JOURNAL OF BACrERIOLOGY, October 1970, p. 313-322Copyright 0 1970 American Society for Microbiology

Cellular Control of the Synthesis and Activity ofthe Bacterial Luminescent System1

KENNETH H. NEALSON, TERRY PLATT, AND J. WOODLAND HASTINGS

Biological Laboratories, Harvard University, Cambridge, Massachusetts, 02138 andMarine Biological Laboratory, Woods Hole, Massachusetts 02534

Received for publication 30 April 1970

In bioluminescent bacteria growing in shake flasks, the enzyme luciferase hasbeen shown to be synthesized in a relatively short burst during the period of ex-

ponential growth. The luciferase gene appears to be completely inactive in a freshlyinoculated culture; the pulse of preferential luciferase synthesis which occurs later isthe consequence of its activation at the level of deoxyribonucleic acid transcriptionwhich is attributed to an effect of a "conditioning" of the medium by the growing ofcells. Although cells grown in a minimal medium also exhibit a similar burst of syn-thesis of the luminescent system, the amount of synthesis is quantitatively less, rela-tive to cell mass. Under such conditions, added arginine results in a striking stimu-lation of bioluminescence. This is attributed to a stimulation of existing patterns ofsynthesis and not to induction or derepression per se.

The control of the appearance of luminescenceduring the growth cycle is a spectacular but poorlydescribed and irn-understood feature of thebioluminescent bacteria. The matter of interestin this paper is that the growth of the bacteriaand the development of luminescence do notoccur in concert in liquid cultures. The develop-ment of light emission is delayed and starts onlyduring the middle of the logarithmic phase ofgrowth. Its rise is then far more rapid than is theincrease in cell mass during that time.The phenomenon is illustrated in the plots of

Fig. 1 and 2, which present the data in both expo-

nential and linear form. The growth of the cellsproceeds exponentially over the first few hours;in a complex medium, the doubling time at 25 Cis about 25 min. But the in vivo bioluminescenceof the cells does not increase during that time; itactually decreases and begins to rise only at alater time. Its rise is then very quick; the lightoutput may double about every 4 or 5 min. Atits peak, the luminescent enzyme (luciferase)constitutes between 2 and 5% of the soluble pro-tein of the cell (6).The experiments described in this paper were

carried out in an attempt to better understandthis phenomenon, especially the conditions whichcontrol the synthesis of luciferase. We concludedthat in a freshly inoculated culture the gene con-trolling luciferase synthesis is inactive and that

1 This research was supported by Contract N00014-67-0298-0001 from the Office of Naval Research.

neither luciferase nor its messenger is being syn-thesized. The onset of in vivo luminescence ishypothesized to occur by means of an activationof the luciferase gene resulting in the stimulationof messenger ribonucleic acid (mRNA) synthe-sis, followed by luciferase synthesis. Since thephenomenon occurs without external interven-tion, it must be attributed to a conditioning ofthe medium effected by the growing cells (9). Wehave referred to the phenomenon as "autoinduc-tion."The behavior of cells growing in a minimal

medium has important similarities to that of cellsin a complex medium, but there are also signifi-cant differences. After inoculation into a freshmedium, there is, similarly, a period during whichno luciferase synthesis occurs, followed by a

pulse of relatively rapid synthesis. The autoinduc-tion phenomenon is thus basically similar to thatwhich occurs in complex medium, but it differswith regard to the extent of autoinduction, i.e.,in the amount of synthesis which results. Lesssynthesis of both the luciferase and other co-

factor(s) occurs, so that the in vivo biolumines-cence appears very dim indeed.

Coffey reported that the bioluminescence ofcells growing in minimal medium can be greatlystimulated by the addition of arginine (1). Al-though this effect was referred to as induction,we found that the arginine acts instead by simplyenhancing the autoinduction, stimulating thesynthesis of both luciferase and the other factor(s)

313

Vol. 104, No. 1Printed In U.S.A.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

NEALSON, PLATT, AND HASTINGS

to higher levels. On the basis of cell mass, theluciferase content of such cells is similar to thecontent of cells grown in complex medium.

MATERIALS AND METHODS

Photobacterium fischeri strain MAV has been de-scribed previously (8) and is designated as the wildtype. Mutants were derived from this strain by muta-genesis with nitrosoguanidine (K. H. Nealson, 1969,Ph.D. Thesis, University of Chicago).

