SEE COMMENTARY
Probing a moving target with a plastic unfolding intermediate of an ankyrin-repeat protein Nicolas D. Werbeck* and Laura S. Itzhaki† Medical Research Council Cancer Cell Unit, Hutchison/Medical Research Council Research Centre, Hills Road, Cambridge CB2 0XZ, United Kingdom
Repeat proteins are composed of tandem arrays of 30- to 40residue structural motifs and are characterized by short-range interactions between residues close in sequence. Here we have investigated the equilibrium unfolding of D34, a 426-residue fragment of ankyrinR that comprises 12 ankyrin repeats. We show that D34 unfolds via an intermediate in which the C-terminal half of the protein is structured and the N-terminal half is unstructured. Surprisingly, however, we find that we change the unfolding process when we attempt to probe it. Single-site, moderately destabilizing mutations at the C terminus result in different intermediates dominating. The closer to the C terminus the mutation, the fewer repeats are structured in the intermediate; thus, structure in the intermediate frays from the site of the mutation. This behavior contrasts with the robust unfolding of globular proteins in which mutations can destabilize an intermediate but do not cause a different intermediate to be populated. We suggest that, for large repeat arrays, the energy landscape is very rough, with many different low-energy species containing varying numbers of folded modules so the species that dominates can be altered easily by single, conservative mutations. The multiplicity of partly folded states populated in the equilibrium unfolding of D34 is also mirrored by the kinetic folding mechanism of ankyrin-repeat proteins in which we have observed that parallel pathways are accessible from different initiation sites in the structure. protein engineering 兩 protein folding
R
epeat proteins, such as ankyrin, leucine-rich, and tetratricopeptide repeats, consist of tandem arrays of a structural motif of 20–40 aa that stack in a roughly linear fashion, creating elongated and superhelical architectures. The 33-aa ankyrin motif adopts a helix–turn–helix–loop conformation (1–3). Ankyrin repeats function as protein-interaction modules in a diverse range of cellular processes including control of the cell cycle and transcription (1). Of particular interest are a number of large ankyrin-repeat proteins (ankyrins) that interact with membrane transporters, cell-adhesion molecules, and ion channels and function as intracellular adaptors targeting a diverse range of proteins to specialized membrane domains (4–7). An ankyrin-repeat protein of an ion channel found in hair cells of the inner ear has been proposed to act as a molecular spring that, in response to mechanical stimuli produced by movement or sound, causes the ion channels to open and thereby allow electrical signals to be sent to the brain (8, 9); similar mechanical roles have been proposed for ankyrin-repeat proteins involved in other biological processes (10). The structures of repeat proteins are dominated by interactions between residues close in sequence, either within a repeat or between adjacent repeats. Thus, they are in striking contrast to the more commonly studied globular proteins in which there are many interactions between distant residues giving rise to complex topologies. This peculiar characteristic of repeat proteins raises a number of questions to researchers in protein folding. One issue in particular is the cooperativity of unfolding of repeat structures, given the lack of long-range contacts. To date, proteins containing between three and seven ankyrin repeats have been studied (11–19), and most unfold in an apparent two-state manner at equilibrium (according to the criterion of the equivalence of denaturation
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0610315104
curves obtained by using different spectroscopic probes). Ankyrinrepeat proteins, therefore, do not seem to be any less cooperatively folded than globular proteins of similar size. Analysis by Mello and Barrick (20) of the seven-ankyrin-repeat domain of Notch has suggested that cooperativity in repeat structures arises because the repeat–repeat interaction is highly stabilizing, whereas the repeats themselves are intrinsically unstable. Folding simulations of proteins containing between three and seven ankyrin repeats have predicted that, for proteins composed of larger numbers of ankyrin repeats, subdomains will be decoupled, because the entropic cost of folding is no longer balanced by the enthalpic gain (21). To investigate the extent of long-range cooperativity across multiple repeats, we have looked at the equilibrium unfolding of a large ankyrin-repeat protein, known as D34, comprising the 12 C-terminal ankyrin repeats of the 24-ankyrinrepeat domain of ankyrinR. AnkyrinR is a multidomain protein that consists of an ankyrin domain, a spectrin-binding domain, a death domain, and a C-terminal regulatory domain that links a diverse set of proteins to the membrane-associated spectrin–actin cytoskeleton. The ankyrin domain mediates many of these protein– protein interactions including ion channels, calcium-release channels, and cell-adhesion molecules (4). We find for D34 that a partly folded intermediate is populated under mildly denaturing conditions. We show by mutation that the intermediate contains folded C-terminal repeats and unfolded N-terminal repeats. We show that, when this intermediate is destabilized by