Quorum Sensing and Metabolic State of the Host Control Lysogeny-Lysis Switch of Bacteriophage T1

The dynamics of microbial communities are heavily shaped by bacterium-bacteriophage interactions. But despite the apparent importance of bacteriophages, our understanding of the mechanisms controlling phage dynamics in bacterial populations, and particularly of the differences between the decisions that are made in the dormant lysogenic and active lytic states, remains limited. In this report, we show that enterobacterial phage T1, previously described as a lytic phage, is able to undergo lysogeny. We further demonstrate that the lysogeny-to-lysis decision occurs in response to changes in the density of the bacterial population, mediated by interspecies quorum-sensing signal AI-2, and in the metabolic state of the cell, mediated by cAMP receptor protein. We hypothesize that this strategy enables the phage to maximize its chances of self-amplification and spreading in bacterial population upon induction of the lytic cycle and that it might be common in phage-host interactions.


E. coli ATCC 14155 carries an AI-2-and glucose-inducible prophage.
In coculturing different E. coli isolates with the laboratory E. coli strain W3110 in liquid tryptone broth (TB), we observed that E. coli ATCC 15144 reproducibly lysed after 2 to 3 h in the coculture, although it grew normally in the monoculture (Fig. 1A). Such lysis could also be induced by addition of cell-free supernatants of W3110 to the growing cultures of E. coli ATCC 15144. Because lysis occurred in most but not all cultures in the latter case, we hypothesized that it might have been caused by prophage induction. Indeed, microscopic analysis of the lysates revealed the presence of virion particles (Fig. 1B).
The observed lysis of ATCC 15144 in the presence of W3110 or its supernatant indicated induction by a W3110-secreted signaling molecule. Since AI-2 is the only established quorum-sensing molecule of E. coli, we tested whether lysis occurs in the presence of E. coli W3110 cells deleted for luxS, which encodes AI-2 synthase (26). No lysis was observed in this case or upon addition of W3110 ΔluxS culture supernatant (Fig. 1A), strongly suggesting involvement of AI-2 in prophage induction.
To further verify this hypothesis, we assessed whether synthetic (S)-4,5-dihydroxy-2,3-pentadione (DPD) would be able to induce the prophage in a similar manner. Since DPD molecules spontaneously convert to AI-2 in solution (12,27), we refer to DPD here as DPD/AI-2. In analyzing the growth characteristics of a larger number (n ϭ 50) of independent E. coli ATCC 15144 cultures, we observed a low (ϳ12%) frequency of spontaneous prophage induction even in the absence of any further treatment (Fig. 1C). This lysis rate was dramatically enhanced (reaching ϳ70%) by addition of 30 M DPD/AI-2, a concentration which is in the range of those observed in E. coli cultures (28,29). Importantly, the growth of E. coli ATCC 15144 was not altered by DPD/AI-2 addition (Fig. 1C), and no growth in minimal medium with DPD/AI-2 as a sole carbon source could be observed (data not shown) (30). We thus conclude that AI-2 signaling, rather than its effect on growth, leads to prophage induction.
We observed that prophage could also be induced in E. coli ATCC 15144 cultures growing in TB by addition of glucose (Fig. 1C). Notably, 2-deoxy-D-glucose (2-DG), a nonmetabolizable analogue of glucose that is nevertheless imported and phosphorylated by the PTS (31,32), also induced cell lysis (Fig. 1C), indicating that the observed induction was most likely caused by the PTS signaling rather than by glucose metabolism. Generally, prophage induction and cell lysis in response to AI-2 and glucose occurred between 60 and 130 min after addition of the respective compounds to the growing culture (see Fig. S1A in the supplemental material). No lysis was observed before or after these time points during our experiments. Prophage induction resulted in the release of 8 ϫ 10 9 to 1.15 ϫ 10 10 phage particles per ml, with no apparent Means of results of a minimum of 12 independent replicates are shown; error bars represent standard deviations. P values were calculated using the Mann-Whitney test (****, P Ͻ 0.0001; ***, P Ͻ 0.0005; *, P Ͻ 0.05; ns, not significant). difference in phage titers between spontaneous or AI-2-containing supernatantmediated or glucose-mediated prophage inductions (Fig. S1B).
