Efficient Genome Engineering of a Virulent Klebsiella Bacteriophage Using CRISPR-Cas9

In the present study, we have addressed efficient, time-saving, and cost-effective CRISPR-based phage genome editing of Klebsiella phage, which has the potential to significantly expand our knowledge of phage-host interactions and to promote applications of phage therapy. The distribution of sgRNA activity was first evaluated in phages. Short homologous arms were proven to be enough to introduce point mutation, small frameshift deletion, gene deletion, and swap into phages, and weak sgRNAs were proven useful for precise phage genome editing but failed to select random recombinants, all of which makes the CRISPR-based phage genome-editing method easier to use.

ABSTRACT

Klebsiella pneumoniae is one of the most common nosocomial opportunistic pathogens and usually exhibits multiple-drug resistance. Phage therapy, a potential therapeutic to replace or supplement antibiotics, has attracted much attention. However, very few Klebsiella phages have been well characterized because of the lack of efficient genome-editing tools. Here, Cas9 from Streptococcus pyogenes and a single guide RNA (sgRNA) were used to modify a virulent Klebsiella bacteriophage, phiKpS2. We first evaluated the distribution of sgRNA activity in phages and proved that it is largely inconsistent with the predicted activity from current models trained on eukaryotic cell data sets. A simple CRISPR-based phage genome-editing procedure was developed based on the discovery that homologous arms as short as 30 to 60 bp were sufficient to introduce point mutation, gene deletion, and swap. We also demonstrated that weak sgRNAs could be used for precise phage genome editing but failed to select random recombinants, possibly because inefficient cleavage can be tolerated through continuous repair by homologous recombination with the uncut genomes. Small frameshift deletion was proved to be an efficient way to evaluate the essentiality of phage genes. By using the abovementioned strategies, a putative promoter and nine genes of phiKpS2 were successfully deleted. Interestingly, the holin gene can be deleted with little effect on phiKpS2 infection, but the reason is not yet clear. This study established an efficient, time-saving, and cost-effective procedure for phage genome editing, which is expected to significantly promote the development of bacteriophage therapy.

IMPORTANCE In the present study, we have addressed efficient, time-saving, and cost-effective CRISPR-based phage genome editing of Klebsiella phage, which has the potential to significantly expand our knowledge of phage-host interactions and to promote applications of phage therapy. The distribution of sgRNA activity was first evaluated in phages. Short homologous arms were proven to be enough to introduce point mutation, small frameshift deletion, gene deletion, and swap into phages, and weak sgRNAs were proven useful for precise phage genome editing but failed to select random recombinants, all of which makes the CRISPR-based phage genome-editing method easier to use.

INTRODUCTION

Bacteriophages are the most numerous biological entities in the biosphere, and they play key roles in biogeochemical cycles, microbial evolution, and pathogenicity (1). Bacteriophages are also highly versatile and adaptable to a great number of applications, such as phage display (2), phage-assisted directed evolution (3), and phage therapy (4). However, only a few phages are well understood, and 75% of genes coding for virus proteins in the Earth's virome have no homologue in public databases (5). One of the obstacles is the difficulty in engineering phage genomes, especially for virulent phages.

Klebsiella is one of the most common nosocomial opportunistic pathogens. In particular, the multiple-drug-resistant (MDR) strain of K. pneumoniae has been widely spread (6). It is one of the six most common causes of drug-resistant hospital infections identified by the Infectious Diseases Society of America (7). There is an urgent need for antibiotic alternatives to treat MDR K. pneumoniae infection. It was demonstrated that treatment with phages could provide significantly increased survival in mice for K. pneumoniae infections (8), but a lack of efficient phage genome-editing tools largely limits our knowledge about Klebsiella phages. Very recently, a recombineering system based on λ red recombinase for modification of Klebsiella bacteriophage genomes was established (9), where a recombination frequency from 9 to 80% could be achieved depending on the different genes. However, a second infection to obtain the mutant phage is unavoidable, as the plaques in the first infection contain a mixture of wild-type and mutant phages.

The CRISPR-Cas system, an acquired immunity system against invasive genetic elements, has been extensively used to modify all kinds of organisms (10). The type II CRISPR system from Streptococcus pyogenes is one of the most intensively studied CRISPR systems (11), and it requires three minimal components: the CRISPR-associated nuclease Cas9 (SpCas9), an auxiliary trans-activating crRNA (tracrRNA), and a specificity-determining CRISPR RNA (crRNA). Besides that, a protospacer adjacent motif (PAM; 5′-NGG-3′) next to the target site is also required for target recognition by the SpCas9-crRNA-tracrRNA complex. Furthermore, crRNA and tracrRNA can also be fused to generate a single guide RNA (sgRNA) by mimicking the natural crRNA-tracrRNA hybrid (12). A properly designed sgRNA can be more efficient and is now commonly used (12). Until now, there have been few reports on the use of the CRISPR-Cas system for phage genome editing (13–17), especially for the SpCas9-sgRNA system. There are still some unanswered questions about the SpCas9-sgRNA system for phage genome modification. Selection of highly active sgRNA is the first step. Weak sgRNA activity will result in a high rate of false positives. In phages, variable activity of crRNA has been observed in the type I-E and type II-A CRISPR-Cas systems (15, 17). Nevertheless, information about the distribution of sgRNA activity in phages still remains largely unknown. On the other hand, it seems that there are no certain rules of highly active sgRNA for different Cas9 applications or for the same application in different species (18–20). Whether the current sgRNA prediction models trained from eukaryotic data sets could be used for predicting the sgRNA activity targeting phages also remains unknown. Additionally, there was a report on Lactococcus lactis MG1363 successfully achieving precise phage genome editing via modest crRNA activity, where most plaques contained a mixed population of wild-type and recombinant phages (16). It was reported that no random mutant phages could be selected from the strains of the modest phage resistance phenotype, but no data have been rendered about the number of selected escape phages. It is now thought that weak sgRNAs could be used for precise phage genome editing, but failure to select random mutants is a universal rule for CRISPR-based phage genome editing. Finally, construction of recombination template is always one of the most time-consuming procedures for precise genome editing, and it usually includes several rounds of PCR amplification and restriction enzyme digests (21). These cumbersome steps could be eliminated by direct synthesis of recombination template, but the cost of production of long oligonucleotides will be a limitation (22). A short homologous arm for recombination will make the whole procedure easier and save more time (21). Some recombination systems from phages indeed showed high efficiency for homology-directed repair (HDR) in bacteria, even with a homologous arm as short as 30 to 60 bp (23, 24). Comparative analysis of phage genomes also indicated that genomes of phage are highly plastic and have an unusually high degree of horizontal genetic exchange (25, 26). Very recently, using the CRISPR-Cas9 system, a 50-bp homologous arm has been proven to generate point mutation on T4 phage genomes (17). A similar result has been achieved once for phage T7 and 2972 with the native CRISPR-Cas system of type I-E and II-A (13, 14). Therefore, considering the strong selection of CRISPR systems, it is worth exploring systematically whether a short arm is sufficient for SpCas9-sgRNA-based phage genome editing, including point mutation, gene deletion and swap, and even large fragment deletion.

Based on comparative genome analysis, phage phiKpS2 used in this study belongs to the members of Kp34likevirus within the Autographivirinae subfamily (27). To establish a universal and easy-to-use platform for Klebsiella phage genome editing using the SpCas9-sgRNA system, we focused on the effectiveness of a large number of sgRNAs targeting different genome regions of phiKpS2 and their feasibility for genome editing. The efficiency of short arms for all kinds of operations on the phiKpS2 genome, including point mutation, gene deletion and swap, and small frameshift deletion, then was carefully evaluated. Finally, nine genes of phage phiKpS2 were successfully deleted, and their essentiality was preliminarily determined. The well-established CRISPR-based phage genome editing tools will significantly expand our knowledge about Klebsiella phage, and it also has potential applications for other phages, especially for Enterobacter.

