1 Sam Gellman is in the Department of Chemistry,
University of Wisconsin, Madison, Wisconsin
53706, USA.gellman@chem.wisc.edu
2 Dek Woolfson is in the Centre for Biomolecular
Design & Drug Development, School of Biological Sciences, University of
Sussex, Falmer, BN1 9QG,
UK.dek@biols.susx.ac.uk
A new 20-residue peptide represents the smallest example to date
of cooperatively folded tertiary structure. This achievement provides a new
tool for elucidating protein conformational preferences. The mini-protein
should serve as a fruitful platform for protein design.
On page 425 of this issue, Neidigh et
al.1 describe a remarkable peptide that adopts a small but
well-defined globular shape. Formation of discrete tertiary structure is
generally thought to be the exclusive province of much longer polypeptides.
Although other short peptides with stable tertiary structure have been
reported, the present case stands out in that only natural amino acid residues
are employed and there is no crosslinking via disulfide formation, metal
ion chelation or stabilization through oligomerization. The success of Neidigh
et al.1 appears to be primarily due to a structural motif
the authors dub the 'Trp cage', in which the side chain of a Trp residue is
penned in by several other residues, notably the side chains of prolines.
Development of the Trp cage began during examination of a natural
39-residue peptide from Gila monster saliva. The C-terminal portion of this
peptide showed promise for folding, but structure was observed only in the
presence of the fold-promoting cosolvent 2,2,2-trifluoroethanol (TFE).
Recognizing the prospect for structural optimization, the authors pursued a
series of incremental sequence modifications to enhance the stability of the
folded conformation. These shrewd efforts culminated in a 20-residue Trp cage
that appears fully structured at low temperature in water and displays several
hallmarks of cooperatively folded proteins.
The Trp cage is cooperatively folded The studies of protein structure and folding are mature fields of
endeavor2, and one is led to ask two fundamental questions
regarding the system reported by Neidigh et al.1: (i) Is
the structure completely folded under accessible conditions? (ii) Is folding
cooperative? This latter question is important because for most natural protein
domains, all parts of the molecule undergo the folding transition in concert as
conditions, such as temperature or denaturant concentration, are varied. This
type of cooperativity is regarded as a hallmark of native protein structures.
In such cases, only two 'states' of the protein — the folded and unfolded
states — are present throughout the process. Both states are collections
of conformations, also referred to as ensembles. The unfolded conformations are
flexible and largely unrelated to one another, whereas the folded-state
conformations are closely related. The work of Neidigh et al.1 provides clear answers to these two questions.
Evidence for essentially complete folding into a specific tertiary
structure near 0 °C comes from following independent global measures of
folding — that is, characteristic CD signals and 1H NMR
chemical shifts — as a function of temperature. In both cases the
measured parameters plateau at low temperature to values consistent with a
highly folded state. In addition, amide protons of some buried residues are
protected from H/D exchange in the folded state, another classic sign of
native-like folding. Further support for complete folding in water comes from
the observation that there is very little change in the CD spectrum of the
final Trp cage design when the sample contained 30% TFE. In contrast,
intermediate designs showed substantial enhancements in folding upon addition
of TFE. Although addition of TFE does not necessarily induce complete peptide
folding3, it seems plausible that failure to see a TFE effect
indicates that the extent of folding is already quite high.
What is the evidence for cooperative folding? Sigmoidal dependence of
spectroscopic measurements with changes in temperature or denaturant
concentration is a classic signature of cooperativity. The authors show such
temperature dependencies for both CD and NMR signals. They also use a novel
method4 to test for cooperativity: they compare the rates of
change of characteristic 1H NMR chemical shifts through the
thermal unfolding transition (/T) with the differences
measured for these chemical shifts between the folded state and reference
values for the unfolded state (folded -
unfolded). A linear correlation between these two parameters,
which is observed for the Trp cage, fits a cooperative unfolding transition
between a single folded state and an unfolded ensemble. These comparisons
support the hypothesis that the final Trp cage peptide undergoes a concerted
unfolding transition.
What stabilizes the Trp cage? Once we accept that the Trp cage is highly and cooperatively folded, at
least at low temperature, we are eager to know what holds this little motif
together. Neidigh et al.1 oblige with a solution structure
for the final Trp cage design (Fig. 1a). The
critical tertiary contacts involve the Trp side chain, which is almost
completely buried in the folded state. The atoms packed against the Trp side
chain come largely from Pro residues in the mini-protein. The authors propose
that the Trp−Pro contacts are hydrophobically favorable, and that Pro is
an ideal partner for such contacts because the rigidity of this imino-acid
residue limits the conformational entropy loss associated with folding.
a, The Trp Cage (PDB entry 1L2Y); b, the villin headpiece (PDB entry
1VII); c, the Trp zipper (PDB entry
1HRW); and d, avian pancreatic
polypeptide (PDB entry 1PPT). In each
structure the backbone trace is shown as a ribbon. Aromatic and Pro side chains
are space filled, with aromatic carbon and nitrogen atoms colored gray and
blue, respectively, oxygen is colored red and the atoms of Pro are colored
purple. These structures were generated using MolScript15.
