PrPC directly interacts with proteins involved in signaling pathways.

The cellular prion protein (PrP(C)) is a conserved glycoprotein predominantly expressed in neuronal cells. Its purpose in living cells is still enigmatic. To elucidate on its cellular function, we performed a yeast two-hybrid screen for interactors. We used murine PrP(C) (amino acids 23-231) as bait to search a mouse brain cDNA expression library. Several interaction partners were identified. Three of them with a high homology to known sequences were further characterized. These candidates were the neuronal phosphoprotein synapsin Ib, the adaptor protein Grb2, and the still uncharacterized prion interactor Pint1. The in vivo interaction of the three proteins with PrP(C) was confirmed by co-immunoprecipitation assays with recombinant and authentic proteins in mammalian cells. The binding regions were mapped using truncated PrP constructs. As both synapsin Ib and Grb2 are implicated in neuronal signaling processes, our findings further strengthen the putative role of the prion protein in signal transduction.

The cellular prion protein (PrP C ) is a conserved glycoprotein predominantly expressed in neuronal cells. Its purpose in living cells is still enigmatic. To elucidate on its cellular function, we performed a yeast two-hybrid screen for interactors. We used murine PrP C (amino acids 23-231) as bait to search a mouse brain cDNA expression library. Several interaction partners were identified. Three of them with a high homology to known sequences were further characterized. These candidates were the neuronal phosphoprotein synapsin Ib, the adaptor protein Grb2, and the still uncharacterized prion interactor Pint1. The in vivo interaction of the three proteins with PrP C was confirmed by co-immunoprecipitation assays with recombinant and authentic proteins in mammalian cells. The binding regions were mapped using truncated PrP constructs. As both synapsin Ib and Grb2 are implicated in neuronal signaling processes, our findings further strengthen the putative role of the prion protein in signal transduction.
Prion diseases are fatal neurodegenerative disorders in humans and animals including Creutzfeldt-Jakob disease in humans, scrapie in sheep, and bovine spongiform encephalopathy in cattle. All of these disorders are characterized by the accumulation of an abnormally folded isoform of the cellular prion protein PrP C , denoted PrP Sc , which represents the major component of infectious scrapie prions (1). The formation of PrP Sc from PrP C is accompanied by profound changes in structure and biochemical properties. PrP C , which is rich in ␣-helical regions, is converted into a molecule with mainly ␤-sheeted structure, and PrP Sc becomes partially resistant to proteolytic digestion (1)(2)(3)(4). During biogenesis PrP C transits through the secretory pathway and is modified by the attachment of two N-linked carbohydrate chains and a glycolipid (GPI) 1 anchor connecting it with the cell surface. In addition, two different transmembrane forms have been described (5). The conversion of PrP C into PrP Sc is thought to occur after PrP C has reached the plasma membrane or is re-internalized (6 -8), although the molecular mechanisms of the conversion reaction remains enigmatic. Biosynthesis of PrP C is a prerequisite for PrP Sc formation, as mice lacking PrP are resistant to scrapie (9). Studies in transgenic animals favor a model in which PrP C and PrP Sc interact directly, possibly in a complex with auxiliary factors (1,10).
The physiological function of the cellular prion protein is still unclear. It is mainly expressed on neurons (11) and is evolutionarily highly conserved (12,13). As PrP C can bind Cu(II) ions (14,15) and may have some superoxide dismutase activity (16), it could have a possible protective function against oxidative stress. Mice lacking PrP C show no obvious abnormalities in behavior and reproductivity (17,18). There are some reports of impairments of knockout mice on the neuronal level, such as altered long term potentiation and defects in GABA-ergic receptor-mediated synaptic inhibition (19,20). Other studies found no electrophysiological differences in the hippocampal region of PrP null mice compared with wild type mice (21). Attempts at further insight into the putative function of PrP C were made by searching for interacting proteins using various biophysical assays. Candidate proteins found were Bcl-2 (22), Hsp60 (23), and the 37-kDa LRP (24). Recently, PrP C has been suggested to play a role in signal transduction, as cross-linking leads to activation of the tyrosine kinase Fyn, which is involved in intracellular signaling (25).
As neurons have a high PrP expression, the highest probability of finding interactors of the prion protein should be given by screening proteins expressed in the brain. Therefore, we used a cDNA library derived from brain of adult BALB/c mice to perform a yeast two-hybrid assay with murine PrP.
We identified synapsin Ib, Grb2, and a yet unknown protein with high homology to several EST sequences, to be referred to as Pint1 (for prion interactor 1). Interaction with PrP was confirmed in mammalian cells by co-immunoprecipitation assays. Fractionation assays indicated a common subcellular compartment, where all of the proteins can be present. As synapsin Ib and Grb2 are both involved in neuronal signaling pathways, our findings strongly indicate that the function of PrP C is connected with signal transduction processes.

