Article pubs.acs.org/JPCB
Sub-millisecond Chain Collapse of the Escherichia coli Globin ApoHmpH Li Zhu,†,‡ Neşe Kurt,§,⊥ Jennifer Choi,§,∥ Lisa J. Lapidus,*,† and Silvia Cavagnero*,§ †
Department of Physics and Astronomy, Michigan State University, East Lansing, Michigan 48824, United States Advanced Photonics Center, Southeast University, Nanjing 210096, China § Department of Chemistry, University of Wisconsin-Madison, Madison, Wisconsin 53706, United States ‡
S Supporting Information *
ABSTRACT: Myoglobins are ubiquitous proteins that play a seminal role in oxygen storage, transport, and NO metabolism. The folding mechanism of apomyoglobins from different species has been studied to a fair extent over the last two decades. However, integrated investigations of the entire process, including both the early (sub-ms) and late (ms−s) folding stages, have been missing. Here, we study the folding kinetics of the single-Trp Escherichia coli globin apoHmpH via a combination of continuous-flow microfluidic and stopped-flow approaches. A rich series of molecular events emerges, spanning a very wide temporal range covering more than 7 orders of magnitude, from sub-microseconds to tens of seconds. Variations in fluorescence intensity and spectral shifts reveal that the protein region around Trp120 undergoes a fast collapse within the 8 μs mixing time and gradually reaches a native-like conformation with a half-life of 144 μs from refolding initiation. There are no further fluorescence changes beyond ca. 800 μs, and folding proceeds much more slowly, up to 20 s, with acquisition of the missing helicity (ca. 30%), long after consolidation of core compaction. The picture that emerges is a gradual acquisition of native structure on a free-energy landscape with few large barriers. Interestingly, the single tryptophan, which lies within the main folding core of globins, senses some local structural consolidation events after establishment of native-like core polarity (i.e., likely after core dedydration). In all, this work highlights how the main core of the globin fold is capable of becoming fully native efficiently, on the sub-millisecond time scale.
■
INTRODUCTION The globin fold stands out as one of the most ancient classes of protein structure in Nature, and is ubiquitously represented across all domains of life: prokarya, eukarya, and archaea.1−4 Globins have evolved to serve a set of essential biological functions, spanning from oxygen storage and transport to the metabolism of key signaling agents such as nitric oxide. Among the two main types of globin substructure categories, denoted as 3-3 and 2-2, the 3-3 motif is by far the most widely represented. The 3-3 globins comprise eight helices, typically labeled as A through H, from the N to the C terminus. The noncovalently bound heme, which is essential for function,2 can be removed5 or omitted6 with no effect on the overall integrity of the fold, except for a decreased helicity in the small helix F.7 Some globins differ in sequence by more than 83%, yet share a near identical three-dimensional structure. This amazing robustness toward variations in amino acid sequence is contrasted by the variable thermodynamic stability of the monomeric members of this fold, which spans a 600-fold range across globins from different organisms.8 Due to the diversity in primary structure and conservation in tertiary structure across different organisms, globins are ideal model systems to unveil how amino acid sequence affects folding mechanism, given a unique native topology. Interestingly, the folding pathways of the globins share some important © 2013 American Chemical Society
general features, although they differ in specific details depending on the organism of origin. Apomyoglobin from horse, for instance, undergoes compaction within four major types of events. An initial ultrafast local collapse on a 250 ns time scale is followed by a more global compaction on the 7− 17 μs time scale, and an additional conformational change taking place within 100−500 μs.9−11 The above events, probed by Trp fluorescence intensity, spectral shift, and lifetimes, are accompanied by partial dehydration.12 An additional much slower collapse takes place on the 1−100 ms time scale,13 as assessed by circular dichroism in the far-UV region and smallangle X-ray scattering (SAXS). The latter events take place concurrently with the extrusion of residual hydration water.12 Another well studied monomeric globin, sperm whale apomyoglobin, exhibits similar global features in that an initial ultrafast local conformational change on the ca. 100 ns time scale is followed by a more global collapse on the low microsecond time regime,11 detected by Trp fluorescence. The latter transitions correspond to the formation of a compact core, which persists throughout the rest of the folding timecourse and was characterized at residue-specific resolution, Received: January 6, 2013 Revised: May 30, 2013 Published: June 10, 2013 7868
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
