Multisite Fluorescence in Proteins with Multiple Tryptophan Residues

Time-resolved fluorescence experiments were carried out on a variety of apomyoglobins with one or two tryptophan (Trp) residues located at invariant positions 7 and 14 in the primary sequence. In all cases, the Trp fluorescence kinetics were resolved adequately into two discrete lifetime domains, and decay-associated spectra (DAS) were obtained for each decay component. The DAS resolved for unfolded proteins were indistinguishable by position of the emission maxima and the spectral shapes. The folded proteins revealed noticeable differences in the DAS, which relate to the diverse local environments around the Trp residues in the individual proteins. Furthermore, the DAS of wild-type protein possessing two Trp residues were simulated well by that of one Trp mutants either in the native, molten globule, or unfolded states. Overall, employing Trp fluorescence and site-directed mutagenesis allowed us to highlight the conformational changes induced by the single amino acid replacement and generate novel structural information on equilibrium folding intermediates. Specifically, it was found that conformational fluctuations in the local cluster around the evolutionarily conserved Trp14 are very similar in the native and molten globule states of apomyoglobins. This result indicates that residues in the E and B helices contributing to this cluster are most likely involved in the stabilization of the overall architecture of the structured molten globule intermediate.

Fluorescence methods have evolved into a powerful tool for studies of biological macromolecules (1,2). Intrinsic protein fluorescence, using tryptophan (Trp) as a reporter, provides a sensitive measure of protein tertiary structure and is widely used in protein folding studies (3)(4)(5). Most proteins, however, possess multiple Trp residues and the overall protein emission, naturally, yields only average information on the protein structure. To extract and evaluate the contribution of each reporter, and thereby track the conformational changes occurring in different parts of the macromolecule is a difficult challenge.
Site-directed mutagenesis facilitates these studies by targeting individual Trp residues. Clearly, the replacement of a Trp residue by another non-fluorescent amino acid may help to decompose the intrinsic protein fluorescence into the contributions of each reporter (6,7). A major concern, however, relates to the structural similarity of a natural protein and its designed analog. Indeed, some residues within the protein structure might be forced to occupy alternate, "tense" conformations (8). As a consequence, a point mutation can result in conformational changes either by adjustment of side chain spatial orientations (9), by shuffling conformational microstates toward a population which is less tense (10), or by long-distance structural repercussions on the backbone (11). The fluorescence of the remaining reporter could be affected by these induced changes, which, in any case, need to be evaluated. The present study addresses these issues, using apomyoglobin (apoMb) 1 proteins as examples.
Apomyoglobin, i.e. heme-free myoglobin, is a small, single domain ␣-helical protein, which has been intensively studied over the last few decades by a plethora of biochemical and biophysical techniques (12). The structure of native apoMb is very similar to that of native holomyoglobin (13)(14)(15) and the differences in linear dimension of the native apo-and holoforms are below 12% (16). Taken together, all the available information on (apo)myoglobins contribute to the widespread use of these proteins as model systems in protein folding studies. First, over 700 primary amino acid sequences reported for the myoglobin family provide a sound data base for studies relating protein sequence to chemical and physical properties (17). Second, mutagenesis of myoglobin is well established and allows one to make any requisite amino acid replacements (18 -21). Third, experimental conditions to populate a particular equilibrium state (i.e. native, molten globule, unfolded) of apoMb have been worked out (22) and biophysical characterization of these states has been reported (12).
Most mammalian myoglobins contain two Trp residues at invariant positions 7 and 14 in the amino acid sequence. Both residues reside on the A-helix in the three-dimensional folded structure (Fig. 1). Several attempts to separate the individual Trp contributions to the protein emission have been carried out, either using various quenchers and perturbing reagents (23,24), or tuna apoMb containing only Trp 14 (24 -26), or other apoMbs with various Trp locations (27,28). For native apoproteins it was observed that Trp 7 is more accessible to solvent * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. This paper is dedicated to Oleg B. Ptitsyn who initiated the research, provided the proteins, and was inspirational for early stages of the data analysis.