EXPERIMENTAL PROCEDURES
Generation of Plasmids-All plasmids used for the screening of the cDNA library were included in the Gal4-based yeast two-hybrid system (Matchmaker2 kit, CLONTECH). The ORF of murine PrP-A encoding aa 23-231 (without signal peptides) was amplified by polymerase chain reaction from genomic mouse DNA (derived from N2a cells) using appropriate primers providing BamHI and XhoI restriction sites and ligated in frame into the bait vector pAS2-1. An analogous procedure was used to generate N-and C-terminal fragments of PrP ranging from aa 23-100 or 90 -231, respectively. The mouse brain cDNA library was derived from BALB/c mice (CLONTECH). Inserts were cloned into pACT2 via EcoRI and XhoI sites. For the vaccinia virus expression system (see below), inserts were subcloned from the original plasmids via EcoRI and XhoI sites into the bait plasmid pGBKT7, which encodes an N-terminal c-Myc-derived epitope tag (Myc) and into the prey plasmid pGADT7, encoding an N-terminal hemagglutinin epitope tag (HA). Both contain a T7 promoter (in contrast to pAS2-1 and pACT2) between the Gal4 transcriptional activation/Gal4 DNA binding domains and the HA/Myc epitopes. The use of the T7 promoter leads to the expression of N-terminally tagged inserts without the mentioned Gal4 domains.
Yeast Two-hybrid Screen-For performing the yeast two-hybrid screen, we used a Gal4-based system (Matchmaker2, CLONTECH). The mouse brain cDNA library was purchased from CLONTECH and amplified. Saccharomyces cerevisiae strains Y190 and Y187 have been described (26). Y190 cells were transformed with the bait plasmid followed by large scale transformation of the library, plated on selective media lacking Trp, Leu, His, and Ura, and incubated for 5-10 days. Plates were analyzed by a ␤-galactosidase colony filter lift assay. Blue colonies were restreaked, and positive staining was confirmed by repeating the assay. After isolation of the pACT2 plasmids encoding the library clones, these plasmids were tested for autoactivation of the reporter gene in yeast by co-transforming them with either the empty bait plasmid pAS2-1 or the same plasmid encoding an unrelated negative control (lamin c). Activation of the reporter genes was further confirmed in the yeast strain Y187. Isolates resulting in blue staining without the presence of the PrP bait were excluded from further investigation. All other plasmids were sequenced, and a search for data base homologies was performed using the BLAST algorithm (27). Screening of parts of the library was performed twice independently.
Immunoblot Analysis of N2a and ScN2a Cells-Postnuclear supernatants were produced by rinsing the cells with PBS, followed by the addition of cold lysis buffer A (10 mM Tris-HCl, pH 7.5, 100 mM NaCl, 10 mM EDTA, 0.5% Triton X-100, 0.5% deoxycholate) for 5 min. Cell debris was removed by centrifugation at 14,000 rpm for 2 min. Supernatants were supplemented with proteinase inhibitors (5 mM phenylmethylsulfonyl fluoride, 0.5 mM Pefabloc, and aprotinin; Roche Molecular Biochemicals) and precipitated in 80% methanol at Ϫ20°C. The resulting pellet was solubilized in 200 l of TNE (50 mM Tris, pH 7.5, 150 mM NaCl, 5 mM EDTA), and the protein content was measured by a BCA assay (Pierce) with bovine serum albumin as standard before adding 100 l of SDS loading buffer (7% SDS, 30% glycerine, 20% Mercapto-ethanol, 0.01% bromophenol blue in 90 mM Tris-HCl, pH 6.8) and incubation at 95°C for 5 min. Equal amounts of protein were analyzed on 12.5% SDS-PAGE. Western blotting was performed by transferring the proteins to polyvinylidene difluoride membranes (Amersham Pharmacia Biotech). Hybridization and detection was performed as described below.
Transfection of BHK Cells Using a Vaccinia Virus-based Expression System-BHK cells were grown in BHK-21 medium (Glasgow medium, Life Technologies, Inc.) containing 10% fetal calf serum and antibiotics. At 75% confluence cells were infected with the vaccinia virus strain vT7-3 encoding recombinant T7 polymerase (29) at an multiplicity of infection of 3-5 for 45 min immediately prior to DNA transfection. The infection results in the production of the desired proteins in high amounts, as they are transcribed under the control of the T7 promoter independently of the cellular RNA polymerases. This makes radioactive labeling dispensable and reduces the unspecific binding of cellular proteins. Five g of DNA from both bait and prey plasmids were used for co-transfection of the cells using a calcium phosphate-based method (Stratagene). Cells were incubated for an additional 24 -48 h until harvesting. Transformation efficiency was controlled with a green fluorescent protein plasmid under control of a T7 promoter.
