Evaluating Prion Models Based on Comprehensive …Structure Article Evaluating Prion Models Based on...
Transcript of Evaluating Prion Models Based on Comprehensive …Structure Article Evaluating Prion Models Based on...
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
Structure
Article
Evaluating Prion Models Basedon Comprehensive Mutation Data of Mouse PrPTsuyoshi Shirai,1,2,* Mihoko Saito,1,2,5 Atsushi Kobayashi,3 Masahiro Asano,3 Masaki Hizume,3 Shino Ikeda,3
Kenta Teruya,3,6 Masanori Morita,3,4,7 and Tetsuyuki Kitamoto3,*1Department of Computer Bioscience, Nagahama Institute of Bio-Science and Technology, Nagahama, Shiga 526-0829, Japan2Bioinformatics Research Division, Japan Science and Technology Agency, 5-3, Yonbancho, Chiyoda-ku, Tokyo 102-8666 Japan3Department of Neurological Science, Tohoku University Graduate School of Medicine Research, 2-1 Seiryo-machi, Aoba-ku,
Sendai 980-8575, Japan4Research and Development Division, Benesis Corporation, Kitahama, Chuo-Ku, Osaka 541-850, Japan5Present address: Laboratory of Functional Analysis in Silico, Human Genome Center, The Institute of Medical Science, The University ofTokyo, 4-6-1 Shirokane-dai, Minato-ku, Tokyo 108-8639, Japan6Present address: Department of Chemistry, Graduate School of Medical Science, Kyoto Prefectural University of Medicine, Kita-ku,
Kyoto 603-8334, Japan7Present address: Infectious Pathogen Research Section, Central Research Laboratory, Japan Blood Products Organization,Kobe KIMEC Center Building 8F, 1-5-2 Minatojima-Minamimachi, Chuou-ku, Kobe, 650-0047, Japan
*Correspondence: [email protected] (T.S.), [email protected] (T.K.)
http://dx.doi.org/10.1016/j.str.2013.12.019
SUMMARY
The structural details of the essential entity of priondisease, fibril prion protein (PrPSc), are still elusivedespite the large body of evidence supporting theprion hypothesis. Five major working models ofPrPSc structure, which are not compatible witheach other, have been proposed. However, no sys-tematic evaluation has been performed on thosemodels. We devised a method that combined sys-tematic point mutation with threading on knowl-edge-based amino acid potentials. A comprehensivemutation experiment was performed on mouse prionprotein, and the PrPSc conversion efficiency of eachmutant was examined. The models were evaluatedbased on the mutation data by using the threadingmethod. Although the data turned out to be rathermore consistent with the models that assumed aconversion of the N-terminal region of core PrP intoa b helix than with others, substantial modificationswere also required to further improve the currentmodel based on recent experimental results.
INTRODUCTION
The prion hypothesis assumes conversion of prion protein (PrP)
from the innocuous cellular form (PrPC) into a cytotoxic fibril form
(PrPSc) during infection by prion diseases such as Creutzfeldt-
Jakob disease, scrapie, and bovine spongiform encephalop-
athy (Prusiner, 1991, 1998). The protein in the latter form is
hypothesized to be the sole pathogenic entity that transmits
the disease without mediation by other substances such as
nucleic acids. Much biochemical, biomolecular, and patholog-
ical evidence that supports this hypothesis has been accumu-
lated to date.
Struct
The single difference between PrPC and PrPSc is thought to be
their molecular conformations. Therefore, the three-dimensional
(3D) structure of PrPSc should represent the most essential infor-
mation to understand, treat, and prevent prion diseases. Despite
the importance of the structure, however, little consensus on
PrPSc structure has been established (Diaz-Espinoza and Soto,
2012). It is currently known that, first, the b strand content in-
creases during conversion from PrPC to PrPSc (Caughey et al.,
1991; Gasset et al., 1993). Second, a periodic structure at
4.8 A intervals exists in the PrPSc structure, which is interpreted
to be a long cross-b sheet perpendicular to the fibril axis of PrPSc
(Nguyen et al., 1995; Wille et al., 2009). Third, PrPC to PrPSc con-
version increases the resistance of residues V121–S231 of core
PrP (unless otherwise noted, amino acids and residue numbers
are those of mouse PrP, and the region S102–P240 of PrP is
called core PrP in the following sections) to proteinase K diges-
tion, which is the major hallmark of prion infection, by reflecting
the structural changes from PrPc to PrPSc (Prusiner, 1998). To
date, a series of electron microscopic, X-ray diffraction, and
biochemical studies of PrPs have resulted in several 3D struc-
tural working models of PrPSc that are not compatible with
each other (Figure 1).
The structure of the core region of mouse PrPC (residues
V121–S231) was determined with nuclear magnetic resonance
spectroscopy in 1996 (Figure 1A; Riek et al., 1996). Govaerts
and colleagues proposed the first molecular model of PrPSc
based on the electron microscopic (EM) image of a two-dimen-
sional (2D) crystal in 2004 (Govaerts et al., 2004; Wille et al.,
2002). They detected compatibility between the N-terminal re-
gion of core PrP (residues G89–H176) and the b helix structure
by using a threading method. Based on microscopic images of
the wild-type PrP, a deletion mutant (PrPD141–176), and carbo-
hydrate-depleted proteins, they proposed amodel (called BH1 in
this study; Figure 1B), in which the N-terminal region of core PrP
was converted into a triangular b helix containing four turns (four-
rung model), and was associated with the intact C-terminal
a2-a3 helix bundle. The converted PrP subunits were assembled
into a trimer by using the sidewall of the b helix as an interface,
ure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved 1
Figure 1. Prion Models
(A–G) The PrPC and PrPSc working models, (A) PrPC, (B) BH1, (C) BH2, (D) SP, (E) BH3, (F) IS, and (G) BH4, used in this study are shown schematically. Top: fibril
longitude view, where the BH3, IS, and BH4models are superimposed on the 3D EM image. b strand and a helix are shown as blue and red ribbons, respectively,
and N-glycan moieties are shown as gray wire in each model. Second row: fibril cross-section view of the models and the 3D EM image. Third row: pie graphs of
fractions of consistent (blue), inconsistent (red), and ambiguous (white) sites in coarse evaluation. Fourth row: pie graphs of fractions of sites in fine (threading)
evaluation that are consistent due to PrPSc stabilization (blue), consistent due to PrPC destabilization (light blue), inconsistent (red), and ambiguous (white).
See also Figures S1 and S2 and Table S1.
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
and the PrP trimers were stacked into fibrils by forming an inter-
subunit b helix.
Langedijk and colleagues proposed the second model (BH2;
Figure 1C) in 2006 (Langedijk et al., 2006). They also assumed
conversion of the N-terminal region of core PrP (residues
P104–D143) into a b helix, according to sequence comparisons
and molecular dynamics (MD) simulations. The b helix consisted
of two turns (two-rung model) with eight residues in one edge of
the triangle in contrast to the five residues in the BH1model. The
BH2model was also assumed to form a fibril in the samemanner
as the BH1 model.
DeMarco and Daggett observed extension of a native b sheet
consisting of b1 and b2 strands and recruitment of a nascent b
strand to the b sheet in an MD simulation, and proposed amodel
in which subunits were assembled into a spiral fibril by forming
intersubunit b sheets (SP model; Figure 1D; DeMarco and Dag-
gett, 2004). These three models assumed that conformational
conversion was limited to the N-terminal region of core PrP,
and that the C-terminal a2-a3 helix bundle was largely intact.
The 3D reconstruction of anEM imageof amousePrP fibril con-
verted in vitro obtained byTattumand colleagues demonstrated a
ladder-shaped fibril structure, which seemed very different from
the preceding 2D image (Tattum et al., 2006). Cobb and col-
leagues observed the electron paramagnetic resonance (EPR) of
human PrP fibrils converted in vitro, and proposed an in-register-
2 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reser
stacking model (IS model; Figure 1F), like that also proposed for
several amyloid proteins, based on the interresidue distances
(Cobb et al., 2007; Lu et al., 2007). In thismodel, theC-terminal re-
gion (residuesQ159–K219 inmouse PrP) is converted to a hairpin
structure, and the fibril is formed as a long two-layered b sheet,
which is moderately similar to the 3D EM image.
