Skip to main content
NCBI home page
Biophysical Journal logoLink to Biophysical Journal
. 2019 Sep 18;117(8):1456–1466. doi: 10.1016/j.bpj.2019.09.015

Diverse Folding Pathways of HIV-1 Protease Monomer on a Rugged Energy Landscape

Janghyun Yoo 1, John M Louis 1, Hoi Sung Chung 1,∗
PMCID: PMC6817636  PMID: 31587829

Abstract

The modern energy landscape theory of protein folding predicts multiple folding pathways connecting a myriad of unfolded conformations and a well-defined folded state. However, direct experimental observation of heterogeneous folding pathways is difficult. Naturally evolved proteins typically exhibit a smooth folding energy landscape for fast and efficient folding by avoiding unfavorable kinetic traps. In this case, rapid fluctuations between unfolded conformations result in apparent two-state behavior and make different pathways indistinguishable. However, the landscape roughness can be different, depending on the selection pressures during evolution. Here, we characterize the unusually rugged folding energy landscape of human immunodeficiency virus-1 protease monomer using single-molecule Förster resonance energy transfer spectroscopy. Our data show that fluctuations between unfolded conformations are slow, which enables the experimental observation of heterogeneous folding pathways as predicted by the landscape theory. Although the landscape ruggedness is sensitive to the mutations and fluorophore locations, the folding rate is similar for various protease constructs. The natural evolution of the protease to have a rugged energy landscape likely results from intrinsic pressures to maintain robust folding when human immunodeficiency virus-1 mutates frequently, which is essential for its survival.

Significance

Protein folding is a process by which proteins find their native three-dimensional structure. Typically, naturally evolved proteins exhibit a smooth folding energy landscape for fast and efficient folding by avoiding unfavorable kinetic traps. However, depending on the selection pressures, the energy landscape can be different. In this work, we show that the folding energy landscape of human immunodeficiency virus-1 protease monomer is highly rugged. Using single-molecule Förster resonance energy transfer spectroscopy, we directly demonstrate heterogeneous folding pathways. The distribution of these pathways is sensitive to the mutations and fluorophore locations. The rugged energy landscape likely results from intrinsic pressures to maintain robust folding when the virus mutates frequently, which is essential for the selection and propagation of drug-resistant mutants.

Introduction

The folding energy landscape of a large macromolecule such as a protein is complex and highly multidimensional in general. The energy landscape of a random heteropolymer is glassy, containing multiple local energy minima (termed frustrated). However, according to the principle of minimal frustration, the energy landscape of naturally evolved proteins is tuned to be smooth and funneled toward a deep global minimum of the folded state (1). On this funneled energy landscape, the mechanism of protein folding is described in terms of a few order parameters, such as native interactions that exist in the folded structure (1), as observed in simulations (2, 3). Folding kinetics of small proteins consisting of one domain (single-domain proteins) and exhibiting apparent two states (folded and unfolded) have been explained by diffusion on a one-dimensional free energy surface (2, 3, 4, 5, 6, 7, 8). Despite this macroscopic simplicity, underlying microscopic diversity still persists. The energy landscape theory and simulation show that proteins fold via heterogeneous pathways connecting diverse unfolded conformations and the folded state rather than a small set of well-defined pathways (1, 9, 10, 11, 12, 13). However, diverse microscopic pathways are difficult to detect in experiments. Although there are a myriad of different conformations in the unfolded state, unfolded proteins are flexible and rapidly fluctuate on a timescale of 10–100 ns (14, 15, 16), which removes signatures of diverse folding pathways and results in single exponential folding kinetics (17). The smooth folding energy landscape is postulated to be the result of evolution under a selection pressure that favors fast and efficient folding by avoiding kinetic traps (1). This is supported by examples of slow diffusion on the energy surface (i.e., rough energy landscape) observed for designed proteins (18, 19, 20). However, if the selection pressure does not favor efficient folding, the energy landscape can be rugged, and multiple folding pathways may become apparent (1), as observed for RNA (21, 22) and DNA (23). In this study, we show that human immunodeficiency virus (HIV)-1 protease (PR) has evolved into one such system.

HIV-1 PR is a homodimeric protein that plays an indispensable role in HIV maturation (Fig. S1; (24, 25)). Characterization of folding of the PR monomer using ensemble experiments is complex because the PR exists in various forms because of the low nanomolar dissociation constant of the mature dimer and the intrinsic autoprocessing (self-cleavage) activity, responsible for converting its own precursor to the mature PR (see Fig. S2; Supporting Materials and Methods). We used single-molecule Förster resonance energy transfer (FRET) spectroscopy to examine folding of the mature PR monomer and its precursor prior to N-terminal autoprocessing (Fig. S1). We compare the folding of different substitution mutants labeled with the donor (Alexa Fluor 488) and acceptor (Alexa Fluor 594). Single-molecule FRET experiments at picomolar concentrations allow monitoring monomer folding without any potential interference due to dimerization and autoprocessing. We found various unique signatures of a rugged folding energy landscape for both the mature PR and its precursor, which are not observed in typical two-state proteins. These signatures include the extremely wide unfolded state FRET efficiency distribution, indicating very slow conformational fluctuations in the unfolded state, which allows for direct observation of the diverse folding pathways. Despite the sensitivity of the unfolded state, conformational distribution influenced by the mutation and fluorophore location, overall folding rates of various mutants of the precursor and mature PR are very similar, with the exception of an extremely drug-resistant PR, PR20 (26, 27), which contains 19 drug-resistant/compensatory mutations, exhibiting much slower folding. This robustness of folding likely correlates to virus viability to accommodate frequent naturally selected mutations and those selected under drug pressure. The rough energy landscape, which is not optimized for efficient folding, could be a result of maintaining robust folding of a large number of such mutants (28).

Materials and Methods

Single-molecule experiment

Single-molecule experiments were performed using a confocal microscope system (MicroTime200; PicoQuant, Berlin, Germany) with a 75-μm diameter pinhole, a dichroic beamsplitter (Z488rdc; Chroma Technology, Bellows Falls, VT), and an oil-immersion objective (PlanApo, NA 1.4, × 100; Olympus, Tokyo, Japan). Alexa Fluor 488 was excited by a 485-nm diode laser (LDH-D-C-485; PicoQuant) in the continuous wave mode at 30 μW for the free-diffusion experiment and in the pulsed mode at 0.16 μW for the immobilization experiment, respectively. Alexa Fluor 488 and Alexa Fluor 594 fluorescence was split into two channels using a beamsplitter (585DCXR; Chroma Technology) and focused through optical filters (ET525/50m for Alexa Fluor 488 and E600LP for Alexa Fluor 647; Chroma Technology) onto photon-counting avalanche photodiodes (SPCM-AQR-15; PerkinElmer Optoelectronics, Fremont, CA).