Co-immunoprecipitation-BHK cells co-expressing Myc-tagged PrP and a HA-tagged interactor detected by the yeast two-hybrid screening were rinsed twice with PBS and lysed for 5 min on ice with lysis buffer B (100 mM Tris, pH 8.0, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl, 10 mM phenylmethylsulfonyl fluoride). Cellular debris was pelleted and the supernatants were precleared with 50 l of pre-equilibrated Protein A-Sepharose (Amersham Pharmacia Biotech) for 1 h. The precleared lysates were then incubated with 1 g of HA antibodies (Santa Cruz) for 4 h at 4°C. Protein A-Sepharose was added and incubated for 1 h. Beads were washed five times with cold lysis buffer, and the precipitated proteins were eluted by incubation at 95°C for 5 min in 50 l of sample buffer and used in SDS-PAGE.
Co-precipitation of Full-length Grb2 and Mature PrP-The complete ORF of Grb2 was amplified by reverse transcription-polymerase chain reaction from murine RNA using appropriate primers and subcloned into the mammalian expression vector pcDNA3.1 (Invitrogen). Fulllength murine PrP ranging from aa 1 to 254 was also subcloned into this vector.
Point mutations generating the 3f4 epitope and the 3AV mutation were introduced by site-directed mutagenesis. The 3AV construct has three alanine to valine substitutions and has been described to increase the transmembrane forms of PrP (30). BHK cells were infected and transfected as described above. Cell lysates were subjected to co-precipitation in buffer B using a polyclonal Grb2 antibody (kindly provided by A. Ullrich). Detection of precipitated PrP/Grb2 complexes was done by immunoblotting with PrP mAb 3f4 (Signet Laboratories) and Grb2 mAb 3f2 (Upstate Biotechnology, Inc.).
Gel Electrophoresis and Immunoblotting-Immunoprecipitated proteins were subjected to SDS-PAGE on a gel containing 10% acrylamide and were blotted on a nitrocellulose membrane (Schleicher & Schuell) using a semidry method. Membranes were blocked with 5% nonfat dry milk in PBS containing 0.05% Tween 20. Monoclonal anti-c-Myc and anti-HA antibodies (Santa Cruz) were used at a dilution of 1:1000 in PBS, 0.05% Tween 20 for 2 h at room temperature. Antibodies against synapsin I (mAb A10C; Biozym) and Grb2 (Stressgen) were used at a dilution of 1:2000 and 1:2500, respectively. ␤-Actin antibodies (Sigma) were used 1:5000, the monoclonal PrP antibody 4F2 at 1:1000, and 3f4 at 1:5000. Detection was performed using peroxidase-coupled secondary antibodies and the ECL plus kit (Amersham Pharmacia Biotech).
Preparation of Microsomal Fractions-Confluent N2a and Rin2a cells were scraped off the culture plates in 0.5 M sucrose, 10 mM Tris, pH 7.4, 5 mM EDTA. Cells were disrupted on ice with a Dounce homogenizer, and the efficacy was tested by trypan blue staining. Sucrose was increased to 1.2 M and overlaid with three buffers containing 1.1, 1.0, and 0.5 M sucrose. This gradient was centrifuged for 2 h at 100,000 ϫ g, resulting in opaque bands at 0.8 M (Golgi fraction) and 1.15 M (ER fraction) sucrose. The purity of the fractions was tested by Western blotting with antibodies against Rab6 (Santa Cruz) and calnexin (Stressgen).
RNA Expression Analysis-A RNA dot blot (CLONTECH) containing mRNA from different mouse organs was used to identify tissue-specific expression of Pint1. DNA probes were prepared by labeling the plasmids containing the clones of interest with radioactive [ 32 P]dCTP (Amersham Pharmacia Biotech) using the Megaprime kit (Amersham Pharmacia Biotech). Hybridization and washing was performed with Express-Hyb solution (CLONTECH). Signals were detected by phosphorimaging. An analogous approach was used on a classical mRNA blot (Ambion) representing different murine tissue to determine the size of the mRNA.

Identification of Neuronal Proteins Interacting with the Cellular Prion Protein by a Yeast Two-hybrid Screen-We used a
yeast two-hybrid screen to identify proteins physically interacting with the cellular prion protein. By using a homologous combination of bait and prey proteins both derived from mouse, we improved the probability of detecting putative neuronal interactors of PrP C . The mature sequence of murine PrP C (residues 23-231; Fig. 1B) fused in frame to the Gal4-binding domain was used as bait to screen a cDNA library generated from murine brain fused to the Gal4-transactivation domain.
