Generation of SNCA Cell Models Using Zinc Finger Nuclease (ZFN) Technology for Efficient High-Throughput Drug Screening

Parkinson’s disease (PD) is a progressive neurodegenerative disorder caused by loss of dopaminergic neurons of the substantia nigra. The hallmark of PD is the appearance of neuronal protein aggregations known as Lewy bodies and Lewy neurites, of which α-synuclein forms a major component. Familial PD is rare and is associated with missense mutations of the SNCA gene or increases in gene copy number resulting in SNCA overexpression. This suggests that lowering SNCA expression could be therapeutic for PD. Supporting this hypothesis, SNCA reduction was neuroprotective in cell line and rodent PD models. We developed novel cell lines expressing SNCA fused to the reporter genes luciferase (luc) or GFP with the objective to enable high-throughput compound screening (HTS) for small molecules that can lower SNCA expression. Because SNCA expression is likely regulated by far-upstream elements (including the NACP-REP1 located at 8852 bp upstream of the transcription site), we employed zinc finger nuclease (ZFN) genome editing to insert reporter genes in-frame downstream of the SNCA gene in order to retain native SNCA expression control. This ensured full retention of known and unknown up- and downstream genetic elements controlling SNCA expression. Treatment of cells with the histone deacetylase inhibitor valproic acid (VPA) resulted in significantly increased SNCA-luc and SNCA-GFP expression supporting the use of our cell lines for identifying small molecules altering complex modes of expression control. Cells expressing SNCA-luc treated with a luciferase inhibitor or SNCA siRNA resulted in Z’-scores ≥ 0.75, suggesting the suitability of these cell lines for use in HTS. This study presents a novel use of genome editing for the creation of cell lines expressing α-synuclein fusion constructs entirely under native expression control. These cell lines are well suited for HTS for compounds that lower SNCA expression directly or by acting at long-range sites to the SNCA promoter and 5’-UTR.


Introduction
Parkinson's disease 1 (PARK1) is an autosomal dominant disorder caused by missense mutations and multiplications of the SNCA gene, encoding α-synuclein [1][2][3]. Although missense mutations are rare events, duplications and triplications of the SNCA gene [1][2][3][4][5][6][7][8][9] are found in both familiar and sporadic PD, and have been linked to more than 30 families with PD and parkinsonism [5]. The common occurrences of SNCA genomic multiplications point to the importance of gene dosage and overexpression of wildtype α-synuclein in causing neurodegeneration in α-synucleinopathies [3]. These observations were in line with data showing neuronal toxicity in cell and animal models of α-synuclein overexpression [10][11][12][13][14][15][16][17]. Elevated levels of wild type α-synuclein in patient brains or patient-derived cell lines were also observed in sporadic PD [18][19][20] and in familial PD caused by mutations in PARK2 [14,16,21], GBA [10], and LRRK2 [12]. These observations support the widely held hypothesis that elevated levels of αsynuclein cause death of dopaminergic neurons in PD. Reducing the levels of α-synuclein was neuroprotective in several studies of cellular and animal models of α-synucleinopathies [14,[22][23][24]. Furthermore, SNCA knock-out mouse models showed increased dopamine release with paired stimuli or elevated Ca 2+ , but exhibited no PD phenotypes [25,26]. Therefore, reducing the level of α-synuclein likely delays the onset of PD phenotypes with fewer risks to the recipients. However, specific compounds that reduce the expression and levels of endogenous α-synuclein for therapeutic application have not been identified. One impediment to identifying such compounds is the lack of cell line models that express SNCA in its proper genomic context.
Expression control is complex for SNCA, involving control elements that are located far upstream of the SNCA transcriptional start site, a complicated structure of repeats in the promoter controlling SNCA transactivation and epigenetic expression control. One prominent feature of the SNCA promoter is NACP-REP1, a regulatory element consisting of a complex structure of dinucleotide CT, TA, and CA repeats, flanked by two domains that enhance SNCA expression [27,28]. The NACP-REP1 is located at 8852 bp upstream of the SNCA transcription start site [29]. Polymorphisms at the NACP-REP1 region regulate SNCA expression, and dinucleotide polymorphisms at the NACP-REP1 locus were associated with Parkinson's and Alzheimer's Diseases [30,31]. In addition, a decrease in hypermethylation of the SNCA promoter CpG island has been observed in sporadic PD [32,33], and promoter CpG hypomethylation was correlated with increased α-synuclein expression in a HEK293 cell model [33]. Expression control by far upstream regions can involve complex chromatin loops and epigenetic modifications [34,35]. We reported the generation of cell lines that preserve the entire promoter regulator networks controlling SNCA expression for HTS of small compounds that alter the interactions of the distal elements with the SNCA promoter.
Because of the complexity of SNCA expression control and great length of the SNCA promoter, we employed genome editing to introduce reporter genes downstream of the SNCA locus to create cell line models for identifying SNCA inhibitory compounds. In the present study, we utilized zinc finger nuclease (ZFN) genome editing [36][37][38] to knock-in the desired reporter gene sequences into the SNCA locus. This approach is advantageous to transfecting expression plasmids containing the targeted cDNA due to high integration efficiency and rare integration at off-target sites [36,38]. Additionally, this approach is more likely to report expression from the gene of interest (SNCA) since the entire expression control mechanism is retained. The objective of this study was to generate SH-SY5Y neuroblastoma cell lines edited to express α-synuclein fusions to luciferase or green fluorescent protein (GFP) under the regulation of the complete, native SNCA promoter/enhancer system, and to demonstrate the utility of these cell lines for compound screening.

