Fluorescent fusions of the N protein of phage Mu label DNA damage in living cells

Abstract

The N protein of phage Mu was indicated from studies in Escherichia coli to hold linear Mu chromosomes in a circular conformation by non-covalent association, and thus suggested potentially to bind DNA double-stranded ends. Because of its role in association with linear Mu DNA, we tested whether fluorescent-protein fusions to N might provide a useful tool for labeling DNA damage including double-strand break (DSB) ends in single cells. We compared N-GFP with a biochemically well documented DSB-end binding protein, the Gam protein of phage Mu, also fused to GFP. We find that N-GFP produced in live E. coli forms foci in response to DNA damage induced by radiomimetic drug phleomycin, indicating that it labels damaged DNA. N-GFP also labels specific DSBs created enzymatically by I-SceI double-strand endonuclease, and by X-rays, with the numbers of foci corresponding with the numbers of DSBs generated, indicating DSB labeling. However, whereas N-GFP forms about half as many foci as GamGFP with phleomycin, its labeling of I-SceI-and X-ray-induced DSBs is far less efficient than that of GamGFP. The data imply that N-GFP binds and labels DNA damage including DSBs, but may additionally label phleomycin-induced non-DSB damage, with which DSB-specific GamGFP does not interact. The data indicate that N-GFP labels DNA damage, and may be useful for general, not DSB-specific, DNA-damage detection.

1. Introduction

The study of DNA repair in living cells has benefitted from the use of reagents that specifically bind and label DNA intermediates in DNA-damage and -repair reactions so that they can be quantified in single living cells as fluorescent foci and mapped in genomes using ChIP-seq. Examples include fluorescent-protein fusions of functional or partly functional Escherichia coli DNA-repair proteins including a functional MutL-eGFP protein, which diagnoses mispaired DNA bases as they form [1], partially functional RecA-GFP [2], and SSB-Ypet [3] fusion proteins, both of which label single-stranded DNA as fluorescent foci, and a variety of fluorescent yeast DNA repair and replication proteins [4]. Also highly useful are proteins that bind specific DNA structures but have no catalytic activity, and so “trap” the structure allowing quantification as foci, mapping by ChIP-seq, and also “better-than-genetic” analyses in which all DNA-repair mechanisms are prevented from acting on the DNA structure by the trapping protein, which protects the DNA from other proteins. Examples of DNA-structure-trap proteins include various fluorescent-protein fusions to the double-strand-break (DSB)-end-specific binding protein Gam of phage Mu, which label DSBs as fluorescent foci in E. coli, mouse, and human cells and prevented their exonucleolytic digestion in E. coli and mouse cells [5]. Also, the catalytically defective E. coli four-way DNA-junction (Holliday-junction)-endonuclease RuvCDefGFP or RDG binds and does not cleave Holliday junctions [6]. RDG allows quantification of recombination/repair generated and reversed-replication-fork Holliday junctions as foci, and their mapping in genomes by ChIP-seq [6] in E. coli, and an independently constructed version worked to hold telomere T-loops together in mammalian cells [7]. Though less precise, DNA-damage-induced signals including 53BP1, phospho-P53, and γH2AX are used widely as surrogate markers for detection of damaged DNA, including double-strand-breaks (DSBs), in mammalian cells [8].

We explored here whether another phage protein, the N protein of the transposable phage Mu, might be useful for detection of DNA damage in E. coli cells. The N protein, studied exclusively in living cells, binds the linear phage Mu chromosome non-covalently and holds it, with flanking E. coli DNA from the previous host cell, in a circular conformation during injection [9]. N is injected into the cell with the DNA molecule [10–13]. Phage and E. coli host proteins then catalyze interaction of the phage genome with a DNA target site in the cell [14, 15], the flanking host DNA is digested by the host RecBCD exonuclease [16], and the phage genome is integrated into the E. coli chromosome [17]. These aspects of the biology of N protein in the phage Mu lifestyle suggested that it might bind DSB ends, and so potentially be a useful protein for binding and labeling damaged DNA substrates. Because no biochemistry has been reported for N, it could in principle interact with various damaged (or non-damaged) DNA structures. We constructed an N-GFP fusion and show its DNA-damage labeling properties here.