0.4 ms after refolding initiation, by NMR quench-flow H/D exchange pulse labeling.14 This major kinetic intermediate comprises the A, G, H and part of the B helix.15−17 The intermediate bears some structural heterogeneity18,19 and has small, yet notable, non-native features.20 The fully native state is then reached via a slower conformational transition on the ms-s time scale. Globins from other organisms, studied less extensively, share similar features21 with the exception of slightly faster kinetics22 or moderate variations in the conformation of the major kinetic intermediate.23 The above studies highlight an important unifying feature of the folding timecourse of monomeric globins, namely, the presence of two broad collections of conformational changes: the first taking place on the nanosecond-to-microsecond time scale and involving collapse and structure formation at the A-GH helical interface; and the second occurring on the millisecond-to-second time scale and involving the structural consolidation of the rest of the protein. It was suggested that this broad temporal separation may have functional significance, in that it could have been engineered by Nature to facilitate co- or post-translational heme incorporation, which is believed to require access to a portion of the protein away from the major nonpolar core.24 The folding/unfolding kinetics of this portion of the protein has evolved to be carefully optimized.25,26 Globins from all organisms are characterized by a small number (ca. 6) of highly conserved core residues that do not play any functional role,27 and may therefore be important for folding28,29 and (or) stability.29 These residues are clustered in the main nonpolar core of the protein and typically include 1−3 aromatic amino acids and nonpolar residues with bulky side chains. The highly conserved position 8 of helix H (H8) has been particularly well studied. Its replacement with a variety of noncharged residues in sperm whale apomyoglobin (M131→A, W, V, F, G, Q) leads to only mild perturbations in the stability of the sub-microsecond-forming intermediate but to significant destabilization of the final native structure,30,31 resulting primarily from deceleration of unfolding rates.31,32 Replacement of W14, the only aromatic residue within the highly conserved nonpolar cluster of sperm whale apomyoglobin, with a nonaromatic residue leads to a severe perturbation in the latest kinetic steps of folding32 and irreversible aggregation (V. Bychkova, personal communication). Replacement of W14 with another aromatic residue, F, also leads to an aggregationprone protein.33 These results suggest the importance of having at least one aromatic residue with the correct packing characteristics in the nonfunctional conserved nonpolar cluster, for effective globin foldability. In this work, we target the complete folding timecourse of the monomeric Escherichia coli globin apoHmpH, including its ultrafast sub-microsecond steps. This protein is a particularly interesting representative of the globin fold because it bears a unique aromatic residue, tryptophan, at position 120 (Figure 1), which corresponds to the highly conserved H8 position in globins. This feature enables mechanistic insights on the role of Trp in this position during the early stages of globin folding. ApoHmpH was only studied on the millisecond−second time scale before,22 under experimental conditions different from those employed here. This study uses a combination of microfluidic continuous-flow and stopped-flow methods, mapping over 6 orders of magnitudes of folding time scales, from ca. 10 μs to 40 s. Conveniently, the presence of only 1 tryptophan leads to an easier interpretation of the fluorescence
Figure 1. Cartoon illustrating the three-dimensional structure of HmpH, the N-terminal domain of E. coli Hmp (from the X-ray crystal structure with PDB ID 1GVH22). The native HmpH helices are denoted as A to H, from the N to the C terminus. The side chain of Trp120 is highlighted in magenta. The image was generated with the PyMOL software (DeLano Scientific, LLC).
data than in the case of other globins bearing two (e.g., from sperm whale and horse heart) or more tryptophans. This analysis enables a comprehensive view of the conformational changes leading to the native state. We find that there are at least five kinetically resolvable phases in the folding process, with different probes displaying unique kinetics, and with lifetimes spanning from less than 8 μs to more than 10 s. The faster processes are nearly unaffected by the final denaturant concentration, implying that the solventaccessible surface area buried in each of the early core formation steps is relatively small. Given the rapid nature of the kinetics, any free energy barriers are expected to be low. The final fluorescence intensity amplitude occurs with a half-life of 144 μs, suggesting that the region of the nonpolar core surrounding Trp120 becomes fully native relatively early in folding. The kinetic step corresponding to this event occurs after consolidation of a native-like polarity around Trp120, suggesting that the core undergoes final structural rearrangements after its dehydration. Overall, we interpret our data as indicative of a gradual acquisition of structure on a relatively flat free-energy landscape during the earliest stages of folding. These events are critical, in that they are responsible for the formation of most of the apoHmpH structure.
