Cathepsin D expression level affects alpha-synuclein processing, aggregation, and toxicity in vivo

The human SNCA gene contained in the pcDNA3.1 vector was a kind gift of Prof. P.T. Lansbury. The rat SNCA gene was cloned out of a rat brain cDNA library using the primers GATATCGCCACCATGGATGTGTTCATGAAAGGACTT (forward) and CAAGACTATGAGCCTGAAGCCTCTTCTAGA (reverse), and inserted into the pcDNA3.1 vector (Invitrogen) using the EcoRV and Xba1 restriction sites. pCMV-XL5 vector with and without the human CTSD gene was purchased from Origene's TrueClone collection. All constructs were isolated from DH5alpha E. coli using Qiagen's Maxiprep kit. To create the Ala30Pro, Glu46Lys, Ala53Thr, Ser129Ala, Ser129Asp, and Asp98Ala/Gln99Ala mutants of human SNCA, site directed mutagenesis was performed using Stratagene's QuikChange kit. All constructs were sequence verified prior to use.

The dopaminergic rodent mesencephalic cell culture system (MES 23.5 cells) was a gift of S Appel, Baylor College of Medicine. Cells seeded at a density of 6 million per 10 cm dish onto poly-D-Lysine coated plates (BD Falcon), were transiently transfected using Lipofectamine 2000 (Invitrogen Corp). Cells were routinely transfected with 0.5 μg (per 10 cm dish) of SNCA plasmid (either human wt, rat wt, or human mutant) and with 0–5 μg (per 10 cm dish) of wt CTSD plasmid. The total amount of DNA was normalized to 5.5 μg (per 10 cm dish) using empty pCMV-XL5 vector. 24 hours later, cells were washed with Tris-buffered saline and lysed in 140 mM NaCl, 50 mM Tris-HCl, pH 8.0, 1 mM EDTA, 0.5% Triton-X100, and 1× protease inhibitors, as described [37]. Lysates were centrifuged at 100,000 g for 30 min at 4°C; the top 2/3 of supernatants were removed and frozen in siliconized tubes at -80°C. The activity of LDH (lactate dehydrogenase) in the conditioned medium of transfected MES23.5 cells was routinely assayed; cellular metabolism was measured by the conversion of MTT [(3-(4,5-Dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide) to formazan, as described [38]. For these standard toxicity assays, a positive control leading to 100% cell lysis (0.1% Triton-X 100 treatment) was run in parallel.

The rabbit polyclonal 7071AP, hSA-2 and mSA-1 antibodies (Abs) were raised and affinity-purified at Open Biosystems, Inc. (http://​www.​openbiosystems.​com) against recombinant, full-length aSyn, which had been HPLC- and mass spectrometry-characterized and subjected to amino acid composition and protein concentration analyses [39]. Cell lysates were analysed by SDS-PAGE as described [40]. The monoclonal aSyn antibody, Syn-1, was purchased from BD Clontech, the monoclonal anti-Ser129Phos antibody was from Dako, and the goat polyclonal anti-CathD antibody was from Santa Cruz.

Lysates were analysed for aSyn content by a 384-well quantitative sandwich ELISA (enzyme-linked immune-adsorbant assay), as described in detail elsewhere [39, 40], using hSA-2 Ab for coating and biotinylated Syn-1 Ab as the readout Ab. Color development was carried out by using Fast-p-Nitrophenyl Phosphate (Sigma) and monitored kinetically at OD 405 nm every 5 min for up to 60 min. For serial dilutions of cell lysates on the ELISA plate, blocking buffer containing 0.5% lysate from vector-transfected cells was used as diluent; the same was used to create the corresponding standard curve of recombinant human aSyn (Additional file 1, Fig. S1A). Saturation kinetics were examined for identification of time point(s) where standards and sample dilutions were in the log phase. When analyzing cell lysates by ELISA [hSA2/Syn1-B], we recorded concentrations in MES-hSNCA cells that showed the expected parallelism after serial dilution (e.g., the 1:2000 dilution gave a value approximately twice that of the 1:4000 dilution; Additional file 1, Fig. S1B), that were SNCA cDNA dose-dependent (Additional file 1, Fig. S1C), and that permitted the precise calculation of the total aSyn protein concentration (pg per total protein in μg) expressed in living cells.

