Flavoproteins : studies on flavodoxins and phenol hydroxylase

Flavoproteins play an important role in a variety of catalytic reactions. The chemistry underlying these reactions is quite different from case to case. The basis for this broad reaction spectrum is formed by the flavin. Free flavin is a versatile molecule, capable to undergo many different chemical reactions. The steering of a particular chemical reaction of a flavin in a flavoprotein results from the interaction with the apoprotein. The latter, therefore, determines the specificity of the reaction catalyzed by the flavoprotein. Although flavin is capable to undergo many different reactions, flavoproteins, mainly, can be classified in only four major groups. The flavoproteins studied in this thesis are flavodoxins and phenol hydroxylase. Flavodoxin is an electron transferring protein and therefore belongs to the class of the electrontransferases. Phenol hydroxylase belongs to the class of the mono-oxygenases, as it inserts one oxygen atom of O 2 into the substrate, whereas the other is reduced to water. Firstly, the results on the flavodoxins will be summarized.FlavodoxinsIn chapter 2 the flavodoxin, in its three redox-states, from Desulfovibrio vulgaris (Hildenborough) is investigated by homonuclear two-dimensional NMR techniques. The NMR results are compared to existing X-ray crystallographic data (Watt, et al., 1991). From NOE intensities and the chemical shift values of the flavodoxin spectra in the three redox- states it is concluded that outside the FMN binding site no structural changes occur upon reduction of the flavodoxin. This is in agreement with the X-ray crystallographic data (Watt, et al., 1991), because the only change in structure upon reduction of the flavodoxin was observed near the FMN: the conformation of Gly 61- Asp 62near the FMN N(5) changes upon reduction of the oxidized flavodoxin. The carbonyl oxygen of Gly 61points away from the N(5) group of the FMN in the oxidized state, whereas it points towards the N(5)H in the reduced states, forming a hydrogen bond with the N(5) proton. Although it is impossible from the current NMR data to determine the exact nature of the conformational change of the Gly 61- Asp 62peptide bond in the solution structure, the data are in agreement with the changes observed in the X-ray structures. A similiar change is observed for the flavodoxin from Clostridium MP (Smith, et al., 1977), therefore, the change near the FMN N(5) seems to be important for the function of flavodoxins. As judged from the high chemical shift values of some resonances near the FMN phosphate group, a hydrogen-bonding scheme towards the phosphate is determined. This scheme very much resembles the hydrogen bonds observed in the X-ray structure, except that the amide proton of Gly 13does not seem to be involved in hydrogen bond formation towards the phosphate, whereas the amide proton of Ser 10does not seem to form a hydrogen bond to the phosphate as determined from the X-ray structures. This may mean that the loop which is responsible for binding the FMN phosphate to the flavodoxin is in a different conformation in solution than in the X-ray structure.Upon two-electron reduction of the oxidized flavodoxin the FMN N(1)-C(2) region becomes negatively charged. The resonances of the amide protons of Asp 95and Cys 102show a large down field shift upon two electron reduction. This shift is induced by the negative charge on the N(1)-C(2) region, which forms hydrogen bonds with Asp 95and Cys 102.Upon reduction of the flavodoxin the FMN N(5) becomes protonated. Due to this protonation the N(5) nitrogen changes from a sp 2-type to a sp 3-type nitrogen. This means that the central pyrazine ring is expected to be less aromatic in the reduced state. This is indeed the case, because the ring current effect on the Cαproton of Trp 60, which is located above the pyrazine ring of the FMN, is less pronounced in the two-electron reduced state. Thus less electron density is expected on the pyrazine moiety of the FMN. Whether this is important for the function of the flavodoxin remains unclear.In chapter 3 the flavodoxin from Azotobacter chroococcum is investigated by heteronuclear multi-dimensional NMR spectroscopy. This flavodoxin is involved in the electron transfer to the Fe-protein of the nitrogenase enzyme complex (Yates, 1972). In contrast with the flavodoxin from D. vulgaris this flavodoxin belongs to the long-chain flavodoxins. The complete backbone assignment of the 1H, 13C and 15N resonances of