In the free-diffusion experiment, bursts of fluorescence photons emitted from freely diffusing molecules (40 pM in 50 mM sodium acetate (pH 5)) were collected into 2 ms bins at various urea concentrations. Tween 20 (0.01%) was added to prevent sticking of proteins on a glass coverslip. Data were collected for 1.5–3 h, and 2-ms bins containing 60 photons or more were considered as significant bursts for further analysis. In the immobilization experiment, proteins were immobilized on a biotin-embedded, polyethylene glycol-coated glass coverslip (Bio-01; MicroSurfaces, Englewood, NJ) as described previously (29). 100 mM β-mercaptoethanol and 40 mM cysteamine were added to reduce blinking and bleaching of dyes (30) in both free-diffusion and immobilization experiments. All experiments were performed at room temperature (22°C).

Details of the design, expression, purification, and dye labeling of protein constructs and analyses of single-molecule data are described in the Supporting Materials and Methods.

Results

Heterogeneity in the unfolded state is observed

We first examined FRET efficiency histograms of the folded and unfolded states by free-diffusion experiment at various urea concentrations (Figs. 1 and S3). In this experiment, molecules freely diffuse in solution, and fluorescence bursts are collected from those passing through the focus of laser excitation. The relative fraction of the folded peak at high FRET efficiency (E > 0.8) decreases, whereas that of the unfolded peak (0.4 < E < 0.8) increases as the urea concentration is increased, indicating denaturation of proteins by urea (Fig. S3). We found that the width of the unfolded state of the construct C7/C95, in which dyes are attached at the two termini of the folded part of the protein (31), is much wider than the shot-noise limited width that is determined by the number of photons in a bin (2 ms), whereas the folded state agrees well with the shot-noise limited width (Fig. 1). If the protein is completely unfolded and flexible, as observed for typical single-domain proteins, the width should be the same as that by shot noise. However, the width is broader than the shot-noise limited one even at 6 M urea (Fig. S3 B). The broad distribution suggests that there are relatively stable unfolded substates, with a lifetime longer than the bin time of 2 ms. Typical conformational fluctuation time of unfolded proteins measured by a laser temperature-jump experiment and nanosecond fluorescence correlation spectroscopy is 10–100 ns (14, 15, 16).

Figure 1.

Figure 1

Open in a new tab

Variations of the unfolded state characteristics by mutations and labeling positions measured at 2 M urea. (A) FRET efficiency histograms were fitted to double Gaussian (red: folded, cyan: unfolded, black: sum). (B) Shown are peak center positions, SDs obtained from fitting, and SDs expected from shot noise. FRET efficiency histograms of all urea concentrations are displayed in Fig. S3. To see this figure in color, go online.

Alternatively, the broad distribution can be an artifact induced by labeled fluorophores because a tryptophan residue next to one of the labeling sites can form a complex with fluorophores and quench fluorescence (32, 33). To investigate this possibility, we introduced a single substitution mutation (W6L). We also moved the labeling positions to the termini of the PR (C2/C98) and to the outside of the PR domain (−5C/+5C) (see Fig. S1). Finally, we examined the effect of excessive, but naturally occurring, mutations in PR20.

For the folded state, the mean FRET efficiency is the highest for constructs with labels on C7/C95 pair, and this decreases by moving the labels toward the termini of PR, consistent with residues 1–10 and 91–99 being disordered in the PR monomer (31). The peak center position is almost identical for the constructs with the same labeling positions, and the standard deviation (SD) of the peak is close to the width by shot noise (Fig. 1 B, upper). On the other hand, the unfolded state shows diverse peak center positions, even for the proteins with the same labeling positions (C7 and C95). Moreover, the SD is much larger than the shot noise in all cases. Interestingly, the width decreases as the dyes are away from the folded part (C2/C98 and −5C/+5C in Fig. S3). In addition, the width of the unfolded peak of the construct lacking tryptophan (W6L) is the narrowest and closer to the shot-noise limited width for both the precursor (TFR-PR) and the mature PR (Fig. S3).

The observations that moving the dye labeling positions away from the folded part of the protein and replacing the tryptophan residue at position 6 (W6L) reduces the width of the FRET efficiency distribution suggest that the broad distribution may indeed result from the interaction between the fluorophores and the tryptophan residue. To verify this possibility, we measured the anisotropy values of different constructs and analyzed different FRET efficiency regions of the histograms as shown in Fig. S4 (also see the summary in Table S1). In all three constructs (−5C/+5C, C7/C95, and L6/C7/C95), the anisotropy value increases with the increasing FRET efficiency in the distribution. However, this is not a result of dye-protein interactions but rather that of shorter donor fluorescence lifetimes at higher FRET efficiencies. The extracted donor reorientational correlation time is very similar (1.3–1.5 ns) for all constructs of both the precursor and the mature PR.

Proline isomerization can also cause slow transitions in the unfolded proteins and slow folding. We find all four proline residues within the structured part of the protein to be in trans form. Therefore, proline isomerization is not expected to affect folding significantly. In addition, as the same proline residues are present in all constructs that were used, any potential effect should appear to be the same. The distribution of the unfolded state varying among different constructs rules out this possibility.

Therefore, the changes in the broadness should not be an artifact by direct dye-protein interactions or proline isomerization but rather result from mutations and indirect influence of fluorophores in protein dynamics. The universality of the broadness, which is independent of the additional transframe region (TFR) at the N-terminus of the precursor, suggests that this broadness is encoded in the PR sequence. In addition, the changes of the center peak position and the broadness of the unfolded FRET distribution of PR20 compared to C7/C95 suggest that the drug-resistant mutations not only modulate the structure and function of the folded state but also affect the unfolded state conformational distribution. The existence of the broad unfolded state distribution with slow interconversion between unfolded conformations indicates that the energy landscape of the unfolded state is highly rugged, and the changes in the width of the distribution suggest that this ruggedness is sensitive to mutations and the addition of fluorophores outside the structured region.

Next, to determine FRET efficiency values with less uncertainty (i.e., less shot noise) and probe conformational dynamics on a longer timescale, we immobilized proteins and monitored folding and unfolding transitions. Fig. 2 shows typical donor and acceptor fluorescence trajectories and FRET efficiency trajectories of precursor TFR-PRC7/C95 with 20-ms bin time measured at 2 M urea. In some trajectories, transitions are observed between the folded (E > 0.8) and unfolded (E < 0.8) states (Fig. 2, A and D). On the other hand, photobleaching occurs before any transition in many trajectories because of the very slow folding/unfolding kinetics (Fig. 2, C and E). Slow fluctuations on a timescale of ∼10 s are also observed between different FRET efficiency levels in the unfolded state (Fig. 2 B), which are expected from the broad unfolded state FRET efficiency distribution in Fig. 1 B.

Figure 2.

