Commitment to Lysogeny Is Preceded by a Prolonged Period of Sensitivity to the Late Lytic Regulator Q in Bacteriophage λ

Abstract

A key event in development is the irreversible commitment to a particular cell fate, which may be concurrent with or delayed with respect to the initial cell fate decision. In this work, we use the paradigmatic bacteriophage λ lysis-lysogeny decision circuit to study the timing of commitment. The lysis-lysogeny decision is made based on the expression trajectory of CII. The chosen developmental strategy is manifested by repression of the pR and pL promoters by CI (lysogeny) or by antitermination of late gene expression by Q (lysis). We found that expression of Q in trans from a plasmid at the time of infection resulted in a uniform lytic decision. Furthermore, expression of Q up to 50 min after infection results in lysis of the majority of cells which initially chose lysogenic development. In contrast, expression of Q in cells containing a single chromosomal prophage had no effect on cell growth, indicating commitment to lysogeny. Notably, if the prophage was present in 10 plasmid-borne copies, Q expression resulted in lytic development, suggesting that the cellular phage chromosome number is the critical determinant of the timing of lysogenic commitment. Based on our results, we conclude that (i) the lysogenic decision made by the CI-Cro switch soon after infection can be overruled by ectopic Q expression at least for a time equivalent to one phage life cycle, (ii) the presence of multiple λ chromosomes is a prerequisite for a successful Q-mediated switch from lysogenic to lytic development, and (iii) phage chromosomes within the same cell can reach different decisions.

INTRODUCTION

A key feature of developmental genetic networks is the ability to make a choice between alternative developmental pathways and maintain the decision. Depending on the network, the path from an initial decision to cell fate commitment can be gradual and consist of a series of reversible steps (1, 2), or, at the other extreme, decision and commitment can be obtained by a single all-or-none step (3, 4). The nature of the path to commitment determines important properties such as the adaptability of the system to changes in environmental conditions, as well as the reliability and speed of progression to the new cell fate (reviewed in reference 5).

The lysis-lysogeny decision of bacteriophage λ has long been a paradigm for developmental genetic networks. After infecting their host bacteria, there are two developmental pathways available to temperate bacteriophages such as λ: lytic or lysogenic development (6–8). During lytic development, progeny phage are produced and the host cell is killed. In the lysogenic pathway, a prophage state is established and the phage genome is replicated passively along with that of the host. The key regulatory elements involved in the regulation of these developmental pathways are depicted in Fig. 1. The lysis-lysogeny decision-making process is well understood. Integration of information from several cellular signals, as well as the number of infecting phage particles, results in a specific trajectory of CII protein concentration. The timing of CII accumulation and the abundance of the protein determine the decision by controlling a bistable genetic switch, which is based on mutual repression of the lysogenic regulator CI and the lytic regulator Cro. CII is a transcriptional activator protein which has three major functions. It facilitates production of (i) CI, which is responsible for the establishment and maintenance of lysogeny, (ii) Int, which catalyzes the integration of the λ DNA into the bacterial chromosome, and (iii) the RNA anti-Q, which antagonizes the function of the Q antiterminator protein (9) (Fig. 1).

FIG 1.

Arrangement of the key regulatory elements on the bacteriophage λ chromosome. The pR transcript encodes the transcription-regulatory proteins Cro and CII, the Q antiterminator protein, and the replication proteins O and P. Transcription of the late genes is initiated at the pR′ promoter and require the presence of the Q protein to proceed through the tR′ terminator. The late genes encode proteins required for phage capsid production, DNA packaging, and host cell lysis.

The late lytic genes required for head and tail synthesis, phage assembly, DNA packaging, and lysis are transcribed from the pR′ promoter and require Q for their expression (10, 11). Q and CII are transcribed from the same polycistronic mRNA, but Q activity is significantly delayed compared to CII activity (12, 13). This delay has been suggested to ensure sufficient time for completion of the decision-making process before Q activates expression of the late lytic genes (12). It is, however, not known whether sufficient CII accumulation represents a point of irreversible commitment to lysogeny or whether the decision remains reversible during the progression to the stably inherited prophage state. Here, we addressed this question by investigating how untimely expression of the Q protein influences lysogenic development.