RESULTS AND DISCUSSION

Introduction of SpCas9-sgRNA into K. pneumoniae.

In order to develop and apply CRISPR/Cas9 for Klebsiella phage genome editing, we first constructed a pcas9-sgRNA plasmid and selected targets from different regions of virulent phage phiKpS2. A new spacer was introduced into pcas9-sgRNA through a simple procedure (Fig. 1). These plasmids were then amplified in Escherichia coli DH5α and transformed into K. pneumoniae S2 to generate phiKpS2-resistant derivatives of K. pneumoniae. The CRISPR-Cas9-based phage-resistant strains of K. pneumoniae were used as hosts for phage genome editing.

FIG 1.

New spacer cloning into the plasmid of pcas9-sgRNA system for generating phage-resistant derivatives of K. pneumoniae. (a) The pcas9-sgRNA plasmid used in this study. (b) The method of generating sgRNAs targeting a specific region of phages. Spacers targeting different regions of the phiKpS2 genome were directly synthesized as primers and were annealed to double-stranded DNA fragments with sticky ends precisely fitting the plasmid of pcas9-sgRNA, which could be cleaved by BspQI enzyme, resulting in formation of plasmid pcas9-sgRNA with new targets. (c) Genome organization of the phage phiKpS2.

The overview of CRISPR-Cas9-based phage genome editing in this study is presented in Fig. 2. Once phage genome is injected into the host, a CRISPR-based phage-resistant strain, it will be cut at a specific location by Cas9 protein with the guide of sgRNA, resulting in DNA double-strand breaks (DSBs). The Cas9-induced DSBs can be repaired by nonhomologous end joining (NHEJ) or homology-directed repair (HDR) (28). When a homologous recombination template is available, HDR-mediated repair can be used to introduce specific point mutations, gene deletion, or a desired sequence insertion, resulting in precise genome-editing phages. Otherwise, NHEJ-mediated repair can be used to form all kinds of mutant phages with indel mutation in protospacers or PAMs to escape from the immunity of CRISPR-Cas9.

FIG 2.

Overview of CRISPR-Cas9-based phage genome editing. K. pneumoniae S2 with the plasmid of pcas9-sgRNA was infected by phiKpS2, and then a double-strand break on the genome of phiKpS2 forms with the function of the Cas9-sgRNA complex. Finally, the DSB could be repaired by the pathway of NHEJ or HDR, depending on whether the recombination template is available and a random mutant or precise gene editing phage could be selected from the survival phages.

Evaluation of the efficiency of a large number of sgRNAs for CRISPR-based phage resistance.

A total of 87 sgRNAs targeting 11 putative genes and one putative promoter were designed, and their efficiency of restriction of phage infection was determined. As shown in Fig. 3a, the efficiency of these sgRNAs ranged from 10⁻⁵ to 1 (efficiency of plaquing, or EOP), even targeting the same genes, where more than half of sgRNAs had very low activity (EOP of >0.1), and it was not possible to find obvious differences of GC content or some base composition bias at special positions of the spacer between the high and low activity of sgRNAs (Fig. 3b and c).

FIG 3.

Evaluation of the efficiency of a large number of guide RNAs for CRISPR-based phage resistance. (a) The activity of 87 sgRNAs targeting 11 genes and a putative promoter, ranging from high to low. Activity for the three genes holin, internal core protein, and primase was also listed separately. (b) Sequence alignment of the 20-bp specific guide RNAs and PAMs, all with 20% high or low activity. (c) GC content analysis of all 20% high- or low-activity gRNA and the seed region refers to the 1- to 10-bp region close to the PAM. (d and e) The predicted scores from sgRNA Scorer 2.0 and ge-CRISPR were compared with the observed activity. (f) Correlation analysis of the predicted activity and observed activity.

Whether the activity of these sgRNAs could be predicted by the publicly available sgRNA design tools was then determined. All 87 sgRNAs were scored using sgRNA Scorer 2.0 (29) and ge-CRISPR (30), which were trained from eukaryotic data sets. The predicted score then was compared with the observed activity (Fig. 3d and e), while both of the algorithms had an R² of less than 0.02, demonstrating their poor predictive capability (Fig. 3f).

The highly variable targeting activity of sgRNAs has been observed in many organisms (31). Until now, the molecular mechanisms underlying high guide RNA efficiency, where the rules seem to change depending on the species used, have remained unknown (18, 19). For phages, a variable activity of CRISPR system depending on the spacer used was also observed in some reports (15, 17). The results here clearly demonstrate the limitations of current eukaryotic cell-based sgRNA activity models for predicting sgRNA activity targeting phages. Very recently, a genome-wide activity study in bacteria verified a similar phenomenon for E. coli (32), where it was discovered that the correlation between the observed sgRNA activity targeting the chromosome of E. coli and the predicted activity of current sgRNA models trained from eukaryotic data sets was very low. We failed to find a difference in sequence characteristics with high-activity and low-activity sgRNA, indicating that there are other inhibitors. For example, some DNA modifications on phage genomes and the DNA target site accessibility in eukaryotic cells have been proven to influence the activity of CRISPR/Cas9 (18, 33). Therefore, a more complete investigation of the on-target activity of sgRNA on phages may be necessary.

Mutation pattern of the escape phage from different CRISPR targets.

After infecting three strains of K. pneumoniae with three different highly active sgRNAs (pCas9-priNT3, -icpNT1, and -proT1) targeting early- and late-expressed genes and a putative promoter of phiKpS2, a total of 91 well-isolated phage plaques were selected. The corresponding genomic regions of these escape phages were sequenced, and 90 escape phages showed mutations in the protospacer or PAM (Fig. 4a to c), which indicated the strong selection of our CRISPR system. It is worth noting that none of 32 selected phages escaping from the low-activity sgRNAs (pCas9-proNT1 and -proNT2, which have low EOPs of 0.62 and 0.21) had mutations in the protospacer or PAM (Fig. 4d). Therefore, a highly active sgRNA is required for efficient selection of randomly mutant phages.

FIG 4.

Mutation pattern of the escape phage from different CRISPR targets. (a) A putative promoter with highly active sgRNA proT1. (b) DNA primase with highly active sgRNA priNT3. (c) Internal core protein with highly active sgRNA icpNT1. (d) A putative promoter with two low-activity sgRNAs, proNT1 (1–16) and proNT2 (17–32). All sequences were aligned with the sequence of wild-type phiKpS2 and analyzed using Geneious 11 (Biomatters Ltd., Auckland, New Zealand). All of the mutant bases were labeled with different colors (red for A, green for T, blue for C, and yellow for G).

