|Home | About | Journals | Submit | Contact Us | Français|
The tumor-suppressor morphogen, Patched (Ptc), has extensive homology to the Niemann-Pick-C 1 (NPC1) protein. The NPC disease is a paediatric, progressive and fatal neurodegenerative disorder thought to be due to an abnormal accumulation of cholesterol in neurons. Here, we report that patched mutant adults develop a progressive neurodegenerative disease and their brain contains membranous and lamellar inclusions. There is also a significant reduction in the number of synaptic terminals in the brain of the mutant adults. Interestingly, feeding cholesterol to wild type flies generates inclusions in the brain, but does not cause the disease. However, feeding cholesterol to mutant flies increases synaptic connections and suppresses the disease. Our results suggest that sequestration of cholesterol in the mutant brain in the form of membranous material and inclusions affects available pool of cholesterol for cellular functions. This, in turn, negatively affects the synaptic number and contributes to the disease-state. Consistent with this, in ptc mutants there is a reduction in the pool of cholesterol esters, and cholesterol-mediated suppression of the disease accompanies an increase in cholesterol esters. We further show that Ptc does not function directly in this process since gain-of-function for Hedgehog also induces the same disease with a reduction in the level of cholesterol esters. We believe that loss of function for ptc causes neurodegeneration via two distinct ways: de-repression of genes that interfere with lipid trafficking, and de-repression of genes outside of the lipid trafficking; the functions of both classes of genes ultimately converge on synaptic connections.
Several recent studies have shown that Drosophila can be used as an effective model system to understand human diseases. For instance, fly models for several neurodegenerative diseases such as Parkinson’s disease, Alzheimer’s disease, Huntington’s disease, Fragile-X syndrome, and several spinocerebellar ataxias have been reported (reviewed in Bonini and Fortini, 2003). Ongoing genetic studies with Drosophila neurodegenerative disease models promise to enhance our understanding of disease pathogenesis and generate target lists for future investigations and drug development. Such diseases can also be very helpful in understanding the brain itself. The work described in this paper stems from our unexpected result that the loss of function for patched (ptc) causes a progressive neurodegenerative disease in Drosophila.
The ptc gene encodes a seven-pass transmembrane protein and serves as a receptor for the Hedgehog family of signaling proteins (reviewed in Goodrich and Scott, 1998; Bhat, 1999). In humans, Ptc is a tumor suppressor protein and the loss of function for ptc leads to nevoid basal cell carcinoma, medulloblastoma, and several other tumors and developmental defects (Hahn et al., 1996; Johnson et al., 1996). In Drosophila, the loss of function for ptc results in defects in neuroblast formation and identity specification in the ventral nerve cord, as well as defects in segmentation of the embryonic ectoderm and headcase development (Bhat, 1996; Bejsovec and Wieschaus, 1993; Shyamala and Bhat, 2002). Previous results indicate that Ptc represses Smoothened (Smo), a G-protein-coupled transmembrane protein and the effector of Hh-signaling, from activating downstream target genes (Goodrich and Scott, 1998; Bhat, 1999). In cells where such an activation of downstream genes is required, the interaction of Hh with Ptc relieves the repression of Smo by Ptc, thus allowing Smo to activate downstream genes (such as wingless and also ptc itself).
Ptc has extensive homology to the Niemann Pick C 1 (NPC1) protein. For example, both Ptc and NPC1 proteins are seven-pass membrane proteins of similar sizes, both have sterol-sensing domains located approximately at the middle of the protein, the two proteins also have two of the what are known as permease domains (present in MexD and AcrB bacterial proteins), placed at the C-terminal portion of the protein; the two proteins have significant homology in these domains, including the transmembrane domains (Loftus, et al., 1997). However, any functional sharing between these proteins has not been explored and it remains unknown if this structural similarity translates into and functional sharing in developmental processes or disease or it is just a coincidence.
The NPC disease is a fatal, inherited, paediatric, and progressive neurodegenerative disorder thought to be due to the abnormal accumulation of cholesterol in neurons (Blanchette-Mackie et al., 1988; Suzuki et al., 1995; Carstea et al., 1997; Loftus et al., 1997). A faulty cholesterol homeostasis is thought to play a role in the development of this disease. The disease is either caused by mutations in the NPC1 gene, which encodes a transmembrane protein (Carstea et al., 1997; Loftus et al., 1997), or in the HE1/NPC2 gene, which encodes a cholesterol-binding lysosomal protein (Naureckiene et al., 2000). The exact mechanism, by which these proteins prevent this neurodegenerative disease, or the role of cholesterol in the manifestation of the disease-state, is not understood. While it is thought that in NPC, formation of inclusions in neurons is causing neuronal death, resulting in a disease-state, no direct evidence exists to link the inclusions to disease-state. Additionally, it is not clear if the inclusions are actually caused/constituted by cholesterol. A recent study suggests that in the case of Huntington’s disease, formation of inclusion bodies is a cellular response to decrease the mutant protein and prevent the poisoning of other neurons (Arrasate et al., 2004).
A mouse model for the NPC disease has also been developed (Loftus et al., 1997), but the molecular basis for the disease is not clear in this system as well. In Drosophila, there are two NPC1 genes, dNPC1a and dNPC1b. Mutants for either of the two genes die at the larval stage (Fluegel et al., 2006; Huang et al., 2005). The study from Huang et al (2005) reported that while the brain in these rescued mutants were normal, the malphigian tubule (the kidney equivalent in flies) had the highest sterol accumulation and contained large multilamellar inclusions. However, a recent paper by Phillips et al (2008) reported that null mutants for dNPC1a gene mimic human NPC patients with progressive motor defects and reduced life span; the brains of these mutants were reportedly contain higher levels of cholesterol and multilamellar inclusions. The reasons for discrepancy between the earlier study by Huang et al (2005) and this study is not clear at this moment.
