An efficient CRISPR-Cas12a tool for iterative genome editing, streamlined minimization, and payload engineering of Pseudomonas phage S1

Abstract

The ability to precisely engineer phage genomes is critical for advancing phage-based biotechnology and therapeutic development. Here, we present a high-efficiency, scarless CRISPR-Cas12a genome editing method for a virulent Pseudomonas aeruginosa phage vB_PaeM_SCUT-S1 (S1). Using a two-plasmid system, we achieved near-complete efficiency in gene deletion, point mutation, gene insertion, and replacement. Iterative deletions enabled the identification of 27 non-essential, 16 quasi-essential, and seven essential genes, and culminated in the largest genome reduction (13.9 kb) reported for a P. aeruginosa phage. The minimized phage mutant S1_200L retained infectivity and supported integration of large exogenous gene cassettes (e.g., lacZ, lys009). This system offers a versatile and high-throughput approach for phage genome engineering and rational design of functional phages for synthetic biology and antimicrobial applications.

1. Introduction

Pseudomonas aeruginosa, one of the six ESKAPE pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) [1], is a prevalent multi-drug-resistant (MDR) opportunistic pathogen and poses a serious public health threat globally [2,3]. Due to its high mortality rates among immunocompromised individuals and frequent outbreaks in healthcare settings, P. aeruginosa has been designated a “high-priority” pathogen by the World Health Organization (WHO) [4]. Phage therapy offers a promising alternative to antibiotics for treating MDR bacterial infections, yet natural phages often encode numerous uncharacterized genes, complicating safety evaluations and limiting their clinical utility [[5], [6], [7]]. A thorough understanding of phage genomes and the development of precise genome engineering tools are essential for the advancement of safe, effective phage therapeutics.

To date, two primary strategies have been employed for P. aeruginosa phage genome editing: homologous recombination and CRISPR-Cas-based systems. While homologous recombination is a well-established method, it suffers from low editing efficiencies in P. aeruginosa phage, with low recombination frequencies ranging from 10⁻⁹ to 10⁻⁸ [8]. More recently, CRISPR-Cas systems have emerged as powerful tools for phage genome editing, including in virulent P. aeruginosa phages (Table S1) [9,10]. For instance, the mRNA-targeting CRISPR-Cas13a system (type VI-A), combined with homologous recombination, has been successfully used to edit the P. aeruginosa Podophage PaMx41, which is resistant to DNA-targeting CRISPR-Cas systems [9]. This approach however leaves behind a selectable anti-CRISPR marker (acrVIA1). Additionally, the CRISPR-Cas3 system (type I–C) has been employed to edit virulent Pseudomonas phages [10], but its application is constrained by a strict 5′-TTC-3′ protospacer adjacent motif (PAM).

In our previous study, we developed a scarless, single-step genome editing method based on CRISPR-Cas12a (type V-A), which utilizes a relaxed PAM sequence (5′-YTV-3″, where Y represents C or T, and V represents A, C or G) [[11], [12], [13]]. This system achieved efficient gene deletion, insertion, and replacement in P. aeruginosa, including the deletion of large DNA fragments up to 15 kb and the serial deletion of duplicate gene clusters [11].

In this study, we extended the application of the CRISPR-Cas12a system to the genome editing of virulent P. aeruginosa phages, using phage vB_PaeM_SCUT-S1 (hereafter referred to as phage S1) as a model, which is a headful-packaging member of the Myoviridae family that was previously isolated and characterized in our laboratory [2]. Importantly, headful-packaging Pseudomonas phages account for approximately one-third of all sequenced Pseudomonas phages in GenBank [14], underscoring the broader relevance of studying phage S1.

We demonstrate that this system enabled efficient, one-step gene deletion, point mutation, gene insertion, and replacement without the need for selection markers. Furthermore, we applied this tool to systematically investigate the importance of 48 hypothetical genes through iterative deletion. Our study establishes a versatile and highly efficient genome editing system for P. aeruginosa phages, facilitating deeper understanding of phage genomic structure and gene function, progressive genome minimization, and advancing the potential of phage for therapeutic applications.

2. Results

2.1. A two-plasmid CRISPR-Cas12a system enables efficient genome editing in P. aeruginosa phage S1

This work built on the two-plasmid CRISPR-Cas12a system (pCas12a-λRed/pCrRNA series) that we previously developed [11]. The pCas12a-λRed plasmid carries the FnCas12a gene under the control of the native Cas9 promoter P_Cas, along with the λRed recombinase genes (gam, bet, and exo) under the l-arabinose-inducible promoter P_araB [11]. Since plasmid curing of pCrRNA from the host was no longer required for phage genome editing, we utilized the plasmid pCrRNA-oprM-D [10] as the template, which lacks the sacB gene used in the removal of pCrRNA. pCrRNA-oprM-D expresses crRNA under the strong constitutive promoter P_J23119. To evaluate FnCas12a-mediated DNA cleavage efficiency, we constructed a series of crRNA-expressing plasmids, designated as pCrRNA-X (where X denotes the specific crRNA spacer; see Materials and Methods 4.2). For genome editing applications, we incorporated the donor DNAs into the corresponding pCrRNA-X plasmids to yield editing plasmids, termed pCrRNA-X-Y (where Y indicates the donor DNA; see Materials and Methods 4.2). These plasmids were separately electroporated into P. aeruginosa PAO1 containing pCas12a-λRed to generate the phage propagation strains for spot tests.

