Impact of DJ-1 and Helix 8 on the Proteome and Degradome of Neuron-Like Cells

DJ-1 is an abundant and ubiquitous component of cellular proteomes. DJ-1 supposedly exerts a wide variety of molecular functions, ranging from enzymatic activities as a deglycase, protease, and esterase to chaperone functions. However, a consensus perspective on its molecular function in the cellular context has not yet been reached. Structurally, the C-terminal helix 8 of DJ-1 has been proposed to constitute a propeptide whose proteolytic removal transforms a DJ-1 zymogen to an active hydrolase with potential proteolytic activity. To better understand the cell-contextual functionality of DJ-1 and the role of helix 8, we employed post-mitotically differentiated, neuron-like SH-SY5Y neuroblastoma cells with stable over-expression of full length DJ-1 or DJ-1 lacking helix 8 (ΔH8), either with a native catalytically active site (C106) or an inactive site (C106A active site mutation). Global proteome comparison of cells over-expressing DJ-1 ΔH8 with native or mutated active site cysteine indicated a strong impact on mitochondrial biology. N-terminomic profiling however did not highlight direct protease substrate candidates for DJ-1 ΔH8, but linked DJ-1 to elevated levels of activated lysosomal proteases, albeit presumably in an indirect manner. Finally, we show that DJ-1 ΔH8 loses the deglycation activity of full length DJ-1. Our study further establishes DJ-1 as deglycation enzyme. Helix 8 is essential for the deglycation activity but dispensable for the impact on lysosomal and mitochondrial biology; further illustrating the pleiotropic nature of DJ-1.


Vectors and Cell Transduction
Human DJ-1 (Ensembl: ENSG00000116288, MIM:602533) I.M.A.G.E. cDNA clone IRATp970A044D was used for site directed mutagenesis and generation of the following four different DJ-1 constructs: full length DJ-1 with wild-type C106; full length DJ-1 with active site mutated C106A; DJ-1 lacking helix 8 (C-terminal 15 residues) with wild-type C106; DJ-1 lacking helix 8 (C-terminal 15 residues) with active site mutated C106A. DJ-1 variants were cloned into a bicistronic pMIG expression vector containing an internal ribosomal entry site (IRES) and GFP allowing stoichiometric expression of untagged DJ-1 variants. A three plasmid system was used for the generation of high titer retroviral particles for SH-SY5Y transduction [23]. Successfully transduced cells were selected with 800 µg/ml G418 for three weeks. Subpopulations of each new cell line expressing the four different DJ-1 variants or harboring the empty vector were selected by GFP-based fluorescence-assisted cell sorting (FACS) using a BD Biosciences FACS Aria flow cytometer. To confirm successful genomic integration of the respective DJ-1 constructs, genomic DNA (gDNA) of the established cell lines was isolated using a gDNA extraction kit (Qiagen), DJ-1 gDNA was amplified by PCR with a forward primer binding the vector backbone after the 5 LTTR: TACACCCTAAGCCTCCGCCT and a reverse primer binding in the DJ-1 sequence: AGGCCCCCGGCTTGTAAGA and sequenced with the sequencing primer: CCCTTGAACCTCCTCGTTCGACC.

Cell Culture and Differentiation
SH-SY5Y cells were purchased from LGS standards. Cells were grown in standard Dulbecco's Modified Eagle Medium DMEM/F12, Gibco, Thermo Fischer) supplemented with 10% fetal calf serum, 1% L-glutamine and 1% penicillin/streptomycin on standard plastic cell culture dishes in a sterile incubator (37 • C, 5% CO2). For differentiation a previously published protocol for the generation of a homogenous population of fully differentiated, neurotrophic factor-dependent human neuron-like cells [24] was used with minor modifications: SH-SY5Y cells were seeded at an initial density of 10ˆ4 cells/cm 2 on Advanced cell culture dishes (Greiner). On the following three days 10 µM all-trans retinoic acid (RA) was added in standard medium every 24 h. After three days in the presence of RA, the cells were washed twice with DPBS and grown in serum-free medium supplemented with 50 ng/ml Brain Derived Neurotrophic Factor (BDNF) for four days.

