Introduction

Antibiotics have played an important role in the prevention and treatment of various bacterial infections since their discovery. However, the emergence and rapid spread of antibiotic-resistant (ABR) pathogens are now posing severe threats to human health as they hinder antibiotic efficacy and have led to limitations on available clinical treatment options. A report commissioned by the British government predicted that antimicrobial resistance could cause approximately 10 million deaths and result in a total economic cost of $100 trillion by 2050 if not tackled1. It has also been reported that as of 2019, resistant bacterial infections have directly caused 1.27 million deaths worldwide, while indirectly associated with 5 million more deaths2. The fact that ABR bacteria emerged and spread faster than the discovery of new antibiotics in this decade3 further aggravates the growing burden of antimicrobial resistance (AMR). The emergence of AMR and the lack of antibiotics to treat the emerging ABR bacteria indicate an urgent need to develop novel alternative strategies to combat AMR4.

Bacteriophages (in short, phages), co-discovered by Felix d’Herelle and Frederick Twort5, are viruses that infect and replicate within bacterial host cells. They are ubiquitous and are the most abundant microbiological entities that exist on earth6. Phages have high specificity toward their bacterial host strains, yet do not infect human or animal cells. Due to their bacteriolytic activity, phage therapy against bacterial infections has been introduced since 19197,8,9, but was largely abandoned after the discovery and effective use of antibiotics. It was not until the emergence of ABR bacterial strains that the interest in utilizing phages as therapeutic agents resurged and phage therapy is now considered as one of the effective alternatives to antimicrobial agents10. Many recent pre-clinical and clinical studies have assessed the efficacy and safety of phage therapy and manifested phage therapy as a convincing approach to tackle the emergence of multidrug-resistant pathogens11. In addition to using natural phages for therapeutic purposes, genetically modified phages, such as phages that carry exogenous antimicrobial peptides, CRISPR-Cas system, and sRNA12,13,14,15,16, as well as phage particles that carry genetic elements other than their own genome17,18, have also been developed and successfully employed to treat refractory bacterial infections.

In an effort to address the issue of AMR, our laboratory has previously exploited the specificity of phage-host interaction and the bactericidal activity of CRISPR-Cas13a system. This has led to the development of phage-based antibacterial agents, known as AB-capsids. Utilizing the SaPI (Staphylococcus aureus pathogenicity island) mobile genetic element to load CRISPR-Cas13a into a S. aureus phage capsid, we demonstrated the ability of these AB-capsids to selectively eliminate methicillin-resistant S. aureus (MRSA) in a sequence-specific manner15,19,20. SaPI is a well-studied phage-inducible chromosomal islands (PICI) harbored by S. aureus20. This genetic element possesses the capability to co-opt phage capsids for packaging of its own genetic elements21, facilitating their mobilization between bacterial species. Any foreign DNA fragment carrying SaPI-derived packaging signal can therefore be efficiently encapsulated into phage capsids, allowing for subsequent transduction to other bacteria15,22. However, the cloning of CRISPR-Cas13a onto the SaPI element necessitates allelic exchange, a process that is proven laborious and time-consuming. Furthermore, SaPIs carry inherent virulence potential, raising concerns about the potential horizontal transfer of superantigens and virulence factors during SaPI transduction, thereby complicating clinical management.

Another noteworthy genetic element demonstrated to commandeer phages for self-packaging is the phagemid, typically characterized as a plasmid carrying a packaging site derived from a phage23,24. Essentially, phages recognize the packaging signal present on the phagemid and encapsulate it, resulting in the generation of non-replicative phage particles14. The advantages of employing phagemids as vehicles for mobilizing genetic cargo lie in their ease of modification, their capacity to carry/transduce foreign genes of interest without the risk of unwanted horizontal gene transfer, and their inability to replicate, making them a safer mode of phage therapy25. Nevertheless, this system requires refinement for practical application, addressing aspects such as 1. complexity in phagemid designing (overcoming challenges in cloning and programming CRISPR-Cas13a), 2. host ranges (as the transduction of programmed CRISPR-Cas13a is limited to bacterial hosts susceptible to the helper phage exploited for phagemid packaging), and 3. the efficiency of AB-capsid packaging and transduction (issues like low titer, impaired performance, or contamination could potentially hinder therapeutic applications).

In this study, we adopted the phagemid packaging system, replacing our previous SaPI packaging system, to load a sequence-specific bactericidal CRISPR-Cas13a into the capsids of a broad-host-range phage, referred to as Tan2 in this study, with genome size of about 42.5 kb, to address the problem of MRSA26. To affirm the robustness, we undertook the following measures: 1) packaged CRISPR-Cas13a-loaded phagemids into Tan2 phage capsid, a wide host range temperate phage isolated in-lab (manuscript in preparation); 2) optimized the packaging efficiency of this system; 3) produced pure AB-capsids without natural phage contamination; and 4) evaluated their specific bactericidal activity against clinical S. aureus strains. We envisage our phagemid approach is advantageous, with regard to its ease of construction, transduction efficiency, purity and feasibility as an alternative to antibiotics in the future.

Results

Targeted bactericidal activity of phagemid-based SA-CapsidCas13a against MRSA

We employed a phagemid-based capsid packaging strategy (Fig. 1a) to generate phage capsids carrying programmed CRISPR-Cas13a system, termed as antimicrobial capsid (AB-Capsid) owing to its bactericidal capabilities. Two types of AB-capsids, namely SA-CapsidCas13a::CpR_mecA (containing a CRISPR-Cas13a system targeting methicillin resistance gene mecA) and SA-CapsidCas13a::CpR_nontargeting (non-targeting control), were constructed in the preliminary phase to confirm the relevance of phagemid packaging system in generation of the sequence-specific antimicrobials, AB-capsids.

