extraction method (69). All lipid standards were added to the homogenates prior to extraction (Supplemental Table 2). After phase separation, the organic phase was used for lipidomics and the aqueous phase was processed for determining total protein content using the bicinchoninic acid (BCA) assay (BCA1 kit, Sigma).
For MS measurements, the samples were dissolved in 100 μl methanol containing 0.1% ammonium acetate and subsequently analyzed using a flow-

Supplemental Methods
Epifluorescence. Organs were dissected from adult flies in PBS and, in some cases, stained with DAPI. Organs were imaged using a Zeiss stereomicroscope equipped with an Axiocam MRc5 camera, a Neolumar S 1.5x FWD 30 mm lens and a HBO 100 mercury lamp.

Results
dSap-r has homologous sequence structure and similar expression pattern to mammalian prosaposin.
The Drosophila prosaposin (PSAP) homologue (Saposin-related; dSap-r) was identified by a BLAST screen using the complete human PSAP protein sequence (blast E value 4e-36). The dSap-r locus is located at band 100A7 of the right arm of the third chromosome. There are two predicted transcripts for dSap-r (dSap-r RA and dSap-r RB); dSap-r RB appears to be an in-frame truncation of dSap-r RA (Fig. 1A).
A bl2seq alignment of each human monosaposin (Sap A-D) with the dSap-r protein revealed eight homologous Drosophila saposins (dSaps 1-8). Each dSap contained the six conserved cysteine residues critical for saposin function (Fig. 1B). Five of the dSaps also contained a potential glycosylation signal between the second and third cysteines, previously shown in mammals to be necessary for correct saposin folding and function (21).
Saposin proteins are expressed in the mammalian nervous system and their loss usually causes severe neurodegeneration (22)(23)(24). We therefore investigated the expression pattern of dSap-r in the nervous system. The dSap-r NP7456 GAL4 enhancer-trap insertion was used to drive a membrane localised GFP reporter (mCD8GFP) in a dSap-r-specific expression pattern.
Third instar larvae were dissected and stained with either a glial or neuronal antibody ( -repo or -elav, respectively). The membrane GFP reporter localised around repo-positive nuclei, suggesting that dSap-r is expressed in glial cells (Fig. 1C). The GFP reporter was also shown to faintly surround elav-positive neuronal nuclei, however this is likely to represent expression in glial cells that surround the neuronal cell bodies. To confirm whether dSap-r is expressed in neurons, a nuclear GFP reporter (eIF4AIII:GFP) was driven by the dSap-r NP7456 element. No colocalisation was found between the nuclear GFP reporter and neuronal nuclei, suggesting that neurons have no or very little dSap-r expression ( Fig 1C). The nuclear GFP colocalised with the glial nuclear marker, further confirming dSap-r expression in glia.
In mammalian visceral organs, prosaposin is expressed ubiquitously at low levels; however moderate to high expression levels are found in the jejunum and tubular epithelial cells in the kidney cortex, epithelial cells of the oesophagus, pancreatic duct and bile duct, and the hepatocytes of the liver (22). PSAP has also been shown to promote spermiogenesis and fertility (25,26). Using the mCD8GFP reporter in conjunction with the dSap-r NP7456 enhancer-trap element, dSap-r was shown to be highly expressed in the reproductive system, the digestive system, Malpighian tubules (Drosophila kidney), and the fat bodies (Drosophila liver and adipose tissue) of the adult fly (Fig. S1).
The similar expression pattern of dSap-r and mammalian saposins in cells of the nervous, reproductive, digestive, and renal systems, and the liver is suggestive of conserved functions, and therefore supports the use of Drosophila to model these disorders.

dSap-r mutation causes a reduced longevity and age-dependent deterioration of locomotion.
