Subdomain Architecture and Stability of a Giant Repeat Protein

Sep 20, 2013 - ABSTRACT: Tandem repeat proteins, which are widespread in the human genome, tend to exhibit high stability and favorable expression, an...
0 downloads 0 Views 3MB Size
Article pubs.acs.org/JPCB

Subdomain Architecture and Stability of a Giant Repeat Protein Maksym Tsytlonok,† Pietro Sormanni, Pamela J. E. Rowling, Michele Vendruscolo, and Laura S. Itzhaki* Department of Chemistry, University of Cambridge, Lensfield Road, Cambridge CB2 1EW, United Kingdom S Supporting Information *

ABSTRACT: Tandem repeat proteins, which are widespread in the human genome, tend to exhibit high stability and favorable expression, and hence, they are emerging as promising protein scaffolds in alternative to antibodies in biotechnology. In order to investigate the origin of the stability of these proteins, we dissect the subdomain architecture of the giant repeat protein PR65/A, which comprises 15 α-helical HEAT repeats, using a series of truncations and deletions. We find that the N (HEAT 1−2) and the C (HEAT 14−15) subdomains are not capable of independent folding, but the addition of HEAT 13 to HEAT 14−15 results in an independently stable C-terminal subdomain (HEAT 13−15), which is in turn further stabilized by the inclusion of HEAT 12 (HEAT 12−15). We also further show that the stability of HEAT 13−15 is enhanced by its fusion to HEAT 1−2, and the artificial 5-HEAT-repeat protein thereby created (HEAT NC) behaves like a cooperative multidomain protein. We construct further variants, lacking one or both of the terminal subdomains, and find that such subdomains function as stabilizing caps within full-length PR65/A as in their absence, the central subdomain of the protein unfolds to form non-native β-sheet-like oligomers. Taken together, our results suggest that in full-length PR65/A, the more unstable regions within the central repeats are protected by the adjacent folded repeats, which thus act as gatekeepers by virtue of their greater stability.



INTRODUCTION Tandem repeat proteins, such as ankyrin, TPR (tetratricopeptide), and HEAT (huntingtin, elongation factor 3, PP2A subunit, and the lipid kinase TOR), are characterized by 20- to 50-residue structural motifs repeated multiple times in tandem. Their structures are distinct from those of the more commonly studied globular proteins in that the individual modules of a repeat protein stack to produce highly elongated, onedimensional architectures that lack long-range contacts along the sequence.1,2 The absence of sequence-distant contacts is thought to afford repeat proteins enhanced flexibility, enabling them to stretch, unfold, and refold for efficient and coordinated molecular recognition and to respond to mechanical stresses in the cell. Artificial ankyrin-repeat,3,4 TPR,5 LRR,6 and armadillorepeat7 proteins have been designed based on consensus sequences of these motifs. As these artificial proteins tend to have very high thermodynamic stability, they provide a promising platform for the creation of protein libraries as an alternative to the canonical antibody scaffold.8 Another consequence of the modular architecture is that natural repeat proteins (ankyrin repeats and LRR) are able to accommodate the insertion or deletion of one or more repeats and yet still remain correctly folded.9,10 © 2013 American Chemical Society

Here, we have dissected the subdomain architecture of the giant HEAT-repeat protein PR65/A. HEAT repeats are typically found in arrays of 3−36 repeats, with the 37−39residue HEAT repeat motif comprising two antiparallel αhelices connected by a hairpin and with adjacent motifs being connected by short loops. The 590-residue protein PR65/A, which consists of 15 HEAT repeats, acts as a scaffolding subunit in the heterotrimeric serine/threonine protein phosphatase 2A (PP2A) by facilitating the interactions between PP2A regulatory and catalytic subunits11 (Figure 1A). Our previous analysis of the unfolding mechanism of PR65/A, using sitedirected mutagenesis combined with equilibrium and kinetic experiments, showed that, rather than propagating from one or both ends as observed for many small repeat proteins,12−14 the HEAT-repeat array of PR65/A unfolds by rupturing at multiple distant sites concomitantly, leading to an equilibrium intermediate with noncontiguous folded subdomains (Figure Special Issue: Peter G. Wolynes Festschrift Received: March 7, 2013 Revised: August 17, 2013 Published: September 20, 2013 13029

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

Figure 1. PR65/A subdomain structure and folding. (A) Schematic representation of the structure of the multisubunit enzyme PP2A. The scaffolding subunit, PR65/A, is colored; the catalytic and regulatory subunits are shown in gray. (B) Schematic representation of the structural organization of PR65/A showing the five subdomains (“native”) and the sequence of unfolding events (“intermediates” and “unfolded”), as determined by ref 15; the HEAT repeats that are structured in the intermediates are indicated. (C) Overview of the series of constructs characterized in this study.

1B).15 One of these subdomains comprises the N-terminal two repeats (HEAT 1−2), and another subdomain is formed by the three repeats HEAT 11−13. Here, we find that the Nsubdomain and the C-subdomain (comprising the two repeats HEAT 14−15) are not capable of independent folding; however, the inclusion of HEAT repeat 13 does result in a soluble, independently folded C-terminal fragment (HEAT 13−15), which is stabilized further by the inclusion of HEAT repeat 12. We further show that the stability of HEAT 13−15 can also be enhanced by fusing it to HEAT 1−2. Thus, HEAT 1−2 acts as a stabilizing cap both in a natural context and in an artificial one. The artificial HEAT-repeat protein created by this fusion behaves like a multidomain protein that exhibits cooperative stability, which is promoted by the presence of favorable interactions between the two domains. Lastly, we show that the relatively high stability of the N- and Csubdomains enables them to act as structural gatekeepers that help maintain native interactions in full-length PR65/A, thereby preventing the misfolding of the more unstable HEAT repeats.

