Lysosomal enzyme cathepsin D protects against alpha-synuclein aggregation and toxicity

To investigate the involvement of lysosomal functions in α-syn clearance, we analyzed mice deficient in CD, previously generated by a targeted insertion of the neo marker in exon 4[27]. CD-deficient (Ctsd-/-) mice die at approximately postnatal day 26 (p26) due to a combination of nervous system and systemic abnormalities. Extensive neuron death resulting from activation of both apoptotic and non-apoptotic pathways has been observed in these mice [27‐31]. We examined brains from p21 and p25 Ctsd-/- mice and found significant α-syn accumulation in neuronal cell bodies in p25 Ctsd-/- but not wildtype cortex (Fig. 1a). In contrast to the brains of human lipidoses patients[32] where α-syn aggregates are found in both neurons and glia and co-localize with lipids, in Ctsd-/- brains α-syn accumulations do not co-localize with autofluorescent lipofuscin (data not shown).

Furthermore, α-syn accumulations in Ctsd-/- brains were present in cells co-expressing the neuron marker, neuronal nuclear antigen (NeuN), but not in glial fibrillary acidic protein (GFAP)-immunoreactive astrocytes (Fig. 1b–c). Accumulation of ubiquitinated proteins also occurs in Ctsd-/- cortex (Fig. 1a). A relatively small fraction of neurons (~5%) with intense accumulated α-syn immunostaining also exhibit co-localization with intense accumulated ubiquitin staining, consistent with the observation that a small fraction of α-syn in LB is ubiquitinated [33‐36]. Correspondingly, we found elevated levels of high molecular weight but not monomeric α-syn, and high molecular weight ubiquitinated proteins in both TritonX-100 soluble and insoluble extracts from the cortex of Ctsd-/- mice by western blot analyses compared to Ctsd+/+ mice, similar to what occurs in LB diseases (Fig. 1d–e). Truncated 12 kDa and 10 kDa α-syn fragments are reduced in Ctsd-/- extracts (Fig. 1d–e).

Antibody pulldown experiments indicate that a small fraction of accumulated α-syn in Ctsd-/- brains is mono-ubiquitinated (Fig. 1f). These results are consistent with a small fraction of accumulated α-syn in Ctsd-/- brains being mono-ubiquitinated [33‐36]. Ubiquitin immunostaining is present in nearly all neurons with variable intensity. 5% of those neurons with intense accumulated ubiquitin staining also exhibit intense accumulated α-syn immunostaining in Ctsd-/- brain sections. Conversely, 5% of neurons with intense accumulated α-syn immunostaining also exhibit intense accumulated ubiquitin staining. p21 Ctsd+/+ and Ctsd-/- brains do not exhibit considerable difference in α-syn immunostaining intensity or protein level by western blot analyses. However, ubiquitin accumulation occurs as early as p21 in Ctsd-/- cortical extracts compared to wildtype cortical extracts (Fig. 1g).

Immunostaining with synphilin antibodies have not detected "aggregation-like" structures in the cell body where α-syn is. Nor did we observe increased synphilin in western blot analyses. Thus the α-syn accumulation is unlikely to be induced by synphilin accumulation in Ctsd-/- brains. The cytoplasmic microtubule-associated protein, tau, did not accumulate in Ctsd-/- cortex at p25 compared to wildtype cortex at p25 (data not shown), suggesting that CD deficiency does not have a general effect on the accumulation of all cytoplasmic proteins.

In human LB, 95% of the insoluble α-syn is phosphorylated [33‐35, 37, 38]. We found that anti-phospho-α-syn antibody (from Dr. Iwatsubo [33, 36, 38]) did not recognize the intensely immunoreactive accumulated α-syn in Ctsd-/- brains, although numerous neurons in both wildtype and Ctsd-/- brains exhibited labeling with this antibody. Whether this antibody lacks specificity when used in mouse brain sections or the accumulated α-syn in Ctsd-/- brain is simply not phosphorylated cannot be determined from this immunohistochemical study. Western blot analyses indicate that a few immunoreactive bands are increased in Ctsd-/- extracts compared to wildtype extracts.