Ctsd -/- mice produced by Saftig and colleagues [41] were maintained on a mixed C57BL6J strain background in the animal facility of the Helsinki University, Biomedicum, where food and water were available ad libitum and the light/dark cycle was 12/12 hours. The study protocol was approved by the Ethical Committee of Helsinki University. Mice were decapitated at the age of 24 days and brains were collected. For immunohistochemistry, one hemisphere was immersion fixed in 4 per cent neutral buffered paraformaldehyde, dehydrated and embedded in paraffin, while the other hemisphere was frozen for biochemistry.

Archival paraffin-embedded brain specimens of sheep with congenital ovine NCL (due to a homozygous missense mutation in the CTSD gene) and of three infants with congenital NCL (due to a homozygous duplication event in CTSD and an unknown gene defect leading to absence of CathD in tissue) were analysed by immunohistochemistry, as described [30, 31]. Four μm-thin paraffin sections were dewaxed in xylene and descending alcohol series. Endogenous peroxidase activity was blocked by incubation in methanol containing 1.6% hydrogen peroxide, 30 min at RT. Sections were then rinsed 3 × with PBS and blocked with 15% normal serum, 30 min at RT, before O/N incubation at 4°C with primary antibody. Sections were washed 3 × 5 min with PBS and incubated with biotinylated secondary antibody and avidin coupled peroxidase for 30 min RT (mouse Vectastain Elite ABC kit, Vector laboratories, Peterborough, UK), and stained with aminoethyl carbazole as substrate. Sections were counterstained with hematoxylin. Immunohistochemistry of ovine brain sections was carried out as described in detail elsewhere [37, 42].

The SNCA-expressing and the CathD-deficient Drosophila lines have been described previously [9, 43]. These lines were crossed with each other to obtain flies expressing human aSyn wild-type (WT) in the absence of CathD. Ctsd genotypes were verified using a PCR-amplification of the fly ctsd locus on chromosome 2. Flies expressing either human aSyn WT in wt background and ctsd-null flies not expressing aSyn were used as controls. Flies were fixed in 4 per cent buffered paraformaldehyde, dehydrated and embedded in paraffin. Four mm-thin sections were stained with haematoxylin-eosin.

For initial examination of the central nervous system, frozen brains (without cerebellum and brain stem) from 24-day old wt and ctsd-/- mice were homogenized in 40 mM Tris-Cl, pH 7.4 containing protease inhibitor cocktail (Complete, Amersham) and centrifuged at 1,500 g for 10 min at 4°C. These supernatants were used for Western analyses and called Tris-extracts. For serial extraction experiments, frozen brains (including the brain stem and cerebellum) from 24-day old wt and ctsd-/- mice were homogenized in lysis buffer (STEN LB, 50 mM Tris-HCl, pH 7.5, 140 mM NaCl, 2 mM EDTA, protease inhibitor cocktail, and 0.4 per cent NP-40), as described [37]. Homogenates were centrifuged at 100,000 g for 45 min at 4°C to collect the NP-40 supernatant. The pellet was suspended in STEN LB containing 1% SDS, followed by centrifugation at 100,000 g for 45 min at 12°C to collect the SDS supernatant; the remaining SDS pellet was suspended in 600 μl of STEN LB containing 1% formic acid (FA) and left overnight on a shaker at 4°C.

To investigate whether CathD has any effect on aSyn turnover, we utilized a rodent cell line of mesencephalic origin, MES23.5, which constitutively synthesizes dopamine [38, 44, 45]. At baseline, MES23.5 cells show very low levels of endogenous ctsd and snca gene expression (Fig. 1A). Therefore, we co-expressed untagged wild-type (wt) human SNCA cDNA with increasing amounts of untagged, wt human CTSD in these cells. After 24 hours of co-expression, we observed by Western blotting that as the levels of the mature, 32 kDa CathD protein increased, the signal intensity of the 16 kDa, full-length (FL) aSyn monomer decreased (Fig. 1A). Of note, this decrease in aSyn monomer was not accompanied by an increase in lower molecular weight fragments of aSyn, as examined by longer exposure of Western blots with four different antibodies that had previously shown reactivity with both FL aSyn and C-terminally truncated aSyn species in vivo (Syn-1, 7071AP; hSA-2; mSA-1; Additional file 2, Fig. S2) [39, 46]. Similarly, none of these four antibodies detected any CathD-induced cellular accumulation of insoluble or higher molecular weight aSyn species. Furthermore, we confirmed by LDH and MTT assays that the observed effect of CathD on lowering of aSyn was not caused by a decrease in cellular integrity or metabolic activity. Importantly, co-expression of aSyn with plasmids encoding three other untagged, wild-type, human lysosomal enzymes (i.e., cathepsin E, cathepsin L and sulphotransferase), which were expressed under the same promoter as CTSD and analyzed as above, did not lower the intracellular aSyn concentration (data not shown).