Phage can respond to CRISPR-encoded resistance by rapid evolution through single-nucleotide mutation and deletion either in the protospacer or PAM (34). Usually, single-nucleotide mutation is the most common way for phages to escape from CRISPR interference, and most mutations in the protospacer are located in a 7- to 12-bp seed region flanking the PAM (35, 36). For example, in a previous study which involved using native Streptococcus thermophilus CRISPR-Cas type II-A targeting different regions of phage 2972 (35), 380 out of 382 mutant phages arose due to single-nucleotide mutation, and most of those mutants were found in a 9-bp seed region and the PAM. In this study, all mutants had at least one point mutation located in the protospacer or PAM, as in previous studies. However, there were some mutation patterns inconsistent with the previous studies: (i) for targeting late-expressed structural genes, 5 out of 35 escape phages having double mutations had a second mutation beyond the region of protospacer and PAM; (ii) for targeting a putative promoter of phiKpS2, at least five different mutation patterns existed, including single-nucleotide mutation (4 of 25 mutant phages), double-nucleotide mutation (2 of 25 mutant phages), single-nucleotide replacement and insertion (8 of 25 mutant phages), double-nucleotide replacement, deletion and insertion (6 of 25 mutant phages), and large-fragment deletion (5 of 25 mutant phages); (iii) for targeting the early-expressed primase gene, all mutations were located in just three positions, two GG of PAM for SpCas9 (18 of 30 mutant phages) and the 4th nucleotide flanking the PAM (12 of 30 mutant phages). These results showed that the mutation pattern of phages against CRISPR interference varied largely depending on targeting positions.

The stress of sense mutations could be an explanation for the different mutation patterns of phage escaping from the CRISPR system targeting different genes. Specifically, some mutations beyond the protospacer and PAM in this study were possible to eliminate the bad effects of single-nucleotide mutation, and some existing hotspots on phiKpS2 also may exist because mutations in these positions have little effect on the function of encoded proteins. Similarly, mutation patterns can be diverse when the targeting region is not required for phage infection. In general, the above-described results indicated that the mutation pattern of phages against CRISPR interference was determined not only by the characteristic of CRISPR systems but also by the targeting genes.

Genome engineering of a lytic K. pneumoniae phage, phiKpS2.

Phage with random mutations in the protospacer or PAM could be selected from the SpCas9-sgRNA system in K. pneumoniae. Therefore, it may be possible to directly obtain a precise gene-editing phage by providing a recombination template (14), which must contain a desired mutation to escape from the CRISPR interference and include sequence homologous to the phage genome to be edited. In the present study, a high-copy-number plasmid, pDK6 (37), with a selection marker of the kanamycin-resistant gene, was used to construct all kinds of templates for point mutation, gene deletion, and gene swap.

Introduction of point mutations with short homologous arms.

Three highly active sgRNAs (proT1, priNT3, and icpNT1) targeting different regions of phiKpS2 (Fig. 1) were selected to test the efficiency of point mutation with short homologous arms. All templates were synthesized directly as primers and cloned into the template plasmid with a simple procedure (Fig. 5). The length of homologous arms ranged from 10 bp to 40 bp by an increment of 10 bases (Fig. 6).

FIG 5.

Construction of homologous recombination templates for point mutation, gene deletion, and swap. The recombination templates for point mutation and gene deletion were directly synthesized as primers and annealed to form double DNA fragments with sticky ends precisely fitting pDK6, which could be cleaved by EcoRI and HindIII, resulting in pDK6-PM and pDK6-Del. For gene swap, a pair of primers with short homologous arms and restriction sites of restriction endonuclease were synthesized. After overlap PCR amplification, the fragment was double digested and ligated into the vector pDK6 to form the pDK-GS plasmid. The black color represents the targeting region, and red color represents the desired sequence of mutation. The length of these fragments depends on the length of homologous arms, ranging from 24 bp to 124 bp in the present study.

FIG 6.

Introduction of point mutations with short homologous arms. (a) An example of short homologous arms for point mutation. (b) The efficiency of point mutation with homologous arms ranging from 10 bp to 40 bp, targeting different regions of the phiKpS2 genome. Efficiency of point mutation is defined as the ratio of the number of desired mutants to the number of sampled plaques. In this experiment, at least 8 plaques were randomly sampled each time, and all experiments were repeated three times.

After infecting the hosts bearing the corresponding pcas9-sgRNA plasmid and recombination template, the escape phages were selected and purified, followed by sequencing of the PCR products of corresponding regions. A point mutation with 100% editing efficiency was obtained for all three targets when the length of the homologous arm was 30 bp or more, and even with homologous arms as short as 20 bp, point mutation efficiencies ranging from 28.6% to 81.3% also could be obtained, where fluctuations may occur due to the instability of the repair complex binding to the short arm. All of the results presented here indicated that a homologous arm as short as 30 bp was sufficient to make a point mutation on the genome of phiKpS2.

Gene deletion and swap with short homologous arms using both strong and weak sgRNAs.

By using the same strategy as that described above, template plasmids with short homologous arms for gene deletion and swap were constructed (Fig. 5). First, we tested to delete a 500-bp fragment with short homologous arms of 30, 40, and 50 bp using a highly active sgRNA (proT1). An editing efficiency of 100% could be obtained just with a homologous arm as short as 40 bp (Fig. 7a). An effort then was made to replace the 500-bp fragment with a gene coding for red fluorescent protein (923 bp) using the short homologous arms of 40, 50, and 60 bp. The highest editing efficiency of 87.5% could be obtained with a 60-bp homologous arm, while the editing efficiency decreased to 60.4% and 41.7% with the homologous arms of 50 and 40 bp, respectively (Fig. 7b). Furthermore, an editing efficiency of 76.7% and 56.7% for deletion of 1-kb and 2-kb fragments, respectively, could be obtained with a 40-bp homologous arm (Fig. 7c and d). It was worth noting that the length of the whole template plus the cohesive end of a 40-bp homologous arm for gene deletion was less than 90 bases, as was the directly synthesizing overlap primers for gene swap, where a 60-bp homologous template, a 20-bp primer for amplification of the gene of interest, and a restriction site for ligation were all included in the overlap primers (Fig. 5). As we know, there are still some limitations in high-throughput synthesis of oligonucleotides with a length of >100mer (38, 39), where the cost per base will increase several times compared with that of short oligonucleotides (22). Therefore, the result that a short template was sufficient for CRISPR-based phage genome editing would be significant, as it could be synthesized directly as primers at very low cost and cloned into the template plasmid with a very simple procedure.

FIG 7.

Gene deletion and insertion with short homologous arms. Evaluation of efficiency for gene deletion (a) and insertion (b) with short homologous arms. (c) The efficiency for gene deletion of large fragments with 40-bp homologous arm. Efficiency of gene deletion and insertion is defined as the ratio of the number of desired mutants to the number of sampled plaques. In these experiments, at least 8 plaques were randomly sampled each time, and all experiments were repeated three times. (d) Confirmation of deletions in the genome of recombinant phiKpS2 by PCR. PCR products were migrated on a 1% agarose gel, together with DL10000 DNA marker (TaKaRa Biotech) (lane 1). Lanes 2 to 4, PCR products on wild-type phage and recombinant phiKpS2 for deletion of 0.5-kb, 1.0-kb, and 2-kb fragments. (e) PCR-based detection of deletions (0.5 kb) in the genome of survival phages with low- and high-activity sgRNAs (proNT1, proNT2, and proT1), targeting the putative promoter with (top) and without (bottom) available recombination templates. PCR products were migrated on a 1% agarose gel together with DL2000 DNA marker (TaKaRa Biotech) (lanes 1 and 26). Lanes 2 to 13, PCR products of survivors from proNT1; lanes 14 to 25, PCR products of survivors from proNT2; lanes 27 to 38, PCR products of survivors from proT1. (f) A possible explanation that weak sgRNA was useful for precise phage genome editing but failed to select random recombinant. A weak sgRNA fails to cut all of the copies of phage genomes, and the uncut phage genomes could act as a template for homologous recombination. Therefore, phages could survive without mutation from weak sgRNAs when no template is available, and recombinant phages could be obtained when an external template is available.

