Single-Trp variants of SNase were constructed by introducing a unique tryptophan residue at various positions throughout the structure into a tryptophan-free background containing P47G, P117G, H124L and W140H mutations.24
The residues targeted for Trp substitution are Ile15, Tyr27, Phe61, Phe76, Tyr91, Ala102 and His121, which represent most of the secondary structure elements of SNase (). The resulting single-tryptophan variants are denoted Trp15, Trp27, Trp61, Trp76, Trp91, Trp102 and Trp121 SNase, respectively.
The thermal unfolding transition of each variant was measured by monitoring the changes in the circular dichroism (CD) spectrum at 222 nm (in 25 mM sodium phosphate, 50 mM sodium chloride and 0.5 mM EDTA, pH 7.0). All variants show a cooperative unfolding transition indicative of a two-state mechanism. lists the thermodynamic parameters obtained using standard two-state analysis.44
values range from 3.0 kcal/mol (Trp91 SNase) to 5.2 kcal/mol (Trp121 SNase) at 20 ºC and pH 7.0, which is comparable to that of WT* SNase (6.1 kcal/mol), indicating that all variants maintain a native structure of at least moderate stability. On average, the single-Trp variants are destabilized by 1.8 kcal/mol relative to WT* SNase, but the standard deviation around this average is rather small (±0.8 kcal/mol), which facilitates comparison of their kinetic properties. The substitution of His for Trp140 accounts for 0.4 kcal/mol of this destabilization.24
also lists the CD signal at 222 nm (molar mean-residue ellipticity, [θ]222
) recorded at 15 oC prior to thermal unfolding scans. All variants show a moderate decrease in [θ]222
ranging from 80% (for Trp91) to 99% (Trp102) of the value for WT* SNase. However, the mutations do not give rise to significant changes in the shape of the CD spectra in the far-UV region (195–250 nm; data not shown for brevity), indicating that they do not alter the overall secondary structure content of the protein. Similar effects have previously been reported for other SNase mutants.39
A likely explanation for the variation in [θ]222
is that it reflects in part the mutational changes in aromatic residues, which are known to make some contributions to the far-UV CD spectrum.45
Thermodynamic parameters describing the thermal unfolding transitions of WT* SNase and a series of single-Trp SNasea (25 mM NaPO4, 50 mM NaCl and 0.5 mM EDTA at pH 7.0) monitored by CD at 222 nm.
Fluorescence spectra of WT* SNase and the seven single-Trp variants, along with N
-tryptophanamide (NATA) as a reference, were measured at 15 ºC under native (100 mM sodium acetate) and denaturing (100 mM sodium acetate and 2.4 M GuHCl) conditions at pH 5.2, and in the acid-denatured state (10 mM phosphoric acid, pH 2), using excitation wavelengths of 280 nm and 295 nm ( and ). In the absence of GuHCl, all the spectra exhibit significantly blue-shifted emission maxima (λmax
ranging from 323.4 to 341.7 nm) compared to that of NATA (λmax
~ 351 nm; ), indicating that the tryptophan residues are largely shielded from the solvent in the native state. On the other hand, the fluorescence yields show a great deal of variation among different variants. Trp15, Trp27 and Trp61 exhibit lower fluorescence under native conditions () compared with the corresponding unfolded states (), which can be attributed to specific contacts with polar quenchers in the native structure.35
Based on the structure of wild type SNase,46
likely quenching partners of Trp15, Trp27 and Trp61 are the side chains of Lys24, Lys28, and Gln106, respectively. Under native conditions, the tryptophans at positions 76, 91, 102 and 121 variants exhibit a major enhancement in fluorescence yields () with respect to the corresponding unfolded states (panels (e) and (f)), indicating that these tryptophan residues are at least partially buried in an apolar, solvent-shielded, environment (see Maki et al.24
with regard to Trp76 SNase). While the fluorescence yield of Trp91 SNase is comparable to that of other proteins when excited at 280 nm (), its emission spectrum, normalized with respect to an equimolar solution of NATA, is unusually intense when an excitation wavelength of 295 nm is used (). The Trp91 variant also exhibits red-shifted tryptophan absorption and fluorescence excitation spectra (data not shown), which appear to be responsible for its unusual fluorescence properties. The spectroscopic properties of Trp91 SNase, along with a crystal structure, will be reported elsewhere (K. M., C. Hayden, H.C., Y. C. Park and H.R., in preparation).
