Analysis of SOS-Induced Spontaneous Prophage Induction in Corynebacterium glutamicum at the Single-Cell Level

Abstract

The genome of the Gram-positive soil bacterium Corynebacterium glutamicum ATCC 13032 contains three integrated prophage elements (CGP1 to -3). Recently, it was shown that the large lysogenic prophage CGP3 (∼187 kbp) is excised spontaneously in a small number of cells. In this study, we provide evidence that a spontaneously induced SOS response is partly responsible for the observed spontaneous CGP3 induction. Whereas previous studies focused mainly on the induction of prophages at the population level, we analyzed the spontaneous CGP3 induction at the single-cell level using promoters of phage genes (P_int2 and P_lysin) fused to reporter genes encoding fluorescent proteins. Flow-cytometric analysis revealed a spontaneous CGP3 activity in about 0.01 to 0.08% of the cells grown in standard minimal medium, which displayed a significantly reduced viability. A P_recA-eyfp promoter fusion revealed that a small fraction of C. glutamicum cells (∼0.2%) exhibited a spontaneous induction of the SOS response. Correlation of P_recA to the activity of downstream SOS genes (P_divS and P_recN) confirmed a bona fide induction of this stress response rather than stochastic gene expression. Interestingly, the reporter output of P_recA and CGP3 promoter fusions displayed a positive correlation at the single-cell level (ρ = 0.44 to 0.77). Furthermore, analysis of the P_recA-eyfp/P_int2-e2-crimson strain during growth revealed the highest percentage of spontaneous P_recA and P_int2 activity in the early exponential phase, when fast replication occurs. Based on these studies, we postulate that spontaneously occurring DNA damage induces the SOS response, which in turn triggers the induction of lysogenic prophages.

INTRODUCTION

Genome sequencing projects have revealed a large amount of prophage DNA in bacterial genomes. Although not all prophage DNA accounts for functional prophages, because it includes degenerated phage remnants, this DNA can have a marked impact on bacterial physiology (1). The biotechnological platform organism Corynebacterium glutamicum is a Gram-positive, biotin-auxotroph soil bacterium that is used for the industrial production of more than four million tons of l-glutamate and l-lysine per year (2, 3). As revealed by whole-genome sequencing, C. glutamicum ATCC 13032 possesses three prophages that are integrated into its genome (CGP1 to -3), of which CGP1 and CGP2 are probably degenerated phage remnants (4–6). Previous studies showed that the large prophage CGP3 (187 kb) retains the ability to be excised from the genome and exist as a circular DNA molecule. Interestingly, a small number of wild-type cells showed a much higher copy number of circular phage DNA per cell than the average of the population (5).

Recent studies in Shewanella oneidensis (7) and Streptococcus pneumoniae (8) have provided evidence that sacrificing a small number of cells by spontaneous prophage-induced lysis is beneficial to the remainder of the population. For these species, genomic DNA released into the extracellular space following cell lysis supports biofilm formation and maintenance (9, 10). Although these results shed new light on the spontaneous induction of prophages, the mechanisms governing this general microbiological phenomenon are poorly understood to date.

The best-studied model for prophage-host interactions is the Escherichia coli λ phage model. As far back as the 1950s, spontaneous induction of lysogenic phage λ was observed in E. coli cultures (11). Remarkably, a decrease in spontaneously induced prophages was shown in recombination-deficient E. coli strains (12). It remains unknown, however, whether these events are (i) random events caused by promoter noise or by the stochastic distribution of key regulatory components or (ii) the result of specific induction by intrinsic and/or extrinsic factors. The lysogenic state of phage λ is controlled by cI, the central repressor of phage genes (13). When the integrity of the genome is compromised (as indicated by lesions in the DNA), the protein RecA polymerizes along single-stranded DNA (ssDNA). This nucleoprotein filament of ssDNA and activated RecA* protein catalyzes the autocatalytic cleavage of the repressor LexA, which leads to the derepression of more than 40 SOS genes (SOS response) (14). The life cycle of lambdoid phages is linked to this regulatory pathway. The central repressor cI mimics the autocatalytic center of LexA and thus becomes cleaved upon the induction of the SOS response. Its binding to repressor sites within the prophage is alleviated, leading to an expression of genes responsible for the excision of phage DNA, virion assembly, and release of the infectious phage particles into the extracellular space (13).

In this study, we address the question of whether the spontaneous induction of the lysogenic prophage CGP3 in single C. glutamicum cells is linked to the spontaneous activation of the SOS response. The promoters of genes of the SOS pathway and those encoded by CGP3 were fused to the fluorescent reporter genes eyfp and e2-crimson to analyze the activity of the respective promoters under standard cultivation conditions. Single-cell analysis was performed using flow cytometry and an in-house developed polydimethylsiloxane (PDMS) microfluidic chip setup (15, 16) suitable for observing rare cellular events of interest. We observed a positive correlation between the spontaneous activation of the SOS response and the spontaneous induction of the prophage CGP3, and we postulate a bona fide activation of the SOS response as a prominent trigger leading to prophage excision.