Finally, the efficiency of gene deletion was tested using two weak sgRNAs (proNT1 and proNT2, with EOPs of 0.62 and 0.21, respectively). Results showed that with the help of a 40-bp homologous arm, these weak sgRNAs were sufficient to readily modify the genome of phage phiKpS2 with nearly 100% efficiency, where all plaques contained a mixture of wild-type and mutant phages (Fig. 7e). It was noted that no mutations could be detected from all 32 phages selected from these weak SpCas9-sgRNA systems when the recombination template was unavailable (Fig. 4d). The phenomenon has also been observed in L. lactis for genome editing of lactococcal phage p2 (16). A possible explanation of those results is that inefficient cleavage can be tolerated through continuous repair by the HDR pathway (Fig. 7f). Briefly, a weak sgRNA fails to cut all copies of phage genomes, and the uncut phage genomes could act as a template for homologous recombination. Therefore, phages could survive without mutation from weak sgRNAs when no template is available, and recombinant phages also could be obtained when an external template is available (Fig. 7f). A strong sgRNA will cut all the copies of phage genomes, where phage could survive only via the NHEJ pathway for DSB repairs when no external template is available for homologous recombination (Fig. 2). A similar tolerating mechanism for a weak chromosome-targeting CRISPR-Cas system in E. coli recently has been verified (40).

In summary, a homologous arm as short as 40 bp was sufficient to generate recombinant phages with point mutation and gene deletion with an efficiency near 100%, and a 60-bp homologous arm for gene swap could achieve a high efficiency of 87.5%, which makes the proposed CRISPR-based phage genome editing an efficient, time-saving, and cost-effective tool. This result also clearly indicates that with the help of an external homologous template, a weak sgRNA could still be used to select recombinant phages. Most sgRNAs used in this study showed low activity (Fig. 3a), and some DNA modifications on the genome of phages could also largely reduce the activity of sgRNA (17, 33). Therefore, the fact that sgRNAs with low activity can be used for gene editing is significant and will greatly expand the editable locations of phage genome based on CRISPR technology.

Small frameshift deletion: a rapid and efficient way to detect the essentiality of phage genes.

In the present study, a 2-bp frameshift deletion by recombination was also tested using highly active sgRNAs of proT1, priNT3, icpNT1, H2, and gp2-3, which target five different regions of the phiKpS2 genome. A 40-bp homologous arm was constructed, as mentioned above, which is expected to delete the GG of the SpCas9 PAM NGG. A small deletion with 2 bases that had not been previously obtained from the escape phage in the absence of recombination template is available. Since the gene consists of triplet code, this small deletion usually could inactivate the targeting gene.

Sequence analysis showed that 100% of the escape phages had a 2-bp frameshift deletion within the putative phage promoter, the putative protein gp2, or the holin gene (Fig. 8). It can be expected that a 2-bp frameshift deletion will be successfully formed within the putative phage promoter or gp2 once the mutants with complete or partial deletions in these two regions have been obtained (Fig. 4a). Unexpectedly, a 2-bp frameshift deletion of holin was also obtained, since it is one of three core genes (in order, spanin-holin-endolysin genes), forming the lysis cassette of KP34likevirus (27), which was believed to be involved in bacterial lysis. The results here indicate that the holin gene is unnecessary for phage infection under our experimental conditions.

FIG 8.

Mutation patterns of the escape phages for small frameshift deletion targeting different genes. Increase of EOP calculated by dividing the EOP of the host by that with the homologous arms and that with empty vector.

No 2-bp frameshift deletion was obtained for the primase gene and internal core protein (Fig. 8). Sequence analysis showed that all escape phages from the sgRNA-targeting primase gene had a 3-bp deletion of 5′-CGG-3′ without frameshift, and all of the escape phages for the latter had only point mutations (Fig. 8). Obviously, these results further confirmed that the primase protein and internal core protein are essential for phage survival, which is consistent with their respective important roles in DNA replication and phage architecture (41, 42). Interestingly, the EOPs of hosts with or without recombination templates for the two targets (the primase gene and internal core protein) showed different trends, where the former increased by nearly 10-fold and the latter decreased by 2 to 3 orders of magnitude (Fig. 8). A possible explanation is that A₂₇ (alanine at position 27 was deleted in the escape phages) of the primase is not necessary for protein function, which is consistent with the fact that most mutations are located in A₂₇, as shown in Fig. 4. However, the amino acids within the PAM of internal core protein must be important for protein function, which is also consistent with the fact that only a few mutations are located within the PAM (Fig. 4 and 8). Therefore, for the target of internal core protein, it can be expected that the EOP of a host with a recombination template will significantly decrease, since most broken genomes will be occupied by the HDR complex but no recombinants from it could survive. Additionally, it is surprising that 100% of escape phages from pcas9-sgRNA-icpNT1 had 3-bp deletions with high repair efficiency when a template of a 2-bp deletion was provided. The reason behind this phenomenon is still unclear, which indicates that the homologous recombination mechanism of Cas9-induced phage genome DSBs is somewhat mysterious.

All in all, these results indicate that it is difficult to form a small frameshift deletion in essential phage genes through the HDR pathway due to lethal effects. Therefore, combined with the sequence analysis of escape phages, the strategy of small frameshift deletion shown here could be a way to rapidly and efficiently detect the essentiality of putative phage genes.

Primary evaluation of the essentiality of phiKpS2 genes by genome editing.

Most genes in the genus of KP34likevirus remain uncharacterized (27), as does phiKpS2. Among them, there are no known functions for eight adjacent genes (gp1 to gp8), as shown in Fig. 1. A conserved phage promoter (Fig. 1) was once identified in KP34likevirus using bioinformatics analysis (27), but its activity has not been verified. Additionally, the holin gene is a conserved gene for the lysis cassette in KP34likevirus and its mutant, with a 2-bp frameshift deletion, was easily obtained, which makes it fascinating to explore the essentiality of holin.

Several rotations of deletion for the targeting genes were carried out using the aforementioned strategies. In particular, the eight adjacent genes and putative phage promoter were deleted by two consecutive recombinations using the highly active sgRNA proT1 and gp8-5 (see Table 3). First, a 2-kb fragment was deleted, which included the whole genes from gp2 to gp6, the partial genes of gp1 and gp7, and the conserved phage promoter, and then the entire gp8 gene (1593 bp) was deleted based on the mutant phiKpS2-Δ2k to form the recombinant phiKpS2-Δgp1_8 (Fig. 9a). The holin gene was deleted using a highly active sgRNA H2 (see Table 3) to form a mutant phiKpS2-Δholin (Fig. 9a). All of the mutant phages used here were screened and purified three times in succession. The results showed that the mutant phages phiKpS2-Δgp1_8 and phiKpS2-Δholin both formed normal plaques. One-step growth experiments also showed no obvious difference between the mutants and wild-type phages (Fig. 9b). Therefore, it was concluded that all nine genes and the putative promoter can be deleted without effect on phage phiKpS2 growth. However, the reason behind this is not yet clear. One explanation is that these genes are nonessential for phage growth. Another explanation is that cryptic gene duplications or suppressor mutations exist that allow the phage to overcome the loss of the gene (43).