Figure 2 Fluorescence spectra of WT* SNase and single-Trp variants at 15°C under native and denaturing conditions. Emission spectra of WT*, Trp76, Trp91, Trp102 and Trp121 SNase with excitation at 280 nm (a) and 295 nm (b) in native buffer (100 mM sodium (more ...)
Table 2 Fluorescence emission properties of WT* and the single-Trp SNase variants under native (100 mM sodium acetate, pH 5.2), acid-unfolded (10 mM phosphoric acid, pH 2.0) and GuHCl-unfolded (100 mM sodium acetate, 2.4 M GuHCl, pH 5.2) conditions at 15 ºC. (more ...)
In the presence of 2.4 M GuHCl (), the fluorescence spectra of all single-Trp proteins studied are qualitatively similar to that of NATA, and the variation in the yields among the different proteins (0.80 ± 0.08) and λmax (347.9 ± 0.7 nm) is much smaller than under the native conditions (), indicating that the tryptophan residues are fully exposed to the solvent in the unfolded state. The variation in relative yields for the acid-denatured proteins is somewhat larger (0.87 ± 0.21), but λmax exhibits a similar average (347.2 nm) and narrow standard deviation (1.2 nm), indicating that the tryptophans are also largely exposed to the solvent in the acid-unfolded form. In the case of Trp27, which exhibits one of the lowest relative yields under native conditions (Irel = 0.46), the fluorescence is somewhat depressed even in the acid- and GuHCl-denatured state (Irel = 0.68–0.69), suggesting that the local Trp27-Lys28 quenching interaction persist in the unfolded state. These residues are part of a reverse turn connecting strands II and III of the native structure (), which may be partially formed even under denaturing conditions. Two other tryptophans (Trp140 in WT* SNase and Trp61) show evidence for residual quenching at pH 2, and may also participate in residual local interactions with polar residues. In contrast to the data in GuHCl, the acid-denatured forms of two variants (Trp91 and Trp102) show enhanced fluorescence relative to NATA (), suggesting that these tryptophans remain partially shielded from the solvent, perhaps due to (local) clustering of hydrophobic residues.
Taken together, the fluorescence properties of the variants indicate that all seven engineered trptophan residues are largely buried within the native structure and involved in specific tertiary interactions. Along with the observation of cooperative thermal unfolding transitions, our equilibrium analysis confirms that the single-Trp SNase variants maintain the ability to fold into a native-like structure with a close-packed hydrophobic core despite the fact that we replaced partially or fully buried side chains with a bulky tryptophan.
The refolding reaction of WT* SNase and the single-Trp variants was triggered by a pH jump from 2.0 to 5.2 by tenfold mixing of acid unfolded protein (in 20 mM phosphoric acid, pH 2.0) with refolding buffer (100 mM sodium acetate, pH 5.3) at 15 ºC. The time course of folding was measured by monitoring tryptophan fluorescence (), using a combination of continuous- and stopped-flow methods to cover the time range from ~100 μs to ~10 s. The dead times of the continuous- and stopped-flow instruments, measured under solvent conditions matching those of refolding experiments, were 100–120 μs and 2.4 ms, respectively. The kinetic traces were fitted to a sum of four to six exponential functions. The kinetic parameters thus obtained are listed in Supplementary Table 1
available online). The folding kinetics are similar to those of WT* and Trp76 SNase in terms of the number of kinetic phases and their approximate time constants,24
indicating that all variants follow a similar folding mechanisms. The kinetic scheme of SNase folding proposed on the basis of previous studies on a proline-free variant of this protein,40
as well as those on WT* and Trp76 SNase,24
is represented as (excluding the slow process rate-limited by the isomerization of the peptidyl prolyl bonds; see below):
Figure 3 Refolding kinetic traces of WT* SNase and the seven single-Trp variants induced by a pH-jump from 2.0 to 5.2 at 15°C, measured by the continuous-flow (colored lines) and stopped-flow (circles) methods. Panels (a) and (b) show the kinetics of the (more ...)