We first assessed the cleavage efficiency of FnCas12a targeting the non-essential gene orf073 [7]. The predicted protein product of orf073 is 324 amino acids in length, and its adjacent genes, orf072 (261 amino acids) and orf074 (152 amino acids), both encode hypothetical proteins [2]. Six pCrRNA-X plasmids (pCrRNA-orf073-1 to pCrRNA-orf073-6) were used, each with a high-scoring crRNA spacer (orf073-1 to orf073-6) designed with the Cas-Designer web tool (http://www.rgenome.net/cas-designer/, [15,16]) (Table S2). Cleavage activity was quantified by measuring the efficiency of plaquing (EOP) through spot tests performed in the absence of l-arabinose, as λRed recombinase was not required for this step [17]. EOP was calculated as the ratio of phage titers on P. aeruginosa PAO1 strains carrying pCas12a-λRed and a target-specific pCrRNA-X plasmid to those on strains carrying pCas12a-λRed and the negative control plasmid pCrRNA-NC, which targets neither the phage S1 genome nor the P. aeruginosa PAO1 genome. As shown in Fig. 1A, FnCas12a exhibited variable cleavage efficiency against the S1 genome, with EOP values ranging from 10⁻⁴ to 0.69 (Table S2) depending on the crRNA spacers. The strongest inhibition was observed with spacers orf073-2 and orf073-6. However, orf073-6 resulted in the appearance of isolated colonies on plate (Fig. S1), suggesting potential cytotoxicity likely due to off-target effects. We therefore selected orf073-2 for subsequent genome editing experiments.

Fig. 1.

The CRISPR-Cas12a system enables efficient genome editing of P. aeruginosa phage S1. (A) Spot test results for the cleavage efficiency of six distinct crRNA spacers targeting orf073. Red vertical lines indicate the relative positions of crRNA spacers; the optimal spacer selected for downstream editing is marked with a red asterisk. (B) Top: schematic overview of the two-plasmid CRISPR-Cas12a system for gene deletion. The system comprises a plasmid encoding FnCas12a and λRed recombinase (pCas12a-λRed) and a crRNA-expressing plasmid (pCrRNA-orf073-2-D). Homologous arms (HA1 and HA2) flank the target gene orf073 to facilitate deletion. Red arrows indicate the primers used in plaque PCR. Bottom: Plaque PCR results for the deletion of orf073 using primers YM19062/YM19063. Expected length: 2,369 bp for wild-type and 1,394 bp for deletion mutant. M, Takara DL5000 DNA ladder; lanes 1–24, randomly selected plaques. (C) Schematic diagrams of homologous arms (left), corresponding editing plasmids (middle), and representative plaque PCR results (right) for different editing tasks. HA3 and HA4: for point mutations; HA5 and HA6: for gene insertion; HA1 and HA2: for entire gene replacement; HA1 and HA7: for partial gene replacement. Red arrows indicate the primers used in plaque PCR.

Next, we examined the deletion efficiency of the CRISPR-Cas12a system by targeting the same gene, orf073. We constructed pCrRNA-orf073-2-D, which harbors 500 bp homologous arms (HA1 and HA2) (Fig. 1B). Spot tests and PCR verification showed 100 % deletion efficiency (6/6 plaques), in the absence or presence of l-arabinose induction (Fig. 1B, Table 1). In contrast, spot tests with the control plasmids pCrRNA-NC or pCrRNA-orf073-2 (lacking homologous arms) yielded only wild-type plaques. These results suggest that the combination of FnCas12a, a specific crRNA, and homologous arms is sufficient for the deletion, and that λRed recombinase is dispensable for this task.

Table 1.

Summary of the genome editing in P. aeruginosa phage S1 using CRISPR-Cas12a system.

Target gene	Efficiency				Efficiency after Purification
	Deletion	Point mutations	Insertion	Replacement
orf073	100 % (6/6)	100 % (6/6)	100 % (6/6)a	100 % (6/6)	ND
	30.0 % (9/30, pCrRNA-orf073-2-D50)a, 6.7 % (2/30, pCrRNA-orf073-3-D50)a	ND	ND	ND	ND
orf053	100 % (6/6)a	ND	ND	ND	100 % (6/6)
orf026	ND	ND	100 % (6/6)a	ND	100 % (6/6)

ND, not determined. The lengths of the homologous arms were 500 bp, unless otherwise specified.

^aPhage populations consisting of both wild-type and recombinant phages.

We further tested more complex editing tasks, including point mutation (M), gene insertion (I), and replacement (R). Accordingly, three editing plasmids were constructed utilizing 500 bp homologous arms and the same crRNA spacer (orf073-2): pCrRNA-orf073-2-M-T9C/T10C, which was designed to introduce two-point mutations at nucleotide positions 9 and 10 of orf073 (within the orf073-2 PAM sequence) using homologous arms HA3 and HA4; pCrRNA-orf073-2-I-rfp, which was designed to insert the rfp cassette (786 bp) within the orf073-2 protospacer using homologous arms HA5 and HA6; and pCrRNA-orf073-2-R-rfp, which was designed to replace the entire orf073 with the rfp cassette using homologous arms HA1 and HA2 (Fig. 1C). Spot tests and PCR verification demonstrated point mutations achieved 100 % editing efficiency (6/6 plaques; Fig. 1C and Fig. S2A, Table 1). Gene insertion reached near-complete efficiency in the presence of λRed, with all six plaques exhibiting only minor contamination by wild-type phages. In the absence of λRed, two out of six plaques showed significant contamination of wild-type phages, while the remaining four showed minor contamination (Fig. 1C and Fig. S2A, Table 1). These results indicate that λRed recombinase markedly improves the efficiency of more challenging editing tasks, such as gene insertion.