mutation, then different intermediates can be detected. Thus, for a large repeat array, the energy landscape may be very rough, with many low-energy species containing varying numbers of folded modules; consequently, the species that dominates can be altered easily by single, conservative mutations. This behavior contrasts with the unfolding mechanisms observed for globular proteins that seem to be more robust (i.e., a single mutation can greatly destabilize a partly folded intermediate but cannot cause alternative intermediate species to be populated). The multiplicity of partly folded states populated in the equilibrium unfolding of D34 is also mirrored by the kinetic folding mechanism of repeat proteins in which there are parallel pathways with folding initiated at more than one site in the structure (refs. 21 and 22; unpublished data; and R. D. Hutton, A. R. Lowe, and L.S.I., unpublished data). Results and Discussion Urea-Induced Unfolding of D34 Is Reversible and Involves the Formation of a Partly Structured Intermediate. D34 consists of the 12
C-terminal ankyrin repeats of the ankyrin domain of human ankyrinR (residues 402–827). To facilitate comparison with the Author contributions: N.D.W. and L.S.I. designed research; N.D.W. performed research; N.D.W. and L.S.I. analyzed data; and N.D.W. and L.S.I. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. See Commentary on page 7735. *Present address: Max Planck Institute for Medical Research, Department of Biomolecular Mechanisms, Jahnstrasse 29, 69120 Heidelberg, Germany. †To
whom correspondence should be addressed. E-mail:
[email protected].
© 2007 by The National Academy of Sciences of the USA
PNAS 兩 May 8, 2007 兩 vol. 104 兩 no. 19 兩 7863–7868
BIOPHYSICS
Edited by Peter Wolynes, University of California at San Diego, La Jolla, CA, and approved March 13, 2007 (received for review November 23, 2006)
Fig. 1. Schematic representation of the structure of D34. The single tryptophan at position 600 is shown as well as the positions of the valine residues that were mutated to alanine. The figure was created by using the programs Swiss PDB Viewer and POV-Ray.
crystal structure, the same numbering system for amino acids and ankyrin repeats was used. In a screen of different buffer conditions, D34 was found to be more resistant to aggregation in buffers containing high concentrations of salt, and it was more stable relative to the unfolded state in higher-pH buffers. The conditions chosen for the denaturation experiments, therefore, were 50 mM Tris䡠HCl buffer (pH 8.0)/150 mM NaCl/1 mM DTT at 25°C. The unfolding process was followed by fluorescence of the single tryptophan of D34, located at position 600 in the loop between the sixth and seventh ankyrin repeats (Fig. 1), by using an excitation wavelength of 280 nm. The wavelength of maximum fluorescence is 343 nm for the native state, which is consistent with the solventexposed environment of the tryptophan in the crystal structure (23) (Fig. 2A). The spectrum of the denatured state is red-shifted with a wavelength of maximum fluorescence of 357 nm and is quenched compared with that of the native state (Fig. 2 A). However, at urea concentrations close to the unfolding transition (e.g., 2.5 M urea), the emission spectra are red-shifted but not quenched relative to the native state, suggesting that unfolding of D34 cannot be described by a simple two-state model (Fig. 2 A). The denaturation curves obtained by plotting the fluorescence intensity at various wavelengths underline this observation, particularly at higher wavelengths (Fig. 3). The data can be fitted to a three-state model for the transition between the native state (N) and the unfolded state (U), N º I º U, in which there is a hyperfluorescent, red-shifted intermediate species (I) (Eq. 1). The denaturation curves obtained at higher wavelengths contain more information on the two transitions than those obtained at lower wavelengths, but they are noisier. To overcome this problem, the data were fitted globally (i.e., the denaturation curves at 0.5-nm intervals from 323 to 373 nm were fitted by sharing the m values and midpoints of unfolding, and all of the other parameters were left unconstrained; see Materials and Methods). This approach has been applied successfully elsewhere (24). The parameters obtained are shown in Table 1. The two unfolding transitions occur within a very narrow range of urea
Fig. 2. Fluorescence behavior of D34. (A) Fluorescence spectra of D34 at different urea concentrations: 0 M (——), 2.5 M (– –), and 3.5 M (- - -). (B) Fluorescence properties of the native (——), intermediate (– –), and denatured (- - -) states as estimated from fitting the unfolding data to Eq. 1. The increased fluorescence observed at 3.5 M urea in A compared with that of the denatured state in B arises from the linear increase in the fluorescence of the denatured state as a function of urea concentration. A.U., arbitrary units. 7864 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0610315104
Fig. 3. Urea-induced unfolding of D34 monitored by tryptophan fluorescence. (A) Unfolding monitored at different emission wavelengths of 325 nm (blue), 350 nm (green), 365 nm (orange), and 370 nm (red). Solid lines represent the least-square fitting of the data (see Materials and Methods for details). (B) Overlay of the unfolding (circles) and refolding (squares) denaturation curves monitored at 325 nm (blue and purple) and 365 nm (orange and red). A.U., arbitrary units.