where U and U′ represent the unfolded species, while I1
, I′ and M are intermediate ensembles, and N is the native state. Accumulation of the I1
state on the major folding pathway is necessary to explain the fastest phase (phase 1 and the corresponding rate constant, λ1
). Accumulation of the I2
state accounts for a lag phase (phase 3) observed in the ~10 ms time range and the curvature in the chevron plot for the fourth and dominant folding phase (log λ4 vs.
Phase 4 corresponds to the rate-limiting step of the folding reaction, and the curvature is associated with the accumulation of the I2
state by the relationship λ4
. Similarly, to account for the curvature of the unfolding limb of the chevron plot reported previously,24,40
we introduced a high-energy native-like intermediate, M, which brings about a switch in the rate-limiting step for unfolding (λu
; see below). Accumulation of the I′ state along a parallel folding pathway is necessary to account for the second fastest phase (phase 2, λ2
). In addition, a minor phase slower than λ4
(phase 5, λ5
) is attributed to the interconversion between the I′ and M states (see below).
The slowest phase (phase 6, λ6
< 0.3 s−1
) of WT*, Trp76, Trp102 and Trp121 SNase arises from a small fraction of molecules folding via a parallel pathway that is rate-limited by proline isomerization (typical rate of ~0.01 s−1
However, the Trp15, Trp27 and Trp61 variants exhibits only five detectable phases. Since the rate of their slowest phase, λ5
, is similar to λ6
for the remaining proteins, it probably consists of two overlapping events one of which is rate-limited by the proline isomerization step and the other due to a late step along the minor folding pathway.
Phases 3 and 4 were assigned by using the rate constants and amplitudes of the corresponding phases of WT* and Trp76 SNase as a reference. The phase with the largest change in tryptophan fluorescence (enhancement or quenching, depending on the location of the tryptophan) during folding was assigned to phase 4 (I2
M step in ), whose rate constant ranges from 1.9 s−1
(for Trp15 SNase) and 11 s−1
(for WT* SNase). As previously reported for WT* and Trp76 SNase,24
a lag phase with a rate constant λ3
of 30 to 60 s−1
, which corresponds to the I1
step in , was also observed for the Trp15, Trp61 and Trp102 variants (Trp102 SNase has a somewhat faster lag phase of ~400 s−1
; Supplementary Table 1
). The characteristic feature of this phase is that the sign of the corresponding amplitude is opposite to that of phase 4, since the fluorescence intensity of the I2
state is closer to that of the unfolded rather than the native state.24,40
Although the refolding kinetics of Trp27 SNase lacks a lag phase, the second fastest phase (10.7 s−1
) was assigned to phase 3, since the rate of this process is ~3 times faster than λ4
and comparable to λ3
of the other proteins.
In addition to these four phases (phases 3 – 6) observed in the stopped-flow experiments, the continuous-flow fluorescence measurements revealed one (for Trp15, Trp27, Trp61, Trp91, Trp102, Trp121 and WT* SNase) or two (for Trp76 SNase) faster phases for all proteins (), which are assigned to the U
I′ steps in . Distinct changes in fluorescence occur with time constants ranging from 70 μs to 260 μs. Although the amplitude of these rapid phases is smaller than that of the rate-limiting step, the observed fluorescence enhancement nevertheless suggests that the tryptophan side chains are partially shielded from the solvent in this time regime. The fluorescence change observed in both continuous- and stopped-flow experiments fully accounts for the total fluorescence change associated with folding, without evidence for any missing amplitude (burst phase). The fluorescence intensities extrapolated to a refolding time of zero converge to a value between 1.2 to 1.4 on a scale where the fluorescence of the acid unfolded state is 1, which can be attributed to the fact that the fluorescence intensity of a solvent-exposed tryptophan is ~30% higher at pH 5.2 relative to that at pH 2.0 (measured for NATA at 15 ºC). Thus, the initial fluorescence intensity observed here corresponds to that of the unfolded state under refolding condition (pH 5.2 and 15 ºC).
