Article IMF

YJMBI-64437; No. of pages: 9; 4C: 2, 4, 5, 6

Redefining the Dry Molten Globule State of Proteins

Sabine Neumaier and Thomas Kiefhaber Munich Center for Integrated Protein Science and Chemistry Department, Technische Universität München, Lichtenbergstrasse 4, D-85747 Garching, Germany

Correspondence to Thomas Kiefhaber: [email protected] http://dx.doi.org/10.1016/j.jmb.2014.04.022 Edited by J. Clarke

Abstract Dynamics and function of proteins are governed by the structural and energetic properties of the different states they adopt and the barriers separating them. In earlier work, native-state triplet–triplet energy transfer (TTET) on the villin headpiece subdomain (HP35) revealed an equilibrium between a locked native state and an unlocked native state, which are structurally similar but have different dynamic properties. The locked state is restricted to low amplitude motions, whereas the unlocked state shows increased conformational flexibility and undergoes local unfolding reactions. This classified the unlocked state as a dry molten globule (DMG), which was proposed to represent an expanded native state with loosened side-chain interactions and a solvent-shielded core. To test whether the unlocked state of HP35 is actually expanded compared to the locked state, we performed high-pressure TTET measurements. Increasing pressure shifts the equilibrium from the locked toward the unlocked state, with a small negative reaction volume for unlocking (ΔV 0 = − 1.6 ± 0.5 cm 3/mol). Therefore, rather than being expanded, the unlocked state represents an alternatively packed, compact state, demonstrating that native proteins can exist in several compact folded states, an observation with implications for protein function. The transition state for unlocking/locking, in contrast, has a largely increased volume relative to the locked and unlocked state, with respective activation volumes of 7.1 ± 0.4 cm 3/mol and 8.7 ± 0.9 cm 3/mol, indicating an expansion of the protein during the locking/unlocking transition. The presented results demonstrate the existence of both compact, low-energy and expanded, high-energy DMGs, prompting a broader definition of this state. © 2014 Elsevier Ltd. All rights reserved.

Introduction Proteins typically exist in different conformational states that are important for biological functions such as ligand binding and enzyme activity [1–3]. Partially folded states of proteins have frequently been observed during folding of larger proteins [4] and were also shown to be important for folding of apparent two-state folders [5,6]. The large majority of experimentally observed intermediate states are located on the unfolded side of the major folding barrier separating the ensemble of unfolded states (U) from the native state (N). They are typically collapsed compared to the unfolded state and contain large amounts of secondary structure, but 0022-2836/© 2014 Elsevier Ltd. All rights reserved.

the polypeptide chain is still highly solvent accessible. Due to their properties, these states were named “molten globules” [4,7–9]. Moreover, conformational substates on the native side of the major free energy barrier have been detected during unfolding of several proteins. Time-resolved one-dimensional NMR unfolding experiments on ribonuclease A revealed an intermediate in rapid equilibrium with the native state that transiently accumulates prior to the rate-limiting step for unfolding [10]. This transient unfolding intermediate shows increased side-chain flexibility but has a solvent-shielded protein interior, native secondary structure and a native hydrogen bonding network [10,11]. Similar transient unfolding intermediates were detected during urea-induced J. Mol. Biol. (2014) xx, xxx–xxx

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

2

Dry Molten Globule State of Proteins

unfolding of dihydrofolate reductase by timeresolved 19F NMR experiments [12] and during GdmCl-induced unfolding of monellin by fluorescence resonance energy transfer [13]. It was suggested that these intermediates represent the “dry molten globule” (DMG) state [9–14], which was originally proposed by Shaknovich and Finkelstein based on theoretical considerations [15]. According to their proposal, the DMG is an expanded native state with increased side-chain flexibility and a solvent-inaccessible core that is high in energy and precedes protein unfolding [15]. Native-like states with increased volume relative to the native state were observed in high-pressure folding/unfolding studies on staphylococcal nuclease [16], tendamistat [17], ribonuclease A [18], cold shock protein [19] and an ankyrin repeat domain [20]. However, in contrast to the transiently populated intermediates observed during denaturant-induced unfolding of several proteins [10–13], these expanded states represent transition states for folding/unfolding, that is, high-energy states as originally postulated for the DMG. The properties of these transition states led to the proposal that they also represent DMGs [17]. Results from our work presented here prompt a broader definition of the DMG including both expanded, high-energy states and compact, low-energy conformations, as discussed in a later section. A DMG state in equilibrium with the native state on the folded side of the major folding/unfolding barrier was recently also observed in the villin headpiece subdomain (HP35) by native-state triplet–triplet energy transfer (TTET) experiments (Fig. 1a) [21]. HP35 is a 35-amino-acid polypeptide chain that

