Genome of Enterobacteriophage Lula/phi80 and Insights into Its Ability To Spread in the Laboratory Environment

Abstract

The novel temperate bacteriophage Lula, contaminating laboratory Escherichia coli strains, turned out to be the well-known lambdoid phage phi80. Our previous studies revealed that two characteristics of Lula/phi80 facilitate its spread in the laboratory environment: cryptic lysogen productivity and stealthy infectivity. To understand the genetics/genomics behind these traits, we sequenced and annotated the Lula/phi80 genome, encountering an E. coli-toxic gene revealed as a gap in the sequencing contig and analyzing a few genes in more detail. Lula/phi80's genome layout copies that of lambda, yet homology with other lambdoid phages is mostly limited to the capsid genes. Lula/phi80's DNA is resistant to cutting with several restriction enzymes, suggesting DNA modification, but deletion of the phage's damL gene, coding for DNA adenine methylase, did not make DNA cuttable. The damL mutation of Lula/phi80 also did not change the phage titer in lysogen cultures, whereas the host dam mutation did increase it almost 100-fold. Since the high phage titer in cultures of Lula/phi80 lysogens is apparently in response to endogenous DNA damage, we deleted the only Lula/phi80 SOS-controlled gene, dinL. We found that dinL mutant lysogens release fewer phage in response to endogenous DNA damage but are unchanged in their response to external DNA damage. The toxic gene of Lula/phi80, gamL, encodes an inhibitor of the host ATP-dependent exonucleases, RecBCD and SbcCD. Its own antidote, agt, apparently encoding a modifier protein, was found nearby. Interestingly, Lula/phi80 lysogens are recD and sbcCD phenocopies, so GamL and Agt are part of lysogenic conversion.

INTRODUCTION

The laboratory setting is, by design, an inhospitable environment, as very few species are capable of overcoming the researcher-imposed barriers for growth, multiplication, and spread. The well-known examples of organisms successfully propagating in the laboratory environment are mycoplasmas (76) and HeLa cells (39), both of which cross-contaminate mammalian cell cultures and spread through the laboratories around the world with appalling ease (48). Part of their success is their rapid growth and hardiness, giving them the ability to outgrow and outlive any laboratory strain competitors. Until recently, however, no such creature was known for cultures of rapidly growing microbes, which, precisely because of their speed of growth, effectively deny such chances to a contaminating competitor. This has changed with a description of a temperate phage, Lula, that infects Escherichia coli, converts it into a lysogen, and efficiently spreads in the laboratory setting, stealthily contaminating laboratory strains without researchers' knowledge (75). Lula contamination is apparently widespread yet not broadly recognized, and, because of this combination, it can be a scourge in shared facilities that handle both contaminated cultures (lysogens) and noncontaminated cultures (nonlysogens), resulting in a frustratingly high frequency of lysis and growth inhibition of the latter “for no apparent reason.” Besides being superinfectious, Lula also changes the phenotype of cells in which it established itself as a lysogen, making them extra sensitive to any DNA damage (75), so its lysogenic conversion is not limited to a typical cross-resistance of lysogens to infection by lytic phages.

Besides being the first characterized microbial culture-contaminating agent, Lula exemplifies a distinct, viral strategy of spreading in the laboratory environment; thus, study of Lula provides original insights into the ways viruses breach human-imposed barriers to cross-contamination. This is a topic of critical importance for mammalian cell cultures, where viral cross-contamination among cell lines is pervasive, sometimes with fatal consequences for researchers (37). In a previous paper we analyzed Lula's characteristics that allow it to spread in the laboratory setting and concluded that these are (i) stealthy infectivity at 37°C (lysogenization instead of lytic development), allowing Lula infection to avoid detection, and (ii) cryptic lysogen productivity, yielding normally growing cultures of Lula lysogens carrying extremely high loads of infectious phage (75). The latter property is critically enhanced by Lula's stability in stationary-phase cultures (75). These characteristics could be the consequence of Lula's distinct genome organization relative to that of other temperate phages or could be due to specific genes unique to Lula. In our efforts to understand the genetic and genomic bases for the above-described traits that make Lula so efficient in overcoming the researcher-imposed barriers to its spread in the laboratory environment, in this paper we describe sequencing and annotation of the Lula genome, as well as characterization of several Lula mutants.

MATERIALS AND METHODS

Strains, plasmids and growth conditions.

The E. coli strains and plasmids used in the experiments are described in Tables S1 and S2 in the supplemental material, respectively. Alleles were moved among strains by P1 transduction (63). All strains were propagated in liquid LB with shaking at 200 rpm or on LB plates (containing, per 1 liter, 10 g of peptone and 5 g of yeast extract, 5 g of NaCl [pH 7.4] with 250 μl of 4 M NaOH and 15 g of agar) (63). Antibiotics were used in the following final concentrations: 50 μg/ml for kanamycin (Kan), 100 μg/ml for ampicillin, 10 μg/ml for tetracycline, and 12.5 μg/ml for chloramphenicol.

Sequencing of the phage genome.

A 405-μl portion of high-titer phage stock was treated with 45 μl DNase I buffer and 1 μl DNase I (NEB) at 37°C for 15 min to eliminate E. coli chromosomal DNA. Fifty microliters of 10% SDS was added, and the phage was lysed at 70°C for 10 min, followed by 5 min on ice. Phage DNA was extracted by shaking with 500 μl phenol, with 500 μl phenol-chloroform, and finally with 500 μl chloroform; the aqueous phase was transferred to a fresh tube at each step. The DNA was precipitated twice with 20 μl of 5 M NaCl and 1 ml of 100% ethanol and dissolved first in 500 μl Tris-EDTA (TE) buffer and then in a final volume of 100 μl TE buffer. The DNA concentration was approximately 40 ng/μl, as measured with a fluorometer. As a quality control, Southern blots using the isolated phage DNA as a probe revealed no hybridization to nonlysogen E. coli DNA. Two micrograms of this phage DNA was submitted for shotgun sequencing (done at the W.M. Keck Center for Comparative and Functional Genomics, University of Illinois at Urbana-Champaign). A total of 587 sequence runs of 800 to 1,000 nucleotides (nt) in length were compiled using the program Sequencher 4.9.

Digestion with restriction enzymes.

Standard restriction digestion for phage and other DNA (∼1 μg) was done in 100-μl reaction volumes, using 1 μl NEB enzyme (typically 10 to 20 U) and in the recommended NEB buffer. Digestions were for 1 h at 37°C; prior to gel loading, DNA was concentrated by ethanol precipitation.

Southern blotting.

Approximately 1 μg of total DNA of a strain containing Lula was digested as described above for 4.5 h at 37°C and separated in an agarose gel. The gel was treated for transfer by gentle shaking in 0.2 M HCl for 40 min, followed by 45 min in 0.5 N NaOH and 15 min in Tris-HCl, pH 8.0. After capillary transfer, DNA was cross-linked to the membrane by UV irradiation, and the membrane was incubated in hybridization buffer (5% SDS, 0.5 M sodium phosphate, pH 7.4) at 65°C for 5 h. To probe the membrane, ³²P-labeled BamHI-digested Lula DNA was added to the hybridization buffer and the membrane was hybridized overnight. After three washes in 0.1× hybridization buffer, the membrane was exposed for 9 h and quantified with a PhosphorImager.

Determination of cell titer.

Saturated cultures were serially diluted 10-fold at each step in 1% NaCl. Ten-microliter spots of each dilution were placed onto LB or BBL agar (10 g BBL Trypticase, 5 g NaCl, 250 μl 4 M NaOH, and 15 g agar per liter) plates and incubated overnight at 30°C. A stereomicroscope was used to count the dilutions which contained between 30 and 200 colonies to determine the original titer.

Determination of Lula/phi80 titer.

Saturated cultures of lysogens were pelleted in a microcentrifuge. The supernatant was serially diluted in TM buffer (10 mM Tris-HCl [pH 8.0] and 10 mM MgSO₄), and 10-μl aliquots were placed atop a lawn of susceptible cells on BBL agar. The lawn consisted of 150 μl saturated culture mixed with 4 ml top BBL agar (half TM and half BBL) incubated for 1 h at 30°C before spotting the phage. Plates were incubated overnight at 30°C, and individual plaques were counted to determine the original titer.

UV survival.

Logarithmic-phase cultures were serially diluted in 1% NaCl to 10⁻⁶, and 10-μl portions of the six dilutions (10⁻¹ to 10⁻⁶) were spotted onto BBL agar square plates in six rows (one strain per 36-position plate, with six spots of the same dilution per row) and allowed to dry. The plates were irradiated with a UV gradient perpendicular to the dilution gradient and incubated overnight in the dark at 30°C. The titers in the UV-treated columns were normalized to the titer of the unirradiated column.