The microscopic rate constants (kij
) and the fluorescence intensity of each ensemble were systematically varied until the predicted time course, based on , reproduced the observed kinetic trace. The kinetic parameters previously determined for WT* and Trp76 SNase served as a starting point in this modeling effort (see Maki et al.24
for a more detailed description of kinetic modeling procedures). The apparent rates and the corresponding amplitudes were determined by numeric calculation of the eigenvalues and eigenvectors of the rate matrix corresponding to . The microscopic rate constants and the fluorescence intensities of each species thus obtained are listed in and , respectively. The solid lines in show the calculated time-course of folding for each variant, which is virtually indistinguishable from the curves obtained by nonlinear least-squares fitting of exponentials (data not shown).
Elementary rate constants, k0ij (s−1), estimated by modeling the folding kinetics of WT* SNase and the single-Trp variants, based on .
Fluorescence intensity of each state relative to the acid unfolded state estimated by modeling the folding kinetics of WT* and the single-Trp variants, based on .
A full analysis of the folding mechanism for each variant would require systematic studies as a function of denaturant concentration using multiple spectroscopic probes (e.g., intrinsic and ANS fluorescence), as in our previous analysis of WT* and Trp76 SNase, which is beyond the scope of the present study. However, we were able to obtain reliable estimates of the populations and relative fluorescence yields of the major intermediates encountered during folding by constraining some of the microscopic rate constants defined in on the basis of our previous work on WT* and Trp76 SNase, which included studies as a function of urea concentration.24
In cases where two rate constants are mutually coupled, one of them was set to an arbitrary value within a range of possible values. Parameters thus constrained include the back rates k1U
, as well as the forward rate (kMN
), and the interconversion rates between U and U′ (kUU′
, and kU′U
). Although most kinetic parameters are close to those previously determined, some of the rates for WT* and Trp76 SNase (kU1
of WT*, and kU1
of Trp76 SNase) were modified slightly in order to improve the fit of the kinetic traces in . As in our previous work, we also assumed an arbitrary value of 50,000 s−1
; since the native-like state M is a high-energy intermediate, we can only define the ratio kMN
, based on the roll-over observed in the kinetics of unfolding,24,40 but the individual rate constants cannot be determined independently. The two unfolded ensembles (U and U′) are assumed to interconvert slowly, so that the major and minor folding pathways are independent of each other during the folding reactions. The two largest rates (λ1
) observed for Trp76 SNase reflect the elementary rate constants kU1
, respectively. All other variants show only one phase on the submillisecond time scale (see Supplementary Table 1
), indicating that the initial collapse rate for both the major and minor unfolded species are similar (kU1
). This conclusion is consistent with the observation that all variants exhibit a minor slow phase (phase 5) assigned to refolding of a heterogeneous I′ ensemble, and thus explains the presence of a rapid process corresponding to interconversion between U′ and I′ ensembles. The kU1
values of WT* SNase, which shows only a small decrease in fluorescence on the submillisecond time scale (), are consistent with our previous study using ANS fluorescence.24
After manually optimizing the elementary rate constants to reproduce the observable rates (see Supplementary Material, Table 1
), the intrinsic fluorescence of each state was adjusted so that the time-course calculated from the microscopic rate constants fully reproduces the experimentally obtained kinetic traces (). The fluorescence values for each state, normalized with respect to the acid-unfolded state at pH 2, are listed in . As mentioned above, the values for the U-state are larger than 1, since they reflect the unfolded ensemble under refolding conditions (pH 5.2) resulting in ~30% higher fluorescence compared to pH 2. The fluorescence of the native-like high-energy intermediate M, whose spectroscopic properties cannot be measured directly, was assumed to be the same as that of the N state.40
This modeling procedure led to an excellent fit of the observed kinetic trace for each variant studied (), indicating that is consistent not only with WT* and Trp76 SNase,24
but also describes the folding mechanism of all other single-tryptophan variants studied here.