MATERIALS AND METHODS

Bacterial strains, media, and growth conditions.

The bacterial strains used in this study are listed in Table 1. C. glutamicum ATCC 13032 was used as the wild-type strain; all strains were cultivated at 30°C. For growth experiments, a glycerin stock culture was streaked onto BHI (brain heart infusion; Difco, BD, Heidelberg, Germany) agar plates. Single colonies were used to inoculate 5 ml liquid BHI medium. After cultivation for 8 h, the preculture was used to inoculate 25 ml CGXII minimal medium (1:50) containing 4% glucose (wt/vol) as the carbon source (17). After growth overnight, fresh CGXII medium was inoculated to an optical density at 600 nm (OD₆₀₀) of 1 in 25 ml. For microtiter-scale cultivations, the Biolector microbioreactor system (m2p-labs, Heinsberg, Germany) was utilized (18) using an established protocol for C. glutamicum cultivation (16). Escherichia coli was cultivated in LB (lysogeny broth) medium and on LB agar plates at 37°C. If required, kanamycin was added to the cultivation medium at a concentration of 25 μg/ml for C. glutamicum and 50 μg/ml for E. coli. For induction of the SOS response, mitomycin C (Sigma-Aldrich, Seelze, Germany) was added at the appropriate concentrations at an OD₆₀₀ of 4.

TABLE 1.

Bacterial strains and plasmids used in this study

Strain or plasmid	Characteristics	Source or reference
Strains
E. coli DH5α	supE44 ΔlacU169 (ϕ80lacZDM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1	Invitrogen
C. glutamicum
ATCC 13032	Biotin-auxotrophic wild type	36
ATCC 13032ΔlexA	In-frame deletion of the gene lexA (cg2114)	22
ATCC 13032::PrecA-eyfp	Integration of PrecA-eyfp into the intergenic region between cg1121 and cg1122	This study
Plasmids
pJC1	Kanr, Ampr; C. glutamicum shuttle vector	38
pEKEx2-Ptac-eyfp	Kanr; pEKEx2 containing eyfp with pET16 RBS, under the control of Ptac	23
pAN6-e2-crimson	Kanr; pAN6 derivative for expression of E2-Crimson under the control of the Ptac promoter	6
pK19mobsacB	Kanr, oriVE. coli sacB lacZα	37
pK18mobsacB-cg1121/1122	Kanr, oriVE. coli sacB	6
pK18mobsacB-cg1121/1122-PrecA-eyfp	pJC1 derivative containing the promoter of recA (260 bp) fused to eyfp; the insert includes the promoter of recA and an additional ribosome binding site (pET16) in front of eyfp	This study
pJC1-PdivS-e2-crimson	pJC1 derivative containing the promoter of divS (411 bp) fused to e2-crimson; the insert includes the promoter of divS, 30 bp of the coding sequence, a stop codon, and an additional ribosome binding site (pET16) in front of e2-crimson	This study
pJC1-PrecN-e2-crimson	pJC1 derivative containing the promoter of recN (207 bp) fused to e2-crimson; the insert includes the promoter of recN, 30 bp of the coding sequence, a stop codon, and an additional ribosome binding site (pET16) in front of e2-crimson	This study
pJC1-Pint2-e2-crimson	pJC1 derivative containing the promoter of int2 (250 bp) fused to e2-crimson; the insert includes the promoter of int2, 30 bp of the coding sequence, a stop codon, and an additional ribosome binding site (pET16) in front of e2-crimson	This study
pJC1-Plysin-e2-crimson	pJC1 derivative containing the promoter of lysin (250 bp) fused to e2-crimson; the insert includes the promoter of lysin, 30 bp of the coding sequence, a stop codon, and an additional ribosome binding site (pET16) in front of e2-crimson	This study
pJC1-Pcg2067-e2-crimson	pJC1 derivative containing the promoter of cg2067 (250 bp) fused to e2-crimson; the insert includes the promoter of cg2067, 30 bp of the coding sequence, a stop codon, and an additional ribosome binding site (pET16) in front of e2-crimson	This study

Cloning techniques.

For PCR amplification of DNA used for cloning, KOD HotStart polymerase (Merck Millipore, Darmstadt, Germany) was used. DreamTaq (Fisher Scientific, Schwerte, Germany) was utilized for PCR verification of ligation reactions. Heat shock transformation of E. coli was performed as described previously (19). Transformation of C. glutamicum was performed by electroporation as described previously (20). Isolation and purification of plasmids from E. coli cultures was performed using plasmid isolation and purification kits from Qiagen (Qiagen, Hilden, Germany) and Thermo Scientific (Fisher Scientific, Schwerte, Germany) miniprep kits. Chromosomal DNA of C. glutamicum was isolated as described previously (21). DNA sequencing and oligonucleotide synthesis were performed by Eurofins MWG Operon (Ebersberg, Germany). Plasmids and oligonucleotides used in this work are listed in Table 1 and Table 2, respectively. The in-frame deletion mutant of lexA in the wild-type strain ATCC 13032 was constructed as described by Jochmann et al. (22).