represents the C-terminal, actin-binding domain of the villin protein. It folds independently into a three-helix-bundle structure and shows two-state equilibrium folding/unfolding when the transition is monitored by fluorescence or CD [21,22]. However, native-state TTET experiments revealed an equilibrium between two native states, N and N′, which both have a solvent-shielded core and show identical secondary structure [21]. In these experiments, the triplet donor/acceptor pair xanthone/ naphthylalanine was introduced at different positions in HP35. TTET between xanthone and naphthylalanine is a diffusion-controlled process that requires van der Waals contact between donor and acceptor [23,24]. Thus, in native proteins, local or global unfolding of the structure between the labels is required to enable TTET through loop formation (Fig. 1b). TTET can therefore distinguish between structurally similar states of a protein that differ in their dynamics of unfolding of the region between the labels. In a HP35 variant with the labels at positions 23 and 35 (HP35 Nal23/Xan35), unfolding of helix 3 is required for donor and acceptor to come into van der Waals contact and undergo TTET by loop formation (Fig. 1b). Two TTET processes were observed in the native state of the Nal23/Xan35 variant, indicative of two native conformations. One of the native states, N′, is in rapid equilibrium with a high-energy intermediate, I, that has helix 3 locally unfolded (Fig. 1). This leads to fast TTET from Xan35 to Nal23. From the other native state, N, in contrast, helix 3 cannot unfold locally prior to formation of N′ [21]. The experiments revealed that N and N′ are populated to similar amounts and interconvert on the hundreds of

Fig. 1. (a) Schematic free energy diagram of native-state dynamics in HP35 detected by TTET experiments [21] showing the locked (N), the unlocked (N′) and the partially unfolded state (I) separated from the unfolded state (U) by the major folding/unfolding barrier. (b) Schematic representation of TTET experiments to investigate native-state dynamics in HP35. The TTET labels xanthone (Xan, blue) and naphthylalanine (Nal, red) were placed at the N- and C-terminal ends of helix 3, at positions 35 and 23, respectively. Local unfolding of helix 3 to the partially unfolded intermediate, I, only occurs from N′. The experiments are started by a 4-ns laserflash at 355 nm, which produces the Xan triplet state, indicated by the asterisk. After unfolding of helix 3, loop formation occurs, which brings Xan and Nal into van der Waals contact and leads to TTET from Xan to Nal. These experiments yield the rate constants for unlocking (k− l), locking (kl) and unfolding of helix 3/loop formation (kt), as well as the equilibrium populations of N and N′.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

3

Dry Molten Globule State of Proteins

nanoseconds timescale (Fig. 1b). Further information on the properties of N and N′ came from an HP35 variant with N-terminus and C-terminus labeled (Xan0/Nal35), which revealed that the two native states differ in the relative orientation of N-terminus and C-terminus [21]. The results from four different TTET pairs in combination with results from GdmCl-induced unfolding transitions monitored by CD suggested that N and N′ are structurally similar and have similar solvent-accessible surface areas (SASA), indicating a solvent-shielded core in both states. However, the two native states have largely different dynamic properties [21]. N is rigid and locked into conformations with low amplitude motions, whereas N′ undergoes rapid, large-scale structural fluctuations including local unfolding of helix 3 (Fig. 1). N′ has increased entropy and enthalpy compared to N [21], which indicates that formation of N′ is associated with a loss in interaction energy and an increase in conformational flexibility. Due to these properties, N and N′ were termed the “locked” and “unlocked” native state, respectively [21]. An equilibrium between a locked native state and an unlocked native state of HP35 was also observed subsequently in molecular dynamics (MD) simulations, which were able to reproduce the results from native-state TTET experiments [25]. The increased conformational flexibility and the solvent-shielded core of the unlocked native state, N′, in HP35 are similar to the properties of the transient unfolding intermediates observed in other proteins [10–13] and suggested that N′ represents a DMG state [21]. To test whether the unlocked state, N′, of HP35 is an expanded native state as proposed for a DMG [15], we determined the volume difference between N and N′ in high-pressure TTET measurements. Expansion of the structure without entry of water to the core should lead to a volume increase, since void volumes were shown to increase the overall protein volume [26]. According to Le Chatelier's principle, increasing pressure would thus shift the equilibrium toward the supposedly more compact locked native state N. We performed high-pressure TTET experiments on the HP35 Nal23/Xan35 variant, which yields information on the effect of pressure on the equilibrium between the locked native state and the unlocked native state and on the rate constants for unlocking and locking (see Fig. 1b). The results show that increasing pressure shifts the equilibrium toward the unlocked state, N′, due to a slightly smaller volume of N′ compared to N. Thus, the unlocked state represents a compact, alternatively folded state rather than an expanded native state. Both the locking and unlocking reactions become slower with increasing pressure, indicating that the transition state for locking/unlocking is expanded relative to both N and N′, as originally postulated for a DMG. These results show that DMGs include both expanded, high-energy transition states and compact, low-energy states.