Cell Proliferation Measurement
Real-time cell proliferation was assessed using the xCelligence ® System (Roche, Basel Switzerland). Two thousand cells per well were seeded in E-plates and differentiated as described above. Proliferation was measured in 1 h intervals during the course of differentiation. A BrdU ELISA (Roche Basel Switzerland) was used to determine proliferation after 7 days of differentiation according to the manufacturer's protocol. Briefly, 1 × 10 4 cells per well were seeded in a 96 well plate and differentiated as described above. At day 6 of the differentiation protocol BrdU was added for 24 h before fixation of the cells and immuno-detection and quantification of BrdU incorporation after 30 min incubation with anti-BrdU-POD solution and 30 min incubation with substrate solution. The absorbance of the samples was measured at 370 nm (reference wavelength: 492 nm).

Sample Preparation for Quantitative Proteomics
Sample preparation for liquid chromatography-tandem mass spectrometry (LC-MS/MS)based proteome comparison was essentially performed as described previously [25] including denaturation and alkylation, trypsin digestion, stable isotopic labeling and prefractionation via strong cation exchange chromatography (SCX, replicates 1 and 2) or high pH reversed phase chromatography (hpH-RP, replicate 3).
Differentiated SH-SY5Y DJ-1 WT ∆H8 and SH-SY5Y DJ-1 C106A ∆H8 cells were washed three times with phosphate buffered saline (PBS), harvested with a cell scraper, and lysed on ice in lysis buffer (50 mM Tris pH 7.5, 150 mM NaCl, 1% NP-40, 1% sodium deoxycholate, 0.02% SDS, 1 mM EDTA) with protease inhibitors (10 µM E64d, 1 mM PMSF and 5 mM EDTA) added fresh to avoid protein degradation during lysis. Lysates were heat incubated for five min at 95 • C prior to ultra-sonication for 10 × 30 s. For protein precipitation, cell lysates were incubated with nine volumes of ice-cold acetone and one volume of ethanol for one hour at −80 • C. After centrifugation at 4500 g, protein pellets were washed three times with ice-cold ethanol and reconstituted in 100 mM NaOH using ultra-sonication, followed by pH adjustment to 8.0 with HEPES. For the proteome comparison, samples were labeled by reductive dimethylation of primary amines either "light" (formaldehyde CD2O + sodiumcyanoborohydrid NaBH3CN) or "heavy" (formaldehyde 13CD2O + sodium cyanoborodeuteride NaBD3CN). The label was switched in the second experiment. Samples were pre-fractionated on C18 Hypersep columns using 15%, 25% and 60% acetonitrile in 1% ammonium hydroxide as elution buffers respectively.

Sample Preparation for N-Terminomics
The Terminal Amine Isotopic Labeling of Substrates (TAILS) protocol was modified from [26,27]. Cells were lysed using 5 mM TCEP, 0.1% RapiGest, 100 mM HEPES pH 8.0 and samples were immediately heat denatured for 10 min at 95 • C. Subsequently, DNA was sheared using ultra-sonication for 10 × 30 s. Iodoacetamide (20 mM) was added for alkylation of cysteines and incubated for 30 min at room temperature in the dark. αand ε-amines were labeled for 18 h at room temperature by adding tandem mass tag (TMT) 11 plex reagents (Thermo), which were resuspended in acetonitrile with a protein to TMT reagents mass ratio of 1:8. Following pooling of samples, sequencing grade trypsin (Worthington) was added for digestion in a trypsin to protein mass ratio of 1:50 and the sample was incubated for 2 h at 50 • C. The same amount of trypsin was added again and the sample was incubated for 18 h at 37 • C. Internal peptides were then coupled to an aldehyde-functionalized polymer in the presence of 50 mM NaBH3CN and removed from TMT-blocked N-terminal peptides by ultrafiltration through a 10 kDa MWCO filter (Millipore, Burlington, MA, USA). The flow-through, containing TMT-blocked N-terminal peptides, was desalted using C18 columns (Hypersep, Thermo, Waltham, MA, USA). The sample was pre-fractionated on C18 Hypersep columns using 15%, 25% and 60% acetonitrile in 1% ammonium hydroxide as elution buffers, respectively. For exact ratio determination of Cathepsin B propeptide cleavage in four independent TAILS pilot experiments was analyzed, which were performed as described previously [28].

