Modulation of the enzymatic efficiency of ferredoxin-NADP(H) reductase by the amino acid volume around the catalytic site Matı ´ as A. Musumeci, Adria ´ n K. Arakaki, Daniela V. Rial, Daniela L. Catalano-Dupuy and Eduardo A. Ceccarelli Molecular Biology Division, Instituto de Biologı ´ a Molecular y Celular de Rosario (IBR), Facultad de Ciencias Bioquı ´ micas y Farmace ´ uticas, Universidad Nacional de Rosario, Argentina Ferredoxin (flavodoxin)-NADP(H) reductases (FNRs, EC 1.18.1.2) are a widely distributed class of flavoen- zymes that have non-covalently bound FAD cofactor as a redox center. FNRs participate in a wide variety of redox-based metabolic reactions, transferring elec- trons between obligatory one- and two-electron carri- ers and therefore functioning as a general electron splitter. In non-phototrophic bacteria and eukaryotes, the reaction is driven towards ferredoxin (Fd) reduc- tion, providing reducing power for multiple metabolic pathways, including steroid hydroxylation in mamma- lian mitochondria, nitrite reduction and glutamate synthesis in heterotrophic tissues of vascular plants, radical propagation and scavenging in prokaryotes, and hydrogen and nitrogen fixation in anaerobes (for a review, see [1,2]). In plants, FNR participates in photosynthetic electron transport, reducing Fd at the level of photosystem I, and transferring electrons to NADP + . This process ends with the formation of the NADPH necessary for CO 2 fixation and other biosyn- thetic pathways [2]. The three-dimensional structures of several FNRs have been determined. They display similar structural features, which have been defined as the prototype for a large family of flavoenzymes [3–10]. Plant-type FNRs can be classified into a plastidic class, characterized by Keywords catalytic efficiency; enzyme evolution; ferredoxin; ferredoxin-NADP(H) reductase; oxidoreductases Correspondence E. A. Ceccarelli, Molecular Biology Division, Instituto de Biologı ´ a Molecular y Celular de Rosario (IBR), CONICET, Facultad de Ciencias Bioquı ´ micas y Farmace ´ uticas, Universidad Nacional de Rosario, Suipacha 531, S2002LRK Rosario, Argentina Fax: +54 341 4390465 Tel: +54 341 4351235 E-mail: ceccarelli@ibr.gov.ar (Received 1 November 2007, revised 8 January 2008, accepted 16 January 2008) doi:10.1111/j.1742-4658.2008.06298.x Ferredoxin (flavodoxin)-NADP(H) reductases (FNRs) are ubiquitous flavoenzymes that deliver NADPH or low-potential one-electron donors (ferredoxin, flavodoxin, adrenodoxin) to redox-based metabolic reactions in plastids, mitochondria and bacteria. Plastidic FNRs are quite efficient reductases. In contrast, FNRs from organisms possessing a heterotrophic metabolism or anoxygenic photosynthesis display turnover numbers 20- to 100-fold lower than those of their plastidic and cyanobacterial counterparts. Several structural features of these enzymes have yet to be explained. The residue Y308 in pea FNR is stacked nearly parallel to the re -face of the fla- vin and is highly conserved amongst members of the family. By computing the relative free energy for the lumiflavin–phenol pair at different angles with the relative position found for Y308 in pea FNR, it can be concluded that this amino acid is constrained against the isoalloxazine. This effect is probably caused by amino acids C266 and L268, which face the other side of this tyrosine. Simple and double FNR mutants of these amino acids were obtained and characterized. It was observed that a decrease or increase in the amino acid volume resulted in a decrease in the catalytic efficiency of the enzyme without altering the protein structure. Our results provide exper- imental evidence that the volume of these amino acids participates in the fine-tuning of the catalytic efficiency of the enzyme. Abbreviations Fd, ferredoxin; Fld, flavodoxin; FNR, ferredoxin (flavodoxin)-NADP(H) reductase; IPTG, isopropyl thio-b- D-galactoside. 1350 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS an extended FAD conformation and high catalytic efficiency (turnover numbers in the range 100– 600 s )1 ), and a bacterial class displaying a folded FAD molecule and very low turnover rates (2–27 s )1 ) [2,11]. The K m values for NADP(H), Fd and flavo- doxin (Fld) remain in the low micromolar range for all reductases [2]. Two tyrosine residues interact with each side of the isoalloxazine in plastidic FNRs. On the si-face of the flavin, which is buried within the protein structure, a tyrosine aromatic side-chain (Y89 in pea FNR) makes angles between 54 and 64° with the isoalloxazine in a conformation that is at the energy minimum (Fig. 1A) [12]. This residue participates in an intricate network of interactions that involve other amino acids and the prosthetic group, contributing to the correct position- ing of FAD and the substrate NADP + [12]. The other tyrosine (Y308 in pea FNR) is conserved in all plant- type plastidic FNRs stacked coplanar to the re-face of the isoalloxazine moiety and interacts extensively with it (Fig. 1A) ([1,2,13] and references therein). This tyro- sine has been implicated in catalysis, modulation of the FAD reduction potential, inter- and intra-protein electron transfer processes [14–18] and determination of the specificity and high catalytic efficiency [15,17– 19]. Using NMR techniques, it has been shown that the maize FNR homolog Y314 is perturbed on NADP + binding, as is the carboxyl terminal region of the protein [20]. Recently, experimental evidence for the mobility of the carboxyl terminal backbone region of FNR and, mainly, Y308 has been provided [19], indicating that this movement is essential for obtaining an FNR enzyme with high catalytic efficiency. During catalysis, the nicotinamide ring must move to the re-face of the isoalloxazine moiety for electron transfer to occur. Thus, Bruns and