Synthetic Scaffolds and Protein Assemblies
for Engineering Applications
by
Julie Erin Norville
Submitted to the Department of Electrical Engineering and Computer
Science
in partial fulfillment of the requirements for the degree of
Master of Science in Electrical Engineering and Computer Science
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
September 2004
@ Massachusetts Institute of Technology 2004. All rights reserved.
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
OCT 2 8 2004
Author..
...................
... LIBRARIES
Department of Electrical Engineering and CompuloSIEn
,September 8, 2004
Certified by......
Thiomas F. Knight, Jr.
Senior Research Scientist, MIT Computer Science and Artificial
Intelligence Laboratory
Thesis Supervisor
Certified by . .
.....................
Angela M. Belcher
Associate Professor, MIT Department of Material Science and
Engineering-and 1epartmenja.ePiological Engineering
Te Superir
Accepted by .....
. ...............
Arthur C. Smith
Chairman, Department Committee on Graduate Students
MASSACHUSETTS INSTITUTE
OF TECHNOLOGY
LIBRARIES
.'
I w--
Synthetic Scaffolds and Protein Assemblies
for Engineering Applications
by
Julie Erin Norville
Submitted to the Department of Electrical Engineering and Computer Science
on September 8, 2004, in partial fulfillment of the
requirements for the degree of
Master of Science in Electrical Engineering and Computer Science
Abstract
S-layer proteins, which naturally self-assemble on the exterior of cells, provide an
interesting basis for the creation of synthetic scaffolds. In this thesis, I created a
plasmid which produces a recombinant form of a well characterized S layer protein,
sbpA, which has a number of properties ideal for nanotechnology applications. I also
explored purification of both the native and recombinant forms of sbpA. Together
these preliminary studies are the first, necessary, steps towards quantitative generation of crystallization conditions and the ultimate modifications of the protein form
for a wide variety of engineering applications.
Thesis Supervisor: Thomas F. Knight, Jr.
Title: Senior Research Scientist, MIT Computer Science and Artificial Intelligence
Laboratory
Thesis Supervisor: Angela M. Belcher
Title: Associate Professor, MIT Department of Material Science and Engineering and
Department of Biological Engineering
3
4
0.1
Acknowledgments
This material is based upon work supported under a National Science Foundation
Graduate Research Fellowship. This work was also supported in part by a Vinton
Hayes Graduate Research Fellowship, in part by DARPA/ONR under grant N0001401-1-1060, and in part by a David and Lucille Packard Fellowship.
Any opinions, findings, conclusions or recommendations expressed in this publication are those of the author and do not necessarily reflect the views of the National
Science Foundation.
" To my advisors Thomas Knight, Jr. and Angela Belcher for inspiring and
cultivating 'creative milieus' both in their laboratories and in the minds of their
students.
" To Gerald Sussman for helpful conversations.
" To the students of the Knight lab: Austin Che, Reshma Shetty, and Ed Ouellette
who have helped me in so many ways. A particular thanks to Austin for the
many hours he spent explaining techniques and the fundamentals of molecular
biology.
" To the students and former students of the Belcher Laboratory: Kiley Miller,
Soo-Kwan Lee, Saeeda Jaffar, Daniel Solis, Stephen Kottman, Beau Peele, Ki
Tae Nam, Chung-Yi Chiang, Eric Krauland, Asher Sinensky, Sreekar Bhaviripudi,
Kate Ryan, Andrew Magyar, Brian Reiss, Jifa Qi, Christine Flynn, Yu Huang,
Dong-Soo Yun, and Davide Marini. A particular thanks to Saeeda Jaffar, Kiley
Miller, and Yu Huang for the many helpful discussions that we shared. Thanks
also to Christine Flynn for teaching me the techniques of phage display.
" To Katherine Gibson and Sanjay DeSouza of the Walker lab, Hector Hernandez
of the Drennan lab, and Parick Kiley, Steve Yang, and Andrea Lomander of the
Zhang lab thanks for answering my numerous question and providing help and
advice.
5
*
To my parents and family for their enduring support.
6
Contents
0.1
. . . . . . . . . . . . . . . . . . . .
Acknowledgments . . . . . . . ..
13
1 Introduction
1.1
Problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
13
1.2
Content of the Thesis. . . . . . . . . . . . . . . . . . . . . . . . . . .
16
17
2 Background
2.1
3
5
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
17
2.1.1
The Location of S-Layers: the Surface of Cells . . . . . . . . .
17
2.1.2
Natural Configurations of S-Layers . . . . . . . . . . . . . . .
17
2.1.3
Functions of S-Layers . . . . . . . . . . . . . . . . . . . . . . .
19
S-layers
23
Choosing a Protein
27
4 Experimental Procedures
4.1
4.2
Purification of the Native Form . . . . . . . . . . . . . . . . . . . . .
27
4.1.1
Motivation for Purifying the Native Form
. . . . . . . . . . .
27
4.1.2
S-layer in Relation to the Organism . . . . . . . . . . . . . . .
28
4.1.3
Chaotropic Reagents . . . . . . . . . . . . . . . . . . . . . . .
29
4.1.4
Purification Strategies
. . . . . . . . . . . . . . . . . . . . . .
30
Creating the Recombinant Plasmid . . . . . . . . . . . . . . . . . . .
33
4.2.1
Protocol Overview . . . . . . . . . . . . . . . . . . . . . . . .
33
4.2.2
O ligos . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
33
4.2.3
Miniprep of Genomic DNA . . . . . . . . . . . . . . . . . . . .
34
7
4.3
5
4.2.4
PCR of the Gene sbpA . . . . . . . . . . . . . . . . . . . . . .
34
4.2.5
Cloning of the Gene
. . . . . . . . . . . . . . . . . . . . . . .
34
Purifying the Recombinant Form . . . . . . . . . . . . . . . . . . . .
37
4.3.1
Strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
37
4.3.2
Inclusion Body Based Preparation.
. . . . . . . . . . . . . . .
37
4.3.3
Affinity Based Preparations
. . . . . . . . . . . . . . . . . . .
37
Results
41
5.1
Purification of the Native Form . . . . . . . . . . . . . . . . . . . . .
41
5.1.1
Whole Cell Method . . . . . . . . . . . . . . . . . . . . . . . .
41
5.1.2
Cell Wall Methods
. . . . . . . . . . . . . . . . . . . . . . . .
41
5.2
Creating a Recombinant Plasmid . . . . . . . . . . . . . . . . . . . .
44
5.3
Purifying the Recombinant Form . . . . . . . . . . . . . . . . . . . .
44
5.3.1
Inclusion Body Based Preparation. . . . . . . . . . . . . . . .
44
5.3.2
Affinity Based Preparations
. . . . . . . . . . . . . . . . . . .
45
6 Discussion and Conclusions
6.1
7
47
Purification of the Native Form . . . . . . . . . . . . . . . . . . . . .
47
6.1.1
Chaotropic Reagents . . . . . . . . . . . . . . . . . . . . . . .
47
6.1.2
Whole Cell Method . . . . . . . . . . . . . . . . . . . . . . . .
47
6.1.3
Cell Wall Methods
. . . . . . . . . . . . . . . . . . . . . . . .
48
6.2
Creating a Recombinant Plasmid . . . . . . . . . . . . . . . . . . . .
49
6.3
Purifying the Recombinant Form . . . . . . . . . . . . . . . . . . . .
50
6.3.1
Inclusion Body Based Preparation . . . . . . . . . . . . . . . .
50
6.3.2
Affinity Based Preparations
. . . . . . . . . . . . . . . . . . .
50
Impact and Future Work
51
A Winfree's Work
53
A .1 T heory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
53
A .2 Practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
55
8
List of Figures
1-1
S-layers form regular, nanostructured arrays. Image adapted from [45]
2-1
Relationship between S-layers and cell architecture for different cell
15
types. Image adapted from [40] . . . . . . . . . . . . . . . . . . . . .
18
2-2
From protein chains to crystals
. . . . . . . . . . . . . . . . . . . . .
18
2-3
Most common S-layer symmetry groups. Image adapted from [45] . .
19
2-4
S-layer as a phage receptor. Adapted from [3]
. . . . . . . . . . . . .
20
3-1
Schematic of sbpA assembly: assembly formation, double layer in solution, monolayer on silicon
4-1
. . . . . . . . . . . . . . . . . . . . . . .
25
Basic schematic of sbpA assembly. Depending upon the protocol used
however, the s-layer protein can be removed from either whole cells or
the cell walls, it can be concentrated before the denaturant is removed,
and the cofactor (which is calcium ions) can also theoretically be added
at the same time that the denaturant is removed. . . . . . . . . . . .
4-2
Map of plasmid sbpA-pET28a vector which codes for the protein Sbpa
with an n-terminus hexahistidine tail. . . . . . . . . . . . . . . . . . .
5-1
28
36
Whole Cell Urea Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lanes 2 and 3, sample heated to 70'C. Lanes 4 and 5, sample
heated to 100"C. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5-2
42
Whole Cell GHC1 Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lane 2, dialysis pellet A. Lane 3, dialysis supernatant A.
Lane 4, dialysis pellet B. Lane 5, dialysis supernatant B. . . . . . . .
9
42
5-3
Whole Cell pH Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lane 2, blank. Lane 3, .2M Glycine HCl. Lane 4, blank. Lane
5, .05M Glycine HCI. . . . . . . . . . . . . . . . . . . . . . . . . . . .