■
EXPERIMENTAL METHODS Protein Expression and Purification. The E. coli globin HmpH was expressed in lysogeny broth (LB) medium at 37 °C and purified as described.22 Protein purity was assessed by analytical reversed-phase high-performance liquid chromatography (HPLC)34 and matrix-assisted laser desorption/ionization (MALDI) mass spectrometry (apoHmpH average mass = 15,720 Da). Continuous-Flow Ultrafast Kinetics. ApoHmpH (200 μM) was unfolded in 8 M urea, 125 mM Tris and 500 mM KCl at pH 7.5 (unfolding buffer) and refolded in the same buffer with no urea (refolding buffer). Upon mixing in the continuous-flow apparatus, the urea is diluted ∼100-fold. The urea dependence of the refolding rates (Figure 6) was monitored in the presence of appropriate concentrations of urea in the refolding buffer. Protein refolding experiments were conducted through a microfluidic ultrarapid mixer similar to the 7869
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
258 nm in a fluorimeter from PTI Technology (model QM4 CW).
one developed by Knight et al.35 and Hertzog et al.36,37 and modified by Yao et al.38 The mixer is made from a 500 μmthick fused silica wafer with channels typically etched 10 μm deep and a second 170 μm wafer bonded on top to seal the device. All flows are in the laminar regime and the flow rates and concentration profile can be computed from the applied pressures by mathematical simulations (COMSOL Multiphysics, Stockholm Sweden). The mixing time is defined as the time for the concentration of denaturant to decrease by 80%, and it amounts to ∼8 μs (see Figure 1b, and DeCamp et al.39 and Waldauer et al.40 for a full discussion of how the mixing time is determined experimentally and computationally). The tryptophan is excited with a frequency-doubled Argon-Ion laser at 257 nm (Lexel Laser 95 SHG, Fremont, CA) and fluorescence intensity (Figure 1a) and spectral (Figure 1c) changes were observed at various times beyond mixing using the same confocal instrument described by Lapidus et al.9 The spectral data used in all subsequent analysis were corrected for wavelength-dependent changes in detector sensitivity. All experiments were performed at 20 °C. Temperature stability was maintained within 0.005 °C via a temperature controller (TE Technology model 1600). Stopped-Flow Kinetics. Stopped-flow refolding was performed at 20 °C into refolding buffer (see continuousflow kinetics section), with an SFM-400 MOS-450/AF-CD stopped-flow kinetics spectropolarimeter equipped with an HDX mixer (Bio-Logic, Claix, France). ApoHmpH was unfolded in refolding buffer containing 6 M urea at pH 7.5 (unfolding buffer). This concentration of denaturant is sufficient to fully unfold the protein.22 Over the course of the experiment, the protein was diluted 10-fold into refolding buffer in a 2 × 2 mm Suprasil FC-20 cuvette (Bio-Logic, Claix, France) (4−7 μM final protein concentration). The experimental dead time of the stopped-flow apparatus was 6 ms. A Xe−Hg lamp (Hamamatsu Photonics, Hamamatsu, Japan) with an 8-mm excitation slit width was used. Circular dichroism (CD) was monitored in the far-UV region at 225 nm. Fluorescence intensity and anisotropy data were collected upon exciting Trp120, the only Trp of apoHmpH, at 257 nm (same excitation wavelength as in ultrafast experiments), with a 314-nm cut-on long-pass filter placed on the emission side. Refolding kinetic data were collected for 40 s. The unfolded protein was incubated at 4 °C overnight and equilibrated to 20 °C prior to the stopped-flow experiments. Unfolded controls consisted of urea-unfolded protein diluted into the same ureacontaining buffer. After subtracting the signal from buffer background, the CD refolding trace was fit to a doubleexponential decay relation. The data show that most of the CD amplitude decays within the stopped-flow dead time. This burst phase shows that the majority of apoHmpH’s secondary structure is formed on the sub-millisecond time scale. The observable decay consists of a kinetic phase with a lifetime of 132 ms and a minor slower component with very small amplitude corresponding to a 1−10 s decay. Manual Mixing Experiments. In order to confirm that the fluorescence intensity at ∼1 ms refolding time is the same as the fluorescence of the fully native protein, we performed control manual mixing experiments in a steady-state fluorimeter. The protein was diluted to 3.1 μM in 125 mM Tris and 500 mM KCl at pH 7.5 in the presence of either 0.1 or 8 M urea, under refolding and unfolding conditions, respectively. The spectra were measured upon excitation at