Karplus [3] have proposed that the aromatic side-chain of the carboxyl terminal tyrosine should be displaced to allow the sub- strate to move into the correct position (named ‘in’ conformation). The interaction of the phenol ring of Y308 with the isoalloxazine should be precisely adjusted to facilitate the ‘in’ and ‘out’ conformations of the NADP(H) nicotinamide. A strong interaction of Y308 with the flavin would impede the ability of nico- tinamide to go into the site; meanwhile, a slight inter- action would favor the stacking of the nicotinamide onto the isoalloxazine, thus decreasing the turnover rate of the enzyme, as previously demonstrated with mutant FNRs lacking this amino acid [15,17,21]. By computing ab initio molecular orbital calcula- tions, the geometry of the tyrosine and flavin has been analyzed. It is proposed that Y308 is constrained against the isoalloxazine in a forced conformational arrangement. This arrangement could be a consequence of the influence of amino acids C266 and L268, which face the other side of this tyrosine (see Fig. 1A), forcing it to adopt a more planar orientation with respect to the flavin. C266 is conserved between all FNRs and FNR-like proteins. Homologous residues to L268 are found in the reductases from plant leaves, plant roots, cyanobacteria (blue–green algae) and all algal groups (Chlorophyta, Rhodophyta and Glaucocystophyta). AB Fig. 1. Computer model showing the flavin and Y308 arrangement in FNR. (A) FAD cofactor, Y308 stacked on the re-face of the flavin and amino acids C266, G267 and L268 flanking Y308, as found in pea FNR. (B) Computer graphic based on X-ray diffraction data for pea FNR [21], with the 266–270 loop, FAD prosthetic group and the terminal Y308 shown in dark grey. M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1351 Leu268 is replaced by a serine in the bacterial reducta- ses of subclass I (Azotobacter vinelandii) and by an asparagine in the bacterial reductases of subclass II (Escherichia coli) [22]. The equivalent residues to L268 in other FNR-like enzymes are less conserved, being proline, aspartic acid, serine or alanine. Simple and double FNR mutants of amino acids C266 and L268 were obtained and characterized. It was observed that alteration of the amino acid volume decreases the catalytic efficiency, suggesting that these steric considerations may be a requirement for high catalytic efficiency. The mutations did not produce a significant perturbation of the overall protein structure and did not affect the oxidase activity of the flavo- enzyme. Our results suggest that these amino acids participate in the fine-tuning of enzyme efficiency, modulating the interaction of Y308 and ⁄ or the nicotin- amide with the isoalloxazine. This type of modulation of aromatic residue interactions could be a general strategy occurring in enzyme structures. Results Ab initio molecular orbital calculations The geometries of aromatic amino acids facing the re- face of the flavin were determined using high-resolu- tion plant-type FNR crystal structures. It was observed that these tyrosines always interact in face-to-face posi- tions (Fig. 1A; Table 1). The B ring of the flavin is always involved in this interaction in a nearly parallel position in which the angle formed with the tyrosine phenol and isoalloxazine varies from 0° to 6° in all high-efficiency plastidic FNRs and 15° for the ferre- doxin-NADP(H) reductase from E. coli. To gain a better understanding of this interaction, the geometric preferences of the above-mentioned interaction were analyzed using model molecules and ab initio mole- cular orbital calculations with the restricted Hartree Fock theory level and a 6-311 + G(d,p) basis set. A simplified system was constructed containing lumiflavin (7,8,10-trimethylisoalloxazine), which is an accepted flavin model compound for calculations [23], and phe- nol as the tyrosine R group. This system has been used previously to analyze the geometry of the tyrosine stacked on the si-face of the flavin in FNRs [12]. The single point energies of the flavin–tyrosine system in different conformations were calculated. The arrange- ment of lumiflavin and phenol with the exact geometry found between flavin and Y308 in the crystal structure of pea FNR was used for the initial set-up. Then, dif- ferent arrangements were generated in which the phe- nol placed in this exact position was rotated around the Cc–Cf axis in discrete steps, keeping the orienta- tion of the phenol hydroxyl group and the distance between the aromatic ring centroids constant (see Fig. 2A). This allowed us to obtain arrangements of the two molecules with angles (a) from )75° to 90°. Figure 2B illustrates the differences in potential energy values determined for arrangements of the phe- nol–lumiflavin pair plotted against the angle a,as depicted in Fig. 2A, at a centroid distance of 3.6 or 4.6 A ˚ . The value obtained for the natural geometry of the carboxyl terminal tyrosine in pea FNR (5.8°) was used as reference. These distances were chosen consid- ering the tyrosine–flavin arrangement found in FNRs and because energetically favorable, non-bonded, aro- matic interactions occur in proteins at phenyl ring centroid separations of > 3.4 and < 7 A ˚ [24]. A global energy minimum was theoretically detected between 11° and 22° at a distance between centroids of 3.6 A ˚ . The angle found in E. coli FNR was the closest to the minimum of the plot. In all plastidic FNRs, the position of the tyrosine was near the minimum (repre- sented in Fig. 2B with open circles and a number indi- cating the enzyme). Any position that does not fall within )10° to 37° notoriously decreases the stability of the pair, increasing repulsion, probably as result of steric constraints between the two aromatic rings. When the total energy of the system was