5-4
Bull's eye like patterns appear in cell lysate pellets when some but not
all cells are broken. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
5-5
43
43
Comparison of Native and Recombinant Protein. Lanes 1 and 2, sbpA
from whole cell urea method. Lane 3, Mark12 Unstained Standard
(Invitrogen.) Lanes 4 and 5, SDS extracted whole cells containing the
sbpA/pET28a plasmid which were induced for 3 hours. . . . . . . . .
5-6
Affinity Purification of Soluble Fraction.
45
Lanes 1 through 5, frac-
tions 1 through 5 of 50mM imidazole elution. Lanes 6, 7, 12, and
13, Mark12 Unstained Standard (Invitrogen.)
Lanes 8 through 12,
fractions 1 through 5 of 100mM imidazole elution. Lanes 14 through
18, fractions 1 through 5 of 250mM imidazole elution. . . . . . . . . .
46
A-1 Self-Assembled Circuit Patterns a) Rule tiles, boundary (input and
seed) tiles, and resulting demultiplexer based on a counter type pattern. b) Addressable memory array utilizing two demultiplexers. Image adapted from [10] . . . . . . . . . . . . . . . . . . . . . . . . . . .
54
A-2 Sierpinski triangle assembly. a) Rule tiles (which perform the logic by
implementing an XOR operation) and boundary tiles (which includes
the corner tile) b) Process of growth using the two or more strong
interaction rule from the abstract Tiling Assembly Model. c) Rates
of tile association and disassociation using the kinetic Tiling Assembly
Model. d) Various stages of assembly. Image adapted from [54]
. . .
54
A-3 DNA Tiles a) Colored edge model of DX DNA tile striped assemblies.
b) Model structures of different type A units. c) Assembly lattice
topologies of different tile forms. d) Physical realizations of AB and
ABCD forms, which are shown schematically in part a), measured with
fluid cell AFM (the scale bars are 300nm.) Image adapted from [55] .
10
56
List of Tables
4.1
Synthesized Oligos for PCR . ..
11
. . . . . . . . . . . . . . . . . . . .
34
12
Chapter 1
Introduction
1.1
Problem
For over thirty years, lithography has been the best and least expensive method for
transferring complex instructions for building devices and interconnect onto a surface
in a massively parallel fashion. Throughout this period the process has been continually improved by decreasing the wavelength of light used, which has the effect
of decreasing the size and cost of transistors while increasing their speed. As the
wavelength decreases into the x-ray region, the cost of both sources and optics becomes prohibitively expensive [52]. Although the methods of e-beam lithography and
dip-pen lithography do allow pattern formation on relevant length scales, parallelism
is challenging to build into these processes. Thus, these low throughput processes are
unlikely to replace lithography as a primary pattern formation technique. Processes
like stamping do retain the parallelism of lithography at nanometer length scales, but
in this case alignment becomes a problem. Moreover, if silicon remains the medium
for device fabrication, other problems such as statistical variations in devices due
to doping become a consideration below 50nm. Devices such as carbon nanotubes,
silicon nanowires, and molecular electronics provide alternate possibilities where the
size of the device is no longer the main limitation. Thus, the main requirements for
a replacement technology for lithography are self-alignment on all length scales and
the parallel construction of complex patterns with molecular precision. Chemically
13
precise self-assembly potentially meets these criteria and could replace lithography.
In order for chemically precise self-assembly to supplant lithography, the process
must be more than a series of dyadic or triadic cross-linking or binding interactions
between a number of parts. Similarly, the construction of uniform crystalline lattices
as an underlying pattern is not sufficient if the goal is to create structures that will
template the formation of a functioning circuit.
Instead, chemically precise self-
assembly must be a process where designed complexity is an emergent property of
the starting conditions. Thus the object is not merely building from the bottom up,
but rather building in designed complexity from the bottom up. Ideally, the result
will be synthetic scaffolds that can template the formation of secondary structures,
such as circuits, on their surface. Winfree's work with two-dimensional, non-periodic,
self-assembled DNA structures gives some idea of the directions that I anticipate
(please refer to Appendix A for a review of his relevant work.)
This strategy for creating a self-assembly approach uses a naturally occurring
protein tile as a basis. Biological proteins, unlike normal disordered polymers, consist of molecularly identical and identically ordered heteropolymers of twenty distinct
monomers, the natural amino acids. These highly structured polymers fold into identical, intricate, three dimensional structures with specific surface shapes and chemical
properties. Specific proteins, either engineered or naturally occurring, self-assemble
into regular arrays in one, two, or three dimensions by forming surface binding regions
complementary to one another. I have been motivated in thinking by the so-called
S-layer proteins, a diverse group of proteins which coat the outer surface of many
eubacterial and archaeal cells with a two-dimensional crystalline layer.
My choice of proteins as a material to work with is not accidental. The remarkable
ability of biology to programmably construct proteins using DNA as instructions
allows us to think about very intricate self-assembled structures.
Further, the fabrication of these structures is inexpensive - one can program
cells to do most of the work and one can program cells in some instances to export
the proteins into the culture medium. One might even imagine a film of cells (size
approximately 400 nm) which could be programmed to selectively produce specific
14
model of
-Computer
-layer
S-layers on cel
S-layer on
-
-100
silicon
20 nm
nm
Figure 1-1: S-layers form regular, nanostructured arrays. Image adapted from [45]
proteins at specific spatial locations. Those proteins might then self-assemble to form
locally distinct, non-uniform structures on a surface or within a volume.
Several researchers, notably Uwe Sleytr, have investigated the properties of these
S-layer proteins (some illustrations of which are shown in Figure 1-1.) The sequence
of many of these proteins is known. Some, but not all, can be cloned into E. coli. The
three-dimensional structures of some have been explored with electron microscopy or
x-ray crystallography. The largest monomolecular domains created are of diameter
50Mm [30, 31] (the domains are created with tiles on the order of 13x13 nm 2 .) This
shows the potential scalability of the process using Langmuir Blodgett techniques
at an air-water interface and the feasibility of transfer of the assemblies to solid
substrates. Alternately, when tiles assemble directly onto solid substrates, such as
silicon, relatively uniform coverage can occur with domains ranging in size from 100200nm to 3-10pm in diameter depending on the substrate utilized [18].
15
1.2
Content of the Thesis
S-layer proteins seem to provide a natural and promising model for an engineering
system to create programmed self-assembled structures utilizing protein tiles. The
objective of the Master's thesis is to evaluate whether a particular S-layer protein is
a suitable basis tile for self-assembly. One can break this process into a number of
steps: 1) select the most promising model S-layer protein from the literature, 2) gain
expertise in purifying the native form of the model protein (to utilize as a control
in future experiments), 3) build a plasmid for producing the recombinant form of
the protein, 4) gain expertise in purifying the recombinant form of the protein, and
5) begin exploring crystallization and characterization of native and recombinant
monomolecular assemblies.
16
Chapter 2
Background
2.1
2.1.1
S-layers
The Location of S-Layers: the Surface of Cells
The term S-layer is derived from surface layer, because S-layer proteins form crystalline arrays on the outermost portion of the cell wall [46].
Thus, the outermost
surface layer or S-layer consists of these proteins. S-layers are common in eubacteria
and archaea, though the protein sequences that comprise them have little homology
[40].
What does it mean for the S-layer to be on the outside of the cell? It means
different things for different types of bacterial depending on the architecture of the
cell, as seen in Figure 2-1. In archaea, outside the cell means that the S-layer is
outside the cell membrane. In gram-positive bacteria, outside the cell means outside
the peptidoglycan layer. In gram-negative bacteria, outside the cell means outside
the outer membrane.
2.1.2
Natural Configurations of S-Layers
Typically, several identical S-layer protein chains or subunits assemble to form a
morphological unit [43]. Many morphological units assemble to form the crystalline
sheet of S-layer proteins that cover the cell (see Figure 2-2.)
17
Three major types
I Archaea
cell exterior
Gram positive eubacweda
I
IN
=
I||
cell irtedor
Gram negave
cell eaerior
e
l"-
cell interior
eubactera
oell egenor
eyn L3"WvMbMW
cell
interior
Figure 2-1: Relationship between S-layers and cell architecture for different cell types.
Image adapted from [40]
Protein chains form units,
units form crystals
Protein Chain
or Subunit
Morphological
Unit
Crystalline
Array
cetrtocne
i
e
center-to- con% r latice
spacing: 3-15 nm
Figure 2-2: From protein chains to crystals
18
- -------------------
A diversity of lattice shapes exists
of symmetry exist in S-layer protein lattices: oblique, square, and hexagonal (see
Figure 2-2.)
Square symmetry may be the most appropriate for nanotechnology
applications. Lattice spacing is the center-to-center spacing of the asymmetric units.
Lattice spacings of S-layers, which may be useful for nanotechnological applications,
range in size from 3-l5nm.
2.1.3
Functions of S-Layers
The functions of S-layer proteins have not yet been fully characterized from a molecular standpoint.
Hypothetical functions have been postulated, however.
Campy-
lobacter fetus, which causes spontaneous abortion in sheep and infections in other
mammals, dynamically switches its coat of S-layer proteins during different stages
of infection, perhaps allowing it to stay one step ahead of the immune response [3].
