Embed Size (px)
Citation preview
Current Biology 17, 173–178, January 23, 2007 ª2007 Elsevier Ltd All rights reserved DOI 10.1016/j.cub.2006.10.063
ReportContinuous Molecular Evolutionof Protein-Domain Structuresby Single Amino Acid Changes
Sebastian Meier,1,5,* Pernille R. Jensen,1,6
Charles N. David,2 Jarrod Chapman,3
Thomas W. Holstein,4 Stephan Grzesiek,1
and Suat Ozbek4,*1Department of Structural BiologyBiozentrum, University of BaselKlingelbergstrasse 70CH-4056 BaselSwitzerland2Department Biologie IILudwig-Maximilians-UniversitatGrosshadernerstrasse 2D-82152 Planegg-MartinsriedGermany3Department of EnergyJoint Genome InstituteWalnut Creek, California 945984 Institute of ZoologyDepartment for Molecular Evolution and GenomicsUniversity of HeidelbergIm Neuenheimer Feld 230D-69130 HeidelbergGermany
Summary
Protein structures cluster into families of folds that
can result from extremely different amino acid se-quences [1]. Because the enormous amount of genetic
information generates a limited number of proteinfolds [2], a particular domain structure often assumes
numerous functions. How new protein structures andnew functions evolve under these limitations remains
elusive. Molecular evolution may be driven by the abil-ity of biomacromolecules to adopt multiple conforma-
tions as a bridge between different folds [3–6]. Thiscould allow proteins to explore new structures and
new tasks while part of the structural ensemble retainsthe initial conformation and function as a safeguard
[7]. Here we show that a global structural switch canarise from single amino acid changes in cysteine-
rich domains (CRD) of cnidarian nematocyst proteins.The ability of these CRDs to form two structures with
different disulfide patterns from an identical cysteinepattern is distinctive [8]. By applying a structure-
based mutagenesis approach, we demonstrate thata cysteine-rich domain can interconvert between two
natively occurring domain structures via a bridge statecontaining both structures. Comparing cnidarian CRD
*Correspondence: [email protected] (S.M.), [email protected]
heidelberg.de (S.O.)5 Present address: Institute of Molecular Biology and Physiology,
August Krogh Building, University of Copenhagen, Universitet-
sparken 13, DK-2100 Copenhagen, Denmark.6 Present address: Imagnia AB, Cronquists Gata Building 137, S-214
28 Malmo, Sweden.
sequences leads us to believe that the mutations we
introduced to stabilize each structure reflect the birthof new protein folds in evolution.
Results
Evidence for the conformational diversity of proteinsand the biological relevance of this diversity has gath-ered over the last years. Biological processes are drivenby confined structural fluctuations [9–11] and localsecondary-structure interconversions [7, 12], as wellas tertiary- and quaternary-structural rearrangements,most prominently those involved in folding disorderssuch as prion diseases [13, 14]. Although there is notyet direct proof for the role of conformational diversityin protein evolution, an RNA sequence has been shownto assume two unrelated ribozyme folds with differentactivities in solution [4]. Such ‘‘bridge states’’ that formtwo different folds from a single sequence could evolvetwo different functions upon gene duplication and muta-tion [7, 15]. The existence of such bridge states for theconversion between fundamentally different proteinstructures has been questioned because the chemicaldiversity of 20 amino acids leads to diverse conforma-tional preferences [4].
A key position in the evolutionary tree is assumed bycnidarians, which have one of the longest fossil historiesof all metazoans [16]. The most distinctive feature ofcnidarians is the nematocyst, a specialized organelle,which discharges with nanosecond kinetics uponstimulation [17]. The walls of these osmotically chargedcapsules are stabilized by a covalent crosslinking ofcysteine-rich domains (CRDs). The short CRDs of littlemore than 20 amino acids contain a conserved patternof six cysteines (Figure 1) and form the N- and theC-terminal domain of various minicollagens in Hydra[18, 19]. In addition, they occur in the nematocystouter-wall antigen (NOWA) as a C-terminal octad repeat[20]. NOWA forms globular aggregates, which probablyfunction as positional organizers of minicollagen assem-bly [21]. Minicollagens are expressed as soluble precur-sors with three intramolecular disulfide bonds in theirCRDs. During nematocyst maturation, they crosslinkwith NOWA in an intermolecular disulfide-reshufflingreaction to form the capsule suprastructure [21]. TheN- and C-terminal domains of minicollagen 1 havebeen shown to form different structures with differentdisulfide bridges [8] and a different overall topology.The significance of cnidarians for early metazoan evolu-tion and the structural variation in closely related do-mains make the CRDs from Hydra an attractive modelfor studying the evolution of new protein folds.
