A CI-Independent Form of Replicative Inhibition: Turn Off of Early Replication of Bacteriophage Lambda

Several earlier studies have described an unusual exclusion phenotype exhibited by cells with plasmids carrying a portion of the replication region of phage lambda. Cells exhibiting this inhibition phenotype (IP) prevent the plating of homo-immune and hybrid hetero-immune lambdoid phages. We have attempted to define aspects of IP, and show that it is directed to repλ phages. IP was observed in cells with plasmids containing a λ DNA fragment including oop, encoding a short OOP micro RNA, and part of the lambda origin of replication, oriλ, defined by iteron sequences ITN1-4 and an adjacent high AT-rich sequence. Transcription of the intact oop sequence from its promoter, pO is required for IP, as are iterons ITN3–4, but not the high AT-rich portion of oriλ. The results suggest that IP silencing is directed to theta mode replication initiation from an infecting repλ genome, or an induced repλ prophage. Phage mutations suppressing IP, i.e., Sip, map within, or adjacent to cro or in O, or both. Our results for plasmid based IP suggest the hypothesis that there is a natural mechanism for silencing early theta-mode replication initiation, i.e. the buildup of λ genomes with oop + oriλ+ sequence.


Introduction
Normal cellular immunity to l infection arises upon the lysogenic conversion of E. coli cells by a l prophage. The CI repressor protein encoded by the prophage binds to the o L and o R operator sites, each with three repressor binding sites, e.g., o R 3, o R 2, o R 1, within the imml gene cluster p L -o L -rexB-rexA-cI-p M -o R -p Rcro. CI protein within a l lysogenic cell blocks transcription of the phage genes situated downstream from the major leftward and rightward phage promoters p L and p R [1], both from the resident prophage, or when a homo-immune imml phage infects the cells. The variant lvir efficiently forms plaques on cells lysogenized by l because it carries point mutations v2 in o L , v1 in o R 2, and v3 in o R 1 [2]. Transcription from p R (Fig. 1A) is required for expression of genes cro-cII-O-P, respectively encoding a second repressor (Cro) that binds to o R ; an unstable stimulator (CII) of the establishment mode of cI transcription from promoter p E [3]; and the repl replication initiation cassette including genes O, P, and the origin (oril within O) site, which participate in oril-dependent bidirectional (theta mode) replication initiation. The gene oop, is transcribed from promoter p O [4] (opposite orientation from p R ), partially overlaps the terminal end of cII, and encodes a short selfterminating antisense RNA (OOP) opposing CII expression [5]. Part of oop and p O share a 33 bp region of high sequence homology within lambdoid phages (Fig. S1). The organizational similarity within the region encoding the cII-like-oop-''orf''-O-like-P-like genes for lambdoid phages is shown in Fig. S2.
The dual infection of a l lysogen with two phages, a homoimmune imml phage and a hybrid hetero-immune limm434 phage, each of which share an identical repl replication initiation cassette, revealed that the imm434 phage predominated by 20 + -fold over the imml phage in the cell burst [6]. The impaired replication of the homo-immune imml phage, described as replicative inhibition, which we consider herein ''CI-dependent'' was explained by the assumption that CI repressor molecules made by the l prophage in the co-infected lysogenic cells prevented replication of the homo-immune phage, even when the l replication initiation proteins (gpO and gpP) were provided in trans by the heteroimmune phage. The observations that CI-dependent replicative inhibition was suppressed by mutations in o R causing p R to become insensitive to repression, or by base changes creating new promoter sites downstream from p R , as exemplified by c17 and four ri c (replication inhibition constitutive) mutations [7], provided support for an argument that transcription from p R (transcriptional activation) was required in cis for theta-mode replication initiation, and that replicative inhibition was explained by CI repressor in the lysogen preventing transcriptional activation of replication initiation from the co-infecting imml repl phage.
Plasmids termed ldv were derived from phage lvir [8,9]. They encode the imml and repl regions and are capable of autonomous replication. Early studies with cells transformed with ldv suggested that the cells acquired an unusual immunity or exclusion phenotype [8,10] and inhibited plating by homo-immune phages, including lvir, and hetero-immune hybrid phages as limm434. Some other hetero-immune phages (e.g., limm21 and limm80) that were presumably repl were able to escape the inhibition, i.e., could plate efficiently on cells transformed with ldv [8,10]. The ability of cells with ldv plasmids to inhibit lvir development was rationalized by the suggestion that cells with this plasmid make more CI repressor than would a cell with a single l prophage, and the higher levels of repressor would eventually bind the altered lvir operators [8]. However, CI levels were not actually measured. No explanation was provided for the inhibition of limm434 development. When RNA transcription levels from cells with ldv1 plasmid were measured, it was found that little [10] or no [11] cI transcription was detected, showing that the inhibition of homo-immune infecting phage development by ldv plasmid was not due to CI repressor activity. It was proposed [10] that the ldv-mediated inhibition of infecting repl phage development represents a competition for bacterial protein(s) between the plasmid and an infecting phage, and that the site for the competition was different in the limm21 and limm80 phages that escaped the inhibition.
Independently, Rao and Rogers [12] demonstrated that cells containing a pBR322/l hybrid plasmid that included the imml and repl regions exhibited an inhibition phenotype (referred to herein as ''IP''), that prevented the plating of lvir and limm434 infecting phage, but allowed limm21 to plate at high EOP. They reported isolating mutants of lvir and limm434 which formed plaques at high EOP on cells with the plasmid, but the causative mutations were not further identified. Another inhibition phenotype, termed nonimmune exclusion (NIE) [13], was specific for imml and imm434 phages that were repl. NIE was exhibited by a variety of engineered cells with thermally induced (CI-inactivated) cryptic cI [Ts] prophage deleted Figure 1. Replication-targeted inhibition of repl phage plating. A. Plasmid cloned l DNA fragments used to map the sequence requirement(s) for an inhibition phenotype (IP). B. Genomic region spanning five contiguous and partially homologous genes of phages l and P22 (see Fig. S2). Phage l is naturally missing the orf48 gene between oop and O that is present between oop and 18 in P22 [37,51]. C. Assay for EOP, defined as phage titer on strain 594 (with one of the indicated plasmids) / titer on 594 cells, where plating on 594 = EOP of 1.0. All of the plasmids shown were derived from pBR322. The oop + oril + plasmid used was p27. The DNA substitution of the ''ice'' [16] sequence of l to make plasmid Dice oop + oril + ( = p50) is shown in Fig. S3A. Numbers in brackets represent standard error values. doi:10.1371/journal.pone.0036498.g001 for attL through kil, all genes rightward of P, and had acquired mutations inactivating P [14]. Seven independent l se (suppress exclusion) mutations of l wt (wild type) were isolated from NIE phenotype cells having a cro27 mutation in the cryptic prophage. The se defects were point mutations within o R 2 (se100a, identical to mutation v1; and se101b) and within o R 1 (five mutations represented by se109b, identical to mutation vC1, and at the same site as vs387) [13]. All seven l se isolates exhibited a CI-defective phenotype, complemented for cII and cIII, and were about 10-fold less sensitive to replicative inhibition than l wt or l cI - [13].
We have attempted to understand further the inhibition phenotype(s), IP, by constructing plasmids with portions of repl. By removing imml from plasmids, the conflicting plating data for lvir was eliminated. We have shown that CI-independent, plasmid-dependent IP requires cis acting oril iteron (ITN) sequences [2,15] and oop transcription, and is directed to repl phages. We suggest that the target of IP is early (theta-mode) replication initiation. Phage mutations suppressing IP, i.e., Sip, map within, or adjacent to cro or in O, or both.