Interestingly, the AI-2-dependent prophage induction mechanism seemed to be sensitive to even a small elevation of the AI-2 concentration. While E. coli ATCC 15144 produces AI-2 ( Fig. S1C), its extracellular concentration is apparently not sufficient to induce lysis in most cultures in the mid-exponential phase of growth. However, a modestly higher level of AI-2 produced by the W3110 strain in the ATCC 15144:W3110 mixed cultures or addition of 5 M DPD/AI-2 to growing ATCC 15144 cultures was already enough to activate the lysogeny-lysis switch and cell lysis ( Fig. S1C and D). In contrast, increasing the initial ATCC 15144 inoculum size did not result in significant changes in extracellular AI-2 levels in growing cultures (Fig. S1E), most likely due to the balance of AI-2 production and uptake (12), and thus did not lead to increased phage release.
Phage identification. Sequencing the phage DNA revealed that it shares 99.86% identity with the enterobacterial phage T1. This was highly surprising, since T1 phage had been described as a classical example of a virulent phage (33)(34)(35). Nevertheless, we could invariably detect T1 phage DNA stably propagating within the host E. coli cells without lysis (Fig. S2A). This lysogeny was apparently established in the absence of integration into the bacterial chromosome, as revealed by genome analysis of E. coli ATCC 14155 by the use of both nucleotide BLAST searches against the GenBank database and the phage sequence-seeking PHASTER tool (see Materials and Methods for details). Consistently, no bacterial genome sequences flanking T1 phage genome were observed. Despite its extrachromosomal residence, the prophage is highly stable and ATCC 15144 could not be cured despite repeated sequential restreaking on lysogeny broth (LB) agar plates.
Our analysis of the phage genome identified at least three genes that might be associated with the lysogenic lifestyle. Two of these genes, orf23 and orf65, code for putative transcription regulators. A third gene, orf30, encodes a homolog of Cor, which is involved in lysogeny of N15 phage, leading to surface exclusion of several bacteriophages, including T1 (20). Homologs of Cor are also found in several other lysogenic phages (36,37). Consistent with its expected function, overexpression of Cor resulted in resistance to phage infection in an otherwise T1-susceptible host, E. coli ECOR-4 ( Fig. S3).
Apart from ECOR-4, T1 phage was able to infect several other tested E. coli strains, including ECOR-13, ECOR-16, ECOR-50, and ATCC 11303. In all cases, we observed bacterial colonies appearing in the plaque zones formed by the lytic activity of the phage (Fig. S2B). But although T1 phage DNA was initially detected in these colonies, indicating induction of the lysogenic state, these E. coli strains apparently could not support stable lysogeny as the phage DNA was invariably lost during cultivation in liquid TB medium ( Fig. S2C and D).
In contrast, although the genome of E. coli ATCC 15144 is nearly identical to that of the common laboratory strain BL21(DE3), the latter could not be infected by the phage. The reason for this selectivity remains to be elucidated, but ATCC 15144 carries several open reading frames (ORFs) absent in BL21(DE3), including ORFs encoding remnants of prophages, an IS3 family transposase, adhesins, and parts of secretion systems (see Table S2 in the supplemental material).
Phage-encoded transcription regulator Pir controls AI-2-and PTS-dependent prophage induction. To verify that orf23 encodes a functional transcription regulator-here renamed Pir (for "prophage induction regulator")-we expressed Pir from an IPTG (isopropyl-␤-D-thiogalactopyranoside)-inducible plasmid in E. coli W3110 strains carrying reporter plasmids containing genes encoding green fluorescent protein (GFP) under the control of several phage promoters. Indeed, we observed that Pir activates promoters of hol (orf13, coding for holin) and dam (orf20, coding for Dam methylase) while negatively regulating its own promoter and the promoter of recE (orf29, coding for putative exodeoxyribonuclease VIII) ( Fig. 2A). Since holin is required for cell wall degradation at the end of the lytic cycle (38,39), the positive regulation of hol by Pir indicated an activating role of Pir in the lysogenic-to-lytic transition. Consistent with this role, activities of both the pir promoter and the hol promoter were upregulated during DPD/AI-2-and PTS-mediated prophage induction in E. coli ATCC 15144 (Fig. 2B). Both promoters were also induced by glucose, but not by AI-2, in the phage-free E. coli  Table S2). Changes in pir expression were also detected in the majority of T1-susceptible strains upon their lysogenization (Fig. S2E), but while pir promoter activity was downregulated in some lysogens, it was upregulated instead or did not change in other lysogenized strains. Thus, pir expression does not seem to be directly involved in establishing lysogeny of T1 phage.
