CONSERVED STRESS-PROTECTIVE ACTIVITY BETWEEN PRION PROTEIN AND SHADOO

Shadoo (Sho) is a neuronally expressed glycoprotein of unknown function. Although there is no overall sequence homology to the cellular prion protein (PrP(C)), both proteins contain a highly conserved internal hydrophobic domain (HD) and are tethered to the outer leaflet of the plasma membrane via a C-terminal glycosylphosphatidylinositol anchor. A previous study revealed that Sho can reduce toxicity of a PrP mutant devoid of the HD (PrPΔHD). We have now studied the stress-protective activity of Sho in detail and identified domains involved in this activity. Like PrP(C), Sho protects cells against physiological stressors such as the excitotoxin glutamate. Moreover, both PrP(C) and Sho required the N-terminal domain for this activity; the stress-protective capacity of PrPΔN as well as ShoΔN was significantly impaired. In both proteins, the HD promoted homodimer formation; however, deletion of the HD had different effects. Although ShoΔHD lost its stress-protective activity, PrPΔHD acquired a neurotoxic potential. Finally, we could show that the N-terminal domain of PrP(C) could be functionally replaced by that of Sho, suggesting a similar function of the N termini of Sho and PrP(C). Our study reveals a conserved physiological activity between PrP(C) and Sho to protect cells from stress-induced toxicity and suggests that Sho and PrP(C) might act on similar signaling pathways.


INTRODUCTION
Prion diseases in humans and other mammals are characterized by a conformational transition of the cellular prion protein (PrP C ) into an aberrantly folded isoform, designated scrapie prion protein (PrP Sc ). PrP Sc can form amyloid plaques in the diseased brain and is the major constituent of infectious prions (rev. in (1)(2)(3)(4). Biogenesis of PrP C is essential for the pathogenesis of prion diseases. Mice with a targeted disruption of the PrP gene (Prnp) are resistant to prion diseases and to the propagation of infectious prions (5). Moreover, neuronal expression of PrP C seems to be required to mediate neurotoxic effects of scrapie-prions (6)(7)(8)(9).
Structural studies with recombinantly expressed PrP revealed a large flexibly disordered N-terminal region, containing an octarepeat region, and a structured C-terminal domain (aa 121-231). This autonomously folding domain contains three α-helical regions and a short two-stranded β-sheet (10)(11)(12). The Cterminal domain is characterized by extensive co-and posttranslational modifications, including two N-linked glycans with complex structure, a disulfide bridge and a C-terminal GPI anchor (rev. in (13). Interestingly, the octarepeat region in the N-terminal domain as well as the C-terminal GPI anchor are dispensable for the generation of infectious prions (8,14), but are required for the stressprotective activity of PrP C (9,15).
While in the majority of prion diseases there is a correlation between the accumulation of PrP Sc , formation of infectious prions and neurodegeneration, there are some interesting exceptions. Transgenic mouse models revealed that several PrP mutants can induce neuronal cell death in the absence of infectious prion propagation (rev. in (16). From one class of PrP mutants it emerged that PrP C can acquire a neurotoxic potential by deleting the internal HD (17)(18)(19). Similarly to PrP C , PrPΔHD is complex glycosylated and linked to the plasma membrane via a GPI anchor (20). Moreover, expression of PrP C can completely prevent the neurotoxic activity of PrPΔHD, suggesting that PrP C and PrPΔHD can induce neuroprotective or neurotoxic signaling via binding to the same signal transducing protein(s) (9,19). Indeed, numerous studies in transgenic animals and cultured cells are now supporting the idea that the physiological role of PrP C is to protect neuronal cells against stress-induced cell death (rev. in (21).
In search for homologs/paralogs of PrP C a new gene was identified termed SPRN, encoding for a protein denoted Shadoo (Sho). Sho is highly conserved from fish to mammals. The sequence homology between Sho and PrP is restricted to the internal hydrophobic domain, however, certain features, such as a N-terminal repeat region and a C-terminal glycosylphosphatidylinositol (GPI) anchor are conserved, suggesting that Sho and PrP may be functionally related (22). Experimental evidence for the post-translational modifications and cell surface localization of Sho was first presented for zebrafish Sho (23) and afterwards also for mouse Sho (24). Moreover, it was demonstrated that Sho, similarly to PrP C , can prevent neuronal cell death induced by the expression of PrPΔHD mutants (24).
In this study, we analyzed the stressprotective activity of Sho in detail and defined domains required for this activity. Furthermore, the generation of a Sho-PrP chimeric protein revealed that the N-terminus of Sho can restore the stress-protective activity of a PrP mutant devoid of the N-terminal domain (PrPΔN).