The solution structure of the first CRD of NOWA (NW1)shares the disulfide pattern and overall structure of theN-terminal CRD of minicollagen 1 (Figure 1). The disul-fide pattern (8–20,12–25,16–24) differs completely fromthe one (8–24,12–20,16–25) previously determined forthe minicollagen1 C-terminal CRD (Mcol1C) [22] despite
Current Biology174
a sequence identity of 44% between the first and lastcysteine residues of NW1 and Mcol1C. The twosequences need to turn several times to bring the cyste-ine side chains together. None of the turns, however,coincide between the two structures (Figures 1A and1B). Furthermore, a conserved proline has cis conforma-tion in the NW1 fold but a trans conformation in theMcol1C fold. The NW1 structure has an overall left-handed topology, whereas the Mcol1C structure has
Figure 1. Turn Topology and Hydrogen Bonds in the NW1 and
Mcol1C Folds
(A and B) Disulfide bonds in the NW1 (A) and Mcol1C (B) domains are
depicted as solid lines, and hydrogen bonds are depicted as dotted
lines. Both domain structures are devoid of salt bridges and hydro-
phobic cores. One single hydrogen bond HN12/O’9, which is in a
bII turn in the NW1 structure and in an a helix in the minicollagen
C-terminal domain structure, is common to both domains. The
conformation of Cys20 in the NW1 structure is stabilized in a right-
handed helix conformation by a g-turn with a hydrogen bond
HN21/O’19. This interaction is destabilized in the Mcol1C se-
quence with Pro21 at the position of the putative hydrogen donor.
Also, the b-branched Val11 in the Mcol1C sequence disfavors the
formation of the NW1 fold as it disfavors the formation of a bII turn
(see Figure S1).
(C) Disulfide pattern and sequence alignment of cysteine-rich do-
mains with known structure. Detailed structural comparison (A and
B) indicates mutations (shown in light blue) that should favor
a Mcol1C-like structure over the Mcol1N-like structure in NW1.
a right-handed topology, resulting in an entirely differentappearance of the two structures.
The small cysteine-rich structures exhibit characteris-tics that distinguish them from larger folds but turnout to be beneficial for a mutational study of protein-structure conversion. The NW1 and Mcol1C structuresare both devoid of a hydrophobic core or salt bridges.Because the cysteine pattern is conserved, differentturn propensities of the noncysteine residues mustaccount for the differences in structure between thecysteine-rich domains. Consequently, noncysteine resi-dues in the NW1 and Mcol1C sequences were tested forturn propensities [23] that favor their respective domainstructure over the competing structure (Figure 1). Thispermitted a rational approach to identifying localfeatures, which favor the different domain structures.The strongest effect was predicted for mutations intro-ducing Pro21 (K21P) and Val11 (G11V) into the NW1structure (Figure 1; also Figure S1 in the SupplementalData available online).
Wild-type and mutant forms of the NW1 sequencewere recombinantly expressed as 15N-labeled proteins.Purification yielded one major species for the NW1 wild-type (Figure 2A). Amide resonances of the 1H-15N HSQCwere assigned to give a ‘‘fingerprint’’ of the wild-typestructure. In contrast to the wild-type, the K21P mutantform yielded two prominent separable species I and II ina ratio of 1:3.5 (Figure 2). Mass spectrometry confirmedthat both species have the predicted molecular weightwith three intact disulfide bonds (see SupplementalExperimental Procedures). NMR spectroscopy of bothspecies showed that they are folded and structurallystable. Resonance assignments proved that the twoK21P species differ significantly, and the chemical shiftsin the cysteine-rich core indicate a global structuralchange (Figure 2B).
