Prions amplify through degradation of the VPS10P sorting receptor sortilin

Prion diseases are a group of fatal neurodegenerative disorders caused by prions, which consist mainly of the abnormally folded isoform of prion protein, PrPSc. A pivotal pathogenic event in prion disease is progressive accumulation of prions, or PrPSc, in brains through constitutive conformational conversion of the cellular prion protein, PrPC, into PrPSc. However, the cellular mechanism by which PrPSc is progressively accumulated in prion-infected neurons remains unknown. Here, we show that PrPSc is progressively accumulated in prion-infected cells through degradation of the VPS10P sorting receptor sortilin. We first show that sortilin interacts with PrPC and PrPSc and sorts them to lysosomes for degradation. Consistently, sortilin-knockdown increased PrPSc accumulation in prion-infected cells. In contrast, overexpression of sortilin reduced PrPSc accumulation in prion-infected cells. These results indicate that sortilin negatively regulates PrPSc accumulation in prion-infected cells. The negative role of sortilin in PrPSc accumulation was further confirmed in sortilin-knockout mice infected with prions. The infected mice had accelerated prion disease with early accumulation of PrPSc in their brains. Interestingly, sortilin was reduced in prion-infected cells and mouse brains. Treatment of prion-infected cells with lysosomal inhibitors, but not proteasomal inhibitors, increased the levels of sortilin. Moreover, sortilin was reduced following PrPSc becoming detectable in cells after infection with prions. These results indicate that PrPSc accumulation stimulates sortilin degradation in lysosomes. Taken together, these results show that PrPSc accumulation of itself could impair the sortilin-mediated sorting of PrPC and PrPSc to lysosomes for degradation by stimulating lysosomal degradation of sortilin, eventually leading to progressive accumulation of PrPSc in prion-infected cells.


Introduction
Prion diseases are a group of fatal neurodegenerative disorders, which include Creutzfeldt-Jakob disease in humans and bovine spongiform encephalopathy and scrapie in animals [1]. They are caused by the infectious agents termed prions, which mainly consist of the abnormally folded, amyloidogenic isoform of prion protein, designated PrP Sc . PrP Sc is a β-sheet-rich conformer produced by conformational conversion of the cellular counterpart, PrP C [1]. Intermolecular interaction between PrP C and PrP Sc is essential for the conversion of PrP C into PrP Sc . We and others have shown that mice devoid of PrP C neither developed the disease nor accumulated PrP Sc even after prions were inoculated into their brains [2][3][4][5]. These results indicate that the conversion of PrP C into PrP Sc plays a pivotal role in the pathogenesis of prion disease, and that depletion of PrP C could be therapeutic by preventing the production of PrP Sc .
PrP C is normally located at the cell surface as a glycosylphosphatidylinositol (GPI)anchored glycoprotein [6]. Some endocytosed PrP C molecules are transported to lysosomes for degradation while others are recycled to the cell surface through the endocytic recycling compartments [7]. PrP Sc is also trafficked to lysosomes for degradation [7]. However, the cellular transport mechanism of PrP C and PrP Sc to lysosomes remains unknown. Whether prion infection or PrP Sc impairs the lysosomal trafficking of PrP C and PrP Sc for its progressive propagation is also unknown.
The vacuolar protein sorting-10 protein (VPS10P)-domain receptors, including sortilin, SorLA, SorCS1, SorCS2 and SorCS3, are multi-ligand type I transmembrane proteins abundantly expressed in the brain and involved in neuronal function and viability [8,9]. They function as a cargo receptor to deliver a number of cargo proteins to their subcellular destination through the VPS10P domain in the extracellular luminal N-terminus. Sortilin traffics the amyloid precursor protein (APP)-cleaving enzyme BACE1 [10] and the neurotrophic factor receptors Trks [11]. SorLA directs trafficking of APP into the recycling pathway [12]. SorCS1 also mediates APP transport [13]. Recent lines of evidence indicate that the altered VPS10P receptor-mediated trafficking could be involved in the pathogenesis of neurodegenerative disorders, including Alzheimer's disease (AD) [12][13][14][15] and frontotemporal lobar degeneration (FTLD) [16]. However, the role of VPS10P receptors in the trafficking of PrP C or PrP Sc and in the pathogenesis of prion disease is little known.
In the present study, we show that sortilin has an inhibitory role in PrP Sc accumulation by sorting PrP C and PrP Sc to lysosomes for degradation. Interestingly, however, prion infection stimulates lysosomal degradation of sortilin, indicating that prion infection itself could disturb the inhibitory function of sortilin. We also confirm that sortilin-knockout (KO) mice have accelerated prion disease after infection with RML prions, with early accumulation of PrP Sc in their brains. These results suggest that PrP Sc accumulation may be amplified through PrP Scinduced impairment of the sortilin-mediated lysosomal degradation of PrP C and PrP Sc .