Prior studies found autophagosomes start to accumulate in Ctsd-/- brains as early as p1, compared to Ctsd+/+ age-matched controls [28, 30]. Furthermore, CB immunostaining as well as enzymatic activities are increased as early as p21 in Ctsd-/- brains [28, 30]. Electron microscopy studies demonstrated that CB was associated with irregularly shaped and membrane-bound structures containing electron-dense materials, characterizing them as lysosomes in Ctsd-/- brains [28, 30]. We found that α-syn accumulation does not become prominent until near p25 in Ctsd-/- brains. α-syn accumulation occurs in only a sub-population of neurons that exhibit enhanced Atg8 (AuTophaGy8)/LC3 (light chain 3) staining, indicating that AV accumulation precedes α-syn accumulation (Fig. 2a). We also found that α-syn accumulations are adjacent to, but do not overlap with, immunoreactivity for the AV-associated protein Atg8/LC3 or the lysosomal associated protein CB, suggesting that α-syn accumulation formed outside of autophagosomes and lysosomes (Fig. 2b). Neuronal populations immunoreactive for the apoptotic marker, cleaved caspase-3, are distinct from those with intense α-syn accumulation (Fig. 2c).

While bulk protein degradation appears to be normal in Ctsd-/- mice[27], we found that α-syn mRNA is down-regulated in Ctsd-/- brains at p25 when α-syn accumulation occurs (Fig. 3a). This is consistent with the finding that α-syn mRNA is either unchanged or down-regulated in the majority of sporadic PD cases[39], indicating that α-syn accumulation is unlikely to be the consequence of elevated gene expression. Prior studies reported that CB but not CL protein is up-regulated in Ctsd-/- brains at p23[28]. To better understand the role of CD in selective protein degradation and PD, we analyzed the expression of genes encoding other brain-enriched lysosomal proteases, autophagy-associated factors, proteasome subunits, and genes linked to familial PD. Interestingly, genes encoding other lysosomal cathepsins B, L, F and H (Ctsb, Ctsl, Ctsf, and Ctsh) mRNAs are all up-regulated at p25 (Fig. 3a). This result may suggest a common transcription regulatory mechanism for these cathepsins in response to CD deficiency. Alternatively, the influx of macrophages or microglia into the Ctsd-/- brain at this age may lead to an increase in cathepsin mRNA expression[40]. We also determined that accumulation of autophagosomes in Ctsd-/- neurons is accompanied with transcriptional up-regulation of Atg7 but not Atg12 (Fig. 3a). Both Atg7 and Atg12 are involved in an ubiquitin-like activity important for autophagosome expansion. Up-regulation of Atg7 may indicate an increase of autophagosome production in addition to a blockade of autophagy completion [28, 38, 40]. mRNA of Park2 which encodes Parkin, mutation of which has been found in a subset of autosomal recessive PD;UCHL1, mutation of which has been found in a subset of autosomal dominant PD; and Psmb7 which encodes proteasome 20S core β2 subunit are also modestly up-regulated in response to CD deficiency, indicating a compensatory response to CD deficiency at the level of gene transcription (Fig. 3a).

In addition to deficient macroautophagy, we found an accumulation of GAPDH, a substrate of CMA (Fig. 3b). Interestingly, we also found reduced proteasome activity in Ctsd-/- brain extracts (Fig. 3c), suggesting a functional interaction between the two major α-syn clearance machineries, lysosomes and proteasomes. Accumulation of ubiquitinated proteins in Ctsd-/- brain appears at p21 compared to that in wildtype brains (Fig. 1g, 3e), when proteasome activities are unaltered as determined by the proteasome activity assay (data not shown). So far, none of the proteasome-related proteins we examined, including UCHL1, a gene mutated in familial PD and a ubiquitin hydrolase and E3 ligase; Usp14, a key deubiquitination enzyme; Rpt3, an ATPase regulatory subunit, a4 subunit that is important for the gating into the 20S core particle, and b1 subunit that is part of the proteasome core, were significantly changed as determined by western blot analyses in Ctsd-/- brains (Fig. 3d).