Kunes and colleagues proposed another model (BH3; Fig-
ure 1E) based on the ladder-shaped 3D EM image, in which
the N-terminal region (residues G89–D143) and the C-terminal
region (residues Y154–G227) of core PrP were independently
converted into b helices. The b helices were stacked into rails,
and the linker peptide between the b helices formed the steps
of the ladder (Kunes et al., 2008).
Although these five major working models of PrPSc are
apparently not consistent with each other, they have not been
systematically compared. The fibril nature of PrPSc prevented
elucidation of its structure with ordinary structure determination
methods such as X-ray crystallography. In such cases, a system-
atic site-directedmutagenesis study such as alanine scanning or
b value evaluation can be used to probe structures by correlating
a substituted residue or site position to the effect of substitution
(Fersht and Daggett, 2002; Sidhu and Kossiakoff, 2007). This
method has been applied to investigate protein folding interme-
diates or temporary complex structures for which a direct struc-
ture determination is difficult.
ved
Figure 2. Conversion Efficiencies of A115X
Mutant PrPs
Conversion efficiencies of A115X (X is every
possible amino acid) mutant PrPs.
(A) Western blot analysis of total cell-associated
PrP after deglycosylation with PNGaseF.
(B) Western blot analysis of PK-resistant PrP.
(C) The mean fold increase of conversion efficiency
of each mutant from the wild-type over three inde-
pendent measurements. The conversion efficiency
of wild-type (WT) PrP was assigned as 1. Data are
expressed as mean ± SEM. The color codes for the
fold increase of conversion efficiency are shown at
bottom (red, increased; green, no change; blue,
decreased). The mutants indicated with asterisk,
namely, A115C and A115P, were excluded from the
model evaluation.
(D and E) As a confirmation of constant conversion
efficiency against total amount of PrP, the amount
of PK-resistant PrP with 3F4- and T41-epitope-tags
was plotted against varying total amounts of the
corresponding PrP. The values are relative amounts
of protein as quantified from the gel images shown
in Figure S5. The correlation coefficients in the 3F4-
detection system (D) and the T41-detection system
(E) are 0.93 and 0.97, respectively.
See also Figures S3–S5.
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
In this study, we devised amethod that combines comprehen-
sive point mutation data with threading based on knowledge-
based amino acid potentials, which is used to evaluate the
compatibility of amino acids to a molecular environment in a
defined protein structure (Luthy et al., 1992). A systematic point
mutation experiment was executed onmouse PrP, and the PrPSc
conversion efficiencies in vitro (in-cell conversion) were exam-
ined by using a scrapie-infected mouse cell line.
The results of the experiment on the working models of PrPSc
were further analyzed by using the threading method to judge
whether the models were consistent with the comprehensive
mutational data. With this method, each site of the molecular
models was tested against 20 amino acids for compatibility
with the molecular environment of the site. The results demon-
strated that the mutation data were rather more consistent with
the models that assumed an N-terminal b helix of core PrP
than with the others. However, molecular modeling also indi-
cated that a substantial improvement of the current models is
required to make them consistent with the recently accumulated
experimental data.
RESULTS
Comprehensive Point Mutations of Mouse PrPThe PrPSc conversion efficiencies for comprehensive point mu-
tants of mouse PrP within the region S102–P240 were evaluated.
The mutant genes were expressed in scrapie-infected mouse
Structure 22, 1–12, April 8, 2
neuroblastoma (ScN2a) cells (Figure 2;
Figures S1 and S2 available online). The
PrP expressed and converted into PrPSc
in the ScN2a cell is infective in mice (Butler
et al., 1988; Race et al., 1988). Also, a pos-
itive correlation between the conversion
efficiency of PrP mutants in ScN2a cells and that of the mutants
in transgenic animals has been demonstrated (Kaneko et al.,
1997; Perrier et al., 2002). Therefore, the experimental system
of this study would be useful to evaluate PrPSc working models
on the basis of conversion efficiencies of the PrP mutants.
The conversion efficiencies were calculated as relative
amounts of a mutant PK-resistant PrP molecule, which was cor-
rected by the expression levels of the PrP molecule without PK
digestion to that of the transfectedwild-type (Figure 2). The linear
correlation between expression levels of PrP and the amounts of
converted PrP, or constancy of conversion rate, was confirmed
for the wild-type PrP (Figures 2D and 2E). The transfected PrPs
(PrPC) are readily converted by intrinsic PrPSc nuclei, which sus-
tained in the ScN2a cell. Therefore, the conversion efficiency
should represent the tendency of mutant PrPs to be integrated
into PrPSc, rather than the tendency to spontaneously convert
into PrPSc.
As a result, the residue sites in which point mutations tended
to increase conversion efficiency were mainly localized in
residues K103–A119 and S131–M137 in the primary structure
(Figure 3 and Table S2). The ten most effective sites (with the
highest average of fold-increase of PrPSc conversion over all
amino acids) were A115, A117, K105, K109, M137, K103,
V121, S131, S134, and V111. The ten most efficient mutants
were V121I (23.3-fold increase), K105R (17.1), A115S (16.5),
H176R (15.6), A115W (15.5), A117V (14.8), Y127I (13.4),
A115N (12.7), P104T (12.1), and A115Y (10.6). Thus, the
014 ª2014 Elsevier Ltd All rights reserved 3
Figure 3. Conversion Efficiency of PrP Mutants
The fold differences of PrPSc conversion efficiency of mouse PrP mutants from the wild-type are shown in a heat map (red, increased; green, no change; blue,
decreased). The mutants that were excluded from analyses (conversion efficiency < 0.2) are indicated with ‘‘/’’. The horizontal axis is residues from G89 to P240,
(legend continued on next page)
Structure
Prion Model Evaluation
4 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
N-terminal region of core PrP appeared to play an important
role in PrPSc conversion.
In contrast, mutations in the C-terminal helical regions of PrPC
rarely increased efficiency or reduced protein production by
much, except for few residues close to the helix C termini,
such as T189 or S231. It has been demonstrated that the mu-
tants at S231 are GPI-modified in this experimental system,
therefore might be used for analyses (Hizume et al., 2010). It
was also shown that S230 and S232 were critical for GPI-
anchoring, and some of the mutants (W, Y, R, F, H, M, E, K, Q,
I, L, and C for S230 and W, Y, F, M, I, L, and P for S232) were
not GPI-modified, and were not localized to membrane. These
mutants have been excluded from the analyses due to a low
expression (Figure 3).
Constructing PrPSc Working Model CoordinatesEvaluations of the mutation data against various PrPSc working
models were then attempted. The atomic coordinates of the
models were required for this purpose. Unfortunately, however,
none of the model coordinates turned out to be publicly avail-
able. Therefore, model coordinates were prepared from scratch,
as detailed in the Supplemental Experimental Procedures, refer-
ring to the original reports. Information explained in texts or pre-
sented in graphics was generally not sufficient to reproduce
atomic coordinates. Consequently, those used in this study
resemble models of the proposed ones to a certain extent, and
the authors of this paper are responsible for any discrepancy
from the original models. Other proposed models that are
considered only a small part of PrP or for which enough informa-
tion on coordinates was not provided were not considered in this
study.
The BH1 (Govaerts et al., 2004), BH2 (Langedijk et al., 2006),
SP (DeMarco and Daggett, 2004), BH3 (Kunes et al., 2008),
and IS (Cobb et al., 2007) models were constructed and refined
by combining manual model constructions and energy calcula-
tions. The subunits contained between 74 and 143 amino acid
residues, and they had N-glycan moieties (Man a1-6[(GlcNAc
b1-4)][(Man a1-3)]Man b1-4GlcNAc b1-4-[(Fuc a1-6)]GlcNAc-
Asn; Stimson et al., 1999) at N180 and N196, if possible.
Everymodel had at least 98%of the residues in the generously
arrowed region in the Ramachandran plot and less than 0.22 A
and 2.5� SD from the ideal geometries for bond distance and
angle, respectively (Figure S1 and Table S1). The fibril models
contained from 8 to 16 subunits where at least one of the sub-
units was fully surrounded by other subunits so that all interfaces
could be examined (Figure S2).
Evaluating Models by Threading MethodA simple but probable assumption was made for the model eval-
uation; that is, a point mutation that increased conversion effi-
ciency would destabilize the PrPC structure and/or stabilize the
and vertical axis shows substituted amino acids. The BH4model interface 1 (D143
orange), and 4 (Y168–N173 and D226–S231, green) are indicated above the sequ
and PrPSc working models are also shown below the map. For eachmodel, the to
and blue arrow, respectively; the a helices a1-a3 and b strands b1-b2 of PrPC ar
(blue, positive correlation; green, no significant correlation; red, negative correla
assignments are coordinated with those defined in the literature based on the corr
in this study. See also Table S2.