Results
Sortilin is a novel PrP C -binding protein regulating the surface levels of PrP C To investigate the role of VPS10P cargo receptors in the trafficking of PrP C , we first examined whether or not VPS10P molecules could interact with PrP C . Co-immunoprecipitation assay in neuroblastoma N2aC24 cells showed that SAF61 anti-PrP antibody (Ab) precipitated PrP C with sortilin, but not with other VPS10P molecules (Fig 1A, S1 Fig). PrP C was also co-precipitated with sortilin by anti-sortilin Abs (Fig 1B). GST-pulldown assay using purified recombinant proteins revealed that the VPS10P domain of sortilin fused with GST (GST-VPS10P) successfully pulled down His-tagged full-length recombinant PrP, but not PrP with a deletion of 23-88 residues (Fig 1C), suggesting that the residues 23-88 are important for PrP C to interact with sortilin. SAF61 anti-PrP Ab also co-precipitated full-length mycHis-tagged sortilin expressed in sortilin-KO N2aC24 cells, designated ΔSort#1 cells, but not in PrP-KO N2a cells, N2aΔPrP#1 cells (S2A and S2B Fig). Both types of KO cells were established using the CRISPR-Cas genome editing system. This clearly indicates that PrP C expression is required for sortilin to be co-precipitated by SAF61 anti-PrP Ab, further supporting the interaction of sortilin and PrP C . However, sortilin lacking residues 610-753 was not efficiently co-precipitated with the Ab, compared to other deletion mutants of sortilin (S2A and S2B Fig), suggesting that the residues 610-753 of sortilin are involved in interaction with PrP C . Furthermore, co-immunoprecipitation assay using mouse brain homogenates also revealed an interaction between PrP C and sortilin ( Fig 1D). Immunofluorescence staining of non-permeabilized N2aC24 cells showed co-localization of sortilin and PrP C on the cell surface (Fig 1E and 1F We then knocked down sortilin in N2aC24 cells using two sortilin-specific siRNAs, termed siRNA#1 and #2. Immunostaining of sortilin-knockdown (Sort-KD) cells showed an increase in PrP C expression on the cell surface (Fig 2A). Biotin labeling of surface proteins confirmed the increased surface levels of PrP C in Sort-KD cells (Fig 2B and 2C). Total PrP C levels were also increased in Sort-KD cells (Fig 2B and 2C). However, intracellular PrP C was not increased in Sort-KD cells (Fig 2D), indicating that the surface PrP C is specifically increased in Sort-KD cells. PrP mRNA levels were not increased in Sort-KD cells (Fig 2E), suggesting that the increased surface expression of PrP C might be attributable to the impaired degradation of PrP C in Sort-KD cells. PrP C levels were also significantly increased in the brains of sortilin-KO (Sort1 -/-) mice compared to those in wild-type (WT) mice (S5A and S5B Fig).
PrP C undergoes an endopeptidic cleavage by the ADAM family of metalloproteases, with the C-terminal fragment, designated the C1 fragment, being produced [17,18]. Sort-KD cells . Arrows and arrowheads indicate a nonspecific signal of the degraded fragment of protein G or the light chain of antibodies used in co-immunoprecipitation. Percentages indicate the proportion of the sample loaded. (C) GST-pulldown assay using GST-tagged VPS10P domain of sortilin and His-tagged full-length recombinant PrP or His-tagged recombinant PrP lacking residues 23-88. Percentages indicate the proportion of the sample loaded. (D) Co-immunoprecipitation assay in mouse brain homogenate with SAF61 anti-PrP Ab. Percentages indicate the proportion of the sample loaded. Arrowheads indicate a non-specific signal of the degraded fragment of protein G. (E) Double immunofluorescence staining of PrP C (green) and sortilin (red) in permeabilized or nonpermeabilized N2aC24 cells, with SAF83 anti-PrP Ab and goat polyclonal anti-sortilin Abs. Arrowheads indicate co-localized signals of PrP C and sortilin. Bar, 5 μm. (F) Person's correlation coefficient for cell surface and intracellular co-localization of PrP C and sortilin. produced the C1 fragment more abundantly than N2aC24 cells (S6A Fig). This is probably because PrP C was increased on the cell surface in Sort-KD cells. We also investigated PrP C levels in exosomes of N2aC24 and sortilin-KO ΔSort#1 and ΔSort#2 cells. ΔSort#1 and ΔSort#2 cells also showed an increase in total PrP C levels (S6B Fig). PrP C was significantly higher in exosomes from ΔSort#1 and #2 cells than in those from N2aC24 cells (S6B and S6C Fig). Exosomes were verified by the presence of exosome-specific molecules TSG101 and flotillin and the absence of GM130 and Bcl-2, both of which are not normally included in exosomes (S6B Fig) [19,20].

Sortilin sorts surface PrP C to late endosomes/lysosomes
To address whether or not sortilin could sort surface PrP C to lysosomes for degradation, we first investigated the role of sortilin in internalization of surface PrP C using an Ab-labeling technique. Surface PrP C was labeled with SAF61 anti-PrP Ab at 4˚C in which internalization of membrane proteins is inhibited, and then allowed to be internalized for 2 h at 37˚C. The labeled PrP C was then detected using Alexa Fluoro 488-conjugated anti-mouse IgG Abs. The internalization of the labeled PrP C was slightly but significantly inhibited in Sort-KD cells, compared to that in control N2aC24 cells (Fig 3A and 3B), suggesting that sortilin could be involved in internalization of some portions of PrP C . To further confirm the involvement of sortilin in internalization of PrP C , we biotinylated the surface proteins of N2aC24 cells and sortilin-KO (ΔSort) cells #1 with sulfo-NHS-SS-biotin, whose biotin motif can be removed by reducing agents. We then allowed the biotinylated proteins to be internalized for 2 h, and treated the cells with the membrane-impermeable reducing agent glutathione to remove the biotins only from surface proteins but not from those internalized. The treated cells were lysed, and then biotin-labeled, internalized proteins were purified using avidin-beads, and investigated for internalized PrP C by Western blotting with 6D11 anti-PrP Ab. Strong signals corresponding to the internalized PrP C were detected in N2aC24 cells (Fig 3C and 3D). However, the signals were significantly reduced in ΔSort#1 cells (Fig 3C and 3D). These results reinforce the role of sortilin in internalization of PrP C .
To track the SAF61 anti-PrP Ab-labeled, internalized PrP C in N2aC24 and Sort-KD cells, we immunofluorescently stained both types of cells for internalized PrP C with the late endosome marker Rab9 or the recycling endosome marker Rab11. The labeled PrP C was normally internalized to both late endosomes and recycling endosomes, as observed in N2aC24 cells These results indicate that internalization of PrP C to the recycling endosomes could be independent of sortilin. Furthermore, sortilin could function to sort surface PrP C to the late endosome/lysosome degradation pathway, thereby regulating levels of surface PrP C . Consistent with the results from Sort-KD cells, sortilin-KO ΔSort#1 cells showed higher expression of PrP C than control N2aC24 cells (Fig 3G and 3H). Inhibition of lysosomal enzymes by NH 4 Cl increased PrP C markedly in N2aC24 cells, but only slightly in ΔSort#1 cells (Fig 3G and 3H). PrP C was detected in the LAMP1-positive lysosomes in both cell types after NH 4 Cl treatment (Fig 3I and 3J). However, its lysosomal localization was much less in ΔSort#1 cells than in N2aC24 cells (Fig 3I and 3J). These results confirm that transport of PrP C to lysosomes is disturbed in sortilin-deficient cells, therefore reducing the localization of PrP C in lysosomes and resulting in an increase in PrP C levels in sortilin-deficient cells.
We also investigated localization of PrP C in early endosomes in N2aC24 and Sort-KD cells. PrP C was labeled with SAF61 Ab and spontaneously internalized for 1 h instead of 2 h, which was utilized for detection of internalized PrP C in late endosomes or recycling endosomes, We also investigated the internalization of PrP with a deletion of 23-88 residues, PrPΔ23-88, to lysosomes by establishing PrP-KO N2aΔPrP cells permanently expressing wild-type (WT) mouse PrP C (WT#1 and #2) or PrPΔ23-88 (Δ23-88#1 and #2). Western blotting showed that PrPΔ23-88 was expressed at higher levels than WT PrP C without NH 4  It is thus conceivable that the interaction with sortilin could be important for PrP C transportation to lysosomes for degradation. However, the increased localization of PrPΔ23-88 in lysosomes after NH 4 Cl treatment in Δ23-88#1 and #2 cells suggests that other molecules are also involved in trafficking of PrP C to lysosomes.