TABLE 2.

Oligonucleotides used in this study

Oligonucleotide	Sequence (5′–3′)a	Restriction site
cg2114_del_1	TCCCCCGGGGATCTAGGATCCACATGGAAGCGAACAGAG	SmaI
cg2114_del_2	TGAAGTCCTGCAGCATCA
cg2114_del_3	TGATGCTGCAGGACTTCATGTTGGCGAGTCCATGAG
cg2114_del_4	GCTCTAGAGATCTACTGCAGCGCCACGATATGTGAGAA	XbaI
PrecA_pK18_fwd	GGAATTCAATTGTTCGCATGAAAATTCGAATT	MfeI
PrecA_YFP_rev	GCTCACCATATGTATATCTCCTTTTTTAATTCCTCTTAGTTTTATTGA
YFP_fwd	AAGGAGATATACATATGGTGAGCAAGGGCGA
YFP_pK18_rev	GGAATTCTCGAGTTATCTAGACTTGTACAGCTCGTCCAT	XhoI
divS_fwd	CTAGCTAGCTAGCGTTGGGCAAGGTCTTAACT	NheI
divS_rev	ATGATATCTCCTTCTTAAAGTTTAATCTAGCTCTTTACCCGCATAAAC
crimson_fwd	TAAACTTTAAGAAGGAGATATCATATGGATAGCAC
crimson_rev	CTAGCTAGCTAGCAAAAGAGTTTGTAGAAACGC	NheI
int2_fwd	CGCGGATCCCGGGCGAGAGGGTGAGCGAT	BamHI
int2_rev	CGCCATATGATATCTCCTTCTTAAAGTTCAGAAGCGTGCCCTGTACCTCTCCGA	NdeI
lysin_fwd	CGCGGATCCCCTTCTTTGAGGCTTGATGCCT	BamHI
lysin_rev	CGCCATATGATATCTCCTTCTTAAAGTTCAATTTTTCGGCATTGCGCCTTTAAT	NdeI
cg2067_fwd	CGCGGATCCGAAGCTTTTGAGTCTGTTACTGG	BamHI
cg2067_rev	CGCCATATGATATCTCCTTCTTAAAGTTCACGACCACATCTCCAACGCGCTAAA	NdeI

^aRestriction sites are underlined.

Construction of promoter fusions.

For construction of the promoter fusions of P_int2, P_lysin, and P_cg2067, 250 bp upstream of the coding sequence with an additional 10 codons and the 16-bp ribosomal binding site of pET16 were amplified using the oligonucleotide pairs int2-fwd/int2-rev, lysin-fwd/lysin-rev, and cg2067-fwd/cg2067-rev, respectively. The promoter sequences were ligated into the vector pJC1-crimson-term by restriction with BamHI and NdeI.

For construction of the recA promoter fusion, 260 bp upstream of the coding sequence were amplified from genomic C. glutamicum DNA followed by the 16-bp ribosomal binding site of pET16 using oligonucleotides PrecA_pK18_fwd and PrecA_YFP_rev. eyfp was amplified with oligonucleotides YFP_fwd and YFP_pK18_rev using the plasmid pEKEx2-P_tac-eyfp (23) as the template. The promoter fusion was generated by overlap extension PCR using oligonucleotides PrecA_pK18_fwd and PrecA_YFP_rev. XhoI and MfeI restriction sites were used to ligate the amplicon into the shuttle vector pK18mobsacB_cg1121/22. Promoter fusions of divS and recN were amplified with the oligonucleotides divS_fwd and divS_rev and the oligonucleotides recN_fwd and recN_rev, respectively, using genomic DNA as the template. Additionally, the first 10 codons were amplified along with a stop codon. The coding sequence of the fluorescent protein E2-Crimson (24) was amplified from the plasmid pAN6-e2-crimson with the oligonucleotides crimson_fwd and crimson_rev. Both amplicons were combined by overlap extension PCR. The restriction site NheI was used for ligation into the medium-copy-number vector pJC1.

The P_ptsG-e2-crimson promoter fusion was constructed according to Hentschel et al. (23) by amplifying 725 bp upstream of the ptsG start codon with oligonucleotides PptsG_fwd and PptsG_rev, introducing BamHI and NdeI restriction sites. The PCR fragment was ligated into pJC1-e2-crimson-term after treatment of both with BamHI and NdeI restriction enzymes.

Fluorescence microscopy.

