Bacterial bioluminescence - PNAS

Proc. Natl Acad. Sci. USA Vol. 78, No. 2, pp. 948-952, February 1981 Biophysics

Bacterial bioluminescence: Spectral study of the emitters in the in vitro reaction (low-temperature fluorescence/fluorescent intermediates)

I. B. C. MATHESONt, JOHN LEEt, AND FRANZ MULLERt tBioluminescence Laboratory, Department of Biochemistry, University of Georgia, Athens, Georgia 30602; and *Department of Biochemistry, Agricultural University, Wageningen, The Netherlands

Communicated by W. D. McElroy, October 9, 1980

ABSTRACT Transient fluorescent species are.observed in the bioluminescent reactions of three reduced flavin mononucleotides with aliphatic aldehydes and. oxygen, catalyzed by bacterial luciferase. In each case the. fluorescence spectral distribution. is sims flar to that of the bioluminescence but is readily distinguishable from it on the basis of a significantly greater signal strength. The corrected bioluminescence. maxima using Beneckea Iharveyi luciferase are 479 nm (iso-FMNH ), 490 nm (FMNHi), and 560 nm (2-thio-FMNH2). In an ethanol glass at 77 K, 2-thioriboflavin is fluorescent (OF = 0.03, A = 562 nm). These results are interpreted by a sensitized chemiluminescence mechanism in which the flavins bound to luciferase act as acceptors of excitation energy. For 2-thio-FMNH2I, this acceptor species appears to be the oxidized 2-thio-FMN on the basis of the spectral evidence, whereas for the other flavins, some form of reduced species is a more likely candidate.

Lumazine protein is well fitted for its role in the bioluminescence of P. phosphoreum. It is manufactured in large quantity by the bacterium (7, 9). Its fluorescence quantum yield is 0.6 (10) and its fluorescence emission spectrum is identical to the

bioluminescence (2). The present work is an examination ofthe spectral properties of the in vitro bioluminescence reaction, with FMNH2, isoFMNH2, or 2-thio-FMNH2 as substrate. The results can be rationalized by a common, sensitized chemiluminescence mechanism in which the acceptors or emitters are identified as luciferase-bound flavin species. Such a process is also consistent with the evident function of lumazine protein (2, 7). MATERIALS AND METHODS Bacterial luciferase was isolated and purified by methods as described (11, 12), modified for speed and efficiency by the use of DEAE-Sephacel (Pharmacia) in place of DEAE-cellulose and DEAE-Sepharose as an alternative to Sephadex A-50. The details will be published elsewhere. To remove the last traces of ubiquitous fluorescent impurities in luciferase preparations, the luciferase was eluted from Sephadex A-50 with an eluant of constant ionic strength as an alternative to the salt gradient. The bacteria used were Beneckea harveyi strain 392 (13), P.fischeri strain 399, and P. phosphoreum strain A-13 (2). NADH:FMN oxidoreductase (NADH dehydrogenase) from P.ftscheri 399 was purified by a minor modification of the method of Puget and Michelson (14) and was free of luciferase activity. All chemicals used were of the best commercial grades, and FMN (Fluka, Buchs, Switzerland) was purified (-90%) by DEAE column chromatography. Lumiflavin, iso-FMN, isoriboflavin, 2-thio-FMN, and 2-thioriboflavin were prepared by literature methods. Some flavins used were the gifts of P. Hemmerich (University of Konstanz, Federal Republic of Germany). Reduced flavin species for the low-temperature fluorescence were prepared by reduction of ethanolic FMN solutions in situ (quartz EPR tube) with excess solid NaBH4. Reduction was essentially complete for lumiflavin- and 2-thioriboflavin in a period of a few hours, but isoriboflavin required an overnight period. Fluorescence and bioluminescence spectra were obtained by using a computer-controlled spectrometer (15) and the "SPECOS" software supplied by J. E. Wampler. Spectra were measured in short-pathlength cells (bioluminescence, 1 mm; fluorescence 2-3 mm) to minimize corrections for self-absorption. All spectra are corrected for self-absorption and for spectral sensitivity of the instrument-by reference to a National Bureau of Standards lamp. Bioluminescence was generated by two methods-usually by injection of the photoreduced form of the appropriate flavin into a solution containing luciferase and aldehyde, but in some cases (where indicated) by use of NADH and FMN coupled through oxidoreductase.