§ To whom correspondence should be addressed.  (23). This result agrees well with the large body of structural data on myoglobins (29 -31). The data relating to the buried Trp 14 are less clear, and the reported emission maxima cover a 10-nm range (23,24,32). Serious questions remain as to the contributions of each Trp residue to the intrinsic fluorescence. For instance, both a lack of Trp 7 emission (23, 24) and coequal contributions from both Trp residues (32) have been reported for native apoproteins. With respect to the other conformational states, a few experiments with benign modifications suggested that the emission of Trp 7 dominates in the molten globule state (pH 4.2), i.e. exhibits twice the amplitude of that of Trp 14 (23,24). As to the mutagenesis substituting the indolic residue with a phenylalanine, prior studies reported both a lack of structural changes (21,33) and perturbation of the native structure due to this replacement (20). Clearly, a better understanding of the factors that determine the Trp fluorescence in apoMbs is required to take full advantage of this information. We therefore initiated a study aimed at dissecting the individual Trp contributions using myoglobin variants with desired replacements. Natural proteins were sperm whale and horse myoglobins containing two Trp residues at positions 7 and 14, along with tuna myoglobin possessing only a single Trp 14 . Two specifically engineered proteins were sperm whale myoglobins with only Trp 7 or only Trp 14 and the other Trp residue replaced by phenylalanine. As a technique we employed time-resolved fluorescence spectroscopy utilizing the Trp residue as an intrinsic spectroscopic probe. Indeed, fluorescence resolved on the nanosecond scale provides information on both macromolecular structure and dynamics, including the fluctuation of the microenvironment of the reporter, rearrangement of the intra-and inter-macromolecular interactions, rotational diffusion of the chain segments, etc. Using tryptophan as a reporter offers an ideal means to monitor protein folding mainly due to the high sensitivity of the indole fluorescence to the polarity of the environment. Together with steady state fluorescence experiments and site-directed mutagenesis this allowed us to explore the structural changes occurring in different parts of the protein during equilibrium (un)folding, as well as highlight the effects of single amino acid replacement on the protein structure. Individual Trp contributions to the total apoMb emission are assessed and a correlation between the Trp fluorescence and structural features of the native, molten globule, and unfolded states of apoMb is discussed in the context of interactions between the individual Trp residues and neighboring residues in local structural clusters.

EXPERIMENTAL PROCEDURES
Myoglobins-Horse myoglobin (isolated from heart muscle) was purchased from Sigma. Tuna myoglobin (isolated from dark muscle) was prepared by the procedure as described in Ref. 34. A crude fraction of the sperm whale myoglobin (isolated from skeletal muscles) was a generous gift of Dr. G. B. Postnikova and Dr. R. I. Artyukh (Institute of Biological Physics, Pushchino). Protein obtained from an ammonium sulfate precipitate at 70 -95% saturation was sequentially purified by column chromatography on Sephadex G-75 and DEAE-52 resin in 10 mM Tris-HCl at pH 8.5, and, thereafter, on CM-52 cellulose in 10 mM Tris-HCl at pH 6.5. Mutated sperm whale myoglobins containing a single Trp residue with the other one replaced by phenylalanine were gifts of the Institute of Protein Research, Pushchino. Site-directed mutagenesis of sperm whale myoglobin was carried out as described previously (18 -21).
Apomyoglobins-The heme was removed by 2-butanone extraction (35) followed by gel chromatography on Sephadex G-25 (PD-10 columns, Amersham Pharmacia Biotech). The apoproteins were extensively dialyzed against water at 4°C and lyophilized for storage. Contamination of the apoprotein by myoglobin was assessed spectrophotometrically; no significant absorption was observed in the Soret region. Homogeneity of the apoMb was verified by SDS-polyacrylamide gel electrophoresis (36); a single band was observed for each sample. Protein concentration was estimated by absorbance at 280 nm with a Hewlett-Packard 8452A diode array spectrophotometer. The molar extinction at 280 nm was calculated from the tryptophan and tyrosine composition and the standard values of 5690 and 1280 M Ϫ1 cm Ϫ1 reported for those amino acids (37).
Chemicals and Solutions-All reagents were of reagent grade or better. Ultra-pure urea was purchased from ICN (Irvine, CA) and used without further purification. Experiments were carried out at room temperature in a 10 mM sodium acetate, 10 mM sodium phosphate buffer mixture, containing 30 mM NaCl. The protein concentration was 6 M unless stated. The native and molten globule states were studied at pH 6.5 and 4.2, respectively (22). The unfolded state was achieved by addition of 8 M urea to the native protein solution.
