Interactions of Prion Protein With Intracellular Proteins: So Many Partners and no Consequences?

Abstract

Prion protein (PrP) plays a key role in the pathogenesis of transmissible spongiform encephalopathies (TSEs)—fatal diseases of the central nervous system. Its physiological function as well as exact role in neurodegeneration remain unclear, hence screens for proteins interacting with PrP seem to be the most promising approach to elucidating these issues. PrP is mostly a plasma membrane-anchored extracellular glycoprotein and only a small fraction resides inside the cell, yet the number of identified intracellular partners of PrP is comparable to that of its membranal or extracellular interactors. Since some TSEs are accompanied by significantly increased levels of cytoplasmic PrP and this fraction of the protein has been found to be neurotoxic, it is of particular interest to characterize the intracellular interactome of PrP. It seems reasonable that at elevated cytoplasmic levels, PrP may exert cytotoxic effect by affecting the physiological functions of its intracellular interactors. This review is focused on the cytoplasmic partners of PrP along with possible consequences of their binding.

Introduction

Transmissible spongiform encephalopathies (TSEs) are a group of fatal neurodegenerative diseases including scrapie, bovine spongiform encephalopathy (BSE), and chronic wasting disease (CWD) in animals and Creutzfeldt–Jakob disease (CJD), Gerstmann–Sträussler–Scheinker disease (GSS), and fatal familial insomnia (FFI) in humans. More than 25 years ago these diseases were linked to prion protein (PrP) (McKinley et al. 1983; Prusiner et al. 1984; rev. in Prusiner 1998). Susceptibility to TSEs requires expression of the cellular prion protein (PrP^C), since organisms experimentally devoid of the Prnp gene encoding this protein are resistant to the infection (Büeler et al. 1993). It is also widely accepted that a misfolded form of PrP^C, termed PrP^Sc (the scrapie form of PrP), is the main or even sole component of the TSE infectious agent (rev. in Prusiner 1998). This pathogenic folding, characterized by an increased beta-strand content, leads to the reduction of PrP solubility accompanied by a gain of substantial resistance to proteinase K (PK) digestion. Despite many efforts, the molecular mechanisms of the PrP^C conversion into PrP^Sc and of the PrP-linked neurodegeneration remain elusive. Furthermore, although PrP is highly conserved in mammals, its physiological function is still unclear. PrP^C is ubiquitously expressed in many cell types with the highest level reached in the nervous system (Bendheim et al. 1992). This protein is translated as a 253-aa polypeptide (human PrP) containing at the both termini signal sequences of ca. 22 amino acids each (Fig. 1). The N-terminal signal peptide (ER SP) is removed during entry into the lumen of the endoplasmic reticulum (ER) whereas the C-terminal signal (GPI SP) is cleaved off upon attachment of a glycosylphosphatidylinositol (GPI) moiety (Basler et al. 1986; Stahl et al. 1987; Turk et al. 1988). Additionally, inside the ER, PrP^C may be variably glycosylated at two asparagine residues (N181 and N197) giving rise to di-, mono- and un-glycosylated species (Haraguchi et al. 1989). The N-terminal part of mature PrP^C is basic and unstructured, whereas the C-terminal part forms a slightly acidic globular domain stabilized by a single disulfide bridge (Riek et al. 1997). The central region of the polypeptide, encompassing residues 112–135, is hydrophobic and may function as a transmembrane domain (TM) (Lopez et al. 1990). The N-terminal sequence is flanked (residues 23–28 and 101–106) by cryptic nuclear localization signals (NLS) (Gu et al. 2003). In the N-terminal part, five octapeptide repeats (OR) are localized which constitute the major Cu²⁺-binding site (Hornshaw et al. 1995). Additional copper-binding sites are localized C-terminally to the ORs (Jackson et al. 2001; Jones et al. 2004). Normal proteolytic processing at residues 110–112 generates fragments termed N1 and C1 corresponding to the above-mentioned unstructured and globular parts of the PrP^C molecule, respectively (Chen et al. 1995; Jiménez-Huete et al. 1998). Additional fragments called N2 and C2 are produced in the brain of TSE-infected organisms by proteolytic cleavage at residues 90/91 (Chen et al. 1995).

Fig. 1.

Scheme of human PrP. Signal peptides governing entry into ER lumen (SP ER) and attachment of GPI (SP GPI), octarepeats (OR), transmembrane domain (TM), two nuclear localization signals (NLS), disulfide bridge (S–S), GPI attachment site, two glycosylated asparagine residues (N-Glyc), three α-helices and two β-strands are indicated. Sites of proteolytic processing related to the maturation of PrP and generation of two shortened forms C1 and C2 are marked with scissors

Numerous studies have focused on the identification of molecules interacting with PrP^C or PrP^Sc. Upon the discovery of Cu²⁺ binding by PrP^C accompanied by a superoxide dismutase-like activity it was proposed that this protein may be involved in the homeostasis of copper ions and protection against oxidative stress (Brown et al. 1999). The basic N-terminal region of PrP^C has also been shown to interact with glycosaminoglycans (Caughey et al. 1994; Pan et al. 2002; Warner et al. 2002). Interestingly, PrP^C is a nucleic acid-binding protein capable of interacting with DNA as well as RNA (Nandi 1997; Weiss et al. 1997; Gabus et al. 2001). As a cell-surface molecule, PrP interacts with extracellular or plasma-membrane proteins such as, e.g., dystroglycan, vitronectin, caveolin-1, laminin, laminin receptor/precursor, apolipoprotein, and neural cell adhesion molecules (Keshet et al. 2000; Hajj et al. 2007; rev. in Vana et al. 2007). Basing on some of these observations, PrP^C has been proposed to take part in neuronal differentiation, signal transduction, cytoprotection, and cell adhesion. Surprisingly, it has also been demonstrated that PrP may interact with numerous intracellular proteins, of which the majority are found in the cytosol (summarized in Table 1). This issue will be reviewed in detail below.

