A DNA phosphorothioation-based Dnd defense system provides resistance against various phages and is compatible with the Ssp defense system

ABSTRACT DndABCDE-catalyzed DNA phosphorothioation (PT), in which the nonbridging oxygen is swapped with a sulfur atom, was first identified in the bacterial genome. Usually, this modification gene cluster is paired with a restriction module consisting of DndF, DndG, and DndH. Although the mechanisms for the antiphage activity conferred by this Dnd-related restriction and modification (R-M) system have been well characterized, several features remain unclear, including the antiphage spectrum and potential interference with DNA methylation. Recently, a novel PT-related R-M system, composed of the modification module SspABCD paired with a single restriction enzyme, SspE, was revealed to be widespread in the bacterial kingdom, which aroused our interest in the interaction between Dnd- and Ssp-based R-M systems. In this study, we discussed the action of Dnd-related R-M systems against phages and demonstrated that the host could benefit from the protection provided by Dnd-related R-M systems against infection by various lytic phages as well as temperate phages. However, this defense barrier would fail against lysogenic phages. Interestingly, DNA methylation, even in the consensus sequence recognized by the Dnd system, could not weaken the restriction efficiency. Finally, we explored the interaction between Dnd- and Ssp-based R-M systems and found that these two systems were compatible. This study not only expands our knowledge of Dnd-associated R-M systems but also reveals a complex interaction between different defense barriers that coexist in the cell. IMPORTANCE Recently, we decoded the mechanism of Dnd-related R-M systems against genetic parasites. In the presence of exogenous DNA that lacks PT, the macromolecular machine consisting of DndF, DndG, and DndH undergoes conformational changes to perform DNA binding, translocation, and DNA nicking activities and scavenge the foreign DNA. However, several questions remain unanswered, including questions regarding the antiphage spectrum, potential interference by DNA methylation, and interplay with other PT-dependent R-M systems. Here, we revealed that the host could benefit from Dnd-related R-M systems for a broad range of antiphage activities, regardless of the presence of DNA methylation. Furthermore, we demonstrated that the convergence of Dnd- and Ssp-related R-M systems could confer to the host a stronger antiphage ability through the additive suppression of phage replication. This study not only deepens our understanding of PT-related defense barriers but also expands our knowledge of the arms race between bacteria and their predators.

P hages, whose population is estimated to be 10 31 in the biosphere, are responsi ble for 20%-40% of bacterial mortality every day (1,2). To respond to such a large number of predators, bacteria have evolved several antiphage strategies tar geting different stages of the phage life cycle (3). Innate immune systems, such as DNA methylation-dependent restriction-modification (R-M) systems, bacteriophage exclusion, cyclic-oligonucleotide-based antiphage signaling systems, pyrimidine cyclase systems for anti-phage resistance, and the recently discovered DNA phosphorothioa tion (PT)-based R-M systems, are triggered by phage components (e.g., phage DNA or proteins) to protect the host or act in an altruistic mode to preserve the popu lation (4)(5)(6)(7)(8)(9)(10)(11)(12). In contrast to innate immune systems, clustered regularly interspaced short palindromic repeat (CRISPR)-CRISPR-associated protein (Cas) systems, which take advantage of the molecular memory from prior infections, are classified as an adaptive immune mechanism (13). By hijacking the genetic information from the invaders as spacers into the CRISPR array and deploying the repertoire of spacers, the host can complete the acquisition, expression, and interference stages and can thus effectively recognize and eliminate their phage foes. DNA PT, mediated by DndABCDE, which transfers sulfur atoms from L-cysteine to the microbial genome in a sequence-and stereo-specific manner, was first identified in the bacterial genome, and this system was classified as the Dnd system (14)(15)(16)(17). To build up a line of defense, the dndABCDE gene cluster is usually paired with dndFGH, which acts as a restriction module. In the presence of invaders, DndFGH exerts DNA binding, translocation, and nicking activities to initiate the destructive DNA-shredding program to eliminate genetic parasites (18,19). Interestingly, some DNA methylases, such as Dam, which converts 5′-GATC-3′/5′-GATC-3′ to 5′-G 6m ATC-3′/5′-G 6m ATC-3′, can modify the same motif recognized by DndABCDE, leading to the potential conflict with DndFGH in the presence of exogenous DNA harboring a methyl group in the PT modification motif (11,(20)(21)(22). Recently, a novel PT-associated antiphage system, composed of the modification module SspABCD paired with a single restriction enzyme, SspE, was identified as being widespread in the bacterial kingdom and was classified as an Ssp system. This SspABCD-E system could confer the host with broad-spectrum phage resistance (10,(23)(24)(25). Unlike the previously characterized Dnd systems, which are responsible for the double-stranded DNA PT modification, Ssp systems confer the host with single-stranded DNA PT modification exclusively at the 5′-C PS CA-3′ motifs with much higher frequency. In general, some dnd or ssp clusters lack dndA or sspA; DndA or SspA is instead function ally replaced by other cysteine desulfurases, such as IscS, in bacteria.
Although the molecular mechanism underlying the DNA damage caused by DndFGH has been well studied, its phage resistance spectrum has not been characterized. Furthermore, the effect of the convergence of Dnd-and Ssp-related restriction-modification (Dnd R-M, Ssp R-M) systems is less well known. In this study, we determined the spectrum of phage resistance conferred by DndFGH and demonstrated that the Dnd R-M system could protect the host from various lytic phages (T1, T4, T5, T7, and engineered E. coli phage EEP) as well as temperate phage λ. Although DndFGH could inhibit phage lysogenization effectively, this defense barrier failed to protect the host on prophage induction. Moreover, by taking advantage of the Dnd R-M systems from Bermanella marisrubri RED65, which recognized 5′-GATC-3′/5′-GATC-3′ motifs, we revealed that the methyl group in 5′-G 6m ATC-3′/5′-G 6m ATC-3′ would not inhibit the restriction activity of DndFGH. Finally, we combined Dnd R-M and Ssp R-M systems. We found that these two kinds of PT-associated R-M systems were compatible. These findings not only expand our knowledge of PT-associated R-M systems but also deepen our understanding of the arms race between bacteria and phages.