Plasmid-mediated Inhibition Phenotype (IP)
The bacterial strains, phage, plasmids and primers for modifying plasmids are described in Tables 1, 2, 3. Plasmid pCH1, theoretically identical to the IP plasmid described by Rao and Rogers [12], and deletion derivatives as p25 and others ( Table 2, Fig. 1A) were made to determine which l sequences were responsible for IP. Plasmids pCH1 and p25 inhibited the plating of lvir, but versions deleting imml (including the p R promoter) did not (data not shown). The IP toward repl phage was seen with plasmids as p26 (data not shown), p27, (rop + , oop + , oril + ), p27R (oop + , oril + ), and p50 (Dice oop + oril + ) in Fig. 1C. p50 was deleted for the proposed replication inceptor site ice [16] ( Fig. S3A,C,D), including all l DNA from 31 bp leftward / downstream of the oop sequence (Fig. S3A). Plasmids that were oop + Doril, or oril + but deleted for the t O -oop-p O sequence expressing the self-terminating 77 nt OOP RNA [17] (Fig. S3B), were defective in IP. In contrast, phages where repl was replaced by repP22 as in lcI857 (18,12)P22 escape IP ( Fig. 1C; gene replacements are shown in Fig.'s 1B, S2, S3C). These results strongly suggest that IP is directed to repl phages that employ genes O and P to initiate replication from oril.
The influence of IP on the temporal events for cell lysis and phage burst following thermal induction of a prophage was examined (Fig. 2). None of the four plasmids, p27R, p27Rp O -(oop + p O 2 oril + ), p28 (oril + ) and p29 (t O -oop-p O + ) ( Fig. 2A) prevented phage-dependent cell lysis by an induced repP22 prophage (Fig. 1B). In contrast, vegetative development of the repl prophage was markedly inhibited (as was cell lysis) in cells with the oop + oril + The l prophage genes int-xis-exo-bet-gam-kil in strain Y836 were substituted with bio275 [13]. The strain carries the chromosomal deletion D431 [33] that removes genes rightward from ninB in prophage through moaA in host, including prophage genes orf146 (orf) -Jb2 (i.e., all the late genes required for cell lysis and phage morphogenesis). A map of the cryptic lambda prophage in strain Y836 is drawn in Fig. 4A. b Lysogenic strains show the prophage within the cell by ''( )'' bordering the prophage. c The NinR region deleted by Dnin5 removes l bases 40,503-43,307, i.e., ren-ninA -ninI (including orf -ninC and rap-ninH ); the NinL region substituted by bio275 replaces genes int-xis-hin-exo-bet-gam-kil, representing l bases 27,731-,33,303 [18]. doi:10.1371/journal.pone.0036498.t001 plasmid (Fig. 2B); but, when the plasmid was altered by changing the -10 region for p O, or removing the t O 2oop-p O , or oril regions, no inhibition of repl prophage development was observed, in agreement with the plating results in Fig. 1C. We examined if a cloned intact O gene, repressed at 30uC, but expressed at 39u and 42u, exhibited IP to repl phage plating ( Table 4). The result was similar to that for the D (t O -oop-p O ) oril + plasmid carrying a fragment of O (Fig. 1C), i.e., no significant IP. The plasmid version containing intact O/oril, with cI from imml, reduced the plaque diameter of all four assayed repl phages but the version with a hybrid imml-imm434 cI gene did not. lvir was inhibited for plating at 30u in cells with multiple copies of the O/ oril plasmid version with cI from imml, while limm434cI was not inhibited, suggesting lvir plating remains sensitive to high CI repressor concentration (we made a similar observation with another cI + plasmid [18]).

Dissecting IP sequence requirement(s)
The spacing interval between the t O -oop-p O sequence and oril in p50 was modified by deletion or insertion (Fig. S3D) to learn if the spatial orientation between these two regions was important for IP. All the modified versions of p50, i.e., p51, p51kan, and p52, retained IP ( Fig. S3C-D). We asked if transcription of oop from p O participated in IP by inactivating the -10 region of p O , replacing the sequence ATTAT with GCGCG in p27R to stringently assess a requirement for oop expression from a high copy oril plasmid. The resulting plasmid, p27Rp O 2 (Fig. 3C), no longer expressed oop, as determined by the OOP antisense phenotype/cII inactivation assay (see Materials and Methods) and was defective for IP ( Fig. 3D), suggesting that transcription from p O is essential for IP.
To distinguish whether the transcription of the downstream oop sequence, or just transcription initiation from the p O promoter was required for IP, the coding sequence of oop was modified in plasmid p27R-R45OOP (Fig. 3C). Nucleotides 2-46 of oop were replaced with a randomly chosen sequence, edited to remove internal secondary structure formation. For maintaining the selfterminating stem-loop structure of t O , the distal 31 nucleotides of oop were retained, as was the first base pair of the oop sequence, corresponding to 59 pppG of OOP RNA. p27R-R45OOP was unable to serve as an antisense RNA to inactivate cII and it was defective for IP ( Fig. 3D, columns 1-3). The results with plasmids p27Rp O 2 and p27R-R45OOP suggest that transcription of the intact oop sequence is required for IP, rather than just transcription initiation from p O .

IP silences l replication initiation
Lambda replicates in two stages. The early or bidirectional (theta) mode from oril starts within two minutes following thermal de-repression of a lcI[Ts]857 prophage [28]. The late or rolling circle (sigma) replication mode forms linear DNA concatemers, the Thermally induced repl prophage. The results represent the averages for 2 independent assays. Plasmids within lysogenic cells: square, Po + oop + ori + (results shown for p27R, but identical results were observed for p27); triangle, Po2 oop + ori + ; inverted triangle, D (to-oop-Po) oril + (ITN-AT + ); diamond, cII-oop-Po + Doril; circle, none (no plasmid). The standard deviation is shown for all of the data points, but is too small for visualization in some data intervals. doi:10.1371/journal.pone.0036498.g002   [29][30][31][32]. Skalka et al. [31] stated that replication via the ''early mode occurs only once or twice, after which rolling circle (late) replication predominates.'' They suggested that a direct, internal control gene for the turn-off of early replication either ''does not exist'', or ''must not be expressed in the absence of replication'' because early replication products accumulate (after infection or induction) when concatemer formation is destabilized in l gam mutants, or under fec 2 conditions (involving both l red and host recA mutations). The chromosome in strain Y836 (Table 1; Fig. 4A) has an engineered cryptic l prophage deleted for recombination genes int-xis-exo-bet-gam-kil involved in general and site specific recombination [13] and for genes orf146 ( = orf) -Jb2, including genes required for cell lysis and phage morphogenesis [33], but it encodes the imml and repl regions. Transcription of O-P from p R is prevented at 30u by the cI[Ts]857 encoded temperature sensitive repressor. Inactivating the CI repressor, by shifting cells grown at 30u to 42u, triggers oril-dependent bi-directional replication initiation from the trapped l fragment. Initiated replication forks escape leftward and rightward beyond the l fragment and into the E. coli chromosome. This event is lethal to the cell and was termed Replicative-Killing [7], i.e., RK + phenotype [18,34]. Survivor cells that escape Replicative-Killing (RK 2 mutants) arise within the RK + starting cells and were found to possess mutations that prevented replication initiation from oril [13,14,[33][34][35]. Transducing a dnaB mutation (GrpD55) that prevents l replication initiation (but not E. coli DNA synthesis) into the RK + Y836 cells can fully suppress Replicative-Killing without interfering with gene expression from the induced l fragment [18]. We examined whether plasmids exhibiting the IP phenotype could suppress Replicative-Killing (Fig. 3D, rightward columns e-g). The viability of RK + Y836 cells shifted from 30u to 42u was ,0.00001. Similar results were seen when Y836 was transformed with p27R-R45OOP, p27RDITN1-4, or to a lesser extent with p27Rp O 2 , indicating that these three plasmids do not suppress the RK + phenotype. Cells transformed with plasmids p27R and p27RDAT suppressed Replicative-Killing at 42u, suggesting that they interfered with (silenced) theta-mode replication initiation from the chromosomal l fragment.
We examined if the IP-plasmids could block replication initiation from a thermally induced cI[Ts]857 l fragment within the Y836 chromosome. Replication initiation arising from the oril region of the induced cryptic prophage was assessed by probing for a 1774 bp NdeI fragment ( Fig. 4A-C) following Nde I digestion of the Y836 cell chromosome. The probe to the NdeI fragment overlapped with each of the l fragments in the plasmids introduced into Y836, permitting an internal measure of plasmid copy increase. Theta-mode replication initiation increased by about 3-fold from oril when Y836 cells without a plasmid were shifted from 30u to 42u (Fig. 4C). The oop + oril + plasmid p27R fully inhibited theta mode replication initiation, in full agreement with the data showing that this plasmid blocked Replicative-killing (Fig. 3C). Cells with p27RDITN1-4, with a deletion of the four iterons (but not the AT-rich region) was not inhibitory; whereas, the converse plasmid p27RDAT, modified to remove the high ATrich sequence but containing ITN1-4, was fully inhibitory to theta-mode replication initiation from the prophage oril site. The intensity of the replication increase was not as robust as previously seen ( Fig. 2 in [18]) where the probe was larger and could detect two l prophage restriction fragments (i.e., 3675 bp oril band, and a 4250 bp band showing escape replication), possibly because of the high level of competition for the probe by the l DNA within the plasmids. Two of the 1774 bp bands at 42uC for cells where oril replication initiation was inhibited decreased slightly compared to their 30uC counterparts. This may represent some level of DNA extraction variation, or it could be real and represent fragment destruction resulting from abortive oril replication initiation from the prophage in these strains.