Bacterial luciferase catalyzes a reaction in which FMNH2, in the presence of oxygen and a long-chain aliphatic aldehyde, produces light. The bioluminescence has a broad spectral distribution with a maximum around 490 nm (1). The spectrum is unperturbed by changes in external conditions such as temperature, pH, reactant concentrations, and ionic strength, but the maximum does range over about 15 nm, depending on the type of bacterium from which the luciferase was purified (1, 2). Mitchell and Hastings (3) reported that chemical modification of the flavins has a more marked effect, with iso-FMNH2 giving an uncorrected bioluminescence spectral maximum at 472 nm and 2-thio-FMNH2 giving one at 534 nm. The in vivo bioluminescence spectrum also appears to depend on the type of bacteria, and the maxima range from 472 to 545 nm (1, 4, 5). This would suggest that the emitting species in the cell is not the same as that produced by the in vitro reaction with FMNH2. For example, for types of Photobacterium phosphoreum, the bioluminescence maxima are all at the shortwave-length end of the range (4, 6), and the emitting molecule has been identified in this case as 6,7-dimethyl-8-ribityllumazine bound to a protein (7). This holoprotein is called "lumazine protein." It is obviously of importance to a postulated reaction mechanism that the bioluminescence-emitting species be identified. There are several conditions that such a species must fulfill to qualify. First, it must be there in the reaction, either added to it or identified as a product (at the very least identified as a reasonable hypothetical product). Second, it must have a high fluorescence efficiency sufficient to account for the quantum yield of the bioluminescence reaction [for example, at least 0.1 in the case of the reaction of FMNH2 with the luciferase from Photobacteriumfischeri (8)]. Finally, the fluorescence spectral distribution must be the same as that ofthe bioluminescence. The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U. S. C. §1734 solely to indicate this fact.

948

Biophysics:

Matheson et al.

Proc. Nati. Acad. Sci. USA 78 (1981)

949

Low-temperature emission spectra (liquid N2, 77 K) were obtained by using a quartz EPR tube as the sample cell and a specially constructed quartz Dewar flask. Low-temperature absorption spectra were obtained with an Aminco DW-2 double wavelength spectrophotometer. RESULTS Bioluminescence spectra obtained on reaction of B. harveyi luciferase with the three reduced flavins showed maxima at 479 nm (iso-FMNH2), 490 nm (FMNH2), and 560 nm (2-thioFMNH2) (Fig. 1). The same spectra were produced if the reactions were carried out at a higher temperature (< 15'C) or if the appropriate flavin was reduced by NADH through the oxidoreductase. The differences between the bioluminescence maxima of this work and those of Mitchell and Hastings (3) are largely accounted for by the fact that the spectra of this work were corrected for fluorimeter spectral sensitivity and self-absorption, whereas the earlier work was not. The maximum of the iso-FMNH2 bioluminescence (Fig. 1, curve a) relative to FMNH2 bioluminescence (Fig. 1, curve b) showed an I1-nm blue shift. For P. fischeri luciferase, on the other hand, the NADH-iso-FMN-coupled reaction produced a new contribution at shorter wavelengths (Fig. 1, curve d) in addition to a main band with a distribution like that in Fig. 1, curve b. The intensity of this short-wavelength component varied considerably from experiment to experiment and, therefore, an average curve was plotted. This short-wavelength contribution was not seen when the bioluminescence was induced by photoreduced iso-FMNH2 or under either of the reaction conditions in B. harveyi bioluminescence. The bioluminescence of 2-thio-FMNH2 with P. fischeri luciferase was very similar to that shown for B. harveyi (Fig. 1, curve c). Neither iso-FMNH2 nor 2-thio-FMNH2 produced bioluminescence with P. phosphoreum luciferase. Fig. 2 shows the fluorescence spectra in a rigid ethanol glass at 77 K of the reduced isoriboflavin (Fig. 2A; maximum at 405 nm, OF 0.03), reduced lumiflavin (Fig. 2B; maximum at 492 nm, OF 0.3), and 2-thioriboflavin (Fig. 2C; maximum at 560 nm, OF 0.03). The excitation spectrum for each species is shown on the left side of each panel. The spectrum in Fig. 2B shows a higher resolution than the low-temperature fluorescence spectrum of tetraacetylriboflavin in the neutral form re550

