Landscape of New Nuclease-Containing Antiphage Systems in Escherichia coli and the Counterdefense Roles of Bacteriophage T4 Genome Modifications

ABSTRACT

Many phages, such as T4, protect their genomes against the nucleases of bacterial restriction-modification (R-M) and CRISPR-Cas systems through covalent modification of their genomes. Recent studies have revealed many novel nuclease-containing antiphage systems, raising the question of the role of phage genome modifications in countering these systems. Here, by focusing on phage T4 and its host Escherichia coli, we depicted the landscape of the new nuclease-containing systems in E. coli and demonstrated the roles of T4 genome modifications in countering these systems. Our analysis identified at least 17 nuclease-containing defense systems in E. coli, with type III Druantia being the most abundant system, followed by Zorya, Septu, Gabija, AVAST type 4, and qatABCD. Of these, 8 nuclease-containing systems were found to be active against phage T4 infection. During T4 replication in E. coli, 5-hydroxymethyl dCTP is incorporated into the newly synthesized DNA instead of dCTP. The 5-hydroxymethylcytosines (hmCs) are further modified by glycosylation to form glucosyl-5-hydroxymethylcytosine (ghmC). Our data showed that the ghmC modification of the T4 genome abolished the defense activities of Gabija, Shedu, Restriction-like, type III Druantia, and qatABCD systems. The anti-phage T4 activities of the last two systems can also be counteracted by hmC modification. Interestingly, the Restriction-like system specifically restricts phage T4 containing an hmC-modified genome. The ghmC modification cannot abolish the anti-phage T4 activities of Septu, SspBCDE, and mzaABCDE, although it reduces their efficiency. Our study reveals the multidimensional defense strategies of E. coli nuclease-containing systems and the complex roles of T4 genomic modification in countering these defense systems.

IMPORTANCE Cleavage of foreign DNA is a well-known mechanism used by bacteria to protect themselves from phage infections. Two well-known bacterial defense systems, R-M and CRISPR-Cas, both contain nucleases that cleave the phage genomes through specific mechanisms. However, phages have evolved different strategies to modify their genomes to prevent cleavage. Recent studies have revealed many novel nuclease-containing antiphage systems from various bacteria and archaea. However, no studies have systematically investigated the nuclease-containing antiphage systems of a specific bacterial species. In addition, the role of phage genome modifications in countering these systems remains unknown. Here, by focusing on phage T4 and its host Escherichia coli, we depicted the landscape of the new nuclease-containing systems in E. coli using all 2,289 genomes available in NCBI. Our studies reveal the multidimensional defense strategies of E. coli nuclease-containing systems and the complex roles of genomic modification of phage T4 in countering these defense systems.

INTRODUCTION

The genomic DNAs of many phages, particularly Escherichia coli and Salmonella enterica phages, are hypermodified, and more than 21 modified nucleotides have been identified thus far (1, 2). Many phages encode their own enzymes to modify the genome (3, 4), indicating the important roles of these modifications in their life cycle, which are not yet fully understood. Previous studies have demonstrated that genomic DNA modifications are the main mechanisms by which phages resist bacterial restriction-modification (R-M) defense systems that cleave phage DNAs at specific sites while protecting their own genomes (5). One well-studied example is phage T4, the genomic DNA of which contains glucosyl-5-hydroxymethylcytosine (ghmC) instead of cytosine (6, 7). During T4 DNA replication in E. coli, the 5-hydroxymethyl dCTP, instead of dCTP, is incorporated into the newly synthesized DNA. The 5-hydroxymethylcytosines (hmCs) are further glucosylated by phage-encoded α- and β-glucosyltransferase, making T4 genomic DNA resistant to almost all of the R-M systems (types I to III) (8–11). However, the remaining type IV R-M systems specifically target T4 phages with different genome modifications; for instance, the GmrSD system cleaves ghmC-modified T4 DNA (12), whereas the McrBC system specifically targets hmC-modified DNA (13).

Recent studies revealed that the ghmC modification also endows T4 phage with variable resistance to different types of CRISPR-Cas systems (14–17), the acquired immune systems of bacteria against phages through the assembly of Cas-CRISPR RNA (crRNA) complexes that specifically recognize and cleave phage genomic DNA (3, 18). However, bacteria might express different Cas-crRNA complexes targeting independent sites on the phage genome to defend themselves more effectively against phages with a modified genome (14, 19). Additionally, the partial resistance of phage T4 to CRISPR-Cas systems not only allows for the survival of some E. coli cells but also accelerates the evolution of phages (20–22), highlighting the complex roles of genomic DNA modifications in the coevolution of phages and their hosts.