Table 1.

Intracellular partners of PrP. The table indicates which PrP form was found to bind to the interactor, how the interaction was detected, in which compartment may the interaction occur, and its effect

PrP interactor	Interacting form of PrP/technique	Localization	Effect of interaction	Reference
Neuroglobin	rPrP/affinity	Cytosol	Aggregation	Lechauve et al. (2009)
Synapsin 1b	rPrP, PrPC/Y2H, co-IP, co-fractionation	Cytosol	?	Spielhaupter and Schätzl (2001)
Grb2	rPrP, PrPC/Y2H, co-IP, co-fractionation	Cytosol, nucleus	?	Lysek and Wüthrich (2004), Spielhaupter and Schätzl (2001)
GFAP	rPrP, PrPC, PrPSc/pull-down, overlay, co-IP	Cytosol	?	Dong et al. (2008b), Oesch et al. (1990)
Tubulin	rPrP, PrPC, PrPSc/pull-down, co-IP, co-fractionation, cross-linking, affinity	Cytosol	Oligomerization, aggregation, inhibition of microtubule assembly	Brown et al. (1998), Brown (2000), Dong et al. (2008a), Giorgi et al. (2009), Hachiya et al. (2004a), Keshet et al. (2000), Nieznanski et al. (2005), Nieznanski et al. (2006), Osiecka et al. (2009)
Tau	rPrP, PrPC, PrPSc/pull-down, co-IP	Cytosol	Reduction of binding to tubulin	Han et al. (2005), Han et al. (2006), Wang et al. (2008)
hnRNP A2/B1	rPrP, PrPC/overlay, co-IP	Nucleus/cytosol	?	Strom et al. (2006)
Aldolase C	rPrP, PrPC/overlay	Cytosol	?	Strom et al. (2006)
Bcl-2	rPrP, cytoPrP/Y2H, co-IP, affinity	Cytosol	Aggregation, loss of Bcl-2 function, induction of apoptosis	Kurschner and Morgan (1995), Rambold et al. (2006)
NRAGE	rPrP, cytoPrP, PrPC/pull-down, Y2H, co-IP	Cytosol	Aggregation, affected mitochondrial membrane potential	Bragason and Palsdottir (2005)
Nrf2	rPrP/screen of bacteriophage expression library of brain cDNA	Cytosol, nucleus	?	Yehiely et al. (1997)
CK2	rPrP, PrPC/pull-down, overlay, co-IP, SPR	Cytosol, nucleus, extracellular matrix	Increase of enzymatic activity, phosphorylation of PrP	Meggio et al. (2000), Negro et al. (2000)
14-3-3 protein	rPrP, PrPC, PrPSc ?/pull-down, overlay, co-IP	Cytosol	?	Mei et al. (2009), Satoh et al. (2005)
Mahogunin	cytoPrP, CtmPrP/pull-down, affinity	Cytosol	Aggregation, loss of mahogunin function, neurodegeneration	Chakrabarti and Hegde (2009)
HOXA1	PrPC/microarray, co-IP	Nucleus	?	Satoh et al. (2009)
MPG	PrPC/microarray, co-IP	Nucleus	?	Satoh et al. (2009)
PLK3	PrPC/microarray, co-IP	Cytosol, cell membrane	?	Satoh et al. (2009)

The abbreviations used in this table are: co-IP co-immunoprecipitation, SPR surface plasmon resonance, Y2H yeast two-hybrid screen

Intracellular Localization of PrP

Usually most molecules of the PrP are found anchored by the GPI moiety in detergent-resistant raft domains on the cell surface and hence it is commonly considered extracellular (Stahl et al. 1990; Naslavsky et al. 1997). This is, however, an incomplete picture of PrP biology. Subpopulations of neurons in the hippocampus, thalamus, and neocortex have been found where PrP is localized predominantly to the cytosol (Mironov et al. 2003). This fraction of the protein, not associated with membranous structures, was designated cytPrP. Moreover, due to the central hydrophobic sequence and inefficient translocation, PrP may cross the ER membrane only partially and adopt either of two transmembranal topologies termed ^CtmPrP and ^NtmPrP (Hegde et al. 1998; Yost et al. 1990; Kim et al. 2001; rev. in Chakrabarti et al. 2009). In the former, the N-terminal part of PrP, retaining its signal sequence, is exposed to the cytosol whereas the C-terminal domain resides in the ER lumen. In the latter, the N-terminal region of PrP is inside the ER while the C-terminal domain faces the cytosol. In the brain of healthy organisms, these forms do not exceed 10% whereas in TSE they may constitute up to 30% of all PrP molecules. Mutations within or near the hydrophobic domain are responsible for increased generation of ^CtmPrP. Particularly, the GSS-linked mutations P105L and A117V favor production of ^CtmPrP (Hegde et al. 1998; Kim and Hegde 2002). ER stress and other pathogenic mutations evoking incorrect translocation of PrP may lead to its accumulation in a fully cytosolic form (rev. in Miesbauer et al. 2009). There is no uniform nomenclature for this cytosolically mistargeted protein and in various reports it is termed cyPrP, cytPrP, or cytoPrP, irrespectively of its origin. It was reported that GSS-linked mutants W145Stop and Q160Stop exhibited reduced translocation into the ER and were detected in the cytosol and nucleus (Zanusso et al. 1999; Heske et al. 2004). Associated with FFI and CJD in humans, PrP mutant D177N may be mistargeted to the cytosol as a result of retrograde translocation from the ER (Ma and Lindquist 2001). Usually, in a quality control process, the retro-translocated PrP undergoes degradation by the ubiquitin–proteasome system (Yedidia et al. 2001). Inhibition of the proteasome activity led to accumulation of aggregated PrP in the cytosol (Ma and Lindquist 2001; Ma et al. 2002). Interestingly, proteasome inhibition is observed in TSE-infected cells. Notably, PrP^Sc has been shown to inhibit the proteasome (Kristiansen et al. 2007). It was also demonstrated in yeast two-hybrid screen that PrP interacted with proteasome subunit alpha type-3 (Lim et al. 2006). Furthermore, as mentioned above, the NLS sequences may target PrP into the nucleus (Gu et al. 2003). In scrapie-infected cells, PrP^Sc accumulates in the nucleus and this localization requires a microtubule-dependent transport (Mangé et al. 2004). Hence, particularly at pathologic conditions, intracellular proteins may be exposed to interactions with PrP. Importantly, numerous studies have demonstrated that cytosolic PrP may be neurotoxic (Hegde et al. 1998; Ma et al. 2002; Grenier et al. 2006). Therefore, analysis of these interactions may be critical for elucidation of the molecular mechanism of TSE-related neurodegeneration and also the physiological function of PrP^C.