Surface layers may be part of an evolutionary battle between bacteria and phage (the
viruses that kill them.) In the literature, researchers highlight this property and they
often refer to S-layer proteins as phage receptors [3, 21] (see Figure 2-4.) Suppose that
only the cell membrane, peptidoglycan layer, or outer membrane (depending upon the
type of bacteria-either archaea, gram positive eubacteria, or gram negative eubac19
Phage bound
to s-layer
Figure 2-4: S-layer as a phage receptor. Adapted from [3]
teria) was the interface between the bacteria and the environment. In that case the
phage would have a one shot method for entry into the bacteria as soon as it evolved
a mechanism for getting past those defences. If the cell has an S-layer, specifically a
crystalline covering composed of repeating units, then either the phage or the immune
response must specifically recognize that exact sequence. (An interesting counterpoint
here is the remarkable success achieved in recognizing repeating crystalline forms in
inorganic materials using the technique of phage display. [53, 13, 34]). This mechanism may have evolved because the S-layer is both a conservative means in DNA
coding space to create a cell wall covering that reveals as little information about
the organism as possible provided that the sequence topology is not known to either
the phage or the immune system. Dynamic switching of S-layers may also allow fast
adaptation to changes in environmental conditions on non-evolutionary time scales.
In this case, again, the molecular mechanisms conveying functional advantage are not
known. Researchers have observed that certain organisms change or stop producing
an S-layer under one or more of the following conditions: transfer from an anaerobic
20
to aerobic environment [35], temperature increase [12], or repeated cultivation under
laboratory conditions
21
22
Chapter 3
Choosing a Protein
I selected the protein sbpA from the strain Bacillus sphaericus (B. sphaericus) ATCC
4525 for a variety of reasons. No implicit access-based roadblocks existed for this
choice. The strain (ATCC 4525) and the media (nutrient broth) required for growth
were commercially available from the American Type Culture Collection. The cultivation conditions required by the strain were accessible. It grew at a moderate
temperature (30'C) with shaking in an aerobic environment, was a BLi (no safety
risk) organism, and had a short doubling time [9, 47]. Both native and recombinant
forms of the protein sbpA were well characterized. The DNA and protein sequences
are known and publicly available [14, 15, 25]. Since the spbA protein tetramer formed
a square (measured lattice periodicity 12.3x12.3nm 2 +/-0.8nm [49]) which can selfassemble into arrays, it was perhaps one of the closest natural embodiments of a
Wang tile (-typically rigid square tiles [10]). In addition, the plasmid containing
the entire reading frame of the gene sbpA wass stable in E. coli [24] despite the
fact that B. sphaericus is gram positive and E. coli is gram negative. Thus, the
recombinant protein can be produced in E. coli, which is the standard for genetic
engineering. Protein purification and crystallization techniques exist for both native
[17, 18, 23, 30, 44] and recombinant forms (both native-like and modified) [24, 25].
Thus, sbpA assembly could be a reversible and controlled process, which was not
true for all s-layer proteins. In addition, sbpA shows utility in previously reported
nanotechnology applications. These applications included sbpA assembled on silicon
23
[18], used as a "nanonatural" resist or mask [32], incorporated on substrate surfaces
irrigated by PDMS channels [17], supported lipid membranes (which can also contain
functional membrane proteins) [16], coated liposomes for importation into cells [24],
oriented of single chain antibodies [29], assembled functional fusion protein arrays (of
GFP [24] and an allergen protein sequence [25]), and created (cadmium sulphide [38]
and gold [11]) S-layer templated nanoparticle superlattices. This is useful because my
objective was not to discover new science at this stage but rather to utilize known
science in order to build a system with well defined parameters and consistently reproducible results. The natural characteristics of the protein sbpA which enabled
prior nanotechnology applications, primarily that it presents three distinct surfaces
(1. lateral which binds to other subunits-one on each side of the tetragon, 2. top
which is exposed to the environment in nature, and 3. bottom which is exposed to
the cell in nature) are likely to lead to additional advances in the future. The ability
of sbpA to crystallize on surfaces such as silicon in either single or double layer sheets
depending upon where the surface is hydrophobic or hydrophilic (see Figure 3-1) [18]
is especially useful if assembly occurs on a flat surface, as atomic force microscopy
can then be used as an observational technique. Although the structure of the s-layer
protein from this particular strain has not been solved, the three-dimensional structure of a tetragonal protein from a related strain B. sphaericus P-I has been explored
with 3-D electron microscopy to a resolution on the order of Inm [28]. This is useful
for gaining some insight into what s-layer subunit interactions look like in a rough
sense as the first 35 n-terminal amino acids of sbpA show 97.1 percent identity with
the S-layer protein of B. sphaericus P-1, the next 200 n-terminal amino acids show
80 percent identity, and the remaining amino acids show less than 20 percent identity
[23].
24
The proteins are oriented and have different
connectors at top, bottom, and side
Hydrophilic top
Side connectors
cause the
Hydrophobic
Double-layer
Hydrophilic
Hydrophobic
Hydrophobic
Hydrophilic
bofttm
formation of layers
Monolayer on surface
Hydrophilic
Hydrophobic
Surface (i.e.,
hydrophobic silicon)
Figure 3-1: Schematic of sbpA assembly: assembly formation, double layer in solution,
monolayer on silicon
25
26
Chapter 4
Experimental Procedures
4.1
4.1.1
Purification of the Native Form
Motivation for Purifying the Native Form
It is important to have reliable methods of purifying the native form because it provide
a standard for comparison for the recombinant form. This will be especially useful
as a control for exploring crystallization conditions. Eventually, when the procedure
is optimized for the recombinant protein, the native protein can become the positive
control for further modifications of the protein or to replace novel components. For
purification of the native form the strategy is relatively straightforward. The S-layer
protein is located on the outside of the cell wall, which it binds to. All the other
proteins that make up the cell or could be released into the medium are impurities.
To remove the protein sbpA in a pure state, either the protein is removed from the
whole cell using a denaturant or the cells are broken, the cell walls harvested and
washed, and then the protein is removed from them using a denaturant. For this
thesis I explored a few of these alternatives. Although the general strategy is the
same regardless of the protocol, I will consider the implications of using different
steps such as using a different denaturant or a different type of cell breakage method.
27
Some s-layers proteins can be
disassembled and then reformed into
crystals for nanotech applications
Denaturation method
S-protein
o iw
Add
denaturant
Remove
denaturant
Remove
cell wall
Unfolded
Protein Chain
Refolding
koce-Add
tratecofactor
Refolding
o
4
Formation4
Assembly of subunits
Into subunits
into crystals
Nanotech
application
Figure 4-1: Basic schematic of sbpA assembly. Depending upon the protocol used
however, the s-layer protein can be removed from either whole cells or the cell walls,
it can be concentrated before the denaturant is removed, and the cofactor (which is
calcium ions) can also theoretically be added at the same time that the denaturant
is removed.
4.1.2
S-layer in Relation to the Organism
The first thing one should be consider when contemplating sbpA, the S-layer protein
of B. sphaericus ATCC 4525, is the position and relationship of it with regard to the
rest of the cell. As described in the background section and displayed in Figure 2-1,
gram positive cells such as B. sphaericus have thick peptidoglycan cell walls surrounding them. Peptidoglyclan is a polymer scaffold of peptide cross-linked disaccharides
which surrounds the cell and prevents osmotic lysis. Some types of peptidoglycan
can be cleaved by the enzyme lysozyme (if this was the case it could be utilized as
a strategy for s-layer purification.) However, experimental evidence indicates that
the B. sphaericus ATCC 4525 is lysozyme resistant [36]. Possible evidence for this
property is also described in a paper on fast screening methods for examining S-layer
proteins where S-layer proteins are examined by AFM and TEM after treatment with
lysozyme and sodium azide, since the B. sphaericus ATCC 4525 cells kept their shape
despite this treatment [49]. The partial release of S-layers that they observe is con28
sidered to be evidence "that the peptidoglycan layer of both B. sphaericus 9602 and
B. sphaericus CCM 2177 (=ATCC 4525) becomes at least partially digested during
lysozyme treatment, because we observe a release of S-layers" and in controls for B.
sphaericus 9602 this behavior is not observed in the absence of lysozyme [49]. (A
similar result with the same strain, NTCC 9602, and lysozyme treated cell walls was
previously reported [20].) However, I note that "only during late-exponential-early
stationary phase are free S-layer sheets frequently detected" and that for B. sphaericus "the S-layer protein is synthesized during vegetative growth" [39], which is similar
to the time in which the fast screening experiments are conducted.
4.1.3
Chaotropic Reagents
Two chaotropic reagents, urea and guanidinium hydrochloride (GHCl), were utilized
for these experiments. (Alternatively, sarkosyl, a mild anionic detergent often used for
solubilizing and refolding inclusion bodies may be useful for this application although
it was not explored in these experiments.)
Urea and GHC1 have different advantages and disadvantages for protein denaturation and refolding. Urea can contain cyanate ions which can cause protein modification. In order to minimize this effect urea solutions are always made fresh for
these experiments. Alternatively, some researchers add (1mM) glycine to their (6M)
urea solutions to neutralize the cyanate ions as they typically react more quickly with
glycine residues than with lysine residues [22, 48]. However, urea denatured samples
can be run directly on SDS-PAGE gels. In fact urea gradient gels can be used to
directly explore the folding transitions of proteins. Diluted urea samples (3.5M or
less) are compatible with the Bradford assay (Coomassie or Coomassie Plus, Pierce.)