EXPERIMENTAL PROCEDURES
Reagents and antibodies. All standard chemicals and reagents were purchased from Sigma if not otherwise noted. The following antibodies were used: mouse monoclonal anti-PrP 3F4 antibody (25), mouse monoclonal anti-V5-tag antibody (Invitrogen), mouse monoclonal anti-HA-tag antibody (Covance), anti-active caspase-3 polyclonal antibody (Promega), Cy3 conjugated anti-rabbit IgG antibody (Dianova), Cy3 conjugated anti-mouse IgG antibody (Dianova). For the generation of a rabbit polyclonal anti-Sho antibody recombinant human full-length Sho, expressed in and purified from E.coli was used for the immunization of rabbits (Eurogentec, Belgium). The mounting medium Mowiol (Calbiochem) was supplemented with 4',6diamidino-2-phenylindole (DAPI). Plasmid construction. The cloning of PrP, PrP∆N (∆28-89), PrP∆HD (∆113-133), and PrPS131C were described previously (20,26,27). The human Sho gene was prepared by ligation-based strategy for chemical gene synthesis using the Sloning building block technology (28) (Sloning, Puchheim) and cloned into a pcDNA 3.1/Zeo (+). A V5 tag (GGT AAA CCG ATA CCG AAC CCG CTC CTC GGT CTC GAT TCG ACG) or HA tag (TAC CCA TAC GAT GTT CCA GAT TAC GCT) was inserted between amino acids 124 and 125. Sho was used as a template to introduce the following deletions and mutations by standard PCR cloning techniques. Sho∆N (aa 30-56 deleted), Sho∆HD (aa 68-89 deleted), ShoS87C (serine at aa 87 replaced with cysteine), Sho∆N,S87C (aa 30-56 deleted; serine at aa 87 replaced with cysteine). All amino acid numbers refers to the human Shadoo sequence (GenBank™ association number NM-001012508). The chimeric N-Sho/PrP-C (Sho 1-63 and PrP 89-254) was generated by PCR and cloned into pcDNA 3.1/Zeo (+) vector. Cell culture, transfection, protein deglycosylation, phospholipase-C, tunicamycin, Brefeldin A treatment and Western blotting. Human neuroblastoma cells (SH-SY5Y) were cultured as described earlier (20). Cultured cells were transfected with Lipofectamine plus (Invitrogen) according to the manufacturer instruction. After 24 h cells were washed with cold PBS, scraped off the plate, pelleted and lysed in cold detergent buffer (0.5% Triton X-100, 0.5% sodium deoxycholate in PBS). Endo H (New England Biolabs) and PNGase F (Roche) digestion were done according to manufacturer instruction. To inhibit N-linked glycosylation or transport to the plasma membrane transfected cells were grown in the presence of tunicamycin (0.5µg/ml) or Brefeldin A (1µg/ml) respectively. 24h later cells were analyzed. For Phosphatidylinositolphospholipase C (PIPLC) treatment cells were washed twice with cold PBS. PIPLC (1 U/ml) was added to the cells for 3 h at 4°C. The medium was collected and cells were washed twice with cold PBS and lysed in cold detergent buffer. Proteins present in the medium were precipitated by trichloroacetic acid (TCA), boiled in Laemmeli sample buffer and analyzed by Western blotting as described earlier (29). Apoptosis assay. As described earlier (27) SH-SY5Y cells were grown on cover slips. 24 h after transfection the cells were incubated with glutamate [500µM] for 3 h. The cells were then fixed and activated caspase-3 detected by indirect immunofluorescence using an anti-active caspase-3 antibody. To detect cells undergoing apoptosis, the number of activated caspase-3 positive cells out of at least 1100 transfected cells was determined using a Zeiss Axioscope 2 plus microscope (Carl Zeiss, Göttingen, Germany). Quantifications were based on triplicates of at least three independent experiments. Co-immunoprecipitation. To analyze formation of a mixed PrP/Sho dimer SH-SY5Y cells were co-transfected with PrPS131C and ShoS87C. At 24h post-transfection the cells were harvested and lysed in ice-cold detergent lysis buffer (0.5% Triton X-100, 0.5% sodium deoxycholate in PBS) supplemented with protease inhibitors. Precleared lysates were incubated with αV5 antibody overnight at 4°C. The immunocomplex was precipitated with protein A sepharose beads and analyzed by Western blotting. To analyze formation of Sho trans-dimers separate dishes of SH-SY5Y cells were transfected with either ShoS87C-V5 or ShoS87C-HA. 3h posttransfection cells were extensively washed, trypsinized, mixed together and seeded in one cell culture dish. 24h later the cells were harvested and analyzed as described above. Indirect immunofluorescence microscopy. Transiently transfected SH-SY5Y cells were grown on glass cover slip and fixed 24 h post transfection with 3% PFA for 20 min. Fixed cells were incubated with primary antibody for 45 min at 37°C in PBS containing 1% BSA. After extensive washing with cold PBS, incubation with the Cy3 conjugated secondary antibody followed at 37°C for 30 min. Cells were mounted onto glass slides and examined by fluorescence microscopy. Statistical analysis. Data were expressed as means ± SE. Statistical analysis was carried out using student's t-test. P-values are as follows: * P<0.05, **P<0.005, ***<0.0005.