TABLE 3.

sgRNAs used in this study

sgRNA	EOP	Sequence (5′–3′)	sgRNA	EOP	Sequence (5′–3′)
H1	0.13	GATGATTAAAGTAGGGGACA	Pri4	0.67	CTGCGGCAGGGACGCCGCCG
H2	3.00E−04	ATTAAAGTAGGGGACATGGT	Pri5	7.00E−05	TGACCCGGCAGGTCACGCTG
H3	0.93	TTGGATCTGACCTCGCTACC	Pri6	0.27	TCTAAAGGTCGTGCTCACCG
H5	2.70E−02	TCTGACCTCGCTACCCGGGC	Pri7	4.00E−02	CACGCTTGCCCACGCCGCTG
H6	0.11	CTGCACCTGCCCGGGTAGCG	Pri8	0.40	GCTGCGGCAGGGACGCCGCC
H7	6.00E−04	GGTAACTGCTGCACCTGCCC	Pri9	0.40	ATGCTCGGCAGCAACCGAAG
H8	0.12	CGGTAACTGCTGCACCTGCC	Pri10	0.53	GCATGAACGGACGCAGTGCG
H9	0.11	CGGGCAGGTGCAGCAGTTAC	PriT1	0.33	GCTTAGAGCGTGCAAGCGCC
H10	3.90E−02	TGCAGCAGTTACCGGCGCTA	PriNT1	0.25	ATGCATTACTCCGATTTAAC
H11	3.90E−02	GTTACCGGCGCTACGGTATC	PriNT2	0.58	GCGCTTGCACGCTCTAAGCC
H12	8.67E−04	ACCGGCGCTACGGTATCAGG	PriNT3	3.58E−04	GTTATTGTAGAGCACCCCGG
H13	6.00E−04	ACCTCCTGATACCGTAGCGC	proT1	5.42E−04	CAAGCCTATAGCGTCCTACG
H14	1.30E−02	GCGCTACGGTATCAGGAGGT	proNT1	0.63	AGCGCCCCGTAGGACGCTAT
H15	8.00E−03	TACGGTATCAGGAGGTTGGT	proNT2	0.21	GCATTCACATAGCGCCCCGT
H16	2.70E−03	GGTTGGCAGAGTTAATGAGC	gp1-1	0.10	TAAGTTGAAACAACAATTCG
H17	0.47	CAGAGTTAATGAGCTGGAAC	gp1-2	2.60E−02	GAGGGGCTTGAGAAGTTATC
H18	0.40	CATCACTGCGACTGTGTGCG	gp1-3	5.75E−04	GCCCCGAGTTGAGAAAGCGC
H19	8.70E−04	TGTGTGCGCGGTGCTAACCC	gp2-1	0.29	CGCATCATGACTAATTCAAC
H20	2.30E−02	GCGCGGTGCTAACCCTGGTG	gp2-2	6.00E−02	GTATTCAAGTTAACTGCCGC
H21	6.10E−04	GTAATACGCATTCCACACCA	gp2-3	1.11E−03	AGCTTTACGAATGCTGCCAG
H22	7.30E−04	TGTAATACGCATTCCACACC	gp3-1	0.20	AAGTCCCGGTTATCAAGTGG
H23	2.50E−03	TGTGGAATGCGTATTACAAG	gp3-2	0.63	GTGGTGGGCCTTGTGCCCTG
H24	8.00E−04	GCGTACATTCAAGCTCCTAG	gp3-3	3.63E−02	GCGTCGGAATCGTTGGCCGC
H25	8.00E−03	ATTCAAGCTCCTAGAGGAGC	gp4-1	0.54	GTTTCATATGAAAGCGATAT
H26	0.93	CCTTACGTGCCTGCTCCTCT	gp4-2	0.29	GCCATGAACTCTGGCCCGTG
H27	3.30E−03	CTAGAGGAGCAGGCACGTAA	gp4-3	6.25E−02	TATACCCGATTGCACCACAC
H28	3.30E−03	TAGAGGAGCAGGCACGTAAG	gp5-1	0.50	CATTGTGACCAGCGCTATGC
H29	1.07	GACTATTAAATATGAGTTTA	gp5-2	6.25E−02	GGGCAACGACGTAAACAACC
H30	0.53	AAATATGAGTTTAAGGACTA	gp5-3	0.75	TGCACTGTGCAATAGGGCAC
H31	6.67E−05	TTTAAGGACTAAGGTTATTG	gp6-1	0.67	GGCAGCAACCAACAATCGAA
ICP1	1.07E−04	GGATGCAGATACAAGCAAGG	gp6-2	0.25	GGCGCGAGAGTACAGCGCTA
ICP2	0.60	GAGAGCTGAACGATGCTGGG	gp6-3	0.88	CGCCTGGGTATGACCCTGCG
ICP3	6.67E−03	GGGCGGACCGGAACAGGGCG	gp7-1	8.30E−02	CAAGCGCCAGATAAACGCTG
ICP4	4.00E−02	TACGTGCAGCTAAACCCGCG	gp7-2	0.33	CTGGAAACGCAGGGCAAGTG
ICP5	2.67E−04	CCAGCGCAGCCCGCTCAGCG	gp7-3	0.792	CCCTTGGGAAACATTCAAAG
ICP6	0.60	CGAGATTGCTAAACGCCTGG	gp7-4	0.114	TGAACTCAAGCAAATCGTGC
ICP7	9.33E−05	GGTCCTGGAAGGCTTCACTG	gp7-5	4.29E−02	AGTCCTGCTTACGAACTCGG
ICP8	0.53	CATCGGGGCCAGGATGAACG	gp8-1	1.29	TGGGGTCCAGTAATCACCGC
ICP9	8.00E−03	CGCGCGGCCAGCAATACCGG	gp8-2	0.79	CATTTGCAAGTCCTGGAATG
ICP10	4.00E−05	AATACCGGCGGCAGACATCG	gp8-3	0.75	CGCTGACCATCCCCACACTG
icpNT1	3.47E−04	AGCGGAGCAATCTCTGCATC	gp8-4	5.71E−02	CGCCCAGCATTTGCAAGTCC
Pri1	0.27	TACAGTTACAGGAAGAGCCG	gp8-5	9.29E−03	AGACCCATAAGCGCCTGCAG
Pri2	0.27	CAGCCAGGCCCTCAGCGGCG	gp8-6	0.11	TATAGTCACCTCTAAGCCCG
Pri3	0.27	TGAATCAGCCAGGCCCTCAG

FIG 9.

Primary evaluation of the essentiality of nine putative genes (gp1 to gp8 and holin) of phiKpS2. (a) PCR-based confirmation of deletions in escape phages. PCR products were migrated on a 1% agarose gel, together with DL10000 DNA marker (M1) and DL2000 DNA marker (M2 and M3) (TaKaRa Biotech). (b) One-step growth curve of recombination phages phiKpS2-Δgp1_8 and phiKpS2-Δholin.

Conclusions.

A simple strategy for CRISPR-based phage genome editing was developed based on the discovery that a short homologous arm was enough to introduce point mutation, gene deletion, and swap into Klebsiella phage phiKpS2. As shown here, small frameshift deletion is a potential way to rapidly identify the essentiality of phage genes. By using the above-described strategies, we successfully deleted a putative phage promoter and nine phiKpS2 genes (eight with unknown function and one coding for holin), which accounted for 17% of all annotated phiKpS2 genes and 8% of the whole-genome sequence. Undoubtedly, the SpCas9-sgRNA tool used here established an efficient, time-saving, and cost-effective genome-editing system for Klebsiella phage that has the potential to significantly expand our knowledge of phage-host interactions and to promote applications of phage therapy.

Selection of highly active sgRNA targeting phages may be a challenge. There are 50.6% (44 out 87) sgRNAs having very weak activity (EOP of >0.1) in this study, and the observed activity cannot be predicted using models trained on eukaryotic cell data sets. Interestingly, although they could only provide poor phage resistance phenotypes, weak sgRNAs have been shown here to be useful for precise phage genome editing by homologous recombination with an external homologous template, making CRISPR-based tools easier to use. Apart from the strong selection of the CRISPR-Cas9 system, the method here and other CRISPR-based phage genome-editing strategies (13–17) all largely depend on the high efficiency of the HDR pathway for the repair of phage genome DSBs, but it is not yet clear whether the host's system or the bacteriophage's own system is at work for the DSB repair. With the help of the efficient CRISPR-based phage genome-editing tools, it is believed that more knowledge about the DSB repair system and phage-host interactions will be easily obtained in the future.