600

650

0)

C.)

c\ ._

cc;,

a)

350

400

450

550

500

600

650

Wavelength, nm FIG. 1. In vitro bioluminescence spectra (bandwidth 20 nm) in 50 mM phosphate (pH 7.0). Reactions a-c for B.,harveyi luciferase at 0°C, contained tetradecanal [1% (vol/vol) of a soluition that is 5% of saturation (vol/vol) in methanol]. Curves: a, luciferrase (7.7 mg/ml) and 230 ,uM iso-FMNH2 (initial concentration); b, luciiferase (7.7 mg/ml) and 3 jAM FMNH2; c, luciferase (56 mg/ml) and 45 , 2-thio-FMNH2; and d, P. fischeri luciferase (6.4 mg/ml) and 19 ,uM iso-FMN coupled to 100 ,uM NADH through oxidoreductase in sufficiebnt quantity to give a convenient light level, with 1% saturated decana 1 in methanol at 11°C.

~M

,,,

I

I

I

I

0~~~~~~~~~~~~~

ccs 0)

0~~~~~~~~~~~~~~~~~~~~~~~~~0

350

400

450 500 550 Wavelength, nm

600

650

FIG. 2. Low-temperature fluorescence spectra (bandwidth, 5 nm) in an ethanol glass (77 K). Corrected excitation spectra are on the left; open circles are the absorption spectrum at 77 K (2-thioriboflavin only). (A) Reduced isoriboflavin (A = 0.3 at 455 nm) excited at 380 nm. (B) Reduced lumiflavin (A = 0.3 at 450 nm) excited at 380 nm. (C) 2thioriboflavin (oxidized) (A = 0.42 at 485 nm) excited at 475 nm.

ported by Ghisla et al. (16) but is otherwise similar in shape and in the position of the maximum. The fluorescence maxima in Fig. 2A and C are very similar to the corresponding bioluminesI curvCe d ailu maIVIMa inrmig. niirvc u anti C, roonnp Fiar 1, irc~oiow culi 11za ilzd III respectively.

The total luminescence of 2-thioriboflavin was probably fluyield 0.03 in ethanol glass (77 K). The excitation spectrum matched the low-temperature absorption spectrum (Fig. 2C), suggesting that the emission arose truly from 2-thioriboflavin and not from an impurity. This compound has been reported to be nonfluorescent by Mitchell and Hastings (3) and by Sun et al. (17). In the case of 2-thio-FMN in aqueous solution, a fluorescent impurity [maximum at 535 nm (FMN?)] was present at a level varying from sample to sample; its interference has been noted in a previous study (18). This impurity was present at a negligible level in 2-thioriboflavin. The fluorescence spectrum obtained with reduced isoriboflavin varied with the degree of reduction. The low-temperature spectra are shown (Fig. 3) as a function of the time of NaBH4 reduction (at room temperature) in ethanolic solution. The 0.3-hr spectrum shows a contribution from oxidized isoriboflavin in the 550-nm region and a new, putatively reduced species in the 400-nm region. This short-wavelength species increased with time, but the width of the spectral curve at longer reduction times suggests the presence of more than one reduced orescence, quantum

species. No fluorescence spectrum

(OF < 10-a) could be detected from reduced 2-thioriboflavin in an ethanol glass at 77 K or from