Steady State Fluorescence Measurements-Fluorescence emission spectra were recorded on a SPEX Fluorolog-2 spectrofluorometer (data interval of 0.5 nm, scan speed of 50 nm/min) supplied with DM-3000 software. Emission was measured under "magic angle" conditions in the ratio mode and corrected for the appropriate solvent blanks, as well as for wavelength-dependent bias of the optics and detection system. An excitation wavelength of 295 nm was used in all experiments to avoid tyrosine absorbance.
Time-resolved Fluorescence Measurements-Fluorescence intensity decay measurements were performed using the time-correlated singlephoton counting setup essentially as described in Ref. 38. Samples were excited at 295 nm using a synchronously purged, cavity dumped, frequency doubled dye laser (repetition rate of 4 MHz, pulse width of 5 ps, average UV power Ͻ200 W). The channel width was 85 ps, and the data were collected in 512 channels. The instrumental time resolution limited by the detector (R955 photomultiplier) had a transit time spread of 900 ps. This allows one to resolve correlation times as short as 150 ps. Some samples were re-measured with a microchannel plate photomultiplier with a resolution approaching 50 ps (total optical and electronic transit time spread of 120 ps). This improvement in resolution, however, led to no changes in the recovered parameters.
Fluorescence intensity decay surfaces were collected under magic angle conditions for equal dwell times and by stepping the emission monochromator in increments of 5 nm in the range of 300 -450 nm. One complete measurement included the fluorescence intensity decay of the reference compound, the sample, the background, and again the reference. Reproducibility of the experimental data was found to be satisfactory as judged by the results of duplicate experiments. In all experiments the reference lamp profile and color shift used for convolution analysis was tested with a monoexponential standard (melatonin aqueous solution with fl ϭ 5.2 ns). Finally, over 30 decays were obtained for each sample and further used for the global analysis. In particular, the fluorescence intensity decay was assumed to follow a multiexponential law: where the relative amplitudes, ␣ i , and the decay constants, i , were the numerical parameters to be recovered. Decay-associated spectra were obtained from a global fit of the decay profiles collected at different wavelengths with the common lifetimes i linked across the data set while optimizing the amplitudes ␣ i for each decay (39). To resolve the emission spectra associated with the individual decay constants, the fluorescence intensities at various wavelengths were expressed as ␣ products. The relative contribution of each decay component to the total emission, f i , was calculated as The convolution was compared with the experimental decay by nonlinear least-squares analysis. The best fit between the theoretical curve and the data was evaluated from the plot of weighted residuals, the autocorrelation function of the weighted residuals, and the reduced R 2 value. In all cases, a short-lived, fixed component was employed to compensate for any scattered excitation or color shift. The confidence limits for DAS have been explored previously (40); and for our experimental conditions, we anticipate that fluorescence parameters will be recovered within 10%.

RESULTS
Fluorescence intensity decays were recorded for each sample as a function of wavelength across the emission spectrum from 300 to 450 nm. Over 1000 decays were collected for a variety of conditions, which include native, molten globule and unfolded proteins, and variation of the solvent (e.g. addition of sucrose). Overall, the Trp fluorescence was described adequately with a two-exponential model for both one and two Trp proteins. The reduced R 2 value, characterizing the goodness of the fit, was always less than 1.5, and the weighted residuals and the autocorrelation of the residuals were randomly distributed around zero. Attempted fits of the experimental data to a one-exponential model showed a substantial (Ͼ30%) increase in R 2 value. A three-exponential fit led to no improvement in R 2 and generated the third term with rather short (ϳ0.5 ns) or large (Ͼ7 ns) lifetimes. Since the emission (␣) as well as the relative amplitude (␣) associated with this additional term were always comparable to the experimental accuracy, the simpler two-exponential model was chosen for further consideration. Parameters associated with the time-resolved fluorescence in apoMb natural variants and site-directed mutants are summarized in Table I. The decay-associated spectra resolved for two-and one-Trp proteins are shown in Figs. 2 and 3. To emphasize the difference in DAS generated for a particular sample, the spectra were normalized and displayed in the inset.
