A STUDY OF LYSIS IN BACTERIOPHAGE-INFECTED ESCHERICHIA COLI

ARTHUR BROWN1 2

State University of New York, College of Medicine, Brooklyn, New York

- Received for publication August 22, 1955

Several hypotheses have been proposed to ex-plain the poorly understood mechanism by whichvirulent bacteriophage induces lysis in bacteria(Hershey and Bronfenbrenner, 1952; Luria,1953). It has been suggested that lysis is causedby: 1) the accumulation of an overwhelmingamount of phage within the cell (d'Herelle, 1926);2) the possession by the phage of a lytic enzyme(Sertic, 1929; Anderson, 1945); 3) the osmoticpressure built up inside the cell during infection(Hetler and Bronfenbrenner, 1932); 4) thechanges in surface tension during infection(Bayne-Jones and Sandholzer, 1933), 5) the acti-vation of a host-cell enzyme(s) during infection(Bronfenbrenner and Muckenfuss, 1927; Bordetand Renaux, 1928; Wollman and Woilman, 1936;Pirie, 1939). These hypotheses, derived fromstudies of different phage host systems, are notall mutually exclusive. In addition, it appearspossible that there may be more than one mecha-nism of phage-induced lysis, even in one hostcell. The object of the present study was to in-vestigate at least one possible mechanism indetail in a single host strain, and to compare thismechanism with those associated with otherphages for the same host strain.

MATERIALS AN METHODS

Organism and media. Escherichia coli strainB/r and the phages T,, T3, and Tsr+ that wereused were obtained from Dr. Vernon Bryson ofthe Cold Spring Harbor Biological Laboratory. T,and T3 form large plaques on nutrient agar, andthe infected bacteria lyse en masse within 30 min,while the r+ variants of T6 form small plaques onagar, and take about 4 hr to induce completeclearing of broth cultures. The latent periods forall 3 phages, as strictly defined, are about 21 to 30min (Adams, 1950). Infections of E. coli with

I Present address: Camp Detrick, FrederickMaryland.

' A brief report of this work was presented be-fore the annual meetings of the Society of Ameri-can Bacteriologists, 1955.

T6r+ normally have a short latent period forsome of the cells in the population, but themajority of the culture shows a 4-hr delayed masslysis, described as a lysis inhibition phenomenonby Doermann (1948, 1952).The different media employed for growth of the

bacteria and infection with phage all gave com-parable results. They included nutrient broth atpH 7.2, the same broth with 0.05 per cent glucoseand 0.5 per cent NaCl, a simple synthetic mediumwith lactate as energy source, and the samemedium with glucose as energy source (Adams,1950). When glucose media were used, the pHwas routinely adjusted to 7.2 with NaOH beforeinfection or before lysis was induced. Phagestocks were prepared in nutrient broth and oc-casionally in synthetic medium, and had aminimum titer of 1010 particles per ml.

Measurement of growth and lysis. Lysis experi-ments were carried out in duplicate or triplicatein Wassermann tubes in which lysis was measuredturbidimetrically by eye and graded semi-quanti-tatively from 0 to 3 minus, the latter indicatingcomplete clearing. Plus numbers were used forincreased turbidity over refrigerated controls thatwere assigned a zero value. In several experi-ments, a Klett-Summerson photoelectric color-imeter was used to establish a more quantitativemeasure of growth or lysis. All growth and lysisexperiments were carried out at 37 C exceptwhere noted.

RESULTS

Premature LysisThe search for lysing agents.3 A survey experi-

ment was first carried out to see which chemicalagents could induce premature lysis in T6r+ in-fected bacteria. In this way, it was hoped that (1)the lytic reaction among all the reactions in theinfected cell could be relatively isolated and in-duced at will, and (2) that a hint of the lytic

' A lysing agent is defined as any agent that caninduce premature lysis in phage-infected bacteria.

482

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

LYSIS IN BACTERIOPHAGE-INFECTED E. COLI

mechanism might be obtained by knowing theproperties or mode of action of the inducingchemical.