Escape from IP
We previously showed [18] that marker rescue for imml recombinants was below the level of detection for Y836 dnaB-GrpD55 host cells infected with imm434 phage deleted for l genome regions NinL (int-red-gam recombination functions) and NinR (ren-ninA-ninI, including Orf and Rap) ( Table 1 in [18]). The same result was seen for Y836 recA host cells infected with imm434 versions of NinR + DNinL and DNinR DNinL phages ( Table 2, lines 2-3 in [18]). The GrpD55 locus was suggested linked to dnaB [36], and Horbay [37] subsequently determined by sequence analysis that it represented two missense mutations within dnaB.
oop sequence, which overlaps cII is transcribed leftward from pO. C. Illustrated mutations within the l DNA region in plasmids numbered 1-6 ( Table 2). Plasmid p27R (shown as #1) carries with WT sequence from which other plasmids were derived. In each plasmid the rop gene was deleted to provide higher plasmid copy number per cell to test the stringency of introduced mutations. The ''X'' in #2 inactivates the p O promoter for oop gene; the filled rectangle in #3 (mutation oopR45) substitutes random 45 bp for 45 bp within oop providing a 77nt RNA without internal secondary structure (Fig. S3B); and the gaps in #'s 4-6 are deletions (Table 2). D. Columns (left 'a,' to right 'g'): Lane 'a' shows the plasmid number and common name (Table 2), with plasmid genotype indicated in part C. Lanes 'b' and 'c': EOP of repP22 and repl phages on 594 host cells with indicated plasmid; 'd' summary of the inhibitory effect of a plasmid in 594 cells to the plating of repP22 or repl phages, where NONE is essentially no inhibition of plating and FULL indicates that plaque formation was prevented by the presence of the plasmid. Lanes ''e'' through ''g'' indicate the results of a separate experiment to determine if plasmids #1-5, transformed into strain Y836, can suppress Replicative Killing, which occurs upon prophage induction when the Y836 cells are raised above 38uC. Prophage induction leads to replication initiation from oril within the chromosome, as shown in part A, which is very lethal to cell. Lane 'e' shows the level of cell survival upon shifting the cells to 42uC. The survival of Y836 cells that were diluted and spread on plates incubated at 42uC requires plasmid suppression / interference of replication initiation and cell killing upon de-repression of the prophage in Y836 cells. Two single colonies of each transformant of Y836 cells were inoculated into 20 ml TB +50 ug/ml ampicillin and grown overnight at 30uC. The following day the cultures were subcultured (2.5 ml overnight culture +17.5 ml TB and grown to mid-log (,0.35 A 575nm ), whereupon, cells were diluted into buffer and spread on TB plates that were incubated for 24 hr at 30uC, and onto TB and TBamp50 plates that were incubated at 42uC for 24 hr. Survival to Replicative Killing was assessed by dividing the average cfu/ml at 42uC incubation (the cell titers on both TB and TBamp50 plates were equivalent) by average titer for cell dilutions incubated at 30uC. Lane 'f' is a summary of the plasmid's effect on Replicative Killing of induced Y836 cells, where NONE indicates the cells were killed upon induction, and FULL reflects high cell survival as determined by colony formation at 42u. The values in parentheses show standard error for at least two independent determinations. Lane ''g'' shows the level of each plasmid present in the cells at 30uC (noninduced), immediately prior to shifting cells to 42uC (see legend, Fig. 4). The duplicate cultures processed at time 0 were extracted for DNA using Qiagen DNAeasy Kit, estimating 1.0610 8 cells per 0.1 A 575nm and calculating the amount of cell culture needed for 2.0610 9 cells per DNA preparation. All DNA samples were prepared in duplicate. The gel purified bands for the plasmid DNA present in the 0 time cultures was assessed by hybridization as described in Fig. 4. doi:10.1371/journal.pone.0036498.g003 The dnaB-GrpD55 mutation confers a temperature sensitive phenotype for l plating but does not prevent E. coli replication, cell growth [36]. The EOP of l on strain W3350 dnaB-GrpD55 was significantly reduced compared to W3350 (EOP set = 1.0). The respective EOP's at 30u, 40u or 42u on the dnaB-GrpD55 host were 0.08, 0.01, ,0.0001 (for lcI857); 0.2, 0.002, ,0.0001 (for limm434cI); and 0.4, 0.04, ,0.0001(for limm434DNinR), showing increasing temperature sensitivity for l replication, while the E. coli dnaB-GrpD55 host was able to form effective cell lawns at the elevated temperatures. We define ''free-loader'' coefficient, as a measure of phage progeny for infections at multiplicity of infection (MOI) 5, per the phage progeny from infections at MOI 0.01 (see Discussion). The availability of l recombination functions within an infected cell can influence the free-loader coefficient. W3350 dnaB-GrpD55 recA + cells were infected at MOI's of 5 or 0.01 with l deleted for the NinL, NinR, or both recombination regions, then incubated for 90 minutes at 42u and plated for phage burst. Infections with phages limm434NinL + NinR + , limm434NinR + D-NinL, limm434DNinR NinL + , and limm434 DNinLDNinR yielded respective coefficients of 1065 (+/218 std. error), 502 (+/2 31), 136 (+/210), and 111 (+/2 27), suggesting that the l NinR and NinL recombination functions can influence phage burst from multiply infected cells where the infecting phages are blocked for theta-mode oril-dependent replication initiation by the dnaB-GrpD55 mutation. This result supports our prior suggestion [18] that ori-specific theta-mode replication initiation, dependent upon P-DnaB interaction, can be bypassed in multiply infected cells, i.e., phage replication can likely be driven by intermediates derived via homologous recombination between co-infecting phage genomes.
The results from Fig's 1, 2, 3 and 4 suggest that IP serves to block / silence replication initiation from oril. We examined whether IP could be bypassed, comparing the bursts from singly infected (low MOI, 0.01), or multiply infected (high MOI, 5) cells ( Table 5). Infections of wild type host strains W3350 and 594 at either MOI's of 5 or 0.01 with repl or repP22 phages produced essentially equivalent bursts. A similar result was seen for repP22 phage infections of W3350 dnaB-GrpD55 cells at either MOI 5 or . C. Assay for replication initiation from oril after shifting Y836 culture cells from 30u to 42u to induce transcription and oril replication from the cryptic prophage. Cultures were grown to mid-log and aliquots were removed at time 0 as described in legend, Fig. 3. Thereupon, cultures were transiently swirled in a 60uC water bath and transferred to a 42uC shaking bath for one hour and aliquots were removed. Cell concentration of the 42uC aliquots was based upon the calculations for 30uC 0-time cultures, and DNA was prepared using Qiagen DNAeasy Kit from 2.0610 9 cells. The concentration of extracted DNA was determined by spectrophotometer (A 260nm x DNA dilution X 50 ng/ml). The Y836 cellular DNA (2.5 ug of ethanol precipitated and resuspended DNA) was digested 2 hrs with NdeI and digests were run on horizontal 0.7% agarose gel, followed by Southern transfer of DNA bands. The Southern blot bands for the 1774 bp chromosomal prophage fragments were each scanned 3X using GE Healthcare software program ImageQuant version 5.2 and the region under the peaks was integrated and averaged. The numbers below the bands compare the relative levels of 1774 bp fragment obtained for induced / noninduced sample pairs. Refer to Hayes et al. [18] for detailed hybridization methodology, and for comparing the effect of a cI + repressor expressed from a plasmid on prophage induction, the influence of host recombination defects on replication initiation from oril from the prophage in Y836 cells, and the inhibition of replication initiation from oril by host mutations. doi:10.1371/journal.pone.0036498.g004 0.1. There was essentially no burst (background level) when the repl phage infected W3350 dnaB-GrpD55 cells at an MOI of 0.01; however, the phage burst was equivalent to that on the W3350 cells when the W3350 dnaB-GrpD55 cells were multiply infected at an MOI of 5. Thus, while the altered DnaB protein [GrpD55 allele] interferes with the P-DnaB interaction required for thetamode l replication initiation, it can still apparently drive l or E. coli DNA synthesis that is independent of P. Placing multiple copies of a recombination proficient l genome within a cell appears to bypass the P-DnaB interaction at oril required for the theta-mode of l replication initiation. Similarly, 594 cells with plasmid p27R (oop + oril + ) prevented phage burst from cells infected at MOI 0.01. But when these same cells were infected at MOI 5, IP was suppressed (bypassed). 594 cells with p27Rp O 2 , which is defective for IP, yielded an essentially similar repl phage burst at MOI 0.01 as when 594 cells without the plasmid were infected. These results suggest that IP serves to silence / inhibit theta-mode oril replication initiation and that multiple copies of recombinationproficient l genomes can, at some level, bypass this essential requirement for replication initiation from a single prophage or from one infecting l genome.

