Elements in the λ Immunity Region Regulate Phage Development: Beyond the “Genetic Switch”

Summary

Genetic elements in the bacteriophage λ immunity region contribute to stable maintenance and synchronous induction of the integrated E. coli prophage. There is a bistable switch between lysogenic and lytic growth that is orchestrated by the CI and Cro repressors acting on the lytic (P_L and P_R) and lysogenic (P_RM) promoters, referred to as the Genetic Switch. Other less well-characterized elements in the phage immunity region include the P_LIT promoter and the immunity terminator, T_IMM. The P_LIT promoter is repressed by the bacterial LexA protein in λ lysogens. LexA repressor, like the λ CI repressor, is inactivated during the SOS response to DNA damage, and this regulation ensures that the P_LIT promoter and the lytic P_L and P_R promoters are synchronously activated. Proper RexA and RexB protein levels are critical for the switch from lysogeny to lytic growth. Mutation of P_LIT reduces RexB levels relative to RexA, compromising cellular energetics and causing a 10-fold reduction in lytic phage yield. The RexA and RexB proteins interact with themselves and each other in a bacterial two-hybrid system. We also find that the transcription terminator, T_IMM, is a Rho-independent, intrinsic terminator. Inactivation of T_IMM has minimal effect on λ lysogenization or prophage induction.

Graphical abstract

Accessory elements in the phage λ immunity region include rexA, rexB, and the P_LIT promoter; these elements modulate bacteriophage development as λ switches from lysogenic to lytic growth. P_LIT transcribes only rexB and is repressed by the E. coli LexA repressor. DNA damage inactivates the LexA and λ CI repressors, causing RexB expression to increase as λ switches to lytic development. Lytic growth is compromised if the level of RexB relative to RexA is reduced.

Introduction

The encoded information in the bacteriophage λ immunity region (Figure 1) regulates the lysogenic state of the virus and enables both stable maintenance and synchronous induction of the integrated prophage in E. coli (Echols, 1986; Oppenheim, Kobiler, Stavans, Court, & Adhya, 2005). Well-studied regulatory activites in this region include: repression of the lytic promoters P_L and P_R by the phage CI repressor protein (Ptashne & Hopkins, 1968), activation of the promoter for repressor maintenance, P_RM, by the CI repressor (Meyer & Ptashne, 1980; Ptashne et al., 1976), and autologous repression of P_RM and CI expression at higher repressor concentrations (Fong, Woody, & Gussin, 1993; Meyer, Kleid, & Ptashne, 1975). CI repressor dimers bind cooperatively to operator sites O_L and O_R that overlap the lytic promoters P_L and P_R thus inhibiting transcription of the lytic functions (Johnson, Meyer, & Ptashne, 1979; Ptashne et al., 1980). The CI repressor molecules bound at O_L and O_R interact with each other to mediate DNA looping of the intervening immunity region (Dodd, Shearwin, & Egan, 2005), thereby enhancing repression of P_L and P_R. These protein-DNA interactions make the λ prophage extremely stable in the lysogenic state. If CI repressor levels are reduced, Cro repressor will be made from the P_R promoter. Cro binds at P_RM, turning off cI expression, and activating expression of the lytic P_L and P_R promoters. Thus Cro binding induces an epigenetic change in phage gene expression, with the consequence that the bistable “Genetic Switch” is flipped from the immune configuration to a lytic configuration, which is also quite stable (Eisen et al., 1975; Ptashne, 2004; Ptashne et al., 1980).

Figure 1. Genetic map of bacteriophage λ immunity region.

The major players in the lysis-lysogeny decision are CI repressor and Cro, expressed from the P_RM and P_R operons, respectively. These two repressors bind to sites at the left and right operators, O_L and O_R, to control expression of the lytic and lysogenic promoters. Other genes in the region include rexA and rexB. The P_LIT promoter is within the coding sequence at the end of the rexA gene; thus, rexB is transcribed from two promoters, P_RM and P_LIT, while cI and rexA are transcribed only from P_RM. The black arrows represent the P_RM and P_LIT transcripts. Immediately downstream of the rexB open reading frame is the transcriptional terminator, T_IMM.

Other genetic elements are present in the λ immunity region but are less well characterized. Two other genes, rexA and rexB, are located downstream of cI in the P_RM operon. Benzer (1955), during his fine-structure mapping of the bacteriophage T4rII region, made the observation that a (Rex⁺) λ lysogen prevents the growth of T4rII mutant phage. Both RexA and RexB proteins are necessary for this phenotype. Benzer’s work was seminal to our current understanding of the genetic code and genetic recombination. Although 50 years have passed, the roles of the T4rII proteins and the λ Rex proteins have still not been clearly elucidated. A second promoter, P_LIT, is located in the distal end of the rexA gene and transcribes the rexB region (Hayes, Bull, & Tulloch, 1997; Hayes & Szybalski, 1973; Landsmann, Kroger, & Hobom, 1982; Pirrotta, Ineichen, & Walz, 1980). The rexA and rexB genes encode protein products: RexA is a soluble protein, and RexB is an inner membrane protein predicted to have four integral transmembrane domains (Parma et al., 1992). Transcription immediately beyond rexB is limited by a transcription terminator, T_IMM (Landsmann et al., 1982). To better understand the genetic regulation and evaluate individual contributions of the P_LIT promoter and the immunity terminator, T_IMM, to the overall regulation of this region, we have characterized the behavior of these cis-acting elements individually and in the context of the full immunity region, using genetic reporters. We examined the effect of these mutations on establishment of lysogeny, prophage induction and the switch to lytic growth. Our results provide new insights into the complex regulation of the λ prophage. We show that P_LIT is a member of the E. coli SOS LexA regulon, and that when RexA is present, P_LIT activity is necessary for a robust lytic response. We also show that T_IMM is an intrinsic terminator that prevents transcription originating in the immunity region from reaching the P_L operon. These new findings enhance our understanding of λ development and the transition between lysogenic and lytic growth.

Results

The phage λ P_LIT promoter is a member of the SOS regulon.

We constructed a lacZ reporter in single copy on the E. coli chromosome (Figure 2A). Upstream of lacZ we inserted 138bp of phage λ DNA (λ coordinates 36248–36385), purported to contain a promoter, P_LIT (Landsmann et al., 1982; Pirrotta et al., 1980). The distal end of the λ segment includes the first four codons of the rexB gene fused in frame to the seventh codon of lacZ to form the gene fusion rexB-lacZ. The other end extends 126bp upstream of rexB, and includes the last 36 codons of rexA. Experiments with this fusion, in strain LT1771, shows that this piece of λ DNA activates lacZ expression (Figure 2B–1, Table S2), consistent with previous reports that this DNA region contains the transcription promoter, P_LIT. We identified −35 and −10 promoter elements in this λ DNA segment, as indicated in red in Figure 2C, line 1, and spanning λ coordinates 36315–36344 using information theory (Shultzaberger, Chen, Lewis, & Schneider, 2007), the same mathematics used for sequence logos (Schneider & Stephens, 1990) (for details see Figure S1). These promoter elements do not correspond to the hypothetical promoter described previously by others (Landsmann et al., 1982; Pirrotta et al., 1980), which is indicated in blue in Figure 2C, line 1, and spans λ coordinates 36278–36310. We introduced mutations (underlined in Figure 2C, line 2) into the −10 element of the promoter we identified to generate strain LT1989. This alteration of the −10 sequence abolished lacZ expression (Figure 2B–2, Table S2, Figure S1). This confirms our identification of P_LIT, and at the same time demonstrates that no other promoter exists on this λ DNA segment. The mutant promoter is hereafter referred to as P_LIT-10.