The HSQC spectrum of species I closely resembledthe spectrum of the wild-type. Accordingly, indepen-dent structure determination showed that species Iretains the structure of the wild-type NW1 domain(Figure 2C; also Figure S2). As a result of the geometricconstraints imposed on the sequence by disulfide bond-ing, the g turn topology around Cys20 is retained in theK21P mutant, albeit not stabilized by a hydrogen bond.Species II, on the other hand, assumes a fold that closelyresembles the Mcol1C structure, with disulfide bondsswitched from (8–20,12–25,16–24) to (8–24,12–20,16–25), a right-handed topology instead of a left-handedtopology, and a proline switched from cis to trans(Figure 2C; also Figure S2).
Both purified structures of the K21P mutant are stablein the absence of catalysts; that is, they do not reshuffletheir intramolecular disulfide bonds. This is most likelydue to the enormous activation energy required forunfolding, proline isomerization, and the rearrangementof disulfide bonds. However, the interconversion ofdomain structures I and II does occur in redox buffercontaining reduced and oxidized glutathione (Figure 3).This indicates that conversion occurs via partially or fullyreduced states and demonstrates that the differentstructures are not populated as kinetic traps. Rather,the two domain structures are populated under equilib-rium conditions; the ground-state structure is notunique. Despite differences in high-energy structural
Molecular Evolution of a Protein-Domain Structure175
Figure 2. HPLC and NMR Analysis of an Overlapping Structural Transition
(A) Analytical HPLC runs of NW1 constructs incubated for 72 hr at 298 K in 10 mM KPi buffer (pH 7.5) containing 5 mM oxidized and reduced
glutathione, each.
(B) 1H-15N HSQC spectra of these NW1 constructs at 288 K in KPi (pH 5.5), illustrating spectral similarities within the domain structures and spec-
tral differences between the domain structures shown in (C). The most sensitive reporter on the conformational switch is the 15N upfield shift of
E19 upon formation of a canonical bI turn in domain structure II ([B], bottom).
(C) Domain structure of the lowest-energy NMR conformers out of 100 calculated (see Supplemental Experimental Procedures) for NW1 wild-
type (top) and NW1 G11V, K21P (bottom).
features, both structures have less than 0.8 kcal/molenergy difference in the K21P mutant, which thus formsa bridge state between two fundamentally different ter-tiary structures. The absence of one hydrogen bond inthe K21P mutant destabilizes structure I relative to IIby DDGf > 2.5 kcal/mol. This is in full agreement withthe energetic contribution of few kcal/mol expectedfrom a partially buried hydrogen bond in proteins [24].As a consequence, the equilibrium between domainstructures I and II, estimated by peak volumes in NMRspectroscopy and analytical HPLC, shifts from >95%structure I to only 22% structure I in the K21P mutant.Structure II in the K21P mutant can be further stabilizedrelative to structure I by the introduction of a G11Vmutation. This shifts the equilibrium proportion ofdomain structure II from approximately 78% to approx-imately 95% (Figure 2A), accordingly conferring DDG z1.0 kcal/mol to the stabilization of the new structurerelative to the previous one. A single G11V mutation inNW1, on the other hand, yields domain structures Iand II at a ratio of 70% to 30%, as determined by NMR15N HSQC peak intensities (not shown). In conclusion,the main energetic contribution for the structural switcharises from the removal of a single hydrogen bond, sup-plemented by a smaller contribution from a single intrin-sic amino acid positional potential (Figure S1).
A depiction of the structural conversion from wild-type NW1 via the bridge state K21P to the G11V K21Pmutant is shown in Figure 4. The two point mutations in-duce a nearly complete conversion of structure I in thewild-type sequence into structure II. The bridge statestructures very closely resemble the natively and artifi-cially evolved domains of the NW1 wild-type and doublemutant. The artificially induced structure II is very similarto the naturally evolved Mcol1 C-terminal domain(Figure 4A). An alignment of known minicollagen se-quences demonstrates the striking conservation of theresidues we have identified as conformational switchesbetween the different domain structures of Hydra CRDs(Figure 4B; Figure S3), indicating that the structuralpolarity between N- and C-terminal domains of Mcol1 isconserved in minicollagens. This suggests an evolution-ary scenario in which a bridge state similar to the K21Pmutant has diversified upon gene duplication to formtwo different domains fixed by their respective disulfidepatterns (Figure 4).
