Recombinant proteases
Cathepsin E (CSTE) (R&D #1294-AS), cathepsin D (CTSD) (R&D #1014-AS), cathepsin G (CTSG) (Millipore #219873), cathepsin A (CTSA) (R&D #1049-SE), cathepsin L (CSTL) (Millipore #219402), cathepsin B (CSTB) (Millipore #219364), cathepsin K (CTSK) (Millipore #219461), (CTSS) cathepsin S (R&D #1183-CY), cathepsin V (CTSV) (R&D #1080-CY), asparagine endopeptidase (AEP) (R&D #2199-CY), cathepsin H (CTSH) (R&D #7516-CY-010), cathepsin C (CTSC) (R&D #1071-CY), cathepsin O (CTSO) (Abcam #ab267932), cathepsin F (CTSF) (Abcam #ab240858), and cathepsin X (CTSX) (R&D #934-CY).
Antibodies
The following primary antibodies were used: 1) monoclonal mouse anti-GAPDH (Abcam, #ab8245, 1:2500) and 2) primary mouse anti-flag antibody (F3040; Sigma-Aldrich) at a 1:1000 dilution.
Plasmid constructs, point mutations, and stable lines
Novel doxycycline-inducible, flag-tagged, full length WT Tau, TDP-43, and α-syn lentiviral plasmid constructs were first designed and synthesized (Epoch Life Sciences) using a puromycin resistant plasmid backbone, pTet-O-Ngn2-Puro (Addgene #52047). Using these 3 WT plasmid constructs, individual constructs containing single pathogenic point mutations were also generated. Tau: K257T, N279K, P301S, and S305N. TDP-43: G298S, A315T, A321G, Q331K, and M337V. Α-syn: A30P, E46K, H50Q, G51D, and A53T. With these constructs, lentivirus was made using the psPAX2 packing (Addgene #12260) and pCMV-VSV-G envelope (Addgene #8454) plasmids. To generate stable lines, SH-SY5Y cells already containing the pLenti CMV rtTA3 Blast construct (Addgene #26429) were infected and stable lines (17 in total) were generated using Puromycin 1 mg/mL selection.
In vitro protease cleavage assays
For in vitro cleavage assays, 1 µg of recombinant full-length human 4N2R tau (rPeptide #T-1001-1), full-length TDP-43 (R&D #AP-190), or full-length α-syn (Abcam #ab51189) was incubated with or without 1µM of each protease. Proteases requiring pre-activation were performed, as mentioned in Extended Data Table 2. Depending on the pH setpoint, the following buffers were used: 100 mM sodium citrate pH 3.4, 50 mM sodium acetate pH 4.5 or 5.5 or 100 mM phosphate buffer saline (PBS) pH 7.4. 1mM EDTA and 2mM DTT were also used, and each reaction was performed over 1 hour at 37ºC. The assay was performed in a total volume of 19.5µl. Protease activity was stopped by adding 7.5µl of NuPAGE 4X LDS (Fisher Scientific #NP0007) and 3µl of 10X reducing agent (i.e., 50µM) (Fisher #NP0009). Samples were then immediately denatured for 10 minutes at 80ºC. All samples were run on precast NOVEX 4-12% Bis-Tris gels (Fisher #NP0321PK2) using MES buffer (Fisher #NP0002). The gel was then either fixed in 40% ethanol and 10% acetic acid for silver or transferred onto nitrocellulose membranes for western blotting analysis. Silver staining was performed according to manufacturer’s instructions with SilverQuest silver staining kit (Thermo Fisher #LC6070).
Multiplex substrate profiling by mass spectrometry (MSP-MS)
MSP-MS was performed as previously published by O’Donoghue et al., 201227. A substrate library was designed to cover the sequences for the three human proteins TDP-43, α-syn, and tau. The specific isoforms, their protein length and the UniProt accession numbers selected were as follows: TDP-43 isoform 1 (414 aa, Q13148), α-syn isoform 1 (140 aa, P37840), and tau 2N4R isoform Tau-F (441 aa, P10636-8). To design library peptides, overlapping fragments were chosen in a tiling approach, generally using a length of 18 amino acids and 5 amino acid overlap between fragments . In some cases, highly acidic regions such as 105 – 140 in α-syn, containing many Asp and Glu residues, would have yielded peptides bearing too much negative charge; these fragments were designed with shorter overlapping fragments of length 7 – 16 amino acids. To avoid oxidation of Cys residues that can form non-natural aggregates, all Cys sites were mutated to Ala. Finally, additional basic and spacing residues were appended to the N- and C-termini of each peptide to produce peptides suitable for liquid chromatography tandem mass spectrometric (LC-MS/MS) analysis. The final substrate library contained an equimolar mixture of 77 peptide substrates with a maximum length of 24 amino acids, with the sequences provided in Extended Data Table 1.