concentration, and the m values obtained are 3.5 and 1.7 kcal䡠 mol⫺1䡠M⫺1 (1 kcal ⫽ 4.18 kJ), respectively. Attempts to separate out the two transitions by changing the conditions, such as lowering the temperature to 10°C or adding Na2SO4 (which has been shown to stabilize intermediates), were not successful. The unfolding of D34 is reversible under the conditions used, as shown in three different experiments. First, at each wavelength the denaturation curve of refolded D34 superimposes with the unfolding curve and gives the same m values and midpoints of unfolding within experimental error (Fig. 3B). Second, in CD experiments the refolded protein recovers the helical ellipticity of the native state. Third, in stopped-flow experiments the refolded protein unfolds with the same kinetics as does the native protein (data not shown). There was no change in the unfolding behavior when measurements were made at different protein concentrations: at protein concentrations of 1.1 and 4 M, the same m values and midpoints of unfolding, within experimental error, were obtained, which indicates that none of the species are oligomeric. Transition from the Native State to the Intermediate Involves the Unfolding of Approximately Half of the Ankyrin Repeats of D34.
Because the probe in the fluorescence experiments, W600, is located in the middle of the 46-kDa protein at a distance of ⬇50 Å from both termini, we wondered what types of conformational changes it detects and whether these changes correspond to a substantial unfolding event or the local unfolding of a region in close proximity to the tryptophan. To address these questions, ureainduced unfolding of D34 was monitored by far-UV CD, which is a useful probe of ␣-helical structure. Although the denaturation curve obtained by CD shows a transition occurring in the same range of urea concentration as the transitions observed by fluorescence, in contrast to the fluorescence data there is no obvious indication of an intermediate (Fig. 4). This result can be interpreted in two different ways: (i) the fluorescence-detected intermediate represents a minor conformational change that occurs in the proximity of the tryptophan residue and does not involve the unfolding of (helical) ankyrin repeats, or (ii) the two unfolding transitions occur within too narrow a range of urea concentrations to be resolved in the CD-monitored denaturation curve, because Werbeck and Itzhaki
SEE COMMENTARY
Table 1. Summary of values obtained from fitting fluorescence and CD data to a three-state model Fluorescence
CD
Protein
mI⫺N, kcal䡠mol⫺1䡠M⫺1
D50I⫺N, M
mD⫺I, kcal䡠mol⫺1䡠M⫺1
D50D⫺I, M
mI⫺N, kcal䡠mol⫺1䡠M⫺1
D50I⫺N, M
mD⫺I, kcal䡠mol⫺1䡠M⫺1
D50D⫺I, M
Wild type V420A V486A V519A V585A V651A V717A V750A
3.46 ⫾ 0.75 3.24 ⫾ 0.28 3.14 ⫾ 1.00 2.28 ⫾ 0.03 3.24 ⫾ 0.84 3.74 ⫾ 0.62 5.04 ⫾ 0.98 7.75 ⫾ 0.19
2.64 ⫾ 0.06 2.68 ⫾ 0.05 1.97 ⫾ 0.01 1.78 ⫾ 0.49 1.85 ⫾ 0.01 2.49 ⫾ 0.10 2.48 ⫾ 0.05 2.34 ⫾ 0.01
1.69 ⫾ 0.26 1.86 ⫾ 0.17 2.19 ⫾ 0.17 2.58 ⫾ 0.91 2.21 ⫾ 0.33 1.15 ⫾ 0.07 0.55 ⫾ 0.08 0.48 ⫾ 0.07
2.32 ⫾ 0.11 2.18 ⫾ 0.06 2.59 ⫾ 0.04 2.52 ⫾ 0.02 2.56 ⫾ 0.05 1.65 ⫾ 0.24 1.81 ⫾ 0.08 2.11 ⫾ 0.24
— — 1.86 ⫾ 0.30 5.39 ⫾ 2.51 3.94 ⫾ 0.48 — — —
— — 1.88 ⫾ 0.08 1.87 ⫾ 0.07 1.79 ⫾ 0.03 — — —
— — 2.95 ⫾ 0.15 3.05 ⫾ 0.11 2.84 ⫾ 0.21 — — —
— — 2.68 ⫾ 0.03 2.64 ⫾ 0.01 2.57 ⫾ 0.03 — — —
Errors represent SDs from repeat measurements in the case of the fluorescence data and SEs of the fitted parameters in the case of the CD data.
