Escherichia coli general NAD(P)H:flavin oxidoreductase (Fre) does not have a bound flavin cofactor; its flavin substrates (riboflavin, FMN, and FAD) are believed to bind to it mainly through the isoalloxazine ring. This interaction was real for riboflavin and FMN, but not for FAD, which bound to Fre much tighter than FMN or riboflavin. Computer simulations of Fre·FAD and Fre·FMN complexes showed that FAD adopted an unusual bent conformation, allowing its ribityl side chain and ADP moiety to form an additional 3.28 H-bonds on average with amino acid residues located in the loop connecting Fβ5 and Fα1 of the flavin-binding domain and at the proposed NAD(P)H-binding site. Experimental data supported the overlapping binding sites of FAD and NAD(P)H. AMP, a known competitive inhibitor with respect to NAD(P)H, decreased the affinity of Fre for FAD. FAD behaved as a mixed-type inhibitor with respect to NADPH. The overlapped binding offers a plausible explanation for the largeK m values of Fre for NADH and NADPH when FAD is the electron acceptor. Although Fre reduces FMN faster than it reduces FAD, it preferentially reduces FAD when both FMN and FAD are present. Our data suggest that FAD is a preferred substrate and an inhibitor, suppressing the activities of Fre at low NADH concentrations. Escherichia coli general NAD(P)H:flavin oxidoreductase (Fre) does not have a bound flavin cofactor; its flavin substrates (riboflavin, FMN, and FAD) are believed to bind to it mainly through the isoalloxazine ring. This interaction was real for riboflavin and FMN, but not for FAD, which bound to Fre much tighter than FMN or riboflavin. Computer simulations of Fre·FAD and Fre·FMN complexes showed that FAD adopted an unusual bent conformation, allowing its ribityl side chain and ADP moiety to form an additional 3.28 H-bonds on average with amino acid residues located in the loop connecting Fβ5 and Fα1 of the flavin-binding domain and at the proposed NAD(P)H-binding site. Experimental data supported the overlapping binding sites of FAD and NAD(P)H. AMP, a known competitive inhibitor with respect to NAD(P)H, decreased the affinity of Fre for FAD. FAD behaved as a mixed-type inhibitor with respect to NADPH. The overlapped binding offers a plausible explanation for the largeK m values of Fre for NADH and NADPH when FAD is the electron acceptor. Although Fre reduces FMN faster than it reduces FAD, it preferentially reduces FAD when both FMN and FAD are present. Our data suggest that FAD is a preferred substrate and an inhibitor, suppressing the activities of Fre at low NADH concentrations. NAD(P)H:flavin oxidoreductase riboflavin molecular dynamics potassium phosphate dithiothreitol electrospray ionization mass spectrometry high pressure liquid chromatography Escherichia coli general NAD(P)H:flavin oxidoreductase does not contain any bound flavin cofactor (1Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1987; 262: 12325-12331Abstract Full Text PDF PubMed Google Scholar, 2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar). This property separates Fre1 from the flavin-containing NAD(P)H:flavin oxidoreductases of Vibrio harveyi and V. fischeri that supply FMNH2to bacterial luciferases (3Inouye S. FEBS Lett. 1994; 347: 163-168Crossref PubMed Scopus (64) Google Scholar, 4Lei B. Liu M. Huang S. Tu S.C. J. Bacteriol. 1994; 176: 3552-3558Crossref PubMed Google Scholar). Fre uses either NADH or NADPH as electron donors to reduce FAD, FMN, or riboflavin (Rfl); however, when FAD is the electron acceptor, the K m values for NADH and NADPH are exceptionally large (1Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1987; 262: 12325-12331Abstract Full Text PDF PubMed Google Scholar, 2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). Reduced flavins generated by Fre are believed to have important biological functions. It has been shown that the reduced flavins generated by Fre can regulate the activity of the aerobic ribonucleotide reductases by regenerating or scavenging the Tyr122 radical of the ribonucleotide reductase in vitro (1Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1987; 262: 12325-12331Abstract Full Text PDF PubMed Google Scholar, 6Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1989; 264: 9164-9170Abstract Full Text PDF PubMed Google Scholar). An E. coli fremutant is more susceptible to hydroxyurea, a scavenger of the Tyr122 radical, than the wild-type E. coli (7Covès J. Nivière V. Eschenbrenner M. Fontecave M. J. Biol. Chem. 1993; 268: 18604-18609Abstract Full Text PDF PubMed Google Scholar), pointing to the protective role of Fre for the aerobic ribonucleotide reductase in vivo. Fre produces reduced flavins that can reduce metal ions, including ferrisiderophores (8Covès J. Fontecave M. Eur. J. Biochem. 1993; 211: 635-641Crossref PubMed Scopus (95) Google Scholar), Cob(III)alamin (9Fonseca M. Escalante-Semerena J.C. J. Bacteriol. 2000; 182: 4304-4309Crossref PubMed Scopus (57) Google Scholar), and chromate (10Puzon G.J. Petersen J.N. Roberts A.G. Kramer D.M. Xun L. Biochem. Biophys. Res. Commun. 2002; 294: 76-81Crossref PubMed Scopus (118) Google Scholar). Fre is capable of supplying FADH2 to the FADH2-utilizing monooxygenases (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar, 12Louie T.M. Webster C.M. Xun L. J. Bacteriol. 