Other than R-M and CRISPR-Cas systems, recent studies have revealed many novel antiphage systems from a variety of bacteria and archaea that contain nuclease domains, such as Zorya, Septu, Gabija, Shedu, Druantia, SspABCD-SspE, mzaABCDE, STAND, and qatABCD (23–26). Although all contain at least one nuclease domain, it is unknown whether the nuclease domains are necessary for their antiphage activities and whether these nucleases directly degrade phage genomic DNAs. In addition, no studies have systematically investigated the nuclease-containing antiphage systems of a specific bacterial species.

In focusing on phage T4 and its E. coli host, the aim of this study was to explore the landscape of new nuclease-containing systems in E. coli and the roles of T4 genome modifications in countering these systems. We found abundant nuclease-containing defense systems in E. coli, with type III Druantia being the most abundant system, followed by Zorya, Septu, Gabija, AVAST type 4, and qatABCD. The divergent defense effects of these systems were observed against T4 (ghmC), T4 (hmC), and T4 (C). Overall, genome modifications reduce the defense activities of many nuclease-containing systems, whereas some systems specifically or more efficiently restrict T4 (hmC). Our studies revealed the multidimensional defense strategies of E. coli nuclease-containing systems and the complex roles of genomic modification of phage T4 in countering these defense systems other than simply blocking the degradation of its genomic DNA. Although we did not explore mechanisms of each system in depth due to too many new nuclease-containing systems, our work sets a foundation for further mechanistic dissection of the various defense mechanisms.

RESULTS

Identification of antiphage systems with nuclease domains in E. coli.

Since CRISPR-Cas and R-M systems have been well studied, here we focus on newly identified nuclease-containing defense systems. Twenty types of new nuclease-containing defense systems have been identified from a variety of bacteria and archaea as of October 2020, including 12 systems identified in E. coli (see Table S1 in the supplemental material). To determine whether the remaining 8 of the originally identified 20 defense systems are present in E. coli, we downloaded all 2,289 E. coli genomes available in NCBI through October 2020. The open reading frames of each genome were annotated and analyzed for domain structures with HMMER (27). We identified 5 potential defense systems that share similar domain structures with Gabija, Septu, Shedu, mzaABCDE, and type III Druantia, respectively (Fig. 1A), whereas DISARM (28), AVAST type 3 (25), and Ago (29) were not found. The E. coli defense systems Gabija, Septu, Shedu, and type III Druantia showed very low homology to the corresponding systems reported in other bacteria, whereas mzaABCDE shared 93% sequence similarity with Salmonella enterica (Table S2).

FIG 1.

Identification of antiphage systems with nuclease domains in E. coli. (A and B) Identification of E. coli Gabija, Septu, Shedu, mzaABCDE, and type III Druantia systems. The domain organization (A) of each system is shown, and their defense activities (B) were determined using E. coli MG1655 against 111 different phages as described in Materials and Methods. (C) The abundance of all 17 experimentally confirmed defense systems containing nuclease domains in the 2,289 E. coli genomes.

Using a collection of 816 E. coli strains isolated from different sewage and clinical samples, we were able to clone all the 5 potential defense systems (Fig. 1A and B and Table S3). The coding sequence and native promoter of each system were cloned to the pSEC1 vector (see Materials and Methods), which maintains a low copy number in E. coli cells due to the p15A origin of replication and has been widely used to evaluate the antiphage activities of defense systems (23, 25, 30). The defense activity of these systems was determined using E. coli MG1655 against 111 different phages isolated from farms. E. coli MG1655 containing an empty pSEC1 vector was used as a control, in which no antiphage activity was detected. As shown in Fig. 1B, Gabija and type III Druantia have broad antiphage activities with greater than 10³-fold protection against many phages, and the remaining systems also have defense activities, although they fall within a narrow spectrum.

An analysis on the abundance of all 17 experimentally confirmed E. coli defense systems containing nuclease domains in the 2,289 genomes showed that type III Druantia is the most abundant system, followed by Zorya, Septu, Gabija, AVAST type 4, qatABCD, and hhe, whereas upx, DUF4297-HerA, and Retron-Eco8 are rare (Fig. 1C). More than half of the strains (1,267 of 2,289) contained at least one nuclease-containing system, 10 strains even contained 2 copies of the same system (Table S4), and 424 strains contained more than one nuclease-containing system (Fig. S1). These results demonstrated that divergent nuclease-containing defense systems are employed by E. coli, in addition to the well-known R-M and CRISPR-Cas systems. The abundance of these systems indicated that phage genomic DNAs could be the major targets for E. coli to restrict infection; therefore, it is not surprising that many phage genomes are hypermodified.

To determine the roles of phage genome modifications in countering nuclease-containing systems, we screened our E. coli collections and were able to clone 14 systems (Table S5). Their defense activities were determined with T4 (ghmC), T4 (hmC), and T4 (C) and are discussed below.

Genome modifications of phage T4 cannot abolish the antiphage activities of Septu, SspBCDE, and mzaABCDE.