For phase contrast and fluorescence microscopy, samples were analyzed on 1 to 2% agar pads, which were placed on microscope slides and covered by a coverslip. Images were taken on a Zeiss Axioplan 2 imaging microscope equipped with an AxioCam MRm camera and a Plan-Apochromat ×100 magnification, 1.4 numerical-aperture oil differential interference contrast (DIC) objective. Filter sets 46HE and 63HE were used for imaging enhanced yellow fluorescent protein (eYFP) and E2-Crimson fluorescence, respectively. Digital images were acquired and analyzed with the AxioVision 4.6 software (Zeiss, Göttingen, Germany).

Flow cytometry.

Flow-cytometric measurements and sorting were performed on a FACSAria II (Becton, Dickinson, San Jose, CA) flow cytometer with 488-nm excitation by a blue solid-state laser and 633-nm excitation by a red solid-state laser. Forward-scatter characteristics (FSC) and side-scatter characteristics (SSC) were detected as small- and large-angle scatters of the 488-nm laser, respectively. eYFP fluorescence was detected using a 502-nm long-pass and a 530/30-nm band pass filter set. E2-Crimson fluorescence was detected using a 660/20-nm band pass filter set. Cells were analyzed at a threshold rate of 3,000 to 8,000 events/s and sorted onto BHI agar plates at a threshold rate of 3,000 to 4,000 events/s. Data were analyzed using FlowJo V7.6.5 (Tree Star, Inc., Ashland, OR).

Statistical analysis.

Nonparametric tests for the analysis of correlation were performed with GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, CA). Spearman's rank correlation coefficient (the Pearson correlation coefficient of the ranked variables) was used. A perfect (inverse) correlation takes on ρ values of (−)1; totally independent variables take on ρ values of 0.

Microfluidic cultivation.

C. glutamicum was cultivated in in-house-developed microfluidic cultivation chambers (0.9 μm by 60 μm by 60 μm) arranged in parallel between 10-fold-deeper supply channels. For details on our microfluidic chip setup, see references 15 and 25.

During the experiment, CGXII minimal medium was infused continuously at 300 nl min⁻¹ using a high-precision syringe pump (neMESYS; Cetoni GmbH, Korbussen, Germany) with attached disposable syringes (Omnifix-F Tuberculin, 1 ml; B. Braun Melsungen AG, Melsungen, Germany) to maintain constant environmental conditions. Cell growth and eYFP fluorescence were observed at 10-min intervals by time-lapse imaging with a fully motorized inverted Nikon Eclipse Ti microscope (Nikon GmbH, Düsseldorf, Germany). Chip cultivation was performed at 30°C using a microscope incubator system (PeCon GmbH, Erbach, Germany). The microscope was equipped with a focus assistant (Nikon PFS) to compensate for thermal drift during long-term microscopy, with a Plan Apo λ 100× oil Ph3 DM objective (Nikon GmbH, Düsseldorf, Germany) and a high-speed charge-coupled device (CCD) camera (Andor Clara DR-3041; Andor Technology Plc., Belfast, United Kingdom). An optical filter system (YFPHQ filter system [excitation, 490 to 550 nm; dichroic mirror, 510 nm; absorption filter, 520 to 560 nm]; AHF Analysentechnik AG, Tübingen, Germany) and a mercury light source (Intensilight; Nikon GmbH, Düsseldorf, Germany) were installed for fluorescence microscopy.

RESULTS

The prophage CGP3 is spontaneously induced in single cells.

In recent studies we observed a spontaneous excision of the prophage CGP3 in a small number of C. glutamicum cells cultivated in shake flasks with CGXII minimal medium (5). Transcriptome analysis revealed an upregulation of CGP3 genes (cg1890 to cg2071) upon induction of the SOS response by addition of the DNA-cross-linking antibiotic mitomycin C (A. Heyer and J. Frunzke, personal communication). In an effort to create appropriate tools to monitor prophage activity, we consulted reports on previous microarray experiments to determine which genes are suitable candidates. We constructed plasmid-based promoter fusions of the CGP3 genes cg2071 (integrase, int2), cg1974 (putative lysin), and cg2067 (hypothetical protein) to the coding sequence of the fluorescent protein E2-Crimson. To test their function, wild-type C. glutamicum ATCC 13032 cells were transformed with the promoter fusion constructs and treated with 2 μM mitomycin C to induce DNA lesions and subsequently trigger the SOS response. Samples were analyzed by flow cytometry and fluorescence microscopy (shown for P_int2) (Fig. 1A). The treated samples exhibited a highly induced P_int2 activity and morphological changes. The untreated cultures showed no significant P_int2 activity, yet, in agreement with earlier studies, P_int2 was highly induced in a small number of cells. Flow-cytometric analysis of untreated cultures revealed that a fraction of 0.01 to 0.08% of the cells exhibited a 5- to 160-fold higher P_int2 activity than the bulk of the population (Fig. 1B). Single cells with a high fluorescent signal (phage⁺) and cells showing background fluorescence (phage⁻) were sorted onto BHI agar plates, and their survival was assessed after incubation for 24 h. As expected, the survival rate of phage⁺ cells was significantly below that of phage⁻ cells (survival rates of 23% and 96%, respectively) (Fig. 2). Thus, cells showing an increased P_int2 activity were significantly impaired in their ability to resume growth on plates, which is likely caused by prophage excision and subsequent cell lysis.