Our comprehensive set of kinetic parameters and relative fluorescence values for the different tryptophan probes ( and ) provides new insight into the structural characteristics of the intermediates populated during SNase folding. While the variants studied previously lacked appropriate optical probes for monitoring the rapid formation of the β-barrel domain, we now have unique fluorophores in different locations () that allow us to monitor the formation of the β-barrel and the α-helical domains. For several variants (Trp15, Trp61, Trp76, Trp91, Trp102 and Trp121), the I1
and I′ states populated on the 100 μs time scale exhibit enhanced fluorescence compared with the unfolded ensembles (; ). Since the fluorescence emission spectra under denaturing conditions () indicate that all tryptophans are largely exposed to the solvent in the unfolded state, the initial increase in the fluorescence intensity is likely to reflect partial protection of the tryptophan side chains from the solvent in the I1
-state (the same is true for the I′ state populated along a minor parallel pathway, but for simplicity we will focus on the major pathway). A rapid increase in fluorescence on the submillisecond time scale was reported for several other proteins, including the bacterial immunity protein Im7 and single-Trp variants of ubiquitin and ribonuclease A, and was also attributed to partial burial of a tryptophan residue within the hydrophobic core.17,25,48
In the present study, we have observed a rapid fluorescence enhancement for several individual tryptophan probes in different locations throughout the SNase structure (), including the β-barrel domain (strands I, IV, V and helices H1) and parts of α-helical domain (helix H2 and the N-terminus of helix H3), suggesting that a major portion of the chain undergoes a cooperative collapse within ~100 μs of initiating the folding reaction. This result is fully consistent with the previous folding experiments on WT* and Trp76 SNase monitored via
the hydrophobic dye ANS, which revealed the formation of a loosely packed hydrophobic core in the same time regime.24
The three variants that experience intramolecular fluorescence quenching under native conditions () are especially informative. While the fluorescence of Trp15 and Trp61 SNase is enhanced in the I1
state compared to both U and N states, Trp27 SNase shows a small, but clearly detectable, decrease in fluorescence during the initial folding phase (). In fact, Trp27 SNase is the only one among all single-Trp variants studied with a lower intrinsic relative to U (), indicating that Trp27 comes in contact with an fluorescence in I1
intramolecular quencher; a likely candidate is Lys28, which is within 5 Å of Tyr27 in the structure of SNase.46,47
Interestingly, residue 27 is located in a β-turn connecting strands II and III of the β-barrel, whose amide groups are protected from H/D exchange early in refolding (within 10 ms).37
In addition, turn regions are considered to be likely initiation sites in protein folding in general.49,50
Thus, the observed decrease in fluorescence of Trp27 SNase on the submillisecond time scale provides evidence for rapid formation of the reverse turn within the strand II – strand III hairpin, which brings Trp27 within close contact of a quenching partner, such as Lys28.
For five variants (Trp76, Trp91, Trp102, Trp121, and WT* SNase), the fluorescence of the I2
state populated over the 10–100 ms time range is similar to or slightly higher than that of the I1
intermediate () and undergoes a major increase only during the main folding phase on the 100 ms time scale (). Thus, these tryptophan probes become progressively shielded from the solvent during folding and are fully buried upon formation of the native state (the M-state has negligible population during refolding and only affects the kinetics of unfolding). In the case of WT* SNase, the I2
state is more fluorescent than the U and I1
states, suggesting that the C-terminal region of helix H3 involving Trp140 is recruited to the structured regions at this stage of folding (~50 ms). In addition, previous H/D exchange results showed that stable hydrogen-bonded interactions begin to be formed between 10 and 100 ms, as indicated by protection factors larger than five for residues in strands II and III, as well as a residue in helix H3 (Lys134).37
Despite the native-like secondary structure content of the I2
state, based on the burst phase observed by far-UV CD,41,43
the fluorescence of I2
is closer to that of the unfolded rather than the native state, which gives rise to a lag phase in the appearance of the N-state ( and ), as previously reported for WT SNase and a proline-free variant.40
Our present observation that two variants, Trp15 and Trp61 SNase, exhibit a distinct kinetic phase with a time constant matching that of the lag phase (~30 ms) provides striking confirmation for this mechanism. Thus, the lag phase reflects accumulation of the I2
intermediate, which in the case of the Trp15 and Trp61 variants has a higher intrinsic fluorescence than all other states (), giving rise to a transient increase in fluorescence of these tryptophans before they encounter their quenching partners within the native structure. These observations further confirm that the I1
states are sequential intermediates along a direct path from the unfolded to the native state; if the native state could be reached rapidly along a parallel pathway, bypassing I1
, this would give rise to a fast fluorescence change towards the native level, counteracting the lag phase or transient increase in fluorescence51
(see also refs. 5, 52
for further discussion). It should be noted that the U′
M branch of does not provide a short-circuit, since the I′ → M transition is slower than the rate-limiting (I2
→ M) step along the main folding path.