Suppression of Inhibition Phenotype (Sip) by l mutants and hybrids
We looked for a target of IP by i) characterizing 10 independent (Sip) mutants of lcI857 (Fig. 5); and ii) by screening for IP-escape, testing l mutants and hybrid phages (Fig.'s S4, S5). We first asked if insertion by homologous recombination (of the Amp R oop + oril + plasmid into the infecting phage) was responsible for Sip ( Fig. S6 and Supplemental Methods S1), and eliminated this possibility. The cI -P regions were sequenced for 10 independent Sip phage isolates, and for lcI857cro27 with a null mutation in cro, Alternatively, several of the Sip mutants conferred missense mutations in an 81 codon open reading frame, PreX; these included five Sip mutations (of which Sip6 eliminated the PreX start codon); plus the ''se'' mutations (described above) introduce missense changes into PreX (Fig. S7). PreX can only be expressed via high level establishment mode p E -preX-cI-rexA-rexB mRNA synthesis (i.e., 20-100X level of pM-cI transcription [28,39,40]), requiring CII activation at p E [3]. The p E -cI transcript is antisense to cro, and the possible PreX reading frame from it would overlap 13 codons at the N-terminal end of cI, all of oR/pR region, and 35 codons of cro, and would be expressed from the same reading frame as cro, but the opposite coding strand (Fig. S7).
Since six of the lcI857-derived Sip mutants produced five missense changes in cro (two independent Sip mutations, 8 and 10, each changed base pair 38183 in cro), we examined if any Sip mutants exhibited the lcI857cro27 plating phenotype. Phage lcI857cro27 has the interesting property of forming plaques at 37-39u, but not at 30u or 42u [38,[40][41][42], and of exhibiting a phenotype within infected cells termed Cro lethality (See [40] for a discussion of Cro lethality concept relative to rexA-rexB expression, translational frameshift sites within [43], and possible effect upon [14] high levels of p E -preX-cI-rexA-rexB expression (Fig. S7A,B) from an induced cro-defective l lysogen or infecting phage.) Our isolate of lcI857cro27 carried a single G-A transition (Arg to Gln) at base 38153 in cro (Fig. 5), nullifying cro activity. Only the Sip7 phage shared a nearly similar plating phenotype with lcI857cro27 by forming faint plaques at an EOP of ,10 23 at 30u, tiny-faint plaques at EOP 0.3 at 42u, and 1 mm clear plaques at 37 and 39u. Sip phages 1-6 and 8-10 formed 0.5-1.0 mm turbid plaques on 594 host cells at 30u, and about 1 mm clear plaques at 37u. Only the Sip 4 and 8 phages plated with slightly reduced EOP, i.e., by 3 or 13-fold, at 30u compared to 37u. Alternatively, we asked if lcI857cro27 can escape IP, i.e., if it shares properties with the lcI857Sip phages, and found that the cro27 allele did not confer a Sip phenotype (Table S1). Thus, simply inactivating Cro does not directly confer a Sip phenotype, and so the Sip mutations must have another effect.
The inability of the repl phage lcI857 to escape IP was not modulated by the CI repressor, reflected by equally IP-sensitive repl phages lwt (cI + ), and phenotypically CI 2 (lysogenizationdefective) phages: lcI72 (cI -), and by phages with CI-defective phenotype that escape replicative inhibition, i.e., loR/pR point mutations (lse mutants: 100a, 101b, and 109b (Table 1, Fig. S7C), and lcI90c17 (Table 1), where pR-independent transcription [44,45] arises via the c17 insertion downstream from pR). The repl phages lvir, limm21cI and limm434cI partially escaped IP, plating with EOP's of 0.1 or higher (Fig. S4A), but their plaque sizes were reduced. The sequence of limm434cI was identical to l throughout the cII-O interval (Fig. S5). lvir is mutated in both o R 2 and o R 1 at bases 37979 and 38007 [2,34], respectively, although, it is unclear what other mutations it possesses. The limm21 hybrid had base alterations within the cII-oop overlap (Fig. S5) and a silent TGC to TGT codon change at 39,033 (not shown), one base left of the ITN1 sequence in O.
Plaque size is a qualitative measure of phage development or burst, and we previously found that impeding l replication significantly reduced normal plaque size [18]. Thorough examination revealed that the plaques formed by limm434cI on 594[oop + oril + ] cells were barely visible, i.e., 5% of their normal diameter on 594 host cells (Fig.S4C) and limm21cI plaques were 35% their normal diameter. Plaques formed on 594[oril + ] cells by the repl phages (Fig. S4C) were reduced in plaque diameter by about half, in agreement with the observations that oril + plasmids partially interfere with phage maturation. To help ascertain why the repl phages limm434cI, and to a greater extent limm21cI, partially escaped IP, their oop-rep regions were sequenced (Fig. S5). While phage 434 has three base changes within the oop sequence, the limm434cI hybrid sequence was equivalent to l. The limm21cI hybrid shared the same sequence as phage 21, with an expected altered sequence within cII left of oop, and differences within the oop / cII overlap region (Fig.'s S1, S5). The l/P22 hybrid, i.e., lcI[Ts]857 (18,12)P22 that was insensitive to IP, carried the l version of cII, yet differed: by one base (37673) within oop, by one base (36689) just right of the common -10 sequence (ATTAGG) for the oop promoter p O , and completely diverged rightward from the l sequence at base -19 (38694) within p O , so that the -35 region's for the p O promoters for l and for l/ P22 hybrid were distinct (Fig.S5) as were downstream l genes O -P [2] and P22 genes orf48-18-12 [46] (Fig. S2).
All of the repl phages formed plaques with ,120% larger diameters on 594[oop + ] vs. 594 cells (Fig. S4C), suggesting that OOP RNA can stimulate repl lytic growth. The C-terminal 55 nt including the stop codon for gene cII overlap the 39-end of oop (Fig. S1). The last 17 amino acids of cII are not required for CII activity, but this region is necessary for CII regulation by OOP [5]. The infection of cII + -l phages into cells with plasmids expressing OOP micro RNA, which is antisense to cII [47] (Fig. 2), creates a cII-defective phenotype [48] resulting in clear plaques at 30u even for the hybrid lcI857(18,12)P22. Even our cI + version of limm21 gave turbid plaques on 594, but clear on 594[oop + ] host cells, suggesting that the five base changes within the oop / cII overlap region do not prevent OOP RNA (made from oop + plasmid) from serving as an antisense RNA to cII expression from imm21. Clearly, infecting cII + phages into cells expressing OOP RNA creates a phenotypic cII-defective condition, characterized by no p M -preX-cI-rexA-rexB transcription, no cro antisense RNA, and lytic phage growth. Thus, we did not consider it relevant to evaluate independent missense cIIphages, all of which map left of the cII/oop overlap [3]. In hundreds of cro + cII + prophage induction experiments, for example [4,28,39,40,49], no l-strand transcription attributable to p E was ever detected (Hayes lab results). This result, coupled which with our current understanding of the role of OOP as an antisense regulator of cII expression, suggests that the synthesis of OOP RNA under the conditions described herein will prevent p E transcription from infecting phage or induced prophage. But, an OOP block to p E transcription is insufficient on its own to explain CI-independent IP, i.e., oop + Doril plasmids were defective in IP.
We examined the IP-sensitivity of a phage deleted for cII-oop. The interval between AUG for cII and second codon for O in phage lcI + DcII ( = Doop) [50] was deleted (i.e., l bp 38363-38688; we confirmed by sequencing two isolates). The deletion fused the retained -35 region of the oop promoter, p O (leftward from bp -14 at 38689), with the sequence left of the second codon for cII (bp 38362), changing the -10 region for p O from ATTATG to CATATG, which might still support p O -dependent leftward transcription. The lcI + DcII phage partially escaped IP, forming pinprick-ghost plaques (impractical to quantitate/measure) on 594[oop + -oril + ], considerably smaller than those of limm434cI on the same host (Fig. S4C). The lcI + DcII phage was much more sensitive to copies of oril and formed very much smaller plaques than limm434cI or lcI857 phages on 594[oril + ] and 594[oop p O 2oril + ] cell lawns; yet it was capable of forming large clear plaques at EOP of 1 on 594 and 594[oop + ] cells. Further analysis is needed to explain the paradox that repl phages retaining the cII-oop region are sensitive to IP (requiring OOP and oril) yet their development is not curtailed by the presence of competing oril plasmids; whereas, deleting cII-oop has the opposite effect.