Minimal medium was similar to that of Farghaly(3), containing 30 g of NaCl, 7 g of Na2HPO-7H20,1 g of KH2PO4, 0.5 g of (NH4)2PO4, 0.1 g of MgSO4,and 3 ml of glycerol per liter of distilled water. Com-plex medium was prepared by adding 0.5 g of yeastextract (Difco), 5 g of peptone (Difco), and 3 ml ofglycerol to 1 liter of sea water. For solid media, 11 gof agar (Difco) were added per liter.The experiments were carried out in liquid shake

cultures at 25 to 27 C. Since the amount of lumines-cence varies to some extent with the amount of aera-tion, care was taken to keep the shaking conditions assimilar as possible in different experiments. Using 300-ml flasks with a culture volume of 75 ml, the genera-tion time in the complex medium was approximately25 min, whereas in the minimal medium it was approx-imately 250 min. For large experiments similar condi-tions were found to pertain by using a 500-ml culturevolume in 2-liter flasks.The light measuring equipment employed a photo-

multiplier tube (lP21) enclosed in a lighttight chamberto which the sample could be exposed by a shuttermechanism (7). The tube was operated at 1,000 v,and, after appropriate amplification, the output wasmonitored on an Esterline-Angus Speed Servo re-corder. Light intensity, as established by the stand-ards of Hastings and Weber (7), is expressed inquanta/sec.

Measurements on the cells during growth werecarried out by using portions removed from the shakeflask at the times indicated. Samples were discardedafter use. Cell density was determined at 660 nm byusing test tubes in a Coleman Jr. Spectrophotometerand is expressed in the optical density units of thatinstrument. These measurements were accompaniedby viable cell counts which were carried out by platingon the complex medium after appropriate dilution.Well aerated cells emit continuously at a relativelyconstant intensity, and the in vivo bioluminescencewas measured with 1 ml of such a sample. Cyanide(0.015 M) and aldehyde, which stimulated lumi-nescence under certain conditions, were added (in seawater solutions) by injection from a syringe to the vialpositioned in front of the phototube during the actuallight measurement.

In vitro determinations of luciferase were carriedout by measurements of its activity in extracts ofcells. Culture portions were harvested on membranefilters (Millipore Corp.) or by centrifugation at15,000 X g for 10 min. Extraction was accomplishedby lysis at low ionic strength; after quick freezingand thawing, luciferase was extracted in 0.01 M ethyl-enediaminetetraacetic acid (EDTA) (pH 7.0) with

1O-3 M dithiothreitol added. The luciferase assay wascarried out as previously described (8), by measuringthe initial maximum light intensity (Io) upon mixingwith 1 ml of reduced flavine mononucleotide(FMNH2; 5 X 10-5 M) in the presence of aldehyde(n-decanal) and oxygen. The overall reaction may berepresented by the following equation:

FMNH2 + luciferase + 02+ RCHO -- light + products

The reaction mixture before FMNH2 addition con-tained 0.1 ml of the cell extract (luciferase), 0.1 mlof decanal solution and 1 ml of 0.1% bovine serumalbumin in 0.075 M phosphate buffer (pH 7.5).Reactions were carried out at 22 i 2 C. In all experi-ments, the luciferase activity of the extract is expressedas the activity obtained from 1 ml of the cell culture.

Chemical reagents were of analytical quality whereavailable. Flavine mononucleotide (FMN), aminoacids, and chloramphenicol were obtained fromSigma Chemical Co., St. Louis. Puromycin dihydro-chloride was purchased from Nutritional Biochemi-cals, Cleveland, Ohio. Rifampin was obtained fromCIBA Pharmaceutical Co., Summit, N.J. Platinizedasbestos was obtained from E. H. Sargent and Co.Stock solutions of decanal (K and K Laboratories,Plainview, N.Y.) were prepared by sonic disruptionof a 0.1-ml portion in 10 ml of distilled water. Solu-tions for use were diluted 1 to 100 and prepared fresheach day.

Antibody to luciferase was produced in rabbitsby using purified luciferase as the immunogen(suspended in Freund's adjuvant at a concentrationof 1 mg/ml). The primary 0.5-ml subcutaneous in-jection made in the neck was followed 21 days laterby an identical secondary injection. Serum washarvested 10 days later and found to contain a hightiter of antiluciferase. Since luciferase is enzymaticallyinactive after reaction with antiluciferase, the quantityof luciferase and antigenically cross-reacting material(CRM) could be measured by determining the amountof antiluciferase required to give inhibition. This wasdone using a modification of the method of Tsuji andDavis (11).