The Dnd-related restriction-modification system protects the host from various lytic phages
Recently, we resolved the crystal structure of DndG and the C-terminal domain of DndH (residues 1,313-1,687) from Escherichia coli B7A (18). Based on structural and biochemical evidence, we proved that DndF, DndG, and DndH could form a complex with a molar ratio of 2:2:1 and exert DNA binding, translocation, and DNA nicking activities to protect the host from phage invasion (18). However, the antiphage spectrum of DndFGH is unclear.
To determine the antiphage spectrum, we chose three Dnd R-M systems: the Dnd R-M system from Proteus mirabilis 1166 PMIR (Dnd 1166 R-M) with undefined modification motifs, the Dnd R-M system from E. coli B7A (Dnd B7A R-M) recognizing 5′-G PS AAC-3′/ 5′-G PS TTC-3′, and the Dnd R-M system from B. marisrubri RED65 (Dnd RED65 R-M) recognizing 5′-G PS ATC-3′/5′-G PS ATC-3′ (Fig. 1A) (26,27). First, we cloned dndBCDE-FGH derived from P. mirabilis 1166 PMIR, E. coli B7A, and B. marisrubri RED65 into pACYC184, generating pWHU4386, pWHU4387, and pWHU4388, respectively. By applying liquid chromatography with tandem mass spectrometry (LC-MS/MS), d(G PS A) and d(G PS T) were identified in the genome of P. mirabilis 1166 PMIR as well as DH10B(pWHU4386), implying that the motif in Dnd 1166 was 5′-G PS AAC-3′/5′-G PS TTC-3′ ( Fig. 1A; Fig. S1). As expected, along with the existence of Dnd R-M systems, several protective phenomena, including a decrease in phage titer, a reduction in plaque size, and the absence or delay of bacterial culture collapse, could be observed when the bacteria were challenged with various lytic phages ( Fig. 1B and D). To quantify the protection efficiency conferred by Dnd R-M systems, an efficiency of plating (EOP) assay was performed. Consistent with the results from the phage spotting and growth curve assay, Dnd B7A R-M exhibited the strongest phage resistance ability, providing up to five orders of magnitude protection against all the phages tested here. Compared to Dnd 1166 R-M, which conferred the host with one to three orders of magnitude protection against all the phages tested, Dnd RED65 R-M showed the weakest antiphage activity ( Fig. 1C; Supplementary Data 1). Interestingly, when coincubated with bacteria in liquid broth, T4 exhibited more aggressive activity against all three Dnd R-M systems than other phages. In addition, Dnd 1166 R-M was more vulnerable to T7 (Fig. 1D).
The strong phage resistance exhibited by Dnd 1166 R-M and Dnd B7A R-M prompted us to challenge DH10B(pWHU4386) and DH10B(pWHU4387) with phage cocktails at various multiplicities of infection (MOIs). According to the antiphage performance of Dnd R-M systems during coincubation in the growth curve assay, different phage cocktails were prepared. As shown in Fig. 1E, the Dnd 1166 R-M systems exhibited activity against the phage cocktails consisting of T1, T5, and EEP at an MOI of 0.01, which has been deter mined to be the limit of T5 tolerance for Dnd 1166 R-M systems. However, with the addition of T4 and T7, even at an MOI of 0.001, which is the lowest concentration that we tested, the growth curve collapsed similarly to that of the control group. Again, Dnd B7A R-M systems exhibited strong antiphage activity against the phage cocktails consisting of T1, T5, T7, and EEP even at an MOI of 5, which was the highest concentration that we tested. However, consistent with the phenomenon that we observed in Dnd 1166 R-M systems, the growth curve collapsed when strains carrying Dnd B7A R-M systems were challenged with the all-in-one phage cocktail (Fig. 1F).
These data demonstrated that the Dnd R-M systems could protect their host from various lytic phages, even when facing allied phage forces.

