Electrostatic control of thickness and stiffness in a designed protein fiber.

Abstract

Attempts to design peptide-based fibers from first principles test our understanding of protein folding and assembly, and potentially provide routes to new biomaterials. Several groups have presented such designs based on alpha-helical and beta-strand building blocks. A key issue is this area now is engineering and controlling fiber morphology and related properties. Previously, we have reported the design and characterization of a self-assembling peptide fiber (SAF) system based on alpha-helical coiled-coil building blocks. With preceding designs, the SAFs are thickened, highly ordered structures in which many coiled coils are tightly bundled. As a result, the fibers behave as rigid rods. Here we report successful attempts to design new fibers that are thinner and more flexible by further programming at the amino-acid sequence level. This was done by introducing extended, or "smeared", electrostatic networks of arginine and glutamate residues to the surfaces of the coiled-coil building blocks. Furthermore, using arginine--rather than lysine--in these networks plays a major role in the fiber assembly, presumably by facilitating multidentate intra and intercoiled-coil salt bridges.

Full text

Page 1

Electrostatic Control of Thickness and Stiffness in a Designed
Protein Fiber
David Papapostolou,†Elizabeth H. C. Bromley,†Christopher Bano,†and
Derek N. Woolfson*,†,‡
School of Chemistry, UniVersity of Bristol, Cantock’s Close, Bristol BS8 1TS, U.K., and
Department of Biochemistry, UniVersity of Bristol, UniVersity Walk, Bristol BS8 1TD, U.K.
Received October 12, 2007; E-mail: D.N.Woolfson@bristol.ac.uk
Abstract: Attempts to design peptide-based fibers from first principles test our understanding of protein
folding and assembly, and potentially provide routes to new biomaterials. Several groups have presented
such designs based on R-helical and ?-strand building blocks. A key issue is this area now is engineering
and controlling fiber morphology and related properties. Previously, we have reported the design and
characterization of a self-assembling peptide fiber (SAF) system based on R-helical coiled-coil building
blocks. With preceding designs, the SAFs are thickened, highly ordered structures in which many coiled
coils are tightly bundled. As a result, the fibers behave as rigid rods. Here we report successful attempts
to design new fibers that are thinner and more flexible by further programming at the amino-acid sequence
level. This was done by introducing extended, or “smeared”, electrostatic networks of arginine and glutamate
residues to the surfaces of the coiled-coil building blocks. Furthermore, using argininesrather than lysinesin
these networks plays a major role in the fiber assembly, presumably by facilitating multidentate intra and
intercoiled-coil salt bridges.
Introduction
The de novo design of bioinspired fibrous materials is
challenging, but has potential applications in nanobiotechnology,
which range from ordered arrays for nanoelectronics to three-
dimensional (3D) scaffolds for tissue engineering.1–3Indeed,
considerable effort has been made to develop new self-
assembling systems based on nucleic acids, peptides, and
proteins.1–4Through this work, impressive arrays of discrete
nanoscale objects and nano-to-mesoscale biomaterials have been
constructed. However, and particularly for peptide- and protein-
based materials, programming and controlling the morphology
(shape, size, and related physical properties) of such materials
has proven difficult.
Regarding peptide-based fibrous assemblies,1–3many of these
are based on ?-structured peptide assemblies, or amyloid-like
structures,1,5,6and will not be discussed in detail in this paper.
The R-helix has also proven to be a valuable building block, or
tecton,7for constructing fibrous biomaterials de novo, and this
has spawned a rapidly growing subfield;8–14this area has been
reviewed recently.3The leucine-zipper motif, which is a specific
type of dimeric R-helical coiled coil, has been particularly
exploited in this area. Leucine zippers (LZs) make ideal tectons
for nanoscale assemblies and nanostructured biomaterials for
several reasons: LZs are among the most straightforward and
best-understood peptide–peptide interaction motifs;15as a result,
reliable rules for LZ design and assembly are available;16LZs
are small motifs (∼30 amino acids long) that are synthetically
accessible; when folded, LZs have the approximate dimensions
of 4 nm × 2 nm cylinders; and finally, building with LZs is
scalable, that is, tandem LZ repeats (which can be noncovalently
assembled or covalently linked) can be used to generate
structures that span the nanometer or micrometer regimes, but
still reflect the underlying nanoscale tectons.7,16,17
†School of Chemistry.