RESULTSIn a freshly inoculated culture, light emission

does not begin to increase until a considerableamount of growth has occurred (Fig. 1 and 2).Several explanations have been suggested to ac-count for this. The explanations can be groupedin four categories, depending on whether thecontrol involved is hypothesized to operate atthe level of substrate control, enzyme activation,translation, or transcription. It has been proposedthat the lag is due to an inhibitor in the mediumwhich is removed by a "conditioning" of themedium during growth (9). Although this hy-pothesis seems inadequate to explain the experi-mental data which we have obtained, it could beaccomodated into any one of the four categoriesproposed.

314 J. BAcTERIOL.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CONTROL OF LUMINESCENT SYSTEM

Let us consider first the possibility of substratecontrol. It is possible that the failure of cells toemit light during the early stages after inoculationmight simply be due to a competition for elec-trons (2, 4, 10). Since FMNH2 (the substrate forthe bioluminescent oxidation) derives from thereducing power of the cell, it is possible that acontrol mechanism exists which channels elec-trons during growth. The requirements for reduc-ing power for adenosine triphosphate generationand biosynthesis might then be preferentiallyaccomodated during the early stages of growthin a shake flask. A prediction of this theory is thatactive luciferase is synthesized at all times alongwith other enzymes and cellular constituents butthat it is inactive because of the unavailability ofits substrate. The experiment of Fig. 1 refutesthis particular hypothesis because there is no in-crease in the total extractable luciferase activity;during the first 3 to 4 hr of growth it remainsvirtually constant. This might be due to a precisebalance between synthesis and breakdown;alternatively, the luciferase present in the cells atthe time they are inoculated into the fresh me-dium (zero hours) might be distributed withoutloss among the progeny during growth. As willbe described later, the latter alternative is sup-ported by experiments with inhibitors of pro-tein synthesis.

Figure 1 also shows that the spectacular rise ofthe in vivo luminescence can be attributed to andclosely parallels the appearance of extractableactive luciferase. At the same time, it is evidentand obligatory that most or all other enzymes,unlike luciferase, are synthesized in concert withgrowth. This has been shown to be true for glu-cose-6-phosphate dehydrogenase (K. H. Neal-son, 1969, Ph.D. Thesis, University of Chicago),and also for a flavine reductase (W. C. Duane,1969, Ph.D. Thesis, University of Illinois,Urbana).Another prediction of this first hypothesis is

than an inhibitor such as cyanide, which blockselectron flow through cytochromes to oxygen,should greatly stimulate luminescence duringthe "eclipse" period but not at times when in vivoluminescence is maximal. Actually, cyanide isknown to stimulate luminescence under low oxy-gen tension (K. L. Van Schouwenburg, 1938,Ph.D. Thesis, Delft, Holland; reference 4)implying that a competition for electrons doesexist under these conditions.We therefore measured the effect of cyanide

upon luciferase at various times during growthand found a differential stimulation (Fig. 1 and3). But the stimulatory effect evidently relates tothe interesting and previously unexplained de-crease in the in vivo luminescence which occurs

TIME-HOURSFIG. 1. Time course of growth and luminescence of

P. fischeri MAV cells in complex medium. The ordinatevalues, plotted on a logarithmic scale, should be multi-plied by the factors specified to obtain the experimentalvalues. Growth was measured both by viable cell count(108 cells/ml) and cell density (CD; X I). Luminescence(2 X 1010 quanta per sec per ml) was measured bothin vitro (the activity of the extractable luciferase) and invivo. In the latter case, the luminescence was alsomeasured subsequent to the addition of 0.015 M potas-sium cyanide and aldehyde (decanal). These additionsstimulated light emission during the first and last fewhours, but not during the intervening hours when activeluciJerase synthesis was taking place. The kinetics ofthe response to aldehyde are illustrated in the inset.

4 8 12TIME - HOURS

FIG. 2. Experiment of Fig. I plotted on a linearscale, better illustrating the "pulse" nature of theluciferase synthesis and in vivo luminescence. Theordinate values were all normalized to the maximumandplotted as the per cent ofthat value.

VOL. 104, 1970 315

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

NEALSON, PLATT, AND HASTINGS

during the first few hours. Thus, electron channel-ing does occur during this period, and it can beredirected, presumably by cyanide inhibition ofrespiration.At the same time, the stimulation by cyanide

never results in an in vivo activity exceeding thatwhich was present at the time of inoculation,thereby supporting the thesis that there is no in-crease in the quantity of the luciferease duringthe eclipse period. Actually, cyanide alone failsto restore the original in vivo activity, but if alde-hyde (n-decanal) is added after the cyanide stimu-lation has come to a steady state (about 1 min),an increase occurs to a level of light emissioncomparable to that which the cells exhibited atthe time of inoculation (Fig. 1 and 3).