Fig 1.

Utilization of prophage promoter fusions for monitoring spontaneous CGP3 induction. (A) Histogram of eYFP fluorescence of C. glutamicum/pJC1-P_int2-e2-crimson with 2 μM mitomycin C (red line) and without mitomycin C (gray line) with additional fluorescence microscopy images showing the spontaneous and induced activity of P_int2. The insets show uninduced and induced cells carrying the plasmid pJC1-P_int2-e2-crimson. The green arrow indicates spontaneous induction of P_int2 occurring under standard cultivation conditions. (B) Scatter plot of the strain C. glutamicum/pJC1-P_int2-e2-crimson. A total of 100,000 cells were analyzed for their size characteristics (forward scatter) and their fluorescent properties. Even under noninducing conditions, a small fraction of cells exhibited an increased reporter signal (red dots) in comparison to the bulk of the population (black dots). Cells were cultivated in CGXII medium plus 4% glucose and analyzed at an OD₆₀₀ of 4.

Fig 2.

Viability assay of C. glutamicum cells which exhibit an induction of CGP3 and/or of the SOS response. Single cells were sorted onto agar plates, and the percent survival was determined as the fraction of cells able to form colonies. Nonfluorescent cells (−/−), cells with either a positive P_recA-eyfp or a positive P_int2-e2-crimson signal (+), and cells with both signals (+/+) were sorted onto separate agar plates. Colony growth was analyzed after 24 h.

Spontaneous P_recA induction in single cells.

Since the host SOS response is a prominent trigger of lysogenic phages, we constructed a promoter fusion of the recA promoter to eyfp and integrated it into the genome of C. glutamicum ATCC 13032 at the intergenic region of cg1121 and cg1122 to test for a spontaneous SOS induction. As proof of principle, the P_recA-eyfp strain was cultivated in microtiter scale and the SOS response was induced by addition of mitomycin C in increasing concentrations. A strain with a truncation of lexA, the repressor of SOS genes, served as a reference strain which exhibits a maximally induced SOS response (Fig. 3A, black bar). At low concentrations (15 nM and 100 nM) of mitomycin C and in the ΔlexA strain, P_recA activity showed a bimodal distribution, with the majority of cells showing a slight increase in reporter signal and a smaller fraction of cells shifted toward an even higher signal. This bimodal state was not observed at higher concentrations (500 nM and 1,000 nM mitomycin C) (Fig. 3B). We subjected cells with the integrated P_recA-eyfp promoter fusion to flow-cytometric analysis to get more detailed insight into single-cell dynamics of the SOS response in C. glutamicum populations. Under standard cultivation conditions, we observed a spontaneous activity, analogous to the activities of CGP3 promoters (Fig. 3C). About 0.07 to 0.2% of cells showed a 12- to 18-fold increased reporter signal (SOS⁺ cells). These SOS⁺ cells had a reduced survival rate (recovery rate of 46% after sorting on BHI agar plates) (Fig. 2). It was tempting to hypothesize that these rare events may act as a bet-hedging strategy to ensure an increased fitness under changing environmental conditions. We sorted SOS⁺ cells on agar plates with different DNA damaging conditions yet saw no increased fitness under the tested conditions (data not shown).

Fig 3.

P_recA promoter fusion for the analysis of SOS induction in single cells. (A) Dose response plotted as mean eYFP fluorescence at increasing mitomycin C (MmC) concentrations. (B) Offset histogram of C. glutamicum::P_recA-eyfp cells subjected to increasing concentrations of mitomycin C. C. glutamicum ΔlexA was used as a reference strain which exhibits maximal induction of the SOS response. (C) Scatter plot of the strain C. glutamicum::P_recA-eyfp cultivated under nonstressful conditions. A total of 100,000 cells were analyzed for their size characteristics (forward scatter) and their fluorescent properties. A small fraction of cells exhibits an increased reporter signal (green dots) in comparison to the bulk of the population (black dots). Cells were cultivated in CGXII medium plus 4% glucose in the Biolector system until the stationary phase.

Spontaneous P_recA activity reflects a bona fide SOS response.