Figure 2. Identification, characterization and regulation of the P_LIT promoter.

A. A β-galactosidase reporter was used to characterize the P_LIT promoter. The distal end of the rexA gene containing the P_LIT promoter region and the beginning of the rexB gene were inserted at the lac operon such that the first four codons of rexB are fused in frame to the seventh codon of lacZ. B. Normalized β-galactosidase units obtained with the reporter diagrammed in A using different bacterial strains. In each case the open bars are values obtained in the absence of Mitomycin C, and the gray bars are values obtained after 2.5 hours of growth in the presence of 1μg/ml drug. Results for seven different genetic configurations in the six strains tested are presented, as indicated below the bar graph. The numbers 1, 2, 3, 4 in the bar graph correspond to the DNA sequences 1, 2, 3, 4 in Figure 2C. 1. the reporter with the wild-type promoter and LexA binding site, in a wild-type bacterial host (LT1771); 2. the reporter with the P_LIT-10 promoter mutation, in a wild-type host (LT1989); 3. the P_LIT reporter with the mutant LexA binding site, in a wild-type host (KM18); 4. the P_LIT reporter with the improved LexA binding site, in a wild-type host (KM17); 5. the wild-type reporter in a bacterial strain deleted for the lexA gene (KM14); 6. the wild-type reporter in a bacterial strain deleted for the recA gene (KM12). Numerical data are in Table S2. C. The P_LIT promoter sequence indicating the LexA binding site and mutations. The region of λ DNA containing the P_LIT promoter is illustrated (λ coordinates 36346–36275). Line 1 represents the wild type sequence. The end of the rexA gene is indicated by the TAA translation stop codon in brackets. The −35 and −10 elements of the promoter we identified as P_LIT are indicated in red type, the promoter previously predicted by others is indicated in blue. The LexA binding site is indicated with a black box. 2. The four base changes made to create the P_LIT-10 mutation are underlined (the G->A change downstream of the −10 is a neutral change made to facilitate identification of the mutant with PCR); 3. the underlined base changes destroy the LexA binding site; 4. the underlined base changes create an improved LexA binding site. The sequences were analyzed using sequence walkers, see Supporting Information Figure S1.

It has been reported that transcription of rexB continues during prophage induction, when CI repressor is inactivated (Hayes et al., 1997; Hayes & Szybalski, 1973; Liu, Jiang, Gu, & Roberts, 2013). Sequence analysis of the region upstream of rexB had suggested the presence of a LexA repressor binding site (Hayes et al., 1997). LexA binding sites are well-defined and LexA binding reduces transcription from the DNA damage inducible promoters they regulate (Lewis, Harlow, Gregg-Jolly, & Mount, 1994). We again used information theory to analyze the region and identified a possible LexA binding site overlapping the P_LIT promoter (Figure 2C, line 1, indicated by the black box; Figure S1. To characterize LexA control of P_LIT, we compared P_LIT with a known LexA-controlled promoter, P_SFI. Strain LT1771, as well as containing the P_LIT-lacZ reporter (Figure 2A) also contains a luciferase reporter construct, P_SFI-luc, in which luciferase expression is under control of LexA via the SOS-inducible promoter P_SFI (Li, Thomason, Sawitzke, Costantino, & Court, 2013). We found that β-galactosidase expression from P_LIT increased when the LT1771 cells were challenged with the DNA damaging agent Mitomycin C (see Figure 2B–1 and Table S2). In the same experiment, luciferase expression from P_SFI also increased as expected (see Table S2). To ask whether the putative P_LIT LexA binding site was accurately identified, we made two types of mutations in its DNA sequence (Figure 2C, lines 3 and 4, respectively, with modified bases underlined; Figure S1); one to destroy the site (strain KM18), and one to improve it (strain KM17). These mutations were designed using information theory and have not been published elsewhere. The mutation that destroyed the binding site eliminated LexA-dependent regulation of the P_LIT promoter and resulted in an ~3-fold higher level of lacZ expression (Figure 2B–3, Table S2). The mutation that improved the LexA binding site resulted in P_LIT being more tightly regulated by LexA, as evidenced by a lower level of expression, independent of the presence of Mitomycin C (Figure 2B–4, Table S2). We tested the response of the P_LIT reporter in cells deleted for the lexA gene (strain KM14); this deletion results in constitutive SOS expression, and also results in constitutive P_LIT expression independent of Mitomycin C addition (Figure 2B–5, Table S2). When the P_LIT-lacZ reporter with the improved LexA binding site was tested in a strain lacking the lexA gene (LT2312, Table S2) the level of P_LIT expression was high and independent of drug addition. RecA function is required for SOS induction (Kuzminov, 1999): RecA protein bound to single-stranded DNA arising as a result of DNA damage forms a complex, activated RecA (RexA*) (Ghodke et al., 2019), which acts as a co-protease to promote LexA represser autocleavage (Little, 1982). When recA was deleted from strain LT1771 (strain KM12) the P_LIT promoter showed little response to Mitomycin C (Figure 2B–6, Table S2).

The phage λ transcription terminator T_IMM is an intrinsic, Rho-independent terminator.

Immediately downstream of the rexB stop codon, phage λ DNA contains a G/C rich region that, in the RNA transcript, can form a 6 bp G/C stem followed by a run of seven U residues interspersed with a single C (U₄CU₃) (Figure 3A, B; Figure S1). This sequence is typical of an intrinsic Rho-independent transcription terminator. Because this terminator is located in the immunity region, it has been named T_IMM (Daniels, 1983), and is thought to be the site where transcripts initiating at the P_RM and P_LIT promoters terminate (Hayes & Szybalski, 1973; Landsmann et al., 1982). We used a firefly luciferase reporter (Figure 3A) to characterize this transcription terminator. In this reporter construct, the P_LIT promoter drives expression of rexB with the T_IMM terminator immediately downstream as it is found in the native phage configuration. Beyond the terminator, we placed the luciferase reporter (luc) (Figure 3). Mutations that abolish the potential for G-C pairing in the stem (Figure 3C, Figure S1) result in >4-fold higher luciferase activity when compared to the wild type sequence (Figure 3B vs. 3C; GM33 vs GM15; Table S3A; Figure S1). Hence our data demonstrate that the λ DNA sequence downstream of rexB behaves as a transcription terminator in vivo when the stem is intact, blocking ~76% of the transcription from P_LIT.

Figure 3. Identification, characterization and regulation of the T_IMM transcription terminator.