The screening of the library revealed several yeast clones resulting in positive signals in the ␤-galactosidase assay. Sequencing of the isolated plasmids and search for homologous sequences in data bases were performed. Most of the isolates were homologous to proteins described to give typically false positive results in a yeast two-hybrid assay, such as mitochondrial proteins or proteins of the citric acid cycle (31) and were therefore excluded from further investigations. However, some clones not belonging to the above groups showed an extremely high homology to known proteins and/or published gene sequences and were chosen for further biochemical characterization. One was coding for the murine growth factor receptorbound protein 2 (Grb2) (32). Grb2 contains a central SH2 domain flanked by an N-and a C-terminal SH3 domain ( Fig.  2A). Structurally our isolate includes about half of the central SH2 and the C-terminal SH3 domain from aa 85 to aa 217. The second clone represents the D/F region of murine synapsin Ib ranging from residue 460 to the stop codon at aa 670 (Fig. 2B). The third isolate was coding for an ORF encoding a protein of 162 aa residues (Fig. 3A). Significant parts of its sequence are identical to several EST clones (Fig. 3B). A stretch of 90 aa has an almost complete homology to one derived from human brain and one from human heart (GenBank accession nos. AAC19158 and AAH01072). The same part of the sequence additionally has a striking similarity to several other hypothetical proteins of humans and Caenorhabditis elegans. Of note, no similarity was found for the C-terminal part of the isolated protein. As this protein was not yet described in data bases, we designated it Pint1 (for prion interactor 1, GenBank accession number AY029599).
In summary, by performing a yeast two-hybrid screen with murine PrP-(23-231) and a murine neuronal cDNA library, we were able to identify several proteins that are expressed in neurons and have the ability to physically interact with PrP C in yeast.
PrP Co-precipitates with Synapsin Ib, Grb2, and Pint1 Recombinantly Expressed in Mammalian Cells-To confirm the interaction of the identified proteins with PrP by a method completely unrelated to the yeast interaction trap, we used a vaccinia virus-based expression system in mammalian cells. A vaccinia strain encoding the T7 polymerase was used to infect BHK cells harboring plasmids containing both PrP and the putative interactors under the control of a T7 promoter. The infection led to high expression of the desired proteins in the cytoplasm, fused to either an N-terminal c-Myc (bait) or HA epitope (c-DNA isolates). They were not fused to the Gal4 transcriptional activator or DNA binding domains that were present in yeast. Lysates of the cells were subjected to co- immunoprecipitation, and the co-precipitated bait protein (PrP) was detected by immunoblot. The full-length construct PrP-(23-231) was used to test whether the interaction of synapsin Ib, Grb2 and Pint1 could be reproduced in mammalian cells. A band indicating the co-precipitation of PrP-23-231 is visible with all three isolates (Fig. 4A, lanes 2, 4, and 6). No signal is detectable in cells co-expressing the yeast-isolates with the unrelated negative control (Myc-tagged rabies virus M protein; lanes 1, 3, and 5) and in cells expressing only Myctagged PrP-(23-231) or mock-infected cells (lane 7 and 8). Neither the bait nor the prey proteins showed an intrinsic affinity to Protein A-Sepharose or to the Myc epitope per se. Additionally, when lysates of cells that separately expressed PrP and the interactors were mixed after lysis, no signals indicating a co-precipitation of PrP were detectable (data not shown).
Our data show that PrP-(23-231) can specifically interact with synapsin Ib, Grb2, and Pint1, not only in yeast but also in mammalian cells.
Synapsin Ib and Grb2, but Not Pint1, Interact with Both the N-and the C-terminal Part of PrP-After the initial screening of the library with full-length PrP-(23-231), the identified clones were subjected to binding assays with truncated constructs of PrP to determine the region of interaction. Two constructs encoding the N-terminal (aa 23-100) or the C-terminal part of PrP (aa 90 -231) were used to test their ability to interact with the identified clones in BHK cells by co-immunoprecipitation. PrP-(23-100) includes the putative copper-binding octarepeats and, having no defined NMR structure, is thought to be highly flexible, whereas PrP-(90 -231) contains the ordered, mainly ␣-helical structure (33). Consequently, we tested our isolates for interaction with these two parts of the protein.
In summary, we have shown that synapsin Ib and Grb2 can physically interact with both PrP-(23-100) and PrP-(90 -231), whereas the binding motif to Pint1 is encoded in PrP-(90 -231). These findings indicate that PrP-(23-231) has at least two distinct sites enabling it to interact with synapsin Ib and Grb2.