In summary, the decrease in the in vivo ac-tivity during the eclipse period is not due to the

I 2 3 4TIME- HOURS

FIG. 3. Details of the in vivo luminescence of theexperiment ofFig. I during the first few hours, showingthe stimulation by cyanide and aldehyde, individuallyand together. The luminescence of 1-ml samples of thecells was measured at the times indicated, and then 0.1ml of either 0.015 M KCN in sea water or a saturateddecanal solution was injected. When both were added,the cyanide was added first, and after the maximalresponse (-1 min) the aldehyde was added. The re-sponse to aldehyde was rapid and the luminescencethen declined rather quickly (inset, Fig. 1).

loss of luciferase, but to a combination of twoother factors: (i) substrate (electron) deficiencywhich can be restored by cyanide, and (ii) theloss or dilution out of the endogenous cellularaldehyde or aldehyde factor. With regard to thephenomenon under consideration, namely theapparent delay in the synthesis of new luciferase,the fact that there is no increase in luminescence(either in vivo or in vitro) during the first fewhours cannot be attributed to substrate control.The second possibility is that luciferase syn-

thesis actually does accompany cell growth, butthat the enzyme is produced in an inactive formrequiring only some specific activation step. Onthe assumption that the hypothetical zymogenis antigenically reactive with antiluciferase, thishypothesis predicts that antigenically CRMwould be produced during the early phases ofgrowth and that its quantity would be propor-tional to cell mass. The titer of CRM (as deter-mined by activity inhibition with anti-luciferase)closely parallels the extractable activity (Fig. 4).

1 2 3 4 5TIME- HOURS

FIG. 4. The kinetics of the appearance of anti-genically cross-reacting material (CRM) to antibodyprepared against highly purified MAV luciferase (8),showing that it closely parallels the activity ofextracta-ble luciferase. Culture and growth conditions similarto those of the experiment ofFig. 1. Values ofordinatesalso the same as in Fig. 1; the CRM values werenormalized.

316 J. BACTERIOL.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CONTROL OF LUMINESCENT SYSTEM

This hypothesis has thus not been supported byexperiment.The third possibility to consider is that trans-

lational control exists. It might be supposed thatthe mRNA for luciferase is uniformly synthesizedas growth proceeds but is not active; that it ac-cumulates and then is called into activity during aspecific phase of growth. A prediction of thishypothesis is that the rapid rise of luminescenceshould be sensitive to inhibitors of protein syn-thesis (chloramphenicol, puromycin, and kana-mycin) but not to those which block synthesis ofmRNA (rifampin or actinomycin D). This is notthe case (Fig. 5 and 6); as measured by its extract-able activity, the synthesis of luciferase is blockedby both types of inhibitors, thus supporting thehypothesis that the phenomenon is due to acontrol mechanism which operates at the level oftranscription. However, the luciferase present atthe time the inhibitor is added persists as extract-able activity.The in vivo luminescence is also very sensitive

to the inhibitors (Fig. 5 and 6). Actually, not onlyis the further increase in luminescence blocked; amarked decrease in the bioluminescence occursafter the addition of either one of the inhibitors.Since the extractable luciferase activity does not

w

zw

(J

TIME - HOURS

FIG. 5. Effect of chloramphenicol on the in vivoluminescence (a) and the in vitro luciferase levels (b)when added at various times, after inoculation, A, B, C,and D, as indicated by arrows. The culture and growthconditions were similar to those ofexperiment ofFig. 1.The control culture (0) was a 500 ml broth culturegrowing in a 2-liter flask. For each addition of chlor-amphenicol, 50 ml of culture was removed and placedin a 300-ml flask containing 5 mg of chloramphenicol.The letters A to D serve to identify the cultures cor-

responding to the various times ofaddition.

z

n

z x

J

TIME - HOURS

FIG. 6. Effect of rifampin on the in vivo lumines-cence (a) and the extractable luciferase levels (b) whenadded at various times during the growth cycle. Theexperimental procedure was similar to that of theexperiment of Fig. 5. A 500-ml broth culture growingin a 2-liter flask was used as the control. At each timeindicated, 50 ml was removed and placed in a 300-mlflask containing I mg ofrifampin.