The Dnd system could protect the host from lysogenization of λ but not prophage induction
Compared to the lytic phage, temperate phages, such as λ, which can remain dormant as prophages in E. coli and become reactivated to achieve proliferation and lyse the host, have a more complex life cycle. This prompted us to investigate the potential resistance ability conferred by the Dnd R-M system against λ during the process of lysogenization as well as prophage induction.

Research Article mBio
First, we evaluated the EOP for phage λ invading the strains harboring Dnd B7A , Dnd 1166 , or Dnd RED65 R-M systems. Phage λ could generate visible plaques on DH10B(Dnd B7A R-M), DH10B(Dnd 1166 R-M), and DH10B(Dnd RED65 R-M) with EOPs of (1.64 ± 0.58) × 10 −2 , (3.05 ± 0.22) × 10 −1 , and (3.13 ± 1.67) × 10 −2 , respectively, indicating that the Dnd R-M system could protect the host from invasion by temperate phages such as λ ( Fig. 2A). After identifying the λ resistance ability conferred by Dnd R-M, we began to evaluate the effectiveness of Dnd R-M in preventing phage lysogenization. To generate lysogens, we mixed DH10B with phage λ at an MOI of 1 and incubated it on LB agar plates. Colonies were checked by PCR to identify successful lysogenization. The efficiency of lysogenization reached 12.50% ± 3.23% for DH10B without the protective effect of Dnd R-M. For DH10B(Dnd 1166 R-M), the lysogenization efficiency was reduced to 2.78% ± 2.15%, and no lysogens were detected in strains harboring Dnd B7A or Dnd RED65 R-M (Fig.  2B). These observations indicated that Dnd R-M systems could prevent the temperate phage from lysogenizing the host.
The other characteristic that distinguished temperate phages from lytic phages is that the prophage persisting as dormant components of host cells could be "woken up" to lyse the host cells. For some defense systems, such as CRISPR-Cas systems, the phage DNA can be eliminated during prophage induction. Thus, the parasitized hosts can be "cured" (28). However, no such data are available for Dnd R-M systems. Therefore, we examined the potential effect of Dnd R-M on this "internal foe". For efficient induction, we used λcI857 instead of λ. This phage can be triggered at 42°C due to the temperaturesensitive cI mutation (29)(30)(31). Unlike the CRISPR-Cas defense barrier, Dnd R-M systems failed to provide effective protection when we changed the temperature from 28°C to 42°C. As shown in Fig. 2C, the presence of λcI857 at 28°C or the Dnd system at 42°C did not influence the growth rate of DH10B. In sharp contrast, the OD 600 dropped steeply to nearly 0 after 1.5 hours at the elevated temperature and remained steady in all the strains harboring λcI857 regardless of the existence of Dnd R-M systems. Furthermore, we calculated the survival rate by plating the bacteria on LB agar plates and incubating them at 28°C or 42°C. After induction, the survival rates were (0 ± 0), (1.01 ± 1.56) × 10 −8 , (0 ± 0), and (0 ± 0) for strains carrying pACYC184, pWHU4386, pWHU4387, and pWHU4388, respectively (Fig. 2D). Meanwhile, we monitored progeny production by λcI857 during induction. For strains equipped with Dnd R-M systems, λcI857 began to lyse the cells at 30 minutes and released ~25 viral particles per cell during the first 1.5 hours after temperature shift, which was consistent with the phenomenon we observed in DH10B(λcI857, pACYC184) (Fig. 2E).
Considering that the efficiency of the Dnd R-M system could be impaired due to the rise in temperature (11), the λ phage resistance ability of Dnd R-M at 28°C and 42°C was assessed to rule out the potential influence of temperature variation. We found that the rise in temperature did not impair the protection efficiency of the Dnd 1166 or Dnd B7A R-M systems, and the EOP was (3.05 ± 0.22) × 10 −1 or (3.26 ± 0.63) × 10 −1 and (1.64 ± 0.58) × 10 −2 or (1.93 ± 0.71) × 10 −2 for the Dnd 1166 and Dnd B7A R-M systems at 28°C or 42°C, respectively. However, the EOP increased by approximately one order of magnitude from (3.13 ± 1.67) × 10 −2 to (3.64 ± 0.47) × 10 −1 for the Dnd RED65 R-M systems as the tempera ture increased. Considering that this value was close to that from Dnd 1166 R-M, we hypothesized that this weakened defense system could still provide sufficient protection ( Fig. 2A; Fig. S2). Thus, the lack of protection against prophage induction was not caused by the temperature variation.