Intracellular Partners of PrP

Below are listed intracellular proteins shown to interact with PrP, particularly those partners for which published reports may imply direct binding. Brief description of the physiological functions of the interactors is provided to facilitate prediction of possible consequences of their binding to PrP.

Neuroglobin

Neuroglobin is a heme protein distantly related to myoglobin and hemoglobin, present in the cytosol of retina and brain neurons (rev. in Greenberg et al. 2008). This protein may be involved in neuroprotection in response to ischemia and hypoxia, and against amyloid-beta toxicity in Alzheimer’s disease.

Recently, a direct interaction between neuroglobin and PrP was demonstrated by light-scattering-monitored co-aggregation of purified proteins (Lechauve et al. 2009). Interestingly, the aggregates were formed without substantial changes in the PrP secondary structure, as judged by infrared spectroscopy. The unstructured N-terminal part of PrP was responsible for the binding. The authors did not observe aggregation using the C-terminal domain of PrP or when full-length PrP was mixed with myoglobin. Formation of aggregates with neuroglobin was reduced by increased salt concentration, indicating an ionic character of the binding. The interaction seems to occur in vivo since both proteins were found to colocalize in the perinuclear region of retina cells. It remains to be elucidated whether co-aggregation with PrP may affect the neuroprotective functions of neuroglobin in the cell.

Synapsin

Synapsins are a family of adaptor-like phosphoproteins which can bind to the surface of small synaptic vesicles as well as to the cytoskeletal structures: actin filaments and microtubules (rev. in Ferreira and Rapoport 2002). These proteins are found in the cytoplasm of nerve terminals and play an important role in synaptogenesis, synaptic vesicle turnover, and neurotransmitter release.

Interaction of PrP23–231 with synapsin 1b was discovered by means of a yeast two-hybrid screen of murine brain cDNA library (Spielhaupter and Schätzl 2001). This observation was further confirmed by co-immunoprecipitation from kidney cells transfected with plasmids encoding both proteins. The binding sites for synapsin were localized in the N- and C-terminal parts of PrP. Notably, both proteins are present at high concentrations in the presynaptic nerve terminals. Furthermore, synapsin co-fractionated with PrP^C in the microsomal fractions of N2a cells, suggesting that these proteins may interact in vivo at intracellular vesicles. One can speculate that PrP, by interacting with synapsin, may affect synaptic signal transduction. Interestingly, reduced level of synapsin 1 was detected in the brain of scrapie-infected mice (Sisó et al. 2002).

Grb2

Growth factor receptor-bound protein 2 (Grb2) is a key molecule in the transduction of intracellular signal which links activated cell surface receptors to downstream targets of the Ras signaling pathway (rev. in Giubellino et al. 2008). This adapter protein is ubiquitously expressed and found both in cytosolic and nuclear fractions. Grb2-mediated signaling is critical for actin-based cell motility and cell cycle progression; it is also involved in angiogenesis, vasculogenesis, and oncogenesis.

Interaction of PrP with Grb2 was first reported by Spielhaupter and Schätzl (2001), as was the interaction with synapsin (see above). As for synapsin, the interaction with Grb2 was identified in the yeast two-hybrid screen and confirmed by co-immunoprecipitation and co-fractionation. In the immunoprecipitation assays, the binding sites for Grb2 were initially localized both in the N- and C-terminal parts of PrP. Surprisingly, PrP mutations P102L and P105L, linked to GSS and localized within the potential Grb-2-binding motif XPXXP, had no effect on co-immunoprecipitation of the proteins. Subsequently, the binding site for C-terminal SH3 (Src homology domain 3) domain of Grb2 was mapped within the PrP sequence 100–109 by means of NMR, circular dichroism, and fluorescence spectroscopy (Lysek and Wüthrich 2004). For this binding a K _d value of 5.5 μM was determined. Contrary to the previous report, in experiments on purified recombinant PrP90–231, the P102L and P105L substitutions abolished the interaction. Since the docking surfaces of Grb2 and PrP bear complementary charges the binding seems to be electrostatic. Notably, Grb2 interacts also with synapsin, which enables formation of a complex composed of all three proteins. The binding to Grb2 and synapsin suggests an influence of PrP on diverse signal transduction pathways.

Tubulin

The recent decade of studies has provided numerous reports evidencing association and even direct interaction of PrP with tubulin—the major component of microtubular cytoskeleton. Heterodimers composed of α- and β-tubulin assemble into tubular polymers called microtubules (rev. in Nogales 2000). Networks of these structures form the cellular cytoskeleton and serve as tracks for active transport. Besides intracellular trafficking and determination of cell morphology, microtubules play a pivotal role in chromosome segregation, cell division, secretion, and cell motility. Microtubules are very dynamic structures that can switch stochastically between assembly and disassembly phases. This property, termed dynamic instability, is crucial to microtubule functions. The assembly and stability of microtubules are regulated by a numerous group of molecules referred to as microtubule associated proteins (MAPs) (rev. in Amos and Schlieper 2005).