Ethics Statement
No animal or human participants were used in this research.

ATXN2 cell lines
The HEK293 cell line expressing ATXN2-luc was previously established by our lab [39]. Briefly, HEK293 cells were stably transfected with plasmid pGL2-5A3 to create cell lines H2 and S2, respectively. Plasmid pGL2-5A3 includes -1062 to +660 of the ATXN2 gene ending on the first CAG of the CAG tract, upstream of the luciferase gene. The luciferase start codon was mutated to CTG to fuse the ataxin-2 fragment with luciferase. The luciferase gene is followed by 1019 bp of ATXN2 3'-UTR and downstream sequence (+4098 to +5116). ATXN2 bp positions are relative to the transcription start site (TSS) as described in Scoles et al. [39]. Selection was accomplished using hygromycin.

Construction of the right and left ZFN-FokI expression plasmids
To generate cell lines expressing full-length α-syn-luc (α-synuclein-luciferase) and α-syn-GFP (αsynuclein-green fluorescent protein) fusion proteins under the regulation of the native regulatory elements, we used the Zinc Finger Nuclease (ZFN) genome editing method [36]. The ZFN plasmids that we produced create a double strand break (DSB) at the SNCA locus that in the presence of a customized donor DNA, directly inserts a reporter gene (luc or GFP) in frame with the last codon of the SNCA gene (Fig 1). We employed a pair of custom-made ZFN-FokI plasmids (Sigma Aldrich) that specifically cleaved at a site 59 bp downstream of the targeted locus (S1 Fig). The ZFN plasmids were verified by Cel-1 nuclease assay using PCR primers flanking the predicted breakage point which generated an uncut PCR product of 329 bp (S1  Generation of stable SH-SY5Y cell lines expressing α-syn-GFP or αsyn-luc To generate SH-SY5Y cell lines expressing α-syn-luc or α-syn-GFP, SH-SY5Y cells were transfected with a cocktail of ZFN and donor plasmids using Invitrogen lipofectamine 2000. The transfected cells were selected with 10 μg/ml puromycin (InvivoGen). Puromycin selected colonies from each positive well were plated in new tissue culture dishes for storage and cell line characterization using RT-PCR, qPCR, Western blots, SNCA siRNA, and VPA induction. Three cell lines were created and designated GFP12, Luc6B, and Luc6B-5 for their expression of GFP (GFP12) or luciferase (Luc6B), respectively.