In this section, we provided evidence that Dnd R-M systems could protect the host from lysogenization of λ but not prophage induction.
This result led us to use T1 phages to evaluate the response of Dnd RED65 R-M systems to methylated DNA. First, we determined the modification motif of the DNA methylase (protein_ID = YP_003925.1) encoded by the T1 phage. The genomic DNA of T1 prepared from JW3350 remained sensitive to DpnI, indicating that this Dam methyltransferase could convert 5′-GATC-3′/5′-GATC-3′ to 5′-G 6m ATC-3′/5′-G 6m ATC-3′, which is consistent with previous research (33,34) (Fig. S4C). To prepare T1 with a "naked" genome, we deleted dam from the T1 genome using the CRISPR-Cas system to generate CC20 ( Fig.  S4A and B). The enzyme digestion assay demonstrated that the genomic DNA from CC20 prepared from JW3350 lost resistance to DpnI and showed restored sensitivity to MboI, indicating that all the 5′-GATC-3′ sites in the CC20 genome remained unmethylated (Fig. 3A). On challenging DH10B(pACYC184) or DH10B(Dnd RED65 R-M) with T1 or CC20, the EOP as well as the efficiency of the center of infection (ECOI) were determined by calculating the ratio of the phage titer measured in DH10B(Dnd RED65 R-M) to that Research Article mBio measured in DH10B(pACYC184) (Fig. 3B and C). Surprisingly, DH10B(Dnd RED65 R-M) showed similar resistance to both CC20 and T1, with an EOP of (1.11 ± 0.07) × 10 −1 versus (1.04 ± 0.13) × 10 −1 and an ECOI of (1.23 ± 0.28) × 10 −1 versus (1.07 ± 0.19) × 10 −1 , respectively, implying that DndFGH could scavenge invaders regardless of DNA methylation. Furthermore, to explore the potential influence of the Dnd RED65 R-M system on phage proliferation in a single life cycle in more detail, one-step growth curves were generated for DH10B(pACYC184) and DH10B(Dnd RED65 R-M) infected with CC20 and T1, respectively (Fig. 3D). Compared with that of T1, the burst size of CC20 decreased to approximately 60% (48.75 ± 8.65 versus 29.24 ± 3.02) on infection of DH10B(Dnd RED65 R-M). However, without the protection of the Dnd RED65 R-M system, the burst sizes of T1 and CC20 were 52.26 ± 3.10 and 31.93 ± 2.73, respectively, indicating that the decrease was caused by dam deletion rather than the effect of the Dnd RED65 R-M system. These data demonstrated that DNA methylation would not interfere with the action of Dnd R-M systems even if both systems recognized the consensus sequence. Surprisingly, the presence of Dnd RED65 R-M systems showed no significant influence on the burst size of T1 phage (P value = 0.9235), which seemed contradictory to the previous results (18). We proposed that this phenomenon could be attributed to the relatively weak defense activity of Dnd RED65 R-M. To verify this hypothesis, we rechecked the one-step growth curve of T1 when DH10B(Dnd B7A R-M), which exhibited the strongest protection, was treated as the host. Consistent with the previous results, Dnd B7A R-M conferred the host with powerful protection with an ECOI of (3.12 ± 0.27) × 10 −3 , indicating that there was only a narrow chance for T1 to produce at least one active progeny in the presence of Dnd B7A R-M. However, the burst size of T1 decreased slightly from 52.26 ± 3.10 to 40.65 ± 5.95 (P value 0.0185) ( Fig. S5A and B). These data indicated that the Dnd R-M system exhibited antiphage activity mainly in the early stage of phage infection. Once the escaped phages launched the replication process, the protection conferred by DndFGH was limited.