A. Transcription from P_LIT through rexB was monitored with this reporter in which the firefly luc gene is placed beyond the string of Ts defining the end of T_IMM. The delta sign indicates that the region upstream of rexA including the cI gene and P_RM promoter are absent from this construct. The DNA sequence between rexB and the luc gene is shown, with the stop codon of rexB indicated in cyan and the ATG start codon of luciferase indicated in green. The sequence of the T_IMM terminator is underlined, and the forward slash indicates where DNA was deleted from LT732 to form this reporter (see construction details in Supporting Information). B. A schematic of the wild-type structure of the λ immunity region transcription terminator, T_IMM, formed in the RNA, with the rexB stop codon indicated in cyan. The normalized luciferase measurement and standard deviation (n=11 independent replicates) obtained from the reporter illustrated in 3A containing this wild-type terminator sequence is also indicated (strain GM15). C. Mutations made in the T_IMM stem to prevent pairing are indicated in red, and the resulting luciferase measurement and standard deviation (n=11 independent replicates) from the corresponding reporter in strain GM33 is reported. D. Wobble positions at the end of the rexB coding sequence in the A-rich region immediately upstream of the stem-loop were converted from adenine to guanine nucleotides, while maintaining the wild type rexB amino acid sequence. Mutations are indicated in red. The luciferase measurement and standard deviation (n=6 independent replicates) from the reporter containing these mutations is indicated (strain GM32). Numerical data are in Table S3.

Some E. coli intrinsic terminators have an A-rich region in the RNA immediately upstream of the stem that contributes to their strength (Chen et al., 2013). The distal end of the rexB transcript is A-rich and just upstream of the stem. We made silent wobble changes of adenine to guanine in the last four codons of rexB (Figure 3D) to determine whether this A-rich region affects T_IMM. We found that these mutations reduced termination efficiency to ~33% of wild type (Figure 3B vs. 3D; Table S3A, GM32 vs GM15; Figure S1) even though the G-C rich stem-loop and the downstream run of U’s were not altered.

λ T_IMM has been reported to be Rho-dependent (Cardinale et al., 2008) although it has the characteristic structure of an intrinsic terminator. To assess the dependence of T_IMM on Rho protein, we assayed its ability to terminate transcription from P_LIT in a rho15::bla mutant (Singh et al., 2016) background or in the presence of bicyclomycin (BCM), a specific Rho inhibitor (Zwiefka, Kohn, & Widger, 1993). T_IMM was fully functional for termination in the rho15::bla mutant background (Table S3A, GM37 vs GM38). Transcription from P_LIT also terminated efficiently in the presence of BCM, an inhibitor of Rho termination (Table S3A). Using the same two concentrations of BCM, we demonstrated a 2.7-fold reduction in transcription termination at a known Rho-dependent terminator, T_R1 (Table S3B, strain LT1055), the previously reported reduction with the rho15 allele was 2.1-fold (Court et al., 1980). Collectively, our data are consistent with T_IMM belonging to the class of intrinsic transcription terminators that contain an upstream A-tract contributing to terminator strength (Chen et al., 2013).

Analysis of relative transcription strength of the P_RM and P_LIT promoters.

To compare the relative strengths of the P_RM and P_LIT promoters, we made a luciferase reporter construct containing a λ DNA fragment carrying both promoters (strain LT2056, Figure 4A). When both P_RM and P_LIT are active, the luciferase signal is ~2×10⁷ RLU (Figure 4B, Table S4). We also modified this construct by generating point mutations that inactivated the −10 regions of P_RM and P_LIT, either individually or together (see Experimental Procedures and Figure S1). Surprisingly, the P_RM promoter (LT2068) and the P_LIT promoter (LT2078) have similar strengths (Figure 4B) and the luciferase activities from P_RM (~6×10⁶ RLU) and P_LIT (~4 ×10⁶ RLU) are not significantly different. When both promoters are present the signal is roughly additive. Removal of both promoters by mutation in the same construct (strain LT2079) reduced the signal more than 100-fold. Note that these reporters express functional CI repressor protein, which activates the basal level of P_RM 10-fold (Ptashne et al., 1976).

Figure 4. A comparison of the relative strengths of the P_RM and P_LIT promoters using a luciferase reporter.

A. This reporter is similar to the reporter illustrated in Figure 3A but also contains P_RM and the cI gene. All reporters were mutant for the T_IMM stem (i.e. contained the mutations illustrated in Figure 3C). B. Versions of this reporter containing both the P_RM and P_LIT promoters (strain LT2056), either P_RM (LT2068) or P_LIT (LT2078), or neither promoter (LT2079), were assayed at 32ºC. As described in Experimental Procedures, point mutations in the −10 region were used to inactivate the promoters. Numerical data are in Table S4.

Analysis of the effect of the T_IMM stem mutation on P_RM and P_LIT transcription.

To determine the impact of preventing transcription termination at T_IMM on the lytic promoters, we used variants of the dual P_L and P_R reporter described in the legend of Figure 5A (Svenningsen, Costantino, Court, & Adhya, 2005). We see increased expression of the luc gene beyond P_L (Figure 5, Table S5) when T_IMM is defective and CI repressor is active, however, there is no corresponding increase in lacZ expression from P_R (Table S5). This suggests that the increased luciferase activity beyond P_L might be due to upstream transcription proceeding beyond T_IMM into the P_L operon. To determine whether the luciferase transcription originated from P_L or upstream of it, we made point mutations in the −10 region of P_L (Figure S1) and measured luciferase with and without the terminator T_IMM (Figure 5B). Even with the P_L-10 mutation, the luciferase signal remained higher in the T_IMM stem mutant, consistent with transcription read-through from P_RM and/or P_LIT.

Figure 5. A dual luciferase and β-galactosidase reporter used to determine the effect of the T_IMM stem mutation on the lytic promoters P_L and P_R.

A. Schematic of a dual reporter (Svenningsen et al., 2005), with the phage λ immunity region inserted at the E. coli lac operon. The lacI and the lac promoter are removed, while leaving the lacZ ribosome-binding site and the rest of the lacZYA operon intact such that expression of lacZ is driven from the P_R lytic promoter. Additionally, the firefly luc gene is fused to N and is transcribed from the P_L lytic promoter; note that luc may also be expressed from P_LIT and/or P_RM. Versions of this reporter were used to determine the effect of mutations in T_IMM and P_L on lacZ and luc expression. B. Luciferase activity from the luc gene downstream of P_L measured for four different strains, varying T_IMM transcription termination activity and P_L promoter activity: 1. LT732, wild-type T_IMM terminator and wild-type P_L; 2. GM3, stem mutation in T_IMM terminator and wild-type P_L; 3. LT2093, wild-type T_IMM terminator and P_L-10; and 4. LT2093, stem mutation in T_IMM terminator and P_L-10. Assays were performed at 32°C, with CI repressor present (under immune conditions). Numerical data for both P_L-luc and P_R-lacZ are in Table S5.

Effect of the P_LIT and T_IMM mutations on phage λ development.

The immunity region controls both the establishment and maintenance of the prophage state. Thus, we looked at possible developmental defects caused by the P_LIT and T_IMM mutations compared to wild-type λ during establishment and maintenance of lysogeny, as well as the switch to lytic growth in response to UV-induced DNA damage.