To further understand how CD activity influences α-syn homeostasis, we examined whether enhancing CD expression can reduce α-syn aggregation. We adopted the simple culture system developed by McLean and colleagues in which an α-syn-green fluorescent protein (GFP) fusion protein (α-syn-GFP) forms visible aggregates in cells when co-expressed with synphilin[41]. We transfected H4 neuroglioma cells with α-syn-GFP, synphilin, and CD. As a control, cells were transfected with α-syn-GFP, synphilin and empty vector pcDNA3.1. Approximately 50% of control transfected cells exhibited α-syn aggregates, consistent with published findings[41]. Remarkably, transfection of CD together with α-syn-GFP and synphilin led to less than 20% of transfected cells exhibiting α-syn aggregates (Fig. 4a). To address the issue of whether CD could affect synphilin levels, we performed immunocytochemistry experiments to compare α-syn+synphilin+vector transfected cells and α-syn+synphilin+CD transfected cells. We chose this approach because very low transfection efficiency was achievable (less than 5%) in this aggregation assay. Thus, we cannot perform western blot analyses or quantitative RT-PCR assays for this experiment. Double immunostaining for CD and the V5 epitope tag on synphilin and double immunostaining for CD and α-syn were performed (Figs. 4b and 4c). We quantified the number of transfected cells with α-syn aggregates and those without α-syn aggregates. Many more cells that were transfected with synphilin+α-syn+vector exhibited α-syn aggregates in comparison to cells transfected with synphilin+α-syn+CD. Nonetheless, the levels of synphilin immunostaining were indistinguishable in cells transfected with synphilin+α-syn+vector versus cells transfected with synphilin+α-syn+CD. This is true in cells with or without aggregates. We conclude that CD reduction of α-syn aggregation is unlikely to be secondary to a reduction in synphilin levels.

Excessive α-syn induces neuron death in cell cultures, and in a variety of genetic and viral delivery based animal models [42‐45]. To examine the potential of elevating CD level as a means to reduce α-syn-induced cell death, we transfected human neuroblastoma SHSY5Y cells with α-syn-GFP, in the absence or presence of increased CD expression (Fig. 5a). Similar to previous studies of α-syn overexpression in yeast, worms and rat neurons[44], we found that overexpression of α-syn-GFP induced significant cell death in SHSY5Y cells. Co-transfection of human CD provided significant protection against α-syn-GFP overexpression-induced cell death (Fig. 5a). Quantitative real-time RT-PCR confirmed the up-regulation of α-syn mRNA by α-syn transfection, and lack of effect of CD transfection on α-syn mRNA level (Fig. 5b). Furthermore, co-expression of CD with α-syn-GFP in human neuroblastoma SHSY5Y cells produced a cleavage product of α-syn-GFP and reduced endogenous monomeric 17 kDa α-syn (Fig. 5c).

α-syn is rich in hydrophobic amino acids (52%) and is natively unfolded. CD has a known specificity in recognizing hydrophobic residues[11]. Although α-syn contains many putative CD cleavage sites, the main cleavage occurs at Y125[11]. We found that CD is also protective against PD-causing mutant α-syn-GFP induced-cell death in SHSY5Y cells (Fig. 5d). In contrast, mutating α-syn-GFP at the putative CD cleavage site Y125[11] results in an α-syn mutant that induces cell death that resists neuroprotection by elevated CD (Fig. 5d). Furthermore, CD is ineffective at attenuating chloroquine- or staurosporine-induced cell death (Fig. 5d). Western blot analyses confirmed the up-regulation of transfected genes (Fig. 5e).