Nutrient broth cultures of aerated E. coli strainB/r were started by inoculation of 100 to 1,000organisms. When the turbidity reached approxi-mately 4 X 108 bacteria per ml (checked by curveof turbidity vs. plate count), half of the culturewas infected with Ter+ in a ratio of 5 phages to1 bacterium. At 8 min, 16 min, and 1 hr afterinfection, 2 ml quantities of the infected cultureswere dispensed in Wassermann tubes containing

TABLE 1The ability of various agents to induce premature

lysis in T.r+ infected Escherichia coli

Turbid-Turbid- Time ity* Time

Agent ity for Unin- forInfectedt Lysis fected Lysis

(Control)

Refrigerated con-trol............. 0 0

Incubated control . 0 -? 2+Chloroform........ 3- 3-5 1- 60Ether (peroxide

free)............. 3- 7-10 1- 60Phenol 0.25%...... 3- 7-10 1- 60Phenol 2.5%....., 0 - 0 -Ethyl acetate...... 3- 10-15 1- 60Amyl acetate...... 3- 10-15 1- 60Xylene............. 3- 20 1- 60Toluene ........... 3- 20 1- 60KCN 0.01M ....... 3- 25 1- 601KCN 0.001M......0 0Cetyl pyridinium

chloride 0.01%... 3- 3-5 1- 40Cetyl pyridinium

chloride 0.001%.. 3- 5 1- 60Formaldehyde0.025%...........0 0

Formaldehyde0.25%.............3- 45 1- 60

Formaldehyde2.5%.............0 0

* Turbidity is 0 when set at 4 X 10$ bacteria/ml;3- is complete clearing; 2+ is maximum turbidityattained.

t Agent was added at 8 min, 16 min or 1 hr afterinfection. No cultures lysed when agent addedat 8 min; the 16-min and 1-hr results are identicaland are listed as one.

t Incubated control never cleared in this pro-cedure but lysed in the normal 4 hr, if aerated.

the agents listed in tables 1 and 2. The timeintervals chosen for contact of the infected cul-tures with the agents permitted observationsduring the first and second half of the latentperiod of phage infection. In only T6r+ infectionscould observations be extended beyond the nor-mal latent period because of the lysis inhibitionphenomenon already referred to earlier. Whenpoorly soluble agents were used, they were mixedby blowing through a 1 ml pipette for about 5 to10 sec. The uninfected controls were diluted bythe same amount as were the infected culturesand distributed similarly in tubes containing thetest agents. From boththe infectedand uninfectedseries, one tube with no added agents was kept at2 to 4 C as a turbidity reference arbitrarily as-signed a value of zero. A similar tube from eachseries was incubated at 37 C with the other tubesin the test, so as to have normal growth, and in-fected cell controls. It is important to note that afew reports have appeared (Woilman and Woll-man, 1936), whose results are confirmed here, thatsome bacteriostatic agents can cause a partiallysis of uninfected bacteria.The agents having an effect as inducers of pre-

mature lysis in at least one concentration arelisted in table 1 and those having no effect can befound in table 2.None of the agents added 8 min after infection

(during the first half of the latent period) effectedany change in turbidity; when added 16 min and1 hr after infection (second half of latent period),results were equally positive for producing pre-mature lysis of phage infected E. coli. It appearsfrom table 1 that the most effective agents werechloroform, ether and some of the other fatsolvents, and the cationic surface active agent,cetyl pyridinium chloride. Cyanide was somewhatless effective. The other agents listed as inducingpremature lysis were all much slower in action.The latter group were later studied for their effecton E. coli infected with T,. No effect was ob-served. The fast-acting group was able to inducepremature lysis in T8 infections as it did in TXinfections, with the exception that the cyanideeffect was enhanced. Repetition of this experi-ment in aerated tubes produced identical resultsin both T3 and Tor+ infected cells. The failure ofcertain poisons which inhibit energy systems(fluoride, azide, dinitrophenol, etc.) to induce pre-mature lysis, and the positive effect of otherpoisons indicates that the system studied here is

48319561

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BROWVN

TABLE 2Compounds failing to induce premature lysis in Tsr+ infected Escherichia coli

Agent

"Tween 80".....................................Sodium taurocholate...........................Chlorine........................................Sodium iodoacetate.............................Sodium fluoride.................................2l4 Dinitrophenol..............................Sodium azide...................................Mercuric chloride...............................Penicillin......................................Streptomycin ...................................Chloromyetin ................................Trypsin (Armour, crystallized)..................Papain (Merck).................................Lecithinase (dialyzed and lyopholized Welch

toxin) ........................................Alcohol.........................................Acetone.........................................Glycerol.......................................Ethylene glycol.................................Mineral oil.....................................Hydrochloric acid...............................Sodium hydroxide...............................

Concentration

0.002%0.0001% 0.001%0.0001% 0.001%0.0001M 0.001M0.0001M 0.001M0.0001M 0.001M0.0001M 0.001M0.005% 0.01%0.5 u*/ml 5 u*/ml0.5 u*/ml 5 u*/ml0.5 u*/ml 5 u*/ml1 mg/ml

Supernatant from 0.5% suspension

1%1%1%1%

to pH 3.5to pH above 10

5%5%5%5%

* u = unit; 1 unit streptomycin = 1 ,ug; 1 unit chloromycetin = 1 ,ug.

different from the lysis-from-without type ofsystem described elsewhere (Cohen, 1949), andthat the effect of the poisons is complex.The premature lytic reaction induced so effec-

tively by the fat solvents was studied becausethese solvents did not, at the same time, inacti-vate any of the phages studied. In addition, theypermit a partial isolation of the lytic reaction.