MATERIALS AND METHODS

Bacterial strains, bacteriophages, and culture conditions.

K. pneumoniae S2 (44) and its derivatives were kept in 20% (wt/wt) glycerol stock solution at −70°C. The strain of K. pneumoniae S2 was used as an indicator strain for phiKpS2 reproduction. Bacteriophage phiKpS2 and its derivatives were kept at 4°C, and the whole-genome sequence of phiKpS2 is available from GenBank under accession number KX587949. The open reading frame (ORF) of genome phage phiKpS2 was identified by GeneMark.hmm with heuristic models (45), and the identified ORF was annotated by BLASTP against the NCBI nonredundant (nr) database (Table 1). E. coli DH5α was used to produce chemically competent cells for cloning. All strains in this study were grown at 37°C in LB medium. For solid media, agar (1.2%, wt/vol) was added to LB, and for double-layer plaque assays, a bottom layer of the medium was supplemented with 1.2% (wt/vol) agar. A top layer of the medium was supplemented with 0.6% (wt/vol) agar.

TABLE 1.

Phages phiKpS2 predicted genes and gene products

ORF	Location (nta)	Product size (aaa)	Homologue	GenBank accession no.	Sequence identity (%)
gp1	1403–1618	71	Hypothetical protein kpv74_01	APZ82713.1	99
gp2	1678–2265	195	Hypothetical protein F19_03	YP_009006023.1	98
gp3	2262–2393	43	Hypothetical protein SU552A_03	YP_009204792.1	100
gp4	2443–2676	77	Hypothetical protein kpv71_05	YP_009302709.1	99
gp5	2669–2932	87	Hypothetical protein kpv41_05	YP_009188747.1	99
gp6	2941–3120	59	Hypothetical protein SU552A_06	YP_009204795.1	98
gp7	3308–3652	114	Hypothetical protein kpv74_08	APZ82720.1	99
gp8	3780–5372	530	Hypothetical protein F19_09	YP_009006029.2	95
gp9	5372–6418	348	Putative peptidase	YP_003347662.1	98
gp10	6421–6672	83	Hypothetical protein KPRIO2015_10	AOT23849.1	85 (QCb, 62)
gp11	6669–6890	73	Hypothetical protein SU552A_11	YP_009204800.1	95
gp12	6887–7051	54	Hypothetical protein kpv41_13	YP_009188755.1	56
gp13	7055–7840	261	DNA primase	YP_009188325.1	99
gp14	7843–8031	62	Hypothetical protein	YP_003347666.1	98
gp15	8032–9312	426	Putative DNA helicase	YP_009280683.1	99
gp16	9365–9520	51	Hypothetical protein kpv41_17	YP_009188759.1	98
gp17	9517–11928	803	Putative DNA polymerase	YP_009188761.1	96
gp18	11925–12146	73	Hypothetical protein KPRIO2015_17	AOT23856.1	95
gp19	12308–13351	347	Putative phosphoesterase	APZ82732.1	99
gp20	13323–13532	69	Hypothetical protein kpv41_22	YP_009188764.1	100 (QC, 79)
gp21	13584–14408	274	Large tegument protein	YP_009204811.1	98
gp22	14461–14691	76	Hypothetical protein kpv41_24	YP_009188766.1	97
gp23	14720–15091	123	Hypothetical protein kpv71_26	YP_009302730.1	98
gp24	15094–15291	65	Hypothetical protein kpv41_27	YP_009188769.1	94
gp25	15291–15452	53	Hypothetical protein SU552A_28	YP_009204817.1	100
gp26	15452–16420	322	Putative 5′-3′ exonuclease	AOZ65238.1	99
gp27	16401–16577	58	Hypothetical protein kpv475_28	YP_009280697.1	98
gp28	16571–16801	76	Hypothetical protein	BAS69540.1	38
gp29	16798–17220	140	Putative DNA endonuclease VII	YP_009006047.1	99
gp30	17217–17711	164	Polynucleotide kinase/phosphatase	YP_009204821.1	96
gp31	17708–18022	104	Hypothetical protein kpv48_34	AOZ65242.1	100
gp32	18164–20632	822	DNA-dependent RNA polymerase	YP_009026413.1	99
gp33	20642–20980	112	Hypothetical protein F19_30	YP_009006052.1	98
gp34	21004–21444	146	Hypothetical protein F19_31	YP_009006053.2	99
gp35	21441–21704	87	Hypothetical protein	YP_003347631.1	100
gp36	21714–23309	531	Head-tail connector protein	YP_003347633.2	98
gp37	23324–24166	280	Putative scaffolding protein	YP_009188780.1	99
gp38	24192–25214	340	Putative capsid protein	YP_009188781.1	99
gp39	25226–25408	60	Hypothetical protein SU552A_40	YP_009204829.1	100
gp40	25497–26057	186	Putative tail tubular protein A	YP_009188783.1	99
gp41	26067–28427	786	Putative tail tubular protein B	YP_009188784.1	98
gp42	28429–29016	195	Putative internal virion protein B	YP_009204832.1	100
gp43	29034–31718	894	Hypothetical protein SU552A_44	YP_009204833.1	99
gp44	31769–35467	1232	Putative internal core protein	YP_009199928.1	98
gp45	35469–38309	946	Putative tail fiber protein	YP_009188788.1	78 (QC, 36)
gp46	38311–38613	100	Putative DNA maturase A	YP_003347644.1	100
gp47	38613–40469	618	Putative DNA maturase B	YP_009188791.1	99
gp48	40469–40843	124	Hypothetical protein kpv71_48	YP_009302752.1	100
gp49	40855–41037	60	Hypothetical protein kpv41_51	YP_009188793.1	100
gp50	41037–41441	134	Putative spanin	YP_009204840.1	99
gp51	41434–41685	83	Putative holin	YP_009199935.1	98
gp52	41669–42268	199	Putative endolysin	YP_009006073.1	98
gp53	42279–44024	581	Putative tail fiber protein	APZ82768.1	97

^ant, nucleotide; aa, amino acid.

^bQC, query coverage; only the values of QC under 95% are shown.

Plasmid constructions.

Plasmid pcas9-sgRNA carries a wild-type cas9 gene from S. pyogenes with a trc promoter, an sgRNA (+85) gene under a strong constitutive promoter of BBa_J23119 (SpeI), a broad-host-range, medium-copy-number replicon from pBHRK18, and a chloramphenicol resistance marker (Cm^r). The sgRNA biobrick contains three parts: a 20-bp based-pairing region complementing the gene sequence of interest, a hairpin region for Cas9 protein binding, and a 40-bp terminator. Two restriction sites of type II restriction endonuclease BspQI were inserted in sgRNA biobrick so a new spacer could be easily cloned. The 20-bp target site complementary sequence, with another 3 bases fitting the sticky end, was synthesized directly as primers. The pairs of primers were subsequently annealed to obtain a double-stranded inserted fragment precisely fitting the pcas9-sgRNA vector, which could be cleaved by BspQI enzyme, resulting in formation of plasmid pcas9-sgRNA with new target.

