the ,b-d-galactosidase escherichia coli' …jb.asm.org/content/76/1/1.full.pdf · ·...

REGULATION OF ENZYMATIC ACTIVITY IN THE INTACT CELL: THE,B-D-GALACTOSIDASE OF ESCHERICHIA COLI'

BORIS ROTMAN2Institute for Enzyme Research and Department of Genetics, University of Wisconsin,

Madison, Wisconsin

Received for publication November 1, 1957

Large differences in enzymatic activity areoften found between intact cells and extracts.Lacking a better explanation, it has frequentlybeen assumed that the penetration of the sub-strate is the rate limiting step in the intact cells.The experiments reported here were designed totest several hypotheses which might account forthe discrepancy.The 3-galactosidase (lactase) in Escherichia

coli was selected as a model system because of thesensitivity and ease of assay, together with theextensive knowledge of the induction, isolation,and general properties of the system (Lederberg,1950; Cohn and Monod, 1951; Kuby and Lardy,1953; Rotman and Spiegelman, 1954). Deere(1939) reported that drying E. coli-mutabileappeared to "activate" the lactase. Lederberg(1950) studied the ,B-galactosidase of E. colistrain K-12 with a chromogenic substrate andfound discrepancies of 10- to 47-fold between theactivity of the intact cells and of disrupted cellsor extracts. In addition, Lederberg found that theactivity of the intact cell was protected from pHchanges and inhibitory cations.The hypotheses tested in this investigation

were: passive diffusion, inhibitors complexedwith the enzyme, and product inhibition of theenzyme. All three were ruled out by the datapresented below. Consequently, a more com-patible proposal is given which assumes theexistence of a specifically induced penetrationmechanism. During the course of this work,Monod and his associates postulated an inducibleenzymelike mechanism for specific control overgalactoside penetration and termed the agent a"permease" (see Cohen and Monod, 1957). The

1 Paper no. 676 of the Department of Genetics.Supported by an American Cancer Society Insti-tutional Grant and by a grant from the UnitedStates Public Health Service.

2 Present address: Instituto de Quimica Fisio-logica, Borgono 1470, Escuela de Medicina,Santiago, Chile.

experiments described here were completed beforethe data supporting the permease hypothesisbecame available and therefore were not designedto test this hypothesis. Nevertheless, when ourfindings and the permease data were compared,basic similarities and differences between thetwo hypothetical penetration systems becameapparent. Possible ways of reconciling the twoviews are discussed.

MATERIALS AND METHODS

Bacterial strains and culture conditions. To avoidthe additional variable of adaptation, the strainW-1317, a constitutive mutant (K-12) of E. coliwas employed for most of the experiments. Thismutant, obtained by selection on neolactoseagar, synthesizes large amounts of g-galactosidasein the absence of external inducer (Lederberg,1951). Experiments involving adaptation wereperformed with the parent K-12 stock. Davis'(1949) minimal medium with 0.1 per cent glycerolas the carbon source was used throughout. Unlessotherwise specified, the cells were grown in con-tinuous culture in a modified chemostat (Rotman,1955a) with an average generation time of 3 hr.For most of the experiments, 12 ml of culturewere withdrawn from the growth tube. The cellswere then washed twice with 20 ml of cold 0.02M sodium phosphate buffer at pH 7.2 by centrif-ugation at 14,000 X G for 3 min in the highspeed head of an International centrifuge. Thecells were resuspended in 12 ml of the same bufferto give an average nitrogen content of 70 j,g perml.Some experiments required the use of bacteria

freed of exogenous salts. Repeated washing withwater proved inefficient in that the bacteriaceased to pack in the centrifuge even at highspeed. Consequently, the cells were passedthrough an ion exchange column as described byRotman (1956). The deionization procedure andsubsequent storage in distilled water at 0 C didnot impair cell viability and recoveries were

1

on June 4, 2018 by guesthttp://jb.asm

.org/D

ownloaded from

http://jb.asm.org/

ROTMAN

quantitative. After this treatment, less than0.04 ppm of sodium or potassium were found to bepresent in the supernatant by flame photometry.

Chemical reagents.3 o-Nitrophenyl-f-D-galacto-pyranoside was synthesized according to themethod of Seidman and Link (1950) or procuredfrom Mann Laboratories. p-Nitrophenyl-f-D-galactopyranoside, p-nitrophenyl-a-L-arabino-pyranoside, and o-nitrophenyl-a-L-arabinopyran-oside were kindly supplied by Dr. K. P. Link.Methyl-f3-D-thiogalactopyranoside was obtainedthrough the kind assistance of Dr. A. Novick.Galactose (Pfanstiehl) was recrystallized fourtimes from aqueous ethanol and dried at 60 Cunder vacuum for 24 hr. Lysozyme and ribo-nuclease were obtained from Armour Co. Thelactose was from Difco Laboratories, but allother reagents were analytical grade. Waterwas deionized and then glass distilled.

Analytical methods. 3-D-Galactosidase wasmeasured with ONPG as a chromogenic substrate(Lederberg, 1950). The test solution (0.05 to 0.2ml) was added to 2 ml of M/1000 ONPG dissolvedin the buffer described above. The reaction wasstopped after 1 to 15 min incubation at 30 C byaddition of 10 ml of 0.1 M sodium carbonate.The o-nitrophenol was determined colorimetri-cally at 420 m,u. Enzymatic activity was expressedas m,umoles ONPG hydrolyzed per minute.

Benzene-treated cells were prepared by shaking2 ml of a washed suspension of cells with 0.1 ml ofbenzene. After incubation for 10 min at 30 C,0.05 ml was withdrawn from the tube and assayedas indicated above. A 0.2 ml sample of the un-treated suspension was used to assay the enzy-matic activity of the intact cells. The cells werecentrifuged and removed after stopping thereaction with sodium carbonate. Blanks wereprepared by adding the cells after the sodiumcarbonate. The enzymatic activity obtained whencells were treated with benzene was taken as thestandard. The activity of the intact cells showedlinear kinetics on a Michaelis-Menten plot up to aconcentration of 3.4 X 10-3 M ONPG, therebyconfirming and extending Lederberg's (1950)original observations.