same unfolding behavior with a hyperfluorescent intermediate (Fig. 5). This result, therefore, is inconsistent with hypothesis i, in which the formation of the intermediate corresponds to a minor change in conformation local to the tryptophan residue. To test hypothesis ii (in which both transitions detected by fluorescence involve major conformational changes but only a single transition is observed by CD because the respective midpoints are very close together), we looked at the effect of conservative mutations throughout the protein. Seven valine residues, each located at position 18 of their respective ankyrin repeat, were replaced by alanine to obtain information across the whole of the structure (Fig. 1). The mutations delete both intra- and inter-repeat contacts. Three of the mutants (V486A, V519A, and V585A), all located in the N-terminal half of the protein, showed a distinct separation of the two unfolding transitions compared with the wild
BIOPHYSICS
the signal changes associated with the two transitions are of the same sign using this probe (unlike the fluorescence data for which, at least at the higher wavelengths, the changes occur in opposite directions). D34 has, in addition to the 12 ankyrin repeats, a C-terminal extension comprising an unstructured loop that folds back onto the concave surface of the six C-terminal ankyrin repeats (repeats 19–24). The most C-terminal residue within this loop that is resolved in the crystal structure, E812, is only 6 Å away from W600 (Fig. 1). Disruption of the packing interactions between the loop and the ankyrin repeats before the unfolding of the repeats themselves could explain the unfolding behavior observed, in accordance with hypothesis i. The model was tested by using a truncated variant of D34 that lacks the C-terminal extension (⌬D34, residues 402–798). However, although this variant is destabilized relative to the wild type, it nevertheless displays qualitatively the
Fig. 4. Valine-to-alanine mutations in the N-terminal half of the protein affect the first unfolding transition. For clarity, two representative mutants, V486A (green) and V585A (blue), are shown and compared with wild type (red). Solid lines represent the least-square fitting of the data. (A) Fluorescence-monitored denaturation curves with the data at an emission wavelength of 365 nm are shown, although the solid lines represent the globalfitting results. (B) CD-monitored denaturation curves. A.U., arbitrary units; res, residue.
Werbeck and Itzhaki
Fig. 5. Valine-to-alanine mutations in the C-terminal half of the protein increase the mI⫺N value, indicative of a process involving the unfolding of a greater number of repeats. For clarity, only two mutants, V717A (green) and V750A (blue), are shown and compared with wild type (red). Solid lines represent results from least-squares fitting. (A) Fluorescence-monitored denaturation curves with the data at an emission wavelength of 365 nm are shown, although the solid lines represent the global-fitting results. The truncated variant, ⌬D34 (residues 402–798), is also shown in orange. (B) CDmonitored denaturation curves. A.U., arbitrary units; res, residue. PNAS 兩 May 8, 2007 兩 vol. 104 兩 no. 19 兩 7865
change in the environment of W600 (i.e., hypothesis i). Rather, the results indicate that the intermediate has a significantly different ␣-helical content from that of the native state and, therefore, that some of the ankyrin repeats are unfolded in the intermediate. The ellipticity of the intermediate obtained from fitting the CD data from V486A, V519A, and V585 to a three-state model suggests that the first transition represents the unfolding of approximately 6 of the 12 repeats. The relative sizes of the m values of the two transitions (which should correlate with the change in solventaccessible surface area on unfolding) seem to conflict with this interpretation of the CD data, in that if both transitions correspond to the unfolding of a similar number of repeats, then their m values should be similar. However, similar m values will only be observed if the transitions are equally cooperative; below we present evidence suggesting that this is not the case for D34. Finally, we do not know at this stage without further experiments why the intermediate is hyperfluorescent. Interestingly, a hyperfluorescent intermediate is also observed in kinetic folding and unfolding studies, and it seems to have some structural properties in common with the equilibrium intermediate (unpublished data). Hyperfluorescent intermediates have been observed for other proteins (25), the best characterized being the ␣-helical protein Im7 (26, 27). C-Terminal Mutations Result in the Population of Different Unfolding Intermediates. Mutations in the C-terminal half of the protein
Fig. 6. Hypothetical curves for the unfolding of the native state to the intermediate (red) and the intermediate to the denatured state (blue), overlaid on the observed denaturation curve (gray symbols with the results of the global fit in green). (A) Wild-type D34. (B) V585A. (C) V750A. The profiles shown correspond to an emission wavelength of 365 nm, although the results are taken from the global fitting. The hypothetical denaturation curves for the two transitions were created by using the parameters obtained from the global fitting. A.U., arbitrary units.