Conditions affecting the lytic reaction. The effectof the phage to bacteria ratio was studied as a

factor in the reaction by varying the amount ofinoculated phage, while keeping the bacterialpopulation constant at about 4 X 108 ml. It wasfound that a minimum input ratio of 0.5 phage to1 bacterium was necessary for obtaining at leastpartial lysis when the fat solvents were added 17min after infection. This figure is expected on thebasis of the number of lysed cells necessary fordetecting a drop in turbidity by eye estimation.Phage adsorption appears to be at least one

necessary prerequisite of the premature lytic re-

action as implied from the results in the firstexperiment. First, uninfected cells only lysedslightly in the presence of the lysing agents indi-cating that phage host interaction is needed for

the induced lytic reaction to occur. Furthermore,when T6r+ phage was inoculated into broth con-

taiing a Tor+ resistant variant of E. coli in a

ratio up to 15 to 1, chloroform, ether, or phenolfail to bring about lysis beyond that of controls,even after 1 hr of incubation. On the other hand,the fact that the lysing agents failed to inducepremature lysis in the first half of the latentperiod implies that phage adsorption and otherearly events following infection are not enough toprepare the bacteria for premature lysis by thelysing agents. It was also found that heat-killedbacteria (60 C, 30 min) which are allowed toadsorb more than 5 phage particles per bacteriumdo not lyse in the presence of chloroform or ether.This, too, shows that some factor other thanphage adsorption is necessary before lysis can beinduced.

Characteristics of the Induced PrematureLysis Reaction

The effect of temperature and pH. The effect oftemperature on the lytic reaction was studied intwo ways. First, a temperature inactivation endpoint was sought. Infected cultures (5 phages per

0.02%0.01%

0.01 M0.01 M

0.01 M0.1%50 u*/ml50 u*/ml50 u*/ml

2 mg/ml10%10%10%10%

484 [VOL. 71

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

LYSIS IN BACTERIOPHAGE-INFECTED E. COLI

bacterium) were exposed to various temperaturesfor different times at 13, 16, and 18 min, and 1 hrafter infection. The lysing agent was added im-mediately after the heating, when the tempera-ture was brought back to 37 C. Lysis was meas-ured by eye estimation. It was found that heatingTor+ infected cells at 60 C for 30 min or 70 C for10 min at the designated time intervals after in-fection prevented chloroform or ether from in-ducing premature lysis.A second type of temperature study was car-

ried out as follows: Bacterial cultures were pre-pared and infected as described in the first experi-ment, except that the cultures were aerated.After 14 min and again in 1 hr, the infectedcultures were placed in a water bath at tempera-tures varying from 2 to 70 C for 4 min. When theagents used for inducing premature lysis wereadded, they were at the same temperature as thecultures except for those cultures above 50 C, inwhich instances, they were left at 37 C. Lysis wasmeasured in a photoelectric colorimeter and thetime necessary for 50 per cent lysis to occur wasrecorded. The graph shown in figure 1 in which 10is the value representing the greatest relative ac-tivity indicates that there was an increase in ac-

tivity with the rise in temperature until about37 C, after which the reaction slowed until therewas complete failure of lysis at 70 C. A Qlo valueof 2 to 3 can be calculated which, together with

pH5.5 6.0 6.5 7.0 7.5 8.0 8.5 9.0 9.5 1

10

TEMPERATUMPRATUR

9 0~~~~~~----PH

b

> 7-

6

ture> ad pH ajusted14mi af4C agJ a

2%

0 20 30 50 60 70-TEMPERATURE C

Figure 1. Effect of temperature and pH on the

lytic reaction induced prematurely by chloro-

form on Tsr+-infected Escherichia coli. Tempera-

ture and pH adjusted 14 min after infection.

Chloroform added 18 min after infection.

t Activity measured by time of 50% lysis on

turbidity of 4 X 108 bacteria/ml. 10 represents

greatest relative activity.