To ask about effects of the P_LIT and T_IMM mutations on the ability of phage λ to form lysogens, we used a λ phage carrying a gene encoding KanR (Henry & Cronan, 1991; St-Pierre & Endy, 2008), which allows selection of lysogens as drug resistant colonies. Lysogenization frequencies for both low (0.1) and high (5) multiplicities of infection (m.o.i.) were determined (Table S6). At a multiplicy of 5 phage particles per cell, more than 20% of the infected cells were successfully lysogenized, independent of genotype, while at low multiplicity, between 0.2–0.3% of the infected cells were lysogenized, independent of genotype. Both the wild-type λ values and those of the mutants are not significantly different from each other or from those observed by Kourilsky (1973) in his seminal studies. Thus, neither mutation confers a defect on the ability of the phage to enter the prophage state. We also found that lysogens carrying either of these mutations grow normally in L broth (Figure S2).

We examined the stability of the lysogens and the switch to lytic growth in response to various doses of UV. At higher UV doses (>10 J/m²), we found no difference between lysogens of the λ wild-type and the λ T_IMM mutant, where each gave a yield of ~250 phage/bacterium (Figure 6; Table 2). At very low UV doses, however, a slightly higher level of phage release is seen for the T_IMM mutant phage. This effect is most significant in the absence of UV (Figure 6).

Figure 6. UV induction curves for λ lysogens inducible by DNA damage.

UV induction of λ lysogens was performed as described in Experimental Procedures. LT445, MG1655(λ) (●) (n=7); LT2003, MG1655(λ T_IMM-) (Δ) (n=4); LT1997, MG1655(λ P_LIT-10) (○) (n=5), with SEM shown. The higher level of spontaneous phage for the T_IMM- mutant at t=0 and the reduced final phage yield of the P_LIT-10 mutant are evident. Titers of the lysates were determined at 2hr post UV irradiation by plating on A584; numerical data for this time point are in Table 2.

Table 2.

Effect of P_LIT-10 and T_IMM- mutations on prophage induction in response to UV-mediated DNA damage

Lysogenic strain	Prophage genotype	Phage/Bacterium†
LT445	WT λ	264 ± 41 (n=7)
LT1997	λ PLIT-10	25 ± 12 (n=5)
LT2003	λ TIMM-	278 ± 57 (n=4)
LT1676	λ PLIT+ rexA<>cat	224 ± 33 (n=3)
LT1677	λ PLIT+ ¾rexB<>cat	231 ± 59 (n=3)
LT 2156	λ PLIT-10 rexA<>cat	241 ± 38 (n=4)
LT 2157	λ PLIT-10 ¾rexB<>cat	264 ± 72 (n=4)

^†Cell titers were determined as described in Experimental Procedures. Lysates were titered 2 hours post UV irradiation. SD shown. Note that the cat insertions in both rexA and rexB are non-polar in-frame gene replacements. The rexA<>cat allele retains P_LIT+, and the cat insertion in rexB leaves the A-rich region upstream of T_IMM intact.

UV induction of the λ P_LIT-10 mutant lysogen (strain LT1997) yielded an ~10-fold lower phage titer compared to that for wild-type λ (Figure 6; Table 2). The RexB protein is normally expressed from two promoters, P_RM and P_LIT, while the RexA protein is only expressed from P_RM (see Figure 1). The relative expression levels of RexB and RexA may be important for phage production. To determine whether the level of RexA protein relative to RexB is involved in the poor response to UV induction, we examined the requirement for Rex function during induction of the prophage (Table 2). Removing RexA and/or RexB function restored normal phage yields on induction in response to a fully inducing UV dose. Thus, when both Rex functions are present, the relative levels of RexB to RexA appear critical and a ratio of RexB greater than RexA is needed to efficiently produce phage.

To ask whether the P_LIT-10 defect during prophage induction is specifically dependent on UV-mediated DNA damage, we examined temperature induction of a λ cI857 lysogen containing the P_LIT-10 mutation. Heat inactivation of the temperature-sensitive repressor allows lysogens with this allele to enter the lytic state without experiencing DNA damage or a requirement for RecA co-protease (Kuzminov, 1999). When the P_LIT-10 cI857 prophage is induced to undergo lytic growth by raising the temperature, it also displayed an ~10-fold reduced phage yield relative to P_LIT⁺, like UV induction of the P_LIT-10 mutant. Thus, the P_LIT-10 defect in phage yield is independent of the method of λ induction. Coincident with the reduction in phage yield, P_LIT-10 cI857 also lysed prematurely relative to P_LIT⁺ (Figure 7A). A Rex effect on the timing of lysis for induced lysogens has been previously observed (Campbell & Rolfe, 1975; Rolfe & Campbell, 1977). The λ S holin protein controls the timing of cellular lysis during the λ lytic cycle (Wang, Smith, & Young, 2000). To explore the relationship between phage yield and premature lysis caused by the P_LIT-10 mutation, we asked whether the premature lysis phenotype required a functional S holin, by introducing an Sam7 mutation into the cI857 P_LIT-10 prophage. The Sam7 mutation prevents λ-mediated cell lysis from occurring (Wang et al., 2000), and an induced culture with this mutation must be lysed manually. In a Sam7 background both cell growth and phage production continue until cells are lysed from without; and an extended growth period following prophage induction can result in unusually high phage titers. In the P_LIT-10 mutant background, the Sam7 allele prevented premature lysis, but did not restore normal growth of the lysogen, nor did it restore phage yield to P_LIT+ levels (Figure 7B, see figure legend for titers). Thus, premature lysis itself is not the cause of the reduced phage yield.

Figure 7. Lysis curves for heat-inducible λ lysogens.

High temperature induction of λ cI857 lysogens was performed as described in Experimental Procedures. A. LT447, MG1655(cI857) (●) and LT2228, MG1655(P_LIT-10 cI857) (○). Average phage yield for the MG1655(cI857) culture was 1.4×10¹⁰/ml±3.7×10⁹/ml (n=5), while that of the MG1655(cI857 P_LIT-10) culture was 2.5×10⁹/ml±1.0 ×10⁹/ml (n=5); SD shown for five independent replicates. B. LT2228, MG1655(P_LIT-10 cI857) (○); LT2231, MG1655(P_LIT-10 cI857 Sam7) (♦). The same LT2228 lysis curve presented in A is also shown in this graph for comparison. CHCl₃ was added to the Sam7 cultures at 55min (indicated by the vertical arrow), after which lysis ensued. Average phage yield for the MG1655(P_LIT-10 cI857 Sam7) culture was 3.0×10⁹/ml±2.4 ×10⁸/ml; SD shown for three independent replicates.

RexB complementation restores normal prophage behavior:

Since the P_LIT-10 phenotype requires active rexA and rexB functions, an imbalance between the levels of RexA and RexB proteins may be responsible for the phage induction and growth defects observed in the P_LIT-10 mutant background, as originally suggested by Parma et al. (1992). To confirm that the P_LIT-10 problems result from a reduced level of RexB protein in bacteria making normal levels of RexA protein, we expressed an additional copy of rexB in trans from the E. coli P_BAD promoter by adding arabinose, and then induced the P_LIT-10 cI857 lysogen at high temperature. Expression of the second copy of rexB fully complemented both the abnormal lysis profile and the reduced phage yield phenotype observed in the lysogen with the P_LIT-10 mutation (Figure 8A), consistent with our supposition that an altered ratio of the proteins causes defective phage development. In contrast, in cI857 lysogens having normal physiological levels of both Rex proteins, expression of the P_BAD-rexB construct had no effect on either parameter (Figure 8B).