T4 spotting.

Suspensions of phage T4 and T4 2 mutants were diluted in TM buffer. Plates were made of Hershey agar (13 g tryptone, 2 g sodium citrate, 8 g NaCl, 1.3 g glucose, 4 mg thiamine, and 15 g agar per liter) (8). The lawn was created by subculturing cells 1:100 in LB broth (supplemented with 100 μg/ml ampicillin and 1 mM IPTG [isopropyl-β-d-thiogalactopyranoside] as needed). Cultures were grown to an optical density at 600 nm (OD₆₀₀) of 0.2 to 0.25, and 110 μl of cells was mixed with 1.4 ml top agar (equal volumes of TM buffer and Hershey agar with 1 mM IPTG) and poured onto a small (60-mm) plate. The lawn was dried for 20 min at room temperature before spotting the T4 dilutions. The plates were incubated overnight at 37°C.

Construction of mutations in Lula.

Gene knockouts in Lula/phi80 were made by the Datsenko-Wanner method of deletion-replacement (24) in lysogens of AB1157. The entire gene was replaced with a chloramphenicol resistance cassette to allow for backcrossing into different backgrounds (either by P1 transduction into lysogens or by direct infection with Lula). See Table S1 in the supplemental material for more details.

Lambda-pal phage methods. (i) Phage genotypes.

Lambda-pal is MMS1632 pal571 cI857 b1453 Chi+76. b1453 is a deletion spanning the int, xis, exo, bet, and gam genes (38). The pal571 palindrome was described previously (11). DKC14 is red270 gam210 cI857 ChiC (Stahl lab collection).

(ii) Phage lambda titer determination.

Plating cultures were grown from overnight cultures diluted 100-fold in tryptone broth (TB) (10 g of tryptone and 5 g of NaCl per 1 liter with the pH adjusted to 7.2 with 4 M NaOH) supplemented with 0.1% maltose at 37°C. One milliliter of late-log-phase culture was spun down and resuspended in 0.5 ml of TM buffer. A 0.5-ml portion of the plating culture was mixed with 3 ml of top BBL agar and poured on a freshly prepared 25-ml BBL plate, and the plates were dried for 30 min before spotting phage. Phages were serially diluted in TM buffer, and 10-μl dilutions were spotted on the plates with the plating culture in top agar. Plates were incubated overnight at 35°C. High-titer phage stocks were purified according to the one-plate protocol (45) with the host JC7623.

Nucleotide sequence accession number.

The sequence of the Lula/phi80 genome has been deposited in GenBank under accession number JX871397.

RESULTS

Sequencing of the Lula/phi80 genome.

We sequenced the Lula genome by the Sanger technique via construction of a shotgun library from isolated virion (linear) DNA. Instead of the expected single contig, the reads assembled into two separate contigs, 29,331 bp and 16,250 bp in length, both with a 10-fold coverage. Two of the four ends of the two contigs carried the same 12 nucleotide sequence GGGCGGCGACCT (one on the 3′ end and the other on the 5′ end), apparently identifying the two ends of the opened cos site of the virion DNA (which also happens to be the sequence of the “sticky ends” of bacteriophage lambda). We reasoned that the two other ends defined the gap in Lula DNA spanning a gene(s) that could not be cloned in E. coli. We closed this 569-nt-long gap by sequencing a PCR product across the gap, which allowed us to link the two contigs into a single one.

The total Lula genome turned out to comprise 46,150 bp, in line with other lambdoid phages (48,502 bp for lambda and 46,375 bp for N15). While there were no similar genomes in the database at the time of Lula sequencing, several sequenced genes of the well-known but now forgotten lambdoid phage from the 1960s and 1970s called phi80 (77) were identical to the corresponding Lula sequences. In fact, although it was never published, the phi80 genome was sequenced by the members of the Blattner lab in Wisconsin (27), who used its DNA to troubleshoot a variety of sequencing technologies in the 1980s and who have made the resultant sequence available to individual researchers. Comparing the Wisconsin phi80 and Lula sequences revealed complete identity of the two genomes, down to the last nucleotide. It was remarkable that, after so many years of spreading in various laboratories (our “Illinois” phage came originally from Arizona via Oregon), there was not a single nucleotide change in the two sequences.

Sequence analysis and annotation of Lula/phi80 genome was as previously described (56, 72). The overall GC content of Lula/phi80 is 52%, which is similar to E. coli's 51%. No tRNA genes were found by tRNAscan (57).

Overall genome layout.

As already mentioned, the linear (virion) Lula/phi80 genome starts and ends with the 12-nt complementary (“sticky”) ends indicating a cut cos site, which shows during sequencing as 12-bp direct repeats bracketing the genome. The genome features an attachment site roughly in the middle of the chromosome near the integrase (int) gene, four major promoters (pL, pRm, pR, and pR′), and at least nine transcription terminators. The disposition of promoters and terminators is mostly preserved between the Lula/phi80 and lambda genomes, which means that the direction of expression and composition of the major gene clusters are also preserved. Thus, the overall layout of the Lula/phi80 genome is very much lambda-like (Fig. 1), with the linearized-at-cos genome starting at the left with the cluster of head functions, followed by, in order, a cluster of tail functions, a cluster of infection specificity functions, site-specific recombination functions, homologous recombination functions, regulation and replication initiation functions, and finally, on the right end of the chromosome, the host cell lysis functions. The only apparent exception to the lambda order of genes is the reversal of the integrase-containing region in the middle of the Lula/phi80 chromosome, perhaps because of the presence of the gene for Pin invertase nearby. At the same time, the level of identity at the nucleotide level is limited: Lula/phi80 and lambda share only ∼18.5 kbp of common DNA, with an average level of identity in this DNA at just below 70%. Sequence identity is mostly clustered in the morphogenesis (head and tail) part of the genome, with the rest of the two genomes highly divergent but harboring occasional islands of identity (Fig. 1), a result which is in line with the early physical analysis (32).

Fig 1.

The layout of the Lula/phi80 genome versus that of the lambda genome. The two genomes, which are 46,150 bp and 48,502 bp in length, respectively, are shown in their virion forms, linearized at the cos site, with the terminase genes on the left. Black arrows, genes transcribed from left to right; gray arrows, genes transcribed from right to left. The names of the important/interesting genes are indicated. The regions of shared sequence identity (averaging 70%) are indicated by gray parallelograms. The genome region labels are as follows: Head, head assembly and DNA packaging (terminase); Tail, tail assembly; Legs, tail fibers; Att, attachment site and integration/excision; Rec, homologous recombination and host exonuclease inhibition; Reg-Rep, regulation/decision making and replication initiation; Lysis, host cell lysis.

Preservation of the overall genome layout is especially striking in the control region (Fig. 2). This region is the location of the following important actions: (i) prophage repression, (ii) lysis/lysogeny decision-making, (iii) replication initiation, and (iv) transcription antitermination. The control region spans approximately 11 kbp both in Lula/phi80 and in lambda, but the shared homology between Lula/phi80 and lambda (again, ∼70% identity) is limited to the total of ∼1.5 kbp of DNA (Fig. 2), split between the 600-bp ppp (ninI) gene (coding for a nonessential phage protein phosphatase; this homology was noted earlier [14, 46]), the 250-bp 5′-terminal half of the Rz gene (coding for the lysis protein), and 800 bp spanning the postiteron 3′ half of the O gene and the 5′ half of the P gene (suggesting that the corresponding parts of the O and P proteins physically interact). None of the control genes or sequence elements show any similarity between the two phages, with the exception of a weak homology between the Q genes. Specifically for our goal, though, the overall genome structure of Lula/phi80 does not provide any insight into its ability to spread in the laboratory, so next we considered specific genes of Lula/phi80.

Fig 2.

Comparison of the control regions of Lula/phi80 and lambda. The region bracketed by kil on the left and Rz on the right is shown, with the (mostly) control genes indicated above the genome line and control sequence elements below the genome line. The gray parallelograms indicate the three regions to which the ∼70% sequence identity between the two genomes is limited.

Annotation of the Lula/phi80 genome.

A brief annotation of the major features of the Lula/phi80 genome is in Table 1. Even though Lula/phi80 has its own history of genetic research, with many genes known by their (nonconsecutive) numbers (77), we decided to part from that time-honored nomenclature, instead numbering the Lula/phi80 genes consecutively, following the standards of modern phage genomics. The numbering starts at the left end with the gene for the small terminase subunit as gene 1, but we also decided to assign Lula/phi80 genes “second” names according to lambda (or other lambdoid phage) gene nomenclature, as more biologically meaningful ones.

Table 1.