To determine the antidefense effects of genomic DNA modifications against nuclease-based defense systems, we first deleted the Alpha-gt and Beta-gt genes encoding α- and β-glucosyltransferase, respectively, to generate a T4 mutant, T4 (hmC), which contains hmCs within its genomic DNA (Fig. S2A and Table S6). The T4 (C), which contains unmodified cytosines, was constructed by deleting gene 56 encoding a cytosine methyltransferase that is responsible for the hmC modification (Fig. S2B) (31). The genomic DNA modifications of T4 (ghmC), T4 (hmC), and T4 (C) were confirmed by incubating phage genomic DNAs with two nucleases, which specifically degrade DNAs with different modifications as reported previously (Fig. S2C) (12, 15).

Since the growth of the T4 mutants, especially T4 (C), in E. coli cells containing R-M and CRISPR-Cas systems was severely impaired, we therefore used E. coli DH10B, which lacks these systems, as a host to determine the antiphage activities of the new nuclease-containing defense systems. Plaque assays showed that Septu, mzaABCDE, and SspBCDE systems confer various degrees of protection on E. coli DH10B against T4 (ghmC), T4 (hmC), and T4 (C) (Fig. 2A). As a control, E. coli DH10B cells containing an empty vector showed no antiphage activity (Fig. 2A). The Septu and mzaABCDE systems showed similar antiphage activities against T4 (hmC) and T4 (C) phages, with 10³- to 10⁴-fold protection compared to the empty vector control. However, the defense activity of the mzaABCDE system decreased more strongly than that of Septu when T4 (ghmC) was used for infection (10⁴-fold versus 10¹-fold reduction) (Fig. 2A).

FIG 2.

Genome modifications of phage T4 cannot abolish the antiphage activities of Septu, SspBCDE, and mzaABCDE. (A) Genetic compositions and defense activities of E. coli Septu, SspBCDE, and mzaABCDE systems. The nuclease domains of each system are highlighted in red, and the putative active sites or the conserved amino acids in the nuclease domain are indicated with red stars. The representative results of plaque assays show the defense activities of each system against T4 (ghmC), T4 (hmC), and T4 (C). The number of phages (10¹ to 10⁶ PFU) used for plaque assays is indicated on top of each panel. (B to D) The antiphage activities of wild-type and mutated Septu (B), SspBCDE (C), and mzaABCDE (D) systems against phages T4 (ghmC), T4 (hmC), and T4 (C) are presented as fold reduction in efficiency of plating (EOP) (see Materials and Methods for the details). Data are represented as mean ± SD from three independent assays. ns, not significant; *, P < 0.05; **, P < 0.01; ***, P < 0.001; ****, P < 0.0001 (analysis of variance).

The SspABCD-SspE system was identified recently in Vibrio cyclitrophicus, E. coli, and Streptomyces yokosukanensis, and it restricts phages by introducing nicks into genomic DNAs via the nickase SspE (24). Proteins SspA, -B, -C, and -D work together to replace the oxygen atoms in the genomic DNA backbone with sulfurs in a sequence-specific manner, known as phosphorothioate (PT) modification, which stimulates the nickase activity of SspE. Both PT modification and the SspE nickase are necessary for the antiphage activity of the V. cyclitrophicus SspABCD-SspE system (24). Interestingly, the E. coli SspABCD-SspE system cloned from our E. coli collection does not have the sspA gene (Fig. 2A), and the SspBCDE cassette is active against T4 phages (Fig. 2A). However, antiphage efficiencies depend on the modifications of phage genomic DNA, with 10⁵-fold, 10⁴-fold, and 10³-fold reductions for phages T4 (C), T4 (hmC), and T4 (ghmC), respectively (Fig. 2A). These results revealed that E. coli SspBCDE uses a different mechanism for defense against T4 infection than that of the V. cyclitrophicus SspABCD-SspE system, which relies on PT modification.

To determine whether the antiphage activities of each system depend on nuclease activity, we mutated the putative active sites or the conserved amino acids in the nuclease domain of each system to alanine (Fig. 2). Gene B of the Septu system encodes the PF13395 HNH endonuclease domain, which contains a conserved (D/E) HxxP motif. A single point mutation at H in the (D/E) HxxP motif inactivates the endonuclease of Streptococcus pyogenes Cas9 (32). We therefore mutated H75A in the (D/E) HxxP motif as well as K103A, which is also conserved in the PF13395 domain. Our results showed that mutation at site H75A or K103A completely abolished the defense activity of Septu against phage T4 regardless of genomic DNA modifications (Fig. 2B).

A previous study found that a single point mutation at G in the DGQQR motif and a single point mutation at H in the EHxxP motif of S. yokosukanensis SspE severely impaired its GTPase and nicking activities, respectively (24). Our results showed that mutation G130A in the DGQQR motif of SspE abolished the defense activity of E. coli SspBCDE against T4 phages (Fig. 2C). Interestingly, mutation H688A in the conserved EHxxP motif of the SspE nickase catalytic center completely abolished the defense activities against T4 (C) and T4 (ghmC) but did not affect the activity against T4 (hmC) (Fig. 2C). These results indicated that both GTPase and nickase activities of SspE are necessary for SspBCDE antiphage activity. However, the nickase activities of SspE may depend on the modifications of phage genomic DNA, and mutation of the key residue may severely decrease its nickase only against specific modifications such as hmC. More experiments are needed to further investigate its mechanism.