Valproic acid treatment
Valproic acid (VPA) treatments were performed according to Choi et al [40], Equal numbers of Luc6B cells were plated in 384-well tissue culture plates and grown in DMEM/FBS/puromycin containing the indicated concentrations of VPA for the designated times. Luciferase assays were performed according to the protocol provided in the Promega Bright-Glo Luciferase Assay System Kit (Cat # E2620). VPA was purchased from Sigma Aldrich.
For the GFP12 cell line, 5000 cells were plated in 384-well, plate containing the indicated concentrations of VPA in phenol red free DMEM/10% FBS/10 μg/ml puromycin. After 72 hours, GFP fluorescent intensity was measured using a Beckmann DTX880 plate reader, then MTT viability assays (Promega's CellTiter96Q kit, cat #5421) were performed. For Western blots and qPCR, equal numbers of GFP12 or Luc6B cells were cultured in six-well dishes overnight. The next day, cells were treated with VPA at 0, 5 or 10 mM for 48 hrs. After treatment, cells were harvested and evaluated by qPCR and Western blotting.

RT-PCR and quantitative RT-PCR
Total RNA was extracted using the RNeasy Mini Kit (Qiagen Inc., USA) according to the manufacturer's protocol. DNAse I treated RNA samples were used to synthesize cDNAs using the Pro-toScript cDNA First Strand cDNA Synthesis Kit (New England Biolabs Inc., USA). Synthesized cDNAs were used for RT-PCR and qPCR analyses. GAPDH amplification was conducted in parallel as an internal control for RNA quality and was employed to evaluate the quality of the reverse transcriptase reactions. Quantitative RT-PCR was performed in QuantStudio 12K (Life Technologies, Inc., USA) with the Power SYBR Green PCR Master Mix (Applied Biosystems Inc, USA). PCR reaction mixtures contained SYBR Green PCR Master Mix and 0.5 pmol primers and PCR amplification was carried out for 45 cycles: denaturation at 95°C for 10 sec, annealing at 60°C for 10 sec and extension at 72°C for 40 sec. The threshold cycle for each sample was chosen from the linear range, converted to a starting quantity by interpolation from a standard curve run on the same plate for each set of primers. All gene expression levels were normalized to the GAPDH mRNA levels. Primer pairs designed for RT-PCR and qPCR are listed in (S1 Table).
Control and SNCA siRNA transfection GFP12 and Luc6B cells were cultured and maintained in DMEM media with 10% fetal bovine serum containing 10 μg/ml puromycin. For siRNA-mediated depletion of α-synuclein and αsyn mRNA, cells were cultured in 6 well dishes overnight and siRNAs; Control siRNA (Santa Cruz Inc.; cat# sc-37007) and α-synuclein siRNA (Santa Cruz Inc.; cat# sc-29619) at a concentration of 100 nM were transfected into GFP12 or Luc6B cells using lipofectamine 2000 (Invitrogen Inc.), according to the manufacturer's protocol. At five days post-transfection, cells were harvested for Western blot analyses.

Results
The human SNCA gene is predicted to generate 11 isoforms resulting in gene products ranging from 67 aa to 140 aa (ENSEMBL.ORG, GENE CODE: ENSG00000145335). Some of these are tissue-specific. The largest isoform group, which consist of 140 aa, was shown to accumulate in dopaminergic neurons of PD patients as Lewy bodies or Lewy neurites [18][19][20]43,44]. The luciferase-/or GFP-puromycin marker was inserted between the last coding codon GCC and the stop codon TAA (GCC/TAA) in exon 6 of the SNCA gene.

