Search

IntroductionCellular communication is partly mediated through the modulation of protein activity, structure and dynamics by lipids. In contrast to the biochemical aspects of lipid signalling, relatively little is known about the physical properties of the "signal" lipids (lipids involved in cellular signalling) in membranes and their interaction with membrane proteins. Knowledge about these properties contributes to the understanding of the molecular mechanism of intracellular communication. The main objective of this thesis is to evaluate organisational and motional aspects of protein-lipid systems that play a role in this signalling. Fluorescence spectroscopy was employed to distinguish between interacting elements of the protein-lipid binding equilibrium from the non-interacting ones and to reveal their motional properties and characteristics in organisation. The first chapter of this thesis summarises the functional relevance and fundamental mechanisms of protein-lipid interactions and introduces some fluorescopic tools which can be applied in these investigations. In the remaining chapters the biophysical properties of the phosphoinositides phosphatidylinositol (PI), phosphatidylinositol- 4-phosphate (PIP) and phosphatidylinositol-4, 5-biphosphate (PIP 2 ) and diacylglycerol (DG) and their relation to protein kinase C (PKC) and Band 3 (an abundant anion exchanging protein in erythrocyte membranes) form a central theme. Three types of binding are evaluated: 1) peripheral membrane binding of proteins, 2) lipids replacing other lipids at the protein surface and 3) lipids which bind to vacant cofactor sites at the protein. By combining complementary fluorescopic techniques, and by refinement of fluorescence data analysis, new aspects of protein-lipid relations have been elucidated which are summarised below. As a model study and introduction for the protein-lipid studies described in this thesis, Chapter 2 considers the association of lysozyme with acidic vesicles. From the experiments it could be concluded that the protein conformation and dynamics of lysozyme are effected when the protein interacts with acidic membranes. Apart from the general interest in protein-lipid interactions, this study demonstrates that a fractional analysis of time-resolved fluorescence and fluorescence anisotropy decay curves can provide accurate binding curves of protein molecules to lipid interfaces.The interaction of PKC with PS and phosphoinositides and DGVarious aspects of the application of fluorescence anisotropy in the evaluation of lipid motion are introduced in Chapter 3. In that Chapter the motional characteristics of three 1, 6- diphenyl-1, 3, 5-hexatriene (DPH) labelled phosphatidylcholines are compared.Based on the results of this study the effect of PKC on the reorientational properties of fluorescent analogues of DG and PC have been evaluated using a constrained model of analysis (Chapter 4). It appeared that in membranes DPH-DG experiences a larger motional freedom than DPH-PC, which indicates that DG induces a lipid packing irregularity. Defects in membrane organisation lead to a partial exposure of hydrophobic regions of phospholipids to aqueous environment and thereby increase the ground state energy of bilayers (Jain & Zakim, 1987). Therefore the activation energy associated with insertion of PKC into the hydrophobic membrane core could be reduced by DG. Three complementary observations provide evidence that PKC is able to interact with PS containing membranes in the absence of calcium:1) The isotropic rotation of PS containing micelles slows down upon addition of PKC (Chapter 4).2) Micelles and vesicles containing pyrene labelled lipids quench the tryptophan fluorescence of PKC (Chapter 5).3) PKC significantly reduces the collision frequency of pyrene labelled PS (Chapter 5).Addition of calcium results in the binding of PKC to DG (Chapter 6). For this DG-PKC interaction the presence of PS is a prerequisite (Chapters 4, 5, 6). The precise mechanism of this calcium dependence of DG binding remains unclear. Calcium enhances the binding of PKC to PS. As a result of this enhanced PS binding the probability of finding a DG molecule close to PKC may increase since PKC molecules retain longer at the membrane surface. Alternatively, calcium may induce changes in PKC that result in exposure of a shielded lipid cofactor site for DG. As a consequence of the interaction with PKC, the lateral and rotational diffusion and orientational freedom (Chapters 4 and 5) of labelled DG are reduced. The interaction with DG is highly specific, since the motional properties of labelled PC, PS and PI analogues are only moderately effected by PKC (Chapters 4 and 5) and these lipids are