Lula/phi80 sites and ORF annotation summary

Gene no.a	Position (nt)		Length		Reading frame	Identity of encoded proteinb or sequence description/best homolog(s) from other phages or bacteria	N15 homolog
	Start	End	nt	aa
s	1	12				GGGCGGCGACCT, sticky end
s	109	165				Inverted repeat, AGAAAGGAAA(38 N)TTTCCTTTCT
1	190	732	543	181	F1	Terminase small subunit (Nu1)/lambda	gp1
2	710	2629	1920	640	F2	Terminase large subunit (A)/lambda	gp2
3	2632	2835	204	68	F1	Head-tail joining protein (W)/N15, lambda	gp3
4	2835	4424	1590	530	F3	Capsid portal protein (B)/N15, lambda	gp4
5	4408	5748	1341	447	F1	Capsid serine protease (C)/N15, lambda	gp5
6	5761	6090	330	110	F1	Head DNA stabilization protein (D)/N15, lambda	gp7
7	6161	7183	1023	341	F2	Major capsid (head) protein (E)/N15, lambda	gp8
8	7196	7645	450	150	F2	DNA-packaging protein (FI)/N15, lambda	gp9
9	7660	8010	351	117	F1	Head-tail joining protein (FII)/N15, lambda	gp10
10	8023	8604	582	194	F1	Minor tail protein (Z)/Fels-1, Gifsy-1, N15, lambda	gp11
11	8604	8999	396	132	F3	Tail capping + head interface protein (U)/N15, lambda	gp12
12	9173	9559	386	129	F2	Putative protein, internal to gpV
12	9010	9750	741	247	F1	Major tail protein (V)/N15, lambda	gp13
13	9764	10192	429	143	F2	Tail assembly chaperonec (G)/N15, lambda	gp14
14	10246	10515	270	90	F1	Minor tail protein (T)/N15, lambda	gp15
s	10429	10469				Inverted repeat, GTCATCAGCG(21 N)CGCTGATGAC
15	10502	13636	3135	1045	F2	Tail length tape measure protein precursor (H)/Fels-1, lambda, Gifsy-1, Gifsy-2, N15	gp16
16	13636	13971	336	112	F1	Minor tail protein (M)/N15, lambda	gp17
s	13996	14011				Rightward transcription terminator (between M and L)
17	14031	14765	735	245	F3	Minor tail protein (L)/N15, HK022, HK97, lambda	gp18
s	14721	14737				Rightward transcription terminator (at the very end of L)
18	14770	15486	1401	467	F1	Tail fiber cell wall hydrolase (K)/N15, HK022, HK97	gp19
19	15482	16096	615	205	F2	Tail assembly protein (I)/T1, N15, HK022 and HK97	gp20
20	16104	16634	531	177	F3	Putative protein, 24% identity to molybdate-binding protein ModA
21	16630	20205	3576	1192	F1	Host specificity protein, tail fiber protein (J)/HK022, HK97, N15, lambda	gp21
22	20210	20509	300	100	F2	Hypothetical phage protein, identical to HK022 gp24/N15, T1	gp22
23	20512	21183	672	224	F1	Hypothetical phage protein/HK022, N15, T1	gp23
24	21298	21528	231	77	F1	Phage superinfection exclusion protein Cor/HK022, N15, T1	Cor
s	21553	21568				Rightward transcription terminator (after cor)
25	21592	22800	1209	403	F1	Mosaic tail fiber protein/HK97, HK022, T1
s	22746	22761	16			Weak LexA box.
26	22877	23188	312	104	F2	Putative protein, perfect antisense for dinL (the next gene)
27	23169	22933	237	79	R2	DinL (MsgA virulence protein) (macrophage survival gene), a relative of host DinI (DNA damage-inducible protein I)/Gifsy-2
s	23210	23225	16			Strong LexA box in front of dinL/msgA
28	23526	23419	108	36	R2	Conserved hypothetical protein
29	23874	23686	189	63	R2	Tail collar domain protein (gpH)/CP-933H
s	23919	23945	27			Inverted repeat, CTTCTCCTGTTtttacAACAGGAGAAG, in front of pin
30	23951	24466	516	172	F2	DNA invertase, resolvase, site-specific recombinase Pin/e14
s	24623	24639				attP (ACACTTTCTTAAATTGT)
31	24808	26013	1206	402	F1	Experimentally established phi80 integrase (Int)/CP-933H, Gifsy-1
s	25961	25981				Leftward transcription terminator (to stop xis transcription antisensing int)
32	26210	26015	195	65	R3	Experimentally established phi80 excisionase (Xis)
33	26497	26255	243	81	R1	Conserved hypothetical protein
34	26930	26497	429	143	R2	DamL, a putative DNA adenine methylase protein, the C-terminal two-fifths/Salmonella enterica, P1, P7
35	27379	26924	456	152	R1	DamL, a putative DNA adenine methylase protein, the N-terminal three-fifths/S. enterica, P1, P7
36	27959	27528	432	144	R3	Putative protein
37	28630	27953	678	226	R1	DNA exonuclease (Exo), homologous recombination/CP-933K, lambda
38	29544	28630	915	305	R2	RecT, the phage recombinase-annealing protein/Rac
39	29835	29557	278	93	R2	GamLd (host RecBCD/SbcCD enzyme inhibitor), toxic without Agt
40	30811	29846	966	322	R1	Agtd (alleviates Gam toxicity), the antitoxin for the toxic Lula GamL/ Sf6
41	31752	30919	834	278	R2	Putative protein
42	31972	31766	207	69	R1	Kil, FtsZ inhibitor protein (blocks cell division)/Salmonella CT18 prophage, Rac
43	31904	32143	240	80	F2	Putative protein, completely overlaps and antisenses gp44
44	32169	31972	198	66	R2	Putative protein, completely overlaps and antisenses gp43
s	32262	32299				Leftward transcription terminator tL1B
s	32473	32495				Leftward transcription terminator tL1A
45	32790	32497	294	98	R2	Early transcription antiterminator protein N
s	33078	33089				CACGTAAACGTG, nutL core (reverse orientation)
s	33134	33142				TGCTTTTTA, nutL boxA (reverse orientation)
s	33207	33225				oL1
s	33207	33212				pL, −10 box
s	33230	33248				oL2
s	33230	33235				pL, −35 box
s	33273	33292				oL3
46	34454	33591	864	288	R3	“Protein 40A” of phi80 (other phage exclusion function?) (acetate kinase superfamily domain)
47	35355	34594	762	254	R2	CI transcriptional regulator
s	35321	35326				pRM, −10 box
s	35323	35341				oR3
s	35344	35349				pRM, −35 box
s	35347	35365				oR2
s	35360	35365				pR, −35 box
s	35380	35398				oR1
s	35383	35388				pR, −10 box
48	35419	35637	219	73	F1	Cro-like transcriptional regulator
49	35671	36210	540	180	F1	CII decision-making protein
RNA	36229	36288				oop RNA (M3 RNA)
50	36299	37204	906	302	F2	Replication initiation protein (O), analog of host DnaA/lambda
s	36663	36677				ori iteron 4
s	36683	36698				ori iteron 3
s	36704	36719				ori iteron 2
s	36725	36739				ori iteron 1
51	37204	37890	687	229	F1	Replication initiation protein (P), analog of host DnaC/lambda
52	37957	38673	717	239	F1	Conserved hypothetical protein /CdtI	gp45
s	38786	38809	24			Inverted repeat, CGCATTTAAGgggaCTTAAATGCG
53	38897	39349	453	151	F2	Putative conserved phage protein
54	39519	40184	666	222	F3	Phage protein phosphatases (PPP) (NinI, AKA Nin221)/lambda, P1, P7
55	40180	40818	639	213	F1	DNA junction-specific endonuclease NinG (Rap, [Nin204])/lambda, P22
56	40818	41021	204	68	F3	Restriction alleviation protein, Lar family
57	41137	41943	807	269	F1	Late transcription antitermination protein (Q)/Gifsy-2, lambda
s	41989	41994				pR′, −35 box
s	42012	42017				pR′, −10 box
s	42070	42098				Rightward transcription terminator (after Q)
s	42362	42385				Rightward transcription terminator (before S)
58	42425	42646	222	74	F2	Phage lysis protein (S), holin/DLP12, Qin
59	42627	43118	492	164	F3	Phage lysozyme (R) (Lysis protein)/DLP12, PS119, PS34
60	43118	43573	456	152	F2	Phage Rz lysis protein, endopeptidase (Rz)/Sf6, DLP12, lambda
61	43338	43520	183	61	F3	Phage lipoprotein (Rz1) precursor/HK97, Sf6, lambda
s	43609	43631				Rightward transcription terminator (after Rz)
62	43739	44290	552	184	F2	Phage regulatory protein Rha, DNA synthesis inhibition in integration host factor-negative hosts
63	44979	44368	612	204	R2	Hypothetical protein
64	44960	45187	228	76	F2	Fur-regulated protein 1
65	45528	45232	297	99	R2	Putative protein
66	45352	46077	726	242	F1	Conserved hypothetical protein/phiKO2	gp59
s	46139	46150				GGGCGGCGACCT, sticky end

^as, DNA site.