950

Biophysics: Matheson et al. 350

400

450

500

Ca~~~~~~~~

35 404-0'0

Proc. Nad Acad. Sci. USA 78 (1981) 600 550 During the course ofthe reaction, the intensity ofthe contriTTTT | l Z |bution at the shorter wavelength decreased with a corresponding increase in the longer-wavelength region. The short-wavelength fluorescence (Fig. 4, curve a) was in the same spectral region as was the fluorescence of the reduced isoriboflavin at low temperature (Fig. 2A), but the data quality is such that only a general similarity between the bioluminescence and the fluorescent intermediate spectra could be discerned. Fig. 4, curve b, shows the fluorescence for the FMNH2 reaction. A small contribution due to the reaction product, FMN,

550

600

Wavelength, nm FIG. 3. Low-temperature fluorescence spectra (bandwidth, 5 nm) for reduced isoriboflavin in ethanol glass (77 K). E3pectra shown as a function of time of NaBH4 reduction at room tempoerature. Curves: a, 0.3 hr; b, 0.5 hr.; c, 29 hr.

iso-FMNH2, FMNH2, 2-thio-FMNH2, 2-thic)-FMNH', or 2thio-FMN in aqueous solution at room temp)erature, with or without the presence of luciferase (B..harveyii, A2w= 3). This was in contrast to the observations of Becvmr et al. (19), who have reported that the luciferase-bound FM NH2 was weakly fluorescent in the 530-nm region. On initiation of the bioluminescence reactiion with reduced

flavin, a new fluorescent species made its apt:earance and was readily distinguishable from the bioluminescernce by a signal intensity that was several times greater. Fig. 4 sh,OWS the transient

fluorescent spectra observed after the addition ofeach ofthe re4a) flavins to B. harveyialuciferase and tetiradecanal. These duced spectra were obtained with maximized excitaltion, and in each case the fluorescence signal intensity was excit.ation wavelength with dependent. The fluorescence of the reaction imixture vlnFMNH2 (Fig. 4, curve a) was made up of twc major contributions. The longer-wavelength contribution was attributed to isoFMN, the final product, because the excitationi spectrum for the fluorescence at 550 nm corresponded to that of iso-FMN and because the fluorescence ofthe mixture, afterscompletion ofthe bioluminescence, was the same as iso-FMN. 450

500

550

600

650

was subtracted from this spectrum. This was carried out by measuring the emission level at 580 nm (where the fluorescent transient did not emit) with the assumption that this emission was due entirely to FMN and by subtracting a suitably scaled file FMN spectrum from the experimental spectrum. A particularly fluorescent impurity-free sample of B. harveyi luciferase, A28J/A4, = 300, was used to collect these data so that no subtraction of the luciferase contribution was required. The same spectrum was obtained when B. harveyi luciferase samples of even higher A ratio were used. The fluorescence shown in Fig. 4, curve b, was about 2 to 3 times stronger than the "background" bioluminescence and decayed at about half the rate of the bioluminescence. The transient fluorescence in the 2-thio-FMNH2 bioluminescence was obtained after subtraction ofbackground fluorescence due to the protein (Fig. 4, curve c). This spectrum was measured several minutes after initiation of the reaction so that the bioluminescence contribution was negligible. This spectrum also had a contribution in the short-wavelength 535-nm region from the fluorescent impurity present in 2-thio-FMN mentioned above. It should be noted that the fluorescence spectra during the course of the bioluminescence reaction were weak and hard to measure. Therefore exact matching ofspectral maxima with Fig. 1, curves a, b, and c, was not obtainable, but the bioluminescence and fluorescence for each flavin were in the same general region. The fluorescence spectral change during the course of reaction was slower than the rate ofdecrease ofbioluminescence intensity, in all cases decreasing at about half the rate of decrease of the bioluminescence intensity. These experiments were done several times with different preparations of luciferase and of 2-thio-FMN containing higher levels of impurity contributions. The deconvoluted spectra were the same as in Fig. 4. In the absence of added aldehyde, a low level ofbioluminescence was generated and was attributed to the presence of "endogeneous aldehyde" in the luciferase preparation. However, no transient fluorescence could be detected above this bioluminescence intensity. The added aldehyde seemed to be required for formation of these transients.