Domain analysis revealed that mzaD of the mzaABCDE system belongs to the PD-(D/E) XK nuclease superfamily (33). We therefore mutated the first aspartate residue of the PD-(D/E) XK motif, D187, in mzaD, which is critical for its nuclease activity, as well as the conserved proline residue P297. Plaque assays showed that the mutations at sites D187A and P297A abolished the antiphage activity against T4 (ghmC). However, the mzaABCDE mutants still showed antiphage activities against T4 (C) and T4 (hmC), although with significant decreases, especially for T4 (C) (Fig. 2D). These results indicated that modifications of genomic DNA affect mzaD nuclease activity, which is critical for mzaABCDE antiphage activity.

Glucosylation of hmC in the phage T4 genome abolishes Gabija and Shedu defense activities.

We found that Gabija and Shedu protected E. coli DH10B cells against both T4 (C) and T4 (hmC) phages. However, their defense activities were completely blocked by T4 (ghmC) (Fig. 3A and B). Shedu showed higher defense activity against T4 (C) than T4 (hmC) (10³-fold versus 10²-fold reduction compared to empty vector, Fig. 3B). Interestingly, Gabija showed higher defense efficiency against T4 (hmC) than T4 (C) (10³-fold versus 10¹-fold reduction, Fig. 3A).

FIG 3.

Glucosylation of hmC in phage T4 genome abolishes Gabija and Shedu defense activities. (A and B) Genetic compositions and defense activities of E. coli Gabija (A) and Shedu (B) against T4 (ghmC), T4 (hmC), and T4 (C). The nuclease domains are highlighted in red, and the putative active sites in the nuclease domain are indicated with red stars. The representative results of plaque assays are shown. (C and D) The antiphage activities of wild-type (WT) and mutated Gabija (C) and Shedu (D) against phages T4 (hmC) and T4 (C) are presented as fold reduction in EOP compared to the empty vector control. Data are represented as mean ± SD. **, P < 0.01; ****, P < 0.0001 (Student’s t test).

A previous study has shown that a single mutation at E379A of Bacillus cereus GajA completely abolished its endonuclease activity (26). Therefore, we mutated the conserved E466A residues in the E. coli GajA, which is equivalent to E379A in Bacillus cereus GajA. Plaque assays showed that the E466A mutation completely abolished the defense activity of Gabija against both T4 (C) and T4 (hmC) (Fig. 3A and C), indicating that the antiphage activity of E. coli Gabija depends on the nuclease activity of GajA.

Domain analysis revealed that Shedu contains a DUF4263 domain, which also belongs to the PD-(D/E) XK nuclease superfamily (33). However, the PD-(D/E) XK motif is not as well conserved as mzaD. Therefore, we mutated the conserved Q110A residues to determine the critical site for the Shedu antiphage activity. Plaque assays showed that mutations at site Q110A also completely abolished the defense activity of Shedu (Fig. 3B and D). However, further experiments are needed to determine whether this point mutation affects the nuclease activity of Shedu. These results indicated that both the Gabija and Shedu systems restrict T4 (C) and T4 (hmC) by directly targeting phage DNAs, and ghmC modification endows T4 with resistance to these systems.

The Restriction-like defense system specifically targets phage T4 (hmC).

The Restriction-like defense system was first identified in 2020 and is composed of four proteins containing the nuclease domain, ATPase domain, methylase domain, helicase domain, and several uncharacterized domains (Fig. 4A) (25). This system has been shown to be active against phages P1, λ, and M13 with variable efficiency but inactive against phages T2, T3, T4, T5, and T7 (25). Our plaque assays confirmed that the Restriction-like system lacks defense activity against T4 (ghmC) (Fig. 4B). However, we found that it can specifically protect E. coli DH10B cells against phage T4 (hmC), with 10²-fold protection, but not against phage T4 (C) (Fig. 4B). The A protein of the Restriction-like defense system contains a res subunit of the type III restriction enzyme (pfam04851), whose nuclease active site is unclear. However, 13 residues in the nuclease domain are highly conserved (>80% identity), such as Q109, G132, G134, K135, D230, and H233 (Fig. S3). The plaque assays showed that mutations changing the site G132A, K135A, H233A, or T269A in the nuclease domain did not affect the antiphage activity of the Restriction-like system against phage T4 (hmC). Interestingly, mutations at sites Q109A and D230A even increased the defense activity (Fig. 4C). These results indicated that the anti-T4 (hmC) activity of the Restriction-like defense system most likely does not depend on nuclease activity. However, more nuclease essays are needed to further confirm this speculation.

