Consider a Spherical Mad Cow: Physical
modeling of amyloid diseases
D.L. Cox, Physics, UC Davis
R.R.P. Singh, Physics, UC Davis
K.Kunes, S. Dai, J. Romnes, N.R. Hayre, C. Trevisan, Physics, UC
Davis
A. Slepoy, Sandia Labs
R. Kulkarni, Physics, Virginia Polytechnic University
D. Mobley, Pharmaceutical Chemistry, UCSF
F. Pazmandi, Sidney Austin LLP (Patent Law, Intellectual Property)
S. Yang, U. Chicago Med z
S. Clark, Oregon State
E. Olson, Central College Iowa
H. Levine, J. Onuchic, Center for Theoretial Biological Physics,
UCSD
Support: NIH (Seed award from regional Alzheimer’s center), NSF
(NEAT IGERT, CTBP, ICAM), US Army CDMRC
Outline
I. What is ``amyloid matter’’?
II. Biophysics vs. Biological Physics
III. The Central Dogma and Protein Folding
IV. Amyloids: where physics can clearly make inroads
-``Precision biology’’ in Huntington’s and prion diseases
V. Prions: Why they are interesting
VI. Prions: Intrinsically slow?
VII.Prions: Molecular models
VIII.Recent developments p53 (Cancer gene) and AIDS
What we are talking about is novel
materials…..
Plaques: this is your
mind on amyloid…
Alzheimer’s
Parkinson’s
Huntington’s
Kuru (prion)
Fatality is correlated with
plaques. Plaques are bundles
of fibrils
(Pictures: Feaney lab, Harvard)
Model cross section
Ab42 Fibrils
(H. Lashuel)
Diameter: ~10 nm
Fibrils have been
used to template
nanowires!
Fibrils: Protein
Nanotubes/nanofibers!
(Model: H. Saibil; see
also Perutz)
Amyloid diseases: 20 + known; Not known—
mechanism for cell death, toxicity… BUT, in at
least two cases, ``precision biology’’…
• Most
proteins involved
are of unknown function
• Incidence rates for
Huntington’s and Prion
diseases are quite
reproducible, in a way
I will define more later
• Alone among these
diseases, prion diseases
can be infectious.
Recent developments: amyloid in cancer,
AIDS?
CANCER:
• Two separate regions of the p53 protein, for which mutations play a
role in 50% of known cancers undergo amyloid aggregation
• In aggregate form, the p53 cannot carry out programmed cell death or cell
cycle reset functions to arrest tumors.
AIDS:
•
•
•
Naturally forming fragments of the HPAP protein found in semen bind
strongly to the HIV virus.
Evidence suggests that this enhances infectivity of the virus by 10x or
more in lab animal studies.
May explain greater susceptibility of women to infection by men
Biophysics vs. Biological Physics
• Tradition: biophysics = importation of techniques from
physical sciences to study of biological problems –
frequently a one way flow of both ideas and people.
• Example: first physicist in UC Davis Biophysics Grad
Group (in existence since 1961) = me!
Biological Physics
• ``Ask not what physics can do for biology, but what biology can do
for physics’’ S. Ulam via Hans Frauenfelder
• Study biological problems as interesting problems of the physical
world, and ask interesting questions.
• Practical: two way flow of ideas
Caveats:
• ``Just because you can throw a dog off of a roof to
measure g does NOT mean you are doing biological
physics’’ L. Pelliti
• Just as physicists can study materials with physics ideas
and methods, physicists can study biological systems
with physics ideas and methods
• Prerequisites = humility + courage. We have some
unique techniques and styles of inquiry, but we are not
the only smart people in the world…. BUT don’t be
cowed by jargon or expertise…
• Martin Perl, Nobel Laureate and t-lepton discoverer:
``When you enter a new field, don’t know too much, and
watch out for fast talkers!’’
