Structural Organization of Mammalian Prions as Probed by Limited Proteolysis

Elucidation of the structure of PrPSc continues to be one major challenge in prion research. The mechanism of propagation of these infectious agents will not be understood until their structure is solved. Given that high resolution techniques such as NMR or X-ray crystallography cannot be used, a number of lower resolution analytical approaches have been attempted. Thus, limited proteolysis has been successfully used to pinpoint flexible regions within prion multimers (PrPSc). However, the presence of covalently attached sugar antennae and glycosylphosphatidylinositol (GPI) moieties makes mass spectrometry-based analysis impractical. In order to surmount these difficulties we analyzed PrPSc from transgenic mice expressing prion protein (PrP) lacking the GPI membrane anchor. Such animals produce prions that are devoid of the GPI anchor and sugar antennae, and, thereby, permit the detection and location of flexible, proteinase K (PK) susceptible regions by Western blot and mass spectrometry-based analysis. GPI-less PrPSc samples were digested with PK. PK-resistant peptides were identified, and found to correspond to molecules cleaved at positions 81, 85, 89, 116, 118, 133, 134, 141, 152, 153, 162, 169 and 179. The first 10 peptides (to position 153), match very well with PK cleavage sites we previously identified in wild type PrPSc. These results reinforce the hypothesis that the structure of PrPSc consists of a series of highly PK-resistant β-sheet strands connected by short flexible PK-sensitive loops and turns. A sizeable C-terminal stretch of PrPSc is highly resistant to PK and therefore perhaps also contains β-sheet secondary structure.


Introduction
Prions are the etiological agents responsible for a diverse set of transmissible fatal neurodegerative diseases of humans and animals, characterized by an abnormal accumulation of prion protein (PrP) [1,2], primarily in the brain. Prions replicate by converting the normal non-infectious cellular prion protein (PrP C ) into a prion (PrP Sc ), via a poorly characterized post-translational conformational transformation. In mice, PrP contains approximately 209 amino acids (numbered 23-231 after cleavage of a 22mer signal peptide) and has four covalent post-translational modifications: two asparagine N-linked glycans at residues N 180 and N 196 , a disulfide bridge between residues C 178 -C 213 and a glycosylphosphatidylinositol (GPI) anchor attached to the Cterminus of the protein (residue S 231 ) [2,3]. Mouse PrP C is a monomer, while PrP Sc is a heterogeneous multimer [2,3]. There have been no demonstrated covalent differences between mouse PrP Sc and PrP C . The only difference between PrP Sc and PrP C is conformational; they are isoforms [2].
The structure of folded, monomeric, recombinant PrP, highly likely to be identical to that of PrP C , has been solved by NMR spectroscopy [4] and X-ray crystallography [5]. In contrast, the structure of PrP Sc remains unclear because the insolubility of PrP Sc and the failure to crystallize the heterogeneous PrP Sc multimers prevent the application of the mentioned high resolution analytical techniques. However, a variety of lower resolution instrumental techniques have provided some information about the structure of PrP Sc . Unlike PrP C , PrP Sc is partially resistant to proteinase K (PK) digestion [2,6]. The secondary structure of PrP C is largely composed of unstructured and a-helical regions, while PrP Sc is largely composed of b-sheet with little, if any, a-helix [7,8,9]. The structure of PrP Sc has also been studied using electron microscopybased analysis of two-dimensional crystals of the PK resistant core of Syrian hamster (SHa) PrP Sc (PrP27-30) [10,11] and mass spectrometry(MS)-based analysis of hydrogen/deuterium exchange [9]. Although theoretical models for PrP Sc have been proposed [10,12], there is an insufficient amount of experimental data to reach a definitive consensus.
In a previous study, we used limited proteolysis to elucidate structural features of PrP Sc [13]. Conformational parameters such as surface exposure of amino acids, flexibility, and local interactions correlate well with limited proteolysis. Peptide bonds located within b-strands are resistant to proteolytic cleavage, whereas peptide bonds within loops and, more rarely, a-helices may be cleaved [14]. Therefore, the targets for limited proteolysis are locally unfolded or highly flexible segments [14]. In our previous study [13], we demonstrated the usefulness of combining limited proteolysis and mass spectrometry (MS) to obtain structural information about two strains of hamster PrP Sc . We concluded that the amino-terminal half of PrP Sc features a series of short PK-resistant stretches, presumably b-strands, interspersed with short PK-sensitive stretches, likely loops and turns. Unfortunately, the structural information was largely limited to the Nterminal portion of the protein, as a consequence of the covalent attachment of the heterogeneous GPI anchor and the heterogeneous asparagine-linked sugar antennae to amino acids in the Cterminal portion of the molecule, which prevented MS-based analysis of this part of the molecule.
Here we extended our studies of the structure of PrP Sc , by using transgenic (tg) mice expressing PrP C lacking the GPI anchor (GPI 2 ) [15]. The GPI 2 PrP Sc produced by these mice is fully infectious, lacks the GPI anchor, and is largely unglycosylated, which reduces the heterogeneity in the C-terminal portion of the molecule [15,16]. These properties make it ideal to carry out structural studies, and have allowed us to obtain, for the first time, a complete survey of the whole PrP Sc sequence, regarding its susceptibility to proteolysis.