DISCUSSION

c 450

500

550

650

600

Wavelength, nm FIG. 4. Fluorescent intermediates (bandwidthL, 10 nm) observed during the course of in vitro bioluminescent react ions, all in 50 mM phosphate (pH 7.0) for B.harvoyi luciferase and witth 1% (vol/vol) of a Curves: a, 5% saturated solution of tetradecanal in methanol at luciferase (1.7 mg/ml) and 88 M iso-FMNH2;bF ml) and 5 ,uM FMNH2; and c, luciferase (12 mg/ml) and6.6MM 2-thioFMNH2. Excitation wavelengths: iso-FMNH2, 35C) nm; FMNH2, 350 nm; and 2-thio-FMNH2, 475 nm.

0rC.

A transient fluorescent species is formed on reaction ofreduced flavins, oxygen, and luciferase with aldehyde, and it has the properties that would be required ofthe bioluminescence emitter. Comparison of the spectra for the appropriate flavins (Fig. 4) shows in all cases a generally similar spectral distribution to the bioluminescence (Fig. 1). In each respective case, the bioluminescence spectrum is also similar to the spectrum of reduced

isoriboflavin, reduced lumiflavin, and 2-thio-riboflavin in a rigid glass. Ghisla et al. (20) and Muller et al. (21) have shown that several dihydroflavin species have spectra obtained in a rigid glass, bound to enzyme or free in solution, that are com-

patible in wavelength with the bioluminescence spectral distribution. In addition to FMNH2, these include species substituted at the 4a position (20), N(5)-adducts (22, 23), and 3,4dihydroflavins (21). Any of these reduced species is a possible

Biophysics:

Matheson et aL

candidate for the bioluminescence emitter. Oxidized flavin species seem unlikely to be the emitters in the FMNH2 and isoFMNH2 reactions because the spectral shifts seem to be unreasonably large [530-nm fluorescence maximum for FMN with bioluminescence at 490 nm and 550-nm fluorescence maximum with bioluminescence at 479 nm (or less in the oxidoreductasecoupled reaction)]. This is particularly true in the case of isoFMN. However, in the case of 2-thio-FMNH2-induced bioluminescence, the near coincidence of the fluorescence maximum of 2-thioriboflavin in a rigid glass, with both the maximum of the bioluminescence and of the transient fluorescence, suggests that the emitter is the oxidized species or some species derived from it; although N(5)-adducts of 2-thio-4a,5-dihydro-FMN are also reported to have fluorescence maxima in the 500- to 600nm region (22). Thus, the present results suggest that the bioluminescent emitter is probably a reduced flavin species for the iso-FMNH2 and FMNH2 reactions and an oxidized flavin species in the 2thio-FMNH2 reaction. We propose that the luciferase (E)-catalyzed oxygenation of reduced flavin and aldehyde produces an energy-rich chemical species, EX, whose decomposition in the presence ofa suitable acceptor A, results in excitation of A to its electronic excited state, A*. The overall scheme, is an extension ofthat of Hastings and Gibson (24), E + FH2--* E FH2 E FH2 + 02+ RCHO -- EX EX + A