Creation of SH-SY5Y cell lines expressing α-synuclein-GFP
To eliminate the production of free GFP or luciferase, we generated donor plasmids containing no ATG initiation on the reporter gene, the SNCA-GFP-2A-Puro and SNCA-luc-2A-Puro donor plasmids (Fig 1). The SNCA-GFP-2A-Puro or SNCA-luc-2A-Puro donor plasmid was cotransfected with the right and left Zinc Finger Nuclease (ZFN) plasmids (Fig 1) into SH-SY5Y cells. Transfected SH-SY5Y cells were transferred to 6-well dishes, and selected with 10 μg/ml puromycin. After 1-2 weeks of puromycin selection, a mixed population of colonies was formed. The colonies in each well were pooled and recultured for characterization and storage. Using these approaches, we created one SH-SY5Y cell line that expressed α-syn-GFP; we labeled this cell line GFP12 (Fig 2, S2 Fig) and two cell lines that expressed α-syn-luc fusion protein, we labeled these cell lines Luc6B and Luc6B-5. Since the Luc6B-5 expressed a lower amount of αsyn-luc mRNA than the Luc6B (S3 Fig), the Luc6B cell line was selected for detailed analyses.
To confirm whether the GFP12 cell line expressed α-syn-GFP, cells were plated in 6-well plates, and fluorescent images were taken 48 hours later. Fig 2 shows that all puromycinselected cells were GFP positive suggesting that 100% of the GFP12 cells expressed GFP (Fig 2). RT-PCR analysis using the F600/GFP-R1 and purF5/e6R1 primer pairs of RNA samples from the GFP12 cell line produced the expected PCR products for each primer pair (Fig 3C). The F600/ GFPR2 primer pair is located at the α-syn mRNA/ GFP junction, while the purF5/e6R1 primer pair is located at the puromycin/3'-UTR α-syn mRNA junction. The positive RT-PCR bands showed that the GFP-2A-Puro reporter gene cassette was properly inserted into the SNCA locus.
Generation of SH-SY5Y cell lines expressing full-length α-synucleinluciferase (α-syn-luc) fusion proteins Transfection of SH-SY5Y cells with ZFN-FokI plasmid and the SNCA-Luc-2A-Puro donor plasmid resulted in two α-syn-luc expressing cell lines, designated Luc6B and Luc6B-5. We selected RT-PCRs using primer pairs F600/LucR1 and purF4/e6R1, and purF5/e6R1 ( Fig  4B) generated RT-PCR products that were absent from untransfected cells. The sizes of these PCR products were consistent with the predicted PCR fragment for these primer pairs ( Fig  4C). These results confirmed that the luc-2A-Puro cassette was inserted correctly into the  targeted SNCA locus. RT-PCR using the control GAPDH primer pair showed that RNA samples extracted from the Luc6B and untransfected SH-SY5Y cells were of high quality and quantitatively equal.
Luciferase antibody detected a band of about 77 kDa predicted for the α-syn-luc fusion protein in protein extracts from the Luc6B cell line (Fig 4D, lane 2) but only a 60 kDa band from SH-SY5Y cells transfected with the pGL2-CMV-luc positive control plasmid (Fig 4D, lane 3). No additional high MW band was detected in the untransfected SH-SY5Y cells (Fig 4D, lane  1). The α-synuclein antibody (211) detected the endogenous 18 and 77 kDa bands, predicted sizes of untagged α-synuclein and α-syn-luc, respectively, from lysates of Luc6B cells transfected with a control siRNA (Fig 4E, lanes 4 and 5). The observation of both the untagged αsynuclein (α-syn) and tagged α-syn-luc proteins indicated that only one copy of the SNCA locus was modified by ZFN. The α-syn 211 antibody failed to detect both the endogenous 18 and 77 kDa α-synuclein bands in protein extracts from Luc6B cells transfected with 100 nM of α-syn siRNA (Fig 4E, lanes 6-7). These observations confirm that the Luc6B cell line specifically expressed the α-syn-luc fusion protein.

Effects of valproic acid on Luc6B and GFP12 expression of α-synuclein
VPA is a histone deacetylase (HDAC) inhibitor that modifies gene expression via epigenetic mechanisms [45,46]. HDAC inhibitors modify gene expression by increasing the level of histone acetylation, generally resulting in increased gene expression. VPA increased the levels of endogenous α-synuclein in SH-SY5Y and was found to be neuroprotective against rotenone cytotoxicity [40,46] and mouse cerebral neuron-enriched cultures [45]. VPA-induced increase of α-synuclein was accompanied by increased protection against glutamate-induced neurotoxicity [47]. We hypothesized that VPA treatment of GFP12 and Luc6B cells would increase αsyn-GFP and α-syn-luc expression, while also realizing that VPA treatment does not necessarily increase the expression of all genes.