A possible evolutionary bridge carrying only one of thetwo mutations identified here is found in the C-terminalCRD of Hydra minicollagen 7 (Figure 4B; Figure S3).Interestingly, in minicollagen sequences of the moreprimordial sea anemone Nematostella vectensis, bothmutations can be found separately (Figure S3). Clearly,
Current Biology176
these evolved folds retain metastability in accordancewith the need to undergo structural changes upon nem-atocyst wall maturation. The fact that the NOWA octare-peat domain, which spontaneously undergoes disulfide-dependent self assembly [21], has uniform CRD folds(S.M. et al., unpublished data) strongly points to a homo-philic polymerization mechanism. Reoxidation of a solu-tion containing two differently folded CRDs (NW1 wt andMcol1C wt) and subsequent PAGE and mass spectro-metric analysis supports the notion of a spontaneoushomodimerization between CRDs of only the NW1wild-type fold. The invention of a second fold to inhibita premature elongation of the collagen polymer mighttherefore have facilitated a controlled step during finalnematocyst maturation (Figure S4). In addition, a homo-philic propagation process is presumed to result ina more extended and flexible network that adapts tothe microtubule cage harboring the growing nematocystvesicle and thus can accommodate the unique mechan-ical properties of the nematocyst.
Discussion
We demonstrate that two different naturally occurringtertiary structures of cysteine-rich domains from Hydraminicollagens are linked in sequence space by singleamino acid changes via a bridge state sequence adopt-ing both native structures. The determinants of the two
Figure 3. Conversion of Domain Structure I into Equilibrium Quanti-
ties of I and II
(A) The structural conversion is catalyzed by refolding buffer
containing 5 mM oxidized and reduced glutathione at pH 7.0 and
288 K and is monitored by 1H-15N HSQC spectra.
(B) Kinetics of structural conversion. 1H-15N HSQC peak intensities
are shown for resonances of the spectral section displayed in (A);
the averages of these intensities are fit to monoexponentials. Appar-
ent rate constants for the unfolding of domain I and the formation of
domain II are 0.125 h21 and 0.121 h21, respectively, under the given
conditions.
domain structures are strikingly conserved amongcnidarian minicollagen sequences. This argues for animportant gene-duplication and -differentiation step inthe genetic history of cnidarians for the invention ofnovel molecular phenotypes, thus pointing to the rele-vance of the continuous mutational paths retracedhere. Although previous experimental proof has beensparse, local conformational variations and secondary-structure fluctuations in a stable tertiary context [7, 12]have pointed to the possibility of global structuralswitches in proteins. The only known example of a globaltertiary-structure switch in a particular protein sequencewithout change in solvent conditions has been the inter-conversion of soluble protein and aggregating b-sheet-rich conformation in misfolding diseases. As a result ofthe dynamic instability of the folds involved, bridgestates between different structures in solution have,however, been inherently harder to detect than struc-tural populations that self-replicate and get trapped byaggregation in protein-misfolding diseases. Notably,though, this capability to self-replicate may play a devel-opmental and evolutionary role because the structuraldiversity of prion proteins has been implicated in molec-ular memory formation [14].
The use of disulfide-linked domains in our studyallowed the purification and characterization of differentstructural states that would freely interconvert for poly-peptide chains without disulfide linkage. In addition, thedomains used in this study retain marginal stability tofulfil their biological function. The structural switch ina small disulfide-rich sequence from Hydra is achievedby exclusive tailoring of dihedral angle preferencesand hydrogen-bonding properties of the mutated resi-dues, as indicated by the fact that sidechain interactionsapart from the disulfide bridges are essentially absent inthese small domains. Presumably, the structural evolu-tion of larger protein folds will further depend on a co-evolution of amino acid pairs involved in long-rangesidechain interactions [25]. In addition, a larger confor-mational space may become accessible to marginallystable proteins, which are not densely packed with di-sulfide bonds. Previously, different protein folds havebeen predicted to be separated by only a few aminoacids or even to overlap in sequence space [15, 26]. The-oretical models of protein structure and evolution thushave questioned the uniqueness of the ground-statestructure and have pointed to the relevance of avoidingoverly stable ‘‘mutational traps’’ in order to maintain theevolvability of the sequence [15], which is in agreementwith our findings. We conclude that theoretical modelingin conjunction with the experiments presented hereprovides strong support that smooth transitions maybe a widespread feature in the evolution of naturallyoccurring protein folds.