type. This behavior was apparent in both the fluorescence- and CD-monitored denaturation curves (Fig. 4). Fitting of the fluorescence data and the CD data gave the same m values and midpoints of unfolding as each other, within error (Table 1). From the data fitting it can be seen that the separation of the two transitions arises because the mutations have a destabilizing affect on the first transition only, with a decrease in the midpoint of 0.5–1.5 M compared with wild type, whereas the second midpoint is only slightly altered if at all. The small changes in the second midpoint on mutation may arise from an underestimation of this parameter for the wild type because of the overlap in the two transitions; a similar explanation may also apply to the different m values obtained for the wild type compared with the mutants. A fourth mutation, V420A, located in the first repeat of D34 (repeat 13), has only a very small effect on the unfolding of D34; therefore, it is not discussed further. It is easier to understand the effect of the mutations by plotting out the hypothetical transitions corresponding to the unfolding of the native state to the intermediate and the intermediate to the denatured state, as shown at the 365-nm emission wavelength in Fig. 6 overlaid on the observed denaturation curve. The mutations shift the N-to-I transition to a lower midpoint while not changing the I-to-U transition, which results in a separation of the two transitions and the intermediate being populated to a greater extent; consequently, the hyperfluorescence of I is more apparent in the denaturation curve. It is difficult to reconcile the behavior of the N-terminal mutants with a model in which the first transition corresponds to only a local 7866 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0610315104
(V651A, V717A, and V750A) have a very different effect on the unfolding behavior of D34 compared with N-terminal mutations. Instead of separating the two unfolding transitions, the mutations resulted in a large increase in the m value of the first transition, which was apparent by both fluorescence and CD. Fitting the fluorescence data to the three-state model gave a very steep first transition characterized by a large mI⫺N and a D50I⫺N shifted slightly to lower urea concentrations (by 0.2–0.35 M depending on the mutation) relative to wild type (Fig. 5 and Table 1). The second transition was very broad and characterized by a small mD⫺I. We note that the mI⫺N value is highest for V750A, followed by V717A and then V651A. The effect of the mutations can be understood most easily by examining the plots of the hypothetical N-to-I and I-to-U transition, as shown in Fig. 6 for the data at 365-nm emission. The effect of the mutations is seen also in the shape of the CD denaturation curves, in that the observed transition appears steeper compared with that of the wild type (Fig. 5). Although only one transition can be clearly discerned, the CD data do not fit well to a two-state model, consistent with there being two transitions. The reason why the two transitions are more apparent by fluorescence than by CD is that the fluorescence of the intermediate is very different from that of the native and denatured states (at higher wavelengths at least), whereas the ellipticity of the intermediate is partway between that of the native and denatured states. In summary, the results suggest that the transition from the native state to the intermediate involves a larger change in solventexposed surface area for the C-terminal mutants than for the wild type and that the more C-terminal the location of the mutation is, the greater this change in solvent-accessible surface area is. At the same time, the mD⫺I value decreases the more C-terminal the mutation. The results are summarized in Fig. 7. We can conclude that the C-terminal mutations cause different unfolding intermediates to be populated. Model for the Equilibrium Unfolding of D34. The data presented in
this article are consistent with a model (Fig. 8) in which the six N-terminal ankyrin repeats of wild-type D34 unfold first at equilibrium, followed by the six C-terminal repeats. We emphasize that six is only an approximation. Mutation of residues in the N-terminal repeats results in the separation of the midpoints of the two unfolding transitions, because the N-terminal moiety is destabilized and the C-terminal moiety is not. Thus far, therefore, the protein behaves in a similar way to globular proteins. However, destabiliWerbeck and Itzhaki
zation of the C-terminal repeats after mutation results in a much more dramatic change in the unfolding of the protein. For these mutants, the first transition now corresponds to the formation of a different intermediate consisting of a smaller number of folded repeats. The more C-terminal the location of the mutation is, the greater the number of repeats that unfold in the first transition. It is interesting that, as judged by the m values, the transition from the native state to the intermediate in wild-type D34 seems to be cooperative, whereas the transition from the intermediate to the denatured state is not. This result suggests that the N-terminal half of the protein is more cooperatively folded than the C-terminal half. In the schematic shown in Fig. 8, we interpret this lack of cooperativity in the unfolding of the C-terminal half of the wild type by proposing that intermediates containing varying numbers of folded C-terminal repeats, similar to those we observe in the unfolding of the C-terminal mutants, are also populated (albeit to a lesser extent) in the wild type. A mutation will destabilize the native state and those intermediate states that contain folded structure at that site but not an intermediate that is unstructured at that site; hence, the latter intermediate will dominate in the mutant.