Results and Discussion Effect of pressure on the unlocking/locking reaction in HP35 To test whether the unlocked state of HP35 represents an expanded native state, we determined the volume difference between N and N′ in highpressure TTET experiments. Additionally, the TTET experiments yield the volume changes for formation of the transition state for the unlocking/locking process. The experiments were performed on a HP35 variant labeled with the triplet donor xanthone at position 35, which is a phenylalanine in the wild-type sequence, and the triplet acceptor naphthylalanine replacing the tryptophan residue at position 23 (Nal23/Xan35 variant). Our previous results showed that the labels in these positions have only a minor effect on HP35 stability [21]. In the HP35 Nal23/Xan35 variant, the TTET labels are located at the opposite ends of helix 3 (Fig. 1). Local unfolding of helix 3 leading to TTET can only occur in the unlocked state, N′, but not in N (Fig. 1), which leads to the kinetic mechanism for TTET shown in Fig. 1b. Since both N and N′ are significantly populated in equilibrium and I represents a high-energy intermediate that is populated to only small amounts, this mechanism leads to double exponential kinetics for native-state TTET in the Nal23/Xan35 variant at ambient pressure (Fig. 2 and Supplementary Information) with apparent rate constants of λ1 = 1.0 × 10 −6 s −1 (τ1 = 1/λ1 = 1 μs) and λ2 = 5.9 × 10 − 6 s − 1 (τ2 = 1/λ2 = 170 ns) (Fig. 3a). From the apparent rate constants and their amplitudes (A1 = 0.25, A2 = 0.75; Fig. 3b), the microscopic rate constants for locking (kl) and unlocking (k−l) and TTET through loop formation (kt) can be determined using the analytical solution of the kinetic model shown in Fig. 1b [see Eqs. (S1) to (S5) in Supplementary Information], which yields k−l = (1.8 ± 0.1) × 10 6 s −1, kl = (1.3 ± 0.1) × 10 6 s − 1 and kt = (3.2 ± 0.2) × 10 6 s − 1. These values result in an equilibrium constant for unlocking (Keq = [N′]/[N] = k− l/kl) of 1.33, corresponding to an equilibrium population of 57% N′ at 0.1 MPa. The rate constant kt reports on TTET from N′ through unfolding of helix 3 and loop formation. Since I is only populated to small amounts, a single rate constant, kt, is observed for this process, which is a function of the rate constants of local unfolding and folding of helix 3, kop and kcl, respectively, and the rate constant for loop formation, kc, according to (see Fig. 1b): kt ¼

k op  k c k cl þ k c

ð1Þ

Equation (1) is identical with the formalism that describes hydrogen exchange kinetics in folded proteins, which proceeds by the same kinetic

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

4

Dry Molten Globule State of Proteins

Fig. 2. Pressure dependence of the unlocking/locking dynamics in HP35 Nal23/Xan35 in 10 mM potassium phosphate at pH 7.0 (measured at 0.1 MPa) and 5 °C. The Xan triplet state was produced by a 4-ns laserflash at 355 nm. TTET from Xan to Nal leads to a decay in Xan triplet absorbance at 590 nm. The signal was normalized between 1 and 0. The gray trace shows the Xan triplet decay in the donor-only HP35 Xan35 protein as a reference. All curves were fitted with double exponential functions. The residuals of the fits are shown for 0.1 MPa and 390 MPa. The sum of two exponentials is sufficient to describe the data over the whole pressure range. Additionally, a minor slow kinetic phase with the intrinsic triplet lifetime of Xan is observed due to small amounts of aggregated protein. This phase is not considered in data analysis. In addition, all kinetics were fitted globally including the pressure dependence of all microscopic rate constants, ki (see Fig. 3 and Supplementary Information). For clarity, the kinetics at selected pressures between 0.1 and 390 MPa are displayed. The global fit, however, included the kinetics at all 13 measured pressures (cf. Fig. 3).

mechanism that underlies TTET but on a much slower timescale [27]. Figure 2 shows the effect of pressure between 0.1 and 390 MPa (1–3900 bar) on the TTET kinetics in the HP35 Nal23/Xan35 variant. The Xan triplet decay curves can be described by the sum of two exponentials under all conditions indicating that both N and N′ are significantly populated over the range of pressures tested. The unfolded state (U), in contrast, does not become populated to detectable amounts, as indicated by the absence of a third kinetic phase [21,28]. Both kinetic phases and their amplitudes are sensitive to pressure (Fig. 3a and b). Both apparent rate constants, λ1 and λ2, decrease with increasing pressure (Fig. 3a) whereas the amplitude of the faster kinetic phase (A2) increases at the expense of amplitude of the slower phase (A1; Fig. 3b). The analytical solution of the kinetic model for nativestate TTET in the Nal23/Xan35 variant [Eqs. (S1) to (S5)] was fitted globally to the pressure dependence of the TTET kinetics (Fig. 2) assuming a linear effect of pressure on the logarithm of the microscopic rate constants kl, k− l and kt according to ∂ ln k i ΔV 0‡ ¼− ∂p RT