Supplemental Data
Supplemental Data include the Experimental Procedures, four
figures, and three tables and can be found online at http://www.
current-biology.com/cgi/content/full/17/2/173/DC1/.
Acknowledgments
We thank M. Rogowski for the recording of mass spectra. We thank
J. Engel, J. Stetefeld, D. Haussinger, and M. Allan for carefully read-
ing the manuscript and for helpful discussions. This research was
Molecular Evolution of a Protein-Domain Structure177
Figure 4. Structural and Disulfide Switch upon Mutation of G11V and K21P in NW1
(A) The lowest-energy NMR conformers out of 100 calculated (see Supplemental Experimental Procedures) are shown as ribbon representations
with cystines depicted in yellow and mutated residues 11 and 21 shown in red. Single point mutations of the bridge-state sequences can induce
an abundance of more than 95% of either fold. Experimental details on NMR structure determination as well as PDB identifiers are given in the
Supplemental Data; NMR structure bundles are shown in Figure S2. The comparison of the double mutant structure with the Mcol1C structure
illustrates the identity of the synthetic with the natural fold (right). The only remaining difference in the mutated NW1 sequence is a glycine (G23)
that induces a bII turn in position i + 3 of the synthetic NW1 sequence [23]; in comparison, there is a bI turn in Mcol1C.
(B) Sequence alignment of N- and C-terminal cysteine-rich domains of minicollagens from Hydra. Mutations shown to switch the domain struc-
ture are highlighted in light blue. The alignment indicates that the polarity between N- and C-terminal domains is conserved in minicollagens
because identified determinants of the different structures are conserved.
funded by the Swiss National Fund. Sequencing of Hydra ESTs was
carried out by the Genome Sequencing Center, Washington Univer-
sity, St Louis with support from the National Science Foundation.
The sequencing, assembly and annotation of the Nematostella
vectensis genome was performed under the auspices of the U.S. De-
partment of Energy’s Office of Science, Biological and Environmen-
tal Research Program and by the University of California, Lawrence
Livermore National Laboratory under Contract No. W-7405-Eng-48,
the Lawrence Berkeley National Laboratory under contract No. DE-
AC02-05CH11231, and the Los Alamos National Laboratory under
contract No. DE-AC52-06NA25396.
Received: September 13, 2006
Revised: October 25, 2006
Accepted: October 26, 2006
Published: January 22, 2007
References
1. Orengo, C.A., Michie, A.D., Jones, S., Jones, D.T., Swindells,
M.B., and Thornton, J.M. (1997). CATH-a hierarchic classifica-
tion of protein domain structures. Structure 5, 1093–1108.
2. Chothia, C. (1992). Proteins. One thousand families for the
molecular biologist. Nature 357, 543–544.
3. Jensen, R.A. (1976). Enzyme recruitment in evolution of new
function. Annu. Rev. Microbiol. 30, 409–425.
4. Schultes, E.A., and Bartel, D.P. (2000). One sequence, two ribo-
zymes: Implications for the emergence of new ribozyme folds.
Science 289, 448–452.
5. James, L.C., and Tawfik, D.S. (2003). Conformational diversity
and protein evolution–a 60-year-old hypothesis revisited.
Trends Biochem. Sci. 28, 361–368.
6. Aharoni, A., Gaidukov, L., Khersonsky, O., Mc, Q.G.S., Rood-
veldt, C., and Tawfik, D.S. (2005). The ‘evolvability’ of promiscu-
ous protein functions. Nat. Genet. 37, 73–76.
7. Cordes, M.H., Burton, R.E., Walsh, N.P., McKnight, C.J., and
Sauer, R.T. (2000). An evolutionary bridge to a new protein
fold. Nat. Struct. Biol. 7, 1129–1132.
8. Milbradt, A.G., Boulegue, C., Moroder, L., and Renner, C. (2005).
The two cysteine-rich head domains of minicollagen from Hydra
nematocysts differ in their cystine framework and overall fold
despite an identical cysteine sequence pattern. J. Mol. Biol.
354, 591–600.
