Bacteriophage genome-wide transposon mutagenesis

Summary

Bacteriophage genomes are densely packed with coding sequences and frequently encode genes of unknown function. Unbiased phage functional genomics approaches are therefore needed, particularly for large lytic phages. Here, we harness the mariner transposase to develop phage transposon mutagenesis and sequencing (TnSeq), enabling pooled sequencing to identify both fitness-conferring and dispensable genes. Using the Pseudomonas aeruginosa-infecting nucleus-forming jumbo phage ΦKZ (280,334 bp; 371 predicted genes), we show that ~110 genes are fitness-conferring via phage TnSeq, identifying many known and previously unknown essential genes. Moreover, this phage carries ~261 non-essential genes, including some capsid and tail proteins, many of which are important for fitness across different clinical isolates or conditions. Phage TnSeq was also extended to a base-modified phage. Together, phage TnSeq is a scalable technology that can identify essential phage genes, generate knockouts in all non-essential genes, and sensitively assign the quantitative fitness contributions of every gene in parallel.

Graphical Abstract

A technique enabling phage transposon mutagenesis and sequencing reveals essential and fitness-related genes, including many previously unknown ones, in both standard and base-modified phages.

Introduction

Bacteriophage genomes are rich in genes of unknown function and exhibit extensive genetic and proteomic diversity. Many fundamental discoveries likely remain hidden in uncharacterized phage genes¹. However, few tools exist to enable unbiased genome-wide studies to bring phages into the functional genomics era. Most challenging perhaps are large lytic phages (e.g. jumbo phages with genomes >200 kb), where most genes are of unknown function². For most of these phages, even the basic inventory of genes essential for replication is unknown.

Here we focus on ΦKZ, a model member of the Chimalliviridae family with distinctive infection biology^3,4. ΦKZ has many notable features, including two distinct multi-subunit RNA polymerase complexes^5–8, a lipid-based early phage infection vesicle^9–11, and a protein-based phage “nucleus”^12,13 that enables resistance to nucleases¹⁴. It remains unclear which genes are strictly required for this core biology and which are dispensable but conditionally important, for example in specific hosts or environments. A method that could both rapidly identify all essential phage genes and simultaneously generate a library of knockouts in all other genes would greatly accelerate phage-host mechanistic dissection.

Targeted phage genome manipulation tools such as homology-directed recombination (HDR) and CRISPR-Cas counter-selection have been widely used¹⁵ albeit at low throughput. Recombineering is also effective and has recently been adapted to genome-wide mutagenesis¹⁶, and in vitro genome manipulation and rebooting have expanded the toolkit further¹⁷. RNA-targeting Cas13 has been particularly useful for selecting for spontaneous^14,18 or HDR-derived phage mutants¹⁹. To enable bacteriophage genome modification with the ease afforded to bacteria by transposon insertion sequencing (TnSeq), we sought to develop an analogous genome-wide insertional mutagenesis method for phages.

Here, we deploy mariner transposition in the nucleus-forming jumbo phage ΦKZ, using an anti-CRISPR (acrVIA1) plaque selectable marker and CRISPR-Cas13a–mediated enrichment to isolate thousands of independently transposed phages. This approach was also extended to the related KTN4 phage and an unrelated phage with base modifications, YuA, using the recently identified END nuclease and anti-END gene²⁰ to select for transposed phages. High-throughput mapping of insertion sites revealed all essential genes in the genome and yielded a large collection of phage mutants for conditional phenotypic screens. Phage TnSeq requires no large-scale oligo or guide RNA design, mariner transposase is active in many bacterial hosts²¹ and inserts transposons semi-randomly^22,23. Together, these features should enable broad adoption of phage TnSeq as an easy-to-implement, low-cost genome-wide mutagenesis platform which can, in principle, be executed with any combination of a strong phage defense system and cognate anti-defense molecule.

Results

Establishing transposition in a nucleus-forming jumbo phage

ΦKZ protects its genome within a protein-based nucleus-like structure during infection. To enable transposition inside this compartment, we fused the Mariner transposase to an sfCherry fluorescent reporter and to the nucleus-localized phage protein Nlp1^12,24 (gp152) (Figure 1A). Upon induction of this fusion and infection with ΦKZ, fluorescence microscopy revealed co-localization of sfCherry–Mariner–Nlp1 with DAPI-stained phage DNA, indicating import of the transposase into the phage nucleus (Figure 1B).

Figure 1. Transposon mutagenesis of a nucleus-forming jumbo phage.

(A) Schematic illustrating the ΦKZ transposon mutagenesis strategy. The Nlp1–Mariner transposase fusion localizes to the phage nucleus, where a transposon encoding acrVIA1 inserts into the phage genome, generating insertional mutants. Output phages are subsequently subjected to Cas13-based counter-selection.

(B) Fluorescence microscopy of P. aeruginosa infected with ΦKZ expressing sfCherry-tagged transposase (red) and stained with DAPI (blue). White arrows indicate co-localization of transposase with the phage nucleus in infected cells.

(C) Quantification of ΦKZ replication efficiency in the presence of single versus dual Cas13a crRNAs.

(D) Representative plaque phenotypes of transposed ΦKZ forming on dual Cas13a targeting strains.

(E) Efficiency of plating (EOP) of ΦKZ on dual Cas13a strains following transposition with variants V1 (acrVIA1 with a phage early promoter), V2 (acrVIA1 lacking promoter), or V3 (derived from V1 with modified primer binding sites). All dark grey bars are at the limit of detection, where a pseudocount of 1 was used for calculating detection limits. See also Figure S1.

The initial transposon donor (V1) carried a selectable acrVIA1 cassette flanked by a ΦKZ consensus early promoter^8,25 and a terminator, to drive early acrVIA1 expression (Table S1). For selection, we co-expressed Listeria seeligeri Cas13a (LseCas13a)^19,26 with one of two crRNAs: one targeting orf120 (major capsid) and one targeting orf55 (a nvRNAP subunit). Each crRNA alone strongly restricted phage infection but was individually escapable (Figure 1C). We therefore designed a single repeat–spacer–repeat–spacer array expressing both crRNAs, which efficiently blocked infections by previously isolated escaper phages (Figure S1A). No escaper mutants were detected (limit of detection (LOD) <2.5 × 10⁻¹⁰ pfu/mL), so this dual crRNA array was used for all subsequent selections (Figure 1C, Figure S1A–B).

To generate a transposed phage population, we infected a log-phase culture expressing Mariner–Nlp1 with ΦKZ for 16 h and then selected output phages on the dual-crRNA Cas13a strain using full-plate infections. Single plaques were isolated, and acrVIA1 was detected by PCR (Figure S1C). Plaque formation required both the transposase and the transposon donor (Figure 1D–E, Figure S1D). Whole-genome sequencing of four independently isolated phages confirmed bona fide transposon insertions in orf303, orf208, orf241, and orf252. In each case, two inverted repeats flanked an insertion at a TA dinucleotide in the ΦKZ genome (schematized in Figure S1C).

Transposition using the V1 donor occurred at a rate of ~1 in 3 × 10⁷ phages, demonstrating that transposition is inefficient but feasible in a large population (Figure 1E). No plaques were detected when the donor plasmid was present without transposase (LOD: 1.3 × 10⁻¹⁰ pfu/mL (Figure 1E)). A second donor (V2) lacking the early promoter yielded slightly lower efficiency (1 in 6.7 × 10⁷ phages; Figure 1E), consistent with acrVIA1 expression relying on endogenous promoters. To facilitate compatibility with bacterial Tn libraries, we engineered a third donor (V3, derived from V1) that altered the transposon inverted repeat junctions to introduce orthogonal primer binding sites. Transposition efficiency with V3 (1 in 7 × 10⁷ phages) was similar to V1 and V2 and remained strictly dependent on the Nlp1–Mariner fusion (LOD: 3.9 × 10⁻¹¹ pfu/mL). Transposition was detected across a range of input phage concentrations and was most efficient at a relatively low input (~2 × 10⁴ pfu/mL; Figure S1E). We therefore used the V3 donor to generate a pool of mutants for insertion sequencing.

A ΦKZ phage TnSeq library with genome-wide coverage

To build independent pools of transposon mutants, we performed ten biological replicate transposition infections (Figure S2). Mutants from each reaction were enriched on the Cas13a counter-selection host, and fractions from all ten pools were combined to generate a single mutant library. We isolated DNA, prepared sequencing libraries, and amplified transposon–genome junctions for Illumina sequencing (Methods). We obtained ~3.4 million high-quality reads, 90% of which contained transposon sequence. Of these, 55.7% mapped to the ΦKZ genome and the remainder to the V3 donor plasmid. Across the ΦKZ genome, 6,667 unique insertion sites were identified (Figure 2A), corresponding to a mean of one insertion every 42 bp over the 280,334 bp genome.

Figure 2. Genome-wide map of ΦKZ gene essentiality.

(A) Genome-wide distribution of transposon (Tn) insertion reads across the ΦKZ genome. All forward reads are shown without filtering, with a pseudocount added to sites with single reads for log-scale visualization. The ΦKZ genome map is displayed below, with genes in the forward direction in dark gray and reverse in light gray.

(B) Schematic summary of gene essentiality across the ΦKZ genome, as determined by insertion sequencing (TnSeq) analysis. Dark red represents low insertion index (shown in log scale), while dark blue represents a high insertion index. See also Figures S3, S4, S5 and Tables S2 and S3.