■
RESULTS AND DISCUSSION General Considerations on Time Scales. This study samples conformational changes over more than 6 orders of magnitude before the apoHmpH protein becomes fully folded. Trp fluorescence emission intensity and spectral shifts were monitored in a microfluidic continuous-flow mixing device on the microsecond−millisecond time scale, and far-UV CD and steady-state fluorescence anisotropy were recorded on the millisecond−second time scale. The above techniques reveal the presence of at least five kinetic phases including a final structural rearrangement establishing a native environment around Trp120 with a half-life of 144 μs. Folding of apoHmpH from Microseconds to Milliseconds. The very early stages of apoHmpH folding were monitored by time-resolved Trp fluorescence spectra using a microfluidic ultrafast mixer. Figure 2a shows the basic layout of
Figure 2. (a) Mapping of the fluorescence variations of apoHmpHcontaining solutions in a microfluidic mixer. Contour plot illustrating the apoHmpH Trp fluorescence emission intensity near the mixing region. The denatured protein (in 8 M urea) enters from the top and is constricted to a narrow jet by side channel flows of refolding buffer at ∼80 μm on the y-axis. The protein then continues down the exit channel as it folds. (b) Fluorescence intensity of NATA flowing along the center streamline of the exit channel when mixed with 400 mM KI, a fluorescence quencher of NATA. The solid line denotes the concentration of KI relative to its final concentration, calculated independently using COMSOL38 (left y axis), and the solid circles are the experimentally determined ratio of NATA fluorescence intensity in the absence and presence of KI (I/I0, right y axis). (c) Fluorescence emission spectra of apoHmpH observed along the jet as the protein folds.
the apparatus. The protein denatured in 8 M urea enters from the top and is mixed with buffer coming from either side at ∼80 μm on the y-axis. The buffer flows ∼100 times faster than the protein, constricting the protein to a narrow (∼100 nm) jet and allowing the light denaturant molecules to diffuse out of the jet quickly. The flow along the exit channel (below 80 μm) is constant, allowing a simple conversion of distance to refolding time after mixing. Figure 2b shows the calculated concentration of potassium iodide (KI), a diffusion-limited fluorescence 7870
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
quencher41,42 (solid line), and the experimentally determined relative fluorescence of N-acetyl tryptophan amide (NATA) in the presence of 400 mM KI (quenching rate constant of NATA by KI:41 kq = 3.8 × 109 M−1 s−1). The NATA timecourse profile shows that mixing is complete in about 8 μs. Given that this mixer is laminar (nonturbulent), optical measurements can be made throughout the mixing process, so any “burst phase” will be observed as a rapid phase with a rise/decay of about the mixing time (see, for example, the insets of Figure 4b,c). Once the position of the protein jet is determined from Figure 2a, data are collected at various points along the exit channel using a spectrograph and charge-coupled device (CCD) camera. Figure 2c shows a collection of Trp fluorescence emission spectra of apoHmpH as the protein flows through the jet and folds. This illustration shows that apoHmpH undergoes both fluorescence intensity variations and spectral shifts, upon reaching its native state. Figure 3a shows the typical
components are significant. Higher-order SVD components have singular values less than 15% of the first two components. The first significant SVD component (dark gray line in Figure 4a) represents the average fluorescence spectrum over
Figure 3. (a) Typical fluorescence emission spectrum from NATA. The spikes in the spectrum are due to an optical artifact of the confocal microscope. (b) Trp fluorescence emission spectra of apoHmpH 1 ms after refolding initiation in a microfluidic mixer (lines) and at equilibrium, after several minutes from refolding initiation (points), at 20 °C. The denaturant concentrations are 8 M (black) and 0.1 M (gray) urea. The amplitude axis of the kinetic data was adjusted to enable optimal lineshape comparisons with the equilibrium data.