2. Material and Methods

2.1. Strains, media, cloning and chromosomal expression cassettes encoding N and N-GFP

E. coli strains used are shown in Table 1. They were grown in LBH rich [18], TB [19] or M9 minimal medium [20]. Antibiotics were added when necessary at the following concentrations: 25 μg/ml chloramphenicol, 50 μg/ml kanamycin or 20 ug/ml carbenicillin. P1 transductions were as described [21]. The native Mu-gam gene with EcoRI and BamHI cloning sites was synthesized (GenScript, NJ) and subcloned into pNT3-SD plasmid under the control of the tac promoter. The phage Mu N gene was amplified from the Mu genome with primers P1 and P2 (Table 1), digested with BamHI and SalI, purified, and subcloned into the BamHI and SalI sites of pLC3 [6] to create pHL1. To construct an N-gfp inducible plasmid, the Mu N gene was amplified from the Mu genome with primers P1 and P3 digested with BamHI and SalI and gfp was amplified from P tac gfp [6] with primers P4 and P5 and restriction digested with SalI and HindIII. Purified PCR products were then subcloned into the BamHI and HindIII site of pLC3 [6] to create pHL2. The N or N-gfp expression cassette containing the doxycycline-inducible P_N25tetO promoter and chloramphenicol resistance (cat) gene was moved into the chromosome of E. coli strain SMR21475 by short-homology phage lambda Red-mediated recombineering using primers P6 and P7 (Table 1).

Table 1.

Escherichia coli K12 strains, plasmids and primers used in this study

Name	Relevant Genotype	Reference or Source
Plasmids
pHL1	pET28a PN25teto SD N FRTcatFRT	This study
pHL2	pET28a PN25teto SD N-GFP FRTcatFRT	This study
pCP20	FLP recombinase vector	[31]
pKD46	ori101 repA101TS PBAD-gam-bet-exo AmpR	[32]
Ptac	pNT3	[25]
Ptac-Gam	pNT3-Gam	This study
E. coli strains
RTC0013	MG1655 ΔrecB: :Kan	[33]
CH30	MG1655 rph-1	[34]
CH5557	ΔaraBAD567 ΔattL::PBADI-SceI ΔlacA::I-SceIsite FRTKanFRT	[19]
CH7090	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetO SD N-GFP AlacΔ::I-SceIsite FRTKanFRT	SMR22141 x P1(CH5557)
CH7099	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRT PN25tetO SD N-GFP ΔlacA::I-SceIsite FRT	CH7090 x pCP20, shifted to 37°C
SMR11768	zie3936.8::3ChiKanI-SceI = I-site C	[35]
SMR12725	yddJ I-SceIsite-Kan 3Chi = I-site D	[35]
SMR12724	MG1655z1 Δattλ::PN25tetR FRTKanFRT	[5]
SMR14328	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT	[5]
SMR14334	MG1655 ΔaraBAD567 Δattλ::PBADzfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetOgam-gfp	[5]
SMR14338	MG1655 ΔaraBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetOgam-gfp	[5]
SMR14354	MG1655 ΔaraBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetOgam-gfp I-site C	[5]
SMR14362	MG1655 ΔaraBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetOgam-gfp I-site D	[5]
SMR21417	MG1655 rph-1 Δattλ::PN25tetR FRTKanFRT	CH30 x P1(SMR12724)
SMR21475	MG1655 rph-1Δattλ::PN25tetR FRT	SMR21417 x pCP20, shifted to 37°C SMR21475 x pKD46, λ
SMR22131	MG1655 rph-1 Δattλ::PN25tetR FRT ΔattTn7::PN25tetO SD-N FRTcatFRT	Red-mediated short- homology recombineering from pHL1 using primers P6 and P7
SMR22132	MG1655 rph-1 Δattλ::PN25tetR FRT ΔattTn7::PN25tetO SD-N-GFP FRTcatFRT	SMR21475 x pKD46, X Red-mediated short- homology recombineering from pHL2 using primers P6 and P7
SMR22141	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetO SD N-GFP	SMR14328 x P1(SMR22132)
SMR22146	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetO SD N-GFP I-site C	SMR22141 x P1(SMR11768)
SMR22147	MG1655 Δ araBAD567 Δattλ::PBADI-SceI zfd2509.2::PN25tetR FRT ΔattTn7::FRTcatFRT PN25tetO SD N-GFP I-site D	SMR22141 x P1(SMR12725)
SMR25027	MG1655 rph-1 Δattλ::PN25tetR FRT ΔattTn7::PN25tetO SD-N-GFP FRTcatFRT [Ptac]	SMR22132 x Ptac
SMR25029	MG1655 rph-1 Δattλ::PN25tetR FRT ΔattTn7::PN25tetO SD-N-GFP FRTcatFRT [Ptac-Gam]	SMR22132 x Ptac-Gam
Primers	Sequences
P1	GGATTTCCAGGATCCATGTTTGAAGATGCTTTAAATGCCG
P2	GGATTTCCAGTCGACTTACTCATTAATCACCTCAATACT
P3	GGATTTCCAGTCGACCTCATTAATCACCTCAATACTGACA
P4	GGATTTCCAGT CGACGCT AT CGACGAAAACAAACAGAAAG
P5	GGATTTCCAAAGCTTTTATTTGTATAGTTCATCCATGCCA
P6	CGTAACCTGGCAAAATCGGTTACGGTTGAGTAATAAATGGA TGCCGGCGTAGAGGATCGAGATCT
P7	ACGGTTTGTCACATGGAGTTGGCAGGATGTTTGATTAAAAA CATAGCAGCAGCCAACTCAGCTTC