was used, and the excitation and emission bandwidths were 5 nm. Wavelength scans between 320 and 370 nm were performed for each sample at a scan speed of 1 nm s−1. For HEAT NC, the denaturation curves were plotted at 1 nm intervals between 325 and 370 nm, and the curves were globally fitted (using GraphPad Prism 5.0) to a three-state equation in which the fluorescence intensities of the folded, intermediate, and unfolded states have a linear dependence on the denaturant concentration. The m values and midpoints of unfolding were shared between the data sets; all other parameters were not constrained. This approach has been applied successfully elsewhere.18,19 Circular Dichroism. Sample preparation was the same as that used for fluorescence except that the final protein concentration was 1.5 μM. Samples were incubated at 25 °C for 2 h in 25 mM MES at pH 6.5, 1 mM DTE or 50 mM HEPES at pH 7.5, 150 mM NaCl, 1 mM DTE, and different urea concentrations. Far-UV CD spectra between 200 and 280 nm were then collected on a Chirascan circular dichroism spectrometer (Applied Photophysics) using a 0.2 cm path length cuvette thermostatted with a water bath. Stopped-Flow Fluorescence Measurements. Kinetic experiments were performed on an Applied Photophysics SX18MV stopped-flow instrument with a water bath connected to the observation cell (Thermo Neslab). The excitation wavelength was 280 nm, and emission was monitored at wavelengths above 320 nm by using a glass cutoff filter. For the unfolding experiments, protein in 50 mM HEPES at pH 7.5, 150 mM NaCl, and 1 mM DTE was rapidly mixed in a 1:10 ratio with buffer containing urea. For the refolding experiments, protein was unfolded in 5 M urea for 2 h before mixing with buffer containing low concentrations of urea. Excitation was at 280 nm using a 2 mm slit width. The protein concentration after mixing was 1 μM. Approximately six traces were obtained at each denaturant concentration, which were overlaid, averaged, and then fitted using GraphPad Prism 5.0. Native PAGE. Samples at 5 μM protein concentration were incubated in 50 mM MES at pH 6.5 containing various concentrations of urea for 2 h before diluting two-fold with



METHODS Molecular Biology. The expression vector for PR65/A with an N-terminal hexahistidine tag was a kind gift from David Barford (ICR, London, U.K.). Truncated variants were made using the QuickChange site-directed mutagenesis kit (Stratagene) and confirmed by sequencing. Protein Expression and Purification. Truncation and deletion variants of PR65/A were expressed and purified as described for the full-length protein.16 The protein concentration was determined spectrophotometrically.17 Protein was frozen after purification and stored at −80 °C. Fluorescence Spectroscopy. The fluorescence was measured using a Perkin-Elmer luminescence spectrometer LS55. Urea samples were prepared using a Hamilton Microlab 500 Series. Protein stock was added to a final concentration of 2−5 μM in 50 mM MES buffer at pH 6.5, 1 mM DTT or 50 mM HEPES at pH 7.5, 150 mM NaCl, and 1 mM DTT. The samples were equilibrated at 25 °C for 2 h before measurement. The 1 cm path length cuvette was thermostatted using a water bath (Thermo Neslab). An excitation wavelength of 280 nm 13030

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

loading buffer and running on a Novex tris-glycine gel (Invitrogen).



RESULTS AND DISCUSSION Presence of Flanking Folded Subdomains Preventing the Misfolding of Internal HEAT Repeats. PR65/A is a giant repeat protein made up of five subdomains, HEAT 1−2, HEAT 3−7, HEAT 8−10, HEAT 11−13, and HEAT 14−15 (Figure 1B).15 The N-terminal subdomain of PR65/A, comprising HEAT repeats 1 and 2, is structured in the major equilibrium unfolding intermediate, with the two adjacent subdomains, comprising HEAT repeats 3−10, being unstructured (Figure 1B), and a similar intermediate is also populated in kinetic unfolding experiments.15 The protein is destabilized upon deletion of HEAT 1−2, indicating that this subdomain provides a stabilizing cap for the HEAT repeat array (Figure S1, Supporting Information).15 Here, we investigated the effect of deleting both the N- and the C-terminal subdomains (Figure 1C). Previous analysis by far-UV CD and native PAGE showed that the variant HEAT 1−9.5 (comprising the first nine and a half repeats, residues 1−378) populates an intermediate state at ∼2.5−3.5 M urea concentrations that is α-helical in structure and monomeric.15 In contrast, upon truncation of both the Nterminus and the C-terminus, the resulting variant, HEAT 3− 9.5 (residues 81−378), forms oligomers at ∼1−1.5 M urea concentration (Figure S2A, Supporting Information), and the CD spectrum indicates that this state does not have α-helical structure and is β-sheet-like (Figure 2A,C and Table S1, Supporting Information). At higher urea concentrations, the far-UV CD spectrum of HEAT 3−9.5 reverts to the double minimum typical of α-helical structure (Figure 2C and Table S1, Supporting Information), and it is monomeric (Figure S2A, Supporting Information), indicating that the non-native oligomers dissociate and the resulting intermediate recovers its native-like α-helical structure. The misfolding of HEAT 3− 9.5 is pH-sensitive as there is no evidence for the formation of the oligomers at a higher pH of 7.5 (our standard buffer condition is pH 6.5) (Figures 2B and S2B, Supporting Information). Because we have never previously observed non-native oligomers either for full-length PR65/A or for the fragments HEAT 1−9.5, HEAT 1−11, HEAT 1−12, or HEAT 1−13,15 these results indicate that misfolding is prevented in the longer fragments and in the full-length protein by the presence of the N-terminal capping subdomain HEAT 1−2. Independently Folded C-Terminal Subdomain Formed by HEAT 13−15. Our previous analysis of the unfolding kinetics of PR65/A revealed that the C-terminal subdomain, HEAT 14−15, unfolds in a distinct step after the unfolding of the internal repeats HEAT 3−10 but before the unfolding of its adjacent subdomain, HEAT 11−13 (Figure 1B).15 Here, we investigated whether the HEAT 14−15 subdomain could be produced in isolation. We made three different constructs consisting of HEAT repeats 14−15, each one having slightly different boundary conditions, but we could not express any of them in E. coli, indicating that interactions with the adjacent subdomain HEAT 11−13 are required for HEAT 14− 15 to fold. The inclusion of HEAT 13 results in the soluble expression of C-terminal fragment HEAT 13−15 (residues 474−589) (Figure 1C). Analytical gel filtration shows that HEAT 13−15 oligomerizes under the standard conditions used for our folding analysis of full-length protein (MES buffer, pH 6.5), whereas at a higher pH (HEPES buffer, pH 7.5), HEAT