^bLambda nomenclature, where applicable.

^cRoger Hendrix, personal communication (October 2009).

^dExperiments, this lab.

We identified a total of 79 open reading frames (ORFs) (not shown) and proposed 66 actual genes (Table 1), with 51 of them producing proteins with either known or apparent functions and another 15 of them coding for putative proteins, some of them conserved with other phages, or whose synthesis was experimentally demonstrated before. Of the phage proteins with known functions, 18 are involved in capsid formation and DNA packaging (head, tail, and terminase), five are involved in the infection process and its specificity (tail fibers, superinfection immunity, and cell wall hydrolase), 12 are involved in various aspects of the phage DNA metabolism (replication, site-specific recombination, repair, and modification), five are involved in transcription regulation (repression, antitermination, and decision-making), and four are involved with cell lysis. The remaining “miscellaneous” proteins of Lula/phi80 are represented by such typical phage functions as Kil (cell division inhibitor) (17, 81) and Lar (restriction alleviation) (44), common phage functions such as phage protein phosphatase (PPP) (13, 14), and some unknown functions such as “protein 40A” (judging by its position in the same operon with the cI repressor, its function may be exclusion of other lytic phages), a molybdate-binding protein, and regulatory protein Rha.

The capsid proteins of Lula/phi80 are those with most recognizable homology to other lambdoid phages, especially lambda and N15, with sequence identity ranging from 30% to 90% depending on the protein and homology to N15 generally higher than that to lambda. However, as mentioned above, such a high level of homology to lambda and N15 is generally limited to the capsid region and is already absent in the tail fiber region, which codes for mosaic proteins with relatedness to several phages, mostly HK022, HK97, lambda, N15, and T1. It should be mentioned that the two major tail fiber genes, 21 (lambda J) and 25, bracket three genes, 22, 23, and 24 (cor), that Lula/phi80 strongly shares with only four other phages: HK022, N15, ES18, and T1. All four phages use the host siderophore receptor FhuA (also known as TonA) for attachment, which the phage-produced and cell-exported Cor lipoprotein then plugs to prevent superinfection (91). Since the surrounding tail fiber proteins are shared among many other phages with various receptor specificities, it was argued that homologs of gp22 and gp23 determine the phage specificity to FhuA (91).

The DNA metabolism proteins of Lula/phi80 are weakly related to those of various phages, including CP-933H, Gifsy-1, Gifsy-2, P1, P7, P22, lambda, and e14. They include an integrase with excisionase, DNA exonuclease and annealing protein (the RecT type) for homologous recombination, and lambda-related replication-initiation proteins O and P. Finally, among the cell lysis proteins of Lula/phi80, the highest homology is to DLP12 and Sf6. Thus, the Lula/phi80 genome is a classic case of phage mosaic genomes, with various genes having recognizable relatedness to corresponding genes of other phages but, with the exception of the morphogenesis region, no extended homology to any particular phage. Thus, the overall gene content of Lula/phi80 provides no insights into its ability to spread in the laboratory.

We reasoned that if we were to find the genetic basis of Lula/phi80's stealthy infectivity and covert productivity, we should concentrate on atypical genes (those found only in some phages), whose products could be involved in the regulatory aspects of DNA metabolism. A few Lula/phi80 genes fall under this category: gene 27 (a homolog of the host DNA damage-inducible gene dinI), gene 30 (encoding an invertase, Pin), gene 34/35 (a homolog of the host dam DNA adenine methylase gene), and perhaps the toxic gene 39, which apparently caused the sequencing gap between nucleotides 29332 and 29900. In search for insights into Lula/phi80 infectivity, we targeted some of these genes for further analysis.

Deletion of genes 34/35 (damL).

We reported before (75) that the virion DNA of Lula/phi80 is cut by BamHI, but our attempts to cut it with several other common restriction enzymes, such as EcoRI and HindIII, were unsuccessful (Fig. 3A and C; see Fig. S1 in the supplemental material). The suspected DNA modification was absent from the prophage Lula/phi80 DNA, inserted into the host chromosome, as this DNA became cuttable by both EcoRI and HindIII (Fig. 3A). Curiously, the enzymes that we originally found to cut Lula/phi80 virion DNA, BamHI (G′GATTC), HaeIII (GG′CC), and AvaII (G′GWCC), all have the GG dinucleotides in their recognition sequence. At the same time, some other enzymes with GG dinucleotides in their sequences, such as RsrII (CG′GWCCG) or StuI (AGG′CCT), still failed to cut under conditions that would cut similarly isolated lambda DNA to completion.

Fig 3.

Lula/phi80 virion DNA modification and the ΔdamL mutant. Ethidium bromide (EtBr)-stained gels are framed and inverted for clarity. (A) Virion Lula/phi80 DNA is not cuttable by many restriction enzymes that cut genomic DNA from lysogens. Left and central panels, inverted pictures of EtBr-stained gels; right panel, Southern blot of the genomic DNA digest, hybridized with whole-genome Lula probe. (B) Scheme of the damL (34/35) gene and its position in the Lula/phi80 genome. The genome of Lula/phi80 is shown in lambda genome notation, opened at the cos site, with the att site indicated in the middle. Gene 34 starts with two Phe residues, while gene 35 ends with two Phe residues, with a +1 nucleotide frameshift between the two genes. (C) The virion DNA from the ΔdamL phage is equally uncuttable by EcoRI and HindIII. (D) The Δdam host has a 100-fold-higher titer of Lula/phi80 in overnight cultures of lysogens. Strains: Host WT Lula WT, EL103; Host WT Lula damL, EL124; Host dam Lula WT, EL125; Host dam Lula damL, EL123.

DNA of large lytic phages, such as T4, is resistant to restriction by the host enzymes due to extensive modification (hydroxymethylation and glucosylation of cytosines) (9). The virion Lula/phi80 DNA could be similarly resistant to restriction if the phage modifies its own DNA during the lytic cycle, for example, by producing a relaxed-specificity methylase. Interestingly, like several other enteric phages (T1 [79], P1 [18], and T2/T4 [9]), Lula/phi80 carries a DNA adenine methylase gene. It is a split gene in this phage, with the N-terminal 3/5 of it in gene 35 and the remaining C-terminal 2/5 of it in gene 34 (Fig. 3B), separated by an apparent ribosome slippage site in the run of 8 A's. The corresponding fused (in silico) product of the two genes is 46% identical to the host Dam methylase that methylates “A” in GATC sites, so we called the fused gene 34/35 damL (dam of Lula/phi80). The specificity of the DamL methylase is not known, but a precise deletion of gene 34 and 35 ORFs did not make Lula/phi80 virion DNA cuttable by the enzymes that failed to cut virion DNA of wild-type (WT) phage (Fig. 3C), demonstrating that DamL is not responsible for the resistance of the virion Lula/phi80 DNA to restriction enzymes. Suspecting a general inhibitor in phage DNA preparations, we used 10 times more restriction enzyme, which helped with some enzymes, such as MluI and SphI, but did not help with EcoRI or HindIII (see Fig. S1 in the supplemental material), so some Lula/phi80 virion DNA modification is still not ruled out.

We noticed that the Δ34/35 (ΔdamL) Lula/phi80 lysogen formed smaller colonies, indicating a growth defect. In the case of the other phages mentioned above, the phage-encoded Dam methylase targets the same GATC site methylated by the host Dam enzyme, supplementing the host enzyme activity and making this methylation possible on phage DNA that may carry other modifications (9). Our results above were also consistent with DamL of Lula/phi80 being a narrow-specificity enzyme, methylating the same GATC sites that its host homolog does. To test whether Lula/phi80 uses its own DamL methylase to supplement host GATC methylation during lysogeny, we measured the titers of the wild-type and ΔdamL Lula/phi80 in cultures of lysogens of wild-type and Δdam mutant E. coli. We found that the Lula/phi80 ΔdamL defect increases the lysogen culture titers only modestly, in contrast to the host dam mutation, which increases the titer by almost two orders of magnitude (Fig. 3D). There is no further increase in phage titer if the dam host is a lysogen of the ΔdamL Lula/phi80 mutant (Fig. 3D). Thus, these data make it unlikely that Lula/phi80 supplements the host methylase with its own enzyme, although the host Dam methylase does reduce the frequency of spontaneous Lula/phi80 induction, probably by reducing SOS induction (70). It is known that dam mutations cause lambda lysogens of E. coli to release more free phage during growth in culture (61).