To define confident gene insertions, we first examined genes expected to be essential. Many insertion sites yielded low sequencing depth, suggesting that insertion occurred but rendered phages unfit. We therefore required ≥3 reads per insertion for downstream analysis. In addition, insertions within the first or last 10% of a gene were excluded to avoid misclassifying overlapping genes or insertions that might not fully disrupt gene function. Using these criteria, ~30% of ΦKZ genes lacked disruptive insertions, whereas ~70% tolerated insertions (Figure 2B, S3).

ΦKZ phage TnSeq identifies essential genes

As an initial accuracy check, we focused on large genes, which should, by chance, accumulate many insertions unless they are essential. Tailed phage genomes often encode large essential structural and polymerase proteins. In our ΦKZ TnSeq dataset, all 16 genes >2,154 bp and 30 of 34 genes >1,560 bp lacked disruptive insertions (Figure 2B), strongly indicating essentiality given the genome-wide insertion frequency (one unique insertion every 42 bp). We also created a TnSeq library in a closely related phage (96.7% nucleotide identity), KTN4 (Figure S4) and observed that the essentiality profile across both phages was nearly identical. Below, we highlight examples of ΦKZ essential genes in tail, capsid, and conserved non-virion modules with all features reported in Figure 3A.

Figure 3. Discovery of ΦKZ gene functions using Tn mutagenesis.

(A) Heat map showing gene fitness scores compared with antisense oligonucleotide (ASO) assay results. Gene conservation and virion localization are indicated. Genes are rank-ordered by essentiality. Yellow represents high insertion index or no fitness defect by ASO (−1,0). Colors proceed to purple as insertion frequency goes down or as fitness defect emerges due to ASO knockdown (2,3).

(B–C) Examples of Tn insertion read density across representative genomic regions (orf 53–59 and 141–144). See also Figure S5.

Tail genes.

The largest gene in ΦKZ, orf181, encodes a lysozyme-containing tail protein likely serving as the tape measure²⁷. orf181 and its neighbor orf182 (encoding a baseplate protein) both lack transposon insertions (Figure 2B). Given the 6,714-bp size of orf181, its complete lack of insertions makes it a high-confidence essential gene. Other tail-associated genes lacking Tn insertions encode putative tail fibers (gp145, gp146), a baseplate wedge protein (gp87), a paralogous baseplate protein family (gp131–135)²⁷, the tail sheath^28,29 (gp29), and the tail tube (gp30) (Figure 2B). Virion-associated proteins previously of unknown function such as gp27 (DUF7193) and a low-copy virion protein gp128 also lack insertions. A recent cryo-EM structure of the ΦKZ virion demonstrated that these two proteins are both in the baseplate³⁰. Virion localization and functional annotations described in Figure 3A were supported by previously published mass spectrometry datasets^31–33, HHpred/PSI-BLAST analyses, and a high-resolution structure³⁰.

Head genes.

Several large genes that appear essential by phage TnSeq encode for capsid assembly, DNA packaging, or injected proteins. These include the major capsid protein (gp120)^32,34, the portal (gp129)^32,35, and the large terminase (gp25)³² (Figure 2B). Mass spectrometry of a tail-less ΦKZ mutant previously revealed >50 distinct polypeptides incorporated into the head, many of which are proteolytically processed by an essential (per phage TnSeq) head protease, gp175^32,36. Many head proteins (gp92, 93, 94, 95, 96, 97, 153, 162, 163; PF12699.12 (C-terminus)^10,32) are members of a paralogous family^10,33 and some of them are injected during infection¹⁰. Interestingly, several paralogous genes (e.g., orf92, 93, 96, 162, 163) readily tolerate insertions, whereas genes encoding other high-copy inner body proteins³² such as orf89 and orf90, along with genes encoding the vRNAP (orf178, 149, 180, 80)^8,37 appear essential by phage TnSeq.

To validate a subset of phage TnSeq-based essentiality calls, we used Cas13a to target transcripts encoding head proteins and isolated escape mutants. Strong targeting yielded phages with large deletions in non-essential genes or resulted in point mutations, if the genes are essential. Deletions were obtained in several genes predicted to be non-essential by phage TnSeq (orf93, 94, 162, 163, 203, 303), while targeting of genes predicted to be essential (orf89, 90, 177) did not result in deletions (Figure S5), corroborating their essentiality. Notably, orf95, orf97, orf157, and orf85–86 tolerated deletions despite lacking transposon insertions, likely due to a mix of polar effects and fitness contributions during competition, both to be discussed below.

Conserved Chimalliviridae Proteins:

In addition to gene size, evolutionary conservation is expected to be correlated strongly with essentiality. A previous study defined 20 ΦKZ genes that are conserved across Chimalliviridae but not found in other phages³⁸. Of the “Prichard 20”, 19 appear essential or strongly fitness conferring by phage TnSeq (15 with zero insertions; 4 with only 1–3 insertions; Figure 3A, Tables S2, S3). The lone exception is the paralogous protein family encoded in multiple copies (orf93/162/163). As an example of an essential conserved gene, Chimallin A (orf54), the phage nucleus shell protein^39,40, was previously shown to be refractory to knockout in ΦKZ¹⁹ and essential upon translational repression in the related E. coli phage Goslar⁹. No Tn insertions were identified in this gene. Other conserved genes encoding protein importers (imp1/orf69; imp2/orf47; imp3/orf59; imp4/orf48), non-virion RNA polymerase (nvRNAP) subunits (orf123, orf74, orf68, orf71–73, orf55–56.1), and the nucleus-associated imp6/chmC/orf67^24,41 also lack disruptive insertions (Figures 2B, 3A). Additional conserved non-virion genes, such as orf165 (a putative SbcCD-like ABC ATPase with SMC homology), also lacked insertions. A subset of these genes were also queried for essentiality previously via ASO-based translational repression⁴², with generally consistent results and some exceptions. Figure 3A summarizes the concordance between phage TnSeq fitness scores and ASO knockdown.

Polar Effects and Transcriptional Architecture

Insertional approaches in bacteria and phages are susceptible to polar effects, where an insertion disrupts expression of downstream genes in the same transcriptional unit. Polar effects are more severe when transcription is driven by few transcription start sites (TSSs), and less problematic when multiple TSSs distribute transcription across densely packed loci. Although phages are sometimes depicted as having few promoters, recent long-read direct RNA sequencing identified 149 TSSs across the ΦKZ genome²⁵, in line with earlier predictions of 134 operons⁸.

To ask whether abundant TSSs mitigate polar effects in ΦKZ, we again examined the chmA (orf54) gene. orf53 and chmA appear in a putative operon with a TSS driving orf53-chmA, yet multiple transposon insertions are present in orf53 while none occur in chmA (Figure 3B). This indicates that Tn insertion in orf53 does not abrogate chmA transcription. An independent TSS is found directly upstream of chmA, likely sustaining chmA transcription even when orf53 is disrupted. orf55 (nvRNAP subunit) also appears essential, while orf56 and orf57 are non-essential and orf58–59 appear essential (Figure 3B and Tables S2, S3). A TSS upstream of orf59 (imp3) supports the interpretation that the absence of insertions in orf58 reflects essentiality of gp58, rather than a polar effect on imp3. This pattern in a tightly packed locus underscores how multiple TSSs can insulate neighboring essential genes from polar effects.

Another example of polar effects being mitigated arises in the lysis cassette. Biochemical studies show that gp144 is a highly active peptidoglycan hydrolase and the likely endolysin^43,44. A recent report further identified gp142 as the holin and gp143 as a lysis regulator⁴⁵. By phage TnSeq, orf142 and orf144 lack well-sequenced insertions and appear essential, whereas orf143 is disrupted by five insertions in its 534 bp coding sequence. Multiple TSSs in this region are again identified, upstream of orf139, orf143, and orf144²⁵ (Figure 3C). Thus, many potential polar effects are mitigated by the underlying transcriptional architecture. Importantly, abundant TSSs are not unique to ΦKZ: several smaller Pseudomonas phages profiled by the same RNA sequencing method also exhibit numerous TSSs (e.g., 14–1, 50 TSSs; LUZ24, 16; LUZ19, 9; PAK_P3, 75; YuA, 21)²⁵.

By contrast, polar effects are expected when genes are expressed from a single polycistronic transcript without internal TSSs. Two clear examples are the ΦKZ intron-encoded homing endonucleases gp72 and gp179, both lying within RNA polymerase operons. ASO knockdown data suggested that gp72 is essential while gp179 is not⁴², while both genes appear essential by phage TnSeq (no insertions). Moreover, a gp179 homolog in ΦPA3 is non-essential but implicated in phage–phage competition⁴⁶. To resolve this discrepancy, we used Cas13a to target each gene and readily isolated deletion mutants in orf72 and orf179, demonstrating that neither gene is strictly essential for ΦKZ replication (Figure S5C). These loci thus exemplify polar effects: transposon insertion in orf72 or orf179 likely terminates transcriptional readthrough, blocking expression of the downstream essential genes (orf73 or orf178). Similar cases likely exist elsewhere in the ΦKZ genome. Cas13-based targeting of individual genes coupled to sequencing of escape mutants provides a straightforward way to discriminate true essential genes from polar-effect artifacts uncovered by phage TnSeq.