What can biological physics ask and say
about amyloid diseases,
OR, ``Consider a spherical mad cow…’’
• Are these diseases of aging, or just slow? (Spontaneous prion
disease = ``thermodynamically unlucky). Namely, can we
physically model the onset distributions of disease…
•What are the ``organizing principles’’ of the relevant protein
structures (fibrils, oligomers…..) and how can we constrain
these from data and make falsifiable predictions?
•Do the structures suggest/correlate with toxicity mechanisms?
(time permitting…)
•Can the structures be modified by interacting with other
atoms/molecules? (time permitting….)
•For experiment: frontier nanoscience—how to reliably probe
biological matter at the nanoscale in complex
environments????
First, a brief tour of the central
dogma and proteins
Or, ``Biological Physics, the early
years…’’
A reminder: In some ways,
biological phyiscs is not new!
•
1940s: Erwin Schroedinger wrote What is Life? In which he looked at the
major problems of biology from the perspective of a physicist. Among other
things-he predicted that DNA would turn out to be an aperiodic crystal as it
did-
``The non-physicist cannot be expected to grasp - let alone to appreciate
the relevance of - the difference in `statistical structure' stated in
terms so abstract I have just used. To give the statement life and
colour, let me anticipate …[that].. the most essential part of a living
cell - the chromosome fibre - may suitably be called an aperiodic
crystal. In physics we have dealt hitherto only with periodic crystals.
‘’
•
Lots of physicists were inspired at this time to dive into biological
problems (Perutz, Crick, Delbruck, Monod……)
Central Dogma of Molecular Biology I
(Schroedinger to Watson to Crick to…)
Who ordered
That? Each
Biological amino
Acid comes from
Triplets of the four
Base nucleotides—
G,A,T,C for DNA
G,A,U,C for RNA
Central Dogma II – Elements of
Protein Structure
Proteins are polymers formed from
The 20 amino acids=RESIDUES,
coded for from DNA/RNA. They can,
depending upon the side chain, be polar (but uncharged), charged, or hydrophobic
Most common secondary
structures:
a-Helices – Amide carbonyl
Hydrogen bonds every 3-4 residues
 b-structure—can be parallel or
or antiparallel
Lends to aggregation on edges
No constraint on where bonding
arises along the sequence
Paradox: Typical ensembles of random
heteropolymers  act like a glassy system-• Typical random heteropolymers DON’T fold into compact
shapes…Frustration = unsatisfiable competing interactions arise
from i) putting hydrophobic residues out
• How on earth do we get ``well folded’’ useful biological proteins in
reasonable time scales (milliseconds to seconds)?
• Levinthal Paradox: (c.f. Wikipedia…): ``In 1969 Cyrus Levinthal
noted that, because of the very large number of degrees of freedom
in an unfolded polypeptide chain, the molecule has an astronomical
number of possible conformations. (The estimate 10300 appears in
the original article). If the protein is to attain its corrected folded
configuration by sequentially sampling all the possible
conformations, it would require a time longer than the age of
universe to arrive at its correct native conformation. This is true even
if conformations are sampled at rapid (nanosecond or picosecond)
rates. ‘’
The protein folding problem!!!
Organizing Principle: ``Minimal
Frustration’’ (Wolynes, Onuchic;
Shakhnovich, Dill)
``Frustration’’
• Evolution selects for minimal
frustration of sequence, which
in model simulations leads to a
funneled landscape—
• Formation of partially folded
molten globule largely erases
Levinthal paradox (to within a
few orders of magnitude)
• Correlations of nearbye states
on landscape erases the rest
• Sufficient stabilization of
minimum with respect to mean
Onuchic and Wolynes
``Designed Stability’’ of glassy continuum  fast
folding and well folded proteins
Bryngelson and Wolynes, PNAS 1987
So what about amyloid proteins?
• Most have large
stretches of random
structure or are
completely random!
• Stabilization of
structure apparently
comes with
aggregation.