Replicative inhibition
We previously showed that the hybrid phage lcI857(18,12)P22, with the repl region swapped by repP22, was extremely sensitive to CI-dependent replicative inhibition, and by comparison, lcI72, the l se mutants, and lcI90c17 were respectively 4.6, 27-76, and 173 fold less sensitive [13]. This result illustrates that CIdependent replicative inhibition does not directly target the rep region, but rather, transcriptional activation of rep. In contrast, the repP22 phage escaped CI-independent replicative inhibition; whereas, the repl phages as lcI72, the l se mutants, and lcI90 c17 were fully sensitive. Therefore, we would assert that the CIdependent (blocking transcriptional activation of the rep region) and the CI-independent (IP directed theta mode replication silencing) forms of replicative inhibition are completely distinct, and that their mechanisms are likely different, even if they share the same end result.

Requirement for IP
We have provided additional understanding of the observation, termed here IP (Inhibition Phenotype), whereby host cells with plasmids containing the oop-oril region of the lambda genome inhibited phage plating. This region includes several cis-acting target sites, for example, the iteron sequences, ITN1-4, bound by O protein and sites for promoter, p O , and terminator, t O , for the 77nt OOP micro RNA (Fig's S1,2,3B,5) [51]. In summary: i) Plasmids containing the l t O -oop-p O through oril DNA sequence inhibited the development of repl infecting, or an induced lcI857 prophage, and neither the oop nor oril regions, separately, could account for IP. ii) IP was independent of the activity of l repressors CI and Cro, iii) A l/P22 hybrid with repP22 was insensitive to plasmids containing the t O -oop-p O l and oril DNA sequences, suggesting that IP is directed to a repl function. iv) Sequence analysis revealed that the l/P22 hybrid contained imml, an essentially intact (one base change) oop sequence, a hybrid p O promoter with a l -10 region and P22 -35 region, and the substitution of l genes O-P with P22 genes orf48-18-12 [37,51]. v) OOP RNA synthesis from the oop + plasmids channeled both the l/P22 and limm21 phages into a lytic mode to form clear plaques, suggesting the level of OOP RNA made was sufficient to serve as an antisense regulator of cII expression from the p R transcript(s). [5,47]. vi) A dissection of the contributions to IP revealed that an oop + plasmid deleted for the AT rich region of oril was fully functional for IP, oop + plasmids deleted for ITN1-4 or ITN3-4 were defective for IP, and oril + -containing plasmids substituted for 45bp within oop, or inactivating the p O promoter for oop transcription, were defective for IP.

Phage escape from IP
In summary: 1) Two types of full escape from oop + oril + plasmiddependent-IP were observed: i) substitution of O-P in l by orf48-18212 in the l/P22 hybrid (Fig. S5) enabled the hybrid to escape IP, even though its cII expression was inactivated by OOP RNA; and ii) Sip mutations within or near cro or in O suppressed IP. 2) Some repl phage partially escaped oop + oril + plasmid-dependent IP, but phage development was retarded (as evidenced by reduced EOP and plaque size). 3) Phages that could escape CI-dependent replicative inhibition were unable to suppress IP. This result refutes a hypothesis that natural or mutational events that increase transcription from p R , e.g., by limiting Cro or CI binding to o R , or introducing downstream promoters, will augment transcriptional activation of oril, and in turn promote theta-mode-oril-dependent replication initiation, and suppress IP. Another explanation is needed. Anderl and Klein [52] suggested that if the ratio of DNA:O protein is increased, theta-mode replication initiation will be inhibited due to titration of O protein, which suggests that plasmid-borne oril iteron sites could act as competitor origins, sequestering the O protein made by infecting repl phages. The ''handcuffing'' analogy for dimer formation [25] between O proteins binding to the iteron sequences in several oril sites could serve as a model for blocking the formation / completion / processing of a preprimosomal complex. The minimum molar ratio [53] of O protein:oril (termed O-some [54] complex) that was required for strand unwinding was 20:1. When additional oril regions are present, or if multiple interacting O-oril complexes are formed, it is unlikely that this molar ratio will be achieved. Our results suggest that handcuffing cannot account for IP, even if multiple oril targets bind excess O protein. Cells with multiple copies of two plasmids lacking oop sequence, but encoding an intact gene O/oril, did not reduce EOP, i.e., exhibit IP, whether or not O was expressed.