Deletion of gene 27 (dinL).

The increased UV sensitivity of Lula/phi80 lysogens, compared with that of nonlysogens (Fig. 4A) or lambda lysogens (75), could be due to the action of phage genes, attuned to the host DNA damage response. The DNA damage response in E. coli is called the SOS response (55), and it is induced whenever the homologous pairing and strand exchange protein RecA forms filaments on single-stranded DNA (ssDNA) (53, 71). In E. coli, the genes responding to DNA damage are controlled by the LexA repressor, which is destroyed during the SOS induction by RecA filaments. The LexA repressor has a well-defined DNA binding site with the consensus sequence TACTGTATATATATACAGTA, where the underlined pair of trinucleotides, as well as the precise distance between them, is absolutely conserved (31). Phage lambda has several potential LexA binding sites in its genome that could be repressing its terminase and Gam and Int protein production in the absence of DNA damage. Perhaps the best-known LexA binding site in the lambda genome is found in the control region, between the decision-making gene cII and the replication initiation gene O (51). Not only could the binding of LexA to this site inhibit expression of the replication initiation protein, but it also inhibits the production of the oop transcript (its likely main target), which runs antisense to the cII transcript. The effect of this antisensing is that in the SOS-induced cells, production of the cII protein is inhibited, strongly favoring lytic development over lysogenization, as demonstrated by oop RNA overproduction experiments (87).

Fig 4.

Lack of phenotypes of the ΔdinL mutant of Lula/phi80. (A) UV survival curves. All the values are means of three independent measurements ± standard errors of the means (SEM). Strains: Nonlysogen, AB1157; WT Lula/phi80, EL103; ΔdinL Lula/phi80, EL113. (B) Phage titer at 60 min or 180 min postinduction with 0, 4, or 10 J/m² UV. All the values are means of four independent measurements ± SEM. Strains: WT phage, EL103; ΔdinL phage, EL113. (C) Cell titer at 60 or 180 min postinduction with 0, 4, or 10 J/m² UV. All the values are means of four independent measurements ± SEM. Strains are the same as in panel B. (D) UV sensitivity of the WT-Lula WT lysogen versus that of the ΔdinI-Lula ΔdinL lysogen. The WT and ΔdinI nonlysogens are shown for comparison. All the values are means of five independent measurements ± SEM. Strains: ΔdinI-no prophage, EL120; WT-no prophage, AB1157; WT-Lula WT, EL103; ΔdinI-Lula ΔdinL, EL122.

Lula/phi80, like lambda, has at least four potential LexA binding sites. Two potential LexA binding sites are in the regulatory region, one controlling the oop RNA production downstream of cII, as in lambda (Fig. 2), and the other in the middle of the cI repressor gene. Curiously, binding of LexA to the latter site in cells without DNA damage should reduce production of the phage repressor, making lysogens easier to induce spontaneously but (potentially) harder to induce in response to DNA damage. In contrast to lambda, Lula/phi80 also has a high-affinity LexA binding site (TACTGTtTATtTATACAGTG) in front of gene 27, which is a homolog of a gene called msgA (“macrophage survival gene A”) found in several temperate phages. This gene is implicated in resistance to macrophages in Salmonella (35) and is related to the E. coli SOS-inducible gene dinI (51), which is also present in temperate phages (10). Since Lula/phi80 gene 27 shares 37% amino acid identity with E. coli dinI, we decided to call it dinL (DNA damage-inducible gene of Lula/phi80). There is one more LexA binding site, with a lower affinity, just downstream of dinL, apparently controlling the expression of gene 26, whose mRNA is antisense for dinL mRNA.

In vitro, the DinI protein of E. coli stabilizes RecA filaments (without enhancing their formation) (59, 60), at the same time inhibiting the RecA filament-mediated self-cleavage of the LexA repressor and the UmuD protein (93, 94). The in vivo behavior of dinI mutants of E. coli is also consistent with the idea that DinI is an SOS response modulator, enhancing the DNA repair function of RecA while decreasing its SOS-inducing function (74). If Lula supplements host DinI with its own DinL, amplifying SOS modulation during DNA damage, then inactivation of dinL should make the prophage even more lethal after DNA damage. To test this idea, we deleted dinL in a lysogen. In contrast to our expectations, the UV sensitivities of the Lula/phi80 wild-type and ΔdinL lysogens were the same (Fig. 4A). Moreover, we detected no difference between the ΔdinL Lula/phi80 lysogen and the wild-type lysogen in titers of the phage released after UV induction at two different times and with two different UV doses (Fig. 4B). Under the same conditions, the titer of surviving cells also showed no significant difference (Fig. 4C). Thus, the status of the dinL gene has no influence on the gross parameters of Lula/phi80 lysogen induction by UV.

Only when we added the effect of ΔdinI mutation of the host did we detect a small difference in UV sensitivity between the WT-Lula WT lysogen and the ΔdinI-Lula ΔdinL lysogen (Fig. 4D). The latter was more UV sensitive, as would be expected if DinI of the host and DinL of the prophage work together either to reduce the SOS response or to augment the recombinational repair, or both. However, our efforts to distinguish the effects of these two mutations were not successful, as the effects of the single mutants were mostly not statistically significant (see Fig. S2 in the supplemental material). Thus, under the conditions we tried, the dinI-dinL status of the lysogen contributes little to the lytic infection in response to exogenous DNA damage.

We noticed that during growth of a lysogen with no exogenous DNA damage, a dinL mutation in the prophage reduced the free phage titer severalfold (see Fig. S3 in the supplemental material). We reported before that the recA mutation decreases Lula/phi80 titers in the lysogen supernatants by almost five orders of magnitude (75), which is understood in terms of phi80 repressor cleavage catalyzed by RecA filaments (28). Now we found that the lexA3 mutation, which renders the host SOS response noninducible (54, 65), actually increases supernatant titer of wild-type Lula/phi80 by at least one order of magnitude (see Fig. S3 in the supplemental material). This increased titer in the lexA3 host is again decreased severalfold if the Lula/phi80 lysogen carries the ΔdinL mutation (see Fig. S3 in the supplemental material). We conclude that (i) the functional dinL⁺ gene elevates phage output of the lysogen in response to endogenous (background) DNA damage and (ii) Lula/phi80 has at least one more LexA-sensitive gene, with the product of this gene inhibiting lytic induction. This could be the cI repressor itself, reacting to the uncleavable LexA due to the LexA inhibition of its synthesis proposed above, or gene 26 running antisense to dinL.

Mutagenesis of the nonessential genes.

To isolate Lula/phi80 mutants that would not be as sensitive to UV as lysogens of the wild-type phage, we mutagenized a Lula/phi80 lysogen with the pRL27 insertional kan cassette (49), pooled the library of independent inserts, isolated the total phage from the supernatant of the overnight culture of this pool, used it to infect a nonlysogen, and plated for individual kanamycin-resistant colonies. Pools of these Kan-resistant lysogens were plated and UV irradiated with doses of 40 or 80 J/m², but all the rare survivors showed wild-type (high) sensitivity to UV.

Although we failed to isolate Lula/phi80 mutants with defective reaction to DNA damage, we realized that, by insisting that mutant phages complete the full infection cycle (first lytic induction and then phage production, a new infection, and establishment of a new lysogeny), we had isolated Lula/phi80 mutants with mutations in nonessential functions (called “accessory genes” in lambda [19]). Sequencing five of these mutants confirmed our reasoning. Two of the five were different hits in the same gene (gene 20, coding for a putative protein), bracketed by the tail assembly gene I (19) and the tail fiber gene J (21), one insert was at the very C terminus of gene 9 encoding the head-tail joining protein FII, one landed in the long intergenic region between Q and S (57 and 58), and the last one was in gene 63, again coding for a hypothetical protein. Thus, pRL27 mutagenesis of a Lula/phi80 lysogen, with the subsequent isolation of phage and relysogenization, can be used to isolate mutants with mutations in the nonessential Lula/phi80 functions.

GamL and its modifier Agt.

Gene 39 demanded attention because it caused the 569-nt-long gap in the Lula/phi80 genome contig (with otherwise 10-fold coverage on both sides), apparently coding for a toxic protein. Cloning of gene 39 confirmed that its expression noticeably inhibits wild-type E. coli cells and is quite toxic for recA mutant cells, especially if overexpressed (Fig. 5A). This inhibition was the likely reason why gene 39 caused the gap in the Lula/phi80 sequencing contig, as our sequencing approach entailed cloning random genome pieces in a multicopy plasmid and recovering the plasmid library in a recA mutant.

Fig 5.