MSP-MS reactions were prepared with proteases at 1-70 nM concentration with a library concentration of 500 nM for each peptide (except CTSF at 1 µM) and incubated at 37 °C. Buffers were sodium acetate buffer (50 mM) containing 5 mM DTT and 1 mM EDTA for pH 4.5 and pH 5.5, or sodium citrate (50 mM), containing 5 mM DTT and 1 mM EDTA at pH 3.5. Reactions were monitored at two time points in an end point screening format. Each aliquot taken at a given time point was immediately desalted with C18 zip tips (Millipore-Sigma), and then freeze-dried. Samples were re-suspended in 0.1% formic acid in HPLC-grade water for LC-MS/MS analysis.
Reaction conditions for each protease followed manufacturer recommendations for pre-activation and then were assayed at the concentrations and pH buffer conditions mentioned in Extended Data Table 2. For CTSA, a matched CTSL-only sample was also prepared for this reaction for use as a negative control in cleavage identification. The final CTSA reaction contained 41 nM CTSA with a background of 8.25 nM CTSL, treated with E64 (440 nM), and the matched reaction contained only CTSL and E64.
Mass spectrometry
Peptide sequencing by LC-MS/MS was performed on an QExactive Plus mass spectrometer (Thermo), equipped with a nanoACQUITY (Waters) ultraperformance LC (UPLC) system and an EASY-Spray ion source (Thermo). Reversed-phase chromatography was carried out with an EASY-Spray PepMap C18 column (Thermo, ES800; 3μm bead size, 75μm by 150 mm). Chromatography was performed at a 600-nl/min flow rate during sample loading for 14 min, and then at a 400-nl/min flow rate for peptide separation over 90 min with a linear gradient of 2 to 35% (vol/vol) acetonitrile in 0.1% formic acid. Peptide fragmentation was performed by higher-energy collisional dissociation (HCD) on the six most intense precursor ions, with a minimum of 2,000 counts, with an isolation width of 2.0 m/z and a minimum normalized collision energy of 25. Data were analyzed using Protein Prospector software, v.6.2.1 (http://prospector.ucsf.edu/prospector/mshome.htm, UCSF) using published methods27. The peptide cleavage data were then output as 8-mer sequences that spanned the P4 – P4’ sites for each verified cleavage site (Supplementary Data 1).
Hierarchical clustering of protein level cleavages from MSP-MS
The percentage contribution of each enzyme to the total number of cleavages in each protein was calculated by dividing the number of cleavage sites for each enzyme with the sum of all enzyme cleavage sites. Since our substrates α-syn, TDP-43 and tau have different amino acid lengths, in order to compare the cleavage patterns, we next normalized the percentage contribution of enzyme cleavages for each substrate with the mean of total percentage contribution for each enzyme across all substrates. Using these mean-normalized values for each enzyme within each substrate, we then performed an unsupervised hierarchical clustering in Python using the clustermap function within the data visualization library called “seaborn” (https://seaborn.pydata.org/index.html).
Correlation analysis of enzyme cleavage profiles
In order to compare the cleavage profile for each enzyme within each protein substrate, we computed the correlation matrix with the raw cleavage data obtained from MSP-MS and then plotted this matrix as a heatmap using the correlation function within seaborn in Python. The diagonal correlation matrix was then plotted in the form of heatmap.
In silico analysis of MSP-MS data
The resulting peptide identifications via MSP-MS were assessed for specific cleavage in the enzyme-treated sample by subtracting the results from a no-enzyme control incubation. Both endopeptidase and exopeptidase cleavage sites were readily identified. Using these sites, we generated full sequence cleavage maps of α-syn, TDP-43 and tau. We also created IceLogos diagrams of each cathepsin to identify which amino acids were favored at the P4’ through P4 positions of each enzyme’s specific cleavage motif45. We further validated our findings using the PROSPER protease prediction algorithm server (https://prosper.erc.monash.edu.au/)8. To determine which mutations might alter protease cleavage, we also conferred with the MEROPS database (https://www.ebi.ac.uk/merops/), assessing the amino acid frequency at the P4’-P4 positions, where all known cathepsins (cathepsins A, B, D, E, F, K, L, O, S, V, X and AEP) were previously found to cleave7. Using the results of our MEROPS and PROSPER searches as well as our experimental results, we identified regions of α-syn, TDP-43 and tau where pathogenic mutations overlapped with protease cleavage sites. Using this information, we designed fluorogenic peptides to test for our activity assays.