Accumulation of PrP Sc in GPI-anchorless Mice
Homozygous GPI-anchorless PrP mice were inoculated at 6 weeks of age with the RML strain of murine-adapted scrapie. Three-hundred sixty-five days post-inoculation, the mice were humanely euthanized. Their brains were surgically removed for further biochemical processing. The presence of PrP Sc was confirmed by digesting a portion of some of these brains, after suitable homogenization, with proteinase K (PK) and analyzing the result by Western blot (Figure 1A and S1). The PK treatment yielded the characteristic PK resistant core protein, referred to as PrP27-30 in PK-treated wild-type PrP Sc , although in this case its apparent MW is lower, given the lack of GPI and sugars. Histological analysis of brains from several of the infected transgenic mice showed a characteristic PrP accumulation pattern, as previously described [15,16], with hyaline deposits arranged radially around blood vessels. Those deposits were strongly immunoreactive to PrP monoclonal antibody 6H4. Deposits were also located submeningeally, subventricularly and scattered in the neuropil ( Figure 1B). In order to verify that the GPI -PrP Sc was infective, a group of ten wild-type (C57BL/6) mice were inoculated with brain homogenate prepared from one of the infected transgenic mice. All ten of these wild-type mice became ill with clinical signs characteristic of the RML strain of murineadapted scrapie and were humanely euthanized. The incubation period of the disease was 154615 days post-inoculation ( Figure 1C).

Identification of PK Cleavage Sites in GPI-anchorless PrP Sc by Mass Spectrometric Detection
We isolated PK-resistant PrP Sc fragments from infected GPI 2 brains. Purity of this material was assessed by SDS-PAGE followed by Coomassie staining ( Figure S2). Using a high resolution Tricine/SDS-PAGE system [17], we compared the distribution of these fragments with that of fragments present in PK-treated unpurified GPI 2 infected brain homogenate, and found them to be similar, which demonstrates that our purification process isolates all of the PK-resistant fragments ( Figure S3). GPI 2 PrP Sc , unlike wild-type PrP Sc , permits the use of MS to accurately identify all PK cleavage sites. This allowed us to analyze samples by Western blot (WB) and by MS.