decrease, this cannot be attributed to luciferaseinactivation. An interesting feature is that thedecrease is more pronounced when inhibitor isadded after autoinduction than if added before.When added afterwards, there is a precipitousdecline in bioluminescence to levels even belowthat of parallel cultures to which the inhibitor hadbeen added at an earlier time. But as noted above,the extractable luciferase remains constant. Thein vivo luminescence of such inhibited cellscould not be stimulated by the addition of cyanideand aldehyde. Thus, the decline after the additionof inhibitors cannot be attributed to the samefactors which are responsible for the spontaneoustransient eclipse of luminescence which occurs innewly inoculated cultures. There is no apparentexplanation available at the present time to ac-count for the in vivo inhibition.The experiments with rifampin and chloram-

phenicol thus indicate that both the luciferasemessenger and the protein itself are synthesizedlate and at a rate faster than other cellular com-ponents, supporting the hypothesis that the con-trol involved is indeed at the level of transcrip-tion.

Studies with cells grown in a minimal mediumgive additional support to this hypothesis (Fig.7). Under these conditions, the cells not onlygrow much more slowly; they also emit muchless light. However, when cyanide and aldehydeare added to the growing cells, it is possible todemonstrate that an appreciable amount ofluciferase is actually being synthesized. Its pres-

VOL. 104, 1970 317

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

NEALSON, PLATT, AND HASTINGS

TIME - HOURS

FIG. 7. Time course oJ growth and luminescenceoJ P. fischeri (MAV) cells in minimal medium. Theordiniate values, plotted on a logarithmic scale, shouldbe multiplied by the factors specified to obtain theexperimental values. Growth was much slower than inconmplex medium and was measured both by viable cellcount (101 cells per ml) and cell density (CD; X1).The fact that in minimal medium there are more cellsper optical density unit at 660 nm is attributed mostlyto the fact that cells growing in minimal medium areconsiderably smaller (microscopic observations). Lu-minescentce (2 X 108 quanta per sec per ml) was meas-ured both in vitro and in vivo. Stimulation by eithercyanide plus aldehyde or aldehyde alone shows that thein vivo emission, which was never very great, cati be sostimulated at any and all times to give an accurate indexofthe luciferase content.

ence and the kinetics of and quantity of its syn-thesis are measured by in vitro assays of lucif-erase activity in extracts.The kinetics of luciferase synthesis are similar

to those which occur in complex medium, in thesense that there is a period after inoculation whenno luciferase synthesis occurs, followed by a burstduring which its synthesis is more rapid thangrowth. This indicates that similar control phe-nomena occur in both media. A plot of the dataof Fig. 7 on a linear scale is shown in Fig. 8.The stimulation which occurs upon the addi-

tion of cyanide and aldehyde indicates that boththe reduced substrate and the endogenous alde-hyde are limiting in cells growing on a minimalmedium, comparable to the situation which wasobserved to occur transiently after inoculationinto fresh complex medium. However, in a mini-mal medium, there is a difference in that theluciferase is never fully active in vivo. The alde-hyde factor and reducing power limit lumines-cence throughout.

Another reason for the failure of cells growing

in a minimal medium to emit a very brightluminescence is related to the fact that the in vivoluminescence peaks at a time when the celldensity is relatively low. Cells grown in minimalmedium are smaller than those grown in com-plex, and the actual cell mass at the time of peakluminescence is quite low. Thus, the burst ofluminescence phenomenon for cells growing incomplex medium (Fig. 2) is even more dramaticin the case of cells growing in the minimal me-dium (Fig. 8). It truly comes and goes during theperiod of exponential growth.

Coffey (1) reported that luminous bacteriagrown in a minimal medium are stimulated toemit much more light after the addition of argi-nine. Arginine appears to be unique in its abilityto evoke this striking response and to act selec-tively upon the luminescent system; no stimula-tion of growth has been detected, even at highconcentrations.We found that the response is attributable not

only to the synthesis of luciferase; after arginineaddition the luminescence of the cells is no longerstimulable by aldehyde, indicating that it alsoinvolves the synthesis of that factor. Argininedoes not alter the stimulability of luminescenceby cyanide. The effect of arginine upon the in vivoluminescence, and its stimulability by aldehydeis shown in Fig. 9. It should be added that thisresponse is fully blocked by inhibitors of RNAand protein synthesis. The data when inhibitorswere added after the primary arginine addition(at 10 hr) have not been plotted; the results wereclosely comparable to those shown in Fig. 5 and 6.The decline in the whole cell luminescence

which occurs subsequent to the maximum canalso be shown to be attributable to a loss of thealdehyde factor, as judged by stimulability (Fig.9). Moreover, the decline in the in vivo lumines-cence can be reversed by a secondary addition of

tO0- 25 C

N VITROw 8 I VIVO

40-

0

TIME - HOURS

FIG. 8. Experiment of Fig. 7 plotted on a linearscale showing that the "pulse" of luciferase synthesisoccurs quite early in the growth, as estimated by thecell density. Thle ordinate values were normalized andare plotted as percent of that value.