Figure 2

Open in a new tab

Representative fluorescence (left) and FRET efficiency (right) trajectories (20 ms bin time) of precursor TFR-PRC7/C95 measured at 2 M urea. (A–E) Solid lines drawn in the middle of noisy raw trajectories are the average photon count rates of each divided segment (left) and the FRET efficiencies corrected for background, donor leak into the acceptor channel, and γ-factor (right). The FRET efficiencies of the donor-only segments were corrected for only background. Red and green arrows in the fluorescence trajectories (left) indicate photobleaching of the acceptor and donor, respectively. To see this figure in color, go online.

Fig. 3 shows FRET efficiency histograms constructed from immobilization trajectories. FRET efficiency values were calculated from the photons in the first segment of each trajectory (i.e., each molecule), the distributions of which properly reflect the relative populations of the folded and unfolded states. By calculating the FRET efficiency using a large number of photons in a long segment rather than those in a bin in the free-diffusion experiment (Figs. 1 and S3), broadening by shot noise is significantly reduced, and the constructed FRET efficiency distribution is close to the real distribution. Indeed, the FRET efficiency histograms at the native (0 M urea) and the most denaturing condition (8 M GdmCl) show very narrow distributions of the folded and unfolded states (Fig. S5), respectively, similar to those of a two-state folder, protein G (29), indicating the broad unfolded state distributions at other urea concentrations reflect the real, broad conformational distribution. Overall, the trend of broadness changes in the histograms from the immobilization experiment agrees well with that of the free-diffusion experiment, which lends support to the reliability of the detailed analysis of the immobilization data in the following sections.

Figure 3.

Figure 3

Open in a new tab

FRET efficiency histograms obtained from immobilization experiments. (A–F) Histograms were constructed from the mean FRET efficiencies calculated from the initial segment of each trajectory of (A) TFR-PR−5C/+5C, (B) PR−5C/+5C, (C) TFR-PRC7/C95, (D) PR20C7/C95, (E) TFR-PRL6/C7/C95, and (F) PRL6/C7/C95. Urea concentration (red) and the number of trajectories analyzed (N, black) are indicated on the left side of the histogram. The FRET efficiency was corrected for background, donor leak into the acceptor channel, and γ-factor (see Supporting Materials and Methods). Yellow and red bars in the FRET efficiency histograms indicate the unfolded and folded states, respectively. To see this figure in color, go online.

To summarize, the broad unfolded state FRET efficiency distribution is originated from the protein conformation rather than an artifact of fluorophore labeling. This result indicates that in addition to the completely unfolded state, there may be additional, relatively stable unfolded substates, which are largely disordered but contain residual structures or are more collapsed compared to the completely disordered state so that they exhibit higher FRET efficiencies. This observation is consistent with a previous NMR study showing extra weak resonances, in addition to the resonances corresponding to the unfolded PR in a heteronuclear single quantum coherence spectrum at 5 M GdmCl (34).

Existence of stable unfolded substates and direct observation of diverse folding pathways

Fig. 4 shows transition maps constructed by the FRET efficiency values before and after changes in FRET efficiency (transitions) (see Fig. S6 for the transition maps at different urea concentrations). The majority of the transitions of PR−5C/+5C and its precursor (TFR-PR−5C/+5C) are those between the folded and unfolded states (indicated by red dots in Fig. 4, A and B), and there are only a small number of transitions within the unfolded state (blue dots). On the other hand, more transitions within the unfolded state are observed for the precursor of C7/C95 (Fig. 4 C). This observation contrasts with the simple transition map of a two-state folder, such as protein G (29), which exhibits transitions mostly between the folded and unfolded states (after excluding transitions caused by the spectral shift of Alexa Fluor 488). Moreover, these transitions within the unfolded state are clustered. At 2 and 2.5 M urea, the clusters appear at E (initial) ∼0.6 and E (final) ∼0.8 and vice versa (Figs. 4 C and S6 C). This result suggests that there are multiple stable unfolded substates. Although these two clusters would not be well-defined states, we call these low-E (E ∼0.6) and high-E (E ∼0.8) unfolded states, for the sake of convenience for discussion in this article. Because the high-E unfolded state is in between the low-E state and the folded state, it may be an intermediate state. However, transitions occur directly between the folded and both high- and low-E unfolded states (red dots in Fig. 4 C). Therefore, the high-E unfolded state is not an on-pathway intermediate state but another unfolded substate, possibly with residual structures by non-native interactions. (The flexibility of the high-E unfolded state is also as high as that of the low-E unfolded state as discussed in the next section). The detection of transitions between the folded state and the unfolded state with a very wide range of the FRET efficiency means that diverse folding and unfolding pathways are directly observed because the folding pathways from the unfolded states with significantly different FRET efficiency values will not be the same. This is a consequence of a very rugged energy landscape, in which the FRET efficiency of the unfolded state is not rapidly averaged out.

Figure 4.

Figure 4

Open in a new tab

Transition map constructed from the FRET efficiencies before (E (initial)) and after (E (final)) transitions in the immobilization experiments. (A–F) Transition maps and FRET efficiency histograms in Fig. 3 are shown for the various PR constructs: (A) TFR-PR−5C/+5C, (B) PR−5C/+5C, (C) TFR-PRC7/C95, (D) PR20C7/C95, (E) TFR-PRL6/C7/C95, and (F) PRL6/C7/C95. Urea concentration is indicated on the upper left corner of the FRET efficiency histogram. Red dots indicate transitions between the folded and unfolded states, and blue dots indicate transitions between unfolded substates. Transitions with a significant FRET efficiency difference (ΔE) compared with the noise (SD of the segments, SD) are plotted (ΔE > 1.64 SD) to exclude small fluctuations caused by a drift of the FRET efficiency. Black vertical and horizontal dashed lines guide the center of the FRET efficiencies of the folded and unfolded states. Green vertical and horizontal dashed lines in (C–E) indicate the unfolded states with high FRET efficiencies. The complete data set at all urea concentrations are displayed in Fig. S6. To see this figure in color, go online.

The FRET efficiency distribution of L6/C7/C95 is much narrower and localized at E ∼0.5–0.65 depending on the urea concentration (Figs. 4, E and F and S6, E and F). Although the equilibrium population of the high-E unfolded state of L6/C7/C95 is very low, interestingly, transitions are observed between E ∼0.6 and 0.8 for both precursor and mature proteins of L6/C7/C95. This result indicates that although the FRET efficiency distribution of the unfolded state is narrow for L6/C7/C95, the mutation W6L does not completely eliminate the ruggedness of the energy landscape.

In the case of PR20C7/C95, a relatively smaller number of transitions between the folded and unfolded states are observed because of the slow folding/unfolding kinetics (Figs. 4 D and S6 D). On the other hand, transitions within the unfolded states are frequently observed similar to C7/C95 and L6/C7/C95, indicating that the ruggedness of the unfolded state energy landscape persists in this highly mutated multi-drug-resistant variant.