This proposal merits a close examination. One noteworthy aspect of the
truncated Trp cage sequence is its high Pro content (4 of the 20 residues are
Pro), which is some four times greater than the average occurrence of the
residue in natural proteins. In apparent contradiction to this, however, the
sequence contains twice as much Gly (3 out of 20 residues) — the least
rigid residue — as the average protein. This intrinsic flexibility is
supposed to counter stable folding, but perhaps the Gly residues in the Trp
cage are critical to allow contortion of the backbone at certain places in
order to better define and, in effect, fine tune the tertiary fold. Indeed,
measured across all protein structures, Gly residues adopt a much wider
spectrum of backbone conformations than any other residue. No doubt related to
this, Gly is also one of the most conserved residues at specific sites in
protein structures. Consistent with these ideas, all three Gly residues in the
Trp cage adopt conformations outside the - and -regions of the
Ramachandran plot, which are the most populated areas for non-Gly residues.
Thus, an odd alliance of Trp, Pro and Gly in the Trp cage conspires to
produce an organized fold cemented by tightly knit buried side chain
interactions. Such arrangements are clearly important in specifying and
stabilizing competently folded proteins. However, in systems as small as the
Trp cage, the key is to make a limited number of such interactions count. The
battle here is against the effects of conformational entropy, which is lost
during folding because of restrictions placed on both the backbone and side
chains. Clearly, the inclusion of Pro must help, but is it enough?
Another source of contributions to protein stability is from
interactions in the unfolded state that may reduce the entropic cost of
folding. Neidigh et al.1 present some evidence for this in
both the early and final iterations of the Trp-cage design process. Again, Trp
and Pro are involved; specifically, a cluster involving Trp 25−Pro 31
interaction may exist in some population of the unfolded ensemble. An embryonic
and potentially native-like interaction like this should facilitate folding.
Such clustering and possibilities have been observed and discussed for other
systems5.
Neidigh et al.1 note that Trp−Pro
interactions play a key role in the formation of complexes bewteen proline-rich
ligands and WW domains. This protein−protein recognition theme is
actually more general, involving other widespread recognition modules, such as
SH3 domains, and other aromatic side chains (such as the highly conserved Tyr
in WW domains6). So, are the favored side chain contacts between
aromatic residues and/or Pro a general lesson in specifying and maintaining
minimal tertiary folds? The answer is: possibly. For instance, a fragment of
the villin headpiece — which at 36 residues is a close runner-up to the
Trp cage for the smallest independently folded unit — has a cluster of
three Phe residues in its core7 (Fig.
1b); many thermophilic proteins contain aromatic clusters,
whereas their mesophilic counterparts do not8; and a recently
designed -hairpin based on two cross-strand Trp−Trp interactions
(Fig. 1c) forms a highly twisted but very stable
structure9. Turning to natural, intact structures and to
aromatic−Pro interactions specifically, the 36-residue avian pancreatic
polypeptide (aPP) and its related hormones represent the smallest tertiary
folding units that are not stabilized through crosslinking10. The
structure of aPP features a polyproline-II helix followed by an -helix
(Fig. 1d). Contact between these two helices
occurs largely between the three Pro residues (positions 2, 5 and 8) and
nonpolar residues on one face of the -helix, including Leu 17, Phe 20,
Leu 24, Tyr 27 and Val 30. The contacts between the aromatic and Pro residues
in this interface (Fig. 1d) provide food for
thought on the role of such juxtapositions.
Prospects for mini-proteins The development of the Trp cage is an important achievement for many
reasons. We still have a great deal to learn about why specific amino acid
sequences fold as they do, from both thermodynamic and kinetic perspectives.
Small folding motifs offer excellent systems for detailed analysis because the
number of factors that influence folded state stability is reduced, which
benefits both experimental and computational studies of folding. In addition, a
short sequence facilitates the use of chemical synthesis to generate analogs
using not only the 20 standard amino acids, but also unnatural side chains and
backbone elements. Such molecules would allow one to explore the origins of
conformational stability in ways that are impossible with the proteinogenic
amino acids11. For example, replacing a methyl side chain
(alanine) with ethyl can probe the contribution of incremental nonpolar surface
burial to conformational stability, and replacing a backbone amide linkage with
an ester can probe the contribution of specific hydrogen bonds. In addition to
their use for fundamental structural studies, small, robust tertiary motifs
provide ideal scaffolds upon which interesting functions may be built. This
utility has been recently demonstated with aPP-derived peptides12, and it is easy to imagine analogous studies with Trp cage peptides
or other mini-proteins.
The Trp cage represents a new benchmark in protein design13,
14. It is hard to see how a well-defined tertiary fold could
comprise fewer than 20 residues. The exciting challenge offered by the Trp cage
is development of alternative folding patterns with comparable length and then
endowing them with interesting functions.
/
- and 
-regions of the
Ramachandran plot, which are the most populated areas for non-Gly residues.