Fig. 1: Generation of phagemid-based bactericidal capsids carrying mecA-targeting CRISPR-Cas13a.
figure 1

a Schematic representation of the generation of the phagemid-based SA-capsid. S. aureus cells with an integrated prophage in their chromosome are transformed with the phagemid carrying the phage packaging site. Post-transformation, mitomycin C induction facilitates the excision of the prophage genome and initiates the translation of phage structural proteins, leading to phage assembly. This process results in the phagemid being loaded into capsids, as the phagemid carries packaging signals that can be recognized by phage, thereby packaging them and yielding phagemid-based SA-capsid. b Gene structure of the Tan2 phage packaging site, consisting of genes encoding the proteins RinB, RinA, TerS, and TerL. c Schematic representation of the generation of bactericidal phagemid-based SA-Capsid Cas13a:: CpR_mecA. The phagemid carrying mecA-targeting spacer, CRISPR-Cas13a, phage packaging site genes, and chloramphenicol-resistant (CpR) gene is packaged into capsids, resulting in phagemid-based SA-Capsid Cas13a::CpR_mecA. d Schematic depiction of targeted killing assay by soft agar overlay method. The packaged capsid filtrate and the host S. aureus bacterial culture were mixed along with TSB soft top agar, poured on the top of chloramphenicol TSA plate, and incubated for 12 h at 37 °C to examine the transduction efficacy and any sequence-specific bactericidal activity. e MRSA strain USA300 or MSSA strain USA300ΔmecA were independently treated with mecA-targeting and non-targeting capsids. The representative plate images (left panel) and the colony counts of bacterial grown on these plates (right panel) are shown. Note that the mecA-targeting SA-CapsidCas13a killed the target cells MRSA that carries target gene mecA. Mean log CFU/ml of surviving target strain USA300 transformed with mecA-targeting/non-targeting capsids were compared with that of non-target strain USA300ΔmecA via Student’s t test (n = 3). ***p  <  0.001; ****p  <  0.0001. Error bars represent standard deviation of the mean.

Full size image

For this purpose, we constructed an E. coli-S. aureus shuttle phagemid vectors that include sequences encoding phage packaging signals [rinB, rinA, terS and terL spanning about 2.81 kb (Fig.1b, Supplementary Note 1)]. These vectors were designed to carry either mecA-targeting LshCas13a (pLK12_mecA) or non-targeting LshCas13a (pLK12_null). The chloramphenicol-resistant (CpR) gene serves as the selection marker for this E. coli-S. aureus shuttle vector (Fig. 1c). The constructed phagemid was transformed into a prophage-integrated S. aureus strain RN4220 as a host cell (RN4220ΦTan2 WT). Subsequently, the host cell is chemically induced with mitomycin C to produce respective AB-Capsids, SA-CapsidCas13a::CpR_mecA and SA-CapsidCas13a::CpR_nontargeting. In essence, the induction of the host cell with mitomycin C initiates the synthetic processes of the phage. This involves excising the prophage genome and initiating the translation of phage structural proteins, leading to phage assembly. The transformed phagemids are subsequently packaged into phage capsids due to the presence of phage packaging sites, resulting in the generation of phage capsids loaded with LshCas13a. The genetic materials carried on the phagemid, including LshCas13a, can then be delivered into target bacterial cells through standard phage transduction.

We assessed the bactericidal efficacy of the two AB-Capsids, SA-CapsidCas13a::CpR_mecA and SA-CapsidCas13a::CpR_nontargeting, against the target strain MRSA USA300 carrying mecA and its isogenic mecA deletion mutant (USA300ΔmecA) (Fig. 1d). SA-CapsidCas13a::CpR_mecA effectively inhibited USA300 but not USA300ΔmecA, demonstrating mecA-specific targeting. As shown in Fig. 1e, SA-CapsidCas13a::CpR_mecA almost completely inhibited the growth of the target host USA300 carrying the target gene mecA but did not inhibit the growth of the mecA-deleted mutant USA300ΔmecA. In contrast, their respective non-targeting controls, SA-CapsidCas13a::CpR_nontargeting, almost did not exhibit an inhibitory effect against both USA300 and USA300ΔmecA. These results align with the notion that SA-CapsidCas13a::CpR_mecA successfully transduced mecA-targeted CRISPR-Cas13a into the target bacteria, and the subsequent expression of CRISPR-Cas13a within the cell executed sequence-specific killing. This underscores its capability as a targeted bactericidal system, ensuring no harm to nonrelevant bacteria, as we discussed previously15.

The influence of phagemid copy number on packaging efficiency

After verifying the feasibility of phagemid packaging system in generation of AB-capsids with targeted bactericidal activity against MRSA, we sought to improve the yield of AB-capsids, recognizing their potential importance for future clinical applications. Our initial focus was on evaluating the impact of phagemid copy number on the packaging efficiency of AB-capsids. To explore this aspect, we constructed four phagemids, pLK7-623, pLK11-KAT, pLK7-608, and pLK12_1, carrying different staphylococcal replication origins (ori): repC 187C-A, repC Δ183-362, repM, and repB, respectively 27,28. Additionally, all these phagemids feature a chloramphenicol-resistant (CpR) marker and a kanamycin-resistant (KmR) marker for selection in S. aureus and E. coli, respectively. The phagemids are also incorporated with packaging site genes of phage Tan2. The genome lengths of pLK7-623, pLK11-KAT, pLK7-608, and pLK12-1 were 7051 bp, 6412 bp, 6871 bp, and 7161 bp, respectively. The copy numbers of these phagemids were measured using real-time PCR and normalized against chromosome copy number of the host cell RN4220ΦTan2 WT. Our results showed that the phagemids copy numbers ranged from 3 to 45 copies (Fig. 2). Correspondingly, their transduction/packaging efficiency varied from 1.73 to 7.99 log TFU/ml (Fig. 2). Our result revealed a positive correlation between phagemid copy number and transduction efficiency with a correlation coefficient of 0.86 (p < 0.05). Notably, pLK7-623, with ori repC 187 C-A, boasting the highest copy number in this study, exhibited the highest level of transduction activity. Conversely, repC Δ183-362, governing the lowest copy number in pLK7-608, contributed to the lowest packaging efficacy of the generated AB capsids. It’s worth noting that no product were detected in the negative control (NC), which is a strain without any phagemid, as confirmed by qPCR.