not able to quench PKC fluorescence at low concentrations (Chapter 6). However, labelled phosphoinositides PIP and PIP 2 partly mimic the properties of DG. They bind with high affinity to PKC and the acyl chain dynamics of (dipyr)PIP and (dipyr)DG are equally effected by PKC (Chapters 5 and 6). In addition, like is known for DG, only a very limited number of phosphoinositides bind to one PKC molecule (Chapter 6). Double labelling experiments and replacement studies suggest that PIP and DG can bind simultaneously to PKC while PIP 2 and DG cannot (Chapters 5 and 6). In addition, it appeared that only PIP 2 and DG effectively activate PKC. Combined, these results indicate a cofactor role of PIP 2 for PKC.Although the role of PIP 2 in the activation of PKC has to be established in vivo , its properties do fit excellently to a messenger function. Radioactive labelling studies with H-inositol and P-phosphate revealed that the turnover of PIP 2 is extremely rapid and in resting cells its level is low (reviewed by Hokin, 1985). The major part of PIP 2 present at resting conditions, may even be not available for PKC activation. Several studies have shown that the inositol lipids are heterogeneously distributed in the different cellular membranes and that only a small fraction of the total inositol lipid content is available for enzymes like phospholipase C (see for a review Downes & Michell, 1985). Like most other signalling lipids the PIP 2 concentration increases transiently to high levels within seconds after hormone stimulation of cells (Hansen et al., 1986). The dual role of PIP 2 as a DG-precursor and as a direct PKC-activator opens new avenues for the differentiation of signals via the various PKC families. The cell responses that induce the PLC catalysed hydrolysis of phosphoinositides always result in the co-appearance of DG and calcium. Several PKC-isoforms, however, do no need calcium for their activity. Other lipid cofactors like PIP 2 may function as cofactors for these calcium independent protein kinases. For the calcium dependent protein kinases an inositol triphosphate (IP 3 ) mediated calcium release would imply the production of DG.When calcium is provided via calcium channels in the plasma membrane or via sphingosine phosphate signalling (Dessai et al., 1992), no DG is released and other lipid activators like PIP 2 can take over the cofactor role. In this DG-independent activation, the temporal availability of calcium in the cytosol is the main switch for PKC activity. In this respect, it would be interesting to know if there are situations in the cell in which the PIP-kinase and (phosphoinositide-specific) phospholipase C activities are uncoupled. In that case the levels of PIP 2 can be regulated independently from those of DG, which would make the double role of PIP 2 in the regulation of PKC activity even more attractive (Figure 1).Heterogeneous distribution of phosphoinositides in membranesIn Chapter 7, the interaction of the phosphoinositides with human erythrocyte Band 3 protein was characterised using resonance energy transfer. The efficiency of quenching of the fluorescence from tryptophan residues in the protein by pyrene lipids in the membrane decreased in the order pyrPIP 2 , pyrPIP, pyrPI and pyrPC indicating different spatial distributions of these lipids with respect to Band 3. Global analysis of the quenched tryptophan fluorescence decays yielded indices for the spatial distributions of the different lipids. It appeared that the charged lipids PIP 2 and PIP spend longer time in the vicinity of Band 3 than PI and PC. In investigations complementary to these protein fluorescence observations, the monomer/excimer fluorescence of pyrene labelled phosphoinositides in DOPC membranes was evaluated in the absence and presence of Band 3 protein (Chapter 8). In combination with theoretical refinements of the model for lipid migration, these experiments revealed estimates for the lateral dynamics and lipid-lipid repulsion in membranes. In addition, the relative binding constants and the minimum number of Band 3 sites possessing affinity for the phosphoinositides could be roughly estimated. Consistent with intuitive expectations the lipid-lipid repulsion appeared highly dependent on the amount of negative charge of the lipid headgroups. Combining these experiments indicates that lipid-lipid repulsion and the Band 3-lipid interactions induce a non- random distribution of the Phosphoinositides. This non-random distribution will probably effect the accessibility of the inositol lipids for phospholipase C and for other phosphoinositide dependent proteins like PKC and may result in the functional heterogeneity of the phosphoinositides (Downes & Michell, 1985).