2002; 184: 3492-3500Crossref PubMed Scopus (111) Google Scholar). Despite these apparent biological functions, the genuine physiological role of Fre remains unclear. Recently, Ingelman et al. (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar) reported the crystal structure of Fre with or without bound Rfl, revealing that Fre is similar to members of the ferredoxin:NADP+ reductase family in structure, although the similarities in amino acid sequence between Fre and members of the ferredoxin:NADP+ reductase family are low. The crystal structure shows that Fre is organized into an N-terminal flavin-binding domain and a C-terminal NAD(P)H-binding domain; the secondary structures of the two domains are labeled as F and N, respectively. The most interesting feature is that the loop connecting Fβ5 and Fα1 in the flavin-binding domain that normally interacts with the ADP moiety of FAD in other ferredoxin:NADP+ reductases is exceptionally short in Fre (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar). Consequently, although FAD is a bound cofactor in most ferredoxin:NADP+ reductase proteins, all three flavin substrates of Fre do not remain bound (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar). Because the crystal structure of the Fre·Rfl complex revealed the interactions between the isoalloxazine ring of Rfl and the flavin-binding domain (12Louie T.M. Webster C.M. Xun L. J. Bacteriol. 2002; 184: 3492-3500Crossref PubMed Scopus (111) Google Scholar) and because the K m values of Fre for Rfl, FMN, and FAD are very similar (2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar), it has been proposed that all three flavins mainly interact with residues of the flavin-binding domain through the isoalloxazine ring. However, direct experimental data are unavailable to support the proposed interactions of Fre with FMN and FAD. In this report we characterized FAD and FMN binding to Fre. TheK d values of Fre with FAD and FMN were determined. The value with FAD was much smaller than that with FMN or Rfl. Molecular dynamics (MD) simulations of Fre with FAD predicted that the ribityl side chain and the ADP moiety of FAD confer extra stability to the Fre·FAD complex. Experimental data supported the simulation model. Studies on the physiological roles of the high affinity of Fre for FAD revealed that FAD is a preferred substrate and inhibitor of Fre. An in vivo model for regulating Fre activities in responding to O2 supply is proposed. All reagents were of the highest purity available and were purchased from Sigma, Aldrich, or Fisher Scientific Co. HpaB, an FADH2-utilizing 4-hydroxyphenylacetate 3-monooxygenase, was overproduced and purified from E. coliBL21(DE3) carrying pES2 (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar). Overexpression of the clonedfre gene in E. coli BL21(DE3)(pES1) and purification of Fre were done primarily as reported previously (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar). To ensure Fre of the highest purity, a phenyl-agarose chromatography was added to the previously reported procedures. The ammonium sulfate precipitated proteins were dissolved in 20 mm KPi buffer (pH 7.0) containing 1 mm DTT with 25% saturation of ammonium sulfate and loaded onto a phenyl-agarose (Sigma) column (1.5 × 12.5 cm) equilibrated with the same buffer. The proteins were eluted with a linear gradient of ammonium sulfate (25% to 0%, 200 ml) in KPi buffer with 1 mm DTT at a flow rate of 1 ml·min−1. Active fractions eluted with ∼10% saturation of ammonium sulfate were pooled together and dialyzed against several changes of 20 mm KPi buffer (pH 7.0) with 1 mm DTT overnight. The sample was then purified by going through a Bioscale Q column and Superdex 75 column as reported previously (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar). Purified protein was analyzed by SDS-PAGE (14Laemmli U.K. Nature. 1970; 227: 680-685Crossref PubMed Scopus (207018) Google Scholar). ESI-MS was done with a Waters Micromass ZQ mass spectrometer. Fre samples in 10 mm Tris-HCl buffer (pH 7.5) were acidified with formic acid to a final concentration of 0.1% (v/v) and infused into the electrospray ionization source at a flow rate of 10 μl·min−1. The electrospray ionization source temperature and the desolvation temperature were maintained at 90 °C and 150 °C, respectively. The capillary voltage and the cone voltage were 3400 V and 35 V, respectively. Spectra were scanned fromm/z 800 to 3000 at a rate of 1 scan/s. Protein mass spectrum was deconvoluted using the MaxEnt software (Waters). NAD(P)H:flavin oxidoreductase activity was determined spectrophotometrically by monitoring the oxidation of NADH (ε340 = 6220m−1·cm−1) in 20 mmKPi buffer (pH 7.0) containing 450 μm NADH and 10 μm FMN or FAD at 30 °C. One unit of NAD(P)H:flavin oxidoreductase activity was defined as the oxidation of 1 nmol NADH/min. A Fre-HpaB coupled assay was performed with 50 nmFre and 10 μm HpaB in 20 μl of 20 mm KPi buffer (pH 7.0) containing 80 μm NADH and 500 μm 4-hydroxyphenylacetate, with either 5 μmeach of FAD and FMN or only 5 μm FAD. The reaction was incubated at 30 °C for 30 min, and the amount of 3,4-dihydroxyphenylacetate produced was measured by a previously reported HPLC method (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar). A HPLC system equipped with a photodiode array detector (Waters) and a Biosep Sec-S3000 size-exclusion column (7.8 × 300 mm; Phenomenex) was used to estimate the fraction of Fre molecules that contained tightly bound flavins. Pure Fre (200 μg) was eluted from the column with an isocratic flow of 20 mm KPi (pH 7.0) buffer with 1 mm DTT at 0.5 ml·min−1. The amount of flavin co-eluted with Fre was estimated by comparing the elution peak area of Fre derived at 450 nm with the elution peak area of free flavin standards. The same HPLC system equipped with a Delta Pak C18-300Å reverse phase column (3.9 × 150 mm; Waters) was used to determine the nature of the tightly bound flavin. Fre sample (80 μg) was eluted from the column with a 33-ml linear gradient of acetonitrile in 0.1% trifluoroacetic acid (0–70%) with a flow rate of 1 ml·min−1. Retention time of the flavin dissociated from Fre was compared with those of free flavin standards. The dissociation constants, K d, of the Fre·FAD and Fre·FMN complexes were determined by a spectrofluorometric