Lula/phi80 genes 39 and 40, coding for GamL and its modifier Agt. (A) Overexpressed gene 39 is toxic to cells, especially if they are recA mutants. Strains: recA⁺, ER119; recA⁻, DH5α Z1. (B) Toxicity of the overexpressed gene 39 is alleviated by overexpression of gene 40, which has no effect on its own. The cells are all recA mutants (DH5α Z1). Plasmids: vector, pLAC11; 40, pER17; 39, pER18; 39 + 40, pER19. (C) Expression of gene 39, alone or together with gene 40, makes E. coli cells ExoV⁻ phenocopies, as shown by spotting of T4 2 mutant phages as a test for ExoV status of E. coli. Top, wild-type AB1157 cells are ExoV⁺ and therefore are nonpermissive for the T4 2 mutant, while the recBC mutants are ExoV⁻ and therefore allow T4 2 mutant plating. Bottom, plasmids are as in panel B. The lawn is wild-type strain ALS225. (D) Expression of GamL or Agt, alone or together, has little influence on the recombinational repair capacity of E. coli, as tested by sensitivity to UV. Plasmids are as in panel B, while the strain is ALS225.

Toxin genes are usually accompanied by cognate antitoxin genes, to provide the organism with immunity from toxicity. Indeed, gene 40, just upstream of gene 39, apparently codes for such an antitoxin, because when the two genes are cloned together, the inhibition by gene 39 is significantly relieved, even in recA mutant cells (Fig. 5B). Although true toxin-antitoxin pairs are ubiquitous and recognizable (89, 92), a BLAST search failed to provide a clue to the possible functions of genes 39 and 40, since the only homology found was to other conserved phage proteins of unknown functions. Toxin-antitoxin pairs are typically functions that contribute to the stability of plasmids (89), but since the Lula/phi80 prophage resides in the chromosome, rather than being a free plasmid, like its N15 relative (73), the utility of such a pair to Lula/phi80, even if indeed confirmed at the protein level as a stable toxin and unstable antitoxin, is not apparent. It could contribute to the development of persisters (50) of its lysogenic host, which would increase the chances of population survival after lethal damage to growing cells. Recently, type III toxin-antitoxin pairs that work to prevent infection of the cell by lytic phages were described (33), giving one more rationale for a toxin-antitoxin system benefit to a temperate phage.

A potential function of gene 39 was suggested by its position in the Lula/phi80 genome just upstream of recT, which codes for strand annealing/exchange homologous recombination activity. The analogous position in the phage lambda genome, upstream of the lambda strand annealing/exchange homologous recombination activity bet, is occupied by gam (22, 23), coding for an inhibitor of the host RecBCD helicase/nuclease (AKA ExoV) (43); a similar gene arrangement was proposed for phi80, based on genetic studies (64). In E. coli cells, the RecBCD enzyme has the highest affinity to double-stranded DNA ends and degrades linear DNA starting from such ends with the highest known rate for linear DNA degradation, at least 1 kbp/s (26). Therefore, phages with linear genomes protect their DNA from this host enzyme, either by inhibiting RecBCD directly or by capping double-strand ends with special proteins (3, 34, 78). In phage T4, the double-strand end-capping protein is the product of gene 2; as a result, T4 2 mutant phages cannot plate on wild-type E. coli but can plate if the RecBCD nuclease/helicase is mutationally inactivated (the ExoV⁻ phenotype of recBC mutants) (85) (Fig. 5C, top). Although the Lula/phi80 gp39 and lambda Gam proteins show no relatedness, plating of T4 2 mutant phage confirmed that gp39 indeed inhibits the host RecBCD exonuclease, in both the presence and the absence of gene 40 (Fig. 5C bottom). Therefore, we gave gene 39 the second name “gamL” to reflect its likely in vivo function in Lula/phi80. We also gave gene 40 the mnemonic name “agt” (alleviates GamL toxicity). Interestingly, lambda Gam exists in a long version and a short version, with the short version being quite toxic for E. coli cells if overexpressed (81), so in the case of lambda Gam it can be argued that Gam and Agt functions have been fused in a single polypeptide.

Lambda Gam makes a significant contribution to phage lambda productivity during lytic infection in the RecBCD⁺ hosts, as the yield of gam mutant infections is only 40% of that of gam⁺ infections (30). Because of this consideration, measuring the levels of free phage titers in cultures of ΔgamL Lula/phi80 lysogens could be misleading, reflecting mostly the decreased productivity of lytic development rather than a less frequent induction. Therefore, instead of deleting gamL, we decided to look at the nature of the RecBCD defect that GamL induces. There are two types of recBCD-defective mutants of E. coli: the recD mutants are reduced for linear DNA degradation power but still retain all recombinational repair capabilities of the wild-type strain, while the recBC mutants are deficient in both linear DNA degradation and recombinational repair (4). For example, expression of lambda Gam in E. coli cells sensitizes them to UV radiation (34, 66), suggesting inhibition of recombinational repair. To test whether expression of GamL turns cells into recD mutant or recBC mutant phenocopies, we tested the UV sensitivity of cells expressing either GamL or Agt alone or the two together. We found that expression of GamL alone slightly reduces the cells' resistance to UV, whereas expression of Agt alone or of GamL and Agt together is inconsequential for UV resistance (Fig. 5D), indicating normal recombinational repair capability and, therefore, the RecBC⁺ status of all combinations. We conclude that expression of GamL modifies the host RecBCD to inhibit linear DNA degradation but not the recombinational repair functions of the enzyme, effectively turning E. coli cells into recD mutant phenocopies.

gamL and agt are expressed in Lula lysogens.

“Lysogenic conversion” is the change in the host cell phenotype as a consequence of lysogenization with a prophage. Immunity to infection by self or by phages with the same immunity region is an example of lysogenic conversion, afforded by the prophage repressor (cI, in case of lambdoid phages). Directly downstream of the cI repressor gene of phage lambda are the rexAB genes, which allow lambda prophage to kill the cell upon lytic development of other phages (69). This “other lytic phage exclusion” function is also typical of lysogenic conversion, except that in case of lambda, wild-type T4 already developed an antidote to it, in the form of its rII cistron (83). Interestingly, just downstream of its cI gene, Lula/phi80 has a single gene of unknown function (gene 46), historically known as gene 40AB (67), with a BLAST-identified “acetate kinase superfamily domain,” which could be Lula/phi80's “other lytic phage exclusion” function.

A gene 2 mutant of T4 plates equally poorly on nonlysogens and lambda lysogens, indicating no expression of the lambda gam gene from lambda prophage (62). In contrast to lambda lysogens, Lula/phi80 lysogens allow T4 2 mutants to plate (Fig. 6A), suggesting that Gam (plus Agt) is produced by the prophage as well. We also found that Lula/phi80 lysogens exclude wild-type T4 (Fig. 6A), perhaps by gp46. The curious fact that Lula/phi80 lysogens exclude wild-type T4 yet allow plating of a T4 2 mutant, which is defective in the pilot protein that protects the ends of the T4 linear chromosome (52), not only means that Lula/phi80 prophage inactivates ExoV activity of the host RecBCD enzyme to allow T4 2 development but also suggests that the “other lytic phage exclusion” function of Lula/phi80 recognizes the T4 gp2 protein and somehow kills the cell that expresses this specific protein. In any case, ExoV inactivation is a part of the Lula/phi80-induced lysogenic conversion.

Fig 6.

The Lula/phi80 lysogen is an ExoV⁻ SbcCD⁻ phenocopy. (A) The Lula/phi80 lysogen allows T4 2 mutant plating but restricts T4 WT phage. Strains: WT nonlysogen, AB1157; WT lysogen, EL103; recBC nonlysogen, SK129; recBC lysogen, ER114. (B) Gene 39 (Lula/phi80 gamL), is the only gene required for plating of lambda-pal. On the right is a scheme of the truncated phi80 DNA in DH5α and the constructs we built. Lawns were made with the following strains: (i) recBC(Ts), SK129; (ii) DH5α Δ(attP-46), L-264; (iii) L-264/pTOPO:2; (iv) L-264/pER17; (v) L-264/pER18; and (vi) L-264/pER19. The phage is lambda-pal, MMS1632. (C) Titer of lambda-pal on a recA mutant in the AB1157 background, carrying various additional mutations or ExoV-inactivating plasmids. Strains according to bar number on the right: 1. DH5α; 2, DH5α/pER19; 3, JC10287; 4, AK10; 5, L-293; 6, JC10287 Lula/phi80 lysogen; 7, JC10287/pER19; 8, JC10287/pKM590; 9, JC10287/pKM590.

