The Comprehensive Native Interactome of a Fully Functional Tagged Prion Protein

The enumeration of the interaction partners of the cellular prion protein, PrPC, may help clarifying its elusive molecular function. Here we added a carboxy proximal myc epitope tag to PrPC. When expressed in transgenic mice, PrPmyc carried a GPI anchor, was targeted to lipid rafts, and was glycosylated similarly to PrPC. PrPmyc antagonized the toxicity of truncated PrP, restored prion infectibility of PrPC-deficient mice, and was physically incorporated into PrPSc aggregates, indicating that it possessed all functional characteristics of genuine PrPC. We then immunopurified myc epitope-containing protein complexes from PrPmyc transgenic mouse brains. Gentle differential elution with epitope-mimetic decapeptides, or a scrambled version thereof, yielded 96 specifically released proteins. Quantitative mass spectrometry with isotope-coded tags identified seven proteins which co-eluted equimolarly with PrPC and may represent component of a multiprotein complex. Selected PrPC interactors were validated using independent methods. Several of these proteins appear to exert functions in axomyelinic maintenance.


Introduction
The cellular prion protein, PrP C , is required for susceptibility to prion infections [1,2], for prion toxicity [3], and for prion transport within the body [4]. PrP C is a conserved glycoprotein that is anchored to the cell surface through a covalently attached glycosyl phosphatidyl inositol (GPI) residue [5]. PrP C undergoes a complex biogenesis encompassing co-translational secretion into the lumen of the endoplasmic reticulum, cleavage of an Nterminal signal peptide, addition of complex N-linked carbohydrate chains at two sites [6], addition of a preformed GPI anchor at its very C-terminus (Ser 230 ), and removal of a C-terminal oligopeptide.
Despite the detailed chemical knowledge described above, the molecular details of the process by which PrP C is converted into a disease-associated homologue, PrP Sc , are unclear [7]. Likewise, the chain of events emanating from prion infections and leading to neurodegenerative changes and clinical signs is unknown. Lastly, the physiological function of PrP C is unclear [8]. Most of the above processes may require interactions with proteins other than PrP, yet the nature of such interaction partners is largely unknown. The present study was initiated as an approach to discovering the functionally relevant interaction partners of PrP C .
Several diverse approaches have been used in the past to achieve the latter goals. In some instances, however, the techniques employed were not sufficiently sensitive or were fraught with other problems. Classical two-hybrid screens, in which fusion proteins leads to biological readouts in the cytosol of yeast, tend to produce when applied to membrane proteins like PrP c . The same holds true for cross-linking experiments, in which proteins resident in the same micro-environment may become linked together even if they do not functionally interact with each others.
In order to avoid the problems described above, and to minimize any interference with the conditions existing in vivo, we isolated native protein complexes containing PrP C and characterized them by mass spectrometry. The addition of epitope tags, for which high-affinity antibodies are available, has proven instrumental for the study of many supramolecular complexes. The engineering of appropriate tags into the proteins of choice yields ''molecular handles'' through which multi-component complexes can be immunoprecipitated and highly purified. PrP C lends itself to this approach as a particularly attractive bait, as its highresolution structure is known [9] and thereby allows for the rational design of tags. If the precipitating antibodies are directed against linear, non-conformational epitopes within the tag, epitope-mimetic peptides can release the protein complexes in a highly specific way under non-denaturing conditions. The introduction of a tag is also a promising starting point for identifying functionally relevant complexes since it preserves protein interactions that occur in the same region of an anti-PrP antibody.
GFP-PrP C fusion proteins have proved useful for determining the subcellular distribution and trafficking of normal and mutated prion protein [10,11,12]. However, the suitability of GFP to the proteomic approach delineated above is limited. GFP is a bulky, highly structured and rigid tag whose molecular weight exceeds that of PrP C . Therefore we reasoned that GFP may distort the composition of any native multiprotein complex that encompasses PrP C .
In the present study, we have tagged the C-terminus of mouse PrP C with the human ''myc-tag''. The resulting chimaeric protein, termed PrP myc , was used to immunoprecipitate and characterize the supramolecular complex containing the prion protein from transgenic mice. Using immunoprecipitation and mass spectrometry, we have identified a set of proteins associated with PrP myc . Since the conversion of cellular prion protein PrP C into the proteinase K-resistant isoform PrP Sc is the central pathogenic process in prion diseases, we investigated whether PrP myc can be converted into PrP Sc . Our results indicate that C-terminally myctagged prions can contribute to prion infectivity and to neurotoxicity. Therefore, myc tagged PrP Sc may also allow for identification of proteins interacting with PrP Sc .