To quantify precisely the efficiency of CathD-induced reduction of aSyn in our cell culture model, we employed a sandwich-type enzyme-linked immunoabsorbent assay (ELISA [hSA2/Syn1-B pair]), that is both specific and sensitive for the quantification of cellular aSyn concentrations (Additional file 1, Fig. S1; [39, 40]). As shown in Fig. 1B, we consistently recorded a significant, dose-dependent decrease in soluble aSyn with ectopic CathD expression, which peaked at 5 μg CTSD cDNA per 10 cm dish. Importantly, the effect of CathD on aSyn occurred in living cells, because the addition of the specific CathD inhibitor, pepstatin A, to our lysis buffer did not alter the outcome (Fig. 1B).

Heritable PD has been linked to distinct point mutations within the SNCA gene; furthermore, sustained phosphorylation of aSyn at residue Ser129 has been found in brain specimens of virtually all human synucleinopathies [14, 47]. Therefore, we next examined the effect of human CathD on these aSyn variants by expressing degradation efficiency as a percentage of the signal obtained from each relevant mutant in the absence of CathD. As shown in Fig. 1C, CathD over-expression efficiently promoted the degradation of the three known PD-linked aSyn mutants, i.e., Ala30Pro, Glu46Lys and Ala53Thr. Furthermore, CathD over-expression also efficiently degraded two other aSyn variants, which we engineered to abrogate and mimic phosphorylation at Ser129, i.e., Ser129Ala and Ser129Asp, respectively. Similarly, exogenous CathD was able to lower the levels of full-length rat aSyn, using the same paradigm (Fig. 1C).

According to Cuervo and colleagues, residues Asp98 and Gln99 represent a putative motif by which aSyn is recognized by the hsc70 chaperone complex preceding chaperone-mediated autophagy and degradation [28]. To investigate the importance of this motif for CathD in lowering aSyn ex vivo, MES23.5 cells were transfected with an aSyn variant where these two critical residues were changed to Alanine (DQ/AA-aSyn); CathD expression reduced DQ/AA aSyn as efficiently as wt aSyn in these MES23.5 cells when analyzed 24 hours after transfection (Fig. 1C).

To determine the role of CathD on endogenous aSyn metabolism in vivo, we utilized a well characterized ctsd knock-out (ctsd-/-) mouse model. These mice carry two null alleles of murine ctsd and develop a fatal NCL leading to premature death at day 26 ± 1 [33, 41]. We first examined aSyn levels in the brain of ctsd-/- mice and their wild-type (wt) littermates by SDS/PAGE and Western blotting under both non-reducing (Fig. 2A) and reducing conditions (Fig. 2B). We found that the 16 kDa aSyn monomer was slightly but consistently less abundant in Tris-extracts of ctsd-/- brain (without cerebellum and brain stem) when compared with littermate controls (n = 4 per group for non-reducing conditions; n = 16 per group for reducing conditions). In these extracts, we found no evidence of truncated aSyn species by Western blotting (Fig. 2A, B; and data not shown). In parallel, the amount of monomeric, full-length β-synuclein protein (17 kDa) appeared unchanged in ctsd-/- versus wt control mouse brains (Fig. 2C), although higher molecular weight species (and possibly truncated variants) appeared increased (Fig. 2C; note, the latter findings could not be corroborated with another Ab due to lack of availability). In addition to the somewhat unexpected reduction of the 16 kDa aSyn protein in ctsd-null brains, we observed that the mouse monoclonal aSyn Ab, Syn-1, also visualized bands that migrated at molecular weights of 25 kDa, 50 kDa (reducing conditions; Fig. 2B) and > 150 kDa positions (non-reducing, Fig. 2A). These bands were interpreted as endogenous immunoglobulin chains, as they were also detected by anti-mouse secondary Ab alone (not shown).