Struct
PrPSc structure, and the highly influential sites would be in
different molecular environments in PrPC and PrPSc.
To avoid biases from aberrant PrP samples, the data on
mutants were discarded when a defect in protein expression,
glycosylation, or conversion into PK-resistant formwas detected
(Figures 2 and S2). The confirmation of glycosylation in mutant
PrP should guarantee proper cellular sorting of the proteins,
and that of conversion into PK-resistant form was required to
exclude folding-deficient mutants of PrPC.
PrPs have been known to interact with various intracellular co-
factors (Caughey et al., 2009). For that reason, some of the PrP
mutants might have affected the conversion efficiency owing
to the interactions with cofactors rather than destabilization or
stabilization of PrPC or PrPSc. However, the mechanisms and in-
teractions of cofactors that directly affect PrPC to PrPSc conver-
sion have not been sufficiently clarified at this point. Additionally,
PrPC is known to convert into PrPSc through several intermedi-
ates, although the structural detail is not clarified (Stohr et al.,
2008). Because the fitness of mutant sequences to initial (PrPC)
and final (PrPSc) structures have been evaluated, the presented
analysis is analogous to an equilibrium evaluation, and the inter-
actions in intermediate structures have not been explicitly taken
into account.
As already mentioned, transfected PrPs are readily converted
by intrinsic PrPSc templates in the conversion system of this
study. It implies that the energies of intermediate structures are
very much lowered. Without this condition, the conversion effi-
ciency might be largely ruled by conversion kinetics depending
on the formation and stability of intermediate structures. The
consistency of this analysis obviously depends on an important
condition, that is, PrPC to PrPSc conversion pathway is not
impaired by a mutation. If a mutant makes PrPSc state unreach-
able, then PrPSc is not observed. For this reason, the data of the
mutants with low conversion efficiency (less than 0.2) were
totally discarded, although these mutants might contain consid-
erable number of interesting cases, in which PrPSc structure was
largely destabilized without affecting conversion pathway. Addi-
tionally, PrPC to PrPSc conversion has not been observed in N2a
cells, the parent strain of ScN2a (Hiraga et al., 2009). Thus, the
most influential cofactor in this system is thought to be PrPSc
itself. In this study, therefore, the conversion efficiencies of
mutants were interpreted entirely as a function of initial (PrPC)
and final (PrPC) structures, although interactions in intermediate
structures and with cofactors remain to be a major limitation of
current evaluation.
The evaluation was made in both coarse and fine scopes. In
the former evaluation, the sites that made a larger difference in
conversion efficiency and in different molecular environments
between PrPC and PrPSc in each PrPSc working model were
thought to be consistent with the corresponding model. As a
result, the excesses of the fraction of consistent sites over those
–R155, red), 2 (P104–A115, blue), 3 (K105–A115, G125–R135, andD143–M153;
ence. The secondary structure and effective correlation coefficient of the PrPC
p is the scheme of the secondary structure (a helix and b strand are red cylinder
e indicated), and the bottom is a heat map of effective correlation coefficients
tion; the color code is identical to that in Figure 1C). The secondary structure
esponding original models, even if they are distorted in the models constructed
ure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved 5
Table 1. PrPSc Model Evaluation
Model Evaluationsa
Model Clearance
for PeripheralsbCourse Evaluation (%) Fine (Threading) Evaluation
Protease Restriction Site
Consistency (%)
Consistent Inconsistent Excess Consistent (%) Inconsistent (%) Excess (%) AECC
Exposed
on ModelcExposed or
out of Modeld NT NG SS CT
PrPC 18.4 44.7 �26.3 �0.26
PrPSc
BH1 37.2 21.3 16.0 50.0 7.5 42.6 �0.09 28.6 40.0 no hard
BH2 55.3 26.6 28.7 52.2 14.1 38.1 �0.13 18.8 48.0
SP 40.4 14.9 25.5 49.4 11.5 37.9 �0.10 18.8 48.0 no
BH3 56.4 24.5 31.9 52.1 12.8 39.4 �0.12 52.4 60.0 no
IS 54.3 26.6 27.7 38.1 23.8 14.3 0.00 33.3 92.0 hard
BH4 53.2 25.5 27.7 54.4 13.0 41.3 �0.19 50.0 68.0aPercentage of sites for which mutation data are consistent or inconsistent with each model, excess of consistent fraction over inconsistent fraction,
and average of effective correlation coefficients over sites in each model (AECC).bJudgment on whether a model can accommodate an N-terminal peptide (NT), N-glycans (NG), disulfide bond (SS), and C-terminal peptide, and
GPI (CT).cPercentage of protease restriction sites of PrPSc modeled as accessible residues in each model.dPercentage of protease restriction sites of PrPSc modeled as accessible residues or not in each model.
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
of inconsistent sites were 16%, 29%, 26%, 32%, and 28%of the
total for BH1, BH2, SP, BH3, and ISmodels, respectively (Table 1
and Figure 4). The value for BH1 was significantly lower than that
of the other models.
In the fine evaluation, threading profiles were calculated for
each model and used to correlate the difference in amino acid
fitness (the difference in the pseudo-free energy of amino acids
before and after point mutation) to the changes in PrPSc conver-
sion efficiency (Ota et al., 2001). A site was regarded as
consistent if the conversion efficiency increment was negatively
correlated to the difference in pseudo-free energy on the PrPSc
working model (because it stabilized PrPSc) and/or positively
correlated to that on PrPC model (because it destabilized
PrPC). The excesses of the consistent site fraction over that of
the inconsistent site were 43%, 38%, 38%, 39%, and 14% of
the total for BH1, BH2, SP, BH3, and IS models (Table 1 and Fig-
ure 4). In this case, the value for IS was significantly lower than
that for the other models.
The averages of correlation coefficients over all sites in each
model were �0.09, �0.13, �0.12, �0.10, and 0.0 for BH1,
BH2, SP, BH3, and IS models, respectively. Altogether, the
mutation data appeared to be relatively coherent with the BH2
or BH3 models, followed by the SP model (Figure 4).
Interestingly, correlation analysis showed that destabilization
of the PrPC structure did not always increase conversion effi-
ciency. The consistent site fraction was much smaller than that
of inconsistent site fractions for PrPC (Table 1). A close inspec-
tion of the result suggested that this wasmainly due to the contri-
bution of mutations that destabilized the C-terminal a2-a3 helix
bundle (Figure 5). The result implied that the helix bundle struc-
ture is required for the PrPSc conversion process.
Another explanation of this result might be provided by the
recently found interaction between the C-terminal a2-a3 helix
bundle and the octarepeat at N-terminal region of full-length
PrP (Spevacek et al., 2013). It was suggested that the mutations
at C-terminal a2-a3 helix bundle promoted PrPSc conversion by
6 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reser
disrupting the interaction with octarepeat region, rather than by
destabilizing the helix bundle. At this point of time, however,
evaluation of the comprehensive mutation data under the pres-
ence of the structural model of interacting octarepeat should
remain as an issue in the future, because atomic details of the
interaction are not provided.
Constructing Refined PrPSc Working ModelAlthough the BH2 or BH3 models were judged to be compatible
with the mutation data, the triple-helix fibril structure of the BH2
model and ladder structure of the BH3 model did not resemble
the 3D and 2D EM images, respectively, of the mouse PrP fibril
(Govaerts et al., 2004; Tattum et al., 2006). Therefore, modifica-
tion of the BH2 model was attempted to construct a model more
consistent with the recent experimental data (BH4 model; Fig-
ure 1G). The BH2 model was selected as a starting point
because preservation of the C-terminal a2-a3 helix bundle was
suggested by the analyses described above.
Themodeling process of the BH4model is detailed in the Sup-
plemental Experimental Procedures (Figure S2G). Briefly, the
N-terminal b helices of core PrP were arranged so that they
comprised the rails of a ladder, and the a2-a3 helix bundles
were placed on steps of the ladder. The interfaces between sub-
units were selected according to the experimental results of
PrPC to PrPC/PrPSc interactions obtained with the hydrogen ex-
change, antibody inhibition, and peptide-array methods (Rigter
et al., 2007, 2009; Solforosi et al., 2007).