FIG 4.

The Restriction-like defense system specifically targets phage T4 (hmC). (A) The cartoon shows the genetic composition and protein domain of the Restriction-like system. The main functional domains are indicated in different colors. (B) The representative results of plaque assays show the defense activity of the Restriction-like system against T4 (ghmC), T4 (hmC), and T4 (C). The number of phages (10¹ to 10⁶ PFU) used for plaque assays is indicated on top of each panel. (C) Phage plaque assays on E. coli DH10B cells expressing wild-type or mutated Restriction-like systems. E. coli cells containing empty vectors were used as controls. The mutation sites Q109A, G132A, K135A, D230A, H233A, and T269A in the nuclease domain are indicated.

Druantia and qatABCD systems confer protection only against phage T4 (C).

The Druantia defense system was discovered in 2018 and is characterized by a very large protein, DruE (~2,000 amino acids long and containing an unannotated domain, DUF1998), accompanied by 1 to 4 accessory proteins (23). Three types of Druantia systems (I to III) have been reported based on the number of accessory proteins; however, in our current study, only types I and III were found in E. coli, with type III being the dominant system (Fig. S4A). The type I Druantia system was shown to have defense activity against wild-type phage T4, but its DruE does not contain a nuclease domain (23). Therefore, we cloned the type III Druantia, which contains only one accessory protein and is the most abundant nuclease-containing system in E. coli (Fig. 1C). We found that the type III Druantia system has no defense activity against T4 (ghmC) and T4 (hmC) but is highly active against phage T4 (C), with 10⁶-fold protection (Fig. 5A). These results revealed a defense and counterdefense mechanism between E. coli Druantia and phage T4. Namely, the phage resists type III Druantia via ghmC modification of its genome, whereas E. coli cells employ type I Druantia to overcome this resistance (23). Another nuclease-containing system that targets only phage T4 (C) rather than T4 (ghmC) and T4 (hmC) is qatABCD (Fig. 5B), which was identified in 2020 as a defense system against phages P1, T3, and λ but not against phages T2, T4, T5, and T7 (25). However, the activity of qatABCD against T4 (C) was quite low compared to that of the type III Druantia system (10¹-fold versus 10⁶-fold reduction, Fig. 5A and B).

FIG 5.

Druantia and qatABCD systems confer protection only against phage T4 (C). (A and B) The genetic composition and defense activities of E. coli Druantia (A) and qatABCD (B) against T4 (ghmC), T4 (hmC), and T4 (C). The genetic composition of each system, the nuclease domains (highlighted in red), and the putative nuclease active sites (red stars) are shown on top of each panel. The representative results of plaque assays are shown at the bottom. (C and D) The defense activities of wild-type and mutated Druantia (C) and qatABCD (D) against phage T4 (C) are presented as fold reduction in EOP compared to the empty vector control. Data are represented as mean ± SD from three independent assays. **, P < 0.01; ****, P < 0.0001 (Student’s test or analysis of variance).

Similar to the A protein of the Restriction-like defense system, DruH of type III Druantia also contains a res subunit of the type III restriction enzyme domain (pfam04851). The plaque assays showed that the K124A mutation completely abolished the defense activity of type III Druantia against phage T4 (C) (Fig. 5A and C). Domain analysis revealed that qatD of the qatABCD system contains a pfam01026 domain and belongs to the metalloenzyme superfamily. A previous study of E. coli TatD DNase, a member of the metalloenzyme superfamily, has showed that sites D203 and H62 are the binding site of metal ions and a critical catalytic site, respectively (34). The sites H59 and D193 of qatD are equivalent to TatD DNase H62 and D203, respectively. We found that mutations at sites H59A and D193A completely abolished the defense activity of qatABCD against T4 (C) (Fig. 5B and D). These results indicated that nuclease activity is necessary for qatABCD to defend E. coli cells against phage T4 (C).

Six nuclease-containing systems lack defense activities against phage T4.

We found six recently identified nuclease-containing systems, namely, Zorya, hhe, ppl, AVAST type 4, Retron Ec78, and Retron Ec67-like, which lack defense activities against phages T4 (ghmC), T4 (hmC), and T4 (C) (Fig. 6A). There are two types of Zorya systems, both of which contain a nuclease domain. We were able to clone type II Zorya, which contains an HNH nuclease domain, from our E. coli collection. Plaque assays showed that type II Zorya has defense activity against phage T7, and mutating the putative active site in the nuclease domain significantly reduced the defense activity (Fig. 6A and B). However, type II Zorya is not active against phage T4 regardless of the genome modifications (Fig. 6A), indicating that T4 possesses counterdefense mechanisms other than genome modifications.

FIG 6.