Theta-mode replication silencing by IP
The loading of DnaB onto ssDNA, formed by strand separation within the high-AT-rich region of oril, was suggested to mark the end of the initiation phase of l theta mode DNA replication [55]. Previously, we confirmed that theta-mode oril-dependent prophage replication initiation, which requires P interaction with, and loading of, DnaB, was inhibited if the host carried the dnaB-GrpD55 mutation, yet there was no obvious influence of this allele on E. coli DNA propagation [18]. Herein, we observed that both theta-mode replication from oril, and its manifestation, i.e., the Replicative Killing of induced cells (dependent upon triggering theta-mode replication from a trapped, defective l prophage) was prevented in cells with plasmids exhibiting IP. Both observations strongly suggest that theta-mode replication initiation is silenced, in trans, by the oop + oril + plasmids. Blocks to theta-mode replication initiation from an infecting phage, by cellular oop + oril + plasmid copies or by the chromosomal dnaB-GrpD55 mutation, could be bypassed by multiply infecting such cells with l. This result is not without precedent. Freifelder et al. [56] infected nonpermissive cells at MOI's between 0.01 and 40 with lcI857Pam3 phages that were variously inactivated for integration or Red recombination functions. For their Int + Red + variant, they showed an increase in phage burst of 240-fold between MOI's of 0.01 (transmission coefficient 0.001) and 10 (transmission coefficient of 0.24), yet the lcI857Pam3 phage was unable to form plaques on nonpermissive cells; and in our hands the Pam3 mutation reverts at a frequency of ,10 27 . Freifelder et al. [56] concluded that if recombination is reduced, the ability to produce mature phage was markedly reduced. McMillin and Russo [57] reported that under conditions which block l DNA duplication, unduplicated l can mature, including molecules which have recombined in the host. Stahl et al. [58] extended this observation, coining the term ''freeloader'' phage to describe phage produced under replicationblocked conditions, whose synthesis depended upon bacterial and phage recombination systems. We borrowed this concept, using ''free-loader coefficient'' to describe the influence of phage recombination functions on l progeny from infected dnaB-GrpD55 cells in which the infecting phage genome cannot initiate theta-mode replication. We showed that phage recombination functions from both NinL and NinR regions can influence by up to ten-fold the phage progeny released from multiply infected dnaB-GrpD55 host cells, supporting the Freifelder et al. [56] conclusion. Sclafani and Wechsler [59] showed that at low MOI, no l particles were produced in cells lacking a functional dnaB product; yet at high MOI, a significant proportion of the cells can produce phage. Thus, the bypass of an oril replication block in multiply infected cells could depend upon a recombination-driven replication shunt, possibly analogous to the replisome invasion mechanism described by Poteete [60]. It is recognized that if a cell contains $2 circularized l genomes, recombination between the monomers can produce an invading strand which could lead to rolling circle replication, independent of oril [61]. Presumably, recombination / replication intermediates can be formed that produce packageable, concatemeric DNA by the introduction of a nick into one of the DNA strands of a l monomer, enabling rolling circle replication initiating from the 39-OH end of the nick, or by recombination between homologous l DNA segments. It was proposed that double-strand break repair recombination intermediates in E. coli are capable of initiating and undergoing DNA replication [62,63]. It is possible that the circularized l genomes produce linear multimers, formed by the rolling circle type of plasmid replication dependent on the RecF recombination pathway [64][65][66][67].
The potential to bypass theta-mode replication initiation via recombination suggests that there is no obligatory order / mechanism for triggering late mode l replication from the early oril-dependent replication products. Alternatively, the extensive evidence for a shift from early to a late replication mode supports the possibility that some natural mechanism can inhibit early theta-mode replication initiation. Two events come to mind where theta-mode replication initiation is undesired and would best be silenced. Theta-mode bidirectional replication forks arising from a l DNA copy that is integrating, or has integrated, into the host chromosome will kill the potential lysogen via the escape replication (Replicative Killing). The initiation of theta replication from linear concatemeric DNA might inhibit genomic DNA packaging into the phage head. Our results for plasmid based IP suggest that there is a natural mechanism for silencing theta-mode replication initiation, i.e. the buildup of l genomes with oop + oril + sequence.

Toward a mechanism for IP
There are a number of ways oop expression could influence transcriptional activation of oril: i) OOP antisense RNA binding the p R transcript could promote degradation of the downstream cII-O-P transcript, in turn limiting transcriptional activation of oril and O-P expression. ii) Cells expressing OOP antisense RNA can nullify CII formation, eliminating p E -preX-cI-rexA-rexB transcription and the (little appreciated) potential of this mRNA to permit a) high CI repressor buildup, b) hypothetical orf-preX expression, or c) high level p E -promoted antisense RNA to cro expression, in turn, reducing Cro buildup and interference with transcription from p R (Fig. S7). Since the repP22 phage lcI857 (18,12)P22 was insensitive to IP, yet almost fully shared the same cI-p R -cro-cII-oop sequence as repl phages, it seems unlikely that the contribution of oop to CI-independent IP simply involves OOP serving as an antisense RNA to the p R -cII mRNA, or events that increase transcription from p R , but they might explain why cells with an oop + plasmid can stimulate phage maturation (i.e., support larger plaques). Overall, the results suggest that OOP RNA expression from an oop-oril DNA template increases the sensitivity of repl genomes to competing oril sequences, with the outcome of silencing theta mode replication initiation from the oril sites. This is a new idea in search of an explanation. Some form of molecular coupling between oop expression and oril may serve to block the formation or completion of the preprimosomal complex. Several old observations remain a mystery regarding the regulation of oop expression. A low level of p O transcription arises from a repressed prophage [4], which, if extrapolated would additively increase the level of OOP in cells with multiple oop + plasmids. This low level transcript was discovered because its expression increased about 40-fold between 5 to 12 minutes following the thermal induction of a cryptic l prophage (as in Fig. 4) [4,28]. The increase was linked to phage replication, since a prophage deleted for P showed no OOP increase [4], nor was there an increase from intact l prophages in cells with Ts host dnaB or dnaG genes, or prophage with O, P, or oril mutations [49] which we have confirmed by sequence analysis. While one might explain this as a gene dosage effect, the level of induced oop expression was about the same from an induced defective prophage [49] as from an induced lcI857Sam7 prophage defective for cell lysis (Table S2), where we typically see between 30 -200 + fold increase in phage particles; or when l was induced in cells with a Ts dnaE mutation blocking DNA fork progression [49]. This coupling between replication events at oril, and oop expression, still requires an explanation.