analyzed at a centroid– centroid distance of 4.6 A ˚ , a minimum was observed at 40° and a shallow low-energy region was detected from 20° to 55°. Moreover, all differences in potential energy values obtained for arrangements at 4.6 A ˚ between angles from )20° to 85° were equal or lower than the energy calculated for the observed arrangements found in plastidic FNR enzymes in nature (Fig. 2B). All FNRs Table 1. Angles and distances between the tyrosine interacting with the re-face of the flavin and the isoalloxazine B ring obtained from FNR crystal structures. FNR source Type Maximal angle (deg) a Centroid distance (A ˚ ) b PDB ID Reference Paprika Plastidic 5.09 3.70 1sm4 [6] Spinach Plastidic 0.01 3.65 1fnb [3] Anabaena Plastidic 5.40 3.60 1que [4] Pea Plastidic 5.80 3.65 1qg0 [21] Maize Plastidic 1.60 3.65 1gaw [7] Synechococcus sp. Plastidic 0.40 3.50 2b5o Unpublished E. coli Bacterial 15.00 3.60 1fdr [10] a Angle formed between the tyrosine and the re-face of isoalloxa- zine, measured as shown in Fig. 2A. b Distance (d ) from the center of the phenol ring to the center of the proximal flavin ring, as shown in Fig. 2A. Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al. 1352 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS displayed geometries for the re-face tyrosine phenol and flavin falling near or into the minimum energy valley with a centroid separation of about 3.6 A ˚ . However, if the tyrosine were able to move away from the flavin, a more stable arrangement was possible between both aromatic rings, allowing them to gain up to approxi- mately 5.8 kcalÆmol )1 of stabilization energy, as calcu- lated from Fig. 2B. Thus, it may be inferred that the position of the re-face tyrosine in FNRs is not governed by the energetic minimum of the pairwise flavin–phenol interaction. By analyzing the crystal structure of pea FNR, it was deduced that Y308 is constrained against the isoalloxazine in an unfavorable conformational arrangement by the influence of amino acids C266 and L268. These residues face the other side of this tyrosine and are members of a conserved loop (266CGLKG270) that shapes part of the FNR catalytic site (see Fig. 1A,B). They may force Y308 to adopt a more pla- nar orientation with respect to the flavin. The overall result is a less stable conformational arrangement. Design and construction of C266, G267 and L268 single and double FNR mutants Five single mutants of C266, G267 and L268 and a double mutant of C266 and L268 were successfully constructed and confirmed by DNA sequencing. The design of the mutants was intended to preserve the amino acid character and to modify only the relative volume of their R groups. The expression of the FNR mutants as soluble cyto- solic proteins in E. coli was analyzed using SDS-PAGE and western blot (not shown). The expression levels of FNR mutants C266AL268A, C266A and L268V were similar to those of recombinant wild-type FNR. These recombinant enzymes were largely recovered in the sol- uble fraction after the induction of protein expression at 25 °C, disruption of E. coli cells and fractionation by centrifugation. In contrast, replacement of either G267 with a valine or C266 with a leucine or methio- nine produced a notorious precipitation of the expressed polypeptide. FNR mutants C266L and C266M were successfully expressed at 15 °C during 16 h with 0.1 mm isopropyl thio-b-d-galactoside (IPTG). Both C266L and C266M mutant enzymes showed a higher FAD release rate [4.8 · 10 )2 and 2.6 · 10 )2 lmolÆFADÆh )1 Æ(lmolÆFNR) )1 , respectively] than the wild-type enzyme [1.1 · 10 )5 lmolÆFADÆ h )1 Æ(lmolÆFNR) )1 ], as determined by measuring the increase in FAD fluorescence [22] after incubating the enzymes for 5 h at 25 °C. These observations suggest a weaker FAD interaction with the apoprotein, and may explain the difficulties in obtaining these enzymes in soluble form during protein expression in E. coli. Attempts to purify mutant enzyme G267V were unsuc- cessful and no further analysis was possible. All reductase variants were excised from the carrier protein using thrombin protease and, after chromato- graphy on nickel-nitrilotriacetic acid agarose, were obtained in homogeneous form as judged by SDS- PAGE (not shown). FAD content and spectral properties Analysis of the UV–visible absorption properties of the different FNR mutants showed small changes, AB Fig. 2. Computed relative free energy calculations for the lumiflavin–phenol interaction. (A) Scheme of the coordinate system used to define the relative positions of phenol and lumiflavin, as found for Y308 and flavin in pea FNR. a, Dihedral interplanar angle between rings (for clar- ity, only three positions are shown); d, distance between ring centroids. (B) Relative free energy of the arrangement shown in (A) as a func- tion of the stated a angles at fixed distances of 3.6 A ˚ ( , full line) and 4.6 A ˚ ( , broken line). Open symbols indicate the observed values for the different plastidic FNRs as follows: 1, paprika; 2, spinach; 3, Anabaena ; 4, pea; 5, maize; 6, Synechococcus sp.; 7, E. coli. Ab initio molecular orbital calculations were performed as described in Materials and methods. M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1353 indicating slight variations in the environment of the flavin prosthetic group. All proteins displayed a typical FNR spectrum with maxima at approximately 380 and 460 nm and shoulders at 430 and 470 nm. The absor- bance maximum of the transition band of the wild- type enzyme at 386 nm was shifted slightly to 381 nm in the C266L, C266M and C266A mutants, but not in the L268V mutant. At 459 nm, all changes were within the detected error (Fig. 3A). These shifts may indicate modification of the isoalloxazine environment, although none of the amino acids that directly interact with the flavin