However, attempts to use the plasmid pCrRNA-orf073-2-R-rfp for gene replacement yielded no positive transformants in PAO1. Further analysis revealed the consistent loss of an approximately 1–2 kb DNA fragment encompassing the crRNA, the rfp cassette, and both homologous arms upon electroporation (Fig. S2B). Notably, HA2 contained the entire orf074. Additional investigation found a predicted promoter (5′-AGGTCGTCACTCTACCGGCGGTATGGACGAACTGTATAAATAATAGTGAC-3′) spanning the 3′ end of the rfp cassette and the adjacent spacer region 18 bp upstream of orf074 [18], along with a potential ribosome binding site (5′-AGGAGTT-3′) 6 bp upstream of orf074 [19]. These features suggest that pCrRNA-orf073-2-R-rfp may have inadvertently driven unintended orf074 expression, which could have caused the observed cytotoxicity. To address this issue, we employed two strategies: 1) incorporation of a transcriptional terminator downstream of the rfp cassette, resulting in the plasmid pCrRNA-orf073-2-R-rfp-rrnB T1; and 2) exclusion of the entire orf074 coding sequence by shifting HA2 185 bp upstream to generate HA7, which retained the 3′ end of orf073 and the 5′ end of orf074, yielding the plasmid pCrRNA-orf073-2-R-rfp-2. Gene replacement efficiencies reached 100 % (6/6 plaques) for both modified plasmids, independent of λRed expression (Fig. 1C and Fig. S2C, Table 1), indicating that λRed recombinase is again dispensable for this editing task, at least when a high-activity crRNA is used.

Taken together, these results establish CRISPR-Cas12a as a robust and versatile tool for phage S1 genome editing, supporting efficient gene deletion, point mutation, gene insertion, and replacement using 500 bp homologous arms. Furthermore, λRed recombinase enhanced the efficiency of gene insertion. For simplicity, λRed was consistently induced with l-arabinose in subsequent editing tasks.

2.2. Investigating the role of holin in phage S1 using CRISPR-Cas12a

Holin is a membrane protein that facilitates host cell lysis by forming micron-scale holes in the inner membrane, thereby enabling the release of endolysin into the periplasm to degrade the peptidoglycan [20,21]. This prompted us to investigate the role of the holin gene (orf053) [2] in phage S1 using our CRISPR-Cas12a system.

We selected a high-activity crRNA targeting orf053 (pCrRNA-orf053-3, EOP = 10⁻⁵; Fig. 2A, Table S2) and constructed the corresponding deletion plasmid pCrRNA-orf053-3-D. Spot tests confirmed that the deletion mutants remained viable and retained plaque-forming ability. PCR verification showed nearly 100 % deletion efficiency, but all six plaques exhibiting variant amounts of contamination by wild-type phages (Fig. S3A). A phage mutant isolate was further purified through three sequential rounds of infection on PAO1 carrying pCas12a-λRed and pCrRNA-orf053-3, yielding a stable holin-deficient mutant, S1Δorf053 (Fig. S3B). Compared to the wild-type phage S1, the S1Δorf053 mutant formed visibly smaller plaques (Fig. S4). One-step growth curve analysis revealed that S1Δorf053 exhibited an extended latent period and slower rise period, along with a reduced burst size of 26 pfu/infected cell, compared to 113 pfu/infected cell in wild-type (Fig. 2B).

Fig. 2.

CRISPR–Cas12a-mediated deletion of the holin (orf053) gene in phage S1. (A) Spot test results for evaluating the activity of distinct crRNA spacers targeting the holin (orf053) gene. Red vertical lines indicate the relative positions of crRNA spacers; the optimal crRNA spacer used for gene deletion is marked with a red asterisk. (B) One-step growth curve of wild-type S1 and S1Δorf053 mutant. (C) Growth inhibition of PAO1 by S1Δorf053 at different multiplicities of infection (MOIs). Data represent mean values ± standard deviation (SD) from three biological replicates.

We further assessed the effects of S1Δorf053 against PAO1 planktonic cultures across a range of multiplicities of infection (MOIs) from 0.01 to 100 (Fig. 2C). S1Δorf053 efficiently inhibited PAO1 growth during the early phase of infection (0–12 h), with inhibition rates increasing from 68 % to 97 % as MOI rose from 0.01 to 100 at 6 h post-infection. However, growth inhibition declined during the later phase (12–23.5 h), likely due to the emergence of phage-insensitive or resistant strains [2]. Interestingly, higher MOIs were associated with a more pronounced decline in inhibition efficiency over time. By 23.5 h, inhibition rates had dropped to 24 %–30 % at MOIs of 0.1–100, whereas a low MOI of 0.01 maintained a relatively higher inhibition rate of 43 %. Thus, an MOI of 0.01 was selected for subsequent assays.