Each unfolded substate is highly flexible

If the high-E unfolded state contains a significant amount of residual structures, the unfolded chain may be less flexible than the completely disordered state. To compare the flexibility of conformations in each state, we used two-dimensional (2D) FRET efficiency-lifetime analysis (35, 36, 37). When the distance between the two fluorophores is fixed (i.e., rigid structure), the FRET efficiency and the donor lifetime are related as τD/τD0 = 1 − E, where τD and τD0 are the donor lifetimes in the presence and absence of the acceptor, respectively (38). In this case, the distribution appears on the diagonal of a 2D plot. On the other hand, when there is a distribution of the donor-acceptor distance, with sufficiently fast fluctuations so that transitions do not appear in the binned trajectories, the FRET efficiency and donor lifetime is differently related as τD/τD0 = 1 − E + σc2/(1 − E) (35). σc2 is the variance of the FRET efficiency of the underlying conformational distribution (note that this variance is not related to the width of the peaks in a FRET efficiency histogram). Therefore, this variance value reflects the relative conformational flexibility of a state.

Fig. 5, A and B shows the 2D FRET efficiency-lifetime distribution of TFR-PR−5C/+5C and its mature counterpart (see Fig. S7, A and B for the data at other urea concentrations). Both the folded and unfolded states are shifted upward from the diagonal. The large σc2 values of the unfolded state of 0.14–0.15 indicate a large conformational flexibility of the unfolded state. The unfolded polypeptide chain has been very well described by the Gaussian chain model (39, 40, 41, 42, 43, 44). σc2 value expected from a Gaussian chain with the FRET efficiency of 0.4–0.5 is 0.13, which is very close to the experimental values (Figs. 5 and S7). In addition to the unfolded state, the folded state distribution is also shifted from the diagonal (σc2 = 0.04). This shift results from the flexibility of the unfolded residues at the N- and C-termini in this construct, which also lowers the FRET efficiency of the folded state compared to those of the other constructs (C7/C95 and C2/C98). The distance between the two residues flanking the folded region is very short, as indicated by the very high FRET efficiency of the folded state of C7/C95. Therefore, the shift of the 2D distribution resulting from the disordered part in the folded state of −5C/+5C, which consists of the residues −5 to 10 and 91–104, also lies near the curve, corresponding to a Gaussian chain in Figs. 5, A and B and S7, A and B.

Figure 5.

Figure 5

Open in a new tab

2D FRET efficiency-donor lifetime analysis. The distributions were constructed using the values of the FRET efficiency and donor lifetime of the first segment of each trajectory of (A) TFR-PR−5C/+5C, (B) PR−5C/+5C, (C) TFR-PRC7/C95, (D) PR20C7/C95, (E) TFR-PRL6/C7/C95, and (F) PRL6/C7/C95. Segments containing more than 3000 photons were included in the analysis. The distributions are shifted upward from the diagonal line, indicating the distributions of the distance between the donor and acceptor that fluctuate rapidly (see text). The data inside the rectangles (yellow/blue: unfolded; red: folded) were used to calculate the average FRET efficiency and lifetime values for the estimation of the variance of the FRET efficiency due to these distance distributions (Table S2). The red curve shows the lifetime dependence on the FRET efficiency of a Gaussian chain. Urea concentrations are indicated on the upper left corner. The complete data set at all urea concentrations are displayed in Fig. S7. To see this figure in color, go online.

Compared to −5C/+5C, the flexibility of the folded state of C7/C95, L6/C7/C95, and PR20C7/C95 (Figs. 5, C–F and S7, C–F) is very small (σc2 = 0.01), indicating that the dye labels are proximal to the folded region of the protein, consistent with the very high FRET efficiency (E > 0.95). The broad distribution along the lifetime axis results from the error in the lifetime determination for individual trajectory segments because of the small number of donor photons when the FRET efficiency is very high. On the other hand, σc2 values of the unfolded state (Figs. 5, C–F and S7, C–F, yellow rectangles) is 0.13–0.14, indicating that the flexibility of the unfolded state is similar to that of −5C/+5C. σc2 values of the high-E unfolded state (Figs. 5, C and D and S7, C and D, blue rectangles) are 0.08–0.13, smaller than those of the low-E unfolded state (Table S2). However, the variance value expected from the Gaussian chain with this FRET efficiency range (E = 0.6–0.8) is 0.06–0.11, which is similar to the experimental values. In the case that a significant amount of residual secondary or tertiary structures are present in the high-E unfolded state, the Gaussian chain model may not describe the chain dynamics of this state properly. However, the large positive shift in the 2D plot indicates that molecules in this higher FRET efficiency range should also be largely disordered and very flexible.

Similar folding kinetics of various mutants indicate robustness of PR folding

Because of the slow folding and unfolding kinetics, transitions do not frequently occur, and no transition is observed in many trajectories, which makes it difficult to carry out conventional kinetic analysis using the distributions of the residence times in a state. Therefore, we first determined the equilibrium fraction of the folded state by counting the folded molecules in the histogram constructed from the FRET efficiency values of the first segment of the trajectories of immobilized molecules (Fig. 3). Using this value, the rate coefficients were obtained using the maximal likelihood method (see Fig. S8; Supporting Materials and Methods). In this analysis, transitions within the unfolded state were ignored, and unfolded substates were merged into a single unfolded state. The results are summarized in Fig. 6 and Table S3.

Figure 6.

Figure 6

Open in a new tab

Equilibrium and kinetic parameters of folding. (Left) TFR-PR−5C/+5C (light green) and PR−5C/+5C (orange) are shown. (Right) TFR-PRL6/C7/C95 (light green), PRL6/C7/C95 (orange), TFR-PRC7/C95 (blue), and PR20C7/C95 (purple) are shown. (A and B) The fraction of folded molecules (pF) calculated from the histogram in Fig. 3A and the sum of folding and unfolding rate coefficients (B) are shown. (C) Folding (kF, open circle) and unfolding (kU, solid square) rate coefficients were fitted to RTln kF = RTln kF0 + mF [urea] and RTkU = RTln kU0 + mU [urea], respectively. R = 8.3145 J mol−1 K−1 is the gas constant, and T (295.15 K) is the temperature. Fitting parameters are listed in Table S3. Errors in (A) are SDs (σ) of the binomial distribution (σ2 = pF(1 − pF)/N, where N is the number of molecules in Fig. 3), and errors in (B and C) are SDs calculated from the curvature at the maximum of the likelihood functions. To see this figure in color, go online.