The homologous donor plasmids were constructed by cloning donor DNA into pDK6 vector. As most of the templates used in this study were characterized by short arms, they could be synthesized directly as primers with a sticky end fitting the vector. Therefore, the homologous donor plasmids pDK-PM and pDK-Del for point mutation and gene deletion could be easily constructed without any PCR amplifications. A pair of primers with short homologous arms and restriction sites of restriction endonuclease was synthesized directly to construct the templates for gene swap. After overlap PCR amplification, the fragment was double digested and ligated into the vector pDK6 to form the pDK-GS plasmid.

All of the primers and sgRNAs used in this study are listed in Table 2 and Table 3, respectively.

TABLE 2.

Primers for PCR and oligonucleotides used in this study

Primer	Sequencea (5′–3′)	Purpose
priG20T-F	agctGGCAGGGACGACGCCGGGGT	Point mutation
priG40T-F	agctCGGTGCTGCGGCAGGGACGACGCCGGGGTGCTCTACAATA	Point mutation
priG60T-F	agctGGCGCGCTTTCGGTGCTGCGGCAGGGACGACGCCGGGGTGCTCTACAATAACCCGGATGC	Point mutation
priG80T-F	agctTGGGGCAGAGGGCGCGCTTTCGGTGCTGCGGCAGGGACGACGCCGGGGTGCTCTACAATAACCCGGATGCCTGGGAATAT	Point mutation
priG20T-R	aattACCCCGGCGTCGTCCCTGCC	Point mutation
priG40T-R	aattTATTGTAGAGCACCCCGGCGTCGTCCCTGCCGCAGCACCG	Point mutation
priG60T-R	aattGCATCCGGGTTATTGTAGAGCACCCCGGCGTCGTCCCTGCCGCAGCACCGAAAGCGCGCC	Point mutation
priG80T-R	aattATATTCCCAGGCATCCGGGTTATTGTAGAGCACCCCGGCGTCGTCCCTGCCGCAGCACCGAAAGCGCGCCCTCTGCCCCA	Point mutation
proG20T-F	agctGTCCTACGGGTCGCTATGTG	Point mutation
proG40T-F	agctGCCTATAGCGTCCTACGGGTCGCTATGTGAATGCAACTAG	Point mutation
proG60T-F	agctTATAGTAGCAAGCCTATAGCGTCCTACGGGTCGCTATGTGAATGCAACTAGCGTATGAGG	Point mutation
proG80T-F	agctGGTGAGTACGTATAGTAGCAAGCCTATAGCGTCCTACGGGTCGCTATGTGAATGCAACTAGCGTATGAGGTTTCATATGA	Point mutation
proG20T-R	aattCACATAGCGACCCGTAGGAC	Point mutation
proG40T-R	aattCTAGTTGCATTCACATAGCGACCCGTAGGACGCTATAGGC	Point mutation
proG60T-R	aattCCTCATACGCTAGTTGCATTCACATAGCGACCCGTAGGACGCTATAGGCTTGCTACTATA	Point mutation
proG80T-R	aattTCATATGAAACCTCATACGCTAGTTGCATTCACATAGCGACCCGTAGGACGCTATAGGCTTGCTACTATACGTACTCACC	Point mutation
icpG20T-F	agctTTCCGGTGCAGGATGCAGAG	Point mutation
priG40T-F	agctGGTCGCGCTGTTCCGGTGCAGGATGCAGAGATTGCTCCGC	Point mutation
priG60T-F	agctCGACGTAAACGGTCGCGCTGTTCCGGTGCAGGATGCAGAGATTGCTCCGCTGGTTGATGC	Point mutation
priG80T-F	agctCTTATAGCGCCGACGTAAACGGTCGCGCTGTTCCGGTGCAGGATGCAGAGATTGCTCCGCTGGTTGATGCTTACCGTCGC	Point mutation
priG20T-R	aattCTCTGCATCCTGCACCGGAA	Point mutation
priG40T-R	aattGCGGAGCAATCTCTGCATCCTGCACCGGAACAGCGCGACC	Point mutation
priG60T-R	aattGCATCAACCAGCGGAGCAATCTCTGCATCCTGCACCGGAACAGCGCGACCGTTTACGTCG	Point mutation
priG80T-R	aattGCGACGGTAAGCATCAACCAGCGGAGCAATCTCTGCATCCTGCACCGGAACAGCGCGACCGTTTACGTCGGCGCTATAAG	Point mutation
F-del30proT1	agctGTTAAAAACGCCATCAAGAAGGCTAAAGAGTGGTTAACGTGTTCAACATCATTGTGACCA	Gene deletion
R-del30proT1	aattTGGTCACAATGATGTTGAACACGTTAACCACTCTTTAGCCTTCTTGATGGCGTTTTTAAC	Gene deletion
F-del40proT1	agctGCTGGGGCGCGTTAAAAACGCCATCAAGAAGGCTAAAGAGTGGTTAACGTGTTCAACATCATTGTGACCAGCGCTATGCT	Gene deletion
R-del40proT1	aattAGCATAGCGCTGGTCACAATGATGTTGAACACGTTAACCACTCTTTAGCCTTCTTGATGGCGTTTTTAACGCGCCCCAGC	Gene deletion
F-del50proT1	agctGCTTGCAAGCGCTGGGGCGCGTTAAAAACGCCATCAAGAAGGCTAAAGAGTGGTTAACGTGTTCAACATCATTGTGACCAGCGCTATGCTGGTGCTGGGC	Gene deletion
R-del50proT1	aattGCCCAGCACCAGCATAGCGCTGGTCACAATGATGTTGAACACGTTAACCACTCTTTAGCCTTCTTGATGGCGTTTTTAACGCGCCCCAGCGCTTGCAAGC	Gene deletion
Del1k40-F	agctGTACTCTGCTGCCTATCAAGTTCAACAAAGAGTCCGGCAACAACGATATTTGAGGTGGGGTATGCACGGAAAGAATCCTG	Gene deletion
Del1k40-R	aattCAGGATTCTTTCCGTGCATACCCCACCTCAAATATCGTTGTTGCCGGACTCTTTGTTGAACTTGATAGGCAGCAGAGTAC	Gene deletion
Del2k40-F	agctAATCATTAACTTTGAGGTGCAATATGAAATACAGAGATAATCGTGCCGGTCGAGCGTAGAATCCTCCGAGTTCGTAAGCA	Gene deletion
Del2k40-R	aattTGCTTACGAACTCGGAGGATTCTACGCTCGACCGGCACGATTATCTCTGTATTTCATATTGCACCTCAAAGTTAATGATT	Gene deletion
Delgp8-F	agctATTACCAGAAGGTGCAAACGCACTGCCAATGAAATGTTCGATGTTCTTGAATCCGCACGGGATTGATATGCAGCTGCTCT	Gene deletion
Delgp8-R	aattAGAGCAGCTGCATATCAATCCCGTGCGGATTCAAGAACATCGAACATTTCATTGGCAGTGCGTTTGCACCTTCTGGTAAT	Gene deletion
Delgp51-F	agctTCGTAAACTGTTCAGGAAGAAGGGGCAGCAGGATGATTAAATGAGTTTAAGGACTAAGGTTATTGCGGCCCTCACGGGGGCC	Gene deletion
Delgp51-R	aattGGCCCCCGTGAGGGCCGCAATAACCTTAGTCCTTAAACTCATTTAATCATCCTGCTGCCCCTTCTTCCTGAACAGTTTACGA	Gene deletion
pro80GG-F	agctGGTGAGTACGTATAGTAGCAAGCCTATAGCGTCCTACGGCGCTATGTGAATGCAACTAGCGTATGAGGTTTCATATGA	Small frameshift deletion
pro80GG-R	aattTCATATGAAACCTCATACGCTAGTTGCATTCACATAGCGCCGTAGGACGCTATAGGCTTGCTACTATACGTACTCACC	Small frameshift deletion
pri80GG-F	agctTGGGGCAGAGGGCGCGCTTTCGGTGCTGCGGCAGGGACGGCCGGGGTGCTCTACAATAACCCGGATGCCTGGGAATAT	Small frameshift deletion
pri80GG-R	aattATATTCCCAGGCATCCGGGTTATTGTAGAGCACCCCGGCCGTCCCTGCCGCAGCACCGAAAGCGCGCCCTCTGCCCCA	Small frameshift deletion
ICP80GG-F	agctCGCTTATAGCGCCGACGTAAACGGTCGCGCTGTTCCGGTGGGATGCAGAGATTGCTCCGCTGGTTGATGCTTACCGTCGC	Small frameshift deletion
ICP80GG-R	aattGCGACGGTAAGCATCAACCAGCGGAGCAATCTCTGCATCCCACCGGAACAGCGCGACCGTTTACGTCGGCGCTATAAGCG	Small frameshift deletion
H280GG-F	agctAAGAAGGGGCAGCAGGATGATTAAAGTAGGGGACATGGTTATCTGACCTCGCTACCCGGGCAGGTGCAGCAGTTACCGGC	Small frameshift deletion
H280GG-R	aattGCCGGTAACTGCTGCACCTGCCCGGGTAGCGAGGTCAGATAACCATGTCCCCTACTTTAATCATCCTGCTGCCCCTTCTT	Small frameshift deletion
gp280GG-F	agctATCATGACTAATTCAACTGGTAAAGTATTCAAGTTAACTGGCTGGCAGCATTCGTAAAGCTCTGGGCGATGTAGTGGAAG	Small frameshift deletion
gp280GG-R	aattCTTCCACTACATCGCCCAGAGCTTTACGAATGCTGCCAGCCAGTTAACTTGAATACTTTACCAGTTGAATTAGTCATGAT	Small frameshift deletion
In40RFP-F	CCGgaattcGCTGGGGCGCGTTAAAAACGCCATCAAGAAGGCTAAAGAGGACTGGAAAGCGGGCAGTGA	Gene swap
In40RFP-R	GCtctagaAGCATAGCGCTGGTCACAATGATGTTGAACACGTTAACCATTAAGCACCGGTGGAGTGACG	Gene swap
In50RFP-F	CCGgaattcGCTTGCAAGCGCTGGGGCGCGTTAAAAACGCCATCAAGAAGGCTAAAGAGGACTGGAAAGCGGGCAGTGA	Gene swap
In50RFP-R	GctctagaGCCCAGCACCAGCATAGCGCTGGTCACAATGATGTTGAACACGTTAACCATTAAGCACCGGTGGAGTGACG	Gene swap
In60RFP-F	CCGgaattcGATGCAGTGCGCTTGCAAGCGCTGGGGCGCGTTAAAAACGCCATCAAGAAGGCTAAAGAGGACTGGAAAGCGGGCAGTGA	Gene swap
In60RFP-R	GCtctagaTTACGTCGTTGCCCAGCACCAGCATAGCGCTGGTCACAATGATGTTGAACACGTTAACCATTAAGCACCGGTGGAGTGACG	Gene swap
F-sgRNA	GAGTCAGCTAGGAGGTGACTAAC	Spacer cloning check
R-sgRNA	TGCCATTGGGATATATCAACGG	Spacer cloning check
Test-Pro-F	GCTGGGCATCGAACTGGAC	Phage mutant check
Test-Pro-R	GCATCAGCAGCGTCTCAGG	Phage mutant check
F-priNT3	CCGACCGCACCTATACGGTC	Phage mutant check
R-priNT3	ACGACACCGCATGAACGGAC	Phage mutant check
F-icpNT1	GCCCAAGCTATGCTGGACGG	Phage mutant check
R-icpNT1	TGCTGCCGGATTCTCGTACG	Phage mutant check
F2-Ba7	CGGCTAACCTGGGCACTACC	Phage mutant check
R2-Ba7	CGGAAGTCTACGCCGTTGCc	Phage mutant check
Holin-F	GTTACAGTCTTATCCTCGCCGG	Phage mutant check
Holin-R	GTCTTTGTAGGCGGTAAGGCTC	Phage mutant check
F-Ba7	GTTCCACGGTAGTGCAGTGGG	Phage mutant check
R-gp8	GTGCATACCACTACCCAGCGC	Phage mutant check