Identification of PK Cleavage Sites in GPI-anchorless PrP Sc by Western Blot
In parallel we used Tricine-SDS-PAGE [17] followed by WB to analyze the PK-digested GPI -PrP Sc ( Figure 3). When the WB was probed with the antibody #51 (epitope G 92 -K 100 ), just one wide band (,17 kDa) was observed, suggesting a set of cleavage products near G 89 with no C-terminally truncated fragments. A blot probed with the W226 antibody (epitope W 144 -N 152 ), revealed three additional faint bands (,14.6, 13 and 12 kDa), suggesting three PK cleavage sites between the epitopes of these antibodies. Probing with the C-terminal R1 antibody (epitope Y 225 -S 230 ) revealed three more bands (,10.2, 8 and 6.7 kDa), suggesting three additional cleavage sites near residues Y 149 , P 164 and V 175 . These bands agree quite well with our MS-based analysis (vide supra). In order to exclude the possibility that the observed PKresistant fragments are the result of the known preference of PK of certain amino acid residues, rather than structural constraints, we subjected a similar amount of freshly refolded, recombinant MoPrP to cleavage by PK. A concentration of PK much lower than that used with mouse GPI -PrP Sc , 1 mg/ml, completely destroyed all PrP, leaving no PK-resistant fragments larger than 3.5 kDa ( Figure S5). Only PK concentrations below 1 mg/ml yielded some partially resistant fragments, whose sizes do not match those of PK-treated GPI -PrP Sc .

Kinetics of PK Digestion in GPI-anchorless PrP Sc
We performed a PK-digestion time course to determine the relationship of these peptides to one another. A time-dependent reduction in intensity of all PK-resistant bands was observed ( Figure 4). The intensities of the 17, 14.6, 13, 12, and 6.7 kDa bands decreased steadily over time. By 240 minutes the intensities of the 17 and 10.2 kDa bands are nearly equal and by 360 minutes the intensity of the 17, 10.2 and 8 kDa bands are similar. These results are consistent with a progressive digestion of GPI -PrP Sc from the N-terminus. This further suggests that different PKresistant fragments are not from different sub-populations of GPI -PrP Sc , instead they are derived from a larger common GPI -PrP Sc peptide.

PK Cleavage Analysis After Partial Unfolding of GPIanchorless PrP Sc
The above observations were confirmed when the GPI 2 PrP Sc was partially unfolded with increasing concentrations of guanidine prior to PK cleavage, following the procedure of Kocisko et al. [18]. These authors have shown that partial unfolding of PrP Sc with up to 2.5-3 M guanidine is reversible upon dialysis. GPI -PrP Sc became more susceptible to proteolytic digestion in a guanidine-concentration dependent manner. At concentrations above 1 M, the 10.2 and, to a lesser extent, 12 and 8 kDa bands (N 152 -S 232 /M 153 -S 232 , G 141 -S 232 , and Y 162 -S 232 ) predominate. Above 3 M guanidine, which renders the unfolding irreversible [18], almost no PK-resistant material remains ( Figure 5). These results mirror those of the PK time course (vide supra), i.e. all of the bands are derived from the progressive N-terminal digestion of a progenitor peptide. In their original report, Kocisko et al. identified in SHaPrP Sc partially unfolded with guanidine, a highly stable PKresistant core starts before position 143 and continues to the Cterminus [18]. Sajnani et al. also detected a resistant SHaPrP Sc core starting at position 139/142 [13].