The analysis shows that the folded population (stability) and the folding and unfolding rates are similar between the precursor and mature proteins for both −5C/+5C and L6/C7/C95 constructs. The free-diffusion data in Fig. S3, A and C also suggest that the stability of the precursor and mature proteins are similar for these constructs. The slopes of the rate coefficients in Fig. 6 C show that the folding rate depends strongly on the urea concentration, whereas the unfolding rate is almost independent of it. More importantly, despite the very different FRET efficiency distributions of the unfolded state, the extrapolated folding rates at 0 M urea (kF0) are not so different for various constructs. kF0 varies only by a factor of 2–3 (Table S3). In other words, the ruggedness of the unfolded state energy surface is very sensitive to mutations, but the overall folding rate is much less sensitive, indicating robustness of folding. The extracted kinetic parameters are consistent with an ensemble measurement of folding that is coupled with dimerization at high protein concentration (45).

On the other hand, the folding rate of PR20C7/C95 is more than an order of magnitude slower than those of other constructs, whereas the unfolding rate of this construct is comparable to that of the wild-type (C7/C95). This suggests that drug-resistant mutations in PR20 have a significant impact on the folding barrier height, whereas its unfolded state energy landscape is still rugged compared to the wild-type.

Discussion

In this study, we have described folding and characteristics of the unfolded state of various constructs of HIV-1 PR with different dye labeling positions. Our results reveal that the folding process of the PR monomer is complex because of the rugged energy landscape. We observed a very broad FRET efficiency distribution of the unfolded state when dyes are labeled flanking the folded region in the monomer form (C7/C95, Figs. 1 and S3). Although broad unfolded state distributions have been observed for several proteins such as ribonuclease H (46) and adenylase kinase (47), to our knowledge, HIV-1 PR presents the most extreme case. Moreover, the broadness is very sensitive to the mutations and the position of dye labels. The distribution is narrowed by replacing tryptophan 6 with leucine (L6/C7/C95, Figs. 1 and S3 C), which may suggest that the broad distribution is an artifact caused by direct interactions between Trp and dyes. However, the anisotropy measurement rules out this possibility (Fig. S4; Table S1). Therefore, the broad FRET efficiency peak of the unfolded state indicates that the conformational fluctuations of the unfolded PR monomer are slow. The sensitivity of the unfolded state distribution to small changes such as a single mutation and dye labeling is actually one of the characteristics of the rugged energy landscape (1, 17).

Immobilization experiments show that the interconversion time of conformations within the unfolded state can be as long as ∼10 s (Fig. 2), which is much slower than typical conformational fluctuation times of 10–100 ns for chemically denatured and intrinsically disordered proteins (15, 16, 42). This long fluctuation time suggests clusters of unfolded conformations that are separated by high free energy barriers, meaning that the unfolded state energy landscape is very rugged for HIV-1 PR. This result is consistent with the previous NMR observations of residual or partially folded structures at a high (5 M) GdmCl concentration (34).

The rugged energy landscape allows for observing various unique features of folding, as we discuss below. The energy landscape theory describes that, in general, heterogeneous folding pathways exist as seen in molecular dynamics simulations of small fast-folding proteins (8, 48), although some proteins exhibit relatively well-defined folding pathways (11, 49). In most experimental methods, however, distinguishing diverse folding pathways is very difficult because of fast conformational fluctuations in the unfolded state and signal averaging. In single-molecule mechanical unfolding studies of SH3 domain and cold shock protein B using optical tweezers and atomic force microscopy, respectively, different unfolding pathways have been observed (50, 51, 52), although both proteins exhibit two states in ensemble measurements. Marqusee and co-workers have shown that the two pathways of SH3 domain switch, depending on the amount and direction of pulling force (50, 51, 53). Schönfelder et al. (52) have observed that multiple intermediate states of cold shock protein B connect the folded and unfolded states in various ways, and therefore, different combinations of intermediates appear in each unfolding transition. In addition, Caldarini et al. (54) have detected multiple pathways of mature HIV-1 PR during folding and unfolding in optical tweezer experiments. In this study, the majority of transitions showed a single-step (i.e., two-state) or two-step (i.e., three-state) transition, and more complex transitions were observed for a minor fraction of trajectories. However, simulation studies have shown that the folding mechanism can change depending on the amount of force and pulling direction because the experimental reaction coordinate, which is the pulling coordinate, does not necessarily coincide with the intrinsic folding reaction coordinate in the absence of force (55, 56). Pulling force can also selectively stabilize intermediates, which do not appear in bulk measurement with chemical denaturant (57). Perhaps unbiased heterogeneous folding pathways can be directly observed by monitoring individual folding transition paths (i.e., barrier crossing) in equilibrium single-molecule FRET trajectories. However, the transition path time is short, and until now, only the average times have been measured for several proteins (19, 20, 58). Alternatively, if the energy landscape is highly rugged, slow interconversions between unfolded state conformations may allow direct observation of diverse folding pathways in equilibrium trajectories. Indeed, transition maps in Figs. 4 and S6 clearly show the diversity of folding pathways and their modulation. The unfolded state peak is the broadest for C7/C95 (Fig. 4 C), and transitions occur between the FRET efficiency of the folded state and virtually all FRET efficiency values of the broad unfolded state (Figs. 4 C and S6 C). The distribution of the transitions become more clustered as the unfolded state distribution becomes narrower (L6/C7/C95, Figs. 4, E and F and S6, E and F).

The extremely broad FRET distribution of the unfolded state indicates conformational fluctuations are much slower compared to the measurement timescale. However, the positive shift of the unfolded state distribution in the 2D FRET efficiency-donor lifetime plot is large and aligns well with that of a Gaussian chain (Figs. 5 and S7). This result indicates that there are large barriers between the clusters of unfolded conformations (unfolded substates) with different FRET efficiencies, but the conformational fluctuations within each unfolded substate are very fast.

To summarize, various observations made by single-molecule FRET in this study indicate that the energy landscape of HIV-1 PR is highly rugged, which enables direct observation of heterogeneous folding pathways. This raises a question: what causes this ruggedness even though the PR is also a naturally evolved protein? Generally, proteins have evolved to have a smoother energy landscape that will enhance folding efficiency. However, a smooth energy landscape may adversely affect folding of the PR. The infidelity (error-prone) of reverse transcriptase results in frequently occurring mutations (50 among 99 residues mutate), even in the absence of drugs (28), and progressively, up to 19 mutations accumulate within one variant as in the extremely drug-resistant mutant PR20. The energy landscape optimized for efficient folding of a small subset of these mutants will fail to fold the rest of the mutants and limits the scope of HIV survival. Instead, marginally optimizing the energy landscape elevates the possibility to fold more mutants robustly and thus increases the adaptability of the virus. The consequence of this marginal optimization is the increased ruggedness, which results in slow folding. This hypothesis is supported by our observation that the folding rate extrapolated to the native condition (0 M urea) is insensitive to the location of dye labels and a small number of mutations (Fig. 6; Table S3), indicating robust folding, in contrast to the high sensitivity of the unfolded state energy landscape. In addition, the folding rate of HIV-1 PR at the native condition is indeed slow (5–10 s) compared to those of many other small single-domain proteins (range 1 μs to ms) (59, 60), presumably exhibiting a smooth energy landscape. Our results show that a folding energy landscape can be largely modulated in response to selection pressures during evolution as in HIV.