2.2. Phage lambda red gam plaque assays and UV-sensitivity tests for RecBCD nuclease activity

Plaque assays of λΔb1453 Chi⁰ (deleted for the λred and gam genes) were performed as previously [5]. For UV light-sensitivity assays, saturated overnight cultures grown in LBH were diluted 1:100 into fresh LBH medium and grown at 37°C for 90 min with shaking. 100 ng/ml doxycycline was added to induce GamGFP, N, or N-GFP production. After 2 hours induction at 37°C with shaking, cultures were diluted and plated onto LBH solid medium containing 100 ng/ml doxycycline. The LBH plates were irradiated with different UV doses, and incubated at 37°C overnight in the dark, then colonies counted the next day. Control cultures without doxycycline were treated otherwise identically.

2.3. XO-seq and data analysis

The XO-seq protocol, which monitors exonucleolytic DNA loss from DSB ends by whole-genome sequencing, was performed as described [19]. Briefly, TB medium containing 100 ng/mL doxycycline was inoculated at 1:100 dilution with saturated overnight culture and grown at 37°C shaking to an absorbance (OD600) ~0.1, or approximately 2 hours. After reaching logarithmic growth, the 0-minute samples were collected and 0.2% arabinose was added to induce a chromosomal DSB with P_BADI-SceI. Samples were collected 60 minutes after arabinose induction and DNA isolation was performed using the Promega Wizard Genomic DNA Purification kit. The Nextera XT protocol and sample preparation kit was used to prepare libraries for sequencing and alignment on an Illumina MiSeq. The bedGraph files that contain the log2 ratio of normalized coverage in 2Kb bins were generated using deepTools [22]. Plots were generated by R software. All sequencing data are available in the European Nucleotide Archive (ENA) under study accession no. PRJEB28408.

2.4. I-SceI- and DNA-damaging agent-induced focus assays

For cleavage with I-SceI double-strand endonuclease produced from the chromosomal Δattλ::P_BADI-SceI cassette [23, 24], saturated overnight cultures of E. coli carrying chromosomal expression cassettes encoding GamGFP or N-GFP were diluted 1:100 in fresh LBH medium and incubated with 100 ng/ml doxycycline for 1 hour with shaking at 37°C to induce expression. 0.1% arabinose was added to induce I-SceI production and the cultures incubated at 37°C with shaking for 3 hours. Cells were fixed with 2% formaldehyde at room temperature for 20 minutes, then washed twice using PBS, and fluorescent foci visualized using a GE healthcare DeltaVision deconvolution microscope as described [6, 25], and foci quantified with ImageJ software.