The first study implying an association of PrP with tubulin was performed employing a synthetic peptide corresponding to PrP residues 106–126 (Brown et al. 1998). This neurotoxic peptide bound to α- and β-tubulin from brain extracts. Furthermore, it was shown by turbidity measurements that PrP106–126 inhibited tau-dependent microtubule assembly (Brown 2000). The peptide carrying mutation A117V linked to GSS was much more efficient in the inhibition than the wild-type PrP106–126. In the absence of tau neither peptide affected tubulin polymerization. Surprisingly, in subsequent studies (see below) this region of PrP molecule has been shown to lack a tubulin-binding ability. One cannot exclude that in the earlier studies the PrP106–126 peptide interacted in fact with tau protein bound to tubulin. In subsequent studies, complexes of tubulin with native full-length PrP were also reported but interpreted as a result of an indirect interaction (Keshet et al. 2000) or even an artifact of the experimental procedures used (Schmitt-Ulms et al. 2004). Colocalization with the microtubular cytoskeleton has been demonstrated for an N-terminal fragment of PrP produced proteolytically in neuroblastoma N2a cells (Hachiya et al. 2004a). In the same study, co-immunoprecipitation of this part of the PrP molecule with tubulin was observed in homogenates of N2a cells. The PrP region encompassing residues 1–91 was identified as responsible for the colocalization with microtubules. Moreover, nocodazole treatment that disrupts the microtubular network, but not latrunculin A sequestering monomeric actin, affected the intracellular distribution of PrP. The association of PrP with microtubules was proposed to be related to the active transport of this protein in neurons. It was shown that retrograde and anterograde transport of PrP is dependent on microtubule-associated molecular motors belonging to the dynein and kinesin families (Hachiya et al. 2004b). Employing deletion mutants, PrP sequences 23–33 and 53–91 were identified as indispensable for the retrograde and anterograde transport of the molecule, respectively. Also, nuclear targeting of PrP^Sc has been shown to require an intact microtubule network (Mangé et al. 2004). By covalent cross-linking of purified proteins, we have demonstrated for the first time a direct interaction between PrP and tubulin (Nieznanski et al. 2005). Moreover, we found that PrP co-sedimented with microtubules. A convincing evidence for the in vivo interaction was co-purification of PrP with tubulin from brain homogenate, following a specific multi-step procedure allowing isolation of highly purified tubulin. It is commonly accepted that such a preparation contains only traces of MAPs—molecules of physiological function tightly bound to tubulin. Only the full-length PrP, but not its in vivo-generated proteolytic fragment C1 lacking residues 23–110/112, co-purified with tubulin. In a subsequent study, we showed that the direct binding of PrP to tubulin led to the inhibition of microtubule assembly by inducing tubulin oligomerization (Nieznanski et al. 2006). Furthermore, we observed that the oligomers associated into large aggregates. Employing proteolytic fragments of PrP and a panel of PrP deletion mutants, we identified two microtubule-binding regions within the N-terminal part of the molecule, with residues 23–32 constituting the major binding site and residues 101–110 representing a weak one (Osiecka et al. 2009). Moreover, we found that the sequence 23–32 was responsible for inducing tubulin oligomerization and inhibition of microtubule formation. The effects of PrP were also confirmed at the cellular level (Osiecka et al. 2009). The sequence 23–30, guided into the cytoplasm by the signal peptide 1–22, caused disintegration of the microtubular cytoskeleton and congregation of tubulin in the perinuclear area of human epithelial cells. It is plausible that at low cytosolic concentrations PrP may, similarly to some MAPs, modulate the dynamic instability of microtubules, whereas at high, pathologic levels it may aggregate tubulin causing disassembly of the microtubular cytoskeleton and cell death. Notably, it was demonstrated that tubulin co-immunoprecipitated with PrP^Sc in brain homogenates of scrapie-infected hamsters (Dong et al. 2008a).

Tau

Tau belongs to a family of structural MAPs that regulate the dynamics of microtubular cytoskeleton (rev. in Avila et al. 2004). Tau is a phosphoprotein mainly expressed in neural cells where it promotes assembly of microtubules and is engaged in their stabilization. Phosphorylation of tau at specific residues negatively regulates its binding to tubulin. Additionally, hyperphosphorylated-tau aggregates into insoluble neurofibrillary tangles which are a hallmark of Alzheimer’s and other neurodegenerative diseases.

Interaction between purified recombinant tau and PrP as well as between the native proteins in brain extracts has been demonstrated by means of pull-down assays and co-immunoprecipitations (Han et al. 2006). It was also shown that tau may bind PrP^Sc in brain extracts from scrapie-infected hamsters. Furthermore, tau and PrP colocalized in Chinese hamster ovary (CHO) cells co-transfected with plasmids encoding both proteins. The interaction site was roughly mapped within the PrP region 23–91. Deletion of the octarepeats significantly reduced the binding, implying an involvement of this region of the PrP molecule. Surprisingly, the PrP–tau binding was strengthened by introduction of the P102L mutation linked to GSS (Wang et al. 2008). The interaction sites with PrP were mapped within tau regions 1–91 and 186–283 (Wang et al. 2008), of which the latter is also involved in binding to tubulin (rev. in Avila et al. 2004). Therefore, the previously reported reduction in tau–microtubule association in the presence of PrP (Han et al. 2005) may result from blocking of the tubulin-binding region of tau by PrP. Since the tau region 186–279 is also known to promote microtubule assembly (rev. in Avila et al. 2004), its interaction with PrP may affect the physiological function of this MAP and hence cell viability.

GFAP

Glial fibrillary acidic protein (GFAP) is a major component of intermediate filaments (rev. in Messing and Brenner 2003). This cytoskeletal protein is abundantly expressed in astrocytes where it is involved in determination of cell shape.