The blue, green, and red colors refer to the native, molten globule, and unfolded states, respectively. Figs. 2 and 3 demonstrate vividly that unfolded proteins (red curves) yielded two-component DAS, which exhibit similar positions of emission maxima ( max ϳ 347 nm) and spectral shape. Variation of solvent viscosity, i.e. the addition of sucrose up to 23 wt %, showed no significant effect on the time-resolved fluorescence as evidenced by the data presented in Figs. 2C. Hence, the complex fluorescence kinetics observed for Trp residues in the unfolded proteins cannot be explained by the effect of solvent relaxation alone (41). In fact, the strong similarity in DAS recovered for the entire set of samples hints at common features of the electronic relaxation of the indolic residue in the unfolded polypeptide chains.
In contrast, the DAS recorded for folded proteins exhibited noticeable differences, which might relate to the specific local interactions between the Trp residues and neighboring amino acids in a local structural cluster. Fig. 2, A and B, show the DAS of natural proteins containing two Trp residues, when measured in the native (blue curves) and molten globule (green curves) states. Considering native apoMb from sperm whale as an example ( Fig. 2A), one observes that the spectrum associated with the short-lived decay component ( fl ϭ 1.3 ns) consti-  from horse (Fig. 2B). In fact, the visual difference in DAS generated for these variants arises from only small deviations in the recovered decay parameters (Table I). However, slight variations in the folded structures of these proteins were previously observed by high-resolution diffraction methods (29). Importantly, the contribution of the short-lived component to the total decay ␣ short (see Equation 1) in both native proteins is about 80%, and the two-component DAS are separated by 5 nm. Furthermore, both proteins show similar changes in DAS when adopting the molten globule state. The spectrum of the new long-lived component ( fl Ϸ 5.5 ns) centered at 335 nm becomes dominant in emission and the apparent population of shortlived species is reduced to 50%. Fig. 3 displays the DAS generated for proteins containing only a single Trp residue. Altogether, DAS resolved for Trp 14 under native conditions are indistinguishable by spectral shape and the position of emission maximum at ϳ330 nm (Fig. 3, A  and C). In contrast, spectra of the Trp 7 are clearly separated by 5 nm with the maxima located at ϳ330 and ϳ335 nm (Fig. 3B). Furthermore, one-Trp proteins exhibit different changes in DAS when adopting the molten globule state. Specifically, no changes in DAS were observed for Trp 14 , indicating the absence of any dynamic changes in the local environments of this residue. The fluorescence of the Trp 7 , however, changes substantially and a new long-lived component ( fl ϭ 5.7 ns) becomes dominant in the emission of the molten globules. The spectra resolved for the Trp 7 in the molten globule state are still separated by 5 nm. Surprisingly, the folded proteins containing only Trp 14 show considerable differences in time-resolved fluorescence. For instance, the emission of the native apoMb from tuna (Fig. 3C) derives the largest contribution from the longlived decay component ( fl ϭ 5.2 ns), whereas the fluorescence of Trp 14 in the sperm whale mutant (Fig. 3A) is dominated by the short-lived component ( fl ϭ 1.3 ns). Both samples, however, exhibit DAS, which are indistinguishable by the positions of the emission maxima or spectral shape, and are not appreciably affected by the native-molten globule transition. DISCUSSION Apomyoglobins, either the natural variants or site-directed mutants, investigated in the present study display complex heterogeneity in Trp emission as evidenced by the two lifetimes required to describe the time-resolved fluorescence data. Given the complexity observed for one-Trp proteins, it appears that there is no simple way to associate a particular component of a fluorescence decay of a multi-Trp protein with a particular Trp residue, as suggested previously for apoMb (44). Indeed, each Trp residue in apoMb proteins reports on complex (non-exponential) electronic relaxation in the native, molten globule, or unfolded states.