Fig. 2: Relationship between phagemid copy number and transduction efficiency (log TFU/ml).
figure 2

The x-axis indicates copy number comparison per chromosome for phagemids with different staphylococcal replication origins (ori) (n = at least 4): repC 187 C-A (pLK7-623), repB (pLK12_1), repM (pLK11-KAT), and repC Δ183-362 (pLK7-608). NC represents the negative control without phagemid. The y-axis indicates the transduction efficiency of phagemids (in log TFU/ml) with different ori. The correlation coefficient (r2) calculated using Pearson’s correlation was determined to be 0.86 (p < 0.05). Error bars represent standard deviation of the mean.

Full size image

Efficient packaging is not reliant solely on terS

The inadvertent production of natural phage Tan2 during the generation of phagemid-based AB-capsids raises concerns, as it indicates the presence of contaminating particles with non-sequence-specific bactericidal activity. To ensure the generation of a pure collection of CRISPR-Cas13a-loaded antibacterial capsids (AB-capsids), we initially knocked out the expected phage packaging site terS from the Tan2 prophage lysogenized in the host RN4220ΦTan2 WT. We anticipated that the contaminating natural Tan2 phage would be eliminated if phagemid-based AB-capsids were induced from the terS-knockout mutant RN4220ΦTan2ΔTerS, where the packaging signals on phage DNA would be diminished (Fig. 3a).

Fig. 3: Impact of TerS packaging site on transduction efficiency and natural phage contamination.
figure 3

a Schematic representation illustrating the reduction in natural phage contamination during the packaging of phagemids when the terS gene is knocked out in the integrated prophage within host cells. b Quantification of transduced colony-forming unit (TFU) and plaque-forming unit (PFU) of capsids packaged in host cells, RN4220ΦTan2 WT, and RN4220ΦTan2ΔTerS, transformed with either of the two phagemids, pLK12::TerS and pLK12::TerL-TerS-RinA-RinB, or without phagemid transformation (n = 3). White bars and black bars represent log PFU/ml and log TFU/ml, respectively. Mean differences in log TFU/ml or log PFU/ml were determined using Student’s t test. *p  <  0.05; **p  <  0.01. Error bars represent standard deviation of the mean.

Full size image

As a proof of concept, two different phagemids, pLK12::TerL-TerS-RinA-RinB (carrying a putative complete packaging gene set of 7161 bp long) and pLK12::TerS (carrying only terS, length of 4967 bp), were transformed into RN4220ΦTan2 WT or its terS-knockout mutant RN4220ΦTan2ΔTerS. After mitomycin C induction, we collected the lysates, and assessed the packaging efficiencies of AB-capsids and potentially contaminating natural Tan2 phages, indicated by transduced colony-forming units (TFU) and plaque-forming units (PFU), respectively. Results showed that there was no observable difference in the amount of AB-capsid production and Tan2 phage contamination in the RN4220ΦTan2WT, whether using pLK12::TerL-TerS-RinA-RinB or pLK12::TerS (Fig. 3b). Instead, mitomycin C induction of terS-deleted mutant RN4220ΦTan2ΔTerS transformed with pLK12::TerS generated a significantly lower number of natural Tan2 phages (a 4-orders reduction; p < 0.01) compared to that of phagemid pLK12::TerL-TerS-RinA-RinB. However, this is accompanied by a drop of about 2 orders in the generated AB-capsids (p < 0.01) compared to using the same phagemid pLK12::TerS in the original host cell, RN4220ΦTan2 WT (Fig. 3b). These results indicate that efficient packaging is not reliant solely on terS, and the putative complete packaging gene set need to be intensively studied.

terL-terS-rinA-rinB gene set is essential for efficient packaging

Upon learning that both the deletion of the packaging site gene terS on prophage and the cloning of different combinations of packaging site genes onto phagemid resulted in distinctive numbers of AB-capsids/contaminating Tan2 phages, we speculated an interplay between packaging site genes residing on prophages and those cloned onto phagemids in packaging efficiency. To clarify, we knocked out different sets of packaging site genes sequentially to generate three additional host cells: RN4220ΦTan2ΔTerL-TerS, RN4220ΦTan2ΔTerL-TerS-RinA, and RN4220ΦTan2ΔTerL-TerS-RinA-RinB, in addition to RN4220ΦTan2 WT and RN4220ΦTan2ΔTerS (Fig. 4a). Meanwhile, apart from pLK12::TerS and pLK12::TerL-TerS-RinA-RinB, three phagemids were constructed: pLK12::TerL-TerS-RinA, pLK12::TerL-TerS, and pLK12::Empty (vector without packaging site gene). All phagemids carried same ori for S. aureus and E. coli, as well as the specific resistant markers as aforementioned (Fig. 4b). This provides us with a total of 25 combinations of host cells/phagemids, serving as potential systems for AB-capsid generation.

Fig. 4: Elimination of contaminating natural phage during phagemid-capsid packaging.
figure 4

a Schematic maps depicting wild-type or various knockout mutants (packaging site genes deletion from the prophages of host cells) constructed in this study. b Schematic maps illustrating the packaging site genes retained on each constructed phagemid. c Comparison of PFU or TFU of SA-Capsids packaged by five phagemids (pLK12::TerS, pLK12::TerL-TerS, pLK12::TerL-TerS-RinA, pLK12::TerL-TerS-RinA-RinB, and pLK12::Empty as a negative control), transformed into five different host cells (RN4220ΦTan2 WT, RN4220ΦTan2ΔTerS, RN4220ΦTan2ΔTerL-TerS, RN4220ΦTan2ΔTerL-TerS-RinA, RN4220ΦTan2ΔTerL-TerS-RinA-RinB). This resulted in a total of twenty-five combinations. White and black bars represent log PFU/ml and log TFU/ml, respectively (n = 3). n.d. not detected. Error bars represent standard deviation of the mean.