To further confirm the function of Pir, we depleted pir mRNAs in the host E. coli ATCC 15144 strain with CRISPR interference (CRISPRi). CRISPRi allows sequence-specific inhibition of gene expression by blocking transcription with a deactivated Cas9 (40,41). In order to deplete pir mRNA, cells were cotransformed with two plasmids encoding the anhydrotetracycline (aTc)-inducible dCas9 protein (a mutation of Cas9 without endonuclease activity) and constitutively expressed single guide RNA (sgRNA) targeting the coding region of pir. To ensure the effective repression of pir transcription, dCas9 expression was induced in both overnight and day cultures of E. coli by addition of 5 ng/ml aTc. The resulting CRISPRi-mediated downregulation of pir expression (Fig. S4A) abolished induction of T1 prophage and cell lysis upon addition of AI-2 (in the form of W3110 cell-free supernatant) or glucose ( Fig. 2D; see also Fig. S4B and C). Consistently, no upregulation of hol expression was observed upon addition of W3110 supernatant (Fig. 2E), although the activity of the hol promoter was generally elevated in the presence of the CRISPRi system. This basal elevation was apparently due to the protein burden caused by dCas9 expression, which was also observed in the E. coli W3110 strain carrying no T1 prophage ( Fig. S4D and E). Despite its apparent role in the activation of lysis in response to AI-2, the activity of Pir did not seem to be directly affected by AI-2 (Fig. S5).
We additionally showed that Orf65 is a transcriptional regulator that represses pir while positively autoregulating its own expression, whereas the activity of orf65 promoter is independent of Pir ( Fig. S6A to C). No changes in orf65 expression were detected upon addition of W3110 supernatant or AI-2 (Fig. S6D) or during phage induction (data not shown). Thus, Orf65 is unlikely to play a role in the lysogeny-lysis transition, but pir repression by Orf65 might possibly stabilize T1 lysogeny.
PTS-mediated glucose uptake activates pir expression via CRP. Since elevated pir expression was seen upon addition of glucose even in the BL21(DE3) strain lacking the prophage (Fig. 2C), the underlying regulatory mechanism must be provided by the host. Given the dependence of its activity on the PTS signaling, we considered CRP to be a potential regulator. Indeed, abolishing cAMP production in the cell by deleting the adenylate cyclase gene (cyaA), which decouples CRP activity from the PTS-mediated sugar uptake, resulted in increased pir expression combined with loss of sensitivity to glucose (Fig. 3A). The effect could be partially complemented by supplementing the growth medium with cAMP. These results show that in the absence of glucose, CRP in its cAMP-bound state inhibited pir expression and that such inhibition was relieved in the presence of decreasing cAMP levels when glucose was taken up. The repression of the pir promoter by cAMP-bound CRP could be directly confirmed using an in vitro transcription-translation system (Fig. 3B). Notably, dam and recE promoters were also inhibited by cAMP-CRP (Fig. S7), suggesting a more global mechanism of CRP-mediated regulation of T1 phage genes.

DISCUSSION
Phages are known to employ various distinct strategies to ensure their effective propagation. These can be generally classified into lytic and lysogenic life cycles (13). Virulent (lytic) phages use their hosts for rapid propagation, releasing phage particles into the environment via lysis of the host. In contrast, temperate (lysogenic) phages replicate with their hosts until the extracellular conditions favor phage release and reinfection of the new hosts. Elaborate molecular machineries that control the initial decision between lysis and lysogeny upon host infection, as well as the subsequent lysogeny-lysis decision to enter the lytic cycle, have been described for temperate phages such as lambda (42). However, given the extremely high diversity of bacterio-phages in nature, the majority of regulatory mechanisms underlying lysogeny-lysis decisions likely have yet to be discovered.
