https://aaas.confex.com/aaas/2015/webprogram/Paper14853.html
Session: Emerging Trends in Visualizing Physical Models and Rapid Prototyping for Biological Systems
TITLE: Physical Biomodeling and Foldable, Coarse-Grained Physical Model of Polypeptide Chain
Sunday, 15 February 2015: 8:30 AM-11:30 AM
Room 230C (San Jose Convention Center)
AUTHOR: Promita Chakraborty
ABSTRACT: Physical Biomodeling is a new area of exploration at the interface of computer science and the biological systems. While tremendous advances have been made in computational biology, the cutting-edge 3D printing provides unprecedented opportunities for a third angle into the landscape, thus uncovering this new computational space for modeling that has remained unexplored so far. We tie together these concepts of form-specific physical-digital interfaces. With these principles, a new computational paradigm emerges for physical-digital interfaces for studying of biological phenomenon (e.g. protein folding timesteps) that focuses on shape and dynamics. We define 3 categories, and 6 processes connecting these 3 categories. These categories and processes together form the basis of the philosophy behind the field of Physical Biomodeling. Through exploring Processes 1-6, we see that a relationship-triangle exists between the experimental data from natural systems (N), the computational models for biosystems (C), and scaled, accurate physical models of biosystems (P). Processes 1 and 2 have already existed in the literature for a long time. Physical Biomodeling brings forth Processes 3-6 that will provide a new way to look at the old problems in biology. We arrive at a computational space at the intersection of N, C and P that has so far remained unexplored because of the difficulty in designing and fabricating accurate, scaled physical models of biosystems.
As a first step towards building computer-augmented physical polypeptide chain model that may have applications in structural biology and drug design, we have designed, fabricated and validated a dimensionally accurate, physical model of the polypeptide chain (called Peppytide), that has a flexible backbone where the dihedral bonds are rotationally constrained to match their molecular counterparts. Biological systems involve complex phenomena, and encapsulating these characteristics within a physical body is thought to be difficult. For example, representing the polypeptide chain, a generalized protein chain, along with its complex degrees-of-freedom, by a physical scaled model that will fold dimensionally-accurately, was thought to be quite difficult before the Peppytide project proved otherwise. It opens up new possibilities and challenges with guiding principles that can be extended to build other form-specific physical bio-models. Now with 3D-printing technologies, and possibilities for CAD-cum-biocomputation platforms, we are poised to explore this new domain of study.
AAAS feb15 2015 Physical Biomodeling and Foldable, Coarse-Grained Physical Model of Polypeptide Chain
1. Physical Biomodeling and
Foldable, Coarse-Grained
Physical Model of Polypeptide
Chain
Promita Chakraborty
Feb 15, 2015
AAAS Annual Meeting 2015, San Jose
Symposium: Emerging Trends inVisualizing Physical Models
and Rapid Prototyping for Biological Systems
1
2. 2
A vision: Dynamic physical models of
macromolecules that fold and convey
information
3. BioTable
Computer interface and physical models
• BioTable is a computer monitor
• An interactive unit for translation
• Idea was to detect foldable
models and molecule-molecule
interaction with computers +
head-mount cameras
3
30. Lessons learned for these studies
• Foldable macromolecules: Can they be built at all and folded
with accuracy?
• How can models interact with computer without
cumbersome designs, but retaining accuracy?
• There exists no computational/CAD platform for physical
models to biocomputation platforms
6 Color-coded nucleotides
Coarse-graining
37. Ramachandran plot: a comparison
PDB data
Chakraborty and Zuckermann, PNAS,Vol. 110, No. 33, 2013.
Peppytides at
0.7 RVDW
with rotational
barrier constraints
ϕ ψ
38. Atom-radii = 0.6 RVDW Atom-radii = 0.7 RVDW Atom-radii = 0.8 RVDW
Measured at 5˚ intervals
Ramachandran plot generated using
approx. 80,000 structure files from
Protein Data bank
14
39. O
H
N
15
A scaled model: Peppytide
At Lawrence Hall of Science, UC Berkeley
93,000,000 times magnified
40. 18.4Å
Measuring folding accuracy
6.75(~18.362Å)
alphaC1 to alphaC13
pid: 2ZTA chain B {AA-16} KNYHLENEVARLKKL
16
Hydrogen
bond
Measuring N1 to N5
13.4 Å
12.9 Å
4.853” ± 0.044”
(~13.202 ± 0.121Å)
4.853” ± 0.103”
(~13.202 ± 0.281Å)
pid: 202J
Representing with magnets
• acceptor/donor as N/S pole
O–N distance is typically 3.00 ±
0.12 Å, from an α-helix crystal
structure
In the Peppytide model, the O–N
distance is 1.17 ̋ ± 0.04 ̋
(equivalent to 3.18 ± 0.11 Å)
Alpha helix
Parallel beta sheet
43. Physical Biomodeling: a new field of
exploration
19
Chakraborty, PhD Dissertation, 2014. Example: BioTable
Example: Peppytide
Example: MD in grid*
*Chakraborty, Jha, Katz, Phil.Trans. R. Soc A, 2009 @CCT, LSU
44. Physical Biomodeling: a new field of
exploration
• Precision biomodels as
scientific tool for
computational
modeling
20
Chakraborty, PhD Dissertation, 2014.