Figure 2. Unfolding and oligomerization of HEAT 3−9.5. (A, B) Comparison of the urea-induced denaturation curves, monitored by ellipticity at 222 nm, of HEAT 3−9.5 (open circles) and of HEAT 1− 9.5 (filled circles). The unfolding was measured at two different pH values, 50 mM MES at pH 6.5 (A) and 50 mM HEPES at pH 7.5 (B). (C) CD spectra of HEAT 3−9.5 at pH 6.5 as a function of urea concentration, illustrating the non-native features of the protein at pH 6.5 and 1 M urea. In all cases, the protein concentration was 3 μM.

13−15 is monomeric (Figure S3A, Supporting Information). Both the monomeric and the oligomeric forms of HEAT 13− 15 display the double minima in the far-UV CD spectra characteristic of α-helical structure (Figure S3B, Supporting Information). Urea-induced denaturation curves of HEAT 13− 15 (at the pH 7.5 conditions under which it is monomeric) show a single transition and yield the same m values and midpoints of unfolding within experimental error when monitored by CD and fluorescence (HEAT 13−15 has a single tryptophan residue at residue 477), which suggests that HEAT 13−15 unfolds in a cooperative manner (Figure 3A). The increase in stability of the C-terminal moiety resulting from the inclusion of additional repeats is particularly pronounced when HEAT 12 is added to the array (HEAT 12−15, residues 435−589) (Figure 1C). The denaturation profile of HEAT 12−15 shows a single transition, with a large shift in the midpoint of unfolding from 1.6 ± 0.02 M urea for HEAT 13−15 to 3.4 ± 0.04 M urea for HEAT 12−15 (Figure 3B). The m value, however, decreases from 2.7 ± 0.1 kcal mol−1 M−1 for HEAT 13−15 to 1.3 ± 0.1 kcal mol−1 M−1 for HEAT 12−15, suggesting that the unfolding of HEAT 12−15 is not a two-state process. Consequently, the midpoint and m obtained from the fitting must be considered as apparent values only. 13031

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

Figure 3. Urea-induced denaturation of HEAT 13−15 and HEAT 12−15. (A) Urea denaturation profiles of HEAT 13−15 monitored by fluorescence (black) and CD (gray). The fluorescence data are plotted at an emission wavelength of 360 nm, and the CD data are plotted at 222 nm. The data were fitted to a two-state equation with the following m values and midpoints: m = 2.7 ± 0.3 kcal mol−1 M−1, D50 = 1.7 ± 0.04 M and m = 2.8 ± 0.1 kcal mol−1 M−1, D50 = 1.6 ± 0.02 M for fluorescence and CD, respectively. (B) Comparison of the CD-monitored denaturation profiles of HEAT 12−15 (open symbols) and HEAT 13−15 (closed symbols). The protein concentration was 1 μM. The data for HEAT 12−15 were fitted to a two-state equation with m = 1.3 ± 0.1 kcal mol−1 M−1 and D50 = 3.4 ± 0.04 M.

Unfolding and Refolding Kinetics of HEAT 13−15. Next, the unfolding and refolding kinetics of HEAT 13−15 was investigated by using stopped-flow fluorescence. The unfolding kinetics involves an increase in fluorescence, and the traces can be fitted to a single exponential phase (Figure S4 (Supporting Information) and Figure 4). The refolding kinetics can be fitted

Figure 5. Denaturation of HEAT NC. Denaturation profiles of HEAT NC (filled symbols) monitored by fluorescence (A) and CD (B). The fluorescence data are plotted at an emission wavelength of 340 nm, and the CD data are plotted at 222 nm. The fluorescence data, for which two transitions are clearly visible, are fitted to a three-state equation. The denaturation profiles for HEAT 13−15 (open symbols) are shown for comparison.

suggest that the first transition corresponds to the unfolding of repeats 13−15 because it is associated with an increase in fluorescence similar to what is observed for the HEAT 13−15 construct. Therefore, the free energy of unfolding of repeats 13−15 in the presence of folded HEAT 1−2 is calculated to be 9.0 ± 0.1 kcal mol−1, which is 4.4 kcal mol−1 greater than the free energy of unfolding of HEAT 13−15 in isolation (4.6 ± 0.4 kcal mol−1). Thus, HEAT 13−15 gains substantial stability by fusion to the N-cap. To investigate the three-state unfolding of HEAT NC further, we attempted experiments with ANS (8-anilino-1napthalenesulphonic acid), which changes fluorescence intensity and emission wavelength upon binding exposed hydrophobic patches. Unexpectedly, we found that both HEAT NC and full-length PR65/A bind ANS under native conditions (i.e., 0 M urea) in which they are folded and monomeric (data not shown). They also bind ANS at urea concentrations in which they are partly folded and at the same urea concentrations and higher protein concentrations at which they are oligomeric. The binding of ANS to the native, monomeric state of PR65/A may be understood in view of the fact that several other proteins have been observed to behave similarly, including a number of molecular chaperones, intrinsically disordered proteins, and also two ankyrin-repeat proteins, p16INK4a20 and IκBα.21 It is possible that binding to ANS is a consequence of the elongated, nonglobular structure of PR65/A. As ANS binds to the native monomeric form of the protein, this probe does not provide us with further information on the unfolding and oligomerization