Ribose was determined by the oreinol test(Kerr and Seraidarian, 1945) and desoxyribose

3 Abbreviations: ONPG = o-nitrophenyl-/3-D-galactopyranoside, DNA = desoxyribose nucleicacid, RNA = ribose nucleic acid,,-galactosidase= O-D-galactosidase.

by the diphenylamine test (Sevag et al., 1940).Separation of nucleosides, nucleotides, and freebases was accomplished by the method of Cohn(1950). Protein was measured by the biuretreaction (Gornall et al., 1949).

EXPERIMENTAL RESULTS AND CONCLUSIONS

Increase in enzymatic activity by treatment withsolvents. (1) Benzene-In agreement with otherinvestigators (Lederberg, 1950; Koppel et al.,1953) treatment of the cells with benzene ortoluene in 0.02 M Na-phosphate buffer, pH 7 to7.3, yielded enzymatic activity similar to thatobtained with cell-free extracts prepared bysonic disintegration or drying in vacuo. Fiveminute treatments with 5 per cent (v/v) benzene,including shaking once or twice by hand, broughtthe enzymatic activity of the suspension to itsmaximal value without releasing more than 5 percent of the enzyme into the supernatant. Thebulk of the enzyme was retained by the "cells."Vigorous mechanical shaking provided maximalactivity in about 15 sec at room temperature.The reaction appeared to be temperature inde-pendent in the range 0 to 40 C.

Benzene-treated cells do not appear differentfrom normal cells under phase contrast micros-copy, but when Giemsa stain is used the treatedcells are less basophilic than the normal ones,suggesting the loss of ribonucleic acid.Under the electron microscope the difference

becomes evident as shown in figure 1. Thedestruction of the cell wall in the benzene-treatedcells is quite apparent but the possibility that thedrying process might have increased the damagemust be considered.

In contrast with the f-galactosidase of intactcells, the benzene-treated cells and the freeenzyme were similar in respect to pH optimum,K, value, and inhibition by alkali metal ions.The only detectable difference was that in ourhands benzene-treated cells did not bind signifi-cant amounts of specific rabbit antienzyme.Throughout this work, therefore, benzene-treatedcells were used as reference for assay of totalenzymatic activity.

(2) Other solvents-The low solubility in waterof benzene or toluene limits somewhat the useful-ness of these solvents. Therefore a search for amore suitable agent was conducted. Isoamylalcohol was chosen because of its higher solubilityat low temperatures and because it has a low

2 [VOL. 76


.org/D

ownloaded from

http://jb.asm.org/

REGULATION OF ENZYMATIC ACTIVITY IN THE INTACT CELL

Figure 1.* Top. Normal cells of Escherichia coli. Center. Cells after 10 min treatment with benzene.Bottom. Benzene-treated cells as above but incubated 10 more min with substrate, alizarin-,B-D-galacto-pyranoside.

* Uranium shadowing. Electron photomicrographs courtesy of Dr. Paul J. Kaesberg.

19581 3


.org/D

ownloaded from

http://jb.asm.org/

ROTMAN

ultraviolet absorption and would not interferewith the spectrophotometric estimation ofpurines and pyrimidines. Table 1 shows that onlyrelatively insoluble hydrocarbons or aliphaticalcohols give maximal activation under the con-ditions employed.

Since isoamyl alcohol disappears during theprocess of activation, a minimum of about 6 X10-12 g of isoamyl alcohol per cell was necessaryat 0 C for maximal enzymatic activity.

Physiological state of the cells influencingenzymatic activity. From Lederberg's (1950) workand our earlier experiments it was evident thatthe ratios between enzymatic activities of ex-tracts and intact cells were not reproducible. Asshown in table 2, the variable was found to be theage of the cultures. Stationary cells show nearlythe same enzymatic activity as their extracts,whereas cells in the exponential phase of growthexhibit only a small fraction of their potentialactivity, e. g., the activity ratio between theextract and the intact cells is 50 to 100. By usinga chemostat and keeping the culture in constantgrowth it was possible to obtain reproduciblevalues for the activity ratio. Furthermore, it wasestablished that large variations in growth ratein the chemostat did not influence this ratiosignificantly (table 2). This observation supports

TABLE 1

Increase in enzymatic activity of cultures bytreatment with solvents

One ml of bacterial suspension, prepared inphosphate buffer as described in Methods, was in-cubated for 10 min with 0.2 ml of the solventtested. The ,3-galactosidase activity is expressedas per cent of the activity obtained with benzene.

Solvent ,6-Galacto-sidase

No addition.........................Benzene, toluene, xylene.............Ethyl alcohol.......................s-Butyl alcohol.....................t-Butyl alcohol......................1-Pentanol ...........................

Isoamyl alcohol.....................Octyl alcohol.......................5% Phenol..........................5% Na Lauryl sulfate................Ethyl ether..........................Acetone ..............................

3.610033

529796583237788

TABLE 2

Relationship between age of cultures and enzymiaticactivity of whole cells

,3-Galactosidase activity was determined inwashed cells with and without benzene treatmentas described in Methods. The activity is expressedas the ratio between benzene-treated cell and in-tact cell activities.

Ratio ,a-Ga-Growth Conditions and Age of Cultures lactosidase

Activity

Stationary, aerated, 12 hr............ 21Stationary, not aerated, 24 hr........ 24Stationary, not aerated, 37 hr........ 5Stationary, not aerated, 47 hr........ 3Chemostat, 1 hr generation time. 173Chemostat, 1.4 hr generation time.... 166Chemostat, 7.7 hr generation time... . 124Chemostat, 11.1 hr generation time... 161

the assumption that in the steady state the vastmajority of the cells are actively growing even atlow generation times.