Four different interfaces were assumed for a subunit in the
BH4 model. The interfaces 1 and 2 were at the top and bottom
of the N-terminal b helix of core PrP, and were also assumed in
the BH1, BH2, and BH3 models (Figure 6A). Interface 3 provided
lateral interaction between b helices from neighboring subunits,
which was also hypothesized in the BH1 and BH2 models. By
using interfaces 1, 2, and 3, like the BH2 model, the fibril of the
BH4 model could be constructed as observed in the 2D EM
image (Figure S6A). The additional interface 4 was located
ved
Figure 4. Evaluating PrPSc Working Models
The PrPSc working models are plotted (A) against
the excess of the consistent site fractions in
coarse (horizontal) and fine evaluations (vertical),
and (B) against the excess of the consistent site
fraction in the fine evaluation and the average of
effective correlation coefficients over the sites in
each model (AECC).
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
between a2-a3 helix bundles, which was suggested from the
peptide-array experiment for sheep PrP. The ladder-shaped fibril
could be formed with interfaces 1, 2, and 4.
The excess of consistent over inconsistent site fractions for
the BH4model was 28% (coarse evaluation), and 42% (fine eval-
uation). The average of the efficient correlation coefficient was
�0.19. All of these values were comparable to the best values
from other models (Figure 4). Also in this model, the experimen-
tally detected interfaces between PrPC and PrPSc (Rigter et al.,
2007, 2009; Solforosi et al., 2007), which might be important to
initiate the template-assisted conversion (Figure 6B), and highly
influential sites observed in the mutation experiment (Figure 6C)
were thoroughly used for intersubunit interactions.
Additionally, the models were examined by referring to the re-
sults of protease restriction of mouse GPI-less PrPSc (Sajnani
et al., 2008; Vazquez-Fernandez et al., 2012). The percentages
of protease restriction sites, which were solvent-accessible in
each model, were relatively large for the BH3 and BH4 models
(Table 1). When the out-of-model sites were also counted as cor-
rect protease restriction sites, because those sites might be in a
disordered region, the ISmodel was themost consistent with the
experimental results because most of the protease restriction
sites were outside of this model. Under this criterion, the BH4
model was the second most consistent among the other models
(Table 1).
DISCUSSION
Coherence of Mutation Data to b Helix ModelsThe five major PrPSc working models were evaluated using the
comprehensive mutation data of mouse PrP and compared to
each other in atomic detail. An intriguingmethodology that corre-
lated comprehensive mutation data to threading profiles was
devised for this purpose. The experimental data revealed high
consistency of particular models—those that involved conver-
sion of the N-terminal region of core PrP into a b helix.
Preparation of in vivo PrPSc samples that are suitable for
the ordinary structure determination methods is quite difficult.
Therefore, some of the PrPSc working models were based on
the analyses of in vitro generated PrP fibrils. The difference be-
tween converted PrP structures grown in vitro and in vivo has
long been discussed, mainly in terms of their infectivity, because
PrP converted in vitro (in cell-free conversion systems) tended to
demonstrate very low infectivity (Baskakov and Breydo, 2007;
Collinge and Clarke, 2007; Prusiner, 1998).
Structure 22, 1–12, April 8, 20
Nevertheless, several lines of evidence
have suggested common structural fea-
tures shared by PrPSc and in vitro-con-
verted PrP. For example, the 3D EM
image, on which the BH3 model was based, was obtained
from an in vitro generated PrP fibril. However, the fibril shared
a similar morphology with the PrPSc fibril extracted from the
brains of prion-infected animals (Sim and Caughey, 2009), and
the antibody against the in vitro fibril component was shown to
be reactive with PrPSc from human and mouse prion-infected
brains (Makarava et al., 2010).
EPR spectra endorsing the IS model were also obtained from
PrP fibrils prepared in vitro (Cobb et al., 2007; Lu et al., 2007). It
was shown that the PrPSc extracted from the brain of a prion-
infected animal demonstrated an H/D exchange profile consis-
tent with the IS model (Smirnovas et al., 2011). Additionally,
recent studies also demonstrated that some of the PrPSc-spe-
cific antibodies were reactive with PrPs converted in vitro (Bia-
sini et al., 2008), and highly infective PrPSc can be generated
in vitro (Diaz-Espinoza and Soto, 2010; Shikiya and Bartz,
2011). In this study, therefore, the five major proposed models
and two EM images were used for model evaluation by
assuming that they shared structural features with infective
PrPSc.
At this point, however, the conformational details of the PrPSc
working models are mostly theoretical, except that the IS model
is based on rigid geometric data, distance constraints between
residues, obtained by EPR spectroscopy of human PrP (Cobb
et al., 2007; Lu et al., 2007). The IS model assumed a drastic
structural change that involved complete refolding of a helices
and their conversion into in-register b sheets. The absence of a
helices in PrPSc was recently proposed based on the reassign-
ment of an infrared absorbance at 1,650 cm�1, which was inter-
preted to assign 10%–17% of the a helix in PrPSc in previous
studies, to a certain nonprotein component (Baron et al.,
2011). As mentioned above, the H/D exchange mass spectros-
copy experiments detected strong protection of the C-terminal
regions in mouse PrPSc (residues W144–A223) extracted from
the brain of a prion-infected animal, which could be explained
only by a tightly packed b sheet structure (Lu et al., 2007; Smir-
novas et al., 2011). The result of protease restriction of mouse
GPI-less PrPSc, by indicating strong protection in the C-terminal
region of core PrP, was consistent with this model (Sajnani et al.,
2008; Vazquez-Fernandez et al., 2012). The in-register-stacking
model provides common structural bases for PrPSc and several
amyloid proteins (Luhrs et al., 2005; Sawaya et al., 2007). There-
fore, it is quite interesting to investigate why the mutation data
indicated a preference for the b-helical models rather than the
in-register-stacking b sheet model.
14 ª2014 Elsevier Ltd All rights reserved 7
Figure 5. Evaluating Mutation Data on PrPC
Structure
The fraction (A) and position (B) of consistent (blue),
ambiguous (white), and inconsistent (red) sites on
PrPC. The inconsistent sites the on a helix and other
residues are shown as light and dark red, respec-
tively, in (A). The a helices a1-a3, b strands b1-b2,
and N and C termini (N and C) are indicated in (B).
N-glycans (gray wire), and disulfide bonds (yellow
ball-and-stick) are also shown in (B).
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
The mutation data appeared to be incompatible with the IS
model on two points. First, the data suggested that stabilization
of the C-terminal a helices promoted conversion because 71%
of inconsistent sites were found on the a helices of PrPC (Fig-
ure 5). Second andmore significantly, the sites that largely affect
conversion efficiency were localized to the N-terminal region of
core PrP (residues �V160) and involved in b sheet formation in
the b-helix-based models (Figure 3), while the corresponding re-
gion was assumed to be disordered in the IS model (Smirnovas
et al., 2011).
Several lines of evidence have accumulated for involvement of
the N-terminal regions of core PrP in the formation of PrPSc.
Mouse-hamster chimeric PrP revealed that residues N99–P104
were required for PrPSc conversion (Hara et al., 2012). The
motif-grafted antibodies against peptides W88–V111 and
R135–P157 were shown to prevent mouse PrPSc conversion,
and those against K23–T33, W98–H110, and P136–N158 pre-
vented mouse PrPC to PrPSc interaction (Moroncini et al., 2004;
Solforosi et al., 2007). Similarly, peptide array experiments de-
tected that sheep PrP residues W102–K104, P140–Y148,
Y152–R154, Y165–R167, N177–V179, and S225–Y228 (W98–
K100, P136–W144, Y148–R150, Y161–R163, N173–V175, and
S221–Y224 in mouse PrP) were the self-interface of PrPC to
PrPSc (Rigter et al., 2007, 2009). A supporting role for the N-ter-
minal region of core PrP, for instance, once involved in PrPSc
conversion but totally disordered in the mature fibril, would be
possible but highly improbable. Therefore, a PrPSc working
model that explicitly involves the N-terminal region of core PrP
is required.
Suggestions for Multiple Conformations for PrPSc
Both direct and indirect clues for multiple conformations of PrPSc
have been accumulating. The amyloid fibril structure of the
fungus prion protein HET-s was determined using solid-state
nuclear magnetic resonance (Wasmer et al., 2008). Unlike
the previously determined amyloid fibril structure, the core
region of HET-s assumed a b-solenoid fold, which resembled a
b helix to a certain extent, and used intersubunit stacking of
b olenoids in the fibril formation. Although HET-s is not a homo-
8 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved
log of mammalian PrP, the experimental
structure demonstrated that the in-regis-
ter-stacking b sheet like the IS model is
not the sole architecture of amyloid fibrils.