Figure 4. Urea dependence of the unfolding and refolding kinetics of HEAT 13−15 and HEAT NC. The unfolding and refolding kinetics were monitored using stopped-flow fluorescence. Rate constants for unfolding and refolding of HEAT 13−15 are shown with open symbols, and those of HEAT NC are shown with filled symbols.

to the sum of three exponentials, comprising a fast phase associated with increasing fluorescence and two slower phases associated with decreasing fluorescence (Figure S4 (Supporting Information) and Figure 4). HEAT 13−15 Stabilized by Fusion to the N-Terminal Capping Subdomain HEAT 1−2. In order to test whether the capping repeats can only stabilize their adjacent repeats or whether they are also capable of stabilizing other repeats, we designed an artificial protein consisting of HEAT 1−2 and HEAT 13−15 linked by two glycine residues (referred to subsequently as HEAT NC) (Figure 1C). Because HEAT 13− 15 is monomeric at pH 7.5, all of the experiments for HEAT NC were carried out at this pH to allow direct comparison of the two constructs. In contrast to HEAT 13−15, HEAT NC populates a hyperfluorescent intermediate at equilibrium (Figure 5). The fit to a three-state model (involving the native state, N, the denatured state, U, and an intermediate, I) gives an m value for the first transition, m(N→I), of 3.9 ± 0.06 kcal mol−1 M−1 and a midpoint, D50(N→I), of 2.3 ± 0.01 M. Taken together, our data 13032

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

Figure 6. Aggregation propensity of HEAT 3−9.5. Comparison of the intrinsic aggregation propensity (lower panel) and the structurally corrected aggregation propensity (upper panel) of HEAT 3−9.5. The regions of high intrinsic aggregation propensity (red−orange) are not exposed to the solvent. As the structure of HEAT 3−9.5 is not available, we show the corresponding part in the structure of the full-length PR65/A (part of which is shown in gray at the sides). The figure was prepared with UCSF Chimera,32 and the profiles themselves are in Figure S7, Supporting Information.

of the protein. We have instead carried out native gel analysis of HEAT NC at two protein concentrations, 2 and 10 μM, as a function of urea (Figure S5A,B, Supporting Information). These show that at 2 μM protein concentration, similar to the concentrations at which the fluorescence- and CD-monitored denaturation curves were performed, HEAT NC remains monomeric throughout the urea titration, whereas at 10 μM protein concentration, HEAT NC forms oligomers in mild urea. Moreover, when we look at the fluorescence-monitored urea denaturation curve, we see a shift to higher urea in the midpoint of the second unfolding transition when the protein concentration is increased from 2 to 10 μM (Figure S5C, Supporting Information). We conclude that the intermediate state of HEAT NC populated in urea is prone to oligomerization. Stepwise Unfolding Kinetics of HEAT NC. We also investigated the unfolding and refolding kinetics of HEAT NC. Above ∼6 M urea, the unfolding kinetic traces can be fitted to the sum of two exponentials with a fast, major phase associated with increasing fluorescence and a slower, minor phase associated with decreasing fluorescence (Figure S6 (Supporting Information) and Figure 4). The amplitude of the minor phase decreases as the urea concentration is lowered, and below 6 M urea, only the phase associated with increasing fluorescence can be observed (Figure S6, Supporting Information). It is interesting to compare the unfolding kinetics of HEAT NC with that of HEAT 13−15. Above 6 M urea, both constructs exhibit a fast phase associated with increasing fluorescence, but the slow phase seen for HEAT NC, associated with decreasing fluorescence, is absent from the unfolding kinetics of HEAT 13−15, suggesting that this slow phase corresponds to the

unfolding of HEAT repeats 1−2 of HEAT NC at high urea concentrations. The urea dependence of the fast phase for HEAT NC shows a distinct downward curvature, contrasting with the linear urea dependence observed for HEAT 13−15. A downward curvature is indicative of movement of the transition state along the reaction coordinate in accordance with Hammond behavior, or alternatively interpreted as a switch between two sequential transition states separated by a highenergy intermediate.22,23 Previously, we showed that for repeat proteins, this type of downward curvature can be rationalized in terms of the unzipping of an increasing number of repeats in the rate-limiting step with decreasing urea concentration.24,25 For HEAT NC, it appears that the high-energy intermediate is stabilized at high urea concentrations, and consequently, whereas only one unfolding phase is observed at low urea concentrations, both transition states (i.e., biphasic kinetics) are observed at high urea concentrations. The refolding kinetics of HEAT NC (Figure 4) is very similar to that of HEAT 13−15, suggesting that in HEAT NC, folding of the N-cap is not ratelimiting. Differences in unfolding but not in refolding have been observed previously for consensus-designed ankyrin repeats and TPRs of increasing lengths.26−28 Analysis of the Aggregation Propensity of PR65/A. Despite the fact that PR65/A is a stable protein composed of subdomains that share a very high degree of structural similarity, not all of our attempted truncations could be expressed. Some could not be expressed at all or were expressed in very low yield. This observation is presumably a consequence of the fact that during the expression process, these variants remain unfolded or they misfold and are consequently degraded. Moreover, we find that two truncated variants 13033