LC-MS/MS
LC-MS/MS of the global proteome comparison was performed with a Q Exactive PLUS System (Thermo Fisher) coupled to an Easy nLC 1000 (Thermo Fisher) as described previously [29].
LC-MS/MS of the TMT TAILS experiment was measured on an Orbitrap Eclipse mass spectrometer (Thermo Fisher) coupled to an Easy nLC 1200 (Thermo Fisher). Pre-columns with 100 µm ID were self-packed with 3µm C18 AQ (Dr. Maisch) to a length of 2 cm. A 75 µm Picofrit column (New Objective) was self-packed with 1.9 µm C18 AQ (Dr. Maisch) to a length of 20 cm. Buffer A consisted of 0.1% formic acid, buffer B consisted of 80% acetonitrile in 0.1% formic acid. The samples were separated using a 70 min linear gradient from 10% to 38% B followed by a 5 min linear gradient from 38% to 47% buffer B. The mass spectrometer was operated in data dependent acquisition mode with a TMT MS2 quantitation method. A survey scan from 400-1600 m/z at 120 K resolution was followed by MS2 events up to 2 s. Standard precursor filter options from the TMT MS2 method editor node were used. For MS2 scans, peptides were fragmented using higher energy collision dissociation (HCD) with CE 38, maximum injection time 54 msec at 30 K resolution with TMT and TMTpro resolution enhancement activated.

Proteomic Data Analysis
For proteome comparisons, raw files were analyzed using the default settings of MaxQuant [30] Version 1.6.0.16 except for the following parameter changes: The labels DimethylLys4, DimethylNterm4 and DimethylLys8, DimethylNterm8 were set. iBAQ and no variable modifications were chosen and protein quantification was allowed for label min ratio count 1. The data were compared to a complete human reviewed database without isoforms downloaded from UniProt in June 2017 containing 20188 entries. For proteome comparisons the parameters: enzyme Trypsin, digestion mode specific (0 missed cleavages) and "re-quantify" were chosen.
For the TMT TAILS analysis MaxQuant Version 1.6.12.0 was used. Reporter ion MS2 with TMT11 plex and semi-tryptic enzyme specificity (ArgC) were chosen. TAILS data were analyzed by filtering semi-tryptic peptides, more specifically, no lysine or arginine prior to the peptide sequence were allowed and initiator methionines were also excluded. Subsequently, data were z-score normalized and linear models for microarray data (LIMMA package from R) was used to detect differentially abundant peptides. Cleavage motif analysis was performed using an online tool for protease specificity characterization [31]. The Dimethylation TAILS experiments were analyzed using MaxQuant Version 1.6.14.0. Standard Quantitation with appropriate multiplicity was chosen depending on the experimental design: DimethNterm0, 4 or 8 and DimethLys0, 4 or 8 and semi-tryptic enzyme specificity (ArgC) was chosen.

Protein Glycation
Differentiated SH-SY5Y cells were incubated with 5 mM methylglyoxal (MGO) for 2 h under normal cell culture conditions. Cells were washed with DPBS three times and whole cell lysates were prepared as described for Western Blot. Protein glycation was probed by immunoblotting for MGO as described above.