‡Department of Biochemistry.
(1) MacPhee, C. E.; Woolfson, D. N. Curr. Opin. Solid State Mater. Sci.
2004, 8, 141–149.
(2) Fairman, R.; Akerfeldt, K. S. Curr. Opin. Struct. Biol. 2005, 15, 453–
463.
(3) Woolfson, D. N.; Ryadnov, M. G. Curr. Opin. Chem. Biol. 2006, 10,
559–567.
(4) Jaeger, L.; Chworos, A. Curr. Opin. Struct. Biol. 2006, 16, 531–543.
(5) Zhang, S. G.; Marini, D. M.; Hwang, W.; Santoso, S. Curr. Opin.
Chem. Biol. 2002, 6, 865–871.
(6) Rajagopal, K.; Schneider, J. P. Curr. Opin. Struct. Biol. 2004, 14,
480–486.
(7) Bromley, E. H. C.; Channon, K.; Moutevelis, E.; Woolfson, D. N.
ACS Chem. Biol. 2008, 3, 38–50.
(8) Kojima, S.; Kuriki, Y.; Yoshida, T.; Yazaki, K.; Miura, K. Proc. Jpn.
Acad., Ser. B 1997, 73, 7–11.
(9) Pandya, M. J.; Spooner, G. M.; Sunde, M.; Thorpe, J. R.; Rodger, A.;
Woolfson, D. N. Biochemistry 2000, 39, 8728–8734.
(10) Potekhin, S. A.; Melnik, T. N.; Popov, V.; Lanina, N. F.; Vazina,
A. A.; Rigler, P.; Verdini, A. S.; Corradin, G.; Kajava, A. V. Chem.
Biol. 2001, 8, 1025–1032.
(11) Zimenkov, Y.; Conticello, V. P.; Guo, L.; Thiyagarajan, P. Tetrahedron
2004, 60, 7237–7246.
(12) Frost, D. W. H.; Yip, C. M.; Chakrabartty, A. Biopolymers 2005, 80,
26–33.
(13) Lazar, K. L.; Miller-Auer, H.; Getz, G. S.; Orgel, J.; Meredith, S. C.
Biochemistry 2005, 44, 12681–12689.
(14) Wagner, D. E.; Phillips, C. L.; Ali, W. M.; Nybakken, G. E.; Crawford,
E. D.; Schwab, A. D.; Smith, W. F.; Fairman, R. Proc. Natl. Acad.
Sci. U.S.A. 2005, 102, 12656–12661.
(15) Mason, J. M.; Arndt, K. M. ChemBioChem 2004, 5, 170–176.
(16) Woolfson, D. N. AdV. Protein Chem. 2005, 70, 79–112.
(17) Ryadnov, M. G.; Woolfson, D. N. J. Am. Chem. Soc. 2007, 129,
14074–14081.