There is genetic evidence that, besides the ExoV activity of RecBCD, lambda Gam also inactivates the host SbcCD enzyme (47), a relative of eukaryotic SMC proteins (82) and a helicase/nuclease (16) that attacks hairpins formed by inverted palindromes (15). In the classic E. coli K-12 AB1157 background (5), either a recD mutation (ExoV⁻ phenotype) or a sbcCD mutation (as was shown before [11]), as well as Lula/phi80 prophage (as we show now), all allow full-titer plating of a lambda red gam mutant (deficient both in the phage-encoded homologous recombination system and in inhibition of the host ExoV) carrying a large palindrome (lambda-pal) (Table 2, top four rows). Since we knew that the Lula/phi80 prophage makes AB1157 strain ExoV⁻, we wondered if the lysogens could also become sbcCD mutant phenocopies. In an unrelated line of research, we observed that a recA⁺ derivative of the standard BRL-developed DH5α (6), which is genetically recD⁺ sbcCD⁺ but carries a truncated phi80 prophage (27), nevertheless plates lambda-pal (Table 2, strain L-41). At the same time, DH5α recA⁺ is not an ExoV⁻ phenocopy (Table 2, L-41), suggesting that lambda-pal permissivity is due to an sbcCD mutant phenocopy of the strain, which we confirmed (Table 2, compare strains L-41 and L-243). This SbcCD⁻ phenotype was due to the truncated phi80 prophage, as deleting it made the strain restrict lambda-pal plating, but introducing the sbcCD mutation in the latter strain restored the plating (Table 2, strains L-41, L-262, and L-263). Finally, we found that to produce the lambda-pal-nonpermissive phenotype, only 10 kbp between attP and gene 46 in the truncated phi80 has to be removed (Table 2 [strain L-264] and Fig. 6B). The deletion strain starts plating lambda-pal again if genes 33 to 40 of Lula/phi80 are introduced back (4.4 kbp on plasmid TOPO:2) (Fig. 6B) or if GamL, alone or in combination with Agt (but not Agt alone) is produced off a plasmid (Table 2 [two bottom rows] and Fig. 6B). All the observations with strain L-41 supported the idea that GamL of Lula/phi80 inhibits the SbcCD enzyme.

Table 2.

Plating of T4 2 mutant (for ExoV status) and lambda-pal (for SbcCD status) on strains with the recD or sbcCD mutations or carrying various Lula/phi80 sequences^a

Background	Strain	Relevant genotypeb	ExoV phenotype	Lambda-pal titer
AB1157	AB1157	WT	+	106
	AK74	recD	−	1010
	L-310	sbcCD	+	1010
	AB1157(Lula)	phi80+	−	1010
DH5α recA+	L-41	phi80t	+	1010
	L-243	phi80t sbcCD	+	1010
	L-262	Δphi80t	NDc	107
	L-263	Δphi80t sbcCD	+	1010
	L-264	Δ(attP-46) phi80t	+	107
	L-264/pER17	Δ(attP-46) phi80t Agt+++	ND	107
	L-264/pEAK73	Δ(attP-46) phi80t GamL+++	−	1010

^aPalindrome phage, MMS1632 pal571 cI1857 b1453 k + 76. Control phage DKC14 plated with equal efficiency in these experiments.

^bphi80t, truncated phi80, as in DH10B (27).

^cND, not done.

Because red gam lambda mutants cannot plate on recA mutant hosts (30, 95), the original DH5α strain, which is a recA mutant, does not plate lambda-pal (Fig. 6C, bar 1) but becomes permissive for lambda-pal if the GamL and Agt proteins of Lula/phi80 are overproduced (Fig. 6C, bar 2). The interplay of recD and sbcCD mutant phenotypes in palindrome permissivity could be resolved in a recA mutant background, as was first shown for the phi80-free E. coli K-12 by Chalker et al. (11). We confirmed their findings in our recA mutant nonlysogen: the original recA mutant does not plate lambda-pal (Fig. 6C, bar 3), a recA recD double mutant plates it poorly (Fig. 6C, bar 4), and the addition of the sbcCD mutation allows normal plating of the palindrome-containing phage, as we show for a Lula/phi80 lysogen of this recA mutant (Fig. 6C, bars 5 and 6). This lambda-pal-permissive phenotype of the Lula/phi80 lysogen in the recA mutant background is reproduced by expression of only two Lula/phi80 proteins, GamL and Agt, in this recA mutant (Fig. 6C, bar 7); we also show that the same phenotype is produced by overexpression of the Gam protein of phage lambda (Fig. 6C, bars 8 and 9), which is consistent with previous genetic data (47). We conclude that the Lula/phi80 lysogen makes recA mutant cells both recD and sbcCD phenocopies, while the truncated phi80 lysogen in DH5α makes cells only sbcCD mutant phenocopies but not ExoV⁻ ones (they plate lambda-pal only if made RecA⁺).

Overall, we found that Lula/phi80 lysogens are both recD and sbcCD mutant phenocopies, due to expression of a single Lula/phi80 gene, gamL, while expression of the nearby agt makes expression of gamL less toxic for the host. Since GamL and Agt are produced during lysogeny, they are candidates for activities that underlie Lula/phi80's high lysogen output, although how this may happen mechanistically is unclear at this point.

DISCUSSION

The Lula prophage was found to contaminate a significant fraction of E. coli laboratory strains, stealthily spreading in the laboratory environment—a highly unusual feat (75). In fact, Lula contamination is so pervasive that at the time of its initial sequence analysis, its tail fiber protein gene (gene 21) returned 99% identity to two human loci (LOC338829 and LOC392563; records of these human genes were later discontinued). While characterizing Lula's physiology, we found at least two of its features that contribute to this phage's ability to breach human-imposed barriers to cross-contamination: cryptic lysogen productivity and stealthy infectivity (see the introduction) (75). Since nucleotide sequences of a few Lula restriction fragments revealed no homologies in the databases current at that time, we sequenced the genome of this new temperate phage in order to seek explanations for Lula's cross-contaminating powers in its genome structure/content. The whole 46,150-bp sequence of the Lula genome did not match anything in the database, but turned out to be a 100% match to the unpublished genome of phi80, a lambdoid phage from the 1960s and 1970s (77) that, apparently, “decided to take dwelling in the lab” rather than to fall into oblivion. The overall layout of the Lula/phi80 genome turned out to be very much lambda-like, even though the limited homology between the two genomes (∼70% nucleotide identity) was mostly concentrated in the leftmost 1/3 of the chromosome, in the morphogenesis region (Fig. 1), consistent with the earlier physical analysis (32) and confirming a general pattern of genome relatedness between several lambdoid phages (86). Especially striking is the similarity of the control region layout, further emphasized by there being no sequence relatedness between the control genes themselves (Fig. 2). Not only is the overall layout of the multiple important elements of the region essentially the same, but the general dimensions, including the sizes of the analogous regulatory genes and the distances between various regulatory sites, are carefully preserved. The only explanation for this remarkable layout conservation that we can think of is that in vivo the whole region functions as a nucleoprotein complex of a particular structure, and this particular structure is critical for its functionality, while the actual sequences of either DNA sites or proteins that interact with these sites are not important. It was noted before on several occasions that lambdoid phages, while sharing only limited sequence homology, have the same genome structure (7, 12, 41); the general colinearity of the lambda and Lula/phi80 genetic maps was specifically noted (77).

Since none of the 66 proposed genes of Lula/phi80 offered an obvious explanation for the phage's stealthy infectivity or cryptic lysogen productivity, we further investigated three phage genes, damL, dinL, and gamL, whose products could have played a controlling function in the DNA metabolism of the phage or the host cell. Deletion of damL does not make Lula/phi80's virion DNA cuttable, but it increases the phage titer in saturated cultures of the corresponding lysogen severalfold, making DamL a negative factor in the phage's spontaneous lytic induction. Deletion of the SOS-regulated dinL does not make Lula/phi80 lysogens more UV resistant and has no influence on either the released phage titer or the decrease in cell titer after UV induction, although it does decrease severalfold the spontaneous phage titer in lysogen cultures, making DinL a positive factor. Finally, GamL acts to reduce ExoV activity of the host RecBCD nuclease/helicase without affecting the host's recombinational repair capability and also inhibits the host SbcCD enzyme, allowing propagation of phages carrying large palindromes. At the same time, the Lula/phi80 genome does not have palindromes of its own. If lambda's Gam mutant is considered an example, inactivation of GamL should have significantly reduced the titer of Lula/phi80 in cultures of lysogens, but that would be explained by GamL's important role in normal lytic development (phage DNA replication is more productive in the absence of ExoV). The production of the GamL protein is inhibitory for the host cells without coproduction of its neighbor Agt and kills the recA mutant bacteria. Surprisingly, GamL and Agt are also expressed in lysogens (maybe because Lula/phi80 has only two terminators downstream of N, while lambda has five [Fig. 2]), so the RecBCD and SbcCD inactivation is a part of Lula/phi80 lysogenic conversion (see below).