In a modifier screen, we had previously isolated a hypomorphic mutation of ptc, ptcheadless (ptchdl), which, in combination with several alleles of ptc, produced few adult escapers. We also found that a viable GAL4 insertion line (ptcgal4), which in combination with null alleles of ptc, produced viable adults. These ptc mutant adults developed a progressive NPC-like neurodegenerative disease and their brain contained lamellar inclusions. The levels of cholesterol esters in these individuals were also reduced. We also observed a significant reduction in the number of synaptic terminals in the brain of the mutant adults. Our results show that feeding cholesterol enhances the synaptic connections and suppresses the disease; this also restores the levels of cholesterol esters. The results further show that Ptc does not function directly in this process but via repressing Smoothened (Smo). Thus, the gain of function for Hedgehog also induces the same disease, and the disease and the loss of synaptic connections in ptc mutants can be suppressed by simultaneously reducing the dosage of smo. These results provide novel insight into the role of synaptic connections and their maintenance in neural diseases.
The ptcgal4 line has been previously described. It is an insertion of the gal4 gene in the regulatory sequence of the ptc gene. ptchdl allele was isolated in an F1 modifier screen (Shyamala and Bhat, 2002). Other stocks used: Df (2R) NP3/CyO (ptcdf), hs-hh, smo1, ptcdf, smo1 and ptcH84, smo1. Other ptc alleles used: ptcIN108, ptcH84, ptcS2, ptcIIR87, ptcIF85, ptcIIC84, ptc9B28 and UAS-ptc. For wild type, Oregon R, Canton-S flies and white1118 were used. For analysis, unless otherwise stated, we used ptcgal4/ptcdf. All the experiments were done at 22 °C except for those involving Hs-hh, which were done at 18 °C and 29 °C (see text for details).
Four Newly eclosed females and four newly eclosed males of wild type or ptc mutants were placed in a cornmeal-agar vial and their locomotor activity and death were recorded every three days for four weeks. Vials were changed once in every three days. The testing of the flies for stage 1 or 2/3 defects was done by transferring flies to new vials to avoid any interference due to stickiness in older vials. One or two trials were conducted double blind in all experiments. For rescue of the phenotype, we used a UAS-ptc transgene (Shyamala and Bhat, 2002). This transgene was introduced to ptcgal4/ptcdf background and the adults were monitored for the progression of the disease. Following are the details of the staging in our locomoter assay:
Please refer to the details and movie clips shown in Supplementary data.
To determine if overloading the cholesterol metabolism in wild type would result in the formation of inclusions in the brain and cause behavioural defects, newly eclosed wild type adults were fed with a small amount of yeast paste containing 5% cholesterol (W/W; Sigma, stored at -70 °C) placed on top of the food medium once a day (the vials were changed once every day) and their locomotor activity was monitored once a day. Cholesterol was manually mixed with yeast paste thoroughly until detectable cholesterol crystals were not observed. It is possible that the cholesterol is not dissolved in the food; our idea is to make it thoroughly dispersed in the food. For EM analysis, flies fed with yeast paste with or without cholesterol for 4-5 days were used.
To determine the effect of feeding cholesterol to ptc mutant flies, newly eclosed four females and four males of wild type or ptc mutants were placed in a vial of 1% Agarose containing 1% sucrose solution. A small amount of yeast paste with or without 5% cholesterol was placed on top of the Agarose medium (this feeding regimen in wild type also causes inclusion formation similar to the feeding regimen in experiment I discussed above). Flies were transferred into new vials once every three days. The number of dead flies was recorded on the first, fifth, ninth, and fourteenth day when the experiment was terminated. The testing of the flies for stage 1 or 2/3 defects was done in new vials.
Wild type and ptc mutants, aged three weeks, were used in this experiment. Three separate experiments with triplicates for wild type and mutant, fed with normal food or food with cholesterol, were done. Three measurements of each sample were taken and then averaged. To measure cholesterol, the Cholesterol Quantification kit from Biovision was used. Each sample started out with twenty heads (10 males+10 females) homogenized in 200 uL of 1% Triton X-100 in 100% chloroform. After the procedure, the extract was dissolved in 200 uL of reaction buffer from which 25 uL were taken for the fluorometric reaction.
The qPCR was done in triplicates of 100 heads each for wild type and ptcgal/ptcdf, aged 2 weeks. Note that mutant flies used were not selected for only those that show the strong phenotype, but a collection of those that show stage 1, stage 2 and stage 3 defects. Heads were dissected and immediately frozen in liquid nitrogen. The total RNA was isolated using the RNAqueous kit from Ambion. The cDNA synthesis and real-time PCR were done in the molecular Genomics Core facility at UTMB using the SYBR Green dye. We used the RpL32 gene as a standard. We used primers RpL32-92F (ATCGATATGCTAAGCTGTCGCAC) and RpL32-182R (GGCATCAGATACTGTCCCTTGAAG) with amplicon of 91 bp. For the ptc gene we used primers PTC-852F (TCTCGGATCTTTACATACGCACC) and PTC-971R (GGGACTGGAATACTGATCGCAG) with amplicon of 120 bp. For data analysis we have used the comparative CT method (Livak and Schmittgen, 2001). The graph shows the results as a percentage of wild type expression level (which is 100%).
Hs-hh transgenic flies were raised at 18 °C and the progeny were also kept at 18°C throughout the developmental stages. Immediately following eclosion one set of adults were transferred to 29 °C and the second set was kept at 18 °C. Each set contained ten vials; each vial had four males and four females. Flies were transferred into new vials once every three days. After three weeks, flies were scored for locomotor defects and lethality.
Hs-hh flies and control were either kept at 18 °C or shifted to 29 °C immediately after eclosion. They were kept at these temperatures for 2 weeks. For each experiment 60 flies total (20 flies per experiment, in triplicates) were used. The levels of cholesterol were determined using the cholesterol measuring kit as above.
To determine if reducing the dosage of smo suppresses the disease, we generated ptcdf, smo1 and ptcH84, smo1 double mutants. These mutants were then crossed to ptcgal4 and the progeny adults of the genotype ptcdf, smo1/ptc,+ and ptcH84, smo1/ptc,+ were followed for the progression of the disease.