Current Biology178
9. Volkman, B.F., Lipson, D., Wemmer, D.E., and Kern, D. (2001).
Two-state allosteric behavior in a single-domain signaling
protein. Science 291, 2429–2433.
10. James, L.C., Roversi, P., and Tawfik, D.S. (2003). Antibody multi-
specificity mediated by conformational diversity. Science 299,
1362–1367.
11. Pauling, L. (1940). A theory of the structure and process of
formation of antibodies. J. Am. Chem. Soc. 62, 2643–2657.
12. Van Dorn, L.O., Newlove, T., Chang, S., Ingram, W.M., and
Cordes, M.H. (2006). Relationship between sequence determi-
nants of stability for two natural homologous proteins with
different folds. Biochemistry 45, 10542–10553.
13. Bennett, M.J., Schlunegger, M.P., and Eisenberg, D. (1995). 3D
domain swapping: A mechanism for oligomer assembly. Protein
Sci. 4, 2455–2468.
14. Shorter, J., and Lindquist, S. (2005). Prions as adaptive conduits
of memory and inheritance. Nat. Rev. Genet. 6, 435–450.
15. Bornberg-Bauer, E. (1997). How are model protein structures
distributed in sequence space? Biophys. J. 73, 2393–2403.
16. Scrutton, C.T. (1979). Early fossil cnidarians. In The Origin of
Major Invertebrate Groups, M.R. House, ed. (London: Academic
Press), pp. 161–207.
17. Nuchter, T., Benoit, M., Engel, U., Ozbek, S., and Holstein, T.W.
(2006). Nanosecond-scale kinetics of nematocyst discharge.
Curr. Biol. 16, R316–R318.
18. Kurz, E.M., Holstein, T.W., Petri, B.M., Engel, J., and David, C.N.
(1991). Mini-collagens in hydra nematocytes. J. Cell Biol. 115,
1159–1169.
19. Engel, U., Pertz, O., Fauser, C., Engel, J., David, C.N., and Hol-
stein, T.W. (2001). A switch in disulfide linkage during minicolla-
gen assembly in Hydra nematocysts. EMBO J. 20, 3063–3073.
20. Engel, U., Ozbek, S., Streitwolf-Engel, R., Petri, B., Lottspeich,
F., and Holstein, T.W. (2002). Nowa, a novel protein with minicol-
lagen Cys-rich domains, is involved in nematocyst formation in
Hydra. J. Cell Sci. 115, 3923–3934.
21. Ozbek, S., Pokidysheva, E., Schwager, M., Schulthess, T., Tariq,
N., Barth, D., Milbradt, A.G., Moroder, L., Engel, J., and Holstein,
T.W. (2004). The glycoprotein NOWA and minicollagens are part
of a disulfide-linked polymer that forms the cnidarian nemato-
cyst wall. J. Biol. Chem. 279, 52016–52023.
22. Meier, S., Haussinger, D., Pokidysheva, E., Bachinger, H.P., and
Grzesiek, S. (2004). Determination of a high-precision NMR
structure of the minicollagen cysteine rich domain from Hydra
and characterization of its disulfide bond formation. FEBS
Lett. 569, 112–116.
23. Hutchinson, E.G., and Thornton, J.M. (1994). A revised set of
potentials for beta-turn formation in proteins. Protein Sci. 3,
2207–2216.
24. Deechongkit, S., Nguyen, H., Powers, E.T., Dawson, P.E., Grue-
bele, M., and Kelly, J.W. (2004). Context-dependent contribu-
tions of backbone hydrogen bonding to beta-sheet folding
energetics. Nature 430, 101–105.
25. Socolich, M., Lockless, S.W., Russ, W.P., Lee, H., Gardner, K.H.,
and Ranganathan, R. (2005). Evolutionary information for
specifying a protein fold. Nature 437, 512–518.
26. Babajide, A., Farber, R., Hofacker, I.L., Inman, J., Lapedes, A.S.,
and Stadler, P.F. (2001). Exploring protein sequence space
using knowledge-based potentials. J. Theor. Biol. 212, 35–46.
Accession Numbers
Coordinates have been deposited in the Protein Data Bank,
www.pdb.org, under PDB ID codes 2HM3 (NW1), 2HM4 (NW1
K21P, species I), 2HM5 (NW1 K21P, species II), and 2HM6 (NW1
G11V, K21P).