Qualitatively, urea acts as a hard denaturant, thus discreet folding transitions occur
at defined concentrations. GHC1 is more time consuming to use for SDS-PAGE gel
assays as one must either be dilute, dialyze before analysis, or isolate the protein
from GHCl by trichloroacetic acid precipitation [33]. Diluted GHC1 samples (3M or
less) are compatible with the Bradford assay (Coomassie or Coomassie Plus, Pierce.)
Qualitatively, GHCl acts as a soft denaturant, with wider folding transitions. It is
29
impossible to know de novo which denaturant will provide a larger percentage of soluble, active protein. This property can only be determined with an assay-in this case
determining if crystallization occurs. It is noted that GHC1 is used most frequently
in B. sphaericus purification applications, however.
4.1.4
Purification Strategies
Two approaches towards the peptidoglycan layer can be made.
It can either be
broken and the contents of the cell washed away or it can be left intact and the
external S-layer protein removed while the cell remains intact.
Whole Cell Method
First, we will consider the method where the cell remains intact as this is theoretically
the simpler of the two. In [42] it is suggested that intact S-layer fragments can be
removed from intact cells (of most strains) by incubation in low concentrations of
urea (0.5M.) This procedure describes a method for removing S-layer subunits at
high concentrations of urea. B. sphaericus was induced from glycerol stocks with a
sterile toothpick and grown with shaking for 12 to 16 hours at 30'C in nutrient broth
(DIFCO.) They were then diluted 1:100 in nutrient broth and grown with shaking
until they reached a absorbance of OD600=.78. They were then harvested at 4500g in
a Beckman Coulter J-20 XP centrifuge at 4C. In order to remove external impurities,
B. sphaericus cells were rinsed 2-3 times in 30mL TrisHCl pH 7.4. (At this point the
cells could be stored at -80'C and after thawing for 30 minutes, the cell pellet was
again rinsed in 30mL of TrisHCl pH 7.4 (to remove impurities from cells that had
broken during the freezing process.)) If the cells were not thoroughly rinsed a protein
band on the order of 40kD was sometimes present. The cells were incubated in a
chaotropic buffer (8M Urea, 10mM Tris, pH 8.0, or 5M GHCl, 50mM Tris, pH 8.0)
for 2 hours at room temperature. The cells were then removed by centrifugation
at 10,000g for 30 minutes in a Beckman Coulter J-20 XP centrifuge at 4'C using a
25.50 rotor. (At this point the GHCl treated samples were dialyzed against water
30
overnight. The dialyzed sample was centrifuged at 15,000 g for 30 minutes.) The
purity of the isolation was then assessed qualitatively by running the preparation on
a NuPage NOVEX 4-12 percent Bis-Tris gel, using MES buffer, LDS sample buffer,
and Mark12 unstained markers (Invitrogen.)
Afterwards the gel was stained with
Simply Blue Safe Stain (Invitrogen) and dried.
A pH method was also explored. In this case all steps were the same except the
cells were incubated in either .2M or .05M Glycine hydrochloride instead of a urea
buffer.
Cell Wall Methods
Cell Breakage Techniques
Two cell breakage techniques were explored in these experiments, both of which
are utilized in the literature in protocols where S layers are isolated. Sonication with a
macroprobe (in a Branson sonifier) is one method. Approximately 1.85g (wet weight)
of cells were resuspended in 30ml of 50mM Tris-HCl buffer pH 7.4 (amount harvested
from 500mL of culture.) Sonication occurred in a 50ml polypropylene centrifuge tube
(Corning) chilled in an ice water bath for 2 minutes at 75 output and a 50 percent
duty cyle as suggested in [26].
Cell breakage can be observed qualitatively by visually examining the pellet after a
10,000g centrifugation. Generally, the broken cells can be separated be separated from
the unbroken ones by gentle vortexing of the pellet or by differential centrifugation (a
spin at 1000-4500g to remove unbroken cells followed by a spin at 10,000g or higher
isolate cell walls.) An alternate method used in these experiments is the French Press.
In this case cells are broken by passing through a small orifice in with a large pressure
differential on either side. The cell suspension was passaged twice through a chilled
French Press (Thermo Spectronic) as suggested in [41] at medium pressure.
Purification Technique
For these experiments the protocol was modified from [26] and [30]. French pressed
or sonicated cells (using the above procedures) were centrifuged at 11,500xg for 15
minutes at 4'C in a Beckman Coulter J-20 XP centrifuge. The cells were resuspended
31
in 30mL 50mM Tris-HCl buffer pH 7.4 containing 0.5 percent Triton X-100 (SurfactAmps X-100, Pierce Biotechnology) (this type of Triton X-100 has been highly purified
and is stored under nitrogen to prevent oxidation.) The Triton X-100 wash is used to
solubilize the plasma membranes which are still attached to the cell walls [41]. The
bottom portion of the pellet containing intact cells (larger for sonicated cells than
for French pressed cells) was discarded and the cell walls retained for further washing
(the two portions were separated by gentle vortexing.) The plasma membranes were
extracted again by resuspension in the Triton X-100 buffer and pelleting by the same
parameters as previously. The cell walls were washed two additional times in 30mL
50mM Tris-HCl buffer pH 7.4. S-layer protein was extracted with 1ml of chaotropic
reagent (5MGHCl or 8M Urea) per gram original biomass.
lml aliquots of each
extraction were placed in polyallomer ultracetrifuge tubes (347287, Beckman Coulter)
and were centrifuged for 15 minutes at 30,000rpm
(
40,000) in a Beckman Coulter
tabletop ultracentrifuge with MLA130 rotor. This step removed the peptidoglycan
layer. Urea extracted S layer protein was run directly on a gel (Novex 4-12 percent
Bis-Tris as described previously.) Next the extracted layers were dialyzed (Slide-ALyzer 10K MWCO Dialysis Cassette) against distilled water first (2 changes of 500ml,
2 hours at room temperature, overnight at 4*C) before proceeding to recrystallization
experiments.
32
4.2
Creating the Recombinant Plasmid
I have successfully engineered a construct which will be utilized for sbpA protein
production. The following section describes the process.
4.2.1
Protocol Overview
The following is the methodology for cloning the gene sbpA into E. coli:
1. Synthesize the standard oligos sbpA28 and sbpA29 for isolating the gene sbpA.
2. Prepare genomic DNA from the organism B. sphaericus ATCC 4525.
3. Isolate the gene using PCR.
4. Subclone the gel purified gene into a TOPO vector.
5. Confirm presence of the sbpA gene using colony PCR.
6. Using purified sbpA TOPO vector, perform the restriction digests.
7. Ligate into digested pET28a vector.
8. Transform into DH5a cells, confirm presence of sbpA, and miniprep the plasmid.
9. Transform the sbpA pET28a vector into BL21(DE3)pLysS cells.
4.2.2
Oligos
Oligo Synthesis
Oligonucleotides were synthesized (394 synthesizer, Applied Biosystems.) Table 5.1
displays the sequences of the synthesized oligos which were obtained from [25]. A
40nmoles CE cycle was used. After synthesis, the oligos were deprotected by heating
in concentrated ammonium hydroxide at 80"C for 1 hour. The liquid was then evaporated at room temperature for 30 minutes. The oligos were dissolved in 1mL TE
buffer (10mM Tris, 1mM EDTA, pH 8.) They were stored at -20"C.
33
Oligo Name
sbpA28
sbpA29
Sequence
5'-CGCGGATCCCATATGGCGCAAGTAAACGACTATAACAAAATC-3'
5'-CGCGGATCCTTATTTTGTAATTGTTACTGTTAATTCAGC-3'
Table 4.1: Synthesized Oligos for PCR
4.2.3
Miniprep of Genomic DNA
The genomic DNA from B. sphaericus ATCC 4525 was prepared using the standard
method of [2]. The concentration and purity of the DNA was quantified using UV
spectroscopy (Hewlett Packard 8453.) The genomic DNA was stored at -20 0C.
4.2.4
PCR of the Gene sbpA
Standard forward (sbpA28) and reverse (sbpA29) primers, were used to PCR the
gene sbpA. PCR was performed utilizing PCR Supermix High Fidelity (Invitrogen)
on a PTS-200 Thermal Cycler with hot lid (MJ Research.) The protocol for PCR was
designed so that the anneal temperature was below the minimum melting temperature of the interaction between the oligos and the gene (50 0C.) The duration of the
extension step was chosen based on the size of the gene. The PCR reactions were run
on a 1 percent agarose gel. The gel was examined utilizing a UV Transilluminator
(UVP.) DNA of the proper length was removed and then extracted using a Qiaquick
Gel Extraction Kit (Qiagen.)
4.2.5
Cloning of the Gene
Subcloning into a TOPO Vector
Using the gel extracted form of sbpA produced by PCR, sbpA was directly cloned
using a TOPO TA Kit (Invitrogen.)
This was accomplished by transforming the
sbpA TOPO vector into competent cells (One Shot DH5a-T1, Invitrogen) which
were grown overnight on kanamycin plates. Colony PCR was performed a number
of colonies to verify the presence of the sbpA insert. For the PCR step of the colony
PCR the original forward and reverse primers were used as was the program for the
original PCR of the gene. Glycerol stocks were made from an assortment of successful
34
transforms to inoculate cultures for subsequent steps. The sbpA TOPO plasmid was
purified using a Fast Plasmid Mini Prep (Eppendorf.)