Flies were fixed in freshly prepared 4% paraformaldehyde and 2.5% glutaraldehyde in 1M sodium cacodylate buffer overnight at 4 °C. The samples were treated with 1% osmium tetroxide in 1M sodium cacodylate buffer containing 1.5% potassium ferrocyanide for 1 hr. They were dehydrated in graded series of alcohol and embedded in Epon, 70-90nm thin sections were cut and stained with 2% urinyl acetate and lead citrate. The sections were analyzed with a transmission electron microscope. To determine if the brain in the ptc mutant has neurodegeneration at the gross morphological level, the brains were dissected in PBS, fixed, embedded and sections were stained with 1% toluidine blue solution. These sections were then examined under microscopy.
Adult brains were stained with an antibody against Ptc (Forbes and Spradling, 1996) and an antibody against Syt (Littleton et al., 1993). Wild type or ptc mutant flies were kept under the same condition as above during the cholesterol feeding experiment. After 0 day (the same day as the flies eclosed), 8 days, and 16 days, flies were anesthetized and the brains were dissected in ice-cold PBS. The brains were treated with 1mg/ml Collagenase type VII in PBS for 15 minutes. After rinsing twice in PBS containing Triton X100, 0.05% (PBST) the brains were fixed in 4% paraformaldehyde for 1 hour. These brains were washed in PBST, three times each for 10 min and were blocked in 1% Normal Goat Serum (NGS) for three hours at 4°C. The brains were then incubated with rabbit anti-Syt or Ptc antibody in PBST and 10% NGS for two days at 4°C. After three hours of washing, the brains were again blocked for one hour at 4°C and were incubated with the secondary antibody (Alexa Fluor 488 labelled goat anti-rabbit) for two days at 4°C. After washing in PBST for three hours at RT, the brains were mounted on slides using Vectashield and analyzed under the ApoTome Imaging System. Using the Axio Vision software, we created the custom method for counting the number and area of Syt staining in the brain. For regular confocal imaging, the brains were incubated with a FITC-labeled secondary antibody after Syt staining. To determine the number of Syt-positive connections in ptc, smo/ptc, + flies, 16-day-old flies were examined.
We had previously isolated a mutation in the ptc gene, ptcheadless (ptchdl; Shyamala and Bhat, 2002); while ~99% of the homozygous individuals die as late pupa with partially missing headcases, ~1% made it to adulthood with most of them showing headcase defects (Shyamala and Bhat, 2002). These escapers showed aberrant locomotor activities as adults in an age-dependent manner. We also examined individuals that are transheterozygous for a known viable allele of ptc, ptcgal4, where the gal4 gene is inserted into the regulatory region of ptc (Wilder and Perrimon, 1995), and ptc null alleles or ptc deficiencies (ptcdf). About 3% of these individuals also showed the same headcase defects with partially missing headcase (Fig. 1C) as ptchdl while the remaining 97% were morphologically normal (Fig. 1B) similar to wild type (Fig. 1A). We sought to examine if part of the brain is missing in those individuals that have headcase defects. The head capsule in Drosophila is generated from the eye-antennal disc (Haynie and Bryant, 1986). We found that in these individuals, the brain cells are not eliminated; instead, the brain corresponding to the missing half of the headcase is located more interiorly in the thorax (Fig. 1E). Despite this gross morphological defect and the abnormal brain location, these flies are normal when they eclose and are fertile and reproduce (see below).
While most of the ptcgal/ptcdf flies are normal when they eclose, including those with partially missing headcase, they soon show signs of a locomoter disease: sluggishness and difficulty walking, particularly affected is the movement and coordination of the legs. We designate this as stage 1 of the disease (see Supplementary data). These locomotor defects progress (Fig. 1F), and their walking become unsteady, and they are unable to hold on to smooth surfaces against gravity (stage 2; see Supplementary data). About a quarter of these flies become paralyzed (stage 3, see Supplementary data) and die between the second and the fifth week. These defects were progressive in nature and by 30 days of time, ~15% of the flies were dead, nearly 30% had stage 2/3 defects and another 20% were showing strong stage 1 defects (Fig. 1F). Note that the remaining 35% of the mutant flies also showed a weaker stage 1 defect. Wild type control flies showed ~2% lethality over a period of four weeks but none exhibited any locomotor defects. The progressive lethality and the locomotor defects are fully rescued with a ptc transgene (UAS-ptc; ptcgal4/ptcdf), indicating that the defects are due to loss of function for ptc. Note that the stronger ptchdl allele has grown stronger since its isolation and now has a fully penetrant headcase defect and dies as late pupae with perhaps additional defects (ptc is required for multiple tissue development), we have not been able to rescue the homozygous ptchdl pupal lethality with any of our gal4 drivers and have not found an appropriate driver.
We also examined the progression of the disease in the small percentage of the mutant individuals that show the head defects with part of their brain stuck within the thorax. We also determined if the mutant individuals with head capsule defects show a higher incidence of the behavioral defects. A comparison of the frequency of incidence and the progression of the disease between the two groups of mutant adults reveals that the mutant adults that are normal-looking developed the disease as frequently as flies with eye-headcase defects. Similarly, the percentage of mutant adults with headcase defects not developing any severe symptoms of the disease was the same as the mutant adults without any morphological defects. Thus, the above behavioral symptoms are unlikely related to any developmental or head/brain morphological defects in these mutants. This conclusion is consistent with the fact that the behavioral symptoms are rarely observed in the mutant adults when they are born, but develop only later and they are also progressive in nature.
We also found that ptchdl in trans to several other alleles of ptc are viable (between 3-5% of such viable adults show head capsule defects). The penetrance of the locomotor defects (stage 2) in adult flies transheterozygous for ptchdl and several different alleles of ptc are as follows: ptchdl/ptcS2 (0%), ptchdl/ptcIIR87 (9%), ptchdl/ptcIF85 (7%), ptchdl/ptcIIC84 (12%), ptchdl/ptc9B28 (4%).