Patients with saposin deficiency die prematurely, usually within the first decade of life. To test whether dSap-r mutants also die prematurely, we first generated deletions of the dSap-r locus via an imprecise P-element mobilisation strategy. The dSap-r C27 allele is a deletion of the first two exons of the dSap-r locus ( Figure 1A). We also identified a PiggBac transposon insertion, dSap-r PBac , in the 4 th exon of the dSap-r locus ( Figure 1A). To assess the effect of each dSap-r mutation on dSap-r transcript levels, RT-PCR was performed on transheterozygous combinations of these two alleles, and combinations of these alleles in trans to a deficiency chromosome uncovering the dSap-r locus ( Fig. 1A and 2C). The dSap-r transcript was almost undetectable in dSap-r C27 /Df mutants and is likely to reflect dSap-rRB levels, as the dSap-rRA start site is deleted in this mutant. In contrast, dSap-r transcript levels in the dSap-r PBac mutant were indistinguishable from wild type; however, the dSap-r primers were designed to anneal upstream of the dSap-r PBac insertion. Production of saposins from the dSap-r locus prior to the transposon insertion is therefore possible. Using our dSap-r mutants, a longevity assay was performed. Each of the three dSap-r allelic combinations caused a reduction in longevity compared to wild type flies ( Fig. 2A). Median survival for wild type flies was approximately 35 days, whereas 50% of dSap-r mutants survived only 6 days (dSap-r C27 /Df), 15 days (dSap-r C27/PBac ) or 18 days (dSap-r PBac /Df). This suggests that the dSap-r C27 allele is more severe than the dSap-r PBac allele. Taken together with the RT-PCR results, this suggests that the dSap-r C27 allele is likely a strong loss-of-function mutation, whereas the dSap-r PBac allele may produce dSap-r protein with reduced or aberrant function.
To confirm that the dSap-r mutations were responsible for the reduced longevity, flies carrying a UAS-dSap-r transgene were generated to allow GAL4-driven rescue of dSap-r longevity. Ubiquitous expression of dSap-r using either Act5c-GAL4 or tubulin-GAL4 resulted in lethality prior to third instar; we therefore used dSap-r expression driven by the neuronal 1407-GAL4 ( Fig. 2A) or the glial repo-GAL4 (Fig. 2B), either of which did not induce lethality, for rescue experiments. Expression of dSap-r in neurons of all dSapr mutant combinations resulted in a substantial rescue of longevity and in most cases longevity was equivalent to the transgene control. Glial dSap-r expression was driven in the dSap-r C27/PBac mutants only, due to greater genetic ease of repo-GAL4 recombination using the dSap-r PBac allele. Glial dSap-r expression also provided a substantial rescue of dSap-r C27/PBac longevity, however, not to the same degree as the neuronal 1407-GAL4 expression. These results confirm that the reduced longevity in dSap-r mutants was due to dSap-r mutations.
During the longevity analysis it was observed that the dSap-r flies showed an age-dependent decline in locomotion; the 20+ day-old dSap-r mutants rarely climbed the vial during spontaneous activity. To quantify this loss of climbing behaviour, 5-day and 22-day-old female flies were tapped to the bottom of a measuring cylinder and their climbing response was captured by video for 45 seconds. Calculation of climbing speed revealed that the 22-day-old dSap-r C27 /Df mutants were only able to climb at 5% of their 5-day-old speed compared to maintenance of 36% of 5-day climbing speed for the dSap-r C27 /+ controls. The dSap-r C27/PBac mutants showed an intermediate phenotype maintaining 16% of their 5-day climbing ability at 22-days old. These data confirm an age-dependent deterioration of climbing behaviour in the dSap-r mutants (Fig. 2D).

Age-dependent neurodegeneration in dSap-r mutants.
The reduced longevity and age-dependent deterioration of locomotion in dSap-r mutants is suggestive of progressive neurodegeneration. In Drosophila, vacuolarisation of the brain is a hallmark of neurodegeneration (27,20), which can be quantified from 1 m tissue sections at the light microscopy level. Horizontal sections were taken from 5-day and 22-day-old fly heads and stained with toluidine blue. Vacuolarisation was specifically observed in regions of sensory function, mainly the antennal lobes, the eye and optic lobes (Fig. 3A). Vacuole number was quantified for these sensory regions ( Fig. 3B & C) where we observed that vacuole number in most 5-dayold dSap-r mutants was not significantly different from controls, with the exception of limited vacuolarisation observed in the antennal lobes of the dSap-r C27 /Df mutants. As the flies aged, the number of vacuoles increased in all dSap-r mutants, indicative of progressive neurodegeneration. In the visual system, vacuole number increased by 4 -18 fold in the dSap-r mutants compared to less than a 1.5 fold increase in the controls. In the olfactory system (antennal lobes), the controls showed a modest 1.5 -3-fold increase in vacuole number. In contrast, the dSap-r PBac /Df and dSap-r C27/PBac mutants showed a massive 6-fold and 16-fold increase in vacuole number, respectively. Although the dSap-r C27 /Df mutants showed the greatest number of vacuoles at 22-days old, the increase was only 3-fold due to the increased severity of vacuolarisation at 5-days old.

