CRISPR-prime editing, a versatile genetic tool to create specific mutations with a single nucleotide resolution in Leptospira

ABSTRACT

Leptospirosis, caused by pathogenic bacteria from the genus Leptospira, is a global zoonosis responsible for more than one million human cases and 60,000 deaths annually. The disease also affects many domestic animal species. Historically, genetic manipulation of Leptospira has been difficult to perform, resulting in limited knowledge on pathogenic mechanisms of disease and the identification of virulence factors. The application of CRISPR/Cas9 and its variations have helped fill these gaps but the generation of knockout mutants remains challenging because double-strand breaks (DSBs) inflicted by Cas9 nuclease are lethal to Leptospira cells. The novel CRISPR prime editing (PE) strategy is the first precise genome-editing technology that allows deletions, insertions, and base substitutions without introducing DSBs. This revolutionary technique utilizes a nickase Cas9 that cleaves a single strand of DNA, coupled with an engineered reverse transcriptase and a modified single-guide RNA (termed prime editing guide RNA) containing an extended 3′ end with the desired edits. We demonstrate the application of CRISPR-PE in both saprophytic and pathogenic Leptospira from multiple species and serovars by introducing deletions or insertions into target DNA with a remarkable precision of just one nucleotide. Additionally, we demonstrate the ability to genetically manipulate Leptospira borgpetersenii, a prevalent pathogenic species of humans, domestic cattle, and wildlife animals. Rapid plasmid loss by mutated strains in liquid culture allows for the generation of knockout strains without selective markers, which can be readily used to elucidate virulence factors and develop optimized bacterin and/or live vaccines against leptospirosis.

IMPORTANCE

Leptospirosis is a geographically widespread bacterial zoonosis. Genetic manipulation of pathogenic Leptospira spp. has been laborious and difficult to perform, limiting our ability to understand how leptospires cause disease. The application of the CRISPR/Cas9 system to Leptospira enhanced our ability to generate knockdown and knockout mutants; however, the latter remains challenging. Here, we demonstrate the application of the CRISPR prime editing technique in Leptospira, allowing the generation of knockout mutants in several pathogenic species, with mutations comprising just a single nucleotide resolution. Notably, we generated a mutant in the Leptospira borgpetersenii background, a prevalent pathogenic species of humans and cattle. Our application of this method opens new avenues for studying pathogenic mechanisms of Leptospira and the identification of virulence factors across multiple species. These methods can also be used to facilitate the generation of marker-less knockout strains for updated and improved bacterin and/or live vaccines.

INTRODUCTION

Leptospirosis is a global, re-emerging neglected zoonotic disease caused by pathogenic species of the genus Leptospira (1–3). The genus comprises >40 pathogenic species and hundreds of serovars. Hosts may contract infection through penetration of abraded or sodden skin, as well as intact mucosa by leptospires, which then disseminate hematogenously to target organs (4, 5). Leptospirosis accounts for more than one million human cases and 60,000 deaths per year worldwide (6), and incidence rates are likely underestimated (7, 8). Recent occurrences of extreme climatic events are anticipated to exacerbate its global burden (9).

In urban environments, rodents serve as the principal reservoir host while exhibiting asymptomatic chronic infection (2, 10, 11). Notably, rodents harbor leptospires within their kidneys, excreting substantial concentrations of virulent leptospires via urine (12, 13). This contributes to the perpetuation of the pathogen and facilitates environmental spillover, increasing the risk of infection for new hosts (3, 14). Leptospirosis has been comprehensively documented across many domestic species including cattle, swine, sheep, goats, horses, and dogs (3, 15, 16). Bovine leptospirosis is also transmitted by genital discharges and causes significant financial losses due to abortion, stillbirths, suboptimal reproductive performance, and diminished milk production (17–19).

Mutagenesis of Leptospira spp. is critical to elucidating pathogenic mechanisms of infection and virulence. More importantly, targeted mutagenesis can lead to the development of universally protective subunit vaccines as well as more efficacious differentiating infected from vaccinated animals (DIVA) bacterin vaccines. However, mutagenesis of Leptospira species for decades lagged behind that of other bacteria, employing less reliable and time-consuming methods including random transposon mutagenesis (20, 21) or site-directed allelic exchange (22, 23).

The application of the Streptococcus pyogenes CRISPR/Cas (clustered regularly interspaced short palindromic repeat/CRISPR associated) system and its variants to Leptospira has facilitated the generation of mutants and subsequently expanded our knowledge of Leptospira biology, host-pathogen interactions, and virulence factors (24–26). The original iteration of the CRISPR/Cas9 system comprised the expression of the programmable endonuclease Cas9 that targets genomic sites with a single-guide RNA sequence (sgRNA) to create a double-strand break (DSB) in the genome (27–29). However, DSBs are lethal to many prokaryotes (30, 31) including Leptospira (32, 33). The DSB lethality to leptospires was overcome by the application of CRISPR-interference (CRISPRi) methodologies (34, 35). Alternatively, the concurrent expression of CRISPR/Cas9 with the non-homologous end-joining machinery from Mycobacterium smegmatis facilitated DSB repair, permitting the generation of knockout mutants in pathogenic Leptospira (33).