10.1021/ja0778444 CCC: $40.75  XXXX American Chemical Society
Published on Web 03/25/2008
J. AM. CHEM. SOC. XXXX, xxx, 000 9 A

Page 2

Previously, we have reported the design, assembly, and
experimental characterization of a new class of fibrous bioma-
terials based on LZ peptides 28-residues long.9In this so-called
self-assembling fiber (SAF) system, two complementary LZ
peptides are designed to coassemble to form sticky-ended dimers
which, in turn, are engineered to foster end-to-end assembly
into long, noncovalently linked R-helical coiled-coil fibrils. In
practice, and through some as yet undetermined mechanism,
these fibrils bundle to form stiff, thickened fibers (Figure 1A
and B). The final dimensions of these matured fibers are ∼40
nm thick and 1–15 µm long. The fibers behave as stiff rods
that do not bend over their entire lengths; therefore, their
persistence lengths must be greater than the experimentally
observed lengths, that is, >15 µm.18
Over the past five years, we have reported redesigned and
engineered SAF systems with improved stability,19,20of altered
fiber morphology and general properties,19,21,22and with ap-
pended functions.18,23Most significantly for the work presented
here, however, we have described the subtle redesign of the
first-generation SAFs,9to give more stable and better ordered
second-generation SAFs. The redesign process rests on the
simple assumption that the outer surfaces of the coiled-coil fibrils
interact somehow to give thickening, with the focus on potential
charge–charge interactions. Specifically, discrete, complemen-
tary charge clusters are introduced to the surfaces of the
interacting coiled coilssnamely, a pair of negatively charged
aspartates on one peptide, to complement an identically spaced
pair of positively charged arginine side chains on the other.20
This promotes further thickening (to give ∼70 nm fibers), and
improves assembly and thermal stability. Interestingly, matured
second-generation fibers also display considerable internal order,
observed as sharp patterns by X-ray fiber diffraction, and regular
striations on the outer surfaces in TEM (Figure 1B).24Moreover,
the striation pattern precisely matches the ∼4.2 nm length
expected for the folded LZ tectons. These recent data indicate
that the introduced interfibril charge–charge interactions are
acting as designed to specify fibril-fibril interactions.24
These particular properties distinguish the SAFs from some
of the other coiled-coil-based peptide-fiber systems, which
show thickening, but with little or no apparent internal organiza-
tion and ultrastructure;3,11,14,25,26similar ripening behavior has
also been observed for recently engineered self-assembled
collagens.27–29
Our successes in the design and redesign of the SAF system
and the useful, albeit serendipitous, discovery of fiber thickening
encouraged us to probe thickening further. As described herein,
we have attempted to create thinner and more flexible peptide
fibers. We show that this can be done by further programming
charge–charge interactions on the surfaces of the underlying
coiled-coil units. Specifically, the surfaces were supercharged
to make them highly hydrophilic with smeared, rather than
discrete, charged patches. As a result fibers can be thinned down
to ∼10 nm and made much more flexible such that they wrap
around one another to form loose, but nongelling, networks.
The redesigns come at a price, however, as some fiber stability
and the internal order appear to have been lost.
Ionic- and polyelectrolyte-based self-assemblies are well
established in the preparation of both hard and soft multicom-
ponent, nanostructured materials. Indeed, the depth and breadth
of these fields preclude a detailed survey here. Instead, we refer
the reader to excellent reviews by others.30–34In short, the
approaches taken employ simplexes, layer-by-layer deposition,
the self- and coassembly of block copolymers, and polyelec-
trolyte-amphiphile interactions. By and large, these center on
combining molecules of opposite charge, i.e., either all anionic
or all cationic. The approach outlined herein, although it adds
to this effort in some respects, is distinct in that we have
rationally designed folded peptide structures that present faces
(18) Smith, A. M.; Acquah, S. F. A.; Bone, N.; Kroto, H. W.; Ryadnov,
M. G.; Stevens, M. S. P.; Walton, D. R. M.; Woolfson, D. N. Angew.
Chem., Int. Ed. 2004, 44, 325–328.
(19) Ryadnov, M. G.; Woolfson, D. N. Nat. Mater. 2003, 2, 329–332.
(20) Smith, A. M.; Banwell, E. F.; Edwards, W. R.; Pandya, M. J.;
Woolfson, D. N. AdV. Funct. Mater. 2006, 16, 1022–1030.
(21) Ryadnov, M. G.; Woolfson, D. N. Angew. Chem., Int. Ed. 2003, 42,
3021–3023.
(22) Ryadnov, M. G.; Woolfson, D. N. J. Am. Chem. Soc. 2005, 127,
12407–12415.