Here, we describe a novel lysogenic state of T1 phage, which has been previously viewed as a typical virulent phage, showing that this phage can be stably propagated in the extrachromosomal state in E. coli host cells. Furthermore, our results suggest that this prophage state integrates information about bacterial population density and the metabolic state of the host to control its switch from lysogeny to lysis. This regulation apparently occurs at the level of a phage-encoded transcriptional regulator, Pir (Fig. 4). In the absence of external stimuli, a steady state of pir expression is ensured by the self-inhibitory regulation mediated by Pir, along with additional inhibition of its promoter by Orf65 and by cAMP-CRP. Increased sugar influx relieves CRP-mediated inhibition of pir expression, resulting in Pir-dependent activation of the lytic cycle. pir is similarly upregulated in the presence of AI-2, likely explaining the dependence of lysis on bacterial culture density. Although the precise mechanism of AI-2-mediated prophage induction remains unknown, our observation that transcription of pir is induced by AI-2 in prophage-carrying strain ATCC 15144 but not in prophage-free strain Since the concentration of quorum-sensing molecules is essentially a function of bacterial population density (43), it would be expected that lysogenic phages have evolved mechanisms to adopt this signal for their lysogeny-lysis decisions. Indeed, recent studies have shown that quorum sensing led to prophage induction in Vibrio cholerae and Enterococcus faecalis (24,25). In the case of V. cholerae, induction occurred in response to the presence of 3,5-dimethylpyrazin (DPO), an intraspecies signaling molecule, and was mediated by a homologue of the host quorum-sensing activator VqmA (24). In the case of E. faecalis, prophage induction mediated by AI-2 was observed, similarly to the case described here for E. coli, but the underlying regulatory circuit was not investigated (25).
Equally interesting is the ability of T1 phage to effectively respond to changes in host metabolism. Although the nutrition state is known to affect lysis-lysogeny decisions (15,42,(44)(45)(46), this is the first example of the phage being able to "eavesdrop" on the host's metabolic signaling system to decide whether the conditions are favorable for prophage induction. Given that Pir homologues are found in phages infecting not only E. coli but also Salmonella, Citrobacter, Klebsiella, Cronobacter, and Pantoea species (see Table S3 in the supplemental material), the mechanisms of prophage induction described in this report may be common among Enterobacteriaceae viruses.

MATERIALS AND METHODS
Bacterial strains and culture conditions. All strains and plasmids used in this study are listed in Table 1. E. coli strains were grown in liquid tryptone broth (TB) medium (10 g tryptone and 5 g NaCl per liter) or in lysogeny broth (LB) medium (10 g tryptone, 10 g NaCl, and 5 g yeast extract per liter) supplemented with antibiotics where necessary.
Plasmids. Genes of T1 phage were amplified from the purified T1 phage DNA using primers listed in Table S1 in the supplemental material. The PCR products were purified, digested with SacI and XbaI enzymes (NEB, USA), and ligated into pTrc99A expression vector. For construction of fluorescent reporters, promoter regions (up to 150 bp upstream and 15 bp downstream of the start codon) of genes of interest were amplified from the purified T1 phage DNA and cloned into pUA66 vector using XhoI and BamHI cloning sites.
CRISPR interference. CRISPR interference-mediated inhibition of pir (orf23) expression was performed as described previously (40) Table S1). The base pairing region was designed to target the pir gene on T1 phage DNA on the coding strand adjacent to protospacer motif AGG. dCas9 protein expression was induced with 5 ng/ml aTc in both overnight and day cultures.

Real-time quantitative PCR (RT-qPCR).
Total mRNA isolation and RT-qPCR for confirmation of CRISPRi-mediated pir mRNA depletion were carried out as described previously (47). A 200-ng volume of total mRNA was used in every reaction. Primers used for pir and cor mRNA detection in the samples are listed in Table S1.