318 J. BACTrERIOL.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CONTROL OF LUMINESCENT SYSTEM

TIME - HOURS

FiG. 9. Stimulation by arginine of the in vivo bio-luminescence ofcells growing in a minimal medium. Theaddition of arginine (250 ,ug/ml, final concentration)at the time indicated results in an increase in the in vivobioluminescence. This stimulation can be accountedfor in part by the increased synthesis of luciferase (seeFig. 10) and in part to the synthesis of the aldehydefactor. After the arginine response the addition of aIde-hyde fails to stimulate. After the peak of luminescence,there is not only a cessation of synthesis of the lumi-nescent system components; the in vivo luminescence it-selfdramatically declines and can again be stimulated byaldehyde. As shown, luminescence can then be stimula-ted by arginine; chloramphenicol and rifampin, whenadded along with arginine, again block this effect.

arginine. This stimulation can be attributed tothe synthesis of both luciferase and aldehydefactor, and again the response is blocked byinhibitors ofRNA and protein synthesis (Fig. 9).

In spite of the relatively dim emission of cellsgrowing in minimal medium, the amount ofluciferase synthesized after the autoinduction isquantitatively equal to about 5% (on the basis ofcell mass or protein) of that which is formed in acomplex medium. On the same basis, the amountof luciferase formed after arginine stimulation(Fig. 10) is about twice that which is synthesizedin a complex medium (Table 1).

Coffey (1) had concluded (and with good rea-son) that the arginine stimulation constitutes acase of enzyme induction. When arginine is addedto a culture which has already entered or com-pleted the autoinduction phase, the response israpid and similar to induction. But an authenticinducer should be equally effective, irrespectiveof when it is added. The experiments of Fig. 10show that this is clearly not the case in this system,since the time at which the onset of luciferasesynthesis occurs is essentially independent of boththe time of arginine addition and the amountadded. The arginine effect is essentially a potentia-tion of the autoinduction which occurs in itsabsence; it augments the amount but does notinitiate the synthesis of luciferase and the alde-hyde factor.

TIME - HOURS

FIG. 10. Effect of varying time and quantity of arginine addition upon the synthesis of extractableluciferase. Arginine was added at the three times indicated in the three concentrations noted (10, 100, and1000 ,ug/ml, final). As measured by cell density (X) no differences were detected in the growth rates in thedifferent experiments. The time at which the onset of luciferase synthesis occurred was also unaffected. Moreluciferase was synthesized in the flasks containing the higher arginine concentrations, but the time of additionof the arginine had no influence. Luciferase activity is expressed as cells extracted per ml. Ordinates: celldensity (X 1) and luminescence (2 X 1010 quanta/sec).

VOL. 104, 1970 319

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

NEALSON, PLATT, AND HASTINGS

TABLE 1. Levels ofin vivo luminescence and extractable luciferase obtained during growth on complex,minimal, and arginine-supplemented minimal medias

Luminescence Protein in Quanta Quanta Per QatMedia intensities Cell density Viable count crude

in | Qsec per(quanta per at 660 nm cells per ml crude extract per sec optical per sec per

sec per ml) (m/l e el density unit Mg o rti

ComplexIn vivo 4 X 1012 12.5 5 X 109 1.8 8 X 102 3 X 10") 2.2 X 1012In vitro 4 X 101" 12.5 5 X 109 1.8 8 X 102 3 X 1010 2.2 X 1012

MinimalIn vivo 4.0 X 108 0.15 4 X 108 0.002 0.01 2.5 X 109 1010In vitro 5.0 X 10' 0.15 4 X 10' 0.002 0.1 3.1 X 10' 1.3 X 1011

Minimal + arginineIn vivo 1.2 X 1010 0.4 109 0.005 12 3 X 1010 2.4 X 1012In vitro 2.0 X 1010 0.4 109 0.005 20 5 X 1010 4 X 1012

a Measurements were made in each experiment at the time when the luminescence per cell was at amaximum. In minimal medium, for example, luminescence peaked at a low cell density and the cells weresmaller (compare cell density with the viable cell count). In the last two columns, it is shown that, basedon cell mass or extractable protein, the luciferase content of cells grown in minimal is about 5% of thosegrown in complex, whereas those grown in minimal plus arginine contain almost twice as much as thecells in complex.