The DAS of unfolded proteins reveal strong similarities in fluorescence kinetics. The two decay components with lifetimes of 1.6(Ϯ0.3) ns and 4.8(Ϯ0.2) ns contribute almost equally to the total decay and their spectra resemble each other closely. Since the fluorescence is affected predominantly by local interactions, it is instructive to inspect the primary amino acid sequences as to clues about the origin of the Trp emission heterogeneity in the unfolded chains. Table II shows that the local neighborhoods around the Trp residues in the different myoglobins are quite diverse and residues of different nature (aliphatic, polar, non-polar, proton donor, or acceptor) are found in equivalent positions. Overall, a 74% amino acid identity is observed between horse and sperm whale myoglobin for the 19 neighboring positions, while only 35% identity is present between sperm whale and tuna myoglobin. Some of the varying residues (e.g. lysine, histidine, and asparagine) are known to be potential quenchers of Trp fluorescence (45). Therefore, one explanation for the emission heterogeneity may include the multitude of Trp isomers or rotational isomers of nearby quenching residues. The similar DAS positions, however, point to a quasi-homogeneous environment of the apparent fluorescent species; thus, the interconversion between any possible Trp isomers or conformational substates of quenchers has to occur during the lifetime of the Trp fluorescence (ϳ3 ns). The molecular mechanism of this averaging in the unfolded chain might be a rotation or hopping of side chains from one isomeric well to another, which is known to occur on a ϳ5 ns scale (38). A substantial increase in the solvent viscosity, however, leads to no changes in the Trp fluorescence, indicating that conformational exchange has very little effect, if any at all, on the population of these possible isomers. In addition, very similar Trp decay parameters were recovered on numerous occasions for unfolded proteins of different nature (46). Altogether, these notions strongly suggest a common origin for the DAS in the unfolded polypeptide chains. An alternative explanation could be provided by the formation of "excited state complexes" (exciplex formation, electron or proton transfer, etc.), i.e. by the interaction of the excited indolic residue with backbone atoms or the solvent (46 -48). The kinetic scheme including reversible excited state reactions is known to result in a two-exponential analytical solution, and its numerical parameters (time constants and amplitudes) relate in a complex manner to the different rate constants involved (46,47). The short-lived component in the fluorescence decay of a fluorophore might be roughly associated with the quenching of the fluorescence due to complex formation, whereas the long-lived component reports on the delayed fluorescence following dissociation of the complex. This hypothetical mechanism is consistence with the independence of the Trp fluorescence on primary sequence and/or the intrachain micro-Brownian dynamics. It also agrees well with the independence of the fluorescence kinetics on emission wavelength (47). Furthermore, our preliminary analysis employing the apoMb decay data provides reasonable rate constants for the forward (complex formation) and reverse TABLE III Local structural cluster around tryptophan residues in myoglobins The contact matrix was calculated using the atomic coordinates for the following PDB structures: horse (1ymb), sperm whale (2myb) and tuna (1myt). Contacts were calculated for a sphere of 5 Å around the Trp residue. Similar clusters around Trp7 (dashed box) and Trp14 (solid box) were found in myoglobins from human (2mm1), asian elephant (1emy), pig (1myg), loggerhead sea turtle (1lhs) and common seal (1mbs). (complex dissociation) reactions. Perhaps, given the observed complexity, it is prudent to assume that both mechanisms, conformational heterogeneity and excited state reactions, could contribute to some extent to the complex relaxation of the Trp excited state in the unfolded polypeptide chains. To further address this issue experiments using variation of the excitation wavelength and/or addition of external quencher may be extremely useful (47). The folded proteins reveal noticeable differences in the timeresolved fluorescence, which most likely report on the diversity of the Trp local environments in the individual proteins. For instance, folded variants possessing only Trp 14 exhibit very similar changes in DAS during protein (un)folding, although their individual spectra are different and the long-lived component is dominant in tuna apoMb, in contrast to the sperm whale mutant. In an endeavor to understand the origin of those differences we inspected the NMR and crystal structures reported for the myoglobin family. Specifically, we calculated a matrix of van der Waals contacts of the Trp side chain with neighboring amino acids within a surrounding sphere of 5 Å. The residues in the local structural cluster around the indole ring (and the helix they are located in), as well as the number of atoms in contact and the closest distance are listed in Table  III. As can be appreciated, Trp 14 is surrounded by aliphatic (Leu 69 , Leu 72 , Leu 76 ), non-polar (Gly 73 ), polar (Met 131 or Met 115 ), and charged (Lys 77 ) residues, which reside on the E (residues 58 -79), G (residues 100 -119), and H (residues 124 -150) helices. Moreover, the local cluster around Trp 14 as well as that around Trp 7 (when the latter is present) was found to be invariant in mammalian myoglobins from human, asian elephant, horse, pig, loggerhead sea turtle, sperm whale, common seal and differs only slightly in myoglobins from other species (sea hair and tuna). An unusual case, however, occurs in tuna myoglobin where a proline residue is