titration method using a spectrofluorometer (Jobin Ybon Fluorlog). A 48 μm Fre solution in 20 mmKPi buffer (pH 7.4) was titrated with various amounts of flavin added from a 50 μm stock solution, and the change in fluorescence after each addition of flavin was recorded. The excitation wavelength was set at 280 nm, and fluorescence emission of Fre was recorded from 300 to 400 nm at 5-nm intervals. Both excitation and emission monochromator slit widths were set at 5 nm. The fluorescence intensity of Fre was plotted against the ratio of initial concentration of FAD to Fre and fitted with the following equation Ifit=κI0Fre0+xI0−I1Equation 1 in which κ is a scaling constant to compare the model with experimental data, I 0 is the fluorescence intensity of Fre, [Fre]0 is the initial concentration of Fre at that titration point, I 1 is the fluorescence intensity of the Fre·FAD complex, and x is the final concentration of the Fre·FAD complex. The value ofx is calculated by the following equation. x=Fre0+FAD0+Kd−Fre0+FAD0+Kd2−4Fre0FAD02Equation 2 The effect of AMP on the K d of Fre for FAD was also determined. The experimental setup was identical to that described above, except that 0.18 mm AMP was included in the solution. The initial geometries of the Fre·FAD and Fre·FMN complexes were based on the crystal structure of the Fre·Rfl complex (from Dr. Vincent Nivière of the Université Joseph Fourier). The complete protein sequence was reconstituted and annealed by MD simulations using the package GROMACS (15Berendsen H.J.C. van der Spoel D. van Drunen R. Comput. Phys. Commun. 1995; 91: 43-56Crossref Scopus (7194) Google Scholar, 16Lindahl E. Hess B. van der Spoel D. J. Mol. Model. 2001; 7: 306-317Crossref Google Scholar) running on a dual-CPU Linux work station. MD simulations were performed in the presence of 4895 water molecules of simple point charge as solvent. Periodical conditions were imposed for the triclinic simulation box. The GROMOS-96 43a1 force field was used for the Fre·FMN simulation, whereas the GROMACS force field was used for the Fre·FAD simulation. After the initial energy minimization and subsequent relaxation runs, the system was allowed to further equilibrate for another 800 ps with Berendsen-type temperature (300 K) and pressure (1.0 atm) coupling to an external bath. A 1-ns trajectory was recorded for each of the Fre·FAD and Fre·FMN runs. Hydrogen bonding analyses between Fre and the flavin substrate were performed. A hydrogen bond is registered when the distance between the hydrogen bonding donor (OH and NH) and acceptor (O: and N:) is less than 0.25 nm. Root mean square deviation clustering analysis was performed every 5 ps along the trajectory to locate the representative structure for the complexes. In the full-linkage clustering algorithm, a structure is added to the existing cluster when its distance to any element of the cluster is less than the 0.1-nm cutoff. The 5-ps structures were categorized according to their root mean square deviation values. A structure from the middle of the root mean square deviation distribution was extracted from the MD trajectory as the representative configuration. The inhibitory effects of FAD on Fre activity were examined in the presence of NADPH and FMN. Reciprocal initial velocities were plotted against the reciprocal substrate (NADPH or FMN) concentrations at various fixed concentrations of FAD to determine the nature of the inhibitions. Inhibition constants were then determined from the Michaelis-Menten plots fitted with Equation 3 (for competitive inhibition) and Equation 4 (for mixed inhibition) Vo=Vmax[S]KM1+[I]Ki+[S]Equation 3 Vo=Vmax[S]KM1+[I]Ki+1+[I]Ki′[S]Equation 4 using the GraFit 5.0 program (Erithacus Software Ltd.). The apparent kinetic parameters of Fre for the NADPH-FMN substrate pair were also determined from Michaelis-Menten plots fitted with the following equation Vo=Vmax[S]KM+[S]Equation 5 using the GraFit program. Forty-five mg of Fre was purified from 289 mg of protein in the cell extracts. The protein was purified to apparent homogeneity. The purified Fre was acidified and analyzed by ESI-MS; the molecular weight of Fre was determined to be 26,115 ± 5, which is practically identical to Fre's theoretical molecular weight of 26,111 calculated from the amino acid sequence. No other major molecular mass was detected by ESI-MS, further validating the purity of the Fre preparation. The purified Fre had a specific activity of 69,230 ± 295 units·mg−1 when reducing FMN with NADH as the electron donor at 30 °C. The highly concentrated and pure Fre had a pale yellow color, suggesting the possibility of flavins being bound to Fre. The specific flavin bound to Fre was further tested. When 200 μg (∼8 nmol) of Fre was loaded onto an HPLC size-exclusion column, we detected a small amount of flavin co-eluted with the Fre based on the absorption spectrum recorded by the photodiode array detector (Fig.1). The flavin peak was centered at the beginning of the protein peak, consistent with the slightly larger molecular weight of the Fre·flavin complex as compared with that of Fre alone. After integrating the peak area of the flavin peak that co-eluted with Fre and comparing it with the peak area of free FMN standards, our data showed that ∼0.13 nmol equivalent of FMN was co-eluted with Fre. When Fre was loaded onto a C18 reverse phase HPLC column, the Fre-bound flavin was dissociated from the protein, with a retention time of 9.938 min (Fig.2). A typical absorption spectrum of a flavin molecule was clearly detected (Fig. 2, inset). The respective retention times of free FAD, FMN, and Rfl standards were 9.985, 10.155, and 10.898 min, suggesting that the Fre-bound flavin is possibly FAD. This possibility was further examined by loading 25 μl of a mixture of Fre (0.14 μm) with either