(23) Ryadnov, M. G.; Woolfson, D. N. J. Am. Chem. Soc. 2004, 126, 7454–
7455.
(24) Papapostolou, D.; Smith, A. M.; Atkins, E. D. T.; Oliver, S. J.;
Ryadnov, M. G.; Serpell, L. C.; Woolfson, D. N. Proc. Natl. Acad.
Sci. U.S.A. 2007, 104, 10853–10858.
(25) Kajava, A. V.; Potekhin, S. A.; Corradin, G.; Leapman, R. D. J. Pept.
Sci. 2004, 10, 291–297.
(26) Zimenkov, Y.; Dublin, S. N.; Ni, R.; Tu, R. S.; Breedveld, V.;
Apkarian, R. P.; Conticello, V. P. J. Am. Chem. Soc. 2006, 128, 6770–
6771.
(27) (a) Paramonov, S. E.; Gauba, V.; Hartgerink, J. D. Macromolecules
2005, 38, 7555–7561. (b) Rele, S.; Song, Y.; Apkarian, R. P.; Qu, Z.;
Conticello, V. P.; Chaikof, E. L. J. Am. Chem. Soc. 2007, 129, 14780–
14787.
(28) Koide, T.; Homma, D. L.; Asada, S.; Kitagawa, K. Bioorg. Med. Chem.
Lett. 2005, 15, 5230–5233.
(29) Kotch, F. W.; Raines, R. T. Proc. Natl. Acad. Sci. U.S.A. 2006, 103,
3028–3033.
(30) Faul, C. F. J.; Antonietti, M. AdV. Mater. 2003, 15, 673–683.
(31) Stuart, M. A. C.; Hofs, B.; Voets, I. K.; de Keizer, A. Curr. Opin.
Colloid Interface Sci. 2005, 10, 30–36.
(32) Johnston, A. P. R.; Cortez, C.; Angelatos, A. S.; Caruso, F. Curr.
Opin. Colloid Interface Sci. 2006, 11, 203–209.
(33) Harada, A.; Kataoka, K. Prog. Polym. Sci. 2006, 31, 949–982.
(34) Hales, K.; Pochan, D. J. Curr. Opin. Colloid Interface Sci. 2006, 11,
330–336.
Figure 1. Transmission electron micrographs of standard SAFs (A and
B), the SAF-SC-4 (C and D) and SAF-SC-3 (E and F) fibers, stained with
1% uranyl acetate. (A, C, and E): low-magnification images (scale bars: 1
µm); (B, D, and F): high-magnification images (scale bar: 50 nm).
B J. AM. CHEM. SOC. 9 VOL. xxx, NO. xx, XXXX
ARTICLES
Papapostolou et al.

Page 3

that incorporate both anionic and cationic charges to alter the
higher-order assembly of the peptide blocks.
Results and Discussion
Peptide Design. The SAF-SuperCharged (SAF-pSC) peptides
described herein (see Table 1 for peptide sequences) follow
sticky-end design principles similar to those reported elsewhere
(Figure 2).9Briefly, the self-assembly fiber (SAF) peptides are
based on the heptad repeat associated with coiled-coil protein
sequences. This is often denoted abcdefg with hydrophobic
residues at the a and d sites, which form the hydrophobic core
of the coiled-coil helical bundle. Good design rules are now
available for coiled-coil structures.16For example, the combina-
tion of isoleucine at a and leucine at d best specifies dimeric
leucine zippers (LZs). The helical interface can be cemented
further with interhelical (intracoiled coil) charge–charge interac-
tions between successive g and e sites. Together these rules
give the general framework of +IxxL-x (in a gabcdef register)
for a LZ sequence; where x is usually made a polar residue or
alanine. Tandem repeats of 3–5 such heptads tend to associate
to give stably folded structures. Further specificity can be
achieved, although at the expense of some stability, by
incorporating hydrogen-bond-forming pairs of asparagine sides
at a sites in the hydrophobic core. A combination of such
asparagine residues at offset a sites, and complementary
interhelical charges was used to achieve the sticky-ended
heterodimer designed to propagate the SAFS.