ð2Þ

where ΔVi0‡ represents the molar activation volume, that is, the molar volume change between the ground state and the transition state of the respective reaction (see Fig. 4a). The global fit reveals that all reactions become slower with increasing pres-

sure (Fig. 3a) with activation volumes of 7.1 ± 0.4 cm 3/mol (11.9 Å 3/molecule), 8.7 ± 0.9 cm 3/mol (14.5 Å 3 /molecule) a n d 3 . 5 ± 0 . 4 c m 3 / m o l (5.9 Å 3 /molecule) for unlocking (k− l), locking (kl) and triplet transfer through unfolding of helix 3 and loop formation (kt), respectively. This result shows that the transition state for the unlocking/ locking reaction has a larger volume than both the locked native state and the unlocked native state (Fig. 4a). From the molar activation volumes for the unlocking (ΔV−0‡l ) and locking (ΔVl0‡) reactions, the molar reaction volume for unlocking (ΔV 0) can be obtained using Eq. (3) ∂ ln K eq ¼− ∂p

kl 0‡ ΔV 0‡ ΔV 0 k −l l −ΔV −l ¼− ¼− ∂p RT RT

∂ ln

ð3Þ

Figure 3c shows that the equilibrium constant, K eq , for forming N′ increases with increasing pressure due to a negative reaction volume of ΔV 0 = − 1.6 ± 0.5 cm 3 /mol (− 2.7 Å 3 /molecule) indicating that the unlocked state has a slightly smaller volume than N. This negative reaction volume leads to an increase in the population of N′ from 57% at 0.1 MPa to 64% at 390 MPa. Consequently, N′ does not represent an expanded form of the native state, as originally postulated for a DMG. Rather, the unlocked state is a compact folded state with even a slightly smaller volume than the native state.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

5

Dry Molten Globule State of Proteins

Interpretation of the observed activation volume for the reaction that leads to TTET through unfolding of helix 3 and loop formation is not straightforward, since the observed single rate constant for this process (kt) is a function of kop, kcl and kc according to Eq. (1), which simplifies for two limiting cases. If kc ≫ kcl, Eq. (1) becomes k t ¼ k op

ð4Þ

corresponding to the EX1 limit in the hydrogen exchange formalism. The limiting case of kcl ≫ kc results in kt ¼

k op  k c ¼ K op  k c k cl

ð5Þ

corresponding to the EX2 limit. Based on our data, we cannot decide whether TTET from N′ occurs in the EX1, EX2 or the intermediate regime and we can thus not interpret the positive activation volume observed for this reaction. However, the observed value of ΔV t0‡ = 3.5 ± 0.4 cm 3/mol is similar to the reaction volume observed for unfolding of the central part of Ala-based helical peptides (ΔV 0 = 3.3 cm 3/mol) [29]. This may indicate contributions from the helix opening/ closing equilibrium (Kop) in an EX2 regime [Eq. (5)], since it was shown that loop formation in the unfolded state (kc) has only a small activation volume around 0.5 cm 3/mol [29] and should thus have only minor contributions to ΔVt0‡ in an EX2 regime. A possible contribution to the observed volume changes arises from the effect of pressure on the pKa values of the buffer and of HP35. To facilitate comparison with our previous studies, we used phosphate-buffered solutions. Due to the effect of pressure on the pKa value of phosphate [30], the pH of the solutions decreases from 7.0 to 5.8 in the pressure range applied in our experiments. Since HP35 does not contain any His residues, this drop in pH should not lead to changes in the protonation state of the protein. Protonation of the acidic side chains and of the C-terminus should not be affected by the decrease in pH, especially since their pKa values also decrease by 0.6 units in the applied pressure range [30]. These considerations were confirmed by pH-dependent TTET measurements on the Nal23/Xan35 variant, which revealed that the kinetics are independent of pH in the range between pH 5.8 and 7.0 (Fig. S1). Structural and energetic properties of the unlocking/locking transition Comparison of the NMR and X-ray structures of HP35 with experimental results from native-state Fig. 3. Pressure dependence of (a) the two observable rate constants λ1 (●) and λ2 (○) and (b) their corresponding amplitudes A1 (●) and A2 (○) obtained from double exponential fits to the kinetics traces at each pressure (Fig. 2). Global fitting of the kinetic traces at all pressures according to Eqs. (S3) to (S6) yields the observable rate constants λ1 and λ2 and respective amplitudes A1 and A2 [black lines in (a) and(b)], as well as the microscopic rate constants for unlocking, k− l (red), locking, kl (blue), and helix 3 unfolding/loop formation, kt (green). The fit gives rate constants of k− 1 = (1.8 ± 0.1) × 10 6 s − 1 and kl = (1.3 ± 0.1) × 10 6 s − 1 and kt = (3.2 ± 0.2) × 10 6 s − 1 at 0.1 MPa. The slopes of the pressure dependencies of the microscopic rate constants yield activation volumes of 0‡ ΔV− l = 7.1 ± 0.4 cm 3/mol, ΔV l0‡ = 8.7 ± 0.9 cm 3/mol and ΔV t0‡ = 3.5 ± 0.4 cm 3/mol [Eq. (2)]. (c) Pressure dependence of the equilibrium constant for formation of N′ (Keq = [N]′/[N] = k− l/kl) calculated from the results of the global fit. The slope yields ΔV 0 = − 1.6 ± 0.5 cm 3/mol [Eq. (3)].