To further investigate the role of CD activity in α-syn clearance in vivo, we generated transgenic C. elegans expressing a human α-syn and GFP fusion protein in body wall muscle cells. In these worms, human α-syn::GFP forms aggregates as worms develop and age (Fig. 6a). As in mammalian cells[46], co-expression of the worm TOR-2 protein chaperone ameliorated the formation of α-syn::GFP aggregates (Fig. 6b). Importantly, this established a genetic background within which enhancement of α-syn aggregation could be more readily visualized by RNA interference (RNAi). Using bacterial RNAi feeding to specifically target the C. elegans ortholog of Ctsd, we knocked down Ctsd in α-syn::GFP + TOR-2 transgenic worms. RNAi targeting of Ctsd led to a return of fluorescent aggregates over time (Fig. 6c), and RNAi of Ctsd did not affect tor-2 mRNA levels as examined by real-time RT-PCR (Fig. 6f). Thus, Ctsd deficiency led to α-syn aggregation in both mice and worms.

We further investigated whether human CD attenuates the loss of dopaminergic neurons in a C. elegans model of α-syn-induced neurodegeneration[44]. Overexpression of α-syn led to DA neuron death, as evidenced by the finding that only 16% of 7 d old adult α-syn expressing worms (n = 270 worms analyzed) displayed normal numbers of DA neurons (Fig. 6d, g). In contrast, co-overexpression of human CD significantly protected against DA neurodegeneration, since 30% of same-staged animals (n = 270) exhibited wildtype DA neurons (Fig. 6e, g; p < 0.001, Fisher Exact Test). Overexpression of enzymatic mutants of CD (D295R and F229I) [16, 47], or related human cathepsin gene products, CB or CL, did not attenuate DA neuron death in this in vivo assay (Fig. 6g), thereby suggesting a specific role of CD in neuroprotection against α-syn-induced DA cell death, as well as an essential role of CD enzymatic activity in this neuroprotection.

We genotyped littermates from Ctsd+/- breeding. We used Ctsd+/+, Ctsd+/- and Ctsd-/- littermates on C57BL6 background for all experiments.

Brains were placed in Bouin's fixative overnight at 4°C followed by paraffin embedding. 5 μm thick sections were used for immunohistochemical studies. The following antibodies were used: mouse anti-NeuN (Chemicon), mouse anti-α-syn (BD Transduction Lab), sheep anti-α-syn (Chemicon), goat anti-CD (Santa Cruz), mouse anti-GFAP (Chemicon), rabbit anti-Ub (Dako), mouse anti-Ub (FK2, Biomol), mouse anti-Ub (Chemicon, clone Ubi-1), mouse anti-Ub (Zymed), mouse anti-synaptophysin (Chemicon), goat anti-cathepsin B (Santa Cruz), rabbit anti-active caspase 3 (Chemicon), and rabbit-anti-GAPDH (Cell Signaling). Horseradish peroxidase conjugated donkey derived secondary antibodies were used at 1:2000 (Jackson ImmunoResearch). The sections were then incubated with tyramide signal amplification (TSA) plus detection solution with Cy3 or fluorescein tyramide according to the manufacturer's instruction (PerkinElmer Life Sciences), followed by bisbenzimide staining of DNA. For chromogenic immunohistochemistry, we incubated the sections with a biotinylated secondary antibody (Vector Laboratories) followed by ABC reagent (ABC kit, Vector Laboratories). We visualized the immunoreaction by treating the sections in 0.05% diaminobenzidine (DAB) with hydrogen peroxide. The images were taken using a Leica TCS SP5 confocal microscope or a Zeiss Axiocam CCD camera on a 100 W Axioscope bright field and fluorescence microscope.