PrP C shifts to raft domains in sortilin-KO cells
To gain insight into the mechanism of sortilin-mediated sorting of PrP C to lysosomes, we investigated membrane microdomain localization of PrP C in N2aC24 and sortilin-KO N2aC24 cells, ΔSort#1 and ΔSort#2 cells, by detergent-based biochemical membrane fractionation. PrP C was detected in both detergent-resistant membrane (DRM) and detergent-soluble membrane fractions, that is raft and non-raft fractions, in N2aC24 cells, with higher amounts of PrP C in raft fractions (63.0%) than in non-raft fractions (37.0%) (Fig 4A and 4B). However, in ΔSort#1 and #2 cells, PrP C in non-raft fractions was reduced to 11.9 and 14.9%, respectively (Fig 4A and 4B). Instead, PrP C was increased in raft fractions (Fig 4A and 4B). These results suggest that sortilin could shift the localization of PrP C to non-raft domains from raft domains. Raft-resident protein flotillin-2 was observed in raft fractions in N2aC24 and ΔSort cells ( Fig  4C), ruling out the possibility that lack of sortilin could affect the membrane microdomain integrity leading to the shift in the location of surface PrP C in ΔSort cells. Sortilin was predominantly detected in non-raft fractions in N2aC24 and ΔSort cells (Fig 4D), indicating that purified using avidin-beads and subjected to Western blotting with 6D11 anti-PrP Ab. (D) Signal intensities in each lane in Sort-KD cells were evaluated against that of total PrP signals in N2aC24 cells. Data are means ± SD of 3 independent experiments. * p < 0.05, ** p < 0.01, *** p < 0.001. (E, F) Double immunofluorescence staining of the internalized, SAF61-labeled PrP C (green) and the late endosome marker Rab9 (red) or the recycling endosome marker Rab11 (red) in permeabilized N2aC24 cells transfected with control or sortilin-specific siRNAs. Bar, 10 μm. (G) Western blotting for PrP C and sortilin in N2aC24 and sortilin-KO (ΔSort#1) cells after 12 h-treatment with or without 20 mM NH 4 Cl. (H) Quantification of PrP C in (G) after normalization against β-actin. Signal intensity in NH 4 Cl-treated cells was evaluated against that in NH 4 Cl-untreated N2aC24 cells. Data are means ± SD of 3 independent experiments. *** p < 0.001. (I) Double immunofluorescence staining of PrP C and the lysosome marker LAMP1 in N2aC24 and sortilin-KO (ΔSort#1) cells after 12 h-treatment with 20 mM NH 4 Cl. Bar, 10 μm. (J) Pearson's correlation coefficient for co-localization of PrP C and LAMP1 in N2aC24 cells treated (n = 106) and untreated (n = 151) with NH 4 Cl and in ΔSort#1 cells treated (n = 140) and untreated (n = 166) with NH 4 Cl. Data are means ± SD. *** p < 0.001. These results indicate that residues 23-88 are important for PrP C to be retained at non-raft domains. We then investigated membrane microdomain localization of PrP C in the brains of Sort1 -/and WT mice. Localization of PrP C at raft and non-raft fractions was observed in WT brains (S9C and S9D Fig). However, PrP C was increased in raft fractions in Sort1 -/brains (S9C and S9D Fig). Taken together, these results suggest that sortilin might function to retain surface PrP C in non-raft domains and sort the non-raft PrP C to the late endosome/lysosome degradation pathway through interaction with residues 23-88 of PrP C . We also investigated membrane microdomain location of PrP molecules in prion-infected N2aC24L1-3 cells. In contrast to PrP C detected in raft and non-raft domains in uninfected N2aC24 cells (Fig 4A and 4B), total PrP molecules and PK-resistant PrP Sc were predominantly detected in raft fractions in prion-infected N2aC24L1-3 cells (S10A and S10B Fig). These results suggest that prion infection accumulates PrP Sc and PrP C in raft domains.