Author Contributions

J.Y., J.M.L., and H.S.C. designed and performed research and wrote the manuscript. J.Y. and H.S.C. analyzed data.

Acknowledgments

We thank W. A. Eaton, A. Szabo, and R. B. Best for numerous helpful discussions and comments, A. Aniana for technical assistance, and J. Lloyd for mass spectrometry.

This work was supported by the Intramural Research Program of the National Institute of Diabetes and Digestive and Kidney Diseases of National Institutes of Health.

Editor: Elizabeth Rhoades.

Footnotes

Supporting Material can be found online at https://doi.org/10.1016/j.bpj.2019.09.015.

Supporting Material

Document S1. Supporting Materials and Methods, Figs. S1–S8, and Tables S1–S3
mmc1.pdf (577.7KB, pdf)
Document S2. Article plus Supporting Material
mmc2.pdf (1.8MB, pdf)

Supporting Citations

References (61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78) appear in the Supporting Material.

References

  • 1.Bryngelson J.D., Onuchic J.N., Wolynes P.G. Funnels, pathways, and the energy landscape of protein folding: a synthesis. Proteins. 1995;21:167–195. doi: 10.1002/prot.340210302. [DOI] [PubMed] [Google Scholar]
  • 2.Gin B.C., Garrahan J.P., Geissler P.L. The limited role of nonnative contacts in the folding pathways of a lattice protein. J. Mol. Biol. 2009;392:1303–1314. doi: 10.1016/j.jmb.2009.06.058. [DOI] [PubMed] [Google Scholar]
  • 3.Best R.B., Hummer G., Eaton W.A. Native contacts determine protein folding mechanisms in atomistic simulations. Proc. Natl. Acad. Sci. USA. 2013;110:17874–17879. doi: 10.1073/pnas.1311599110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Bryngelson J.D., Wolynes P.G. Intermediates and barrier crossing in a random energy model (with applications to protein folding) J. Phys. Chem. 1989;93:6902–6915. [Google Scholar]
  • 5.Klimov D.K., Thirumalai D. Viscosity dependence of the folding rates of proteins. Phys. Rev. Lett. 1997;79:317–320. [Google Scholar]
  • 6.Best R.B., Hummer G. Diffusion models of protein folding. Phys. Chem. Chem. Phys. 2011;13:16902–16911. doi: 10.1039/c1cp21541h. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Zheng W., Best R.B. Reduction of all-atom protein folding dynamics to one-dimensional diffusion. J. Phys. Chem. B. 2015;119:15247–15255. doi: 10.1021/acs.jpcb.5b09741. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Lindorff-Larsen K., Piana S., Shaw D.E. How fast-folding proteins fold. Science. 2011;334:517–520. doi: 10.1126/science.1208351. [DOI] [PubMed] [Google Scholar]
  • 9.Dill K.A., Chan H.S. From Levinthal to pathways to funnels. Nat. Struct. Biol. 1997;4:10–19. doi: 10.1038/nsb0197-10. [DOI] [PubMed] [Google Scholar]
  • 10.Eaton W.A., Wolynes P.G. Theory, simulations, and experiments show that proteins fold by multiple pathways. Proc. Natl. Acad. Sci. USA. 2017;114:E9759–E9760. doi: 10.1073/pnas.1716444114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Englander S.W., Mayne L. The case for defined protein folding pathways. Proc. Natl. Acad. Sci. USA. 2017;114:8253–8258. doi: 10.1073/pnas.1706196114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Klimov D.K., Thirumalai D. Lattice models for proteins reveal multiple folding nuclei for nucleation-collapse mechanism. J. Mol. Biol. 1998;282:471–492. doi: 10.1006/jmbi.1998.1997. [DOI] [PubMed] [Google Scholar]
  • 13.Klimov D.K., Thirumalai D. Multiple protein folding nuclei and the transition state ensemble in two-state proteins. Proteins. 2001;43:465–475. doi: 10.1002/prot.1058. [DOI] [PubMed] [Google Scholar]
  • 14.Sadqi M., Lapidus L.J., Muñoz V. How fast is protein hydrophobic collapse? Proc. Natl. Acad. Sci. USA. 2003;100:12117–12122. doi: 10.1073/pnas.2033863100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Nettels D., Gopich I.V., Schuler B. Ultrafast dynamics of protein collapse from single-molecule photon statistics. Proc. Natl. Acad. Sci. USA. 2007;104:2655–2660. doi: 10.1073/pnas.0611093104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Schuler B., Hofmann H. Single-molecule spectroscopy of protein folding dynamics--expanding scope and timescales. Curr. Opin. Struct. Biol. 2013;23:36–47. doi: 10.1016/j.sbi.2012.10.008. [DOI] [PubMed] [Google Scholar]
  • 17.Oliveberg M., Wolynes P.G. The experimental survey of protein-folding energy landscapes. Q. Rev. Biophys. 2005;38:245–288. doi: 10.1017/S0033583506004185. [DOI] [PubMed] [Google Scholar]
  • 18.Sadqi M., de Alba E., Muñoz V. A designed protein as experimental model of primordial folding. Proc. Natl. Acad. Sci. USA. 2009;106:4127–4132. doi: 10.1073/pnas.0812108106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Chung H.S., Eaton W.A. Single-molecule fluorescence probes dynamics of barrier crossing. Nature. 2013;502:685–688. doi: 10.1038/nature12649. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Chung H.S., Piana-Agostinetti S., Eaton W.A. Structural origin of slow diffusion in protein folding. Science. 2015;349:1504–1510. doi: 10.1126/science.aab1369. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 21.Solomatin S.V., Greenfeld M., Herschlag D. Multiple native states reveal persistent ruggedness of an RNA folding landscape. Nature. 2010;463:681–684. doi: 10.1038/nature08717. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Okumus B., Wilson T.J., Ha T. Vesicle encapsulation studies reveal that single molecule ribozyme heterogeneities are intrinsic. Biophys. J. 2004;87:2798–2806. doi: 10.1529/biophysj.104.045971. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Hyeon C., Lee J., Thirumalai D. Hidden complexity in the isomerization dynamics of Holliday junctions. Nat. Chem. 2012;4:907–914. doi: 10.1038/nchem.1463. [DOI] [PubMed] [Google Scholar]
  • 24.Louis J.M., Ishima R., Weber I.T. HIV-1 protease: structure, dynamics, and inhibition. Adv. Pharmacol. 2007;55:261–298. doi: 10.1016/S1054-3589(07)55008-8. [DOI] [PubMed] [Google Scholar]