In addition to the above considerations, two other design
principles were incorporated into the new SAF-pSC peptides:
first, after the heterodimeric Velcro system reported by O’Shea
and Kim,35complementary, interhelical (intracoiled coil) charge–
charge interactions were introduced by making all of the e and
g positions of one peptide basic, and those of the other, acidic
(Figure 2). Thus, for the basic SAF-pSC peptides, C3B and C4B,
arginine was placed at every e and g position; while their acidic
partners, C3A and C4A, had glutamate at all these sites. Second,
to make the resulting coiled coils as soluble as possible, and
introduce controllable aggregation, charged residues were placed
at the b, c and f positions of the coiled-coil repeat, i.e., on the
outmost surfaces of the coiled coils. To avoid intrahelical
repulsions, we introduced these to allow interconnected networks
of intrahelical salt bridges by maximizing the i to i+3 and i to
i+4 juxtapositions of oppositely charged side chains (Figure
2). This follows the work of Burkhard and colleagues who
present a 2-heptad-long oligomerizing coiled-coil-like peptide
bearing similar networks of intra- and intermolecular salt
bridges.36–38Finally, two parent, two-peptide systems were
designed, SAF-SuperCharged3 (C3) and SAF-SuperCharged4
(C4), differing by the length of the peptides: the C4 peptides
were four-heptads, or 28-residues long, while the C3 peptides
were shorter by one heptad. Using these design principles, we
aimed to follow on from our work on the SAF system,20,24to
further understand the phenomena of fiber thickening and order,
and to attempt to design thinner, more-flexible fibers.
Visualizing SAF-SuperCharged Fibers by Electron Micro-
scopy. Consistent with the design principles, when peptides C4A
and C4B were mixed and incubated at 20 °C overnight (in 10
mM MOPS, pH 7.5), fibers were observed using negative-stain
Transmission Electron Microscopy (TEM), Figures 1 C and D.
In contrast, the individual peptides gave only amorphous
aggregates (data not shown). The resulting C4A + C4B fibers
were 2 to 3 times thinner than the standard SAFs, Figures 1 A
and B, with average widths ∼ 25 nm (24.7 ( 4.9 nm; n ) 113
measurements) compared with 43 ( 9.3 nm for the first-
generation SAFs, and 69 ( 18.5 nm for second-generation
SAFs.20Assuming that the influence of the core coiled-coil
residues—i.e., those at a, d, e and g—is minimal, this suggests
that the charge distribution on the outer surface of component
coiled-coil fibrils is indeed a key feature in fiber thickening,20,24
and that fiber width can be programmed either way (thicker or
thinner) within the peptide sequence.
Interestingly and in contrast to the standard SAFs, the C4A
+ C4B fibers were flexible, and appeared to lack internal order
as judged by the absence of the aforementioned striation patterns
in the stained TEM images (Figures 1B and D). Nonetheless,
individual fibers did interact with each other to form loose
bundles of two or three fibers (Figure 2C and D), and
interconnected into networks. As a result, though individual
fibers did not exceed 5 µm, these intertwined structures were
10 – 15 µm in length. This behavior is more reminiscent of the
entangled polymers,39rather than the discrete rod-behavior
typically observed for the standard SAFs.
(35) O’Shea, E. K.; Lumb, K. J.; Kim, P. S. Curr. Biol. 1993, 3, 658–667.
(36) Burkhard, P.; Meier, M.; Lustig, A. Protein Sci. 2000, 9, 2294–2301.
(37) Burkhard, P.; Ivaninskii, S.; Lustig, A. J. Mol. Biol. 2002, 318, 901–
910.
(38) Meier, M.; Lustig, A.; Aebi, U.; Burkhard, P. J.Struct. Biol. 2002,
137, 65–72.
(39) Jones, R. A. Soft Condensed Matter; Oxford University Press:New
York, 2002; pp 90–91.