In order to further substantiate and expand these observations, we carried out a serial fractionation protocol of whole mouse brain (including cerebellum and brain stem) using an initial NP-40 detergent extraction step followed by a SDS extraction step of the remaining NP-40 pellets. There, we blinded the operator regarding the genotype, analyzed fractions by both immunoblotting with multiple Abs (Fig. 3A, 3B) and by sandwich ELISA (Fig. 3C), and then confirmed the genotype by probing for the presence (or absence) of CathD in all specimens (Fig. 3A). In agreement with the results obtained in Tris extracts, our serial fractionation and ELISA quantification demonstrated a reduction – rather than an increase – in the levels of soluble aSyn in ctsd-/- mouse brains (n = 8) when compared with their WT (n = 4) littermates; the reduction measured 18.6 and 20.0 per cent in NP-40 and SDS fractions, respectively (Fig. 3C). In addition, using densitometry we quantified the low molecular weight species of aSyn (migrating at 12.5 and 10 kDa positions) that were detected in NP-40 extracts by mSA-1 (Fig. 3A; Additional file 2, Fig. S2), but we found no apparent difference between the two genotypes (not shown).

To complete the serial extraction protocol, we next searched for the presence of insoluble aSyn species in the formic acid extracts of SDS pellets from the ctsd-/- mouse brains shown in Fig. 3. There, we specifically searched for HMW aggregates of aSyn (referred to as oligomers) employing our most sensitive Abs (Fig. 4A). Independent Western blotting with hSA-2 and mSA-1 (Fig. 4A) and subsequent densitometry of the ratio, aSyn to actin, revealed more insoluble, HMW oligomers in ctsd-/- than in WT mouse brain (Fig. 4B). Of note, these oligomers migrated predominantly at the interface of the stacking gel and running gel; their difference between the two genotypes was estimated to be ~15 per cent, as detected by both Abs (Fig. 4A and 4B; and data not shown). Intriguingly, these formic acid extracted aSyn oligomers evidently did not co-localize with the HMW ubiquitin positive smears, as the latter were predominantly found in SDS extracts (Fig 4C).

Our biochemical data showed a reduction of detergent soluble aSyn and an increase of formic-acid soluble aSyn species in ctsd-/- mouse brains. In human PD, DLB and MSA brain, the formic acid-soluble or urea-soluble fractions contain HMW aSyn aggregates. Therefore, we next explored whether any aSyn aggregates were also detectable in brains of 24-day old ctsd-/- mice by immunohistochemistry (Fig. 5). To avoid any cross-reactivity of the secondary anti-mouse Ab with endogenous immunoglobulins, we employed our rabbit polyclonal, affinity-purified anti-aSyn Abs (mSA-1 and hSA-2), whose binding specificity for mammalian aSyn was confirmed using snca-null tissue (Additional file 2, Fig. S2) [39, 40, 48, 49]. As shown in Fig. 5, we recorded prominent aSyn staining in the neuropil of the cerebral cortex, thalamus and cerebellum of control mice, as expected from its known presynaptic localization (Fig. 5D to 5F). In contrast, brain sections of ctsd-/- mice, processed in parallel, showed a less prominently stained neuropil, particularly in the thalamus (Fig. 5A, B). Intriguingly, in these CathD-deficient mice, we also detected aSyn-positive aggregates within select neurons of the deep cortical laminae, superior colliculi, the subiculum, as well as in the thalamus, deep cerebellar nuclei, and even within white matter tracts (Fig. 5A–C). Intra-neuronal cytoplasmic accumulations appeared large, while those within the thalamus were relatively small, thereby suggesting their localization within neurites. We concluded from these collective mouse data that: one, the distribution and processing of endogenous aSyn is altered in ctsd-/- mice; and two, murine CathD deficiency promotes an early-onset NCL phenotype that features the accumulation of a small pool of aSyn within an insoluble brain compartment.