FMN or FAD (0.24 μm) onto the HPLC size-exclusion column. About 25% of FAD in the Fre-FAD mixture was co-eluted with Fre, but only 2.2% of FMN in the Fre-FMN mixture was co-eluted with Fre.Figure 2HPLC analysis of Fre by a C18reverse phase column. Fre sample (80 μg) was eluted from the column with a 33-ml linear gradient of acetonitrile in 0.1% trifluoroacetic acid (0–70%) with a flow rate of 1 ml·min−1. A minor peak with maximal absorption at 450 nm (dashed line) was eluted from the column at 9.938 min, whereas a major peak with maximal absorption at 280 nm (solid line) was eluted from the column at 26.0 min. Inset,absorption spectrum of the 9.938-min peak with maximal absorptions at 372 and 450 nm.View Large Image Figure ViewerDownload (PPT) Aliquots of FAD solution were added to a 48 μm Fre solution in 20 mm KPi (pH 7.4) buffer. The binding of FAD to Fre quenched the fluorescence of Fre, resulting in a gradual decrease in the fluorescence intensity of Fre (Fig. 3,inset). The K d for FAD was determined to be 29 ± 9 nm after plotting the fluorescence intensity of Fre against the ratio of the added FAD concentration to the initial Fre concentration (Fig. 3). Using a similar methodology, the K d for FMN was determined to be 1.5 ± 0.2 μm. The lowK d value for FAD is not consistent with the previous hypothesis that the isoalloxazine rings of the flavin substrates provide the major determinant for the binding of the three flavins to Fre and that the flavins have similar affinity for Fre (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar). We hypothesized that the ribityl moiety and the ADP moiety of FAD might interact with other structural elements in Fre and provide extra stability for the Fre·FAD complex. To test this hypothesis, the structure of the Fre·FAD complex was modeled by MD simulations. MD simulations showed that FAD could adopt an unusual bent conformation (Fig. 4 A). Amino acid residues located in the loop connecting Fβ5 and Fα1 in the flavin-binding domain (e.g. Gly65 and Asn70) and at the proposed NAD(P)H-binding site (e.g. Thr112, Gln143, and Arg202) were shown to form additional H-bonding with the ribityl side chain and the ADP moiety of FAD (Fig. 4 B). Specifically, the MD trajectory showed that there were 8.03 ± 2.13 (3,600 (average) 0.5-ps snapshots of a 900-ns trajectory with S.D.) hydrogen bonds between the Fre protein matrix and the FAD substrate, whereas there were only 4.75 ± 1.72 hydrogen bonds between Fre and FMN. The number of hydrogen bonds appeared to vary with time, reflecting the dynamic nature of the Fre·FAD complex. This is illustrated in Fig. 4 B, in which we show two snapshots of the hydrogen-bonding network around FAD. The computer modeling of the Fre·FAD complex suggested that the ADP moiety of FAD interacts dynamically with amino acid residues located at the proposed NAD(P)H-binding site. This hypothesis was further examined. TheK d value for FAD was determined to be 83 ± 25 nm in the presence of 0.18 mm AMP, a known competitive inhibitor of Fre with respect to NAD(P)H (with reportedK i values ranging from 0.3 to 0.5 mm) (2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). This K d value is almost 3 times larger than the K d value determined in the absence of AMP. The effect of FAD as an inhibitor for Fre activity was studied using NADPH as the electron donor and FMN as the major electron acceptor. It has been reported that the K m,NADPH of Fre is 14,000 μm with a k cat of 16 s−1 if FAD is the sole electron acceptor (5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). Because the highest concentration of NADPH used in our inhibitory study was 415 μm, Fre did not have any detectable activity for FAD reduction due to the large K m,NADPH value. Because FAD was practically not a substrate for Fre under the assay conditions, its inhibitory effect on FMN reduction with NADPH was studied. Fre activity was determined as a function of NADPH concentration (83–415 μm) in the presence of a fixed concentration of FMN (10 μm) at several concentrations of FAD. The double-reciprocal plots showed a series of lines converged at a point to the left of the y axis (Fig.5 A), indicating that FAD is a mixed-type inhibitor with respect to NADPH with a K iof 0.43 ± 0.07 μm and a Ki' of 0.59 ± 0.17 μm; the apparent K m,NADPH was 418 ± 41 μm. When Fre activity was determined as a function of FMN concentration (2–10 μm) in the presence of a fixed concentration of NADPH (200 μm) at several concentrations of FAD, FAD exhibited a typical competitive inhibition effect (Fig 5 B), with a K i of 0.10 ± 0.01 μm and an apparent K m,FMN of 2.0 ± 0.2 μm. The effect of FAD on the activity of Fre for FMN reduction with NADH as the electron donor was also studied. Because FAD reduction by Fre is much slower than FMN reduction when NADH is less than 100 μm (5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar), the effect of FAD was studied by measuring the Fre activity under a fixed concentration of NADH (80 μm) and FMN (10 μm), with various FAD concentrations. FAD at 1 μm reduced the rate of NADH consumption of Fre by 13%, and at 10 μm, FAD reduced the rate by 45% (Fig.6). The decrease was shown to be due to the preferential reduction of FAD when the amount of produced FADH2 was measured by HpaB, an FADH2-utilizing 4-hydroxyphenylacetate 3-monooxygenase that does not use FMNH2 (11Xun L. Sandvik E.R. Appl. Environ. Microbiol. 