In Vitro System of Differentiated SH SY5Y Neuron-Like Cells
To probe for putative, natural neuronal DJ-1 proteolytic substrates we overexpressed different DJ-1 variants ( Figure S1A) in the human neuroblastoma cell line SH-SY5Y with a comparatively low background of endogenous DJ-1 ( Figure 1A). The putative active protease form of DJ-1, DJ-1 WT ∆H8, lacks the 15 C-terminal amino acid residues 175-189 forming helix 8 (H8), which can block substrate access to the catalytic site. In the inactivated form, DJ-1 C106A ∆H8, the proposed catalytic nucleophile cysteine-106 (C106) was mutated to alanine ( Figure S1B). For comparison, full length DJ-1 WT and DJ-1 C106A and the empty vector were overexpressed as well. All variants were expressed with an IRES-GFP. Subpopulations of transduced cells with equal GFP signal intensity were selected using FACS, and five different cell lines with stable overexpression of the different DJ-1 variants or the empty vector were established. Although the same gates were used for the sorting of all cell lines, higher protein expression of the full length variants in comparison to the truncated variants was observed ( Figure 1A). The different overexpression levels need to be taken into consideration when directly comparing the impact of full length DJ-1 and DJ-1∆H8 on cellular behavior.
We consider differentiated SH-SY5Y cells to be a more adequate model than nondifferentiated cells to investigate cell-physiological roles of proteins such as DJ-1; as highlighted by a proteome study comparing undifferentiated and differentiated SH-SY5Y cells, which substantiated profound differences of their respective proteome biology [39]. To this end, we differentiated the aforementioned transduced SH-SY5Y cell lines stably overexpressing different DJ-1 variants, using a protocol adapted from [24] including three days of sequential retinoic acid (RA) and subsequently four days of Brain Derived Neurotrophic Factor (BDNF) treatment ( Figure 1B). Untransduced differentiated SH-SY5Y cells displayed formation and extension of neurite-like structures ( Figure 1C) and dramatically reduced proliferation as shown by real-time monitoring and BrdU incorporation ( Figure 1D,E). We observed a higher expression of the mature neuronal markers Growth-associated Protein 43 (GAP43) and β3-Tubulin following differentiation ( Figure 1F). Hence, we refer to these cells as neuron-like, post-mitotically differentiated cells. Each transduced cell line could be differentiated into neuron-like cells with neurite-like structures ( Figure S1C). Cells 2021, 10, x FOR PEER REVIEW 6 of 15 We consider differentiated SH-SY5Y cells to be a more adequate model than nondifferentiated cells to investigate cell-physiological roles of proteins such as DJ-1; as highlighted by a proteome study comparing undifferentiated and differentiated SH-SY5Y cells, which substantiated profound differences of their respective proteome biology [39]. To this end, we differentiated the aforementioned transduced SH-SY5Y cell lines stably over-expressing different DJ-1 variants, using a protocol adapted from [24] including three days of sequential retinoic acid (RA) and subsequently four days of Brain Derived Neurotrophic Factor (BDNF) treatment ( Figure 1B). Untransduced differentiated SH-SY5Y cells displayed formation and extension of neurite-like structures ( Figure 1C) and dramatically reduced proliferation as shown by real-time monitoring and BrdU incorporation ( Figure 1D,E). We observed a higher expression of the mature neuronal markers Growth-associated Protein 43 (GAP43) and β3-Tubulin following differentiation ( Figure 1F). Hence, we refer to these cells as neuron-like, post-mitotically differentiated cells. Each transduced cell line could be differentiated into neuron-like cells with neurite-like structures ( Figure S1C).

Degradome Analysis
To detect putative proteolytic substrates of DJ-1 and to assess the influence of DJ-1 on the cellular degradome, we performed TAILS, an N-terminomic technique for the identification and quantification of native and proteolytically generated protein N-termini [22]. In brief, protein N-termini and N-termini generated by proteolytic cleavage are

Degradome Analysis
To detect putative proteolytic substrates of DJ-1 and to assess the influence of DJ-1 on the cellular degradome, we performed TAILS, an N-terminomic technique for the identification and quantification of native and proteolytically generated protein N-termini [22]. In brief, protein N-termini and N-termini generated by proteolytic cleavage are chemically protected and tagged with a tandem mass tag (TMT). Proteins are then subjected to tryptic digestion. The thereby generated Neo N-Termini are not chemically protected and can be depleted using an amine-reactive polymer. We compared differentiated SH-SY5Y cells overexpressing the catalytically active form DJ-1 WT ∆H8 to cells overexpressing the catalytically inactive form DJ-1 C106A ∆H8 in three replicates. We chose an 11plex tandem mass tag (TMT)-based labeling approach. We identified a total of 2223 peptides, which either represent native protein N-termini or proteolytically generated N-termini (Supplementary  Table S1).
To identify differentially abundant peptides, the LIMMA package from R was used. As an initial step to identify enriched or underrepresented cleavage sites in connection with overexpression of different DJ-1 types, we considered N-terminal peptides with a non-adjusted p-value < 0.05. As shown in Figure 2A, we detected 71 peptides enriched in DJ-1 WT ∆H8 overexpressing cells as compared to DJ-1 C106A ∆H8 overexpressing cells that fulfill this criterion. However, none of these peptides fit to the previously described sequence specificity of DJ-1: V↓A, valine in P1 and alanine in P1' (Supplementary Table S1, sheet "enriched cleavage sites"). A summarized cleavage site motif is shown in Figure 2B, which shows the occurrence of amino acids of differentially abundant cleaved peptides.
We note that P2 mostly comprises aliphatic residues, whereas P1 and P1 are predominantly composed of small amino acids. This motif is reminiscent of cysteine cathepsins; e.g., cathepsins B and L [40]. To investigate a potential link to lysosomal biology we probed for Cathepsin B (CTSB) activity by immunoblot. Cathepsin B is a potent lysosomal protease whose activation includes removal of its 62 residue N-terminal domain. We verified a higher amount of activated CTSB in DJ-1 WT ∆H8 overexpressing cells compared to DJ-1 C106A ∆H8 or empty vector control cells ( Figure 2C). Analysis of TAILS peptides confirmed significantly more Cathepsin B propeptide cleavage in DJ-1 WT ∆H8 relative to DJ-1 C106A ∆H8 overexpressing cells ( Figure S2). A higher amount of the CTSB active form was also found in cells overexpressing full length DJ-1 WT compared to DJ-1 C106A or empty vector control cells ( Figure 2D). In conclusion, the affected degradome of DJ-1 WT ∆H8 overexpressing cells is dominated by cysteine cathepsin-type proteolysis when compared to the degradome of DJ-1 C106A ∆H8 overexpressing cells. CTSB activation upon DJ-1 overexpression is dependent on C106 but independent of H8.