Fig. 8. Model for the equilibrium unfolding of D34. (A) Wild-type D34 unfolds via an intermediate (shown in color) in which approximately the six N-terminal repeats are unfolded and the six C-terminal repeats are folded. Unfolding of this intermediate is not cooperative, and the presence of other intermediates in this transition is denoted by the species shown in gray. (B) A mutation in the Nterminal half of the protein does not alter the relative populations of the intermediates. (C) A mutation in the C-terminal half of the protein destabilizes the major intermediate, causing a different intermediate to be predominant (as shown in color), which contains fewer folded ankyrin repeats.
Werbeck and Itzhaki
SEE COMMENTARY
Pivotal Role for the Interface Between the Middle Two Repeats of D34.
The observation that the unfolding behavior of D34 is different depending on whether the mutation is located in the six N-terminal or C-terminal ankyrin repeats supports the estimation from the CD data of six repeats unfolding in the first transition. The pivotal role of the loop region between the sixth and seventh repeats (repeats 18 and 19) was noted previously in proteolysis experiments on D34 in which each six-repeat half of the protein was identified as a stable subdomain (29) and also in steered molecular dynamics simulations showing that the interactions between repeats 18 and 19 were disrupted first before the global unfolding of the protein (8). Atomic force microscopy studies of a 12-ankyrin-repeat fragment of ankyrinB indicate that six repeats unfold cooperatively before the sequential unfolding of the remaining repeats (30), and the subdomain structure may have implications for ankyrin function, because two substrate-binding sites have been identified in the 24-repeat ankyrin domain and the sites act in a synergistic manner (31). It will be interesting to determine whether ankyrin variants that have altered subdomain structures, such as the C-terminal mutations of D34 described here, respond differently to mechanical unfolding. Energy Landscape of D34. The presence of multiple intermediates implies a very rough energy landscape with many energetically similar minima, the relative populations of which are consequently affected by small perturbations such as conservative mutations. In this way, the ensemble of intermediates can be described as plastic. The picture that arises from the study of D34 fits well into the bigger picture of the folding and stability of ankyrin-repeat proteins in that the intrinsic instability of an isolated repeat gives rise to the energetic coupling of the repeats, which leads to cooperative unfolding transitions and folding mechanisms polarized within one part of the structure. The mutational studies in this work support this view: destabilizing the (C-terminal) repeats, by decreasing the intrinsic repeat stability or the stability of the inter-repeat interface, causes more repeats to unfold in one transition, because the stability of a repeat is sensitive to the folding status of the neighboring repeats. Previous studies of the seven-ankyrin-repeat Notch domain indicated a breakdown in cooperativity after mutation of residues in repeats six or seven (28). The lack of cooperativity was detected by a small discrepancy in the denaturation curves monitored by different probes. Likewise, in studies of the seven-ankyrin-repeat protein gankyrin we have observed deviations in two-state behavior upon mutation within a localized part of the structure (R. D. Hutton and L.S.I., unpublished data). By contrast, in our study we PNAS 兩 May 8, 2007 兩 vol. 104 兩 no. 19 兩 7867
BIOPHYSICS
Fig. 7. Summary of the equilibrium denaturation of wild-type D34 and mutants. Error bars represent SDs from repeated measurements. The data are also presented in Table 1. Values of mI⫺N are shown with a solid line, and those of mD-I are shown with a dashed line.
The less cooperative unfolding of the C-terminal half of D34 may be, in part, a consequence of the C-terminal extension loop. The loop folds back on the ankyrin structure and makes numerous van der Waals contacts as well as hydrogen bonds with residues in the C-terminal repeats (Fig. 1). These interactions may stabilize subdomains of the C-terminal half of the protein, causing this region to unfold in a noncooperative manner. The results obtained for the truncated variant ⌬D34 support this view. ⌬D34 behaves in a qualitatively similar manner to the C-terminal mutants in that the m value of the first transition is larger than that of the wild type, and the m value for the second transition is smaller. It is also interesting that for all of the C-terminal mutants, there is still a high degree of cooperativity in the first transition (as judged by the m values) even though this transition now involves the unfolding of a greater number of repeats than for the wild type. Therefore, six does not seem to be the maximum number of repeats that can unfold in a cooperative manner. We have not attempted to quantify the model in a manner similar to the analysis by Bradley and Barrick for Notch (28), because the lack of cooperativity observed for the unfolding of the C-terminal subdomains means that true free-energy changes cannot be calculated.