We next examined the brain sections of ptc mutants that showed significant locomotor deficits (stage 3) by Toluidine Blue staining. For this purpose, we used the strongest allele, ptchdl (this mutation no longer produces homozygotes and over ptcdf is embryonic lethal; over ptcgal, it shows a weaker locomotor defect, therefore, we have not used this mutation extensively in this study). As shown in Fig. 2B, the brain of ptchdl mutants (stage 2-3) had holes in several areas (long arrows), which were absent in wild type (Fig. 2A). We did not see these holes in the mutants when they are showing the stage 1 of the disease. We also examined brains from ptchdl mutant individuals by EM. As shown in Fig. 2D, we found regions of the brain with degenerating neurons and holes (Fig. 2D, star). The brain in these individuals had droplets of lipids (arrows), which were observed at much less frequently in wild type (the ratio between wild type and the mutant is about 1:10; N=6 brains examined); these lipid droplets are also found in the weaker ptcgal4/ptcdf flies (data not shown) but only slightly more than wild type. In this weaker allelic combination, we did not find any significant holes in the brain as we saw in the stronger combination, however, the brains of stage 3 individuals (and not stage 1) in the weaker combination were fragile (as judged during dissection) and somewhat smaller compared to wild type.
We stained wild type adult brains with an antibody against Ptc (Forbes and Spradling, 1996). As shown in Fig. 3A-C, Ptc is present in a large number of axon tracts; it is present in the dendritic fibres of the mushroom body calyx and the KC cells but the rest of the mushroom body was negative (Fig. 3B). It is also expressed in nerve fibres of the antennal lobe, antennal nerve bundle (Fig. 3C) and secondary neuronal projections of the retina and several other structures (data not shown). Ptc is not present in the protocerebral bridge. We could not stain ptc mutant brains with this antibody since our supply of this antibody was exhausted. Since whole mount RNA in situ results are often not reliable in a hypomorphic mutant scenario as the case here, we examined the levels of expression of ptc RNA in the mutant using quantitative PCR. As shown in Fig. 3E, the levels of ptc RNA in the mutant (ptcgal/ptcdf) appears to be about 67% of the wild type. Taken together with the result that a ptc transgene expressed from ptcgal4 fully rescues the disease in the ptc mutant (UAS-ptc; ptcgal4/ptcdf), we conclude that the normal function of Ptc is affected in the mutant, which then leads to the development of the disease in adults.
To determine if the progressive locomotor deficits in ptc mutant adults indeed have a neural basis, we performed EM analysis of the brain sections from affected individuals. As shown in Fig. 4, at the ultrastructural level, brain sections in flies (ptcgal/ptcdf) with severe locomotor defects (about 2-3 weeks old flies) showed the presence of lamellar inclusions in neurons (Fig. 4b-h). We found that brain sections contained 2.7-48 inclusions per 100 μm2 (the number of sections examined per brain was nine, and five different brains; typical were about 6 inclusions per 100 μm2). We also found brain sections that had no inclusions, which may represent an area of the brain not subjected to any ptc effect. Additionally, the brain in a mutant with severe locomotor defects showed the presence of extensive membranous material (Fig. 4c, d, e). These membranous materials appear to be the precursors for inclusions, i.e., they appear to organize into lamellar bodies (Fig. 4c-h). However, it appears that not all of the membranous material eventually becomes inclusions. Some of the inclusions in the stronger ptchdl mutants (Fig. 4i-n) resemble those found in the brains of NPC mutant mice and humans (Fig. 4i-k; note the dense material of unknown nature at the central regions of the inclusions) and in diseases caused by the accumulation of gangliosides (Fig. 4l-n), collectively called gangliosidoses (Glomset et al. 1995; Suzuki et al. 1995; Gravel et al. 1995). In wild type, in mutant flies that show no locomoter defects, mutant flies in their first week, or mutant flies that were rescued using a UAS-ptc transgene, we rarely find any inclusions (~0.3 in an area of 100 μm2, number of sections examined is five per brain and four different brains).
Because of the above results, we thought that there might be an abnormal accumulation of cholesterol in the mutant brain. While Filipin staining is a method that has been used to show accumulation of cholesterol, we realized that it works well in a monolayer of cells in tissue culture but it is not reliable for the analysis of the adult brain because of accessibility, autoflorescence, and the very rapid decay of the signal. Therefore, to determine if the inclusions in the ptc mutant brain are due to an abnormal cholesterol metabolism, we took a different approach. We fed wild type flies with yeast-paste containing 5% cholesterol for one week. We reasoned that a large amount of cholesterol might overload the system and the excess cholesterol might accumulate in the E/L system. Indeed, feeding 5% cholesterol in the diet resulted in the formation of membranous material and lamellar inclusions in the brains of wild type adults (Fig. 4o and p). We found 5.7-65 inclusions per 100 μm2 (the number of sections examined was six per brain and four different brains; typically these brains had about 12 inclusions per 100 μm2). Less than one (~0.3) inclusion/100 μm2 was found in age-matched wild type flies fed similarly with yeast paste without any cholesterol (Fig. 4q). We monitored these flies for any locomotor and abnormal behavioural changes. However, these wild type flies fed with cholesterol did not show a ptc-like locomotor or any other types of defects. Nor they died prematurely. We conclude that the excess of cholesterol leads to the formation of membranous material and lamellar inclusions in the brain; however, their presence alone does not cause the disease-state. By extension, it seems likely that the presence of inclusions per se in the mutant brain is not responsible for the disease-state or causes any type of locomoter defects.