Figure 4. (upper panel) First (dark gray) and second (light gray) SVD components of the fluorescence emission spectra monitoring the refolding of apoHmpH into a ∼ 0 M urea buffer in a microfluidic mixer. The singular values of these components are 2.95 × 106 (dark gray) and 4.71 × 105 (light gray). (central panel) Change in amplitude of the first SVD components as a function of refolding time relative to the first SVD component of the protein mixed into 8 M urea (unfolded control). (lower panel) Change in amplitude of the second SVD component as a function of refolding time. This data is not compared to the unfolded control. For central and lower panels, the black line is the fit of the timecourse starting at 12 μs to two exponentials. The inset plot highlights the changes in signal amplitude of the SVD component during the mixing time to show the burst phase. In central panel, the spike at t ∼ 0 is an instrumental artifact due to slight misalignment in the time domain of the refolding and control experiment. The burst phase should be viewed as the difference in signal between t < 0 and t ∼ 8 μs.
fluorescence spectrum of NATA, and Figure 3b shows the fluorescence spectra of unfolded and folding protein 1 ms after mixing. The agreement with spectra collected when the folding equilibrium is fully established (upon manual mixing of the unfolded protein solution in refolding buffer in a commercial steady-state fluorimeter, and incubation for several min; also displayed in Figure 3) shows that the fluorescence signal equilibrated at the native state level by 1 ms within error. Folding trajectories are analyzed by a global analysis of the time-dependent fluorescence emission spectra with singular value decomposition (SVD), using the native function in Matlab. Singular value decomposition determines orthogonal vectors in both wavelength and time that can reconstruct the entire data set, and rank-orders the significance of the vectors. Most components are simply noise and only the first two
all times and includes a significant decrease in signal during mixing due to the formation of the jet. To overcome this complication, the effect of the jet formation was eliminated via a point-by-point ratio of the first SVD component of two data sets for (a) the unfolded protein mixed into folding buffer (i.e., 0 M urea), and (b) the unfolded protein dissolved into the 7871
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
same denaturing buffer (8 M urea). The timecourse of decay of the total fluorescence emission, shown in Figure 4b, has a small burst phase (change in signal during mixing, Figure 4b inset), a rapid decay and a much slower rise with a half-life of 144 μs. The rate constants for all the kinetic phases are given in Table 1. Table 1. Kinetic Rate Constants and Corresponding Fractional Amplitudes for the Refolding Timecourse of apoHmpH Detected by a Variety of Spectroscopic Probes detection method Continuous Flow fluorescence total intensity burst phase fluorescence spectral shift burst phase fluorescence total intensity decay fluorescence spectral shift decays Fluorescence total intensity rise Stopped Flow far-UV CD burst phase far-UV CD decay
rate constant (s‑1)
fractional amplitudea
>1.25 × 105
1.04 ± 0.1
>1.25 × 105
0.55 ± 0.05
9.7 2.8 1.2 4.8
± ± ± ±
3.9 0.1 0.1 1.5
× × × ×
104 104 104 103
>166 7.6 ± 1.2 0.06 ± 0.03
− 1.00 ± 0.1 0.28 ± 0.03 0.17 ± 0.03 0.96 ± 0.1 0.70 ± 0.02 0.22 ± 0.02 0.08 ± 0.03
a
Spectral shift, total intensity, and far-UV CD fractional amplitudes were assessed independently as follows. Total decay amplitudes (i.e., amplitudes at the end of the decay minus amplitudes of the unfolded state) were first determined. Decay amplitudes associated with each kinetic component were then assessed from preexponential factors, except for the case of the burst phases and fluorescence total intensity where amplitudes were estimated by visual data inspection. Fractional amplitudes of each kinetic phase were finally assessed as the ratios between the relevant decay amplitude and the total decay amplitude.
Figure 5. (a) Cartoon illustrating all the amino acids within 7 Å distance (dark green) of Trp120 (magenta) in E. coli HmpH. (b) Analogous cartoon focusing (in dark green) on K11, D118, K122, and Y124, which are the amino acid types known to quench Trp in proteins that lie within 7 Å of Trp120 (magenta). The Trp environment mapped here employs the structural scaffold of native HmpH, for simplicity. Hence, the helical content of the cartoons does not necessarily reflect the precise secondary structure adopted by the protein on the sub-millisecond time scale. The HmpH helices are labeled as A to H, from the N to the C terminus. The HmpH structure (1GVH PDB ID) was drawn with the PyMOL software (DeLano Scientific, LLC).