Transgenic mice expressing C-terminally tagged PrP
We tagged the murine prion protein by introducing a human myc epitope tag (EQKLISEEDL) at its C terminus next to Ser 230 and amino proximally to the C-terminal signal sequence for the GPI anchor (Fig. 1A). As the minimal myc epitope tag consists of only 10 amino acids, we reasoned that it might not interfere with the geometry and proper folding of PrP C , and with its function. The human myc epitope tag was detectable by both monoclonal anti-myc antibodies 9E10 and 4A6 [13]. To guarantee correct GPI linkage of this fusion protein, the sequence comprising Ser 230 and its four immediately preceding N-proximal amino acids was duplicated after the tag. The resulting fusion molecule was termed PrP myc .
Preliminary analyses of PrP myc transfected cells indicate that the biosynthesis, processing, and trafficking of the resulting fusion protein were indistinguishable from those of endogenous PrP C (data not shown).
To generate transgenic mice expressing C-terminally tagged PrP, PrP myc was ligated into the 'half-genomic' phgPrP backbone, driven by the endogenous Prnp promoter [14]. Pronuclear injections of linearized purified DNA were performed into fertilized oocytes derived from a B6D2F16B6;129S5-Prnp o/o mating. Four founder mice were identified by PCR analysis using primers TAP 20 (59-CCG ATG TGA AGA TGA TGG AGC) and myc 22 (59-CCG TCG ATC GGA TTC AGA TCC) specific for the myc-tag amplicon. The two highest-expressing lines, termed Tg(PrP myc )940Zbz and Tg(PrP myc )941Zbz (henceforth Tg940 and Tg941 for brevity) were chosen for further propagation.
Southern blot analysis revealed that Tg940 and Tg941 mice harbored 6 copies and 1 copy of the transgene per haploid genome, respectively (Fig. 1B). Northern blot analysis performed on total RNA from brains of PrP myc mice confirmed transcription of transgenic PrP myc (Fig. 1C). Transgenic mice expressing PrP myc did not show any anatomical or behavioral abnormalities, survived in health for .700 days, and did not show any neurohistological changes. We monitored weight and food uptake until adolescence. Transgenic mice had shiny fur indicative of good general health, and reproduced with frequency and litter sizes comparable to wild-type mice (data not shown). We did not recognize any difference in locomotor activity from wild-type mice over a period of .2 years.
To obtain transgenic strains that only expressed PrP myc yet no endogenous PrP, both transgenic founders Tg940 and Tg941 were crossed twice to Prnp o/o mice. Transgene expression in brain and spleen of these mice was analyzed by Western blotting using anti-PrP antibody POM1 [15], and mouse monoclonal anti-myc antibody 9E10. Tg940 mice lacking PrP C (henceforth termed Tg940 PrP o=o myc ) expressed 1.6 fold more of PrP myc protein in brain than wild-type mice (Fig. 1D), but had lower expression levels of the transgene in spleen (about 0.5 fold of Prnp +/o mice, data not shown). Expression of PrP myc in Tg941 PrP o=o myc was approximately 0.33 fold in brain and 2-fold in spleen of PrP C expression in Prnp +/o mice (data not shown). Tg940 and Tg941 exhibited a three-banded pattern very similar to PrP C glycoforms  in wild type mice (Fig. 1E).

PrP myc is localized within detergent resistant membranes (DRMs)
We isolated DRMs from Tg940 brain tissue by gradient centrifugation [16]. A series of fifteen individual fractions was carefully removed from the tubes after centrifugation of typical DRM preparations from mouse cerebella of Tg940 PrP o=o myc , and analyzed by Western blotting. The quality of the preparations was monitored using the control proteins flotillin 2 is known to reside in DRMs [17,18]. PrP myc was found to reside in the same fractions as these proteins, confirming its localization in these specialized membrane domains (Fig. 1F). Therefore, the subcellular localization of PrP myc was similar to that of endogenous PrP C .