• Whether in fibril or
oligomer form, there
is cross-beta structure
Fibril axis  ↑ b-sheets
Cross b-Structure - Evidence
from Experiment
• 8 different amyloid diseases, x-ray diffraction off of fibers of fibrils
• fiber axis vertical - peak at +/- 2/C, C ~ 4.8-5 angstroms
• Cross b structure (Pauling, 1951)
• Also see bin circular dichroism, FTIR
(From M. Sunde et al., Journal of Molecular Biology
Volume 273, Issue 3, 31 October 1997, Pages 729-739)
Why is b-sheet prone to aggregation?
unstructured
Unsatisfied
H-bonds Allows
Aggregation
Critical nucleus
elongation
Amyloids can be GOOD!
• Curli amyloids present in bacteria
(E. Coli image from Chapman lab,
U. Mich.) - can participate in
stationary phase survival
mechanism (Bad-may play a role
in infection…)
• Spider silk manufacture has been
argued to be pH switched alpha ->
beta amyloid self assembly
• Amyloids appear to be part of
some insect eggs (silk fibroin)
• Controlled reversible amyloid
scaffolds useful in tissue
engineering (e.g., J. Schneider, U.
Delaware)
• ``Prions’’ in yeast/fungus convey
useful, heritable traits (outside the
genome!!!)
The Beautiful? Organizing
principles for amyloid matter
Organizing Principle 1: Extend minimal
frustration in well ordered proteins by
domain swapping
•
•
•
Link native contacts on one
monomer to corresponding native
contacts on another (champion-D.
Eisenberg)
Example - human cystatin at left
(Janowski et al., Nat Struct Bio
[2001])
Theory: extends minimal
frustration concept to aggregates Yang, Cho, Levy, Cheung, Levine,
Onuchic, Wolynes, PNAS 2004
Organizing Principle 2: Steric Zippers (D.
Eisenberg group, Nature 2007)
•
•
Synthesized lots
of fragments
from amyloidogenic proteins
Fibrils from
combination of
beta sheet
stacking plus
steric zipper
(interlocking of
well packed side
chains)
Organizing principle 3: Amyloid stucture
from monomeric motifs
1T3D
PrPSc
Overlay
model
1T3D stacked in silico
•Appears in multiple bacterial enzymes and insect antifreeze proteins (~11 on PDB)
•``Who ordered that?’’ Unlike a-helix which Pauling predicted prior to discovery
and has local hydrogen bonding (residue j bonds to residue j+3 or j+4) b-helix is
very nonlocal (residue j bonds to j+18)
• b-helix structures easily bond into aggregate (edge-to-edge bonding of monomers)
Example: Huntington’s: one of 10
neurodegenerative diseases associated with
genetically acquired added repeats of glutamine (polyQ
diseases)
•Model with elongation of equilib.
nucleus
•Growth ~ [PQ]M+2 t2 for simple
elongation from critical nucleus of
size M. Here: M=1!!!!
Intrinsically Slow!!!
Extrapolate slow PolyQ lag
kinetics to physiological concentrations (Chen, Ferrone,
Zoghbi & Orr, Ann Rev Neurosci 2000
Wetzel, PNAS 2002)
Theory as a `probe’ of possible sub-observable
structure I
• All atom molecular dynamics
•PolyQ diseases have
critical
(MD
= Newton’s laws for
insert number of ~36approxmate force fields) used
to probe stability of left handed
for PolyQ (folding to
•Aggregation studiesb-helix
of PolyQ
helix not possible!) (CHARMM)
suggest critical nucleus
of 1 (!)
• drms = mean square deviation
monomer (Chen, Ferrone,
of atoms from starting
Wetzel, PNAS 2002)positions.