Binding of Grb2 to PrP Is Independent of the Canonical SH3 Recognition Motif-SH3 domains are known to interact with proline-rich proteins bearing the consensus motif XPXXP. Such a site is present in wild type PrP at position 101-105, but absent in two mutations leading to Gerstmann-Strä ussler-Scheinker syndrome (GSS) in humans (P102L and P105L, respectively) (1). We showed above that Grb2 is able to bind both to PrP-(23-100) and to PrP-(90 -231). To avoid binding to the putative N-terminal binding site, we introduced the GSS mutations only in C-terminal PrP-(90 -231) and tested the influence on Grb2 binding by co-precipitation (Fig. 4B). As seen in lanes 9 -11, neither of the two mutations inhibited binding under these experimental conditions.
Consequently, interaction of Grb2 with the C terminus of wild type and mutated PrP seems to be independent of the presence of a classical XPXXP motif.
Co-precipitation of Mature and Authentic PrP with Fulllength Grb2-Next, we tested whether the mature and authentic forms of Grb2 and PrP also interact. Therefore, BHK cells were transfected with plasmids encoding either full-length Grb2 or PrP C including the N-and C-terminal signal sequences or both constructs were transfected simultaneously. Cells were lysed, and a co-immunoprecipitation was performed using polyclonal Grb2 antibodies (Fig. 6). By immunoblotting, no PrP C was detectable in precipitates of cells expressing only one pro-

FIG. 4. PrP co-immunoprecipitates with the yeast two-hybrid isolates in mammalian cells. A, co-immunoprecipitation of Myc-tagged
PrP-(23-231) with HA-tagged Y2H isolates. Vaccinia virus-infected BHK cells expressing the HA-tagged yeast two-hybrid isolates coding for synapsin Ib (Syn), Pint1 (Pint1), or Grb2 (Grb) and co-expressing either Myc-tagged PrP-(23-231) or Myc-tagged rabies virus matrix protein (RM) as a negative control were subjected to immunoprecipitation with rat anti-HA mAb, and the lysates were analyzed by immunoblot. Detection of the co-precipitated bait (in this case PrP-(23-231) or rabies M, respectively) was done by mouse anti-Myc mAb. PrP-(23-231) gives rise to a band of about 29 kDa in all lanes with the isolates (lanes 2, 4, and 6). No signal is detectable with the rabies M negative control (lanes 1, 3, and 5). Cells expressing only PrP-(23-231) without any prey constructs and mock-infected cells (lanes 7 and 8, respectively) 4 -6) is N-terminally truncated, and PrP-(23-100) (lanes 7-9) lacks the C-terminal part. Of note, mouse anti-HA mAb instead of rat mAb (as used in the previous experiment) was used for the co-immunoprecipitation, giving rise to an unspecific band of 25 kDa, which is also visible in the mock control (lane 10, marked by an asterisk). Molecular size markers are indicated on the right. tein (lanes 1 and 2) or in the mixed lysates of those cells (lane 5). This indicates that the observed interaction was present before lysis and is not just an artifact occurring during cell lysis. A strong PrP signal was visible in cells co-expressing Grb2 with wild type PrP C and PrP containing the transmembrane-favoring mutations 3AV (described in Ref. 30 (lanes 3  and 4). Additionally, two HA-tagged PrP variants (HA inserted at positions 23 and 92, respectively; lanes 6 and 7) co-precipitated with Grb2. Of note, the inserted tags resulted in slightly slower migration of PrP C in those lanes.
Our data indicate that not only PrP-(23-231), but also mature and authentic PrP C , can specifically interact with fulllength Grb2 in living mammalian cells.
Grb2 and Synapsin I Co-fractionate with PrP in Microsomal Fractions-To determine the subcellular compartment where a possible interaction of the detected interactors with PrP C could occur, we prepared microsomes from N2a cells and Rin2a cells (Fig. 7). These cell lines were chosen because they showed high expression of synapsin. First, the integrity of ER and Golgi fractions was tested by staining with the markers Rab6 and calnexin. Duplicate fractions representing either Golgi or ER vesicles were then analyzed for the presence of PrP, Grb2, and synapsin I. PrP C was found in Golgi fractions of N2a cells. No detectable PrP signal was present in the ER fractions. Grb2 was detectable solely in Golgi fractions of both Rin2a and N2a cells with no visible staining in ER. Strong signals for synapsin I were present in Golgi microsomes and weaker signals in ER fractions.
This assay indicates that the proteins do fractionate at identical subcellular compartments, presumably Golgi vesicles.
We conclude that the interaction of the proteins could not only occur on the outer leaflet of the cell membrane, but also at intracellular vesicles.