The novel CRISPR prime editing (PE) strategy was first utilized by Anzalone et al. (36) to edit human cells, followed by its application to several model organisms, such as mouse embryos (37, 38), organoid lines (39, 40), plants (41), and Escherichia coli (42). CRISPR-PE is the first precise genome-editing technology that allows deletion, insertions, and base substitutions without introducing DSBs (43) as it relies on a nickase Cas9 (Cas9n) that cleaves only a single strand of DNA. In CRISPR-PE, Cas9n is genetically coupled with an engineered reverse transcriptase (RT). The system also utilizes a modified sgRNA (termed prime editing guide RNA [PEgRNA]) that includes a primer-binding sequence (PBS) and reverse transcription template (RTT) (36). This PEgRNA allows not only target recognition by protospacer base-pairing but also the incorporation of desired edits in the DNA target dictated by the RTT sequence.

Here, we showcase the utilization of CRISPR-PE in both saprophytic and pathogenic Leptospira species by introducing deletions or insertions into target DNA with a remarkable precision of just one nucleotide, resulting in the generation of knock-out mutations. After establishing desired mutations, plasmid instability favored rapid plasmid curing, resulting in mutants without selective markers. This demonstrates not only the first application of the innovative CRISPR-PE technique in developing knockout strains across various species and serovars of Leptospira spp. but also the first time that this tool has been applied to spirochetal species.

This advancement is expected to both accelerate studies on Leptospira pathogenesis and virulence factors across multiple species and transform the Leptospira vaccinology field by generating marker-less knockout strains for updated and improved bacterins and/or live vaccine strains.

RESULTS

Validation of CRISPR-PE in the Leptospira biflexa model

The nucleotide sequence encoding the Cas9-RT fusion protein was amplified from the pCRISPR-PE-bacteria (42) and mixed with an amplicon containing the pMaOri backbone along with the cas9 promoter, previously amplified from the pMaOriCas9 plasmid (32), for Gibson Assembly. The recombinant plasmid was named pMaOriPE and was designed to allow the constitutive expression of the fusion protein Cas9n-RT, which is responsible for the prime-editing strategy, and to enable the incorporation of PEgRNA cassettes into a NotI restriction site. The PEgRNA contains 3′ extension sequences comprising a PBS and an RTT (Fig. 1A, panel a). The desired nucleotide mutation, comprising an insertion, deletion, or substitution, is included in the RTT. After protospacer adjacent motif (PAM; 5′ NGG 3′) recognition and PEgRNA base-pairing to the target strand due to the 20-nt protospacer, Cas9n cleaves the PAM-containing strand of the DNA, releasing the adjacent sequence. After nicking, the PBS hybridizes to the single-stranded DNA, the 3’ OH of which serves as a primer for RT, allowing reverse transcription of DNA containing the desired edits based on the RTT sequence (Fig. 1A, panels b and c). In this work, mutations were specifically designed to disrupt the PAM sequence so that the Cas9n target motif was disrupted once the mutation was introduced, thereby avoiding continued Cas9n activity after the establishment of the desired mutation (42).

Fig 1.

CRISPR prime editing mechanism of action and validation of Cas9n-RT expression in L. biflexa. (A) The fusion protein Cas9n-RT, upon PAM NGG recognition, recognizes the desired genomic targets by base-pairing due to the 20-nt protospacer within the PEgRNA (a). Cas9n cleaves the PAM-containing strand of the DNA, releasing the adjacent sequence (b). The PEgRNA also contains a 3′ extension comprising the PBS and RTT (reverse transcriptase template) sequences, containing the desired edits, specifically designed to disrupt the NGG sequence. The PBS hybridizes to the single-stranded DNA, whose 3′ OH group serves as a primer for RT, allowing the reverse transcription of the new DNA containing the desired edits based on the RTT sequence (c). An equilibrium exists between the 3′ (containing the edit) and the 5′ flap (unedited). Excision of the 3′ flap by cellular nucleases leads to reversion to wild-type (WT) sequence, which will in turn be subjected again to CRISPR-PE action until the desired mutation is achieved. Cleavage of the unedited 5′ flap leads to the establishment of the PAM-disruptive mutation (d). (B) Expression of the fusion protein Cas9n-RT was validated in the surrogate saprophyte L. biflexa. Plasmid pMaOriPE was delivered by conjugation, colonies were randomly selected, and cells were grown in liquid Ellinghausen–McCullough–Johnson–Harris (EMJH) with spectinomycin. Cell lysates were analyzed by SDS-PAGE (left panel) and immunoblotting with anti-Cas9 antibodies (right panel). WT cells or cells containing pMaOri.dCas9 were used as controls.