We next fed cholesterol to ptc mutant flies (ptcgal/ptcdf). To ensure that flies ate only the yeast paste containing cholesterol, we placed them in Agarose vials (instead of the corn meal food) and the yeast paste containing 5% cholesterol was placed on top of the Agarose medium (note that there is no endogenous synthesis of cholesterol in Drosophila). While this is a nutrient-poor diet and causes lethality in wild type flies as well (see Fig. 5), which is likely due to the stressful condition (see Discussion), it standardizes the feeding experiments since flies eat only the yeast paste as opposed to eating also the corn meal fly food. This allowed us to measure the effect of cholesterol more reliably. The disadvantage is that flies start to die after two weeks (both wild type and mutant) because of poor nutrition; therefore the flies can be reliably monitored for about two weeks. However, since the wild type do not exhibit any locomoter defects even under this stressful condition and we evaluate flies for the three different stages of locomoter defects, we think that the data are valid even with lethality in wild type. Thus, the advantages outweigh the disadvantages of this method of feeding cholesterol. We found that feeding mutant flies (ptcgal/ptcdf) with 5% cholesterol resulted in a suppression of the locomotor defects, progression of the disease and lethality (Fig. 5; compare normal food versus cholesterol food; note that normal food means food without cholesterol). Feeding 10% cholesterol did not increase the survival rate, and a 1% cholesterol diet had no significant effect. These results were reproducible and consistent (see Discussion). These mutant flies fed with cholesterol had typically 8 inclusions per 100 μm2 area. These inclusions in mutants (or in wild type) fed with cholesterol, however, were much larger compared to mutants not fed with cholesterol.
In order to explore the relationship between cholesterol and the disease-state, we next assayed the levels of cholesterol in wild type and the mutant at intervals of day 1, day 7 and day 14. We also measured the levels in flies fed with cholesterol. We measured the total cholesterol, free cholesterol and cholesterol esters. We thought that measuring the levels of cholesterol esters is particularly useful since in NPC disease, it is thought that unesterified cholesterol accumulates in the EL system of neurons; if the situation is similar in ptc mutants, we would expect a reduction in the levels of cholesterol esters in the mutant. For assaying cholesterol, we used a cholesterol quantification kit (Biovision). As shown in Fig. 6, several results are of interest. First, in wild type fed with normal food (without cholesterol), we found a gradual increase in the levels of cholesterol, all three forms: total, free and esters, over the measured intervals from day 1 to day 14 (Fig. 6). The increase in levels from day 1 and day 14 is most pronounced in esters of cholesterol, which is about 2.5 times; however, this increase from day 1 to day 7 is about 2 times, indicating that the level is reaching plateau from day 7 to day 14. Second, in wild type the levels of esters do not appear to be significantly increased with cholesterol feeding, although there is an increase in the free and the total cholesterol levels (Fig. 6). This is consistent with the finding that when wild type flies are fed with cholesterol, multi-lamellar inclusions appear in their brain in large numbers, likely a consequence of accumulation of unesterified cholesterol in the E/L system. This also suggests a tight regulation of levels/formation of cholesterol ester in wild type and also a similarity to the NPC disease in vertebrates.
Third, the levels of esters are significantly reduced in the mutant from day 1 itself, with about half the levels in wild type. The levels of esters appear to reach a plateau in the mutant by day 7 since there is not much difference between day 7 and day 14 in the levels of cholesterol esters (Fig. 6). Compared to wild type, however, this level difference is the maximum at day 14, nearly 2.5 times, whereas at day 1 or day 7, the differences are much less (see Fig. 6). Fourth, with cholesterol feeding to mutants, the levels of cholesterol increase significantly and by day 14, it is slightly more than the level in wild type (Fig. 6). These results show that in the mutant, the levels of esterified cholesterol is reduced, however, feeding cholesterol restores the level in mutants. Since there is an increase in the level of total cholesterol as well, we cannot be certain that the rescue observed is only due to an increase in the esters, although it is more likely that esters are responsible.
Given the above results, we decided to examine the mutant brains with an antibody against Drosophila Synaptotagmin I (Syt) to determine if the synaptic terminals are affected in their brain. Synaptic function is known to mediate neuronal survival during development (Nguyen et al., 2001). Pre-synaptic proteins such as Synuclein have been directly implicated in neurodegenerative diseases (Surguchov et al., 2001). Pre-synaptic nerve terminals are enriched with pre-synaptic vesicles and Syt is localized to these vesicles, where it is one of the principal proteins. Syt appears to play an important role in both the exocytosis and endocytosis of these pre-synaptic vesicles (Sudhof and Rizo, 1996; Nicholson-Tomishima and Ryan, 2004). Thus, an active pre-synaptic nerve terminal is enriched for Syt and we reasoned that Syt staining using an anti-Syt antibody (Littleton et al., 1993) should give us a measure of the synaptic connectivity in the brain.
We selected the dendritic-rich calyx (ca) and the protocerebral bridge (pr br) to evaluate Syt-positive terminals in mutant brains. This is due to the fact that Ptc is present in the calyx but not in the pr br (see Fig. 3) and thus they provide an excellent test structure (calyx) and an internal control (pr br). Furthermore, unlike the central complex, which is a large area in the brain, both the calyx and the pr br are discrete structures and therefore the synaptic connections can be reliably quantified. Using a Zeiss ApoTome microscope and the Axiovision computer program, we counted the number of Syt-positive “areas” and the μm2 of the counted Syt “areas” in each slice-frame across the calyx and the pr br of Syt-stained adult brains. Wild type flies fed with normal food (yeast paste), wild type flies fed with yeast paste containing cholesterol, ptc mutant flies fed with normal yeast paste, and ptc mutant flies fed with yeast paste containing cholesterol were examined. We also examined wild type and mutant flies that were 0-days old (the same day the flies were eclosed), 8-days old, and 16-days old, grown in Agarose with yeast paste that had either cholesterol or no cholesterol.
As shown in Fig. 7, several findings are of interest. In wild type, we found that the number of Syt-positive areas is the highest soon after eclosion (~1300) and decreases by eight days (~800) and further decreases by 16 days (~600). However, the Syt-positive areas in the pr br increase from about 300 when eclosed to ~450 by eight days but drops to ~300 by 16 days. This may be due to changes in synaptic density. While we observed a decrease in the Syt areas in the calyx in 8-day-old wild type flies fed with cholesterol (from ~1000 to 900), there was an up-regulation in 16-day-old flies with cholesterol feeding (from ~670 to 800). In the pr br, however, the differences were not very significant with or without cholesterol, although cholesterol seems to increase the number slightly in 16-day-old flies.