RNA Tissue Distribution of Pint1 Overlaps with That Known for PrP C -To identify the organs where Pint1 is expressed, we examined an mRNA dot blot representing different murine tissues by probing it with a Pint1 DNA fragment. A strong expression was detectable in brain as well as in heart, thyroid, and testis (Fig. 8). A faint signal was visible in muscle cells, liver, pancreas, and kidney. No transcripts were detectable in embryonic development stages and in ovary, uterus, eye, lung, and spleen. The hybridization of a classical Northern blot bearing mRNA of different murine tissue revealed that the length of the corresponding mRNA has a uniform size of about 3.8 kilobases in all tissues with a positive signal (data not shown). The tissue distribution of PrP mRNA has been published previously. A high expression was reported in brain, lung, kidney, and heart, and a weaker expression was seen in liver, spleen, and other organs (34).
Our data indicate that Pint1 is not only expressed in neuronal tissue, but also in a variety of other organs. The pattern is partially overlapping with the known mRNA tissue distribution of PrP C , indicating a coexpression in these organs.
Analysis of the Endogenous Expression of PrP, Grb2, and Synapsin I in Neuronal Cells-As we were able to show that synapsin Ib and PrP C can physically interact, we looked for cells with high endogenous levels of synapsin I. Different cell types were screened for the presence of synapsin I by immunoblot analysis. As expected, expression was detectable in cells of neuronal origin (e.g. neuroblastoma and hypothalamic cells) as well as in cells with neuroendocrine function, like insulinoma cells or ␤-cells of the pancreas, whereas cells derived from other tissues, like fibroblasts, lacked detectable synapsin I (data not shown). These findings are in line with the known distribution of synaptic-like vesicles in secretory cells (35).
By comparing normal and persistently prion-infected N2a cells, we investigated whether expression of synapsin I and PrP C are correlated to prion production. Our initial studies indicated a high level of synapsin I in persistently prion-infected ScN2a cells and a weaker expression in uninfected N2a cells (data not shown). However, a closer analysis showed that FIG. 6. Co-precipitation of Grb2 with mature, fully glycosylated PrP. BHK cells co-expressing full-length Grb2 together with different PrP-constructs were subjected to co-immunoprecipitation with a polyclonal Grb2 antibody. The precipitates were analyzed for the presence of PrP C with 3f4 mAb (A). The co-expressed proteins are indicated above the lanes. Lanes 1and 2 represent precipitates of cells expressing either Grb2 or PrP alone. Aliquots of these lysates were mixed and analyzed in lane 5. In all other lanes, a signal for coprecipitated full-length PrP C is detectable. B, Grb2 was detected as a control for successful precipitation. the increased expression of synapsin I was not strictly dependent on the presence of PrP Sc , but rose with the density of cultured cells (Fig. 9A); cells lysed at 20, 40, 80, and 100% confluence were analyzed by immunoblot for the amount of synapsin I. Analyzing the same lysates for PrP expression revealed also a density dependence both in infected and in uninfected cells (Fig. 9C). The intracellular amount of Grb2 was not influenced by the state of confluence (Fig. 9B). Other cell types also showed no difference in their Grb2 amount (data not shown). Of note, not only the levels of PrP but also those of synapsin I were higher in PrP-overproducing ScN2a cells as compared with wild type N2a cells. Again, Grb2 levels did not vary between infected and wild type cells.
In summary, we were able to show that expression of both synapsin I and PrP C is dependent on the cell density of cultured neuronal cells. DISCUSSION Several different screening methods have been performed previously to identify proteins that interact with PrP C or PrP Sc . These assays have revealed some more or less well characterized binding partners for the prion protein. Nevertheless, the function of PrP C is still enigmatic. A lot of these efforts have been made in heterologous systems with, e.g., PrP and cDNA libraries from differing species and prey proteins that were fused to several additional proteins, such as alkaline phosphatase or GST. We tried to circumvent this by using a highly homologous system, where both the PrP C and the cDNA library were derived from mouse thereby eliminating a possible "species barrier." The cDNA was derived from total murine brain, where PrP C is known to be strongly expressed. Finally, no additional fusion parts apart from the obligatory AD/BD domains were present. In the confirmation studies in mammalian cells, these domains were even completely absent.
As PrP C is a protein predominantly present on the outer leaflet of the plasma membrane, there remains the question how an extracellular protein can come into contact with intracellular counterparts. Recently, convincing evidence was presented that the prion protein has the unusual intrinsic ability to exhibit different membrane topologies (5). The majority of PrP is completely secreted into the ER lumen and attached to the outer cell membrane via a GPI anchor ( sec PrP). Nevertheless, transmembrane isoforms with either the N terminus or the C terminus in the ER lumen and the other end facing the cytosol have also been identified ( Ntm PrP and Ctm PrP, respectively). PrP Ctm represents a possible "loop-back" mechanism of the GPI-anchored form through the lipid bilayer (36). Therefore, interaction of PrP C with proteins present in the cytosol is possible and the search for these proteins was feasible.