The presence of a burst phase in total fluorescence intensity shows that there are molecular events taking place within 8 μs that lead to a variation of the environment surrounding Trp120. Given that a burst phase is observed for both the fluorescence emission and spectral shift data (diagnostic of local polarity, see analysis of second SVD component below), we ascribe these initial events to a hydrophobic collapse of the protein chain. A subsequent fluorescence intensity kinetic phase with a rate constant of 9.7 ± 0.4 × 104 s−1 shows a decay in the fluorescent signal. The fluorescence intensity from any given fluorophore is modulated by the extent of nonradiative events leading to quenching43 mostly via excited-state proton and electron transfer from selected amino acids coming in close proximity of the fluorophore.43,44 Hence, variations in fluorescence intensity can be interpreted as due to the establishment of specific short-range intramolecular interactions within a few angstroms of the fluorophore. Therefore, we ascribe the timedependent quenching corresponding to the first resolvable fluorescence intensity kinetic phase to the establishment of specific short-range contacts around Trp120. Figure 5a provides a pictorial representation of the apoHmpH residues within the representative quenching-relevant distance of 7 Å from Trp120. These amino acids, shown in dark green, belong to portions of the A, G, H, and E helices. Among them are K11, D118, K122 and Y124, i.e., residue types known to quench Trp fluorescence in proteins.44 The latter four residues belong to the apoHmpH helices A and H, as shown in Figure 5b. A final slower fluorescence intensity rise is also observed, with a rate constant
of 4.8 ± 1.5 × 103 s−1 (corresponding to a half-life of 144 μs). This kinetic phase occurs after Trp120 has already reached a native-like polar environment (see second SVD component data, discussed below). Hence, we ascribe the kinetic phase with 144 μs half-life to the fluorescence intensity dequenching resulting from final structural rearrangements within the Trp120 core environment defined in Figure 5b, including the portions of helices A and H surrounding Trp120. The second significant SVD component (gray line in Figure 4c) represents the spectral shift to lower wavelengths. This second component followed as a function of time shows both a burst phase (see Figure 4c inset) and a double exponential decay. Shifts in fluorescence emission spectra are well-known indicators of electronic excited-state transition dipole moment stabilization due to an increase in environment polarity (red shifts) over a time scale shorter than the fluorophore’s lifetime (∼ns), or excited state dipole moment destabilization due to generation of a more nonpolar environment (blue shifts).45 7872
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
0 M, the slow rise in the first component was too slow and had an amplitude that was too small to be reliably fitted. We performed curve fitting of the fast decay of the first SVD component (open squares in Figure 6) at [urea] ≤ 2 M according to the relation
The fastest observable events (Table 1) correspond to an increase in intensity and a blue spectral shift occurring within the 8 μs mixing time (burst phase), as shown in Figures 4b,c. We ascribe this shift to a reduced polarity in the environment around the fluorophore’s excited state as the protein switches from 8 M urea to nearly denaturant-free refolding conditions. A control experiment based on subjecting NATA to the same variations in solvent conditions as the refolding experiment revealed no spectral shifts either after 8 μs from mixing initiation in the microfluidic device or at equilibrium after manual mixing. Therefore, we conclude that the observed burst-phase blue shift is primarily due to a decrease in the polarity of the Trp120 side chain environment resulting from a collapse within this region of the protein. Interestingly, this collapse is likely accompanied by the establishment of some specific interactions around Trp120 within the mixing time, given the small burst phase in fluorescence emission. Ultrafast collapse on the sub-microsecond time scale has been previously observed for several proteins via both burst-phase fluorescence spectral shifts 9 and Fö r ster resonance energy transfer (FRET).46,47 It has also been detected by fluorescence lifetime10,24,48 or FRET49 measurements in ultrafast temperature-jump experiments with extremely short (∼ns) dead times. After the initial burst-phase spectral shift, the two subsequent kinetic phases involving spectral shifts correspond to the further consolidation of a fully nonpolar structure around Trp120. These events likely involve the establishment of short-range contacts and possibly also core dehydration, given that this region of a native globin fold is known to be dry.50 Importantly, the two observed spectral shift kinetic phases are followed by no further fluorescence variations, suggesting that the local environment of the Trp is native-like with a half-life of 144 μs. The observed rate constants for the first two SVD components as a function of final denaturant concentration are plotted in Figure 6. Each component was fitted independently to two exponentials. A global fit to the same exponential rates as those of the independent fits, for both SVD components at the same urea concentration (with amplitudes treated as adjustable parameters), produced significantly inferior results (see Figure S1), suggesting that each rate constant corresponds to an entirely separate event. For [urea] >