Although there is an equilibrium between the 3′ (containing the edit) and the 5′ flap (unedited; Fig. 1A, panel d), the extended 3′ flap is thermodynamically less likely to hybridize to the unedited complementary strand. Excision of the 3′ flap by cellular nucleases leads to the reversion to the wild-type (WT) sequence, which in turn will be subjected again to CRISPR-PE action until the desired mutation is achieved. On the other hand, cleavage of the unedited 5′ flap leads to the establishment of the PAM-disruptive mutation, and insertions or deletions of one or two nucleotides result in a frameshift that leads to premature stop codons and subsequently, gene knockout (42, 43).

To validate the expression of the Cas9n-RT protein, the surrogate saprophyte L. biflexa model was used. Recombinant cell lysates were analyzed by SDS-PAGE (Fig. 1B, left panel) and immunoblotting with anti-Cas9 antibodies (Fig. 1B, right panel). A ~280 kDa band was found exclusively in cells containing the pMaOriPE plasmid; the band also reacted to anti-Cas9 antibodies. No reactivity was observed in the WT cells, and a band of 160 kDa, corresponding to dead Cas9 (dCas9), was detected in the control cells containing the pMaOri.dCas9 plasmid (32). These results validate the successful expression of Cas9n-RT in saprophytic Leptospira.

Establishment of deletion and insertion mutations in the β-galactosidase gene

Four distinct PEgRNA constructs were created to induce one or two nucleotide insertions or deletions in the L. biflexa β-galactosidase gene. A summary of the constructs and the effect of each mutation on the target protein translation are presented in Table 1. Complete PEgRNA sequences, highlighting the protospacer, PBS, and RTT sequences, are listed in Table S2.

TABLE 1.

PEgRNA targets, designed mutations and resulting stop codon in Leptospira spp.

PEgRNA	Target	Mutation	Resulting stop codon
PEgRNAβgal-1	L. biflexa β-galactosidase	One nucleotide deletion G (406)	TGA (491–493)
PEgRNAβgal-2	L. biflexa β-galactosidase	Two nucleotide deletion GG (406–407)	TGA (417–419)
PEgRNAβgal+1	L. biflexa β-galactosidase	One nucleotide insertion T (between 406 and 407)	TGA (417–419)
PEgRNAβgal+2	L. biflexa β-galactosidase	Two nucleotide insertion TT (between 406 and 407)	TGA (491–493)
PEgRNAlipL32-1	L. interrogans and L. borgpetersenii lipL32	One nucleotide deletion G (211)	TAG (215–217)
PEgRNAlipL32-2	L. interrogans and L. borgpetersenii lipL32	Two nucleotide deletion GG (211–212)	TGA (273–275)

Recombinant plasmids were delivered to L. biflexa by conjugation (44). After colony incubation with X-gal solution, only blue colonies were observed for cells containing no PEgRNA, whereas heterogeneous phenotypes were observed in cells containing PEgRNA. Most colonies displayed a blueish core with white surroundings, similar to the results of mixed populations in colonies as reported by Tong et al. (42) in E. coli. This suggests that longer incubation times favor the appearance of mutant cells. White- or blue-colored colonies were picked from each group, and cells were grown in liquid media for qualitatively reassessing β-galactosidase activity. As depicted in Fig. 2A, most colonies displayed a mixed population of WT and mutant cells, and only one colony containing the PEgRNA for deleting two nucleotides (culture “d”) had completely abolished enzyme activity. Repeat experiments with quantitative evaluation of newly selected colonies confirmed that the intermediate enzyme activity phenotype was due to mixed populations (Fig. S1), as normalized cell numbers resulted in various degrees of enzymatic activity when compared to a concentration curve of WT L. biflexa.

Fig 2.

Recovery of deletion and insertion knockout mutants for β-galactosidase in L. biflexa. (A) Colonies were retrieved from plates, grown in liquid media, and recovered by centrifugation. Leptospires were resuspended in X-gal solution (50 µg/mL) in PBS and incubated for color development. Cells containing the pMaOriPE plasmid with no PEgRNA were used as a control of maximal enzyme activity. (B) Distinct mixed cultures containing the plasmids for either deletion (βgal-1 and −2) or insertion (βgal+1 and +2) were diluted and plated for individual colony formation. These were reassessed regarding β-galactosidase activity by X-gal substrate. Null phenotype (white) colonies were picked and grown in liquid media, and the desired PAM disruptive mutations (C) were confirmed by sequencing (D) in the knocked-out clones. Representative chromatograms are shown.