MATERIALS AND METHODS

Strain and plasmids construction.

The oligonucleotides used in the construction of strains and plasmids are shown in Table 1. In the construction of plasmid pBADQ, the Q-coding region from bacteriophage λ was amplified by PCR using primers QUP and QDN. The resulting DNA fragment was digested with NheI and PstI and inserted into plasmid pBAD24 (14) using the same sites. The sequence of the cloned region was verified. Plasmid pSEM3058 was constructed by inserting the attB site, fused to a DNA sequence encoding the P5 variant of the fast-folding GFPmut2 protein (15), into plasmid pLG338 (16). Escherichia coli MG1655 cells carrying plasmid pSEM3058 were infected by wild-type λ (λ^WT). Integration of the λ chromosome into pSEM3058 but not into the chromosome was verified by PCR in the selected clones. To create plasmid pSEM3141B, we first ligated the EcoRV-PstI fragment of pEM7/Zeo (Invitrogen), which contained the zeocin resistance cassette, with the PstI-PvuII fragment of pMOD-3〈R6Kγori/MCS〉, which contained the R6Kγ replication origin. The plasmid was maintained in Pir1 cells (Invitrogen), which supplied the replication protein π. To add the attB site, the resulting plasmid was PCR amplified using the ATTBR6 and ZEORI primers, and the PCR product was digested with EcoRI. The DNA fragment obtained was circularized by T4 DNA ligase to generate pSEM3141B. Pir1 cells which carried the pSEM3141B plasmid were infected by λ^WT. Integration of a single λ chromosome into pSEM3141B and the absence of the prophage at the chromosomal attB site in the selected clones (Pir1/pSEM3141Bλ) were verified by restriction analysis of pSEM3141Bλ and by PCR, respectively.

TABLE 1.

Oligonucleotides used in the construction of strains and plasmids

Name	Sequence (5′→3′)
ATTB1GFP	TTTTGAATTCCCTGCTTTTTTATACTAACTTGACTGCAGCTTTAAGAAGGAGATATACATATGGC
GFPDNBGLII	TTTTTAGATCTTTATTTGTAGAGCTCATCCATGCCAT
ATTBR6	TTTTGAATTCCCTGCTTTTTTATACTAACTTGACTGCAGGCATGCAAGCTTTAAAAGCCT
ZEORI	GGAAGGAATTCTCAGTCCTGCTCCTCGGCCAC
QDN	TTTTCTGCAGATCATGCCGTTAATATGTTGCCATCCGT
QUP	TTTTGCTAGCAAGCTCTTGC CCATAAAGCA GATGAACTTC
CIIUP	TTTTTGAATTCTTCTTTAGATCTAGGAGGCCCTTATGGTTCGTGCAAACAAACGCAACGAGGC
CIIDN	TTTTTGGATCCTTCTTTACCGGTGAACTCCATCTGGATTTGTTCAGAACGC
CYSG RT F	TCGTCGCATCTTCTGTAACG
CYSG RT R	GAGGAGACCGCTACCATGAG

Q-induced assembly of phage particles in cells carrying multiple prophages.

To test whether Q expression would allow production of functional phage particles, lysogenic cells carrying the pBADQ plasmid were grown in LB medium containing appropriate antibiotics and 10 mM MgSO₄ to an optical density at 600 nm (OD₆₀₀) of 0.3 at 30°C. The cultures were split, and Q expression was induced in one set of cultures by the addition of 0.4% arabinose. After overnight incubation, cells and cell debris were removed by centrifugation. The cleared supernatant was treated with chloroform, and its phage content was determined by determining the titers of the lysates on maltose-induced E. coli MG1655 cells using the soft-agar overlay method and counting PFU.

Determination of the number of λ prophage copies.