We used the in vitro system developed by McLean and colleagues in which an α-syn-green fluorescent protein (GFP) fusion protein (α-syn-GFP) becomes truncated at the C-terminus, cleaving off GFP, to form visible aggregates in cells when co-expressed with synphilin[41]. We transfected H4 neuroglioma cells with α-syn-GFP and synphilin, and either the empty vector pcDNA3.1 or CD. 24 h after transfection, cells were fixed and stained with a mouse monoclonal antibody against α-syn (1:1000; BD transduction), and a secondary Alexa 488-conjugated goat anti-mouse antibody (1:500; Jackson ImmunoResearch). An observer blind to the transfection conditions scored neurons as positive or negative for α-syn aggregates visible with a 20X objective under a fluorescent microscope. Three independent experiments were carried out with 4 replicates per experiment. Student t-test was used to compare transfection with empty vector versus transfection with CD.

The human neuroblastoma SHSY5Y cells were transfected in triplicate by vector alone, a-syn-GFP, pCMV-CD, or co-transfected by a-syn-GFP (or A53T, A30P, Y125A mutant a-syn) and pCMV-CD (or mutant CD) by Amaxa method as described by the vendor. Transfection efficiency was 80% as assessed by cells with or without GFP. 72 h after transfection, cells were harvested. For chloroquine (10 mM) treatment, the chemicals were added 48 h after transfection, cells were harvested 42 h later. Live cells were counted by trypan blue exclusion. Relative cell survival was calculated as number of live cells after transfection by pCMV-CD and/or a-syn-GFP divided by live cells after transfection by vector alone. Elevated expression of a-syn-GFP or cathepsins was confirmed by western blot analyses using whole cell extracts.

We homogenized wildtype and Ctsd-/- cortex (n ≥ 3 each genotype) in 10 volumes of ice-cold lyses buffer (50 mM Tris-HCl pH 7.4, 175 mM NaCl, 5 mM EDTA), sonicated for 10 sec, add TritonX-100 to 1% and incubated for 30 min on ice. We then centrifuged homogenates at 15,000 g for 15 min at 4°C to separate supernatants (fractions soluble in 1% TritonX-100) and pellets (TritonX-100-insoluble fractions)[43]. Pellets were resuspended in lyses buffer containing 2% SDS. Western blotting for each sample was done at least twice. The antibodies used were described in Immunohistochemistry.

Total brain RNA was isolated from p25 mice using RNA-STAT60 (Tel-Test, Friendswood, TX). Total RNA (2 μg) was then reverse transcribed using Applied Biosystems GeneAmp Gold RNA PCR Reagent Kit (Foster City, CA). Real-time PCR reactions were setup in duplicate using TaqMan gene assays and amplified in an Applied Biosystems Step-One instrument. ΔΔCCT curves were generated using 18S TaqMan gene assays as internal standards. Quantitative PCR results are shown as standard deviation of from 3 different amplifications from RNA reverse transcribed from 3 different animals. Individual gene assay kits were purchased from Applied Biosystems for each of the RNAs analyzed. Paired t-tests were conducted on RQ (relative quantity) values for each group to determine significance.

We analyzed the proteasome activities using the TritonX-100-soluble fractions. The assay buffer consists of 50 mM Tris (pH7.5), 2.5 mM EGTA, 20% glycerol, 1 mM DTT, 0.05% NP-40, 50 μM substrate. Lactacystin was used at a final concentration of 10 μM to block proteasome activities as negative controls. Fluorescence was read at 5 min intervals for 2 h, at an excitation wavelength of 380 nm and an emission wavelength of 460 nM. Assays were done in triplicate, each using n ≥ 3 mice per genotype.

Nematodes were maintained following standard procedures[69]. Worms expressing α-syn alone UA49 [baInl2; P_unc-54::α-syn::gfp, rol-6 (su1006)] or with tor-2 [UA50; baInl3; P _unc-54 ::α-syn::gfp, P _unc-54 ::tor-2, rol-6 (su1006)] were created, integrated into the genome to generate an isogenic line, and out-crossed four times. We used the worm line that overexpresses TOR-2 protein (a worm homolog of human torsinA) and a-syn fused to GFP in the body wall muscle cells because these cells are much larger than neurons for detecting a-syn aggregation. Moreover, C. elegans dopaminergic neurons have been shown to be refractory to RNAi. Using this isogenic line, we knocked down the worm Ctsd ortholog by RNAi, and scored for the return of a-syn aggregates over the course of development and aging.