IntroductionCellular communication is partly mediated through the modulation of protein activity, structure and dynamics by lipids. In contrast to the biochemical aspects of lipid signalling, relatively little is known about the physical properties of the "signal" lipids (lipids involved in cellular signalling) in membranes and their interaction with membrane proteins. Knowledge about these properties contributes to the understanding of the molecular mechanism of intracellular communication. The main objective of this thesis is to evaluate organisational and motional aspects of protein-lipid systems that play a role in this signalling. Fluorescence spectroscopy was employed to distinguish between interacting elements of the protein-lipid binding equilibrium from the non-interacting ones and to reveal their motional properties and characteristics in organisation. The first chapter of this thesis summarises the functional relevance and fundamental mechanisms of protein-lipid interactions and introduces some fluorescopic tools which can be applied in these investigations. In the remaining chapters the biophysical properties of the phosphoinositides phosphatidylinositol (PI), phosphatidylinositol- 4-phosphate (PIP) and phosphatidylinositol-4, 5-biphosphate (PIP 2 ) and diacylglycerol (DG) and their relation to protein kinase C (PKC) and Band 3 (an abundant anion exchanging protein in erythrocyte membranes) form a central theme. Three types of binding are evaluated: 1) peripheral membrane binding of proteins, 2) lipids replacing other lipids at the protein surface and 3) lipids which bind to vacant cofactor sites at the protein. By combining complementary fluorescopic techniques, and by refinement of fluorescence data analysis, new aspects of protein-lipid relations have been elucidated which are summarised below. As a model study and introduction for the protein-lipid studies described in this thesis, Chapter 2 considers the association of lysozyme with acidic vesicles. From the experiments it could be concluded that the protein conformation and dynamics of lysozyme are effected when the protein interacts with acidic membranes. Apart from the general interest in protein-lipid interactions, this study demonstrates that a fractional analysis of time-resolved fluorescence and fluorescence anisotropy decay curves can provide accurate binding curves of protein molecules to lipid interfaces.The interaction of PKC with PS and phosphoinositides and DGVarious aspects of the application of fluorescence anisotropy in the evaluation of lipid motion are introduced in Chapter 3. In that Chapter the motional characteristics of three 1, 6- diphenyl-1, 3, 5-hexatriene (DPH) labelled phosphatidylcholines are compared.Based on the results of this study the effect of PKC on the reorientational properties of fluorescent analogues of DG and PC have been evaluated using a constrained model of analysis (Chapter 4). It appeared that in membranes DPH-DG experiences a larger motional freedom than DPH-PC, which indicates that DG induces a lipid packing irregularity. Defects in membrane organisation lead to a partial exposure of hydrophobic regions of phospholipids to aqueous environment and thereby increase the ground state energy of bilayers (Jain & Zakim, 1987). Therefore the activation energy associated with insertion of PKC into the hydrophobic membrane core could be reduced by DG. Three complementary observations provide evidence that PKC is able to interact with PS containing membranes in the absence of calcium:1) The isotropic rotation of PS containing micelles slows down upon addition of PKC (Chapter 4).2) Micelles and vesicles containing pyrene labelled lipids quench the tryptophan fluorescence of PKC (Chapter 5).3) PKC significantly reduces the collision frequency of pyrene labelled PS (Chapter 5).Addition of calcium results in the binding of PKC to DG (Chapter 6). For this DG-PKC interaction the presence of PS is a prerequisite (Chapters 4, 5, 6). The precise mechanism of this calcium dependence of DG binding remains unclear. Calcium enhances the binding of PKC to PS. As a result of this enhanced PS binding the probability of finding a DG molecule close to PKC may increase since PKC molecules retain longer at the membrane surface. Alternatively, calcium may induce changes in PKC that result in exposure of a shielded lipid cofactor site for DG. As a consequence of the interaction with PKC, the lateral and rotational diffusion and orientational freedom (Chapters 4 and 5) of labelled DG are reduced. The interaction with DG is highly specific, since the motional properties of labelled PC, PS and PI analogues are only moderately effected by PKC (Chapters 4 and 5) and these lipids are not able to quench PKC fluorescence at low concentrations (Chapter 6). However, labelled phosphoinositides PIP and PIP 2 partly mimic the properties of DG. They bind with high affinity to PKC and the acyl chain dynamics of (dipyr)PIP and (dipyr)DG are equally effected by PKC (Chapters 5 and 6). In addition, like is known for DG, only a very limited number of phosphoinositides bind to one PKC molecule (Chapter 6). Double labelling experiments and replacement studies suggest that PIP and DG can bind simultaneously to PKC while PIP 2 and DG cannot (Chapters 5 and 6). In addition, it appeared that only PIP 2 and DG effectively activate PKC. Combined, these results indicate a cofactor role of PIP 2 for PKC.Although the role of PIP 2 in the activation of PKC has to be established in vivo , its properties do fit excellently to a messenger function. Radioactive labelling studies with H-inositol and P-phosphate