Passive-diffusion hypothesis. Induced biosyn-thesis of enzyme (adaptation) and cellular expres-sion of enzymatic content. An opportunity to testa simple passive diffusion hypothesis is offered bythe fact that the ratio of intact cell activity toextract activity changes with the 3-galactosidasecontent of the cell (Rickenberg et al., 1953). The,-galactosidase content per cell can be altered(Monod et al., 1951) by varying the concentrationof a suitable inducer of the enzyme in the wildtype E. coli.The proposed diffusion hypothesis postulates

that the rate of penetration of the substrate is therate-limiting step when the cell expresses onlya fraction of its enzyme content. The hypothesiswas tested under the following assumptions:(a) Inasmuch as passive diffusion is under test, itcan be assumed that the rate of penetration isconstant if conditions of "gratuity" are usedthroughout the experiment, i. e., the medium hasa constant chemical composition except for theconcentration of the nonmetabolized inducer(Monod and Cohn, 1952). Other possible pene-tration mechanisms which fulfill this assumptionwill not be discriminated by our test; (b) Thephysical properties of the enzyme inside the cellare similar to those of the enzyme in cell-freeextracts; and (e) The concentration of substratearound the cell remains practically constant,

4 [VOL. 76


.org/D

ownloaded from

http://jb.asm.org/


i. e., diffusion rate in the medium is much largerthan the entry rate of the substrate. This assump-tion is valid because benzene-treated cells exhibitactivities up to 170-fold larger than intact cells.If we recall that a benzene-treated cell containsthe same amount of enzyme enclosed in practicallythe same volume as an intact cell, we have toconclude that the rate at which substrate reachesthe periphery of the cell is not the limiting step.It should be noted that for the purpose of ourdiscussion "penetration of the cell by the sub-strate" could mean also arrival of the substrateat the enzyme site.The same concentration of substrate was used

in all the assays and it remained practically con-stant throughout the experiment.The hypothesis thus formulated predicts the

curves illustrated in figure 2A. The predictionsare: (a) At low levels of induction, the intact cellshould express its full enzyme content until sub-strate penetration becomes rate limiting, i. e.,throughout this range, enzymatic activity of cellsand extracts should be the same, and, (b) Aftersubstrate penetration becomes rate limiting asindicated by the lower activity of cells comparedwith extracts, the enzymatic activity of theintact cells should remain constant. Thereforethe ratio of extract activity to cell activity shouldincrease progressively with the increase in theenzyme content of the cell.A test of these predictions is shown in figure

2. The experimental curves in figure 2B differed

A. Theoretical curves

radically from the theoretical (2A). In no portionof the former was either prediction satisfied. Thusthe intact cell expressed little or none of theenzyme content when this was small and athigher induction levels expressed a constantfraction of the enzyme synthesized. The intactcell activity did not reach a plateau but increasedsteadily with increasing enzyme content. Finally,the ratio of extract to cell activity reached amaximum long before the enzyme content did.Thus one must conclude that the passive diffusionhypothesis as stated here is incompatible with thefacts. Any other hypotheses that do not involve achange in the rate of substrate entry duringenzyme induction would be equally incompatible.The same test was applied to the constitutive

strain which synthesizes as much enzyme in theabsence of inducer as does the parental stock inthe presence of optimal concentrations of potentinducers. The enzyme content of the constitutivestrain was varied by inhibition of the f3-galac-tosidase synthesis with lactose. It was found thatthe lower the galactosidase content, the largerwas the fraction of enzymatic activity expressedby the intact cells. Therefore passive diffusionhere too must be ruled out as a possible penetra-tion mechanism.

The enzyme complex hypothesis. The secondhypothesis to be tested was the existence in thecell of an enzyme complex which reduced enzy-matic activity. In this case, the increased activityof extracts or benzene treated cells would result

Enzyme per ml of culture(extracted) mjmoles ONPG/min

Figure 2. j3-Galactosidase activity of whole cells as a function of their jl-galactosidase content. Theo-retical curves constructed according to the passive-diffusion hypothesis described in the text. Curve I.,3-galactosidase activity of whole cells in m,umoles of o-nitrophenyl-3-D-galactopyranoside (ONPG)hydrolyzed per min per ml of culture. Curve II. Ratio ,B-galactosidase activity of extract over ,1-galac-tosidase activity of whole cells. Increasing concentrations of lactose or methyl-,8-D-thiogalactopyrano-side were used to induce the cultures.

1958] 5


.org/D

ownloaded from

http://jb.asm.org/

ROTMAN

from the dissociation of the complex. Dissociationcould occur when the internal milieu of the cellwas no longer protected from external exchange.

Previous experiments (Rotman, 1955b) hadsuggested that RNA might be the complexingagent. In the present work, the test of this sup-

position was confined to (a) attempts to isolatethe hypothetical enzyme-RNA complex fromthe cells and to show its in vitro dissociation witha comparable increase in enzymatic activity, and(b) attempts to demonstrate the inhibitory powerof the RNA in vivo and in vitro.The hypothesis could be extended to include

other complexing molecules such as the enzyme

itself (Kaplan, 1954; Bonner, 1955). A strict testof the latter however would require methods toanalyze intracellular structure not available atpresent.

Leakage of breakdown products of RNA. Asmentioned before, the lowered basophilia ofbenzene-treated cells suggested a substantialloss of nucleic acids. Spectrophotometric andchemical analysis confirmed that as much as 80per cent of the total RNA content of the cell waspresent outside. The RNA loss was accounted forby comparable amounts of breakdown productsof RNA. Small amounts of free bases and poly-merized material were among the breakdownproducts. The relative composition of productsdepended upon the conditions of the benzenetreatment. No detectable amounts of DNA or

protein were found outside the benzene cells.Isoamyl alcohol, like benzene, caused a release ofbreakdown products of RNA.RNA breakdown products were also found to

be released from the cells upon aging with a con-

comitant increase in enzymatic activity. Thecells remain viable throughout the aging process.

Borek et al. (1955) reported also that the viablecount of a cell suspension can remain constantwhile nearly 40 per cent of the RNA is excretedas breakdown products in the supernatant.During aging, addition of Mg++ or Ca-++ ions

in 8.2 X 10-4 M concentration prevented theleakage of breakdown products of RNA as wellas the increase in enzymatic activity. Storage indistilled water at 0 to 4 C was also effective inpreventing leakage.The relationship between RNA release and

increase in enzymatic activity of the intact cellswas tested further using lysozyme, polymyxin B,and heating at 51 C. Polymyxin (Newton, 1956)

and heating at 51 C (Califano, 1952) have beenpreviously shown to cause leakage of depoly-merized form of RNA. Lysozyme alone has notbeen reported to cause any direct effect on E. coliat neutral pH, although recently it has beenshown to cause lysis at pH 7.6 in the presence ofVersene (Repaske, 1956). Treatment with lyso-zyme released half of the RNA in polymerizedform precipitable by HC104. This effect of lyso-zyme was inhibited 70 per cent by 8 X 10-5 Muranyl acetate.