Discussion
We present a complete survey of susceptibility to limited proteolysis of a PrP Sc strain ( Figure S6). The map of PKsusceptible spots: 116-118, 133-134, 141, 152-153, 162, 169, and 179, strongly suggests regions corresponding to loops and turns, while nicks at 81, 85, and 89 signal the frontier between the structured C-terminal and unstructured N-terminal domains of PrP Sc . Given the high proportion of b-sheet secondary sctructure derived from FTIR analyses, it is logical to conclude that PKresistant stretches flanking these spots most likely are strands of bsheet.
Our results are in excellent agreement with our previous studies of wild-type PrP Sc [13]. Our experiments with two different SHaPrP Sc strains showed the sequence stretches 23-86 (263K), 23-101 (Dy), 117-119, 131-142, and the region around 154 ( = mouse M 153 ) to be sensitive to PK. In the present study, besides confirming these regions as being PK-sensitive, we identified three additional PK cleavage sites in the C-terminal region of GPI -PrP Sc (Y 162 , S 169 and V 179 ).
We did not find evidence of any PK-resistant peptide with an Nterminus beginning beyond V 179 . This is not a consequence of technical limitations, since the Tricine-based SDS-PAGE allows identification of peptides as small as 3.5 kDa (Figure 3). Instead, either this region is completely resistant to PK, or no stable PKresistant cores remain if PK cleaves beyond that point.
Our results also agree with several studies describing aminoterminally truncated PK-resistant peptides in human CJD PrP Sc .    [21]. This suggests that synthetic prions and PrP Sc share key structural elements, which would explain the capacity of recombinant PrP fibrils to change their conformation, via a ''deformed templating'' mechanism, to that of PrP Sc [22]. In contrast, relatively few C-terminally truncated peptides have been described. Notari et al. reported two human CJD PrP Sc peptides truncated near position 228 [23]. Stahl et al. also reported the presence of a peptide truncated at position 228 in PK-treated SHaPrP Sc [24]. The absence of such fragments in our study could be explained by slight differences in sample preparation, or perhaps by the fact that the absence of the GPI-anchor might have an effect on nearby residues. This conspicuous absence of the C-terminally truncated peptides is a reflection of the stability of the C-terminal region, in GPI 2 PrP Sc appears to be the most stable part of the molecule, which is inconsistent with the presence of substantial stretches of ahelical secondary structure in that region. Our results agree with Smirnovas et al., who showed the C-terminus of GPI -PrP Sc to exhibit extremely low rates of H/D exchange, typical of extensive H-bonding (b-sheet) [9]. These authors showed that an FTIR absorbance band (,1,660 cm 21 ) previously assigned to a-helical secondary structure in PrP Sc is also present in the spectrum of recombinant PrP amyloid fibrils, which contain no a-helices, and therefore cannot be taken as evidence of the presence of a-helical structure. They concluded that GPI 2 PrP Sc consists of a series of b-sheet stretches connected by short loops and/or turns, in agreement with our conclusions. Some stretches exhibiting a somewhat higher exchange rate, suggested to overlap with loops/ turns, such as 133-148 or 81-118, are consistent with flexible stretches identified in our study, although discrepancies also exist. The limited resolution of both analytical techniques prevents a more exhaustive comparison, but overall both of them agree.
GPI -PrP Sc fibrils are about 3-5 nm wide ( [25] and our unpublished results). This constraint means that each PrP Sc monomer must be coiled in such a way as to fit approximately 140-145 residues (,G 85 -S 232 ) into this width. To do so, PrP Sc monomers must necessarily adopt a multi-layer architecture, as seen in SH3 fibers [26] or the HET-s fungal prion domain [27]. The HET-s prion domain packs 70 residues into two b-strands alternating with turns and loops [27]. Wille et al. have suggested that PrP Sc fibrils are composed of four rungs of b-strands, based on their interpretation of X-ray diffraction patterns [28]. In this model, each rung would comprise ,36-37 residues. Positions N 152 -M 153 lie near the middle of the G 85 -S 232 sequence, so it is tempting to speculate that they might be located at an exposed position at the border between rungs. This might explain why the N 152 -S 232 and/or M 153 -S 232 fragment emerges as the most conspicuous PK-resistant fragment after prolonged treatment with PK or partial unfolding with guanidine (Figures 4 and 5). Positions A 116 -G 118 might be the border between the two most aminoterminal rungs (approximately G 85 -A 115 and A 119 -E 151 ). On the other hand, our results are partially inconsistent with the location  assigned by Govaerts et al., using threading algorithms, to residues K 100 -P 104 and E 145 -R 163 , placed in loops and not rungs [10]. Our data show that the stretches formed by residues K 100 -P 104 , N 142 -E 151 , and Y 154 -Y 161 , are PK-resistant, i.e., likely part of a b-strand rung ( Figure 2 and Table 1).
In summary, our data support a PrP Sc structure consisting of a series of highly PK-resistant b-sheet strands interspersed with PKsensitive short flexible loops and turns. Furthermore, the region comprising ,V 179 to the C-terminus of PrP Sc is probably composed primarily of b-sheet, as it is highly resistant to PK. Our data are consistent with our previous results (263K and Dy strains) and those of other researchers using SHaPrP Sc . Furthermore, they are consistent with those observed for human CJD PrP Sc , which suggests that the myriad human, hamster and mouse prions share a common basic structure.