Sho mutants devoid of the N-terminal domain or the internal hydrophobic domain are complex glycosylated and targeted to the outer leaflet of the plasma membrane via a GPI anchor
In a previous study we have shown that zebrafish Sho expressed in mammalian cells is a complex glycosylated, GPI-anchored protein present at the outer leaflet of the plasma membrane (23). For a functional characterization of mammalian Sho we cloned the human homolog and several mutants as outlined in Fig.  1A. The design of the Sho mutants was based on PrP mutants that have been characterized previously. PrPΔN, a mutant lacking large parts of the intrinsically disordered N-terminus (Δaa 27-89), can promote propagation of infectious prions (14), but shows an impaired neuroprotective activity (9,15). PrPΔHD lacks the highly conserved internal hydrophobic domain (Δaa 113-133) and is characterized by a neurotoxic potential that can be blocked by the co-expression of wild type (wt) PrP C (9,(17)(18)(19). Importantly, these PrP mutants are complex glycosylated and targeted to the outer surface of the plasma membrane via a GPI anchor (20).
First, we expressed the different Sho constructs and the corresponding PrP mutants in human SH-SY5Y cells to analyze the co-and post-translational modifications. To evaluate the modification with N-linked glycans, cell lysates were treated with PNGase F to remove all Nlinked glycans. The apparent shift in electrophoretic mobility after PNGase F digestion indicated that all constructs are modified with N-linked glycans (Fig. 1C). The glycans were resistant to Endo H digestion (data not shown), indicating that all constructs are modified with N-linked glycans of complex structure.
To test for a C-terminal GPI anchor, live cells were incubated with phosphatidylinositolspecific phospholipase C (PIPLC) to release GPI-anchored proteins from the cell surface. Indeed, after PIPLC treatment the relative amount of the PrP constructs in the cell lysates (L) decreased concomitantly with the appearance of the respective proteins in the cell culture supernatant (M) ( Fig. 2A). Similarly, PIPLC treatment reduced the relative amount of the Sho constructs present in the cell lysates. In parallel, Sho and ShoΔN were found in the supernatant of PIPLC-treated cells ( Fig. 2A). Curiously, we never detected significant amounts of ShoΔHD in the cell culture medium after PIPLC treatment, neither using the V5 or our newly generated polyclonal anti-Sho antisera. So far, we have not been able to unravel the underlying mechanisms for this phenomenon. However, when we performed indirect immunofluorescence of non-permeabilized cells we were able to demonstrate localization of all constructs, including ShoΔHD, at the outer surface of the plasma membrane (Fig. 2B).
In sum, these data revealed that wild type Sho as well as Sho mutants devoid of the N-by guest on March 17, 2020 http://www.jbc.org/ Downloaded from terminal domain or the internal HD are complex glycosylated and tethered to the outer leaflet of the plasma membrane via a GPI anchor.