Restriction Digests
The primers sbpA28 and sbpA29 inserted an NdeI and BamHI restriction site into the
sbpA PCR construct. However, since there is a BamHI site directly in front of the NdeI
site and the NdeI site requires at least a 7 base pair overhang on either side, the NdeI
digestion must be performed first. For the first digestion, miniprepped sbpA TOPO,
NdeI restriction enzyme (New England Biolabs), and Buffer 4 (New England Biolabs)
were combined, incubated for 2 hours at 37 0C, and then the restriction enzyme was
heat inactivated at 80"C. The second digestion was then performed using the NdeI
digested sbpA TOPO vector, BamHI restriction enzyme (New England Biolabs), and
BamHI buffer (New England Biolabs.) Since the BamHI restriction enzyme cannot
be heat inactivated, the DNA was isolated using a Min Elute PCR Purification Kit
(Qiagen) The digestion was confirmed by running the products on an agarose gel.
The pET28a vector (Novagen) is the final destination of the sbpA gene. A double
digest was performed using the pET28a vector, NdeI, BamHI, and BamHI buffer.
The digested pET28a vector was gel purified (Qiaquick Gel Extraction Kit) after
examination with the UV transilluminator.
Creation of the Final Vector and Transformed Cell Line
The digested sbpATOPO and digested gel purified pET28a were ligated using Quick
T4 DNA ligase (New England Biolabs). They were then transformed into DH5a
competent cells. The presence of sbpA was confirmed by colony PCR and the absence of TOPO vector was confirmed by the lack of growth on ampicillin selective
LB plates. Transformed colonies that met both criteria were grown up and their
plasmids miniprepped.
Plasmids of the appropriate length were transformed into
BL21(DE3)pLysS cells. Colony PCR was performed on selected transformed colonies.
Glycerol stocks were made from culture derived from colonies meeting these criteria.
35
Eco R I(7)
Psi 1(8975)
Pst
I 133)
INor
1(8711)
His tag
T7 terminator
f1 origin
kan sequence
Eco R I (7320)
S-layer protein
pET28-SBPA
Pst I(6851)
Xba 1(5744)
Eco R I (5610)
9053
CoEl
pBR322 origin
bP
Xba 1 (5492)
Psit 1 (5474)__
thrombin
His
tag
start codon
rbs
T7 promoter
lac I
Xba 1(5233)
lac operator
Figure 4-2: Map of plasmid sbpA-pET28a vector which codes for the protein Sbpa
with an n-terminus hexahistidine tail.
36
4.3
4.3.1
Purifying the Recombinant Form
Strategy
The sbpA-pET28a vector produces both soluble histidine tagged protein and inclusion
bodies. The form of the protein leads to different purification strategies.
4.3.2
Inclusion Body Based Preparation
The cells were broken and the inclusion bodies solubilized according to [5, 6, 7] with
the exceptions described below. In
[5]
the cells were resuspended in 30ml of Buffer
A and 0.2 percent DOC was replaced with 0.1 percent Triton X-100 (Surfact-Amps
X-100, Pierce Biotechnology.) In [6] the centrifuge used was a Beckman Coulter J-20
XP. In [7] the 0.2 percent DOC was replaced with 0.1 percent Triton X-100, a Dremel
with stainless steel drill attachment was used instead of a Tissue-Tearor, and instead
of adding a mixture of buffer A and sarkosyl either buffer A containing 5M GHC1 or
8M urea was used.
4.3.3
Affinity Based Preparations
For affinity based purifications the specificity of the engineered hexahistidine tag
for Ni-NTA is exploited. In these procedures, Ni-NTA Agarose (Qiagen) is utilized
because it allows protein binding to occur in bulk solution (allowing the protein
to have access to all binding sites on the agarose beaks), whereas elution can be
performed in a column (allowing protein to be concentrated as it elutes in the disk
shaped step gradient between different concentrations of imidizole.) For this method
we use gravity flow columns (Biorad) because they allow one to minimize dilution
(which occurs in more automated setups.)
Limitations of Approach for Ni-NTA Affinity Based Purification
The sbpA-pET28a vector was designed in order to simplify production and purification
of the recombinant form of sbpA. However, in choosing this vector design a number
37
of tradeoffs were made. The sbpA-pET28a vector utilizes the hexahistadine tag built
into the pET28a vector, meaning that it did not need to be created and built into
the construction.
Hexahistadine tags are the standard protein form used for Ni-
NTA affinity purifications. However, increasing the number of histidines in the tag to
seven allows for purer protein isolations because it allows washes and elution at higher
imidazole concentrations [1]. Standard hexahistadine Ni-NTA purifications can allow
up to 95 percent pure isolations. Higher imidazole concentrations remove natural
protein contaminants that bind to the column. Using more than seven histidines
makes the protein difficult to elute from the column [1].
A n-terminal histidine
tag was chosen rather than a c-terminal one. As shown in the experiments the nterminal histidine tag is accessible for purification of the soluble fraction of the protein.
Soluble fractions are more likely to be folded correctly and contain active forms of
the protein. Solubilization by denaturing reagents of insoluble forms and purification
under denaturing conditions is also a possibility. A limitation of this purification
technique is that if prematurely terminated proteins are present, they will be purified
with the intact protein but may act as contaminants in subsequent crystallization
processes. In the literature, is has been determined that the C-terminal is not located
on the surface. If the C-terminus was chosen as the location for the histidine tag,
soluble protein purifications would be unlikely to work. Denaturation based methods
most likely would be required so that that the histidine tag would be exposed to the
column. Also, the hidden c-terminal tag may affect protein folding.
Soluble Fraction
For standard soluble protein affinity based purifications (which have not yet been
modified to include the S layer-specific procedures), the protocols [5, 6] with the
modifications given above are used. However there are several additional changes.
Since the protein will be purified on a Ni-NTA column which EDTA strips, Buffer A'
(which is made without EDTA) is used instead of Buffer A. The supernatant produced
in [6] is then subject to affinity purification with the Ni-NTA beads.
The following protocol is that used in Sue-Hwa Lin's module at the Cold Spring
38
Harbor Protein Purification Course except the phosphate buffer has been replaced by
a Tris buffer. The Ni-NTA beads are washed in column with ten volumes Washing
Buffer (Buffer A'=50mM Tris-HCl, pH=7.9, 50mM NaCl, 5 percent glycerol + 250mM
NaCl + 25mM Imidazole + 0.1 percent Triton X-100). Afterwards, the beads are
transferred to a 50ml tube. The supernatant is transferred to the 50ml tube. The
beads and the supernatant are mixed gently by shaking in an ice water bath for 1-2
hours. Briefly spin down the beads at 1g or let them settle due to gravity and save
the flow through for a protein assay. Wash the beads with Washing Buffer twice. The
first time use a Washing Buffer that includes 0.1 percent Triton X-100. The second
time use a Washing Buffer that does not contain Triton X-100 (as it will interfere
with the Bradford assay.) The beads are loaded into the column. Stepwise elution
is then performed with an increasing amount of Imidazole in the Washing Buffer
(without Triton X-100). After each fraction is collected there is a 1-2 minute pause
in order to achieve good equilibration. The column should be eluted by 6-8 fractions
(half the bead volume) each of Washing Buffer containing the following Imidazole
concentration (50mM, 100mM, 250mM) and lacking Triton X-100. If desired there
can be a final elution with EDTA to confirm that all of the protein has been eluted.
Inclusion Body Fraction
Affinity based purification of the inclusion body fraction is a combination of the
inclusion based purification and the affinity purification of the soluble fraction. In
this case the inclusion based purification is performed to completion except Buffer A
is replaced with Buffer A' (though if desired Buffer A can be utilized in [5] in order
to produce a cleaner inclusion body preparation.) At the end of procedure [7], the
inclusion bodies are solubilized in a chaotropic buffer, either Buffer A'G (Buffer A'
containing 5M GHCl) or Buffer A'U (Buffer A' containing 8M urea.)
At this point the affinity purification protocol given above can be used except
Buffer A'G or Buffer A'U should be substituted for Buffer A'.
39
Further Preparations for Crystallization
If affinity purification with Ni-NTA beads is used before crystallization it may be
necessary to concentrate the protein containing fractions by diafiltration before subsequent steps as the beads have a maximum capacity of 5-10mg/ml and there is likely
to be some dilution in the elution procedure.
40
Chapter 5
Results
5.1
5.1.1
Purification of the Native Form
Whole Cell Method
Figures 5-1 and 5-2 compare the urea and GHCl whole cell methods. The GHCl was
somewhat harder to work with because the samples needed to be treated (in this case
dialyzed before the gel could be run.) The GHC1 solution is diluted due to dialysis.
The pH dissociation method produced lower yields of the protein than the chaotropic
methods (see Figure 5-3.)
5.1.2
Cell Wall Methods
Sonication using the parameters specified in Chapter 4 did not break all cells, especially in the case of gram positive cells of B. sphaericus which are not susceptible to
lysozyme (as opposed to E. coli.)
The pellet from whole B. sphaericus ATCC 4525 cells will generally appear grayish. After sonication the pellet generally looks like a bull's eye (see Figure 5-4), with
a grayish area in the center (this is composed of unbroken cells which are more dense)
surrounded by a whiter area. The whiter area is composed of cell walls from broken
cells.
Approximately 50 to 75 percent of cells were broken using the sonication procedure
41
200kDa -- )
-- Theoretical Molecular
116.3kDa --- P
Mass = 132kDa
Figure 5-1: Whole Cell Urea Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lanes 2 and 3, sample heated to 70"C. Lanes 4 and 5, sample heated to
1000C.