Ethics Statement
Animal experiments were carried out in accordance with the European Union Council Directive 86/609/EEC. The procedures and animal care were governed by a protocol that was approved by the Institutional Ethics Committee of the University of Santiago de Compostela. All efforts were made to minimize the suffering of the animals.

Animals
Transgenic heterozygous GPI-anchorless (GPI -) PrP mice (tg44(+/2)) were a generous gift from Bruce Chesebro, Rocky Mountain Laboratories, NIH, Montana, USA. Mice were crossed to obtain homozygous GPIanimals (tg442/2), which were identified by tail DNA analysis using the PCR protocol described by Chesebro et al. [15]. Homozygous animals were bred and expression of GPI -PrP confirmed by Western blot ( Figure S1). Female mice were intracerebrally inoculated at six weeks of age with 20 ml of a 2% RML-infected mouse brain homogenate (BH), kindly provided by Juan María Torres, CISA, Madrid, Spain. After 365 days post inoculation, the asymptomatic mice [16] were euthanized, their brains surgically removed, rinsed in PBS, and stored at 280uC until needed.

Preparation of Brain Homogenates and Isolation of GPIanchorless PrP Sc
Mouse BH, 10% w/v, were prepared in PBS, 5% sarkosyl, using a dounce homogenizer (Wheaton Industries Inc, NJ, USA), followed by one pulse of sonication to clarify the homogenate, with an ultrasonic homogenizer probe (Cole Parmer Instrument CO., Chicago IL, USA). GPI 2 PrP Sc was isolated using the method of Baron et al. [8]. During the purification, total PrP Sc was treated with 10 mg/ml of proteinase K. The final GPI 2 PrP Sc pellet was resuspended in 100 ml of deionised water or in 20 ml of a 6 M guanidine solution (final concentration 1.75 mg/ml). The stock suspension was stored at 4uC. Its purity was assessed by Coomassie stained SDS-PAGE gel and estimated to be ,95% pure. The yield of GPI -PrP Sc was ,35 mg per brain (BCA protein assay).

Recombinant PrP
Recombinant Mouse PrP(23-231) was expressed in E. coli, and purified and refolded in-column on an NTA affinity column (GE Healthcare, Uppsala, Sweden), as previously described [29]. Refolded protein was dialyzed against 10 mM sodium phosphate buffer pH 5.8 and then against d.i. water.

Limited Proteolysis
Aliquots of BH (10% in PBS, 5% Sarkosyl) were digested with PK (Sigma-Aldrich, St. Louis, MO, USA) in 20 mM Tris-HCl pH 8.5 at 37uC for 1 h unless otherwise stated. Digestion was stopped by addition of Pefabloc (Fluka, Buchs, Switzerland) to a final concentration of 2 mM. Deglycosylation was carried out with 2 ml of PNGase F solution (New England Biolabs, Ipswich, MA, USA) at 37uC for 48 h, according to the manufacturer's instructions.
Digestion with PK After Partial Unfolding with Guanidinehcl (Gnd) Samples of BH (5 ml) were mixed with an equal volume of an appropriate aqueous Gnd solution to yield the desired final Gnd concentration and then incubated at 37uC for 1 h. After incubating, the samples were diluted with buffer (20 mM Tris-HCl pH 8.5) to yield a 0.4 M Gnd solution, which were then treated with PK (25 mg/ml) for 1 h at 37uC. The digestion was stopped by adding Pefabloc (2 mM final concentration) and the protein was precipitated by addition of ice-cold methanol (85% final concentration). The resulting pellets were resuspended in 9 ml of deionized water, and deglycosylated with PNGase F (vide supra).