E. coli Zorya, hhe, ppl, AVAST type 4, Retron Ec78, and Retron Ec67 systems lack defense activities against phage T4. (A) The genetic composition and defense activities of each system against phages T4 (ghmC), T4 (hmC), and T4 (C). The genetic compositions, the main functional domains, and the putative nuclease active sites (red stars) are shown on the left side. The antiphage activities were determined by plaque assays, and representative results are shown. (B to F) Antiphage activities of wild type and mutants of each system against phage T7 or phage #76. Data are represented as mean ± SD from three independent assays. *, P < 0.05; **, P < 0.01; ***, P < 0.001 (Student’s test or analysis of variance).

The AVAST type 4 system contains an Mrr-like nuclease domain and has defense activity against phages T3, T7, and ΦV-1 (25). In our study, we also found that it can protect E. coli cells against phage T7 but with very low defense activity, which was abolished when the putative active site in the nuclease domain was mutated (Fig. 6A and C). Similarly, we did not find any defense activity of AVAST type 4 against T4 (ghmC), T4 (hmC), or T4 (C) (Fig. 6A).

The hhe system contains a Vsr (very short patch repair) endonuclease domain and is active against phages λ, T3, T7, and ΦV-1 (25). The ppl system contains a nuclease domain at its NH terminus and was also shown to be active against phages λ, T3, T7, and ΦV-1 (25). In our study, we found that both systems have very weak defense activities against phage T7, and mutations at sites E1731A of hhe and H12A of ppl abolished the antiphage activities of each system (Fig. 6A, D, and E). Similar to type II Zorya and type 4 AVAST, both the hhe and ppl systems are incapable of restricting T4 infection regardless of T4 (ghmC), T4 (hmC), or T4 (C) (Fig. 6A).

Retron Ec78 has an HNH nuclease domain and was shown to have defense activity against phage T5, which was abolished when the putative active site (H57A) in the nuclease domain was mutated (25, 30). However, our phage T5 stock, which showed sensitivity to Retron Ec86 in our recent study (35), can completely tolerate Ec78. The defense activity of Ec78 was confirmed using phage #76, which was completely abolished when the nuclease active site H57A was mutated (Fig. 6A and F). However, Ec78 did not show defense activity against either wild-type T4 or the mutants. In addition, we identified a Retron system in E. coli that contains a predicted HNH nuclease domain and has a genetic organization similar to that of Retron Ec67. However, this Ec67-like system did not show defense activity against T4 (ghmC), T4 (hmC), or T4 (C) (Fig. 6A).

Previous studies indicated that capsid-targeted internal protein 1 (IPI) of T4, which is packaged in mature phage particles and injected into host cells during infection, can protect T4 genomic DNA from type IV R-M nucleases (36). To determine whether IPI gives T4 phage resistance to these six nuclease-containing systems, we deleted the IPI gene from the wild-type T4 genome (Fig. S5A). Plaque assays showed that the T4 ΔIPI mutant is still resistant to these nuclease-containing systems (Fig. S5B), indicating that T4 phage might employ other mechanisms to evade these defense systems.

DISCUSSION

Phage T4 is a widely used model to explore the mechanisms of phages in resisting the nucleases of E. coli R-M systems, leading to the discovery that the ghmC modification is one of the most powerful covalent modifications that make T4 DNA resistant to most nucleases of type I to III R-M systems (8–11). This finding has led to the concept that modifying genomic DNA could be a key strategy of phages to counteract nuclease-based bacterial defense systems. Recently, many new defense systems containing nuclease domains have been identified from various bacteria and archaea (23, 25); however, many of them have not been identified or experimentally validated in E. coli. Here, we systematically analyzed the nuclease-containing defense systems employed by E. coli and the corresponding counterdefense activities of T4 genome modifications.

By analyzing all genome sequences available in NCBI, we found that E. coli employs at least 17 of 20 recently identified nuclease-containing systems to protect itself against phages. We were able to clone 14 systems from our E. coli collections and test their protection activities. In particular, Gabija, Septu, Shedu, mzaABCDE, and type III Druantia were first identified and experimentally confirmed in E. coli. The Gabija system has broad antiphage activity (97 of 111 phages), whereas Septu and Shedu inhibit only 10 to 11 phages. Druantia systems are divided into three types (I to III), and only type I has been experimentally validated in E. coli (23). Our results showed that Druantia is the most abundant nuclease-containing system in E. coli (667 of 2,289 strains). However, only types I and III were found, with type III being the dominant type (639 of 667 strains). We found that the type III Druantia also has broad antiphage activity against 70 out of the 111 tested phages, indicating that it is a key defense system of E. coli. Another system worth mentioning in our studies is SspBCDE, a homologous defense system of V. cyclitrophicus SspABCD-SspE that restricts phages by introducing nicks to the genomic DNAs by SspE (24). The nickase activity of SspE is stimulated by PT modification, which is performed by SspA, -B, -C, and -D. Both PT modification and SspE of V. cyclitrophicus are necessary for its antiphage activity (24). However, we found that SspBCDE, which is defective in PT modification due to the lack of SspA, restricts infection by the wild-type T4 phage as well as that by the two mutants, indicating the different defense mechanisms of the E. coli SspBCDE system.