2000; 66: 481-486Crossref PubMed Scopus (90) Google Scholar). HpaB was used in excess to ensure ultimate usage of FADH2 produced in the assay. After a 30-min incubation, 1.6 nmol of NADH was consumed, and 1.7 ± 0.1 nmol (average of three samples ± S.D.) of 3,4-dihydroxyphenylacetate was produced with FAD alone in the assay; whereas 1.4 ± 0.1 nmol of the product was formed in the presence of equal concentrations of FMN and FAD. Therefore, about 82% of the NADH was used to reduce FAD under the assay conditions containing 5 μm each of FAD and FMN, with the remainder used for FMN reduction. Fre had a much higher affinity for FAD than for FMN. TheK d value for the Fre·FAD complex (29 ± 9 nm) was 52 times lower than that of FMN (1.5 ± 0.2 μm). The K d values for Rfl and lumichrome, a flavin analog, are 3.6 and 0.5 μm (17Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1996; 271: 16656-16661Abstract Full Text Full Text PDF PubMed Scopus (29) Google Scholar, 18Nivière V. Vanoni M.A. Zanetti G. Fontecave M. Biochemistry. 1998; 37: 11879-11887Crossref PubMed Scopus (28) Google Scholar). Thus, among its flavin substrates, Fre binds FAD the most tightly. The tight binding was also observed by the co-elution of FAD with Fre from the HPLC size-exclusion chromatography (Fig. 1). Results of the MD simulations (Fig. 4 A) showed that not only the isoalloxazine ring of FAD but also the ribityl side chain and the ADP moiety were involved in binding to Fre, offering theoretical support for tighter binding for FAD than for FMN of Rfl. On the basis of the crystallized Fre·Rfl structure, which shows only one direct H-bond from the 2′-OH of the ribityl side chain to the carbonyl oxygen of Pro47(on Fβ4) (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar), and the similar K m values for Rfl, FMN, and FAD, it has been hypothesized that Fre binds Rfl, FMN, and FAD mainly through the isoalloxazine ring (1Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1987; 262: 12325-12331Abstract Full Text PDF PubMed Google Scholar, 13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar). The proposed interaction is real for Rfl and FMN, but not for FAD. The MD simulations predicted the formation of 3.28 extra H-bonds on average from the ribityl side chain and ADP moiety of FAD with Fre (Fig. 4 B). Considering the energy of a typical H-bond to be 2.9–7.2 kcal/mol (19Voet D. Voet J.G. Biochemistry. 2nd Ed. John Wiley & Sons, New York1995: 176Google Scholar), the extra H-bonds in the Fre·FAD complex contribute to a significant stabilization energy (enthalpy) of 9.5–23.6 kcal/mol, in comparison with the Fre·FMN complex. This stabilization energy should be viewed as an upper bound, because entropy (e.g. the bent conformation and larger size of the FAD can increase the entropy of the Fre·FAD complex) will reduce the stabilization. This view is consistent with the extra stabilization energy (free energy) of 2.3 kcal/mol calculated from the determined K d values of the Fre·FAD complex relative to the Fre·FMN complex. The bent conformation of FAD predicted by MD simulations is similar to the conformation adopted by the FAD prosthetic group in the E. coli flavodoxin reductase, in which FAD exhibits a U-shaped conformation (20Ingelman M. Bianchi V. Eklund H. J. Mol. Biol. 1997; 268: 147-157Crossref PubMed Scopus (122) Google Scholar). Interestingly, the flavodoxin reductase also lacks a full-size adenosine-interacting loop connecting Fβ5 and Fα1, as does Fre. Structure-based sequence alignment of Fre with the flavodoxin reductase shows that two flavodoxin reductase residues that H-bond with the ribityl side chain and the adenine ring of the FAD prosthetic group (Arg50 and Thr116) are conserved in Fre (Arg46 and Thr112) (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar, 20Ingelman M. Bianchi V. Eklund H. J. Mol. Biol. 1997; 268: 147-157Crossref PubMed Scopus (122) Google Scholar). The hydrogen bonding of Thr112 with FAD is captured by snapshot of MD simulations (Fig. 4), whereas the dynamic interaction of Arg46 with FAD is not presented in the snapshot. The flavodoxin reductase has a Trp248 residue at the C terminus, which provides additional interaction with the adenine ring of FAD. Such a Trp residue is not present in the Fre C terminus and may partly explain why Fre does not bind FAD permanently, whereas the flavodoxin reductase does. Unlike the experimentally verified flavin-binding site, the NAD(P)H-binding site of Fre is only a proposed model (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar) based on comparisons with different ferredoxin:NADP+ reductase·AMP complexes (21Serre L. Vellieux F.M. Medina M. Gomez-Moreno C. Fontecilla-Camps J.C. Frey M. J. Mol. Biol. 1996; 263: 20-39Crossref PubMed Scopus (133) Google Scholar, 22Karplus P.A. Daniels M.J. Herriott J.R. Science. 1991; 251: 60-66Crossref PubMed Scopus (462) Google Scholar). The adenine ring and the ribose group of the 2′-phospho-ADP moiety of NADPH are proposed to bind to Fre near the amino ends of the α-helices of the NAD(P)H-binding domain, and Arg202 and His144 may bind the pyrophosphate group of NAD(P)H (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar). Interestingly, our MD simulations also predicted amino acid residues at these regions, such as Gln143 and Arg202, to form H-bonds with the ADP moiety of FAD (Fig.4 B). The overlapping binding sites of FAD and NAD(P)H were supported by several lines of results. Firstly, 0.18 mmAMP, a known competitive inhibitor with respect to the NAD(P)H (2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar), increased the K d value of the Fre·FAD complex from 29 nm to 83 nm. In comparison, NADP+ or 3-aminopyridine adenosine dinucleotide phosphate, an NADPH analog, has no effect on the K d of Fre for Rfl (18Nivière V. Vanoni M.A. Zanetti G. Fontecave M. Biochemistry. 