Next we tested whether spontaneous P_recA activity and reduced viability were indicative of a bona fide SOS response caused by potentially lethal DNA damage. The reporter strain harboring the integrated promoter fusion P_recA-eyfp was transformed with plasmids carrying promoter fusions of the two SOS-responsive genes divS and recN, respectively, fused to the autofluorescent reporter gene e2-crimson. Both genes were previously shown to be upregulated in the deletion mutant ΔlexA (22) or after induction with mitomycin C. Analysis of the dual reporter strain (P_recA-eyfp integrated into the genome, P_int2-e2-crimson plasmid-borne) by fluorescence microscopy and flow cytometry revealed a correlation of the P_recA signal to P_divS as well as P_recN signals. After gating of SOS⁺ cells, the signals of both reporters were correlated in single cells by using Spearman's rank correlation coefficient (Fig. 4). Both promoter fusions showed a high correlation (P_divS, ρ = 0.78, P < 0.0001, n = 165; P_recN, ρ = 0.85, P < 0.0001, n = 131) at the single-cell level. The control promoter fusion P_ptsG-e2-crimson was constructed and introduced into the P_recA-eyfp strain to exclude high correlation values due to factors other than sharing the same regulation. SOS⁺ cells were gated and their P_recA signal correlated to the P_ptsG reporter signal. Both signals displayed a low correlation (ρ = 0.36, P < 0.0001, n = 248). Thus, the strong correlation of P_recA to P_divS and P_recN activities confirmed that a spontaneous P_recA activity in single cells leads to a bona fide induction of the downstream SOS cascade.

Fig 4.

Correlation of P_recA activity to the activity of promoters of downstream SOS genes. Scatter plots of C. glutamicum::P_recA-eyfp/pJC1-P_divS-e2-crimson (A), C. glutamicum P_recA-eyfp/pJC1-P_recN-e2-crimson (B), and C. glutamicum P_recA-eyfp/pJC1-P_ptsG-e2-crimson (C) are shown; C. glutamicum P_recA-eyfp/pJC1-P_ptsG-e2-crimson served as a control, reflecting the correlation of recA expression and an SOS-independent promoter. Cells with high eYFP fluorescence (spontaneous P_recA-eyfp cells) were gated and are displayed in the dot plot of eYFP fluorescence versus E2-Crimson fluorescence. Spearman's rank correlation coefficient (ρ) was calculated using GraphPad Prism 6. Cells were cultivated in CGXII medium plus 4% glucose and analyzed at an OD₆₀₀ of 4.

Time-lapse analysis of P_recA activation dynamics.

SOS⁺ cells did not display a uniform fate, with some cells being able to survive on agar plates whereas others were not viable. For a time-resolved analysis of the SOS induction in single cells, we cultivated the P_recA-eyfp strain in an in-house-developed PDMS microfluidic system enabling spatiotemporal analysis of growing microcolonies by automated time-lapse microscopy. Single cells were seeded into the microfluidic cultivation chambers and cultivated for several generations in standard CGXII minimal medium under nonstressful conditions; images were acquired every 10 min. Again, we observed different fates of SOS⁺ cells (see Movies S1 to S3 in the supplemental material). Some cells showed a high and continuous reporter signal together with an elongated cell morphology and growth inhibition caused by an activated SOS response (Fig. 5; also, see Movies S1 to S3 in the supplemental material). These bacteria represent cells which undergo severe DNA damage, triggering the SOS response. However, other cells merely showed a pulse of the reporter along with an unaltered growth and cell morphology (Fig. 5; also, see Movies S1 and S2 in the supplemental material). This output either represents activity due to the stochastic binding of repressor molecules or might be triggered by DNA damage which is repaired before a full-blown SOS response is initiated. Furthermore, we observed an additional cell fate which consisted of high induction of the reporter coupled with an elongated cell morphology, growth inhibition, and cell branching (see Movie S3 in the supplemental material). This is analogous to C. glutamicum cells which continuously overexpress divS under the control of the promoter P_aceA (26), indicating a high concentration of DivS in these cells.

Fig 5.

Fluorescence traces of P_recA activity in single cells observed by live cell imaging. Cells harboring the integrated P_recA-eyfp promoter fusion were seeded into the microfluidic chip and cultivated for 24 h; a representative microcolony is shown. (A) Microscopic images of microcolony during cultivation. Single cells were analyzed for their fluorescent output and assigned a unique cell ID (colored arrow). (B) Course of fluorescence for single cells (colored lines) as well as for the entire microcolony (mean value, plotted as a dotted line; standard deviations are shown as a gray area). Single cells either showed no fluorescence (green trace), showed a high induction of P_recA (red and light green trace), displayed a pulsing behavior with no further promoter activity (violet and pink traces), or displayed a pulsing behavior followed by a high reporter signal (cyan and orange traces).

Correlating prophage activity to an induced SOS response in single cells.