Mixed cultures from each PEgRNA group were diluted and spread onto EMJH plates. White/blue screening confirmed the existence of both wild type and knockout phenotypes (Fig. 2B). Representative plates are presented in Fig. S2A. Next, white colonies from each plate were selected and re-validated with X-gal substrate (Fig. S2B). DNA from mutant cultures was extracted, and the β-galactosidase gene was sequenced to confirm the presence of each designed mutation (Fig. 2C and D).

Results demonstrate the efficacy of CRISPR-PE as a versatile tool for generating specific, permanent mutations in the leptospiral genome.

lipL32 gene knockout in Leptospira interrogans strains

CRISPR-PE was next evaluated in pathogenic Leptospira spp. PEgRNA cassettes were constructed to create one or two nucleotide deletions in the conserved and pathogen-specific gene lipL32 of three different strains of L. interrogans: serovar Copenhageni strain Fiocruz L1-130, serogroup Icterohaemorrhagiae strain R47, and serogroup Canicola strain LAD-1. Colonies recovered from agar plates after 21 days were selected for evaluation of LipL32 protein expression. For L1-130 (Fig. 3A), R47 (Fig. 3B), and LAD-1 (Fig. 3C), all selected colonies containing either the PEgRNA cassette for one or two nucleotide deletions had no expression of LipL32 detectable by SDS-PAGE. Immunoblotting with anti-LipL32 antibodies confirmed the absence of LipL32 expression (Fig. 3D, E and F). The expression of Cas9n-RT was readily detected in leptospires containing the CRISPR-PE plasmids, albeit less evident in LAD-1. Cells containing pMaOriPE with no PEgRNA and a knockdown CRISPRi mutant in L. interrogans for LipL32 (44) were used as positive and negative controls for LipL32 expression, respectively. L. interrogans Fiocruz L1-130 cells with the plasmid pMaOriPE containing the PEgRNAlipl32-1 were grown sequentially in liquid media in the absence of spectinomycin to assess plasmid stability. After just one passage (P1), a prominent decrease in Cas9n-RT detection was observed, culminating with no detectable protein at passage 2 (P2), indicating plasmid instability (Fig. S3).

Fig 3.

Confirmation of lipL32 knockout clones in different strains and serogroups of L. interrogans. (A–C) Colonies of distinct L. interrogans strains obtained on plates after conjugation with pMaOriPE with no PEgRNA, or with the construct for one or two nucleotide deletions in the lipL32 gene were grown in liquid HAN media and evaluated by SDS-PAGE. Distinguishing bands for LipL32 and Cas9n-RT can be visualized. Abolished LipL32 expression was further confirmed by immunoblotting with anti-LipL32 and anti-LipL41 antibodies (D–F). A CRISPRi knockdown strain in Fiocruz L1-130 (dCas9sgRNAlipL32) was used as no-LipL32 expression control. (G) A representation showing the genetic element in the constructs for mutation where the desired PAM (NGG) disruptive mutation is reverse transcribed in the target DNA based on the RTT sequence. (H) The lipL32 gene was amplified from all mutants and sequenced to confirm desired mutations. Representative chromatograms are shown.

The lipL32 gene for each clone was amplified and sequenced to validate desired mutations (Fig. 3G), confirming that all clones displayed the expected mutation and resulting premature stop codons (Table 1). Representative chromatograms are illustrated in Fig. 3H.

Generation of a knockout mutant in Leptospira borgpetersenii

To date, mutagenesis has never been accomplished in L. borgpetersenii, a prevalent pathogenic species of humans, domestic cattle, and wildlife animal species (45–47). After conjugation of L. borgpetersenii serogroup Ballum strain LR131 with PE plasmids, alone or containing each PEgRNA, colonies were observed as soon as 14 days (Fig. 4A). Colonies were retrieved at 21 days post-inoculation (Fig. 4B) to ensure maximal editing efficiencies. Abolished LipL32 protein expression was observed by SDS-PAGE (Fig. 4C) and confirmed by immunoblotting (Fig. 4D), demonstrating that CRISPR-PE is applicable across distinct Leptospira species and serovars.

Fig 4.

L. borgpetersenii lipL32 deletions result in gene knockout. (A) Enumeration of colonies recovered after L. borgpetersenii conjugation with pMaOriPE alone or with PEgRNA construction for one or two nucleotides deletion. The averages of enumerated colonies from two separate plates are displayed. Error bars represent SD. (B) Colony morphology after growth for 21 days to ensure maximal editing efficiency. Distinct colonies were selected, grown in liquid media, and evaluated for expression of LipL32 by SDS-PAGE (C) and immunoblotting (D). Cell lysates of a lipL32 knockdown in L. interrogans Fiocruz L1-130 (dCas9sgRNAlipL32) were used as control.