dSap-r mutants have age-dependent lysosomal storage defects
LSDs are characterised by lysosomal dysfunction leading to the storage of undegraded material in the lysosomes (28). Therefore, to assess the degree of lysosomal dysfunction and storage in dSap-r mutants, western blotting was performed using two lysosomal antibodies: anti-Arl-8 and anti-Cathepsin-L.
Arl-8 is an Arf-like GTPase that localises to the lysosomes (29). The Arl-8 antibody was used to probe a western blot containing soluble protein from 5day and 22-day-old flies to investigate the degree of lysosomal storage in dSap-r mutants (Fig. 4A). In 5-day-old flies, Arl-8 levels were similar between wild type and dSap-r mutants. However, in 22-day-old dSap-r flies, Arl-8 levels were increased compared to wild type and the 5-day-old samples. This suggests an age-dependent accumulation of lysosomal material in dSap-r mutants and implies that lysosomes are more abundant and/or swollen in dSap-r mutants, which is consistent with the human disease.
To determine whether this increased abundance of Arl-8 is suggestive of general lysosomal dysfunction, western blot analysis was repeated using an antibody against a lysosomal enzyme (Cathepsin-L). Like all Cathepsins, Cathepsin-L is synthesised as an inactive precursor that is cleaved in the lysosome to form its mature, active form (30). Therefore, if lysosomal function is disturbed in dSap-r mutants, Cathepsin-L processing should be less efficient leading to an accumulation of the unprocessed, larger form. Western blot analysis of 5-day and 22-day-old lysates revealed an increase in both the unprocessed and processed forms of Cathepsin-L in all dSap-r mutants compared to age-matched controls, but no change in the ratio of unprocessed:processed Cathepsin-L (Fig. 4A). This supports the lysosomal storage phenotype shown by Arl-8 western analysis, but suggests that lysosomal function has not been completely perturbed.
In addition to western blotting, the Arl-8 antibody was used for immunohistochemistry on wild type and dSap-r C27 /Df adult brains. An overall increase in fluorescence was observed in the aged dSap-r brains, particularly in central brain regions and the antennal lobes (Fig. 4B). Together with our light microscopical analysis of brain sections, this reveals that regions of severe neurodegeneration coincide with the regions of the CNS showing the most lysosomal storage.
A common characteristic of LSDs is the accumulation of membranous storage material known as multilamellar bodies (MLBs) and multivesicular bodies (MVBs). We investigated whether the dSap-r C27 /Df mutants carried these hallmark signs. Due to the severe degeneration occurring around the antennal lobes (Fig. 3A), this region of the brain was the main focus for ultrastructural analyses. Electron micrographs of 22-day-old dSap-r C27 /Df neurons revealed an abundance of electron dense storage material that has a complex morphology; some regions were populated by electron-lucent droplets, some contained variable numbers of MLBs, and others were densely packed with MVBs (Fig. 4C). The ultrastructural nature of storage material in saposin deficiency patients and mammalian models has been shown to be highly variable, with the presence of MLBs, MVBs and electron-lucent droplets (11,12,31,32). The ultrastructural pathology observed in the dSap-r mutants is therefore consistent with the reported disease phenotype, which further supports the use of this model for investigating saposin deficiency.

Sphingolipids accumulate in dSap-r mutant brains
Saposin deficiency leads to the accumulation of a complex array of sphingolipids and sphingolipid intermediates. To determine the nature of sphingolipid perturbation in dSap-r mutants, we monitored Hexosyl-Ceramides (HexCer), Ceramide phosphorylethanolamine (CerPE), Ceramide (Cer) and Sphingosine (the breakdown product of Ceramide; 33) in conjunction with a lipidomic analysis of all major membrane phospholipids in the brain (Supplemental Table 1).