Recently, cryo-EM observation revealed
that HET-s forms a triple helix fibril, which
resembles the BH1 or BH2 models (Mizuno
et al., 2011).
The PrPSc strains isolated from infected mouse brains re-
vealed significant differences in infrared absorbance and prote-
ase digestion patterns (Caughey et al., 1998). In addition, a large
variation in fibril morphology of PrPSc strains was detected in EM
analyses (Sim and Caughey, 2009). To date, two different types
of EM images have been observed for PrP fibrils. The 2D image
showed 3- (or 6-) fold symmetric fibrils, where N- and C-terminal
regions resided inside and outside the fibril, respectively (Wille
et al., 2002). The 3D reconstructed image, however, revealed a
2-fold symmetric ladder-shaped structure of a mouse PrP fibril
(Tattum et al., 2006), which resembled some of the twisted fila-
ment EM images observed for the mouse PrPSc (Sim and
Caughey, 2009).
These images are not apparently consistent with each other.
The current PrPSc working models were originally devised to
be consistent with one of these images; the BH1, BH2, and SP
models were consistent with the former, and the BH3 with the
latter. The IS model is rather close to the 3D image, although a
periodic structure with�60 A intervals in the 3D EM image would
not be explained by simple in-register stacking of b strands
(Figure 1F).
Thus, the lines of evidence suggest the requirement of multiple
PrPSc models to explain the observed polymorphism.
Reconciling b Helix Models to Experimental DataBecause PrPSc conversion occurred in scrapie-infected cells in
the mutation experiment of this study, the results might reflect
an in vivo structure of PrPSc. Thus, reconciliation of the current
b-helix-based model was also attempted in this study to find
an alternative model of the PrPSc structure. It was intended to re-
move the questionable features of the current models, which
were observed in the modeling processes or would be raised
in light of recent experimental data, as far as possible.
For example, although mouse PrPC has been shown to
convert to PrPSc with core PrP (L108–L243) only, a conversion
also took place for the intrinsic cellular form of PrP, which con-
sists of highly disordered N-terminal residues of full-length PrP
(K23–K109), N-glycans, and a glycosylphosphatidylinositol
(GPI) moiety attached to the C terminus (Chesebro et al., 2005;
Figure 6. BH4 Model
(A) Scheme of the BH4 model. The b helix and a helix are shown in blue triangles and red cylinders, respectively. Interfaces 1–4 are indicated by magenta dotted
lines. Interfaces 3 and 4 exist between subunits that are related by 3- and 2-fold axes in triple helix and ladder fibrils, respectively.
(B and C) The sites (B) implicated for PrPC to PrPSc interface (blue), and (C) detected to be highly influential in the mutation experiment (orange) are indicated on
subunit A of the BH4 model. The cross between subunits A and B in (C) indicates a potential domain (a2-a3 helix bundle) swap point.
See also Figure S6.
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
Rogers et al., 1993). In addition, the C178–C213 disulfide bond is
known to be important for PrP conversion (Herrmann and
Caughey, 1998). Therefore, a plausible model should have
enough space for these molecular peripherals. In this regard,
the BH1 and SP models were questionable because they did
not provide sufficient space for the N-terminal residues of full-
length PrP to escape. In addition, the disulfide bond could not
be formed in the BH3 model (Table 1).
On the other hand, the detected core structure of theN-glycan
of PrP is considerably large (Stimson et al., 1999), and its packing
in a molecular model is not a minor problem. For the BH1 model,
spaces left for N-glycans were not sufficient due to the close
packing of a2-a3 helix bundles. In the IS model, although the
N-glycans point toward the outside of the complex, the in-regis-
ter conformation inevitably placed both asparagines in a line,
making it is quite difficult to avoid an atomic crash between
N-glycans if PrP is highly modified with N-glycans (Figure S2E).
The BH2 model was selected as the starting model for refine-
ment because it seemed to be free from these problems
(Table 1). Taking these problems in account, the N-terminal b
helix of core PrP of the model was thought to comprise the rails
of the 3D EM image because this part is essential for fibril exten-
sion (Figure 5A). The C-terminal a2-a3 helix bundle was put into
the steps. The helix bundles from the subunits on different rails
were arranged so that the experimentally detected self-interface
residues were used for the interactions, and so that the helix
bundles could be swapped between subunits (Figure 5C). This
is because dissociation between the N-terminal b1-a1-b2 and
C-terminal a2-a3 domains of core PrP was detected in the initial-
Struct
izing step for sheep PrPSc formation, implying that a domain-
swapped PrP dimer is a building block of PrPSc (Eghiaian et al.,
2007; Luhrs et al., 2005).
The persistence of PrPC a helices in PrPSc might be critical
because PrPC with PrPSc interaction was shown to be required
for a formation of intermediate structure in the initial step of con-
version (Rigter et al., 2007, 2009; Solforosi et al., 2007). It implied
that at least part of the interface in PrPSc consists of the native
PrPC structure. In the BH4model, the C-terminal a helices, which
contained the region implied for PrPC with PrPSc interaction, take
part in the intersubunit interactions. Therefore, this model can
explain the introduction of PrPC to PrPSc through this region.
Fitting a model to both 2D and 3D EM images was another
intricate problem, for which none of the current models was
appropriate. The analogy of crystal growth has been often
mentioned for PrPSc formation: inoculation of a small fragment
of a PrPSc fibril as a seeding nucleus initiates conversion of
PrPC by causing crystal-like growth (Sasaki et al., 2008). There-
fore, the apparently incompatible features between the EM
images were interpreted to be due to a difference in crystal
systems. The BH4 model could explain the ladder-shaped 3D
EM image if interface 4 was used. In the ladder structure, the
sidewall of the b helix was open for interfacing, so that a trimer
that resembled the 2D EM image could be formed readily in
the same manner as the BH1 and BH2 models, when interface
3 was used. Thus, switching between interfaces was assumed
to explain different EM images, although this analogy of
protein crystallization might not explain entire of the prion
polymorphism.
ure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved 9
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
As the result, the BH4 model ranked highest among the cur-
rent models in evaluations against the mutation data by using
most of the suggested self-interface residues and the highly
influential sites detected in the mutation experiment for contacts
with neighboring subunits (Figures 6C and 6D). It could also
explain both 2D and 3D EM images, except that the model thick-
ness appeared to exceed that of the 3D EM image if the C-termi-
nal helices were fully intact (Figure S6).
The BH4 model might partially explain the structural polymor-
phism of PrPSc with preferential usage of interfaces. The confor-
mational polymorphism among prion strains has been well
known (Caughey et al., 1998; Safar et al., 1998). However, the
structural differences at atomic detail and the direct cause of
the polymorphism are not understood. A similar structural poly-
morphism depending on pH difference has been observed in
HET-s amyloid fibril (Mizuno et al., 2011). Because the stability
of each interface might differ depending on environmental con-
ditions, such as pH or amino acid sequence variation among
organisms, the subunits of BH4 model would result in different
fibril structures under different conditions, because a protein
can be crystallized into various crystal systems depending
on crystallization conditions. Therefore, the BH4 model might
serve as the 3D model of PrPSc that can partially explain the
conformational polymorphism. Because the PrPSc conversion
was executed in ScN2a cell, the evaluated compatibility of the
models should be that for RML scrapie prion, and it still remained
a possibility that a similar comprehensive mutation study would
provide a different result for other strains. Consequently, appli-
cation of the introduced method to other prion strains would
be quite interesting for further study.
ConclusionsThe comprehensivemutation study of mouse PrP highlighted the
involvement of the N-terminal region of core PrP in structure con-
version, which was compatible with the b helix models. The
revised BH4 model could explain the results of mutation exper-
iments of this study, PrPC to PrPSc interactions, and PrPSc fibril
polymorphism better than the previous models. The method
introduced in this article is rather elaborate because it requires
the existence of comprehensive mutation data and model struc-
tures. Nevertheless, this method would be generally applicable
for these cases in which current structural biology methods are
hardly applicable.