Together, these findings suggest that the holin gene, or the orf053 gene, is not essential for S1 viability under laboratory conditions, consistent with observations in Coliphage T7 [22] and Klebsiella phage phiKpS2 [23]. It is speculated that phage S1 may possess functionally redundancy genes that compensate for the loss of orf053 [23]. Nevertheless, orf053 still appears to play a significant role in host cell lysis efficiency and phage propagation.

2.3. Studying hypothetical genes and constructing a genome-minimized phage S1 using the CRISPR-Cas12a system

We next employed the CRISPR-Cas12a system to investigate the functional importance of hypothetical genes in phage S1, and to generate phages with progressively reduced genomes through iterative gene deletion, as outlined in Fig. 3A. Among the 94 annotated genes in S1, 55 (58.5 %) are classified as hypothetical genes [2]. We excluded seven genes encoding small proteins (≤189 bp; orf002, orf003, orf013, orf023, orf076, orf092, and orf093; totaling 1,116 bp) due to their limited contribution to the overall genome size. Of the remaining 48 hypothetical genes, six (orf010, orf014, orf072, orf073, orf081, and orf086) were previously identified as non-essential and four (orf043, orf055, orf075, and orf078) as essential [2,7]. This left 38 hypothetical genes (orf001, orf004, orf006-orf009, orf011, orf012, orf015-orf019, orf022, orf027, orf031, orf040, orf050, orf054, orf061, orf063, orf064, orf066-orf068, orf071, orf074, orf079, orf080, orf082-orf085, and orf087-orf091) as candidate targets for genome minimization in this study.

Fig. 3.

Iterative gene deletion in phage S1 using the CRISPR–Cas12a system. (A) Schematic workflow for iterative gene deletion approach. (B) Metagenomic sequencing analysis showing relative sequencing abundance (gene depth/genome depth) of deletable genes during serial phage transfers. T: transfer; n: number of transfers. ∗: undesigned genes. Non-essential gene: gene absent in the isolated mutant phage genomes. Quasi-essential gene: gene in the deletable gene sets but retained in the isolated mutant phage genomes. L: large fragment deletion. (C) Growth inhibition of PAO1 by phage S1 and S1 mutants at MOI = 0.01. Data represent mean values ± standard deviation (SD) from three biological replicates. L: large plaque; S: small plaque; wt: wild-type. (D) Genome map of phage S1, with regions deleted in S1_200L indicated by red rectangles.

To systematically interrogate these 38 hypothetical genes, we designed a total of 103 high-scoring crRNA spacers (Table S2), with 2–6 spacers per gene, approximating one spacer per 100–167 bp to ensure complete gene coverage [24]. We also designed corresponding donor DNA templates with shorter HAs of 50 bp to facilitate high-throughput library construction. As shown in Table 1 and Fig. S5, we confirmed that the CRISPR-Cas12a system supports efficient gene deletion using 50 bp HAs, as demonstrated by the successful deletion of orf073 with either a high-activity crRNA (pCrRNA-orf073-2, EOP = 10⁻⁴) or a low-activity crRNA (pCrRNA-orf073-3, EOP = 0.18).

To construct a plasmid library for iterative gene deletion, we synthesized and pooled 103 crRNA fragments, each comprising a crRNA spacer and flanking 50 bp homologous arms (HAs), and assembled them into pCrRNA-X-D50 plasmids using Gibson assembly (see Materials and Methods 4.2). The resulting plasmid library was electroporated into P. aeruginosa PAO1 cells carrying the pCas12a-λRed plasmid, generating a total of 1.06 × 10⁷ colonies, with an average of 1.03 × 10⁵ colonies per individual plasmid. Next-generation sequencing (NGS) confirmed that all 103 crRNA constructs were represented, achieving 100 % coverage at a sequencing depth of 2.4 × 10⁶, indicating the completeness of this library.

To identify dispensable genes, we infected the plasmid library-containing cells with wild-type phage S1 and conducted 260 serial transfers, corresponding to over 2,400 phage generations. Metagenomic sequencing of phage populations collected at the 50th, 100th, 150th, 200th, and 260th transfers (sequencing depth: 1.4 × 10⁴ to 2.1 × 10⁴) revealed 43 genes as deletable (Fig. 3B–Table S3). Among these, 31 corresponded to the 38 hypothetical genes we initially targeted. An additional 11 non-targeted genes were lost (or deletable) likely as part of large genomic fragment deletions, while one non-targeted gene (orf069, a putative structural protein gene [2]) showed decreased relative sequencing abundance (gene depth/genome depth) for unknown reasons. Interestingly, the deletable 11 non-targeted genes included five of the seven small proteins that were not initially targeted (orf002, orf003, orf013, orf092, and orf093).

To identify phage S1 mutants with the smallest genome, we selected three large plaques (S1_150L, S1_200L, and S1_260L) and three small plaques (S1_150S, S1_200S, and S1_260S) from the 150th, 200th, and 260th transfers, respectively. Lytic kinetics revealed that large-scale genome reduction impacted phage infectivity or propagation; however, plaque size did not correlate with lytic activity (Fig. 3C). Compared to wild-type S1, all mutants exhibited slightly lower bacterial inhibition during the early phase (1–3 h), followed by enhanced suppression during the later phase (5–18 h), with inhibition rates ranging from 82 % to 96 % at 6 h post-infection. Ultimately, inhibition levels became comparable after 18 h, with rates declining to 31 %–35 % by 23.5 h post-infection.