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

6

Dry Molten Globule State of Proteins

Fig. 4. (a) Schematic representation of the volume changes for the equilibrium between the locked native state (N) and the unlocked native state (N′) of HP35. Data were taken from the results of the pressure dependence of native-state TTET shown in Figs. 2 and 3. (b) Comparison of the X-ray structure (1YRF, blue) and the NMR structure (1VII, green) of HP35. The two hydrogen bonds that are present in the X-ray structure but missing in the NMR structure are displayed. Ser15 has two different orientations in the X-ray structure.

TTET and with results from MD simulations [25] provides a structural and energetic picture on the locking/unlocking equilibrium in native HP35. The NMR structure of HP35 [31] differs significantly from the X-ray structure [32], particularly in respect to the relative orientation of the three helices (Fig. 4b). This results in differences in packing of the core and leads to the formation of two additional hydrogen bonds between the side chains of Asp3/Arg14 and Thr13/ Ser15 in the X-ray structure compared to the NMR structure [32]. It was argued that these differences may be due to the poorer resolution of the NMR structure [32]. However, a comparison with results from TTET experiments suggests that these differences may be due to the population of different native states of HP35 in the crystal and in solution. Crystal packing may favor the locked native state (N), whereas both the locked state and the unlocked state (N′) should be nearly equally populated in solution (Fig. 3c). The observed time constant for interconversion between N and N′ of about 300 ns [1/(k−l + kl)] is fast on the NMR timescale, which results in an averaging of the NMR resonances of the two states. The differences between the crystal and the solution structure in respect to the relative orientation of the three helices is compatible with the differences between N and N′ derived from the results of TTET kinetics on various HP35 variants. These experiments showed that N-terminus and C-terminus move relative to each other in the transition from N to N′ and that helix 3 becomes more flexible, indicating different packing interactions between helix 3 and the core [21]. Comparing the structural and energetic

differences between N and N′ suggests that re-orientation of the helices in N′ leads to non-optimized geometry and orientation of intramolecular interactions, which results in the observed loss in interaction energy and increase in conformational entropy [21]. However, the structural rearrangements in N′ also yield a well-packed structure as judged by the slightly reduced volume of N′ compared to N (Fig. 4a). Results from MD simulations also suggested that N corresponds to the X-ray structure of HP35 and that N′ mainly differs from N in the structure of helix 3 [25,33]. The simulations proposed that the C-terminal part of helix 3 is partially unfolded in N′, which seems unlikely, since hydrogen/deuterium exchange studies showed that the amide protons in the C-terminal part of helix 3 have protection factors in the range 10 3 –10 4 [22]. Since N and N′ are in fast exchange and the two states are nearly equally populated, a protection factor of only about 2 would be expected for amide protons that are protected in N but free to exchange in N′. In addition, α-helices were shown to become more stable with increasing pressure due to a smaller volume of the helical state compared to the coil state [29]. This effect would result in a stabilization of the C-terminal helix with increasing pressure and would favor N, although void volumes may be eliminated in N′ by partial unfolding of helix 3. The properties of the partially unfolded state observed in MD simulations are rather compatible with the properties of the high-energy intermediate (I in Fig. 1) with helix 3 unfolded or partially unfolded, from which native-state TTET occurs in the Nal23/ Xan35 variant of HP35.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

7

Dry Molten Globule State of Proteins

In the comparison of the different HP35 structures, it should be noted that the NMR structure was solved for the wild-type HP36 protein, whereas a “pseudowild-type” X-ray structure was reported for an N27H variant of HP35. The MD simulations, in contrast, where performed on a fast folding variant of HP35 [25], which is based on the N27H variant and additionally has the two lysyl residues at positions 24 and 29 replaced by norleucines. The native state of this variant was shown to have reduced SASA of residue 24 compared to the N27H variant and a largely increased unfolding rate constant [34], which suggests major changes in the free energy landscape between the native state and the transition state for unfolding. Whereas N and N′ have similar volumes, the transition state for locking/unlocking has a larger volume than both N and N′ (Fig. 4a), indicating that the structural rearrangements during the locking/ unlocking process require a transient expansion of the molecule. Recent results on the volume changes associated with the helix–coil transition in peptides showed that the transition state for adding/removing a single helical residue to an existing helix has an increased volume of 2 cm 3/mol compared to the helical state and the coil state [29]. This increased volume was mainly ascribed to the presence of an unsatisfied hydrogen bond in the transition state. The breaking of the two hydrogen bonds between Asp3/Arg14 and Thr13/Ser15 between N and N′ may thus contribute up to 4 cm 3/mol to the increased volume of the transition state. The remaining volume increase in the transition state is likely caused by formation of void volumes due to an expansion of the molecule, which was shown to be the major origin for the increase in volume upon formation of the native state during protein folding [26]. A compact and an expanded DMG state Our results show that different types of DMG states exist on the free energy landscape for HP35, prompting us to introduce a broader definition of the DMG state. The DMG state was initially proposed to represent an expanded native state with looser packing interactions and increased flexibility that is required for unfolding and was predicted to be a high-energy state [15]. Our experiments reveal that the transition state for the locking/unlocking reaction in HP35 represents an expanded high-energy state (Fig. 4a) that has increased entropy and less favorable enthalpy but the similar SASA as the locked native state [21]. These features are evidence of a high-energy state with a dry core, weakened interactions and increased flexibility and thus satisfy the definition of the DMG as originally proposed. Transition states with increased volume compared to the native state observed for folding/ unfolding of several proteins [16–20] were also