For experiments with DNA-damaging agent phleomycin, X-irradiation, or gamma irradiation (Cs-137 source), saturated overnight cultures were diluted 1:100 in fresh LBH (for phleomycin) or M9 0.4% glucose medium (for X-irradiation or gamma irradiation), and grown for 1 hour with shaking at 37°C. 100 ng/ml doxycycline was added to induce GamGFP or N-GFP production. The cultures were grown for an additional hour with shaking at 37°C, induced with 100μM IPTG (only for strains containing Ptac or Ptac-Gam), then treated with 10 μg/ml phleomycin or different doses of irradiation with a Faxitron X-ray machine or radioactive Cs-137 source. Following treatment with either phleomycin or radiation, cells were incubated with shaking at 37°C for 3 hours, then fixed, visualized, and images taken and analyzed as described above.

3. Results

3.1. N and N-GFP do not inhibit RecBCD exonuclease activity strongly in living cells

We constructed a GFP fusion of the phage Mu N protein placed under the control of the doxycycline-inducible P_N25tetO promoter at a non-genic site in the E. coli chromosome, and an isogenic construct encoding N without GFP (Materials and Methods). We tested N-GFP for DNA-damage labeling ability with various DSB-inducing agents, and assayed the ability of N and N-GFP to block RecBCD exonuclease activity in living cells.

We used three assays to test whether N or N-GFP inhibits RecBCD exonuclease activity in living cells (Fig. 1): plaque size of phage λred gam mutants; sensitivity to ultra-violet (UV) light; and whole-genome sequencing XO-seq to measure RecBCD resection dynamics from I-SceI endonuclease-induced DSB ends. In the λ plaque assay, λred gam mutant phage form small plaques on wild-type (RecBCD⁺) E. coli (Fig. 1A) because RecBCD prevents λ rolling-circle replication, so burst (and plaque) size depend on relatively inefficient RecABC-mediated recombination [26]. The λred gam phage form large plaques on E. coli ΔrecB cells, which lack RecBCD nuclease and so allow rolling-circle replication and large bursts and plaques [26](Fig. 1A). We found that induction of neither N nor N-GFP protein allowed production of large plaques (Fig. 1A), indicating that RecBCD nuclease remained mostly functional with N or N-GFP produced in cells. Similarly, producing N or N-GFP caused only very slight sensitivity to UV-induced DNA damage relative to that in wild-type cells (Fig. 1B); not the severe UV-sensitivity observed in RecBCD-nuclease-deficient ΔrecB cells (Fig. 1B). In addition, using XO-seq (eXOnuclease sequencing), a method that measures exonucleolytic degradation of I-SceI-induced DSBs by RecBCD by quantifying reads from whole-genome sequencing [19], we observed that N-GFP production does not prevent I-SceI-induced DNA degradation (Fig. 1D). Altogether, the data indicate that production of neither N nor N-GFP is strongly inhibitory to RecBCD nuclease in E. coli.

Fig. 1.

RecBCD nuclease activity in cells producing N and N-GFP. A. Plaques of phage λred gam mutants are small on RecBCD-exonuclease-proficient (WT) cells and large on RecBCD-exonuclease-deficient (ΔrecB) cells (upper- and lower-left panels, respectively), in which phage λred gam mutants produce more packaged DNA because of their ability to do rolling-circle replication, uninhibited by RecBCD nuclease. Plaques produced on WT cells with the N- and N-GFP-encoding chromosomal expression cassettes remain relatively small, though perhaps somewhat larger than those on cells without the N and N-GFP cassettes, both when the proteins are, and are not, induced. The inducible P_N25tetO promoter used is somewhat leaky [5], such that the slight increase in plaque size even without induction might reflect leaky production of N and N-GFP. B. Production of N or N-GFP does not cause the strong UV sensitivity observed with RecBCD-deficient (ΔrecB) cells, indicating that most DSB ends are not protected from RecBCD by N or N-GFP production. C. The two strains were compared to show that P_N25tetR has no effect on UV sensitivity (p > 0.5 for each dose tested, two-tailed unpaired t-test). Strain CH30 is used as the control in experiments in Fig. 1A,B. D. Production of N-GFP does not block I-SceI DSB-induced degradation measured by XO-Seq in CH7099; n=2, representative plot shown. Red line: I-SceI endonuclease site.

