Using Computational Biology to Select Protein Segments for Structures that Illuminate Amyloid Diseas

1
Using Computational Biology to Select Protein
Segments for Structures that Illuminate Amyloid
Disease
UCLA Rebecca Nelson, Michael Sawaya, Shilpa
Sambashivan, Magdalena Ivanova, Stuart Sievers,
Michael Thompson, Marcin Aposotol Zhefeng Guo,
Jed Wiltzius, Heather McFarlane, Duilio Cascio,
Daniel Anderson, David Eisenberg ESRF Christian
Riekel, Anders Madsen Univ. of Washington John
Karanicolis, David Baker
2
Using Computational Biology to Select Protein
Segments for Structures that Illuminate Amyloid
Disease

• What are the common features of amyloid
  fibrils/diseases ?

3
Using Computational Biology to Select Protein
Segments for Structures that Illuminate Amyloid
Disease

• What are the common features of amyloid
  fibrils/diseases ?
• How can we grow crystals from proteins involved
  in amyloid diseases ?
• Initial example Sup35 from yeast
• Computational method for other disease proteins

4
Using Computational Biology to Select Protein
Segments for Structures that Illuminate Amyloid
Disease

• What are the common features of amyloid
  fibrils/diseases ?
• How can we grow crystals from proteins involved
  in amyloid diseases ?
• Initial example Sup35 from yeast
• Computational method for other disease proteins
• 12 structures of amyloid spines the questions
  they (partly) answer about the amyloid state
• Is there a common structure for amyloid spines ?
• What is the sequence signal for amyloid formation
  ?
• What is the nature of the native-to-amyloid
  conversion ?
• What accounts for prion/amyloid strains ?
• What makes a protein self-complementary ?

5
Amyloid

• Unbranched, extracellular protein fibrils
• Associated with varied diseases (e.g. CJD
  Alzheimers, Dialysis-related amyloidosis)
• Binds Congo Red with green birefringence
• Formed from globular proteins at low pH
• Cross-b diffraction pattern shows b strands
  perpendicular to fiber axis common spine
• Nucleation-dependent fibril formation

Pathol- ogists
Bio- physicists
Kishimoto, Namba et al (2004)
6
Amyloid-Related Diseases
Amyloid (24)
Prion (infectious)
Other
Disease
Protein
Disease
Protein
Disease
Protein
7
The initial structure Sup35

• Yeast prion-like protein (Saccharomyces
  cerevisiae) (Reed Wickner)
• Functions in translational fidelity
• As amyloid, gives PSI phenotype
• Model for studying amyloidogenesis and prion
  diseases (Lindquist, Weissman)

Prion-determining domain
Middle domain
Elongation factor homology domain
8
For the yeast prion, Sup35, what is smallest
segment that forms amyloid fibrils ?
Research of Melinda Balbirnie and Rebecca Nelson
9
Sup 35 peptides form fibrils and microcrystals
In both micro-crystals and fibrils, b-strands
are normal to the long axis.
NNQQ
GNNQQNY
NNQQ
NNQQNY
GNNQQNY
30-50 mg/mL (60-100 mM) citrate, PEG, isopropanol
1 mg/mL (1.2 mM) in H2O
1 mg/mL (20 mM) in H2O
15-30 mg/mL (19-38 mM) HEPES, acetate, Zn/Cd
GNNQQNY fibrils exhibit properties of amyloid
fibrils dye binding, cooperative aggregation
kinetics, stability, cross-ß diffraction
GNNQQNY crystals are very dense, dehydrated, and
extremely stable
10 mg/mL (12 mM) in H2O
10
Conventional size vs. microcrystals
NNQQNY crystal. Shown on the same scale
A small lysozyme crystal
900 unit cells wide. 100,000 unit cells long
35,000 microcrystals could fit in this small
lysozyme crystal
11
Dissect a single needle from a cluster and glue
it to a capillary
50 mm
50 mm
12
Microfocus is required to reduce background noise
100 mm beam diameter Standard at home or
synchrotron Only a fraction of incoming X-rays
impinge crystal. High background obscures
reflections
1 mm beam diameter ESRF ID13 All X-rays impinge
crystal Low background, good I/s.
13
GNNQQNY Structure Resolution 1.8 A R 0.18
Rfree 0.19 Only short segment needed for
amyloid spine Pair-of-sheets bonded by
intersheet van der Waals steric
zipper Uncharged, low complexity sequences
favored
Nelson et al. Nature, 436, 554-558 (2005)
14
Structure of GNNQQNY
Steric Zipper Enmeshing of sidechains No
H-bonds between sheets in dry interface
H-bonds up and down sheets
15
Structure of GNNQQNY
Steric Zipper Enmeshing of sidechains No
H-bonds between sheets in dry interface
H-bonds up and down sheets
Quantify complementarity with Sc parameter
Lawrence Colman, 1993 For
proteases/inhibitors Sc 0.73
0.03 Antibody-antigens Sc 0.66 0.02 For dry
interface of GNNQQNY Sc 0.86
16
Structure of GNNQQNY
17
What we learned from GNNQQNY

• A 7 amino-acid peptide is sufficient to form
  amyloid structure

18
What we learned from GNNQQNY

• A 7 amino-acid peptide is sufficient to form
  amyloid structure
• The fundamental unit of this amyloid is 2 sheets,
  tightly interdigitated by steric zipper