postulated to represent DMG states [17]. These results suggest that conformational transitions out of the native state that involve either unfolding or structural rearrangements among alternatively folded states require an initial expansion of the structure involving the type of DMG state originally postulated by Shaknovich and Finkelstein [15]. Most properties of the unlocked state, N′, of HP35 also satisfy the characteristics of a DMG: increased flexibility, native-like SASA and emergence on the native side of the major folding/unfolding barrier. However, N′ has similar free energy and volume as the locked native state, N, which is surprising given the weakened interactions and increased flexibility of N′ compared to N. This finding demonstrates the existence of a compact, low-energy DMG state and implies that there are different ways to tightly pack the HP35 structure. Repacking between N and N′ leads to a compensation between entropic and enthalpic contributions to HP35 stability [21], resulting in similar free energies of the two states (Fig. 1a). The transiently populated intermediates observed during unfolding of several proteins [10–13] have properties similar to the unlocked state of HP35. In the light of our results, these intermediates may also represent compact DMG states rather than expanded states, which would imply that compact DMG states are commonly present on the native side of the major folding/unfolding barrier and become transiently populated during denaturant-induced unfolding. High-pressure NMR [26,35–40], EPR [41,42] and fluorescence studies [43–45] also detected compact, alternatively folded states on the native side of the major folding/unfolding barrier. Again, comparison with our results suggests that these pressure-induced states may represent compact and flexible unlocked DMG states. Unlocking of the optimized intramolecular interactions seems to be required prior to local and global protein unfolding. Similarly, during folding, an unlocked, compact DMG state seems to be more easily accessible from the unfolded state and precedes optimization of the geometry and the energetics of the intramolecular interactions. The increase in conformational flexibility and the ability to undergo large-scale conformational fluctuations from the unlocked state may be important for protein function such as binding and catalysis. Modulating the equilibrium between these compact folded states by binding of ligands may be the structural basis for regulation of protein function. Single molecule fluorescence studies on several proteins demonstrated that enzymes cycle between active and inactive states [46,47], compatible with our observation of distinct states with different dynamic properties in HP35. It would be interesting to determine whether the actin-binding properties of villin are correlated with one of the two states of HP35.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

8

Dry Molten Globule State of Proteins

Keywords: villin headpiece subdomain; protein dynamics; high-pressure experiments; triplet–triplet energy transfer; protein folding

Materials and Methods Sample preparation The Nal23/Xan23 variant of HP35 was synthesized and purified to N 98% purity as previously described [21]. 1-Naphthylalanine was introduced as non-natural amino acid during peptide synthesis. 9-Oxoxanthene-2-carboxylic acid was synthesized as previously described [48] and coupled to the β-amino group of α,β-diaminopropionic acid.

Abbreviations used: DMG, dry molten globule; TTET, triplet–triplet energy transfer; SASA, solvent-accessible surface areas; MD, molecular dynamics.

TTET experiments The xanthone triplet state is produced by a 4-ns laserflash at 355 nm (Quantel Nd:YAG Brilliant Laser) and detected by its absorbance band at 590 nm in a Laser Flash Photolysis Reaction Analyzer (LKS) from Applied Photophysics (Leatherhead, UK). A high-pressure cell in combination with a high-pressure setup from ISS (Urbana-Champaign, Illinois) was used for high-pressure measurements. All measurements were performed in 10 mM potassium phosphate buffer, pH 7 at 5 °C. Peptide concentrations were 50 μM and were determined by xanthone absorbance at 343 nm with ε = 3900 M −1 cm − 1. All samples were degassed prior to measurement. The triplet donor lifetime in the absence of acceptor was about 10 μs. The xanthone triplet lifetime was shown to be independent of pressure [29]. For all measurements, five kinetic traces were averaged and data were fitted double exponentially. An additional slow phase that has the same lifetime as a HP35 donor-only variant (HP35 Xan35) is observed with less than 10% of the total amplitude. This slow phase is probably due to small amounts of aggregated peptides and is not considered in analysis. Global fitting of the traces was performed using Eqs. (S4) to (S6). The ProFit software package (QuantumSoft, Zürich, Switzerland) was used for all data fitting.

Acknowledgements We thank George Rose, Lillian Chong, Buzz Baldwin, Cathy Royer, Andreas Reiner and Jeremy Sloan for discussion and comments on the manuscript and Jeremy Sloan for preparation of Fig. 4b. This work was supported by the Deutsche Forschungsgemeinschaft (SFB749-A8 and Center for Integrated Protein Science Munich).