There are two types of Gam proteins known: those that bind and modify the RecBCD (and SbcCD) enzymes directly and those that bind double-stranded DNA ends, thus denying access to the RecBCD enzyme. Phage lambda Gam protein is the prototype of the former, mimicking in its structure a piece of duplex DNA (20), while the Gam protein of phage Mu, a distant relative of the eukaryotic Ku DNA end-binding proteins (21) is the prototype of the latter (1, 3). Since expression of GamL does not sensitize E. coli to DNA damage, it is likely that GamL acts by binding to and modifying the enzymes rather than making DNA ends inaccessible to DNA repair functions.

Finally, even though gene 30, coding for invertase Pin, could be responsible for the original inversion of the integrase-containing region in the Lula/phi80 genome relative to the lambda genome, we did not experiment with it, mostly because there is no evidence for continuing periodic inversions in the Lula/phi80 genome and no obvious candidates for a pair of Pin recognition sequences. We suspect that one of the Pin recognition sites has been mutated/deleted, and so Pin cannot catalyze inversion in Lula/phi80 anymore. A candidate for the remaining Pin site is the palindrome in front of the pin gene (Table 1). It is also possible that Pin catalyzes resolution of Lula/phi80 circular dimers/multimers, for which a single site per genome will suffice.

Genes behind Lula/phi80's ability to survive in the laboratory.

As was already mentioned, Lula/phi80's silent and efficient spread in the laboratory environment depends on at least two factors: its cryptic lysogen productivity and its stealthy infectivity (75). Cryptic lysogen productivity depends on Lula/phi80's opportunistic induction and its stability in laboratory protocols. Stealthy infectivity is due to the known temperature sensitivity of Lula/phi80 vegetative development and the much higher frequency of lysogenization compared with that of lambda. We will try to relate these properties to Lula/phi80's specific genome determinants.

We argued before (75) that Lula/phi80's cryptic lysogen productivity during normal growth, compared to that of other prophages, is the result of its increased prophage induction in response to spontaneous chromosomal damage—what can be called “opportunistic induction.” Whereas other prophages are not induced by spontaneous chromosomal damage events and are induced only by massive exogenous DNA damage and only after a significant delay, Lula/phi80 reacts instantly to any but the lowest doses of exogenous DNA damage and, apparently, goes lytic even from a small fraction of endogenous chromosomal damage events, such as the simultaneous collapse of two or more replication forks in a cell. After finding an SOS-inducible gene in the Lula/phi80 genome, dinL, we suspected it to be the “DNA damage sensor” of the phage, but deletion of dinL did not change Lula/phi80's “overreaction” to exogenous DNA damage, lowering the supernatant titer of the phage only severalfold. In the absence of any other apparent SOS-inducible gene, the major DNA damage sensor may be Lula's cI repressor itself, which is known to be the easiest repressor for activated RecA to cleave in vitro out of several lambdoid phage repressors, as it is still efficiently cleaved by the RecA430 mutant protein, which cannot even cleave the LexA repressor (25, 28). In combination with the suspected LexA repression of the cI production in uninduced cells, this may make Lula/phi80's cI the main DNA damage sensor of the prophage. If this sensor is set up to react even to a fraction of spontaneous DNA damage events, this will contribute to the high supernatant phage titer in the cultures of Lula/phi80 lysogens.

The second important contributor to the cryptic lysogen productivity of Lula/phi80 lysogens is the ability of the phage to survive in lysogen cultures of any density or age. The titer of infectious lambda phage rapidly drops in saturated cultures of both nonlysogens and (lambda) lysogens, but the Lula/phi80 titer is stable under the same conditions (75). This is likely a reflection of Lula/phi80's known ability to infect only metabolically active cells, which are capable of energy-driven enzymatic reactions in the cell envelope (36). Curiously, the only other well-known phage that infects only metabolically active cells is T1 (36, 88); this is in contrast to the case for most other phages, which are known to inject their DNA into stationary-phase or heat-inactivated cells, through cell membranes of lysed cells, or even through artificial membrane-like surfaces (88). The nature of this energy-dependent irreversible step in the case of T1 and Lula/phi80 is postulated to be formation of a covalent bond between the tip of the phage tail and the phage receptor (36, 88), which for both phages happens to be the outer membrane receptor FhuA (TonA) for transport of hydroxamate-type siderophores (29, 36, 42). The specific T1 and Lula/phi80 virion proteins that bind to FhuA or catalyze the formation of the bond are not known. Remarkably, homology between the two phages is limited to only three genes (22, 23, and 24 in Lula/phi80 [Table 1]) in the tail fiber region of the genome. In fact, these three genes are shared between all phages (HK022, N15, ES18, T1, and Lula/phi80) that are known to bind FhuA (91). Moreover, gene 24 is known to code for superinfection exclusion lipoprotein Cor (the postinfection plug for FhuA) (90), so we propose that the phage protein that enzymatically connects with the FhuA receptor is either gp22 (a 100-amino-acid [aa] protein with subdomains related to multidrug resistance proteins and to dehydrogenases) or gp23 (a 224-aa protein related to O-acetylhomoserine sulfhydrylases and cystathionine gamma-synthases). These may be the genetic determinants of Lula/phi80 preservation in cultures.

The stealthy infectivity of Lula/phi80 under standard laboratory conditions is due to at least two features. The first is its much lower optimal temperature for vegetative development (30°C, compared with lambda's 40°C). In fact, Lula/phi80 development at 42°C is so poor that this temperature can be used for transduction from lysogens if reinfection with Lula/phi80 is to be avoided (75, 84). Curiously, this temperature sensitivity of Lula/phi80 contributes to its much longer infection cycle at 37°C compared with lambda (25, 75), explaining the difficulty of detecting Lula/phi80's lytic development in contaminated cultures grown at the optimal E. coli temperatures (which are in the 37 to 40°C range [40]). The temperature-sensitive components of Lula/phi80 are not known, but they are unlikely to reside in the 1/4 of the chromosome transcribed from right to left (gray arrows in Fig. 1), as these genes are transcribed with the same (immediate/constant) temporal pattern during both lambda and Lula/phi80 infections (58), and this transcription from the pL promoter is much higher at 42°C than at 35°C (2). Not only do the genes transcribed from the opposite strand (black arrows in Fig. 1) show a different temporal transcription pattern in both phages (delayed initiation followed by 100-fold induction), but the timing of this induction at 37°C is two times longer in Lula/phi80 than in lambda (58), indicating that the temperature sensitivity of Lula/phi80 development starts at the level of the rightward transcription and ultimately results in late mRNA production. Indeed, both the lysis genes and the morphogenesis genes in Lula/phi80 are transcribed much more poorly at 42°C than at 35°C (2). However, the defect in the rightward transcription seems to start at pR, rather than at pR′, as even initiation of DNA replication is delayed at 42°C relative to 35°C by some 20 min (2). Thus, the temperature sensitivity of the rightward transcription in Lula/phi80 may be due to failure of N antitermination, rather than Q antitermination. Intriguingly, the temperature sensitivity of phi80 is even more complex, as studies with lambda-phi80 hybrids show that there are at least two additional heat-sensitive determinants in both the head and tail morphogenesis regions, in homologs of the A and L genes (lambda nomenclature) (77, 80). Why so many stages of Lula/phi80 development are blocked or delayed at higher temperatures is a mystery.

The second feature contributing to stealthy infectivity is Lula/phi80's much higher frequency of lysogeny than lambda: we estimate it at 1/10, with lambda's at 1/1,000, at 37°C (75). That is, 9/10 of Lula/phi80 infections still result in cell lysis, but the survival of 10% lysogenic cells makes the ongoing infection much less conspicuous, especially on plates (75). This feature is definitely defined by the Lula/phi80′ control circuit, but comparison with the control region of lambda does not yield any differences that could provide a clue (Fig. 2). If lambda's control circuit is the model, then the lysis/lysogeny balance should depend mostly on protein cII (the activator of cI repressor synthesis) or cIII (the inhibitor of FtsH protease, which degrades cII) (68). The apparent absence of cIII in the current Lula/phi80 annotation is due to our inability to identify which of the several small ORFs between kil and N (see the supplemental material) encodes this genetically demonstrated function (77). Since the rightward transcription from pR is proposed to be temperature sensitive in Lula/phi80 (see above), one needs also to measure the ratio of lysis to lysogeny at different temperatures, if only to test for the possibility that the preference for lysogeny is observed only at higher temperatures and is a simple reflection of inhibition of the rightward transcription at these temperatures.