Furthermore, when the release of RNA bypolymyxin was inhibited by high concentrationsof the antibiotic or by divalent cations (Newton,1956) the enzymatic activity was inhibitedproportionally. Cells treated with either lyso-zyme or polymyxin exhibited about 15 percent of the maximal f-galactosidase activity. Onthe other hand, when both agents were used inthe order indicated maximal activity was ob-tained. Reversing the order or adding the twotogether resulted in reduced activity (table 3).Cells heated at 51 C did not show additive effectswith either lysozyme or polymyxin. Lysozymetreatment resulted in no detectable loss in viablecells, whereas 14 min at 51 C killed 85 per centof the cells and polymyxin left only a few sur-vivors.As shown in figure 3, treatment with any of

these agents resulted in a release of RNA pro-portional to the increase in ,B-galactosidase.Although this proportionality between RNA

excretion and 3-galactosidase activity was foundwithin every set of experiments involving a singleactivating agent, the amount of RNA releaseddid not correspond to the same enzymatic activityfor experiments with different activators. Forinstance, when isoamyl alcohol released 237,ug of RNA/mg dry cells with an increase inenzymatic activity of 4520 mumoles ONPG/minmg dry cells, aging in 0.02 M sodium phosphatebuffer for 8 hr released 23.6 ,ug of RNA/mg drycells with an increase of 119 mAmoles ONPG/minmg dry cells.

This lack of correlation between activatingagents suggests that the release of RNA and theincrease in enzymatic activity could be the resultof a more general effect, i. e., loss of permeability,rather than a combination of RNA and enzymeas postulated before (Rotman, 1955b). Furtherevidence of this hypothesis is given below.

Preparation of inhibited enzyme in vitro. It was

[VOL. 76


.org/D

ownloaded from

http://jb.asm.org/


TABLE 3

Synergistic effect of lysozyme and polymyxin uponthe increase in enzymatic activity of whole cellsChemostat-grown cells were washed and re-

suspended in phosphate buffer as described inMethods. Three tubes containing 2 ml of cellsuspension plus the additions indicated belowwere incubated 30 min at 37 C. After this firstincubation, 0.1 ml of each tube was used to deter-mine j3-galactosidase activity and the rest of thecells were spun down and resuspended in 2 ml ofthe same buffer. Aliquots of each tube, with andwithout additions, were incubated at 37 C foranother 30 min. ,-Galactosidase was determinedin each aliquot at the end of this second period ofincubation. The results are expressed as per centof the,-galactosidase activity of benzene-treatedcells. Activity of cells before incubation at 37 Cwas 1.1 per cent. Activity of supernatants fromtubes 2 and 3 after the second incubation was 1.2and 0.3 per cent, respectively.

,-Galacto-sidaseTube No. Additions RiveRate

Firstincubation

1 0.2 ml H20 4.72 0.2 ml polymyxin B 22.6

(0. 5 mg/ml)3 0.2 ml lysozyme (12.5 14.9

mg/ml)Second

incubation1 No additions 8.02 No additions 22.62 0.1 ml lysozyme (50 57.5

mg/ml)3 No additions 20.03 0.1 ml polymyxin B 100.5

(0.5 mg/ml)

03-

(N 02

a

0.1

.0.I,)

C

5,4

0 50 100Increase in enzymatic activity mpmoles/min/ml

Figure S. Relationship between release of RNA-breakdown products in the supernatants andactivation of the ,B-galactosidase of whole cells.A. Cells aged in 0.02 M Na-PO4 (pH 7.35 at 26 C)for 3 hr at 37 C. B. Cells treated with polymyxinB in decreasing concentration. The lowest increasein activity corresponds to 830 ,ug of polymyxinper ml, the highest to 210,Ag per ml. C. Cells heatedat 51 C in 0.02 M Na-PO4 (pH 7.2 at 25 C) for 15min.

rnaximumn activity.-'Sin the extract

possible to prepare extracts containing,-galacto-sidase in an inhibited form. The inhibition couldbe overcome by addition of salts or by incubationat 37 C (figure 4). The key to the preparation ofinhibited extracts is to avgid the use of saltsthroughout the extraction procedure. The cellswere collected in a refrigerated Sharples centri-fuge, and after three washings with water at15,000 rpm in a Spinco (head no. 20), they were

ground with dry ice and lyophilized. The extractswere prepared simply by grinding 0.5 g of drypowder in a cold mortar with 5 ml of cold water

Molarity of Na-PC4 (pH 6.47)

Figure 4. Increase in the il-galactosidase activityof inhibited cell-free extracts by addition of salts.See the text for preparation and assay of inhibitedextracts. The pH of the buffer was measured at0 C and it was chosen to match the pH of theextract.

and eliminating the debris by 8 min centrifugationat 10,000 rpm in the high speed head of an

International centrifuge.These extracts were assayed at 0 C by ad(ding

1958] 7


.org/D

ownloaded from

http://jb.asm.org/

ROTMAN

0.05 ml of 0.05 M ONPG (previously equilibratedat 0 C) to 0.1 ml of extract and stopping thereaction with 10 ml of 0.1 M Na2CO3. The assaywas proportional to incubation time up to 6 min.

3-Galactosidase activity in these inhibite(dextracts was not diminished by 60 min centrifu-gation at 40,000 rpm in the no. 20 head of aSpinco centrifuge. Thus, the association of theenzyme with a particulate matter is ruled out asa cause of inhibition.That the inhibition in these extracts is not of

the known type produced by alkali metal ions(Lederberg, 1950) was shown by the fact thatthe initial salt concentration affected the enzy-matic activity of aliquots of extract reachingidentical final conditions prior to assay (figure 4).The relationship between the inhibited ex-

tracts and the physiological problem in questionis obscure because the cells which had beentreated previously with isoamyl alcohol, andwere therefore capable of expressing their entireenzyme content, nevertheless yielded inhibitedextracts.

Inhibitors of 3-galactosidase released by the cells.According to the enzyme-complex hypothesis thecomplexing agent should be found free afteractivation of the enzyme. An unsuccessfulsearch for the complexing agent was conductedamong the excretion products of cultures whichhad leaked most of their RNA with a proportionalincrease in ,B-galactosidase activity. A 5-L cultureof the constitutive mutant was treated withisoamyl alcohol and the supernatant was concen-trated from the frozen state under vacuum.