Non-essential genes revealed by dense insertion coverage

Genes with abundant, evenly distributed insertions can be confidently classified as non-essential. orf39 encodes the PhuZ tubulin required for nucleus positioning^12,13,47,48 and contains many insertions across its length (Figure 2B, 3A) and has been previously deleted¹⁹. ChmB (gp2), a protein proposed to cooperate with PhuZ to facilitate capsid treadmilling and docking at the phage nucleus⁴⁹, likewise appears non-essential, suggesting that core aspects of DNA packaging remain to be fully explained.

Other genes with many Tn insertions (Figure 2B, 3A) and known non-essential functions encode: a ribosome binding protein (gp14)⁵⁰, an injected protein¹⁹ (gp93), Thoeris anti-defense protein 1⁵¹ (Tad1, gp184), a cluster of uncharacterized non-essential proteins gp206–216¹⁹, the activator of Juk defense (gp241)⁵², and proteins encoded nearby (gp237, gp240, 242)⁵². Dip (gp37), a previously studied RNA degradosome inhibitor⁵³, also appears dispensable. Interestingly, many virion proteins in ΦKZ appear essential, but notable exceptions include a helicase gp203 and gp303 (unknown function), both of which are proteolytically processed and packaged into the head³². gp203 and gp93 co-pellet with phage genomic DNA after capsid disruption³², indicating tight association with packaged DNA despite being non-essential under the conditions tested. Minor capsid proteins with conditional essentiality will be discussed below.

Small genes were particularly likely to tolerate insertions. Of 210 genes shorter than 500 bp (arbitrarily defined as “small”), 185 contain at least one insertion, while just 25 lack insertions (Figure 2B). Because small genes are more likely to be misclassified as essential by chance, this likely underestimates the true number of non-essential small genes. Even so, our insertion data strongly indicate that approximately half of ΦKZ genes are both small and non-essential under the laboratory conditions used here (37 °C, LB media, high aeration). Across the full genome, roughly 70% of genes appear non-essential under these conditions. Many of these non-essential genes may be conditionally essential in specific hosts or environments (examples are presented in the next section)—phenotypes that phage Tn mutagenesis is uniquely poised to uncover.

Minor capsid proteins are required at environmental temperature

P. aeruginosa is frequently isolated from soil, freshwater, and human-associated environments^54,55. We therefore queried whether any phage genes are dispensable at 37 °C but required at a lower environmental temperature (18 °C). We first subjected a randomly chosen set of 90 transposed mutants to infection at 18 °C. Four mutants that grew poorly at 18 °C were identified (Figure S6A) and sequenced alongside a control mutant that grew normally. Insertions mapped to orf57, orf162, orf199 (two redundant insertions), and orf244 (Figure S6B). Transposon insertions in orf244 and orf199 displayed decreases in plaque formation at 18 °C upon subsequent validation (Figure S6B). Additionally, we subjected the pooled transposon library to liquid growth at 18 °C and sequenced the entire output to identify enrichment or depletion. This method revealed that in addition to again identifying orf244, insertions in orf166 and orf31 also decreased phage replication at this temperature (Figure 4A). orf244 and orf31 encode structural proteins, while orf166 is of unknown function. Deletions of orf244 or orf166 impeded ΦKZ replication at 18 °C, but mutants were competent for replication at 37 °C (Figure 4B). While gp166 is of unknown function, gp244 decorates the outside of the phage capsid³⁴, perhaps stabilizing the structure under environmental temperatures. The orf244::Tn mutant completely failed to replicate at 18 °C, a defect rescued by expressing orf244³³ in trans (Figure 4C).

Figure 4. ΦKZ carries genes that are essential under different conditions.

(A) ΦKZ mutant fitness at 18 °C compared to 37 °C.

(B) Single step growth curve of different phages at 37 °C compared to 18 °C.

(C) Capsid structure and plaque assay of ΦKZ orf244::Tn in the presence/absence of orf244 in trans.

(D) Efficiency of plating (EOP) of ΦKZ wild type and Δnlp1 in the presence of CBASS with indicated genes expressed from pHERD30T plasmid. Control phages JBD67 and JBD67Δacb2 confirm CBASS targeting specificity, while Acb2 expression neutralizes CBASS activity.

(E) Quantitative PCR analysis of ΦKZ wild type and ΦKZ Δnlp1 replication in the presence and absence of CBASS.

(F) Plot showing changes in frequency of each ΦKZ gene in the presence/absence of EcoRI and PfJuk. Plaque assay showing ΦKZ and ΦKZΔorf105 in the presence of PfJuk.

(G) Number of phages and insertion index of each ΦKZ gene over time. A pseudocount of 10-4 is used for genes with an insertion index of zero. Limit of detection (LOD) is as shown as dotted line.

See also Figure S6. All data are represented by mean ± s.d.

To understand the role of other minor capsid proteins in this phage, we next checked what phage TnSeq and a recent cryo-EM structure of the capsid reveal³⁴. The cryo-EM structure revealed ~10 “minor capsid” proteins whose N-termini weave into major capsid hexamers³⁴. Most of the genes encoding these minor capsid proteins, including orf35, orf93, orf162, orf184 (tad1), and orf244, appear non-essential by phage TnSeq, at 37 °C. Interestingly, only three minor capsid genes appear essential at 37 °C: orf28, orf91, and orf119. The structural data provide an explanation for this, revealing a junction between vertex-binding complexes in which gp28 from one complex interacts with gp91 and gp119 from the adjacent complex³⁴. These findings therefore demonstrate multiple roles for minor capsid proteins: i) three minor capsid proteins that form an essential (at 37 °C) complex in the capsid, ii) additional minor capsid proteins, including a capsid decorating protein gp244, that are dispensable at 37 °C, but essential at 18 °C, and iii) non-essential proteins whose N-terminal domains are wedged into the capsid with C-terminal domains that are liberated by a head protease³² and injected during the next round of infection (i.e. gp93, gp184/Tad1)^33,51. The transposed phage mutant library and pooled TnSeq can thus be used to rapidly identify genes whose contributions emerge only in ecologically relevant conditions—a likely major evolutionary pressure for P. aeruginosa phages⁵⁶.

Nlp1 is required for NucC resistance

Phage Tn mutagenesis enables systematic screens for strain- or condition-specific phenotypes. To identify phage genes that are conditionally essential for genome protection, replication, repair, or stability, we arrayed ~1,500 transposed ΦKZ mutants in 96-well format and plated the ΦKZ mutants against two nucleases: EcoRI and the Type III-C CBASS system with a NucC effector. Wild-type ΦKZ naturally resists both^14,24. We hypothesized that mutants with increased EPI vesicle or nuclear permeability might become sensitive to EcoRI, a ~30 kDa protein that cannot normally access the phage genome but cleaves ΦKZ DNA if granted access to it in vivo and in vitro¹⁴. However, none of the arrayed mutants were restricted by EcoRI, indicating that genomic segregation from the cytoplasm—and thus resistance to both EcoRI and the endogenous PAO1 Type I RM system^14,57—is robustly maintained across all tested mutants.

By contrast, one ΦKZ mutant from the arrayed collection completely lost the ability to replicate on strains expressing a Type III-C CBASS system (Figure 4D). This system senses unknown phage factors, produces cAAA nucleotide signals, and activates NucC, a nuclease effector previously implicated in blocking a ΦKZ-like phage in Serratia⁵⁸. In that context, degradation of the host genome outside the phage nucleus was proposed to halt phage progression. We were therefore particularly interested in mutants that become sensitive to NucC, because ΦKZ-like phages degrade the host genome^47,48.

The NucC-sensitive mutant carried a Tn insertion in nlp1 (nucleus-localized protein 1), encoding a protein homologous to phage T4 UvsX⁵⁹. UvsX is a recombinase that is not strictly essential in T4 but its loss confers sensitivity to DNA damage⁵⁹. To confirm this result, nlp1 was deleted by replacing it with acrVIA1 (ΦKZ nlp1::acrVIA1, hereafter referred to as Δnlp1) under complementation with wild-type nlp1 in trans. Like the Tn mutant, the Δnlp1 phage was sensitive to Type III-C CBASS NucC, and this sensitivity was rescued by Nlp1 expression in trans (Figure 4D). Sensitivity was dependent on canonical CBASS activity because Acb2, a cyclic-nucleotide sponge that sequesters cAAA⁶⁰, rescued the Δnlp1 phage. In contrast, Nlp1 expression did not rescue CBASS NucC sensitivity of an unrelated phage (JBD67Δacb2), indicating that Nlp1 is not a general CBASS inhibitor (Figure 4D).

During infection, NucC activity modestly reduced WT ΦKZ DNA levels yet completely abolished DNA replication in the Δnlp1 mutant (Figure 4E). This outcome is specific to NucC, as a Δnlp1 phage remains fully resistant to EcoRI (Figure S6C). The absence of phage DNA replication initiation during NucC phage targeting suggests that the promiscuous NucC may target momentarily exposed phage DNA in the early-middle stages of infection, despite generally protective organelles. Attempts to visualize NucC localization were unsuccessful because fluorescent tagging rendered NucC inactive. Nonetheless, the genetic data support a model in which Nlp1 recombinase activity repairs NucC-induced phage DNA damage, analogous to UvsX-mediated recombination of fragmented T4 genomes after UV exposure or Cas9/Cas12 cleavage⁶¹. While NucC is often viewed as anti-phage through host genome destruction, this activity is not intrinsically anti-phage. Our data suggest that NucC directly damages phage DNA and that ΦKZ relies on Nlp1 to ameliorate this assault.