19
What we learned from GNNQQNY

• A 7 amino-acid peptide is sufficient to form
  amyloid structure
• The fundamental unit of this amyloid is 2 sheets,
  tightly interdigitated by steric zipper
• The interdigitating sheets present a high barrier
  to formation (slow formation and slow return)

20
Structure of GNNQQNY
Steric Zipper Enmeshing of sidechains No
H-bonds between sheets in dry interface
H-bonds up and down sheets
21
What we learned from GNNQQNY

• A 7 amino-acid peptide is sufficient to form
  amyloid structure
• The fundamental unit of amyloid is 2 sheets,
  tightly interdigitated by steric zipper
• The interdigitating sheets present a high barrier
  to formation (slow formation and slow return)
• Once exposed, 4 GNNQQNY peptides can nucleate
  fibril formation

22
Problem How do we find the short peptide (or
peptides) in a protein that cause that protein to
form amyloid-like fibrils ?Answer Turn to
computational biology Bowie et al. Science
(1991)
3D Profile Threading From Structure to
Sequence
23
The 3D Profile method to find other
amyloid-forming peptides
RosettaDesign
Research of Michael Thompson and Stuart Sievers
in collaboration with J. Karanicolis D. Baker,
U.Wash
24
Myoglobin
SQAIIH
Thompson, MJ et al., PNAS, 2006
25
Lysozyme
IFQINS
Thompson, MJ et al., PNAS, 2006
26
Microcrystals from amyloid-forming peptides
To date 40 peptides
27
Summary Microcrystal Structures of
Fibril-Forming Peptides

• In every amyloid-disease-related protein, we can
  find at least one short segment that forms
  fibrils (and crystals !)

28
Summary Microcrystal Structures of
Fibril-Forming Peptides

• In every amyloid-disease-related protein, we can
  find at least one short segment that forms
  fibrils (and crystals !)
• There are fundamental similarities shown by the
  crystals closely interacting sheets, with dry,
  steric zipper interfaces. Thus the fibrils of
  several conformational diseases have a common
  atomic basis.

29
Summary Microcrystal Structures of
Fibril-Forming Peptides

• In every conformational disease-related protein,
  we can find at least one short segment that forms
  fibrils (and crystals !)
• There are fundamental similarities shown by the
  crystals closely interacting sheets, with dry,
  steric zipper interfaces. Thus several
  conformational diseases have a common atomic
  basis.
• The structures suggest how proteins convert to
  the fibril form can seed other molecules into
  the fibril by exposing their steric zipper
  segments.

30
How are the microcrystal structures of the
cross-b spine related to the actual fibrils ?
Fibil consists of a spine of 1 pair or a small
number of pairs of sheets
Fibril consists of a spine of a pair of sheets,
with swapped domains on the periphery (Sambashivan
et al. Nature, 2005)
Q10 insert
with Q10 insert
Fibrils active
Spine is a pair of sheets built from VEALYL on
B chain
31
Zipper-Spine Model of Domain-Swapped Amyloid
Protofilament of RNase A with Q10 insert
RNase A enzymatic activity in the fibrils proves
that native-like structure exists in the fibril
-Sambashivan et al. Nature (2005)
32
Fibrils of RNase A with inserts in N- and
C-terminal Hinge Loops
QQQQQQQQQQ In C-terminal loop
QQQQ in N-terminal loop
NNQQNY (Sup35 sequence) in C-terminal loop
GQQQQQQQG In C-terminal loop
Research of Yanshun Liu and Shilpa Sambashivan
33
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ?
34
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface
35
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ?
36
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity
37
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ?
38
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ? For
RNase A, opening of the domains form-
mation of spine with peripheral swapping
39
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ? For
RNase A, opening of the domains form-
mation of spine with peripheral swapping What is
the structural basis for prion strains ?
40
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibiliry with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ? For
RNase A, opening of the domains form-
mation of spine with peripheral swapping What is
the structural basis for prion strains ?
41
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ? For
RNase A, opening of the domains form-
mation of spine with peripheral swapping What is
the structural basis for prion strains ?
Possibly spine polymorphs
42
Amyloid Fundamental Structural Questions
Is there a common structure for the cross-b
spine of amyloid-like fribrils ? A
pair-of-sheets with a dry steric-zipper
interface Is there a sequence signature for
the cross-b spine ? Yes, but subtle
compatibility with the steric zipper
interface low complexity What is the nature of
the conversion from native to fibril ? For
RNase A, opening of the domains form-
mation of spine with peripheral swapping What is
the structural basis for prion strains ?
Possibly spine polymorphs What is the basis for
self-complementation of proteins ?
43
Self-Complementation in Amyoid
Spine Steric zipper favors low complexity
sequences
In some cases, sidechain stacks (polar zippers)
Core Domain swapping
44
UCLA-ESRF Amyloid Structure Team
Rebecca Nelson
Shilpa Sambashivan
45
UCLA Amyloid Structure Team cont.
Zhefeng Guo T7EI
Marcin Apostol Prion protein
Magdalena Ivanova a-synuclein, insulin
Jed Wiltzius, IAPP
Heather Mcfarlane, Sup35
46
Amyloid Computation Team
Mike Thompson
Stuart Sievers
Jon Karanicolas, David Baker at Univ. of
Washington