found at position 16 in the primary sequence (Table II). This disturbs the A helix (49), and another mutation, most likely needed to stabilize the native structure, relates to the substitution of an aliphatic residue at position 21 by the aromatic residue Tyr 21 . Importantly, Tyr 21 becomes a major contact residue for Trp 14 in tuna myoglobin (Table III). Furthermore, the x-ray structure of tuna myoglobin (31) reveals a hydrogen bond between Trp 14 and Tyr 21 (Fig. 4). Thus, the observed differences in DAS recovered for tuna apoMb and the sperm whale mutant containing only Trp 14 reflect the substantial differences in the specific interactions of Trp 14 with the neighboring amino acids. It may well be the case that the dominant long-lived component in the time-resolved fluorescence of tuna apoMb reports on the interactions of the Trp 14 with Tyr 21 , e.g. on the formation of the excited state complex. The latter provides a reasonable explanation for both invariant DAS and time constants ( i ) measured for those proteins, and different fractional amplitudes (␣ i ) of i components. Note that the valuable advantage of DAS derives from the assignment of a particular decay component to a spectral region. In this sense, the two decay components observed in the emission of tuna apoMb cannot be assigned to fully unfolded and folded states, as suggested in Ref. 25. Clearly, the emission spectra of folded and unfolded proteins are separated by ϳ20 nm, whereas the DAS of tuna apoMb either in the native or the molten globule state are clustered around 330 nm indicating that the apparent fluorescent species are both buried in a hydrophobic native-like environment.
A key question concerns the use of mutagenesis to decompose the protein fluorescence into the contributions of individual reporters; in particular it has to be established whether the structure of the wild-type protein is strictly retained in the The decay spectra were added in accord with the values of lifetimes, i.e. "short with short" and "long with long." The blue and green refer to the native and the molten globule states, respectively. The best agreement was achieved for the following compositions: 0.4W7F ϩ 0.6W14F (native state) and 0.3W7F ϩ 0.7W14F (molten globule state). The DAS of the one-Trp mutants are shown in Fig. 3, A and B. designed analog. Overall, the substitution of a Trp residue in apoMb with a phenylalanine has very little effect on the global fold as evidenced by near and far UV circular dichroism studies (20,21,33). In the present case, however, sperm whale apoMb and its one-Trp mutants exhibit discernible differences in steady state fluorescence. Each mutant gave an emission spectrum with higher fluorescence intensity than wild-type protein (data not shown). It appears that the point mutation Trp 3 Phe slightly affects the distribution of local conformations in a particular structural region involving the A helix in myoglobins. Specifically, since the Phe side chain is about 25% smaller than that of Trp, the local conformations around the "guest" can be adjusted, allowing for changes in the orientation of the A helix and/or an escape of the remaining Trp from a close amino acid quencher. Despite the above complications, valuable information regarding the contributions of individual reporters to the protein emission can be obtained by simulating the DAS of the wild-type protein as a summation of spectra of the one-Trp mutants. In fact, a substantial contribution of Trp 7 to the overall emission is evident from the 5-nm split of DAS generated for two-Trp proteins (Fig. 2). Note that the Trp 14 reveals the DAS, which are indistinguishable by position of the emission maximum (Fig. 3). Furthermore, we find that it is possible to simulate adequately the DAS of native apoMb by assuming almost equal contributions from each one-Trp mutant. In particular, within an accuracy of 20%, both the total emission spectrum and both decay spectra with time constants of 1.3 and 4.5 ns can be simulated (Fig. 5A). The observed discrepancy in amplitudes of the simulated and observed DAS indicates, however, that certain native interactions involving Trp residues are missing in the one-Trp mutants. Better agreement was obtained for the less rigid molten globules, assuming that Trp 7 exhibits almost twice the amplitude of the Trp 14 emission (Fig. 5B). This finding indicates that Trp 7 is more responsive to the conformational changes, which distinguish the molten globule and native states, than Trp 14 . At this juncture, some remarks as to previous studies on apoMbs can be made. It was suggested earlier that under alkaline conditions (pH 8 -9) the fluorescence of Trp 14 is 2-fold higher than that of Trp 7 , and on lowering the pH to 4 it exhibits a 2-fold decrease due to the dynamic quenching by the ionized residue His 119 (23,33,44). Inspection of the myoglobin structure, however, reveals that the distance between Trp 14 and His 119 is over 9 Å, which is rather large for effective dynamic quenching in the rigid native protein. Furthermore, His 12 and His 24 are located at similar distances of 7-10 Å, and could also be potential quench-ers of the Trp fluorescence. More troubling, the fluorescence of apoMb variants strongly depends on protein concentration as shown in Fig. 6. This finding corroborates our previous report (38) concerning the strong tendency of native apoMb toward aggregation. As to the partially folded forms (i.e. molten globule state), their propensity to aggregate is a well established fact (50). Therefore, it is very likely, that earlier studies of apoMb fluorescence were complicated and influenced by protein aggregation. For instance, this may explain the discrepancy in interpretation of data obtained in seemingly similar studies, both employing the Trp fluorescence (20,21).