EP-A*

in which P represents products, FH2 is the reduced flavin, and, for the cases we have studied here, A is the emitter (some candidates are listed above). For the in vivo reaction, A is lumazine protein for P. phosphoreum and probably an equivalent yellow fluorescent protein in the yellow bacteria of Ruby and Nealson (5). A feature of this scheme is that the fluorescent transient does not form until after reaction of E FH2 with RCHO. It is possible that the fluorescence reported by Becvar et al. (19) is a consequence of some enzyme turnover on the addition of dithionite. The kinetic consequences of the above scheme and the observation that the fluorescent transient decays at about half of the rate of the bioluminescence for all three flavins will be described elsewhere. Chemical model studies of Bruice and coworkers (25, 26) have shown that synthetic 4a-peroxydihydroflavins give rise to chemiluminescence on reaction with aldehydes. The most recent work of Shepherd and Bruice (27) shows that the decomposition of 4a-(alkyl-peroxy)flavin, a species that is the product of reaction between 4a-hydroperoxyflavin and aliphatic aldehyde, gives rise to an excited species whose energy may be transferred to added fluorescent molecules. The findings ofthis work are consistent with the observations of Shepherd and Bruice in that both demonstrate energy transfer to species of widely varying fluorescence energy. A somewhat different mechanism involving a 4a-substituted flavin has been proposed by Hastings and coworkers (28). This mechanism requires that the 4a-hydroperoxydihydro-FMN reacts with aldehyde to form the 4a-hydroxydihydro-FMN directly in its electronic excited state. The result reported here for FMNH2 bioluminescence is not inconsistent with this proposal, because the 4a-hydroxydihydro-FMN and other dihydroflavins have similar fluorescence spectra. However, it is difficult to see how such a mechanism can lead to population of excited states of

Proc. NatL Acad. Sci. USA 78 (1981)

951

such widely varying energy. These energies correspond to wavelengths varying from a bioluminescence maximum of 400 nm for the oxidoreductase-coupled reaction of iso-FMN to 560 nm for the 2-thio-FMNH2 reaction. In the scheme proposed here, several possible processes may be considered for the mechanism of acceptor excitation. The EX may form an excited donor with energy equivalent to about 350 nm, which is deposited onto the acceptor by Forster longrange energy transfer. However, overlap between donor emission and acceptor absorption is required at 350-400 nm for the reduced flavins or derived species, 414 nm for lumazine protein, and 490 nm for 2-thio-FMN. The lumazine protein absorption has little spectral overlap with the FMNH2 bioluminescence; yet, in addition to causing the blue shift of the bioluminescence spectrum when added to the in vitro bioluminescence reaction, it also increases the bioluminescence light yield (2). Other possible processes are exchange transfer or electron transfer, both of which depend on spectral or redox properties well satisfied by the proposed acceptors. The observations of Bruice and coworkers referred to above can be explained adequately by a sensitized chemiluminescence mechanism, where the acceptor is a minor component of the reaction mixture (27). It is relevant to the chemical requirements for producing EX to note that iso-FMNH2 and 2-thio-FMNH2 are not substrates for bioluminescence with P. phosphoreum luciferase, and isoFMNH2 gives very little spectral shift with P.ftscheri. It is not possible to eliminate the possibility that iso-FMNH2 and 2-thioFMNH2 are both reducing the ubiquitous minor impurity in the B. harveyi luciferase preparation, and this reduced flavoprotein reacts to produce the donor EX; the luciferase-bound, unreacted iso-FMNH2 (or derivative) and luciferase-bound 2-thioFMN can still be acceptors. Then the iso-FMNH2 bioluminescence is a mixture of emissions from two acceptors, one derived from flavin and the other from isoflavin, accounting for the broadening ofthe spectrum to the blue (Fig. 1, curve a). The postulate that the emitting complex is a dihydroflavin of the form EP-FMNH2 or the oxidized species EP-2-thio-FMN has to be reconciled with the fact that no fluorescence is observed on anaerobically mixing native luciferase with FMNH2, iso-FMNH2, or 2-thio-FMN. However, there is some evidence to suggest that luciferase undergoes a conformation change during reaction (29), so it is feasible that EP has a binding site in which these flavins become fluorescent. If the lumazine is the natural emitter and because flavins are benzalumazine derivatives, it is hardly surprising that flavins can also function in a similar capacity as an energy acceptor. Flavoproteins are notoriously indiscriminate in regard to coenzyme structure; luciferase also can light up with whatever it can find. We thank Dr. Ron Makula and James Linn for operating the fermentation facility, Martha G. Elrod and Bruce Gibson for technical assistance, and Dr. D. DerVartanian for measuring the low-temperature absorption in Fig. 2C. We also thank Prof. Dr. W Pfleiderer, University of Konstanz, for giving us some 2-thiolumazine derivatives for study; although these results are not reported on in this present study, the information led us back to the discovery of the 2-thio-FMN fluorescence. This work was supported by National Science Foundation Grant PCM79-11064 and National Institutes of Health Grant GM-28139. 1. Seliger, H. H. & Morton, R. A. (1968) in Photophysiulogy, ed. Giese, A. C. (Academic, New York), Vol. 4, pp. 253-314. 2. Gast, R. & Lee, J. (1978) Proc. NatL Acad. Sci. USA 75, 833-837. 3. Mitchell, G. & Hastings, J. W (1969) J. BioL Chem. 244, 2572-2576. 4. Spruit-van der Burg, A. (1950) Biochim. Biophys.Acta 5, 175-178. 5. Ruby, E. G. & Nealson, K. H. (1977) Science 196, 432-434.