Brains from 5-day old flies revealed a consistent increase in sphingolipids (Supplemental Table 1), particularly sphingosine, in all dSap-r mutants examined (Fig. 4D). Notably, comparison of the sphingosine:ceramide ratios in controls and mutants revealed a striking imbalance between these sphingolipid intermediates in the dSap-r C27/Df mutants (Fig. 4E). We do not observe a significant change in HexCer across the mutant combinations (Supplemental Table 1).
Changes in particular phospholipid classes were observed. However, unlike the sphingolipids, there is no specific and consistent trend for changes in phospholipid classes across the different dSap-r mutant combinations tested (Supplemental Table 1).

dSap-r C27 /Df CNS soma are enlarged with storage.
During ultrastructural analyses of the CNS we observed that dSap-r C27 /Df neuronal soma were consistently and grossly enlarged compared to wild type.
To quantify this difference, the soma boundaries of each cell were marked using ImageJ and the soma area measured (Fig. 5A). Measuring the average soma area revealed an almost 2.5-fold increase in dSap-r C27 /Df mutants compared to wild type (Fig. 5B). When normalised to nucleus area, the dSapr C27 /Df soma were almost 2-fold larger suggesting that nuclear area is also increased in dSap-r C27 /Df neurons (Fig. 5C). We also observed cell enlargement and increased storage in glial cells of the dSap-r C27 /Df adult CNS (Supplemental Fig. 2).

Loss of dSap-r function causes a deterioration of visual function
We have shown that sensory regions of the brain are particularly vulnerable to dSap-r mutation. To assess the effect of dSap-r mutation on the function of sensory neurons, we investigated the integrity of the visual system in dSap-r flies. Electron micrographs revealed an abundance of stored material in photoreceptor neurons of the fly retina (Fig. 6A), similar in morphology to that found in the antennal lobe neurons (Fig. 4C). High magnification images showed a variable phenotype in the rhabdomeres, the light responsive component of the eye. Most rhabdomeres showed massive accumulation of electron-dense material yet some were found to be structurally intact (Fig.   6A).
To determine whether the photoreceptor neurons and their underlying optic lobe synapses were functionally intact, we measured the ability of the fly eye to respond to light using the electroretinogram (ERG). The ERG is a classic method for investigating photoreceptor and lamina neuron function by measuring the summed potential difference on the surface of the fly eye in response to light pulses (34,35). Different components of the ERG trace represent the net influence of different parts of the visual system: the initial decline in potential reflects the depolarisation of the photoreceptor neurons; the on-and off-transients are a result of synaptic transmission with the L1 and L2 lamina neurons; and the recovery phase denotes the feedback from the optic lobe to the photoreceptor neurons required for efficient repolarisation of the photoreceptors (34)(35)(36)(37). Therefore, the integrity of the different components of visual transduction can be assessed using this simple approach.
ERG recordings in 5-day-old dSap-r C27 /Df mutants were not significantly different from controls in all components measured (Fig. 6B & C). However, 22-day-old dSap-r C27 /Df mutants showed a severe deterioration of all components of the ERG compared to controls. The dSap-r C27 /Df ERG potential deteriorated to 46% of its 5-day value, whereas the dSap-r C27 /+ control ERG potential only decreased to 77% of its 5-day value. This severe deterioration of the ERG amplitude in dSap-r C27 /Df mutants was also matched by a 5-fold increase in recovery time after termination of the light pulse. This suggests that neuronal function in both the retina and underlying optic lobe is severely compromised in dSap-r mutants.

Calcium homeostasis is defective in dSap-r mutants
Calcium homeostasis defects have been highlighted as one aspect of cellular pathology implicated in the sphinoglipidoses (38)(39)(40). For example, in NPC1 cells lysosomal calcium levels have been shown to be abnormally low, and by increasing cytosolic calcium levels NPC pathology in both NPC1 cells and mouse models can be substantially rescued (40). Calcium homeostasis defects have also been implicated in other sphingolipidoses (Gaucher's, Sandhoff and GM 1 -gangliosidosis); however, an increase in cytosolic calcium levels was pathological in these cases (41)(42)(43)(44).