Full size image

The results revealed that certain combinations (Fig. 4c) did not produce either plaques or colonies. These included phagemid pLK12::Empty packaged using host cell RN4220ΦTan2ΔTerS, phagemids pLK12::Empty and pLK12::TerS packaged using host cell RN4220ΦTan2ΔTerL-TerS, as well as phagemids pLK12::Empty, pLK12::TerS and pLK12::TerL-TerS packaged using either host cell RN4220ΦTan2ΔTerL-TerS-RinA or RN4220ΦTan2ΔTerL-TerS-RinA-RinB. This suggests a deficiency in packaging signals for AB-capsid production. Conversely, pure AB-capsids were produced (with no plaque formation) in four specific combinations: phagemid pLK12::TerL-TerS packaged using host cell RN4220ΦTan2ΔTerL-TerS, phagemid pLK12::TerL-TerS-RinA packaged using host cell RN4220ΦTan2ΔTerL-TerS-RinA, and the two phagemids pLK12::TerL-TerS-RinA-RinB and pLK12::TerL-TerS-RinA packaged using host cell RN4220ΦTan2ΔTerL-TerS-RinA-RinB (Fig. 4c). Notably, the transduction efficiency, as indicated by TFU, decreased by approximately two orders of magnitude for these combinations compared to the highest TFU achieved (e.g. combination of pLK12::TerL-TerS-RinA-RinB and RN4220ΦTan2ΔTerS) in this assay. Nonetheless, phagemids containing the whole terL-terS-rinA-rinB gene set consistently demonstrated higher transduction efficiency across various host cell types. This dependency of transduction efficacy on packaging site emphases the significance of a comprehensive gene set of packaging signals for controlled and efficient transduction, indicating that the terL-terS-rinA-rinB gene set is essential for optimal phagemid packaging.

Sequence-specific killing of S. aureus using phagemid-based SA-CapsidCas13a

As shown in the aforementioned findings, packaging of pLK12::TerL-TerS-RinA-RinB using host cell RN4220ΦTan2ΔTerS-TerL resulted in relatively lower level of wild-type phage contamination (Fig. 4c). However, the introduction of CRISPR-Cas13a uniformly reduced AB-capsid production by approximately one order of magnitude across all combinations29, hindering a clear assessment of its bactericidal efficacy. Despite this global reduction, the use of phagemid pLK12::TerL-TerS-RinA-RinB in conjunction with host cells RN4220ΦTan2WT exhibited a milder decrease in production, positioning it as a promising candidate for downstream study. Our primary focus was on developing a phagemid-based SA-CapsidCas13a system targeting various genes of S. aureus. The success of this approach would pave the way for selectively eliminating targeted bacteria carrying specific genes. To achieve this, we initially integrated the chloramphenicol-resistance (CpR) gene cat from pIMAY, functional in both S. aureus and E. coli, and the tetracycline resistance (TetR) gene tetM from SaPIbov2bap::tet30 into the pLK12::TerL-TerS-RinA-RinB phagemid. This integration ensures the broad applicability of the generated SA-CapsidCas13a across a diverse spectrum of clinical S. aureus strains, considering that the examined 512 clinical isolates rarely possess both of these genes simultaneously (manuscript in preparation).

Subsequently, we designed nine spacer sequences, each 25-bp long, targeting distinct antibiotic resistance genes of S. aureus. These spacers were individually integrated into the aforementioned phagemid, along with CRISPR-Cas13a, resulting in the development of a phagemid-based SA-CapsidCas13a::CpR-TetR_X series, where X represents the targeted antibiotic resistance genes. The specified genes encompass aph(2’), aadD, aph(3’), aac(6’), ermB, fusC, mphC, mecA, and tetK. Notably, aph(2’), aph(3’), and aac(6’) are associated with resistance to aminoglycoside antibiotics (such as gentamicin and tobramycin); aadD to aminoglycoside antibiotic (such as streptomycin); ermB to macrolide antibiotics (such as erythromycin); fusC to fusidic acid; mphC to macrolide antibiotics; mecA to beta-lactam antibiotics; and tetK to tetracycline. The non-targeting AB-capsid (SA-CapsidCas13a::CpR-TetR_Non-T), lacking any spacers, was utilized as a negative control. Subsequent spot assays were conducted to assess the sequence-specific bacterial killing activities against four clinical MRSA isolates carrying antibiotic resistance genes that encompass all nine targeted genes, and laboratory MSSA strain RN4220 lacking any of the nine genes (Fig. 5a). As depicted in Fig. 5b–f, all nine SA-CapsidCas13a::CpR-TetR_X constructs specifically eradicated the corresponding strains carrying the target genes, while the growth of RN4220 lacking any target gene remained unaffected. The non-targeting AB-capsid control demonstrated no bacterial killing effect against all tested strains.

Fig. 5: Sequence-specific bacterial killing by phagemid-based SA-CapsidCas13a.
figure 5

a Schematic representation of the experimental method to assess capsid-mediated bactericidal activity (spot test). Various clinical isolates of S. aureus were added to TSB top agar and poured onto the TSA-Cp plate. After the top agar solidified, tenfold serial dilutions of phagemid-based SA-CapsidCas13a::CpR;TetR_X (where X represents target genes) were spotted on different S. aureus clinical isolate bacterial lawns and incubated for 12 hours at 37 °C to visualize bactericidal activity. b–e Infection experiments were conducted against clinical isolates of S. aureus JMUB1278, JMUB4958, JMUB3007, and JMUB4975 using SA-CapsidCas13a::CpR;TetR_X (X represents the targeted antibiotic resistance genes). The test results were assessed by observing bacterial growth on TSB top agar plates supplemented with Cp. Anibiotic resistance gene presence in clinical isolates is highlighted in red font. Each gene listed on the left of plate scans signifies the presence of target spacers in the phagemid-based SA-CapsidCas13a::CpR;TetR_X corresponding to the genes. Non-T represents non-targeting capsids that lack spacers and serve as control. f Spot assay was performed against S. aureus laboratory strain RN4220 as a control that does not carry any of the resistance genes that is targeted in this work. No growth arrested by antimicrobial capsid was observed in this strain.