were modified. The FAD content of the wild-type and mutant enzymes was determined by release in the presence of 0.2% SDS [25]. FAD : poly- peptide stoichiometry values of 0.85–0.99 were calcu- lated for all mutants (not shown). Therefore, the amino acid changes introduced do not prevent assem- bly of the prosthetic group and do not impede the pro- duction of a folded protein, although, as mentioned above, they may affect the protein folding process. CD spectra were recorded for wild-type and mutant FNRs in an effort to assess the impact of the amino acid changes on the structural integrity of the reducta- ses. Wild-type and mutant FNRs had very similar spectra, exhibiting a negative region from 204 to 240 nm, with a minimum similar for all proteins, and a positive ellipticity at 202 nm (Fig. 3B). The near-UV and visible CD spectra (Fig. 3C) of the proteins were also very similar, showing the typical spectrum for FNR [26], with positive ellipticity in the region of the first flavin visible absorption band, and with a peak at approximately 380 nm for the wild-type enzyme and 370 nm for the mutant proteins. This is consistent with the alteration observed in the absorbance spectra of the mutants. A less intense band of negative ellipticity was observed in the region of the second flavin visible band at 470 nm for the wild-type enzyme and mutant proteins (Fig. 3C). In the near-UV region, all FNRs exhibited very strong, sharp positive and negative sig- nals at 271 and 286 nm, respectively. A similar strong signal at 272 nm, observed in the CD spectrum of E. coli Fld oxidoreductase, has been attributed to the stacked interactions between FAD and one or more aromatic residues [27]. The introduced mutations in FNR did not alter the position of this near-UV band. Some changes in intensity were observed in the FNR mutants, indicating some perturbation of the symmetry relationships between the isoalloxazine chromophore Fig. 3. Absorbance and CD spectra of wild-type and mutant FNRs. Absorbance (A) and CD (B, C) spectra of wild-type FNR (thick line), L268V (thin line), C266AL268A (thick dotted line), C266A (thin dot- ted line), C266L (thick broken line) and C266M (thin broken line). For spectra at 200–250 nm (B), the optical path length was 0.2 cm and the protein concentration of FNRs was 0.5 l M. For spectra at 250–600 nm (C), the optical path length was 1 cm and the protein concentration of FNRs was 5 l M. Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al. 1354 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS and either the carboxyl terminal tyrosine side-chain or the surrounding protein environment. Together, these results clearly indicate that these mutations introduced only local changes in the flavin microenvironment. Interaction with substrates and steady-state kinetics The alterations in the flavin absorption spectrum and the intrinsic FAD fluorescence were used as described previously [19,25,28] to determine the binding con- stants for the FNR–NADP + complexes. The differen- tial spectral changes obtained by incubation of the wild-type and mutant enzymes with NADP + are shown in Fig. 4. NADP + binding to the L268V mutant provoked spectral changes similar in shape and intensity to that of the wild-type enzyme, with the recognized maximum at 510 nm (Fig. 4). In contrast, mutants C266L, C266AL268A, C266A and C266M showed progressive changes in shape and maxima of the differential spectra, indicating a modification in the way in which the nucleotide interacts with the flavin and ⁄ or its environment. Unexpectedly, dissociation constants for NADP + were not significantly affected in any of the mutants for either NADP + or Fd (Tables 2 and 3, respectively). The only exception was the double mutant, which showed an increase in the K d value for the enzyme–NADP + complex. It has been documented that the intensity of the FNR– NADP + differential spectrum peak at about 510 nm correlates with the nicotinamide interaction on the re-face of the isoalloxazine [17,25,28,29]. It may be inferred from the spectral data presented that the inter- action of the NADP + nicotinamide with the flavin is considerably disturbed, probably as a result of changes introduced by the mutations in the environment of the prosthetic group. As a result of the important changes observed for each mutant in the differential spectra elicited by NADP + , it was decided to use an alterna- tive procedure to determine the affinity constant for the nucleotide. The dissociation constants of the FNR–NADP + (Table 2) and FNR–Fd (Table 3) com- plexes were estimated by measuring flavin fluorescence and flavoprotein fluorescence quenching, respectively, after the addition of each substrate, as described in Materials and methods. Similarly, the K d value of the FNR–Fd complex was determined in the presence of NADP + (Table 3). As shown in Table 2, values obtained for the binding of NADP + are in good agreement with those determined by differential spec- troscopy. In our hands and using this methodology, a significant decrease (7.6-fold) in the affinity of FNR for Fd was detected when NADP + was added at a sat- urating concentration, compared with the respective affinity in the absence of substrate (Table 3). Interest- ingly, in all cases, the mutations introduced diminished or completely abolished the observed effect. The catalytic properties of the different FNR mutants were determined for two different enzymatic reactions. The observed values for k cat and K m and the calculated k cat ⁄ K m value for NADPH and Fd are summarized in Tables 2 and 3. L268V FNR displayed k cat and K m values for the diaphorase reaction in the region of 0.8 and 2.0 times those observed for the wild-type enzyme. Similarly, the decrease in Fd was about 0.7 times that observed with the wild-type enzyme. By