revealed that the turnover of PIP 2 is extremely rapid and in resting cells its level is low (reviewed by Hokin, 1985). The major part of PIP 2 present at resting conditions, may even be not available for PKC activation. Several studies have shown that the inositol lipids are heterogeneously distributed in the different cellular membranes and that only a small fraction of the total inositol lipid content is available for enzymes like phospholipase C (see for a review Downes & Michell, 1985). Like most other signalling lipids the PIP 2 concentration increases transiently to high levels within seconds after hormone stimulation of cells (Hansen et al., 1986). The dual role of PIP 2 as a DG-precursor and as a direct PKC-activator opens new avenues for the differentiation of signals via the various PKC families. The cell responses that induce the PLC catalysed hydrolysis of phosphoinositides always result in the co-appearance of DG and calcium. Several PKC-isoforms, however, do no need calcium for their activity. Other lipid cofactors like PIP 2 may function as cofactors for these calcium independent protein kinases. For the calcium dependent protein kinases an inositol triphosphate (IP 3 ) mediated calcium release would imply the production of DG.When calcium is provided via calcium channels in the plasma membrane or via sphingosine phosphate signalling (Dessai et al., 1992), no DG is released and other lipid activators like PIP 2 can take over the cofactor role. In this DG-independent activation, the temporal availability of calcium in the cytosol is the main switch for PKC activity. In this respect, it would be interesting to know if there are situations in the cell in which the PIP-kinase and (phosphoinositide-specific) phospholipase C activities are uncoupled. In that case the levels of PIP 2 can be regulated independently from those of DG, which would make the double role of PIP 2 in the regulation of PKC activity even more attractive (Figure 1).Heterogeneous distribution of phosphoinositides in membranesIn Chapter 7, the interaction of the phosphoinositides with human erythrocyte Band 3 protein was characterised using resonance energy transfer. The efficiency of quenching of the fluorescence from tryptophan residues in the protein by pyrene lipids in the membrane decreased in the order pyrPIP 2 , pyrPIP, pyrPI and pyrPC indicating different spatial distributions of these lipids with respect to Band 3. Global analysis of the quenched tryptophan fluorescence decays yielded indices for the spatial distributions of the different lipids. It appeared that the charged lipids PIP 2 and PIP spend longer time in the vicinity of Band 3 than PI and PC. In investigations complementary to these protein fluorescence observations, the monomer/excimer fluorescence of pyrene labelled phosphoinositides in DOPC membranes was evaluated in the absence and presence of Band 3 protein (Chapter 8). In combination with theoretical refinements of the model for lipid migration, these experiments revealed estimates for the lateral dynamics and lipid-lipid repulsion in membranes. In addition, the relative binding constants and the minimum number of Band 3 sites possessing affinity for the phosphoinositides could be roughly estimated. Consistent with intuitive expectations the lipid-lipid repulsion appeared highly dependent on the amount of negative charge of the lipid headgroups. Combining these experiments indicates that lipid-lipid repulsion and the Band 3-lipid interactions induce a non- random distribution of the Phosphoinositides. This non-random distribution will probably effect the accessibility of the inositol lipids for phospholipase C and for other phosphoinositide dependent proteins like PKC and may result in the functional heterogeneity of the phosphoinositides (Downes & Michell, 1985).

Refinements in technique and data analysis have opened new avenues for a detailed interpretation of protein fluorescence. What is more, by combining new insights in protein structure and dynamics with improved knowledge of photophysics of biological chromophores, the coupling between structure-function relationships and fluorescence properties is beginning to have a firm foundation. As a consequence of this interdisciplinary marriage, bidirectional information can be inferred from fluorescence spectroscopy of proteins.In chapter 2 we could explain the time-resolved fluorescence parameters of FAD in glutathione reductase and lipoamide dehydrogenase by a different interplay between conformational substates in both enzymes. The existence of these states monitored by fluorescence relaxation spectroscopy, allowed us to postulate a new model of substrate binding prior to catalysis (Fig.1).In this model, the majority of enzyme molecules in solution is present in an inaccessible conformation to substrate or cofactor ("closed" conformation, A in Fig.1). Structural (equilibrium) fluctuations give rise to a minor population of enzyme molecules with a proper conformation to bind substrate (" open " conformation, B in Fig.1). The equilibrium between "open" and "closed" enzyme conformations is shifted towards the "open" form in the presence of excess substrate or cofactor molecules (C in Fig.1) which reduce the flavin on the enzyme (D in Fig.1). The structural transition can be the rate limiting step in catalysis if the activation barrier between the "open" and "closed" enzyme structures is sufficiently large.From time-resolved fluorescence anisotropy measurements as function of temperature one can distinguish between reorientational motion and energy transfer. In both dimeric proteins intersubunit