952

Biophysics: Matheson et al.

6. Fitzgerald, J. M. & Lee, J. (1978) in Microbial Ecology, eds. Loutit, M. W. & Miles, J. A. R. (Springer, Berlin), pp. 40-41. 7. Koka, P. & Lee, J. (1979) Proc. Natl Acad. Sci. USA 76,3068-3072. 8. Lee, J. (1972) Biochemistry 11, 3350-3359. 9. Small, E. D., Koka, P. & Lee, J. (1980) J. Biol Chem. 255, 8804-8810. 10. Visser, A. J. W. G. & Lee, J. (1980) Biochemistry 19, 4366-4372. 11. Lee, J. & Murphy, C. L. (1975) Biochemistry 14, 2259-2268. 12. Gast, R., Neering, I. R. & Lee, J. (1978) Biochem. Biophys. Res. Commun. 80, 14-21. 13. Reichelt, J. L. & Baumann, P. (1973) Arch. Mikrobiol 94, 283-330. 14. Puget, K. & Michelson, A. M. (1972) Biochimie 54, 1197-1204. 15. Wampler, J. E. (1978) in Bioluminescence in Action, ed. Herring, P. J. (Academic, London), pp. 1-48. 16. Ghisla, S., Massey, V., Lhoste, J.-M. & Mayhew, S. G. (1974) Biochemistry 13, 589-597. 17. Sun, M., Moore, T. A. & Song, P. S. (1972)J. Am. Chem. Soc. 94, 1730-1740. 18. Eley, M., Lee, J., Lhoste, J.-M., Lee, C. Y., Cormier, M. J. & Hemmerich, P. (1970) Biochemistry 9, 2902-2908.

Proc. NatL Acad. Sci. USA 78 (1981) 19. Becvar, J. E., Baldwin, T. O., Nicoli, M. Z. & Hastings, J. W. (1976) in Flavins and Flavoproteins, ed. Singer, T. P. (Elsevier, Amsterdam), pp. 94-100. 20. Ghisla, S. (1980) Methods EnzymoL 66, 360-373. 21. Miller, F., Massey, V., Heizmann, C., Lhoste, J.-M. & Gould, D. C. (1969) Eur. J. Biochem. 9, 392-401. 22. Ghisla, S., Massey, V. & Choong, Y. S. (1979)J. Biot Chem. 254, 10662-10669. 23. Massey, V., Ghisla, S. & Kieschke, K. (1980) J. Biol. Chem. 255, 2798-2806. 24. Hastings, J. W. & Gibson, Q. H. (1963) J. BioL Chem. 238, 2537-2554. 25. Kemal, C. & Bruice, T. C. (1977)J. Am. Chem. Soc. 99, 7064-7067. 26. Kemal, C. & Bruice, T. C. (1976) Proc. Natl. Acad. Sci. USA 73, 995-999. 27. Shepherd, P. T. & Bruice, T. C. (1980) J. Am. Chem. Soc. 102, 7774-7776. 28. Balny, C. & Hastings, J. W. (1973) Biochemistry 14, 4719-4723. 29. Nicoli, M. Z., Baldwin, T. O., Beevar, J. E. & Hastings, J. W. (1976) in Flavins and Flavoproteins, ed. Singer, T. P. (Elsevier, Amsterdam), pp. 87-93.