Genome sequencing of these plaques revealed size reductions ranging from 16.1 % to 21.1 % (10.6 kb–13.9 kb) compared to the wild-type genome (66,086 bp) (Table S4). While a few deletions occurred interspersed throughout the central genome region, large contiguous deletions were predominantly localized to the genome ends (Fig. S6). Notably, two large genomic DNA segments, orf001-orf019 (8.9 kb) and orf081-orf094 (5.1 kb), were almost entirely deletable, except for orf005, which encodes a putative terminase large subunit involved in DNA packaging [2]. In addition, we identified 99 unique point mutations across all mutants, with 20.2 % occurring within the protospacer or PAM regions of quasi-essential genes (genes within the deletable gene sets but retained in the isolated mutant phage genomes) and essential genes (targeted hypothetical genes not detected in the deletable gene sets) (Fig. S6), likely resulting from CRISPR-escape mutations (CEMs) [24].

The most extensively reduced genome was observed in S1_200L with 13.9 kb deleted (Fig. S6, Table S4). This variant contained four single-gene deletions (orf004, orf019, orf022, and orf063), four large multi-gene fragment deletions (orf001-orf003, orf006-orf017, orf073-orf074, and orf081-orf094), and one gene disruption (orf069, 37/894 bp deletion), collectively accounting for 36 of the 43 identified deletable genes (Fig. 3D).

To evaluate the genome's capacity for exogenous functional gene insertions, we engineered S1_200L to carry either a β-galactosidase cassette (lacZ, 3,149 bp) [11] or a lysin cassette (lys009, 821 bp) [25]. Each cassette was inserted downstream of orf026, a putative major structural gene, to avoid disruption of other essential functions (Fig. 4A and Fig. S7A). Transmission electron microscopy (TEM) revealed no detectable differences in capsid diameter or tail length among wild-type S1, S1_200L, and insertion mutants (S1_200L_lacZ and S1_200L_lys009), indicating that the 13.9 kb deletion and 3.1 kb insertion did not compromise phage morphology (Fig. 4B and Fig. S7B). It is worth noting that the two distinct morphologies observed in Fig. 4B correspond to the extended and contracted states of the tail structure of phage S1, a member of the Myoviridae family characterized by long, contractile tails [26].

Fig. 4.

Genome editing and functional assessment of the minimized phage S1_200L. (A) Schematic diagram of S1_200L and recombinant phage mutants engineered to express either lacZ or lys009. Red triangle denotes the crRNA spacer. (B) Transmission electron microscopy images of phages S1, S1_200L, S1_200L _lacZ, and S1_200L _lys009 negatively stained by 2 % phosphotungstic acid. (C) Growth inhibition of PAO1 by phage S1 and its engineered mutants at MOI = 0.01. (D) One-step growth curve of S1_200L and its mutants. Data represent mean values ± standard deviation (SD) from three biological replicates.

Lytic kinetics revealed that S1_200L_lys009 retained similar inhibitory activity to S1_200L (Fig. 4C). However, S1_200L_lacZ displayed reduced inhibition during the early phase, with an inhibition rate of 41 % at 6 h post-infection, followed by enhanced inhibition in late stage, reaching 66 % at 23.5 h (Fig. 4C). Furthermore, one-step growth curve analysis revealed that S1_200L_lacZ had a comparable latent period (45 min) but a reduced burst sizes (38 pfu/infected cell) compared to S1_200L (110 pfu/infected cell), while S1_200L_lys009 had a shorter latent period (40 min) and a moderately reduced burst size (92 pfu/infected cell) (Fig. 4D).

Taken together, these results highlight that our CRISPR-Cas12a-based iterative genome engineering strategy enables extensive genome minimization and payload insertion without compromising phage structural integrity or replication capacity, providing a versatile method for engineering phages with customized functionalities.

3. Discussion

In this study, we developed an efficient, scarless genome editing method for P. aeruginosa phage S1 using a two-plasmid CRISPR-Cas12a system, including gene deletion, point mutation, gene insertion, and gene replacement, with near-complete efficiency in a single step. Compared to our previously developed CRISPR RNA-guided transposon insertion system (<1 % efficiency, improved to 8.3 % with the selection marker acrVA1) [7], the CRISPR-Cas12a system markedly enhanced editing performance.

Consistent with previous studies in Escherichia coli phage T4 [27,28] and P. aeruginosa phages PJNP013, PJNP029, and PJNP053 [29], CRISPR-Cas12a outperformed CRISPR-Cas9 in phage S1 genome editing (Table 1, Table S5). It is also noteworthy that CRISPR-Cas12a can achieve 6.7 %–30.0 % editing efficiency using short (50 bp) homologous arms, compared to the near-complete efficiency observed with 500 bp homologous arms. The ability to perform edits with short homologous arms is particularly advantageous for high-throughput and iterative genome engineering workflows [23,24]. Interestingly, although the S1 genome has a moderate GC content (55.4 %) that should theoretically favor SpCas9 due to its preference for G-rich PAM sequences [30,31], FnCas12a demonstrated greater targeting flexibility, recognizing ∼16 PAM sites per 100 bp compared to ∼13 for SpCas9. Using this system, we successfully performed high-throughput, iterative gene deletion, identified 27 non-essential, 16 quasi-essential, and seven essential genes (Fig. 3B and D–Table S3), and achieving a cumulative 13.9 kb genome deletion, the largest deletion reported in a P. aeruginosa phage to date [14,29,32]. Moreover, the minimized genome phage mutant, S1_200L, supported the insertion of a 3.1 kb exogenous functional gene cassette, representing the largest known phage genome insertion using CRISPR-Cas tools [9,17,24,29].