The Dnd restriction-modification system could be compatible with SspABCD-E
Recently, we discovered a novel Ssp R-M system that consists of SspABCD-E (10). The single restriction enzyme SspE could sense the sulfur atom on 5′-C PS CA-3′ generated by SspABCD and exert nicking activity toward unmodified DNA, thus protecting the host from various phages (23). The different mechanisms of the Dnd-and Ssp-based defense barriers prompted us to examine the compatibility of these two kinds of systems.
To address this subject, we chose E. coli MG1655-PT, which was derived by integrat ing sspBCD-E from E. coli 3234/A into the genome of MG1655, as well as Salmonella enterica serovar Cerro 87, which carried dndBCDE-FGH, as model strains. We intro duced pACYC184, pWHU4386, pWHU4387, and pWHU4388 into MG1655 or MG1655-PT. Additionally, we transferred pACYC184 and pWHU3638, which were derived from pACYC184 and carried sspBCD-E from E. coli 3234/A into S. enterica serovar Cerro 87 as well as the associated dndB-H-deficient strain XTG103 to generate strains with four different statuses: (Dnd−, Ssp−), (Dnd+, Ssp−), (Dnd−, Ssp+), and (Dnd+, Ssp+). First, we determined the growth rate of these strains to check for potential conflict between these two kinds of PT-based systems. Similar growth curves were observed using either LB broth or M9 broth, suggesting that the combination of these two kinds of PT-related R-M systems did not inhibit cell growth (Fig. S6). Consistent with the previous results, the Dnd R-M systems exhibited strong phage resistance, and phage plaques could hardly be seen on LB plates coated with Dnd R-M system-armed cells (Fig. 4A). An additive phenomenon could be observed for MG1655 harboring both Dnd and Ssp R-M systems ( Fig. 4A; Fig. S7A and B). Since the additive effect could hardly be seen in the S. enterica strains, which might have been due to the overpowering phage resistance conferred by the endogenous Dnd R-M system, we calculated the ECOI to assess the effect of this combination in S. enterica. Compared to the protection from Dnd R-M or Ssp R-M alone against the PT1 phage, which is a lytic S. enterica phage, with ECOIs of (1.51 ± 0.11) × 10 −2 enterica strains harboring Dnd and/or Ssp R-M systems, respectively. All experiments were performed three times. **P < 0.01, ***P < 0.001, ****P < 0.0001, ns, not significant.

Research Article mBio
and (1.07 ± 0.18) × 10 −1 , respectively, the combination of these two systems significantly decreased the ECOI to (7.94 ± 1.48) × 10 −3 (Fig. 4B). Similar phenomena were observed in the MG1655-series strains when we challenged the strains with T7 phages (Fig. 4B). The increase in phage resistance caused by the combination of the Dnd and Ssp R-M systems prompted us to perform an qRT-PCR assay, which is an effective and simple method (9,35), to evaluate the phage DNA replication rates in the strains harboring different defense systems. In sharp contrast to the rapid replication of phage DNA in the strain lacking the protection of PT-related R-M systems, the presence of Dnd R-M or Ssp R-M alone significantly reduced or inhibited phage DNA replication. Further enhancement in protection efficiency could be observed with the convergence of these two kinds of defense barriers (Fig. 4C). Although there was no statistically significant difference between Cerro 87(pACYC184) and Cerro 87(pWHU3638), the DNA replication rate of PT1 from Cerro 87(pWHU3638) was consistently slower than that of PT1 from Cerro 87(pACYC184) from 30 minutes after infection, implying an additive defense effect due to the coexistence of Dnd and Ssp R-M systems.