Sho protects cells from apoptotic cell death induced by excitotoxic stress or the expression of a neurotoxic PrP mutant
To evaluate the stress-protective capacity of Sho, we employed two different stress paradigms, exposition of cells to the excitotoxin glutamate and expression of the neurotoxic PrP mutant PrPΔHD. To identify cells undergoing apoptosis, transiently transfected SH-SY5Y cells were analyzed by indirect immunofluorescence using an antibody specific for activated caspase-3. Please note that SH-SY5Y cells are characterized by low levels of endogenous PrP C (Fig. 3A, right panel, pcDNA, 3F4). Consistent with previous findings (9), PrP C was able to protect cells against excitotoxic cell death, while the deletion of the intrinsically disordered Nterminal domain lead to a loss of this activity (Fig. 3A). Similarly to PrP C , full length Sho displayed a stress-protective capacity, which was abolished by the deletion of the N-terminal domain (Fig. 3A). These findings indicate that deletion of the N-terminal domain of PrP or Sho had similar consequences. However, deletion of the internal HD had distinct effects. As previously documented in transgenic mice and cell culture models (9,(17)(18)(19), PrPΔHD acquires a toxic activity (Fig. 3A). ShoΔHD lost its stressprotective activity, however, in contrast to PrPΔHD it did not induce cell death (Fig. 3A).
In a next step we employed expression of PrPΔHD as a toxic insult. As expected from previous studies, toxic effects of PrPΔHD expression were alleviated by the co-expression of PrP C (Fig. 3B). Corroborating earlier results with primary neurons (24), Sho was also able to suppress toxicity induced by PrPΔHD (Fig 3B). However, neither PrPΔN nor ShoΔN prevented apoptosis in cells expressing PrPΔHD. Similarly, ShoΔHD could not interfere with the toxic effects of PrPΔHD (Fig. 3B).

The hydrophobic domain mediates homodimerization of Sho
In a previous study we could show that the HD mediates dimerization of PrP C and is part of the dimer interface (9). Since the HD of Sho and PrP are highly homologous (44,4%, Fig.  1B), we tested whether the HD of Sho has a similar activity. To stabilize a possible Sho dimer, we replaced serine 87 by cysteine (Fig.   4A). In case the HD is part of a dimer interface, cysteine 87 could form an intermolecular disulfide bond, which is stable in SDS buffers under non-reducing conditions. Thus, dimer formation can be analyzed by SDS-PAGE followed by Western blotting by preparing protein lysates with sample buffer devoid of reducing agents, such as 2-mercaptoethanol (ß-ME) or dithiothreitol (DTT). SH-SY5Y cells were transiently transfected with ShoS87C. As a control, PrPS131C was analyzed in parallel. Protein lysates were boiled in Laemmli sample buffer with or without 2-mercaptoethanol and analyzed by Western blotting. In line with earlier results, PrP dimers were detected in lysates prepared from PrPS131C-expressing SH-SY5Y cells when reducing agents were omitted from the sample buffer (Fig. 4B). Similarly, in extracts from cells expressing ShoS87C a slower migrating species appeared with a molecular mass indicative for a Sho dimer (Fig. 4B). In the presence of 2-mercaptoethanol the migration pattern of ShoS87C was identical to that of wild type Sho, indicating that the introduced cysteine residue has induced the formation of an intermolecular disulfide bond. To assess the role of the N-terminal domain in dimer formation we deleted residues 30-56. As shown in figure 4C ShoΔN,S87C formed dimers similarly to ShoS87C. Treatment of cell lysates with Endo H or PNGase F revealed that ShoS87C is modified with N-linked glycans of complex structure, similarly to wild type Sho (Fig. 4D). In addition, both ShoS87C and PrPS131C were released from the cell surface of life cells after incubation with PIPLC, revealing the presence of a Cterminal GPI anchor (Fig. 4E).
Next we tested whether dimer formation required N-linked glyosylation and occurs during transit through the secretory pathway. Transiently transfected cells expressing either PrPS131C or ShoS87C were cultivated in the presence of tunicamycin or brefeldin A and dimer formation was analyzed by Western blotting. Indeed, homodimers of PrP or Sho could be detected under both conditions (Fig.  5A). These data indicated that dimer formation can obviously occur as soon as the proteins enter the ER lumen. However, we also addressed the possibility that dimer formation occurs in trans at the plasma membrane between Sho molecules located on adjacent cells. For this purpose we generated a ShoS87C construct with a Cterminal HA tag instead of the V5 tag. Separately transfected cells expressing either ShoS87C-V5 or ShoS87C-HA were mixed and co-cultivated for additional 24 h. The density was chosen to allow cell-cell contact. The coimmunoprecipitation analysis did not indicate the formation of trans-dimers (Fig. 5B). However, we cannot exclude the possibility that Sho transdimers can form in vivo under certain condition.
The HD of Sho and PrP are highly similar, thus we analyzed whether it might promote formation of a mixed PrP/Sho dimer. Transiently transfected cells expressing both PrPS131C and ShoS87C were lysed and Sho immunopurified with the V5 antibody. The immunopellet was then analyzed by Western blotting. Sho homodimers were efficiently detected with this approach, however, we were not able to show the formation of mixed PrP/Sho dimers (Fig. 5C).