In the ptc mutant, we found a decrease in the number of Syt areas in the calyx even in newly eclosed flies (Fig. 7). By 8 days, the number was decreased to ~700 (compared to ~825 in wild type) and to 400 by 16 days (versus ~700 in wild type). However, the numbers in pr br are not significantly affected in the ptc mutant brain, though there is a slight decrease by 16 days of age. With cholesterol feeding, the number of Syt areas in the calyx increased slightly in 8 days, but this increase or a block in the decrease was most pronounced by 16 days. The pr br did not show any significant difference with or without cholesterol.
However, the strength of the Syt-positive areas (as measured by the μm2 of the areas), followed a more complex pattern, perhaps indicative of the synaptic density, but in general, showed an increase with cholesterol both in wild type and the mutant (Fig. 7). The strength was also enhanced in the pr br with cholesterol in wild type; however, the effect was less pronounced in the ptc mutant and there was not much effect in 16-day samples in the mutant. Representative examples to illustrate these results are shown in Fig. 7a-d for 0-day, Fig. 7a-i for 8-day and Fig. 7a-h for 16-day time periods. These results also indicate that the loss of function for ptc does not affect Syt expression per se. It must be noted that in some instances, there is a decrease in the number of staining areas while the size of the counted areas increases (e. g. day 7, WT cholesterol-fed flies in Fig 7). It seems possible that this represents an increase in synaptic density (see also below at the EM level) with cholesterol, which is particularly pronounced in wild type. Also, since this particular mutant combination does not show any obvious neurodegeneration, the reduction in Syt therefore is unlikely due to a loss of neurons per se; moreover, feeding cholesterol rescues the loss of Syt connections in the mutant.
We next sought to examine synaptic connections in the brain of mutant individuals using EM. We were unable to exactly pinpoint the calyx and the pr br in the brain by EM (there is no EM atlas available for the fly brain). Instead, we decided to examine the central brain area to see if there is any difference in the synaptic connections between wild type and the mutant with and without cholesterol feeding. While the precise location of the EM section in this region between samples is not accurate (it will be impossible to perfectly match), we tried to focus as best as we can on the same area and level. We took flies that were 3 weeks old, both wild type and mutants (ptcgal/ptcdf), raised under the same condition, fed with either normal food or food with 5% cholesterol. The mutant flies were all showing the stage 3 of the disease and the experiments were all done at the same time. As shown in Fig. 8, a significant decrease in the synaptic number in the brains of ptc mutants (Fig. 8b) compared to wild type (Fig. 8a) was observed. The number was much higher in the brains of ptc mutant flies fed with cholesterol. The number of synaptic connections and the density also appears to be higher in wild type flies fed with cholesterol. At this point, we do not know the significance of an increase in synaptic density; on the other hand, it is difficult to envision that synaptic density will have no relevance to the functional aspect of a synaptic connection. This is an area that needs to explore further. We also note that EM examination of mutant flies that do not show stage 2/3 of the disease have on the average about 70 (plus or minus 5) connections in the same region (data not shown), suggesting that these flies also likely suffer from loss of synaptic connections. These results are consistent with the Syt staining results and thus argues that the Syt staining and quantification using our procedure is a reliable way to determine synaptic number and perhaps density as well.
We next wanted to determine if Ptc functions directly and independent of Smo or via regulating Smo. The canonical way Ptc signaling works is that Ptc represses Smo from activating downstream genes. An interaction of Hh with Ptc relieves this repression of Smo by Ptc. Smo then triggers the activation of downstream genes (Goodrich and Scott, 1998; Bhat, 1999). Accordingly, gain of function Hh induces the same phenotypes as loss of function Ptc. Therefore, if Ptc functions via repressing Smo from activating downstream genes that would otherwise cause the neurodegenerative disease, the same ptc-phenotype is likely produced by a gain of function hh. If the role of Ptc is direct, say via vesicular trafficking of neurotransmitters (Ptc is present in vesicles and has been implicated in vesicular trafficking of Smo, see Martin et al., 2001; Strutt et al., 2001), altering the activity of hh is unlikely to cause a neurodegenerative disease. The caveat here is that a high level of Hh might titrate out Ptc and still can cause the disease.
When adults that carry the hh transgene under the control of heat shock 70-gene promoter (Hs-hh) were continuously cultured at 29 °C, these individuals developed the same locomotor defects as ptc mutant adults (Fig. 9A). Moreover, as shown in Fig. 8e, EM examination of brain sections from these individuals revealed that their neurons had inclusions similar to ptc mutants (19.7-65 inclusions per 100 μm2, n=5, from four brains); the number of synaptic connections was also negatively affected in Hs-hh flies raised at 29 °C (Fig. 9h). In the control flies where the Hs-hh flies were raised at 18°C, we found 1.3-9.3 inclusions per 100 μm2 (n=5, from two brains), however, the synaptic connections were the same as in wild type (data not shown). Age-matched wild type flies kept at 29 °C under the same condition and the same way as Hs-hh did not have inclusions (Fig. 8f) and the number of synaptic connections was marginally lower compared to wild type kept at room temperature (Fig. 8h).
We further examined if the levels of cholesterol is affected in gain of function hh. As shown in Fig. 9B, we found that Hs-hh flies kept at 29 °C had approximately half the levels of cholesterol esters compared to similarly kept wild type flies. There is also a significant reduction in the total and free cholesterol in these Hs-hh flies compared to wild type. At 18 °C, we observed an unexplained spike in the levels in Hs-hh flies compared to wild type (Fig. 9B). Since flies do poorly at 29 °C when they are grown under stressful agarose-only media conditions, we did not perform rescue experiments in Hs-hh flies with cholesterol feeding.