In a separate arrayed screen against the PfJuk⁵² (jumbo phage killer) defense system, which restricts ΦKZ, we observed robust restriction of most mutants but identified one “escaper” that grew well. Sequencing revealed a Tn insertion in orf241, encoding the previously described Juk activator⁵². In a parallel pooled library sequencing screen, we also excitingly observed insertions that either enhanced or limited phage replication. For example, insertions in the PfJuk activator orf241 enhanced phage replication, while insertions in a gene of unknown function orf105 limited phage replication on PfJuk (Figure 4F). Conversely, when the same library was used to infect a strain expressing EcoRI, no phage mutants were enriched or depleted (Figure 4F), lending confidence to PfJuk-specific hits. Plating of an orf105 mutant on PfJuk showed some reduction in phage titer compared to wild-type (Figure 4F), consistent with gp105 being a PfJuk antagonist. Additional genes were also identified as putative activators and inhibitors in the same dataset. Thus, the transposed mutant collection can rapidly and sensitively uncover defense inhibitor and activator alleles, even when redundancies are present, that quantitatively shift phage outcomes when in competition with all other phage mutants.

Identification of other fitness-conferring genes

The absence of Tn insertions in nlp1 in the pooled phage TnSeq library suggested that it is a fitness-conferring gene, although the results here show that it is not strictly essential. Indeed, Δnlp1 phages replicated DNA slower (Figure 4E), and formed smaller plaques than wild-type ΦKZ, which was restored by Nlp1 complementation (Figure S6D), a phenotype reminiscent of T4 uvsX mutants⁵⁹. Tn insertions in nlp1 were therefore likely present at low frequency in the initial transposed population but were outcompeted during propagation. In a direct competition assay with WT phage over 4 h in liquid culture, Δnlp1 showed a competitive index of 0.05, confirming a fitness defect. Other acr insertions in orf2/chmB, orf184/tad1, orf39/phuZ, or orf241 exhibited competitive indices of ~0.35–2.0 (Figure S6E). nlp1 thus exemplifies a gene for which full deletion is possible, but the mutant drops out during phage competition or in the presence of CBASS NucC.

This observation motivated us to determine if there were other genes that were similarly not essential for replication in PAO1, but fitness-conferring and thus mutants drop out over time. We therefore sequenced transposon insertions prior to CRISPR-Cas13a selection (t=0h), and at t=5h and t=20h post-selection. Over this time, transposed phages go from being a minority of the population to dominating the population, by virtue of CRISPR-Cas13a selection (Figure 4G). We identified an additional 15 genes (Figure 4G, orf3, 44.1, 56.1, 75, 76.4, 79, 85, 95, 97, 106, 132, 156, 179, 188, 228) that had an intermediate insertional tolerance at t=5h, clustering with nlp1, and all appeared to lack insertions by 20h (i.e. these phages generally failed to proliferate, Figure 4G). This suggests that an intermediate group of genes exist that are fitness-conferring like nlp1 but not strictly essential, which time-resolved phage Tn-seq is able to identify. Multiple genes from this category are indeed able to be knocked out or interrupted. orf75, for example, tolerates insertional inactivation (Figure S5C), is a core Chimalliviridae gene³⁸ encoding an SF2 family helicase, is encoded downstream of the nvRNAP, and localizes to the phage nucleus with an unknown function¹⁰. orf85 tolerated Cas13a-selected deletions (Figure S5C) and encodes a capsid protein that associates with the aforementioned essential vertex binding complex (i.e. with essential gp91, gp28, and gp119). orf95 and orf97 encode high-copy inner body proteins that belong to the large paralogous gene family PF12699.12 and are injected during infection^10,33. orf95 tolerated Cas13a-selected deletions along with neighboring orf93–94, as did orf97, but insertions in orf95 appear less fit during growth of the entire library (Figure 4G) and an orf95::acr mutant also has a competition defect when competing with WT ΦKZ (Figure S6E). These results suggest that genes which are clearly not strictly essential can have fitness costs that are resolved with this modified approach to phage TnSeq.

Together, these findings highlight the value of combining pooled end-point phage TnSeq with time-resolved TnSeq to map essential/fitness-conferring genes, paired with an arrayed mutant library to interrogate conditional phenotypes and recover mutants that are underrepresented in pooled selections.

KZ Tn library reveals non-essential genes important in infecting wild isolates

Many non-essential ΦKZ genes cluster within a broad region of largely contiguous non-essential genes spanning from ~orf184 through orf304 and then orf1–orf20 on the left arm (Figure 2B). ΦKZ is circularly permuted and its genome ends in NCBI are arbitrary. This region roughly corresponds to the portion absent from a reduced size P. aeruginosa jumbo phage, ΦEL^35,62. This observation suggests that this region encodes host-interacting or defense-related adaptations rather than core replication modules. We therefore executed arrayed replica plating plaque assays with 1,500 mutant plaques on two isolates (PA1032, Env201), while also infecting clinical (MRSN369569, PA1032, UCSF E16) or environmental (EnvBC13, EnvBC15, Env201) strains with the pooled ΦKZ mutant library (Figure 5A). Gene insertions that impact fitness were identified with high throughput sequencing. Notably, the Tn insertions in the phage are stable and these strains need not be equipped with CRISPR-Cas13a to maintain selective pressure.

Figure 5. ΦKZ Tn-Seq library reveal phage genes in wild isolates.

(A) Plot showing changes in frequency of each ΦKZ gene in the clinical isolate indicated compared to PAO1. 10-fold dilution spot titration is shown above the plot to demonstrate phage titer on the indicated strain.

(B) Adsorption of orf298::Tn compared to wild-type ΦKZ in isolate PA1032 and Env201 compared to PAO1

(C) Adsorption and 2-hour replication profile of orf302::Tn compared to wild-type ΦKZ in the environmental isolate Env201, compared to PAO1. Data shown as mean ± s.d.

Insertions in orf298 were identified from both the arrayed plating experiments and this gene was strikingly required for optimal phage fitness in all six isolates tested (Figure 5A). Conversely, many insertions were tolerated in this gene (28 insertions across its 1,008-bp coding region) when the phage is replicating on PAO1 (Figure 2A). gp298 contains a Kelch domain and a predicted N-acetylneuraminate epimerase domain and has been identified at low abundance in the virion by mass spectrometry^31–33. We therefore wondered if gp298 is a tail protein required for adsorption to wild isolates but not in a lab strain. Indeed, orf298::Tn mutant failed to adsorb to PA1032 and Env201 but adsorbed normally to PAO1 (Figure 5B). These data support gp298 as a previously unrecognized tail-associated factor required for adsorption and/or receptor access on a subset of wild P. aeruginosa strains. Notably, orf298 is not located near canonical tail genes; instead, it clusters with other non-essential genes of unknown function.

In contrast to orf298, which limited phage replication across all six tested isolates, some genes were observed to be conditionally required in a subset of tested isolates. Specifically, orf10–17 and orf57 were conditionally essential in two of the six tested isolates (EnvBC13 and EnvBC15), while orf227 was conditionally essential in two others (UCSF E16 and MRSN369569). There were some genes that were conditionally essential in only one of the four tested isolates: orf233 in PA1032 and orf201 and orf213 in MRSN369569, orf302 in Env201. A phage with an insertion in orf302 was isolated and adsorbed normally, but failed to produce progeny, consistent with an Env201-specific intracellular defense that is absent in PAO1 (Figure 5C). Overall, these data identify host-specific adaptations encoded by non-essential ΦKZ genes and demonstrate that the pooled transposon library can identify their function in an unbiased manner. Future work can leverage these data to identify anti-jumbo phage defenses operating in wild isolates.

Transposition using a non-CRISPR-based system

To achieve CRISPR-independent selection for transposed phages, we applied a recently described antiphage defence system called the END nucleases that broadly targets modified phages²⁰. END nuclease can be inhibited by an anti-END inhibitor identified in distinct families of P. aeruginosa phages²⁰. In principle this could establish an “out-of-the-box” method for transposition of any modified DNA phage. The Yuavirus phage family is a group of phages that have not been well studied but possess a genome with modified thymidine⁶³ and is sensitive to END^PaCF1. We therefore constructed a phage transposon library in YuA, using the same protocol as that used for ΦKZ, but selecting with END^PaCF1. Selection for transposed phages was executed at both 30 °C (Figure 6) and 37 °C (Figure S7), yielding highly similar outcomes. Results from 30 °C selection will be discussed here.

Figure 6. Genome-wide map of YuA gene essentiality.

(A) Genome-wide distribution of transposon (Tn) insertion reads across the YuA genome. All forward reads are shown without filtering, with a pseudocount added to sites with single reads for log-scale visualization. The YuA genome map is displayed below. The log scale zoom across two genomic sites as shown. Blue = forward reads, red = reverse reads

(B) Schematic summary of gene essentiality across the YuA genome, as determined by insertion sequencing (TnSeq) analysis. See also Figure S7.