1
200kDa
116.3kDa
-
)P
2
3
45
Theoretical Molecular
Mass = 132kDa
Figure 5-2: Whole Cell GHC1 Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lane 2, dialysis pellet A. Lane 3, dialysis supernatant A. Lane 4, dialysis pellet
B. Lane 5, dialysis supernatant B.
42
1
200kDa
116.3kDa
234
5
Theoretical Molecular
Mass = 132kDa
10
Figure 5-3: Whole Cell pH Method. Lane 1, Mark12 Unstained Standard (Invitrogen.) Lane 2, blank. Lane 3, .2M Glycine HCl. Lane 4, blank. Lane 5, .05M Glycine
HCl.
Bull's Eye Pellet
Cell Walls
Unbroken Cells
Figure 5-4: Bull's eye like patterns appear in cell lysate pellets when some but not
all cells are broken.
43
described. The sonication sheared the genomic DNA (in fact, in some protocols for
gram negative bacteria it is used solely for this purpose rather than for cell breakage.)
Treatment with a French Press produces uniform breakage, resulting in a mostly
white pellet. The pellets from French Pressed cells were viscous, indicating that a
large percentages of genomic DNA remained intact following the treatment.
Purification Technique
French pressed cells produced very pure preparations because few whole cells were
present, and thus the contaminating interior and exterior proteins could be washed
away. For sonication procedures it was more challenging to separate the whole cells,
which could lead to some impurities. GHC preparations must be dialyzed, diluted, or
the protein precipitated by trichloroacetic acid before they can be run on SDS-PAGE
gels.
5.2
Creating a Recombinant Plasmid
Forward and reverse sequencing indicated that the specified plasmid was created. In
addition, PCR reactions always generated linear DNA of the expected length.
5.3
Purifying the Recombinant Form
In the gel shown in Figure 5-5, the native and recombinant forms are shown side by
side.
5.3.1
Inclusion Body Based Preparation
The inclusion bodies contained the majority of sbpA produced. Optimally, recombinant protein can be isolated from inclusion bodies at 95 percent purity.
44
3
12
Native
---
v
-
4
5
Recombinant
."
Figure 5-5: Comparison of Native and Recombinant Protein. Lanes 1 and 2, sbpA
from whole cell urea method. Lane 3, Mark12 Unstained Standard (Invitrogen.)
Lanes 4 and 5, SDS extracted whole cells containing the sbpA/pET28a plasmid which
were induced for 3 hours.
5.3.2
Affinity Based Preparations
Soluble Fraction
Figure 5-6 shows a standard soluble fraction imidazole elution from a Ni-NTA column.
The purest protein elutes in the 2nd and 3rd fraction of the 250mM imidazole series.
Some of the impurities that remain at this point may be truncated protein that
contains a histidine tag.
Inclusion Body Fraction
Inclusion body based purification produces sbpA at high purity. The use of inclusion
bodies as a starting material acts as an additional purification step that precedes the
affinity purification step.
45
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 1718
Figure 5-6: Affinity Purification of Soluble Fraction. Lanes 1 through 5, fractions
1 through 5 of 50mM imidazole elution. Lanes 6, 7, 12, and 13, Mark12 Unstained
Standard (Invitrogen.) Lanes 8 through 12, fractions 1 through 5 of 100mM imidazole
elution. Lanes 14 through 18, fractions 1 through 5 of 250mM imidazole elution.
46
Chapter 6
Discussion and Conclusions
6.1
6.1.1
Purification of the Native Form
Chaotropic Reagents
The best way to remove cyanate ions from urea buffers is by running the solution
through mixed bed ion exchange resin. I will use this method in the future.
6.1.2
Whole Cell Method
The success of using chaotropic reagents for these experiments depended upon the
extent that they affected the integrity of the cells-which would have the effect of
releasing protein contaminants. Alternate chemicals can be used for whole cell purification if they share the property of maintaining cell integrity. In this case an analytic
gel should be run before proceeding to crystallization steps in order to ensure that
the purity of the native sbpA is above 95 percent. Also, if the protein is too dilute,
the sample may need to be concentrated and refolded.
Though it was not utilized in these experiments, a method for quantitative staining
and scanning of SDS gels is described in [8]. I plan to utilize that method in the future.
pH dissociation methods were less efficient than chaotrope based methods because
they may only allow the harvest of S-layers which have been released from cells prior to
experimentation, a process that occurs naturally to some degree in late logarithmic
47
phase. The reference [20] suggests that for a different strain of the same species
(NTCC 9602) the addition of pH lowering reagents can disrupt the structure of Slayer proteins, though it does not release them from the peptidoglycan layer, and
thus from the natural binding partner of the S-layer protein-most likely a cell wall
polymer attached to the peptidoglycan layer [23].
Urea buffer was the simplest for experimentation. The whole cell method was fast,
though some impurities were present. In future crystallization trials I will determine if
this level of purity is sufficient. If it is, this will be my method of choice for producing
native protein as a control reagent.
6.1.3
Cell Wall Methods
Improving the efficiency of sonication as a cell breakage technique is possible though
the tradeoffs must be considered. Increasing the duration and intensity of the sonication, for example 80-95 output for 6 to 7.5 minutes as was performed for a different
strain of the same species [19], while continuing to cool in an ice water bath can increase the fraction of cells that are broken, though some will remain unbroken. Harsh
sonication can heat the sample and possibly shear the protein. Milder sonication
conditions produce larger cell wall fragments whereas harsher conditions increase the
yield [26] of broken cells.
The primary advantages of sonication are that it is a relatively fast technique and
that it shears DNA.
Though the yield of broken cells is higher, treatment with a French Press takes
much longer than sonication. For B. sphaericus breakage of an equal sized starting
pellet using a French Press provides a higher yield of protein because fewer unbroken cells are present and need to be removed for further processing. For very pure
protein preparations, shearing or removal of the DNA remaining in French pressed
preparations is recommended (to remove protein that is either trapped in the DNA
or present due to affinity.)
The optimal procedure for highest purity and yield of protein for Cell Wall type
purifications is probably a combined method in which the French Press procedure is
48
utilized first to break the cells and this is followed by a light sonication procedure
which causes the genomic DNA to be sheared and removed with the supernatant.
For example 3 x 20s, on 20 percent power with the macrotip should accomplish the
objective.
Purification Technique
For exploring crystallization, we would like to have standard methods for predicting protein concentration before performing assays. Typically for crystallization
protocols protein solution contains 2.5 mg/ml protein after dialysis against distilled
water and removal of aggregated protein by centrifugation [31]. Thus it probably
contains 10 to 20mg/ml before dialysis. Approximately 16 percent of wet weight cells
(biomass) is protein (for E. coli.) According to [4, 39], S-layer proteins constitute
between 5 to 15 percents of total protein of S-layer producing cells when they are in
the exponential phase. This means that a minimum of 8 percent of the wet weight
biomass is S-layer protein provided that the culture has not been overgrown with
non-S-layer producing cells. Thus in order to calculate the amount of denaturant
the wet weight (g) of the starting pellet should be multiplied by .08 (8 percent) and
divided by .010 g/ml.
If on average a .95g wet weight pellet is produced per liter of culture, then the
30ml of urea buffer used to solubilize the S layer protein from pellet derived from 4
liters of culture in the following reference [19] is in line with this order of magnitude
calculation.
6.2
Creating a Recombinant Plasmid
Mass spectroscopy of purified sbpA will demonstrate that the intended protein is in
fact created from this plasmid and that no additional problems exist downstream.
49
6.3
6.3.1
Purifying the Recombinant Form
Inclusion Body Based Preparation
This procedure differs from the procedure cited in [27] and referenced in [25]. This
procedure described in this thesis utilizes a lysozyme present in the cell line and a
freeze thaw cycle to break the cell wall. The two procedures will be compared to
verify that the freeze cycle does not harm the sbpA and the inability to completely
isolate the peptidoglycan does not create interference with competing sbpA assembly
substrates.
This method will probably become the basis for preparative sbpA production,
although these steps will be followed by gel filtration chromatography, lyophilization
of the protein, and refolding at high concentration.
6.3.2
Affinity Based Preparations
The Tris buffers used for these purifications are not optimal because they contain
primary amines. They were selected in this case because phosphate forms an insoluble
precipitate with calcium, which is the cofactor of sbpA that causes crystallization of
the tetramers. In the future a buffer that does not contain primary amines such
as MOPS or HEPES will be used. The protein gel demonstrates that this procedure
produces high purity sbpA, although truncated forms remain as impurities (which may
be a problem for crystallization.) Thus, this technique does not provide significant
advantages over an inclusion body based preparation followed by gel filtration.
50
Chapter 7
Impact and Future Work
This project has allowed me to assess the feasibility of converting a natural S-layer
protein into an engineering building block. Though the protein can be reliably harvested at high purity, there is still much to be done. The next critical steps are to
establish reliable conditions for crystallization and observation. If these steps are
completed satisfactorily, then three challenges arise:
* creating a set of alternate lateral junctions which can be used in register to
perform logic so that the tiles can act as boundary and rule tiles
e creating and positioning a set of vertical linkers that can bind to inorganic
substrates or other substances above and below the subunit so that real circuit
formation is a possibility
o creating rules for the protein sequence design so that the alternate DNA sequences can simply be switched out combinatorially as the design dictates.