Full size image

A whole-genome sequencing analysis of the tested clinical strains unveiled variable resistance genes within their genomes. Strain JMUB1278 carries aac(2’), aac(6’), and mecA genes, strain JMUB4958 carries fusC, strain JMUB3007 carries aadD and mecA, while strain JMUB4975 carries aph(3’), ermB, mecA, and tetK. As expected, phagemid-based SA-CapsidCas13a selectively inhibited the growth of bacteria carrying the target gene to which the spacer complemented (highlighted in red). The growth of bacteria not targeted by SA-CapsidCas13a remained unaffected. This outcome not only validates the efficacy of our approach but also emphasizes its specificity in addressing antibiotic resistance at the genetic level. The unaffected growth of bacteria not targeted by SA-CapsidCas13a further reinforces the precision and discriminative nature of our strategy. The sequence-specific bacterial killing nature of SA-CapsidCas13a, targeting any desired genes, holds significant promise for tailored therapeutic interventions.

Discussion

The exploration of phage therapy as a compassionate and effective alternative to combat ABR infections has gained significant momentum. Numerous pre-clinical and clinical studies have showcased positive outcomes, sparking global interest in this therapeutic approach30. However, persistent challenges associated with phage therapy necessitate further consideration and intervention. A primary concern revolves around biosafety issues linked to phage use27. While phages are incapable of infecting human cells, their potential to proliferate within host bacteria during therapy raises concerns about unintentional gene transfer and the emergence of new species among colonized bacteria through viral transduction. Another drawback of natural phage application lies in their indiscriminate killing activity against target bacterial strains. Phages possess the ability to sterilize all bacteria within their host range, a trait that, while advantageous in clinical applications, could pose a risk to the homeostasis of the human microbiome28. To address these concerns, the exploration of programmable non-replicative phages with an inherent bactericidal activity emerges as an ideal option. Our laboratory has previously succeeded in generating a non-replicative antibacterial agent loaded with CRISPR-Cas13a targeting mecA using SaPI packaging system15. However, this system is relatively laborious and time consuming as it involved genetic modification at the chromosomal level. As an improvement to our AB-Capsids synthetic method, a phagemid-based system is introduced to encapsulate CRISPR-Cas13a within phage capsids for targeted delivery into S. aureus.

Phagemids, characterized as plasmids carrying phage-derived components, have previously demonstrated their effectiveness for DNA delivery into both bacteria and mammalian cells14,31,32,33. In this study, a phagemid was constructed by cloning phage-encoded packaging signals (terL, terS, rinA and rinB)34,35 onto a plasmid vector. rinA/rinB are transcriptional regulators that activate the transcription of prophage packaging genes35, terS recognizes and cleaves DNA concatemers, while terL translocates phage DNA into the capsids36. The phagemid system served as a vehicle to deliver a bactericidal genetic cargo (programmed CRISPR-Cas13a) into the capsids of a candidate broad-host-range phage isolated in-house (manuscript in preparation). This ensures that our developed SA-CapsidCas13a can be applied across a diverse range of clinically relevant S. aureus strains. This packaging method demonstrated remarkable efficacy in generating non-replicative phage-like antibacterials with minimal impurities. A particularly intriguing aspect of our findings is the non-replicative nature of SA-CapsidCas13a packaged using the phagemid system. This characteristic not only alleviates safety concerns associated with traditional phage therapy but also ensures ease in phage selection, yielding a well-purified product and facilitating direct and accurate analyses. The implications of this advancement extend to the enhanced safety profile and practical utility of our engineered phage-based system for targeted bacterial eradication. One key aspect explored in this study was the optimization of the packaging efficiency of AB-Capsids, considering the typically high number of phages/phage-like particles needed during treatment courses. A 12-hour dosing of 109 PFU/ml phages has been suggested based on a clinical trial evaluating Myoviridae bacteriophages (AB-SA01) as adjunctive therapy in patients infected with S. aureus37. This underscores the importance of establishing an efficient packaging system for the generation of our SA-CapsidCas13a. The investigation into the influence of phagemid copy number revealed a positive correlation between copy number and transduction efficiency. Phagemid with the highest copy number, pLK7-623, exhibited the highest transduction activity (Fig. 2). This finding suggests that increasing the copy number of the phagemid can enhance the yield of AB-Capsids, a crucial factor for the scalability and feasibility of future clinical applications. Additionally, the interplay between packaging site genes on prophages and those cloned onto phagemids was further explored to enhance packaging efficiency, in reference to a previously reported study that demonstrated the role of packaging signals in producing pure preparations of phage-based transducing particles38. The results revealed specific combinations of host cells and phagemids that generated pure AB-Capsids without contamination (Fig. 4). This underscores the importance of a comprehensive set of packaging signals for controlled and efficient transduction. Specific combinations, such as RN4220ΦTan2ΔTerS-TerL host cells and pLK12::TerL-TerS-RinA-RinB phagemid, exhibited a milder decrease in AB-capsids yield with a relatively lower level of wild-type phage contamination, positioning them as promising candidates for further exploration29.

The applicability of the optimized phagemid system was demonstrated in the successful development of SA-CapsidCas13a series for sequence-specific killing of S. aureus strains carrying antibiotic resistance genes. The specificity of the approach was validated by the selective eradication of targeted strains while leaving non-targeted strains unaffected (Fig. 5). This precision in addressing antibiotic resistance at the genetic level holds great promise for tailored therapeutic interventions, allowing for the elimination of ABR strains while preserving non-targeted bacterial populations.