The number of λ prophage copies in SY822(λ) (17) and MG1655(pSEM3058λ) cells was determined by real-time quantitative PCR (qPCR) using two primer sets targeting λ and E. coli DNA, respectively, as described for determination of bacterial plasmid copy number (9). Briefly, total DNA from bacterial cells was purified with the PureLink genomic DNA minikit (Invitrogen), and the DNA concentration was measured on a NanoDrop spectrophotometer (Thermo Scientific) and adjusted to 19 ng/μl. The CFX96 instrument (Bio-Rad) was used for qPCR amplification and detection. The qPCR mixture was prepared in triplicates of 25 μl reaction mixture in Hard-Shell 96-well skirted PCR plates (Bio-Rad) and sealed with optical adhesive covers (Applied Biosystems). Each reaction well contained 5 μl of template DNA, 12.5 μl of 2× SsoAdvanced SYBR green Supermix (Bio-Rad), and 0.2 μM (each) forward and reverse primers. qPCR primer sets amplifying the λ cII gene and the E. coli cysG (10) gene are listed in Table 1.

Timed induction of Q expression.

MG1655 cells containing pBADQ were grown overnight in MOPS (morpholinepropanesulfonic acid)-PO₃ medium (1× MOPS salts without phosphate [18], 0.2% maltose, 0.5% glycerol, 100 μg/ml ampicillin, 1 μg/ml thiamine, 10 mM MgSO₄, 5 mM NaH₂PO₃), pelleted, and resuspended in 10 mM MgSO₄. Cells and λ^WT phage were mixed 1:6 in 13-ml conical tubes and incubated for 30 min at 20°C to allow adsorption. Infection was initiated by the addition of preheated MOPS-PO₃ medium and transfer of the tube to a shaking 37°C water bath. After 5 min, a 100-fold surplus of heat-killed, maltose-induced MG1655 cells and sodium citrate to a final concentration of 50 mM were added to inactivate any free phage and eliminate further infection of the live cells. At the indicated times, 0.5-ml aliquots were transferred to a new preheated tube containing arabinose at a final concentration of 0.4%. After 60 min, all samples were diluted in 1× M63 salts containing 50 mM Na-citrate chelator to further prevent infection of live cells by newly released progeny phage, and samples were plated on TB plates containing 50 mM Na-citrate, 100 μg/ml ampicillin, and 0.4% arabinose. Aliquots that had not been exposed to arabinose during the course of the experiment were plated on similar plates without arabinose. Approximately 1% of the cells survived infection in the sample that received arabinose at the earliest time point, 8 min postinfection. To determine whether these survivors represented lysogens or uninfected cells, a total of 131 of the colonies from three independent experiments were tested for their immunity to λ^cIb221 by cross-streaking; 56% (±12%) were not immune, indicating that they had escaped infection (or had survived an abortive infection). As we expect the number of uninfected cells per sample to be independent of the time of arabinose addition, the number of uninfected cells found in the samples that received arabinose at the 8-min time point was subtracted from the number of survivor cells at all the time points. The data shown in Fig. 4 thus represent the number of immune colonies from a sample receiving arabinose at the indicated time postinfection divided by the number of immune colonies in the original sample which did not receive arabinose.

FIG 4.

Effect of the timing of Q expression on the lysis-lysogeny decision. E. coli MG1655 cells carrying pBADQ were infected by wild-type λ at an API of 6. Arabinose (0.4%) was added at the indicated time points, and after 60 min postinfection, the mixture was plated to allow colony formation by the surviving cells. Bars indicate the fraction of cells which maintained the lysogenic decision upon Q expression (number of λ-immune colonies from samples where arabinose was added at the indicated time postinfection divided by the number of λ-immune colonies from samples without arabinose addition). Error bars represent one standard deviation from the mean based on data from three independent cultures.

RESULTS

Premature Q expression promotes lytic development.