Evaluation of mutant phenotypes in the hamster model of infection

LipL32 knockout mutants (KO32) and isogenic WT control cultures of L. interrogans serogroup Canicola strain LAD-1 and L. borgpetersenii serogroup Ballum strain LR131 were re-validated prior to animal challenge (Fig. 5A). Animals infected with either control cells or KO32 mutants displayed weight loss or stagnation at day 5 post-inoculation, with severe symptoms of acute leptospirosis being observed in all animals by day 6 (Fig. 5B and C). All animals were humanely euthanized on day 6 (Fig. 5D and E), except for one animal infected with a lipL32 mutant in the LR131 background which was euthanized on day 5 (Fig. 5E). Control and knockout leptospires were culturable from the kidney and liver macerates of all animals.

Fig 5.

Animal infection with mutant strains and plasmid curing. Control and lipL32 knockout (KO32) strains for L. interrogans serogroup Canicola strain LAD-1 and L. borgpetersenii serovar Arborea strain LR131 were evaluated by immunoblotting to confirm the lack of target protein (A). Hamsters (n = 4) were intraperitoneally infected with the 10⁸ cells of each strain and monitored daily for acute leptospirosis symptoms and weight loss (B and C). Negative control animals were injected with HAN medium. Animals were humanely euthanized when endpoint criteria were met, including significant weight loss and/or additional clinical signs, such as blood on paws/nose/urogenital tract and lethargy (D and E). Bacterial burden quantification was performed by quantitative PCR (qPCR) on total DNA extracted from blood at day 3, as well as from the liver and kidney from euthanized animals. Bacterial loads, expressed by genome equivalents (GEq), were evaluated by analysis of variance (ANOVA). Specific contrasts of interest were determined between least square means. Error bars represent SE (*P < 0.05, **P < 0.01, and ***P < 0.001) (F). Isolated leptospires from kidney (K) and liver (L) were re-evaluated by immunoblotting (G). To assess plasmid loss, cells were plated onto HAN plates, and individual colonies were evaluated by PCR with primers targeting the PEgRNA cassette. A PCR against lipL32 was used as an amplification positive control. A purified plasmid containing PEgRNA (C+) and a non-template control (NTC) were also included (H). Plasmid elimination was also confirmed by immunoblotting with anti-Cas9 antibodies against LR131 cell lysates. Samples consisted of: cells containing the pMaOriPE plasmid alone or with PEgRNAlipL32-1 (grown in the presence of spectinomycin), cells used for animal experiments grown in the absence of antibiotics (KO32 AE), recovered mutants from the kidney, and five clonal isolates from agar plates (I).

Differences in hematogenous kinetics between control LAD-1 and LR131 were observed with significantly less LR131 circulating in blood at day 3 (P < 0.001). Relative to the control, the lipL32 mutation only affected the blood bacterial burden in LAD-1, resulting in less circulating bacteria in LAD-1 KO32 (P < 0.001). Foamy macrophages found previously to be associated with virulence were identified in both control and KO32 mutants for both species. The LR131 bacterial burden was significantly higher in target organs, most prominently in the liver than LAD-1 in both the control (P < 0.05) and KO32 (P < 0.001) groups (Fig. 5F). While not significant, a trend was observed where hamsters infected with LAD-1 KO32 had higher leptospiral loads in the liver and kidneys than those infected with control cells. For strain LR131, lipL32 mutation also resulted in slight but significant increases in liver bacterial burdens (P < 0.01).

Recovery of marker-less knockout strains

Recovered lipL32 mutants from kidney cultures (Fig. 5G) were seeded onto HAN plates, and colonies of LR131 KO32 (n = 5) and LAD-1 (n = 3) mutants were selected for growth in liquid media in the absence or presence of spectinomycin. All cultures grew exclusively in the absence of antibiotics, denoting plasmid loss in the population, and were therefore subsequently evaluated by PCR. No bands corresponding to the PEgRNA cassette were observed in any clones following PCR with primers specifically targeting the constructs (Fig. 5H, upper panel); a control PCR targeting lipL32 in all the clones resulted in positive amplification (lower panel). LR131 KO32 was then evaluated by immunoblotting with anti-Cas9 antibodies to detect the presence of Cas9n-RT. Protein reactivity was observed in those cells containing pMaOriPE plasmids (either with no PEgRNA or with PEgRNA targeting lipL32) which were grown in the presence of spectinomycin. The cell population used for the animal experiment (KO32 AE), which was grown in media with no antibiotics, as well as the mutants recovered from the kidney (P0 kidney) displayed a complete lack of detectable Cas9n-RT protein. This confirmed plasmid instability in media without selective pressure for L. borgpetersenii as observed for L. interrogans previously in this work (Fig. S3). As expected, all colonies selected from plates that were inoculated with kidney cultures had no detectable expression of the target protein.