Testing the functionality of PrP myc
Tg940 PrP o=o myc were crossed with the TgF35 line of mice expressing N-proximally truncated PrP, henceforth referred to as PrP DF . PrP DF mice suffer from degeneration of the cerebellar granular layer, leukoencephalopathy, and death at about 100 days of age [19,20,21]. This phenotype is dose-dependently counteracted by endogenous or transgenic co-expression of wild-type PrP C , presumably because of a competing activity supplied by PrP C .
If the tagged protein PrP myc is functional and appropriately localized, it should also rescue PrP DF mice from neurodegeneration. Indeed, Tg940 PrP o=o myc expressing PrP DF survived for 551673 days (n = 5; Fig. 1G) and maintained a normal weight throughout their lifetime. Mice were examined twice per week for neurological symptoms and scored as described [19], yet did not show clinical signs of CNS disease at any time. Furthermore, they did not develop histopathological changes in brain or other organs (data not shown), suggesting that PrP myc is functional in vivo. Age and sex-matched PrP DF siblings died between 12 and 14 weeks of age (mean survival: 9567 days, n = 5; Fig. 1G).
In contrast, double-transgenic mice of the lower expressing line (Tg941) were not completely rescued and began to show first signs of illness around day 280. Some animals had to be sacrificed at the age of 12 months due to hind leg paresis (mean survival 391657 days, n = 9; Fig. 1G). As Tg941 PrP o=o myc mice express about one-third of the PrP myc found in brains of Tg940 PrP o=o myc mice, this indicates that the action of PrP myc , like that of PrP C , is dose-dependent.
Neuropathology in inoculated PrP z=o myc mice To assess whether PrP myc can be converted into myc-tagged protease-resistant PrP Sc myc , PrP z=o myc and PrP o=o myc mice from lines Tg940 and Tg941 were inoculated with mouse-adapted sheep prions (RML strain, passage 5). After low dose intraperitoneal (ip) inoculation with 10 3 IU or intracerebral (ic) inoculation with 300 IU of RML5 brain homogenate, Tg940 PrP z=o myc mice showed signs of CNS dysfunction at 250692 (n = 5/5) and 236676 (n = 6/6) days post inoculation (dpi), respectively ( Fig. 2A and B). Mice expressing less PrP myc in brain (Tg941) developed signs of CNS dysfunction and terminal scrapie disease more slowly, at 316620 (n = 4/4) days after low-dose intracerebral inoculation ( Fig. 2B and Table S1).
Brain homogenates prepared from terminally sick Tg940 PrP z=o myc mice were inoculated ic into tga20 mice overexpressing PrP C [14] to test for infectivity in an in-vivo mouse assay. All of the tga20 mice developed neurological signs of terminal scrapie at around 80 dpi (Table S1). Prion infection was confirmed by immunochemical and histopathological analysis in all terminally sick mice. PrP z=o myc mice developed neurological dysfunction and terminal disease significantly earlier than Prnp +/o mice: the mean incubation time was 27669 days for Prnp +/o (n = 6) and 226613 days for Tg940 PrP z=o myc mice (n = 8) after high dose ic inoculation ( Fig. 2C and Table S1). Therefore, PrP myc contributes to, rather than interfering with, prion pathogenesis in Prnp +/o mice.
In all terminally sick PrP z=o myc mice tested we detected proteinase K (PK) resistant material in brain and spleen after ic or ip inoculation with RML prions. To distinguish between wild-type PrP Sc and PrP Sc myc we stained Western blots of brain homogenates with an anti-myc antibody (Fig. 2D). PK-resistant PrP Sc myc was clearly detectable under these conditions, indicating that PrP myc itself is convertible, and suggesting that this phenomenon contributed to the shortened incubation periods in PrP z=o myc mice. Comparison of immunohistochemically stained brain sections of terminal Prnp +/o and Tg940 PrP z=o myc mice did not reveal any striking differences in the extent and topography of reactive astrocytic gliosis, vacuolar degeneration and PrP aggregates ( Neuropathology in inoculated PrP o=o myc mice To investigate whether PrP myc can be converted into myctagged PK-resistant PrP Sc myc even in the absence of a wild-type PrP allele, we inoculated PrP o=o myc mice with RML prions. No PrP Sc was detected in brain and spleen at 50 to 100 days after ic or ip inoculation, yet 8 of 34 (23%) PrP o=o myc mice eventually developed a progressive neurological syndrome clinically indistinguishable from scrapie after RML inoculation (Table S2). Brain homogenate from these sick mice was then used to inoculate a second generation of Tg940 PrP o=o myc mice. Western blot analysis of brain homogenate from these second-passage ic-inoculated Tg940 PrP o=o myc mice revealed PK-resistant PrP; these mice had clinical signs of scrapie and developed vacuolation in the neuropil, intense astrogliosis, and abundant PrP aggregates ( Fig. 3A-C). For control, Tg940 PrP o=o myc mice were inoculated with non-infectious brain homogenate. These mice showed no evidence of vacuolar degeneration or nerve cell loss, and only mild astrogliosis when aged ( Fig. 3D-F).
As an additional method to distinguish between PrP Sc derived from wild-type PrP and PrP myc we performed histoblot analysis of cryosections of terminal Tg940 PrP o=o myc mice and Tg940 PrP z=o myc mice ( Fig. 3G-I). Using anti-PrP (POM1) and anti-myc (4A6) antibodies, we could specifically detect PK-resistant PrP in terminal C57BL/6 mice, Tg940 PrP o=o myc and Tg940 PrP z=o myc mice. This technique allowed us to map the distribution of PrP Sc in different transgenic mice.
We then investigated whether PrP myc infectivity would increase upon serial transmission, as frequently observed in strain adaptation [22]. Brain homogenate derived from RML-inoculated Tg940 PrP z=o myc mice was passaged into Tg940 PrP o=o myc mice which all got sick after 590656 days (n = 3) (Table S3). One of these second-passage mice was used as a source for a third passage into 5 Tg940 PrP o=o myc mice. All of them show similar neurological signs as in the second passage, but with a shorter incubation period of 367638 (n = 5), which is suggestive of strain adaptation (Table S3).
We then tested whether deposition of PrP Sc accompanies prion replication, defined as increase in prion infectivity. Samples from Tg940 PrP o=o myc mice after the second passage were used to infect the PK1 subclone of N2a neuroblastoma cells in the Scrapie cell assay in endpoint format (SCEPA [23]). As shown in the Fig. 3 J the titer for the PrP Sc myc is the same as the standard RML.