energy transfer takes place between the FAD prosthetic groups. Furthermore, restricted reorientational motion of the flavin is only revealed in glutathione reductase. It is very likely that this flexible isoalloxazine is present in a minor population of enzyme molecules with a displaced Tyr197 making the flavin accessible to the cofactor (NADPH). This catalytically important enzyme species is not detected by X-ray crystallography and its observation requires a technique with high dynamic range of intensity such as time-resolved fluorescence. In order to derive the distance and relative orientation between the flavins, separate models describing fluorescence anisotropy must be applied to both proteins. The geometric arrangement of the flavins as derived from the analysis of the results is in excellent agreement with crystallographic data.Chapter 3 of this thesis deals with the dynamic properties of the flavin environment in Azotobacter vinelandii lipoamide dehydrogenase as determined by steady-state fluorescence spectroscopy. As model systems the native enzyme and a deletion mutant lacking the 14 C-terminal amino acids were chosen. By fluorescence quenching and temperature dependent spectroscopy it was shown that the mutant has a flavin site accessible to solvent in contrast to wild type lipoamide dehydrogenase. We have to conclude from these observations and enzyme kinetic studies [1] that the C-terminal polypeptide of lipoamide-dehydrogenase is folded back onto the dehydrolipoamide binding site. This result is in agreement with the crystal structure of the Pseudomonas putida enzyme (A. Mattevi, personal communication) but in discrepancy with the disordered C-terminal tail of crystalline A.vinelandii lipoamide dehydrogenase [2]. In cryogenic solvents different dipolar relaxation behavior was found for mutant and native enzymes. The flavin environment in the wild type enzyme was rigid on a nanosecond timescale while a conversion from slow to rapid relaxation was observed in the mutant protein from 203 to 303 K. The contributions to dipolar relaxation in the mutant enzyme were found to arise from solvent and protein dipoles. In aqueous solution, rapid relaxation was observed in both proteins indicating that rapid protein fluctuations are damped in viscogens. Because of the difference in dipolar relaxation in wild type and mutant proteins, the unique spatial arrangement of the two flavins makes this biological system an appropriate object to study the relationships between homo-energy transfer, dipolar relaxation and excitation energy. In this context a compartmental theoretical model is developed in chapter 4 on the basis of existing qualitative descriptions [3, 4]. The model was verified by global analysis of fluorescence anisotropy decay surfaces. The geometry and dynamics derived from the results were in excellent agreement with the protein-crystal geometry [2] and with the dipolar relaxation elaborated in chapter 3. In addition, thermodynamic parameters of equilibrium fluctuations in the protein structures could be determined from the maximum entropy analysis of the fluorescence decay. In agreement with Frauenfelder's concept of tiers of conformational substates [5, 6], the existence of two classes of states were established in lipoamide dehydrogenase (CS 1and CS 3). Thermal quenching of flavin fluorescence was related to transitions between conformational substates of category CS 3by comparing activation energy values.In chapter 5 red- and blue-edge spectroscopy was applied to the electron-transferring flavoprotein from Megasphaera elsdenii which contains two FAD molecules within one subunit. Energy transfer between the two flavins was observed from timeresolved fluorescence depolarization experiments. We could demonstrate for the first time that energy transfer is not observable (as theoretically predicted) at the blue-edge of the emission band in an inhomogeneously broadened system. Interflavin distance and the relative orientation of the flavins could be determined for this protein with unknown threedimensional structure.In general, owing to a physical model based on conformational substates in proteins it is possible to use the complex fluorescence decay patterns in enzymes to explain certain aspects of catalysis. The analysis of the fluorescence decay in distributions of lifetimes by the maximum entropy method is then the logical choice since the information on conformational states is contained in the multiplicity of the distribution pattern. The advantage of the maximum entropy method is that no a priori knowledge of the complex protein system needs to be encoded. According to the fluctuation-dissipation theorema [5] it is possible to monitor catalytically important protein conformations without substrate present. In order to test this hypothesis, fluorescence studies of binary enzyme-substrate complexes must be undertaken in the future. Catalytically important conformational substates will then be preferentially populated in contrast to the free enzyme.We have also demonstrated that the validity of physical models on energy transfer in relation to dipolar relaxation can be tested in relatively complex biological systems. Precise goniometric information in bichromophoric proteins can be derived from fluorescence anisotropy relaxation. Temperature and excitation energy-dependent studies of the fluorescence anisotropy provide a diagnostic method to distinguish reorientational motion from energy transfer. The described experimental and analytical methods have broad applicability to other biological systems containing identical fluorophores.