Systemic Impact of DJ-1 ∆H8 on the Cellular Proteome
Some of the observed altered cleavage events that could not have been generated by cysteine cathepsins, might be explained by a systemic effect of DJ-1 WT ∆H8 overexpression. To approach this question, we conducted a quantitative proteome comparison of cells overexpressing DJ-1 WT ∆H8 and cells overexpressing DJ-1 C106A ∆H8 (see methods). A total number of 2334, 2933 and 4421 proteins were identified in the first, second and third experiment, respectively. An overlap of 2053 proteins was consistently found in all three experiments ( Figure S3A). DJ-1 WT ∆H8/ DJ-1 C106A ∆H8 normalized ratios showed a normal distribution in all three experiments ( Figure S3B). Alterations of protein abundances were calculated as log2 fold change (FC) values of DJ-1 WT ∆H8/DJ-1 C106A ∆H8 normalized ratios. To define proteins with altered abundance, LIMMA was used. Only proteins with an abundance change of more than 50% (FC> |log2(1.5)|) in two of three experiments and with a non-adjusted p-value ≤ 0.05 were considered to be significantly altered (Supplementary  Table S2). We found 689 proteins with significantly higher abundance and 243 proteins with significantly lower abundance in DJ-1 WT ∆H8 overexpressing cells compared to DJ-1 C106A ∆H8 overexpressing cells ( Figure 3A). To interpret the pronounced effect of DJ-1 WT ∆H8 on the proteome of neuron-like cells, all proteins significantly altered in abundance were clustered into gene ontology biological processes (GO BPs). The most significantly enriched GO BP terms are shown in Figure 3B,C. Proteins with a lower abundance in DJ-1 WT ∆H8 are involved e.g., in membrane raft organization, beta amyloid metabolism, and integrin mediated signaling ( Figure 3B). Proteins enriched in DJ-1 WT ∆H8 primarily play a role in mitochondrial processes such as mitochondrial transcription, mitochondrial translation, respiration, and mitochondrial metabolic processes including fatty acid beta oxidation and mitochondrial RNA metabolism. Many studies have shown a role of DJ-1 in maintaining proper mitochondrial function (reviewed in [41]). Here we show that Cterminally truncated DJ-1 lacking H8 affects mitochondrial processes in a C106-dependent manner as well.

Systemic Impact of DJ-1 ΔH8 on the Cellular Proteome
Some of the observed altered cleavage events that could not have been generated by cysteine cathepsins, might be explained by a systemic effect of DJ-1 WT ΔH8 overexpression. To approach this question, we conducted a quantitative proteome comparison of cells overexpressing DJ-1 WT ΔH8 and cells overexpressing DJ-1 C106A ΔH8 (see methods). A total number of 2334, 2933 and 4421 proteins were identified in the first, second and third experiment, respectively. An overlap of 2053 proteins was consistently found in all three experiments ( Figure S3A). DJ-1 WT ΔH8/ DJ-1 C106A ΔH8 normalized ratios showed a normal distribution in all three experiments ( Figure S3B). Alterations of protein abundances were calculated as log2 fold change (FC) values of DJ-1 WT ΔH8/DJ-1 C106A ΔH8 normalized ratios. To define proteins with altered abundance, LIMMA was used. Only proteins with an abundance change of For the investigation of a putative indirect effect of DJ-1 WT ∆H8 on the degradome that we observed in the TAILS experiment, we reviewed the proteins significantly higher or lower in the quantitative proteome comparison in detail. We found several proteases with altered abundance (Figure 4). Seven mitochondrial proteases, such as Neurolysin and Mitochondrial processing peptidase subunit beta, increased in DJ-1 WT ∆H8 overexpressing cells. Also, two lysosomal proteases, namely Dipeptidyl peptidase 1 and Gamma-glutamyl hydrolase, showed higher abundance in DJ-1 WT ∆H8 overexpressing cells. In conclusion, the observed effect on the cellular degradome can partially be explained by the altered abundance of other proteases, including non-cysteine-type lysosomal proteases and mitochondrial proteases.
in membrane raft organization, beta amyloid metabolism, and integrin mediated signaling ( Figure 3B). Proteins enriched in DJ-1 WT ΔH8 primarily play a role in mitochondrial processes such as mitochondrial transcription, mitochondrial translation, respiration, and mitochondrial metabolic processes including fatty acid beta oxidation and mitochondrial RNA metabolism. Many studies have shown a role of DJ-1 in maintaining proper mitochondrial function (reviewed in [41]). Here we show that C-terminally truncated DJ-1 lacking H8 affects mitochondrial processes in a C106-dependent manner as well.