the oxidized A. chroococcum flavodoxin is given in chapter 3. For this assignment the gradient enhanced versions of the CBCANH (Grzesiek and Bax, 1992b) and the CBCA(CO)NH (Grzesiek and Bax, 1992a) experiments were essential. How the enhancement was incorporated in the original experiments is also explained in chapter 3. From the NMR data the secondary structure elements could be determined. The secondary structure of the A. chroococcum flavodoxin consists of a five stranded parallel β-sheet and five α-helices. The outer strands of the β-sheet show some deviation from a regular extended conformation. The outer strand β4/β6 is interrupted by a loop region. This extra loop is typical for the long-chain flavodoxins. It is proposed in this chapter that the extra negatively charged amino acid residues in this loop contribute to the very low redox-potential found for this flavodoxin. From a titration with the Fe-protein of the nitrogenase complex, it is concluded that a small loop near the FMN binding site, that is not present in most other flavodoxins, is important for the complexation with the Fe-protein. This loop, Gly 65-Glu 71, thus may be important for electron transfer to the nitrogenase enzyme complex. In chapter 3 it is shown that electrostatic interactions are important for the complex formation. Also it is found that MgADP influences this complexation, probably caused by a change in conformation of the Fe-protein upon ADP binding. One of the helices from the flavodoxin, helix α1, consists of many positively charged residues. Such a positively charged helix is generally not observed in other flavodoxins. Besides possible stabilisation of the negatively charged FMN phosphate, this helix may be important for the interaction with the electron donor of the flavodoxin. Future studies will have to determine the function of this helix.Phenol hydroxylaseFor the conversion of phenol to catechol a whole set of successive reactions has to be performed by phenol hydroxylase (Scheme 1.2) (Maeda-Yorita and Massey, 1993). The step in which an oxygen atom actually is inserted in the substrate is the attack of the peroxide function of the C(4a)-hydroperoxyflavin intermediate (intermediate I, Scheme 1.2) on the substrate. This attack is believed to proceed by an electrophilic attack of this peroxide function on the substrate. It is found by Maeda-Yorita and Massey (1993), at least for resorcinol as a substrate, that this step is one of the slowest in the reaction cycle of phenol hydroxylase. It is therefore that this step is expected to be important in determining the overall reaction rate. Bearing this in mind, the conversion of a series of phenolic substrates was analysed in combination with theoretically calculated parameters, using semi-empirical methods.Chapter 4 shows that the ortho -hydroxylation of 3-fluorophenol by phenol hydroxylase is regioselective. This means that the two ortho positions, C2 and C6, of the phenol are not equally hydroxylated. This regioselectivity is observed to be pH dependent; the C6/C2 ortho -hydroxylation ratio decreases with increasing pH. At pH values below 5.5 this ratio has a maximum of 6.7, whereas a minimum of 2.2 is reached for pH values above 7.5. A second observation is that the rate of ortho -hydroxylation increases with increasing pH. The pH dependency observed for both effects, has a p K a value of 6.5. In order to get insight in the basis of this pHdependent regioselectivity and rate of the ortho -hydroxylation, binding studies using 19F-NMR and molecular orbital calculations are performed. Binding of 3-fluorophenol to oxidized phenol hydroxylase clearly showed a 19F-NMR resonance for the bound substrate. This broad, inhomogeneous resonance shifted to higher field values upon reducing the enzyme in the absence of molecular oxygen. Whether or not the 19F-NMR resonance in the reduced enzyme also has an inhomogeneous lineshape could not be detected due to partial overlap with the resonance of the substrate free in solution. When the pH is increased a similar shift to higher field is observed for 3- fluorophenol free in solution. Molecular orbital calculations showed that the electron density in the HOMO of the reaction center C6 is about seven times higher than of C2 for 3- fluorophenol, whereas equal HOMO density is calculated for 3-fluorophenolate. In combination with the high field shift of the 19F resonance of the bound substrate upon enzyme reduction, it was assumed that deprotonation of the phenolic substrate causes the observed