Having established promoter fusions for the analysis of spontaneous CGP3 induction and SOS response, we combined promoter fusions of recA and of the prophage genes to correlate them at the single-cell level. To this end, the P_recA-eyfp strain was transformed with the P_int2, P_lysin, and P_cg2067 promoter fusions to e2-crimson. Analysis by flow cytometry again showed the occurrence of spontaneously activated cells for both types of promoter fusions. Fluorescence microscopy revealed that not all cells showed an activation of the prophage promoters when P_recA was active (Fig. 6). To measure the interdependence of both signals, the Spearman rank correlation coefficient was calculated for both. The highest correlation to SOS⁺ cells was observed for P_cg2067 and P_int2 (P_cg2067, ρ = 0.77, P < 0.0001, n = 687; P_int2, ρ = 0.57, P < 0.0001, n = 828) (Fig. 6). P_lysin showed the lowest correlation (ρ = 0.44, P < 0.0001, n = 790) under the tested conditions. Even though the correlation coefficient is not able to reveal a causal link between two processes, it did show a high correlation of the promoters of recA and the prophage genes. Reciprocal analysis of phage⁺ cells was performed as well and gave nearly identical values (P_cg2067, ρ = 0.72, P < 0.0001, n = 119; P_int2, ρ = 0.58, P < 0.0001, n = 565; P_lysin, ρ = 0.45, P < 0.0001, n = 759) (data not shown). This correlation suggests a link between spontaneously induced SOS response and spontaneous activity of CGP3 in single cells. Nevertheless, correlation was lower than that between P_recA and P_divS/P_recN, indicating that others factors besides the SOS response might influence the activity of the prophage promoters.

Fig 6.

Correlation of P_recA reporter activity to the activity of CGP3 genes. Scatter plots of C. glutamicum P_recA-eyfp/pJC1-P_int2-e2-crimson (A), C. glutamicum P_recA-eyfp/pJC1-P_lysin-e2-crimson (B), and C. glutamicum P_recA-eyfp/pJC1-P_cg2067-e2-crimson (C) are shown. Cells with high eYFP fluorescence (SOS⁺ cells) were gated and their E2-Crimson fluorescence plotted against their eYFP fluorescence. Spearman's rank correlation coefficient (ρ) was calculated using GraphPad Prism 6. (D) Fluorescence microscopy analysis of cells on agar pads showing the spontaneous induction of SOS and prophage reporters. Cells were cultivated in CGXII medium plus 4% glucose and analyzed at an OD₆₀₀ of 4.

Impact of growth phase on spontaneous SOS and prophage activity.

The P_recA-eyfp/pJC1-P_int2-e2-crimson strain was cultivated in shake flasks, and samples were analyzed until the cells reached stationary growth phase (Fig. 7A) to eliminate the rigidity of measurements at single time points and better assess the characteristics of the reporters during growth. The P_recA signal showed maximal intensity at the transition to and during the early phase of exponential growth. Activity of P_int2 showed the same behavior. We expected this parallel activation of both promoters, if the SOS response and prophage induction are somehow linked. The number of spontaneously activated cells was measured for all time points as well (Fig. 7B). Whereas the peaks of both reporters' fluorescent output behaved similarly, the maximal number of spontaneously activated phage⁺ and SOS⁺ cells showed a temporal disparity. The increase of SOS⁺ cells was observed 2 h before the relative amount of phage⁺ cells increased.

Fig 7.

Analysis of activity and spontaneous activation of P_recA and P_int2 during growth. C. glutamicum P_recA-eyfp/pJC1-P_int2-e2-crimson was cultivated in the Biolector microcultivation system. (A) OD₆₀₀ (black line) and mean eYFP (green line) and E2-Crimson (red line) fluorescence; (B) percentage of cells with a spontaneous P_recA (green line) and P_int2 (red line) induction. The arrow indicates the delay between the peak for maximum spontaneous P_recA and maximum spontaneous P_int2 activity in single cells. Cells were cultivated in CGXII medium plus 4% glucose.

DISCUSSION

The present study on the induction of lysogenic prophages was performed on a population-wide scale. It was shown that the induction of lambdoid phages is typically linked to the host's SOS response. The underlying bistable switch (13), simplified by the action of the repressor of phage genes cI, is turned toward lytic growth when the host's SOS response becomes activated. Spontaneous induction of prophages had been observed as far back as the 1950s (11), yet studies since then have not explored this phenomenon in more detail at the single-cell level. Due to the general link between the SOS response and prophage induction, it was tempting to speculate that the cause of spontaneous prophage induction lies in a spontaneously induced SOS response. The data shown in this study suggest that a small fraction of C. glutamicum cells grown under standard conditions spontaneously induced expression of prophage genes and that this activation is caused in part by the spontaneous activity of the SOS response in single cells.

During the cultivation of wild-type cells under standard cultivation conditions, we observed single cells that induced the SOS response spontaneously. As the occurrence of spontaneous DNA breakage has previously been reported in studies on E. coli (27), we tested this in C. glutamicum. In the E. coli studies, use of the SOS-inducible promoter of the cell division inhibitor gene sulA fused to gfp revealed a spontaneous SOS response in about 0.9% of cells. The rate of spontaneous SOS induction that we measured with P_recA-eyfp lies at about 0.2%. When other LexA-regulated promoters of C. glutamicum, such as P_divS, P_recN, and P_cglIM, were used, the fraction of cells exhibiting a spontaneous SOS response lay between 0.1 and 0.5% (data not shown). Considering that further studies in E. coli using promoter fusions of lexA, recA, and umuDC showed spontaneous promoter activity in 0.09 to 3.1% of cells (28), our findings are consistent with those reported previously.