Several fractions obtained by precipitation ofthe concentrated supernatant with ethanol weretested for inhibition of cell-free enzyme or ben-zene-treated cells. None was found inhibitory.Likewise yeast RNA was not inhibitory in eithersystem. The effect of highly polymerized RNAfrom E. coli could not be determined. No reportsof a successful method of preparation could befound in the literature. The difficulty probablystems from the presence of active RNAase(s) inthe extracts (Manson, 1953).

The product inhibition hypothesis. This hypoth-esis explains the reduced activity of the intactcell in terms of inhibition of the enzyme by thereaction products. Galactose, one of the products,is known to be a competitive inhibitor of theenzymatic hydrolysis of ONPG. High concen-trations of galactose could accumulate in the

intact cell if its excretion rate happened to besmaller than the hydrolysis rate of ONPG. Theincreasing concentration of galactose wouldinhibit the enzymatic reaction progressivelyuntil a steady state had been reached in which therates of hydrolysis and excretion of galactoseremained constant. This hypothesis predicts thatthe initial enzymatic activity of the intact cellbefore accumulation of reaction products wouldbe larger. The prediction was tested by followingthe initial rates of hydrolysis of ONPG by a con-centrated cell suspension in a recording spectro-photometer of very high sensitivity. A doublebeam apparatus which recorded the difference inoptical density between 440 and 480 m,u wasused (Chance, 1956). These experiments wereperformed in collaboration with Dr. BrittonChance at the University of Pennsylvania.The data summarized in figure 5 show that a

steady rate of hydrolysis had been attained by the

4-'3

a)

b

c

L.

time (sec)Figure 5. Hydrolysis of o-nitrophenyl-,3-D-

galactopyranoside (ONPG) by intact cells meas-ured in a double-beam recording spectropho-tometer. The instrument recorded the differencebetween optical densities at 440 and 480 m,u.

8 [VOL. 76


.org/D

ownloaded from

http://jb.asm.org/


cells as early as one second after substrate addi-tion. Extrapolation to zero time permitted theestimate that the concentration of o-nitrophenol(and therefore of galactose) in the total suspen-

sion at the end of the first second could not haveexceeded 1 X 106 M. Since the product inhibitionhypothesis would demand that a steady state of

enzymatic inhibition should already have beenestablished at this time one may ask if theamount of galactose that was present was

sufficient to have brought this about. The intra-cellular volume of the suspended cells was deter-mined to be 6.3 X 10-3 ml per ml suspension.Thus even were all the galactose that had beenproduced in one second still within the cells,its concentration could have been no more than(1 X 10-6)/ (6.3 X 10-3) = 1.6 X 10-4 M. Butthe data of Kuby and Lardy (1953) for the invitro inhibition of highly purified ,B-galactosidaseshowed that 0.4 M galactose is required for 95per cent inhibition in the presence of 2.28 X 10-4M ONPG (the amount used in the present experi-ments). It follows then that this concentration ofgalactose is (0.4)/(1.6 X 10-4) = 2500-foldgreater than could have been generated in thecells in one second. In other words, the amountof galactose that actually was present wouldhave had to be crammed into ½1500 of the totalcell volume to have been inhibitory. (Assuminga significant effusion of galactose from the cellsduring this time, the fraction would be even

lower.) But this is patently impossible when it

is considered that the highly purified f-galac-tosidase represents 0.5 to 1 per cent of the totalcell protein (Rotman and Spiegelman, 1954) sothat the volume of enzyme itself is greater than½f500 of the total cell volume.It should be noted that the only assumptions

underlying this argument are that the enzyme

inside the cell behaves as the purified enzyme

and that the substrate concentration inside thecell is the same as that outside.

Penetration mechanim hypothesis. p- ando-Nitrophenyl-a-L-arabinopyranoside have alsobeen found to act as substrates for the f-galac-tosidase (Lederberg, private communication).These compounds have the same ring structureas the f3-D-galactopyranosides. They differstrikingly in that intact cells exhibited the same

activity against them as did extracts (table 4).The full expression of enzyme content by theintact cell excludes immediately the idea of an

altered (Kaplan, 1954), inhibited (Rotman,1955b), or stacked enzyme (Bonner, 1955). It isnot likely that the enzyme would be compoundedin such a form that binding sites for the arabi-nosides would be free but those for the galacto-sides would be obstructed. The same reasoningapplies to the stacked enzyme.

In addition, the difference in behavior of theintact cell toward the two classes of substratescannot readily be ascribed to different rates ofpassive diffusion because they are so similar struc-turally. It therefore seems more likely that a se-

TABLE 4

Effect of Na and K ions in the hydrolysis of fl-D-galactosides and a-L-arabinosides by jl-galactosidase

#-Galactosidase activity was determined as described in Methods at 10-3 M substrate concentration.Na or K phosphate (0.02 M) buffer pH 7.2 (25 C) were used in the assay mixture. The results on the

effect of ions are given as the ratio j3-galactosidase activity in Na+ buffer over ,8-galactosidase activityin K+ buffer. The values for induced wild type whole cells are the results of a single experiment.

Ratio Na+ Activity/K+ Activity mumoles of Substrate Hydrolyzed permmn per t&g N in Na+ Buffer

Culture Substrate Substrate

ONPG* ONPA* ONPG* ONPA*

Constitutive whole cells..1.03 3.30 1.73 0.72Wild type induced whole cells.1.01 3.35 1.79 0.62

C6H6-treated constitutive. 1.75 5.55 66.7 1.69

C6Hs-treated induced wild type. 1.83 4.10 57.8 1.36

Noninduced wild type whole cells 1.35 - 0.03 UndetectablePurified j8-galactosidase. 1.78 5.60 960 27.9

* ONPG = o-nitrophenyl-I3-D-galactoside; ONPA = o-nitrophenyl-a-L-arabinoside.

91958]


.org/D

ownloaded from

http://jb.asm.org/

ROTMANV

lective penetration mechanism is involved. To ac-count for the increase in enzymatic activity of theintact cell with increase in enzyme content, onewould have to assume that the penetration mech-nism is induced together with the enzyme (figure2B). Because of its inducibility and specificity,one might visualize the penetration mechanism asenzymatic in nature although it could equallywell act as a specific "keyhole" at the cell surface.

