14 JULY 2017 • VOL 357 ISSUE 6347 133 SCIENCE sciencemag.org
By Derek N. Woolfson,1,2,3 Emily G. Baker,1
Gail J. Bartlett1
How does theamino acidsequenceof aproteinchaindetermineandmain- tain its three-dimensional folded state? Answering this question—a key aspect of the protein-folding problem (1)—would help to explain
how multiple noncovalent interactions conspire to assemble and stabilize complicated
biomolecular structures; to predict protein
structure and function from sequence for
proteins that cannot be characterized experimentally; and to design new protein
structures that do not exist in nature (2).
On page 168 of this issue, Rocklin et al. use
parallel protein design on a massive scale
to create thousands of miniprotein variants
and to determine what sequences specify
and stabilize these structures (3). The work
opens up considerable possibilities for protein folding and design.
Miniproteins are polypeptides shorter
than 40 to 50 residues with stable tertiary
structures (folds) that contain a limited
number of secondary structure elements,
such as a helices and b strands. By contrast, larger proteins have hundreds of
amino acids that are often arranged in complex three-dimensional structures. Thus,
miniproteins simplify the protein-folding
problem and potentially allow in-depth examinations of sequence-structure-stability
relationships without complications from
larger protein contexts. Unfortunately, only
a few miniproteins that are stable without
covalent cross-links or stabilizing metal ions
are currently available for such studies (4).
In their study, Rocklin et al. combine
high-throughput DNA synthesis and clon-
ing (5, 6) with methods for selecting stably
folded proteins (7–9). They implement the
latter by expressing libraries of miniproteins
on the surface of yeast; tagging the displayed
proteins with a fluorescent dye; and discrim-
inating between stable and unstable folds
through their ability to resist or succumb
to protease treatment, respectively (see the
figure). Proteins that survive are rescued by
fluorescence-activated cell sorting and then
identified by deep sequencing. However, the
team’s experiments go beyond a yes/no meas-
ure of protein resilience, providing a semi-
quantitative measure of stability.
To demonstrate the approach, the authors
first apply their method to many variants of
a small number of known miniproteins. With
the method established, they turn their attention to four classes of de novo miniproteins,
which they design computationally using
Rosetta (10): aaa, babb, abba, and bbabb
folds, where each Greek letter represents an
a helix or a b strand in the peptide string.
To cover swaths of sequence space, the team
generate diverse libraries with minimal se-
quence identity between members.
They then use iterative rounds of protease
selection and stability scoring, testing different hypotheses, and introducing tweaks to the
design methods and protocols at each stage.
The value of these tweaks is apparent from
the improved success rate—the proportion of
stable proteins in the starting library—which
reaches 87% for one target. However, both the
initial and final design success rates depend
critically on the fold being targeted, with the
aaa fold proving easiest and the abba fold
most difficult to optimize.
Through sequence analyses of many thousands of these new and also existing miniprotein folds from other studies, the authors
highlight several key sequence and structural
features. First, a long-established basic tenet
of protein folding and design shines through:
the importance of burying nonpolar surfaces.
This is not surprising, but Rocklin et al. quantify the effect, showing that stable variants
require more than 30 Å2 for each residue of
Second, of the initial computational de-
signs, those containing peptide fragments ge-
ometrically similar to ones known from the
thetic conductors is the absence of a band
gap in the former. Electrons can strongly in-
teract with holes in gapless graphene (6), and
this process changes the “sign” of the velocity
renormalization correction compared with
the case of electron-electron interaction.
Strong electron-hole interactions may
cause the electronic liquid in graphene to
become highly viscous (10). The mutual viscous friction forces electrons and holes to
move together, so that the effective charge
contributing to the low-frequency optical
response of the electron liquid is diminished. In theory, this effect should inhibit
plasmons but enable another type of collective excitation—energy waves or “demons”
(11)—to exist at small v and q (see the figure). The thermal photocurrent mapping
technique devised by Lundeberg et al. (2,
4) appears to be particularly promising for
detection of these elusive modes.
Lundeberg et al. point out that their
method of determining the nonlocal
complex conductivity s9(v,q)+ is99(v,q) is
applicable to other quantum materials,
including low-dimensional conductors,
superconductors, and Weyl semimetals.
A technical precondition for such experiments is the ability to use nano-optical
imaging at cryogenic temperatures, which
recently became available (12). Plasmonic
imaging in graphene at liquid helium temperature are also highly desirable because
scattering by phonons in this regime will
be reduced, whereas the observables associated with many-body physics are expected to be enhanced. We anticipate that
future studies will address yet another
unresolved issue pertaining to the analysis of the linewidth of plasmonic modes in
graphene that is determined by the ratio
s9(v,q) /s99(v,q). Plasmonic images, including those reported in (2–4), prompt us to
reimagine the sheer scope of unresolved
problems that can be tackled with this innovative experimental approach. j
1. D. N. Basov, R. D. Averitt, D. van der Marel, M. Dressel, K.
Haule, Rev. Mod. Phys. 83, 471 (2011).
2. M. B. Lundeberg et al ., Science 357, 187 (2017).
3. D. N. Basov, M. M. Fogler, F. J. Garcia de Abajo, Science 354,
4. P.Alonso-González et al., Nat. Nano.12,31(2017).
5. D. C. Elias et al ., Nat. Phys. 7, 701 (2011).
6. J. González, F. Guinea, M. A. H. Vozmediano, Phys.Rev.Lett.
77, 3589 (1996).
7. A. H. Castro Neto, F. Guinea, N. M. R. Peres, K. S. Novoselov,
A. K. Geim, Rev. Mod. Phys. 81, 109 (2009).
8. D. N.Basov,M. M.Fogler, A.Lanzara,F. Wang, Y.Zhang, Rev.
Mod. Phys. 86, 959 (2014).
9. C. F. Hirjibehedin, A. Pinczuk, B. S. Dennis, L. N. Pfeiffer, K.
W. West, Phys. Rev. B 65, 161309 (2002).
10. R. Krishna Kumar et al ., arXiv:1703.06672v1 (2017).
11. Z. Sun, D. N. Basov, M. M. Fogler, Phys.Rev.Lett. 117,
12. A. S. Mc Leod etal. , Nat.Phys. 13, 80 (2017).
How do miniproteins fold?
A high-throughput study yields libraries of miniproteins
that help to explain how proteins are stabilized
1School of Chemistry, University of Bristol, Cantock’s Close,
Bristol BS8 1TS, UK. 2School of Biochemistry, University of
Bristol, Medical Sciences Building, University Walk, Bristol BS8
1TD, UK. 3Bristol BioDesign Institute, University of Bristol, Life
Sciences Building, Tyndall Avenue, Bristol BS8 1TQ, UK.
“...Rocklin et al. have taken
driven protein design,
selection, and optimization
to new heights…”