1998; 37: 11879-11887Crossref PubMed Scopus (28) Google Scholar). AMP affects the binding of FAD to Fre probably by competing with the ADP moiety of FAD for the NAD(P)H-binding site, leading to an increased K d. Secondly, kinetic analysis showed that FAD behaves as a competitive inhibitor with respect to FMN and a mixed-type inhibitor with respect to NADPH (Fig. 5). Competitive inhibition of FAD upon FMN is expected because FAD should compete with FMN for the flavin-binding site. However, the mixed-type inhibition upon NADPH supports the MD simulation model. Lumichrome, an Fre-competitive inhibitor with respect to Rfl, is an uncompetitive inhibitor with respect to NADPH (2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar). Hence, lumichrome does not inhibit the formation of Fre·NADPH complex. The interactions between the ADP moiety of FAD and residues at the NAD(P)H-binding site may inhibit the formation of Fre·NADPH complex, leading to the observed mixed-type inhibitory effect. The bent conformation of FAD also provides a plausible explanation for the extremely large K m,NADH (301 μm) and K m,NADPH (14,000 μm) values during FAD reduction, in comparison with the correspondingK m values for Rfl and FMN reduction (1Fontecave M. Eliasson R. Reichard P. J. Biol. Chem. 1987; 262: 12325-12331Abstract Full Text PDF PubMed Google Scholar, 2Fieschi F. Nivière V. Frier C. Decout J.-L. Fontecave M. J. Biol. Chem. 1995; 270: 30392-30400Abstract Full Text Full Text PDF PubMed Scopus (95) Google Scholar, 5Nivière V. Fieschi F. Decout J.-L. Fontecave M. J. Biol. Chem. 1999; 274: 18252-18260Abstract Full Text Full Text PDF PubMed Scopus (38) Google Scholar). Because FAD binds tightly to Fre through both the isoalloxazine ring and the ADP moiety (Fig. 4), NADH has to compete with the ADP moiety of FAD for the binding site. Consequently, a higher concentration of NADH is required to outcompete the ADP moiety of FAD for the NAD(P)H-binding site, leading to a larger K m,NADH value than that for Rfl and FMN reduction. In the bent conformation, the negatively charged pyrophosphate group of FAD is also located in close proximity to the proposed binding site for the 2′-phosphate group of NADPH (13Ingelman M. Ramaswamy S. Nivière V. Fontecave M. Eklund H. Biochemistry. 1999; 38: 7040-7049Crossref PubMed Scopus (77) Google Scholar) (Fig. 4 A). The highly negatively charged pyrophosphate group may destabilize the binding of NADPH through repulsion with the negatively charged 2′-phosphate group of NADPH. Thus, theK m,NADPH value is 46 times larger than theK m,NADH when FAD is the electron acceptor. The single phosphate group of FMN may also destabilize NADPH binding to Fre by the same mechanism, resulting in a K m,NADPH of 418 ± 41 μm (calculated from Fig. 5); whereas theK m,NADH is merely 14.9 ± 0.8 μmduring FMN reduction (data not shown). The high affinity of Fre for FAD may imply that the in vivoFre activity is much lower than previously expected under aerobic conditions. NADH and NADPH concentrations inside aerobically growingE. coli have been reported in the range of 20 and 150 μm, respectively (23Penfound T. Foster J.W. Neidhardt F.C. Escherichia coli and Salmonella Cellular and Molecular Biology. ASM Press, Washington, D. C.1996: 721-730Google Scholar). Intracellular free Rfl, FMN, and FAD concentrations have not been clearly defined. Only one study reports the internal free FAD concentration of Amphibacillus xylanus to be 13 μm (24Ohnishi K. Nimura Y. Yokoyama K. Hidaka M. Masaki H. Uchimura T. Suzuki H. Uozumi T. Kozaki M. Komagata K. Nishino T. J. Biol. Chem. 1994; 269: 31418-31423Abstract Full Text PDF PubMed Google Scholar). Intracellular Rfl concentration should be much lower than those of FMN and FAD because Rfl should be transformed into the coenzyme forms FMN and FAD in order to fulfill its metabolic purpose (25Bacher A. Eberhardt S. Richter G. Neidhardt F.C. Escherichia coli and Salmonella Cellular and Molecular Biology. ASM Press, Washington, D. C.1996: 657-664Google Scholar). Assuming the internal free FMN and FAD concentrations to be in the order of ∼10 μm, then in vivo Fre activity should be mainly due to the preferential reduction of FAD by NADH as demonstrated by our in vitro assay (Fig. 6) and the Fre-HpaB coupled assay. Fre literally cannot use NADPH to reduce FAD in vivo because theK m,NADPH value is much higher than the internal NADPH concentration. Thus, the in vivo Fre activity should be suppressed due to the low internal NADH concentration in comparison with the K m,NADH value. The low in vivoFre activity is advantageous to the organism because this prevents unrestrained production of FADH2, wasteful reoxidation of the labile FADH2, and excessive production of detrimental reactive oxygen species such as H2O2 (26Gibson Q.H. Hastings J.W. Biochem. J. 1962; 83: 368-377Crossref PubMed Scopus (128) Google Scholar). In summary, this study demonstrated that FAD binds more tightly to Fre than either FMN or Rfl does. The tight binding likely makes FAD the preferred substrate under in vivo conditions (Fig. 6). Experimental data and MD simulations suggested that the tight binding is due to an unusual bent conformation of FAD, allowing additional interactions between FAD and Fre. The ADP moiety of FAD likely competes with NAD(P)H for binding to Fre, explaining the largeK m values for NADH and NADPH during FAD reduction.E. coli is a facultative anaerobe, growing both aerobically and anaerobically. When E. coli grows with sufficient O2 supply, the intracellular NADH concentration is about 20 μm, or 2% of total NADH and NAD+ (23Penfound T. Foster J.W. Neidhardt F.C. Escherichia coli and Salmonella Cellular and Molecular Biology. ASM Press, Washington, D. C.1996: 721-730Google Scholar). The low NADH concentration supports very low Fre activity for FAD reduction, and FAD also inhibits FMN reduction by Fre. Thus, our data suggest that Fre activities are likely suppressed in aerobically growing E. coli cells. We thank Dr. Vincent Nivière of the Université Joseph Fourier for valuable discussions, Dr. Yong Liu of Washington State University for assistance in the ESI-MS experiment, and Christopher M. Webster for assistance in experiments.