Refinements in technique and data analysis have opened new avenues for a detailed interpretation of protein fluorescence. What is more, by combining new insights in protein structure and dynamics with improved knowledge of photophysics of biological chromophores, the coupling between structure-function relationships and fluorescence properties is beginning to have a firm foundation. As a consequence of this interdisciplinary marriage, bidirectional information can be inferred from fluorescence spectroscopy of proteins.In chapter 2 we could explain the time-resolved fluorescence parameters of FAD in glutathione reductase and lipoamide dehydrogenase by a different interplay between conformational substates in both enzymes. The existence of these states monitored by fluorescence relaxation spectroscopy, allowed us to postulate a new model of substrate binding prior to catalysis (Fig.1).In this model, the majority of enzyme molecules in solution is present in an inaccessible conformation to substrate or cofactor ("closed" conformation, A in Fig.1). Structural (equilibrium) fluctuations give rise to a minor population of enzyme molecules with a proper conformation to bind substrate (" open " conformation, B in Fig.1). The equilibrium between "open" and "closed" enzyme conformations is shifted towards the "open" form in the presence of excess substrate or cofactor molecules (C in Fig.1) which reduce the flavin on the enzyme (D in Fig.1). The structural transition can be the rate limiting step in catalysis if the activation barrier between the "open" and "closed" enzyme structures is sufficiently large.From time-resolved fluorescence anisotropy measurements as function of temperature one can distinguish between reorientational motion and energy transfer. In both dimeric proteins intersubunit energy transfer takes place between the FAD prosthetic groups. Furthermore, restricted reorientational motion of the flavin is only revealed in glutathione reductase. It is very likely that this flexible isoalloxazine is present in a minor population of enzyme molecules with a displaced Tyr197 making the flavin accessible to the cofactor (NADPH). This catalytically important enzyme species is not detected by X-ray crystallography and its observation requires a technique with high dynamic range of intensity such as time-resolved fluorescence. In order to derive the distance and relative orientation between the flavins, separate models describing fluorescence anisotropy must be applied to both proteins. The geometric arrangement of the flavins as derived from the analysis of the results is in excellent agreement with crystallographic data.Chapter 3 of this thesis deals with the dynamic properties of the flavin environment in Azotobacter vinelandii lipoamide dehydrogenase as determined by steady-state fluorescence spectroscopy. As model systems the native enzyme and a deletion mutant lacking the 14 C-terminal amino acids were chosen. By fluorescence quenching and temperature dependent spectroscopy it was shown that the mutant has a flavin site accessible to solvent in contrast to wild type lipoamide dehydrogenase. We have to conclude from these observations and enzyme kinetic studies [1] that the C-terminal polypeptide of lipoamide-dehydrogenase is folded back onto the dehydrolipoamide binding site. This result is in agreement with the crystal structure of the Pseudomonas putida enzyme (A. Mattevi, personal communication) but in discrepancy with the disordered C-terminal tail of crystalline A.vinelandii lipoamide dehydrogenase [2]. In cryogenic solvents different dipolar relaxation behavior was found for mutant and native enzymes. The flavin environment in the wild type enzyme was rigid on a nanosecond timescale while a conversion from slow to rapid relaxation was observed in the mutant protein from 203 to 303 K. The contributions to dipolar relaxation in the mutant enzyme were found to arise from solvent and protein dipoles. In aqueous solution, rapid relaxation was observed in both proteins indicating that rapid protein fluctuations are damped in viscogens. Because of the difference in dipolar relaxation in wild type and mutant proteins, the unique spatial arrangement of the two flavins makes this biological system an appropriate object to study the relationships between homo-energy transfer, dipolar relaxation and excitation energy. In this context a compartmental theoretical model is developed in chapter 4 on the basis of existing qualitative descriptions [3, 4]. The model was verified by global analysis of fluorescence anisotropy decay surfaces. The geometry and dynamics derived from the results were in excellent agreement with the protein-crystal geometry [2] and with the dipolar relaxation elaborated in chapter 3. In addition, thermodynamic parameters of equilibrium fluctuations in the protein structures could be determined from the maximum entropy analysis of the fluorescence decay. In agreement with Frauenfelder's concept of tiers of conformational substates [5, 6], the existence of two classes of states were established in lipoamide dehydrogenase (CS 1and CS 3). Thermal quenching of flavin fluorescence was related to transitions between conformational substates of category CS 3by comparing activation energy values.In chapter 5 red- and blue-edge spectroscopy was applied to the electron-transferring flavoprotein from Megasphaera elsdenii which contains two FAD molecules within one subunit. Energy transfer between the two flavins was observed from timeresolved fluorescence depolarization experiments. We could demonstrate for the first time that energy transfer is not observable (as theoretically predicted) at the blue-edge of the emission band in an inhomogeneously broadened system. Interflavin distance and the relative orientation of the flavins could be determined for this protein with unknown threedimensional structure.In general, owing to a physical model based on conformational substates in proteins it is possible to use the complex fluorescence decay patterns in enzymes to explain certain aspects of catalysis. The analysis of the fluorescence decay in distributions of lifetimes by the maximum entropy method is then the logical choice since the information on conformational states is contained in the multiplicity of the distribution pattern. The advantage of the maximum entropy method is that no a priori knowledge of the complex protein system needs to be encoded. According to the fluctuation-dissipation theorema [5] it is possible to monitor catalytically important protein conformations without substrate present. In order to test this hypothesis, fluorescence studies of binary enzyme-substrate complexes must be undertaken in the future. Catalytically important conformational substates will then be preferentially populated in contrast to the free enzyme.We have also demonstrated that the validity of physical models on energy transfer in relation to dipolar relaxation can be tested in relatively complex biological systems. Precise goniometric information in bichromophoric proteins can be derived from fluorescence anisotropy relaxation. Temperature and excitation energy-dependent studies of the fluorescence anisotropy provide a diagnostic method to distinguish reorientational motion from energy transfer. The described experimental and analytical methods have broad applicability to other biological systems containing identical fluorophores.