TABLE 3The effect of some pOi8onl8 on the premature lytic

reaction induced by chloroform, ether or phenolin T6r+ infected Escherichia colt

Chemical Turbidityt

Refrigerated control (no agent)... 0Incubated control (no agent)..... 0:Mercuric chloride, 0.05%......... 0Chlorine, 0.001%................ 0Formaldehyde, 3%.............. 0Sodium fluoride, 0.01 M.......... 3-Sodium fluoride, 0.001M.3-Sodium azide, 0.01 M............ 3-Sodium azide, 0.001 m........... 3-Dinitrophenol, 0.001 M.3-Dinitrophenol, 0.0001M.3-

* 14 min after infection, cultures were equi-librated 4 min with the poisons before lysingagents were added.

t Turbidity is 0 when suspension is set at4 X 108 bacteria/ml; 3- represents completeclearing.

t Control never lysed in this procedure, butlysed in 4 hr if aerated.

previous temperature data, is entirely consistentwith the characteristics of an enzymatic reaction.The next experiment was carried out to estab-

lish the pH limits and the pH optimum for thelytic reaction. After the culture was infected asin previous experiments, the pH of the cultureswas adjusted with sodium hydroxide 14 min afterinfection to the values shown in figure 1. Threeminutes following this adjustment, chloroformwas added. The loss in turbidity was noted andthe time necessary for 50 per cent lysis to occurwas recorded. The results of this experiment(figure 1) show that a pH of 6.4 and 10.0 preventthe chloroform-induced lytic reaction and thatthere is a sharp rise as the curve approaches pH7.5 once activity is evident. An optimum rangefor the reaction exists between pH 6.8 and 7.6.

The effect of poisons. The effect of poisons onthe lytic reaction was studied in an effort togather data for elucidating the nature and, ifpossible, the specificity of the reaction. Cultureswere infected as before. Fourteen minutes afterinfection, the cultures were dispensed into tubescontaining some selected poisons, and wereequilibrated for 4 min after shaking. Chloroform,ether or phenol were then added to induce pre-mature lysis. The results in table 3 show that

19561 485

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BROWN

formaldehyde, chlorine and mercuric chloride("general" enzymatic poisons) are effective inpreventing lysis induced by the lysing agents,while the more "specific" enzymatic poisons,dinitrophenol, sodium azide, and sodium fluoride,fail in preventing the lytic reaction. The "general"poisons could inhibit the lytic reaction even whenadded in the second half of the latent period.In addition to supporting the notion of the in-

volvement of an enzyme reaction, these data pro-vide evidence that the process of virus reproduc-tion is to some extent independent of that oflysis (Foster, 1948; Adams, 1954). This is truebecause the specific poisons tested are known toinhibit virus reproduction (Price, 1947; Cohen,1949; Ackermann and Johnson, 1953) but in thepresent experiments when added in the second butnot the first half of the latent period, they fail toinhibit lysis. Obviously, after a certain point isreached, lysis will occur even though phage re-production has ceased. Similar data have been ob-tained for the normal lytic reaction (no lysingagents used), especially in Ts infections where itcould be conveniently demonstrated because ofthe short time between infection and mass lysis.It would appear from the above that lysis whichoccurs normally and lysis induced prematurely bylysing agents probably have the same basicmechaim.Experiments th Untreated and Ultracentrifuged

Phage Lysates4The next major series of experiments were de-

signed with the following aims in mind: (1) toseparate the lysin from the infected cell system forstudy in a more isolated system; (2) to study itsrelation to the phage and to the host cell; (3) tocharacterize the lytic reaction involving the sepa-rated lysin by the same criteria as were used forcharacterizing lysis in the infected cell. The lastpoint is of particular importance for identifyingan unknown, crude, separated enzyme with thatinvolved in the infected cell. It is well known thatbacteria, including E. coli, posess intracellularenzymes such as proteinases, nucleases, etc.(Cohen, 1949). None of these, or any separatedenzymes from other sources, are known which arecapable of lysing living, or phage-infected bac-teria, except possibly, lysozyme. Trypsin, papain

4For reasons already given and soon to be notedfurther, the terms lysin and enzyme are used inter-changeably.

and lecithinase, for example, failed to lyse T1, T3or Tor+ infected E. coli in the experiments re-ported here. Since living uninfected bacteria alsoproved resistant to these enzymes and to thosepossibly present in phage lysates, heat-killed andchemically killed E. coli cultures were chosen asthe best substrates available. It is recognized,however, that several enzymes, including perhapsthe intracellular enzymes of E. coli, may lysekilled cells yet have no relation to those involvedin the lysis of phage infected cells. To establishthat the enzyme(s) present in a phage lysate-killed cell system is the same as that involved inphage-infected bacteria, the characteristics of theenzymes in both systems must be nearly identical.Such a comparative study was made.

The substrate and its specificity. The first experi-ments in this series met with failure because heat-killed E. coli failed to lyse in the presence of freshTr+ lysates. The cells had been heated in brothor phosphate buffer (pH 7.0) at 60 C for 30 minor at 70 C for 10 min. No lytic activity could bedemonstrated on such cells even after 2 hr ofincubation. To check that the method of heatingwas not responsible for this failure, trypsin orpapain were added to heated suspensions. Com-plete clearing occurred in 3 to 5 min.