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

(HEAT 3−9.5 and HEAT 13−15), among those that could be expressed, form oligomers at nearly physiological conditions (MES buffer, pH 6.5) and low μM protein concentration. The oligomerization behavior of the two variants is different. HEAT 3−9.5 forms β-sheet-like oligomers, whereas HEAT 13−15 forms native-like α-helical oligomers, as was also observed previously for the full-length protein when the urea concentration was between 3 and 8 M.15 To further investigate this issue, we calculated the aggregation propensities of the full-length PR65/A and of the two variants that oligomerize. In the analysis, we consider two different kinds of aggregation propensities. The first type, the intrinsic aggregation propensity, is obtained using the Zyggregator method29,30 from the knowledge of the amino acid sequence and represents the propensity of aggregation from the unfolded state. It is important to stress that, in contrast to other algorithms, Zyggregator is not biased toward fibrillar, β-sheet aggregation, but its score encompasses any aggregation pathway and can be regarded as a “stickiness” score. The second type, the structurally corrected aggregation propensity, is calculated using both the sequence and the structure of the protein (see the Supporting Information) and represents the propensity to aggregate from the native state. Because only residues exposed to the solvent contribute to the structurally corrected aggregation propensity, the two profiles can differ substantially. The analysis of the intrinsic aggregation propensity of the full-length protein reveals a number of potential hotspots for aggregation (i.e., having a Zyggregator score larger than one; see Figure S7, Supporting Information). Such behavior is expected as it is common for stable, ordered proteins to have some hydrophobic regions in their sequences that initiate the folding and help stabilize the native state (see, for example, ref 31). In particular, there are three regions predicted to be extremely aggregation-promoting as they constitute a large number of residues and they have high Zyggregator scores. The first region spans from THR 74 to VAL 78 (second α-helix of HEAT repeat 2), the second region from GLY 137 to VAL 152 (HEAT repeat 4), and the third region from HIS 496 to LYS 518 (in HEAT repeats 13 and 14). In contrast to the intrinsic profile, the structurally corrected profile only has a few isolated residues with scores larger than one (Figure S7, Supporting Information), in agreement with the fact that PR65/A is a stable, soluble protein, in which aggregation-promoting regions are not exposed to the solvent. Interestingly, the two highest scoring aggregation-prone regions in the intrinsic profile are close to the termini of the two variants that oligomerize, HEAT 3−9.5 and HEAT 13−15 (Figures 6, 7, and S7, Supporting Information). Furthermore, variants HEAT 1−2, HEAT 1−2.5, and HEAT 1−3, which contain the first aggregation-prone region but not the second, were not expressed in E. coli nor were HEAT 1−4 and HEAT 1−5, which contain both the first and second regions; HEAT 1−6 could be expressed with extremely low yield, consistent with the second region being farther away from the C-terminus of this construct compared with those of HEAT 1−4 and HEAT 1−5. The second region, 137−152, corresponds to most of the fourth α-helix (and part of its hairpin) of the variant HEAT 3−9.5. The third region, 496−518, is the whole of the second α-helix and the beginning of the third α-helix of the variant HEAT 13−15 (Figure S7, Supporting Information). We first consider the case of HEAT 3−9.5. The fact that this construct lacks the N-terminal capping subdomain suggests that

Figure 7. Aggregation propensity of HEAT 13−15. Comparison of the intrinsic aggregation propensity (lower panel) and the structurally corrected aggregation propensity (upper panel) of HEAT 13−15. The upper panel shows that some residues belonging to the regions of high intrinsic aggregation propensity (red−orange) are not exposed to the solvent in this variant (see also Figure S7, Supporting Information). The figure was prepared with UCSF Chimera.32

the stability of the first repeats in HEAT 3−9.5 (namely, HEAT repeats 3−5) is diminished in the truncated variant. Indeed, the missing subdomain HEAT 1−2 not only is one of the most resistant to unfolding15 (Figure 1B) but it also exerts a stabilizing effect (Figure S1, Supporting Information). As a result, thermal fluctuations about the folded state are likely to be larger in the HEAT 3−9.5 variant than in the full-length protein, especially in HEAT 4 repeat, which contains the aggregation-promoting region. Thus, on average, the aggregation-promoting region in HEAT repeat 4 is more exposed to the solvent, and hence, it can trigger the aggregation process of the truncated variant. The formation of non-native β-sheet-like oligomers is governed by the competition between enthalpic terms, which favor the regular stacking of the aggregates, and entropic terms, which disfavor the reduction of the available conformational space.33 In the case of native-like oligomers, additional energetic terms, associated with the breaking of native contacts, further disfavor fibril formation. The balance of this competition is strongly dependent on the protein concentration and experimental conditions. In the case of full-length PR65/A, the presence of the stable subdomains HEAT 1−2 and HEAT 11−13 (Figure 1) shifts this equilibrium in favor of the native state, and in fact, even when the experimental conditions are highly destabilizing, the full-length protein only forms native-like oligomers.15 By contrast, the truncated variant HEAT 3−9.5, which lacks both of the stabilizing (gatekeeper) domains, is able to form βsheet-like oligomers at less destabilizing conditions. Another interesting aspect is the pH dependence of the oligomerization of HEAT 3−9.5, which forms oligomers at pH 6.5 but does not at pH 7.5 (Figures 2 and S2, Supporting Information). Part of the explanation could lie in the fact that at pH 6.5, the protein is less stable, resulting in enhanced fluctuations about the native state and greater exposure of aggregation-promoting residues to the solvent. Then, oligomers are not formed at pH 7.5, even at higher urea concentration, 13034