can readily observe the population of an intermediate in the unfolding of the wild-type protein by fluorescence and, for some of the mutants, by CD also, making D34 a useful model system for further studies of cooperativity in linear-repeat proteins. Moreover, because it is very well behaved, particularly considering its large size, D34 is particularly suited to such studies. It unfolds in a reversible manner and without misfolding or aggregation; this feature may be another way in which the behavior of repeat proteins differs from that of globular proteins, since the unfolding of large globular proteins is often found to be irreversible or only partly reversible because intermediate species are populated that are prone to aggregation. Computer simulations of ankyrin repeats have made a number of predictions about their properties, including that the cooperativity breaks down when the protein contains more than six or seven repeats (21). The results presented here, as well as those on the seven-ankyrin-repeat proteins Notch and gankyrin, bear out this prediction. However, the behavior of the C-terminal mutants of D34 suggests that as many as 11 ankyrin repeats can unfold in a single step, and it is possible that a variant could be designed in which all 12 repeats of D34 unfold together. Thus, the potential degree of cooperativity in ankyrin-repeat proteins seems to be even greater than perhaps was first envisaged, and the ability to modify their folding properties in a regular way is a particularly exciting prospect for future research. Materials and Methods Chemicals were obtained from Sigma–Aldrich or BDH, and enzymes and oligonucleotides were obtained from Sigma–Aldrich or Stratagene. Sequencing was performed by MRC Geneservice (Cambridge, U.K.). Protein Expression and Purification. D34 contains the 12 C-terminal ankyrin repeats of the ankyrin domain of human ankyrinR (residues 402–827) and was a kind gift from P. Michaely (University of Texas Southwestern Medical Center, Dallas, TX). It was recombinantly expressed and purified as described (23), except NaCl was used in the purification buffers instead of NaBr. Site-directed mutagenesis was performed with the QuikChange kit (Stratagene), and successful clones were identified by sequencing. Purified protein was dialyzed into Tris䡠HCl buffer (pH 8.0), 150 mM NaCl, and 1 mM DTT at 4°C overnight and stored at ⫺80°C. Protein purity was checked by SDS gel electrophoresis and mass spectrometry, and concentration was determined from the absorbance at 280 nm. Fluorescence Experiments. Aliquots of urea solutions were prepared
by dispensing the appropriate volumes of 10 M urea solution in buffer [50 mM Tris䡠HCl buffer (pH 8.0)/150 mM NaCl/1 mM DTT] and buffer alone by using a Hamilton MicroLab M. Protein stock in buffer was then added to a final concentration of 2.2 M. The samples were equilibrated at 25°C for 2 h before measurement. Fluorescence measurements were made by using a PerkinElmer luminescence spectrometer LS55. The cell was thermostated to 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16.
Mosavi LK, Cammett TJ, Desrosiers DC, Peng ZY (2004) Protein Sci 13:1435–1448. Sedgwick SG, Smerdon SJ (1999) Trends Biochem Sci 24:311–316. Main ER, Lowe AR, Mochrie SG, Jackson SE, Regan L (2005) Curr Opin Struct Biol 15:464–471. Bennett V, Baines AJ (2001) Physiol Rev 81:1353–1392. Mohler PJ, Gramolini AO, Bennett V (2002) J Cell Sci 115:1565–1566. Denker SP, Barber DL (2002) Curr Opin Cell Biol 14:214–220. Rubtsov AM, Lopina OD (2000) FEBS Lett 482:1–5. Sotomayor M, Corey DP, Schulten K (2005) Structure (London) 13:669–682. Corey DP, Garcia-Anoveros J, Holt JR, Kwan KY, Lin SY, Vollrath MA, Amalfitano A, Cheung EL, Derfler BH, Duggan A, et al. (2004) Nature 432:723–730. Howard J, Bechstedt S (2004) Curr Biol 14:R224–R226. Zeeb M, Ro ¨sner H, Zeslawski W, Canet D, Holak TA, Balbach J (2002) J Mol Biol 315:447–457. Ferreiro DU, Cervantes CF, Truhlar SM, Cho SS, Wolynes PG, Komives EA (2007) J Mol Biol 365:1201–1216. Zweifel ME, Barrick D (2001) Biochemistry 40:14357–14367. Zhang B, Peng ZY (2000) J Mol Biol 299:1121–1132. Tang KS, Guralnick BJ, Wang WK, Fersht AR, Itzhaki LS (1999) J Mol Biol 285:1869–1886. Lowe AR, Itzhaki LS (2007) J Mol Biol 365:1245–1255.