Indeed, our screening revealed several potential interactors. They all belong to different protein classes and have clearly distinct purposes in the cell. A sequence comparison of the isolates did not reveal a shared motif that could represent a putative common PrP-binding site (data not shown). Obviously, PrP interaction is not restricted to a specific protein family.
Characterization of Pint1-The protein showing the strongest interaction with PrP has not been characterized previously. The complete ORF of this protein, now called Pint1, is still unknown. The identified sequence encodes a putative protein with a length of 120 amino acids. The original start codon is missing, and at present it is not known how long the N-terminal part might be. It is interesting that the coding region contains a stretch of ϳ50 amino acids showing a 100% identity to EST clones from human tissues and a strong homology to some sequences of C. elegans (Fig. 3B). Unfortunately, no proteins with definitely known functions are among those homologues, but the striking homologies of murine Pint1 to human and C. elegans sequences indicate a high evolutionary conservation and therefore an important function. Proteins harboring this conserved domain may belong to a common family. The lack of homology to the EST clones in the C-terminal part of the protein is not caused by a frameshift, but could be the result of alternative splicing. We investigated the tissue distribution of Pint1 by an RNA dot blot analysis. Because not all organs gave rise to specific signals, we concluded that this protein is not transcribed from a housekeeping gene. As expected, a strong signal was detectable in brain, but also in heart (in accordance to the second best IMAGE homologue), thyroid, and testis. No high expression in embryonal development was detectable, arguing against an essential function during embryogenesis. The expression pattern of Pint1 is partially overlapping with that known for PrP C ; therefore, an interaction of these two proteins could also occur in cells outside the central nervous system.
Synapsin I and Grb2 Are Feasible Candidates for Interaction with PrP-Synapsins are a family of highly conserved neuronal proteins, which contribute to the major class of proteins that are associated with small synaptic vesicles (37). They play an important role in the formation of synapses and in the regulation of neurotransmitter release (38). Their putative function is to cross-link synaptic vesicles to each other and to cytoskeletal proteins, thereby providing a mechanism for controlling vesicle release. This is regulated by phosphorylation by a range of diverse kinases, like calcium/calmodulin-dependent protein kinase I and protein kinase A. Synapsin I can be additionally phosphorylated at its C terminus by calcium/calmodulin kinase II, which is also associated with synaptic vesicles (39). As the PrP-interacting region of synapsin Ib contains some of these described phosphorylation sites, it could be possible that their interaction is also regulated by phosphorylation. Of note, recombinant PrP can also be the template for phosphorylation in vitro (40), although this has never been detected in vivo.
The putative biological significance of the PrP-synapsin interaction is further strengthened by the known tissue distribution of both proteins. Synapsin I is very abundant in neuronal tissues, but also in other cell types involved in exocytosis, such as cells of the endocrine system (35). The data we have obtained from cell culture studies suggest that there could be a coregulation of expression of PrP C and synapsin I in neuronal cells, as both increase with growing cell density. A possible explanation would be that both proteins contribute to mechanisms of cell-to-cell contact. Of note, one of the first symptoms of a scrapie-infected brain is synaptic loss (41), setting in before spongiosis and astrocytosis. Mice lacking synapsins are viable and show no abnormalities in gross well being (42). Of note, an electrophysiological increase in paired pulse facilitation and short term synaptic plasticity has been described (43), a phenomenon also observed in mice lacking PrP (19). Whether the course of scrapie disease is impaired in this mice is under investigation. Our findings could lead to the development of new therapeutic targets against the fatal prion diseases.
A strong indication for a likely in vivo interaction of the proteins also comes from our cell fractionation experiments, where we observed an co-fractionation of PrP C and synapsin I in Golgi fractions. Other groups detected intense PrP staining at synaptic vesicles and the presynaptic membrane by immuno-electron microscopy (44). Thus, the mentioned proteins are not only co-expressed, but also co-localize at nerve terminals, especially synaptic membranes. Therefore, interaction of PrP with synapsin I could be involved in the regulation of the release or the recycling of synaptic vesicles. Of note, synapsin I is a well characterized interactor of the third protein we found in our screen, namely Grb2 (45).