E. coli killed with chloroform proved to be asatisfactory substrate for enzymatic study. Brothgrown E. coli cultures were treated with chloro-form for 30 to 45 min, subsequently washed twicein broth or phosphate buffer pH 7.2 and stored at4 C. Nutrient agar-grown cultures treated andwashed similarly served equally well. These cellstoo were lysed by trypsin as well as by the Ter+and T3 lysates. A possible explanation for thesuccess of chloroform-killed cells as a substratefor the lysin activity may be found in thediscussion.

Separation of phage and lysin. An attempt wasmade to separate the enzyme activity from thephage. A lysate was prepared by inducing pre-mature lysis with chloroform in a Tsr+ infectedpopulation of strain B/r in nutrient broth 20 miiafter infection. A lysate naturally produced 4 hrafter infection gave identical results. The chloro-form induced lysate was collected, the chloroformdrawn off by vacuum, and the phage titer deter-mined by plaque count. The titer of enzyme ac-tivity was determined as described below. Half ofthe lysate was then centrifuged in a Spinco ultra-centrifuge at 40,000 rpm (100,000 X G) for 60

486 [VOL. 71

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

LYSIS IN BACTERIOPHAGE-INFECTED E. COLI

TABLE 4The effect of ultracentrifugation* on the phage titer

and lytic activity of T.r+ lysatesPhage Titert Enzyme Titert

Source Before After Before Aftercentrifuga- centrifuga- centrif- centrif-

tion tion ugation ugation

Supernatantfrom unin-fected Esch-erichia coli... None - None

T,r+ Lysate... 4 X 1010 6 X 107 80 80

* 40,000 rpm (100,000 X G) for 60 min.t Measured in plaque-forming units per ml.$ Measured as reciprocal of dilution, producing

50% lysis in 15 min on chloroform-killed E. coliat a turbidity set at 4 X 108 bacteria/ml.

min. The phage and enzyme titers were againdetermined and compared with the lysates beforecentrifugation. The sedimented phage was re-suspended to volume and its titer determined asa control on the infectivity lost from the super-natant. The enzyme titer of the supernate of acentrifuged uninfected culture of strain B/r wasalso determined to see whether phage host inter-action was necessary to demonstrate enzyme ac-tivity. The titer of the enzyme(s) is arbitrarilydefined as the reciprocal of the dilution of thelysate capable of showing 50 per cent lysis within15 min on chloroform-killed E. coli at a beginningturbidity equivalent to 4 X 108 bacteria per ml.

Typical results presented in table 4 show thatsupernatants of uninfected strain B/r have nodemonstrable lytic activity, while the phagelysates possess such activity. Further, more than99 per cent of the infectivity was lost after ultra-centrifugation while the lysin titer remained un-changed. Although the data do not appear in thetable, it was found that the sedimented phagecontained the lost infectivity. It was thereforeconcluded that the demonstrated lysin arose fromthe host cell and could be determined by the meth-ods used here only after phage-host interaction.

Characterization of the reaction of ultracen-trifuged phage lysates. Experiments were next per-formed to test the effect of temperatures, pH, andselected poisons on the activity of phage lysateswhen ultracentrifuged as described above. Thetests were carried out on chloroform-killed E.coli, initially at a turbidity equivalent to 4 X 108viable bacteria per ml. In all other respects, the

pH

'.

,

Figure 2. Effect of temperature and pH on thelytic reaction induced by ultracentrifuged T.r+lysates on chloroform-killed Escherichia coli.Lysate and substrate adjusted to temperature andpH 4 min before observation started.

t Activity measured by time of 50O lysis ofsuspension equivalent to 4 X 108 bacteria/ml. 10represents greatest relative activity.

conditions for the tests were the same as thosedescribed for the study of enzyme activity on theinfected cultures. The data are presented graphi-cally in figure 2. A comparison of these resultswith those found in figure 1 show a markedsimilarity in the enzyme activity involved in bothsystems. In addition, the effects of the poisons onthe killed cell system were identical with thoseshown in table 3 for the infected cell system. Itappears that the activity involved in both systemsis enzymic, and that the enzymes involved areprobably the same.