Sortilin interacts with PrP Sc and facilitates its degradation
We also assessed the role of sortilin in the degradation of PrP Sc . Protein interaction assay using protein-G-conjugated magnetic beads in prion-infected N2aC24L1-3 cells showed that PrP Sc and sortilin were co-collected by anti-sortilin Abs (Fig 5A), suggesting that sortilin interacts with PrP Sc . siRNA-mediated knockdown of sortilin increased PrP Sc in N2aC24L1-3 cells ( Fig  5B and 5C). In contrast, overexpression of sortilin reduced PrP Sc in N2aC24L1-3 cells (Fig 5D  and 5E). These results indicate that sortilin could also be involved in PrP Sc degradation.
We then evaluated the degradation kinetics of PrP Sc in prion-infected cells with or without sortilin. For this study, it is important to prevent the de novo production of PrP Sc from PrP C . PrP siRNA#1 and #2 reduced PrP C in N2aC24 and ΔSort#1 cells from 24 h after transfection, to less than 10% of that in control siRNA-transfected N2aC24 and ΔSort#1 cells ( Fig 6A).
These results indicate that the de novo production of PrP Sc from PrP C could be negligible from 24 h after transfection with PrP siRNAs in these cells even after infection with prions. We thus investigated PrP Sc in RML-infected N2aC24 (N2aC24/RML) and ΔSort#1 (ΔSort/RML) cells at 36, 48 and 60 h after transfection with PrP siRNAs. Control siRNA did not affect PrP Sc levels in these cells (S11A and S11B Fig). However, PrP Sc was decreased after transfection with PrP siRNAs (Fig 6B). In N2aC24/RML cells 60 h after transfection with PrP siRNAs, PrP Sc was reduced to less than 20% of that in control siRNA-transfected N2aC24/RML cells (siRNA#1, 16.5%; siRNA#2, 17.1%) (Fig 6B and 6C). However, significantly higher levels of PrP Sc were still observed in ΔSort/RML cells 60 h after transfection with PrP siRNAs (siRNA#1, 45.2%; siRNA#2, 38.5%) (Fig 6B and 6C). Similar results were obtained in 22L prion-infected N2aC24 and ΔSort#1 cells (S11C and S11D Fig). These results indicate that sortilin is also involved in the degradation of PrP Sc .

Prion infection cell-autonomously reduces sortilin after PrP Sc production
We then asked if prion infection could affect sortilin. Interestingly, sortilin was reduced in N2aC24L1-3 cells to 52% of that in N2aC24 cells (Fig 7A and 7B). However, other VPS10P molecules were not reduced (S12A and S12B Fig). The reduced levels of sortilin were recovered in cured N2aC24L1-3 cells, which were cured from prion infection after treatment with SAF32 anti-PrP Ab, to that in N2aC24 cells (Fig 7A and 7B). Sortilin mRNA was similarly expressed between N2aC24 and N2aC24L1-3 cells (S12C Fig). ScN2a cells, N2a cells persistently infected with RML prions, also expressed sortilin less than N2a cells (S12D and S12E Fig). Moreover, sortilin was reduced in the brains of terminally ill mice infected with RML and 22L prions to 46.3 and 45.6%, respectively, compared to uninfected mouse brains (Fig 7C and  7D). These results indicate that prion infection could reduce sortilin.
We then treated uninfected N2aC24 and infected N2aC24L1-3 cells with inhibitors to lysosomes (NH 4 Cl and concanamycin A) or proteasomes (MG132). Treatment with NH 4 Cl or concanamycin A increased sortilin in uninfected cells (S13A-S13D Fig). However, sortilin was much more increased in infected cells after treatment with NH 4 Cl and concanamycin A (S13A-S13D Fig). In contrast, MG132 did not affect sortilin levels (S13A and S13B Fig). These results suggest that sortilin could be degraded in lysosomes, and that the lysosomal degradation of sortilin could be stimulated in prion-infected cells.   We also monitored the levels of sortilin in N2aC24 cells freshly infected with RML prions. There was no significant decrease in sortilin by 3 days post-infection (dpi) while PrP Sc was obviously detectable (Fig 7E and 7F). Sortilin was decreased at 6 dpi (Fig 7E and 7F). These results suggest that prion infection reduces sortilin, and that the sortilin reduction is preceded by PrP Sc production. We also performed double immunofluorescence staining for sortilin and PrP Sc in freshly infected N2aC24 cells at 9 dpi. PrP Sc was specifically stained using the mAb132, which was demonstrated to specifically visualize PrP Sc under partially denatured conditions [21]. Since the subcellular positions of sortilin and PrP Sc might differ vertically in infected cells, 6 horizontally serial images at 1 μm interval were used to detect sortilin and PrP Sc . In cells displaying green fluorescence for PrP Sc , little or no sortilin (red fluorescence) was detected in any slices ( Fig 7G). In contrast, bright red fluorescence for sortilin was observed only in the cells negative for PrP Sc (Fig 7G). These results indicate that prion infection could reduce sortilin in a cell-autonomous fashion after PrP Sc accumulation.
Sortilin is known to interact with and transport Trk receptors to the cell surface, thereby enhancing nerve growth factor (NGF) signaling leading to activation of MAP kinases [11]. To investigate whether or not prion infection could affect the function of sortilin, we stimulated uninfected N2aC24 and prion-infected N2aC24L1-3 cells with NGF. Phosphorylated ERK1/2 was increased in N2aC24 and N2aC24L1-3 cells after stimulation (S14A-S14C Fig). However, the levels of phosphorylated ERK1/2 were significantly lower in N2aC24L1-3 cells than in N2aC24 cells (S14A-S14C Fig). These results indicate that NGF signaling is impaired in N2aC24L1-3 cells, suggesting that sortilin might be functionally disturbed in prion-infected cells.