DISCUSSION
In this study, we proved that similar to Ssp R-M systems, DndABCDE-FGH can provide a solid defense against various phages, even phage cocktails. Interestingly, the antiphage and antiplasmid preferences were quite different among different Dnd R-M systems, even when they shared the same DNA motifs. However, although DndFGH exhibited strong resistance to invasion by temperate phages, such as λ, this fortress was not infallible against attack by "internal enemies". This was reasonable considering the action mode of R-M systems, which distinguish self-and nonself-DNA based on the specific modification tags. A prophage that is successfully integrated into the genome is recognized as self-DNA and modified as well as replicated together with the lysogen. With the protection provided by the PT modification, prophages can escape restriction by Dnd R-M systems. Notably, researchers recently proved that the PT modification could influence the transcription of the genes encoded by the prophage and alter the binding affinity of the repressor associated with prophage activation, indicating a more complex scenario of the interaction between Dnd systems and prophages (36).
Surprisingly, the methyl group modified in the consensus motifs recognized by Dnd systems did not impair the restriction efficiency of DndFGH, although there was a conflict between Dam methylase and DndABCDE (27). This phenomenon suppor ted the idea that DndFGH exclusively sensed PT tags. However, another possibility could not be ignored: although the full methylation of the T1 genome could be proved by enzyme digestion, few unmethylated 5′-GATC-3′/5′-GATC-3′ motifs, which were randomly distributed in the T1 genome, could be detected using [ 3 H]-labeled AdoMet (34). These unmodified motifs might become potential targets for DndFGH. Interestingly, the deletion of dam from the T1 genome was responsible for the decrease in phage burst sizes even in the strains containing Dam of bacterial origin (Protein ID: NP_417846.1). Although these two Dam methylases could both modify 5′-GATC-3′/ 5′-GATC-3′ to generate 5′-G 6m ATC-3′/5′-G 6m ATC-3′, no significant sequence similarity was found between these two kinds of Dam methylases. This prompted us to check the modification status of the CC20 progeny released from BW25113 (Fig. S4C). Based on the findings that (i) the DNA was still sensitive to MboI, although higher bands could be observed compared to those from the phages prepared from JW3350 and (ii) the DNA was also sensitive to DpnI but remained more intact than the DNA from the MboI-treated group, we proposed that the T1 phage DNA could only be partially modified by the Dam of bacterial origin, which has been proven to be a highly efficient methyltransferase (37). The methylation status might be associated with the efficiency of progeny production.
In addition, we demonstrated that Dnd and Ssp R-M could work together to suppress phage DNA replication. It is worth mentioning that it is possible to apply this combina tion strategy in more species due to the widespread distribution of Dnd and Ssp systems in the bacterial kingdom, including in Streptomyces, Bacillus, and Halomonas, which can be used as microbial chassis for industrial products. Recently, we identified a novel defense module consisting of SspFGH, which shared no significant sequence identity with DndFGH, and the combination of SspFGH with SspE could confer the host with stronger protection against phages (38). These compatibility phenomena revealed the high inclusivity of the PT-based R-M systems and the potential for developing these basic elements as an antiphage package in the field of biosynthesis.
Taken together, the results demonstrated the strong antiphage activities provided by Dnd R-M systems, covering a broad spectrum of phages, and showed that DndFGH could overcome steric clashes due to DNA methylation. In addition, PT-related R-M systems could coexist and be compatible with each other (Fig. 5). These data not only deepen our understanding of PT-based R-M systems but also describe the complex interplay between bacteria and their phage predators.

Bacterial strains, bacteriophages, plasmids, and media
All of the strains, bacteriophages, and plasmids used in this study are listed in Table S1 and Table S2. The primers used in this study are listed in Table S2. The Escherichia coli and Salmonella enterica strains were routinely grown at 37°C or 28°C in Luria-Bertani (LB) broth or on LB agar plates unless otherwise indicated. Proteus mirabilis strains were grown in LB broth at 37°C. B. marisrubri strains were grown in DSMZ Medium 514c at 28°C. When necessary, an appropriate concentration of antibiotic (100 µg/mL ampicillin and/or 25 µg/mL chloramphenicol) was added to the medium.

Plasmid construction
For the construction of pWHU4386, a 16,029-bp fragment containing the dndBCDEFGH gene cluster was amplified from the genomic DNA of P. mirabilis 1166 PMIR using the primer pair 1166F/1166R (Table S2). The fragment was inserted into the HindIII-BamHIdigested pACYC184 plasmid using a ClonExpress Ultra One Step Cloning Kit (Vazyme), generating pWHU4386. For the construction of pWHU4387, the 8,286-bp and 8,321-bp fragments containing the whole dndBCDEFGH gene cluster were amplified from the genomic DNA of E. coli B7A using the primer pairs B7A-LL/B7A-LR and B7A-RL/B7A-RR (Table S2), respectively. The two fragments were inserted into the HindIII-BamHI-digested pACYC184 plasmid through Gibson assembly cloning, yielding pWHU4387.
For the construction of pWHU4388, the 8,137-bp and 7,637-bp fragments contain ing the whole dndBCDEFGH gene cluster were amplified from the genomic DNA of B. marisrubri RED65 using the primer pairs RED-LL/RED-LR and RED-RL/RED-RR (Table  S2), respectively. The two fragments were inserted into the HindIII-BamHI-digested pACYC184 plasmid through Gibson assembly cloning.
For the construction of pWHU3638, an 8,069-bp fragment containing the sspBCDE gene cluster was amplified from the genomic DNA of E. coli 3234/A using the primer pair 3234F/3234R (Table S2). The fragment was inserted into the SalI-BamHI-digested pACYC184 plasmid, generating pWHU3638.
For the construction of pWHU710, both 66-bp homologous sequences were amplified from the genomic DNA of T1 phage using the primer pairs Dam-LL/Dam-LR and Dam-RL/ Dam-RR (Table S2). The two fragments were inserted into the HindIII-BamHI-digested pBluescript II SK+ through Gibson assembly cloning, yielding pWHU710.
For the construction of pWHU711, first, the spacer targeting dam of T1 was prepared by annealing the primer pair Spacer-F/Spacer-R (Table S2). Then, the spacer was ligated into BsaI-digested pCas9 using T4 DNA ligase (NEB), generating pWHU711.