As a comparison, a CRISPR-Cas9-based iterative genome reduction strategy was previously applied to E. coli phages T7 and T4 and Salmonella enterica phages seszw and selz [24]. However, when applied to P. aeruginosa phage S1, CRISPR-Cas9 produced mixed populations of wild-type and recombinant phages for orf073 deletion, even when using 500 bp homologous arms, and failed to select high-activity sgRNAs (Table S5). In contrast, CRISPR-Cas12a allowed efficient selection of high-activity crRNAs and consistently generated pure orf073 deletion phages using 500 bp homologous arms (Table 1 and Table S2), highlighting its advantages for minimizing phage S1. Furthermore, during the preparation of our manuscript, CRISPR-Cas12a was also used to minimize the genome of P. aeruginosa phage PJNP053 [29]. In that study, 33 hypothetical genes (33.3 % of the total) were grouped into four regions (A–D) and deleted stepwise, with only the combined deletions of regions AB or AC tolerated, resulting in a maximum genome reduction of 5,215 bp (7.1 %). While informative, this regional deletion strategy was not designed to assess individual gene function. By contrast, our study established a CRISPR-Cas12a-based iterative deletion strategy that enables functional analysis at the single-gene level.

Our iterative deletion strategy also provided insights into phage genome organization. Essentiality mapping of the S1 genome revealed a non-uniform distribution of genes, with putatively non-essential genes clustered at the genome ends [2]. This pattern is not unique to S1 and has been observed in other headful-packaging phages, such as E. coli phages P1 and T4, as well as the direct terminal repeat (DTR) phage T7 [6,33,34]. By contrast, E. coli phage λ exhibits a distinct organization, with essential genes preferentially located at both genome ends [6]. These observations suggest that different phages adopt diverse genome organizational strategies, likely reflecting evolutionary and functional constraints.

To explore gene function, we performed BLASTP analysis on 48 hypothetical genes, identifying seven non-essential, four quasi-essential, and six essential genes likely involved in DNA replication, recombination and repair, nucleotide metabolism, or phage structure and packaging (Table S6). The remaining 31 genes (18 non-essential, 12 quasi-essential, and one essential) remain unannotated. Promoter and terminator predictions using PromoterHunter [18] and ARNold [35] indicated that only eight genes possess both regulatory elements, while most only have predicted promoters, implying dependance on downstream terminators. Notably, orf090 lacks both, suggesting possible co-transcribed with adjacent genes. These predictions require future experimental validation.

We also uncovered potential functional redundancy. The orf053 gene (holin) was found to be dispensable, likely compensated by the essential gene orf054, which shares 99 % identity with the putative holin in Pseudomonas phage phiLCL12 [36]. Similarly, orf017 and orf018 could be deletable individually but not simultaneously (Fig. S6), suggesting redundant functions. These two genes share an 84 % amino acid identity.

Altogether, the CRISPR-Cas12a system provides a robust and versatile toolkit for phage genome engineering. Its high efficiency and scarless editing capability make it ideally suited for both basic research and biotechnological applications. In particular, by expanding the genetic payload capacity of the phages for PAO1, the toolkit paves the way for customized phage applications, including engineered phages targeting MDR pathogens such as P. aeruginosa [37,38].

4. Materials and Methods

4.1. Bacterial strains, phages and media

E. coli DH5α was used for plasmid construction, E. coli S17-1 served as the conjugation donor strain, and E. coli XL1-Blue was used for crRNA library construction. P. aeruginosa PAO1 was used as the host strain. P. aeruginosa phage vB_PaeM_SCUT-S1 (S1, GenBank MK340760) [2] was used as the model organism to edit (Table S7). All the strains were cultured in lysogeny broth (LB) medium at 37 °C. The medium was supplemented with tetracycline or gentamicin (at final concentration of 10 μg/mL for E. coli and 50 μg/mL for P. aeruginosa) unless otherwise stated and 1.5 % agar was added to the culture medium for plate cultivation. Phage plating was prepared using the overlay agar method, with LB containing 0.6 % and 1.5 % agar used for the top and the base agar containing 2 mM CaCl₂, respectively. Plasmid pCas12a-λRed [11], pCrRNA-oprM-D [11], pT008 (unpublished), pCOLA-PesaS-PASr-RBS-RFP-rrnbT1 [39], pCrRNA-phzM-I-lacZ [11], pT022 (unpublished), and pET32a-Lys009-His-tag [25] were stocked in our lab. KOD FX DNA polymerase was purchased from TOYOBO Biotechnology (Dalian, China). Q5 DNA polymerase, restriction enzymes, T4 DNA ligase, and Gibson assembly reagents [40] were purchased from New England Biolabs, Inc. (Beverly, MA, USA). DNA markers were purchased from Takara Bio Inc. (Dalian, China). The chemical reagents were of analytic grade and were purchased from Sigma-Aldrich (St. Louis, MO, USA) or Sangon Biotechnology (Shanghai, China). The kits for DNA purification and plasmid mini-prep were purchased from Tiangen (Beijing, China), and the kit for gel recovery was purchased from Magen (Guangdong, China). Information about all the bacterial strains, phage, and plasmids are listed in Table S7, and all primers are listed in Table S8.