In vitro transcription-translation assay. Regulation of pir, dam, and recE promoters controlling gfp expression mediated by CRP was assayed as described previously (48,49). Cell-free lysates were prepared by sonication from the E. coli MG1655 ΔcyaA Δcrp strain, in order to avoid contamination with cellular cAMP and CRP. CRP was purified from E. coli strain BL21(DE3) carrying plasmid pYC75 expressing CRP with a C-His6 tag under the control of a T5-lac promoter using Protino Ni-TED packed columns (Machery Nagel, Germany; catalogue number 745100.50). Purified protein was concentrated by filtration with an Amicon Ultra-0.5 centrifugal filter unit (Millipore, Germany; catalogue number UFC501008). The promoter assay reaction mixture (10 l) contained reaction buffer (10 mM magnesium glutamate; 10 mM ammonium glutamate; 130 mM potassium glutamate; 1.   (27) and glucose or 2-deoxy-D-glucose (Sigma-Aldrich, Germany) were added for final concentrations of 30 M and 0.2%, respectively, followed by further incubation at 37°C with shaking. As an alternative to DPD/AI-2, 10 l of E. coli W3110 cell-free supernatant was used. Prophage induction and the resulting cell lysis were visually observed after 60 to 130 min by the dramatic drop in the optical density (OD) of the test culture (OD at 600 nm [OD 600 ], approximately 0.1 to 0.2) compared to that of the control (OD 600 , approximately 0.6 to 0.9). The phage titer after prophage induction and E. coli culture lysis was determined as the PFU level per milliliter using the Gratia method (50).
Genomic DNA extraction, sequencing, and analysis. E. coli ATCC 15144 was grown overnight in 5 ml LB at 37°C with shaking. The cells were spun down (5 min, 4,000 rpm), and the genomic DNA was extracted using a NucleoSpin microbial DNA purification kit (Macherey-Nagel, Germany; catalogue number 740235.50). T1 phage DNA was extracted upon DPD/AI-2-induced prophage induction and cell lysis using a phage DNA isolation kit (Norgen Biotek Corp., Canada; catalogue number 46800).
DNA concentration was measured using a Qubit 4 fluorometer (Invitrogen, USA; catalogue number Q3326) and a Qubit double-stranded DNA (dsDNA) high-sensitivity (HS) assay kit (Invitrogen, USA; catalogue number Q322851). The DNA library for sequencing was prepared using a NEBNext Ultra II FS DNA library preparation kit for Illumina (NEB, USA; catalogue number 3E7805L) and NEBNext multiplex oligonucleotides for Illumina (96 Index Primers) (NEB, USA; catalogue number E6609S). Library quality was verified using an Agilent Technology 2100 Bioanalyzer (Agilent Technologies, USA) and an Agilent high-sensitivity DNA kit (Agilent Technologies, USA; catalogue number 5067-4626). DNA was sequenced with an Illumina MiniSeq system (Illumina, USA).
Quality controlled, trimmed reads were assembled de novo into sequences of contigs with Ͼ100-fold coverage (for the E. coli DSM 15144 genome) and into a single contig with Ͼ4,000-fold coverage (for T1 phage genome) using Geneious 12.0 (N 50 ϭ 114904) (51). The absence of T1 phage sequences in the bacterial genome was confirmed using nucleotide BLAST against the GenBank database and additionally by analyzing the sequence with PHASTER (Phage Search Tool Enhanced Release) (52). The complete T1 phage genome was analyzed for the presence of bacterial sequences flanking its terminal ends.
Transmission electron microscopy. Transmission electron microscopy (performed with a JEM-1400 transmission electron microscope [Jeol, USA]) was used to visualize phage particles in E. coli ATCC 15144 lysates as described previously (53).
Flow cytometry. Promoter activities of orf13 (hol), orf20 (dam), orf23 (pir), orf29 (recE), and orf65 were assayed using plasmid-based reporters containing the respective promoter regions fused to gfp (see the description of the molecular cloning techniques for more details). Bacterial cultures were grown in TB as described above (supplemented with IPTG or 5 ng/ml aTc where necessary) to an OD 600 of 0.6 or until the onset of cellular lysis during the prophage induction. The samples were then diluted 1:100 in tethering buffer (10 mM KH 2 PO 4 , 100 M EDTA, 1 M L-methionine, 10 mM sodium lactate, pH 7.0), and fluorescence was measured with a BD LSRFortessa SORP cell analyzer (BD Biosciences, Germany). AI-2 quantification in culture supernatants was performed as described previously using a biosensor strain deficient in AI-2 production (29).
Data availability. The contigs used in this work were deposited in GenBank under accession number PRJNA526015. T1 phage genome sequence data determined in this work were deposited in GenBank under accession number MN153797.

ACKNOWLEDGMENTS
We thank Anna Potapova for help with transmission electron microscopy experiments.
This work was supported by the Max Planck Society.