Active-site dynamics of flavodoxins.

Until recently the conformational analysis of biomolecular structures was based on experiments performed with solid phase samples (X-ray diffraction, Fourier Transform Infrared spectroscopy, etc.). However, the development of techniques like nuclear magnetic resonance and time-resolved polarized fluorescence spectroscopy enables us to examine structure and dynamics of biomolecules in solution. Furthermore, molecular dynamics calculations of biomolecules in solvent environment can be used to examine processes which are difficult to approach experimentally.This thesis deals with the dynamic properties of the active site of several flavodoxins. These proteins transfer electrons from one redox protein to another, whereby flavodoxin shuttles between fully reduced (hydroquinone) and semiquinone states. The mechanism of electron transfer is not yet elucidated, but it has become clearer in the last few years that both static and dynamic properties of the proteins influence the electron transfer rates.In Chapter 2 the reduced flavin fluorescence characteristics of different flavodoxins are shown: Desulfovibrio gigas, Desulfovibrio vulgaris, Clostridium beijerinckii, and Megasphaera elsdenii flavodoxin. The results are compared with those of reduced flavin mononucleotide in solution. The fluorescence decays are analyzed both as a sum of exponentials with classical iterative nonlinear least- squares reconvolution and using the inverted Laplace transform accomplished by the maximum entropy method. These two approaches yield highly similar results. The fluorescence decay is always composed of three lifetime distributions, where the shortest fluorescence lifetime is influenced by the protein environment. The results indicate that the active site of Desulfovibrio flavodoxins resembles that of reduced flavin in solution more than in the other two flavodoxins. Based on the results of the fluorescence decays of reduced flavodoxins, the proteins can be classified into two groups: the rubrum -like flavodoxins (the two Desulfovibrio proteins) and the pasteurianum -like flavodoxins (the other two proteins). It is suggested that the position of the shortest lifetime can be used as a marker for the environment of the reduced flavin. Analysis of the fluorescence anisotropy decays shows that the proteinbound reduced flavin is immobilized within the apoprotein matrix. The rotational correlation time of the rubrum class of flavodoxins is somewhat longer than for pasteurianum flavodoxins which is explained by a small difference in molecular weight and volume. The dynamic properties of reduced flavin bound in flavodoxin indicate that a specific fixed rotation of the flavin may be required for electron transfer between flavodoxin and other redox proteins.Chapter 3 describes the fluorescence intensity decay of oxidized clostridial flavodoxin as analyzed using the maximum entropy method. Two short lifetimes (30 ps and 0.5 ns) are demonstrated which are not present in the fluorescence decay of oxidized flavin mononucleotide in solution. The third fluorescence lifetime (4.8 ns) coincides with the single lifetime found in free oxidized flavin. The fluorescence lifetime distribution is highly temperature dependent. At higher temperatures (more than 20 °C) the contribution of the 4.8 ns lifetime component increases dramatically. This behaviour can be explained by a small shift in equilibrium from protein-bound (low fluorescence quantum yield) to dissociated oxidized flavin (high quantum yield). This is confirmed by adding stoechiometric amounts of apoflavodoxin to the flavodoxin sample. The equilibrium then shifts to protein-bound flavin, even at elevated temperatures (more than 30 °C). The dissociation constant is determined, K _{<font size="-2">D</font>} = 2.61*10 <sup><font size="-2">-10</font></SUP>M <sup><font size="-2">-1</font></SUP>(at 20 °C). Addition of an excess of apoflavodoxin did not result in the complete disappearance of the 4.8 ns fluorescence lifetime. Collisional quenching experiments show that this flavin moiety in the flavodoxin fluorescence decay is solvent accessible, since it is significantly quenched by iodide. Associative analysis of the fluorescence and fluorescence anisotropy decays shows that the shortest fluorescence lifetime is probably coupled to a rotational correlation time similar to overall protein tumbling. However, the lifetime is far too short to contain accurate information on relatively slow protein tumbling. A part of the flavodoxin solution seems to be composed of dissociated flavin since the 4.8 ns lifetime can be linked to the rotational correlation time of free flavin in solution. This fluorescence lifetime, however, is also partly coupled to a correlation time slightly longer than that of free FMN. This moiety might then be intermediate