A previous study showed that constitutive expression of the late genes from a promoter located between genes P and Q in the λ qin101 mutant gives clear plaques, indicating failure to lysogenize (19). However, the λ qin101 Q mutant gave small plaques but lysogenized normally (19). These observations suggested that expression of Q makes the decision independent of the CII expression trajectory. To test this hypothesis we constructed a plasmid which allowed arabinose-controlled expression of the Q protein (pBADQ). In the absence of arabinose, infection of pBADQ-containing E. coli MG1655 cells by wild-type λ (λ^WT) phage resulted in turbid plaques. However, in the presence of arabinose, the plaques obtained were uniformly clear (Fig. 2). Thus, premature expression of Q indeed appears to overrule CII-mediated decision-making. We note that the ability of the arabinose-induced cells to form plaques proves that functional phage progeny is produced from cells that contain Q at the time of infection.

FIG 2.

Growth of wild-type λ on host cells expressing the Q protein. Top agar containing 0.4% l-arabinose was mixed with E. coli MG1655 cells carrying the pBADQ plasmids preadsorbed with wild-type λ phage and poured on the surfaces of TB plates containing 100 μg/ml ampicillin and 0.4% arabinose. Clear plaques were formed (right panel), indicating that the lytic pathway was chosen. In the absence of arabinose (left panel) turbid plaques were formed.

A stably inherited prophage is insensitive to Q expression.

Interestingly, expression of the late genes in a single-copy λ lysogen had no visible effect on colony growth. To confirm that cell growth is indeed unaffected by Q expression, we compared the growth of λ lysogens to that of the parental MG1655 cells in liquid medium, in both the presence and absence of Q expression (Fig. 3A, squares). Induction of Q expression in E. coli MG1655 by 0.4% arabinose had no visible effect on growth, even if the strain contained a single-copy λ prophage.

FIG 3.

Effect of Q expression on cell growth and phage production. (A) Cells were grown overnight and diluted in fresh LB medium (containing 50 μg/ml ampicillin and 30 μg/ml kanamycin) at the zero time point. At 65 min, l-arabinose was added at 0.4% concentration (solid symbols). Control cultures were grown without l-arabinose (empty symbols). The optical densities of the cultures were recorded at different times. Triangles and dashed lines represent E. coli MG1655 cells carrying plasmids pBADQ and pSEM3058. Squares represent cells carrying a λ prophage (integrated into the chromosomal attachment site) and plasmids pBADQ and pLG338. Circles represent cells carrying plasmids pBADQ and pSEM3058λ. pSEM3058λ carries a single λ prophage and is present at about 10 copies per cell. (B) Cells were grown and induced as for panel A but using a range of arabinose concentrations. After 23 h of growth, optical densities of the cultures were recorded (circles, right axis), and the phage content of the lysates was tested by plaque assays (squares, left axis).

Induction of Q expression postinfection can alter the lysis-lysogeny decision.

As shown in Fig. 2, the Q protein was able to direct the decision toward lytic development when present in the cells at the time of infection. However, in cells carrying a stably inherited prophage, expression of Q had no effect (Fig. 3). Thus, at some point between the time of infection and the stably inherited prophage state, the infected cell commits to lysogeny and becomes insensitive to expression of Q and the late lytic genes.

To get insight into the timing of the decision and commitment process, we tested how long Q expression can affect the lysis-lysogeny decision after infection. Cells carrying the pBADQ plasmid were mixed with λ^WT at an average phage input (API) of 6 (to favor lysogeny). To synchronize the cells with regard to infection time, the mixture was incubated at 20°C to allow adsorption of the phage particles to the host cells (see Materials and Methods) before it was moved to 37°C to allow injection of phage DNA into cells. Furthermore, any remaining free phage particles were inactivated after 5 min at 37°C by the addition of a 100-fold surplus of heat-killed, maltose-induced MG1655 E. coli cells in sodium citrate to prevent further infection of the live cells. At different times postinfection, an aliquot of the synchronized culture was transferred to a culture tube containing arabinose to induce Q expression. The fractions of cells that chose the lysogenic pathway in the arabinose-induced subcultures as well as the original, uninduced culture were recorded by plating for colonies at 60 min postinfection (see the schematic of the experiment in Fig. 4). Subsequently, surviving cells which had become lysogenic for λ were distinguished from those which had escaped infection entirely by testing the colonies for immunity to λ^cI−.