Further evidence in support of the abovehypothesis was obtained by a study of metal ionsknown to inhibit the galactosidase in vitro.Lederberg (1950), using ONPG as a substrate,reported that Na ions did not affect the intactcell activity although they could activate3-galactosidase in extracts as compared with K

ions. We have extended these results usingp-nitrophenyl-f3-D-galactopyranoside. Again, no

effect of ions upon intact cell activity towardthis substrate could be noted but the effects ofNa and K were precisely reversed in extracts.When the arabinosides were employed, metalions were found to influence both intact cellsand extracts (table 4). A similar situation was

encountered in noninduced cells with only back-ground enzyme.

Regardless of the nature of the substrate,intact cells did not exhibit any /3-galactosidaseactivity whatever when assayed in distilled water.The activity was restored upon addition ofmaleic or phosphoric buffers. Although Rb, K,Na, Li, etc. were innocuous toward intact cellswhen they were acting upon ,B-galactosides,triethanolammonium ions were potent inhibitors.Kuby and Lardy (1953) reported that this ioninhibits the purified /-galactosidase.From these results it seems that the metal ions

are inert to the intact cell activity on /-galac-tosides not because the ions cannot reach theenzyme site, but because they do not interferewith the penetration mechanism. This idea isreinforced by the inhibitory effect of Tris bufferwhich should not exist if the ions cannot reachthe enzyme site. The a-L-arabinosides wouldhave such affinity for the ,B-galactoside penetra-tion mechanism as to permit a relatively rapidentrance. Thus, for the arabinosides, the,B-galactosidase would be the rate limiting stepin vivo. The possibility of a separate penetrationmechanism specific for the arabinosides can notbe excluded. Nevertheless, its postulation addsnothing to our hypothesis. On the contrary,

to set aside the j3-galactoside penetrationmechanism, it would be necessary to accept theexistence of a specific penetration mechanism forarabinosides but not for galactosides.The fact that intact cells placed in a mixture

of ONPG and 0.1 M Na2CO3 still can hydrolyze aminute amount of ONPG is additional evidencein favor of the penetration mechanism. Theexperiment indicates that ONPG penetrates thecell faster than OH- ions do. Similarly, HCI,trichloracetic acid, or formaldehyde do not stopthe hydrolysis in vivo completely.The inference from the a-arabinoside experi-

ments is valid only if the /3-galactosidase is themain enzyme capable of splitting a-L-arabi-nosides. This assumption is strongly supportedby two lines of experimental data: (a) The ratiobetween the rates of hydrolysis of o-nitrophenyl-a-L-arabinopyranoside and ONPG, underconstant conditions, is the same for benzene-activated cells, crude extracts, or purifiedf-galactosidase. The average ratio at 10-3 Mconcentration of substrate for benzene-activatedcells was 39.4 and for purified enzyme, 34.3. Theratio given by Kuby and Lardy (1953) forhighly purified ,B-galactosidase is 38. (b) Cellsinduced with L-arabinose (a poor inducer) showtraces of ,B-galactosidase and no detectablea-L-arabinosidase (to be expected because of the30-fold difference in sensitivity). Extracts of,B-galactosidase constitutive mutants grown inthe presence of succinate and L-arabinose showthe same hydrolysis ratio of arabinosides andgalactosides as given above. If an active a-L-arabinosidase exists it would have to be in-duced by galactose but not arabinose, a veryunlikely event.Another possible objection to the induced

penetration mechanism hypothesis stems fromthe fact that a-L-arabinosides are hydrolyzed byextracted enzyme at a smaller rate than /3-galac-tosides. The results with arabinosides could thenbe interpreted as if the ,B-galactosidase becomesthe rate limiting step at low enzymatic activities.This interpretation is incorrect because wholecells with the same enzyme content, in terms ofmoles of substrate hydrolyzed per min whetherit was a galactoside or an arabinoside, show re-duced activity only when #-galactosides wereused.An analysis of the data obtained in the previous

experiments from the standpoint of the inducible

10 [VOL. 76


.org/D

ownloaded from

http://jb.asm.org/


penetration mechanism is warranted. Moreover,such analysis yields a forecast of the propertiesof the hypothetical penetration mechanism whichshould be useful for its future isolation.The increase in enzymatic activity, using

ONPG, which occurs with organic solvents,lysozyme, polymyxin, heat, or aging of the cellwould have to be explained by a nonspecificincrease in permeability. This assumption wouldfit well with the proportional excretion of RNA-breakdown products which occurs concomitantly,indicating a loss of the ability to retain cytoplasm.

Interpreting the results shown in figure 2Bunder the inducible penetration mechanismhypothesis one can conclude: (a) The data con-forms with the hypothesis. (b) The gB-galacto-sidase is induced at lower inducer concentrationthan the penetration mechanism to account forthe early part of the curve in which the enzymecontent increases more rapidly than the intactcell activity. (c) The penetration mechanism,after a certain period, is induced linearly withinducer concentration, at the same rate as theenzyme. This accounts for the proportionalincrease of the intact cell activity.The inhibited enzyme found in extracts cannot

be readily explained under any hypothesisbecause of its presence in extracts of cells whichcould display activity. The possibility of anartifact is not unlikely.The experiments with short reaction times

performed at Dr. Chance's laboratory do notcontradict the inducible penetration mechanismhypothesis. They would indicate that the entryof the substrate by way of the penetrationmechanism follows zero order kinetics from itsinitial velocity.The inertness of alkali metal ions for the en-

zyme in vivo indicates that, in contrast with the,B-galactosidase, the penetration mechanism,which in vivo is the rate limiting step, is notinfluenced by alkali metal ions. Nevertheless,the penetration mechanism needs some salts tofunction since it is inhibited by triethanolaminebuffer.