k = k H2O exp(m∓[urea]/RT )
where k is the denaturant-dependent folding rate constant, kH2O is the folding rate constant in the absence of denaturant, and m∓ is the kinetic m value for folding. We obtained m∓ = −0.16 ± 0.05 kcal mol−1 M−1. This small value supports the establishment of short-range contacts (including those with K11, D118, K122 and Y124 in the core surrounding Trp120) involving the burial of only a moderate amount of surface area. This result is consistent with the already generically collapsed nature of the earlier state populated after the 8 μs spectral-shift burst phase (not sensed by fluorescence intensity). The fast rate constants in Figure 6 are also consistent with an extremely low activation barrier, if any, for all the microsecond kinetic phases.51−53 Folding from Milliseconds to Seconds. The slowest steps of apoHmpH refolding, on the millisond−second time scale, were monitored by stopped-flow CD and Trp fluorescence emission (Figures 7 and S2, and Table 1). The majority of the native CD signal at 225 nm is formed within instrument dead time of 6 ms. There is also an observable decay with a rate constant of 7.6 ± 1.2 s−1, indicating that the α-helices become fully native-like only on the second time scale. On the other hand, Trp fluorescence emission does not show any observable kinetics (Figure 7b), showing that Trp120 has already reached its native environment before the 6 ms dead time, fully consistent with the data of Figure 4. Due to the time resolution of the stopped-flow apparatus, we cannot definitively show whether secondary structure forms simultaneously with the establishment of Trp native environment. On the other hand, it is clear that most of the helical content and Trp’s fully native contacts are formed before 6 ms. In summary, the stopped-flow data are consistent with the results from the ultrafast kinetic refolding performed in the microfluidic mixer, and additionally show that a small amount of secondary structure forms late in the folding process. Overall Features of apoHmpH Refolding. The rate constants for all the apoHmpH refolding kinetic steps probed by microfluidic and stopped-flow mixing are displayed in Figure 8. As a complement, Figure 9 highlights all the molecular events corresponding to each of the observed kinetic steps. The events illustrated in Figure 9 provide a concise overview of the timedependent folding process. This representation is particularly useful because it is model-free, i.e., it applies whether or not the protein folds via sequential or parallel routes. Based on the analysis of all the kinetics phases, apoHmpH displays at least five refolding steps: (1) rapid hydrophobic collapse due to an initial change in protein conformation occurring within the microfluidic mixing time, supported by the burst-phase fluorescence emission intensity and spectral shift; (2) formation of specific intramolecular contacts around Trp120, causing a variation in overall fluorescence intensity on the 10− 100 μs time scale; (3) further consolidation of the tryptophan nonpolar environment detected via biexponential spectral shifts and possibly involving dehydration; (4) formation of the final native contacts around Trp120 with a half-life of 144 μs; (5) formation of a residual amount of secondary structure that had
Figure 6. Dependence of the rate constants for the fastest kinetic phases of apoHmpH refolding on denaturant concentration. The open squares and triangle denote values deduced from total fluorescence intensity (first most significant SVD component), and the solid circles and diamonds illustrate values for the fluorescence spectral shifts (second most significant SVD component). Error bars represent the experimental uncertainly for two independent experiments performed on different chips. 7873
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
Figure 9. Schematic overview of the major conformational changes occurring on widely different time scales during the folding of apoHmpH. The dashed lines denote events whose kinetic profiles could not be explicitly resolved because they occur within the dead time of the relevant technique.