We assumed the SOS⁺ cells to be impaired in their survival due to potentially lethal DNA damage. The survival rate of SOS⁺ cells lay at 46%. Our microfluidic studies showed that an induction of the SOS response leads to the arrest of cell growth in some cells, probably caused by irreparable DNA damage, whereas other cells are able to resume growth after they exhibit P_recA activity. This activity in turn did not necessarily lead to a full-blown SOS response, as some cells showed the SOS phenotype; others, however, were either unaffected in their growth with a mere pulse of P_recA activity or were inhibited in their growth but able to recover from the SOS-induced inhibition of cell division. The presence of some cells showing a short pulse of reporter signal, but no growth inhibition, might hint at stochastic fluctuations in the binding of the repressor LexA or at DNA damage which is so minor that it is repaired before a full-blown SOS response is stimulated.

Studies in E. coli have revealed that SOS genes show a heterogeneous expression which is independent of RecA and based on stochastic factors and binding affinities of LexA to SOS boxes (28). To confirm that the spontaneous induction of P_recA is caused by a bona fide SOS response, we transformed the P_recA reporter strain with plasmids carrying transcriptional fusions of divS and recN promoters, respectively. The degree of correlation gained by calculating Spearman's ranked correlation coefficient is strong (P_divS, ρ = 0.78; P_recN, ρ = 0.78). Whereas this is a strong correlation, higher values of ρ (>0.9) might be expected for causal relationships, as they are present within the SOS cascade. Even though recA is upregulated after cells encounter DNA damage, this damage might be repaired before the cell division inhibitor divS is induced, thus avoiding a premature inhibition of growth. Our microfluidic experiments showed that this might be the case, as a high number of SOS⁺ cells are not inhibited in their growth.

The same applies to the correlation values of the prophage reporters which lie below those observed for the correlation to the SOS reporters. An induced SOS response need not seal the fate of genomic excision for CGP3. Rather, an immunity to low levels of DNA damage and an induction upon accumulation or lasting presence of DNA damage would constitute a more reliable warning system telling the prophage when to “leave the sinking ship.” If CGP3 were induced by every event of SOS response, the integration of the prophage would be a rather unstable situation. Thus, a threshold-based model of excision (29), as described for well-studied lambdoid phages, helps to explain the observed deviation from a perfect correlation. Nevertheless, future studies will elucidate the possible influence of so-far unknown regulators that play a role in the induction of CGP3. Microarray studies, for example, revealed that the putative regulator Cg2040 exhibits an inhibitory effect on a set of prophage genes when overexpressed (A. Heyer and J. Frunzke, personal communication).

Finally, we tested our dual promoter fusion strains (P_recA-eyfp/pJC1-P_int2-e2-crimson) during a standard flask cultivation experiment to analyze their expression during the course of growth. We observed that P_recA and P_int2 promoters reached their peaks of activation during exponential growth phase. It is reasonable to assume that this phase of rapid cell growth gives rise to intrinsic DNA damage, which is produced by native DNA polymerases. Indeed, it has been reported that replication fork breakage is a major contributor to double-strand breaks (DSBs), which in turn activate the SOS response (30, 31). The measurement of spontaneous activation of the SOS response and prophage activity showed temporally separated peaks (Fig. 7B). As this disparity occurs in a small number of cells, it is clear that this effect would be masked in the bulk measurement of reporter output, as shown in Fig. 7A.

While the consequence of spontaneously inducing prophages has been reported for Shewanella oneidensis (7) and Streptococcus pneumoniae (8), the effects on C. glutamicum on a population-wide scale remain unknown. In studies by Bossi and coworkers (32), a spontaneous induction of prophages led to a competitive fitness of the population against other bacteria. Spontaneous phage release is seen as a strategy to maintain the lysogenic status of the prophage. Immunity sequences within the prophage lead to an immunity to superinfection. Interestingly, the genomic CGP3 locus possesses a restriction-modification system (cglIM, cglIR, and cglIIM) which might be used as a classic toxin/antitoxin module (33) to fend off infecting bacteriophages. An example of “bacterial altruism” was recently described (34) in which Shiga toxin produced by a small number of spontaneously induced prophage works as a positive selective force and benefits the population. The production of toxins by C. glutamicum has not been reported, but the release of DNA and proteins into the extracellular matrix as a source of nutrients for the rest of the population or as biofilm matrix for natural microbial communities is a possible scenario. Choosing irreparable or sustained DNA damage as trigger or selection marker to single out “weak individuals” would be a practical approach (35).

Future studies will aim at a more detailed investigation of the phenomenon of spontaneous prophage excision. We have established a first link between the host's SOS response and the excision of the genomically integrated CGP3 prophage in single bacterial cells. Further studies will assess the level at which the two phenomena are linked and which other regulatory pathways may feed into the prophage's decision to excise from the host genome.