DISCUSSION

Four hypotheses which could account for thepartial expression of the galactosidase in vivohave been proposed. Of them, only the one whichpostulates an inducible penetration mechanismspecific for ,B-galactosides was found compatible

with all the data. The experiments which ruleout the other three hypotheses are conclusiveand the few assumptions used in them were sup-ported by subsequent experimental evidence.In other instances, specific induction of trans-

port systems, required to bring substrate to theenzyme site, have been postulated. It was usedto explain (Barret et al., 1953) the process ofcitrate utilization in Pseudomonas fluorescens andin Aerobacter aerogenes (Davis, 1956). Thus, inA. aerogenes intracellular accumulation of citratewas shown to occur only after specific induction.Monod and his collaborators have demonstrateddirectly in Escherichia coli a number of specificsystems which accumulate amino acids (Cohenand Rickenberg, 1955) and an inducible one whichaccumulates ,B-thiogalactosides (Rickenberg etal., 1956). They have postulated an enzymelike"permease" which controls the entrance of thesubstance. The exit rate of the substance wasassumed to depend upon the internal concentra-tion. Thus it is explained how an accumulationoccurs and how a steady state between entry andexit is obtained. The "permease" which accu-mulates ,B-thiogalactosides has been shown to beinduced specifically by the same inducers as the,B-galactosidase although it is clearly differentfrom the latter. The different substances accu-mulated by the system compete with each otherin the same fashion as substrates compete for theenzyme.

According to Monod the "permease" mecha-nism of accumulation can explain the reduced,B-galactosidase activity of the whole cell ofE. coli. The "permease" would be the rate limit-ing step in the process of bringing the substrateto the ,B-galactosidase site.

Although Monod's scheme apparently agreeswith the results obtained here there are seriousdiscrepancies which must be clarified beforedrawing conclusions. The accumulation ofmethyl-f3-D-thiogalactoside is inhibited by 2,4-dinitrophenol, azide or by the absence of a sourceof energy; whereas, the hydrolysis of ONPG isnot affected at all under the same conditions. Themain evidence obtained by Rickenberg et al.(1956) to show the role of the "permease" in thehydrolysis of 3-galactosides is the correlationbetween "permease" and ,B-galactosidase in vivoobtained upon aging of the cells. Both are shownto decrease proportionally. The results obtainedhere are quite opposite inasmuch as the ,3-galac-

1958] 11


.org/D

ownloaded from

http://jb.asm.org/

ROTMAN

tosidase activity in the intact cell increases upon

aging (table 2).More recently additional evidence (Cohen and

Monod, 1957) for the role of the permease in thehydrolysis of ,B-galactosides was presented. It isbased on the correlation between the two systemsin cells with different permease activity as a resultof inhibition of the permease synthesis withp-fluorophenylalanine.

Whether the discrepancies between Monod'sresults and ours are real or due to differences instrains or conditions will have to be tested. Inview of these differences, the constitutive mutantstrain used in our experiments was subsequentlyexamined for the presence of permease and was

found to have a constitutive one inasmuch as

noninduced cells accumulate radioactive methyl-,B-D-thiogalactoside. The strain can be distin-guished from the "cryptic" constitutive becauseit is lactose positive and because of the kineticsof ONPG hydrolysis (Cohen and Monod, 1957).Therefore, the possibility that we have studieda different penetration system than did Monodseems very unlikely.Assuming that the discrepancies between

Monod's results and ours are real, they could bereconciled by postulating two different enzyme-

like systems. One would govern the entry ofspecific substrates into the cell; the second wouldcause the entered material to accumulate in highconcentration. The permeation system would bespecifically induced but would not require ap-

preciable energy to function. The substratebrought in by this system could reach theg-galactosidase site or could contact the accumu-

lation system. The accumulation system, thoughrequiring energy, need not be inducible or highlyspecific in its function to fit our scheme.A support for this hypothesis comes from recent

experiments of Monod (1956). He reported a

competition between inducer and inhibitor for a

site or substance which remains constant inamount per cell during induction. He concludesthat the site or substance must be distinct fromeither ,B-galactosidase itself or the permease

uptake system. In our scheme this site or sub-stance would correspond to the second enzyme-

like system and it would indicate that it is non-

inducible.Off hand, this hypothesis with two enzymelike

systems appears very complex. However, if theaccumulation system is represented by the cell

membrane which could control the exit of thesubstrate and therefore its accumulation, thehypothesis becomes amenable to test. If theassumption is correct, in the presence of azide ordinitrophenol the rate of exit of an accumulated,3-galactoside should increase appreciably. Therate of exit could be determined easily in theabsence of external substrate.To postulate, as Cohen and Monod (1957)

have done, that azide or dinitrophenol inhibits"only the energy coupling which allows thepermease reaction to function as a pump" servesonly to describe the facts.

Thus, the utilization of ,B-galactosides inEscherichia coli would be regulated by the induc-ible penetration mechanism, a flexible system. Itpermits the full expression of the enzyme in cellswith very little of it or in cells which have enteredthe stationary phase. Such cells might need theuse of greater synthesizing ability to start growingagain as is illustrated by the faster rate of RNA,DNA, or protein synthesis in the lag-phase cells.Furthermore, it would indicate that the normalcell synthesizes simultaneously the enzyme andthe mechanism which will control its activity.The smaller quantity of RNA present in

stationary cells (Morse and Carter, 1949) cannow be explained by the large excretion of RNAin the form of depolymerized products whichoccurs with age. The remarkable fact that cellscan lose most of their RNA and still remainviable poses new questions as to the role of RNA.For the enzyme chemist it should be of interest

that the activating effects of Na and K ionsdepend upon whether o-nitrophenyl-f-D-galac-toside or p-nitrophenyl-3-D-galactoside is thesubstrate. With the analogous a-L-arabinosides,the same phenomenon is not observed. For bothof these substrates, Na ions activate the enzymeas compared with K ions, although to a differentextent in each case.

ACKNOWLEDGMENTS

The author is most grateful to Professors H. A.Lardy and J. Lederberg for their helpful discus-sions and criticism, and to Professor B. Chancefor his valuable cooperation in some of the experi-ments.

SUMMARY

Four hypotheses were tested to explain themechanism of the reduced ,B-galactosidase

12 [voL. 76


.org/D

ownloaded from

http://jb.asm.org/


activity of whole cells of Escherichia coli. Passivediffusion, product inhibition, and inhibitors com-pounded with the enzyme were ruled out. Thedata are compatible with the presence of aninducible penetration mechanism. The penetra-tion mechanism is said to be specific for ,B-galac-tosides and to be induced together with the,B-galactosidase. Its properties differ from thoseof an inducible system which accumulates3-thiogalactosides in E. coli. The penetrationmechanism hypothesis offers a flexible regulationsystem for the enzymatic activity of the cell.The assay of whole cells with a-L-arabinosidespermits the bypass of the penetration mechanism.

REFERENCESBARRET, J. T., LARSON, A. D., AND KALLIO, R. E.