YuA has 944 possible TA insertion sites out of 58,663 bp, which corresponds to one possible insertion site every 62 bp. Of the 944 possible insertion sites, 807 sites were within coding regions. Many intergenic insertions were also observed but here we focus on coding gene interruptions. Sequencing of the YuA library showed that of these 807 possible insertion sites (i.e. 1614 possible insertions in both directions), a total of 350 sites had at least one read (21.7%, Figure 6). We observed that YuA follows a grouping of two essential gene clusters (DNA replication and virion morphogenesis/lysis) and one non-essential cluster with some interesting exceptions (Figure 6A). First, the DNA replication locus likely extends through a series of genes that do not tolerate insertions, spanning from gp2 to gp41 (note: previous literature and NCBI record refers to genes as ‘gpX’), the last gene in the locus being an essential 3’−5’ exonuclease (with similarity to the DNA polymerase III exonuclease domain in E. coli). This locus also encodes genes expected to be required to enact DNA modifications in this phage, which all appear to lack well-sequenced insertions. The gene previously suggested to be at the end of the DNA replication locus, annotated as a DNA integrase (gp26), is an essential DNA primase, likely required for DNA replication. The only non-essential exception in this locus is gp28, a small gene downstream of the DNA primase that appears non-essential, tolerating 6 insertions across 366 bp, with high read depth (Figure 6A, left inset). This gene has no predicted function.

Downstream of the DNA replication locus, YuA has a collection of non-essential genes. gp42 and gp45 are non-essential genes that have insertions at most TA sites within those genes and many reads per insertion (Figure 6A, right inset). Well-sequenced insertions are also identified in gp28, gp46, and in gp47 (although this gene only has a single TA site). Consistent with this, random plaque picking and phage whole genome sequencing revealed viable phage mutants with insertions in gp42, 45, 46, 47. gp43, 48, and 9 comprise other putative non-essential genes that tolerate insertions for which sequencing reads are detected. However, gp44 is notable in this locus as a putative fitness conferring gene as it has just 1 insertion (3 TA sites total) that sequence to very low read depth (Figure 6A, right inset). This stands out relative to its gene neighbors that have very well sequenced insertions. gp44 has a high confidence prediction as a GGDEF-domain containing diguanylate cyclase, an enzyme family that generates cyclic-di-GMP. This suggests that although this locus is generally dispensable in PAO1 and may antagonize phage defense elements in other strains (i.e. gp45 is predicted to be an ArdB-like restriction enzyme inhibitor), that the gp44 DGC is required for replication in PAO1. Downstream of this locus are genes encoding the structure proteins of the virion, starting with gp50 (small terminase). From that gene to the end of the genome, gp77, we generally see very few well sequenced insertions suggesting all virion proteins are essential.

With YuA, we show that a phage transposon library can be constructed using a non-CRISPR based system and can be done on a phage with modified thymidine. We also showed that for a small phage such as YuA, many of the genes are essential or fitness conferring, with only 6/77 annotated genes (7%) of the genome being clearly dispensable.

Discussion

Phage functional genomics is beginning to match the scale and sophistication of bacterial genetics, but tools remain limited, particularly for large and complex phage genomes. Here, we establish a generalizable phage transposon mutagenesis and phage TnSeq platform that (i) maps essential and fitness-conferring genes genome-wide, (ii) generates a library of knockouts in nearly all non-essential genes, and (iii) identifies safe harbor loci suitable for insertion and expression of transgenes.

By directing mariner transposition into the ΦKZ nucleus and selecting for insertions using an anti-CRISPR and CRISPR-Cas13a, we demonstrate a rapid and scalable workflow for phage genome mutagenesis. Because mariner is widely used in bacteria, requires no host-specific factors, and inserts semi-randomly, we anticipate that this approach will be broadly adaptable across bacteria–phage pairs. The transposon donor can be easily reconfigured, such as by removing or modifying promoters or terminators, to profile therapeutic phages and identify optimal genomic contexts for stable expression of payload genes such as anti-defense factors or therapeutic cargo.

Genome-wide knockout approaches that generate permanent changes in phage genomes, such as the one presented here, are complementary to recently developed protein knockdown approaches based on antisense oligomers (ASOs) and dCas13d-mediated translation interference (CRISPRi-ART)^42,64. These perturbation tools are powerful for titrating essential gene products that cannot be deleted outright. Tn insertions are not ideal for fine-tuned modulation of essential gene expression and are limited in some cases by polar effects. Moving forward, we anticipate that the recent explosion in defense and anti-defense system discovery means that one can transpose, and select for transposed plaque formation, with any phage of interest. We demonstrate feasibility for the idea by extending our selection for mutagenized phage to the END nuclease, paired with a transposed anti-END gene. This opens the door for facile transposition of modified phages. Supporting the potential for phage transposon mutagenesis to operate broadly, two recent preprints have also implemented this method in E. coli and Serratia/Prodigiosinella phages to great success^65,66. mariner or Tn5 mutagenesis transposition approaches were used with similar CRISPR-Cas/anti-CRISPR selection methods.

One of the most striking outcomes of this work is the revelation that a majority of ΦKZ genes are dispensable under standard laboratory conditions. Some of these non-essential genes encode tail (e.g. gp298) or capsid-associated proteins (e.g. gp244) with host- or temperature-specific roles, inhibitors (e.g. gp302) or activators (e.g. gp241) of bacterial defense systems, and a recombinase (Nlp1/gp152) that becomes critical in the presence of a specific nuclease. The ease with which transposon mutagenesis can interrogate this abundant pool of non-essential genes sets the stage for systematic dissection of phage accessory and anti-defense functions.

Smaller phages, like YuA, likely encode a lower fraction of non-essential genes than ΦKZ, yet the total number of non-essential genes per phage may still be modest enough that a complete collection of knockout mutants could be arrayed rapidly. This makes it feasible to screen for defense inhibitors, defense activators, host-range factors, and conditionally essential genes across multiple hosts, temperatures, and selective pressures. As interest grows in phage diversity, basic biology, genome engineering, and therapeutic applications, we anticipate that phage Tn mutagenesis will provide a broadly useful resource and methodology. The essentiality maps, mutant libraries, and design principles described here should be transferable to other phage systems, enabling the community to move from sequence catalogs to functional insight at scale.

Limitations

Phage TnSeq, as implemented here, requires a robust defense system that is difficult for phages to escape and a cognate anti-defense factor that can be encoded on the transposon. These requirements may constrain applicability for some phage–host systems. Nevertheless, the programmability of CRISPR-Cas systems and the growing catalog of anti-CRISPR proteins⁶⁷, along with many other defense and anti-defense genes, provide a flexible toolbox for adapting selection schemes.

A second, inherent limitation of insertional mutagenesis in prokaryotes and phages is polar effects: insertions can appear to reveal essential genes when they instead disrupt expression of neighboring essential genes. We show that the high density of TSSs in ΦKZ alleviates many potential polar effects, but we also identify clear cases where phage TnSeq misclassifies non-essential homing endonucleases as essential due to operon structure. In some regions, we identified insertion direction biases suggesting that perhaps transcription readthrough can alleviate or induce confounders. Using Tn donors that lack transcriptional terminators or that incorporate unidirectional terminators could reduce some polar effects. Consequently, putative essential genes of interest should be confirmed by complementary approaches such as targeted knockdowns or targeted deletions.

Resource Availability

Lead contact:

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Joseph Bondy-Denomy (joseph.bondy-denomy@ucsf.edu).

Materials availability:

Plasmids and strains generated in this study are available from the lead contact upon request. Plasmids have been deposited in Addgene (https://www.addgene.org/browse/article/28263768/).

Data and code availability:

Insertion sequencing data and ΦKZ whole-genome sequencing data will be deposited in a public repository and accession numbers will be provided upon publication. Custom analysis scripts are available from the lead contact upon request.

STAR Methods

EXPERIMENTAL MODEL AND SUBJECT DETAILS

Bacterial strains

Experiments were performed in Pseudomonas aeruginosa PAO1, PAO1 derivatives, clinical and environmental isolates. Bacteria were grown in LB broth at 37 °C with aeration. Where appropriate, media were supplemented with gentamicin (50 μg/mL) or carbenicillin (250 μg/mL).

Phages

The jumbo phage ΦKZ and derivative mutants generated in this study were propagated on P. aeruginosa PAO1 or specified host strains. For host-range and conditional screens, a panel of wild P. aeruginosa isolates, including clinical isolate PA1032 and environmental isolate Env201, were used as indicated.

METHOD DETAILS

Construction of mariner transposase expression strains

The mariner transposase gene was amplified from pBTK30 (kindly provided by the Lory lab) and assembled into a Tn7 plasmid carrying sfCherry–Nlp1. For construction of FLAG-tagged transposase, plasmid p30T expressing sfCherry-Mariner–Nlp1 was linearized with primers oAWC088/oAWC021 (Table S1; all primers used are in Table S1), and the resulting backbone was ligated with gBlock oAWC093 using NEBuilder HiFi DNA Assembly (New England Biolabs).

All plasmids were sequence-verified (Quintara Biosciences) and introduced into P. aeruginosa PAO1 by electroporation together with pTNS3 for integration at the Tn7 site⁶⁸. Transformants were selected on LB agar containing 50 μg/mL gentamicin. The Flp recombinase plasmid pFLP was then introduced into these strains to excise the gentamicin resistance cassette. Colonies were counter-selected on sucrose, and loss of gentamicin resistance was confirmed by replica plating.

Construction of acrVIA1 transposon donors (V1–V3)

For construction of acrVIA1 transposon donors V1 and V2, p30HERDT (abbreviated p30T) was linearized with primers oAWC047/oAWC048. gBlocks encoding distinct constructs (oAWC122 and oAWC123 for donors V1 and V2, respectively) were ligated into the linearized backbone using HiFi Assembly and transformed into E. coli XL1-Blue.

For the V3 donor used in pooled TnSeq experiments, the p20T plasmid was linearized instead of p30T and ligated with gBlocks oAWC146 and oAWC147 using HiFi Assembly. All constructs were confirmed by Sanger sequencing and subsequently transformed into PAO1 strains carrying the mariner transposase, as described above.