Mass Spectrometry
NanoLC/ESI/MS analysis was done with an Applied Biosystems (AB SCIEX, Framingham, MA) model QStar Pulsar equipped with a Proxeon Biosystems (Odense, Denmark) nanoelectrospray source. Samples of the Gnd stock solution (vide supra) were loaded automatically onto a C-18 trapping cartridge and chromatographed on a reversed-phase column (Vydac Everest 238EV5.07515, 75 mm 6 150 mm) fitted with a coated spray tip (FS360-50-5-CE; New Objective, Inc.). A nanoflow LC system (Dionex, Sunnyvale, CA) with autosampler, column switching device, loading pump, and nanoflow solvent delivery system was used. Elution solvents were A (0.5% acetic acid in water) and B (0.5% acetic acid in 80% acetonitrile/20% water). Samples were eluted at 250 nL/min using a binary gradient (8% B at 0 min to 80% B in a 30 min linear gradient, held at 80% B for 5 min, then back to 8% B for 15 minutes). The QStar Pulsar was externally calibrated daily with human [Glu1]-fibrinopeptide B.
In parallel, 1 mL of the Gnd stock solution was mixed with with 49 mL of sinapinic acid (SA) solution (10 mg/mL SA dissolved in 30% ACN with 0.3% TFA) and analyzed by MALDI-TOF. One half mL aliquots were deposited using the dried-droplet method onto a 384 Opti-TOF MALDI plate (Applied Biosystems, Foster City, CA, USA). MALDI analysis was performed in a 4800 MALDI-TOF/TOF analyzer (Applied Biosystems, Foster City, CA, USA). MS spectra were acquired in linear mode (20 kV source) with a Nd:YAG, (355 nm) laser, and averaging 500 laser shots. The mass of the peptide M 153 -S 232 (9573 Da) was determined by an iterative calibration approach, using insulin (m/z = 5733), ribonuclease A (m/z = 13682) and lysozyme (m/ z = 14305), (Sigma-Aldrich, St. Louis, MO) as internal standards. Then, the signals from the M 153 -S 232 (9573 Da), G 89 -S 232 (16371 Da), and G 81 -S 232 (17148 Da) peptides were used to calibrate the rest of peaks in the spectrum. Masses were matched to PrP fragments with the help of GPMAW 6.0 software (Lighthouse, Odense, Denmark).

Immunohistochemistry
Immediately after extraction, the brain was fixed in formalin and then sliced into four transversal sections by cutting the brain caudally and rostrally to the midbrain and at the level of the basal nuclei. The sections were dehydrated by equilibration in solutions of progressively higher ethanol concentration and then equilibrated with xylene before being embedded in paraffin. Haematoxylineosin was used to stain the 4 mm thick sections. Additional sections were mounted on 3-triethoxysilyl-propylamine-coated glass slides for immunohistochemical (IHC) studies. These brain sections were deparaffinised, immersed in formic acid containing peroxidase inhibitors, and autoclaved prior to IHC analysis. These autoclaved samples were washed, treated with proteinase K, washed again, and then incubated overnight with the antibody 6H4 (1:2000, Prionics AG, Schlieren, Switzerland). The sections were developed using the DAKO EnVision system and 3,39diaminobenzidine as the chromogenic substrate. Figure S1 Western blot of unpurified GPI 2 PrP Sc 2/+ PK. Both samples were treated with PNGase F. WB was probed with the #51 antibody. (TIF) Figure S2 Characterization of isolated GPI 2 PrP Sc . 10 ml of sample were loaded and separated in a 15% gel by SDS-PAGE. The gel was stained by Coomassie blue. The molecular weight of the GPI-less PrP27-30 is ,16750 Da. (TIF) Figure S3 Western blot of PK-resistant fragments. In unpurified (1) and purified GPI-PrP Sc (2). Both samples were digested with proteinase K, 25 mg/ml and 10 mg/ml, respectively, treated with PNGase F and resolved on a Tricine-SDS-PAGE gel. WB was probed with the R1 antibody.