1953 The nature of the adaptive lag ofPseudomonas fluorescens toward citrate. J.Bacteriol., 65, 187-192.

BONNER, D. M. 1955 Aspects of enzyme forma-tion. In Amino acid metabolism, pp. 193-197.Edited by W. D. McElroy and H. B. Glass.Johns Hopkins Press, Baltimore.

BOREK, E., RYAN, A., AND ROCKENBACH, J. 1955Nucleic acid metabolism in relation to thelysogenic phenomenon. J. Bacteriol., 69,460-467.

CALIFANO, L. 1952 Liberation d'acide nu-cl6ique par les cellules bacteriennes sousl'action de la chaleur. Bull. World Health.Organization, 6, 19-34.

CHANCE, B. 1956 Interaction of adenosinedi-phosphate with the respiratory chain. InEnzymes: units of biological structure andfunction, pp. 447-463. Edited by 0. H. Gae-bler. Academic Press, Inc., New York.

COHEN, G. N. AND MONOD, J. 1957 Bacterialpermeases. Bacteriol. Revs., 21, 169-194.

COHEN, G. N. AND RICKENBERG, H. V. 1955Existence d'accepteurs specifiques pour lesaminoacides chez Escherichia coli. Comp.rend., 240, 2086-2088.

COHN, M. AND MONOD, J. 1951 Purification etpropi6t6s de la j3-galactosidase (lactase)d'Escherichia coli. Biochim. et Biophys.Acta, 7, 153-174.

COHN, W. E. 1950 The anion-exchange sepa-ration of ribonucleotides. J. Am. Chem.Soc., 72, 1471-1478.

DAVIs, B. D. 1949 Isolation of biochemicallydeficient mutants of bacteria by means ofpenicillin. Proc. Natl. Acad. Sci. U. S.,35, 1-3.

DAVIS, B. D. 1956 Relations between enzymesand permeability (membrane transport) in

bacteria. In Enzymes: units of biologicalstructure and function, pp. 509-522. Edited by0. H. Gaebler. Academic Press, Inc., NewYork.

DEERE, C. J. 1939 On the "activation" of thelactase of Escherichia coli-mutabile. J.Bacteriol., 37, 473-483.

GORNALL, A. G., BARDAWILL, C. J., AND DAVID,M. M. 1949 Determination of serum pro-teins by means of the biuret reaction. J.Biol. Chem., 177, 751-766.

KAPLAN, J. G. 1954 The alteration of intra-cellular enzymes. J. Gen. Physiol., 38,197-211.

KERR, S. E. AND SERAIDARIAN, K. 1945 Theseparation of the purine nucleosides from freepurines and the determination of the purinesand ribose in these fractions. J. Biol. Chem.159, 211-225.

KOPPEL, J. L., PORTER, C. J., AND CROCKER, B. F.1953 The mechanism of the synthesis ofenzymes. J. Gen. Physiol., 36, 703-722.

KUBY, S. A. AND LARDY, H. A. 1953 Purificationand kinetics of O3-D-galactosidase fromEscherichia coli, strain K-12. J. Am. Chem.Soc., 75, 890-896.

LEDERBERG, J. 1950 The fl-D-galactosidase ofEscherichia coli, strain K-12. J. Bacteriol.,60, 381-382.

LEDERBERG, J. 1951 Genetics studies with bac-teria. In Genetics in the 20th century, pp.263-289. Edited by L. C. Dunn. MacMillanPress, New York.

MANSON, L. A. 1953 The metabolism of ribo-nucleic acid in normal and bacteriophageinfected Escherichia coli. J. Bacteriol.,66, 703-711.

MONOD, J. 1956 Remarks on the mechanism ofenzyme induction. In Enzymes: units ofbiological structure and function, pp. 7-29.Edited by 0. H. Gaebler. Academic Press,Inc., New York.

MONOD, J. AND COHN, M. 1952 La biosynthe-ses induite des enzymes (adaptation en-

zymatique). Advances in Enzymol., 13,67-119.

MONOD, J., COHEN-BAzIRE, G., AND COHN, M.1951 Sur la biosynthese de la,-galactosidase(lactase) chez Escherichia coli. La specificit6de l'induction. Biochim. et Biophys. Acta,7, 585-599.

MORSE, M. L. AND CARTER, C. E. 1949 The

synthesis of nucleic acids in cultures of

Escherichia coli, strains B and B/r. J.Bacteriol., 58, 317-326.

NEWTON, B. A. 1956 The properties and mode

1958] 13


.org/D

ownloaded from

http://jb.asm.org/

14 ROTMAN

of action of the polymyxins. Bacteriol.Revs., 20, 14-27.

REPASKE, R. 1956 Lysis of gram-negativebacteria by lysozyme. Biochim. et Biophys.Acta, 22, 189-191.

RICKENBERG, H. V., COHEN, G. N., BUTTIN, G.,AND MONOD, J. 1956 La galactosideper-mease d'Escherichia coli. Ann. inst. Pasteur,91, 829-857.

RICKENBERG, H. V., YANOFSKY, C., AND BONNER,D. M. 1953 Enzymatic deadaptation. J.Bacteriol., 66, 683-687.

ROTMAN, B. 1955a A simplified device for con-tinuous growth of microorganisms. J. Bac-teriol., 70, 485-486.

ROTMAN, B. 1955b Factors controlling the

[VOL. 76

enzymatic activity of intact cells ofEscherichia coli. Bacteriol. Proc., 1955,p. 133.

ROTMAN, B. 1956 On the mechanism of soniclysis of bacteria. J. Bacteriol., 72, 827-830.

ROTMAN, B. AND SPIEGELMAN, S. 1954 On theorigin of the carbon in the induced synthesisof g-galactosidase in Escherichia coli. J.Bacteriol., 68, 419-429.

SEIDMAN, M. AND LINK, K. P. 1950 o-Nitro-phenyl-$-D-galactopyranoside and its tetra-acetate. J. Am. Chem. Soc., 72, 4324.

SEVAG, M. G., SMOLENS, J., AND LACKMAN, D. B.1940 The nucleic acid content and distri-bution in Streptococcus pyogenes. J. Biol.Chem., 134, 523-529.


.org/D

ownloaded from

http://jb.asm.org/

the ,b-d-galactosidase escherichia coli' …jb.asm.org/content/76/1/1.full.pdf · ·...

Documents