That Ptc does not play a direct role is also indicated by our findings that flies transheterozygous for ptcS2 (an allele that carries mutation in the sterol-sensing domain of Ptc; Strutt et al. 2001), and ptcgal4, or ptcS2 and ptchdl, do not show any significant locomotor or brain defects. Moreover, the above result that ptc loss of function defects can be induced in hh gain of function individuals at any time in adult life by simply shifting them to 29 °C indicates that the defects in ptc mutants are not due to any developmental abnormality.
We also sought to obtain further evidence that Ptc functions via regulating Smo in the brain. Therefore, we examined the effect of reducing the dosage of smo in ptc mutant background. As shown in Fig.10A, reducing the dosage of smo significantly suppressed the death rate of ptc mutants. Consistent with this, the synaptic connections as well as the density (areas) were also up regulated in ptc, smo/ptc, + adults (Fig. 10B and C).
Here, we show that loss of function for ptc causes a neurodegenerative disease that has similarity to the NPC disease in vertebrates. This involvement of Ptc in a neurodegenerative disease defines a novel role for Ptc. The similarity to the NPC disease is consistent with the fact that NPC1 and Ptc share significant sequence homology. Our results also show that the loss of function for Ptc results in a reduction in the levels of cholesterol esters in the brain. Loss of Ptc function also progressively affects the number of Syt-positive synaptic connections in such structures as the dendritic-rich calyx of the brain. The fact that levels of cholesterol esters, and the number of connections in the mutant brain can be restored by feeding cholesterol, and that the disease can be suppressed by feeding cholesterol argues a primary role for cholesterol esters and synaptic connections in the progression of the disease. These results are also supported by the EM data that the number of connections are reduced in the mutant brain, which can be restored by feeding cholesterol. These results also reveal a role for cholesterol esters in forming and or maintaining synaptic connections, either directly or indirectly. Moreover, our results clearly show that inclusions per se are not the problem since wild type flies can have the inclusion but that does not cause the disease. These, we believe, are novel results and shed new insight into the basic science of neurodegenerative diseases in general. Our work also demonstrates new tools to analyze the adult brain that should generally be applicable in adult brain development and functional studies.
There are two NPC1 genes in Drosophila, NPC1a and NPC1b. The Ptc protein shares significant homology to the NPC1 proteins in Drosophila or to the vertebrate NPC1 protein. Two previous studies showed that null mutants for NPC1a die as first instar larvae, but these mutant larvae can be rescued to adulthood by feeding them cholesterol (Flugel et al., 2006; Huang et al. 2005). The study from Huang et al (2005) reported that while the brain in these rescued mutants were normal, the malphigian tubule (the kidney equivalent in flies) had the highest sterol accumulation and contained large multi-lamellar inclusions. These inclusions were strikingly similar to what we have found in the brain of ptc mutants. Moreover, in NPC disease, it is thought that unesterified cholesterol accumulates in the E/L system. Our results indicate that there is a reduction in the pool of esterified cholesterol. These results suggest that the cholesterol metabolism is affected in both NPC disease and ptc mutants.
These results further suggest that NPC1 genes in flies may play a smaller role in the brain. This possibility is consistent with the more recent results by Phillips et al (2008). These authors reported that null mutants for dNPC1a gene mimic human NPC patients with progressive motor defects and reduced life span; the brains of these mutants reportedly contain higher levels of cholesterol and multi-lamellar inclusions. The reasons for this discrepancy between the earlier study by Huang et al (2005) and this study by Phillips et al (2008) is not clear at this moment; it is also not clear if there is any redundancy between dNPC1b and dNPC1a.
What is the relationship between ptc and dNPC1? We have observed that ptc mutant individuals that are also heterozygous for NPC1a (ptc/ptc; NPC1a/+) do not show an enhancement of the disease (data not shown). While we have not examined if the double mutants between ptc and dNPC1 mutants show an enhanced neurodegenerative phenotype, it may be that one needs to eliminate also both NPC1 genes (dNPC1a and dNPC1b) in order to observe a strong brain phenotype. Nonetheless, it seems likely that the ptc-neurodegenerative disease in flies is closest to the NPC-disease in vertebrates. In vertebrates, since loss of function for ptc die as embryos, we do not know if Ptc has any role in preventing a neurodegenerative disease; however, at least in Drosophila, Ptc may function in the same pathway as the NPC1 proteins. Additional work is needed to fully determine the relationship between dNPC1a and dNPC1b as well as between dNPC1 genes and ptc.
The role of Hh-Ptc-Smo signaling in development and disease has been examined in many different studies over the last several years. This study reveals that Ptc has a role in preventing a neurodegenerative disease. This disease is not due to a strange allele of ptc but due to the loss of function for ptc. We observed the disease in several different allelic combinations of ptc, including combinations of known loss of function alleles and deficiencies (which also rules out a background-mediated locomotor defect). The phenotype can also be rescued by a ptc transgene. Ptc is widely expressed in the adult brain, and at the same time, the expression of Hh is restricted to mostly olfactory lobes (data not shown), suggesting that Ptc plays a repressive role in the development of the disease. This is also consistent with the result that a gain of function for Hh causes the same disease as the loss of function for ptc. Moreover, the fact that inducing Hh in the brain at any point in the adult life will cause the disease indicates that the disease in the loss of function for ptc is not due to some developmental abnormality. Our results with Hs-hh also indicate that Ptc functions indirectly via repressing Smo. This conclusion is consistent with the result that reducing the dosage of smo alleviates the disease.