EXPERIMENTAL PROCEDURES
Preparation of PrP Point Mutants and PrPSc Conversion Efficiency
Assay
The open reading frame of the mouse PrP gene was cloned into a pBluescript
plasmid (Stratagene), and 19 single amino acid substitutions at every amino
acid residue from codon 102 to 240 were introduced into the mouse PrP
gene by site-directed PCR mutagenesis (Stratagene) as described (Ikeda
et al., 2008). To detect specifically the transfected mutant PrP against endog-
enous mouse PrP, the epitope for the monoclonal antibody 3F4 (L108M and
V111M) or that of T41 (V214I and Q218E) was also introduced into the mutant
PrP gene (Kascsak et al., 1987; Muramoto et al., 2000). These 3F4- or T41-
epitope-tagged mutant PrP fragments were cloned into the expression vector
pSPOX (Scott et al., 1992). A scrapie-infected mouse neuroblastoma cell line
(ScN2a) was kindly provided by Dr. Stanley B. Prusiner (Butler et al., 1988).
ScN2a cells were transiently transfected with the plasmid constructs using
the FuGENE6 transfection reagent (Roche Diagnostics) and harvested 48 hr
10 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights rese
after transfection. For detection of the total cell-associated PrP, the cell lysates
were digested with 300 U of PNGase F (New England Biolabs) at 37�C for 2 hr.
For detection of the proteinase K (PK)-resistant core of PrPSc, the cell lysates
were digested with 20 mg/ml PK at 37�C for 30 min and ultracentrifuged at
100,0003 g at 20�C for 1 hr, and then the pellets were resuspended in sample
buffer. The samples were subjected to 13% SDS-PAGE and western blotting
as described (Asano et al., 2006; Ikeda et al., 2008). The signal intensities of
western blotting were quantified, and the amounts of PK-resistant PrP mole-
cules were corrected by expression levels of the PrP molecule without PK
digestion to normalize them against transfection and expression efficiencies.
The conversion efficiencies were calculated as the relative amount of the
mutant PK-resistant PrP molecules to that of the transfected wild-type (Fig-
ure 2). The glycosylation on every mutant was also examined by PNGase F
(New England Biolabs) digestion at 37�C for 2 hr.
The experiments for the mutants at 19 residue sites, namely, A115, A117,
N180, T182, N196, T198, R228, R229, S230, S231, S232, T233, V234, L235,
F236, S237, S238, P239, and P240, were repeated three times, and those at
12 sites, namely, A112, G113, A114, V120, G125, G126, Y127, L129, S131,
W144, H176, and D177 were executed twice from transfection procedure, to
examine reproducibility. For other sites, the data were collected from a single
transfection experiment. Reproducibility of conversion efficiency evaluation
was judged to be fine from the data of the triplicate or duplicate experiments.
The example of reproducibility is shown for the A115 site in Figure 2C.
A constancy of conversion efficiency against total expression level of PrP
was examined by expressing varying amounts of wild-type PrP with the
3F4- or T41-epitope-tags in the ScN2a cells (Figures 2D and 2E).
PrPSc Working Model Construction
Five of the PrPSc working models were constructed mostly from scratch by
referring to the descriptions in the relevant literature (Figures 1 and S5; Cobb
et al., 2007; DeMarco and Daggett, 2004; Govaerts et al., 2004; Langedijk
et al., 2006; Tattum et al., 2006). MOE (Ryoka Systems) and COOT/REFMAC5
applications were used for the manual construction and refinement of the
models (CCP4, 1994; Emsley et al., 2010; Murshudov et al., 1997). The core
of the N-glycan moieties (Man a1-6[(GlcNAc b1-4)][(Man a1-3)]Man b1-
4GlcNAc b1-4-[Fuc a1-6]GlcNAc-Asn) was modeled at N180 and N196, and
a disulfide bond was made between C178 and C213 whenever possible.
The models were evaluated by using Verify3D and PROCHECK (Figure S1
and Table S1). The construction procedures for each model are detailed in
the Supplemental Experimental Procedures. Molecular graphics presenta-
tions were prepared using CHIMERA (Pettersen et al., 2004).
Evaluation of Point Mutation Effects on 3D Models
The mutation data (Figure 3 and Table S2) were used to evaluate the PrPSc
working models. To avoid possible biases from aberrant samples, the data
of the mutant PrPs were discarded when they showed very low conversion
efficiency or a defect in glycosylation. The mutant PrPs with very low conver-
sion efficiency were excluded because a defect in protein folding was sus-
pected. The 1,331 (47.9%) mutant and wild-type sites out of a total 2,780
(20 amino acids for 139 sites) were used for the model evaluation (Figure 3).
The confirmation of proper glycosylation on mutant PrPs was used to ensure
the processing of mutant proteins through Golgi apparatus.
The evaluation wasmade in both coarse and fine scopes. In the coarse eval-
uation, the amino acid residues in the models were assigned to surface, inte-
rior, or interface categories. If the average conversion efficiency of a site
ranked within the top 50% and the categories of the site differed between
PrPC and PrPSc working models or ranked within the bottom 30% and stayed
in same category, the site was thought to be consistent with the corresponding
model. If a site ranked in the top 50% and the site categories did not differ or
ranked in the bottom 30% and the categories were different, it was considered
inconsistent. The other sites were assigned as ambivalent (Table 1).
In the fine evaluation, threading profiles were calculated for each model and
used to correlate the difference in fitness (difference in virtual free energy) of
amino acids to the changes in PrPSc conversion efficiency on point mutation
(Ota et al., 2001). The tendency toward stabilization/destabilization of a
point mutation was evaluated as DG = {score of substituted amino acid in
model}� {score of wild-type amino acid in model} for PrPC and PrPSc working
models. Then, the effective correlation coefficient rpi = �rilog pi between DG
rved
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
and the logarithm of conversion efficiency logDT were evaluated, where pi and
ri are the p value and the correlation coefficient of site i, respectively. Sites with
rpi < �0.1 (promoting conversion by stabilizing PrPSc) or rpi > 0.1 (preventing
conversion by stabilizing PrPSc) were regarded as consistent or inconsistent
with the corresponding PrPSc working model, respectively. For PrPC experi-
mental structures, a site was consistent or inconsistent when rpi > 0.1 (promot-
ing conversion by destabilizing PrPC) or rpi < �0.1 (preventing conversion by
destabilizing PrPC), respectively (Table 1). The average of the correlation coef-
ficients over the sites, Si rpi/n, was also evaluated for each PrPSc working
model, where n was the number of counted sites in each model.
Additionally, the models were evaluated by referring to the results of prote-
ase restriction of mouse GPI-less PrPSc (Sajnani et al., 2008; Vazquez-Fernan-
dez et al., 2012). The percentage of exposed protease restriction sites was
obtained for each model. Because disordered residues might be potential re-
striction sites, the percentage of restriction sites, which were exposed on a
model or were not modeled, was also evaluated for each model (Table 1).
The methods are detailed in the Supplemental Experimental Procedures.
SUPPLEMENTAL INFORMATION
Supplemental Information includes Supplemental Experimental Procedures,
six figures, and two tables and can be found with this article online at http://
dx.doi.org/10.1016/j.str.2013.12.019.
AUTHOR CONTRIBUTIONS
T.S. and M.S. designed and executed the computational analyses; A.K., M.A.,
M.H., S.I., K.T., andM.M. performed themutation and assay experiments; K.T,
M.M., and T.K. designed the experiments; and T.S. wrote the manuscript. All
authors commented on the manuscript.
ACKNOWLEDGMENTS
This work was partly supported by research grants from the Institute for Bioin-
formatics Research and Development (BIRD); the Japan Science and Technol-
ogy Agency; and Platform for Drug Design, Informatics, and Structural
Lifescience (PDIS), theMinistry of Education, Culture, Sports, Science & Tech-
nology in Japan.
Received: August 21, 2013
Revised: December 21, 2013
Accepted: December 28, 2013
Published: February 20, 2014
REFERENCES
Asano, M., Mohri, S., Ironside, J.W., Ito, M., Tamaoki, N., and Kitamoto, T.
(2006). vCJD prion acquires altered virulence through trans-species infection.
Biochem. Biophys. Res. Commun. 342, 293–299.
Baron, G.S., Hughson, A.G., Raymond, G.J., Offerdahl, D.K., Barton, K.A.,
Raymond, L.D., Dorward, D.W., and Caughey, B. (2011). Effect of glycans
and the glycophosphatidylinositol anchor on strain dependent conformations
of scrapie prion protein: improved purifications and infrared spectra.