46. Possibilities
22
• a different approach to study the same
problem
• protein-protein interaction
• CAD-cum-biocomputation platform
does not exist yet
• Study of misfolded proteins and
aggregates
• Exploring other types of polymers
Chakraborty, PhD Dissertation, 2014.
• Enabling CAD-Bioplatform-3DPrint
• Design→ 3dPrint → Fold paradigm
• A viable input device for molecular
chains
• A viable output device for
molecular chains
• Self-folding
48. 40!
How to make them?
Open source!
Make Magazine -- Projects!
Peppytides
49. Acknowledgments and Contacts
• QuezyLab
• Collaboration with UCSF Science
Health Education Partnership
• Collaboration with Foothill College,
Los Altos
25
• Shantenu Jha, Daniel Katz (CCT, LSU, now at
Rutgers U. and Argonne/U. Chicago respectively)
• Deborah Tatar, Steve Harrison, Francis Quek (VT)
• Ronald Zuckermann (LBNL), DoE (Office of Basic
Energy Sciences), Defense Threat Reduction Agency
(DTRA)
• Alexey Onufriev (VT), Joseph DeRisi (UCSF)
• Molecular Foundry, Lawrence Berkeley National
Lab
• Virginia Tech, Dept. of Computer Sc.
• Lawrence Hall of Science Museum
• Industry Collaboration and support by Autodesk
Inc.
www.quezylab.com
promita@quezylab.com
50. Peppytide videoshoot Joe DeRisi’s lab, UCSF
Berkeley Lab booth,
Berkeley Solano Fest,
2013
Bay Area 2014
Prof. Robert Stroud
testing for beta-turns,
UCSF, Dec 2013
Congress Offices, Capitol Hill,Washington DC
Jun 2014
LBL team with Congressman Jerry McNerney
Nanobio Summer Camp 2014, Foothill
College
52. Existing models
28
CPK models
Dreiding stereomodel
Ball-and-stick
Center for
BioMolecular
Modeling, Milwaukee
School of Engineering
http://www.
3dmoleculardesigns.com/
Scripps Physical Model
Service, Scripps Research
Institute
http://models.scripps.edu/
beta sheet alpha helix
DNA double helix
55. • beta beta alpha motif
• pid: 1FSD 28 amino acids
identities (36 to 39 percent) and P values scores that no sequence information from
any protein motif was used in our sequence
scoring function.
In order to examine the robustness of the
computed sequence, we used the sequence of
FSD-1 as the starting point of a Monte Carlo
simulated annealing run. The Monte Carlo
search revealed high scoring, suboptimal se-
quences in the neighborhood of the optimal
solution (4). The energy spread from the
ground-state solution to the 1000th most
stable sequence is about 5 kcal/mol, an indi-
cation that the density of states is high. The
amino acids comprising the core of the mol-
ecule, with the exception of position 7, are
essentially invariant (Fig. 1). Almost all of
the sequence variation occurs at surface po-
sitions, and typically involves conservative
changes. Asn14
, which is predicted to form a
stabilizing hydrogen bond to the helix back-
D-1 structures. (A) Stereoview of the second zinc
ues and zinc binding site. (B) Stereoview of the
1. For clarity, only side chains from residues 3, 5, 8,
re created with MOLMOL (38).
traints, structural statistics, and atomic root-mean-
annealing structures, SA is the average structure
onAugust15,2012www.sciencemag.orgdedfrom
binding His positions of Zif268, are more
than 80 percent buried, and the Ala at
position 5 is 100 percent buried but the Lys
at position 8 is more than 60 percent ex-
posed to solvent (Fig. 2). The other bound-
ary positions demonstrate the steric con-
straints on buried residues by packing similar
side chains in an arrangement similar to that
of Zif268 (Fig. 2). The calculated optimal
configuration for core and boundary residues
buries 1150 Å2
of nonpolar surface area.
On the helix surface, the algorithm places
Asn14
with a hydrogen bond between its
side-chain carbonyl oxygen and the back-
bone amide proton of residue 16. The eight
charged residues on the helix form three
pairs of hydrogen bonds, although in our
coiled-coil designs, helical surface hydrogen
the overall helix propensity of the sequenc
(5). Positions 4 and 11 on the exposed she
surface were selected by the program to b
Thr, one of the best -sheet forming res
dues (21).
Alignment of the sequences for FSD
and Zif268 (Fig. 1) indicates that only 6
the 28 residues (21 percent) are identic
and only 11 (39 percent) are similar. Four
the identities are in the buried cluster, whic
is consistent with the expectation that bu
ied residues are more conserved than so
vent-exposed residues for a given motif (22
A BLAST (23) search of the FSD-1 s
quence against the nonredundant prote
sequence database of the National Cent
for Biotechnology Information did not r
veal any zinc finger protein sequences. Fu
N-term
C-term
Dahiyat, B.I. and S.L. Mayo, De Novo Protein Design: Fully Automated Sequence Selection. Science, 1997. 278(82)31
FSD-1 denovo structure