To investigate if the dSap-r sphingolipidosis model showed a calcium homeostasis defects, we measured ERGs in dSap-r mutants overexpressing the plasma membrane calcium exchanger (CalX) in the fly eye using the NINAE (Rhodopsin 1) promoter. CalX transports calcium ions out of the neuron in exchange for sodium ions (45). Therefore, if dSap-r mutants have a reduced cytosolic calcium level, CalX expression should exacerbate the degeneration and function of the photoreceptors; however, the ERG would be substantially rescued if dSap-r mutation causes an increased cytosolic calcium level.
Overexpression of CalX in an otherwise wild type fly resulted in a modest degeneration of the ERG to 81% of its 5-day level. When CalX was expressed in a dSap-r C27 /Df background, there was no effect on the 5-day-old dSap-r C27 /Df ERG. However, as the flies aged, CalX expression induced a decrease in dSap-r C27 /Df vision to only 9% of its 5-day-old level; this is in contrast to the deterioration of the dSap-r C27 /Df ERG to 64% of its 5-day-old level, and therefore suggests that the dSap-r mutation may cause an abnormally reduced cytosolic calcium level.

Discussion
To further our understanding of saposin deficiency disease, we generated a Drosophila model. The Drosophila prosaposin homologue dSap-r was revealed to contain eight saposin-like domains, which all contained the sixcysteine primary sequence common to all saposins (this investigation, 46).
We showed that most of the dSap peptides contain a predicted glycosylation site, a feature critical for saposin folding and function (21). The sequence similarities between the mammalian and Drosophila saposins suggested that we had identified the correct ortholog.
To further characterise dSap-r, we investigated its expression pattern using GFP reporters. Both membrane and nuclear GFP reporters confirmed the expression of dSap-r in glia, as reported by (47). Our findings also suggested that, unlike in mammals, dSap-r is either not expressed in neurons or is expressed at low/undetectable levels.
Although dSap-r expression has been shown to occur in glia, this does not discount the presence of dSap-r protein in neurons. Mammalian prosaposin is abundant in secretory fluids, including cerebrospinal fluid, seminal fluid, milk, bile and pancreatic fluid (48)(49)(50)(51); in fact prosaposin is one of the main secretory products of Sertoli cells in the male reproductive system (52).
Therefore, we propose that Drosophila glia may provide dSap-r to neurons by glial secretion and neuronal uptake. This notion is supported by our findings that dSap-r longevity can be substantially rescued by expressing dSap-r directly in neurons, which suggests a neuronal requirement for dSap-r. This is further supported by our TEM analyses showing more severe lysosomal storage and consequential increase in cell size of the neurons compared to the glia. This non-autonomous function of dSap-r has important implications for potential therapeutic strategies as it suggests that a source of prosaposin secreting cells could provide a successful intervention for these conditions. This investigation also revealed high dSap-r expression in the male and female reproductive organs, the digestive system, Drosophila renal system (Malpighian tubules), and the Drosophila liver equivalent (the fat bodies).
Mammalian prosaposin has a role in spermiogenesis and improving fertility (25,26,53,54), and is also relatively abundant in subsets of cells of the small intestine, kidney and liver (22). Therefore, in addition to having a conserved primary sequence, dSap-r also has a conserved tissue expression suggesting a conserved function.
After confirming the identity of the Drosophila prosaposin orthologue and its expression pattern, we generated a dSap-r loss-of-function model and assessed its pathology. Ultrastructural analysis revealed extensive storage in the dSap-r nervous system (MLBs, MVBs and lipid droplets), characteristic of LSD pathology. Lysosomal storage was also confirmed by western blotting, which showed progressive accumulation of two lysosomal markers (Arl8 and Cathepsin-L) in the dSap-r mutants, and mass spectrometry revealed early accumulation and imbalance of sphingolipid intermediates in dSap-r mutants.
The dSap-r C27/Df mutants were the most severe mutants in all assays tested.
They also showed the most severe imbalance of sphingosine and ceramide, whilst having similar levels of sphingosine and lower levels of ceramide than the other mutant allelic combinations. This therefore supports previous findings that the ratio of sphingolipids is more important than the total levels of individual sphingolipids when considering the effect on pathology (54)(55)(56)(57)(58).