Table 1. SAF-SuperCharged Peptide Sequences
Figure 2. Design principles for the SAF-pSC peptides. Helical-wheel
representation of the heptad repeats for the A- and B-series peptides.
Potential salt bridges are shown: interhelical (solid arrows); intrahelical
(dashed arrows); intercoiled coil (dotted arrows).
J. AM. CHEM. SOC. 9 VOL. xxx, NO. xx, XXXX
C
De Novo Protein Design at the Amino Acid Level
ARTICLES

Page 4

Fibers were also observed for the mixture of the shorter, C3A
and C3B, peptides (Figures 1E and F). These were shorter and
thinner than the C4A + C4B fibers (average length, 989 nm (
318 nm; width, 10.7 nm ( 1.3 nm; n ) 54). The tendency of
C3A+C3B fibers to interact and to form networks was also
increased (Figure 1E). Indeed, the shorter system in particular
is reminiscent of entangled fibrous networks formed by amy-
loidogenic peptides,40–42which are based on ?-sheet structures.
The fibers made from the shorter peptides showed an increased
tendency to be damaged by the electron beam, which made it
difficult to collect very high-magnification images. We propose
that this, together with the thinner fibers that the C3 peptides
form, is related to the less-extensivesand, therefore less
stabilizingsoverhang in the designed sticky-ended dimers
compared with those for the longer C4 system.
Despite the apparent wrapping and entanglement of fibers,
neither SAF-pSC mixture formed gels at any concentration
tested. This suggests that the fibers do not branch, or otherwise
form noncovalently cross-linked networks.
The absence of striations, and therefore the apparent lack of
internal order in the supercharged SAFs compared with our
recently described second-generation fibers,24is interesting and
noteworthy. Our contention is that this is consistent with our
design principles as follows. The second-generation fibers have
a relatively small number of charged residues on their outer
surfaces, which are otherwise alanine and glutamine rich.
Moreover, and compared with the first-generation SAFs, the
charged residues are paired and placed such that they can
potentially form a complementary, self-solvating, local, elelctro-
static networks within the matured fibers. By contrast the entire
surfaces of the SAF-pSC coiled coils have charged side chains.
We posit that: (1) this makes them difficult to completely
desolvate; therefore, we anticipate that the fibers formed are
internally wet and (2) that multiple, isoenergetic intercoiled-
coil interactions can be made within these fibers, compared with
the second-generations SAFs where more specific intercoiled-
coil interactions should be made. These increased possibilities
in the SAF-pSC system should not lead to thicker fibers,
however, as thickening would increasingly incur the desolvation
penalty, and/or other penalties associated with removing water
from bulk.
Solution-Phase Characterization of SAF-pSC Mixtures. Cir-
cular dichroism (CD) spectroscopy is usually the technique of
choice to study the folding of peptides into R-helical structures.
Indeed, the CD spectra of SAF-pSC3 and SAF-pSC4 mixtures
were typical of R-helices, as characterized by double minima
at 208 and 222 nm (Figure 3). Nonetheless, this was curious as
all other SAFssas well helical fibers presented by otherssshow
distorted R-helical spectra in which signals are attenuated,
particularly the 208 nm signal, and red-shifted.9These are the
hallmarks of light scattering associated with large and sometimes
chiral assemblies in solution. This suggested subtly different
assemblies in the SAF-pSC systems compared with the previous
generations. Possibilities are that the thinner fibers simply do
not scatter as much light, or that both the individual peptides
and the mixtures fold to R-helical structures of some type. To
test these two possibilities, the CD spectra of the individual
peptides and mixtures were recorded under different conditions
as follows.