contrast, mutations in C266 produced a more dramatic effect on the catalytic properties of FNR. Replacement of C266 with a methionine, which implies a volume increase of 55.5 A ˚ 3 , decreased the k cat value by more than 99.8% and increased the K m value for diaphorase activity three-fold. C266AL268A FNR, in which substitutions produced an amino acid volume decrease of 96.2 A ˚ 3 , also showed a major disruption in catalytic function, with a k cat reduction of more than 99% and a 20-fold increase in K m . The introduced changes resulted in a 1300- and 2200-fold decrease in the catalytic efficiency of C266M and C266AL268A, respectively. The correlation between the catalytic efficiency changes caused by the mutations and the different amino acid physicochemical properties was investi- gated. All introduced mutations substituted a polar Fig. 4. Interaction of wild-type and mutant FNRs with NADP + . Dif- ferential spectra of the wild-type FNR (thick line), L268V (thin line), C266AL268A (thick dotted line), C266A (thin dotted line), C266L (thick broken line) and C266M (thin broken line) elicited by NADP + binding, as obtained from the mathematical subtraction of the absorption spectra in the absence and presence of 0.3 m M NADP + . M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1355 neutral amino acid by non-polar residues without a change in the net charge. The catalytic efficiencies of the different enzymes were plotted as a function of the absolute change of hydropathy according to Kyte and Doolittle [30] (Fig. 5A), the octanol–water partition coefficient (log P) [31] and volume [32]. No correla- tions were found with changes in hydropathy (Fig. 5A) and log P (Fig. 5B). In contrast, the absolute change in volume correlated with the decrease in catalytic effi- ciency (Fig. 5C). An alteration (increase or decrease) in volume of the amino acid at position 266 induced a decrease in catalytic efficiency of the enzyme. Mutants with higher volume changes in this residue were more affected. The value for the reduction in catalytic effi- ciency previously obtained by replacement of spinach FNR C272 (homolog to pea FNR C266) with a serine [28] was included (open symbols in Fig. 5A–C and white bar in Fig. 5D). Mutations in C266 that decreased the amino acid volume resulted in a moderate increase in catalytic effi- ciency for the activity of cytochrome c reductase. These results originate from the higher relative decrease in K m than k cat, with respect to the corre- sponding values observed in the wild-type enzyme, and consequently inferences from the calculated changes in catalytic efficiency may not be appropriate [33]. The accumulated data reviewed by Mattevi et al. [34] indicate that the rate-limiting step in the oxygen reactivity of flavoproteins is the first electron transfer step from the two-electron-reduced flavin to mole- cular oxygen. In this context, the oxidase activity of the wild-type and mutant FNRs was investigated at Table 2. Kinetic parameters for the diaphorase reaction of the wild-type (WT) and mutant FNRs, and dissociation constants for the different FNR–NADP + complexes. Potassium ferricyanide reduction was measured using the diaphorase assay of Zanetti [68] in 50 mM Tris ⁄ HCl (pH 8.0). FNR form DV a (A ˚ 3 ) K m (NADPH) (l M) k cat (s )1 ) k cat ⁄ K m (lM )1 Æs )1 ) DDG MUT ⁄ WT b (kcalÆmol )1 ) K d (NADP + ) c (lM) K d (NADP + ) d (lM) WT 0 15 ± 2 374 ± 22 25 ± 5 0.00 41 ± 2 38 ± 5 C266A )21.5 193 ± 32 39 ± 2 0.20 ± 0.04 2.86 22 ± 2 31 ± 5 L268V )25.0 31 ± 1 308 ± 15 9.97 ± 0.50 0.54 37 ± 2 31 ± 4 C266AL268A )96.2 299 ± 42 3.5 ± 0.5 0.011 ± 0.003 4.58 87 ± 6 120 ± 8 C266L 53.2 16 ± 2 2.32 ± 0.04 0.14 ± 0.01 3.07 29 ± 4 18 ± 1 C266M 55.5 44 ± 5 0.82 ± 0.04 0.018 ± 0.002 4.28 31 ± 2 18 ± 2 a Volume change of the R amino acid groups introduced by the mutations was determined following the standard radii and volumes calcu- lated by Tsai et al. [32]. b DDG MUT ⁄ WT indicates the energy barrier introduced by the mutations to the catalytic efficiency of FNR calculated by the following equation: DDG MUT ⁄ WT = )RT ln(k cat ⁄ K m ) MUT ⁄ (k cat ⁄ K m ) WT . c Determined by differential spectra using  15 lM flavoproteins in 50 m M Tris ⁄ HCl (pH 8.0) at 25 °C. Absorbance differences (DA at 510 nm for the wild-type and L268V mutant FNRs, and at 390 nm for the C266A, C266AL268A, C266L and C266M mutant FNRs) were measured and plotted against increasing NADP + concentration. The data were fitted to a theoretical equation for a 1 : 1 complex. d Determined by fluorescence spectroscopy using oxidized flavoproteins at 8.5 lM in 50 mM Tris ⁄ HCl (pH 8.0) at 25 °C, as described in Materials and methods. Table 3. Kinetic parameters for cytochrome c reductase of the wild-type (WT) and mutant FNRs, and dissociation constants for the com- plexes of the different FNR forms with Fd. Cytochrome c reduction was followed at 550 nm (e 550 =19mM )1 Æcm )1 ) as described in Materials and methods. ND, not determined. FNR form DV a (A ˚ 3 ) K m (Fd) (l M) k cat (s )1 ) k cat ⁄ K m (lM )1 Æs )1 ) K d for the FNR–Fd complex (lM) In the presence of NADP + (K dP ) b In the absence of NADP + (K dA ) b K dP ⁄ K dA WT 0 2.2 ± 0.4 1.62 ± 0.14 0.73 ± 0.19 5.16 ± 0.25 0.68 ± 0.02 7.6 C266A –21.5 0.20 ± 0.03 0.74 ± 0.03 3.70 ± 0.70 2.78 ± 0.20 1.02 ± 0.13 2.8 L268V –25.0 4.6 ± 0.8 1.08 ± 0.08 0.23 ± 0.05 2.80 ± 0.30 2.69 ± 0.10 1.0 C266AL268A –96.2 0.004 ± 0.001 0.016 ± 0.0009 4.00 ± 1.22 2.74 ± 0.24 2.20 ± 0.26 1.3 C266L 53.2 ND ND ND 2.86 ± 0.22 2.88 ± 0.19 1 C266M 55.5 ND ND ND 3.53 ± 0.39 2.77 ± 0.15 1.3 a Volume change of the R amino acid groups introduced by mutations was determined following the standard radii and volumes calculated by Tsai et al. [32]. b Determined by fluorescence spectroscopy using oxidized flavoproteins at 3 lM in 50 mM Tris ⁄ HCl (pH 8.0) at 25 °Cin the absence or presence of 0.3 m M NADP + , as described in Materials and methods. Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al. 1356 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS saturating NADPH concentration. As shown in Table 4, wild-type and mutant enzymes displayed simi- lar oxidase activities, indicating that no changes are evident in this process on mutation of the FNR resi- dues under study. Thermal analysis of protein unfolding for wild-type and mutant FNRs Thermal denaturation determined by CD was used to measure the stability of the FNR mutants. Based on measurements over a range of temperatures (shown in Fig. 6), parameters such as the midpoint of the unfold- ing transition melting point (T m ) were calculated, and are shown in Table 5. Curves were also analyzed on the basis of the two-state model [35], and the corre- sponding DS m values (entropy change at T m ) were calculated from the slopes of DG versus T at midpoint temperatures [35]. All replacements led to less stable enzymes compared with wild-type FNR. However, mutations that introduced reductions in amino acid volume caused slight to moderate changes in stability with respect to the wild-type enzyme (– 0.76 to )0.94 kcalÆmol )1 ). Using the foldx algorithm [36], the C266A L268A L268V C266A C272S WT C266L C266M –25.0 –21.5 –16.3 0.0 53.2 55.5 –96.2 ΔV (Å 3 ) FNRs ΔV (Å 3 ) (absolute value) 1 ΔHydropathy (absolute value) ΔlogP (absolute value) 0 20 40 60 80 100 0.01 0.1 10 100 0234 0.01 0.1 1 10 100 0.0 0.1 0.2 0.3 0.4 0.5 k cat /K m (%) k cat /K m (%) 1 AB CD Fig. 5. Catalytic efficiencies of wild-type and mutant FNRs plotted as a function of different amino acid physicochemical properties. The cat- alytic efficiencies of wild-type and mutant FNRs from Table 2 (percentage of the wild-type enzyme) are plotted as a function of the absolute changes in hydropathy [30] (A), octanol–water partition coefficient [31] (B), volume [32] (C) and volume change in C266 (filled bars) and L268 (hatched bar) mutants (D). The FNR mutant C272S from spinach showed a k cat ⁄ K m value five-fold lower than that of the wild-type reductase (0.40 versus 14.28 l M )1 Æs –l ) [28], and is represented by an open symbol in (A), (B) and (C) and a white bar in (D). Substitution of C with S introduces a volume change of )16.3 A ˚ 3 . Table 4. Oxidase activity of the wild-type (WT) and mutant FNRs. Oxidase activity was followed by NADPH oxidation, as described in Materials and methods. FNR form DV (A ˚ 3 ) Oxidase activity (s )1 ) WT 0 0.10 ± 0.01 C266A )21.5 0.09 ± 0.01 L268V )25 0.09 ± 0.01 C266AL268A )96.2 0.08 ± 0.009 C266L 53.2 0.11 ± 0.01 C266M 55.5 0.14 ± 0.01 M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1357 direct effect of mutations that replace native amino acids with alanine on the overall stability of the pro- tein was evaluated. A theoretical DDG value of )1.02 kcalÆmol )1 was obtained for the C266A mutant, in complete agreement with our experimental results. When the amino acid mutation induced a volume increase, important destabilizations were experimen- tally observed. C266L and C266M exhibited lower DDG values: )8.50 and )6.80 kcalÆmol )1 , respectively. These outcomes indicate that although little influence is exerted by residue substitutions on the destabiliza- tion of the secondary and tertiary structure (see Fig. 3) there is a considerable difference in thermal energy change between the wild-type enzyme and mutants with replacements that increase volume. Discussion The role of the aromatic residue interacting with the re-face of the flavin in FNR-like enzymes has been analyzed, and a variety of functions have been pro- posed [14,16,18,19,37,38]. In previous publications, mechanistic evidence has been presented that the inter- action of the nicotinamide of substrate NADP + with the isoalloxazine is modulated by the terminal tyrosine (Y308 in pea FNR) [15,17,18]. During binding of NADP + , the terminal tyrosine should be removed from its resting place to allow the nicotinamide to move into a productive position [21]. This exchange between Y308 and the NADP + nicotinamide has been experimentally indicated as the enzyme rate-limiting step [18]. Evidence has recently been presented that the mobility of the carboxyl terminal region is essential for obtaining high catalytic rates [19]. Ab initio calcula- tions and mutagenesis studies were performed on the FNR enzyme with the aim of obtaining a better under- standing of the structural and functional role of this tyrosine and the interacting amino acids C266, G267 and L268. The data support the hypothesis that the aromatic interaction between the flavin, Y308 and the nicotinamide of NADP + is precisely tuned by selecting amino acids that face the other side of the tyrosine phenol ring. The specific volumes of the above-men- tioned residues condition the arrangement of Y308 and the nicotinamide of NADP + in the catalytic site. Non-covalent aromatic interactions are essential to protein–ligand recognition [39]. Furthermore, they are widespread in biomolecules, clusters, organic ⁄ biomo- lecular crystals and, more recently, in the building of nanomaterials [40]. In proteins, the rings of trypto- phan, tyrosine, phenylalanine and histidine participate either in the interaction with hydrogen donors (p–H interaction) or binding with other aromatic rings (p–p interactions) [41]. The latter interactions are observed in a great variety of geometries. The edge–face geome- try is commonly found between aromatic residues in proteins. Other two-stacked orientations are also estab- lished, including one in which the interacting rings are offset and stacked near-planar, and arrangements of face-to-face stacked aromatic rings [42]. By analyzing the crystal structure of FNRs, it was found that the inter-ring orientational angles between the re-face aromatic ring and flavins were quite con- stant and always positioned at a limiting distance of 3.6 A ˚ . Our ab initio calculations indicated that Y308 in pea FNR adopts a conformation close to minimum Table 5. Thermodynamic parameters derived from the thermally induced unfolding curves of wild-type (WT) and mutant FNRs. The data of Fig. 6 were analyzed assuming a two-state approximation as described previously [67]. FNR form DV a (A ˚ 3 ) T m ( 0 C) DS m (kcalÆmol )1 Ædeg )1 ) DDG (kcalÆmol )1 ) WT 0 64.7 ± 0.2 0.61 ± 0.04 C266A )21.5 63.2 ± 0.3 0.71 ± 0.03 )0.94 L268V )25.0 63.5 ± 0.1 )0.40 ± 0.01 )0.76 C266AL268A )96.2 63.5 ± 0.1 0.69 ± 0.01 )0.77 C266L 53.2 50.8 ± 0.1 0.38 ± 0.01 )8.50 C266M 55.5 53.6 ± 0.2 0.43 ± 0.02 )6.80 a Volume change of the R amino acid groups introduced by the mutations was determined following the standard radii and vol- umes calculated by Tsai et al. [32]. Fig. 6. Thermal unfolding of wild-type and mutant FNRs monitored by CD. CD melting curves were recorded at 280 nm, using a pro- tein concentration of 3 l M in 50 mM potassium phosphate (pH 8.0), whilst the temperature of the sample was increased at a uniform rate of 1 °CÆmin )1 (from 25 to 80 °C). Wild-type FNR (thick line), L268V (thin line), C266AL268A (thick dotted line), C266A (thin dot- ted line), C266L (thick broken line) and C266M (thin broken line) are shown. Enzyme efficiency modulated by amino acid volume M. A. Musumeci et al. 1358 FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS energy for a distance of 3.6 A ˚ . However, when calcula- tions were performed with aromatic rings stacked at 4.6 A ˚ , a lower energy minimum was obtained. These results suggest that, if more freedom were available for the arrangement, the aromatic ring of the tyrosine would adopt a T-shaped geometry, with increased stabilization of the pair. In all plastidic FNRs, Y308 homologs are close to the calculated minimum at 3.6 A ˚ , supporting the theoretical data obtained. More- over, it may be inferred from these observations that the orientation of Y308 with respect to the flavin is mainly governed by the aromatic interaction without involvement of attractive forces from the other side of tyrosine. The relative stability of planar and T-shaped aromatic interactions has been studied extensively, but consolidated conclusions are still being debated. The accumulated evidence indicates that the T-shaped structure is likely to be more stable than the planar stacked structure, as calculated for model systems [42]. Tyrosines have been found to interact with flavins in a myriad of arrangements, including, for example, spa- tial T-shaped arrangements [12,34], planar parallel and displaced stacks [1,2,13,43,44] and even near-90° T-shaped orientation [45], demonstrating that the sur- rounding environment can condition these arrange- ments. It has been observed previously that Y89 in pea FNR, which faces the si-face of the flavin in a T-shaped geometry of 54°, is close to the global energy minimum [12]. Similar conclusions have been found for the phenol side-chain of the si-face tyrosine of sev- eral FNR family flavoproteins ([12,46] and references therein). Our calculations also indicate that the tyro- sine–flavin bacterial arrangement in E. coli FNR is 1.24 kcalÆmol )1 more stable than that observed for the same pair in plastidic pea FNR (open circle numbered 4 in Fig. 2). Thus, tyrosine displacement for nicotin- amide binding should be easier in pea FNR than in the bacterial enzyme. As this movement was postulated to be the rate-limiting step for catalysis [18], the differ- ences in stability may account for the distinct turnover numbers that are 20- to 100-fold lower for bacterial enzymes than their plastidic and cyanobacterial coun- terparts. Our mutants enabled the observed results to be interpreted in terms of protein structure, thermody- namics and function. The C266 mutants are of particu- lar interest because this residue has functional homologs in all FNR-like structures. Moreover, the cysteine and glycine at this position are part of one of the consensus sequences that define the structural fam- ily [1,11]. As anticipated, the final tertiary structure of the mutants, with the exception of G267V, was rela- tively unchanged, as shown by the fact that mutations in FNR did not alter the near-UV band of the CD spectra. A small perturbation of isoalloxazine was detected by CD and UV–visible spectrophotometry. Flavin electronic transitions in the 300–600 nm region originate from p–p transitions [26]. Thus, changes in the CD spectra are expected to occur on modification of the interaction of Y308 with the flavin. Our mutants displayed variations at 370–380 nm, correlating with the changes observed in that region of the UV–visible spectra. Mutations may induce either a change in the interaction strength between the flavin and Y308 or a displacement of the ‘in’ and ‘out’ equilibrium of the Y308 phenol ring [15,17,21], which could not be detected by crystal structure analysis. Alteration of the flavin environment was more noticeable when the differential spectra elicited by NADP + binding were analyzed. These changes were closely related to the magnitude of the changes intro- duced with respect to the wild-type enzyme. Substitu- tion of C266 with the bulky methionine completely reverted the shape of the differential spectrum of the wild-type enzyme with NADP + , producing a profile quite similar to that already obtained for the wild-type FNR from Anabaena variabilis when the nucleotide is bound [47]. The absence of the characteristic band at 510 nm for the flavin–nicotinamide interaction has been explained by the observation that the C-terminal tyrosine in this enzyme has a reduced degree of ‘out’ conformations relative to other plastidic FNRs. Conse- quently, our observations may account for a reduced interaction of the nicotinamide with the flavin in the C266M mutant. Moreover, spectral changes on NADP + binding to L268V are coincident with the dif- ferential spectra previously obtained for the Anabaena variabilis FNR mutant L263A [47]. The K d values obtained for NADP + binding to the mutants were only slightly modified, with the exception of the double mutant. It can be concluded that the interactions with the adenine and phosphate regions of NADP + are conserved, and that the observed alteration is probably the result of a change in the position or extent of inter- action between the flavin and the nicotinamide. Kinetic analysis of the mutants indicates that the cysteine sulfhydryl group is by no means essential for catalysis, as documented previously [28]. Replacement of C266 by any aliphatic residue produced enzymes that, even when notoriously affected in catalysis, were still active. When the cysteine was substituted with a methionine, providing a sulfur atom in a nearby posi- tion, a functional enzyme was also obtained. Sulfur– flavin interactions have been proposed and analyzed by computational studies and experimental means [48,49]. These studies have indicated the existence of M. A. Musumeci et al. Enzyme efficiency modulated by amino acid volume FEBS Journal 275 (2008) 1350–1366 ª 2008 The Authors Journal compilation ª 2008 FEBS 1359 [...]...Enzyme efficiency modulated by amino acid volume M A Musumeci et al an interaction of sulfur with the electron-deficient pyrimidine moiety of the flavin ring system In the case of Fld from Clostridium beijerinckii, a methionine is ˚ located 3.9 A from the flavin A methionine to alanine mutation of this enzyme reduced the flavin binding energy by 0.5 kcalÆmol)1 [50] In pea FNR, the sulfhy˚ dryl group of C266... the mutation in position 266 and the decrease in enzyme catalytic efficiency Remarkably, the value for the catalytic efficiency previously obtained by replacement of the homologous residue in spinach FNR (C272) with a serine [28] fits perfectly on our graph, and follows the trend of the other mutants Our theoretical calculations indicate that a more stable arrangement of the flavin and tyrosine would provide... that the probabilities of compensatory mutations that involve small changes in amino acid volumes are higher [53] These observations may be taken to support the intuitive idea that small changes may produce a lesser effect on protein structure, consistent with that observed in the protein stability of our mutants Therefore, these conclusions strengthen our hypothesis that the catalytic efficiencies of the. .. et al the effect on the Michaelis constant being higher than that on the activity of the enzyme Although the ultimate catalytic mechanism of Fd reduction by FNR is not known, it has been suggested that electron transfer from the two-electron substrate NADPH to Fd proceeds in an ordered pathway, in which the observed Km value results from the sum of the Km values for the successive interaction of the. .. related to the volume of the mutated amino acids It has been proposed that, in Anabaena FNR, the 261–265 loop (which is equivalent to the 266–270 region in pea FNR, see Fig 1) is involved in determining coenzyme specificity [47] A triple mutant of amino acids T155G ⁄ A160T ⁄ L263P produced a marked retraction of the above-mentioned loop, resulting in a decrease in catalytic efficiency and a relaxation of enzyme... kcalÆmol)1 of stabilization energy This value is in good agreement with the energy barrier introduced by the mutations to the catalytic efficiency of FNR, as shown in Table 2 Data from the L268V mutant also support our hypothesis, although the change in kcat was smaller than that observed for enzymes mutated at position 266 L268V FNR shows a catalytic efficiency of about 40% with respect to that of the wild-type... binding rather than being directly involved in the hydride transfer reaction [21] Our results, together with the amino acid arrangement observed in the catalytic site, indicate that C266 may constrain the nicotinamide and or the terminal tyrosine against the flavin It has been described that the binding of NADH to lactate dehydrogenase conditions the nicotinamide glycosidic bond torsion angle, altering the. .. molecular theory calculations were carried out at the Restricted Hartree Fock theory level with pc gamess V7.0 accessible at http://classic.chem.msu.su/gran/ gamess/index.html using a 6-311 + G(d,p) basis set The volume change of the R amino acid groups introduced by mutations was calculated following the standard radii and volumes calculated by Tsai et al [32], assuming a reduced state of the cysteine Amino. .. stabilize the protein structure The importance of amino acid volume in relation to non-synonymous substitutions in proteins was envisaged several years ago [51] When globin sequences were analyzed, it was observed that the total sequence volume in conserved proteins was quite constant, with variations of 2–3% [52] The variation in amino acid volume per internal position is in the region of 13% and... the flavin can result in dramatic alterations of the reactivity towards oxygen The reaction with flavins is quite complex and still not completely understood However, following the analysis of Mattevi et al [34], it can be inferred that the amino acid changes introduced in FNR in this study did not modify the thermodynamic driving force of the reaction of the enzyme with oxygen, suggesting that substitutions . Modulation of the enzymatic efficiency of ferredoxin-NADP(H) reductase by the amino acid volume around the catalytic site Matı ´ as A exper- imental evidence that the volume of these amino acids participates in the fine-tuning of the catalytic efficiency of the enzyme. Abbreviations Fd,