Construction of pHERD30T expression plasmids

To generate gene inserts in p30T, the plasmid backbone was first amplified with primers WX_016 and WX_017. Genes of interest were amplified with the appropriate primer pairs and assembled into the backbone using HiFi Assembly. Constructs were cloned into XL1-Blue, sequence-verified, and introduced into PAO1 by electroporation.

Construction of CRISPR-Cas13 guide plasmids

For cloning of Cas13 guide arrays for single or dual targeting, the plasmid backbone was amplified with primers oAWC004/oAWC024. Complementary oligonucleotides (Table S1) encoding the desired crRNAs (100 μM each) were annealed in Duplex Buffer A and ligated into the backbone using HiFi Assembly.

For Cas13 guide plasmids used to target candidate essential or non-essential genes in ΦKZ (to isolate escaper mutants), the Cas13 backbone plasmid was amplified with primers that incorporated the crRNA sequence directly. Products were assembled with HiFi and cloned into XL1-Blue. All guide plasmids were confirmed by sequencing and transformed into PAO1 or PAO1 attTn7::cas13aLse as appropriate.

Generation of transposed ΦKZ arrays

Transposed ΦKZ phages were selected on a Cas13 strain expressing dual crRNAs targeting orf120 and orf55. Individual plaques were picked into SM buffer. In total, 1,500 plaques were collected from 20 plates in a 96-well format, with plate corners reserved for wild-type ΦKZ controls and two central wells containing no phage. The phage array was replicated onto lawns of different P. aeruginosa isolates and incubated overnight at 37 °C to assess host-range and conditional phenotypes. Note that replica plating was not carried out at 18 °C due to limited bacterial and phage growth at low temperature conditions; instead, manual spotting of phage mutants using 2 μL of phages was carried out.

Cas13a-mediated isolation of ΦKZ deletion mutants

High-titer ΦKZ stocks were plated on induced PAO1 attTn7::cas13aLse strains carrying targeting guides as described above. Escaper plaques were picked and propagated on PAO1 attTn7::cas13aLse expressing the same guides. Regions spanning the putative deletion were amplified using primers flanking the targeted locus and sequenced to confirm the nature of the deletions or point mutations.

Deletion of individual ΦKZ genes by acrVIA1 replacement

Homology-directed constructs for gene replacement with acrVIA1 were synthesized (Twist Bioscience) with compatible overhangs and cloned into NheI-digested p30T using NEB HiFi Gibson assembly. Constructs were selected with gentamicin and confirmed by sequencing.

Phage recombinants were generated and selected similarly to a previous report¹⁹. PAO1 strains harboring these plasmids were grown overnight, sub-cultured 1:100 into fresh LB, and grown at 37 °C to an OD₆₀₀ of 0.6. Cultures were diluted 1:100, and ~10³ PFU of ΦKZ were added for overnight infection. The following day, cultures were centrifuged at 10,000 × g for 2 min, and chloroform was added to the supernatant at a 1:8 (v/v) ratio. After ≥15 min at room temperature, lysates were centrifuged again to clarify.

For selection of recombinants, 50 μL of lysate were mixed with 300 μL of PAO1 attTn7::cas13aLse + p30T crRNA^55/120 in the presence of 0.3% arabinose and 1 mM IPTG and plated in top agar. Resistant plaques were picked and propagated on the same strain under the same induction conditions. Candidate recombinant phages were screened by PCR using primers outside the homology arms to confirm acrVIA1 insertion into the phage genome and to exclude plasmid contamination. An internal PCR targeting the wild-type locus was used to monitor residual WT phage. Serial passages and screening were repeated until no WT amplicon was detectable.

Transposition infections for pooled TnSeq

Transposition strains and control strains were grown overnight from glycerol stocks in LB supplemented with either 50 μg/mL gentamicin (V1, V2 donors) or 250 μg/mL carbenicillin (V3 donor) at 37 °C with shaking. Overnight cultures were subcultured 1:100 into LB containing 1 mM IPTG (to induce transposase expression), the appropriate antibiotic, and 10 mM Mg²⁺ and grown for 2 h at 37 °C with shaking.

Log-phase cultures were adjusted to OD₆₀₀ = 0.5 in 3 mL volumes. Bacteria were infected with 10 μL of 2 × 10⁴ PFU/mL ΦKZ at an MOI of approximately 2.22 × 10⁻⁷ and incubated for 16 h at 37 °C with shaking. The next day, 2 mL of each culture were centrifuged at 16,000 × g for 2 min to pellet unlysed cells. Supernatants were transferred to new tubes, mixed with 100 μL chloroform, and incubated at room temperature with shaking for 10 min. After centrifugation at 10,000 × g for 2 min, supernatants were transferred to fresh tubes, extracted once more with 100 μL chloroform, clarified, and stored at 4 °C.

For the genome-wide pooled TnSeq library, 10 independent transposition reactions done using PAO1 attTn7:mariner-sfCherry-nlp1 + p20T-V3 were performed and processed as described above.

Cas13a counter-selection of transposed phages

Transposition output phages were first titered by serial dilution spot assays on Cas13 non-targeting and dual-targeting strains, as well as in the presence of jukAB, to verify identity and exclude contaminants. Typical output titers were ~4 × 10¹⁰ PFU/mL.

To enrich for transposed phages, PAO1 attTn7::cas13aLse + p30T^nlp1 crRNA^55/120 strain was grown to OD₆₀₀ ~1.8 in 3 mL LB and infected at an approximate MOI of 5 using 400 μL of the ~1 × 10¹⁰ PFU/mL transposition output. Counter-selection cultures contained 1 mM IPTG, 0.3% arabinose, 50 μg/mL gentamicin, and 10 mM Mg²⁺ and were incubated overnight at 37 °C with shaking. Phages were harvested the following day by chloroform treatment as described above and stored at 4 °C. Serial dilution spot assays and full-plate infections on the counter-selection strain confirmed enrichment and Cas13 resistance, and titers were typically ~2 × 10¹⁰ PFU/mL.

For pooled TnSeq, enriched phages from the ten reactions were combined in equal PFU amounts.

Phage Tn-Seq in conditional backgrounds

10⁷ - 10⁸ PFU of the pooled ΦKZ transposon library that has been enriched for AcrVIA1 was passed through PAO1 (control) and either wild isolates or PAO1 with different defence systems cloned in. The post selection pool was then collected and phages were extracted using chloroform. Genomic DNA was then extracted and prepared for sequencing as below. From the Biotradis output, frequency of each gene represented in the pool was calculated based on number of reads within a gene divided by total reads. This was then plotted as a dot plot in GraphPad Prism.

Construction of YuA library

The p20T-V3 was first modified to carry the anti-END inhibitor instead of AcrVIA1 using primers WX_380/WX_381 (to amplify the plasmid backbone) and WX_382/WX_383 (to amplify the anti-END inhibitor).

To construct the YuA library, YuA (100–1000 PFU) was input into a log-phase culture of PAO1 attTn7:mariner-FLAG + p20T-V3^anti-END and grown overnight in the presence of carbenicillin and IPTG. Ten independent reactions were carried out. Phages were then extracted using chloroform and the titer was checked. The selection strain, PAO1 attTn7:END was subcultured into 5ml and grown to OD₆₀₀ 1.8 – 2. Transposed phages were then added at MOI ~1; ten independent selections were carried out overnight. The enriched phages were then extracted using chloroform and the titer was checked on both PAO1 and PAO1 attTn7:END.

For analysis of essentiality, we calculated read density based on reads per total TA site within the gene. This takes into consideration smaller genes which have a low number of possible insertion sites but have high read counts, suggesting that these mutants are fit in the absence of the inserted gene, such as gp47 (one unique site) and larger genes that could have a higher read count but spread over multiple insertion sites (such as gp71, 28 unique sites). We were not able to assay gp59 as there were no TA sites in the gene.

Library preparation for transposon insertion sequencing

Phage DNA isolation

Genomic DNA was purified from enriched phage lysates using SDS/proteinase K treatment followed by column purification. Briefly, 50–200 μL of high-titer phage lysate were mixed 1:1 with lysis buffer (10 mM Tris-HCl pH 8.0, 10 mM EDTA, 100 μg/mL proteinase K, 100 μg/mL RNase A, 0.5% SDS), incubated at 37 °C for 30 min and at 55 °C for 30 min, and then processed with the DNA Clean & Concentrator kit (Zymo Research). DNA concentration was measured by spectrophotometry.

Illumina library construction and nested PCR

Transposon junction libraries were prepared using the NEBNext Ultra II FS DNA Library Prep Kit for Illumina with modifications to enrich for transposon–genome junctions. To generate indexed i7 adaptors, oAWC190 was pre-annealed separately with oAWC191, 192, 194, or 195 in 10 mM NaCl at 200 μM. For each pair, 12 μL of each primer were mixed, heated to 95 °C for 5 min, and cooled to 12 °C at 0.1 °C/sec. Annealed oligos were diluted to 15 μM and stored at −20 °C.

For fragmentation, 1,000 ng of phage DNA in 26 μL 10 mM Tris-HCl pH 8.0 were mixed with 7 μL Ultra II FS Reaction Buffer and 2 μL Ultra II FS Enzyme Mix, vortexed, and incubated at 37 °C for 24 min to generate 100–250 bp fragments, followed by 65 °C for 30 min and hold at 4 °C. I7 adapters were ligated by adding 30 μL Ultra II Ligation Master Mix, 1 μL Ligation Enhancer, and 2.5 μL of 15 μM pre-annealed adaptor, followed by incubation at 20 °C for 15 min.