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

interactions between the two domains. These results suggest that HEAT NC may be a ideal starting point for artificial HEAT-repeat proteins to be used as scaffolds in the selection of new binding partners, in a manner similar to that of artificial consensus-designed ankyrin-repeat proteins. In this context, the construction of consensus-designed HEAT repeats was recently attempted, but they were found to not be fully monomeric and to be prone to dimerization.40 Regions of low stability in repeat proteins have been shown to facilitate their post-translational modification41,42 as well as their degradation43 and their translocation between different cellular compartments.9,44 In the case of giant repeat proteins, our previous results suggested that a heterogeneous stability distribution is important for their mode of function.15 This conclusion is illustrated by PR65/A, which is the scaffolding subunit of the heterotrimeric phosphatase PP2A; the Cterminal HEAT repeats bind the catalytic subunit of PP2A, and the N-terminal HEAT repeats bind diverse regulatory subunits (Figure 1A). Uncoupling the folding of the repeats allows some repeats to remain folded when others are unfolded; the relatively lower stability of the central moiety may enable it broaden the search area for binding partners, akin to the “fly-casting” mechanism used to describe folding upon binding of intrinsically disordered proteins to their partners.45,46 In this way, PR65/A can coordinate the binding of distantly located repeats to a diverse set of partner proteins and, thereby, the timely dephosphorylation of the many different PP2A substrates. The results presented here also point to an important consequence of the heterogeneous stability distribution in giant repeat proteins. We find that in the absence of adjacent repeats, the central subdomain of PR65/A partially unfolds to form non-native β-sheet oligomers. In the full-length protein, this central region is the least stable of the four subdomains, and therefore, no intermediate is populated in which the subdomain is unfolded in the absence of adjacent folded subdomains. Thus, cryptic misfolding signals located in the central repeats are protected by the adjacent folded repeats that act as gatekeepers by virtue of their greater stability.

because they dissolve at about 1.5 M urea (Figure 2). However, a more thorough analysis reveals another scenario. A calculation of the side-chain pKa’s and charge distribution performed with pdb2pqr34−36 shows that the total charge of the protein varies from −10 C at pH 6.5 to −12 C at pH 7.5. Therefore, the absolute value of the net charge is diminished at pH 6.5, a fact that is not captured by the Zyggregator analysis (carried out at physiological conditions) but that is widely associated to increased aggregation propensity.33,37−39 The case of HEAT 13−15 is somewhat different. This variant forms native-like oligomers, consistent with the observation that this domain is one of the most stable in the full-length protein (see Figure 1 and ref 15). A few aggregation-promoting residues, no longer protected by the helices of HEAT repeat 12, become exposed to the solvent in the truncated variant and trigger its aggregation (see Figures 7 and S7, Supporting Information). The addition of HEAT 12 to HEAT 13−15 helps shield the aggregation-promoting region 496−518 from the solvent, and it also increases the overall free energy of unfolding of the protein (Figure 3B). Because an increased stability is likely to reduce the thermal fluctuations, the aggregation-promoting residues are less available for forming intermolecular interactions in HEAT 12−15, and indeed, we find that HEAT 12−15 does not oligomerize. This same effect of protection of aggregationpromoting residues is observed when HEAT 1−2 is added to HEAT 13−15 to form HEAT NC, which does not aggregate under native conditions (Figure 5). HEAT 13−15, similarly to HEAT 3−9.5, forms oligomers at pH 6.5 but not at pH 7.5 (Figure S3, Supporting Information). The net charge on the protein varies from −1 C at pH 7.5 to 1 C at pH 6.5; hence, the absolute value does not change. However, the local distribution of the charges does change. In light of the fact that the oligomers are native-like, the charge distribution on the structure should be more important than the one along the sequence in influencing the aggregation behavior. In particular, the difference in charge is due to two histidine residues (at positions 478 and 496, with predicted pKa’s of 6.54 and 7.02) that gain a positive charge at pH 6.5. The side chains of these histidine residues are within 3 Å of negatively charged side chains in the native structure. Therefore, the local net charge in these areas, which are close to the aggregation promoting regions, is changed from −1 C at pH 7.5 to around 0 C at pH 6.5, a factor that favors aggregation.



ASSOCIATED CONTENT

S Supporting Information *

Aggregation propensity calculations, secondary structure prediction results, and additional figures, including equilibrium denaturation of HEAT 3−15, native gels of HEAT 3−9.5, characterization of HEAT 13−15, representative examples of unfolding and refolding kinetic traces for HEAT 13−15 and HEAT NC, oligomerization of HEAT NC in urea, and the predicted aggregation propensity profile of PR65/A. This material is available free of charge via the Internet at http:// pubs.acs.org.



CONCLUSION The modular architecture of repeat proteins makes them remarkably tolerant to major perturbations such as insertion or deletion of one or multiple repeats.9,10 We have exploited this property here to interrogate systematically the structural roles of the four subdomains that make up the giant repeat protein PR65/A by analyzing a series of terminal and internal deletions. We have found that the various subdomains contribute differently to the overall stability and the correct native interactions of the HEAT-repeat stack. More specifically, we have shown that the N- and C-terminal capping subdomains make large contributions to the stability of the protein. Furthermore, this stabilizing effect can also be exerted, in the case of the N-cap, in a non-natural context as the N-cap stabilizes the C-cap when the two subdomains are fused directly. The resulting five-HEAT-repeat array, HEAT NC, behaves like a two-domain protein that exhibits cooperative



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Tel.: +441223763848. Present Address †

M.T.: Vlaams Instituut voor Biotechnologie (VIB), Vrije Universiteit Brussel (VUB), Pleinlaan 2, 1050 Brussels, Belgium. Notes