Prion disease is aggravated in sortilin-KO mice after infection with prions
We then evaluated the effects of sortilin deficiency on the pathogenesis of prion disease using sortilin-KO (Sort1 -/-) mice. Sort1 -/mice were viable and fertile and showed no gross abnormalities [11,22]. Sort1 +/+ (n = 19) and Sort1 -/female mice (n = 24) were intracerebrally inoculated with RML prions. Incubation and survival times were significantly shortened in Sort1 -/mice (Fig 8A, S1 Table). Sort1 -/and Sort1 +/+ mice developed symptoms at 150.9 ± 7.8 and 171.9 ± 6.0 dpi, respectively (Fig 8A, S1 Table). Western blotting also showed earlier accumulation of PrP Sc in the brains of infected Sort1 -/mice. PrP Sc was scarcely detectable in the brains of Sort1 +/+ mice at 45 dpi (Fig 8B and 8C). However, it was obvious in Sort -/mice at 45 dpi (Fig 8B and 8C). PrP Sc levels were still significantly higher in Sort1 -/mice than in Sort1 +/+ mice at 60 and 90 dpi (Fig 8B and 8C). However, no difference in the levels of PrP Sc was observed between Sort1 +/+ and Sort1 -/mice at terminal stage (Fig 8B and 8C). Immunohistochemical analysis of the brains of infected Sort1 +/+ and Sort1 -/mice for PrP Sc showed consistent results. PrP Sc was detectable in a much larger area of the brains of Sort1 -/mice, compared to that in Sort1 +/+ mice, at 60 dpi ( Fig 8D). However, PrP Sc became indistinguishably accumulated throughout the brains of Sort1 -/and Sort1 +/+ mice at terminal stage ( Fig 8D). Similar results were also obtained with Sort1 -/and Sort1 +/+ male mice inoculated with RML prions (S15A-S15D Fig, S2 Table). These results show that sortilin deficiency accelerates prion disease by causing early accumulation of PrP Sc in the brains of mice after infection with prions, reinforcing the inhibitory role of sortilin in the pathogenesis of prion disease. cells was evaluated against that in uninfected cells. Data are means ± SD of 3 independent experiments. n.s., not significant; * p < 0.05. (G) Double immunofluorescent staining of sortilin (red) and PrP Sc (green) in N2aC24 cells 9 days after being freshly infected with RML prions. DAPI was used for nuclear stain (blue). Six serial vertical images of the cells with 1 μm intervals are shown. Arrows indicate PrP Sc -positive cells. Bar, 10 μm. https://doi.org/10.1371/journal.ppat.1006470.g007

