Autophagy–lysosome pathway alterations and alpha-synuclein up-regulation in the subtype of neuronal ceroid lipofuscinosis, CLN5 disease

Neuronal ceroid lipofuscinoses (NCLs) are a group of inherited neurodegenerative lysosomal storage disorders. CLN5 deficiency causes a subtype of NCL, referred to as CLN5 disease. CLN5 is a soluble lysosomal protein with an unclear function in the cell. Increased levels of the autophagy marker protein LC3-II have been reported in several subtypes of NCLs. In this report, we examine whether autophagy is altered in CLN5 disease. We found that the basal level of LC3-II was elevated in both CLN5 disease patient fibroblasts and CLN5-deficient HeLa cells. Further analysis using tandem fluorescent mRFP-GFP-LC3 showed the autophagy flux was increased. We found the alpha-synuclein (α-syn) gene SNCA was highly up-regulated in CLN5 disease patient fibroblasts. The aggregated form of α-syn is well known for its role in the pathogenicity of Parkinson’s disease. Higher α-syn protein levels confirmed the SNCA up-regulation in both patient cells and CLN5 knockdown HeLa cells. Furthermore, α-syn was localized to the vicinity of lysosomes in CLN5 deficient cells, indicating it may have a lysosome-related function. Intriguingly, knocking down SNCA reversed lysosomal perinuclear clustering caused by CLN5 deficiency. These results suggest α-syn may affect lysosomal clustering in non-neuronal cells, similar to its role in presynaptic vesicles in neurons.

cathepsin D 15 diseases. On the other hand, reduced autophagy flux has been found in CLN5 −/− and CLN6 −/− ovine neural cultures 16 . This discrepancy of the latter study may be due to different cell types or animal models used in the studies.
In this report, we use CLN5-deficient NCL human patient skin fibroblasts and CLN5 knockdown (KD) HeLa cells to examine the autophagy-lysosome pathway. The CLN5 gene encodes a lysosomal luminal glycoprotein 17,18 . The function of CLN5 in lysosomes remains elusive. A possible role in endosomal sorting was suggested for human CLN5 19 . Another report suggests a CLN5 orthologue in Dictyostelium has glycoside hydrolase activity 20 .
Here we show in CLN5-deficient cells the basal level of LC3-II is elevated, the autophagy flux is increased, and the expression level of α-syn gene SNCA is up-regulated. α-syn is highly expressed in presynaptic neurons and primarily localized to synaptic vesicles 21,22 . The presence of cytoplasmic inclusions filled with insoluble α-syn aggregates is a hallmark of Parkinson's disease 23 . While α-syn has been implicated in several cellular processes, including synaptic vesicle endocytosis 24 and exocytosis 25 , its exact function remains unclear. Despite being primarily associated with neurodegenerative disorders, both CLN5 and α-syn can be detected in a variety of tissues and cell types [26][27][28][29][30][31] . While this indicates more general roles of CLN5 and α-syn in non-neuronal tissues, there have been few studies performed to investigate these roles. The exogenously overexpressed α-syn has been shown to indirectly affect autophagy through the early secretory pathway protein Rab1a in cell culture systems 32 . Interestingly, we found the endogenous α-syn localizes to the lysosomes in human fibroblasts and HeLa cells. Furthermore, we uncovered a potential role for α-syn in regulating lysosomal positioning.