DISCUSSIONOn the basis of the experiments presented, we

concluded that the control of the synthesis ofbacterial luciferase is exerted at the level of tran-scription. In freshly inoculated cultures, theluciferase gene (or operon) is repressed or inac-tive; during the exponential period of growth itsactivation occurs, and luciferase is then rapidlyand preferentially synthesized. We referred tothis phenomenon as "autoinduction." Thepresent experiments do not allow us to tellwhether the control is negative or positive.Kempner and Hanson (9) demonstrated that

with cells inoculated into a preconditioned me-dium, i.e., a medium in which other cells hadgrown to a point of exponential light production,the characteristic lag in luciferase synthesis wasabolished. We confirmed this observation andagree that the medium is indeed conditioned dur-ing the initial period of growth; however, fromour experiments the exact nature of the condi-tioning is not clear.With regard to the cellular control mechanism,

we would expect that, if the control of luciferasesynthesis does involve endogenous repressorsand a derepression mechanism (or analogouspositive control elements), mutants in whichthese systems are altered could be obtained. Forexample, if the repression of the luciferase genewere relieved (operator or repressor mutations),the mutant might appear as a "luciferase consti-tutive," in which the enzyme was synthesized atall times, in parallel with the growth of the cells.Screening for such mutants presents difficulties,

since the in vivo luminescence also depends onthe aldehyde factor. Nevertheless, it may beanticipated that such a mutant class will occur.

Actually, there appears to be a similar andsimultaneous control on the synthesis of thedifferent components of the bioluminescentsystem, i.e., luciferase and the aldehyde factor.This kind of coordinate control is reminiscent ofthat which occurs in an operon, and it is notunlikely that the bacterial bioluminescent systemwill be so classified.The control of luciferase synthesis is in some

ways analogous to a developmental phenomenon,since pulses of luciferase messenger occur onlyat a certain growth phase. The luminescent systemrepresents an especially interesting and usefulone for studying such phenomena, for both invivo enzyme activity and the in vivo enzyme con-tent can be continuously and instantaneouslymonitored. These experiments have illustratedthe fact that intracellular enzyme may not alwaysbe fully active, showing that an extractable en-zyme activity cannot always be reliably used toevaluate in vivo enzyme activity.These studies have also served to clarify and

emphasize the similarities and differences in thebehavior of the cells in minimal and complexmedia, especially with regard to the autoinduc-tion of the luminescent system. In both mediathe luciferase gene remains inactive during thefirst part of the growth cycle in shake-flasks andis then activated during the time when a burst ofsynthesis occurs. The fact that this same patternoccurs in both media indicates that the cellular

320 J. BAmrRIOL.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

CONTROL OF LUMINESCENT SYSTEM

control phenomenon is essentially similar in thetwo cases.The development of the luminescent system

differs in the two media primarily in quantitativeaspects. In the minimal medium, smaller quan-tities of both the luciferase and the aldehydefactor are found, thereby resulting in a drasticdiminution of the in vivo light emission.

Arginine acts to relieve or remove the block,whatever it may be, which in minimal mediumrestricts the synthesis of the luminescent system,and is unable to act unless autoinduction hasoccurred. So long as arginine was believed to bean inducer of luciferase, it was difficult to under-stand why no effects of arginine addition couldbe demonstrated in the complex medium. Infact, the complex medium contains approximately400 ,ug of arginine per ml. But since arginine hasno effect in the minimal medium upon the dura-tion of the lag period for luciferase synthesis,none would be expected in the complex medium.Moreover, since the components of the lumines-cent system are evidently synthesized in maximalquantities after autoinduction in the complexmedium, no effects on the amount of synthesiswould be anticipated with added arginine.The experiments of Kempner and Hanson (9)

were all carried out in a complex medium, andthey hypothesized that the initial decrease in thein vivo luminescence should be attributed to theaction of an inhibitor present in the medium andthat conditioning involved its removal. It is diffi-cult to reconcile this hypothesis with the fact thatsimilar phenomena occur also in a minimal me-dium, in which the proposed inhibitor wouldprobably not be present (or at least not identical).Moreover, the fact that the transient decline inthe in vivo bioluminescence can be reversed bythe addition of cyanide and aldehyde indicatesthat the luminescent system is in fact not inhibitedduring this period, as they had suggested, but issimply limited by the supply of reduced substrateand aldehyde.The work of Coffey (1) was done with a bac-