The central theme of this thesis is the interaction of FMN with proteins. For one of the proteins studied, the enzyme luciferase from bacteria, further investigations were done on the process of light emission.In chapter 2 and 3 studies are reported on the binding of FMN with relatively simple proteins, the flavodoxins. Flavodoxins were chosen, because they are small proteins with a molecular weight not higher than 23 000. They contain only one equivalent of FMN and consist of only one polypeptide chain. No other prosthetic group is known for the flavodoxins.I n addition to this they can be obtained in high yields from bacterial cultures. These features make the flavodoxins excellent objects for studies on the binding of flavins to proteins.In chapter 2 studies on the binding of FMN by apoflavodoxin from Peptostreptococcus elsdenii are reported. Conclusions were drawn from the dependence on the pH and the NaCl concentration. The rate constant of dissociation depends on the pH, even when extrapolated to zero ionic strength. The titration curve of this rate constant can be explained, by the assumption of the involvment of two protonations, that act highly cooperatively. It should be realised that these protonations do not influence the rate constant of association. But once the complex is formed, the chance of falling apart, if these sites are protonated, is around 6 times higher than without these sites protonated. Furthermore, it was found that the calculated rate constant of association, when extrapolated to zero ionic strength is independent on the pH. At increasing ionic strength this rate constant of association will change. Depending on the pH 1 value, this change is a decrease or an increase in value. At a pH of 3.8 there is almost no change with increasing ionic strength, but above this pH value the rate constant increases, while below this value it decreases with increasing ionic strength. This explains why the combination of a high salt concentration and a low pH is a very effective way of removing the FMN from flavodoxins. This finding might possibly be extrapolated to other flavoproteins as well. By interpreting the results in terms of the Brönsted theory, a net positive charge between 11 and 12 is found on the apoenzyme at low pH. This finding is in agreement with the number of basic amino acid residues in the polypeptide chain.A series of flavin analogues were synthesised and the kinetic parameters of the interaction with Azotobacter vinelandii apoflavodoxin investigated, These studies are presented in chapter 3. Use was made of a fast kinetic method, the temperature jump relaxation technique. The resolution time of the instrument employed is 11 microseconds. All complexes studied revealed only one relaxation process, indicating that within the time limits studied (11 micro seconds - ca. 10 seconds), the association of the flavin and the apoenzyme is a one-step process. This finding is in contrast with an earlier publication by other authors, who detected two relaxation processes. It is shown that the earlier published traces are instrumental artifacts.In chapter 4 the interaction of FMN with an intermediate in the in vitro bacterial bioluminescence reaction is described. The so called "longlived lntermediate", which has been suggested to be an FMN flavoprotein, hao been separated into an apoprotein and free FMN. Because of the high quantum yields of light with :respect to FMN, measured upon reaction of the apoprotein with aldehyde, in vitro bacterial bioluminescence appears to be a sensitised reaction. At a first consideration only FMN could be a likely candidate as a sensitising agent.However, in chapter 5 it is shown that a novel protein, isolated from the bacteria itself will definitely sensitise the in vitro bacterial bioluminescence reaction. This novel protein (BFP) is efficiently fluorescent (quantum yield of fluorescence 0.45) and has an emission maximum at 476 nm. As a result of these observations, it is called the blue fluoreseence protein. By diluting this protein to a concentration of around 1 μM, a spectral shift of the emission maximum is observed. Actually the fluorescence emission spectrum of the protein changes from a spectrum identical to the In vivo bacterial bioluminescence into an emission spectrum identical to the in vitro bacterial bioluminescence emission. Although this means that the emission of this protein could account for both the in vivo and the in vitro emission spectra, it should be mentioned that investigations learned that the chromophore of this protein is not a product of the in vitro reaction. This blue fluorescence protein is the only one of all the emitters proposed so far, that simulates the bluest of the bacterial emissions exactly. Furthermore, the addition of the blue fluorescence protein to the in vitro reaction affects the light emission kinetics it acts as a catalyst), increases the light yield and induces a shift to shorter wavelengths in the bioluminescence emission. Together with the fact that the protein is isolated from the bacteria themselves, these features are strong evidence that this protein is the in vivo emitter.In chapter 6 the purification procedure of the BFP is given in more detail. Furthermore it is shown that it can be isolated from at least two of the four common species of marine bioluminescent bacteria. This suggests that all the bacteria emit their light via the same kind of chemical mechanism. Although the proteins from these two species of bacteria appear to have similar molecular weights, they differ in that the protein from P.fischeri is more tightly associated with the luciferase during the purification procedure than the one from P.phosphoreum and also that its fluorescence excitation maximum is shifted about 10 nm to shorter wavelength. Further investigation should be done in order to learn what the chemical nature of the fluorophore is.