  • 25.Lee S.K., Potempa M., Swanstrom R. The choreography of HIV-1 proteolytic processing and virion assembly. J. Biol. Chem. 2012;287:40867–40874. doi: 10.1074/jbc.R112.399444. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Shen C.H., Chang Y.C., Weber I.T. Conformational variation of an extreme drug resistant mutant of HIV protease. J. Mol. Graph. Model. 2015;62:87–96. doi: 10.1016/j.jmgm.2015.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Louis J.M., Roche J. Evolution under drug pressure remodels the folding free-energy landscape of mature HIV-1 protease. J. Mol. Biol. 2016;428:2780–2792. doi: 10.1016/j.jmb.2016.05.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Weber I.T., Agniswamy J. HIV-1 protease: structural perspectives on drug resistance. Viruses. 2009;1:1110–1136. doi: 10.3390/v1031110. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Chung H.S., Louis J.M., Eaton W.A. Experimental determination of upper bound for transition path times in protein folding from single-molecule photon-by-photon trajectories. Proc. Natl. Acad. Sci. USA. 2009;106:11837–11844. doi: 10.1073/pnas.0901178106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Nettels D., Müller-Späth S., Schuler B. Single-molecule spectroscopy of the temperature-induced collapse of unfolded proteins. Proc. Natl. Acad. Sci. USA. 2009;106:20740–20745. doi: 10.1073/pnas.0900622106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Ishima R., Torchia D.A., Louis J.M. Solution structure of the mature HIV-1 protease monomer: insight into the tertiary fold and stability of a precursor. J. Biol. Chem. 2003;278:43311–43319. doi: 10.1074/jbc.M307549200. [DOI] [PubMed] [Google Scholar]
  • 32.Marmé N., Knemeyer J.P., Wolfrum J. Inter- and intramolecular fluorescence quenching of organic dyes by tryptophan. Bioconjug. Chem. 2003;14:1133–1139. doi: 10.1021/bc0341324. [DOI] [PubMed] [Google Scholar]
  • 33.Chen H., Ahsan S.S., Webb W.W. Mechanisms of quenching of Alexa fluorophores by natural amino acids. J. Am. Chem. Soc. 2010;132:7244–7245. doi: 10.1021/ja100500k. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Bhavesh N.S., Sinha R., Hosur R.V. NMR elucidation of early folding hierarchy in HIV-1 protease. J. Biol. Chem. 2003;278:19980–19985. doi: 10.1074/jbc.M301615200. [DOI] [PubMed] [Google Scholar]
  • 35.Gopich I.V., Szabo A. Theory of the energy transfer efficiency and fluorescence lifetime distribution in single-molecule FRET. Proc. Natl. Acad. Sci. USA. 2012;109:7747–7752. doi: 10.1073/pnas.1205120109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Chung H.S., Meng F., Louis J.M. Oligomerization of the tetramerization domain of p53 probed by two- and three-color single-molecule FRET. Proc. Natl. Acad. Sci. USA. 2017;114:E6812–E6821. doi: 10.1073/pnas.1700357114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Borgia A., Borgia M.B., Schuler B. Extreme disorder in an ultrahigh-affinity protein complex. Nature. 2018;555:61–66. doi: 10.1038/nature25762. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Lakowicz J.R. Third Edition. Springer; New York: 2006. Principles of Fluorescence Spectroscopy. [Google Scholar]
  • 39.Schuler B., Eaton W.A. Protein folding studied by single-molecule FRET. Curr. Opin. Struct. Biol. 2008;18:16–26. doi: 10.1016/j.sbi.2007.12.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Hoffmann A., Kane A., Schuler B. Mapping protein collapse with single-molecule fluorescence and kinetic synchrotron radiation circular dichroism spectroscopy. Proc. Natl. Acad. Sci. USA. 2007;104:105–110. doi: 10.1073/pnas.0604353104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Schuler B., Lipman E.A., Eaton W.A. Probing the free-energy surface for protein folding with single-molecule fluorescence spectroscopy. Nature. 2002;419:743–747. doi: 10.1038/nature01060. [DOI] [PubMed] [Google Scholar]
  • 42.Soranno A., Buchli B., Schuler B. Quantifying internal friction in unfolded and intrinsically disordered proteins with single-molecule spectroscopy. Proc. Natl. Acad. Sci. USA. 2012;109:17800–17806. doi: 10.1073/pnas.1117368109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Müller-Späth S., Soranno A., Schuler B. From the cover: charge interactions can dominate the dimensions of intrinsically disordered proteins. Proc. Natl. Acad. Sci. USA. 2010;107:14609–14614. doi: 10.1073/pnas.1001743107. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 44.Sherman E., Haran G. Coil-globule transition in the denatured state of a small protein. Proc. Natl. Acad. Sci. USA. 2006;103:11539–11543. doi: 10.1073/pnas.0601395103. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Noel A.F., Bilsel O., Matthews C.R. The folding free-energy surface of HIV-1 protease: insights into the thermodynamic basis for resistance to inhibitors. J. Mol. Biol. 2009;387:1002–1016. doi: 10.1016/j.jmb.2008.12.061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Kuzmenkina E.V., Heyes C.D., Nienhaus G.U. Single-molecule Forster resonance energy transfer study of protein dynamics under denaturing conditions. Proc. Natl. Acad. Sci. USA. 2005;102:15471–15476. doi: 10.1073/pnas.0507728102. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Pirchi M., Ziv G., Haran G. Single-molecule fluorescence spectroscopy maps the folding landscape of a large protein. Nat. Commun. 2011;2:493. doi: 10.1038/ncomms1504. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Best R.B., Hummer G. Microscopic interpretation of folding ϕ-values using the transition path ensemble. Proc. Natl. Acad. Sci. USA. 2016;113:3263–3268. doi: 10.1073/pnas.1520864113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Englander S.W., Mayne L. The nature of protein folding pathways. Proc. Natl. Acad. Sci. USA. 2014;111:15873–15880. doi: 10.1073/pnas.1411798111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 50.Jagannathan B., Elms P.J., Marqusee S. Direct observation of a force-induced switch in the anisotropic mechanical unfolding pathway of a protein. Proc. Natl. Acad. Sci. USA. 2012;109:17820–17825. doi: 10.1073/pnas.1201800109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Guinn E.J., Jagannathan B., Marqusee S. Single-molecule chemo-mechanical unfolding reveals multiple transition state barriers in a small single-domain protein. Nat. Commun. 2015;6:6861. doi: 10.1038/ncomms7861. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Schönfelder J., Perez-Jimenez R., Muñoz V. A simple two-state protein unfolds mechanically via multiple heterogeneous pathways at single-molecule resolution. Nat. Commun. 2016;7:11777. doi: 10.1038/ncomms11777. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 53.Zhuravlev P.I., Hinczewski M., Thirumalai D. Force-dependent switch in protein unfolding pathways and transition-state movements. Proc. Natl. Acad. Sci. USA. 2016;113:E715–E724. doi: 10.1073/pnas.1515730113. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 54.Caldarini M., Sonar P., Cecconi C. The complex folding behavior of HIV-1-protease monomer revealed by optical-tweezer single-molecule experiments and molecular dynamics simulations. Biophys. Chem. 2014;195:32–42. doi: 10.1016/j.bpc.2014.08.001. [DOI] [PubMed] [Google Scholar]