In conclusion, based on the results presented here it may be legitimate to speculate to some degree about the formation of the native three-dimensional structure of apoMb during (un) folding. Structured intermediates, specifically, the molten globule state, are assumed to be involved in the folding of small proteins (51,52). Knowledge of the structural features of folded intermediates is of fundamental relevance for the understanding of folding pathways and mechanisms, which govern the overall protein architecture. As to the apoMb proteins, a combination of deuterium exchange and two-dimensional NMR experiments has revealed that the native A, G, and H helices are formed during the early stages of protein folding (53) and are relatively stable in the equilibrium molten globule state (13). Multi-dimensional NMR spectroscopy has yielded information on the backbone conformation of all residues in the molten globule state and allowed an estimation of the extent of helix formation (54). Recently, Förster energy transfer experiments have lent strong support to the hypothesis that A, G, and H helices in the molten globule state constitute a native-like subdomain (55,56). Furthermore, studies on the Trp nanosecond dynamics have demonstrated that both Trp residues in the molten globule state are involved in the formation of helical interfaces and their internal mobility is minimally affected by the "native-molten globule" transition (38). Questions as to the structure in other helical fragments are still under scrutiny. Since the intrinsic fluorescence as resolved on the nanosecond time scale provides information about conformational dynamics, it can be concluded that changes distinguishing the molten globule and native states of apoMb originate from conformational changes distant from Trp 14 . This is supported by the fact that the fluorescence decay parameters of Trp 14 as measured in molten globules are essentially identical to those in native state. Judging by the similarity of the dynamic fluorescence, none of the structural changes occurred in the microenvironment of Trp 14 . In other words, residues in the neighboring local cluster around Trp 14 retain their pivotal native-like positions in the structured molten globule intermediate. This finding suggests that the E helix in the molten globule state of apoMb most likely is located at a native-like distance to the AGH subdomain, exhibiting, however, a lack of rigid secondary structure (54). A similar conclusion can be drawn with respect to the beginning of the B helix, mainly based on the data obtained for tuna apoMb. Clearly, the lack of changes in the time-resolved fluorescence as observed for tuna apoMb indicates that native-like interactions between Trp 14 and Tyr 21 are barely affected by the native-molten globule transition. Altogether, this allows us to conclude that residues in the B and E helices assembling the helical interfaces in the native structure, most likely contribute to the stabilization of the overall structure of the molten globule intermediate. In contrast, the formation of the molten globule state has pronounced effects on the dynamic fluorescence of Trp 7 . The observed fluorescence enhancement and the native-like immobility of Trp 7 in the molten globule state (38) hint at the growing conformational fluctuations within the E-F turn. More specifically, these FIG. 6. pH dependence of the tryptophanyl fluorescence in horse apoMb as monitored for protein concentrations of 5 M (1) and 10 M (2). The excitation wavelength was 295 nm. The fluorescence intensity was integrated over the range of 300 -450 nm and normalized to that at pH 6. changes reflect the altered interactions between Trp 7 and Lys 79 , which is known to be a quencher of the Trp fluorescence (45). Thus, the fluorescence of Trp 7 in apoMb proteins might be a measure for the distance between the E-F turn and the beginning of helix A.
In summary, employing time-resolved fluorescence experiments on Trp residues together with site-directed mutagenesis allows one to probe local structural fluctuations in specified parts of the protein molecule and study correlated conformational effects in the protein matrix induced by the single amino acid replacement. More important, the approach presented here provides a means to extract novel structural information on equilibrium folding intermediates, which is difficult to recover by conventional means.