Construction of CC20
Overnight-cultured E. coli DH10B(pWHU710, pWHU711) was diluted 1:100 in LB medium containing 100 µg/mL ampicillin and 25 µg/mL chloramphenicol at 37°C until the optical density at 600 nm (OD 600 ) reached 0.6-0.8. A 300-µL bacterial culture was mixed with molten soft agar (0.75%) and immediately overlaid on prepoured LB agar (1.5%) plates. Phage T1 stocks were subjected to 10-fold serial dilution with SM buffer (100 mM NaCl, 8 mM MgSO 4 , and 50 mM Tris-Cl pH 7.5) and spotted on the upper layer of the plate. After incubation overnight at 37°C, plaques were verified by PCR using the primer pair Dam-LL/Dam-RR (Table S2). The mutants were purified twice and further confirmed by DNA sequencing.

Phage plaque assays
Overnight-cultured E. coli or S. enterica strains were diluted 1:100 in LB medium containing the appropriate antibiotic at 37°C or 28°C until the OD 600 reached 0.6-0.8. A 300-µL bacterial culture was mixed with molten soft agar (0.75%) and immediately overlaid on prepoured LB agar (1.5%) plates. Phage stocks were subjected to 10-fold serial dilution with SM buffer. Once the top agar solidified, 5 µL of each 10-fold dilu ted phage stock was spotted onto the bacterial lawns and cultured overnight at the indicated temperatures.

Efficiency of plating (EOP) assay
Overnight-cultured E. coli strains were diluted 1:100 in LB medium containing the appropriate antibiotic and incubated until they reached the log phase (~OD 600 = 0.6-0.8). The molten soft agar (0.75%) containing 100 µL of serially diluted phage solution and 300 µL of mid-exponential bacterial culture was poured onto a preprepared LB agar (1.5%) plate and incubated overnight at a fixed temperature. EOP was calculated by dividing PFU/mL (on the tested strain) by PFU/mL (on the control strain).

Efficiency of the center of infection (ECOI) assay
Overnight-cultured E. coli or S. enterica strains were diluted 1:100 in LB broth containing 1 mM CaCl 2 , 1 mM MgCl 2 , and the appropriate antibiotic at 37°C or 28°C until the OD 600 reached 0.5. One milliliter of the bacterial culture was mixed with phage solution to meet the multiplicity of infection (MOI) of 0.1. The mixture was incubated without shaking for 5 minutes for phage absorption. After centrifugation at 10,000× g for 1 minute, the pellet was washed twice with LB broth and resuspended in 1 mL of LB. Samples were subjected to 10-fold serial dilution, mixed with molten soft agar (0.75%) containing the phage-sensitive host, poured onto preprepared LB agar (1.5%) plates, and incubated overnight. The ECOI was calculated by dividing the number of COIs on the sensitive strain by the number of COIs on the resistant strain.

One-step growth curve assay
Overnight-cultured E. coli strains were diluted 1:100 in LB broth containing the appropri ate antibiotic, 1 mM CaCl 2 , and 1 mM MgCl 2 and incubated at 37°C until the OD 600 reached 0.6. Phage was added to the log-phase cells to achieve a final MOI of 0.01. After a 5-minute incubation, 1 mL of cell cultures was centrifuged at 10,000× g for 1 minute to remove the unabsorbed phages. The pellet was washed twice, resuspended in 1 mL of LB, subjected to 10-fold serial dilution, and incubated at 37°C. Then, 100 µL aliquots were taken periodically over 60 minutes from the appropriate dilution gradient, mixed with molten soft agar (0.75%) containing DH10B and overlaid onto preprepared LB agar (1.5%) plates. The burst size of the phage was determined by dividing the PFU/mL in the latent period by the PFU/mL in the plateau period.

Bacterial growth curve assay
Overnight-cultured E. coli or S. enterica strains were diluted 1:100 in LB or M9 medium containing the appropriate antibiotic and incubated until the OD 600 reached 0.5. The log-phase cultures were diluted 1:100 in LB or M9 medium and transferred into a 96-well plate to monitor the growth of samples using a microplate spectrophotometer (Synergy H1MF, BioTek). The OD 600 of each sample was measured at 30-minute intervals during the incubation. For the phage infection growth curve, phages were added at various MOIs to log-phase (OD 600 = 0.5) cells and cocultured in a 96-well plate to monitor the growth of the cells by a microplate spectrophotometer (Synergy H1MF, BioTek).