Discussion
In the present study, we showed that PrP Sc accumulation could be enhanced through PrP Scstimulated degradation of sortilin, a member of the VPS10P sorting receptor family. Sortilin functions as a negative regulator for PrP Sc accumulation by sorting PrP C and PrP Sc to the late endosome/lysosome protein degradation pathway. However, PrP Sc stimulates sortilin to be degraded in lysosomes, thereby disturbing the inhibitory role of sortilin and eventually leading to the further accumulation of PrP Sc .
PrP C is a GPI-anchored membrane protein located in raft domains and, to a lesser extent, in non-raft domains. Some of the PrP C molecules internalized are delivered back to the cell surface directly or indirectly via the recycling endosome compartments and others are transported to lysosomes for degradation [7] (Fig 9A). We showed that sortilin could directly interact with PrP C on the cell surface through the VPS10P domain through the residues 610-753 encompassing cysteine rich 10CCs [23] of sortilin and the N-terminal residues 23-88 of PrP C . Sortilin-knockdown increased PrP C on the cell surface and reduced the localization of PrP C to lysosomes, suggesting that the increased surface expression of PrP C in sortilin-knockdown cells might be caused by the decreased trafficking of PrP C to lysosomes for degradation. Sortilin was predominantly located in non-raft domains, and PrP C accumulated at raft domains and decreased in non-raft domains in sortilin-KO cells. It is thus conceivable that sortilin could function to retain PrP C in non-raft domains and be involved in trafficking of non-raft surface PrP C to lysosomes for degradation (Fig 9A). Low-density lipoprotein receptor-related protein 1 has also been reported as a candidate cargo receptor for the non-raft PrP C [24]. A sortilin-independent pathway may also play a role in PrP C internalization (Fig 9A). We showed that sortilin deficiency increased PrP C at raft domains and shifted PrP C internalization into the recycling endosomes from lysosomes in cells. It is thus conceivable that the internalization of raft PrP C to the recycling endosomes could be sortilin-independent (Fig 9A). Taken together, these results suggest that PrP C located in non-raft domains could be internalized to lysosomes for degradation partly via the sortilin-dependent pathway while an internalization pathway to direct PrP C from raft domains to the endocytic recycling pathway could be sortilin-independent ( Fig 9A).
Sortilin-knockdown increased PrP Sc in prion-infected cells. In contrast, overexpression of sortilin decreased PrP Sc . Moreover, sortilin-KO mice developed the disease earlier than wildtype mice after intracerebral inoculation with RML prions, with earlier accumulation of PrP Sc in their brains. These results indicate that sortilin is a negative regulator for PrP Sc accumulation. Raft domains may be a site for the conversion of PrP C into PrP Sc [25], although the exact site of PrP Sc production remains controversial. Sortilin could retain surface PrP C at non-raft domains and transport it to lysosomes for degradation, thereby reducing PrP C located in raft domains. It is thus possible that the reduction of PrP C in raft domains by sortilin could delay the conversion of PrP C into PrP Sc , eventually leading to less accumulation of PrP Sc . Sortilin also interacts with PrP Sc . Kinetics studies for PrP Sc showed that knockout of sortilin delayed PrP Sc clearance in prion-infected cells. Thus, sortilin also could function to sort PrP Sc for degradation, reducing PrP Sc accumulation. Some PrP Sc molecules are trafficked to lysosomes for degradation via the endolysosomal pathway from the cell surface [21,26,27]. Others are retrogradely transported to the Golgi apparatus where they are subjected to Golgi quality control and trafficked to lysosomes for degradation [28]. Sortilin is expressed on the cell surface and removed from PBS-inoculated Sort1 -/mice aged 16, 20, and 30 weeks were used as controls for 60 and 90 dpi and terminal stage, respectively. Bar, 300 μm. https://doi.org/10.1371/journal.ppat.1006470.g008 Prions amplify via sortilin degradation the Golgi apparatus [29]. Therefore, sortilin might be involved in both degradation trafficking pathways of PrP Sc . However, a large portion of sortilin and PrP Sc molecules differed in their membrane microdomain localization on the cell surface. Sortilin was predominantly detected in non-raft domains. In contrast, PrP Sc was exclusively located in raft domains. It is thus likely that the sortilin-mediated lysosomal degradation of PrP Sc from the cell surface through their direct interaction might be, if any, a minor event. We previously reported that PrP Sc was abundantly detected in the recycling endosomes of prion-infected cells, suggesting that PrP Sc molecules accumulated in the recycling endosomes also might not be directly affected by sortilin.
Sortilin was reduced in both prion-infected cultured cells and mouse brains. Sortilin mRNA was not decreased in prion-infected cells. Reduction of sortilin in prion-infected cells was recovered by treatment with lysosomal inhibitors but not proteasomal inhibitor, suggesting that prion infection could stimulate degradation of sortilin in lysosomes. Immunofluorescent staining of freshly infected cells revealed that sortilin was barely detectable in PrP Scpositive cells but abundant in PrP Sc -negative cells. PrP Sc accumulation preceded the reduction of sortilin. These results suggest that PrP Sc accumulated after prion infection could cause sortilin degradation in lysosomes in a cell-autonomous fashion, and that the enhanced degradation of sortilin could abrogate the negative role of sortilin in PrP Sc accumulation in prion-infected cells. Thus, sortilin-mediated sorting of PrP C and PrP Sc to lysosomes for degradation could be disturbed in prion-infected cells, causing an increase in PrP C and PrP Sc and eventually leading to progressive accumulation of PrP Sc (Fig 9B). In prion-infected cells, due to the disturbed function of sortilin, PrP C might also be increasingly located in raft domains, where PrP C is postulated to efficiently convert into PrP Sc [25], and might be internalized into the recycling endosomes, where PrP Sc was reported to be abundantly detectable [26] (Fig 9B). Increased localization of PrP C in raft domains and in the recycling endosomes in prion-infected cells also could contribute to progressive accumulation of PrP Sc by increasing the conversion of PrP C into PrP Sc . Elucidation of the mechanism by which PrP Sc accumulation stimulates the lysosomal degradation of sortilin would be helpful to further understanding of the mechanism for the progressive accumulation of PrP Sc .
Sortilin also regulates neuronal cell viability by controlling the release of pro-and maturedform of neurotrophins (NTs), such as NGF and brain-derived neurotrophic factor, as does the transport of their receptors, TrkA, TrkB, and TrkC, to the plasma membrane [9]. No gross abnormalities were reported in Sort1 -/- [11,22], probably due to compensatory mechanisms of other family molecules. However, cultured dorsal root ganglion neurons lacking sortilin showed impaired neurite outgrowth upon NGF stimulation [11]. Loss of sortilin also aggravated neurological phenotypes observed in p75 NT receptor (p75 NTR )-KO mice [11]. Sortilin also acts as a co-receptor of p75 NTR against proNTs to transduce cell death signals [30]. We found that sortilin was significantly reduced in prion-infected cells and mouse brains, and that the NGF signaling was disturbed in prion-infected cells. It might thus be interesting to investigate whether or not the sortilin-mediated NT signals are involved in the pathogenesis of prion disease.
It remains controversial whether or not degradation of sortilin might be stimulated in lysosomes in other neurodegenerative diseases. Reduced levels of sortilin have been reported in the limbic and occipital regions of AD brains [31]. To the contrary, increased expression of sortilin has been demonstrated in the temporal cortex of AD brains [32]. Other investigators showed no alteration of sortilin levels in the superior frontal and superior temporal cortices of AD brains [33].
In short, we presented a novel accumulation mechanism of PrP Sc through degradation of sortilin. Prion infection stimulated degradation of sortilin in lysosomes, reducing sortilin levels in prion-infected cells. The reduction of sortilin disturbs its function to sort PrP C and PrP Sc to the late endosomal/lysosomal compartments for degradation. As a result, PrP C is increasingly converted to PrP Sc and PrP Sc degradation is delayed, and eventually PrP Sc progressively accumulates in prion-infected cells. Thus, accelerating the sortilin-mediated lysosomal degradation of PrP C and PrP Sc might be therapeutic in prion diseases.

Cell lines and animals
Cells were maintained at 37˚C with 5% CO 2 in air in Dulbecco's Modified Eagle Medium (DMEM, Invitrogen, Carlsbad, CA) containing 10% fetal bovine serum (FBS). N2aC24, N2aC24L1-3, and cured N2aC24L1-3 cells were previously established elsewhere [26]. N2aC24 cells were cloned from mouse neuroblastoma N2a cells overexpressing exogenous mouse PrP C . N2aC24L1-3 cells were cloned from N2aC24 cells persistently infected with 22L scrapie prions. Cured N2aC24L1-3 cells are cured from prion infection by treatment with SAF32 anti-PrP Ab and then maintained in antibody-free DMEM with 10% FBS. ScN2a cells (kindly gifted from Prof Doh-ura, Tohoku University) were N2a cells persistently infected with RML scrapie prions.