Reagents and media
Growth experiments were carried out using tryptone broth (TB; 10 g Bacto-tryptone and 5 g NaCl per liter), TB plates (TB with 11 g Bacto-agar per liter) and TB top agar (TB with 6.5 g Bactoagar per liter). Ampicillin was added to a final concentration of 50 mg/ml where required. F80 buffer (0.1 M NaCl, 0.01 M Tris-HCl, pH 7.6) was utilized for cell culture and phage dilutions, TE (0.01 M Na 2 EDTA, 0.01 M Tris-HCl pH 7.6) and TE* (TE but with 0.001 M Na 2 EDTA) buffers were used for DNA storage and manipulation of DNA, respectively. TM buffer (0.01 M MgSO 4 , 0.01 M Tris-HCl, pH 7.6) was used in phage burst assays. TBE buffer (0.089 M Boric acid, 0.002 M Na 2 EDTA, 0.089 M Tris-HCl, pH 8) was used to make agarose gels and as running buffer during electrophoresis. Restriction enzymes and T4 DNA ligase were from New England Biolabs. Taq DNA polymerase was from Invitrogen and New England Biolabs. Oligonucleotides were from Sigma Aldrich and Integrated DNA Technologies, Inc. Plasmid DNA was isolated using Promega Wizard Plus SV Mini and Midi prep, or Qiagen miniprep kits. DNA was isolated from gels using the Qiagen gel extraction kit, and reaction fragments were purified using the Qiagen QIA quick PCR purification kit. Table 1 shows the E. coli K-12 and bacteriophage strains and Table 2 and Fig.'s 1, 3, S3 show the plasmids employed. All of the plasmids were derived from plasmid pCH1 [11] prepared by ligating the l34500-41731 BamHI fragment into the unique BamHI site of pBR322. The l sequences are as described by Daniels et al. [2]. The l fragment orientation in pCH1: l base pair 41731 was closest to the N-terminal end of the interrupted tet gene.

Plaque Assay
Repl = lcI857 and repP22 = lcI 857(18,12)P22 infecting phages were plated on several plasmid-containing host cell strains to measure plasmid-mediated inhibition of phage plating. An aliquot (0.25 ml) of a fresh overnight cell culture was mixed with 3 ml of warm TB top agar and 0.1 ml of diluted repl or repP22 phage lysate, and poured over TB or TB+Amp plates. Plates were incubated at 30u overnight and plaques counted. The results were expressed as EOP, i.e. phage titer on 594[test plasmid] / phage titer on plasmid free host 594 cells.

Prophage Induction Assay
The repl and repP22 prophages were thermally induced in lysogenic cells transformed with plasmids containing various l fragments. Lysogenic cells were grown at 30u in 20-ml TB (+/-Amp) in a shaking bath to A575 nm = 0.15. The cI[Ts]857 prophage in the cells was synchronously induced by swirling the culture flask in a 55-60uC water bath for 15 seconds and then transferring to a 42u shaking water bath to denature the repressor. The culture absorbance was monitored at 30 minute intervals over five hours. Each culture assay was repeated, the several results were averaged and the standard error determined.

Phage Burst
Host cells transformed with plasmids containing various l fragments were infected with a repl or a repP22 phage at a high or low MOI. The phage particles released per infected cell (i.e. phage burst) were measured for each infection. Protocol: 16-18 hour culture cells grown at 30u in TB (+/-Amp) were pelleted and resuspended in an equal volume of W80 buffer. A cell aliquot (0.1ml) was mixed with 0.2-ml of ice cold 0.01 M MgCl 2 /CaCl 2 plus an appropriate volume of sterile phage lysate needed for MOIs of 5 or 0.01. The cell-phage infection mix was held on ice for 15 min to permit phage attachment and then transferred (time zero for measuring infective centers) to a stationary 42u air incubator for 10 min to permit phage infection. The cell-phage mixture was pelleted and resuspended (2X) in W80 buffer and the third cell pellet was resuspended in 0.4 ml pre-warmed 42u TB. Half of the resuspended cells (0.2-ml) were inoculated to 20 ml TB (+/-Amp), incubated with shaking at 42u, and aliquots were removed after 65 and 110 min from the time of inoculation to determine phage titer. The second half (0.2-ml) of the washed cell-phage mixture (first held 15 min on ice and then at 42u for 10 min) was immediately pelleted. The supernatant was used to measure the unattached phage remaining after the attachment and infection steps, and the cell pellet was resuspended, diluted, and aliquots were mixed with sensitive cells, top agar, and overlayed on a TB agar plate. Each plaque that arose on the plate was from a potential infective center (an infected cell that has not yet lysed). The phage burst (number of phage released per number of infective centers) was determined for the 65 and 110 min infections, correcting for the phage particles that did not attach to cells.

OOP Phenotype/CII Inactivation Assay
The last 17 codons of cII are not required for CII activity, but are necessary for CII regulation by OOP [5]. The C-terminal 52 nucleotides plus the stop codon for gene cII overlap the 39-end of oop. The expression of OOP antisense RNA from a plasmid prevents lambda CII expression [48], resulting in an otherwise cII + phage producing clear, rather than turbid, plaques. An aliquot (0.3 ml) of stationary phase cells being tested for OOP activity was mixed with 0.1 ml of diluted lcI857(18,12)P22 phage plus 3 ml of warm TB top agar and poured onto TB plates. The plates were incubated overnight at 30u. Plaque morphology was then determined as clear (OOP + ) or turbid (OOP -).

Plasmid Sequence Modification
We supplied primers and DNA template to the service at National Research Council/Plant Biotechnology Institute, Saskatoon to confirm the l-region sequences for the plasmids employed and to verify the mutations introduced into plasmid p27R. PCR mutagenesis was used to modify the t O -oop-p O and oril plasmid DNA sequences using the SOEing technique [68] For mutating the -10 region of the p O promoter in p27R, two primers were made that contained the sequence 59GCGCG39 in place of the wt sequence 59ATTAT39 at l bases 38684-l38688. One primer contained the l-strand sequence l bases 38671-38700 (LPo3) and the other contained the r-strand sequence l bases 38700-38671 (RPo2) ( Table 3). The p27R template was PCR amplified with the mutated primers and with primers LPo1 (59 NdeI site and l bases 38357-38372) and RPo4 (59 EcoRI site and l bases 39172-39153) in a two-step PCR technique. Both for this plasmid and for those described below, the final PCR product was digested with NdeI and EcoRI and ligated into the larger (,2000 bp Amp + ColEI origin) fragment resulting from p27R NdeI and EcoRI digestion. p27R-R45OOP: Bases 2-46 of the oop gene coding sequence in p27R were mutated. Two primers were made to contain ''random'' bases (screened to eliminate secondary structures) replacing l bases 38630-38674 of the wild type oop sequence. One primer contained the l-strand sequence (LROOP3) and the other contained the r-strand sequence (RROOP2) ( Table 3). The p27R template was PCR amplified with the mutated primers and with primers LPo1 and RPo4 (Table 3). p27RDITN1-4: Two hybrid primers were made to delete iterons (ITN) 1-4, each with sequences flanking the iterons. LDITN1-4 contained the l bases 39014-39033 fused to 39120-39144, while RDITN1-4 contained the same sequence on the r-strand (Table 3). These two primers, in conjunction with LPo1 and RPo4, were used for deleting l bases 39044-39119 (i.e. 87 nt of ITNS 1-4). p27RDITN3-4: Two hybrid primers were made for deleting iterons 3 and 4 from p27R. LDITN3-4 contained l bases 39058-39077 fused to 39120-39144, while RDITN3-4 contained the same sequence on the r-strand (Table 3). These two primers along with LPo1 and RPo4 were used to delete l bases 39078-39119 (i.e. 41 nt comprising iterons 3 and 4). pHB27RDAT: Primers LPo1 (59 NdeI site and l38357-38372) and RDAT1 (59 EcoRI site and l39127-39113) were used to amplify the pHB27R l DNA fragment. The resulting PCR fragment was digested with NdeI and EcoRI and cloned into the 2000 bp pBR322 fragment from pHB27R digested with NdeI and EcoRI. The plasmid pHB27RDAT was shown to be deleted for l bases 39,128-39172, removing the AT rich region of oril (Table 2).
Isolation and sequencing Sip mutants lcI857 formed small plaques at a frequency of #10 26 on 594[oop-oril] cells. An individual plaque from ten separate isolations was transferred by a sterile toothpick to 10 ul buffer (10mM Tris-HCL, 10 mM MgCl2, pH 7.6) and spread using sterile paper strips onto a fresh agar overlay of these cells. This procedure was repeated (as many as 13 times) yet always produced plaques that were heterogeneous in size on the 594[oop-oril] cells. Each of the ten independent Sip phages were plated on 594 host cells (without plasmid) and a single plaque was used to prepare a phage lysate. Single plaques arising from these lysates were sequenced from gene cI into P (l bases 37905-39191) using primers LMH29 (37905-37922: 59-CTGCTCTTGTGTTAAT-GG), L22 (38517-38534: TGCTGCTTGCTGTTCTTG), RPG6 (38569-38552: CAATCGAGCCATGTCGTC), and R9+1 (39191-39175: TGGTCAGAGGATTCGCC).