The central theme of this thesis is the interaction of FMN with proteins. For one of the proteins studied, the enzyme luciferase from bacteria, further investigations were done on the process of light emission.In chapter 2 and 3 studies are reported on the binding of FMN with relatively simple proteins, the flavodoxins. Flavodoxins were chosen, because they are small proteins with a molecular weight not higher than 23 000. They contain only one equivalent of FMN and consist of only one polypeptide chain. No other prosthetic group is known for the flavodoxins.I n addition to this they can be obtained in high yields from bacterial cultures. These features make the flavodoxins excellent objects for studies on the binding of flavins to proteins.In chapter 2 studies on the binding of FMN by apoflavodoxin from Peptostreptococcus elsdenii are reported. Conclusions were drawn from the dependence on the pH and the NaCl concentration. The rate constant of dissociation depends on the pH, even when extrapolated to zero ionic strength. The titration curve of this rate constant can be explained, by the assumption of the involvment of two protonations, that act highly cooperatively. It should be realised that these protonations do not influence the rate constant of association. But once the complex is formed, the chance of falling apart, if these sites are protonated, is around 6 times higher than without these sites protonated. Furthermore, it was found that the calculated rate constant of association, when extrapolated to zero ionic strength is independent on the pH. At increasing ionic strength this rate constant of association will change. Depending on the pH 1 value, this change is a decrease or an increase in value. At a pH of 3.8 there is almost no change with increasing ionic strength, but above this pH value the rate constant increases, while below this value it decreases with increasing ionic strength. This explains why the combination of a high salt concentration and a low pH is a very effective way of removing the FMN from flavodoxins. This finding might possibly be extrapolated to other flavoproteins as well. By interpreting the results in terms of the Brönsted theory, a net positive charge between 11 and 12 is found on the apoenzyme at low pH. This finding is in agreement with the number of basic amino acid residues in the polypeptide chain.A series of flavin analogues were synthesised and the kinetic parameters of the interaction with Azotobacter vinelandii apoflavodoxin investigated, These studies are presented in chapter 3. Use was made of a fast kinetic method, the temperature jump relaxation technique. The resolution time of the instrument employed is 11 microseconds. All complexes studied revealed only one relaxation process, indicating that within the time limits studied (11 micro seconds - ca. 10 seconds), the association of the flavin and the apoenzyme is a one-step process. This finding is in contrast with an earlier publication by other authors, who detected two relaxation processes. It is shown that the earlier published traces are instrumental artifacts.In chapter 4 the interaction of FMN with an intermediate in the in vitro bacterial bioluminescence reaction is described. The so called "longlived lntermediate", which has been suggested to be an FMN flavoprotein, hao been separated into an apoprotein and free FMN. Because of the high quantum yields of light with :respect to FMN, measured upon reaction of the apoprotein with aldehyde, in vitro bacterial bioluminescence appears to be a sensitised reaction. At a first consideration only FMN could be a likely candidate as a sensitising agent.However, in chapter 5 it is shown that a novel protein, isolated from the bacteria itself will definitely sensitise the in vitro bacterial bioluminescence reaction. This novel protein (BFP) is efficiently fluorescent (quantum yield of fluorescence 0.45) and has an emission maximum at 476 nm. As a result of these observations, it is called the blue fluoreseence protein. By diluting this protein to a concentration of around 1 μM, a spectral shift of the emission maximum is observed. Actually the fluorescence emission spectrum of the protein changes from a spectrum identical to the In vivo bacterial bioluminescence into an emission spectrum identical to the in vitro bacterial bioluminescence emission. Although this means that the emission of this protein could account for both the in vivo and the in vitro emission spectra, it should be mentioned that investigations learned that the chromophore of this protein is not a product of the in vitro reaction. This blue fluorescence protein is the only one of all the emitters proposed so far, that simulates the bluest of the bacterial emissions exactly. Furthermore, the addition of the blue fluorescence protein to the in vitro reaction affects the light emission kinetics it acts as a catalyst), increases the light yield and induces a shift to shorter wavelengths in the bioluminescence emission. Together with the fact that the protein is isolated from the bacteria themselves, these features are strong evidence that this protein is the in vivo emitter.In chapter 6 the purification procedure of the BFP is given in more detail. Furthermore it is shown that it can be isolated from at least two of the four common species of marine bioluminescent bacteria. This suggests that all the bacteria emit their light via the same kind of chemical mechanism. Although the proteins from these two species of bacteria appear to have similar molecular weights, they differ in that the protein from P.fischeri is more tightly associated with the luciferase during the purification procedure than the one from P.phosphoreum and also that its fluorescence excitation maximum is shifted about 10 nm to shorter wavelength. Further investigation should be done in order to learn what the chemical nature of the fluorophore is.

Er is een onderzoek uitgevoerd naar de bruikbaarheid van de fluorescentiemethode voor het aantonen van Escherichia coli.