For size selection, the reaction was brought to 100 μL with 10 mM Tris-HCl pH 8.0, and 40 μL AMPure XP beads (Beckman Coulter) were added. After 10 min at room temperature and 5 min on a magnetic rack, the supernatant was transferred to a new tube. An additional 20 μL AMPure XP beads were added, incubated 10 min, and the beads washed twice with 200 μL 80% ethanol. Beads were air-dried for 3–5 min and eluted in 16 μL water. Fifteen microliters of eluate containing size-selected DNA were carried forward.

To amplify transposon junctions, 15 μL of size-selected DNA were mixed with 25 μL NEBNext Ultra II Q5 Master Mix, 5 μL 10 μM oAWC196, and 5 μL 10 μM oAWC224 (total 50 μL). PCR was performed with: 98 °C for 30 s; 12 cycles of 98 °C for 10 s, 65 °C for 75 s; 65°C for 5 min; hold at 12 °C. PCR products were cleaned with AMPure XP beads (45 μL beads per 50 μL reaction), washed twice with 80% ethanol, and eluted in 16 μL water (15 μL carried forward).

A second PCR to further enrich junctions and attach i5 adaptors was performed using 15 μL eluate and 35 μL of master mix (25 μL Q5 Master Mix, 5 μL 10 μM oAWC199, and 5 μL of 10 μM oAWC200, 201, 202, or 203). Products were cleaned with AMPure XP beads as above, eluted in 33 μL water, and 30 μL were retained.

Library quality was assessed using the Qubit 1× dsDNA HS Assay (expected 20–50 ng/μL) and Agilent D1000 ScreenTape (expected size ~300–350 bp).

Illumina sequencing

Libraries were sequenced on an Illumina MiSeq i100 instrument using the MiSeq i100 Series 25M Reagent Kit (100-cycle; 20126567). Indexed libraries were pooled and diluted with water to ~4 nM (Qubit-based). A 10× loading concentration of 0.8 nM in 100 μL Resuspension Buffer was prepared according to the manufacturer’s instructions.

A separate indexed sample for ΦKZ whole-genome sequencing was prepared at 0.8 nM and included as a diversity spike-in. The loading mix consisted of 75 μL pooled library and 25 μL spike-in. Runs were configured with 8 cycles for index 1, 8 cycles for index 2, 100 cycles for read 1, and 0 cycles for read 2. Quality control and demultiplexing were performed in Illumina BaseSpace, and FASTQ files for each index were exported for downstream analysis.

Insertion sequencing analysis

Demultiplexed FASTQ reads were filtered for those containing the transposon inverted repeat sequence marking the insertion junction (GACCGGGGACTTATCAGCCAACCTGTTA). Retained reads were trimmed to keep the last 10 nt of the inverted repeat (CAACCTGTTA) plus 35 nt of downstream genomic sequence.

Trimmed reads were mapped to the reverse complement of the ΦKZ reference genome (NC_004629) using the TraDIS toolkit (bacteria_tradis). Insertion sites supported by ≤2 reads were removed before essentiality analysis. Gene-level insertion statistics were obtained using tradis_gene_insert_sites on both raw and filtered insertion datasets; for filtered data, insertions within the first and last 10% of each annotated ORF were excluded. Fitness-conferring genes were annotated using the tradis_essentiality.R script. TraDIS outputs were further processed and visualized in R. For ΦKZ, insertion index was used to determine essentiality. However, for YuA, as the number of TA sites were low and some genes had no/only one TA sites, the read density was used to determine essentiality instead. Read density is determined by reads divided by the number of TA sites within the orf.

Plaque assays

Unless otherwise noted, plaque assays were performed at 37 °C. Strains were grown overnight in LB at 37 °C (EcoRI experiments were typically ~13 h). For spot assays, 100 μL of overnight culture were mixed with 3–5 mL of 0.4% LB top agar supplemented with Mg²⁺ and poured onto LB + Mg plates. After solidification, ΦKZ or derivative phages were serially diluted and 2 μL of each dilution were spotted on the lawn. Plates were incubated overnight at 37 °C.

For complementation assays with the ΦKZ nlp1::acrVIA1 phage, 150 μL of overnight culture were mixed with 10 μL of phage at the desired dilution and 3.5 mL of 0.35% top agar with Mg²⁺, poured on LB + Mg plates containing 50 μg/mL gentamicin, and incubated overnight at 30 °C.

For cold-sensitivity assays, plates were incubated for 48 h at 18 °C. A water-filled plate was placed on top of assay plates to provide weight and limit evaporation.

qPCR for CBASS experiments

Strains were grown overnight at 37 °C and diluted 1:100 into fresh LB the next day. Cultures were grown for 6 h, and ΦKZ was added at MOI ~0.5. At each time point, 100 μL of culture were harvested, washed once with 1× PBS, and pellets were snap-frozen on dry ice and stored at −20 °C.

Genomic DNA was extracted using the Zymo Genomic DNA kit. Approximately 1 ng of DNA was used as input for qPCR with primer pairs 15F/15R (phage) and rpoD_F/rpoD_R (bacterial chromosomal control). Phage DNA levels were first normalized to bacterial DNA (ΔCt), and relative phage DNA abundance was then calculated relative to t = 0.

Adsorption and replication assays

Overnight cultures were diluted 1:100 into fresh LB and grown at 37 °C for ~2.5 h to OD₆₀₀ ≈ 0.6. ΦKZ (10⁴ PFU) was added to each culture. For adsorption measurements, 100 μL of culture were sampled at t = 15 min, mixed with chloroform to lyse cells, and titrated to determine unadsorbed phage. For replication assays, 100 μL of culture were collected at t = 2 h, lysed with chloroform, and titrated on PAO1 to quantify total phage yield.

Competition assays

Mutant phages carrying acrVIA1 were first verified to form plaques on both a non-targeting strain (PAO1 attTn7::cas13aLse + p30T^empty) and a Cas13 single-guide strain PAO1 attTn7::cas13aLse + p30T crRNA¹²⁰, while wild-type ΦKZ formed plaques only on PAO1 attTn7::cas13aLse + p30T^empty. Full-plate infections of mutant and wild-type phage on PAO1 attTn7::cas13aLse + p30T^empty were used to generate starting stocks.

Overnight cultures of PAO1 attTn7::cas13aLse with and without guide RNA and were grown at 37°C. For competition, PAO1 attTn7::cas13aLse + p30T^empty was subcultured 1:100 and grown to OD₆₀₀ ≈ 0.6. Mutant and wild-type phage were mixed at an approximate 1:1 ratio and added to PAO1 attTn7::cas13aLse + p30T^empty at a total MOI of 1 in 150 μL volumes in 96-well plates. Infection was monitored over 4 h in a plate reader.

The initial ratio of mutant to wild-type phage was determined at t = 0 by full-plate infection of PAO1 attTn7::cas13aLse + p30T^empty (total phage) and PAO1 attTn7::cas13aLse + p30T crRNA¹²⁰(mutant-only) at appropriate dilutions for plaque counting. After 4 h, chloroform was added to each well and incubated for ~15 min, and lysates were clarified by centrifugation. The final ratio of mutant to wild-type phage was determined by titration on PAO1 attTn7::cas13aLse + p30T^empty and PAO1 attTn7::cas13aLse + p30T crRNA¹²⁰.

The Competitive Index (CI) was calculated as:

$$\mathit{\text{Competitive Index}}(CI)=\frac{\left(\frac{\mathit{\text{Mutant phage PFU at t}}=0}{\mathit{\text{Total phage PFU at t}}=0}\right)}{\left(\frac{\mathit{\text{Mutant phage PFU at t}}=4h}{\mathit{\text{Total phage PFU at t}}=4h}\right)}$$

A CI of 1 indicates no change in mutant frequency during the assay (mutant fitness equivalent to wild type), CI < 1 indicates a competitive disadvantage, and CI > 1 indicates a competitive advantage.

QUANTIFICATION AND STATISTICAL ANALYSIS

For TnSeq analysis, insertions supported by ≤2 reads were excluded and the first and last 10% of each ORF were masked prior to essentiality calling (results in Table S2). Essential and fitness-conferring genes were identified using the TraDIS essentiality pipeline (tradis_essentiality.R). For calculation of total insertions in orfs, all reads were considered (results in Table S3). Graphical analysis related to TnSeq were all performed in R.

Plaque assays, adsorption/replication assays, qPCR experiments, and competition assays were typically performed with at least three biological replicates unless otherwise indicated in figure legends. Statistical analysis (e.g., calculation of mean, standard deviation, and, where appropriate, significance testing) was performed in GraphPad Prism; details of tests are indicated in the corresponding figure legends.

Supplementary Material

1

Figure S1. Construction and validation of Cas13 dual-guide counter-selection system, related to Figure 1.

(A) Plaque assays of wild-type ΦKZ, ΦKZ phuZ::acrVIA1, and single-guide escapers on targeting and non-targeting Cas13 strains.

(B) Escape frequencies of single versus dual Cas13a crRNAs targeting ΦKZ.

(C) Agarose gel PCR verification of acrVIA1 insertions in transposed plaques. Low-transposition control pools are shown for comparison.

(D) Plaque assays of transposition reactions across targeting and non-targeting hosts.