The authors declare no competing financial interest. 13035

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B



Article

(20) Boice, J. A.; Fairman, R. Structural Characterization of the Tumor Suppressor p16, an Ankyrin-Like Repeat Protein. Protein Sci. 1996, 5, 1776−1784. (21) Croy, C. H.; Bergqvist, S.; Huxford, T.; Ghosh, G.; Komives, E. A. Biophysical Characterization of the Free IκBα Ankyrin Repeat Domain in Solution. Protein Sci. 2004, 13, 1767−1777. (22) Hammond, A. Correlation of Reaction Rates. J. Am. Chem. Soc. 1955, 77, 334−338. (23) Matouschek, A.; Fersht, A. R. Application of Physical Organic Chemistry to Engineered Mutants of Proteins: Hammond Postulate Behavior in the Transition State of Protein Folding. Proc. Natl. Acad. Sci. U.S.A. 1993, 90, 7814−7818. (24) Werbeck, N. D.; Rowling, P. J.; Chellamuthu, V. R.; Itzhaki, L. S. Shifting Transition States in the Unfolding of a Large Ankyrin Repeat Protein. Proc. Natl. Acad. Sci. U.S.A. 2008, 105, 9982−9987. (25) Tsytlonok, M.; Itzhaki, L. S. The How’s and Why’s of Protein Folding Intermediates. Arch. Biochem. Biophys. 2012, 531, 14−23. (26) Javadi, Y.; Main, E. R. Exploring the Folding Energy Landscape of a Series of Designed Consensus Tetratricopeptide Repeat Proteins. Proc. Natl. Acad. Sci. U.S.A. 2009, 106, 17383−17388. (27) Wetzel, S. K.; Settanni, G.; Kenig, M.; Binz, H. K.; Pluckthun, A. Folding and Unfolding Mechanism of Highly Stable Full-Consensus Ankyrin Repeat Proteins. J. Mol. Biol. 2008, 376, 241−257. (28) Main, E. R.; Stott, K.; Jackson, S. E.; Regan, L. Local and LongRange Stability in Tandemly Arrayed Tetratricopeptide Repeats. Proc. Natl. Acad. Sci. U.S.A. 2005, 102, 5721−5726. (29) Tartaglia, G. G.; Pawar, A. P.; Campioni, S.; Dobson, C. M.; Chiti, F.; Vendruscolo, M. Prediction of Aggregation-Prone Regions in Structured Proteins. J. Mol. Biol. 2008, 380, 425−436. (30) Pawar, A. P.; Dubay, K. F.; Zurdo, J.; Chiti, F.; Vendruscolo, M.; Dobson, C. M. Prediction of “Aggregation-Prone” and “AggregationSusceptible” Regions in Proteins Associated with Neurodegenerative Diseases. J. Mol. Biol. 2005, 350, 379−392. (31) Tartaglia, G. G.; Vendruscolo, M. Proteome-Level Interplay between Folding and Aggregation Propensities of Proteins. J. Mol. Biol. 2010, 402, 919−928. (32) Pettersen, E. F.; Goddard, T. D.; Huang, C. C.; Couch, G. S.; Greenblatt, D. M.; Meng, E. C.; Ferrin, T. E. UCSF ChimeraA Visualization System for Exploratory Research and Analysis. J. Comput. Chem. 2004, 25, 1605−1612. (33) Chiti, F.; Dobson, C. M. Protein Misfolding, Functional Amyloid, and Human Disease. Annu. Rev. Biochem. 2006, 75, 333−366. (34) Dolinsky, T. J.; Czodrowski, P.; Li, H.; Nielsen, J. E.; Jensen, J. H.; Klebe, G.; Baker, N. A. Pdb2pqr: Expanding and Upgrading Automated Preparation of Biomolecular Structures for Molecular Simulations. Nucleic Acids Res. 2007, 35, 522−525. (35) Dolinsky, T. J.; Nielsen, J. E.; McCammon, J. A.; Baker, N. A. Pdb2pqr: An Automated Pipeline for the Setup of Poisson− Boltzmann Electrostatics Calculations. Nucleic Acids Res. 2004, 32, 665−667. (36) Li, H.; Robertson, A. D.; Jensen, J. H. Very Fast Empirical Prediction and Rationalization of Protein pKa Values. Proteins 2005, 61, 704−721. (37) Chiti, F.; Stefani, M.; Taddei, N.; Ramponi, G.; Dobson, C. M. Rationalization of the Effects of Mutations on Peptide and Protein Aggregation Rates. Nature 2003, 424, 805−808. (38) Kurnik, M.; Hedberg, L.; Danielsson, J.; Oliveberg, M. Folding without Charges. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 5705−5710. (39) Perchiacca, J. M.; Ladiwala, A. R.; Bhattacharya, M.; Tessier, P. M. Aggregation-Resistant Domain Antibodies Engineered with Charged Mutations near the Edges of the ComplementarityDetermining Regions. Protein Eng., Des. Sel. 2012, 25, 591−601. (40) Urvoas, A.; Guellouz, A.; Valerio-Lepiniec, M.; Graille, M.; Durand, D.; Desravines, D. C.; van Tilbeurgh, H.; Desmadril, M.; Minard, P. Design, Production and Molecular Structure of a New Family of Artificial Alpha-Helicoidal Repeat Proteins (αRep) Based on Thermostable HEAT-Like Repeats. J. Mol. Biol. 2010, 404, 307−327. (41) Coleman, M. L.; McDonough, M. A.; Hewitson, K. S.; Coles, C.; Mecinovic, J.; Edelmann, M.; Cook, K. M.; Cockman, M. E.;

ACKNOWLEDGMENTS M.T. was supported by a studentship from the Medical Research Council of the U.K. (MRC). L.S.I. was supported by the MRC (Grant G1002329) and the Medical Research Foundation of the U.K.