Immunoprecipitation
Cells were lysed in buffer A [20 mM MES-KOH (pH 7.0), 0.15 M KCl, 1 mM DTT, 10% glycerol, 0.2% (w/v) CHAPS] containing protease inhibitor cocktail (Nakalai tesque, Kyoto, Japan). The lysate was cleared by centrifugation for 5 min at 20,000×g at 4˚C and the supernatant was transferred to a new tube. 500 μL of supernatant containing 500 μg of total proteins were incubated with 1 μg of indicated Abs for 2 h. 5 μL of protein-G sepharose (GE healthcare) was added and the mixture was rotated for 4 h at 4˚C. Thereafter, protein-G sepharose was precipitated and the precipitant was washed with buffer A 5 times. The final precipitate with protein-G sepharose was suspended in 50 μL of Laemmli's sample buffer, heated, and subjected to Western blotting to detect proteins of interest with appropriate Abs.

Protein interaction assay using Dynabeads protein-G
For detection of interaction of PrP Sc and sortilin, 15 μL of Dynabeads protein-G (Thermo Fisher Scientific) was added instead of protein-G sepharose. Immunocomplexes were collected using magnet instead of centrifugation and washed with buffer A. The same procedure was repeated 5 times. The finally collected complexes were suspended in 45 μL of buffer B [150 mM NaCl, 50 mM Tris-HCl (pH 7.5), 0.5% (w/v) Triton X-100, 0.5% (w/v) sodium deoxycholate, 1 mM EDTA] containing 100 μg/ml proteinase K (PK) and incubated at 37˚C for 60 min with mixing at 1,100 rpm using Thermomixer (Eppendorf, Hamburg, Germany). The PK-treated samples were denatured in Laemmli's sample buffer and subjected to Western blotting to detect PrP Sc .
For detection of interaction of PrP C and sortilin on the cell surface, cells were washed with ice-cold PBS. After blocking of the cells with buffer C [20 mM MES-KOH (pH 7.0), 0.15 M KCl, 0.25 M sucrose] containing 0.5% (w/v) bovine serum albumin (BSA) for 10 min at 4˚C, the cell surface PrP was labeled with 1 μg/mL of SAF61 anti-PrP Ab in buffer C containing 0.5% (w/v) BSA for 30 min at 4˚C. After washing the cells with ice-cold-PBS, they were suspended in 1 mL of buffer C. To homogenize the cells, the cell suspension was passed through a 27G needle 10 times. The homogenate was centrifuged at 500×g for 10 min at 4˚C and the supernatant was incubated with 15 μL of Dynabeads protein-G with gentle agitation for 2 h at 4˚C. The beads were collected by magnet and washed with buffer C. The collected immunocomplexes on the beads were washed with buffer A 5 times and finally suspended in 50 μL of Laemmli's sample buffer and subjected to Western blotting.
GST pulldown assay 2 μg of GST-tagged VPS10P domain of sortilin, which was pre-bound to 5 μL of glutathione beads, was incubated with 2 μg of His-tagged full-length recombinant PrP or His-tagged recombinant PrP lacking residues 23-88 in buffer A containing protease inhibitor cocktail for 2 h with rotation at 4˚C. The precipitate was washed with buffer A 5 times, suspended in Laemmli's sample buffer, heated, and subjected to Western blotting. GST-tagged VPS10P domain and His-tagged PrPs were detected with rabbit polyclonal anti-sortilin Abs and RGS-His Ab, respectively.

Biotinylation of cell surface proteins
Biotinylation of cell surface proteins was carried out as described elsewhere [26]. In brief, cells (85-95% confluent) were washed with PBS and incubated with Sulfo-NHS-LC Biotin (Thermo Fisher Scientific) in PBS for 30 min at room temperature. The cells were then washed with 0.1 M glycine in PBS and lysed in buffer B. The lysate was mixed with NeutrAvidin UltraLink Resin (Thermo Fisher Scientific) for 4 h at 4˚C and the biotinylated protein-resin complexes were collected by brief microcentrifugation. The complexes were then washed with the buffer and heated at 99˚C for 10 min in Laemmli's sample buffer to separate the biotinylated proteins from the complexes. The biotinylated proteins in the supernatant were subjected to Western blotting.

Immunofluorescence staining
Cells were stained with indicated Abs as described previously [26]. In brief, cells grown on coverslips were fixed in 3% paraformaldehyde (PF) for 15min and treated with 0.1 M glycine in PBS for 10 min. Permeabilization was carried out using 0.1% Triton X-100 in PBS for 4 min at RT. To detect PrP Sc , the cells were treated with 5 M guanidinium thiocyanate for 10 min at RT. After washing with PBS, the cells were incubated with the first Ab in 5% FBS in PBS and then with fluorescent secondary Ab. For detection of PrP Sc , mouse anti-PrP Ab clone 132 [21] (kindly gifted from Prof Horiuchi, Hokkaido University) was used as a first Ab. After washing, the coverslips were mounted with Prolong Gold antifade reagent (Invitrogen). Fluorescence images were obtained using BIOREVO BZ-9000 (Keyence, Osaka, Japan), which is equipped with haze reduction function, which enables production of fluorescent images very similar to those taken by a confocal microscope. To assess the co-localization of proteins of interest, Pearson's correlation coefficient was calculated using Co-localization Plugin (JaCoP) tool in Image J [36].

Internalization assay of antibody-labeled surface PrP
Cells were washed with ice-cold PBS and treated with 1% BSA in PBS for 10 min at 4˚C prior to incubation with the indicated anti-PrP Abs (1 μg/mL) for 10 min at 4˚C in 1% BSA-containing PBS. The cells were then washed with ice-cold PBS and incubated at 37˚C for 2 h. Thereafter, the cells were fixed with 3% paraformaldehyde, permeabilized with 0.1% Triton X-100, and stained with Alexa Fluoro 488 anti-mouse IgG Ab (Thermo Fisher Scientific). Fluorescent signals were observed using BIOREVO BZ-9000 (Keyence) and their intensities were analyzed using BZ-II analyzer (Keyence).

Internalization assay of biotinylated surface PrP
Cells were washed three times with ice-cold PBS (pH7.4) and incubated with 5 mg/mL sulfo-NHS-SS-biotin (Thermo Fisher Scientific) in PBS (pH 7.4) at 4˚C for 10 min to biotinylate cell surface proteins. After washing the cells twice with ice-cold PBS (pH 7.4) and incubation with 50 mM glycine in PBS (pH 7.4), the cells were further washed twice with ice-cold PBS (pH 7.4). Thereafter, the cells were incubated in DMEM medium at 37˚C. After 2 h-incubation, the cells were washed three times with ice-cold PBS (pH 7.4), and biotin was removed from the proteins still on the cell surface by incubating the cells with 100 mM reducing glutathione in PBS (pH 7.4) at 37˚C for 10 min. After washing the cells with ice-cold PBS (pH 7.4), the cells were lysed in buffer B containing protease inhibitor cocktail (Nakalai tesque). The lysate was cleared by centrifugation for 5 min at 20,000×g at 4˚C and the supernatant was transferred to a new tube. 20 mL of NeutrAvidin beads (Thermo Fisher Scientific) was added into the cell lysate containing 300 mg of proteins and the mixture was rotating at 4˚C for 2 h. After washing the beads with lysis buffer, the beads was suspended in 50 mL of SDS-PAGE sample buffer and subjected to Western blotting with 6D11 anti-PrP Ab.

Fractionation of membrane microdomains
Cells grown to~80% confluency in a 35 mm tissue culture dish were suspended in 250 μL of MBS buffer [25 mM MES-NaOH (pH 6.5), 0.15 M NaCl] containing 1% (w/v) Triton X-100, and homogenized by being passed through a 21G-needle 15 times. After centrifugation at 500×g for 5 min at 4˚C, 220 μL of the supernatant was transferred to a new tube and mixed with 220 μL of MBS buffer containing 80% (w/v) sucrose to make 40% (w/v) sucrose. 200 μL of the sample was placed at the bottom of a discontinuous sucrose gradient consisting of 600 μL of 30% (w/v) sucrose and 200 μL of 5% (w/v) sucrose. The sample was centrifuged at 140,000×g for 24 h at 4˚C in an S55S rotor (Hitachi Koki, Tokyo, Japan). Ten fractions (100 μL/fraction) were collected from the top.

Prion infection
Brains were removed from terminally ill wild-type C57BL/6 mice infected with RML prions. A single brain was homogenized (10%, w/v) in PBS using a multi-beads shocker (Yasui Kikai, Osaka, Japan) and then diluted to 1% with PBS. Two 1% (w/v) brain homogenates were mixed to prepare the homogenate of 2 pooled brains and the resulting homogenate was intracerebrally inoculated into a 4-5 week-old mouse with its 20 μL aliquot. The signs for diseaserelated symptoms were evaluated as previously described [37].
For infection of cells, cells were seeded at a density of 2×10 5 cells/well in a 6-well tissue culture plate. At 4 h after cell seeding, the clarified RML-infected brain homogenate [34] containing 50 μg proteins was added to the well and cells were subsequently passaged every 3 days.

Immunohistochemistry
Paraffin-embedded samples were sectioned, deparaffinized, and rehydrated. The samples were autoclaved in 1 mM HCl at 121˚C for 5 min and subsequently washed with PBS. The samples were then digested with 50 μg/mL PK in PBS at 37˚C for 30 min, treated with 3 M guanidine thiocyanate for 10 min, and washed with PBS. After blocking with 5% FBS in PBS for 1 h, the sampled were incubated with 6D11 anti-PrP Ab for 2 h, washed with PBS, and treated with ImmPRESS REAGENT Anti-Mouse IgG (Vector Laboratories, U.S.A) for 30 min. After washing with PBS, the samples were incubated with ImmPACT DAB (Vector Laboratories) for 180 sec for staining.

Western blotting
Western blotting was performed as reported previously [26]. To evaluate protein expression, signals were densitometrically measured using LAS-4000 mini (Fujifilm Co., Tokyo, Japan). The measured intensity of target proteins was normalized against the signal intensity of βactin used as an internal control protein.

NGF stimulation
Cells were removed from a tissue culture dish by pipetting and transferred to a new tube. After washing with DMEM medium, cells were collected by centrifugation at 500×g for 2 min at 4˚C, suspended in 1 mL of DMEM medium containing 2 nM NGF (Thermo Fisher Scientific), transferred into a well of 12 well plate, and incubated for 10 min at 37˚C in a 5% CO 2 incubator. The cells were collected by centrifugation at 500×g for 5 min at 4˚C and washed with icecold PBS. The pellet was lysed in 200 μL of buffer B containing protease inhibitor cocktail (Nakalai tesque). The lysate was cleared by centrifugation at 12,000×g for 2 min at 4˚C and the supernatant was subjected to Western blotting.

Isolation of exosomes
Cells were cultured in 2 mL of DMEM medium containing 10% exosome-depleted FBS (System Bioscience, CA,USA) in a 6 well tissue culture plate for 72 h and the culture medium was collected. The culture medium was centrifuged at 2,000×g for 10 min at 4˚C. The supernatant was passed through 0.22 μm pore filter membrane and the flow-through was centrifuged at 10,000×g for 30 min at 4˚C. Exosomes in the supernatant were collected by ultracentrifugation at 100,000×g for 1 h at 4˚C and washed with PBS. The exosomes were dissolved in 100 μL of Laemmli's sample buffer and subjected to Western blotting.

Statistical analysis
Survival and incubation times are analyzed using the log-rank test. Other data were analyzed using the one-way ANOVA. Signal intensity in Sort1 -/mice was evaluated against that in Sort1 +/+ mice. Data are means ± SD of 4-6 independent brains. n.s., not significant; Ã p < 0.05. (D) Immunohistochemical staining of PrP Sc in the brain hippocampus areas of Sort1 -/and Sort1 +/+ mice at 60 and 90 dpi and at terminal stage. Bar, 300 μm. (TIF)