(E) Quantification of transposition efficiency as a function of input ΦKZ titer (pfu/mL).

Figure S2. Characterization of individual transposition reactions before and after enrichment, related to Figure 2.

(A) Plaque assays comparing ten independent transposition reactions across selection conditions: no selection, dual Cas13a targeting, and JukAB. Wild-type ΦKZ is shown as control.

(B) Full-plate infections (20 μL input) of each transposition reaction.

(C) Post-enrichment plaque assays confirming selective recovery of transposed ΦKZ under Cas13a counter-selection.

Figure S3. Rank-ordered distribution of ΦKZ gene fitness scores, related to Figure 2.

Cumulative ranking of ΦKZ genes based on transposon insertion frequency and essentiality scoring.

Figure S4. Genome-wide map of KTN4 gene essentiality, as determined by insertion sequencing (TnSeq) analysis, related to Figure 2..

Schematic summary of gene essentiality across the KTN4 genome, as determined by insertion sequencing (TnSeq) analysis. Dark red represents low insertion index (shown in log10 scale), while dark blue represents a high insertion index.

Figure S5. Validation of gene essentiality by CRISPR-Cas13a escaper isolation, related to Figures 2 and 3.

(A) Schematic overview of genes targeted by Cas13. Protein function, essentiality, and conservation across jumbo phages and Chimalliviridae are indicated.

(B) Plaque assays showing targeting of wild-type ΦKZ and ΦKZ orf105::acrVIA1 (acrVIA1 control) with crRNAs directed genes at encoding injected structural proteins. The Cas13a dual guide construct targeting orf55 and orf120 are included as control.

(C) Schematic representation of deletion events identified in Cas13 escaper genomes. The annotations start (s), dual (d) and cleavage (c) refer to the position of guide targeting. Numbers shown indicate the base number within the orf that had mutations. Top: virion, bottom: non-virion. Plaque assay showing ΦKZ orf75::Tn compared to wild-type ΦKZ on PAO1 is also included.

Figure S6. ΦKZ mutant analysis, related to Figure 4.

(A) Example images of plaques from arrayed ΦKZ mutants pinned onto P. aeruginosa lawns

(B) Plaque assay of plaques of interest identified at 18 °C

(C) Quantification of in vivo EcoRI restriction of wild-type and ΦKZ Δnlp1. F8 phage is included as an EcoRI-sensitive control unrelated to ΦKZ. Data shown as mean ± s.d.

(D) Full-plate infections comparing ΦKZ and ΦKZ Δnlp1 with or without nlp1 complementation.

(E) Competitive fitness of ΦKZ mutants carrying acrVIA1 insertions in different genes, compared with wild-type phage.

Figure S7. Genome-wide map of YuA essentiality at 37 °C, related to Figure 6. Schematic summary of gene essentiality across the YuA genome, as determined by insertion sequencing (TnSeq) analysis.

2

Table S1. List of oligonucleotides used in this paper.

3

Table S2. Raw TraDIS output for gene essentiality after trimming, related to Figure 3.

4

Table S3. Raw TraDIS output for all insertions across the ΦKZ genome, related to Figure 3.

Key resource table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Bacterial and virus strains
PAO1 p30Tempty	Alan Davidson Lab
PAO1 p30T244-FLAG	Fossati, Mozumdar et al., 202333
PAO1 attTn7::empty	This paper
PAO1 attTn7::jukAB	Yuping et al., 202552
PAO1 attTn7::CBASS III-C	Cao et al., 202360
PAO1 attTn7::CBASS III-C + p30Tempty	This paper
PAO1 attTn7::CBASS III-C + p30Tnlp1	This paper
PAO1 attTn7::CBASS III-C + p30Tacb2	This paper
PAO1 attTn7::empty + p30Tempty	This paper
PAO1 attTn7::empty + p30Tnlp1	This paper
PAO1 attTn7::empty + p30Tacb2	Cao et al., 2023
PAO1 p30TGb135	This paper
PAO1 p30TGb136	This paper
PAO1 p30TGb144	This paper
PAO1 attTn7::mariner-sfCherry-nlp1	This paper
PAO1 attTn7::mariner-sfCherry-nlp1 + p30T-V1	This paper
PAO1 + p30T-V1	This paper
PAO1 attTn7::mariner-sfCherry-nlp1 + p30T-V2	This paper
PAO1 + p30T-V2	This paper
PAO1 attTn7::mariner-FLAG + p20T-V3	This paper
PAO1 attTn7::mariner-sfCherry-nlp1 + p20T-V3	This paper
PAO1 + p20T-V3	This paper
PAO1 attTn7::cas13aLse + p30Tnlp1 crRNA55/120	This paper
PAO1 attTn7::cas13aLse + p30Tempty	Guan et al., 202219
PAO1 attTn7::cas13aLse + p30T crRNA120	Guan et al., 2022
PAO1 attTn7::cas13aLse + p30T crRNA55	Guan et al., 2022
PAO1 attTn7::cas13aLse + p30T crRNA55/120	This paper
PAO1 attTn7::cas13aLse + p30T crRNA86start	This paper
PAO1 attTn7::cas13aLse + p30T crRNA95cleav	This paper
PAO1 attTn7::cas13aLse + p30T crRNA97start	This paper
PAO1 attTn7::cas13aLse + p30T crRNA97cleav	This paper
PAO1 attTn7::cas13aLse + p30T crRNA99start	This paper
PAO1 attTn7::cas13aLse + p30T crRNA119start	This paper
PAO1 attTn7::cas13aLse + p30T crRNA162dual	This paper
PAO1 attTn7::cas13aLse + p30T crRNA163dual	This paper
PAO1 attTn7::cas13aLse + p30T crRNA177start	This paper
PAO1 attTn7::cas13aLse + p30T crRNA303start	This paper
PAO1 attTn7::ENDPaCF1	Yee et al., 202620
PAO1 attTn7::mariner-FLAG + p20T-V3antiEND	This paper
Env201	This paper
PA1032	This paper
EnvBC13	This paper
EnvBC15	This paper
MRSN369569	This paper
UCSF E16	This paper
Phages
ΦKZ	Alan Davidson Lab	AF399011
ΦKZ orf298::Tn	This paper
ΦKZ orf302::Tn	This paper
ΦKZ_orf152::Tn	This paper
ΦKZ nlp1::acr	This paper
ΦKZ orf2::acr	This paper
ΦKZ orf95::acr	This paper
ΦKZ orf93::acr	Guan et al., 2022
ΦKZ orf105::acr	This paper
ΦKZ orf244::acr	This paper
ΦKZorf184::acr	Li et al., 202451
ΦKZorf241::acr	Yuping et al., 2025
JBD67	Alan Davidson Lab	NC_042135.1
JBD67Δacb2	Huiting et al., 202369
YuA	Alan Davidson Lab	NC_010116.1
Biological samples
Chemicals, peptides, and recombinant proteins
Gentamicin	RPI
Carbenicillin	BPBio
IPTG	Fisher
Chloroform	Alta Aesar
Proteinase K	Thermofisher Scientific
RNaseA	Omega
Critical commercial assays
Luna qPCR master mix	New England Biolabs	M3003X
NEBNext Ultra II FS DNA Library Prep Kit	New England Biolabs	E7805
NEBuilder HiFi DNA Assembly Master mix	New England Biolabs	E2621L
DNA Clean and Concentrator Kit	Zymo Research	D4034
AMPure XP Beads	Beckman Coulter	A63880
Qubit 1x dsDNA HS Assay Kit	Thermofisher Scientific	Q33230
Agilent D1000 ScreenTape and Reagents	Agilent Technologies	5067–5582 5067–5583 5067–5602 5067–5586
Deposited data
Experimental models: Cell lines
Experimental models: Organisms/strains
Pseudomonas aeruginosa strain PAO1
Oligonucleotides
See Table S1 for details
Recombinant DNA
p30Tempty	Alan Davidson Lab
p30T crRNA120	Guan et al., 2022
p30T crRNA55	Guan et al., 2022
p30T crRNA55/120	This paper
p30T crRNA86start	This paper
p30T crRNA95cleav	This paper
p30T crRNA97start	This paper
p30T crRNA97cleav	This paper
p30T crRNA99start	This paper
p30T crRNA119start	This paper
p30T crRNA162dual	This paper
p30T crRNA163dual	This paper
p30T crRNA177start	This paper
p30T crRNA303start	This paper
p30T V1	This paper
p30T V2	This paper
p20T V3	This paper
p20T V3-antiEND	This paper
p30T244-FLAG	Fossati, Mozumdar et al., 2023
p30Tnlp1	Huiting et al., 2023
p30Tacb2	Huiting et al., 2023
p30TGb135	This paper
p30TGb136	This paper
p30TGb144	This paper
attTn7::mariner-sfCherry-nlp1	This paper
attTn7::mariner-FLAG	This paper
TNS3	Choi et al., 200768
Software and algorithms
GraphPad PRISM	GraphPad Software
RStudio	R Foundation
TraDIS Toolkit	Barquist et al., 201670
Other

Proposed highlights:

• Transposon mutagenesis with anti-CRISPR selection enables genome-wide phage-TnSeq.

• Model jumbo phage ΦKZ TnSeq reveals ~110 fitness-conferring and ~261 dispensable genes.

• Arrayed mutant screens and TnSeq libraries uncover conditionally essential genes.

• Phage TnSeq extended to base-modified phage using CRISPR-free selection