First, under similar conditions used to study the C4A + C4B
mixture, the individual SAF-pSC4 peptides, C4A and C4B, were
also folded as R-helices with percentage R-helix calculated from
the molar ellipticity values at 222 nm of ∼60% and ∼50%,
respectively (Figure 3A). This autonomous folding probably
reflects the high helix propensities of the peptides; their potential
to form coiled-coil oligomers, despite these being noncognate
(not designed) homo-oligomers; and the large number of
potential i, i + 3 and i, i + 4 intrahelical salt bridges within the
sequences. Again, both spectra were typical of R-helices (albeit
partially folded), with no indication of ?-sheet. Furthermore,
C4A alone was found to be more helical than the SAF-pSC4
mixture (Figure 3A), suggesting that peptide assembly and/or
the subsequent fiber growth resulted in a loss of CD signal.
This led us to study the CD signal of peptide mixtures under
(nonpermissive) conditions that did not lead to fiber growth as
judged by TEM: namely, concentrations below 50 µM in each
peptide, or salt concentrations above 50 mM KCl.
(40) Lamm, M. S.; Rajagopal, K.; Schneider, J. P.; Pochan, D. J. J. Am.
Chem. Soc. 2005, 127, 16692–16700.
(41) Papanikolopoulou, K.; Schoehn, G.; Forge, V.; Forsyth, V. T.; Riekel,
C.; Hernandez, J. F.; Ruigrok, R. W. H.; Mitraki, A. J. Biol. Chem.
2005, 280, 2481–2490.
(42) Baldwin, A. J.; Bader, R.; Christodoulou, J.; MacPhee, C. E.; Dobson,
C. M.; Barker, P. D. J. Am. Chem. Soc. 2006, 128, 2162–2163.
Figure 3. Circular dichroism spectra of the SAF-pSC4 peptides. (A) Individual peptides and mixtures: 100 µM C4A (black broken line); 100 µM C4B (gray
broken line); C4A + C4B mixtures at 10 µM (gray solid line) and 100 µM in each peptide (black solid line). (B) C4A + C4B mixtures: 10 µM in each
peptide and unperturbed (gray solid line); supernatant spun-down from a sample 100 µM in each peptide (black solid line); and 100 µM in each peptide and
incubated with 300 mM KCl (black broken line). (C) C4A + C4B mixture incubated with 300 mM KCl (broken line), and an untreated C4AK + C4BK
mixture (solid line); for these experiments all peptides were at 100 µM.
DJ. AM. CHEM. SOC. 9 VOL. xxx, NO. xx, XXXX
ARTICLES
Papapostolou et al.

Page 5

[n.b. A range of salt concentrations was explored from 0.5
to 500 mM KCl. Fibers were only observed by negative-stain
TEM at 50 mM KCl or below. Under these permissive
conditions fiber morphology (thickness and length) did not
change appreciably within the experimental ranges given above.
Though high, 300 mM KCl was used in the experiments
described below because (1) fibers did not form, and (2) the
CD signals for the individual and mixed peptides showed
plateaus at this concentration.]
First of all, a nonpermissive, 10 µM mixture of SAF-pSC4
peptides showed a larger R-helical content than the more-
concentrated, fiber-producing mixtures (Figure 3B). This indi-
cated that fiber growth, but not the coiled-coil assembly, was
responsible for the loss of signal described above and is
consistent with previous CD data on the SAF systems. To
investigate this further, a C4A + C4B fiber-containing mixture
(at 100 µM in each peptide) was spun down by centrifugation,
and a CD spectrum of the supernatant measured and normalized
against the measured peptide concentration, which was found
to be close to a third of the initial peptide concentration (∼60
µM total peptide). The molar ellipticity of this supernatant was
similar to that for an unperturbed, nonpermissive mixture of
C4A + C4B (10 µM in each peptide) (Figure 3B).
These experiments demonstrate that peptides incorporated into
grown fibers are not soluble and do not contribute fully to the
CD signal, hence, the apparent drop in signal upon fiber
formation. As a result, we conclude that, in the case of the SAF-
pSC system at least, the CD signal does not report on the
peptides incorporated in the fibers, but on smaller assemblies
that are found in the supernatant of the spun-down fiber
preparations. This latter point and the secondary structure of
the peptides incorporated into fibers are addressed in the next
two sections.