RNAi was performed by bacterial feeding as described[70] with the following modification. A Ctsd-specific RNAi feeding clone targeting a distinct portion of the C. elegans open reading frame [R12H7.2 (asp-4); e-value = 1.8e-108] with highest homology to human Ctsd (Geneservice) was grown for 14 h in LB broth with 100 μg/ml ampicillin and seeded onto NGM agar plates containing 1 mM isopropyl β-D-thiogalactoside. After 4 h incubation at 25°C to dry the plates, five gravid adults were then placed onto the corresponding RNAi plates and allowed to lay eggs for 9 h; the resulting age-synchronized worms were analyzed at the indicated stage. RNAi knockdown was performed in duplicate sets of animals and enhancement α-syn misfolding was scored as positive if at least 80% of worms displayed an increased quantity and size of α-syn::GFP aggregates. For each trial, 20 worms were transferred onto a 2% agarose pad, immobilized with 2 mM levamisole, and analyzed using Nikon Eclipse E800 epifluorescence microscope equipped with Endow GFP HYQ filter cube (Chroma Technology). Images were captured with a Cool Snap CCD camera (Photometrics) driven by MetaMorph software (Universal Imaging).

The procedure for total RNA isolation, cDNA preparation, and semi-quantitative RT-PCR was described previously[71]. The following primers were used for the PCR: cdk-5 Primer 1: 5' ggg-gat-gat-gag-ggt-gtt-cca-agc 3'; Primer 2: 5' ggc-gac-cgg-cat-ttg-aga-tct-ctg-c 3'. tor-2 Primer 1: 5' caa-tta-tca-tgc-gtt-ata-caa-ag 3'; Primer 2: 5' cat-tcc-act-tcg-ata-agt-att-g 3'. cdk-5 Primer 1: 5' ggg-gat-gat-gag-ggt-gtt-cca-agc 3'; Primer 2: 5' ggc-gac-cgg-cat-ttg-aga-tct-ctg-c 3'. tor-2 Primer 1: 5' caa-tta-tca-tgc-gtt-ata-caa-ag 3'; Primer 2: 5' cat-tcc-act-tcg-ata-agt-att-g 3'.

For the DA neurodegeneration analysis, strain UA54 [baEx45; P _dat-1 ::α-syn, P _dat-1 ::gfp, P _dat-1 ::CD,rol-6 (su1006)], UA90 [baEx69; P _dat-1 ::α-syn, P _dat-1 ::gfp, P _dat-1 ::CD D295R,rol-6 (su1006)], UA91 [baEx70; P _dat-1 ::α-syn, P _dat-1 ::gfp, P _dat-1 ::CD F229I,rol-6 (su1006)], UA53 [baEx44; P _dat-1 ::α-syn, P _dat-1 ::gfp, P _dat-1 ::CB,rol-6 (su1006)], and UA55 [baEx46; P _dat-1 ::α-syn, P _dat-1 ::gfp, P _dat-1 ::CL,rol-6 (su1006)] were generated by injecting 50 μg/ml of expression plasmid containing the human cathepsin cDNA and 50 μg/ml of rol-6 into an integrated line of UA44 [baInl1; P _dat-1 ::α-syn, P _dat-1 ::gfp[72]]. Three stable lines were randomly selected for neurodegeneration analysis. The 6 anterior DA neurons (4 CEP and 2 ADE neurons) of 30 animals/trial were examined for neurodegeneration when the animals were 7 days old. 90 animals from each of three CD (or CD D295R, CD F229I, CB, and CL) transgenic lines were analyzed (3 lines × 3 trials of 30 animals/trial = 270 total animals scored). Worms displaying at least one degenerative change (dendrite, axon, or cell body loss) were scored as exhibiting degenerating neurons as previously reported [44;72].