  • 55.Best R.B., Paci E., Dudko O.K. Pulling direction as a reaction coordinate for the mechanical unfolding of single molecules. J. Phys. Chem. B. 2008;112:5968–5976. doi: 10.1021/jp075955j. [DOI] [PubMed] [Google Scholar]
  • 56.Graham T.G., Best R.B. Force-induced change in protein unfolding mechanism: discrete or continuous switch? J. Phys. Chem. B. 2011;115:1546–1561. doi: 10.1021/jp110738m. [DOI] [PubMed] [Google Scholar]
  • 57.de Sancho D., Best R.B. Reconciling intermediates in mechanical unfolding experiments with two-state protein folding in bulk. J. Phys. Chem. Lett. 2016;7:3798–3803. doi: 10.1021/acs.jpclett.6b01722. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 58.Chung H.S., McHale K., Eaton W.A. Single-molecule fluorescence experiments determine protein folding transition path times. Science. 2012;335:981–984. doi: 10.1126/science.1215768. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Jackson S.E. How do small single-domain proteins fold? Fold. Des. 1998;3:R81–R91. doi: 10.1016/S1359-0278(98)00033-9. [DOI] [PubMed] [Google Scholar]
  • 60.Kubelka J., Hofrichter J., Eaton W.A. The protein folding ‘speed limit’. Curr. Opin. Struct. Biol. 2004;14:76–88. doi: 10.1016/j.sbi.2004.01.013. [DOI] [PubMed] [Google Scholar]
  • 61.Louis J.M., Webert I.T., Gronenborn A.M. HIV-1 protease: maturation, enzyme specificity, and drug resistance. Adv. Pharmacol. 2000;49:111–146. doi: 10.1016/s1054-3589(00)49025-3. [DOI] [PubMed] [Google Scholar]
  • 62.Louis J.M., Aniana A., Sayer J.M. Inhibition of autoprocessing of natural variants and multidrug resistant mutant precursors of HIV-1 protease by clinical inhibitors. Proc. Natl. Acad. Sci. USA. 2011;108:9072–9077. doi: 10.1073/pnas.1102278108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Sayer J.M., Agniswamy J., Louis J.M. Autocatalytic maturation, physical/chemical properties, and crystal structure of group N HIV-1 protease: relevance to drug resistance. Protein Sci. 2010;19:2055–2072. doi: 10.1002/pro.486. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Louis J.M., Ishima R., Gronenborn A.M. Revisiting monomeric HIV-1 protease. Characterization and redesign for improved properties. J. Biol. Chem. 2003;278:6085–6092. doi: 10.1074/jbc.M209726200. [DOI] [PubMed] [Google Scholar]
  • 65.Ishima R., Torchia D.A., Louis J.M. Mutational and structural studies aimed at characterizing the monomer of HIV-1 protease and its precursor. J. Biol. Chem. 2007;282:17190–17199. doi: 10.1074/jbc.M701304200. [DOI] [PubMed] [Google Scholar]
  • 66.Merchant K.A., Best R.B., Eaton W.A. Characterizing the unfolded states of proteins using single-molecule FRET spectroscopy and molecular simulations. Proc. Natl. Acad. Sci. USA. 2007;104:1528–1533. doi: 10.1073/pnas.0607097104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Liu J., Campos L.A., Muñoz V. Exploring one-state downhill protein folding in single molecules. Proc. Natl. Acad. Sci. USA. 2012;109:179–184. doi: 10.1073/pnas.1111164109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Zheng W., Borgia A., Best R.B. Probing the action of chemical denaturant on an intrinsically disordered protein by simulation and experiment. J. Am. Chem. Soc. 2016;138:11702–11713. doi: 10.1021/jacs.6b05443. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.Borgia A., Zheng W., Schuler B. Consistent view of polypeptide chain expansion in chemical denaturants from multiple experimental methods. J. Am. Chem. Soc. 2016;138:11714–11726. doi: 10.1021/jacs.6b05917. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 70.Koshioka M., Sasaki K., Masuhara H. Time-dependent fluorescence depolarization analysis in three-dimensional microspectroscopy. Appl. Spectrosc. 1995;49:224–228. [Google Scholar]
  • 71.Schaffer J., Volkmer A., Seidel C.A.M. Identification of single molecules in aqueous solution by time-resolved fluorescence anisotropy. J. Phys. Chem. A. 1999;103:331–336. [Google Scholar]
  • 72.Gopich I.V., Szabo A. Theory of single-molecule FRET efficiency histograms. Adv. Chem. Phys. 2012;146:245–297. [Google Scholar]
  • 73.Meng F., Bellaiche M.M.J., Chung H.S. Highly disordered amyloid-β monomer probed by single-molecule FRET and MD simulation. Biophys. J. 2018;114:870–884. doi: 10.1016/j.bpj.2017.12.025. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 74.Chung H.S., Louis J.M., Eaton W.A. Distinguishing between protein dynamics and dye photophysics in single-molecule FRET experiments. Biophys. J. 2010;98:696–706. doi: 10.1016/j.bpj.2009.12.4322. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 75.Chung H.S., Louis J.M., Gopich I.V. Analysis of fluorescence lifetime and energy transfer efficiency in single-molecule photon trajectories of fast-folding proteins. J. Phys. Chem. B. 2016;120:680–699. doi: 10.1021/acs.jpcb.5b11351. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 76.Rabiner L.R. A tutorial on hidden Markov models and selected applications in speech recognition. Proc. IEEE. 1989;77:257–286. [Google Scholar]
  • 77.McKinney S.A., Joo C., Ha T. Analysis of single-molecule FRET trajectories using hidden Markov modeling. Biophys. J. 2006;91:1941–1951. doi: 10.1529/biophysj.106.082487. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 78.Gopich I.V., Szabo A. Decoding the pattern of photon colors in single-molecule FRET. J. Phys. Chem. B. 2009;113:10965–10973. doi: 10.1021/jp903671p. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Document S1. Supporting Materials and Methods, Figs. S1–S8, and Tables S1–S3
mmc1.pdf (577.7KB, pdf)
Document S2. Article plus Supporting Material
mmc2.pdf (1.8MB, pdf)

Articles from Biophysical Journal are provided here courtesy of The Biophysical Society

RESOURCES