Before leaving this section, we are aware that the CD data
for the SAF-pSC and standard SAF systems are possibly in
conflict. Though, we note that the loss of signal observed for
the SAF-pSC mixtures could be attributed to partial attenuation
of the CD signal akin to that observed for the standard SAFs.
More likely, however, the behavior is related to the ability shown
by the SAF-pSC fibers to bundle into large networks that are
more particulate than soluble proteins. In such heterogeneous
systems artifacts can arise in CD spectra from the effects of
differential light scattering and absorption flattening. By contrast,
the standard SAFs form discrete rigid rods that do not appear
to interact in solution or to form networks. In addition, the
individual standard SAF peptides are much less folded than
the individual SAF-pSC peptides and contribute much less to
the solution-phase CD signal.9
Oligomerization of the SAF-pSC Peptides by Analytical
Ultracentrifugation. To investigate the solution-phase assembly
of the SAF-pSC peptides further, we turned to analytical
ultracentrifugation (AUC).
First, as shown in Figure 4, data were recorded at a rotor
speed of 3000 rpm. At this speed only very large assemblies
sediment, and small oligomers of peptides of the mass of the
SAF peptides do not. The sums of the traces for the individual
peptides, which serve as controls, are superimposed on those
for the experimental mixtures. Comparison of the traces for the
individual peptides and fiber-forming mixtures of C4A and C4B
(100 µM in each peptide and without salt) indicated a loss of
approximately two-thirds of the peptide from solution, and a
build up of aggregated material at large radii, Figure 4A. These
data are fully consistent with the aforementioned CD analysis
of the permissive C4 mixtures. In comparison, the data for C4A
+ C4B with 300 mM KClswhich is nonpermissive for fibers
as judged by TEMsshowed only a third of the signal is lost,
and less signal from aggregated material at the bottom of the
cell, Figure 4B.
To probe the oligomer states of the R-helical species that
remained in solution further, we ran a series of conventional
equilibrium AUC experiments at high rotor speeds, at which
small LZ oligomers are known to sediment. For example, Figure
5 shows traces for the individual peptides and mixtures at a
speed of 53000 rpm. These data indicate that for all samples
the soluble peptides formed mixtures of small oligomers. The
masses of C4A, C4B, the theoretical heterodimer and an
averaged trimer are 3757, 3838, 7595 and 11392 Da, respec-
tively. Without salt the global single ideal species fits to the
data for C4A, C4B and the mixture were found to be 4458,
6391 and 8256 Da, respectively, Figure 5A. With 300 mM salt
these increased to 6768, 8075 and 8510 Da, respectively, Figure
5B. This analysis indicates that the individual C4A and C4B
peptides show varying propensities to self-associate into small
oligomers, but not larger structures; and, for the mixtures, the
peptides remaining in solution after fiber formation also form
small oligomers. The apparent increase in solution-phase
molecular weight with salt presumably reflects (1) a reduction
in the electrostatic repulsion between peptides through additional
screening of charge, and/or (2) an increase in the hydrophobic
effect. Given the likely heterogeneity in these samples from the
various associated states possible, we chose not to fit the AUC
data to more-elaborate models for self-associating systems;
Figure 4. Low speed Analytical Ultracentrifugation of the SAF-pSC-4 and
SAF-pSC-4K peptides (3000 rpm). (A) Average of 100 µM C4A and 100
µM C4B (gray line), 100 µM C4A + 100 µM C4B mixture (black line).
(B) Average of 100 µM C4A and 100 µM C4B with 300 mM KCl (gray
line), 100 µM C4A + 100 µM C4B mixture with 300 mM KCl (black
line). (C) Average of 100 µM C4AK and 100 µM C4BK (gray line), 100
µM C4AK + 100 µM C4BK mixture (black line). Traces were aligned
using the outer meniscus’s.
J. AM. CHEM. SOC. 9 VOL. xxx, NO. xx, XXXX
E
De Novo Protein Design at the Amino Acid Level
ARTICLES