51
52
Appendix A
Winfree's Work
A.1
Theory
Winfree has explored the idea of programmed self-assembly as a method for building
non-periodic structures in two dimensions. His work utilizes the results of Wang's
explorations of the tiling problem, which showed that two dimensional tilings can
lead to Turing-universal systems [50, 51].
Winfree's models provide pathways for
building complex logic into the process of self-assembly, allowing one to consider
constructions that could feasibly form scaffolds for an array of digital circuits (see
Figure A-1) and other patterns [10]. Typically his constructions consist of two types
of tiles: boundary tiles and rule tiles. Boundary tiles (sometimes called input tiles, as
they are the first assembled and determine the resulting calculation) and the corner
seed tile generally define the initial conditions for the computation and the physical
boundaries of its extent, whereas rule tiles complete the calculation-thus forming the
bulk of the pattern. Some examples of boundary and rule tiles are shown in the circuit
patterns (see Figure A-1) and the Sierpinski triangle tiling (see Figure A-2). Winfree's
abstract Tiling Assembly Model is Turing-universal, meaning that only small numbers
of tiles are required for complex assembly formation [37]. Typically the threshold for
a tile addition is set for two in this model, i.e., two strong interactions must occur
at one position between a tile and an existing structure for the tile to assume that
position. An illustration of this model is shown in Figure A-2, part b). A kinetic
53
-
., -f
-
-
-
- jl
.
;-
69
909
9
-
.Zq-q 9!=
-
,
9
-
-
-
-
--
-
-
--
--
9
-
4~b)
(a)
(a)
P
0-9
-j
4
W-8
E EE
-9
-9
-l
-J
-5
-
-I
I
-J
-- I
WiSE
-D
-9
-3
D-
pig
6
-I
Figure A-1: Self-Assembled Circuit Patterns a) Rule tiles, boundary (input and seed)
tiles, and resulting demultiplexer based on a counter type pattern. b) Addressable
memory array utilizing two demultiplexers. Image adapted from [10]
boundavylks
t
ca3
strcnith-2 (rong) bond
output
iniput 'K
0+"=) 9+1= 1+01l1
bond
slngth-Inullk
(c)
1b)
IT2
MII
Vr
Figure A-2: Sierpinski triangle assembly. a) Rule tiles (which perform the logic by
implementing an XOR operation) and boundary tiles (which includes the corner tile)
b) Process of growth using the two or more strong interaction rule from the abstract
Tiling Assembly Model. c) Rates of tile association and disassociation using the
kinetic Tiling Assembly Model. d) Various stages of assembly. Image adapted from
[54]
54
Tile Assembly Model considers the implications of reversible chemistry: addition of
tiles is proportional to free monomer concentration and dissociation is exponentially
related to bond strength. An illustration of the implications of this model is shown
in Figure A-2, part c). In this case complex self-assemblies are expected to form if
the following conditions are met, when the growth begins from a seed corner tile:
* monomer tile concentrations remain constant
" assemblies grow by single tile addition
" assembly growth is independent (no assembly fusions) [37].
Together these systems provide standard models and patterns that are highly applicable for future explorations in this field, regardless of manner in which the tile is
physically implemented.
A.2
Practice
Winfree has created tiles out of DNA, using a double crossover type motif, with
tiles approximately 2x4x10-16 nm3 in size [55].
He has produced striped lattices
consisting of repeating patterns of two to four units each (see Figure A-3.) Typically
the largest assemblies are on the order of 2x8 um 2 [55]. Thus far, no well-defined
physical algorithmic assemblies on the order of the complexity of the simulations
(such as counter assemblies, Sierpinski triangles, and Hadamard matrices) have been
formed [37]. Currently, creating the boundary tiles assemblies which are necessary to
provide the starting conditions for complex assembly formation remains a challenge,
primarily because the constraints of the kinetic Tiling Assembly Model are challenging
to achieve in physical reality. The primary problem is that intermediate assemblies
can form in the absence of corner tiles, meaning that assembly often does not grow
by single tile addition in a programmed fashion and tile monomers are expended in
non-programmed assemblies. For programmed self-assembly, perfect boundaries must
be created either before or simultaneously with rule tile assembly in order to serve as
the inputs for secondary two-dimensional growth of a patterned array.
55
a)
A
A
A
A
b)
A
jA
A
AB
A
A
A
AB
A
BAA
A
-
"~
AA
I
IBI
C DAB
A
C
CQ
A
C
B-g
I "D
C
C
A
A
CA
13A4
c)
)
DAE-O AB lattice
Periodicity: 33+/-2nm
DAE-O ABCD lattice
Periodicity: 66+/-5nm
Figure A-3: DNA Tiles a) Colored edge model of DX DNA tile striped assemblies.
b) Model structures of different type A units. c) Assembly lattice topologies of different tile forms. d) Physical realizations of AB and ABCD forms, which are shown
schematically in part a), measured with fluid cell AFM (the scale bars are 300nm.)
Image adapted from [55]
56
Bibliography
[1] Personal communication from Sue-Hwa Lin.
[2] F.M. Ausubel, R. Brent, R.E. Kingston, D.D. Moore, J.G. Seidman, J.A. Smith,
and K. Struhl, editors. Short Protocols in Molecular Biology, chapter Miniprep
of Bacterial Genomic DNA, pages 2-12. Wiley, 4 edition, 1999.
[3] T.J. Beveridge, P.H. Pouwels, M. Sara, A. Kotiranta, K. Lounatmaa, K. Kari,
E. Kerosuo, M. Haapasalo, E.M. Egelseer, I. Schocher, U.B. Sleytr, L. Morelli,
M.L. Callegari, J.F. Nomellini, W.H. Bingle, J. Smit, E. Leibovitz, M. Lemaire,
I. Miras, S. Salamitou, P. Beguin, H. Ohayon, P. Gounon, M. Matuschek, and
S.F. Koval. Functions of S-layers. FEMS microbiology reviews, 20(1-2):99-149,
1997.
[4] R.D. Bowditch, P. Baumann, and A.A. Yousten.
Cloning and sequencing of
the gene encoding a 125-kilodalton surface-layer protein from bacillus sphaericus
2362 and of a related cryptic gene. Journal of Bacteriology, 171(8):4178-4188,
1989.
[5] R.R. Burgess and M.W. Knuth. Strategies for Protein Purificationand Characterization: A Laboratory Course Manual, chapter Breakage of Cells by Sonication, page 213. Cold Spring Harbor Laboratory Press, 1996.
[6] R.R. Burgess and M.W. Knuth. Strategies for Protein Purificationand Characterization: A Laboratory Course Manual,chapter Preparation of Inclusion Bodies
and Soluble Extract, page 213. Cold Spring Harbor Laboratory Press, 1996.
57
[7] R.R. Burgess and M.W. Knuth. Strategies for Protein Purificationand Characterization: A Laboratory Course Manual, chapter Solubilization of the Inclusion
Body with Sarkosyl, pages 214-215. Cold Spring Harbor Laboratory Press, 1996.
[8] R.R. Burgess and M.W. Knuth. Strategies for Protein Purificationand Characterization: A Laboratory Course Manual, chapter Quantitative SDS Gel Staining
and Scanning, pages 229-230. Cold Spring Harbor Laboratory Press, 1996.
[9] American Type Culture Collection. Product information sheet for ATCC 4525.
P.O. Box 1549, Manassis VA 20108, www.atcc.org, Dec. 2002.
[10] M. Cook, P.W.K. Rothemund, and E. Winfree. Self-assembled circuit patterns.
Lecture Notes in Computer Science, 2943:91-107, 2004.
Formation of a gold superlattice on
[11] S. Dieluweit, D. Pum, and U.B. Sleytr.
an S-layer with square lattice symmetry. Supramolecular Science, 5(1-2):15-19,
1998.
[12] E.M. Egelseer, T. Danhorn, M. Pleschberger, C. Hotzy, U.B. Sleytr, and M. Sara.
Characterization of an S-layer glycoprotein produced in the course of S-layer
variation of Bacillus stearothermophilusATCC 12980 and sequencing and cloning
of the sbsD gene encoding the protein. Journal of Bacteriology, 177(1):70-80,
2001.
[13] C.E. Flynn, C.B. Mao, A. Hayhurst, J.L. Williams, G. Georgiou, B. Iverson,
and A.M. Belcher. Synthesis and organization of nanoscale II-VI semiconductor
materials using evolved peptide specificity and viral capsid assembly. Journal of
Materials Chemistry, 13(10):2414-2421, 2003.
[14] National
Center
for Biotechnology
Information.
surface layer protein precursor (sbpA)
gene,
Bacillus sphaericus
complete cds.
Entrez,
http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?db=nucleotideval=6665711.
58
[15] National
face
layer
Center
protein
for
Biotechnology
precursor
Bacillus
Information.
Sur-
sphaericus.
Entrez,
http://www.ncbi.nlm.nih.gov/entrez/viewer.fcgi?db=proteinval=6665712.
[16] P.C. Gufler, D. Pum, U.B. Sleytr, and B. Schuster. Highly robust lipid membranes on crystalline S-layer supports investigated by electrochemical impedance
spectroscopy. Biochimica et Biophysica Acta-Biomembranes, 1661(2):154-165,
2004.
[17] E.S. Gy6rvary, A. O'Riordan, A.J. Quin, G. Redmond, D. Pum, and U.B. Sleytr.