Figure 8. Expression of additional rexB complements P_LIT-10 phenotypes and does not alter growth of a P_LIT+ lysogen.

Lysis curves are shown for temperature induction of λ cI857 lysogens with pBAD-rexB expressed from the arabinose operon. Overnights were subcultured 200-fold into either L broth or L broth containing 0.2% arabinose. A. LT2314, MG1655(P_LIT-10 cI857) pBAD-rexB. (○) growth without arabinose induction of pBAD, average phage titer was 2.4×10⁹ (n=4); (●) growth in the presence of 0.2% arabinose, average phage titer was 1.1×10¹⁰ (n=4). B. LT1961, MG1655(cI857) pBAD-rexB. (○) growth without induction of pBAD, average phage titer was 1.8×10¹⁰ (n=3); (●) growth in the presence of 0.2% arabinose, average phage titer was 1.21×10¹⁰ (n=4). SD shown for at least three independent replicates. Note that lysogens were grown to OD₆₀₀ 0.4 before the temperature shift.

The P_LIT-10 promoter mutation perturbs cellular energy in induced lysogens:

Membrane depolarization and collapse of proton motive force (PMF) are predicted to occur when RexA protein levels are abnormally high relative to RexB levels (Parma et al., 1992; Snyder & McWilliams, 1989). An active PMF is necessary for normal cellular energetics, since ATP synthesis by the F₁F_O membrane ATPase requires the proton gradient component of PMF. Thus, if PMF is abnormal in the P_LIT-10 background, ATP synthesis is predicted to be compromised. Accordingly, we compared ATP levels in both wildtype and P_LIT-10 cI857 temperature-induced lysogens (Figure 9), as a proxy for measuring PMF. Following heat induction, the wildtype cI857 lysogen maintains high ATP levels until lysis occurs, however, when the P_LIT-10 cI857 mutant prophage is induced by a temperature shift, ATP levels decline several-fold following prophage induction, beginning ~15 min after the shift to high temperature. Thus, some aspect of attempted phage development triggers an energetic dysfunction, a similar phenotype as that observed for Rex exclusion of T4rII phage.

Figure 9. Normalized ATP levels in heat-induced lysogens.

High temperature induction and ATP measurements were as described in Experimental Procedures for strains LT447, MG1655(cI857) (●) and LT2228, MG1655(P_LIT-10 cI857) (○). Note that before the temperature shift (t=0) and until t=10min, both cultures maintain equal ATP levels, but by 15min post induction ATP levels have dropped for MG1655(P_LIT-10 cI857). The SD is shown for three independent replicates.

Two-hybrid Analysis of RexA and RexB.

If RexA and RexB interact, as previously proposed (Parma et al., 1992), the ratio of the two proteins would determine which multimeric protein species would be likely to form during prophage induction and phage development. To explore possible RexA-RexB interactions, we used the Bacterial Adenylate Cyclase Two-Hybrid (BACTH) system, which is performed in E. coli and allows detection of protein-protein interactions, including those with membrane proteins (Karimova, Gauliard, Davi, Ouellette, & Ladant, 2017; Karimova, Ullmann, & Ladant, 2001; Ouellette, Karimova, Davi, & Ladant, 2017). We constructed eight plasmids by inserting the rexA and rexB genes into each of the four cyclase vectors as gene fusions, containing either the T25 adenylate cyclase domain on the N- or C-terminus of the Rex protein, or the T18 adenylate cyclase domain on the N- or C-terminus of the Rex protein (see diagram of fusions in Figure 10 and Table S7 for bacterial strains). Those pairs of Rex fusions that interact generate functional adenylate cyclase and cAMP. The presence of cAMP enables expression of β-galactosidase from the lac operon (Figure 10). Indeed, we found evidence that the RexA and RexB proteins do interact, both with themselves and with each other. All four plasmid pairs of RexB tested against itself were positive in the β-galactosidase assay, independent of the location of the adenylate cyclase domains (Figure 10 a-d), while two of the four plasmids pairs of RexA tested against itself were positive in the β-galactosidase assay (Figure 10 e-h). Surprisingly, only two of the eight possible RexB by RexA plasmid pairs were weakly positive in the β-galactosidase assay (Figure 10 i-p). We also examined the colony phenotype of all sixteen strains containing the pairs of plasmids on MacConkey Maltose agar. The level of β-galactosidase measured for each pair (Figure 10, Table S7) correlated well with the appearance of red colonies indicating maltose utilization, which is dependent on cAMP (Figure S3).

Figure 10. Protein-protein interactions for RexA and RexB.

β-galactosidase was measured in a cya mutant strain (BTH101) containing a pair of compatible plasmids, one KanR and one AmpR. The KanR plasmids contained the DNA encoding the T25 fragment (aa 1–224) of the Bordatella pertussis adenylate cyclase protein fused to either the rexB or rexA gene at either the N- or C-terminus. The AmpR plasmids contained the DNA encoding the T18 fragment (aa 225–399) of the same adenylate cyclase protein, also fused to the rexB or rexA gene at either the N- or C-terminus. If the Rex proteins interact, the two adenylate cyclase domains functionally interact to generate cAMP, allowing lacZ expression. At least three independent replicates were performed for each strain, and the SEM is shown. The bars in the graph are color-coded in gray-scale, with the locations of the Cya domains on the Rex proteins indicated in the accompanying diagrams, where red indicates T25, cyan indicates T18, and yellow indicates RexA or RexB. Bars (a-d) show RexB-RexB interaction, and bars (e-h) show RexA-RexA interaction. Eight pairwise interactions were tested for RexA-RexB: (i-l) shows the data for RexA fused to the T25 domain and RexB fused to the T18 domain, while (m-p) shows the data for RexB fused to the T25 domain and RexA fused to the T18 domain. Note that (m-p) values are no higher than the background obtained in control assays with BTH101 containing combinations of vectors lacking Rex insertions, which gave ~28 units of β-galactosidase. BTH101 lacking plasmids gave ~4 units of activity. Strains used for the two-hybrid analysis and numerical values for β-galactosidase measurements are in Supporting Information (Table S7). Colony phenotypes on MacConkey Maltose are shown in Figure S3.