Grb2 is an adaptor protein involved in intracellular signal transduction. Its main role seems to be to link signals coming from extracellular and/or transmembrane receptors like neuronal and epidermal growth factor receptors to intracellular signaling molecules (46). Despite the absence of intrinsic enzymatic functions, Grb2 is essential for the formation of signaling complexes, as there are no viable animal knock-out mutants (47). The structure of Grb2 consists of a central SH2 domain flanked by two SH3 domains (Fig. 2A). The SH2 domain is responsible for interaction with tyrosine kinases (48), whereas the SH3 domains can bind to proline-rich motifs. The PrPinteracting region we isolated includes the complete C-terminal SH3 domain, but lacks the N-terminal SH3 domain. Interestingly, both N-terminal as well as C-terminal PrP-fragments can bind to Grb2. As the SH2 domain is not complete in the identified clone, we presume that the few amino acids of this domain do not contribute to the binding. Therefore, two distinct sites in the PrP molecule seem to be suitable for interaction with the C-terminal SH3 domain of Grb2.
Does PrP Meet the Requirements for Grb2 Binding?-The minimal consensus motif for SH3 domain recognition is XPXXP (49). It is present in PrP at position 102-105 (Fig. 1A). The two prolines in these motif are strictly conserved between all known wild type prion sequences of mammalian species without any exception (12,13). Interestingly, two inheritable pathological mutations in humans destroy just this minimal consensus sequence. P102L and P105L both result in GSS with slightly different clinicopathologic patterns (50). We introduced both mutations in our PrP constructs and analyzed whether the binding to Grb2 is influenced. We observed no difference in PrP binding under our experimental conditions, so we conclude that the recognition of PrP can be independent of the presence of a XPXXP motif. This has also been observed with some other Grb2 interactors (51). Thus, the pathogenicity of the GSS mutations is not likely to be caused by lack of this binding. Nevertheless, abolished interaction with other, not yet identified SH3-containing proteins may contribute to the disease. Therefore, agents influencing signal transduction might be considered as putative therapeutic drugs against prion diseases.
As the N-terminal PrP construct 23-100 also binds to Grb2, an additional structure recognized by Grb2 exists in this region. It has been shown that SH3 domains can bind to polyproline type II helices (52), which are characterized by bearing the consensus motif XPG. PrP has three XPG sequences in its N terminus (Fig. 1A, arrows). Indeed, the existence of a polyproline type II loop has recently been described for PrP C in vivo (53). This secondary structure may explain the interaction with PrP-(23-100). As for the XPXXP motif, the XPG sequences also show a complete conservation between all species.
The Newly Identified Proteins in the Context of Previously Described Interactors-It has been shown that PrP C co-immunoprecipitates with caveolin-1␤ and that cross-linking of PrP C leads to phosphorylation of the intracellular signaling molecule Fyn in differentiated 1C11 neuronal cells (25). The co-fractionation of Fyn with PrP C in detergent-resistant complexes has been described for N2a cells (54). As these cells lack caveolin, its presence is not essential for PrP expression. Detergentinsoluble complexes are derived to a large extent from rafts, which may be the equivalent or counterpart to caveolae-like domains. These cholesterol-rich membrane fractions are believed to play a major role in concentrating cell-signaling events. They even may give rise to synaptic vesicles in neurons at the recycling process (55). We showed that both synapsin I and Grb2 co-purify with PrP C in neuronal microsomal vesicles. Whether interaction occurs at exo-or endocytotic vesicles is not distinguishable by our fractionation experiment. However, experiments performed with epidermal growth factor receptor internalization found an association of Grb2 with endocytotic vesicles (56).
Several studies indicate that the cellular prion protein is a dimer in its functional form. In vivo data using covalent crosslinking on brain tissue showed convincing data that PrP dimers might form, but auxiliary factors may be required (57). The potential ability of full-length Grb2 to bind two PrP molecules (one with each SH3 domain) leads us to the hypothesis that this protein could act as a dimerizer of PrP.
In summary, using a yeast two-hybrid screen, we were able to identify neuronally expressed proteins that interact with the cellular prion protein. These interactions were confirmed in vivo by co-immunoprecipitation in mammalian cells. All PrP interactors are present at presynaptic nerve terminals, which are also the region of the highest PrP C concentration in neuronal cells. A major part of intracellular signaling is performed by phosphorylation events. Therefore, binding of PrP C to Grb2, which is involved in this signaling cascades and also to synapsin Ib, related to extracellular signaling, suggests a role for the cellular prion protein in both of these two crucial pathways. In particular, the interaction with an adaptor protein widens the possibility for PrP C to form complexes with a large variety of signaling molecules.
Taken together, our results indicate that PrP C could be involved in signal transduction. Therefore, the inhibition of these interactions might provide future therapeutic approaches against prion diseases.