The intrinsic fluorescence of Trp15 and Trp61 SNase is quenched after the rate-limiting step of the refolding reaction, indicating that the specific side chain contact between these engineered tryptophans and their respective quenching partners (Lys24 and Glu106) is established only upon formation of the native state. The fact that a specific contact between Trp15 and Lys24, which are located on strands I and II of the β-sheet, respectively, is established after the rate-limiting folding step indicates that complete assembly of the β-barrel occurs only during the final stages of folding, long after formation of the strand II/strand III hairpin. The progressive increase in fluorescence in going from U through I1
observed for the Trp15 and Trp61 variants is attributed to partial burial of the side chains in a hydrophobic environment within the collapsed molecule. In fact, a few hydrophobic residues, including Met26 (strand II) and Pro31 (turn between strands II and III), are in contact with Ile15 (strand I) in the crystal structure of H124L SNase.47
Similarly, Phe61 located in helix H1 interacts with hydrophobic residues in helix H1 and H2, including Met65 (helix H1), Val99 (helix H2) and Leu103 (helix H2). Thus, we suggest that the the Trp15 and Trp61 side chains first encounter these hydrophobic residues upon formation of the I1
states, before they engage in specific contacts with the corresponding quenchers (Lys24 and Gln106, respectively). This conclusion is supported by our previous folding experiments on WT* and Trp76 SNase monitored by ANS,24
and suggests that a loosely packed hydrophobic core is formed within the first 100 μs of refolding comprising a large fraction of theβ-barrel domain as well as helix H1 and H2.
Taken together, the folding kinetics as monitored by the various tryptophans suggests a scenario in which the probes are initially buried in a collapsed, but loosely packed ensemble of states, followed by the rate-limiting formation of the densely packed native core mediated by specific tertiary-structural contacts, which bring the tryptophans into close contact with intramolecular quenchers. As in the case of other proteins of similar or larger size,2
the folding mechanism of SNase exhibits several intermediate states populated along parallel folding channels. This kinetic complexity may be a manifestation of a “rough energy landscape”, which can give rise to a large number of alternative pathways that eventually converge toward a common free energy minimum corresponding to the native structure.53–56
According to this landscape view of protein folding, intermediates are thought to accumulate mainly as a result of kinetic trapping or topological frustration. It is quite possible that trapping of misfolded states is responsible for some of the minor species populated along parallel pathways during folding of SNase 24,36,40
, as well as other proteins.21,52,57,58
However, the states populated along the main pathway exhibit features characteristic of productive folding intermediates, including progressive structural organization and increasing stability towards the later stages of folding. Although as many as six exponential phases are required to reproduce the fluorescence-detected time course of folding (see Supplementary Material, Table 1
), the individual phases are well resolved in time, and a simple kinetic scheme with a limited number of populated states is adequate to describe the kinetics. This “chemical kinetics” description can be consolidated with the energy landscape view of protein folding if the free energy surface is subdivided into several regions (macro-states) mutually separated by substantial free energy barriers. The protein can rapidly explore conformational space within each region, which can comprise a broad ensemble of unfolded or partially folded states, but encounters larger kinetic barriers before entering another region of conformational space. This type of free energy surface can thus give rise to multi-exponential folding kinetics.