Our optimized phagemid system offers versatility in design, programmable editing, and robust production capabilities. Moreover, the well-engineered phagemid-based SA-CapsidCas13a demonstrated the ability to sequence-specifically recognize various staphylococcal genes, leading to targeted killing of S. aureus. Beyond its efficacy, our strategy holds promise for diverse applications, including clinical therapy against ABR bacterial infections. Furthermore, it opens avenues for the development of phage or capsid-based vaccines for antigen delivery, the establishment of a capsid library for phage therapy cocktails, targeting toxin genes, and creating simple and cost-effective gene detection or SCCmec typing kits for bacterial infection diagnosis. The versatility and effectiveness of our approach showcase its potential in addressing complex challenges associated with bacterial infections and antibiotic resistance.

While our study presents promising advancements in the development of phagemid-based SA-CapsidCas13a, certain limitations need to be considered for future refinements. Notable challenges, such as phage resistance in bacterial cells, cross-contaminating impurities in lysates, and concentration methods to enrich titers (attributable to the inherent limitation of the burst size of temperate phages), are currently under investigation in ongoing studies. Particularly, phage contamination in our AB-capsids system is hindering a clear presentation of the system’s efficacy. Although the effect of phage-mediated killing in this study has been normalized using nontargeting AB-capsids, the presence of phage contamination needs to be resolved to enhance the clinical applicability of our AB-capsids system. Another consideration is the need for a more extensive evaluation of the off-target effects and unintended consequences of the phagemid-based SA-CapsidCas13a. A comprehensive analysis of the impact on non-target bacterial species and potential alterations in the microbiome is crucial for assessing the broader ecological implications of our engineered system. Furthermore, while our study focused on the bactericidal efficacy against S. aureus, expanding the scope to encompass a broader spectrum of bacterial pathogens would provide a more comprehensive understanding of the versatility and limitations of the phagemid-based SA-CapsidCas13a. This could involve testing against diverse bacterial strains and assessing the potential cross-reactivity of the designed spacers. To improve the clinical translatability of our approach, additional investigations into the pharmacokinetics and pharmacodynamics of phagemid-based SA-CapsidCas13a are also needed.

In conclusion, our study successfully established a robust phagemid-based packaging system for the creation of sequence-specific bactericidal agents, termed the phagemid-based SA-CapsidCas13a series. Meticulous optimization of the phagemid copy number resulted in high transduction efficiency. Importantly, our findings confirmed a positive correlation between the packaging efficiency of phagemid-based SA-CapsidCas13a and the phagemid copy number, intricately dependent on the origin of replication of the phagemid. Addressing a critical concern, we implemented measures to eliminate contamination from natural phages during the production of phagemid-based SA-CapsidCas13a. This was achieved by strategically knocking out essential packaging site genes from the integrated prophage in the host cells, ensuring the purity of our designed phagemid-based SA-CapsidCas13a.

Methods

Bacterial strains and culture conditions

S. aureus strains were grown at 37 °C in tryptic soy broth (TSB; BD Difco, USA)/tryptic soy agar (TSA; BD Difco, USA), with 10 µg/ml of chloramphenicol (Cp) added when appropriate. E. coli strains were grown at 37 °C in Luria-Bertani (LB) broth (BD Difco, USA)/LB agar (BD Difco, USA), with the addition of following antibiotics when appropriate: 30 µg/ml of kanamycin (Km) and/or 10 µg/ml of Cp. All S. aureus and E. coli strains used in this study were listed in Supplementary Table 1.

Lysogenizing RN4220 with phage Tan2

Phage Tan2 was lysogenized into S. aureus strain RN4220 for the construction of host cells used for phagemid packaging. First, 100 µl of RN4220 grown to an OD600 of 0.5 in TSB was added to 4 ml soft agar (TSB, 0.75% agarose, 1 mM CaCl2) and the mixture was overlaid on TSA bottom agar plate. Phage Tan2 suspension was then serially diluted 10-fold (to 10−3) with SM buffer [50 mM Tris-HCl (pH 7.5), 0.1 M NaCl, 7 mM MgSO4·7H2O, 0.01% gelatin] and 100 µl of each dilution was spotted onto the bacterial lawn. After drying, the plate was incubated overnight at 37 °C. Surviving colonies in the spotted area were randomly selected and the lysogenization of phage Tan2 into host RN4220 (RN4220ΦTan2 WT) was confirmed by PCR amplification using primer sets Tan2 F5-s/RN4220 Tan2-as (3’-integration site) and RN4220 Tan2-s/Tan2 F1-as (5’-integration site).

Construction of pac deletion mutant library

Integrated prophage Tan2 (RN4220ΦTan2 WT) with single terS deletion, or combined deletion of terL/terS, terL/terS/rinA, or terL/terS/rinA/rinB were generated through allelic exchange using E. coli/staphylococcal temperature-sensitive plasmid pIMAY39 /pKOR140. For the construction of RN4220ΦTan2ΔTerS, 500–1000 bp upstream and downstream flanking sequences of phage Tan2 terS were PCR-amplified with primer sets attB2-Tan2TerS-Up-F/sacII-Tan2TerS-Up-R and sacII-Tan2TerS-Down-F/attB1-Tan2TerS-Down-R using KOD FX Neo (Toyobo, Japan). The two amplified fragments were digested with sacII, ligated using Ligation high ver. 2 (Toyobo, Japan) and purified from agarose gel. Then, PCR amplification of pKOR1 and the ligated products with primer sets pKOR1 7190-F4/pKOR1 4549-R and Lig-F3/Lig-R4, respectively, were performed using KOD FX Neo. The two DNA fragments were ligated using In-Fusion HD Cloning Kit (TaKaRa, Japan), generating pKOR1-KO-terS. For the construction of RN4220ΦTan2ΔTerL-TerS, RN4220ΦTan2ΔTerL-TerS-RinA, and RN4220ΦTan2ΔTerL-TerS-RinA-RinB, phage Tan2 packaging signals (rinB, rinA, terS and terL) together with its 500–1000 bp upstream and downstream flanking sequences were amplified with 2 different primer sets: Tan2 23519-F2/Tan2 28540-R2 (for pKOR1) and Tan2 23519-F3/Tan2 28540-R3 (for pIMAY) using KOD FX Neo. pKOR1 and pIMAY were also PCR-amplified using the same polymerase with primer set pKOR1 7280-F5/pKOR1 4509-R2 and pIMAY 2858-F/pIMAY 2769-R, respectively. The two phage Tan2 packaging site gene fragments were cloned onto their corresponding plasmids using In-Fusion HD Cloning Kit to first generate pKOR1_Tan2pac and pIMAY_Tan2pac. Following that, inverse PCR excluding terL/terS, terL/terS/rinA, and terL/terS/rinA/rinB, respectively, were performed and the amplified fragments were self-ligated using In-Fusion HD Cloning Kit to finally generate pIMAY_ΔterL/terS, pIMAY_ΔterL/terS/rinA, and pKOR1_ΔterL/terS/rinA/rinB.