During infection, T4 genomic DNA is injected into E. coli cells to initiate genome replication, protein expression, and subsequent virion assembly. Therefore, its genomic DNA is considered the main target for E. coli to restrict infections. It is expected that E. coli cells employ nuclease-containing systems to degrade phage DNAs, whereas phages evolve counterdefense strategies by modifying their genomic DNAs. In fact, we found that the ghmC modification of the phage T4 genome can completely block the defense activities of Gabija, Shedu, Restriction-like, type III Druantia, and qatABCD. The antiphage activities of the last two defense systems can also be counteracted by hmC modification (Fig. 3 to 5). Interestingly, the Gabija system is more effective in restricting T4 (hmC) than T4 (C), and the Restriction-like system inhibits only T4 (hmC) infection (Fig. 3 and 4). A similar phenomenon has been observed in other nuclease-based bacterial defense systems, such as McrA and McrBC, which specifically target T4 phage with hmC-modified DNA due to the unique structure of the nucleases (13, 37, 38). However, the mechanism underlying this preference is unclear. It is also possible that the changes in the modification patterns of the T4 genome could affect the physiology of the phage. The antiphage activities of Septu, SspBCDE, and mzaABCDE were not abolished by ghmC or hmC modifications of the T4 phage genome. In particular, the Septu system showed similar activities against phages T4 (ghmC), T4 (hmC), and T4 (C). Although SspBCDE and mzaABCDE can restrict both wild-type T4 and two mutants, their antiphage activities heavily depend on the extent of the cytosine modifications in the T4 genome (Fig. 2). In addition, there are six nuclease-containing systems that cannot protect E. coli cells against phage T4 regardless of cytosine modifications. Taken together, our data showed that covalent modifications of cytosines confer variable resistance on T4 phage against some nuclease-containing systems, but E. coli can use other nuclease-containing systems to target T4 phage with different modifications. The T4 phage might have other undefined mechanisms in addition to DNA modifications with which to counteract nuclease-containing systems.

In conclusion, our study provides comprehensive insights into the nuclease-containing defense systems employed by E. coli and the corresponding counterdefense activities of phage T4 genome modifications. We found that E. coli employs at least 17 of 20 recently identified nuclease-containing systems to protect itself against phages. The covalent modifications of cytosines give T4 phage variable resistance against some nuclease-containing systems, but E. coli can use other nuclease-containing systems to target T4 phage with different modifications. These findings shed light on the arms race between phages and bacteria and provide new insights into the mechanisms of phage-host interactions.

MATERIALS AND METHODS

Bacteria and phages.

E. coli strains DH5α [hsdR17(r_K⁻ m_K⁺) sup²], DH10B [F⁻ mcrA Δ(mrr-hsdRMS-mcrBC)], MG1655 (F⁻ λ⁻ rph-1), B834 (hsdR_B hsdM_B met thi sup⁰), and B40 (sup¹) were used in this study as described below. E. coli DH5α was used for plasmid construction, and B834 was used for the construction of T4 mutants, T4 (hmC) and T4 (C), as described below. E. coli MG1655 was initially used to determine the antiphage activities of all the nuclease-containing defense systems. The defense activities of nuclease-containing systems against T4 (ghmC), T4 (hmC), and T4 (C) were determined with E. coli DH10B. A collection of 816 E. coli strains from our laboratory stocks was used to amplify nuclease-containing defense systems. Phages T4, T5, and T7 were from our laboratory stocks. A collection of 111 phages, which we recently isolated with E. coli MG1655 as host cells from sewage, was used to determine the antidefense effects of 5 potential defense systems in E. coli. These phages were isolated as described previously (39). Briefly, 58 sewage samples were collected from 58 farms in different cities in China. The samples were processed as previously described and were mixed with 500 μL of E. coli MG1655. After adding 3 mL of melted soft tryptic soy agar (TSA) (0.75% agar), the mixtures were poured into prepared LB plates and incubated overnight at 37°C. We picked out only two plaques from each sewage sample to avoid picking the same kind of phage. Finally, we used 40 E. coli strains to determine the lysis profiles of these phages to further confirm that they were different types of phages based on the difference in their lysis profiles.

Identification of the nuclease-containing defense systems in E. coli.

All complete E. coli genomes (2,289 strains as of February 2021) available at NCBI were downloaded, and the protein-coding sequences were analyzed using Prodigal (release 2.6.3) (40). Protein domains were annotated using HMMER version 33.0 (27). The conserved domains of all 20 nuclease-containing defense systems identified as of February 2021 (see Table S1 in the supplemental material) were collected. The E. coli protein sequences were screened using BLAST+ (release 2.11.0) for the potential defense systems that share similar domain structures with reported nuclease-containing systems. A collection of 816 E. coli strains from our laboratory stocks was used as the templates for the amplification of identified defense systems using primers listed in Table S7. We were able to amplify 14 defense systems (Table S5) from our E. coli collection and clone them to the pSEC1 vector (Table S8), which includes a kanamycin resistance gene and the p15A origin of replication. The generated plasmids were confirmed by sequencing.