Discussion

Our analyses shed new light on the regulation and function of two cis-acting genetic elements in the bacteriophage λ immunity region: the P_LIT promoter and the T_IMM terminator. The P_LIT promoter transcribes a single phage gene, rexB. The rexB gene is also expressed from the P_RM promoter, coordinately with cI and rexA (Figure 1) (Hayes et al., 1997; Hayes & Szybalski, 1973; Parma et al., 1992). We find that failure to initiate transcription from the P_LIT promoter causes defects in lytic growth and phage yield (Figures 6 and 7), but only when both rexA and rexB functions are active (Table 2 and Figure S4). These defects occur irrespective of the mode of CI repressor inactivation, i.e., by DNA damage or by high temperature. Theoretically, failure to establish normal lytic growth might be due to a failure of the CI/Cro bistable switch to fully establish the lytic pathway of development following DNA damage and CI repressor inactivation. Alternatively, the switch to lytic growth may be established, but subsequent phage DNA replication and/or lytic gene expression may be impacted. Since we still see P_LIT-10 mutant defects when CI repressor is inactivated by temperature, our experiments favor the idea that a problem in phage development occurs after the switch to lytic growth has occurred but cannot rigorously exclude some modulating effect on the switch itself.

P_LIT is naturally constitutive and weakly repressed by LexA (Figure 2B, Table S2). Our reporter results (Figure 2, Table S2) show a two-fold activation of P_LIT following LexA cleavage, which is a small effect when compared to other promoters activated by SOS (Simmons, Foti, Cohen, & Walker, 2008). In our λ induction experiments the prophage was reactivated by two different methods: DNA damage, which results in cleavage of both CI and LexA repressors, and high temperature, which inactivates CI, but does not affect LexA. The yield of P_LIT+ phage was the same for both prophage induction methods, thus this set of experiments fails to show any effect of LexA regulation on P_LIT. The two-fold increase of RexB in response to DNA damage and LexA cleavage is small when compared to the 16-fold difference between the active P_LIT+ promoter and the defective P_LIT-10 promoter (Table S2); this likely accounts for our failure to observe a difference in phage yields for SOS induction vs. temperature induction. P_LIT is the second λ promoter reported to be regulated by LexA; the other is P_O, the promoter for a small RNA, oop (Lewis et al., 1994), which is antisense to the cII gene. This cII antisense transcript inhibits cII expression (Krinke & Wulff, 1987) and helps to lock in the lytic state (Kobiler, Koby, Teff, Court, & Oppenheim, 2002). The λ P_LIT and P_O promoters, both with LexA binding sites, should be coordinately upregulated during DNA damage as previously suggested (Hayes et al., 1997). The role of LexA-mediated regulation of these λ promoters during prophage induction in response to DNA damage has not been carefully characterized and thus, remains an enigma. It may be that under some growth condition, this additional layer of LexA-mediated regulation does impact phage development.

Even after CI repressor inactivation there is measureable expression of rexA from P_RM (Liu et al., 2013). Both RexA and RexB protein functions are required for the mutant P_LIT-10 phenotype, as removal of either rex gene restores normal λ yields in the promoter mutant background (Table 2). While the roles of RexA and RexB (i.e. the Rex system) are not well understood, it is known that when expressed together, they can cause cellular defects. Wild type (Rex⁺) λ lysogens prevent the growth of bacteriophage T4rII mutants. Such lysogens, after infection with T4rII phage, suffer energetic defects (Colowick & Colowick, 1983; Sekiguchi, 1966). RexA over-expression relative to RexB causes phenotypes consistent with loss of PMF and cellular energy, even in the absence of T4rII infection (Snyder & McWilliams, 1989). In harmony with this result, Parma et al. (1992) demonstrated that overexpressing RexB relative to RexA rescues growth of T4rII and suggested that under various cellular conditions, an appropriate ratio of RexA to RexB is critical for normal cellular energetics. Our experiments show that the Rex system also affects λ lytic development. Expression of rexB from the P_BAD promoter restores normal growth and yield to the P_LIT-10 mutant phage (Figure 8A); thus, P_LIT is required to maintain a necessary balance of RexB to RexA, and failure to maintain this balance compromises λ growth.

Maintenance of the λ S holin in the closed state requires cellular energy (Grundling, Manson, & Young, 2001); any reduction of PMF and energy levels triggers premature S holin opening and cell lysis. We suspect that a lowered RexB to RexA ratio in the P_LIT-10 mutant lysogen impairs PMF and subsequent phage development. The lower phage yield we observe following induction of the P_LIT-10 lysogen (Figure 6 and Figure 7) may be due to the altered cellular energy state caused by the lower RexB/RexA ratio, reducing ATP levels necessary for phage production (Figure 9). Our results are consistent with the hypothesis that cellular energy defects result from an imbalance in expression of the RexA and RexB proteins. Evidently a primary role of P_LIT is to maintain RexB protein at high levels relative to RexA to prevent loss of cellular energy and thus, “self-exclusion” of the phage, as suggested by Parma et al. (1992).

Parma et al. (1992) showed that RexB resides in the inner membrane, while RexA is cytoplasmic. Our two-hybrid results (Figure 10) show that both RexB and RexA self-multimerize. In the absence of the other Rex protein, neither of these postulated self-interactions causes cellular dysfunction when expressed at physiological levels, and we suspect that each of these self protein complexes has some as-yet unidentified role in the phage life cycle. RexB and RexA also interact with each other, but with reduced affinity or transiently relative to their self-interactions. Overexpression of RexA relative to RexB causes membrane depolarization (Snyder & McWilliams, 1989) and may enhance RexA-RexB interactions. By increasing the ratio of RexA to RexB, the P_LIT-10 mutation would have a similar effect, although of smaller magnitude.

Why does λ maintain the Rex functions? While the Rex system is present during the lysogenic and lytic life cycles, it is not essential for either, however, it may confer some advantage for the phage or its lysogenic host. As an example, (Rex⁺) λ lysogens out-compete non-lysogens in a glucose-limited chemostat (Lin, Bitner, & Edlin, 1977). Alternatively, the Rex system may serve as an environmental sensor, perhaps monitoring cellular energetics, and thus, be sensitive to superinfecting phages. Overexpression of RexAB from a multicopy plasmid prevents growth of wild type T4 as well as other phages (Shinedling, Parma, & Gold, 1987). Phage infection causes a partial depolarization of the cytoplasmic membrane (Labedan & Letellier, 1981) and this PMF perturbation may trigger a conformational change in the Rex system that results in superinfection inhibition.

The λ-like defective Rac prophage found in most E. coli K-12 strains has been shown to have a Rho-dependent terminator in the immunity region (Cardinale et al., 2008), and it was postulated that T_IMM, located in the same region in λ, was also a Rho-dependent terminator. However, our analysis shows that the λ T_IMM looks and behaves like an intrinsic Rho-independent terminator, having a G-C rich hairpin followed by a string of Us in the RNA. Examination of the Rac prophage reveals no analogous hairpin structure in the region between the P_L promoter and the Rac repressor gene. Thus, in contrast to the Rac prophage immunity terminator, we find that λ T_IMM is Rho-independent. Our experiments show that T_IMM is ~76% efficient at terminating transcription, thus similar in strength to the intrinsic terminator λ T_R2, which has a termination efficiency of ~77% (Nojima, Lin, Fujii, & Endo, 2005) and is located downstream of the phage replication genes (Leason & Friedman, 1988). In the absence of a functional T_IMM terminator, transcription originating in the immunity region continues into the P_L operon (Figure 5B), and the prophage may escape repression more easily (Figure 6). However, even with this leaky transcription beyond P_L, cell growth of the T_IMM mutant lysogen is normal in L broth (Figure S2), despite the presence of toxic genes downstream (Sergueev, Yu, Austin, & Court, 2001).