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx.doi.org/10.1016/j.jmb.2014.04.022. Received 27 March 2014; Received in revised form 23 April 2014; Accepted 25 April 2014 Available online xxxx

References [1] Frauenfelder H, Sligar SG, Wolynes PG. The energy landscapes and motions of proteins. Science 1991;254:1598–603. [2] Eisenmesser EZ, Bosco DA, Akke M, Kern D. Enzyme dynamics during catalysis. Science 2002;295:1520–3. [3] Eisenmesser EZ, Millet O, Labeikovsky W, Korzhnev DM, Wolf-Watz M, Bosco DA, et al. Intrinsic dynamics of an enzyme underlies catalysis. Nature 2005;438:117–21. [4] Baldwin RL, Rose GD. Is protein folding hierarchic? II. Folding intermediates and transition states. Trends Biochem Sci 1999;24:77–83. [5] Bachmann A, Kiefhaber T. Apparent two-state tendamistat folding is a sequential process along a defined route. J Mol Biol 2001;306:375–86. [6] Sánchez IE, Kiefhaber T. Evidence for sequential barriers and obligatory intermediates in apparent two-state protein folding. J Mol Biol 2003;325:367–76. [7] Ptitsyn OB, Pain RH, Semisotnov GV, Zerovnik E, Razgulyaev OI. Evidence for a molten globule state as a general intermediate in protein folding. FEBS Lett 1990;262:20–4. [8] Ptitsyn O. Molten globule and protein folding. Adv Protein Chem 1995;47:83–229. [9] Baldwin RL, Rose GD. Molten globules, entropy-driven conformational change and protein folding. Curr Opin Struct Biol 2013;23:4–10. [10] Kiefhaber T, Labhardt AM, Baldwin RL. Direct NMR evidence for an intermediate preceeding the rate-limiting step in the unfolding of ribonuclease A. Nature 1995;375:513–5. [11] Kiefhaber T, Baldwin RL. Kinetics of hydrogen bond breakage in the process of unfolding of ribonuclease A measured by pulsed hydrogen exchange. Proc Natl Acad Sci U S A 1995;92:2657–61. [12] Hoeltzli SD, Frieden C. Stopped-flow NMR spectroscopy: real-time unfolding studies of 6- 19F-tryptophan-labeled Escherichia coli dihydrofolate reductase. Proc Natl Acad Sci U S A 1995;92:9318–22. [13] Jha SK, Udgaonkar JB. Direct evidence for a dry molten globule intermediate during the unfolding of a small protein. Proc Natl Acad Sci U S A 2009;106:12289–94. [14] Baldwin RL, Frieden C, Rose GD. Dry molten globule intermediates and the mechanism of protein unfolding. Proteins Struct Funct Bioinform 2010;78:2725–37. [15] Shaknovich EI, Finkelstein AV. Theory of cooperative transition in protein folding. I. Why denaturation of globular protein is a first-order phase transition. Biopolymers 1989;28:1667–80. [16] Vidugiris GJA, Markley JL, Royer CA. Evidence for a molten globule-like transition state in protein folding from determination of activation volumes. Biochemistry 1995;34:4909–12.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022

9

Dry Molten Globule State of Proteins

[17] Pappenberger G, Saudan C, Becker M, Merbach AE, Kiefhaber T. Denaturant-induced movement of the transition state of protein folding revealed by high pressure stopped-flow measurements. Proc Natl Acad Sci U S A 2000;97:17–22. [18] Panick G, Winter R. Pressure-induced unfolding/refolding of ribonuclease A: static and kinetic Fourier transform infrared spectroscopy study. Biochemistry 2000;39:1862–9. [19] Jacob MH, Saudan C, Holtermann G, Martin A, Perl D, Merbach AE, et al. Water contributes actively to the rapid crossing of a protein unfolding barrier. J Mol Biol 2002;318:837–45. [20] Rouget JB, Aksel T, Roche J, Saldana J-L, Garcia AE, Barrick D, et al. Size and sequence and the volume change of protein folding. J Am Chem Soc 2011;133:6020–7. [21] Reiner A, Henklein P, Kiefhaber T. An unlocking/relocking barrier in conformational fluctuations of villin headpiece subdomain. Proc Natl Acad Sci U S A 2010;107:4955–60. [22] McKnight CJ, Doering DS, Matsudaira PT, Kim PS. A thermostable 35-residue subdomain within villin headpiece. J Mol Biol 1996;260:126–34. [23] Krieger F, Fierz B, Bieri O, Drewello M, Kiefhaber T. Dynamics of unfolded polypeptide chains as model for the earliest steps in protein folding. J Mol Biol 2003;332:265–74. [24] Satzger H, Schmidt B, Root C, Zinth W, Fierz B, Krieger F, et al. Ultrafast quenching of the xanthone triplet by energy transfer: new insight into the intersystem crossing kinetics. J Phys Chem A 2004;108:10072–9. [25] Beauchamp KA, Ensign DL, Das R, Pande VS. Quantitative comparison of villin headpiece subdomain simulations and triplet-triplet energy transfer experiments. Proc Natl Acad Sci U S A 2011;108:12735–9. [26] Roche J, Caro JA, Norberto DR, Barthea P, Roumestand C, Schlessmand JL, et al. Cavities determine the pressure unfolding of proteins. Proc Natl Acad Sci U S A 2012;109:6945–50. [27] Hvidt A, Nielsen SO. Hydrogen exchange in proteins. Adv Protein Chem 1966;21:287–386. [28] Fierz B, Reiner A, Kiefhaber T. Local conformational fluctuations in α-helices measured by fast triplet transfer. Proc Natl Acad Sci U S A 2009;106:1057–62. [29] Neumaier S, Büttner M, Bachmann A, Kiefhaber T. Transition state and ground state properties of the helix-coil transition in peptides deduced from high pressure studies. Proc Natl Acad Sci U S A 2013;110:20988–93. [30] Neumann CRJ, Kauzmann W, Zipp A. Pressure dependence of weak acid ionization in aqueous buffers. J Phys Chem 1973;77:2687–91. [31] McKnight CJ, Doering DS, Kim PS. NMR structure of the 35residue villin headpiece subdomain. Nat Struct Biol 1997;4:180–4. [32] Chiu TK, Kubelka J, Herbst-Irmer R, Eaton WA, Hofrichter J, Davies DR. High-resolution X-ray crystal structures of the villin headpiece subdomain, an ultrafast folding protein. Proc Natl Acad Sci U S A 2005;102:7517–22.