4.2. Plasmid construction

Construction of pCrRNA-NC. The DNA fragment containing the spacer sequence “5′-CACTCTTGAGGGCCACAAAGCTT-3′” was amplified from plasmid pCrRNA-oprM-D [11] using primers YM20067/TT18084; the DNA fragment carrying the remaining vector was amplified from the same plasmid using primers TT17022/YM20068. Then the two DNA fragments were ligated by Gibson assembly to yield plasmid pCrRNA-NC. In this plasmid, the crRNA was expressed as a processed repeat-guide (a 19-nt scaffold plus a 23-nt spacer) through the constitutive promoter P_J23119. Subsequently, a series of pCrRNA-X (where X indicates crRNA spacer) plasmids for testing FnCas12a activity were generated on the basis of pCrRNA-NC, by replacing the 23 bp spacer in pCrRNA-NC with new spacers via Gibson assembly. The spacer on pCrRNA-NC did not target either the phage genome or the host genome (Table S2). crRNA spacers were designed using the web tool Cas-Designer (http://www.rgenome.net/cas-designer/) [15,16], following the same criteria as previously described [11].

Construction of pCrRNA-X-Y (where X indicates crRNA spacer and Y indicates the donor DNA). The fused homologous arms (500 or 50 bp each) were amplified from the S1 genomic DNA by overlap-extension PCR, digested with BamHI and HindIII, and then ligated into the corresponding sites on the pCrRNA-X plasmids using T4 DNA ligase, generating pCrRNA-X-Y.

The rfp cassette (786 bp) was amplified from pT008 (an unpublished rfp expressing plasmid from our lab), the rrnB T1 terminator (150 bp) was amplified from pCOLA-PesaS-PASr-RBS-RFP-rrnbT1 [39], the lacZ cassette (3,149 bp) was amplified from pCrRNA-phzM-I-lacZ [11], the Ptat promoter (155 bp) was amplified from pT022 (an unpublished plasmid from our lab), and the lys009 gene (666 bp) was amplified from pET32a-Lys009-His-tag [25].

Construction of pCrRNA-X-D50 plasmid library. The crRNA plasmid library was constructed as previously described with some modifications [24]. The 193 bp fragment library was synthesized by Sangon Biotechnology (Shanghai, China), including the crRNA scaffold (19 bp), the spacer (23 bp), the vector sequence (31 bp), the homologous arm (HA) template (50 × 2 bp), and the homologous sequence (20 bp). The HA was designed to flank the target gene, with the overlap region of adjacent genes retained, if present. The synthesized sequences were separately amplified by PCR using primers YM23007/008. The pCrRNA backbone was amplified by PCR using primers YM230099/010, and the template was digested with DpnI. Both the plasmid backbone and synthesized fragments were purified by gel electrophoresis, pooled, and assembled using Gibson assembly. The Gibson assembly product was purified using the TlANquick Mini Purification Kit (Tiangen, DP 203) and electroporated into XL1-Blue electrocompetent cells. The cells were diluted and plated on LB plates with gentamycin (10 μg/mL) and incubated overnight at 37 °C. The total number of colony-forming units was approximately 10⁵. The colonies were then scraped off the plates, and the plasmids were extracted from the colonies using the TIANprep Mini Plasmid Kit (Tiangen, DP103).

4.3. Preparation P. aeruginosa PAO1 cells for phage S1 genome editing

The plasmid pCas12a-λRed was transferred into P. aeruginosa PAO1 by conjugation, and pCrRNA series plasmids were transferred by electroporation as previously described [11]. The cells were cultured on LB agar plates containing gentamycin (50 μg/mL) and tetracycline (50 μg/mL). The transformants were confirmed by colony PCR and Sanger sequencing.

The crRNA plasmid library was electroporated into P. aeruginosa PAO1 with pCas12a-λRed, yielding 10⁷ colonies on LB plates containing tetracycline (50 μg/mL) and gentamycin (50 μg/mL). To evaluate the coverage of the designed fragments in the crRNA plasmid library, PCR targeting the fragments was conducted using primers YM23091/092, and the PCR products were sequenced on a NOVAseq 6000 sequencer (Illumina, Sangon Biotech (Shanghai, China) Co., Ltd.). We mapped the sequencing data to the designed fragments and determined the ratio of each fragment with 100 % identity in the library.

4.4. Phage S1 genome editing

Recombinant phages were obtained by the spot test method [41,42]. A strain containing pCas12a-λRed and the pCrRNA was cultured overnight in LB medium supplemented with 2 mM CaCl₂ and selective antibiotics. Subsequently, 1 mL of the culture was transferred and induced with 0.1 % l-arabinose (Ara) for expression of λRed at 37 °C with aeration for 1 h.