The surface charge of nanoparticles or nanocarriers (NCs) plays a critical role in the vehicle function, distribution, and fate in drug delivery applications. Flash NanoPrecipitation (FNP) is a platform for producing block copolymer-stabilized NCs. We show that NC charge (measured as the ζ-potential) can be continuously tuned from +40 to −40 mV by using blends of neutral poly(styrene)-block-poly(ethylene glycol) (PS-b-PEG) with polyelectrolyte block copolymers, anionic poly(styrene)-block-poly(acrylic acid) (PS-b-PAA) or cationic poly(styrene)-block-poly(N,N-dimethylaminoethyl methacrylate) (PS-b-DMAEMA), while simultaneously controlling NC diameters between 40 and 180 nm. The dense polymer brush on the surface of these FNP NCs provides a better platform to test NC surface charge effects on cellular interactions than NC systems in which charge is applied onto hydrophobic surfaces. NC charge gradually increases as more PS-b-PEG is replaced with a polyelectrolyte stabilizer, where sparsely substituted NCs (1–20 wt %) have nearly neutral (|ζ| < 5 mV) followed by a region where ζ-potential increases with increasing polyelectrolyte substitution. The protein binding to negatively charged NCs is low and equivalent to the adsorption on PEG-coated NCs, which are normally considered as the gold standard in "stealth" low protein adsorbing surfaces. In contrast, as little as 1 mol % cationic polymer produces strong protein adsorption, and cellular uptake, even though the ζ-potentials are still near zero, |ζ| < 5 mV. Binding of the NCs to Tib67, HEK293T, and HepG2 cells is distinct. While cationic NCs are taken up by all cell lines, anionic NCs are only taken up by the macrophage-like Tib67 cells. These results are discussed in terms of the protein corona differences on the NCs and the receptor differences between these cell lines. This study shows that ζ-potential alone is inadequate to predict the biological identity of an NC formed by protein corona adsorption and interactions with different types of cells.
The on- and and off-time distributions from fluorescence single-molecule experiments are widely used to extract kinetics parameters with the goal to provide a quantitative description for the molecule's behavior on the ensemble level. Such experiments are inevitably influenced by photobleaching, where the fluorescent probe transitions to a nonemissive state. Yet, it appears that few reports went beyond acknowledging this unavoidable complication; in fact, it has so far been ignored when evaluating off-time distributions. Here, we present a theoretical framework that allows the derivation of analytical equations in which photobleaching kinetics are rigorously incorporated. Unexpectedly, our results indicate that the off-time distribution should be nonexponential even when all the rate processes are single exponential. With the analytical theory understood and demonstrated as easy to implement, such ubiquitous photochemical processes can now be readily included in routine experimental analyses.
The response of solvent to the change of charge or dipole of solute molecules has been intensely studied in recent years 1 . In previous solvation experiments, time dependent fluorescence Stokes shift of dye molecules in different solvents were measured, from which the solvation time for the solvents were determined 1 , 2 . Various theories, from the simple dielectric continuum model to instantaneous solvent normal mode analysis, have been used to relate solvent motions to solvation time 3 , 4 . MD simulations have also been carried out to understand the nature of these solvent motions in the solvation process 5 . However, these time dependent Stokes shift experiments, which measure the solute fluorescence, can only provide an indirect microscopic picture of the relevant solvent motions during the solvation process.
In this work, we demonstrate for the first time massively parallelizable nanoplasmonic structures and integration of electronics in the same substrate in CMOS. We adopt the same "fabless" approach in today's semiconductor industry with absolutely "no change" of either fabrication or processing and show that copper interconnects in an industry standard digital CMOS process (65 nm node) can be exploited to allow subwavelength optical field processing in a massively scalable fashion. We demonstrate this in the context of eliminating all external optics and enabling the first optics-free fully integrated CMOS fluorescence-based biosensor array. The system has massively multiplexed biomolecular sensing capability for DNAs with surface sensitivity comparable to commercial fluorescence readers. The angle and scattering insensitive nature of the filter, relying on coupled surface-plasmon polariton modes, allows us to eliminate all external optics and miniaturize the entire 96-sensor array system (including a LED source) within 0.1 cc of volume. The system demonstrates detection sensitivity of less than 1 molecule/μm2 or zepto moles of quantum dot based fluorophores on the chip surface. The electronic–nanophotonic codesign approach allows us to optimally partition optical and electronic filtering, enabling us to detect fluorescence signal 77 dB lower than the excitation. Such CMOS-based nano-optical systems can lead to novel chip-scale optical sensors for in vitro and in vivo applications.