Protease activity assays
Fluorescence-based protease activity assays were performed in triplicates in black, flat-bottom 384-well plates (Greiner #781091, Fisher Scientific) using custom designed fluorescent α-syn, TDP-43 and tau peptide substrates (Genscript, Extended Data Table 3). Assays were run at 37°C for 10 hours in 50 mM sodium citrate (pH 3.4), 50 mM sodium acetate (pH 4.5 and 5.5) or 50 mM phosphate buffer saline (pH 7.4) buffers containing 2mM DTT and 1mM EDTA. The concentration of cathepsins and fluorescence substrate in all the assays was 20 nM and 20 μM, respectively. The substrates were designed in the form of a fluorophore/quencher paired lysine conjugated with 7-methoxycoumarin-4-acetic acid (MCA) as a fluorophore on the N-terminus, and a lysine conjugated to dinitrophenol (DNP) as a quencher on the C-terminus. In addition, two arginine residues were added to all the substrates in order to increase the solubility of peptides in the buffers. When the substrate is intact, MCA is non-fluorescent due to the presence of the DNP quencher. However, when the substrate is cleaved by means of protease activity, MCA becomes unquenched and fluorescent. This MCA fluorescence, which serves a readout for substrate cleavage, was monitored as a function of time in a Tecan Infinite M Plex plate reader using excitation and emission wavelengths of 328nm and 393nm, respectively.
Raw MCA fluorescence data was normalized from 0-2000 using Microsoft Excel Version 16.49. Normalized data was then input into a kinetic model using Berkeley Madonna Version 10.2.8. to fit the enzyme kinetic data using the Rosenbrock Stiff method. Equations of the model were as follows:
d/dt (C1) = -a*C1*C1-b*C1*C
d/dt (C) = +a*C1*C1+b*C1*C
d/dt (S) = -k*C*S
d/dt (P) = k*C*S
C1 represented the proform of the cathepsin, C represented the mature cathepsin, S represented the full custom designed fluorescent α-syn, TDP-4, and tau peptide substrates, and P represented the cleaved substrate. Initial concentrations of the proform cathepsin and the substrate were 2 nM and 2000 nM, respectively to represent experimental concentrations. Parameters a & b represented cathepsin maturation. Parameter a represented the slow auto-activation mechanism by the proenzyme, while parameter b represented the faster rate of activation that occurs by the mature cathepsin. Initial estimates for parameter a were 0.05 and 0.15, and 0.5 and estimate for parameter b was 1.5. Parameter k represents substrate cleavage. Initial estimates for parameter k were 5.0e-4 and 0.0015. All parameters were constrained to be positive values. Normalized MCA fluorescence data was fit to variable P.
Cell-based protein half-life experiments
SH-SY5Y human neuroblastoma cells were obtained from ATCC (CRL-2266) and maintained 1:1 EMEM/F12 media (ATCC #30-2003/Thermo #11765062) supplemented with 10% FBS and 1% Penicillin-Streptomycin (Thermo Fisher #15140122). Stable lines of all 17 aformentioned constructs were generated by lentiviral infection followed by both blasticidin (5 µg/mL) and puromycin (1mg/mL) selection. To terminally differentiate into neuron-like cells, SH-SY5Y cells were treated as published34. Once differentiated, SH-SY5Y cells were treated with 1mg/mL of doxycycline to induce expression of Flag-tagged α-syn, TDP-43 or tau for 24 hours. On day 2, the media was refreshed without doxycycline and cells were harvested at 0 , 96 or 192 hrs. For inhibitor studies, differentiated SH-SY5Y cells had their media refreshed and supplemented with either 10µM PBS (control), 100nm MG132 (EMD Millipore #474791), 10µM Pepstatin A (Sigma #P5318), 10µM E64D (Tocris #4545), or 20nM Bafilomycin A1 (Tocris #1334).
Western blotting was subsequently performed as previously published34. 40µg of protein were separated on 4–12% SDS-PAGE (Thermo) and then transferred onto nitrocellulose membranes (Bio-Rad). Membranes were blocked at room temperature with Odyssey Blocking Buffer (LI-COR 927-40100) and incubated at 4 °C overnight with primary antibodies and 1 hour at room temperature with appropriate fluorescent secondary antibodies (1:5000) (LI-COR). Immunoreactive bands were visualized using a LI-COR Odyssey CLx image scanner and quantified using ImageJ software.
Statistical analysis
Details of the statistical test used for each experiment is in figure legends along with n and p value. All data is represented as mean ± SD. Statistical analysis was performed using GraphPad Prism 9 (GraphPad Software, La Jolla, California USA). Western blot quantifications for Figures 6B, D, and E as well as Extended Data Figure 4 were analyzed using 2-way, repeated measures ANOVA analyses. Pearson’s correlational coefficient testing was used for the pairwise correlational analyses for Figure 5C-E. A p-value < 0.05 was considered significant. *p<0.05, **p<0.01, or ***p <0.001.