Quality of luciferase signal from the Luc6B cell line
We determined the quality metrics of high-throughput screening (HTS) for the Luc6B cells line using luciferase inhibitor ChemBridge 5553825 and SNCA siRNA. The luciferase inhibitor ChemBridge 5553825 was determined previously as a nontoxic compound potently inhibiting luciferase (Fig 7A). After 24 hr treatment, the means and SDs for 10 μM and 1% DMSO vehicle were 2023±1030 and 108,252±7929, respectively (Fig 7B). Following Zang et al. [48], we computed the Z'-score = 0.75 with Signal/Noise (S/N) = 15.4, Signal/Background (S/B) = 53, and Coefficient of variation (C v) = 6.5%). To calculate the quality metrics of the Luc6B cell line in response to SNCA siRNA expression inhibition, Luc6B cells transfected with control siRNA and SNCA siRNA for 5 days were transferred to 384-well plate (n = 24), and luciferase activity was measured 24 hours later. The means and SD for SNCA siRNA and control siRNA were 5,886±163 and 28,093±1,344, respectively. Using these values, the computed quality metrics for the Luc6B cell line, when treated with SNCA siRNA, were as follows: Z'-score: 0.80, S/ Similarly, Western blots using α-syn antibody 211 show that VPA treatment caused a significant dosage-dependent increase in the expression of both wild type α-synuclein and α-syn-GFP fusion protein by GFP12 cells. * P < 0.05, ** P < 0.01, *** P < 0.001. doi:10.1371/journal.pone.0136930.g005 N:18.8, S/B: 4.8, C v : 5.3% (Fig 7C). The high Z'-scores calculated from inhibiting luciferase activity by ChemBridge 5553825 (Z'-score: 0.75) and SNCA siRNA (Z'-score: 0.80) support the use of the Luc6B cell line in HTS.
Expression of α-synuclein is regulated at multiple points such as the transcriptional, translational, and degradation levels. At the transcriptional level, SNCA expression is regulated by the 400 bp promoter/enhancer domain, NACP-REP1 [27,29,32,33], located at about 8852 bp upstream of the transcription start site [28] which is flanked by two enhancer domains. Standard generation of cell lines using a transfected plasmid with short promoters would miss compounds acting on distant regulatory domains. The GFP12 and Luc6B cell lines, generated by directed insertion of the reporter gene (GFP or luciferase) in-frame with the SNCA gene, has none of the above disadvantages since the tagged α-synuclein expression is under the control of the entire transcription regulatory system. This property will allow for the discovery of compounds that affect all possible regulatory domains/elements of the expression of the SNCA gene.
Chromatin structure influences the overall expression of the transcriptome. Valproic acid is one chromatin remodeling inhibitor that we chose to evaluate using our cell models since it had previously been demonstrated to be neuroprotective in cell and animal models of PD [40,[45][46][47]. Increased levels of both SNCA mRNA (Figs 5C and 6C) and tagged α-synuclein (α-syn-GFP and α-syn-luc) (Figs 5D and 6D) upon treatment with VPA suggested that altered chromatin structure in our cell line model is associated with SNCA activation. The non-specific nature of VPA to act genome-wide is likely why neuroprotective properties for VPA have been observed in PD models despite its known function to increase SNCA expression [40,[45][46][47]. α-Synuclein clearance occurs through either lysosomal or ubiquitin-dependent proteasomal pathways [53][54][55][56][57][58]. Inhibiting either of these two pathways would increase the intracellular levels of α-synuclein. Luc6B cell lines treated with bafilomycin A1 for 24 hours increased luciferase activity (S4 Fig) indicating that the level of α-syn-luc fusion protein had increased. This observation was in line with previous observations that bafilomycin A1 treatment increased αsynuclein [54,57].
RNA binding proteins, microRNAs, and short noncoding RNAs, which interact with either 3' or 5' UTRs, may also regulate the expression of α-synuclein. One such study included a screen of >300,000 small molecules inhibiting SNCA expression, and identified compounds that interfered with iron-regulatory protein 1 (IRP-1) binding to the SNCA 5' stem-loop altering α-synuclein expression in H4 and SH-SY5Y neuroblastoma cells [59,60]. The disadvantage of the study was that the screening assay included only the SNCA 5'-UTR, eliminating the ability to identify inhibitors acting at other SNCA regions. Our GFP12 and Luc6B cell lines offer the advantage of a complete, unmodified 5' UTR and 3' UTR sequences, which allow for the discovery of compounds acting at these SNCA regions to modify SNCA expression or mRNA stability.
Clearance of wild type and accumulated α-synuclein occurs through the lysosomal, endosomal and ubiquitin mediated proteasomal pathways [53,[61][62][63]. The ubiquitin-proteasomal pathway [64] mostly mediates wild type α-synuclein clearance, while clearance of aggregated, misfolded, or accumulated α-synuclein occurs through the autophagy dependent lysosomal pathway [65,66]. Discovery of compounds that can enhance the function of these pathways will reduce the accumulation of α-synuclein levels in the cell. For example, compounds that enhance the enzymatic activity or increase the expression of ATP13A2 enzyme, a protein associated with Parkinson's disease 9 (PARK9), would increase the clearance of accumulated αsynuclein via the lysosomal pathways [67][68][69][70]. Alternatively, compounds that enhanced the ubiquitin-dependent clearance of misfolded E3 ubiquitin ligase Nedd4, or increased Nedd4 expression protected against α-synuclein induced toxicity in Drosophila and reduced α-synuclein accumulation leading to decreased DA cell death in the rat substantia nigra [71].
As accumulation of wild type and mutant α-synuclein is pathogenic in PD and a feature of familial PD and other α-synucleinopathies, compounds that reduce the level of α-synuclein may alleviate and prevent the occurrence of Parkinsonism phenotypes. The Luc6B and GFP12 cell lines offer many advantages over other cell lines expressing only partial SNCA mRNA or partial promoter/enhancer sequences for HTS to discover compounds that reduce the level of α-synuclein. These advantages include the preservation of the intact transcriptional regulatory system, the 5' UTR, and 3' UTR. Therefore, SNCA expression in the Luc6B and GFP12 cell lines is under the control of the complete transcriptional regulatory system. This allows the detection of compounds that specifically affect important SNCA transcriptional domains/elements or compounds that have global effects on the transcription of the SNCA gene that would be missed if cell line used expresses only a partial regulatory DNA segment. Furthermore, the presence of unmodified structures of the 5'UTR and 3'UTR of SNCA-Luc or SNCA-GFP hybrid mRNA allow for discovery of compounds that interfere with regulatory proteins that interact with these two UTR regions.
Additional advantages include the expression of full-length α-synuclein fused with luciferase or GFP, allowing for the discovery of compounds that affect all clearance pathways, be they lysosomal, proteasomal, or endosomal pathways such as iron-responsive protein 1 (IRP-1) or microRNAs [47]. Furthermore, the α-syn-luc or α-syn-GFP fusion protein produced by the Luc6B or GFP12 cell line allows for fine evaluation of SNCA expression with varying compound doses [72].

Conclusions
The high Z'-score of luciferase activity supports the use of the Luc6B cell line for HTS (Fig 7). Furthermore, the GFP12 cell line can be used as an independent cell line assay for validation of compounds, and is amenable to high-content assays and evaluation of compound effects on αsynuclein aggregation and subcellular localization.
The cell lines described in this study represent a unique resource for high-throughput compound screening to discover drugs that reduce cellular α-synuclein protein levels. Finally, the successful generation of the Luc6B and GFP12 cell lines demonstrate that ZFN-mediated knock-in of a reporter gene into a designated gene locus is a valuable tool for generating cell lines for HTS. RT-PCR amplicons of RNAs isolated from Luc6B and Luc6B-5 cells using the F600/lucR1, F600/lucR2 and GAPDH primer pairs. Both primer pairs, F600/lucR1 and F600/lucR2, produced the correct bands at the predicted size for fragments generated by these primer pairs. Lane 1, SH-SY5Y, lane 2, Luc6B-5, and lane 3, Luc6B. These results showed that Luc6B cells expressed a high level of α-syn-luc mRNA than the Luc6B-5 cell line. Therefore, the Luc6B cell line was selected for detailed studies. . and Luc6B cells were cultured in 6-well dishes, and grown in DMEM/FBS medium containing 50 μM retinoic acid for 8 days to differentiate cells into neuron-like cells. Cells were transferred to clean wells every 3-4 days. On the day prior to the experiment, cells were transferred to clean wells. The next day, cells were treated with DMSO, 20 nM, and 200 nM of bafilomycin A1. Luciferase activity was measured 24 hrs later using Promega Luciferase detection kit. Bafilomycin A1 was purchased from Sigma Aldrich. (TIF) S1 Table. Oligonucleotide sequences of PCR primers used in this manuscript for RT-PCR or qPCR. (XLSX)