Ly8is in T8 and T1 InfectiomsComparison of the lytic reaction in Tor+, T3 and

T1 infections. The reactions and data for T-in-duced lysis were almost identical with thosefound for T6r+ infections. Although the searchwas not as extensive, it was shown that the samecompounds that were capable of inducing pre-mature lysis in the T6r+ phage-host system wereeffective agents for the T8 phage-host system, andthose that failed to induce premature lysis, wereequally ineffective in both systems. For example,the same fat solvents and the cationic agent wereeffective as premature lysis inducers in bothsystems, while mercuric chloride, chlorine, sodiumazide, dinitrophenol, etc., failed to induce pre-mature lysis under comparable conditions. In ad-

1956] 487

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

TABLE 5Difference. in lysi8 of Escherichia coli induced by three phages

Phage Ti To Tr+

Effect of chloroform, ether, orphenol* t........................ Inhibited Premature lysis Premature lysis

Effect of sodium citrate, 0.7%t....... Inhibited No effect No effectEffect of ultracentrifuged phase lysateon chloroform-killed E. coli ........ No lysis Lysis: enzymic re- Lysis: enzymic re-

action actionMicroscopic appearance before lysis Swelling of cells Swelling of cells No swelling

frequent infrequent

* Chloroform did not inactivate free T1, T3 or Tsr+.t Effects noted are for agents added in second half of latent period.

dition to these similarities, the pH limits andoptima, the relation of the rate of the lytic re-

action to temperature, and the effect of poisonson the lytic reaction itself are virtually identicalin both T3 and Ter+-induced lysis. The data on

the ultracentrifuged phage lysate-killed cell

system were also similar.A study with Ti-infected populations, however,

immediately revealed striking differences. Chloro-form, ether, or other fat solvents, which inducedpremature lysis in T6r+ and T3 infections, in-hibited completely the lytic reaction in T1 in-fections, even when added only 3 min prior tomass lysis. Since chloroform fails to inactivateeither Ti, T3 or Tr+, the differences are not dueto an effect on the phage. Adams (1949) reportedthat sodium citrate inhibits lysis in T1 infectionswhen added at or before infection. In the experi-ment reported here, sodium citrate was found todelay T1-induced lysis when added at varioustimes after adsorption. It was least inhibitory inthe second half of the latent period (Andrewes andElford, 1932). Sodium citrate was then studied foran effect in T3 and Ter+-infected strain B/r. In0.7 per cent concentration, it fails to affect thenormal or chloroform-induced lytic reaction inthe latter infections, no matter when it is added.

Experiments were carried out to determine theactivity of fresh T, lysates of high titer, bothultracentrifuged and untreated, on heat andchloroform-killed B/r as was done for the otherphages. No significant lytic activity was observedfor 6 separately prepared lysates having titersranging from 3 to 8 X 1010 plaque units per ml,even when the killed cells were suspended di-rectly in the lysates. Thus, there are striking bio-chemical differences between T1-induced lysis on

the one hand, and Ts and Tsr+-induced lysis onthe other.

Microscopic observations. The process of lysiswas observed for the 3 phage infections in hangingdrop preparations, or, less frequently, on agar(Adams, 1950). Swelling of E. coli under theinfluence of certain phages (including T1) has beenobserved by others (Hetler and Bronfenbrenner,1932; Bayne-Jones and Sandholzer, 1933; Del-bruck, 1940). The significance of these observa-tions are in some doubt in view of Delbruck'sobservation (1940) that the presence or absenceof swelling in some phage infections depends onthe number of adsorbed phages. In the experi-ments reported here for T1, T3 and T6r+ infec-tions, where an input ratio of 5 phage to 1 bac-terium was employed, swelling before lysis wasseen often in T1 infections, only occasionally inT3 infections, and never in Tsr+ infections.

All the comparative characteristics mentionedabove for the 3 phage-induced lytic reactions aresummarized in table 5. They indicate that themechanism of lysis in T1 infections is a differentone from that involved in both T, and T,,r+infections. Further work on the T, associatedlysis was postponed until an elucidation of theTor+ lytic mechanism was obtained.

DISCUSSION

One of the main conclusions to be drawn fromthe experiments reported here is that the mecha-nism of lysis in 1 host-cell population probablydepends on the particular infecting or sensitizingagent. This is aptly proven by the data showingmarked differences between T, versus Ts andT6r+ infected E. coli. It implies, therefore, thatthere may be several lytic mechanisms in bacterio-

488 BROVVN [VOL. 71

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

LYSIS IN BACTERIOPHAGE-INFECTED E. COLI

phage-infected bacteria as a group, or, indeed, inany group of systems which involves lysis of thebacterial cell. The detailed studies of the lytic re-actions involved in Ter+ and Ts infections ofhost strain B/r, lead to the conclusions that inthese systems the reaction is enzymic in nature,and that the enzymes probably originated in thehost cell.The following picture represents one possible

mechanism for phage induced lysis in T3 andTor+ infections based on the work of others andthat described here. The reproducing phage par-ticles are conceived of as agents which activateautolysis of the host cell in at least a 2-stepprocess. The process involves a "sensitization"which is followed by a more direct activation ofthe host cell autolysin(s) when the phage par-ticles are at or near completion. It is at this pointthat the premature lysing agents are more efficientthan the phage, and thus hurry the process along.When the lysing agents are added to infectedcultures in the first half of the latent period, theydo not induce premature lysis because phagegrowth is prevented from reaching that point atwhich the final steps of autolysin activation canproceed. It is possible that at least one step inthe activation involves the removal or inactiva-tion of an autolysin inhibitor, as has been de-scribed (Kozloff, 1953) for the activation ofdeoxyribonuclease in phage infected E. coli, butfurther work is necessary before this guess can besubstantiated.

Lysis of E. coli that is more or less associatedwith phage has been described for several diversesystems. These include systems that may be re-garded as the lysis-from-without type (Delbruck,1940), where lysis occurs without increase of in-fectious phage. In this category would fall: (1)the lysis induced by nucleic acid-free phage"ghosts" (Herriott, 1951); (2) the lysis induced byphage when the bacteria, before infection, havebeen deprived of their source of energy by starva-tion (Monod and Wollman, 1947) or by anaero-biosis or energy poisons (Cohen, 1949); (3) thelysis induced in uninfected bacteria by poly-electrolytes, the latter presumably acting likephage in the initial stages of phage infection(Puck, 1953). The relation of this type of lysis towhat occurs in "normal" bacteriophage infectionsis unknown (Adams, 1954).

Admittedly, lysis occurring in the lysis-from-without type systems is somewhat removed from

that occurring in "normal" phage infections. Acloser resemblance to "normal" phage inducedlysis may perhaps be found in the systems where(1) proflavine permits lysis while inhibiting theproduction of mature virus (and not of incompletevirus) as described by Foster (1948), and De-Mars et al. (1953), (2) lysis is induced prematurelyin the second half of the latent period, as describedby Doermann (1952), and the experiments in thepresent paper.The 2-step hypothesis suggested in this discus-

sion for explaining "normal" phage-induced lysisis first based on the assumption that lysis in-duced prematurely and "normal" lysis involve thesame mechanism. Considerable evidence for thishas been presented. Secondly, it is emphasizedthat the hypothesis applies to Ter+ and Tsphage-induced lysis and not to T1-induced lysisbecause of the evidence showing they have differ-ent lytic mechanisms.

In order to gather additional evidence for the2-step hypothesis, exploratory experiments werecarried out to test whether any chemicals couldsubstitute for the phage in the first or "sensi-tizing" step. Of many tried, it was found that 0.7per cent citrate (not growth-inhibitory), andcyanide, both sensitized uninfected E. coli strainB/r to partial lysis (20 per cent turbidity re-mained) when subsequently treated with chloro-form, ether, or 0.25 per cent phenol. The super-natant from these cultures, in several differentexperiments, contained a maximum enzyme titerof 10 on chloroform-killed E. coli. These resultsare suggestive but still equivocal. If the systemcould be improved to give enzymes comparable intiter, in degree of lysis, and reproducibility tothose found in the phage infected bacteria, thiswould be additional evidence in support of the2-step hypothesis.

SUIMMARY

The lysis of Escherichia coli strain B/r bybacteriophages T1, Ts and T6r+ was studiedincluding lysis of infected cells induced prema-turely by lysing agents. Biochemical and somemicroscopic evidence is presented which indicatesthat the mechanism of lysis in a single host strainpopulation depends on the specific infecting agent.The data, for example, show that the mechanismsassociated with Ts and Ter +-induced lysis appearsimilar, but both are different from T1-inducedlysis.

19561 489

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from

BROWN

A close study was made of the lytic reactions in-duced by Ts and Tor +. The effect of temperature,pH, and poisons on the reaction in the infectedcell population was noted. Chloroform-killed, butnot heat-killed, E. coli provided a suitable sub-strate with which to study lysis by ultracen-trifuged phage lysates with respect to the samecharacteristics. The data in both the infected celland killed cell systems suggest a mechanisminvolving the activation of host cell (autolytic)enzyme(s) by the T3 and Tor+ phages.

REFERENCESACxEBMANN, W. W. AND JOHNSON, R. B. 1953

Some energy relations in a host virus system.J. Exptl. Med., 97, 315-322.

ADAms, M. H. 1949 The calcium requirement ofcoliphage Ts. J. Immunol., 62, 505-516.

ADAms, M. H. 1950 Methods of study of bac-terial viruses. Methods in Med. Research,2, 1-73.

ADAMS, M. H. 1954 Abortive infection withviruses. The dynamics of virus and rickettsialinfections. Ed. by F. W. Hartman, F. C.Horsfall, and J. G. Kidd. The Blakiston Co.New York, N. Y.

ANDERSON, T. F. 1945 On a bacteriolytic sub-stance associated with a purified bacterialvirus. J. Cellular Comp. Physiol., 25, 1-15.

ANDREWES, C. H. AND ELFORD, W. J. 1932 The"killing" of bacteria by bacteriophage.Brit. J. Exptl. Pathol., 13, 13-21.

BAYNE-JONES, S. AND SANDHOLZER, L. A. 1933Changes in the shape and size of Bacteriumcoli and Bacillu megatherium under the in-fluence of bacteriophage-a motion photo-micrographic analysis of the mechanism oflysis. J. Exptl. Med., 57, 279-303.

BORDET, J. AND RENAUX, E. 1928 L'Autolysemicrobienne transmissable ou le bacteri-ophage. Ann. inst. Pasteur, 42, 1283-1335.

BRONFENBRENNER, J. AND MUCKENFUSs, B. S.1927 Studies on the bacteriophage ofd'Herelle. VIII. The mechanism of lysis ofdead bacteria in the presence of bacteriophage.J. Exptl. Med., 45, 887-909.

COHEN, S. S. 1949 Growth requirements of bac-terial viruses. Bacteriol. Revs., 13, 1-24.

DELBRUCK, M. 1940 The growth of bacteri-ophage and lysis of the host. J. Gen. Physiol.,23, 643-660.

DzMARs, R. I., LURA, S. E., FISEER, H., ANDLEVINTHAL, C. 1953 The production of in-complete bacteriophage particles by the ac-tion of proflavine and the properties of the in-

complete particles. Ann. inst. Pasteur, 84,113-128.

DOEzRANN, A. H. 1948 Lysis and lysis in-hibition with Escherichia coli bacteriophage.J. Bacteriol., 55, 257-276.

DOERMANN, A. H. 1952 The intracellulargrowth of bacteriophage. I. Liberation ofintracellular bacteriophage T4 by prematurelysis with another phage or with cyanide.J. Gen. Physiol., 35, 645-6.

FoSTER, R. A. C. 1948 An analysis of the ac-tion of proflavine on bacteriophage growth.J. Bacteriol., 56, 795-9.

D'HERELLE, F. 1926 The bacteriophage and itsbehaviour. The Williams & Wilkins Co.,Baltimore, Md.

HiIRRioTr, R. M. 1951 Nucleic acid-free T2virus "ghosts" with specific biological ac-tion. J. Bacteriol., 61, 752-754.

HERSHEY, A. D. AND BRONFENBRENNER, J.1952 Bacterial viruses: Bacteriophages.Viral and Rickettsial Infections of Man.Ed. by T. M. Rivers. 2nd Edition. J. P.Lippincott Co., Philadelphia, Pa.

HETLER, D. M. AND BRONFENBRENNER, J. 1932Further study on the mechanism of transmis-sable lysis of bacteria. Proc. Soc. Exptl.Biol. Med., 29, 806-808.

KOZLOFF, L. M. 1953 Desoxyribonuclease in-hibitor and bacteriophage infection. Fed-eration Proc., 12, 234.

LURA, S. E. 1953 General Virology. John Wileyand Sons, Inc., New York, N. Y.

MONOD, J. AND WOLLMAN, E. 1947 L'inhibi-tion de la croissance et de l'adaptation en-zymatique chez les bacteries infectees par lebacteriophage. Ann. inst. Pasteur, 73, 937-956.

PIRIE, A. 1939 A chemical change occurringduring the lysis of Bact. coli by bacteriophage.Brit. J. Exptl. Pathol., 20, 99-108.

PRCE, W. H. 1947 Bacteriophage formationwithout bacterial growth. III. The effect ofiodoacetate, fluoride, gramicidin and azideon the formation of bacteriophage. J. Gen.Physiol., 31, 135-140.

PucK, T. T. 1953 The first steps of virus inva-sion. Cold Spring Harbor Symposia Quant.Biol., 18, 149-154.

SERTIC, V. 1929 Origine de la lysine d'une racedu bact6riophage. Compt. rend. soc. biol.,100, 477-479.

WOLLMAN, E. AND WOLLMAN, E. 1936 Recher-ches sur le phenomene de Twort-d'HerelleBacteriophagie ou autolyse h6r6do-conta-gieuse. Ann. inst. Pasteur, 56, 137-164.

490 [VOL. 71

on May 26, 2020 by guest

http://jb.asm.org/

Dow

nloaded from