Protective Effect of DJ-1 on Protein Glycation
A few years ago, DJ-1 deglycase activity was reported [14]. The authors suggested that deglycation constitutes the primary cell-physiological function of DJ-1. However, this concept has been challenged [42,43]. Using the in vitro model system of differentiated SH SY5Y cells, we probed for the potential deglycation properties of the four different DJ-1 variants. Differentiated SH-SY5Y cells were exposed to 5 mM methylglyoxal (MGO) for 2 h in culture. Immunoblots of whole protein extracts with an anti-MGO antibody showed a significant decrease of protein glycation in cells overexpressing DJ-1 WT compared to empty vector control cells ( Figure 5A), which was abolished in the DJ-1 C106A mutant. These results confirm a C106-dependent protective effect of DJ-1 on protein glycation in neuron-like cells. In contrast, protein glycation was not decreased in DJ-1∆H8 overexpressing cells compared to empty vector control cells ( Figure 5B). Our findings corroborate that DJ-1 protects proteins from glycation and we show that the C-terminal Helix 8 (H8) is essential for this protective effect.
found several proteases with altered abundance (Figure 4). Seven mitochondrial proteases, such as Neurolysin and Mitochondrial processing peptidase subunit beta, increased in DJ-1 WT ΔH8 overexpressing cells. Also, two lysosomal proteases, namely Dipeptidyl peptidase 1 and Gamma-glutamyl hydrolase, showed higher abundance in DJ-1 WT ΔH8 overexpressing cells. In conclusion, the observed effect on the cellular degradome can partially be explained by the altered abundance of other proteases, including non-cysteine-type lysosomal proteases and mitochondrial proteases.

Protective Effect of DJ-1 on Protein Glycation
A few years ago, DJ-1 deglycase activity was reported [14]. The authors suggested that deglycation constitutes the primary cell-physiological function of DJ-1. However, this concept has been challenged [42,43]. Using the in vitro model system of differentiated SH SY5Y cells, we probed for the potential deglycation properties of the four different DJ-1 variants. Differentiated SH-SY5Y cells were exposed to 5 mM methylglyoxal (MGO) for 2 h in culture. Immunoblots of whole protein extracts with an anti-MGO antibody showed a significant decrease of protein glycation in cells overexpressing DJ-1 WT compared to empty vector control cells ( Figure 5A), which was abolished in the DJ-1 C106A mutant. These results confirm a C106-dependent protective effect of DJ-1 on protein glycation in neuronlike cells. In contrast, protein glycation was not decreased in DJ-1ΔH8 overexpressing cells compared to empty vector control cells ( Figure 5B). Our findings corroborate that DJ-1 protects proteins from glycation and we show that the C-terminal Helix 8 (H8) is essential for this protective effect.

Discussion
DJ-1 plays a role in tumor progression (reviewed in [44]) and DJ-1 loss-of-function is associated with autosomal early-onset Parkinson's disease [1]. DJ-1 protects cells from oxidative stress [7], but its precise function remains elusive. The aim of the present study is Immunoblots of whole cell lysates of untreated and MGO treated cells expressing different DJ-1 variants with anti-MGO antibody and GAPDH as loading control: right: quantification of signal intensity of MGO-modified proteins relative to empty vector control (pMIG empty) with ImageJ of three independent experiments for full length variants and two independent experiments for C-terminally truncated variants (** p ≤ 0.01 by two tailed t-test on independent groups).