Figure 4 shows the fraction of lysogens in the arabinose-induced subcultures relative to the number of lysogens in the original culture, which did not contain arabinose. It is evident that expression of Q in trans shortly after infection resulted in a reduction of the lysogenic cell fraction by more than 90%, in agreement with the plaque phenotype shown in Fig. 2. Interestingly, we found that most cells remained sensitive to Q expression long after the initial lysis-lysogeny decision had occurred. Specifically, induction of Q expression resulted in cell death even at 50 min postinfection (corresponding to a full lytic life cycle for λ) in more than half of the cells which otherwise would have continued the path to lysogeny (Fig. 4).

Q expression in cells carrying 10 stably inherited prophages results in lysis.

Expression of Q in trans resulted in very different outcomes depending on whether it occurred at 50 min postinfection (the last time point in Fig. 4) or many generations postinfection (single-copy lysogen [Fig. 3A]). One important difference between lysogenic cells in these two situations is that a freshly infected cell carries multiple phage λ genomes per host chromosome, because the phage genome is initially replicated every 2 to 3 min following infection, regardless of whether the outcome of infection will by lysis or lysogeny (20). In contrast, the established lysogen carries only one λ genome per host chromosome. To test whether the number of λ genomes accounts for the insensitivity of the established single-copy lysogen to Q induction, we constructed a strain which contains about 10 plasmid-borne copies of the λ prophage [MG1655(pSEM3058λ)]. The 10:1 phage gene dosage relative to that of host genes was confirmed by quantitative real-time PCR (see Materials and Methods).

The result shown in Fig. 3 (circles) demonstrates that, unlike with the single-copy prophage, the growth of the strain carrying 10 copies of the λ prophage was completely inhibited after induction of Q expression, and later cell lysis was observed. Induction of the pBAD promoter by arabinose is all or none; that is, subsaturating concentrations of arabinose result in a mixed population of cells with fully induced Q transcription and cells with uninduced levels of Q (21). As expected, growth inhibition depended on the concentration of arabinose in the medium and occurred in the dynamic range of the pBAD system (14) (Fig. 3B, circles). Thus, the number of phage chromosomes is a critical determinant of the effect of Q expression.

Q induction results in the production of infectious phage particles.

The above observations raise the important question of whether late gene induction by Q results in the production of infectious phage particles, that is, true lytic development, or simply cause host cell killing without phage production. Testing the cell-free lysates by plaque assay showed that the number of infectious λ particles in the lysates increases with the concentration of arabinose used (Fig. 3B, squares), and a large number of infectious phage particles are produced at saturating arabinose concentrations in the multicopy λ lysogen (∼2 × 10⁹ PFU/ml), while the uninduced control contained only about 6 × 10⁵ PFU/ml. Thus, expression of Q in the multicopy lysogen induces true lytic development resulting in phage particles capable of plaque formation.

Q-induced production of phage particles in the presence of different topological forms of prophage DNA.

Our results (Fig. 3 and 4) suggest that Q can induce the lytic cycle in lysogens that contain more than one prophage. One possible explanation of the results in Fig. 4 is that multiple lysogens containing tandemly integrated λ chromosomes at the chromosomal attB site remain sensitive to Q induction indefinitely. To test this possibility, we compared the effect of Q production in single and triple lysogens carrying the pBADQ plasmid (Fig. 5). This comparison was done using recA mutant hosts to prevent the loss of prophage copies by the host recombination machinery and to prevent induction by the SOS response. As expected, a low number of phages were observed in both the arabinose-induced and uninduced cultures of the single lysogen. About half of the phages formed clear plaques, indicating that induction was primarily the result of mutation (17). Similar results were obtained for the uninduced culture of the triple lysogen. In the presence of arabinose, phage production from the triple lysogen increased substantially, and the produced phages uniformly formed turbid plaques. However, Q induction in the strain containing the triple lysogen produced several orders of magnitude fewer phage than induction in the strain carrying the 10-copy plasmid-borne prophage (Fig. 5). We note that the strain carrying the plasmid-borne prophage, in contrast to the single and triple lysogens, expresses the lambda receptor (lamB⁺). Thus, phage production from the plasmid-borne lysogen is presumably somewhat higher, as a fraction of the released phage will be lost due to superinfection.