^aLowercase letters represent sticky ends or restriction enzyme sites.

Evaluation of the activity of sgRNA.

As mentioned earlier, a total of 87 sgRNAs targeting 11 different genes and a promoter of phiKpS2 were designed. All 87 pcas9-sgRNA plasmids were amplified in E. coli DH5α and then transformed into K. pneumoniae S2 by electroporation to form CRISPR-based phage-resistant strains. The activity of sgRNA was evaluated by determining the efficiency of restriction of phage infection of these CRISPR-based phage-resistant strains. The efficiency of plaquing (EOP) was used to determine the resistance level, which was defined by dividing the phage titer obtained from the resistant strain by the phage titer obtained from the wild-type strain.

The softwares sgRNA_Scorer_2.0 (29) and ge-CRISPR (30) were used to predict the activity of sgRNA, after which the observed activity was compared with the predicted activity to assess the ability of publicly available design tools to predict the efficiency of gRNAs. Both the Pearson and Spearman correlations were evaluated.

Isolation and characterization of recombinant phages.

All of the recombinant phages were isolated by the double-layer plate method as described before (44). Briefly, 0.1 ml of properly diluted phage suspensions and 0.9 ml of bacterial cultures with an optical density greater than 1.8 were added to 8 ml of warm soft agar (0.6%), mixed, and poured on petri dishes to form plaques. The plate was incubated at 37°C for about 12 to 16 h. Well-isolated individual plaques were picked up and transferred to PCR tubes containing 150 ml of sterile phage buffer (50 mM Tris-HCl, pH 7.5, 100 mM NaCl, 8 mM MgSO₄). If necessary, the procedure was repeated three times to further purify the recombination phages. In the present study, the total input phage mainly depends on the activity of sgRNA. Strong sgRNAs require more input phage, while weak sgRNAs require less input phage. Briefly, the total input phage requires adjustments according to the various conditions to obtain well-isolated plaques.

The isolated phage could be used as the template to amplify the phage protospacer region by PCR. The PCR products were sequenced using the ABI 3730xl DNA analyzer at Sangon Biotech Co., Ltd. (Shanghai, China). Sequences were aligned with the wild-type phiKpS2 sequence and analyzed using Geneious 11 (Biomatters Ltd., Auckland, New Zealand).

One-step growth curve of phage phiKpS2 was carried out to evaluate the essentiality of some phiKpS2 genes. The indicator strain was incubated for 7 h at 37°C in LB medium. Cells were harvested and added to a fresh seed medium and then were mixed with phage suspensions to achieve a multiplicity of infection of 1. Phages were allowed to adsorb for 5 min at 37°C, and then the mixture was incubated at 37°C with agitation at 200 rpm. Samples were taken at 15-min time intervals and centrifuged (12,000 rpm for 5 min) for titration. Phage titers were determined by the double-layer plate method.

Statistical analysis.

Statistical analyses were performed using OriginPro 9.1 (OriginLab, Inc.). Values are represented as means ± standard deviations from independent experiments (at least 3 replicates). Statistical significance was evaluated using Student's t test with P values of <0.05.

Accession number(s).

The whole-genome sequence of phiKpS2 is available in GenBank under accession number KX587949. The sequence for pcas9-sgRNA has been deposited in GenBank under accession number MH319949. Additional accession numbers are listed in Table 1.