GFAP was one of the first identified cytosolic interactors of PrP. It was demonstrated in an overlay assay (ligand blotting) that both purified PrP^C and PrP^Sc bound to GFAP from hamster brain homogenates (Oesch et al. 1990). Recombinant GFAP did not bind to PrP in those studies, which was explained by the lack of an N-terminal sequence of GFAP and/or a lack of posttranslational modifications of GFAP. The interaction between PrP and GFAP was subsequently confirmed by co-immunoprecipitations from normal and scrapie-infected brains and in pull-down experiments with purified recombinant proteins (Dong et al. 2008b). The interaction site on PrP was mapped within the C-terminal fragment encompassing residues 91–230. Accordingly, PK-treated PrP^Sc, corresponding to the globular domain, retained the ability to interact with GFAP. It is not known whether PrP may affect formation of intermediate filaments and what could be the physiological or pathological effect of this interaction. Interestingly, GFAP was found to be more abundant in the brain of scrapie-infected than healthy animals (Dormont et al. 1981; Mackenzie 1983). On the other hand, mice deficient in GFAP did not exhibit changes in TSE susceptibility or production of PrP^Sc (Tatzelt et al. 1996).

hnRNP A2/B1 and Aldolase C

In an overlay assay, it was also found that PrP may interact with hnRNP A2/B1 and aldolase C (Strom et al. 2006). Interestingly, all those proteins are able to bind RNA. Heterogeneous nuclear ribonucleoprotein A2/B1 (hnRNP A2/B1) belongs to a family of RNA-binding proteins involved in mRNA nucleo-cytoplasmic transport and splicing (rev. in Weighardt et al. 1996). hnRNP A2/B1 is expressed in various cell types, including neurons. Aldolases are glycolytic enzymes catalyzing the aldol cleavage of fructose-1,6-bisphosphate. Aldolase C/zebrin II is an isoform expressed predominantly in astrocytes and Purkinje cells (Thompson et al. 1982) which also exhibits mRNA binding activity (Cañete-Soler et al. 2005).

On ligand blots, purified recombinant PrP23–231 bound to the both proteins from the cytosolic fraction of brain homogenate obtained from Prnp-knockout mice (Strom et al. 2006). Moreover, hnRNP A2/B1 co-immunoprecipitated with PrP^C in the cytosolic fraction of human Hodgkin’s lymphoma cells. It is plausible that the binding to hnRNP may be related to the postulated involvement of PrP in nucleic acid metabolism. Further studies are required to determine the consequences of these interactions. Interestingly, in the brain of BSE-infected mice at the terminal stage of disease, an increased concentration of a long transcript of the aldolase C gene was reported (Dandoy-Dron et al. 2000).

Nrf2

Nuclear factor erythroid 2-related factor 2 (Nrf2) is a member of the family of basic leucine zipper transcription factors, which controls expression of antioxidant and detoxifying genes (rev. in Motohashi and Yamamoto 2004; Nguyen et al. 2009). Nrf2 is sequestered in the cytoplasm by the actin-associated protein Keap 1. In response to oxidative stress, Nrf2 is released from Keap 1, which allows for its translocation from the cytoplasm to the nucleus and activation of target gene expression. Nrf2 is ubiquitously expressed in a wide range of tissues and, notably, is involved in neuroprotection.

Interaction with Nrf2 has been found in a screen of a bacteriophage expression library of mouse brain cDNA using mouse PrP1–232 fused with alkaline phosphatase as a probe (Yehiely et al. 1997). At present, it is equally plausible that PrP may interact with the cytosolic and the nuclear fraction of Nrf2. It would be of particular interest to check whether PrP can tether Nrf2 in an inactive form in the cytosol.

Bcl-2

B-cell lymphoma 2 protein (Bcl-2) belongs to the group of molecules that regulate cell survival and death (rev. in Wong and Puthalakath 2008; Szegezdi et al. 2009). Bcl-2 is an anti-apoptotic protein which binds to pro-apoptotic members of the Bcl-2 family and antagonizes their functions. Bcl-2 prevents the release of cytochrome c into the cytoplasm thereby blocking activation of caspase cascade and apoptotic death. Furthermore, Bcl-2 is involved in the regulation of intracellular calcium homeostasis. This ubiquitously expressed oncoprotein is anchored on the cytoplasmic face of the nuclear envelope, ER, and mitochondrial membrane. Notably, expression of Bcl-2 is induced in the brain following ischemia (rev. in Soane and Fiskum 2005). This observation is related to the Bcl-2 neuroprotective functions that prevent ischemic death of neurons.

Selective binding of PrP to Bcl-2 has been demonstrated in a yeast two-hybrid screen of a murine cerebellar cDNA library (Kurschner and Morgan 1995). In this system, mouse PrP72–254 interacted with Bcl-2 but not with Bax or A1—other members of the Bcl-2 family. Surprisingly, the authors were unable to demonstrate co-immunoprecipitation of Bcl-2 with PrP. Subsequent studies employing PrP constructs directed to different cellular compartments showed that cytosolically localized PrP lacking the N- and C-terminal signal sequences associated with Bcl-2 (Rambold et al. 2006). Bcl-2 and this cytosolic PrP could be co-immunoprecipitated from transiently transfected neuroblastoma cells. Notably, Bcl-2 co-aggregated with cytosolic PrP indicating that this protein may sequester Bcl-2 leading to a loss of its anti-apoptotic function. In fact, PrP aggregated in the cytosol induced apoptosis in neurons. Overexpression of Bcl-2 interfered with the toxicity of the cytosolic fraction of PrP. Binding of cytosolic PrP to Bcl-2, recruitment of Bcl-2 into PrP aggregates, as well as PrP-induced apoptosis were prevented by chaperones Hsp70 and Hsp40, which were also shown to interact with PrP. Notably, aggregation of Bcl-2 may be related to the pathology of TSE since, in contrast to healthy animals, Bcl-2 has been found in a detergent-insoluble fraction of brain extracts of scrapie-infected mice. According to the results of Kurschner and Morgan (1995), Bcl-2 did not co-immunoprecipitate with PrP from cells expressing wild-type PrP. Only PrP missorted to the cytosol was cytotoxic. Inhibition of proteasome significantly enhanced the apoptotic effect of cytosolic PrP. PrP constructs targeted to the nucleus, mitochondria, or ER did not affect cell viability. Experiments on deletion mutants, including the TSE-linked mutants Q160Stop and W145Stop, have revealed that region 115–146 of PrP is responsible for the toxicity of the cytosolic protein. In light of the above observations, it has been postulated that the cytotoxicity of cytosolic PrP results from the co-aggregation of Bcl-2 and loss of Bcl-2 function.

NRAGE

Neurotrophin receptor-interacting MAGE homolog (NRAGE), also known as Dlxin-1 or MAGE-D1, is a pro-apoptotic member of the melanoma-associated antigen family (MAGE) (Salehi et al. 2000). NRAGE interacts with the cytoplasmic domain of the p75 neurothropin receptor, which induces activation of caspases and cell death through the Jun N-terminal kinase (JNK)-dependent mitochondrial pathway (Salehi et al. 2002). Notably, NRAGE is involved in the regulation of the cell cycle and apoptosis during neuronal development.

In a yeast two-hybrid screen of a rat brain cDNA library, Bragason and Palsdottir (2005) have demonstrated that PrP22–231 may interact with NRAGE. The interaction was confirmed in a GST pull-down assay for purified recombinant proteins and by co-immunoprecipitation from COS7 cells co-transfected with plasmids encoding NRAGE and a cytosolic construct of PrP lacking both signal sequences. The docking site for NRAGE was localized to the C-terminal globular domain of PrP (residues 122–231). After proteasome inhibition, PrP^C containing the signal sequences and NRAGE co-aggregated together in the cytosol of co-transfected COS7 cells. Importantly, co-expression of NRAGE and the cytosolic construct of PrP in neuroblastoma (N2a) cells affected mitochondrial membrane potential, which is an indicator of the early stages of apoptosis. This is another study demonstrating that interaction of PrP with a cytosolic protein may induce cell death.

CK2

Casein kinase 2 (CK2) is a pleiotropic Ser/Thr protein kinase, under certain conditions able to phosphorylate tyrosyl residues as well (rev. by Pinna 2002). CK2 is found in almost all tissues and nearly every compartment of eukaryotic cells (rev. in Faust and Montenarh 2000). This ubiquitous enzyme is involved in the regulation of numerous crucial processes including signal transduction, apoptosis, transcription, tumorigenesis, cell division, and proliferation. Furthermore, CK2 plays important roles in regulation of actin and tubulin cytoskeletons (rev. in Canton and Litchfield 2006). CK2 is composed of two constitutively active catalytic subunits and two regulatory ones.

Employing an overlay assay and the surface plasmon resonance technique, Meggio et al. (2000) have demonstrated that PrP interacts with the catalytic subunits of CK2. A K _d of approx. 0.5 μM was calculated for this binding. The C-terminal part of bovine PrP composed of residues 105–242 was identified as an interaction site. Importantly, PrP binding stimulated the catalytic activity of the kinase. For this stimulation, the 25–116 fragment of PrP was responsible, most probably through its N-terminal basic sequence. PrP did not bind to casein kinase 1, highlighting specificity of the interaction. Moreover, it has been demonstrated that bovine PrP is an in vitro substrate of CK2 (Negro et al. 2000). However, the interaction is not a simple substrate binding by the enzyme, since mouse PrP which lacks a phosphorylatable site still interacts with CK2. The interaction was subsequently confirmed in pull-down assays performed for purified recombinant proteins, as well as in co-immunoprecipitation of native proteins from brain extracts (Chen et al. 2008a). It was also shown that CK2 interacted more strongly with unglycosylated PrP than with its glycosylated forms. Furthermore, changes in the expression pattern of CK2 subunits were observed in the brain of scrapie-infected animals and human TSE cases (Chen et al. 2008b). Since CK2 is also an ectokinase found in the extracellular matrix (Canton and Litchfield 2006), it is possible that it may interact not only with cytoPrP or nuclear PrP, but also with PrP^C anchored on the plasma membrane. Further studies are required to check whether PrP enhances CK2 activity also in the cell.

14-3-3 protein

The family of 14-3-3 proteins consists of seven isoforms (β, γ, ε, σ, ζ, τ, and η) expressed most abundantly in neurons and glial cells of the central nervous system (rev. in Berg et al. 2003). These proteins constitute about 1% of the cytosolic proteome of neuronal cells and interact with numerous proteins, including apoptosis regulators, cytoskeletal proteins, signaling proteins, transcription factors, metabolic enzymes, and tumor suppressors.

14-3-3 protein is detected in the cerebrospinal fluid of Creutzfeldt–Jacob-diseased individuals and hence it is used as an ante-mortem marker as well (Hsich et al. 1996). Notably, the ζ isoform of this protein has been found in PrP amyloid deposits of sporadic CJD and variant CJD patients (Richard et al. 2003) whereas isoform ε is a specific component of PrP amyloid plaques found in GSS (Di Fede et al. 2007). These observations suggest that 14-3-3 protein binds to PrP^Sc and/or PrP^C. In fact, 14-3-3 protein has been demonstrated to interact directly with PrP (Satoh et al. 2005). PrP^C co-immunoprecipitated with 14-3-3 protein in human brain extract and was bound by purified human recombinant 14-3-3ζ protein in an overlay assay. Additionally, purified recombinant PrP and 14-3-3 protein interacted on an overlay blot, indicating that the binding was direct. The interaction site with 14-3-3 protein was roughly localized within the N-terminal part of the PrP molecule spanning residues 23–137. Furthermore, immunocytochemical analysis showed that PrP^C and 14-3-3 protein colocalized in mitochondria of cultured neuronal progenitor cells. In the same study, an interaction between 14-3-3 protein and Hsp60, and a colocalization similar to that with PrP^C, were demonstrated suggesting formation of a ternary complex. Recently, the interaction site with 14-3-3 protein has been narrowed to the PrP region 106–126 (Mei et al. 2009). The consequences of PrP binding to 14-3-3 protein remain to be determined.

Mahogunin

Mahogunin (Mgrn) is a cytosolic Really Interesting New Gene (RING) domain-containing protein of ubiquitin ligase activity involved in cell survival signaling (rev. in He et al. 2003a). Although substrates of this ligase are unknown, loss of its function leads to spongiform neurodegeneration of the brain and a change in coat color found in mahoganoid mutant mice (He et al. 2003b).

Very recently, Chakrabarti and Hegde (2009) have reported functional depletion of Mgrn in cultured cells overexpressing cytosolic form of PrP or transmembranal ^CtmPrP. By performing semipermeabilization of these cells the authors demonstrated co-aggregation of Mgrn with cytosolic PrP. Recombinant Mgrn pulled down PrP from brain lysate confirming the interaction. The octarepeats of PrP were found to be involved in the binding since deletion of this region abrogated the ability to co-aggregate Mgrn. Moreover, a synthetic peptide corresponding to the PrP octarepeat sequence could pull down Mgrn. Although PrP undergoes ubiquitination (Yedidia et al. 2001), it is not a substrate for Mgrn (He et al. 2003b). The interaction with cytosolic PrP leads to Mgrn sequestration and triggers pathological changes resembling the effects of Mgrn depletion. These changes may be reversed or prevented by competition for the PrP–Mgrn binding, overexpression of functional Mgrn or redistribution of cytosolic PrP. Hence, it has been proposed that the molecular mechanism of cytotoxicity of cytosolic PrP is closely related to the interaction with Mgrn resulting in a loss of its function.

Chaperones

The search for a molecular chaperone capable of influencing the pathological conversion of PrP^C into PrP^Sc attracts much attention. To date, interactions with numerous chaperones have been reported which seem to be related to normal cellular folding of PrP^C during its biosynthesis. For example, it was demonstrated that PrP bound directly to stress-inducible protein 1 (Martins et al. 1997), αβ-crystallin (Sun et al. 2005), Hsp60 (Edenhofer et al. 1996), and co-immunoprecipitated with protein disulfide isomerase, calnexin, calreticulin, BiP, grp94 (Capellari et al. 1999), Hsp40, and Hsp70 (Rambold et al. 2006). There is no evidence indicating that interaction with PrP may affect function of any of these chaperones.

PrP as Component of Multimolecular Complexes

The number of already identified PrP interactors implies that this protein may be a component of large multimolecular complexes. Some of these complexes seem to contain intracellular proteins. In a proteomic approach, Giorgi et al. (2009) identified proteins co-purifying with the proteinase K-resistant core of PrP^Sc isolated from scrapie-infected hamster brain. The major components of the PrP^Sc-enriched preparations, identified by mass spectrometry, were ferritin heavy chain, calcium/calmodulin-dependent protein kinase α type II (CaMKIIα), apolipoprotein E, and tubulin. Among these proteins only apolipoprotein E is an extracellular protein. Since the purification procedure employed PK treatment, the co-purified proteins seem to acquire protease resistance by co-aggregation with PrP^Sc, with a possible exception of ferritin which is known to be intrinsically PK-resistant (Atkinson et al. 1989). The possible association was confirmed by colocalization of PrP^Sc with CaMKIIα in the cytoplasm of hippocampal neurons and with apolipoprotein E around lateral ventricles and blood vessels of the brain. Interestingly, an increased expression of CaMKIIα has been observed in the brain of scrapie-infected animals (Jin et al. 1999; Cosseddu et al. 2007). Similarly, expression of ferritin, the intracellular protein responsible for sequestration and storage of iron (rev. in Koorts and Viljoen 2007), was elevated in the brain of TSE-infected mice (Kim et al. 2007). Association of PrP^Sc with the heavy and light chains of ferritin had been demonstrated earlier by co-purification and co-immunoprecipitation from CJD brain homogenates (Mishra et al. 2004). The complex could be dissociated with salt suggesting an ionic character of the interactions. Furthermore, PrP^Sc and ferritin were co-endocytosed and co-transported in vesicles across intestinal epithelial cells. It was proposed that ferritin may facilitate intestinal uptake of PrP^Sc from contaminated food. It remains, however, undetermined whether in the above study all co-purifying proteins interacted directly with PrP^Sc.

Complexes with proteins residing in PrP^C vicinity were covalently conserved in the brain by means of time-controlled transcardiac perfusion cross-linking (Schmitt-Ulms et al. 2004). In these complexes, numerous plasma-membrane and extracellular molecules were found: amyloid precursor protein (APP), amyloid beta precursor-like protein 2 (APLP2), neural cell adhesion molecules (N-CAM), opioid binding protein (OBCAM), neurotrimin, nectin-like 1 protein, calsyntenin 1 (CLSTN1), myelin-associated glycoprotein (MAG), clusterin precursor (CLU), and dipeptidyl aminopeptidase-like protein 6 (DPP6). Interestingly, in this experimental approach an intracellular protein—dynamin-1 (DNM1)—was also identified as a component of the PrP^C complexes. Notably, dynamin is an enzyme involved in membrane trafficking which interacts with microtubular cytoskeleton (Shpetner and Vallee 1989). Also in this study, it is unclear which of the above proteins were cross-linked directly with PrP^C.

Recently, proteomic profiling of PrP^C interactome was performed by means of microarray technology and again revealed the intracellular partners of this protein (Satoh et al. 2009). An array of approximately 5000 purified human proteins was probed with human PrP spanning residues 23–231. Native-like posttranslational modifications were ensured by expression of the target proteins in the baculovirus system and of PrP in human embryonic kidney cells (HEK293). In this study, 47 novel binding partners were identified, which again suggests formation of multimolecular complexes with PrP. Surprisingly, the majority of these interactors are known as proteins involved in the recognition of nucleic acids. The results of the microarray analysis were verified for selected proteins: homeobox A1 protein (HOXA1), polo-like kinase 3 (PLK3), and methylpurine-DNA glycosylase (MPG) in immunoprecipitation and colocalization experiments. HOXA1 is a transcription factor critical for development of the hindbrain (Carpenter et al. 1993), PLK3 is a Ser/Thr kinase involved in regulation of the entry into mitosis (rev. in Myer et al. 2005), whereas MPG is an enzyme repairing damaged DNA (rev. in Bouziane et al. 1998). All of these proteins underwent co-immunoprecipitation with PrP23-231 in HEK293 cells lysates. Interestingly, DsRed-tagged PrP23-231 transiently expressed in HEK293 cells was found predominantly in the nucleus and cytoplasm and less abundantly on the plasma membrane. In this cell culture, PrP coexisted with HOXA1 and MPG in the nucleus, whereas with PLK3 in the cytosol and on the cell surface.

Concluding Remarks

There is growing evidence implying that cytosolically mislocalized PrP may cause neurodegeneration. Experiments on transgenic mice and cell culture models of neurodegenerative conditions have revealed that PrP residing fully or partially (^CtmPrP) in the cytosol can be neurotoxic. Transgenic mice that generate ^CtmPrP above a threshold level develop a TSE-like disease (Hegde et al. 1999). Forced cytosolic expression of PrP in transgenic mice leads to severe degeneration of cerebellar granule neurons and gliosis (Ma et al. 2002). The mistargeted PrP is also toxic to forebrain neurons, which has been demonstrated for inducible transgenic mice expressing cytoPrP in this region of the brain (Wang et al. 2009). The consequences of cytoPrP accumulation seem to depend on the neuronal context in the brain. The neuropathological changes observed in the 1D4 mouse model expressing cytoPrP varied among the brain regions: cerebellum, hippocampus, and neocortex (Faas et al. 2009). Surprisingly, although cytosolic PrP exhibits a toxic potential, it has never been shown to be infectious. Conflicting results have been obtained for cell cultures expressing cytoPrP. In those studies the mislocalized PrP was cytotoxic (Ma et al. 2002; Rane et al. 2004; Grenier et al. 2006), had no effect on cell viability (Fioriti et al. 2005) or even protected neurons against Bax-mediated apoptosis (Roucou et al. 2003). A cell type-selective toxicity of cytoPrP may be an explanation for some of these discrepancies. Plausibly, cytoPrP may be cytotoxic only at particular conditions and/or concentrations.

How might the cytosolically mistargeted PrP be neurotoxic? One can assume that most of the physiological partners of PrP^C are extracellular or membrane proteins. It is also plausible that at its normal, low concentration in the cytoplasm PrP may modulate the physiological functions of the interacting proteins whereas at elevated, pathological levels it may dysregulate them leading to neurodegeneration. Notably, as demonstrated by Rambold et al. (2006), only PrP accumulated in the cytosol was neurotoxic. Aggregation of intracellular partners induced by PrP has been observed in numerous studies, e.g., for mahogunin, NRAGE, tubulin, and Bcl-2 (summarized in Table 1). This aggregation/co-aggregation may lead to a loss of function of the interactor and may represent a molecular mechanism of toxicity of the PrP mistargeted into the cytosol. Therefore, the interactions of PrP with some cytosolic proteins may be considered as potential targets for drugs (inhibitors of the interaction) in therapy of TSEs. It should be noted that some physiological interactions with intracellular proteins may be related to biogenesis/modifications of PrP (e.g., interactions with chaperones) and trafficking of PrP within the cell (e.g., endocytosis, microtubule-dependent transport). The consequences of the numerous interactions outlined in this review remain to be elucidated, however, one can suppose that most of them will turn out to be deleterious to the cell.

Abbreviations

Bcl-2

B-cell lymphoma 2 protein

BSE

Bovine spongiform encephalopathy

C1

N-terminally truncated form of PrP encompassing residues ~110/112–231

C2

N-terminally truncated form of PrP encompassing residues ~90/91–231

CaMKIIα

Calcium/calmodulin-dependent protein kinase α type II

CJD

Creutzfeldt–Jakob disease

CK2

Casein kinase 2

^CtmPrP

Transmembrane form of PrP with the C-terminus residing in the lumen of ER

CWD

Chronic wasting disease

cyPrP, cytPrP, cytoPrP

PrP residing entirely in the cytosol

ER

Endoplasmic reticulum

FFI

Fatal familial insomnia

GFAP

Glial fibrillary acidic protein

GPI

Glycosylphosphatidylinositol

Grb2

Growth factor receptor-bound protein 2

GSS

Gerstmann–Sträussler–Scheinker disease

HEK

Human embryonic kidney cells

hnRNP

Heterogeneous nuclear ribonucleoprotein

Hsp

Heat shock protein

MAPs

Microtubule-associated proteins

Mgrn

Mahogunin

NLS

Nuclear localization signal

NRAGE

Neurotrophin receptor-interacting MAGE homolog

Nrf2

Nuclear factor erythroid 2-related factor 2

^NtmPrP

Transmembrane form of PrP with the N-terminus residing in the lumen of ER

OR

Octapeptide repeats

PK

Proteinase K

Prnp

Prion protein gene

PrP

Prion protein

PrP^C

Cellular form of PrP

PrP^Sc

Scrapie form of PrP

rPrP

Recombinant PrP

SP

Signal peptide

TM

Transmembrane domain

TSE

Transmissible spongiform encephalopathy