• Two layered b-helix not stable
•This is ~ the minimal
stable
left ns of simulation
within
several
handed beta helical time
turn (18
residues per turn)• Three layered b-helix is stable
out to ~10 ns
•Is the minimal stable PolyQ a
left handed b-helix? (But: Hear
Rappu…)
Stork et al, Biophysical
Journal 2005
From organizing principle to
disease - one possibility oligomers
• Fibrils are not perfectly correlated with
disease (many have plaques with no AD,
some prion diseases have no plaques).
• Fibrils may be protective (collecting
aggregate away from cells)
• Some oligomers can form pores which
permeate membrane and let in excess
calcium.
Ab
1
a
2
3
4
5
6
7
8
9
10
11
12
SOD1
Ab
c
A4V
Arctic (E22G)
a-Synuclein
a-Synuclein
A30P
A53T
Plaques
n
oligomers
Monomer-dimer-tetramer
Protofibrils
Amyloid Fibrils
Lewy Bodies
What about prions?
What is special about prions?
• Prion: Proteinaceous infectious particle (Prusiner 1980s).
• Along among amyloid diseases: infectious as well as sporadic,
inherited possibilities (PrPSc)
• Numerous experiments (radiation damage,
UV/temperature/protease/denaturant insensitivity….) -> NO
nucleic acids (not a virus or bacteria)
• Bolstered by test-tube synthesis of infectious protein only prions
last year (Baskakov, Prusiner et al, Science 2004)
• Prusiner isolated the PrPc protein as key to the disease—mice
with the gene for PrPc knocked out don’t get sick on innoculation
with infectious prion material (PrPSc and PrPC are identical after
full denaturation-same primary sequence!)
• Examples: Scrapies (sheep), Kuru (humans), Creutzfeldt-Jakob
Disease (CJD) (humans), Mad Cow, Chronic Wasting Disease
(deer and elk)
Structure of normal PrPC
(Wuethrich et al, PNAS 97,
8334 (2000); 97, 8340 [2000])
Proposed structure of PrPSc in
one case (Wille et al, PNAS 99,
3993 (2002); Govaerts et al, PNAS 101,
8342[2004])
• ~90-95% homology in mammals
•Observed in all vertebrates
•Binds copper in divalent form—
sites in humans, mice, six in
cattle
Trimer of left-handed beta
helices gives best model
What is special about the prion
diseases?
•
•
•
Alone amongst ANY disease, prions can be spontaneous, heriditary AND
infectious
Prion diseases represent ``precision biology’’: rates of incidence and dose
incubation distributions are highly reproducible. (~1 in 106 in developed
countries get sporadic CJD worldwide). Suggests a purely physico-chemical
model might capture important features of the disease
Simple models can test important questions about the disease from this
perspective that protein conformation (and potentially aggregate structure)
dictate disease dynamics and properties
Two dimensional aggregation
kinetics?
• Prions are membrane bound
• Can there be interesting differences in models
with 2D aggregation?
Model prions in action
Seed introduced:
slow initial conversion
and aggregation
Wait a while: conversion
and aggregation
accelerates
More precision biology? prion
incubation (Slepoy et al., Phys Rev Lett, 2001;
Mobley et al., Biophys. J. 2003)
Distribution of aggregation times
Seed
Aggregate
Fission
Fission adds (short) doubling and translates->
(BSE best fit)
Time to aggregate to critical size N over
peak time
Role of membrane in toxicity and exponential
growth (fission)
• Cheseboro et al, Science,
2005: Engineer transgenic (Tg)
mice with GPI anchor deleted.
• Evidence is that expressed
PrPC transport to membrane
but are sent off between cells.
• Innoculate mice with a
particular lethal dose of PrPSc
for which wild type (WT) mice
get symptoms at ~150 days.
• Tg mice don’t die or get
symptoms out to 600+ days,
but accumulate infectious prion
material in between cells!
WT
Exponential growth also requires the
membrane! (Cox, Singh, Yang)
• Short time elongation kinetics without fission or
autocatalysis gives t2—agrees remarkably well with
Cheseboro et al!
