INSIGHTS | PERSPECTIVES
V G RP
P AG LV
2 Libraries of amino acid sequences are
generated to best ft these structures. Those
sequences are synthesized via high-throughput
DNA synthesis and cloning.
3 The resulting proteins are
expressed on the surface of yeast.
4 Stable variants are selected
based on resistance to treatment
5 The sequences of stable variants are analyzed
to determine sequence–stability relationships, which
are fed back into the design cycle.
1 Miniprotein structures are designed computationally
using a fragment-based approach in Rosetta.
Protein Data Bank of protein structures fared
better in the selection process than those
with more geometrically distant matches; i.e.,
the former gave more stable sequences. This
could be a consequence of using Rosetta to
achieve the design frameworks, given that it
is a fragment-based design approach. In the
future, it will be interesting to see how starting points from parametric and other design
approaches perform (11–13).
Third, one relationship not included or
tweaked during the iterative process—it
simply emerges from the analysis—is the
importance of having charged side chains
at the termini of the a helices that oppose
the terminal partial charges of the helices.
This concurs with studies of model peptides
that form freestanding a helices in solution,
where helix formation is attributed to local
capping effects (14).
Despite the impressive and expansive na-
ture of the study, there are gaps to fill and
more steps to take. Although the authors
have characterized many sequences for the
target designs by circular dichroism spec-
troscopy, size-exclusion chromatography,
and thermal and chemical denaturation, and
have verified a small number of structures
by nuclear magnetic resonance spectros-
copy, more high-resolution structural details
would be welcome—for instance, from x-ray
crystallography. Such structures would allow
sequence-structure-stability relationships to
be rationalized in terms of specific noncova-
lent interactions that likely underlie them.
For example, the study points to stabilizing
roles for aromatic residues at surface-exposed
sites of a helices and b strands in minipro-
teins, which hint at noncovalent interactions
particular to this class of side chain.
In an unrelated but pertinent study, Baker
et al. recently designed, characterized, and
interrogated another monomeric miniprotein, PPa. This miniprotein has a compact
structure comprising a polyproline II
helix and an a helix that are connected
by an intervening loop (15). A key determinant of PPa’s stability comes
from intimate CH-p interactions between tyrosine residues of the a helix
and proline residues of the buttressing polyproline II helix. Studying the
role and interplay of these and other
noncovalent interactions will be critical
for feeding back into and improving computational design methods.
In their study, Rocklin et al. have taken
high-throughput, data-driven protein design, selection, and optimization to new
heights, bringing us closer to solving aspects of the protein-folding problem. A
combination of high-throughput studies
of the sequence-structure-stability relationships described by Rocklin et al. and
drilled-down, fully quantitative examinations of the noncovalent interactions within
(mini)proteins will bring us even closer to
solving this long-standing problem. In turn,
this will facilitate better engineering of natural and de novo proteins. j
REFERENCES AND NOTES
1. K. A. Dill, J. L. MacCallum,Science 338, 1042 (2012).
2. W.R. Taylor, V.Chelliah,S.M.Hollup,J. T.MacDonald,
I. Jonassen, Structure 17, 1244 (2009).
3. G. J. Rocklin et al., Science 357, 168 (2017).
4. S.H.Gellman,D.N. Woolfson, Nat. Struct. Biol. 9,408
5. S.Kosuri, G.M.Church, Nat. Methods11,499(2014).
6. M. G. F. Sun, M. H. Seo, S. Nim, C. Corbi-Verge, P. M. Kim,
Sci. Adv. 2, e1600692 (2016).
7. P. Kristensen, G. Winter, Fold Des. 3, 321 (1998).
8. V. Sieber, A. Pluckthun, F. X. Schmid, Nat. Biotechnol. 16,
9. M. D. Finucane, M. Tuna, J. H. Lees, D. N. Woolfson,
Biochemistry 38, 11604 (1999).
10. R. Das, D. Baker, Annu. Rev. Biochem. 77, 363 (2008).
11. P. S. Huang et al., Science 346, 481 (2014).
12. A.R. Thomson et al., Science 346,485(2014).
13. T. J. Brunette et al. , Nature 528, 580 (2015).
14. E. G. Baker et al. , Nat. Chem. Biol. 11, 221 (2015).
15. E. G. Baker et al. , Nat. Chem. Biol .13, 764 (2017).
D.N.W. and E.G.B. are supported by a Biotechnology and
Biological Sciences Research Council–ERASyn Bio grant (BB/
M005615/1); D.N.W. and G.J.B. are supported by the European
Research Council (340764); and D. N. W. holds a Royal Society
Wolfson Research Merit Award (WM140008).
Protein design cycle
Rocklin et al. use an iterative design cycle to create stable miniproteins. After initially designing
miniprotein folds using computational tools, they express them and test their stability, followed
by further optimization cycles.