regioselectivity. The regioselective ortho -hydroxylation could be caused by different orientations of 3-fluorophenol in the active site, leading to either C6 or C2 hydroxylation. A change in the relative contribution of discrete orientations with changing pH could then explain the observed decrease in C6/C2 hydroxylation ratio with increasing pH. Due to the broad nature of the 19F resonance of the 3-fluorophenol bound to reduced phenol hydroxylase, it is not possible to exclude this possibility. Binding studies of 4-fluorophenol, however, indicate to deprotonation of the phenolic substrate. At pH 6.2 in the reduced phenol hydroxylase a single broad 19F resonance of the bound 4-fluorophenol is observed. In the oxidized phenol hydroxylase the resonance of the bound 4-fluorophenol resonates at the same frequency. If the pH is increased a second resonance, which is shifted to higher field, is observed for the 4-fluorophenol bound to reduced phenol hydroxylase. A high field shift can be explained by the depronation of the substrate bound to the active site. Depronation in combination with frontier orbital theory also provides an explanation for the increase hydroxylation rate with increasing pH. Molecular orbital calculations calculate a less negative energy of the electrons in the HOMO of 3-fluorophenolate than of 3-fluorophenol. Transition-state theory in combination with frontier orbital theory (paragraph 1.3) explains that a substrate with a less negative E(HOMO) is expected to have a higher rate of conversion, assuming that the E(LUMO) of the C(4a)-hydroperoxyflavin does not vary for the different substrates. Based on these results a hypothesis is put forward to explain the pH-dependent effects observed. This hypothesis states that an active site amino acid residue, which acts like a base with a p K a of 6.5, is able to (partial) depronate the hydroxyl group of the phenolic substrate. Upon depronation of the substrate the rate of conversion is increased.In chapter 5 it is indeed observed that the energy of the electrons in the HOMO of the substrates plays an important role in determining the overall rate of their conversion. This is because the ln k cat for the conversion of a series of phenolic substrates correlates with their E(HOMO)'s. The presence of such a correlation strengthens the idea that the electrophilic attack of the C(4a)-hydroperoxyflavin of phenolhydroxylase on the substrate is a major factor in determining the overall rate of catalysis. However, since the correlation is 0.85, it is concluded that additional factors contribute to the fine-tuning of the overall rate of conversion. Future research should reveal which factors.For 3-fluorophenol it is observed that the ortho position with the highest electron density of the HOMO electrons is preferentially hydroxylated (chapter 4). For asymmetric phenolic substrates with similar ortho substituents, i.e. either two hydrogen or two fluorine substituents, it is also observed that the ortho position with the highest density of the HOMO electrons is preferentially hydroxylated. This indicates that chemical reactivity of the two ortho positions for an electrophilic attack of the C(4a)-hydroperoxyflavin intermediate is a factor influencing the regioselectivity of the phenol-hydroxylase-catalysed hydroxylation. However, comparison of the regioselectivity observed to the one predicted on the basis of the calculated chemical reactivity of the two ortho sites in a partially deprotonated phenol, demonstrates that the presence of a meta halogen substituent adjacent to the ortho position decreases the possibilities for hydroxylation of this ortho position. Local dipole- dipole interactions of the meta halogen substituent of the substrate with the enzyme active site, probably cause one of the two binding modes, i.e. C2 or C6 closer to the reactive hydroperoxide group of the C(4a)-hydroperoxyflavin, to occur more likely on a time averaged base. For most phenols with both a hydrogen and a fluorine ortho substituent, the hydrogen substituted ortho position becomes preferentially hydroxylated. However, the fact that 2,5-difluorophenol becomes preferentially hydroxylated at the fluorine substituted ortho position, indicates that probably also an ortho fluorine has an effect on the substrate orientation. Thus besides the difference in chemical reactivity on the ortho positions, also dipolar effects, which orientate the substrate in the active site, seem to play a role in the regioselectivity of the hydroxylation by phenol hydroxylase