between dissociated flavin and flavin which is loosely bound to the protein. Since its fluorescence lifetime is similar to that of free flavin this indicates that the flavin is probably bound to the protein via its ribityl-phosphate sidechain. This is in accordance with studies in which it is demonstrated that the phosphate group is essential for flavin binding to the apoprotein of flavodoxin (Barman and Tollin, Biochemistry 11 (1972) 4746-4754). The correlation time is longer than for free flavin as a result of sterical hindrance.The molecular dynamics of oxidized and fully reduced Cl. beijerinckii flavodoxin are calculated and described in Chapter 4. From the simulations of the flavodoxins in a solvent environment it is clear that the conformation of the active site of the flavodoxin in solution is very similar to the three-dimensional structure as determined by X-ray diffraction studies. However, some differences were found in the backbone conformation of the loop composed of residues 57 to 63. This loop is located near the flavin chromophore. Another difference is the direction of the plane through the reduced isoalloxazine ring system which is tilted about 11° as compared to its initial structure. This might be explained by the fact that the initial structure of semiquinone flavodoxin was used. Another explanation could be the choice of the distribution of the negative charge over flavin atoms N(1) and O(2). On the time scale of the calculations (hundreds of picoseconds), the oxidized as well as reduced flavin is immobilized within the protein matrix. This is in accordance with the experimentally determined initial fluorescence anisotropies. Comparison with earlier results of MD simulations in vacuo demonstrated that solvent inclusion is necessary for a correct description of the motional behaviour of the flavin. From the calculations the timecorrelation functions of the three tryptophan sidechains were determined. Residues Trp6 and Trp95 are almost immobilized in both oxidation states, whereas residue Trp90 has a fast decaying time-correlation function, indicating a large motional freedom. This flexibility, present in both oxidation states, is most pronounced in reduced flavodoxin. A possible role for this flexibility in electron transfer is suggested. Based on the coplanarity of the flavin isoalloxazine ring and the indole sidechain of Trp90 in the initial crystal structures, it might be extracted that the ring systems have correlated motion. However, from the MD calculations it is demonstrated that hardly any correlation in positional fluctuations of both ring systems exists.The tryptophan fluorescence decay kinetics of apo- and holo-flavodoxins (from D. gigas, D. vulgaris, Cl. beijerinckii, and M.elsdenii) are described in Chapters 5 and 6. The fast decaying fluorescence in the holoflavodoxins can be explained by energy transfer from the excited state of the tryptophan to the flavin acceptor. Energy transfer rate constants could be calculated from the known three- dimensional structure of D. vulgaris flavodoxin, which are present in the analysis of the fluorescence decays (see additional results summarized in the Appendix). Based on the steady-state and time-resolved characteristics it is suggested that the remote tryptophan (at about 20 Å from the center of the flavin) is preserved in both Desulfovibrio flavodoxins. Quantification of the fluorescence of flavodoxins with more than one tryptophan residues is extremely complex (tryptophan in aqueous solution already exhibits multi-exponential fluorescence decay). It is noted that a straightforward comparison of the fluorescence decays of apo- and holo-flavodoxins is difficult, since removal of the flavin chromophore results in a changed microenvironment of the tryptophan residues. However, strong indications have been found that in holoflavodoxin samples, always contain a certain amount of apoflavodoxin.In Chapter 7 the rotational correlation times of natural flavin compounds as well as some flavin models are reported. These time-resolved polarized fluorescence experiments were performed to examine the rotational characteristics of flavins in solution. This information can be used for comparison with protein-bound flavins (e.g., in flavodoxin (see Chapter 3)). An extra methyl group at flavin N(3) resulted in a few picoseconds longer correlation, in agreement with the slightly larger molecular volume and illustrating the high time-resolution of the picosecond laser fluorescence system used. Analyzing the decays after excitation in the first and second excited states demonstrated anisotropic motion of the lumiflavin. The angle between the long axis of the isoalloxazine ring system and the emission transition moment is approximately 22°, in fair agreement with the value obtained by Bastiaens et al. (Biophys. J. 63 (1992) 839-853)