A nanoscale heat source suspended in fluids constitutes a highly localized yet mobile system that is far from equilibrium. Remarkably, its translational and rotational dynamics can still be theoretically described by Brownian-type equations of diffusion, a "hot Brownian motion" framework (HBM), while the original formulation of diffusive dynamics premises a system that is at or near thermal equilibrium. The HBM theory predicts a steeper temperature dependence for the nanoscale heat source's rotational dynamics over its translational movements─a breakdown of the equipartition principle. Here, we present the first experiment that consistently assessed the HBM prediction by evaluating the diffusivities resulting from both types of motion on an equal footing. We simultaneously tracked the dynamics of all six translational and rotational degrees of freedom for single gold nanoparticles after laser-induced temperature jumps up to ∼30 K above ambient. Without the need for adjustment parameters, the experimental data were recapitulated by the HBM theory across a panel of particle sizes and heating-laser intensities. Our results thus corroborated the translation-rotation diffusivity divergence predicted by the theory, solidifying its underlying microscopic picture which is expected to have important implications in such applications as photothermal imaging, molecular thermobiology and biophysics, nonequilibrium physics and active matters, as well as chemical dynamics, to name a few.
The identification of the intermediates observed in bond activation reactions involving organometallic complexes on time scales from femtoseconds to milliseconds has been accomplished through the use of ultrafast infrared spectroscopy. C—H bond activation by the molecule Tp * Rh(CO) 2 showed a final activation time of 200 ns in cyclic solvents, indicating a reaction barrier of 8.3 kcal/mol. An important intermediate is the partially dechelated η 2‐Tp * Rh(CO)(S) solvent complex, which was formed 200 ps after the initial photoexcitation. Si—H bond activation by CpM(CO) 3 (M=Mn, Re) showed some product formation in less than 5 ps, indicating that the Si—H activation reaction is barrierless. The activated product was formed on several timescales, from picoseconds to nanoseconds, suggesting that there are different pathways for forming final product which are partitioned by the initial photoexcitation.
Affinity-based fluorescence sensing has been one of the key enabling technologies in biomolecular sensing, used for detection of proteins, DNAs, toxins, bacteria, etc, and remains one of the most sensitive, specific, robust, and widely used diagnostics methodology [1-4]. In absence of high-performance integrated optical filters, miniaturization of a fluorescence sensing system in CMOS has relied on time-resolved techniques with synchronized sources or externally grown optical filters and/or collimators. This paper presents a nanophotonic-electronic co-design approach towards fully-integrated fluorescence biosensor with on-chip copper-interconnect based nanoplasmonic filters. The filters demonstrate a measured extinction ratio of greater than 51dB in the excitation/emission bands for a class of quantum-dot based fluorescence tags. Integrated with these filters, the sensor platform is a correlated double sampling architecture which achieves femtowatt photon sensitivity. Detection sensitivity of 47 zeptomoles of quantum-dots was experimentally demonstrated, making the chip a low-cost, fully integrated, high-performance, and fully scalable biosensor for point-of-care applications.
Subsurface biobarriers can be conceived to attenuate the migration of pathogens by adhesion to mineral surfaces. Candidate biobarrier materials of varied surface characteristics (dolomite, α-alumina, silica, pyrophyllite, and Pyrax (a composite form of pyrophyllite, mica, and silica)) were tested for Escherichia coli adhesive capacity in macroscale continuous-flow columns. Atomic force microscopy (AFM) was used to determine nanoscale interaction energies. Predicted attractive interaction energies correlated well with macroscale adhesive behavior for tested E. coli strains. AFM measurements confirmed ExDLVO model predictions of attachment in the primary minima for E. coli O157:H7 and two environmental isolates E. coli (UCFL-339 and UCFL-348) with MOPS conditioned Pyrax. In macroscale column experiments, pyrophyllite and Pyrax demonstrated significantly higher bacterial retention, higher deposition coefficients and lower initial cell breakthrough values for E. coli O157:H7 than did α-alumina, silica, or dolomite (pyrophyllite, 0.93, 3.56 h-1, 3.2% ODo; Pyrax, 0.95, 3.73 h-1, 2.8% ODo; α-alumina, 0.74, 1.60 h-1, 33% ODo; silica, 0.63, 0.43 h-1, 73% ODo; and dolomite, 0.33, 0.17 h-1, 89% ODo, respectively). Bacterial hydrophilicity impacted cell retention in Pyrax columns with the relatively hydrophobic E. coli isolate UCFL-339 (0.99, 6.13 h-1, 0.4% ODo) retained better than the more hydrophilic E. coli isolate UCFL-348 (0.94, 3.70 h-1, 3.6% ODo). The strong adhesive behavior of Pyrax was attributed to the hydrophobic (ΔGiwi = −32.4 mJ/m2) pyrophyllite component of the mineral. Vicinal water appears poised between the bacterial and the mineral surface during initial attachment. Overall, observed behavior of the various E. coli strains and the selected mineral surfaces was consistent with surface analyses, conducted at both the macro- and nanoscale.
'/%3e%3cuse xlink:href='%23a' transform='rotate(23.036 .093 25.536)'/%3e%3cuse xlink:href='%23a' transform='rotate(45.87 1.273 16.18)'/%3e%3cuse xlink:href='%23a' transform='rotate(69.945 .996 12.078)'/%3e%3cuse xlink:href='%23a' transform='rotate(20.66 -19.689 31.932)'/%3e%3c/svg%3e)
'/%3e%3cuse xlink:href='%23a' transform='rotate(60)'/%3e%3ccircle r='17' fill='%23000095'/%3e%3ccircle r='15' fill='%23fff'/%3e%3c/svg%3e)