• We estimate [PrPC]Tg ~ 28 X [PrPC]WT
Proposed b-helical trimer model for
minimal infectious prion particle (UCSF,
GovaertsLoop
et al. PNAS 2004)
Raw EM image of infectious
prion aggregate—note
faceting!
1THJ
Proposed prion trimer has Signal averaged density
(difference) map—note 3
same size as known
Fold symmetry
bacterial trimer (1THJ)
•What holds the UCSF model together? Known bacterial trimers are held
together by intermonomer Zn bonding (1THJ) or massive hydrogen bond
networks (1T3D)
Stabilize Prion Trimer by Domain Swapping (S.
Yang, H. Levine, J. Onuchic, D.L.C. FASEB J, Nov.
2005)
•
•
•
•
•
All atom MD: Amber 8 on trial
structure energy minimized in
loop region
Domain swap: Grab part of one
monomer and bind it to another—
alternative approach to generating
amyloid (Eisenberg et al 1995)
Domain swapping relaxes
stress in loop: Elastic energy
relaxed by ~2 kBT
Domain swapping adds
hydrogen bonds: now structure
plausibly held together – sufficient
to run MD to ~ 1-2 ns for H-bonds
Alternative domain swap: being
explored by S. Cho, Y. Levy, P.
Wolynes
UCSF or BPT model
(beta helical Prion trimer)
Domain swapped
Prion Trimer (DSTP) model
The plot thickens: test tube grown fibrils
(Saibil et al, JMB 06)
For this model, M129
Contacts D178!!!
Kunes, Clark,
Cox, Singh,
to appear in
Prion
C terminal stability good
• Molecular dynamics simulation out to 10 ns of root mean
square deviation of atoms from starting positions
(subtracting center of mass motion) - black, red known
stable beta helices, blue green our models.
Templating - possible
connection to kinetics
• Roughly, extra H bond to link M129 to H177, N178 in FFI
• Hard to link R177 for dogs to this
• For mice the suspicionFor
is that
`S143N’
change relative to
thisthe
model,
M129
humans leads to a different
preferred
thread
Contacts
D178!!!
Conclusions:
•
•
•
•
•
•
•
Amyloid diseases: emergent and generic COLLECTIVE stabilization
of structure
Can be INTRINSICALLY SLOW (Huntington’s, prions..)
Simple areal aggregation model with little biology accounts for
much within the protein only model for proins!
Membrane mediates toxicity (Chesboro et al, Science 2005) AND
exponential growth via fission or oligomeric autocatalysis (Cox,
Singh, Yang)
Domain swapping of the b-helices in the proposed prion trimer may
stabilize the structure Also suggests a model for strains of prions,
possible understanding of GSS mutations
More complex domain swapping and C-terminal beta helix formation
can possibly explain fibrils and more!!!
Amyloid formation is possibly playing a role in other surprising
places (p53 `cancer’ gene, protein in semen which increases
infectivity of AIDS virus)
Some essential issues to explore in
modeling
• Autocatalysis vs. Autocatalytic
Aggregation (cooperative
conversion) Strong arguments
(Eigen) and data legislate
against autocatalysis at the
monomer level; conversion
upon aggregation is more
sensible (and supported by our
work).
• What aggregate structures and
sizes best correspond to
experiment?
• Fission is critical to explain
exponential runaway (Masel,
2000). Do aggregate shapes
and sizes influence this?
• Can infectious and sporadic
time scales be reconciled in
the models?
More precision biology? prion
incubation (Slepoy et al., Phys Rev Lett, 2001;
Mobley et al., Biophys. J. 2003)
Seed
Aggregate
Fission
`Soft Oligomer’/`Micelle’’
`Hard/Oligomer’
Dependence upon coordination
environment
---->
---------------->
12 years --------------1000 years(!)
---->
(Kuru?)
(CJD?)
|
qc=1
Seeded =
Sporadic
qc=2
Seeded = Seeded
Sporadic
|
qc =3
Sporadic