DISCUSSION

Genetic manipulation of Leptospira spp. has advanced greatly in recent years with the advent of the CRISPR/Cas9 system. Since DSBs introduced by sgRNA-guided Cas9 are lethal to leptospires, the initial approach to mutant recovery was the use of a catalytically inactive dCas9. Blockage of transcription by the sgRNA-dCas9 nucleocomplex (CRISPRi) causes gene silencing (knockdown), but this requires continuous antibiotic pressure and does not disrupt the gene at the genomic level (32, 44). The CRISPR Prime Editing strategy is a new and revolutionary tool that does not require DSB, recombinant templates, or homologous recombination, and it facilitates mutations of a single nucleotide resolution (36, 39, 41). The application of CRISPR-PE to prokaryotic cells also resulted in successful mutant recovery with up to 40% efficiencies in E. coli (42).

In this work, we used CRISPR-PE to insert or delete nucleotides in target genes of saprophytic and pathogenic Leptospira to produce gene knockouts. Target mutations were designed to specifically disrupt the PAM 5′-NGG-3′ motif such that once the mutation was obtained, further DNA cleavage by Cas9n was prevented.

In saprophytic L. biflexa, CRISPR-PE resulted in a mixed population of colonies containing both WT and mutated cells. Subsequent plating of mixed populations recovered null β-galactosidase clones, which were confirmed by sequencing. Mixed populations in colonies were also observed by Tong et al. (42) who concluded that extended incubation times favor the accumulation of editing events by CRISPR-PE in E. coli. Similarly, higher editing efficiencies were recently described in M. smegmatis by Zhang et al. (48) compared to E. coli. The authors hypothesized that the slower growth rate of M. smegmatis favors the accumulation of editing products.

The direct correlation between longer incubation times and editing efficiencies was anticipated to favor mutagenesis in pathogenic Leptospira due to their prolonged incubation times for colony formation (44). In fact, 100% mutant recovery efficiencies were observed for all pathogenic strains tested for mutation and knockout of lipL32. Of note, every single conjugation experiment performed resulted in mutant colony formation since CRISPR-PE is a DSB-free technique and mutant recovery is only dictated by conjugation frequencies. CRISPR-PE mutant recovery in Leptospira is also more efficient than recently described methods that generate knockout mutants using CRISPR/Cas9 and the concurrent expression of the NHEJ proteins LigD and Ku from M. smegmatis (33); from a probabilistic standpoint, the frequency of mutant recovery in this system is dictated not only by the efficiency of conjugation but also by the frequency of recovery from DSB.

The higher editing efficiencies achieved by our PEgRNA constructs in Leptospira are also likely due to slight modifications in design from those described by Tong et al. (42). Both PBS and RTT sequences were 13-nt long, and after the stretch of thymidine, the intrinsic terminator from the B. burgdorferi bmpB gene was included. This strategy not only ensures the termination of PEgRNA transcription but favors its stability because the incorporation of structured RNA motifs to the 3′ terminus of PEgRNAs enhances stability and prevents degradation of the 3′ extension (49). Reverse transcription of the RTT sequence within the PEgRNAs could potentially advance into the guide RNA scaffold resulting in scaffold sequence insertion at DNA targets (36), but no scaffold sequence nucleotides were observed in our clones. Mechanistically, it was conjectured by Anzalone et al. (36) that the incorporation of such nucleotides is minimized since the scaffold sequence in the PEgRNA is mostly inaccessible to reverse transcription due to Cas9 domain binding. Surprisingly, the pMaOriPE plasmid displayed instability in liquid media without selective pressure, contrary to previous CRISPRi (32) and Cas9NHEJ systems (33) in Leptospira spp. Complete plasmid loss, premeasured by abolished Cas9n-RT expression, was noticeable as early as passage 2 in liquid media without antibiotics. This rapid plasmid loss provides additional advantages as marker-less knockout strains can be compared directly with their isogenic parent strains; this is also tremendously advantageous for generating novel DIVA bacterins or even live vaccines which ideally should not possess selective markers.

Mutagenesis and evaluations of virulence phenotypes have traditionally been limited to the L. interrogans background. LipL32 is a high copy number protein, conserved and exclusive to pathogenic strains (50), estimated to account for almost 75% of the outer membrane proteome (51). Knockdown lipL32 mutants in L. interrogans serovar Copenhageni strain Fiocruz L1-130 by CRISPRi (25, 26) resulted in a more virulent strain, most probably due to the upregulation of virulence factors to “fill the niche” left by the absence of most abundant outer membrane protein LipL32 (25). The effect of LipL32 mutation in other serovars and species remained to be determined.