not yet been incorporated into the protein chain, taking place on a much slower time scale. The apoHmpH Free Energy Landscape. Formation of the majority of the secondary structure during the burst phase of the stopped-flow mixer may or may not occur commensurately with any of the four kinetically resolvable sub-millisecond steps, and may therefore take place independently or cooperatively with tertiary structure. Independent structure formation might require a free-energy landscape that depends on more than one reaction coordinate. This behavior is similar to that of α3D, where probe-dependent kinetics was shown to require a multidimensional free-energy folding landscape.54 Alternatively, if all structure formation progressed along a unique free energy coordinate a typical one-dimensional energy landscape would suffice. For instance, fast-folding proteins with probe-dependent folding rate constants such as λ repressor39 were shown to have very low free energy barriers on somewhat rugged one-dimensional energy surfaces. A complementary case is that of proteins with a smooth one-dimensional free energy landscape like BBL, a downhill fast-folder characterized by probe-independent relaxation rate constants but probe-dependent kinetic amplitudes and equilibrium transition midpoints.55 All the above examples comprise proteins whose experimental folding kinetics was complemented by computational modeling of the folding process, which helped establish the degree of ruggedness and the dimensionality of the free energy landscape. This was not possible in the case of apoHmpH given its large size and complex kinetics. On the other hand, the peculiar experimental folding features of apoHmpH lead to the following conclusions. While probe-dependent observables are compatible with both concurrent and sequential structure formation,56 the fact that (i) much of the structure in apoHmpH is formed on the sub-millisecond time scale, (ii) the kinetics of Trp fluorescence intensity and spectral shift are
Figure 7. Stopped-flow refolding kinetic traces for apoHmpH refolding at 20 °C from 4 ms to 8 s. The refolding time course was monitored by (a) far-UV CD at 225 nm, (b) Trp fluorescence emission, and (c) Trp fluorescence anisotropy. The error on the decay rate constant of the CD trace was obtained from two independent experiments.
Figure 8. Summary of apoHmpH refolding rate constants detected by various spectroscopic probes. The long thick bars denote spectral burst phases, and correspond to the mixing time of the microfluidic mixer (black) and the dead time of the stopped-flow apparatus (light gray).
7874
dx.doi.org/10.1021/jp400174e | J. Phys. Chem. B 2013, 117, 7868−7877
The Journal of Physical Chemistry B
Article
inconsistent with the overall-highly-dynamic molten globular structure usually invoked to describe globin folding intermediates populated after a few milliseconds from refolding initiation.6,60
entirely different, and (iii) no phase (other than hydrophobic collapse) is reflected in more than one probe clearly shows that much of the structure is formed gradually (Figure 9), possibly on a multidimensional free energy landscape. These features bear strong similarity to a recent experimental and computational study of Acyl-CoA Binding protein (ACBP), which also revealed many pathways to the native state.46 The slow speed of the last folding steps can either be attributed to high free energy barriers or to traps on a fairly rugged free-energy landscape, similar to what was seen for αlactalbumin, a slow folder of marginal stability.57 Comparisons between free energy barriers (determined at equilibrium) and kinetic folding rate constants58 could help sort this out. In any case, the earlier steps suggest that the landscape prior to the last folding step is relatively flat and rugged, not dominated by large energy barriers. The Mechanism of apoHmpH Core Formation. The unique tryptophan (Trp120) of apoHmpH lies within the putative main core of the globin fold. The fact that Trp120 reaches a native environment within 600−800 μs from refolding initiation (Figures 4, 8, and 9, Table 1) suggests that the core of E. coli apoHmpH becomes fully structured fast, on the sub-millisecond time scale. Hence the main nonpolar region of this bacterial globin has evolved to become native-like efficiently during folding. Interestingly, even after the establishment of native-like polarity within the core (see slow spectral shift phase in Figure 8), probably including local dehydration given that the globin core is dry,50 the Trp120 environment undergoes one additional structural consolidation event (see slow fluorescence total intensity phase in Figure 8, with 144 μs lifetime). This result suggests that, for the bacterial globin fold, establishment of native-like polarity is not the final stage of core formation. Yet, some additional sub-millisecond local structural rearrangements followed by slower secondary structure variations (on the millisecond−second time scale) are necessary for the formation of a fully native structure. Relations between the Folding of apoHmpH and Other Globins. There are some correlations between the folding of apoHmpH and globins from other organisms. For instance, fast initial collapse (