Biochemistry 50, 4479–4490.
Baskakov, I.V., and Breydo, L. (2007). Converting the prion protein: what
makes the protein infectious. Biochim. Biophys. Acta 1772, 692–703.
Biasini, E., Seegulam, M.E., Patti, B.N., Solforosi, L., Medrano, A.Z.,
Christensen, H.M., Senatore, A., Chiesa, R., Williamson, R.A., Harris, D.A.,
et al. (2008). Non-infectious aggregates of the prion protein react with several
PrPSc-directed antibodies. J. Neurochem. 105, 2190–2204.
Butler, D.A., Scott, M.R., Bockman, J.M., Borchelt, D.R., Taraboulos, A.,
Hsiao, K.K., Kingsbury, D.T., and Prusiner, S.B. (1988). Scrapie-infected mu-
rine neuroblastoma cells produce protease-resistant prion proteins. J. Virol.
62, 1558–1564.
Caughey, B.W., Dong, A., Bhat, K.S., Ernst, D., Hayes, S.F., and Caughey,
W.S. (1991). Secondary structure analysis of the scrapie-associated protein
PrP 27-30 in water by infrared spectroscopy. Biochemistry 30, 7672–7680.
Structu
Caughey, B., Raymond, G.J., and Bessen, R.A. (1998). Strain-dependent dif-
ferences in beta-sheet conformations of abnormal prion protein. J. Biol. Chem.
273, 32230–32235.
Caughey, B., Baron, G.S., Chesebro, B., and Jeffrey, M. (2009). Getting a grip
on prions: oligomers, amyloids, and pathological membrane interactions.
Annu. Rev. Biochem. 78, 177–204.
CCP4; Collaborative Computational Project, Number 4 (1994). The CCP4
suite: programs for protein crystallography. Acta Crystallogr. D Biol.
Crystallogr. 50, 760–763.
Chesebro, B., Trifilo, M., Race, R., Meade-White, K., Teng, C., LaCasse, R.,
Raymond, L., Favara, C., Baron, G., Priola, S., et al. (2005). Anchorless prion
protein results in infectious amyloid disease without clinical scrapie. Science
308, 1435–1439.
Cobb, N.J., Sonnichsen, F.D., McHaourab, H., and Surewicz, W.K. (2007).
Molecular architecture of human prion protein amyloid: a parallel, in-register
beta-structure. Proc. Natl. Acad. Sci. USA 104, 18946–18951.
Collinge, J., and Clarke, A.R. (2007). A general model of prion strains and their
pathogenicity. Science 318, 930–936.
DeMarco, M.L., and Daggett, V. (2004). From conversion to aggregation: pro-
tofibril formation of the prion protein. Proc. Natl. Acad. Sci. USA 101, 2293–
2298.
Diaz-Espinoza, R., and Soto, C. (2010). Generation of prions in vitro and the
protein-only hypothesis. Prion 4, 53–59.
Diaz-Espinoza, R., and Soto, C. (2012). High-resolution structure of infectious
prion protein: the final frontier. Nat. Struct. Mol. Biol. 19, 370–377.
Eghiaian, F., Daubenfeld, T., Quenet, Y., van Audenhaege, M., Bouin, A.P., van
der Rest, G., Grosclaude, J., and Rezaei, H. (2007). Diversity in prion protein
oligomerization pathways results from domain expansion as revealed by
hydrogen/deuterium exchange and disulfide linkage. Proc. Natl. Acad. Sci.
USA 104, 7414–7419.
Emsley, P., Lohkamp, B., Scott, W.G., and Cowtan, K. (2010). Features and
development of Coot. Acta Crystallogr. D Biol. Crystallogr. 66, 486–501.
Fersht, A.R., and Daggett, V. (2002). Protein folding and unfolding at atomic
resolution. Cell 108, 573–582.
Gasset, M., Baldwin, M.A., Fletterick, R.J., and Prusiner, S.B. (1993).
Perturbation of the secondary structure of the scrapie prion protein under con-
ditions that alter infectivity. Proc. Natl. Acad. Sci. USA 90, 1–5.
Govaerts, C., Wille, H., Prusiner, S.B., and Cohen, F.E. (2004). Evidence for
assembly of prions with left-handed beta-helices into trimers. Proc. Natl.
Acad. Sci. USA 101, 8342–8347.
Hara, H., Okemoto-Nakamura, Y., Shinkai-Ouchi, F., Hanada, K., Yamakawa,
Y., and Hagiwara, K. (2012). Mouse prion protein (PrP) segment 100 to 104 reg-
ulates conversion of PrP(C) to PrP(Sc) in prion-infected neuroblastoma cells.
J. Virol. 86, 5626–5636.
Herrmann, L.M., and Caughey, B. (1998). The importance of the disulfide bond
in prion protein conversion. Neuroreport 9, 2457–2461.
Hiraga, C., Kobayashi, A., and Kitamoto, T. (2009). The number of octapeptide
repeat affects the expression and conversion of prion protein. Biochem.
Biophys. Res. Commun. 382, 715–719.
Hizume, M., Kobayashi, A., Mizusawa, H., and Kitamoto, T. (2010). Amino acid
conditions near the GPI anchor attachment site of prion protein for the conver-
sion and the GPI anchoring. Biochem. Biophys. Res. Commun. 391, 1681–
1686.
Ikeda, S., Kobayashi, A., and Kitamoto, T. (2008). Thr but Asn of the N-glyco-
sylation sites of PrP is indispensable for its misfolding. Biochem. Biophys. Res.
Commun. 369, 1195–1198.
Kaneko, K., Zulianello, L., Scott, M., Cooper, C.M., Wallace, A.C., James, T.L.,
Cohen, F.E., and Prusiner, S.B. (1997). Evidence for protein X binding to a
discontinuous epitope on the cellular prion protein during scrapie prion prop-
agation. Proc. Natl. Acad. Sci. USA 94, 10069–10074.
Kascsak, R.J., Rubenstein, R., Merz, P.A., Tonna-DeMasi, M., Fersko, R.,
Carp, R.I., Wisniewski, H.M., and Diringer, H. (1987). Mouse polyclonal and
monoclonal antibody to scrapie-associated fibril proteins. J. Virol. 61, 3688–
3693.
re 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights reserved 11
Structure
Prion Model Evaluation
Please cite this article in press as: Shirai et al., Evaluating Prion Models Based on Comprehensive Mutation Data of Mouse PrP, Structure (2014), http://dx.doi.org/10.1016/j.str.2013.12.019
Kunes, K.C., Clark, S.C., Cox, D.L., and Singh, R.R. (2008). Left handed beta
helix models for mammalian prion fibrils. Prion 2, 81–90.
Langedijk, J.P., Fuentes, G., Boshuizen, R., and Bonvin, A.M. (2006). Two-rung
model of a left-handed beta-helix for prions explains species barrier and strain
variation in transmissible spongiform encephalopathies. J. Mol. Biol. 360,
907–920.
Lu, X., Wintrode, P.L., and Surewicz, W.K. (2007). Beta-sheet core of human
prion protein amyloid fibrils as determined by hydrogen/deuterium exchange.
Proc. Natl. Acad. Sci. USA 104, 1510–1515.
Luhrs, T., Ritter, C., Adrian, M., Riek-Loher, D., Bohrmann, B., Dobeli, H.,
Schubert, D., and Riek, R. (2005). 3D structure of Alzheimer’s amyloid-
beta(1-42) fibrils. Proc. Natl. Acad. Sci. USA 102, 17342–17347.
Luthy, R., Bowie, J.U., and Eisenberg, D. (1992). Assessment of protein
models with three-dimensional profiles. Nature 356, 83–85.
Makarava, N., Kovacs, G.G., Bocharova, O., Savtchenko, R., Alexeeva, I.,
Budka, H., Rohwer, R.G., and Baskakov, I.V. (2010). Recombinant prion pro-
tein induces a new transmissible prion disease in wild-type animals. Acta
Neuropathol. 119, 177–187.
Mizuno, N., Baxa, U., and Steven, A.C. (2011). Structural dependence of HET-
s amyloid fibril infectivity assessed by cryoelectron microscopy. Proc. Natl.
Acad. Sci. USA 108, 3252–3257.
Moroncini, G., Kanu, N., Solforosi, L., Abalos, G., Telling, G.C., Head, M.,
Ironside, J., Brockes, J.P., Burton, D.R., and Williamson, R.A. (2004). Motif-
grafted antibodies containing the replicative interface of cellular PrP are spe-
cific for PrPSc. Proc. Natl. Acad. Sci. USA 101, 10404–10409.