Like other Drosophila LSD models, dSap-r mutants die prematurely, with 50% death within 18 days; this equates to a 41-49% reduction in longevity, which is comparable to most Drosophila LSD models (40-55% reduction) (19,20,(62)(63)(64). The dSap-r mutants also showed an age-dependent deterioration of spontaneous and evoked locomotion, which correlated with the progressive nature of vacuolarisation in the sensory regions of the dSap-r nervous system. This may suggest that locomotion deterioration is primarily a result of failed perception of sensory cues rather than motor neuron or muscle degeneration. This notion is supported by a less severe deterioration in dSap-r jump performance using an assay that bypasses sensory neuron input (data not shown). A similar degeneration in sensory function is observed in a Our sphingolipid analyses showed that sphingosine levels were significantly increased in all dSap-r mutant brains. Like in NPC1 cells, this may cause a decrease in lysosomal calcium in dSap-r mutants, which was further exacerbated by overexpression of CalX. Further evidence of a link between sphingolipid metabolism and calcium homeostasis was also provided by (68), who revealed that retinal degeneration caused by imbalances in sphingolipid metabolites was suppressed by mutations in the Drosophila CalX. We have also shown an imbalance in the ratio between two sphingolipid intermediates in dSap-r brains, providing further support of calcium homeostasis defects being a likely cause of dSap-r pathology.
In conclusion, we suggest that in dSap-r mutants an early accumulation of sphingosine, or the imbalance of the sphingosine:ceramide ratio, leads to a decrease in lysosomal calcium. This calcium deficiency causes degeneration of sensory neurons in the dSap-r mutants leading to physiological deterioration. Therapeutic interventions to increase cytosolic calcium levels in saposin-deficient patients may therefore provide a successful route for ameliorating these disorders.   1. Drosophila Sap-r is expressed in glia. A. The Drosophila prosaposin homologue Saposin-related (dSap-r) consists of seven exons and is present on the right arm of the third chromosome (3R) at position 100A7. The dSap-r gene contains two potential transcript start sites (ATG), which result in the production of a transcript consisting of the entire coding sequence (dSap-rRA, 3451 bp) and an in-frame, truncated version (dSap-rRB, 3349 bp) (black lines). The dSap-r C27 allele was generated by P-element mobilisation from the P{GawB} NP7456 line and is a 2.5kb deletion. The dSap-r PBac allele consists of a 5.971 bp piggyBac transposon insertion into the largest dSap-r exon. The deletion in the deficiency line Df(3R)tll-e spans cytogenetic bands 100A2 -100C5, which includes the dSap-r gene. B. A Clustal X alignment of the 8 predicted Drosophila melanogaster saposins (dSaps) and the 4 Homo sapiens saposins (hSaps). Asterisks, conserved cysteine residues; underlining, glycosylation sites in each hSap and possible homologous sites in the dSaps. Blue highlights identical residues; yellow highlights similar residues (80% threshold setting). C. Confocal images of third instar larvae brains expressing mCD8eGFP or eIF4AIIIGFP under the control of dSap-r NP7456 GAL4 and stained with -repo (glial nuclear marker) or -elav (pan neuronal marker). Scale bars: 20 m. Neuronal or glial expression of dSap-r in a wild type background was performed as a transgene control. Longevity was plotted as the percentage of Day 0 flies surviving for each genotype. n > 60 for each genotype. C. cDNA from wild type (+/+), dSapr PBac /Df (PBac/Df) and dSap-r C27 /Df (C27/Df) third instar larvae were amplified using rp49 control and dSap-r primers. D. 5-day-old (white bars) and 22-day-old flies (grey bars) were subjected to climbing assays. The climbing speed of dSap-r C27 heterozygotes (C27/+), dSap-r C27 /Df (C27/Df) and dSap-r C27/PBac (C27/PBac) mutants are shown. Error bars: ± sem. * p< 0.05, ** p< 0.005 and *** p≤ 0.001. Genotypes tested: +/+ (wild type); Df/+, PBac/+ and C27/+ (dSap-r heterozygotes); PBac/Df, C27/Df and C27/PBac (dSap-r mutants). * p<0.05, *** p<0.001 and **** p<0.0001.