In addition to L. interrogans serovar Copenhageni strain Fiocruz L1-130, the efficacy of CRISPR-PE in generating knockout mutants for lipL32 was validated in alternative serotypes including a recent clinical isolate of L. interrogans serogroup Canicola strain LAD-1, which remained virulent in the hamster model of leptospirosis. As demonstrated for lipL32, a single PEgRNA construct can be used to target highly conserved genes in different Leptospira species. Using our CRISPR-PE system, we performed genetic manipulation and evaluation of virulence factors in the alternative pathogenic species L. borgpetersenii, a prevalent pathogen of humans, domestic cattle, and wildlife animal species (46, 52, 53). The hamster model of leptospirosis confirms that L. borgpetersenii strain LR131 also retains the ability to cause acute disease even in the complete absence of LipL32. The evaluation of mutant phenotypes in different species such as L. borgpetersenii, which has a smaller genome than L. interrogans and therefore fewer expected redundant genes (54), is important to determine pathogenic mechanisms of infection that are conserved across different species and serovars. The role of the pathogen-conserved gene lipL32 continues to be enigmatic (50).

Genetic manipulation using CRISPR-PE allowed us to compare different species and serotypes of Leptospira alongside their isogenic mutants in animal models, generating novel insights into mechanisms of infection. Both L. interrogans strain LAD-1 and L. borgpetersenii strain LR131 are both lethal in the hamster model of leptospirosis yet show remarkably different and contrasting bacterial loads in blood at 3 days post-infection compared to organs at day 6 post-infection. The higher organ loads observed for LR131 might indicate a more direct pathophysiological mechanism compared to LAD-1. Regardless, CRISPR-PE provides a mechanism to perform functional genomics to further investigate these phenomena.

Collectively, our results demonstrate the applicability of the novel CRISPR-PE technique to include specific and single-nucleotide resolution mutations into the genomes of multiple species and serotypes of Leptospira. Rapid plasmid loss in liquid culture subsequently facilitates the generation of marker-less knockout strains. This method can thus be used to elucidate pathogenic mechanisms of infection, identify virulence factors, and develop optimized DIVA bacterin and/or live vaccines against leptospirosis.

MATERIALS AND METHODS

Bacterial strain and media

Pathogenic L. interrogans serovar Copenhageni strain Fiocruz L1-130 (55), L. interrogans serogroup Icterohaemorrhagiae strain R47 (56), L. interrogans serogroup Canicola strain LAD-1 (J. E. Sykes, J. Nally et al., unpublished data), and L. borgpetersenii serogroup Ballum strain LR131 (45) were grown in liquid HAN media (57) at 29°C. Saprophytic L. biflexa serovar Patoc strain Patoc1 was cultured in EMJH medium (Difco, BD, Franklin Lakes, NJ). Solid media were prepared by supplementing with 1.2% noble agar (Difco). Where necessary, spectinomycin was added at 40 µg/mL. E. coli strain β2163 (58), auxotrophic for diaminopimelic acid (DAP), was used for general cloning and as conjugation donor cells. E. coli cells were grown in Lysogeny Broth (LB, Difco) media supplemented with DAP (0.3 mM, Sigma).

Construction of CRISPR-PE plasmids and PEgRNA

A plasmid expressing the Cas9 H840A nickase and reverse transcriptase fusion protein (Cas9n-RT), under the control of the native S. pyogenes cas9 promoter, was constructed for prime editing in Leptospira spp. For this, the plasmid pCRISPR-PE-bacteria (42) was purchased from Addgene (Catalog 172715) and used as a template for PCR amplification of the Cas9n-RT coding sequence using with Q5 high fidelity polymerase (New England Biolabs M0491). This plasmid expresses an E. coli codon-optimized fusion protein composed of Cas9n, a 33 amino acid flexible linker, and the reverse transcriptase M-MLV2 (Moloney murine leukemia virus variant [36]). The pMaOri backbone, along with the cas9 promoter, was amplified from the pMaOriCas9 plasmid (32) and used for Gibson Assembly (NEB E2611S) ligation with the Cas9n-RT coding sequence. Reactions were used to transform E. coli β2163 cells, and colonies were randomly selected for evaluation of the presence of the recombinant plasmid, termed pMaOriPE.

For PEgRNA design, the lipL32 promoter was used (32, 59), followed by a 20-nt protospacer, the Cas9 scaffold, a 3′ extension composed of RTT sequence (13-nt for deletion or 14–15-nt for insertion) and a 13-nt PBS, and a stretch of thymidine and the intrinsic terminator from the bmpB gene of B. burgdorferi (60). Protospacers were designed by retrieving sequences of interest from GenBank (https://www.ncbi.nlm.nih.gov/genbank/) and submission to the webserver CRISPRscan (https://www.crisprscan.org/). Protospacer off-targets were analyzed by Cas-OFFinder (http://www.rgenome.net/cas-offinder/). In the case of pathogenic species for lipL32 target constructs, sequences were designed to be functional in both L. interrogans and L. borgpetersenii. PEgRNA cassettes were synthesized by GeneArt (Invitrogen) and amplified by PCR with primers PEgRNA-F and PEgRNA-R. The resulting amplicons were used for Gibson Assembly ligation with NotI digested pMaOriPE. All primers, relevant sequences, protospacers, and 3′ extensions used in this study are listed in Tables S1 and S2.