Discussion
DJ-1 plays a role in tumor progression (reviewed in [44]) and DJ-1 loss-of-function is associated with autosomal early-onset Parkinson's disease [1]. DJ-1 protects cells from oxidative stress [7], but its precise function remains elusive. The aim of the present study is to further our understanding of the cell-physiological roles of DJ-1 and their dependence on helix 8, which supposedly acts as a propeptide. Due to the suggested role of DJ-1 in neurodegenerative diseases, we employed an in vitro model of post-mitotically differentiated, neuron-like cells.
While our N-terminomic analysis revealed differences in the degradome between DJ-1 WT ∆H8 and DJ-1 C106A ∆H8 overexpressing cells, we did not find quantitatively affected N-termini that match the sequence specificity of proteolytically active DJ-1 (V↓A) that was determined in vitro before [17]. We note that this finding does not refute DJ-1 acting as an endoprotease in cellulo. It is possible that putative DJ-1 substrates were absent in differentiated SH-SY5Y cells. Furthermore, it is possible that DJ-1 cleavage products are prone to rapid degradation, hence escaping detection by N-terminomics.
Independent of direct DJ-1 substrates, we found increased levels of several other, mainly mitochondrial and lysosomal, proteases and observed a pronounced, yet indirect impact of DJ-1 on cysteine cathepsin-type proteolysis. This impact depends on the presence of C106 but is independent of H8. Of note, we do not postulate that DJ-1 directly activates Cathepsin B but consider this an indirect, yet dominant effect whose mechanistic underpinning remains beyond the findings of the present study. Our findings are in line with previous reports that link the cell-physiological function of DJ-1 to lysosomal biology: Gao and colleagues reported increased autophagy upon DJ-1 overexpression [45] and other groups showed reduced autophagy upon loss of DJ-1 [46,47]. The link between DJ-1 and autophagy awaits further mechanistic clarification.
The quantitative proteome comparison of DJ-1 WT ∆H8 and DJ-1 C106A ∆H8 overexpressing cells showed an increase of mitochondrial proteins in DJ-1 WT ∆H8. The protective function of DJ-1 on mitochondria and the reduction of reactive oxygen species was observed before in several studies (reviewed in [41]). Proteins more abundant in DJ-1 WT ∆H8 overexpressing cells are involved in biological processes such as mitochondrial translation, complex I assembly and transport into mitochondria, which are important for mitochondrial biogenesis. Hence it is conceivable that DJ-1 is not only linked to the maintenance but also to the biogenesis of mitochondria.
Interestingly, we observed an effect on mitochondrial and lysosomal biology when overexpressing the truncated form DJ-1 WT ∆H8. DJ-1 exists as a homodimer in cells, which is its expected active form. Based on its crystal structure, a novel mode of dimerization mediated by C-terminal helix H8 was suggested [21]. It remains to be determined whether DJ-1 dimerization is dependent on H8 or whether it may also function as a monomer. We observed lower protein amounts of the overexpressed truncated versions in comparison to the DJ-1 full length versions despite an equal expression of GFP in the bicistronic system. This could mean that the half live of DJ-1∆H8 is shorter than that of full length DJ-1.
The accumulation of glycated proteins and lipids, which represent so-called "advanced glycation end products" (AGE), is a feature of neurodegenerative diseases. Proteins and lipids become glycated by the exposure to reducing sugars such as MGO. The accumulation of AGE may contribute to the development of neurodegenerative disease [48]. A protective effect of DJ-1 against protein glycation has been shown in yeast, bacteria [49] and human keratinocytes [50]. Whether DJ-1 detoxifies free MGO by acting as a glyoxalase [12,42,51] or removes adducts formed by MGO on proteins by acting as a deglycase [14,52] is a matter of debate. Here we confirm a C106-dependent protective effect of DJ-1 on protein glycation in neuron-like cells and show that in our cellular system H8 is essential for this effect.
In summary, we used neuron-like SH-SY5Y cells to study the physiological role(s) of DJ-1∆H8. We were unsuccessful in identifying DJ-1 proteolytic substrates, but our findings strengthen the impact of DJ-1 on lysosomal and mitochondrial biology, as well as its protective role against protein glycation. All of these activities depend on the presence of cysteine 106, while the C-terminal Helix 8 (H8) is dispensable for the link to lysosomal biology.

Conflicts of Interest:
The authors declare no conflicting interests.