Identification of PrP myc -containing protein complexes
Crude brain homogenates from Tg940 PrP o=o myc mice were subjected to immunoprecipitation (IP) experiments with paramagnetic microbeads coupled to mouse monoclonal anti-myc antibody (4A6, Upstate, USA). Release of myc-containing protein complexes from beads was carried out by exposing the beads to an excess of the synthetic epitope-mimicking myc peptide described above. Control experiments were carried out to verify the specificity of the eluted proteins, and included (1) incubation of beads with 129S2/SvPas wild-type brains followed by elution with the myc peptide, as well as (2) incubation of beads with Tg940 PrP o=o myc homogenate followed by elution with a scrambled version of the myc peptide. In the eluates from 4A6-coupled beads secretory signal peptide, cleaved after sorting of the precursor to endoplasmic reticulum; repeats: five repeats of eight amino acids; CC: charge cluster; HC: hydrophobic core; H1, H2, H3: a-helices of the globular carboxy-proximal domain; MYC: human myc epitope tag (EQKLISEEDL); MA: membrane anchor of precursor protein, replaced during maturation with glycosyl phosphatidyl inositol anchor. (B) Southern blot analysis of lines Tg940 PrP myc (lanes 1, 2, 6) and Tg941 PrP myc (lanes 3, 5, 7). Lane 4: Tg941 PrP o=o myc mouse co-expressing N-proximally truncated PrP DF . Lane 8: PrP DF mouse. The bands diagnostic for PrP myc and PrP DF were 3039 and 2709 bp, respectively. Numbers of transgenic copies per haploid genome, as determined by quantitation of Southern blot signals against the respective Prnp o genomic band, revealed higher copy numbers in Tg940 PrP myc (#6) than in Tg941 PrP myc mice (#1). (C) Northern blot analysis of individual Tg940 PrP myc and Tg941 PrP myc brains using a Prnp probe. Mice homozygous for the transgenic allele PrP myc (lanes 2, 3, 4 from Tg940 and lanes 8, 11 from Tg941) showed higher levels of PrP myc mRNA than hemizygous mice (lanes 1 and 5 from Tg940 and lanes 6, 9, 10 from Tg941). An actin probe was used as a loading control (lower panel). (D) Similar expression levels of transgenic protein from Tg940 PrP o=o myc , and full-length PrP from 129S2/SvPas wild-type mice, analyzed by Western blotting of total brain homogenate using anti-PrP antibody POM1. (E) Similar glycosylation pattern of full-length PrP from 129S2/SvPas wild-type and PrP myc from Tg940 PrP o=o myc mice. Brain homogenates were subjected to PNGase F treatment as indicated, and analyzed by Western blotting using POM1 antibody to PrP C . (F) Detergent-resistant membrane preparations from cerebella of Tg940 PrP myc transgenic mice showed PrP myc in lipid rafts. PrP myc was detectable by Western blotting in fractions with 5-30% Optiprep. PrP myc resided in the same fractions as flotillin (48 kDa) confirming its localization in DRMs. (G) A genetic in vivo assay for the function of the PrP myc protein. Survival curves of mice expressing PrP DF in absence of full length PrP C and in presence of PrP myc from two transgenic lines. Toxicity of PrP DF was counteracted by PrP myc , leading to a longer survival and suggesting that PrP myc has retained at least some of the function of PrP C . Line PrP DF , Tg940 and Tg941 consisted of 5, 5, and 9 individuals, respectively. doi:10.1371/journal.pone.0004446.g001 in presence of a wild-type PrP allele. Western blot analysis using brain homogenate from an inoculated, terminally sick PrP z=o myc mouse. Antibodies incubated with 129S2/SvPas wild-type brain homogenates, PrP C was not detected, whereas only traces of PrP C were detected in the scrambled-peptide eluate from IPs of Tg940 PrP o=o myc brain homogenates (Fig. 4A).
Inspection of silver-stained gels revealed more protein bands in the specific than in the unspecific elution fraction (Fig. 4B), in particular the PrP myc band exclusively present in the myc-specific eluates from Tg940 PrP o=o myc brain homogenates. The corresponding lanes were cut into slices, proteins were extracted, and tryptic peptides were identified by liquid chromatography followed by tandem mass spectrometry (LC-MS/MS). As a further quality control, we verified that the identified proteins originated from the gel area corresponding to their predicted molecular size. Table S4 lists those proteins that were coprecipitated with PrP myc from transgenic brains, yet were not detected in material immunoprecipitated from wild-type brains and unspecific elution under the same conditions. While 442 individual proteins were detected in both the specific and the nonspecific eluates, and 277 proteins were uniquely present in the nonspecific eluate, 96 proteins were present in the specific eluate but absent from the nonspecific eluate.
We then sought to determine the relative abundance of PrP and the interacting proteins in the specific and unspecific peptide elution fractions by using cleavable isotope-coded affinity tags (cICAT) as a quantitative mass spectrometric technique. In the classical cICAT approach the two labeled fractions contain the same amount of protein. Since this is not the case for the specific and unspecific IP elution fractions, we could only determine the relative ratio of PrP between the specific and the unspecific elution fractions.
The two elution fractions derived from immunoprecipitations of PrP myc and wild-type brains were labeled with the ''heavy'' (cICAT-13C9) and ''light'' (cICAT-12C9) cICAT tags, mixed, and mass/charge (m/z) elution profiles were determined by mass spectrometry. Sequest [24], PeptideProphet [25] and XPRESS were used to identify the proteins and to access the cICAT ratios (Fig. 4C, Table S5). Of the 157 peptide pairs that could be assigned to a heavy/light ratio between 0.1 and 100, seven proteins were found to have a comparable ratio to PrP and, at the same time, were identified as specific proteins by the gel-based approach (Table 1). Any ratios below 1 are indicative of proteins more abundant in the scrambled elution than in the myc-specific elution. Proteins displaying a similar abundance in both samples would yield a ratio of 1, which most probably indicates nonspecific binding to and elution from the beads. The ratio for PrP was about 14, and the proteins listed in Table 1 represent values between 4 and 15.
We then sought to confirm the results of mass spectrometric analyses by immunochemical analyses of selected proteins. Indeed, the identity of PrP, 29,39-cyclic nucleotide 39-phophodiesterase, M6-a and Neurofascin was unambiguously confirmed by Western blot analysis. Fig. 4D shows the characteristic double band of CNPase after myc-peptide elution and a low-intensity band for the scrambled-peptide elution. Western blot analysis with antibodies to Neurofascin 155 and M6a revealed specific bands for the specific-peptide elution but in none of the negative controls ( Fig. 4E-F). The signal for M6a from the specific elution shows two strong bands most probably originating from alternative splicing. For both Neurofascin and M6a, the protein expression level in wt and Tg940 PrP o=o myc brain were approximately the same as illustrated in Fig. 4E-F.