Biomimetic nanostructure fabrication: Nonlithographic lateral patterning and
self-assembly of functional bacterial S-layers at silicon supports. Nano Letters,
3(3):315-319, 2003.
[18] E.S. Gy6rvary, 0. Stein, D. Pum, and U.B. Sleytr. Self-assembly and recrystallization of bacterial S-layer proteins at silicon supports imaged in real time by
atomic force microscopy. Journal of Microscopy, 212(3):300-306, 2003.
[19] A.T. Hastie and C.C. Brinton, Jr. Isolation, characterization, and in vitro assembly of the tetragonally arrayed layer of Bacillus sphaericus. Journal of Bacteriology, 138(3):999-1009, 1979.
[20] A.T. Hastie and C.C. Brinton, Jr. Specific interaction of the tetragonally arrayed
protein layer of Bacillus sphaericus with its peptidoglycan sacculus. Journal of
Bacteriology, 138(3):1010-1021, 1979.
[21] L. Howard and D.J. Tipper. A polypeptide bacteriophage receptor: modified cell
wall protein subunits in bacteriophage-resistant mutants of Bacillus sphaericus
strain P-1. J. Bacteriology, 113(3):1491-1504, 1973.
[22] Y. Hu, S. Chen, M. Xu, and S. Zhang. An improved, inexpensice procedure for
the large-scale purification of recombinant human erythropoietin. Biotechnology
and Applied Biochemistry, 40:89-94, 2004.
59
[23] N. Ilk, P. Kosma, M. Puchberger, E.M. Egelseer, H.F. Mayer, U.B. Sleytr, and
M. Sara. Structural and functional analysis of the secondary cell wall polymer of
Bacillus sphaericus CCM2177 that serves as an S-layer-specific anchor. Journal
of Bacteriology, 181(24):7643-7646, 1999.
[24] N. Ilk, S. Kiipcii, G. Moncayo, S. Klimt, R.C. Ecker, R. Hofer-Warbinek, E.M.
Egelseer, U.B. Sleytr, and M. Sara. A functional chimaeric S-layer-enchanced
green fluorescent protein to follow the uptake of S-layer-coated liposomes into
eukaryotic cells. Biochemical Journal,379(2):441-448, 2004.
[25] N. Ilk, C. Vollenkle, E.M. Egelseer, A. Breitwieser, U.B. Sleytr, and M. Sara.
Molecular characterization of the S-layer gene, sbpA, of Bacillus sphaericus
CCM2177 and production of a functional S-layer fusion protein with the ability
to recrystallize in a defined orientation while presenting a fused allergen. Applied
and Environmental Microbiology, 68(7):3251-3260, 2002.
[26] B. Jahn-Schmid, M. Graninger, M. Glozik, S. Kfipcii, C. Ebner, F.M. Unger,
U.B. Sleytr, and P. Messner. Immunoreactivity of allergen (bet v 1) conjugated
to crystalline bacterial cell surface layers (S-layers). Immunotechnology, 2:103113, 1996.
[27] M. Jarosch, E.M. Egelseer, C. Huber, D. Moll, D. Mattanovich, U.B. Sleytr, and
M. Sara. Analysis of the structure-function relationship of the S-layer protein
SbsC of Bacillus stearothermophilusATCC 12980 by producing truncated forms.
Microbiology, 147:1353-1363, 2001.
[28] J. Leupault, N. Martin, and K. Leonard. Three-dimensional structure of the Tlayer of Bacillus sphaericus P-1. Journal of Bacteriology, 168(1):303-308, 1986.
[29] M. Pleschberger, A. Neubauer, E.M. Egelseer, S. Weigert, B. Lindner, U.B.
Sleytr, S. Muyldermans, and M. Sara. Generation of a functional monomolecular
protein lattice consisting of an S-layer fusion protein comprising the variable
domain of a camel heavy chain antibody. Bioconjugate chemistry, 14(2):440448, 2003.
60
[30]
D. Pum and U.B. Sleytr. Large-scale reconsititution of crystalline bacterial surface layer proteins at the air-water interface and on lipid films. Thin Solid Films,
244:882-886, 1994.
[31] D. Pum and U.B. Sleytr. Anisotropic crystal growth of the S-layer of Bacillus sphaericus CCM 2177 at the air/water interface. Colliods and Surfaces A:
Physiochemical and Engineering Aspects, 102:99-104, 1995.
[32] D. Pum, G. Stangl, C. Sponer, W. Fallmann, and U.B. Sleytr. Deep UV patterning of monolayers of crystalline S-layer protein on silicon surfaces. Colloids
and Surfaces B-Biointerfaces, 8(3):157-162, 1997.
[33] Qiagen, http://www.qiagen.com. Ni-NTA Spin Handbook: For Ni-NTA Spin Kit
and Ni-NTA Spin Columns, February 2003).
[34] B.D. Reiss, C.B. Mao, D.J. Solis, K.S. Ryan, T. Thomson, and A.M. Belcher.
Biological routes to metal alloy ferromagnetic nanostructures.
Nano Letters,
4(6):1127-1132, 2004.
[35] M. Sara, B. Kuen, H.F. Mayer, F. Mandl, K.C. Schuster, and U.B. Sleytr.
Dynamics in oxygen-induced changes in S-layer protein synthesis from Bacillus stearothermophilusPV72 and the S-layer deficient variant T5 in continuous culture and studies of the cell wall composition. Journal of Bacteriology,
178(7):2108-2117, 1996.
[36] M. Sara, K. Moser-Thier, U. Kainz, and U.B. Sleytr. Characterization of S
layers from mesophilic bacillaceae and studies on their protective role towards
muramidases. Archives of Microbiology, 153:209-214), 1990.
[37] R. Schulman, S. Lee, N. Papadakis, and E. Winfree. One dimensional boundaries
for DNA tile self-assembly. Lecture Notes in Computer Science, 2943:108-125,
2004.
61
[38] W. Shenton, D. Pum, U.B. Sleytr, and S. Mann. Synthesis of cadmium sulphide
superlattices using self-assembled bacterial S-layers. Nature, 389(6651):585-587,
1997.
[39] M.S. Sidhu and I. Olsen. S-layers of Bacillus species. Microbiology, 143:10391052, 1997.
[40] U.B. Sleytr and T.J. Beveridge. Bacterial S-layers.
Trends in Microbiology,
7(6):253-260, 1999.
[41] U.B. Sleytr and A.M. Glauert. Ultrastructure of the cell walls of two closely related Clostridia that possess different regular arrays of surface subunits. Journal
of Bacteriology, 126(2):869-882, 1976.
[42] U.B. Sleytr and R. Plohberger. Electron Microscopy at Molecular Dimensions:
State of the Art and Strategies for the Future, chapter The Dynamic Process of
Assembly of Two-Dimensional Arrays of Macromolecules on Bacterial Cell Walls,
pages 36-47. Springer-Verlag, 1980.
[43] U.B. Sleytr and M. Sara. Bacterial and archaeal S-layer proteins: structurefunction relationships and their biotechnological applications. Trends in Biotechnology, 15(1):20-26, 1997.
[44] U.B. Sleytr, M. Sara, Z. Kiipcii, and P. Messner. Structural and chemical characterization of S-layers of selected strains of Bacillus stearothermophilus and
Desulfotomaculum nigrificans. Archives of Microbiology, 146:19-24, 1986.
[45] U.B. Sleytr, M. Sara, P. Messner, and D. Pum. 2D Protein crystals as patterning
elements and building blocks in nanotechnology. http://www.bit.ac.at.
[46] U.B. Sleytr, M. Sara, D. Pum, and B. Schuster. Characterization and use of
crystalline bacterial cell surface layers. Progress in Surface Science, 68:231-278,
2001.
62
[47] N.R. Smith, R.E. Gordon, and F.E. Clark. Aerobic sporeforming bacteria, volume 16 of Agriculture monograph. U.S. Department of Agriculture, 1952.
[48] G.R. Stark. Reactions of cyanate with functional groups of proteins. 3. reactions
with amino and carboxyl groups. Biochemistry, 4(6):1030-1036, 1965.
[49] R. Wahl, J. Raff, S. Selenska-Pobell, M. Mertig, and W. Pompe. A fast screening method for surface layers on Gram-positive bacteria. Biotechnology letters,
23:1485-1490, 2001.
[50] H. Wang. Proving theorems by pattern recognition. II. Bell System Technical
Journal, 40:1-42, 1961.
[51] H. Wang. Proceedings of the Symposium on the Mathematical Theory of Autonoma, chapter Dominoes and the AEA case of the decision problem, pages
23-55. Polytechnic Press, Brooklyn, New York, 1963.
[52] S.R. Whaley. Selection of peptides for binding semiconductor and magnetic materials for the purpose of organizing nanoscales materials. PhD thesis, The University of Texas at Austin, 2001.
[53] S.R. Whaley, D.S. English, E.L. Hu, and A.M. Belcher. Selection of peptides
with semiconductor binding specificity for directed nanocrystal assembly. Nature,
405(6786):665-668, 2000.
[54] E. Winfree and R. Bekbolatov. Proofreading tile sets: Error correction for algorithmic self-assembly. Lecture Notes in Computer Science, 2943:126-144, 2004.
[55] E. Winfree, F. Liu, L.A. Wenzler, and N.C. Seeman. Design and self-assembly
of two-dimensional DNA crystals. Nature, 394(6693):539-544, 1998.
63