Transformation efficiency assay
Fifty nanograms of pBluescript II SK+ isolated from E. coli BW25113 or JW3350 was transformed in parallel into chemically competent cells with or without dndBCDE-FGH modules. After overnight cultivation on LB agar plates with or without ampicillin, the transformation efficiency was measured by dividing CFU/mL (from plates with ampicillin) by CFU/mL (from plates without ampicillin).

Quantification of real-time PCR assay
Log-phase (OD 600 = 0.5) E. coli or S. enterica strains were infected with phage T7 at an MOI of 0.5 or phage PT1 at an MOI of 5. Samples were harvested at each time point, and DNA was extracted using the chloroform-phenol method. For each quantitative RT-PCR (qRT-PCR), 1 ng of total DNA was added as a template. The primer pairs RT-T7-F/RT-T7-R and RT-E-gapA-F/RT-E-gapA-R were used to monitor the abundance of DNA from T7 and E. coli, respectively. For the quantification of DNA abundance from S. enterica and PT1, the primer pairs RT-PT1-F/RT-PT1-R and RT-S-gapA-F/RT-S-gapA-R were used instead (Table S2). Real-time PCR was performed using Taq Pro Universal SYBR qPCR Master Mix (Vazyme) and a QuantStudio 3 Real-Time PCR System (ThermoFisher Scientific) to determine the threshold cycle (Ct). The comparative Ct (2 −ΔΔCt ) method was applied to determine the relative DNA level.

LC-MS/MS analysis of PT modifications
P. mirabilis and E. coli strains were incubated until the OD 600 reached 0.6-0.8. Genomic DNA was extracted, hydrolyzed, dephosphorylated, and detected by liquid chromatogra phy with tandem mass spectrometry (LC-MS/MS) analysis, as previously described (26). Briefly, 20 μg of bacterial genomic DNA, which was extracted using the chloroform-phe nol method, was digested with nuclease P1 followed by dephosphorylation with alkaline phosphatase. After ultrafiltration with a Nanoseq 10K column (PALL, USA), the filtrate was loaded onto a Thermo Hypersil Gold aQ column (150 × 2.1 mm, 3 µm) coupled to a Thermo TSQ Quantum Access MAX mass spectrometer for PT modification detection. Elution was performed at a flow rate of 0.2 mL/minute with solvent A (water containing 0.1% acetic acid) and solvent B (acetonitrile with 0.1% acetic acid) according to the following profile: 97% solvent A for 5 minutes, followed by 97%-94% solvent A over 42 minutes and 94%-2% solvent A in 1 minute.

Lambda lysogenization assay
λ and DH10B were mixed at an MOI of 1 and incubated at room temperature for 15 minutes. The unabsorbed phages were removed by centrifugation at 10,000× g for 1 minute. The culture was diluted to an appropriate gradient and overlaid on LB agar (1.5%) plates. After incubation for 24 hours at 28°C, the colonies were checked by PCR using the primer pair Lambda-F/Lambda-R (Table S2) to detect successful lysogenization.

Lambda induction assay
λcI857 instead of λ was applied in this assay. The overnight-cultured lysogen carrying λcI857 was diluted 1:100 in LB with appropriate antibiotics and incubated at 28°C until the OD 600 reached 0.5.
To assess the survival rate of bacteria under the induction effect of λcI857, cultures were subjected to 10-fold serial dilution and parallelly overlaid on LB agar (1.5%) plates. The plates were incubated at 28°C or 42°C. The survival rate was calculated by dividing CFU/mL (from plates incubated at 42°C) by CFU/mL (from plates incubated at 28°C).
To evaluate the growth rate of the bacteria and the progeny production efficiency, 1 mL of culture was taken at each time point to monitor the OD 600 (ND-100C, MIU LAB) after the temperature was shifted to 42°C. Meanwhile, bacteria were removed by centrifugation at 10,000× g for 1 minute. The supernatant was subjected to 10-fold serial dilution, mixed with DH10B in molten soft agar (0.75%), and immediately overlaid on prepoured LB agar (1.5%) plates. The PFU/mL was calculated after overnight incubation.

DIRECT CONTRIBUTION
This article is a direct contribution from Shi Chen, a Fellow of the American Academy of Microbiology, who arranged for and secured reviews by Nan Peng, Huazhong Agricul tural University, and Jianping Xie, Southwest University.

ADDITIONAL FILES
The following material is available online.