Results
Autophagy flux is increased in CLN5-deficient cells. As an initial step to examine whether the autophagy process might be altered with CLN5 deficiency, we compared the basal levels of an autophagy marker, LC3-II, in fibroblasts from control healthy individuals and fibroblasts derived from CLN5-deficient patients. LC3-II is a lipid-modified form of LC3 that is produced during autophagosome formation and is a commonly used readout for autophagy 33,34 . We found the protein level of LC3-II in CLN5-deficient patient fibroblasts was higher than in control cells (Fig. 1A). To ensure the effects observed were solely due to CLN5 deficiency in patient cells, we generated a CLN5 stable KD cell line with shRNA expression (Fig. 1B). The CLN5 protein level was dramatically reduced in CLN5 stable KD cells. Similar to CLN5-deficient patient cells, an increased LC3-II level was also observed in CLN5 KD HeLa cells. This is consistent with previous studies in various subtypes of NCLs [11][12][13][14][15] . When cells were treated with chloroquine (CQ) to block lysosomal degradation, there was a further accumulation of LC3-II in both wildtype and CLN5KD cells (Fig. 1C). This indicates that the higher basal levels of LC3-II seen in CLN5-deficient cells was not because the lysosome was unable to degrade LC3-II, but actually there was more LC3-II produced 34 . Next, we examined the level of P62, an adaptor protein that links ubiquitinated cargo proteins to LC3-II. Upon autophagy activation, P62 is recruited to autophagosomes and eventually degraded in autolysosomes. On the other hand, accumulation of P62 upon autophagy activation (e.g. by starvation) indicates autophagy impairment. Therefore, the level of P62 is a commonly used measurement for autophagy flux 33 . Unlike LC3-II, we did not observe changes in the basal level of P62 in CLN5-deficient cells (Fig. 1A,B). Next, we followed P62 levels upon starvation to test if CLN5 deficiency blocks autophagic degradation (Fig. 1D,E). Within 1.5 h of starvation, P62 levels were increased in both WT and CLN5 KD HeLa cells (Fig. 1D), suggesting P62 was up-regulated 35 . We then followed long term starvation for up to 8 hours. Starvation was performed in the presence of cycloheximide (to block protein synthesis) and bortezomib (to block proteasome degradation). Under these conditions, P62 was degraded equally efficient in WT and CLN5 KD cells, indicating CLN5 deficient cells were capable of degrading autolysosome materials (Fig. 1E). Similar results were found in control and patient fibroblasts (Fig. S1).
To further monitor the autophagy flux, we expressed mRFP-GFP-LC3 in human fibroblasts. The tandem fluorescent mRFP-GFP-LC3 has been used to follow the maturation progression of autolysosomes from autophagosomes, as the GFP fluorescence will be quenched in a more acidic environment such as autolysosomes, leaving the protein with only the mRFP signal 36 . Compared to control cells, the mRFP-GFP-LC3 emitted a stronger red fluorescent signal in CLN5-deficient cells ( Fig. 2A). In addition, there were more punctate structures with only red fluorescence signals in CLN5 patient cells, suggesting more autolysosomes were present. The Mander's coefficient was shown as a percentage of red fluorescent signal overlapping with green fluorescent signal ( Fig. 2A, right). In patient cells the RFP signal was less overlapped with GFP signal. As control experiments, we also treated control fibroblasts with either lysosome inhibitors or starvation to examine the mRFP-GFP-LC3 response (Fig. 2B). In cells incubated with CQ or bafilomycin A (baf A), the overlaps between mRFP and GFP increased to ~80% (Fig. 2B), compared to no treatment of ~50% ( Fig. 2A, control). On the other hand, when autophagy was induced by starvation (HBSS), the overlaps between mRFP and GFP was reduced to ~30%, similar to what we observed in patient cells. This supports our notion that cells lacking the CLN5 protein have higher basal levels of autophagic activity.
Up-regulation of alpha-synuclein gene SNCA in CLN5-deficient cells. To further elucidate the changes of autophagy pathway in CLN5 disease patient cells, we performed quantitative PCR (qPCR) to follow a set of genes involved in the autophagy pathway (RT 2 Profiler PCR array from SABiosciences). Table 1 lists the genes that were greater than two-fold up or down-regulated (with a p value < 0.05 from three repeats) in CLN5-deficient patient cells. Interestingly, none of the genes directly involved in autophagy initiation or machinery, e.g. ATG genes, were differentially expressed (data not shown). This suggests that the effects on autophagy flux with CLN5 deficiency are most likely through indirect regulation of the autophagy process. Among these genes, SNCA showed the greatest difference compared to control cells with ~90-fold increase in expression under normal cultured conditions. SNCA encodes α-syn, a protein well known for its role in the pathogenicity of Parkinson's disease 23 . Immunoblotting was performed to verify the qPCR results. In normal fibroblasts, α-syn was below the detection level. On the other hand, we were able to detect α-syn expression in CLN5 disease patient fibroblasts (Fig. 3A). Compared to control fibroblasts, HeLa cells synthesized a basal level of α-syn that was readily detectable. In CLN5 stable knockdown cells, α-syn expression level was further increased. Using immunofluorescence microscopy, we were able to visualize α-syn in CLN5-deficient cells (Fig. 3B). Consistent with immunoblotting results, the intensity of α-syn was greatly enhanced in cells lacking CLN5. Stably knocking down CLN5 in HeLa cells recapitulated the increased LC3-II and α-syn levels, indicating that the effects we observed in CLN5 patient fibroblasts were indeed due to CLN5 deficiency. Knocking down CLN5 by transient transfection in control fibroblasts as well as SH-SY5Y cells, however, did not increase the level of α-syn (Fig. S2), suggesting that the effects we observed in CLN5 disease patient cells and CLN5 stable knockdown cells are a consequence of long-term deficiency of CLN5. α-syn is an abundant protein in presynaptic neuronal cells and is associated with synaptic vesicles 21,22 . However, it can also be detected in a variety of tissues [27][28][29][30][31] . Upon exogenous overexpression, α-syn has been reported to localize to several different compartments [37][38][39][40] and cause fragmentation of the Golgi apparatus 32,37 . Our finding of endogenously elevated expression levels of α-syn in CLN5-deficient cells prompted us to examine the localization of α-syn in these cells. We immunostained CLN5 KD cells with GRASP65 for the Golgi apparatus and Lamp1 for lysosomes, we found α-syn was largely colocalized with lysosomes but not the Golgi (Fig. 4A).
Several studies have suggested that α-syn can be degraded in the lysosomes via macroautophagy or the chaperone-mediated autophagy pathway [41][42][43] . Its accumulation has been observed in several lysosomal storage disorders (LSDs) [44][45][46][47] . Therefore, we subsequently analyzed if α-syn was localized in the lumen of the lysosomes as a substrate for degradation. In order to distinguish between outside and inside the lumen, we overexpressed GFP-Rab5 Q79L to induce enlarged endolysosomes 48 . Under normal conditions, endogenous α-syn was not More autolysosomes are present in CLN5-deficient patient fibroblasts. (A) Control or CLN5-deficient fibroblasts that had been transfected with the mRFP-GFP-LC3 plasmid for 24 hours were fixed. GFP and RFP signal were visualized using confocal microscopy. The fraction of RFP/GFP signals that was overlapping was analyzed by the Mander's coefficient using the ImageJ JACoP plugin. In ImageJ, two channels of each image file were first split to two separate images. These two images were then used for plugin JACoP analysis. The threshold of each image was adjusted to reduce the background signals before performing JACoP Mander's coefficient. n = 15 cells, error bar represents SEM. (B) Control fibroblasts that had been transfected with the mRFP-GFP-LC3 plasmid for 24 hours were treated with CQ, baf A, or HBSS for 2 h prior to fixation. The fraction of RFP/GFP signals that was overlapping was analyzed by the Mander's coefficient using the ImageJ JACoP plugin. n = 7 cells, error bar represents SEM. All experiments were repeated at least three times. Scale bar: 20 μm.  present in the lumen (Fig. 4B, top), suggesting it is not subjected to degradation in lysosomes at the level of expression in CLN5-deficient cells. When cells were starved for 2 h to induce autophagy, no α-syn was found in the lumen of enlarged vacuoles either (Fig. 4B, bottom). Subcellular fractionation was performed to further analyze whether α-syn was indeed cytosolic under these conditions (Fig. 4C). β-glucosylceramidase (GBA) is a lysosomal luminal protein and was found in the "P" fraction, indicating the lysosomes in the membrane fraction were intact. In both WT and CLN5 KD HeLa cells, α-syn was found primarily in the soluble cytosolic fraction under normal conditions with or without CQ, indicating α-syn was not subjected to degradation. Interestingly, when cells were starved with HBSS for 2 h, a small fraction of α-syn was detected in the pellet fractions of both cell lines. This suggests α-syn can be degraded in the lysosomes when autophagy is induced. This is consistent with previous findings that α-syn can be degraded in the lysosomes via macroautophagy or the chaperone-mediated autophagy pathway [41][42][43] . Furthermore, there was no difference between WT and CLN5 KD cells in α-syn subcellular distribution, suggesting that increased α-syn in CLN5 deficient cells was not because of dysfunctional lysosomes.
Alpha-synuclein and lysosomal clustering. The lysosomal localization of α-syn we observed reveals a possible function for α-syn in the lysosomal membrane. α-syn has a lipid binding property [49][50][51] and has been shown to cause clustering of synaptic vesicles 52 . Therefore, we examined whether α-syn has any role in the lysosomal membrane in CLN5-deficient cells where SNCA is upregulated. Under normal growing conditions, the lysosomes are distributed throughout the cytoplasm. However, in some LSDs the lysosomes become perinuclear 53,54 .
In CLN5 KD cells, the lysosomes were clustered to the perinuclear region as well (Fig. 5A). Remarkably, when we knocked down SNCA in CLN5 KD cells, the lysosomes were redistributed to the periphery of the cell (Fig. 5A). Figure 5A α-syn panel and 5B show efficient knockdown of SNCA in CLN5 KD cells. The quantification analysis showed that, in CLN5 KD cells, the lysosomal distribution area returned to the WT level when α-syn was reduced by siRNA (Fig. 5C). This suggests that α-syn plays a role in lysosomal clustering, a property similar to that found in synaptic vesicles 52 . We also examined the LC3 and P62 levels when we knocked down SNCA in CLN5 KD cells (Fig. 5D). Unlike its effect on lysosome position, transient SNCA knockdown did not change LC3-II in CLN5 KD cells back to the levels in WT or SNCA knockout cells. This suggests that the increase of autophagy flux in CLN5 deficient cells is not due to SNCA up-regulation.

Discussion
Differential expression of α-syn has been shown in animal models of several NCL subtypes 47,[55][56][57][58] . For example, accumulated α-syn has been found in CLN12 disease Atp13a2 −/− mouse model and Atp13a2 knockdown mouse primary neurons 47,55 . On the other hand, a lower level of α-syn protein was found in the motor cortex of a naturally occurring CLN5-deficient sheep 56 . In addition, some mouse model studies have examined Snca gene expression. Snca expression was found to be substantially reduced in the cortex of Cln1/Cln5 double knockout mice 57 , but not in Cln1 or Cln5 single knockout mice 57,59 . Yet in another study, Snca expression was highly up-regulated in the embryonic cortical neurons in Cln1 knockout mice 58 . This highlights the complexity and variability of the regional and likely age-relevant expression.
In this study, we showed increased autophagy flux in CLN5-deficient cells by employing different measurements to monitor autophagy flux. In particular, the tandem fluorescent mRFP-GFP-LC3 assay demonstrated a higher basal autophagy flux in CLN5-deficient cells. The level of autophagy flux was similar to that of starvation induced autophagy in control cells. Since CLN5 disease is a lysosomal storage disorder, it is reasonable to assume increased autophagy is a compensatory effect in CLN5-deficient cells. Our data showed that lysosomal degradation of P62 was normal in CLN5-deficient cells, suggesting that CLN5 deficiency disrupts lysosome function other than lysosomal proteolysis.
Using PCR array of the autophagy pathway to analyze differential expression, we identified the SNCA gene as one of the autophagy-related genes up-regulated in CLN5 disease patient cells. Whether and how up-regulation of SNCA affects the autophagy process in CLN5-deficient cells remain to be determined. However, transient knocking down SNCA by siRNA in CLN5 KD cells did not reduce the LC3-II back to the level in WT cells, suggesting an indirect role of α-syn in regulating this process. Furthermore, while we observed up-regulation of the SNCA gene in CLN5 disease patient fibroblasts and stable CLN5 KD cells, transient CLN5 KD by siRNA in either fibroblasts or SH-SY5Y cells showed no up-regulation. These results suggest that SNCA up-regulation is a consequence of long-term CLN5 deficiency. A stable knockdown or knockout of CLN5 in SH-SY5Y cells or induced pluripotent stem cell (iPSC)-differentiated neuronal cells will help to elucidate whether SNCA up-regulation has a physiological relevance in NCL diseases.
The potential role of α-syn in lysosomal clustering we observed suggests a possible function of α-syn exists in non-neuronal cells. Using in vitro assays, it was shown that by binding to anionic lipid and synaptobrevin-2, a synaptic vesicle SNARE protein, α-syn caused clustering of the synaptobrevin-2 containing vesicles 25 . The clustering property of α-syn demonstrated herein is in agreement with the previous literature. For example, overexpression of α-syn inhibits synaptic vesicle release, possibly due to a reduced pool of readily releasable synaptic vesicles 60,61 . Similarly, deletion of α-syn increases neurotransmitter release 62 . It will be interesting to examine whether α-syn binds to any lysosomal SNARE proteins and whether the clustering effects of elevated levels of α-syn relate to the lysosome function in CLN5-deficient NCL and other NCLs in general.
Besides α-syn, defects in several proteins involved in endo-lysosomal trafficking, such as retromer subunit VPS35, leucine-rich repeat kinase 2 (LRRK2), Rab7L1, and Rab39B, are known to cause Parkinson's disease [63][64][65][66][67][68] . Supporting a role of α-syn in the endo-lysosome pathway, it has been documented that in cortical neurons α-syn localizes to the vicinity of endocytic compartments and interacts directly with VPS29, SNX1 (two retromer subunits), and Rab4A (an endosomal Rab GTPase) 69 . Whether our finding of α-syn affecting lysosome positioning in HeLa cells is physiologically relevant in neurons and contributes to pathogenicity of Parkinson's disease will need to be further examined.

Materials and Methods
Cell culture and transfections. Cell culture media and reagents were purchased from Gibco and Hyclone. Sample preparation and Immunoblotting. Cells grown on culture dishes were scraped and washed once with 1X phosphate buffered saline (PBS), and centrifuged for 3 min at 1,500 × g. Cell pellets were lysed using RIPA lysis buffer (50 mM Tris pH 8.0, 150 mM NaCl, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS) supplemented with protease inhibitor cocktail (G-Biosciences). After incubation for 30 min on ice, extracts were centrifuged at 20,000 × g for 10 min at 4 °C. The supernatant was collected as the whole cell lysates. Protein concentrations were determined by Bradford assay. Protein samples were separated by SDS-PAGE and transferred to PVDF membranes (Millipore) followed by immunoblotting. For the membranes that were prepared to blot for α-syn, after transfer, the membranes were first fixed with 1% formaldehyde in PBS for 30 min followed by blocking in 5% non-fat milk in TBST. ECL detection was performed according to the manufacturer's instructions (Millipore) and blots were imaged using G-Box (Syngene). Quantification was performed using ImageJ.
Immunofluorescence Microscopy. Cells on coverslips were fixed with 4% formaldehyde for 10 min at room temperature or 100% methanol for 15 min at −20 °C. Blocking, permeabilization, antibody incubations, and washes were done using blocking solution (10% fetal calf serum, 0.1% saponin, and 0.02% sodium azide in PBS). The cells were imaged using either Zeiss LSM-5 PASCAL or LSM-700 laser scanning confocal microscope. Quantification was performed using ImageJ. For comparison between groups, data was analyzed using a Student's t test with two-samples assuming equal variances. For comparison among multiple groups, a one-way ANOVA followed by the Tukey's post hoc test was performed.

Subcellular fractionation.
After treatments, cell pellets were collected, resuspended in isotonic buffer (10 mM HEPES, 10 mM KCl, 1 mM EDTA, 250 mM sucrose, 1 mM DTT, protease inhibitor cocktail), and incubated 10 min on ice. Samples were passed through a 25 G needle with 14 strokes. The post nuclear supernatant (PNS) was collected after centrifugation at 600 × g for 10 min at 4 °C. The PNS was centrifuged at 20,000 × g for 20 min at 4 °C. The supernatant was collected as the "soluble" fraction. The pellet was rinsed once with isotonic buffer and centrifuged again at 20,000 × g for 5 min at 4 °C. The pellet was resuspended in RIPA buffer as the "pellet" fraction. This is the crude membrane fraction that contains organelles (except nucleus and microsomes).

Data Availability
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.