The N-terminal domain of Sho can restore stressprotective activity of PrPΔN
The functional characterization of Sho mutants presented above revealed a critical role of the N-terminal domain for the stressprotective activity of Sho. Similarly, PrPΔN lacks a stress-protective activity in cell culture and animal models (Fig. 3) (9,15). To test for the intriguing possibility of a conserved function of the N-terminal domains of PrP C and Sho we fused the N-terminal domain of Sho to PrPΔN (Fig. 6A). Expression and post-translational modifications of N-Sho/PrP-C were similarly to that of wild type PrP or wild type Sho: the chimeric protein was complex glycosylated and tethered to the outer leaflet of the plasma membrane via a GPI anchor (Fig. 6B, C). We then compared the stress-protective activity of N-Sho/PrP-C to that of PrPΔN. As described above, we used two different toxic conditions: 1. expression of the neurotoxic PrP mutant PrPΔHD and 2. treatment of cells with the excitotoxin glutamate. Indeed, these experiments revealed that the N-terminal domain of Sho can restore the stress-protective capacity of PrPΔN. Similarly to full length Sho or full length PrP, N-Sho/PrP-C interfered with apoptotic cell death of SH-SY5Y cells expressing PrPΔHD (Fig. 6D) or exposed to glutamate (Fig. 6E).

DISCUSSION
Sho is a highly conserved protein found from fish to man. While it has no overall sequence homology to PrP both proteins contain a conserved internal hydrophobic domain and a C-terminal GPI anchor, two elements important for physiological and toxic activities of PrP. Our study presents experimental evidence that Sho and PrP are characterized by a conserved activity to protect cells against stress-induced cell death and suggests that both proteins might act on similar signaling pathways.

The stress-protective activity of Sho is dependent on the N-terminal domain and the internal hydrophobic domain
The first finding of our study was that the stress-protective activity of Sho is not restricted to counteract toxic effects of PrPΔHD, an artificial PrP mutant devoid of the internal HD.
We employed glutamate as a physiologically relevant stressor to show that Sho can efficiently protect cells against excitotoxin-induced cell death. Deletion mutants revealed that the stress-protective activity of Sho and PrP seems to be dependent on similar domains, in particular the N-terminal and the internal HD. ShoΔN and ShoΔHD displayed a reduced stress-protective activity, but are complex glycosylated and attached to the outer leaflet of the plasma membrane via a GPI anchor, indicating that the impaired activity is not due to incorrect cellular trafficking.
What could be the role of the N-terminal domain and the internal HD in promoting protective signaling? The N-terminal domain of PrP is intrinsically disordered and it has been shown that intrinsically disordered domains are often involved in protein-protein interactions (30). Thus, it would be an attractive idea to assume that the N-terminal domains of PrP C and Sho mediate interaction with a yet unknown coreceptor required for intracellular signal transmission. Interestingly, a circular dichroism analysis of recombinant Sho indicated that the whole protein might be unstructured (24). The HD is the only domain with significant sequence homologies between Sho and PrP C . Indeed, the HD promoted dimerization of both Sho and PrP C and was part of the dimer interface. In this context is worth mentioning that dimerization is a common feature of many cell surface receptors, therefore it can be speculated that dimer formation is involved in signal transmission of PrP C -and Sho-dependent pathways. Notably, formation of both PrP and Sho homodimers is not dependent on N-linked glycosylation and already occurs during transit through the secretory pathway. Similarly, PrP and Sho mutants devoid of the N-terminal domain are still capable of forming homodimers, however, they have an impaired stress-protective activity.

Distinct activity of PrPΔHD
Studies in transgenic mice revealed the unexpected finding that PrP can acquire a neurotoxic potential simply by deleting the internal hydrophobic domain (17)(18)(19). Importantly, the neurotoxic potential of PrPΔHD is independent of the propagation of infectious prions, a phenomenon also seen for other neurotoxic PrP mutants (rev. in (16). While the underlying mechanisms of PrPΔHD-induced toxicity are still elusive, co-expression of wild type PrP C completely prevents toxic effects of PrPΔHD. Based on this intriguing observation it has been hypothesized that the stress-protective signaling of PrP C and the neurotoxic signaling of PrPΔHD are transmitted through a common coreceptor, which remains to be identified (9,(17)(18)(19). Interestingly, co-transfection experiments with PrP-deficient cerebellar granule neurons indicated that Sho has a PrP C -like activity to alleviate toxic effects of PrPΔHD expression (24). We have been able to recapitulate the toxic activity of PrPΔHD expression in our cell culture model and to demonstrate the protective activity of PrP and Sho against PrPΔHD-induced toxicity. In addition, we could show that ShoΔHD lost its ability to protect against stressinduced cell death, however, ShoΔHD did not acquire a toxic activity, at least not under the experimental conditions tested. Thus, it appears that Sho and PrP share a stress-protective activity, however, the ability to adopt a toxic conformation seems to be specific for PrP. In this context it is tempting to speculate that this ability might be linked to the unique feature of PrP C to convert into PrP Sc .

The N-terminal domain of Sho can functionally replace that of PrP
Our study emphasized a similar activity of Sho and PrP to protect cells against excitotoxin-and PrPΔHD-induced toxicity and revealed a crucial role for the N-terminal domain proximal to the HD: Both PrPΔN and ShoΔN showed a significantly reduced stress-protective activity. To address the possibility that both proteins interact with similar co-factors through their N-termini, we fused the N-terminal domain of Sho to PrPΔN. The chimeric protein N-Sho/PrP-C had a stress-protective activity similarly to that of wtSho or wtPrP, indicating that the N-terminal domain of Sho could functionally replace that of PrP. There is neither sequence homology between the N-terminal domains of Sho and PrP, nor an indication that the N-terminal domain of Sho is able to bind copper. The only similarity seems to be that both domains are intrinsically disordered. Based on the idea that intrinsically disordered domains are involved in protein-protein interactions, it appears plausible to assume that the N-terminal domain of PrP and Sho mediates interaction with the same co-receptor, required for stressprotective signaling. Further studies are now required to identify the putative co-receptor and signaling pathway(s) activated by PrP and Sho.

FOOTNOTES
The work was supported by grants from the Deutsche Forschungsgemeinschaft (SFB 596), the Max Planck Society, the BMBF (BioDisc, DIP5.1) and a PhD fellowship from the DAAD to V.S. J.T gratefully acknowledges support by the German Center for Neurodegenerative diseases (DZNE). We thank Ulrike Resenberger for critical reading the manuscript and Veronika Müller for technical help.