Assay for replication initiation from induced cryptic l prophage
The method is described in [18], only herein, chromosomal DNA was digested with NdeI, not BstEII. Figure S1 Aligned conserved sequence regions for 23 lambdoid phages. Sequence regions were searched using a 33 nt region of sequence similarity between HK620 and l (''sequence 5'' in [79]). The bases in red show greater than 90% sequence homology. The sequence of OOP spans positions -90 (terminator end) through -10 (59end). The termination sequence for lambda gene cII, extending from the left, is at position -33. Position 1 is set as the ATG start for lambda gene O, for P22 orf48 homologue as hkaW, EC_CP1693_21), or a HK097 gp53 homologue orf54 (see Fig. S2) [80]. An annotated version of this data was provided in the review  [80], and gene p43 in HK97, representing 162 nt (NC_002167). This figure was redrawn with modification from [51]. (TIF) Figure S3 Influence of spacing between oop and oril on repl-inhibition. Influence of spacing between oop and oril on repl inhibition. A. Plasmid p50 substitutes E. coli DNA from the specialized transducing phage lspi156 for the ''ice'' sequence of l ( Table 2) and was made by cloning the 684 bp EcoRV-EcoRI fragment from lspi156Dnin5 [96] into the equivalent sites in pBR322 [69]. B. The stable predicted secondary structures of OOP RNA were obtained using the IDI SciTools OligoAnalyzer 3.0 website. C. EOP of repl and repP22 phages on host cells with modified Dice oop + oril + plasmids. The averaged data is shown. (Near identical results were seen for each of the plasmids transformed into E. coli strain W3350, where standard errors were negligible for the repl phage, and ranged between ,0.1 to 0.28 for the repP22 phage on the different transformed cells.) D. Plasmid modifications to p50: l DNA fragments in which the DNA interval between oop and oril was varied by deletion or insertion ( Table 2). (TIF) Figure S4 Plating-sensitivity to cells exhibiting inhibition phenotype (IP) and relative plaque size on cell lawns. A. Variation in susceptibility of repl phages to the IP. A 0.3 ml aliquot of fresh overnight stationary phase 594[p27R] cells (grown in TB+50 ug/ml Amp) were mixed with 0.1 ml of test phage and 3.0 ml of molten top agar and poured onto a TB plate. Plates were incubated overnight at 30uC and resulting pfu were counted. EOP was calculated as the titer on strain 594[p27R]/titer on 594. The results represent the average of at least two independent assays. Averaged EOP's and standard errors values were: lWT (wild type), 5 27 . Notes: 1) The downstream promoter in lcI90c17 was apparently not strong enough to suppress IP.

Supporting Information
2) The plasmids employed in earlier studies [8,10,12] inhibited lvir, but each included cI repressor gene. We show ( Table 4) that lvir was inhibited for plating at 30u in cells with multiple copies of the O/oril plasmid version with cI from imml; whereas, Fig S4A shows lvir is only partially inhibited by cells with oop + oril + plasmids without cI, thus, CI availability to bind oR can increase repl phage sensitivity to IP. B. Portion of l map showing region of DNA substitution for the imm21 and imm434 hybrid phages and the portion of lDNA present in plasmids transformed into strain 594. C. Strain 594 was grown overnight to stationary phase in TB [18]; alternatively, 594 transformed with one of the plasmids, shown in part B, was grown overnight in TB+Amp (50 ug/ml). The culture cells (0.25 ml) were mixed with 0.1 ml of phage lysate dilution plus 3 ml TB top agar [18], poured on TB agar plates, and incubated overnight at 30uC. Phage plaque sizes were determined using a tissue culture (inverted) microscope at 46magnification with an eyepiece grid. Each grid interval was 0.045 mm at 46magnification. Plaque diameters were measured as grid units, i.e., grids/plaque. Approximately 30 plaques were measured per assay phage on each of the host strains and the average plaque diameter and SE were determined. All assays for a given phage were performed in parallel on each of the host strains using same preparation of agar plates. (TIF) Figure S5 Sequence determination for distal cII-oop to O interval for l-hybrid imm434, imm21, and repP22 phages employed. Hybrid phage sequences compared to l. The highlighted/underlined bases differ from l sequence; all data were from this laboratory except sequences for phages 434 and 21; sequence differences rightward from base 38698 are continued in Fig. S1). Phage limm21, which retains the repl sequence, had a silent TGC to TGT codon change (not shown in Fig.'s S5 or S1) at 39,033 (one base left of the ITN1 sequence in O). Lambda = lcI857 (DQ372056) is as in [2]; limm434cI (DQ372053.1); limm21cI (DQ372054.1, being revised); and P22-Lambda hybrid = lcI857(18,12)P22, representing lhy106 from Dr. S. Hilliker (DQ372055.1). The comparative partial sequences for non-hybrid phages 434, 21 and P22 were: phage 434 (GI:14988); phage 21 (GI:4539472), and phage P22 (AF527608.1; GI:21914413; AF217253.1). (TIF) Figure S6 PCR assay for plasmid recombination into l Sip phage within region of l homology. PCR Amplification of lcI857 and SIP Phage Isolates 1-4, from Gene cI Through Gene P. Lanes: 1 & 11, DNA mass ladders from Invitrogen, 2-3, lcI857, 4-5, lcI857 Sip1, 6, lcI857 Sip2, 7-8, lcI857 Sip3, 9-10, lcI857 Sip4. The phages were amplified with primers LMH29 and RPG6 (Methods and Materials). Each PCR was done in duplicate. lcI857 produced the expected 1721 bp fragment. The SIP isolates yielded a 1721 bp fragment, indicating that the p27R plasmid was not integrated into the SIP phage genomes between genes cI and P. (TIF) Figure S7 Sequenced Sip and Se mutations falling within orf-preX. A. Organization for transcription of gene cI from pM and pE. Transcription from pE is 30-100X the level of transcription from pM, [11,28,39,40] and includes an open reading frame preX [14] of 81 codons. Three powerful translational frameshift sites exist within the cI-rexA-rexB operon [14,43] that could influence gene expression from the pE promoter, two arise within the N-terminal end of cI and one within rexA 1 . B. DNA sequence showing potential translation of preX and its overlap with genes / proteins CI and CRO. This figure shows an alternative interpretation for the position of some Sip mutations shown in Fig. 5, which also map within orf-preX. The previously described Se-mutations confer a cIphenotype [13]. The mutations se100a and 101b arise in oR2 and oR1 between the -35 regions for promoters pM and pR, and se109b is representative of four other spontaneous se mutations, arising within oR1 and just left of the -10 region of pR. An alternative interpretation is that se100a, se101b and 109b, respectively, confer G56V, T54K and T46N changes in the putative 81 codon preX orf. (TIF)