FIG 5.

Q-induced production of phage particles in the presence of different topological forms of prophage DNA. E. coli SY822 (recA lamB) carrying a single λ^rex:gfp chromosome and cells carrying three tandemly integrated λ^rex:gfp chromosomes at the chromosomal attB site (17) were transformed with the pBADQ plasmid and grown in the presence (black bars) or absence (gray bars) of 0.4% arabinose. The phage content of overnight cultures was measured by plaque assays. The phage content of induced and uninduced cultures of MG1655 recA lamB⁺ cells carrying pBADQ and pSEM3058λ is shown for comparison. The average phage content of three independent cultures is shown. Error bars indicate the standard deviation from the mean.

Unlike the case of the plasmid-borne prophage, Q induction did not result in growth inhibition of the single or the triple lysogen, indicating that lysis occurs in only a small fraction of the cells.

Spontaneous excision of a plasmid-borne prophage.

Q expression in the multicopy λ lysogen resulted in production of phage particles. Plasmid pSEM3058λ exceeds the DNA size that is packable into λ phage heads, and the kanamycin marker of the plasmid was not transduced by the produced phage progeny (data not shown). Thus, it appears that the λ chromosome must have been excised from the plasmid to allow the production of plaque-forming λ phage progeny. In accordance with these observations, PCR analysis of the attachment sites in uninduced cells carrying pSEM3058λ and pBADQ indicated the presence of both attP and the plasmid-borne attB sites, besides the presence of the attL and attR sites of pSEM3058λ (data not shown), suggesting the presence of both the empty plasmid and the λ-bearing plasmid in the cell population.

To understand whether these results were specific to the plasmid we used, we integrated the λ chromosome into another attB-containing plasmid. This plasmid (pSEM3141Bλ) did not exceed the size restriction for DNA packaging. Integration of a single λ chromosome into the plasmid was confirmed by gel electrophoresis of restriction fragments (Fig. 6, lane 1). This analysis indicated the presence of the empty plasmid along with the plasmid carrying the λ chromosome even after multiple passages of the lysogen and without induction of Q. Bands specific to attP or to the linear form of the λ chromosome packed into the phage heads (cos to cos) were not observed. These results suggest that the prophage is excised spontaneously at a relatively low rate but the excised prophages remain inactive (presumably repressed by CI) in the absence of Q expression.

FIG 6.

Restriction analysis of plasmid pSEM3141Bλ. Plasmid DNA obtained from Pir1 cells carrying pSEM3141Bλ and pBADQ (lanes 1 and 4) and from zeocin-resistant transductants (lanes 2, 3, 5, and 6) was digested with EcoRI (lanes 1 to 3) or with NcoI (lanes 4 to 6) and separated by electrophoresis using an 0.8% SeaKem GTG agarose gel (Cambrex). The GeneRuler DNA ladder mix (Fermentas) was used as a marker (lane M). The DNA fragments labeled attL and attR have the expected size of the junctions between plasmid DNA and the λ chromosome, indicating integration of the λ chromosome into the plasmid. The EcoRI and NcoI fragments containing the attP fragment of the λ chromosome are 5,643 and 3,967 bp long, respectively, and could not observed on the gel. The band corresponding to the empty plasmid (804 bp) is marked by an asterisk, while arrows indicate bands corresponding to the uncleaved pBADQ plasmid, which does not contain any EcoRI sites.

Induction of Q expression in cells carrying pSEM3141Bλ resulted in growth inhibition and lysis, similar to the case with pSEM3058λ. To determine whether the zeocin resistance marker from the plasmid could be transduced by the phage particles, Pir1 cells were infected with the lysate obtained after Q induction or the control lysate form uninduced cells. Transductants that had received the zeocin marker of the plasmid occurred at a low frequency from the lysate of the induced cells but not from that of the uninduced cells. These transductant colonies carried both pSEM3141Bλ and the empty plasmid (Fig. 6, lane 2), confirming the spontaneous excision of the λ chromosome from the plasmid. Thus, we conclude that expression of Q permits the production of plaque-forming phage particles from both chromosomal and extrachromosomal DNA templates.