We note that N- and N-GFP-producing cells show similar slight sensitivity to UV light (Fig. 1B, right, Induced). The similarity of the slight UV sensitivity of N and N-GFP imply that the GFP fusion did not alter N activity in this assay, and that N is likely to be functional in the N-GFP protein. N and N-GFP differ from the Mu Gam and GamGFP proteins, overproduction of either of which blocked RecBCD action on linear DNA strongly and caused severe UV sensitivity similar to that of ΔrecB null-mutant cells, allowed large plaque formation of λred gam mutant phage, and also blocked DSB exonucleolytic resection in mouse cells [5].

3.2. N-GFP foci indicate I-SceI-, X-ray, and gamma ray-induced DSBs in E. coli

We used a chromosomal inducible I-SceI site-specific double-strand endonuclease [23, 24] to generate DSBs at sites in the E. coli chromosome during log-phase growth per [5], to test the possible ability of N-GFP to label or indicate DSB ends. In these experiments, the chromosomes of rapidly growing E. coli, which have multiple replication forks, are cleaved either near to or far from the replication origin, oriC, where there are more or fewer copies of the chromosomal DNA, respectively, generating more or fewer DSBs per cell (Fig. 2A) per [5]. We observed that N-GFP formed foci that correspond with the numbers of I-SceI-induced DSBs, with more foci per cell in cells cleaved near the ori, and fewer foci per cell in cells cleaved near the replication terminus (Fig. 2B,C). There are significantly more foci with ter-proximal cleavage than with No DSB, and more cells having multiple foci with ori-proximal than ter-proximal cleavage (p = 7×10⁻⁴, and 0.002, respectively, two-tailed unpaired t-test, Fig. 2B,C). The data indicate that N-GFP foci correspond with and may label DSBs.

Fig. 2.

N-GFP foci correspond with DSB numbers in single E. coli cells. A. Diagram of E. coli chromosome cleavage with I-SceI endonuclease at engineered DSB cut sites (red arrows) designed to produce more and fewer DSBs per cell in replicating E. coli, because of their proximity to or distance from the replication origin. Proliferating cells have more chromosome copies near ori than near ter, and thus will have more DSBs per cell when cleaved by I-SceI ori-proximally than ter-proximally. B. Representative images show comparison of numbers of DSB foci per cell for GamGFP and N-GFP. Cells have multiple foci when cleaved by I-SceI near ori, and mostly 1 focus per cell when cleaved by I-SceI near ter, with more efficient DSB-detection by GamGFP than N-GFP. C. Quantification of I-SceI-induced N-GFP and GamGFP foci. D. Representative images show that N-GFP foci are positively correlated with DSB-inducing X-ray doses, although with lower detection efficiency than GamGFP. E. Quantification of X-ray-induced N-GFP and GamGFP foci. F. Production of the DSB-specific end-binding protein phage Mu Gam blocks most gamma-ray-induced N-GFP focus accumulation indicating DSB end dependence of these N-GFP foci.

Also showing that N-GFP foci indicate DSBs, we observed that N-GFP forms foci dose-dependently with X-irradiation (Fig. 2D,E), which causes mostly DSBs, among other lesions, in DNA [27]. The numbers of N-GFP foci are positively correlated with doses of X rays (Pearson correlation coefficient r = 0.996, p =0.03), again supporting N-GFP foci representing and/or labeling DSBs in DNA.

For both I-SceI- and X-ray-induced DNA damage, N-GFP formed foci four- to six-times less efficiently than GamGFP at the same DNA-damaging doses (Fig. 2B-E), indicating that GamGFP is more sensitive for detection of DSBs as fluorescent foci in E. coli than N-GFP is. Previous work indicated that GamGFP detects about 70% of DSBs induced in E. coli [5], which would imply that N-GFP may detect only about 12%−18% of DSBs present in the cells.

Additionally, we show that N-GFP focus formation with DSB-generating DNA damage is DSB dependent, by showing that formation of N-GFP foci is inhibited by co-production of the biochemically-demonstrated DSB-end protection protein phage Mu Gam. For these experiments, we used DSB-generating gamma rays, and see that as with I-SceI and X-rays, N-GFP foci are induced by gamma irradiation (Fig. 2F). Moreover, production of DSB-specific binding and trapping protein, Mu Gam [5], reduced most (about 60% of) of gamma-ray-induced N-GFP foci (Fig. 2F) (p=0.02, two-tailed unpaired t-test). The data indicate that interaction of N protein with DSB ends is required for accumulation of N-GFP foci, and support the conclusion that N-GFP can label DSB ends.

3.3. More efficient N-GFP focus formation with phleomycin than I-SceI indicates labeling of DSB and non-DSB structures

We assayed N-GFP foci in cells treated with radiomimetic drug phleomycin, an oxidizing agent that causes single-strand gaps and DSBs in DNA [28]. The data in In Fig. 3A and B show that N-GFP forms foci in response to phleomycin, with more cells displaying foci with 10 μg/ml phleomycin than in untreated cells (p = 1.3×10⁻⁵, two-tailed unpaired t-test), implying DNA-damage labeling by N-GFP. Moreover, unlike focus formation in response to I-SceI and X- and gamma-rays, N-GFP forms foci relatively efficiently with phleomycin compared with GamGFP showing about half as many foci per cell as GamGFP with phleomycin. This ratio is higher than the N-GFP : GamGFP focus ratio with the more purely DSB-inducing X-rays (Fig. 2D,E) and I-SceI (Fig. 2A-C), which showed 4-to-6 times fewer N-GFP foci than GamGFP foci. These data suggest that although N-GFP forms foci in response to DSBs (Fig. 2), it may additionally label non-DSB DNA damage, and that phleomycin produces significant DNA damage other than DSBs at the 10 μg/ml dose used here. Supporting this hypothesis, we observed that producing DSB-end-trapping protein Gam did not reduce phleomycin-induced N-GFP foci significantly (Fig. 3C) (p=0.98, two-tailed unpaired t-test), implying that most of the phleomycin-induced DNA damage recognized by N-GFP is not DSBs, and so cannot be blocked by Gam. This differs from the results with DSB-generating gamma rays (Fig. 2F). The data imply that N-GFP foci are an indicator of DNA damage, and, unlike GamGFP foci, are general, not specific to DSBs.

Fig. 3.

N-GFP foci in phleomycin-treated E. coli imply focus formation in response to DSB and non-DSB DNA damage. A. Representative images show N-GFP foci and GamGFP foci induced by radiomimetic drug phleomycin. B. Quantification of phleomycin-induced N-GFP and GamGFP foci indicates many more N-GFP foci relative to GamGFP foci with phleomycin than was seen with I-SceI- and X-ray-induced DSBs in Fig. 2, suggesting that in addition to DSBs, N-GFP foci indicate non-DSB DNA damage, which GamGFP does not label. C. Gam production does not block most phleomycin-induced N-GFP foci, implying that N-GFP can label phleomycin-induced non-DSB DNA damage, which is not protected by Gam.

4. Discussion

In previous work, we found that GamGFP is an excellent binder and labeler of DSB ends in E. coli and mammalian cells, and inhibited DSB-exonucleolytic resection activity of RecBCD in E. coli and resection exonucleases in mammalian cells [5]. However, we observed somewhat more, and brighter, GamGFP foci in Ku^−/− mouse cells than in Ku⁺ mouse cells, suggesting that GamGFP binding to DSB ends might be somewhat inhibited by competition with Ku [5], a Gam orthologue [29, 30]. Because Ku is a Gam orthologue, and both proteins are predicted to bind DSB ends similarly [29, 30], we wondered whether N might substitute, because N is a suggested linear-DNA-interacting protein [11–13] and shares no apparent sequence identity with Gam and Ku. Thus it might have worked as well for DSB binding, but potentially might not be outcompeted by Ku in mammalian cells, and in Ku⁺ bacterial species such as Bacillus subtilis [30], (because of a presumably different DSB/linear DNA-interacting mechanism).

In this work, we found that N-GFP foci correspond with numbers of DSB ends produced by I-SceI cleavage and X-irradiation (Fig. 2B-E), supporting the utility of N-GFP for labeling DSB ends in E. coli. Although N-GFP foci formed proportionally to DSBs produced by I-SceI and X-rays, N-GFP formed about 4-to-6-times fewer foci per cell than GamGFP with these DSB-inducing treatments (Fig. 2B-E). The data indicate less efficient labeling of DSB ends by N-GFP than GamGFP in E. coli. Because GamGFP was estimated to label about 70% of DSBs in E. coli as visible foci [5], we infer that N-GFP labels 12%−18% as foci. Additionally, N-GFP formed foci relatively more robustly, with respect to GamGFP, in phleomycin-treated cells, with about half as many N-GFP foci induced as GamGFP foci at the same dose (Fig. 3A,B)—many more N-GFP foci than the 4–6-times fewer than GamGFP seen with I-SceI and X-rays (Fig. 2B-E). These data suggest, first, that N-GFP may bind and label DNA damage other than and in addition to DSB ends, which is created in cells treated with phleomycin, and second, that phleomycin causes significant non-DSB DNA damage to cells, which N-GFP detects, but which is undetectable by GamGFP. Supporting this interpretation, blocking DSB ends by co-production with Gam inhibited N-GFP focus accumulation from DSB-causing gamma rays (Fig. 2F), but not substantially from phleomycin (Fig. 3C). The data support the interpretation that much of the phleomycin-induced DNA damage detected by N-GFP is not DSBs. N-GFP may thus be useful for labeling DNA damage, but appears not to be specific for DSBs. N-GFP may be more useful as a general DNA-damage detection reagent.

We also found that N and N-GFP showed only weak or no inhibition of RecBCD nuclease activity on ends of phage λ DNA (Fig 1A), in E. coli DNA damaged by UV light (Fig 1B), and showed no interference with RecBCD resection of I-SceI cleaved DNA shown by XO-seq whole genome sequencing (Fig 1D). This may reflect the relatively inefficient labeling of DSBs by N-GFP relative to GamGFP (Fig. 2) or could result from N interacting with linear DNA transiently in an on/off manner. With only about 12–18% of DSBs estimated to be detected as N-GFP foci (Fig. 2, compare with 70% bound and labeled by GamGFP [5]), most of the DNA is unbound, unprotected and so can be degraded by RecBCD, at least when N/N-GFP are produced with no other phage Mu proteins in E. coli cells, as was done here. This is compatible with the suggested role of N in holding linear phage Mu chromosomes in a non-covalent circular complex [11–13].

In summary, N-GFP may be a useful general DNA-damage-detection reagent that detects DSBs and potentially non-DSB DNA damage. In E. coli GamGFP is more sensitive for DSB-detection, whereas N-GFP may be more useful for DNA-damage types in addition to DSBs. Whether N-GFP performs better or worse than GamGFP for DSB detection in mammalian cells with Ku present, or Ku⁺ bacterial species or yeast, has not been evaluated here, and remains to be examined in future work. However, the N-GFP lack of stringent DSB specificity observed here in E. coli, could complicate interpretations of focus numbers in mammalian cells with or without Ku, making the case for N-GFP as a potential replacement for Gam in mammalian cells less compelling than had it been DSB specific. N-GFP might potentially function as a DNA-damage detector for Ku⁺ cells/species, but our data indicate that the foci could indicate general, not DSB-specific damage to DNA.