To validate these results in a second mammal, we carried out immunohistochemical studies in a sheep colony affected by congenital NCL (Fig. 6) [31]. This phenotype is caused by a homozygous missense mutation in the ovine CTSD gene (Gln934Ala), which renders the enzyme proteolytically inactive despite a 5- to 10-fold over-abundance of the mutant protein in the forebrain [50]. In surveys of sagittal hemibrain sections with our panel of mono- and polyclonal Abs to mammalian aSyn, overall staining of the neuropil appeared relatively normal in affected sheep (Fig. 6A; n = 3 for NCL sheep; n = 1 for heterozygous sheep; n = 2 for wt control sheep). However, staining with our polyclonal Abs for ovine aSyn demonstrated widespread areas of elongated axonal swelling in deep white matter tracts in affected sheep (Fig. 6D) and revealed isolated aSyn aggregates in the thalamus, similar to the mice (not shown). As expected, no axonal abnormalities or aggregates were seen in heterozygous (Fig. 6B) and wt animals (not shown). Intriguingly, when we probed the sections with a well characterized monoclonal Ab to phosphoSer129-aSyn, we observed a small number of neurons in the cingulate cortex that showed intense intracellular labeling in homozygous CTSD mutant sheep (Fig. 6A), but not in heterozygous (Fig. 6B) and wt sheep. Of note, the axonopathy detected by the polyclonal Abs to aSyn was not matched by anti-phosphoSer129-aSyn reactivity. The changes in aSyn reactivity described above appeared specific, as staining of CathD-deficient sheep brain sections with Abs to Alzheimer's-linked phosphorylated tau, and TDP-43 did not reveal any obvious abnormalities (not shown). In contrast, development of sections with a monoclonal anti-ubiquitin Ab revealed extensive inclusion formation that was present in NCL sheep brain throughout the cortex (Fig. 6C), including the cingulate cortex, but was absent in heterozygote and wt animals.

Next, we analyzed serial sections of paraffin-embedded, post mortem human brains from three infants diagnosed with congenital NCL (Fig. 7). One of the three patients carried a homozygous duplication (c.764 dupA) in exon 6 of the CTSD gene, thereby creating a premature stop codon and loss of CathD expression; another patient was his older sibling, who died of the same disease phenotype and thus, most probably, carried the same mutation, while the third, unrelated patient with congenital NCL had no detectable CathD protein present in her tissues although she remains genetically unidentified [30]. The brains of these three children revealed extreme atrophy with massive neuronal loss throughout the cortex, accompanied by generalized glial activation [30, 51]. Using the same Ab panel employed above, we observed markedly reduced aSyn staining in the cortical neuropil of affected infants (Fig. 7B) compared to control brain (Fig. 7A), similar to the changes seen in ctsd-/- mutant mouse brain above. In addition, we identified strongly aSyn-reactive neurons scattered throughout the neocortex (Fig. 7B). These neurons carried cytoplasmic aSyn accumulations that appeared granular and were visible throughout the soma (Fig. 7B). Prominent 'synucleinopathy' was also observed in deep grey matter structures, thalamus (Fig. 7C, D) and basal ganglia of infants with congenital NCL. There, the aggregates varied greatly in size, with some appearing to localize to proximal axons (Fig. 7C), and some to the cell soma (Fig. 7D). Anti-hSA2-positive deposits could also be seen within white matter tracts, thus confirming their axonal localization. Within the cerebellum, aSyn deposits were observed in the granular cell layer and deep white matter. Of note, these inclusions were negative for Ser129-phosphorylation-specific aSyn reactivity, and for ubiquitin (not shown). As expected, no such aSyn abnormalities were seen in the control brain of a neurologically normal infant processed in parallel (Fig. 7A).

Over-expression of aSyn has been shown to be neurotoxic in a variety of cellular and animal models. In order to study the effect of endogenous CathD on aSyn-induced neurotoxicity, we generated flies expressing human aSyn in a ctsd-null background. In agreement with our previous results [9], the retina was normal in one-day-old transgenic flies over-expressing wt human aSyn (Fig. 8A), but showed mild, vacuole-carrying changes by day 30 (Fig. 8B). Flies expressing human aSyn in the absence of CathD also had normal retinas at day one (Fig. 8C), but developed pronounced retinal degeneration within 30 days; these progressive changes included the formation of numerous vacuoles, thinning of the retina and loss of its architecture (Fig. 8D). Importantly, fly CathD deficiency by itself (without aSyn expression) did not cause this neurodegeneration phenotype of the retina at day 30 (Fig. 8E; [43]). These results indicated that neurotoxicity of aSyn is enhanced in the absence of endogenous CathD expression.