RNA polymerases that fail to terminate transcription at T_IMM proceed into a region that includes the P_L promoter and its O_L operators. Under immune conditions, the T_IMM mutant allows an ~three-fold higher expression of the luc gene downstream of P_L (Figure 5B, Table S5). It is not immediately clear why luciferase levels increase in the absence of termination at T_IMM. One possibility is that transcription read-through from upstream promoters in the immunity region (P_RM and P_LIT) proceeds directly into the P_L operon. However, in this scenario, as transcription proceeds into the P_L operon through the left operator sites, CI repressor would be momentarily displaced by RNA polymerase transit, perhaps resulting in a brief derepression of P_L and a burst of expression of the operon. Thus, two sources might contribute to luciferase expression: transcripts originating in the immunity region, and derepression of P_L caused by the act of transcription through the promoter. When T_IMM is defective, we observe the same higher level of luciferase expression whether or not the P_L promoter is functional (Figure 5B, Table S5), thus, the signal we observe does not come from the P_L promoter. These results indicate that even with CI repressor present, some fraction of the transcribing polymerases that initiate upstream of P_L pass through the repressor-DNA complex and proceed into the P_L operon.

We have characterized two genetic elements in the bacteriophage λ immunity region, the promoter P_LIT and the transcription terminator T_IMM, and have investigated their impact on λ development. The bacteriophage λ bistable switch has proven valuable for mathematical modeling of a relatively simple biological system, phage λ development (Arkin, Ross, & McAdams, 1998; Bakk, Metzler, & Sneppen, 2004). Our findings demonstrate additional levels of genetic regulation within the phage immunity region, which impinge on this switch and which can be readily incorporated into these mathematical models to enhance our understanding of the control of viral latency and induction.

Experimental Procedures

Bacterial strains, plasmids and single-strand DNA oligonucleotides:

All strains are Escherichia coli K-12 MG1655 or W3110 derivatives. Relevant genetic elements of strains are described in Table 1; construction details are Table S1. All genetic reporters used in this study are located at the lac operon; the original strains from which these reporters are made are described in the Supplemental section of Svenningsen et al. (2005). Red recombination functions were provided by one of several heat-curable plasmids (Datta, Costantino, & Court, 2006; Murphy & Campellone, 2003; Thomason, Sawitzke, Li, Costantino, & Court, 2014). Strains were made using a combination of recombineering and P1 transduction according to established procedures (Ellis, Yu, DiTizio, & Court, 2001; Sawitzke et al., 2007; Thomason, Costantino, & Court, 2007; Yu et al., 2000). All mutations were confirmed by DNA sequencing on both strands. Single-strand DNA oligonucleotides (ssDNA oligos) were procured from Integrated DNA Technologies as unsalted but otherwise unpurified.. Strain construction details and sequences of ssDNA primers are available upon request.

Table 1.

Escherichia coli K-12 Strains

Strain	Relevant Genotype	Reference/construction
A584	suIII trp::Tn10 TetR	F.W. Stahl
HME45	W3110 gal490 pgl Δ8 cI857Δ(cro-bioA)	(Lee et al., 2001)
NC397	HME45 lacI’<>kan Ter<>cat sacB<>lacZYA	(Svenningsen et al., 2005)
LT445	MG1655(λ)	L. Thomason collection
LT447	MG1655(λ cI857)	L. Thomason collection
LT732	MG1655 lacIo<>kan-Ter<>luc<>‘N pLoL rexB rexA cI857 pRoR cro’<>lacZYA+	derived from NC401 (Svenningsen et al., 2005)
LT1055	MG1655 lacIo<>kan-Ter<>luc<>‘N pLoL rexB rexA cI857 pRoR cro27 cII-lacZ <>lacZYA	derived from NC411 (Svenningsen et al., 2005)
LT1676	MG1655(λ PLIT+ rexA<>cat)	this work
LT1677	MG1655(λ PLIT+ ¾rexB<>cat)	this work
LT1771	MG1655 PLIT+ rexB-lacZ PSFI-luc	this work
LT1873	MG1655 lacIo<>kan-Ter<>luc<>‘N pLoL rexB rexA cat-sacB pRoR cro’<>lacZYA+	his work
LT1875	MG1655 lacIo<>kan-Ter<>luc<>‘N pLoL rexB rexA ΔcI pRoR cro’<>lacZYA+	this work
LT1961	LT447 PBAD-rexB hygroR	this work
LT1964	LT732 PLIT-10	this work
LT1989	LT1771 PLIT-10 rexB-lacZ	this work
LT1997	MG1655(λ PLIT-10)	this work
LT2003	MG1655(λ TIMM-)	this work
LT2056	LT732 ΔPL luc TIMM- ΔrexB	this work
LT2068	LT732 ΔPL luc TIMM- ΔrexB PLIT-10	this work
LT2078	LT732 ΔPL luc TIMM- ΔrexB PRM-10	this work
LT2079	LT732 ΔPL luc TIMM- ΔrexB PLIT-10 PRM-10	this work
LT2093	LT732 PL-10	this work
LT2094	GM3 PL-10	this work
LT2156	MG1655(λ PLIT-10 rexA<>cat)	this work
LT2157	MG1655(λ PLIT-10 ¾rexB<>cat)	this work
LT2162	MG1655(λ PLIT-10 bor<>kan)	this work
LT2163	MG1655(λ TIMM- KO bor<>kan)	this work
LT2228	MG1655(λ PLIT-10 cI857)	this work
LT2232	MG1655(λ PLIT-10 cI857 Sam7)	this work
LT2312	KM17 lexA<>amp	this work
LT2313	MG1655(λ cI857Sam7)	this work
LT2314	LT2228 PBAD-rexB hygroR	this work
KM12	LT1771 Δ(srlA-recA)::Tn10	this work
KM14	LT1771 lexA<>amp	this work
KM17	LT1771 PLIT LexABS+	this work
KM18	LT1771 PLIT LexABS-	this work
GM3	LT732 TIMM-	this work
GM15	MG1655 PLIT+ rexB TIMM+ luc	this work
GM32	MG1655 PLIT rexB(A->G) TIMM luc	this work
GM33	MG1655 PLIT rexB TIMM- luc	this work
GM37	GM15 rho15::bla	this work
GM38	GM33 rho15::bla	this work

Materials and Media:

Bacterial cultures were grown with L broth containing 10g tryptone, 5g yeast extract and 5g NaCl per liter, and on L agar (1.5%). Counter-selection against cat-sacB dual cassette was on L agar containing 6% sucrose and lacking NaCl (Li et al., 2013; Thomason et al., 2014) Phage stocks were maintained in TMG, containing 10mM Tris base, 10 mM MgSO₄ and 0.01% gelatin, pH7.4.