All the constructed knockout plasmids were transformed into E. coli DC10B and selected on LB agar supplemented with 10 µg/ml Cp (for pIMAY-derived plasmids) and 100 µg/ml ampicillin (for pKOR1-derived plasmids). The plasmids were then extracted and verified by Sanger sequencing. Following that, sequence-verified knockout plasmids were electroporated into S. aureus RN4220ΦTan2 WT as previously described41, and the transformants were grown on TSA plate supplemented with 10 µg/ml Cp (for both pKOR1- and pIMAY-derived plasmids) at either 28 °C (for pIMAY-derived plasmids) or 30 °C (for pKOR1-derived plasmids). Single crossover was performed by growing the transformed cells on TSA plate supplemented with 10 µg/ml Cp at non-permissive temperature of 37 °C (for pIMAY-derived plasmids) or 43 °C (for pKOR1-derived plasmids). Finally, double crossover was performed by growing the cells on TSA plate supplemented with 1 μg/ml anhydrotetracycline at 28 °C (for pIMAY-derived plasmids) or 30 °C (for pKOR1-derived plasmids). The construction of RN4220ΦTan2ΔTerS, RN4220ΦTan2ΔTerL-TerS, RN4220ΦTan2ΔTerL-TerS-RinA, and RN4220ΦTan2ΔTerL-TerS-RinA-RinB were further confirmed by PCR and Sanger sequencing.

Construction of phagemids

Constructing pLK12 series

To confirm the applicability of phagemid packaging system in generating AB-capsids, pLK628_mecA and pLK628_null (S. aureus-E. coli shuttle vector cloned with CRISPR-Cas13a with and without mecA-targeting spacers, respectively) were first constructed. These CRISPR-Cas13a-loaded shuttle vectors were composed of two DNA fragments: CRISPR-Cas13a_mecA/CRISPR-Cas13a_null, antibiotic selection marker KmR (aphA-3) and ori of E. coli amplified with primer set pKLC21-2-F/pKLC21-2-R from respective plasmid DNA templates pKLC21_mecA/pKLC21_null; and antibiotic selection marker CpR (cat) and ori of S. aureus amplified from pKLC14 with primer set pKLC14-K-F/pKLC14-K-R. The amplified fragments were assembled using NEBuilder® HiFi DNA Assembly (New England Biolabs, USA) and mecA-targeting (pLK628_mecA) or control (pLK628_null) S. aureus-E. coli shuttle vectors were generated. Next, phage Tan2 packaging site rinB-rinA-terS-terL were amplified with primer set phi1150-1-FW/phi1150-1-RV. The amplified fragment was inserted into pLK628_mecA and pLK628_null, generating phagemids pLK12_mecA and pLK12_null respectively.

To better understand the effects of individual packaging signal on phagemids-based AB-capsids production and natural phage contamination, we constructed a series of pLK12 phagemids carrying different combinations of packaging site genes in the absence of LshCas13a. These phagemids were pLK12_0 (pLK12::Empty), pLK12_1 (pLK12::TerL-TerS-RinA-RinB), pLK12_2 (pLK12::TerS), pLK12_5 (pLK12::TerL-TerS) and pLK12_6 (pLK12::TerL-TerS-RinA). All vectors (plasmid/phagemid) constucted in this study were listed in Supplementary Table 2, while the oligonucleotides used were listed in Supplementary Table 3.

Insertion of different origin of replication (ori)

To evaluate the effect of different ori (associated with plasmid copy number) on phagemid packaging efficiencies, different ori(s) of S. aureus were individually cloned onto phagemid pLK12_1 (pLK12::TerL-TerS-RinA-RinB) in replacement of the native ori repB. The insertion of ori from pT181 cop623, pT181 cop608 and pKAT onto pLK12_1 generated pLK7-623(repC 187C-A), pLK7-608(repC Δ183-362), and pLK11-KAT(repM), respectively30,42.

Insertion of TetM gene and non-targeting LshCas13a

The modification process involved replacing the KmR gene of E. coli on pLK12_1 with the CpR gene cat from the plasmid pIMAY, which functions in both E. coli and S. aureus, resulting in the creation of pLK14. Additionally, the TetR gene tetM43 was inserted into pLK14, leading to the generation of pLK16. The non-targeting CRISPR-Cas13a was then introduced to pLK16, resulting in pLK19_null. A BsaI restriction site was incorporated into pLK19_null to enable the creation of various gene-targeting phagemids. This was achieved by inserting 25-bp spacer sequences at the restriction enzyme cut site, allowing flexibility for future requirements. Circular maps of phagemids pLK12_1, pLK16 and pLK19_null are included in Supplementary Fig. 1.