DISCUSSION

The choice between alternative cell fates is fundamental to biology. Many principles of cellular decision-making were first demonstrated in bacteriophage λ, one of the most comprehensively studied genetic systems (6). The λ genome encodes two developmental gene expression cascades, but the organization of the regulatory network is such that it is impossible to proceed down one developmental pathway without going at least part way down the other (22). One consequence of this organization is that multiple chromosomes are present in the host cell at the time of the decision even in the case of a single infection developing toward lysogeny (20). The choice between the two programs is determined by the interplay of diffusible transcriptional regulators (CI, Cro, and CII) (6). If the regulators can diffuse freely and are present in large enough numbers, then the regulatory outcome is expected to be uniform on all phage DNA molecules present within the cell. However, a recent study suggested that the decision is taken by individual infecting genomes, and the cellular outcome of infection is reached based on the individual decisions (23). Important aspects of the suggested model are that (i) different decisions may occur within a single cell and (ii) the outcome of infection may not be predictable based simply on the temporal development of overall intracellular concentrations of regulatory components. Our results demonstrate that the decision made at the classical CI-Cro switch is not necessarily a point of no return (Fig. 5). By supplying the Q protein in trans, the lysogenic decision could be overruled later than the time at which decision-specific events such as phage DNA integration would normally occur (12, 24). Our results thus suggest that phage particles can be produced from the nonintegrated phage genomes present in cells already containing one or more integrated phage genomes, thus providing experimental support for the theory that phage chromosomes present in the same cell can reach different decisions. However, we have no evidence that chromosomes which followed different fates in our experiments were derived from different infecting genomes.

Critical to this hypothesis is the determination of whether or not infectious phage particles are actually produced from these dual-decision cells. Unfortunately, the experimental setup shown in Fig. 4 does not allow for quantification of released phage particles, both because of the measures taken to inactivate free phage particles in order to synchronize the initial infection (see Materials and Methods) and because the majority of cells in the culture choose the lytic pathway at the time of the initial decision, thereby masking any additional phage release from the minority of cells that initially followed the lysogenic path and later switched to lysis. We note that the low frequency of lysogeny in our experiment (1 to 3%), is most likely due to the presence of a CII binding site on the pBADQ plasmid (at the pAQ promoter), resulting in the titration of cellular CII levels at the time of the initial decision. Nevertheless, our results show that infectious phage particles can be produced as a result of ectopic Q expression in established (multicopy) lysogenic cells. Given that the dual-decision cells observed in Fig. 4 would also contain multiple extrachromosomal λ genomes, they most likely release infectious phage particles as well.

Because expression of Q has no visible effect on single lysogens but induces lysis in cells carrying multiple λ prophages, we conclude that the presence of multiple phage chromosomes is in fact a prerequisite for lytic development. In this study, Q was provided externally from a multicopy plasmid. In the natural system Q is translated from the pR mRNA, the same mRNA that encodes CII. The pR mRNA is produced at a rate of 0.17/s (25) and has a half-life of about 13 min (26), suggesting that it becomes highly abundant after infection and decays slowly after pR transcription is repressed by Cro (in the case of lytic development) or CI (in the case of lysogenic development). CII levels are expected to depend strongly on the presence of pR mRNA because the CII protein is short-lived and actively degraded by host proteases (reviewed in reference 6). Based on the results presented here, we hypothesize that to establish stable lysogeny, CII must accumulate sufficiently to suppress Q production (or activity) until the number of λ chromosomes fall below a critical level required for lytic development. The period of availability of a critical number of phage λ chromosomes thus determines the duration of the noncommitted interval in which the lysogenic course can be turned to lytic development.

The dependence of lytic development on gene dosage may serve to increase the stability of established lysogens by providing protection against consequences of accidental transcription of the q gene or the late genes. Notably, the stability of a double lysogen is significantly reduced compared to that of a single lysogen in the Shiga toxin-encoding lambdoid phage ϕ24_B of E. coli O157 (27).