Discussion
Our understanding of the function of PrP C and its conversion into PrP Sc continues to be sketchy. Genetic experiments have helped defining the domains of PrP C necessary for prion propagation [21] and, with some limitations, for PrP C function [19,26,27,28], yet have failed to identify any further proteins that may be required for this process. However, progress in this field may crucially benefit from enumerating and/or manipulating the PrP-interacting proteome. Towards the latter goals, we have studied the biogenesis, localization in vitro and in vivo of a Cterminally myc-tagged version of PrP C (PrP myc ). Since the physiological function of PrP C is unknown, we used a wellestablished approach of reverse genetics [14] to assay the biological activity of PrP myc . This approach is so far the most proximal surrogate to study the function of PrP. We found PrP myc to be fully functional and substitute dosage-dependently for endogenous PrP in rescuing the neurodegenerative phenotype induced by PrP DF .
Conversion of cellular prion protein PrP C into the diseasecausing isoform PrP Sc is the central pathogenic process in prion diseases [29]. Therefore, any claim of the biological authenticity of a modified PrP protein should be substantiated by its ability to sustain prion replication. We approached this important question in a variety of paradigms. Whereas direct intracerebral inoculation of PrP o=o myc transgenic mice with prions rarely induced scrapie, we found that in the presence of a wild-type Prnp allele PrP myc is converted into a PK-resistant isoform (PrP Sc myc ). The disease of prion-infected PrP z=o myc mice was transmissible by ic inoculation of brain homogenates to wild-type mice and also, importantly, to PrP o=o myc mice. Since it is known, that PrP Sc levels do not necessarily correlate with infectivity titers, we decide to evaluate the infectivity titers by SCEPA and compare to RML, and also in that paradigm PrPmyc behave as normal RML. The latter finding establishes beyond any doubt that PrP myc supports prion replication and scrapie pathogenesis.
In many paradigms, expression of heterologous PrP molecules which differ from the endogenous PrP by as little as one amino acid can profoundly interfere with the overall accumulation of PrP Sc [30,31], suggesting that precise homotypic interactions between PrP molecules are important for PrP Sc accumulation [31,32]. However, when inoculated with the same dose of prions, PrP z=o myc mice developed disease faster than Prnp +/o mice, implying that PrP myc cooperates, rather than interfering, with PrP C in disease pathogenesis. This was unexpected in view of the many instances of interference that have documented to occur even between naturally occurring PrP alleles [12]. If one accepts that interference is brought about by disturbances of the replicative interface of prions, one might speculate that the carboxy terminus of PrP C does not participate to such an interface. The latter conclusion, however, is tempered by another observation. When PrP o=o myc mice were inoculated with RML prions, only few animals developed clinical signs of scrapie. This suggests that the C-terminally modified prion protein presents a ''prion transmission barrier'' to mouse-adapted sheep prions, analogously to the species barriers seen in many natural and experimental prion diseases [33]. The similarities between the amino acid sequence of donor PrP Sc and recipient PrP C play a crucial role in the species barrier [34,35], but the structural understanding of these constraints is still very sketchy. In the PrP myc transgenic model, the species barrier exists if wild-type prions are transmitted into PrP o=o myc animals, but can be overcome if brain homogenates from terminally sick PrP z=o myc mice containing PrP Sc myc is passaged into PrP o=o myc transgenic mice. The successful production of myc-tagged, self-propagating and disease-causing prions paves the way to many studies in vitro and in vivo by taking advantage of the high-affinity reagents available to the myc epitope. For example, the myc-tagged prion inoculum may allow for investigating the fate of inoculated prions in vivo, since PrP myc can be detected and traced by tag-specific antibodies which do not recognize endogenous PrP. In the present study, we provide evidence that PrP myc is useful for probing the PrP Cassociated proteome. We have established a novel method for the specific elution of multiprotein complexes containing PrP myc . We have exploited this method for identifying several candidate proteins which appear to interact with PrP C in vivo. The specificity of these interactions was validated by comparison to wild-type brain eluates and elution with a scrambled peptide. Some of the PrP-interacting proteins describe before and summarized in recent reviews [36,37], including for instance Tubulin, Hsp60 and Laminin, were detected in the specific as well as unspecific elution fraction of our approach and therefore not included into the list of possible candidates.
We utilized a quantitative MS technique, isotope-coded affinity tagging (ICAT), to determine the relative abundance of PrP and other proteins in the various samples, so to identify proteins that might exist in an equimolar complex with PrP C . Such PrP mycinteracting proteins would display an ICAT ratio of specific/ unspecific signals similar to that of PrP C . Based on this mass spectrometric approach, we found a small number of protein candidates equimolarly associated with PrP myc in native brain homogenates.
There are some caveats to the equimolarity filter described above. Supramolecular complexes encompassing PrP C may contain superstoichiometric amounts of accompanying molecules, in which case the ICAT ratios may be skewed. Conversely, if PrP C exists in a free form as well as in a complex, or in several different complexes, the partner proteins may appear to be substoichiometric in an immunoprecipitate. Therefore, even if the seven proteins identified here represent promising candidates, the remaining hits detailed in Table S4 should not be dismissed because of their non-equimolar ICAT ratios.
Two of the latter seven proteins (Q80U89 clathrin linked; Q01853 translational ER ATPase) are not well-characterized and no antibodies to them appear to be available. Chondroitin sulfate proteoglycan core protein was described to strongly inhibit neurite outgrowth of central and peripheral neurons [38]. It was also reported that neurite outgrowth is modulated -at least in culture models -by interactions between PrP C , NCAM and STI-1, which can lead to activation of intracellular signalling pathway [39].
Several PrP myc interactors belong to the families of neuronal glycoproteins and myelin-associated proteins. These include the neuronal membrane glycoprotein M6-a, Neurofascin, and 29,39cyclic nucleotide 39-phophodiesterase (CNP). P0 glycoprotein of compact PNS myelin, myelin-associated glycoprotein (MAG), and others have well-defined roles in the formation, maintenance and degeneration of myelin sheaths [40]. Myelin proteins also appear to mediate signals between the myelin-forming cell and the axon [41]. Current research suggests that CNP is required for maintenance of axon-glial interactions at the nodes of Ranvier in the CNS [42]. The interaction between PrP and CNP may underlie the myelin damage observed in old Prnp o/o mice [43] and in various transgenic PrP deletion mutants age [19,20,21]. In support of this hypothesis, recent studies suggest that myelin integrity may be maintained by a constitutively active neurotrophic protein complex involving PrP C [19].
A possible functional relation between neurofascin and PrP C is particularly intriguing in view of the lethal phenotype of transgenic mice expressing PrP deletion mutants, which display extensive central and peripheral myelin degeneration [19]. Neurofascin 186 (NF186) is expressed prenatally on dorsal root ganglia neurons and it may modulate their adhesive interactions with Schwann cells, which express NF155 postnatally and require it for development of axon-glial paranodal junctions. The major isoform of NF186 inhibits cell adhesion, and this activity may be important in formation of the node of Ranvier [44].
Another enticing candidate for functionally relevant interactions is M6-a, a membrane glycoprotein involved in neuronal differentiation as part of a Ca 2+ channel [45]. The lack of the cellular prion protein was shown to affect Ca 2+ homeostasis in neurons [46], and therefore it is thinkable that PrP C and M6-a are involved in a complex possessing an ion channel-like function.
In addition to identifying the interactors described above, the tools introduced here may allow for studying supramolecular complexes containing the disease-associated prion protein PrP Sc . The biophysical properties and aggregational state of PrP Sc are vastly different from those of PrP C , and there is reason to hypothesize that the PrP Sc interactome will only partially overlap with that of PrP C . Since most prion strains are both neurotropic and lymphotropic [47,48], and inflammatory conditions specify Brain homogenates were incubated with POM1 and 4A6 before or after PK treatment, and showed no PK-resistant PrP. (H) Prion-inoculated Tg940 PrP o=o myc and Tg940 PrP z=o myc mice, treated with PK and incubated with POM1 and 4A6, showed PK-resistant material in brain. (I) Prion-inoculated Prnp +/o , Tg940 PrP z=o myc and Tg940 PrP o=o myc mice treated with PK and untreated were stained with 4A6 anti-myc antibody and show protease-resistant PrP myc in the brain. A terminally sick Prnp +/o mouse was used to control for nonspecific 4A6 signals. (J) SCEPA of brain homogenates of PrP o=o myc and wild-type mouse. Three independent biological replicas of PrP Sc myc and 2 independent biological replicas for RML were analyzed in tenfold dilution steps using 6-12 PK1-containing replica wells for each dilution. Data points indicate the number of infectious tissue culture units per ml of brain homogenates. doi:10.1371/journal.pone.0004446.g003 the tropism of prions [49,50], the interactome of PrP C and PrP Sc in lymphoid organs will also be of interest. The inoculation of wildtype animals with myc-tagged prions may help elucidating the initial events that occur during infection of an animal with prions. Finally, the successful conversion of PrP myc into a protease-resistant moiety may allow for the purification of native PrP Sc -containing complexes using the techniques described above for PrP C . The latter studies may lead to the identification of the elusive chaperones involved in prion propagation, strain barriers and strain adaptation, as well as the crossing of prion species barriers.

Generation and characterization of transgenic mice
The phgPrP-myc plasmid, driven by the endogenous Prnp promoter in the context of the PrP ''half-genomic'' construct (phgPrP) [14], was digested with NotI and SalI to remove its prokaryotic backbone. Pronuclear injections were performed into fertilized oocytes derived from a B6D2F16B6; For Northern blot analyses, RNA was extracted using Trizol (Invitrogen). A randomly 32 P-labeled (Rediprime II Random Prime Labelling System, Amersham Biosciences) restriction fragment encompassing all of exons 1 and 2, all of the ORF and a part of exon 3 (XbaI-fragment) was used as a PrP probe. This probe hybridizes with all wild-type and tagged PrP mRNAs as well as the ''readthrough'' RNA from the disrupted Prnp locus [51].
Southern blot analyses were performed using a 640 bp DNA probe synthesized by incorporation of digoxigenin-11-dUTP (Roche, Switzerland) during PCR using PrP-specific primers and hybridization was performed following established protocols [52]. For the actin control the Northern blot was probed with an inhouse generated mouse beta-actin probe cloned from full-length cDNA.

Rescue of Shmerling's disease
PrP o=o myc mice were crossed with PrP DF [19,21] mice to obtain double transgenic animals with Prnp o/o genotype needed for the experiment described in Fig. 1. Animals were examined twice each week for symptoms of cerebellar dysfunction, including ataxia [53], tremor, weight loss, rough hair coat, and kyphosis. Scoring of neurological signs was performed according to a four-degree clinical score system [19] and mice were euthanized within 3 days of reaching a score of 3.5.

Preparation of DRMs
Brain homogenates were extracted for 1 hour on ice in 1% Triton X-100/25 mM MES/5 mM DTT/2 mM EDTA at pH 7.0 [16] and protease inhibitors. Extracts (500 mg protein/ ml buffer) were mixed with 60% Optiprep TM (Nycomed, Denmark) to reach a final concentration of 40% and overlaid in a SW40 centrifugation tube (Beckman, CA, USA) with a step gradient of 30 and 5% Optiprep TM in MES-buffer. After centrifugation at 35'000 rpm (12 hrs), 9 fractions were collected starting from the top. The raft fraction was obtained from the interphase 5-30% Optiprep TM . Mouse monoclonal anti-PrP antibodies (POM-1) and mouse monoclonal anti-flotillin 2 (BD Transduction, USA) were used to characterize the Optiprep TM fractions by Western blot.

Histopathology and Immunohistochemistry
Organs were fixed in 4% formaldehyde in PBS (pH 7.5) and paraffin-embedded. Two mm brain sections were stained with hematoxylin-eosin (HE). Immunohistochemistry was performed for glial fibrillary acidic protein (activated astrocytes) using a GFAP monoclonal antibody (DAKO, Carpinteria, CA, USA). PrP Sc aggregates were detected on paraffin sections using monoclonal antibody SAF-84. For histological analyses anatomic brain regions were selected according to standard strain-typing protocols (Bruce, 1991, Fraser, 1968. Spongiosis was evaluated on a scale of 0-5 (not detectable, mild, moderate, severe, and status spongiosus). Gliosis and PrP immunoreactivity were scored on a four-degree scale (undetectable, mild, moderate, severe). Histological analyses were performed by investigators blinded to animal identification.

Scrapie cell assay in endpoint format (SCEPA)
Prion-susceptible neuroblastoma cells (subclone N2aPK1) were exposed to 300 ml brain homogenates in 96-well plates for 3 d. Cells were subsequently split three times 1:3 every 2 days, and three times 1:10 every 3 days. After they reached confluence, we filtered 25,000 cells from each well onto the membrane of an ELISPOT plate, treated them with PK (0.5 mg/ml for 90 min at 37uC), denatured, and detected individual infected (PrP Sc -positive) cells by immunocytochemistry using alkaline phosphatase-conjugated POM1 mouse anti-PrP and an alkaline phosphataseconjugated substrate kit (Bio-Rad). We performed serial tenfold dilutions in cell culture medium containing healthy mouse brain homogenate. Scrapie-susceptible PK1 cells were then exposed to dilutions of experimental samples ranging from 10 24 to 10 29 , the same for RML, or to a 10 24 dilution of healthy mouse brain homogenate. Samples were quantified in endpoint format, by counting positive wells according to established methods.

Immunoprecipitations
Brains were homogenized in 0.5% CHAPS and protease inhibitors (Complete; Roche, Switzerland) as described above. Mouse monoclonal anti-myc 4A6 antibody was cross linked to Dynabeads M-280 Sheep anti-Mouse IgG (Dynal, Norway) as recommended by the manufacturer. Four mg of total protein from 5% brain homogenates were diluted to a volume of 1.5 ml of 0.5% CHAPS/NP-40. To precipitate the PrP myc complex, 40 ml of resuspended beads were added and incubated with rotational mixing for 2 hours at 4uC and for 15 min at room temperature. Beads were washed twice in PBS/0.5%CHAPS/NP-40 and twice in PBS/1% CHAPS/NP-40 at 4uC. To elute the complex, beads were incubated for 2 h at 4uC and another 10 min at room temperature with the synthetic specific peptide (c-myc: H-EQKLISEEDL-NH 2 , Roche Diagnostics, Basel, Switzerland) and the scrambled nonspecific peptide (cym: H-IELQKELDES-NH 2 , jct, Berlin, Germany) respectively. Peptides were added in 10-fold molar excess compared to the 4A6 antibody, in a final volume of 380 ml of 1% CHAPS, 1% NP-40.

Tryptic in-gel digestion
Silver stained bands from 12% SDS PAGE were destained and incubated for 1-3 h in 100 mM ammonium bicarbonate (NH 4 HCO 3 , pH 8.0, Sigma) in 50% MeOH at 37uC. The proteins were reduced in 2 mM tris(carboxyethyl)phosphine (TCEPNHCl, Pierce, USA) in 100 mM ammonium bicarbonate at 37uC for 40 min and alkylated with 20 mM iodoacetamide (Fluka, Switzerland) for 30 min at room temperature in the dark. Gel pieces were rinsed twice in 100 mM ammonium bicarbonate, dehydrated in acetonitrile for 10 min, dried under vacuum for 10 min and reswell in 200-400 ng of sequence-grade modified trypsin solution (Promega, Madison, WI, USA) for 15 min at RT. Gel pieces were covered with sufficient amount of 100 mM ammonium bicarbonate buffer containing 2 mM CaCl 2 and incubated overnight at 37uC. Samples were sonicated for 5 min and supernatant was pooled with an additional peptide extraction round with 50% acetonitrile/1% formic acid for 20 min at RT. Samples were dried under vacuum and kept at 220uC whenever they were not used immediately.

ICAT labeling and sample processing
The IP eluate was precipitated by ethanol precipitation and the pellet was dissolved in 100 ml of cICAT labeling buffer (50 mM Tris, pH 8.3; 8 M Urea; 5 mM EDTA; 0.125% SDS and 0.05% RapiGest). The cICAT labeling procedures was performed as described previously [54,55,56]. The control sample was labeled with the light, the specific elution sample with heavy cICAT label (Applied Biosystems, Foster City, CA, USA). Digestion with trypsin (Promega, Madison, WI, USA) was performed at 37uC over night and ICAT-labeled peptides were subsequently purified according to the manufacturer's instructions. ZipTip columns (C18, Millipore, Bedford, USA) were then used for further cleanup of the affinity-purified fraction.

Capillary chromatography and mass spectrometric analysis
Cleaned samples were resuspended in equilibration buffer (3% acetonitrile/0.1 formic acid in MilliQ-water) and loaded onto a microcapillary column constructed by slurry packing 8 cm of reversed-phase (RP) material (Magic C18, 5 mm, 200 Å , Michrom BioResources, Auburn, CA, USA) into a 75 mm fused-silica capillary (BGB Analytik AG, Böckten, Switzerland). Mass spectrometric analyses were performed on an LTQ-FT TM (Thermo Scientific, Bremen, Germany) systems directly coupled to a nanoLC TM HPLC system (eksigent, Dublin, CA, USA) at a flow rate of 200 nl/min. Peptides were eluted with an acetonitrile gradient from 3 to 45% in approximately 55 min and datadependent acquisition of tandem mass spectra was continuously repeated during the course of the analysis. Each high accuracy MS full scan was followed by four MS/MS scans of the four most intense peaks. High mass accuracy data was search with Mascot Integra (Matrix Science, UK) using the UniProt mouse protein data base (ftp.ebi.ac.uk/pub/databases/SPproteomes/fasta/ proteomes/59.M_musculus.fasta.gz), allowing for two missed trypsin cleavage sites and precursor-and fragment ion tolerances of 5 ppm and 0.8 Da, respectively. Peptides from ICAT samples were identified by searching MS/MS spectra against the same mouse protein database using Sequest [24].
PeptidePhrophet was used to assess the validity of peptide assignments. Proteins were filtered using ProteinProphet with a computed overall probability of $0.95 for a protein being present in the sample. Only peptide pairs that had a mass difference of 9.0301 Da were included. Both peptide contained cysteins and belonged to a protein that was identified with an Xcorr value$1.5. Averages and standard deviations were calculated for each protein expression value when multiple peptide measurements were available. We only considered peptides with double and multiple charges, and manually evaluated the expression values by inspecting the areas of integration that the software had chosen and by adjusting them as needed. To calculate protein ratio between different pull down samples, XPRESS [56] was used.

Prion inoculations
8-12 weeks old mice were inoculated intracerebrally (ic) or intraperitoneally (ip) with 3610 6 infectious units (IU) or 10610 6 IU, respectively, of Rocky Mountain Laboratory strain (RML, passage 5.0) brain homogenate, prepared as described [57]. Beginning 50 days after inoculation, mice were examined daily for neurological dysfunction and sacrificed on the day of onset of terminal clinical signs of scrapie. For transmission experiments, mice were inoculated ic with up 30 ml of 10% sonicated brain homogenate. Mice were monitored clinically every other day in order to ascertain the onset of clinical signs and the course of the disease. Clinical signs exacerbated over time and included progressive akinesia, priapism (males), hunchback, and stiff tail. Mice were sacrificed on the day of onset of terminal clinical signs of scrapie, defined as the time point at which they became unable to drink and/or eat.