Insertion of spacers for targeted killing

The construction of the CRISPR-Cas13a system on phagemid pLK19_null, targeting nine drug-resistant genes (aph(2”), aadD, aph(3’), aac(6’), ermB, fusC, mphC, mecA, and tetK), with a non-targeting spacer as a control, involved several steps. Custom-designed 85-mer oligo DNAs, comprising a 25-mer spacer sequence (crRNA) complementary to the conserved region of each target gene, flanked with 30-mer nucleotides at the 5’- and 3’-ends corresponding to the consensus sequences at the point of insertion on phagemid, were created. Subsequently, pLK19_null underwent treatment with the restriction enzyme BsaI-HF, followed by purification with gel electrophoresis. The purified product was then ligated with each of the 85-mer oligo DNAs using NEBuilder HiFi DNA Assembly, resulting in the generation of the pLK19_Cas-X series, where X represents target genes.

Generation of phagemid-based SA-CapsidCas13a

The constructed phagemid vectors were transformed into corresponding host cells according to different experimental purposes: (1) pLK12_mecA and pLK12_null into S. aureus RN4220ΦTan2 WT to confirm the feasibility of loading CRISPR-Cas13a onto phage capsids using phagemid packaging system; (2) pLK12_1, pLK7-623(repC 187C-A), pLK7-608(repC Δ183-362), and pLK11-KAT(repM) into RN4220ΦTan2 WT to evaluate the influence of phagemid copy number on packaging efficiency; (3) pLK12_1 and pLK12_2 into RN4220ΦTan2 WT or RN4220ΦTan2ΔTerS to study the efficacy of prophage terS knockout in minimizing natural phage contamination; (4) pLK12_0, pLK12_1, pLK12_2, pLK12_5 and pLK12_6 into RN4220ΦTan2 WT and its mutant derivatives (RN4220ΦTan2ΔTerL-TerS-RinA-RinB, RN4220ΦTan2ΔTerL-TerS-RinA, RN4220ΦTan2ΔTerL-TerS, RN4220ΦTan2ΔTerS) to understand the interplay between packaging site genes knockout from prophage and cloned onto phagemids; and (5) pLK19_Cas-X series into RN4220ΦTan2 WT to generate phagemid-based SA-CapsidCas13a::CpR-TetR_X series for the evaluation of sequence-specific killing activity of our AB-Capsid.

All constructed phagemids were transformed by electroporation using ELEPO21 electroporator (Nepa Gene, Japan). The following pulse parameters were used: Poring Pulse (voltage: 1,800 V, pulse length: 2.5 ms, pulse interval: 50 ms, number of pulses: 1, Polarity: +); Transfer Pulse (voltage: 100 V, pulse length: 99 ms, pulse interval: 50 ms, number of pulses: 5, polarity: +/−)41. Resulting transformants were recovered at a temperature permissive for plasmid replication (30 °C) for 5 hours and then plated on TSA plates supplemented with Cp. Successful transformation of each phagemid were validated by colony PCR. Next, single PCR-confirmed colonies were inoculated in TSB containing Cp and cultured at 37 °C with shaking. Mitomycin C (FUJIFILM Wako Pure Chemicals, Japan) was added to a final concentration of 2 µg/ml when the bacterial culture reached OD600 of 0.5 to allow prophage excision and subsequent phagemid DNA packaging. Mitomycin C induction was carried out overnight at 30 °C with shaking at 80 rpm. After incubation, the lysates containing phagemid-based SA-CapsidCas13a(s) were passed through a 0.22 µm membrane filter.

Measurement of phagemid copy number

The copy number of phagemids in S. aureus RN4220ΦTan2 WT was measured by real-time PCR and normalized for chromosome copy number. Primer sets RN4220-RT-PCR-FW/RN4220-RT-PCR-RV (chromosome) and pLK12-RT-PCR-FW/pLK12-RT-PCR-RV (phagemids) were used for real-time PCR amplification by MIC qPCR (Bio Molecular Systems, Australia) using TB Green® Premix Ex Taq™ II (TaKaRa, Japan).

Measurement of phage/phagemid-based SA-CapsidCas13a titers

The mitomycin C-induced lysates were serially diluted 10-fold (10−1 to 10−7) with SM buffer. Meanwhile, overnight culture of S. aureus strain RN4220 diluted 1: 100 with TSB broth was incubated with agitation at 37 °C until an OD600 of approximately 0.5. Then, 110 µl of each dilution of the lysates was added to a same volume of bacterial suspension and the mixture was incubated at 37 °C for 20 min. A total of 100 µl of the mixture was added to 4 ml soft agar pre-warmed at 55 °C (2 × 4 ml soft agar) and overlaid onto TSA plate and TSA plate containing Cp. The solidified plates were incubated at 37 °C overnight. Colonies grown on TSA plate containing Cp were counted to determine the transduced colony-forming units (TFU/ml); whereas the plaques formed on drug-free TSA plate were counted for determination of plaque-forming units (PFU/ml).

Targeted killing assay (soft agar overlay)

The target S. aureus strains, USA30044 and USA300ΔmecA, were grown to an OD600 of approximately 0.5. Then, 110 µl of the culture were mixed well with equal volume of SA-CapsidCas13a::CpR_mecA (mecA-targeting) or SA-CapsidCas13a::CpR_nontargeting (control) and the mixture was incubated at 37 °C for 20 min. After incubation, 200 µl of the mixture was added to 4 ml soft agar pre-warmed at 55 °C and overlaid on TSA plate containing Cp. The plates were left at room temperature until the top agar solidified and then incubated at 37 °C overnight. Targeted bacterial killing was observed following overnight incubation.

Targeted killing assay (spot test)

To assess the sequence-specific bactericidal activity of the generated phagemid-based SA-CapsidCas13a, a spot test assay was employed. Selected strains, derived from in-house whole-genome-sequenced S. aureus clinical isolates, were cultured to an OD600 of approximately 0.5. Subsequently, 100 µl of the culture was combined with 4 ml of soft agar and overlaid onto a TSA plate containing Cp and tetracycline. After solidification at room temperature, phagemid-based SA-CapsidCas13a with known titers was adjusted to 106 TFU/ml and subjected to 10-fold serial dilution. Finally, 2 µl of each dilution were spotted onto the bacterial lawn, and the plates were incubated at 37 °C overnight for assessment.