Promoter and transcription terminator mutations:

Point mutations in −10 regions of the P_LIT, P_L and P_RM promoters and the transcription terminator T_IMM were made using ssDNA oligo recombineering and a MAMA PCR screen (Cha, Zarbl, Keohavong, & Thilly, 1992). Neutral changes were introduced as necessary to enable detecion by PCR. The mutation for the P_LIT promoter is shown in Figure 2. The sequence of the P_L-10 mutation is CTTACA; the wild type −10 sequence is GATACT. The sequence of the P_RM—10 mutation is ATCTAG; the wild type −10 is TAGATT. Sequence walkers for all promoter mutations are in Figure S1. Transcription terminator T_IMM mutations are illustrated in Figure 3.

Graphical analysis of data:

Data obtained in experiments were graphed using GraphPad Prism. Note that GraphPad Prism does not show error bars if they are smaller than the size of the symbol used to plot the data, thus, for this reason, several graphs do not show error bars for some points, although at least three independent replicates have been done for every experiment and error bars were plotted for all graphs.

β-Galactosidase Assays:

Cultures were grown in L broth at 30–32°C overnight, diluted 1:200 the next morning, and grown with aeration in baffled flasks in a water bath at the appropriate temperature. When the A600 reached 0.4–0.6, 1ml cultures were rapidly transferred at the indicated times to an Eppendorf tube on wet ice, and β-Galactosidase levels determined (Miller, 1972). Each sample was assayed in duplicate, and the average of three or more independent experiments were determined.

Luciferase Assays:

Luciferase activity was measured using the Promega Luciferase Assay System (catalog no. E1500) according to the company’s directions. Cells were grown and treated as described for the β-galactosidase assays up to the point of addition of lysis buffer. Culture aliquots in L broth (0.1ml) were added directly to 0.4 ml of Cell Culture Lysis Reagent with 2.5mg/ml BSA and 1.25 mg/ml lysozyme. Cell lysate (25μl) was mixed with 0.1ml of luciferase substrate (Promega Luciferase Assay Reagent, catalog no. E151A), incubated for 2.0min and read in a BD Pharmingen Monolight 3010 single sample luminometer for 10s. Each sample was assayed in duplicate, and the average of three or more independent experiments were determined. The relative light unit (RLU) was normalized to the A600.

UV induction for bacteriophage λ lysogens:

Cell preparation: Lysogens and bacteria for plating phage were grown overnight in L broth. Lysogenic cultures were diluted 1/500 into 35ml L broth and grown in a 32°C shaking water bath until OD₆₀₀ was 0.15–0.2. The culture was centrifuged at ~8400rcf in a Sorvall RC5C Plus centrifuge, and the cell pellet was suspended in an equal volume of TMG, then centrifuged again and the cell pellet suspended in half the initial volume. An aliquot was removed and diluted and plated to determine the number of viable cells, which was ~1×10⁸ cells/ml. For titration of phage lysates, 150μl of an overnight culture of the TetR strain A584 was added to 5ml tryptone broth (TB) containing 10mM MgSO4 in a culture tube. Cells were grown for ~2.5hrs in a 37°C roller, then 5ml TMG was added. UV irradiation: The lysogenic suspension was decanted into a sterile 100×15mm petri dish, which was placed on a shaking platform under the UV lamp with the lid on. At t=0 the lid was removed, and the cell suspension was exposed to UV for 30sec. For LT445, LT1997 and LT2003 the suspension was sampled every 5sec. For the other genotypes the suspension was sampled at 30sec, at which time the cells had experienced a UV dose of ~15 J/m². 10μl of each sample of irradiated culture was added to 10ml L-broth in a 50ml baffled Erlenmeyer flask and incubated in a 37°C shaking water bath in the dark. The lysates were diluted appropriately and assayed for phage production after 2hr incubation, on the TetR indicator strain A584 using tryptone plates containing 10μg/ml tetracycline. Note that the tetracycline resistant A584 plating bacteria will grow on this medium while the induced strains are tetracycline sensitive and become inactive when plated.

Heat induction for λ cI857 lysogens.

Overnight λ lysogenic cultures were grown in L broth and diluted 230-fold into 35ml L broth in 125ml baffled Erlenmeyer flasks and propagated in a 32ºC shaking water bath. At OD₆₀₀ 0.2, the flasks were shifted to a 42ºC shaking water bath (t=0), then transferred to 37ºC after 15min. Time points were taken as indicated. After lysis 100μl CHCl₃ was added and incubation continued for an additional 10min. Lysates were diluted and assayed to determine the number of infectious phage particles for each genotype. This heat induction protocol was also followed for the Sam7 phage experiment with the addition of manual lysis at 55min as shown in Figure 7B.

Measuring ATP levels in heat induced lysogens.

The BacTiter-Glo™ Microbial Cell Viability Assay Kit (Promega, catalog no. G8231) was used to measure intracellular ATP levels with an ATP-dependent luciferase assay. The λ cI857 lysogenic cultures were grown at 32ºC to OD₆₀₀ 0.2 and heat induced as described above. Samples were assayed every 10min for the wild type lysogen and every 5min for the P_LIT-10 lysogen. After each OD₆₀₀ reading was taken, aliquots of the cultures were immediately diluted 10-fold into L broth, in order to obtain readings in the linear range, and 100μl of diluted cells was promptly mixed with 100μl BacTiter-Glo™ Reagent, incubated for 2.0min and read in a BD Pharmingen Monolight 3010 single sample luminometer for 10s. Each sample was assayed in duplicate, and the average of three independent experiments was determined. The ATP measurements were normalized to the OD₆₀₀ readings.

Two-hybrid analysis:

The Bacterial Adenylate Cyclase Two Hybrid System Kit (BACTH Kit) was obtained from Euromedex (www.euromedex.com). Recombineering was used to insert the rexA and rexB genes into the four two-hybrid vectors: the low-copy KanR pKT25 and pKNT25 plasmids with the T25 adenylate cyclase domain on the N- and C-terminal domains fused in frame to the gene of interest, respectively. They were also inserted into the high-copy AmpR pUT18 and pUT18C plasmids, with the T18 adenylate cyclase domain fused in frame to the C- and N-terminal domains of the gene of interest, respectively. For BACTH complementation assays, various pairs of T25 (KanR) and T18 (AmpR) plasmids were introduced into the cya mutant strain BTH101 by co-electroporation. The BTH101 derivatives containing these plasmids are listed in Table S7 of the Supporting Information. After outgrowth, 10μl drops of ten-fold serial dilutions of the transformed cells were spotted onto petri plates containing MacConkey Maltose solid medium with kanamycin (30μg/ml) and ampicillin (100μg/ml), or onto L agar containing the same drugs. Plates were incubated at 32°C overnight. Red colonies on MacConkey Maltose indicated maltose utilization, thus a positive interaction between the two proteins of interest that results in interaction of the two adenylate cyclase domains (Figure S3). β-galactosidase assays were used to determine efficiency of interaction among the various hybrid proteins (Figure 10, Table S7). For these assays, colonies picked from the L agar plates containing kanamycin and ampicillin were used to generate overnight cultures, grown in L broth containing the same antibiotics at the same concentrations. The day of the assay, cells were subcultured 200-fold into L broth lacking antibiotics and the protocol described above was followed. At least three independent cultures were measured for each plasmid pair.