Muramoto, T., Tanaka, T., Kitamoto, N., Sano, C., Hayashi, Y., Kutomi, T.,
Yutani, C., and Kitamoto, T. (2000). Analyses of Gerstmann-Straussler syn-
drome with 102Leu219Lys using monoclonal antibodies that specifically
detect human prion protein with 219Glu. Neurosci. Lett. 288, 179–182.
Murshudov, G.N., Vagin, A.A., and Dodson, E.J. (1997). Refinement of macro-
molecular structures by the maximum-likelihood method. Acta Crystallogr.
D Biol. Crystallogr. 53, 240–255.
Nguyen, J.T., Inouye, H., Baldwin, M.A., Fletterick, R.J., Cohen, F.E., Prusiner,
S.B., and Kirschner, D.A. (1995). X-ray diffraction of scrapie prion rods and PrP
peptides. J. Mol. Biol. 252, 412–422.
Ota, M., Isogai, Y., and Nishikawa, K. (2001). Knowledge-based potential
defined for a rotamer library to design protein sequences. Protein Eng. 14,
557–564.
Perrier, V., Kaneko, K., Safar, J., Vergara, J., Tremblay, P., DeArmond, S.J.,
Cohen, F.E., Prusiner, S.B., and Wallace, A.C. (2002). Dominant-negative inhi-
bition of prion replication in transgenic mice. Proc. Natl. Acad. Sci. USA 99,
13079–13084.
Pettersen, E.F., Goddard, T.D., Huang, C.C., Couch, G.S., Greenblatt, D.M.,
Meng, E.C., and Ferrin, T.E. (2004). UCSF Chimera—a visualization system
for exploratory research and analysis. J. Comput. Chem. 25, 1605–1612.
Prusiner, S.B. (1991). Molecular biology of prion diseases. Science 252, 1515–
1522.
Prusiner, S.B. (1998). Prions. Proc. Natl. Acad. Sci. USA 95, 13363–13383.
Race, R.E., Caughey, B., Graham, K., Ernst, D., and Chesebro, B. (1988).
Analyses of frequency of infection, specific infectivity, and prion protein
biosynthesis in scrapie-infected neuroblastoma cell clones. J. Virol. 62,
2845–2849.
Riek, R., Hornemann, S., Wider, G., Billeter, M., Glockshuber, R., and
Wuthrich, K. (1996). NMR structure of the mouse prion protein domain
PrP(121-231). Nature 382, 180–182.
Rigter, A., Langeveld, J.P., Timmers-Parohi, D., Jacobs, J.G., Moonen, P.L.,
and Bossers, A. (2007). Mapping of possible prion protein self-interaction
domains using peptide arrays. BMC Biochem. 8, 6.
Rigter, A., Priem, J., Timmers-Parohi, D., Langeveld, J.P., van Zijderveld, F.G.,
and Bossers, A. (2009). Prion protein self-peptidesmodulate prion interactions
and conversion. BMC Biochem. 10, 29.
12 Structure 22, 1–12, April 8, 2014 ª2014 Elsevier Ltd All rights rese
Rogers, M., Yehiely, F., Scott, M., and Prusiner, S.B. (1993). Conversion of
truncated and elongated prion proteins into the scrapie isoform in cultured
cells. Proc. Natl. Acad. Sci. USA 90, 3182–3186.
Safar, J., Wille, H., Itri, V., Groth, D., Serban, H., Torchia, M., Cohen, F.E., and
Prusiner, S.B. (1998). Eight prion strains have PrP(Sc) molecules with different
conformations. Nat. Med. 4, 1157–1165.
Sajnani, G., Pastrana, M.A., Dynin, I., Onisko, B., and Requena, J.R. (2008).
Scrapie prion protein structural constraints obtained by limited proteolysis
and mass spectrometry. J. Mol. Biol. 382, 88–98.
Sasaki, K., Gaikwad, J., Hashiguchi, S., Kubota, T., Sugimura, K., Kremer, W.,
Kalbitzer, H.R., and Akasaka, K. (2008). Reversible monomer-oligomer transi-
tion in human prion protein. Prion 2, 118–122.
Sawaya, M.R., Sambashivan, S., Nelson, R., Ivanova, M.I., Sievers, S.A.,
Apostol, M.I., Thompson, M.J., Balbirnie, M., Wiltzius, J.J., McFarlane, H.T.,
et al. (2007). Atomic structures of amyloid cross-beta spines reveal varied ste-
ric zippers. Nature 447, 453–457.
Scott, M.R., Kohler, R., Foster, D., and Prusiner, S.B. (1992). Chimeric prion
protein expression in cultured cells and transgenic mice. Protein Sci. 1,
986–997.
Shikiya, R.A., and Bartz, J.C. (2011). In vitro generation of high-titer prions.
J. Virol. 85, 13439–13442.
Sidhu, S.S., and Kossiakoff, A.A. (2007). Exploring and designing protein func-
tion with restricted diversity. Curr. Opin. Chem. Biol. 11, 347–354.
Sim, V.L., and Caughey, B. (2009). Ultrastructures and strain comparison of
under-glycosylated scrapie prion fibrils. Neurobiol. Aging 30, 2031–2042.
Smirnovas, V., Baron, G.S., Offerdahl, D.K., Raymond, G.J., Caughey, B., and
Surewicz, W.K. (2011). Structural organization of brain-derived mammalian
prions examined by hydrogen-deuterium exchange. Nat. Struct. Mol. Biol.
18, 504–506.
Solforosi, L., Bellon, A., Schaller, M., Cruite, J.T., Abalos, G.C., and
Williamson, R.A. (2007). Toward molecular dissection of PrPC-PrPSc interac-
tions. J. Biol. Chem. 282, 7465–7471.
Spevacek, A.R., Evans, E.G., Miller, J.L., Meyer, H.C., Pelton, J.G., and
Millhauser, G.L. (2013). Zinc drives a tertiary fold in the prion protein with famil-
ial disease mutation sites at the interface. Structure 21, 236–246.
Stimson, E., Hope, J., Chong, A., and Burlingame, A.L. (1999). Site-specific
characterization of the N-linked glycans of murine prion protein by high-perfor-
mance liquid chromatography/electrospraymass spectrometry and exoglyco-
sidase digestions. Biochemistry 38, 4885–4895.
Stohr, J., Weinmann, N.,Wille, H., Kaimann, T., Nagel-Steger, L., Birkmann, E.,
Panza, G., Prusiner, S.B., Eigen, M., and Riesner, D. (2008). Mechanisms of
prion protein assembly into amyloid. Proc. Natl. Acad. Sci. USA 105, 2409–
2414.
Tattum, M.H., Cohen-Krausz, S., Thumanu, K., Wharton, C.W., Khalili-Shirazi,
A., Jackson, G.S., Orlova, E.V., Collinge, J., Clarke, A.R., and Saibil, H.R.
(2006). Elongated oligomers assemble into mammalian PrP amyloid fibrils.
J. Mol. Biol. 357, 975–985.
Vazquez-Fernandez, E., Alonso, J., Pastrana, M.A., Ramos, A., Stitz, L., Vidal,
E., Dynin, I., Petsch, B., Silva, C.J., and Requena, J.R. (2012). Structural orga-
nization of mammalian prions as probed by limited proteolysis. PLoS ONE 7,
e50111.
Wasmer, C., Lange, A., VanMelckebeke, H., Siemer, A.B., Riek, R., andMeier,
B.H. (2008). Amyloid fibrils of the HET-s(218-289) prion form a beta solenoid
with a triangular hydrophobic core. Science 319, 1523–1526.
Wille, H., Michelitsch, M.D., Guenebaut, V., Supattapone, S., Serban, A.,
Cohen, F.E., Agard, D.A., and Prusiner, S.B. (2002). Structural studies of the
scrapie prion protein by electron crystallography. Proc. Natl. Acad. Sci. USA
99, 3563–3568.
Wille, H., Bian, W., McDonald, M., Kendall, A., Colby, D.W., Bloch, L., Ollesch,
J., Borovinskiy, A.L., Cohen, F.E., Prusiner, S.B., and Stubbs, G. (2009).
Natural and synthetic prion structure from X-ray fiber diffraction. Proc. Natl.
Acad. Sci. USA 106, 16990–16995.
rved