To perform a spot test, 0.1 mL of the induced bacterial cultures was mixed with 10 mL of pre-warm soft agar (0.6 %, 50 °C) and poured on petri dishes (10 by 10 cm with 6 by 6 grids), dried at room temperature. Then, 10 μL of phage S1 with different dilutions were spotted onto the bacterial lawns and allowed to dry at room temperature for approximately 30 min. The plates were incubated at 37 °C overnight, and the plaques were enumerated the following day. In this step, antibiotics were added to the bottom agar, and the inducer was supplied with the top agar.

4.5. PCR and sequencing analysis of phage mutants

Well-isolated individual plaques were transferred into 1.5 mL Eppedoff tubes containing 200 μL of SM buffer (100 nM NaCl, 8 mM MgSO₄, and 50 mM Tris-HCl, pH 7.5) plus 2 μL chloroform. Following a 1-h incubation at room temperature with mixing every few minutes, the liquid was heat-denatured at 95 °C for 10 min. A 1.6 μL aliquot of the denatured sample served as template for PCR using KOD FX DNA polymerase [7,41]. The amplified DNA was verified via agarose gel electrophoresis and then sent for Sanger sequencing (Sangon Biotech (Shanghai) Co., Ltd. or RuiBiotech (Guangzhou, China)). If necessary, a 100 μL sample was reserved for plaque purification and amplification prior to denaturation.

4.6. Amplification and characterization of recombinant phages

Recombinant phages were amplified as previously described [2,7]. A 10 mL of PAO1 culture (OD₆₀₀ of 0.4–0.6) was infected with 100 μL of the phage lysate sample from the previous step and grown at 37 °C until complete lysis was observed (∼6 h). Subsequently, 200 μL of chloroform was added, and the cells were incubated with shaking at 37 °C for 15 min. The supernatants were collected by centrifugation at 4000×g for 10 min at 4 °C. Chloroform was added to a final concentration of 1 %, and the samples were stored at 4 °C or at −80 °C with 15 % glycerol. The recombinant phages were characterized by one-step growth curve and lysis kinetics as described previously [2].

4.7. Generation and metagenomic sequencing of mutant phage populations

Mutant phage populations were generated as previously described with some modifications [24]. PAO1 harboring pCas12a-λRed and pCrRNA-X-D50 library cells were scraped from the plates using a spreader and suspended in LB liquid medium with 2 mM CaCl₂, tetracycline (50 μg/mL), and gentamycin (50 μg/mL), then stored with 15 % glycerol at −80 °C for subsequent use. A 300 μL aliquot of the frozen bacteria was added to 15 mL of LB medium supplement with 2 mM CaCl₂, antibiotics (50 μg/mL tetracycline and 50 μg/mL gentamycin), and 0.1 % l-arabinose, then incubated overnight at 37 °C. Phage S1 (10¹⁰) was co-cultured with the bacterial solution (10¹⁰) at 37 °C for 6–8 h to produce mutant phages. Subsequently, 1 mL of the supernatant containing the mutant phages was transferred to fresh library cell culture. The remaining supernatant was stored at 4 °C for further analysis. The titer of the mutant phages was measured at appropriate time points. If the mutant phage titer was <10⁵ pfu/mL, the mutant phage population was amplified using PAO1. Total DNA of the mutant phage populations at the 50th, 100th, 150th, 200th, and 260th transfers was sequenced to monitor the deletable gene sets of the phage S1 (Geneplus-Shenzhen).

4.8. Screening, characterization, and genome sequencing of mutants with smallest genome

To obtain isolates of the smallest-genome phages, 10²-10³ pfu of phage lysate (from the 150th, 200th, and 260th transfers) were mixed with 300 μL of PAO1 cells and 5 mL of top LB agar in a tube, then poured onto LB plates. The plates were incubated overnight at 37 °C. One largest plaque and one smallest plaque were selected and purified three times to extract genomic DNA [7], which were then sent for Sanger sequencing as described above.

4.9. Electron microscopy

Phage morphology was examined by transmission electron microscopy (TEM) as previously described with some modifications [2,43]. Briefly, purified phage particles (10⁹-10¹⁰ pfu/mL) were spotted onto 400-mesh carbon-coated grids for 2 min, air-dried, and negatively stained with 2 % phosphotungstic acid (pH 6.5) for 1 min. The grids were imaged using a Thermo Fisher transmission electron microscope at 120 kV. The dimensions of phage particles were measured using Image J.

CRediT authorship contribution statement

Yanmei Liu: Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Zizhen Liang: Methodology, Investigation. Yanyun Jing: Methodology, Investigation. Yanrui Ye: Investigation. Xiaofeng Yang: Writing – review & editing, Supervision, Resources, Funding acquisition, Formal analysis, Data curation. Zhanglin Lin: Writing – review & editing, Supervision, Resources, Project administration, Funding acquisition, Conceptualization.

Funding

This research was funded by National Key R&D Program of China (2022YFC2104800, 2018YFA0901000), the Program for Guangdong Introducing Innovative and Entrepreneurial Teams (2019ZT08Y318), and the Science and Technology Program of Guangzhou City (201803010058).

Declaration of competing interest

Zhangling Lin is an Editorial Board Member for Synthetic and Systems Biotechnology and was not involved in the editorial review or the decision to publish this article. All authors declare that there are no competing interests.

Appendix A. Supplementary data

The following are the Supplementary data to this article: