ARCHIVES OF BIOCHEMISTRY AND BIOPHYSICS

Vol. 337, No. 1, January 1, pp. 89–95, 1997Article No. BB969746

Vibrio harveyi NADPH:FMN Oxidoreductase: Preparationand Characterization of the Apoenzyme andMonomer–Dimer Equilibrium1

Mengyao Liu,*,2 Benfang Lei,* Qizhu Ding,*,3 J. Ching Lee,† and Shiao-Chun Tu*,‡,4

Departments of *Biochemical and Biophysical Sciences and ‡Chemistry, University of Houston, Houston, Texas 77204-5934; and †Department of Human Biological Chemistry & Genetics, The University of Texas Medical Branch, Galveston,Texas 77555-1055

Received June 26, 1996, and in revised form September 30, 1996

with dissociation constants of 1.8 and 3.3 mM, respec-tively. q 1997 Academic Press, Inc.A rapid chromatography method was developed for

Key Words: NADPH:FMN oxidoreductase; flavin re-the preparation of apoenzyme of Vibrio harveyiductase; apoenzyme; dimerization.NADPH:FMN oxidoreductase with §80% yields. The

apoenzyme bound one FMN per enzyme monomer witha dissociation constant of 0.2 mM at 237C. The reconsti-tuted holoenzyme was catalytically as active as the na-

NAD(P)H:flavin oxidoreductases (flavin reductases)tive enzyme. FMN binding resulted in 87 and 92% ofcatalyze the reduction of flavin by NAD(P)H. This classquenching of protein and flavin fluorescence, respec-of enzymes has been proposed to function in the activa-tively, indicating a conformational difference betweention of ribonucleotide reductase (1, 2) and chorismatethe apoprotein and the holoenzyme. Neither riboflavinsynthase (3), reduction of methemoglobin (4, 5), reduc-nor FAD showed any appreciable binding to the cofac-tive iron release from siderophores (6, 7), and oxygentor site of the apoenzyme but both flavins were activeactivation (8). Flavin reductases are also believed tosubstrates for the FMN-containing holoenzyme. 2-provide reduced riboflavin 5*-phosphate (FMNH2)5 inThioFMN bound to the cofactor site of the apoenzymevivo as a substrate for the luciferase-catalyzed biolumi-with an affinity similar to that for FMN binding. Thenescence in luminous bacteria (9–12).holoenzyme reconstituted with 2-thioFMN showed a

509-nm absorption peak, which represents a 19-nm red Free reduced flavin is quite unstable due to rapidshift from the corresponding peak of the free flavin, autooxidation (13). In general, little is known aboutand was catalytically active in using either FMN or 2- how the reduced flavins produced by flavin reductasesthioFMN as a substrate. The holoenzyme showed a are efficiently coupled to other reactions in vivo or inconcentration dependence in molecular sieve chroma- vitro. A NADPH:FMN oxidoreductase (flavin reductasetography corresponding to higher apparent molecular P or FRP) has been isolated some years ago from theweights at higher concentrations. Both the holoen- luminous bacterium Vibrio harveyi (14). We have tar-zyme and the apoenzyme was shown at 47C by equilib- geted this FRP-bacterial luciferase system for detailedrium ultracentrifugation to undergo dimerization investigations regarding the mechanism of interen-

zyme flavin transfer. The genes encoding V. harveyiFRP (15) and luciferase (16) have both been cloned and

1 This work was supported by Grants GM25953 from National In- overexpressed, thus greatly facilitating biochemical,stitutes of Health and E-1030 from The Robert A. Welch Foundation biophysical, and molecular biological studies on theseto S-C.T., and the sedimentation study was supported by Grants H- two enzymes. The crystal structures of V. harveyi lucif-0013 and H-1238 from The Robert A. Welch Foundation and

erase (17) and FRP (18) have also been solved, provid-RR08961 from National Institutes of Health to J.C.L.ing an invaluable structural basis for interpretation of2 Present address: Department of Medicine, Baylor College of Med-

icine, Houston, TX 77030.3 Present address: Department of Chemistry, University of Al-

5 Abbreviations used: FMNH2, fully reduced riboflavin 5*-phos-berta, Edmonton, Alberta, Canada T6G 2G2.4 To whom correspondence should be addressed at Department of phate; FRP, NADPH-FMN oxidoreductase or flavin reductase P;

FRPS, FRP derivative reconstituted from apoenzyme and 2-thi-Biochemical and Biophysical Sciences, University of Houston, Hous-ton, TX 77204-5934. Fax: (713) 743-8351. E-mail: dtu@uh.edu. oFMN.

890003-9861/97 $25.00Copyright q 1997 by Academic Press, Inc.All rights of reproduction in any form reserved.

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

90 LIU ET AL.

onto a Sephadex G-25 column (1 1 30 cm) preequilibrated and elutedexperimental findings at molecular levels and rationalwith 50 mM Pi containing 6 M urea and 1 mM dithiothreitol. To adesign of new experiments. The FRP not only uses fla-200-ml beaker containing 150 ml of 50 mM Pi , the pool of protein peakvin as a substrate but also binds FMN as a cofactor (Ç3 ml) was added with gentle stirring. After 15 min, the refolded

(15) to mediate a ping-pong reaction in a single-enzyme apoenzyme was concentrated to 2 ml by the following procedures.The sample was first loaded on a 0.51 2-cm DEAE-Sepharose columnspectrophotometric assay (11). Strikingly, a sequentialpreequilibrated with 50 mM Pi and then eluted with 20 ml of the samekinetic mechanism was observed in the FRP–lucifer-buffer to remove urea in the column with the apoenzyme remainingase coupled assay monitoring light emission (19). Onbound. The apoenzyme was recovered by elution with 0.5 M Pi . Frac-

the basis of these initial kinetic results, we have pro- tions containing most of the apoenzyme were pooled and passedposed an FRP–luciferase coupling mechanism in which through a Sephadex G-25 column (1 1 30 cm) preequilibrated and

eluted with a desired buffer (0.2 M Pi in this work).the reduced FMN cofactor, which is less sensitive toReconstitution of FRP from apoenzyme and flavins. Reconstitu-autooxidation, rather than the reduced flavin product

tion of holoenzyme was carried out by incubating the apoenzymeof FRP is directly transferred to luciferase for the biolu-with 10 times molar excess FMN or 2-thioFMN for 10 min. Theminescence reaction (19). To further test this working sample was then applied on a Sephadex G-25 column (1 1 30 cm)

hypothesis, it is essential to characterize the nature preequilibrated and eluted with 50 mM Pi to obtain the reconstitutedof interactions of FRP with flavin as a cofactor and a enzyme free from unbound flavin.substrate. Toward these goals, this work first focused Molecular sieve chromatography. FRP holoenzyme, at varying

concentrations, was subjected to molecular sieve chromatographyon the development of a method for the preparation ofusing a Pharmacia FPLC system equipped with a FPLC SephadexFRP apoenzyme in high yield and quality. Subse-75 column. The column was equilibrated and eluted with 50 mM Piquently, the interactions of FMN and 2-thioFMN with at a flow rate of 1.0 ml/min. The column was calibrated by blue

the apoenzyme were characterized by equilibrium and dextran, FMN, and protein standards bovine serum albumin (Mr

68,000), ovalbumin (Mr 43,000), and cytochrome c (Mr 12,380).spectral measurements. The catalytic activities of holo-Equilibrium ultracentrifugation of FRP holo- and apoenzymes.enzymes reconstituted with these two flavins were also

Equilibrium ultracentrifugation experiments were carried out atdetermined.30,000 rpm and 47C using a Beckman XL-A analytical ultracentri-The molecular weight (Mr) of monomeric FRP de-fuge and employing the high-speed, meniscus-depletion procedure

duced from the gene sequence is 26,312 (15). This is in (22). The loading protein concentrations were A280 Å 0.85 and 0.35good accord with the Mr 28,000 determined by SDS– for native and apoenzyme, respectively, in 200 mM Pi . Sedimentation

data were acquired at 4-h intervals until no detectable changes inPAGE but differs from the values of 33,000 to 40,000the distribution of protein concentration over at least 8 h were ob-revealed by earlier molecular sieve chromatographyserved, at which point at least five scans were taken and then aver-(14, 15). On the other hand, our recently completed aged. The data were fitted by nonlinear least squares to both a non

crystal structure of FRP shows that it is a homodimer associating and an associating model according to Eqs. [1] and [2],(18). These findings, taken together, prompted us to respectively,propose a monomer–dimer equilibrium for FRP in so-

Ar Å Aoexp[HrMV w(X2 0 X2o)] / E [1]lution. Molecular sieve and equilibrium ultracentrifu-

gation were also carried out in this work to investigate Ar Å Ao,1exp[HrM1(X2 0 X2o)]

the subunit interaction for the FRP apoenzyme and the/ (Ao,1)n

rKaexp[HrM1rn(X2 0 X2o)] / E, [2]native holoenzyme.

where Ar and Ao are the absorbances at radii X and Xo , respectively;H Å (1 0 vV r)v2/2RT; vV is the partial specific volume of the proteinMATERIALS AND METHODSand is assumed to be 0.74; r is the density of the solvent; v is theangular velocity of the rotor; MV w is the weight-average molecularMaterials. FMN, FAD, riboflavin, and NADPH were purchasedweight; M1 is the monomeric molecular weight; n is the stoichiome-from Sigma. Ultrapure urea was from Fisher. Guanidine hydrochlo-try; Ka is the association constant; and E is the baseline offset. Theride was obtained from Eastman. 2-ThioFMN was synthesized andvalidity of the model was determined by the randomness of residualspurified as described previously (20). FAD was purified by columnand ‘‘goodness of fit’’ calculated bychromatography (21). All phosphate (Pi) buffers used in this work

are at pH 7.0 and consisted of molar fractions of 0.39 sodium mono-base and 0.61 potassium dibase. 1

DOF∑iS residual

standard errorD2

,Purification and assay of FRP. V. harveyi FRP was purified to

apparent homogeneity following published procedures (15). FRP ac-tivities were measured spectrophotometrically by monitoring de-

where DOF is degree of freedom.creases in A340 associated with the oxidation of NADPH. The reac-Other measurements. Protein concentrations were determined bytions were initiated by adding enzyme into 1 ml of 50 mM Pi con-

the Lowry method (23) using bovine serum albumin as a standard.taining 0.16 mM NADPH and 0.05 mM FMN. One unit of enzymeFluorescence emission measurements were obtained at 237C usingactivity is defined as 1 mmol NADPH oxidized/min using e340 Å 6.22a Perkin–Elmer MPF-44 fluorescence spectrophotometer. Technical1 103 M01 cm01 for NADPH.emission spectra were collected without any corrections.Preparation of FRP apoenzyme. FRP apoenzyme was prepared

by removing FMN using molecule sieve chromatography under dena-turing conditions, refolding the FRP apoenzyme, and concentrating RESULTSthe sample by DEAE-Sepharose chromatography. Guanidine hydro-

Preparation of apoenzyme. During the course of thischloride (0.6 g) was added to 1 ml of FRP (5 mg/ml) containing 1 mM

dithiothreitol to denature the enzyme. The sample was then loaded study a number of procedures were tested for the prep-

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

91NADPH-SPECIFIC FLAVIN REDUCTASE APOENZYME AND SUBUNIT INTERACTION

fluorescence was quenched by 87% without any detect-able shift in the emission maximum (Fig. 1B).

Stoichiometry and Kd for FMN binding. Thequenching of protein fluorescence by FMN binding wasused to access the stoichiometry and dissociation con-stant (Kd) for FMN binding. In one experiment, apoen-zyme at 1 mM was titrated with varying amounts ofFMN. The protein fluorescence intensities at 330 nmwere determined and plotted against molar ratios of(FMN)/(enzyme). The extrapolated break point formaximal fluorescence change corresponds to a bindingof 0.85 FMN per apoenzyme monomer (Fig. 2A). Theseresults are consistent with the 1:1 binding of FMN perFRP monomer shown by measurements of protein and

FIG. 1. Fluorescence spectra of FRP holo- and apoenzymes. All FMN contents of native FRP (15) and the crystal struc-measurements were carried out at 237C in 0.2 M Pi , pH 7.0. (A) ture of dimeric FRP (18). In another titration experi-Emission spectra of 5 mM free FMN (solid curve) and 5 mM native ment, the protein fluorescence at 330 nm (excitationFRP (dotted curve; intensity multiplied by 10 for presentation) were

280 nm) of a constant and limiting amount of apoen-measured using 450-nm excitation. (B) Protein fluorescence emis-sions were measured using 280-nm excitation and 4.0 mM apoenzyme zyme was determined in the presence of varying con-without (solid curve) or with 10 mM FMN (dotted curve; intensity centrations of FMN. The reciprocal of DFluorescencemultiplied by 3 for presentation). The flavin-containing sample was was plotted against the reciprocal of FMN concentra-incubated at 237C for 10 min prior to spectral measurement.

tion where DFluorescence is the fluorescence intensityof the apoenzyme in the absence of flavin minus thatof the enzyme at a given FMN concentration. The disso-ciation constant for FMN binding can be calculated asaration of FRP apoenzyme. Dialysis of FRP holoenzymethe negative reciprocal of the intercept on the abscissa.against buffer containing 1 M KBr and increasing levelsA Kd of 0.2 mM was thus obtained at 237C from resultsof urea were found to be inefficient in removing theshown in Fig. 2B (l).FMN cofactor until the urea concentration reached 6

M or higher. Moreover, subsequent refolding of the apo- Flavin cofactor specificity of FRP. FRP activity wasenzyme by gradual removal of urea through dialysis recovered when the apoenzyme was mixed with FMNresulted in substantial protein precipitation. However,yields of soluble apoenzyme could be markedly in-creased if the protein sample in 6 M urea was subjectedto a single-step dilution to substantially and rapidlylower the urea concentration. Another important obser-vation was that the binding of apoenzyme by DEAE-Sepharose was very tight at low phosphate concentra-tions (e.g., 50 mM), but was greatly weakened at 0.5 M

Pi . These findings led us to develop a simple method,as described under Materials and Methods, for thepreparation of FRP apoenzyme with excellent yield andquality. Such a procedure produced apoenzyme with80% yields in just a few hours. The apoenzyme thusobtained was free from FMN (on the basis of no detect-able A445) and, as will be detailed below, can bind astoichiometric amount of FMN to generate a fully ac-tive holoenzyme. FIG. 2. Fluorometric titration of FRP apoenzyme with FMN and

2-thioFMN. (A) A constant amount of apoenzyme at 1 mM was titratedEffects of FMN binding to apoenzyme on protein andwith FMN in 0.2 M Pi , pH 7.0. Emission intensities at 330 nm wereflavin fluorescence. Free FMN is highly fluorescentmeasured, using 280-nm excitation, and are plotted against molarwith an apparent emission peak at 525 nm. A native ratios of (FMN)/(monomeric apoenzyme). The arrow indicates the

FRP sample containing 5 mM FMN, however, showed extrapolated break point for maximal fluorescence change. (B) Apo-enzyme at 50 nM was titrated with FMN (l) or at 70 nM with 2-only 8% of the fluorescence intensity of that for freethioFMN (s). Fluorescence intensities at 330 nm were measured,FMN at the same concentration, but the emission spec-using 280-nm excitation, after the flavin-added samples were incu-trum remained unshifted (Fig. 1A). The protein fluo- bated for 10 min. DFluorescence is defined as the difference of the

rescence of the apoenzyme showed a maximum at 330 emission intensity of apoenzyme minus that with the addition of agiven amount of flavin.nm (excitation at 280 nm). Upon FMN binding, the

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

92 LIU ET AL.

fluorescence was analyzed following the same methoddescribed for the FMN titration (l) shown in Fig. 2B.A Kd of 0.3 mM was thus obtained for the 2-thioFMNbinding (Fig. 2B, s), only slightly larger than the 0.2mM Kd for FMN binding. This high affinity enabled usto prepare an FRP derivative (FRPS) containing a 2-thioFMN cofactor. Figure 3 shows the absorption spec-tra of FRPS and free 2-thioFMN. The binding of 2-thi-oFMN to the apoenzyme resulted in considerable spec-tral changes, especially the shift of the original 490-nm peak to 509 nm with an extinction coefficient of18.4 mM01 cm01 at this wavelength for the bound 2-thioFMN. By measurements of protein and flavin con-tents, the amount of bound 2-thioFMN in FRPS (deter-mined on the basis of e490 Å 20.8 mM01 cm01 for free 2-FIG. 3. Absorption spectrum of FRPS. FRPS was reconstituted from

apoenzyme and 2-thioFMN as described under Materials and Meth- thioFMN after enzyme denaturation by urea) wasods. The spectrum of 8.3 mM enzyme in 0.2 M Pi , pH 7.0, was mea- found to be one flavin per enzyme monomer. There wassured in a 1-cm-path cuvette (solid line). For comparison, the spec-

no detectable spectral change of FRPS after storage fortrum of free 2-thioFMN at the same concentration is also shownseveral months at0207C, while free 2-thioFMN in solu-(dotted line).tion was easily converted to FMN under the same con-ditions.

Values of Km and Vmax for FRP and FRPS. As con-and NADPH. However, no activities were detectedtrols, the Km,FMN, Km,NADPH, and Vmax of the native FRPwhen FAD or riboflavin was added along with NADPHwere determined (Table I). Similar measurements wereto the apoenzyme. FAD or riboflavin at 30 mM did notalso made using FRP reconstituted from apoenzymequench the fluorescence of the apoenzyme. Further-and FMN, and results were essentially the same asmore, the same procedure described earlier for the re-those shown for the native enzyme. Although neitherconstitution of FRP holoenzyme from apoprotein wasriboflavin nor FAD exhibited any detectable binding tofollowed using FAD or riboflavin replacing FMN. Nothe cofactor site of FRP apoenzyme, both flavins arebound flavin was detected in the enzyme sample afteractive as a substrate for FRP. In comparison withsuch treatments. All of these results indicate that theFMN, the Km was increased four- and fivefold and theFRP apoenzyme does not bind FAD or riboflavin as aVmax was lowered by 33 and 40% for FAD and riboflavin,cofactor with any appreciable affinity. However, FADrespectively (Table I). The reconstituted FRPS was ac-and riboflavin are effective substrates (described intive in utilizing either FMN or 2-thioFMN as a sub-more detail below).strate (Table I). The Vmax of FRPS using 2-thioFMNIn contrast, the FRP apoenzyme binds 2-thioFMNsubstrate was 13% of that for the native FRP usingwith an affinity similar to that for FMN. A limitingFMN substrate. The Km,2-thioFMN of 7 mM for FRPS wasamount of apoenzyme (70 nM) was titrated with vary-

ing levels of 2-thioFMN and the quenching of protein only slightly higher than the 5 mM Km,FMN of the native

TABLE I

Substrate and Cofactor Specificitiesa of Native FRP and Reconstituted FRPS

NADPH FMN 2-ThioFMN Riboflavin FAD

Km Vmax Km Vmax Km Vmax Km Vmax Km Vmax

Enzyme (mM) (unit/mg) (mM) (unit/mg) (mM) (unit/mg) (mM) (unit/mg) (mM) (unit/mg)

FRPb 11 107 5 98 §50d 25 39 19 66FRPS

c 3 16 £30d 7 13

a Measurements were carried out in 50 mM Pi , pH 7.0, at 237C.b Values of Km and Vmax for flavin were determined at 0.16 mM NADPH, and those for NADPH were measured at 0.05 mM FMN, both

using 2 1 1008 M FRP per assay.c Values of Km and Vmax for flavin were determined at 0.16 mM NADPH, and those for NADPH were measured at 0.05 mM 2-thioFMN,

both using 2.9 1 1008 M FRPS per assay.d Estimated values. Some degrees of exchange of flavin cofactor with a different flavin added as substrate occurred during the course of

the activity assay.

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

93NADPH-SPECIFIC FLAVIN REDUCTASE APOENZYME AND SUBUNIT INTERACTION

apoenzyme undergoes any detectable slow reactionsuch as denaturation that would affect their self-as-sembly processes.

The sedimentation data were analyzed initially forweight-average molecule weight (MV w) for the entire so-lution column. Values of MV w were determined to be 4.6(4.0, 5.2) 1 104 and 4.3 (4.0, 4.5) 1 104 for native andapoenzyme, respectively. The values in parenthesesrepresent the limits of the 95% confidence interval.Since the monomeric molecular weights of native andapoenzyme are 2.6 1 104, the observed values for MV w

indicate that both native FRP and apoenzyme are capa-ble of self-assembly to larger aggregates. The sedimen-tation data were then fitted to models of various stoichi-ometries and the choice of the appropriate stoichiome-FIG. 4. Effects of FRP concentration on molecular sieve chromatog-try was defined by the goodness of fit and randomnessraphy elution profiles. Experiments were carried out using a Phar-

macia FPLC system and a Sephadex 75 column. Samples (0.5 ml) of the distribution of residuals. Following these proce-containing 22, 1.1, and 0.11 mM FRP were injected into the column dures, the sedimentation data best fit a stoichiometrypreequilibrated and eluted with 50 mM Pi , pH 7.0. Profiles were of 2 for both native FRP (Fig. 5) and apoenzyme (Fig. 6).normalized by setting the maximum A280 at 1.

An increasing value of stoichiometry led to increasinglypoorer goodness of fit (not shown). Thus, both nativeand apoenzyme are capable of undergoing dimerization

FRP. On the other hand, the 3 mM Km,NADPH of FRPS with Kd values of 1.8 and 3.3 mM, respectively.determined at 0.05 mM 2-thioFMN was lower than the11 mM Km,NADPH of the native FRP at 0.05 mM FMN.

DISCUSSIONValues of Km and Vmax cannot be accurately determinedfor FRP and FRPS using 2-thioFMN and FMN, respec- Two general strategies for the preparation of apoen-tively, as a substrate because some degrees of exchange zymes from flavoproteins have been described in thebetween the flavin cofactor and flavin substrate oc-curred during the course of the activity assay. In com-parison with FRP using FMN as a substrate, the Vmax

was estimated to be £30% for FRPS using FMN sub-strate and §50% for FRP using 2-thioFMN substrate.

Subunit interaction. The crystal FRP showed a ho-modimeric structure, yet an earlier molecular sieve ex-periment revealed an apparent Mr of 33,000, higherthan but close to the monomer Mr of 26,312 (15). Twolines of experiments were carried out to test the hy-pothesis for a monomer–dimer equilibrium of FRP.First, FRP at initial concentrations of 22, 1.1, and 0.11mM was subjected to molecular sieve chromatographyat 237C; a well-defined elution peak was observed forall three samples with retention times correspondingto apparent molecular masses of 38, 34, and 30 kDa,respectively (Fig. 4). As a control, V. harveyi luciferaseat 4.8 and 0.6 mM was subjected to the same chromato-graphic treatment but no difference in retention timewas observed. These results indicate the existence of arapid monomer–dimer equilibrium of FRP.

Next, equilibrium ultracentrifugation was employed,using both the native FRP and its apoenzyme, for amore rigorous and quantitative analysis of the pro- FIG. 5. Equilibrium ultracentrifugation of native FRP. Native FRP

(initial A280 Å 0.85) was subjected to centrifugation in 0.2 M Pi , pHposed monomer–dimer equilibrium. Sedimentation7.0, at 30,000 rpm and 47C until reaching equilibrium. Open circlesequilibrium was reached within 24 h and the proteinin the main panel are experimental readings, whereas the solid lineconcentration gradient as a function of radial distance is the theoretical curve generated on the basis of a monomer–dimer

remained constant for at least another 12 h. These ob- equilibrium with a Kd of 1.8 mM. The randomness of the distributionof residuals is shown in the lower panel.servations indicate that neither native FRP nor the

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

94 LIU ET AL.

suggests the importance of the negative charge groupin FMN binding. Interestingly, both riboflavin andFAD exhibited considerable activities as a substratefor FRP (Table I). The Km values for these two flavinswere only four- to fivefold higher than that for FMN.Apparently, the nature of flavin cofactor binding andthat for the flavin substrate binding are significantlydifferent.

Although monomeric FRP has eight tyrosine resi-dues (15), the shape of the fluorescence emission spec-trum of the native enzyme (lmax at 330 nm; Fig. 1)indicates that the sole Trp-212 residue (15) must bethe primary fluorophore under 280-nm excitation. Thebinding of FMN resulted in 87% quenching of proteinfluorescence. The Trp-212 side chain and the isoalloxa-zine ring of the bound FMN have no direct contact andare separated by about 13 A according to the X-raystructure of FRP (18). Therefore, such a high degree ofquenching cannot result from a dark complex forma-tion between Trp-212 and FMN and is unlikely due toa highly efficient energy transfer from tryptophan toFIG. 6. Equilibrium ultracentrifugation of FRP apoenzyme. FRP

apoenzyme (initial A280 Å 0.35) was subjected to centrifugation in FMN. Therefore, FMN binding apparently leads to a0.2 M Pi , pH 7.0, at 30,000 rpm and 47C until reaching equilibrium. conformational change of the enzyme which results inOpen circles in the main panel are experimental readings, whereas the quenching of protein fluorescence.the solid line is the theoretical curve generated on the basis of a

The apoenzyme binds 2-thioFMN with an affinitymonomer–dimer equilibrium with a Kd of 3.3 mM. The randomnessof the distribution of residuals is shown in the lower panel. similar to that for the FMN binding. Moreover, both

FRPS and FRP are catalytically active in using FMNor 2-thioFMN as a substrate. These properties makeFRP and FRPS an invaluable system for the evaluationliterature: weakening the binding of the native cofactorof the mechanism of reduced flavin transfer from flavinand transformation of the bound flavin. Following thereductase to bacterial luciferase. Our preliminary ki-first strategy, acidic pH, KBr, and partial protein un-netic studies suggest that luciferase preferentially uti-folding by urea or guanidine hydrochloride have beenlizes the reduced flavin cofactor, rather than the re-used, individually or in combination, to release theduced flavin product, of FRP for the FRP–luciferasebound flavin (24). An example for the second strategycoupled bioluminescence reaction (19). Luciferase canis the use of CaCl2 to prepare apoproteins from certainuse both FMNH2 and reduced 2-thioFMN as a sub-FAD enzymes by enhancing the hydrolysis of the cofac-strate for bioluminescence but with a lower quantumtor to FMN which dissociates readily (24, 25). Dialysisyield for the latter flavin (26, 27). Moreover, the biolu-is commonly used to remove dissociated or transformedminescence obtained from reduced 2-thioFMN peaks atflavin. The basic principle of the first strategy has been534 nm (uncorrected), considerably red shifted fromadapted in this work to develop a procedure suitablethe 492-nm maximum (uncorrected) observed in thefor obtaining FRP apoenzyme with high yields and ex-FMNH2-initiated emission (26). By using FRP or FRPScellent reconstitutability. It should be noted that a sin-in combination with FMN or 2-thioFMN in the lucifer-gle-step dilution of urea was carried out in the earlyase–flavin reductase coupled assay, the relative contri-stage of treatment to allow protein renaturation, andbutions to luciferase light emission by the reduced fla-the time-consuming dialysis is not employed at all invin cofactor and flavin product of the flavin reductasethis procedure, thus enabling us to complete the apoen-can likely be distinguished and quantified. Such stud-zyme preparation in just a few hours. This not onlyies are currently underway.saves time but also gives high yields and improves the

The X-ray structure of FRP shows that the oxidizedquality of apoenzyme. This procedure may potentiallyFMN cofactor is tightly bound by 16 hydrogen bondsbe adapted for the preparation of apoenzymes from(18). Moreover, only the original hydrogen bond be-other flavoproteins.tween the flavin N5 and the FRP Gly-130 amide hydro-The apoenzyme binds FMN tightly but binding ofgen could be lost as a direct consequence of flavin reduc-FAD or riboflavin at the cofactor site was not detected.tion (18). Hence, it is intriguing how to reconcile theseThe poor binding of FAD probably reflects an intoler-findings with the direct transfer of reduced FMN cofac-ance of the bulky adenosyl group of FAD. The low affin-

ity for riboflavin, which lacks the phosphate end group, tor from FRP to luciferase as suggested by our kinetic

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal

95NADPH-SPECIFIC FLAVIN REDUCTASE APOENZYME AND SUBUNIT INTERACTION

results. We are currently entertaining at least two pos- determined by molecular sieve chromatography (14,15), whereas the enzyme monomer has a molecularsibilities. First, the reduction of FRP cofactor could leadmass of 26 kDa (15). The apparent discrepancy canto a critical conformational change of FRP, thus weak-now be accounted for by the concentration-dependentening the binding of the flavin cofactor. Second, ourdimerization of FRP.kinetic studies suggest that luciferase could form a

transient complex with the reduced FRP. The bindingACKNOWLEDGMENTSof FMNH2 cofactor by FRP in such a complex could be

substantially weakened as a direct (e.g., interactions We appreciate the assistance of Robert Friesen, Lubomir Kovac,and Erica Pyles in the sedimentation study.between luciferase and FRP residues that are required

for cofactor binding) or indirect (e.g., conformationalREFERENCESchange in FRP induced by luciferase binding) conse-

quence of the luciferase–FRP complexation. 1. Fontecave, M., Eliasson, R., and Reichard, P. (1987) J. Biol.Chem. 262, 12325–12331.The equilibrium ultracentrifugation results clearly

2. Coves, J., Niviere, V., Eschenbrenner, M., and Fontecave, M.show that both the native FRP (Fig. 5) and the apoen-(1993) J. Biol. Chem. 268, 18604–18609.zyme (Fig. 6) undergo monomer–dimer equilibrium

3. Hasan, N., and Nestor, E. W. (1978) J. Biol. Chem. 253, 4987–with Kd values of 1.8 and 3.3 mM, respectively, at 47C.4992.

The molecular sieve chromatography profiles of FRP 4. Quandt, K. S., Xu, F., Chen, P., and Hultquist, D. E. (1991) Bio-at 237C and three different concentrations each show chem. Biophys. Res. Commun. 178, 315–321.a single well-defined peak but the retention time in- 5. Chikuba, K., Yubisui, T., Shirabe, K., and Takeshita, M. (1994)creases with decreasing concentration (Fig. 4). These Biochem. Biophys. Res. Commun. 198, 1170–1176.

6. Halle, F., and Meyer, J. (1992) Eur. J. Biochem. 209, 621–627.latter observations not only support the monomer–di-7. Coves, J., and Fontecave, M. (1993) Eur. J. Biochem. 211, 635–mer equilibrium of FRP but also indicate that such an

641.equilibrium is rather rapid. Although the Kd at 237C8. Gaudu, P., Touati, D., Niviere, V., and Fontecave, M. (1994) J.was not determined, it appears not to be drastically

Biol. Chem. 269, 8182–8188.different from that at 47C on the basis of the enzyme9. Duane, W., and Hastings, J. W. (1975) Mol. Cell. Biochem. 6,concentrations tested by the molecular sieve chroma- 53–64.

tography (taking into consideration a dilution factor of 10. Gerlo, E., and Charlier, J. (1975) Eur. J. Biochem. 57, 461–467.approximately two- to threefold during chromatogra- 11. Jablonski, E., and DeLuca, M. (1978) Biochemistry 17, 672–678.phy) (Fig. 4). Considering the low enzyme concentra- 12. Watanabe, H., and Hastings, J. W. (1982) Mol. Cell. Biochem.tions used (2–7 1 108 M) for steady-state kinetic mea- 44, 181–187.

13. Gibson, Q. H., and Hastings, J. W. (1962) Biochem. J. 83, 368–surements for FRP (Table I) and the determination of377.Kd values for flavin binding at the cofactor site (Fig.

14. Jablonski, E., and DeLuca, M. (1977) Biochemistry 16, 2932–2B), both at 237C, the flavin reductase was primarily2936.in the monomeric form under these conditions.

15. Lei, B., Liu, M., Huang, S., and Tu, S.-C. (1994) J. Bacteriol.The enzyme samples (at 4–5 mM) used for fluores- 176, 3552–3558.cence measurements at 237C (Fig. 1) were clearly a 16. Belas, R., Mileham, A., Cohn, D., Hilmen, M., Simon, M., andmixture of monomer and dimer. The high degree of Silverman, M. (1982) Science 218, 791–793.protein fluorescence quenching upon flavin binding 17. Fisher, A. J., Raushel, F. M., Baldwin, T. O., and Rayment, I.

(1995) Biochemistry 34, 6581–6586.(Fig. 1B), however, still indicates that the conformation18. Tanner, J. J., Lei, B., Tu, S.-C., and Krause, K. L. (1996) Bio-of FRP holoenzyme differs from that of the apoenzyme.

chemistry 35, 13531–13539.It is interesting to note that such a difference does not19. Lei, B., and Tu, S.-C. (1994) in Flavins and Flavoproteins (Yagi,result in any substantial change in the Kd for subunit

K., Ed.), pp. 847–850, de Gruyter, Berlin.dimerization. The question as to whether the subunit

20. Fory, W., and Hemmerich, P. (1967) Helv. Chim. Acta 50, 1766–dimerization of FRP is affected by the NADPH binding 1774.awaits further investigation. The binding of FMN to 21. Massey, V., and Swoboda, B. E. P. (1963) Biochem. Z. 338, 474–FRP apoenzyme is accompanied by an 8-nm red shift 484.

22. Yphantis, D. A. (1964) Biochemistry 3, 297–317.of the 445-nm peak (15). The FRP used was at 45 mM;23. Lowry, O. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J.therefore, the absorption spectrum is that for the di-

(1951) J. Biol. Chem. 193, 265–275.meric enzyme. Similarly, the binding of 2-thioFMN to24. Zanetti, G., Cidaria, D., and Curti, B. (1982) Eur. J. Biochem.apoenzyme also resulted in a 19-nm red shift of the

126, 453–458.long-wavelength 490-nm peak. The extents of the ob-25. Komai, H., Massey, V., and Palmer, G. (1969) J. Biol. Chem.

served spectral shifts of 2-thioFMN attributable to 244, 1692–1700.binding by apoenzyme or enzyme dimerization or both 26. Mitchell, G., and Hastings, J. W. (1969) J. Biol. Chem. 244,remain to be resolved. The molecular mass of FRP has 2572–2576.

27. Tu, S.-C. (1982) J. Biol. Chem. 257, 3719–3725.been previously reported to be from 33 to 40 kDa as

AID ABB 9746 / 6b27$$$161 11-25-96 20:32:40 arcal