REFERENCES

(1) D’Andrea, L. D.; Regan, L. TPR Proteins: The Versatile Helix. Trends. Biochem. Sci. 2003, 28, 655−662. (2) Kobe, B.; Kajava, A. V. The Leucine-Rich Repeat as a Protein Recognition Motif. Curr. Opin. Struct. Biol. 2001, 11, 725−732. (3) Mosavi, L. K.; Minor, D. L., Jr.; Peng, Z. Y. Consensus-Derived Structural Determinants of the Ankyrin Repeat Motif. Proc. Natl. Acad. Sci. U.S.A. 2002, 99, 16029−16034. (4) Binz, H. K.; Stumpp, M. T.; Forrer, P.; Amstutz, P.; Pluckthun, A. Designing Repeat Proteins: Well-Expressed, Soluble and Stable Proteins from Combinatorial Libraries of Consensus Ankyrin Repeat Proteins. J. Mol. Biol. 2003, 332, 489−503. (5) Main, E. R.; Xiong, Y.; Cocco, M. J.; D’Andrea, L.; Regan, L. Design of Stable α-Helical Arrays from an Idealized TPR Motif. Structure 2003, 11, 497−508. (6) Stumpp, M. T.; Forrer, P.; Binz, H. K.; Pluckthun, A. Designing Repeat Proteins: Modular Leucine-Rich Repeat Protein Libraries Based on the Mammalian Ribonuclease Inhibitor Family. J. Mol. Biol. 2003, 332, 471−487. (7) Varadamsetty, G.; Tremmel, D.; Hansen, S.; Parmeggiani, F.; Pluckthun, A. Designed Armadillo Repeat Proteins: Library Generation, Characterization and Selection of Peptide Binders with High Specificity. J. Mol. Biol. 2012, 424, 68−87. (8) Boersma, Y. L.; Pluckthun, A. Darpins and Other Repeat Protein Scaffolds: Advances in Engineering and Applications. Curr. Opin. Biotechnol. 2011, 22, 849−857. (9) Kloss, E.; Barrick, D. C-Terminal Deletion of Leucine-Rich Repeats from YopM Reveals a Heterogeneous Distribution of Stability in a Cooperatively Folded Protein. Protein Sci. 2009, 18, 1948−1960. (10) Tripp, K. W.; Barrick, D. The Tolerance of a Modular Protein to Duplication and Deletion of Internal Repeats. J. Mol. Biol. 2004, 344, 169−178. (11) Hemmings, B. A.; Adams-Pearson, C.; Maurer, F.; Muller, P.; Goris, J.; Merlevede, W.; Hofsteenge, J.; Stone, S. R. α- and β-Forms of the 65-Kd Subunit of Protein Phosphatase 2A Have a Similar 39 Amino Acid Repeating Structure. Biochemistry 1990, 29, 3166−3173. (12) Tang, K. S.; Fersht, A. R.; Itzhaki, L. S. Sequential Unfolding of Ankyrin Repeats in Tumor Suppressor p16. Structure 2003, 11, 67−73. (13) Bradley, C. M.; Barrick, D. The Notch Ankyrin Domain Folds via a Discrete, Centralized Pathway. Structure 2006, 14, 1303−1312. (14) Lowe, A. R.; Itzhaki, L. S. Rational Redesign of the Folding Pathway of a Modular Protein. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 2679−2684. (15) Tsytlonok, M.; Craig, P. O.; Sivertsson, E.; Serquera, D.; Perrett, S.; Best, R. B.; Wolynes, P. G.; Itzhaki, L. S. Complex Energy Landscape of a Giant Repeat Protein. Structure 2013, in press. (16) Groves, M. R.; Hanlon, N.; Turowski, P.; Hemmings, B. A.; Barford, D. The Structure of the Protein Phosphatase 2A PR65/A Subunit Reveals the Conformation of Its 15 Tandemly Repeated HEAT Motifs. Cell 1999, 96, 99−110. (17) Gill, S. C.; von Hippel, P. H. Calculation of Protein Extinction Coefficients from Amino Acid Sequence Data. Anal. Biochem. 1989, 182, 319−326. (18) Jemth, P.; Day, R.; Gianni, S.; Khan, F.; Allen, M.; Daggett, V.; Fersht, A. R. The Structure of the Major Transition State for Folding of an FF Domain from Experiment and Simulation. J. Mol. Biol. 2005, 350, 363−378. (19) Werbeck, N. D.; Itzhaki, L. S. Probing a Moving Target with a Plastic Unfolding Intermediate of an Ankyrin-Repeat Protein. Proc. Natl. Acad. Sci. U.S.A. 2007, 104, 7863−7868. 13036

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037

The Journal of Physical Chemistry B

Article

Lancaster, D. E.; Kessler, B. M.; et al. Asparaginyl Hydroxylation of the Notch Ankyrin Repeat Domain by Factor Inhibiting HypoxiaInducible Factor. J. Biol. Chem. 2007, 282, 24027−24038. (42) Low, C.; Homeyer, N.; Weininger, U.; Sticht, H.; Balbach, J. Conformational Switch Upon Phosphorylation: Human CDK Inhibitor p19INK4d between the Native and Partially Folded State. ACS Chem. Biol. 2009, 4, 53−63. (43) Truhlar, S. M.; Mathes, E.; Cervantes, C. F.; Ghosh, G.; Komives, E. A. Pre-Folding IκBα Alters Control of NF-κB Signaling. J. Mol. Biol. 2008, 380, 67−82. (44) Renn, J. P.; Junker, M.; Besingi, R. N.; Braselmann, E.; Clark, P. L. ATP-Independent Control of Autotransporter Virulence Protein Transport via the Folding Properties of the Secreted Protein. Chem. Biol. 2012, 19, 287−296. (45) Shoemaker, B. A.; Portman, J. J.; Wolynes, P. G. Speeding Molecular Recognition by Using the Folding Funnel: The Fly-Casting Mechanism. Proc. Natl. Acad. Sci. U.S.A. 2000, 97, 8868−8873. (46) Trizac, E.; Levy, Y.; Wolynes, P. G. Capillarity Theory for the Fly-Casting Mechanism. Proc. Natl. Acad. Sci. U.S.A. 2010, 107, 2746− 2750.

13037

dx.doi.org/10.1021/jp402360x | J. Phys. Chem. B 2013, 117, 13029−13037