Generation of T4 mutants.

T4 mutants, T4 (hmC) and T4 (C), were generated using CRISPR-Cas12 phage genome editing technology (41). Construction of the pLbCas12a plasmids, which express different crRNAs targeting different genes within the T4 genome, was carried out as described in our previous study (14). Briefly, the spacer fragments were cloned into pLbCas12a under the control of the J23100 promoter to express the corresponding crRNA by Gibson assembly of spacer fragments and EcoRI/XhoI-linearized pLbCas12a vector. The restriction efficiency of each pLbCas12a-spacer plasmid for T4 phage infection was determined by plaque assays and shown in Table S6. The donor plasmids were constructed by cloning the fragments containing the gene of interest flanked by two homologous arms into a XbaI/EcoRI- or KpnI/EcoRI-linearized vector. The primers are available upon request. The CRISPR-LbCas12a plasmid and the corresponding donor plasmid were cotransformed into E. coli DH10B or B834, which was used for phage genome editing. E. coli transformed with the CRISPR-Cas12a plasmid only was used as a control. About 10⁸ PFU of phages was incubated with 200 μL of log-phase E. coli cells at 37°C for 7 min. After adding 3 ml of 0.75% LB top agar containing appropriate antibiotics, the mixture was poured onto a TSA plate and incubated at 37°C overnight. The recombinant T4 phages were confirmed using PCR and sequencing of single plaques.

Phage plaque assay.

Phage plaque assays were carried out as described previously (21). Briefly, 500 μL E. coli fresh culture was mixed with 8 mL top agar (30 g/L tryptic soy broth [TSB], 7.5 g/L agar, and 50 μg/mL kanamycin) and poured onto a TSA plate. Then, 100 μL of 10-fold serially diluted phages was dropped on the plate, and phage plaques were counted after overnight incubation at 37°C. Since the empty vector did not affect the plaque efficiencies of wild-type T4 and the two mutants, we calculated the fold reduction in efficiency of plating (EOP) by dividing the input PFU by the number of plaques produced from infection of E. coli expressing the defense system and expressed it as mean ± standard deviation (SD) from three independent assays. For the defense spectrum experiments, the plasmid expressing the specific defense system was transformed into E. coli MG1655 cells, which were then infected with 10-fold serially diluted phages. In total, 111 different phages isolated from different places in China were used. E. coli MG1655 cells transformed with an empty vector, pSEC1, were used as controls. For T4 phage infection experiments, E. coli DH10B [F⁻ mcrAΔ(mrr-hsdRMS-mcrBC)] cells were used instead of MG1655 (F⁻ λ⁻ rph-1), which contains intact R-M systems that restrict phages T4 (hmC) and T4 (C).

Phage propagation and genomic DNA extraction.

Wild-type T4 and two mutants, T4 (hmC) and T4 (C), were purified as previously described (42, 43). Briefly, log-phase E. coli DH10B cells (~2 × 10⁸ cells per mL) grown on LB medium were infected with phages at a multiplicity of infection (MOI) of 0.1. After 5 h of incubation at 200 rpm at 37°C, the E. coli cells were collected by centrifugation at 30,000 × g for 50 min. The pellets were resuspended with 10 mL of Pi-Mg buffer (26 mM Na₂HPO₄, 68 mM NaCl, 22 mM KH₂PO₄, 1 mM MgSO₄, pH 7.5) containing 10 μg/mL DNase I and incubated at 37°C for 1 h. Phage genomic DNAs were isolated with phenol-chloroform-isoamyl alcohol extraction. The phage pellet was first digested with 100 μL 10% SDS and 100 μL 0.5 mol/L EDTA in a 65°C water bath for 30 min. A mixture of phenol-chloroform-isoamyl alcohol (volume ratio of 25:24:1, 10 mL) was added, and the tube was inverted several times until the emulsion formed. After centrifugation at 10,000 × g at 4°C for 10 min, the supernatant was collected and subjected to a second-round extraction. The genomic DNAs were precipitated by adding isopropanol. After centrifugation at 10,000 × g at 4°C for 10 min, the genomic DNA pellet was washed twice with 1 mL precooled 70% ethanol and dissolved in 200 μL deionized water. The modifications of cytosines in the genomic DNAs of T4 (ghmC), T4 (hmC), and T4 (C) were determined by digestion of genomic DNAs with MspJI (New England Biolabs [NEB]) and AluI (NEB), respectively.

Data availability.

E. coli genomes used in the current study can be found in the NCBI database (https://www.ncbi.nlm.nih.gov/genome/). Data not provided in the supplemental material are available from the corresponding author upon request.