terial strain which would utilize nitrate as the solenitrogen source. He suggested that the failure ofthis organism to emit light on a minimal mediumwith nitrate, as well as the effect of arginine, mightbe a specific property of that strain. However, weobserved that these phenomena are more gen-eral, occurring also with many different strainsof luminous bacteria, including one which willnot utilize nitrate. It will occur also with am-monia as the sole nitrogen source in a minimalmedium, as illustrated by the experimentspresented in this paper.Although the nature of the control mechanism

which acts on the luciferase gene is not known,

its occurrence leads one to expect that the bio-luminescence of these bacteria has some veryspecial biological function. At the same time, theintact luminescent system does not appear to beessential for cells to grow and divide. This is evi-dent when the cells are grown under conditionsin which the repression is maintained semiper-manently (e.g., by repeated subculturing). Lucif-erase is not synthesized under these conditions,and the cells will ultimately "grow out" of theluciferase by dilution. We also found that mu-tants which are deficient in luciferase are fullyviable. This includes both mutants in which theprotein itself is lacking, as judged by the absenceof immunologically cross-reacting proteins andmutants which synthesize defective luciferaseprotein (Cline, personal communication).McElroy and Seliger hypothesized (10) that

the bioluminescent system in the bacteria isvestigial insofar as the light emission is con-cerned. However, we believe that a system con-trolled by truly vestigial genes would not persistwith the characteristics which the luminescentsystem exhibits. For example, at its peak, lucif-erase occurs in large amounts; it constitutes about5% of the soluble protein of the cell (6). Muta-tions in the luminescent system and in the lucif-erase occur readily, both in the laboratory and innature (4). However, activity of and coordinatecontrol over the system persists, suggesting thatthere has been a selection against mutations dele-terious to the system. We thus believe that theluminescent system is in some way advantageousto the bacteria.The precise nature of this function is unknown.

The luminescent system might provide some meta-bolic advantage to the cell, independent of lightemitted (5). On the other hand, the luminescenceitself might be of positive value as, for example,in a symbiotic relationship in which bacterialluminescence is somehow selected for and uti-lized by the host. Certain marine fish possessspecialized luminescent organs in which luminousbacteria are cultured and serve as the source oflight (4). Such a role might constitute not onlythe biological basis for the occurrence andpersistence of the luciferase gene(s) in bacteria;it might also relate to the unusual mechanismwhich controls the synthesis and activity of thisluminescent system.

LITERATURE CITED

1. Coffey, J. J. 1967. Inducible synthesis of bacterial luciferase:Specificity and kinetics of induction. J. Bacteriol. 94:1638-1647.

2. Eymers, J. G., and K. L. Van Schouwenburg. 1937. On theluminescence of bacteria. II. Determination of the oxygenconsumed in the light emitting process of Photobacteriumphosphoreum. Enzymologia 1: 328-340.

321VOL. 104, 1970

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

NEALSON, PLATI, AND HASTINGS

3. Farghaly, A. H. 1950. Factors influencing the growth andlight production of luminous bacteria. J. Cell. Comp.Physiol. 36:165-184.

4. Harvey, E. N. 1952. Bioluminescence. Academic Press Inc.,New York.

5. Hastings, J. W. 1968. Bioluminescence. Annu. Rev. Biochem.37:597-630.

6. Hastings, J. W., W. H. Riley, and J. Massa. 1965. The purifi-cation, properties and chemiluminescent quantum yield ofbacterial luciferase. J. Biol. Chem. 240: 1473-1481.

7. Hastings, J. W. and G. Weber. 1963. Total quantum flux ofisotropic sources. J. Opt. Soc. Amer. 53:1410-1415.

8. Hastings, J. W., K. Weber, J. Friedland, A. Eberhard, G. W.

Mitchell, and A. Gunsalus. 1969. Structurally-distinctbacterial luciferases. Biochemistry 8:4681-4689.

9. Kempner, E. S., and F. E. Hanson. 1968. Aspects of lightproduction by Pholobacterium fischeri. J. Bacteriol. 95:975-979.

10. McElroy, W. D., and H. H. Seliger. 1962. Origin and evolutionof bioluminescence, p. 91-101. In M. Kasha and B. Pull-man, (ed.), Horizons in biochemistry. Academic PressInc., New York.

11. Tsuji, F. 1., and D. L. Davis. 1959. A quantitative photometricmethod for studying the reaction between Cypridina lu-ciferase and specific antibody. J. Immunol. 82:153-160.

322 J. BACTERIOL.

on March 11, 2020 by guest

http://jb.asm.org/

Dow

nloaded from