It is thought that in NPC disease, formation of inclusions in neurons causes neuronal death, which results in a disease-state. However, no direct evidence exists to link the inclusions to disease-state. A previous study suggests that in the case of Huntington’s disease, formation of inclusion bodies is a cellular response to decrease the mutant protein and prevent the poisoning of other neurons (Arrasate et al., 2004). Therefore, formation of such inclusions is a good thing for neurons in slowing down the disease. Our results show that the inclusions in the brain of ptc mutants is due to cholesterol: feeding a large amount of cholesterol to wild type flies causes formation of inclusions in the brain. This also indicates that the presence of inclusions in the brain of ptc mutants is indeed due to an aberrant metabolism or accumulation of cholesterol in neurons. We attempted to quantify the number of inclusions in the entire brain in our EM experiments. However, we discovered that this is a difficult proposition for several reasons. First is the fact that there are no serial section EM maps of the fly brain and it is nearly impossible to pinpoint the precise location within the brain to be able to compare between brains of different genotypes, conditions and age. Second, the inclusions were not uniformly spread in the brain; therefore, we were concerned that our quantification might be misleading. The only way to avoid these two problems and be able to give a precise quantification of inclusions between brains is to do serial EM sections of the entire brain and count the number of non-overlapping inclusions in each brain.
To perform serial EM sections of an entire brain, we made the following calculations. Each EM section is about 100 nm thick and we need to cut about 1,500 sections to cover the entire brain. The area of the each section is about 150,000 μm2. In order to reliably examine the section for lamellar inclusions, we need to examine the sections at a magnification of 5,000 times. At this magnification each image will cover 300 μm2 of an area. This means that we need to take 500 images to cover one entire section at this magnification. Thus, the total number of images one would have to take to examine an entire brain is 1,500 × 500 = 750, 000. This is not practical given that we will have to section a minimum of 3 brains for each sample. Therefore, our quantification of inclusions should be considered approximations and may or may not be representative of the entire brain. However, we want to point out that our quantification of synaptic connections from EM photomicrographs is somewhat of a different scenario since the synaptic connections are generally spread in a uniform fashion within the neuropile (and the brain overall). Therefore, the error due to imprecision of the location, though still exists, it is at a much lower level. Nevertheless, the quantification of the synaptic connections using EM should be again treated as an approximation. Thus, we resorted to examining the number of Syt-positive connections in a discrete structure such as the calyx using the Apotome microscopy between different samples of the brain, a much more accurate analysis.
Because of the technical difficulty with quantifying inclusions between brain samples (see above), we focused on examining synaptic connections in a discrete structure such as the calyx, where we can quantify the number of Syt-positive terminals using our newly developed method and draw meaningful conclusions. Our results indicate that loss of synaptic terminals is a major contributor to the disease-state. In ptc mutants, there is a significant reduction in the Syt-positive synaptic connections. One can argue that the loss of synaptic connections is due to the degenerative disease. However, our results show that loss of synaptic connections and the disease-state in ptc mutants can be suppressed by feeding cholesterol; cholesterol appear to prevent the loss of the synaptic connections and also the progression of the disease.
Normally, esterified cholesterol (LDL) is taken up inside the cell via endocytosis by the endosomal/lysosomal (E/L) system and de-esterified in lysosomes; it is then re-esterified in the cytoplasm and then stored or sent to other cellular components. For instance, the most prominent characteristic of the NPC disease at the cellular level is the blockade of intracellular transport of LDL-derived cholesterol between the E/L compartment and the plasma membrane resulting in the accumulation of unesterified cholesterol in the E/L system (Blanchette-Mackie et al., 1988; Suzuki et al., 1995; Vanier and Suzuki, 1998). This blockade of cholesterol transport appears to be the cause for the formation of inclusion bodies in neurons of NPC human patients or NPC mice. Additionally, the finding that the gene corresponding to NPC2 encodes an ubiquitously-expressed cholesterol binding lysosomal protein, known as HE1 (Naureckie et al, 2000), further supports a role for cholesterol in the disease. On the other hand, in human NPC patients, a combination of diet and cholesterol-lowering drugs had no beneficial effect (Schiffmann, 1996). Therefore, the nature of cholesterol involvement in the genesis of the disease is not known.
Our results suggest that one of the consequences of loss of function for ptc is the accumulation of cholesterol in the cytoplasm leading to the formation of inclusions and membranous material. Furthermore, there is a progressive loss of synaptic connections in the mutant brain. Thus, one possibility is that in ptc mutants, cholesterol becomes limiting, and this leads to a loss of synaptic connections. This is supported by the fact that feeding cholesterol to wild type or mutant flies increases the connections and in mutants, feeding cholesterol suppresses the disease. Results from a few previous studies are also consistent with this possibility. For example, an in vitro study suggested that cholesterol promotes formation of synapses (Mauch et al. 2001). We can also observe this in the adult fly brain. Moreover, apoE has long been suspected to be involved in neurodegenerative loss of synaptic plasticity in the Alzheimer’s disease (Herz and Beffert, 2000). Furthermore, the e4 mutant in the gene for ApoE, an important cholesterol transport protein, is associated with an increased risk of late-onset Alzheimer’s disease; this mutant isoform of the protein is less able to promote neurite outgrowth than other apoE isoforms (Strittmatter et al. 1993). It is not clear, however, if cholesterol promotes synapse formation by directly influencing the membrane property (i.e., fluidity), as a structural component, via regulating gene activity, or via the synthesis of neurosteroids. A decrease in the available cholesterol may increase the fluidity of the plasma membrane of synaptic terminals and make synaptic connections structurally unstable. Alternatively, limiting amounts of cholesterol may affect gene expression or the synthesis of neurosteroids, contributing to the disease-state. We also want to point out that the suppression of the disease with cholesterol is not complete. Thus, we think that loss of function for ptc causes neurodegeneration via two distinct ways: 1) de-repressing genes that interfere with lipid/cholesterol trafficking, and 2) de-repressing genes outside of the lipid trafficking; the functions of both classes of genes ultimately converge on synaptic connections.
We thank Drs. K. Suzuki for comments on the mutant brain pathology, Hugo Bellen for the Syt antibody, Phil Ingham for the anti-Ptc antibody, Michael Sesma, Danilo Tagle and members of the Bhat lab for comments on the manuscript. This work was initially supported by the Ara Parseghian Foundation and mainly by funding from NINDS (NIH).