[33] Beauchamp KA, McGibbon R, Lin Y-S, Pande VS. Simple few-state models reveal hidden complexity in protein folding. Proc Natl Acad Sci U S A 2012;109:17807–13. [34] Kubelka J, Chiu TK, Davies DR, Eaton WA, Hofrichter J. Submicrosecond protein folding. J Mol Biol 2006;359:546–53. [35] Kalbitzer HR, Görler A, Li H, Dubovskii PV, Hengstenberg W, Kowolik C, et al. 15N and 1H NMR study of histidine containing protein (HPr) from Staphylococcus carnosus at high pressure. Protein Sci 2000;9:693–703. [36] Kitahara R, Yamada H, Akasaka K, Wright PE. High pressure reveals that apomyoglobin is an equilibrium mixture from the native to the unfolded. J Mol Biol 2002;320:311–9. [37] Roche J, Dellarole Mariano, Caro JA, Guca E, Norberto DR, Yang Y, et al. Remodeling of the folding free energy landscape of staphylococcal nuclease by cavity-creating mutations. Biochemistry 2012;51:9535–46. [38] Kamatari YO, Smith LJ, Dobson CM, Akasaka K. Cavity hydration as a getaway to unfolding: an NMR study of hen lysozyme at high pressure and low temperature. Biophys Chem 2011;156:24–30. [39] Nisius L, Grzesiek S. Key stabilizing elements of protein structure identified through pressure and temperature perturbation of its hydrogen bond network. Nat Chem 2012;4:711–7. [40] Akasaka K, Kitahara R, Kamatari YO. Exploring the folding free energy landscape with pressure. Arch Biochem Biophys 2013;531:110–5. [41] Lerch MT, Horwitz J, McCoy J, Hubbell WL. Circular dichroism and site-directed spin labeling reveal structural and dynamical features of high-pressure states of myoglobin. Proc Natl Acad Sci U S A 2013;110:E4714–22. [42] McCoy J, Hubbell WL. High-pressure EPR reveals conformational equilibria and volumetric properties of spin label proteins. Proc Natl Acad Sci U S A 2011;108:1331–6. [43] Bismuto E, Sirangelo I, Irace G, Gratton E. Pressure-induced perturbation of apomyoglobin structure: fluorescence studies on native and acid compact forms. Biochemistry 1996;35:1173–8. [44] Vidugiris GJA, Royer CA. Determination of the volume changes for pressure-induced transitions of apomyoglobin between native, molten globule, and unfolded states. Biophys J 1998;75:463–70. [45] Ervin J, Larios E, Osvát S, Schulten K, Gruebele M. What causes hyperfluorescence: folding intermediates or conformationally flexible native states? Biophys J 2002;83:473–83. [46] Lu HP, Xun L, Xie XS. Single-molecule enzymatic dynamics. Science 1998;282:1877–82. [47] De Cremer G, Roeffaers MBJ, Baruah M, Sliwa M, Sels BF, Hofkens J, et al. Dynamic disorder and stepwise deactivation in a chymotryopsin catalyzed hydrolysis reaction. J Am Chem Soc 2007;129:15458–9. [48] Graham R, Lewis JR. Synthesis of 9-oxoxanthen-2-carboxylic acids. J Chem Soc Perkin Trans 1978;1:876–81.

Please cite this article as: Neumaier Sabine, Kiefhaber Thomas, Redefining the Dry Molten Globule State of Proteins, J Mol Biol (2014), http://dx.doi.org/10.1016/j.jmb.2014.04.022