7868 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0610315104
25°C by using a water bath. The excitation wavelength was 280 nm, and the excitation and emission bandwidths were 5.0 nm. Wavelength scans between 320 and 380 nm were performed for each sample at a scan speed of 1 nm䡠s⫺1. Denaturation curves were plotted at 0.5-nm intervals between 322 and 373 nm, and the curves were globally fitted to the following equation assuming a three-state model in which the fluorescence intensity of the folded and unfolded states, FN and FU, respectively, have a linear dependence on denaturant concentration, but the fluorescence intensity of the intermediate, FI, is constant:
冉
冊
关urea兴 ⫺ D50 I⫺N RT 关urea兴 ⫺ D50 D⫺I 䡠 F I ⫹ F Uexp m D⫺I RT F⫽ 关urea兴 ⫺ D50 I⫺N 1 ⫹ exp m I⫺N RT 关urea兴 ⫺ D50 D⫺I 䡠 1 ⫹ exp m D⫺I RT
冉
FN⫹exp m I⫺N
冉
冉 冉
冉
冊冊
冊 冊冊
,
[1]
where F is the observed fluorescence intensity, m is a constant that is proportional to the increase in solvent-accessible surface area between the two states involved in the transition, D50I⫺N and mI⫺N are the midpoint and m value, respectively, for the transition between the native state, N, and the intermediate, I, and D50U⫺I and mU⫺I are the midpoint and m value, respectively, for the transition between I and the unfolded state, U. Data were fitted to this equation by using Graphpad Prism 4.0 with the m and D50 values shared between the data sets; all other parameters were not constrained. CD Experiments. To improve the signal-to-noise ratio for CD measurements, the buffer conditions were modified slightly [20 mM Tris䡠HCl buffer (pH 8.0)/150 mM NaCl/1 mM dithioerythritol] compared with the fluorescence measurements. Protein stocks were dialyzed into this buffer before measurements. Sample preparation was as for fluorescence, but the protein stock concentration was 6–9 M. Samples were incubated at 25°C for 2 h. Far-UV CD spectra were then collected on a Jasco (Easton, MD) J720 spectropolarimeter or an Aviv Circular Dichroism Spectrometer Model 215 with a cuvette of 0.4-cm path length and a thermostated cell. For native proteins (at 0 M urea), at least four spectra between 200 and 250 nm were recorded and averaged to determine whether the spectrum had the expected shape. At all other urea concentrations, the ellipticity at 222 nm was recorded, and at least 10 points were averaged. Where possible, curves were fitted to a three-state model (see Eq. 1) by using Graphpad Prism 4.0. We thank Prof. C. Dobson for use of his CD equipment and Prof. S. Perrett for helpful advice. Research in the laboratory of L.S.I. is supported by the Medical Research Council of the United Kingdom. N.D.W. was supported by the Carl-Duisberg-Stiftung (Germany). 17. Yuan C, Li J, Selby TL, Byeon IJ, Tsai MD (1999) J Mol Biol 294:201–211. 18. Mosavi LK, Williams S, Peng ZY (2002) J Mol Biol 320:165–170. 19. Devi VS, Binz HK, Stumpp MT, Pluckthun A, Bosshard HR, Jelesarov I (2004) Protein Sci 13:2864–2870. 20. Mello CC, Barrick D (2004) Proc Natl Acad Sci USA 101:14102–14107. 21. Ferreiro DU, Cho SS, Komives EA, Wolynes PG (2005) J Mol Biol 354:679–692. 22. Lowe AR, Itzhaki LS (2007) Proc Natl Acad Sci USA 104:2679–2684. 23. Michaely P, Tomchick DR, Machius M, Anderson RG (2002) EMBO J 21:6387–6396. 24. Jemth P, Day R, Gianni S, Khan F, Allen M, Daggett V, Fersht AR (2005) J Mol Biol 350:363–378. 25. Ekblad CM, Wilkinson HR, Schymkowitz JW, Rousseau F, Freund SM, Itzhaki LS (2002) J Mol Biol 320:431–442. 26. Ferguson N, Capaldi AP, James R, Kleanthous C, Radford SE (1999) J Mol Biol 286:1597– 1608. 27. Spence GR, Capaldi AP, Radford SE (2004) J Mol Biol 341:215–226. 28. Bradley CM, Barrick D (2002) J Mol Biol 324:373–386. 29. Michaely P, Bennett V (1993) J Biol Chem 268:22703–22709. 30. Lee G, Abdi K, Jiang Y, Michaely P, Bennett V, Marszalek PE (2006) Nature 440:246–249. 31. Michaely P, Bennett V (1995) J Biol Chem 270:31298–31302.
Werbeck and Itzhaki