Plasmids were delivered to Leptospira spp. by conjugation with recombinant E. coli β2163 cells, as per previously published protocols (44, 61).

Selection and validation of knockout mutants for β-galactosidase in L. biflexa

Plates containing recombinant L. biflexa colonies were evaluated for β-galactosidase activity using X-gal (5-bromo-4-chloro-3-indolyl-beta-D-galacto-pyranoside) chromogenic substrate. Briefly, 200 µL of substrate solution (50 µg/mL) were spread onto plates, which were incubated at 29°C for 16 hours for color development. Next, white or light blue colonies were picked from plates and used to inoculate liquid EMJH media containing 40 µg/mL spectinomycin. Phenotypes were re-assessed by harvesting 10⁹ cells from the media (10,000 × g, 10 min) and resuspending the pellet into 200 µL of X-gal solution (50 µg/mL). Reactions proceeded for 16 hours followed by absorbance readings at 650 nm. Control cells with no PEgRNA were used as a control for maximum enzyme activity. In the case of mixed populations of mutants as characterized by partial β-galactosidase activity, cultures were brought to 10³ cells/mL, and 100 µL of each suspension was spread onto EMJH plates. Colonies were then re-evaluated with X-gal substrate, and those with completely abolished enzymatic activity were harvested for DNA extraction to further confirm desired mutations.

Electrophoresis and immunoblotting

Cultures of Leptospira (mid- to late-log phase, 2–5 × 10⁸/mL) were centrifuged (10,000 × g, 15 min) and washed twice with PBS. Cell lysates were processed for SDS-PAGE on 12% or 4%–15% gradient polyacrylamide gels (BioRad). Proteins were visualized by staining with Sypro Ruby (Invitrogen, CA). For immunoblotting, proteins were electrotransferred onto polyvinylidene difluoride membranes (BioRad) by semidry transfer. Membranes were blocked with SuperBlock (PBS) Blocking Buffer (Thermo) for 1 hour and then incubated with indicated primary antibody diluted in the blocking buffer (1:4,000 for rabbit anti-LipL32 and -LipL41, 1:2,000 for mouse anti-Cas9) for 1 hour at room temperature. Membranes were washed three times with PBS 0.1% Tween 20 (PBS-T) and incubated with horseradish peroxidase-conjugated secondary antibodies (1:4,000) in a blocking buffer for 1 hour at room temperature. Clarity Max ECL (BioRad) was used as a chemiluminescence substrate, and blots were visualized with a ChemiDoc MP Imaging System (BioRad).

Analysis of mutation by DNA sequencing

Total DNA from knockout and control Leptospira spp. cultures were used for PCR of the desired gene using primers flanking the mutation site (Table S1). PCR reactions were visualized on 1% agarose gel. Amplicons were then purified, and the final product was used for Sanger DNA sequencing (62, 63) with the same primers used for amplification. Chromatograms were used for alignment with wild-type sequences by BLASTn.

Animal infection experiments

Weaned female Syrian hamsters (Mesocricetus auratus) were acclimated to the facility a week prior to the challenge at 4–5 weeks of age. Hamsters were monitored daily and had ad libitum access to food and water. Animals (n = 4, weighing 69 ± 12 g) were inoculated intraperitoneally with 10⁸ of leptospires or with 500 µL HAN media (negative control). Each animal was monitored daily, weighed for clinical signs of acute leptospirosis, and humanely euthanized upon weight loss (>10%) and/or observation of additional clinical signs (blood on paws/nose/urogenital tract, lethargy, etc.) as previously described (64). Additionally, animals were bled via retroorbital plexus on day 3 post-infection for quantification of hematogenous dissemination of leptospires and for whole blood smears that were Giemsa stained and evaluated for the presence of foamy macrophages as reported previously (25, 65). One kidney and one liver lobe were harvested and immediately macerated in 5 mL of HAN media plus 5-fluorouracil (5-FU). Suspensions were used to inoculate HAN media plus 5-FU, and cultures were kept at 29°C and monitored daily by dark-field microscopy. Recovered mutants were validated by immunoblotting. An additional section of the kidney and lobe of the liver was harvested and frozen at −80°C for quantitative PCR to measure bacterial burdens, as previously described (66).

Statistics

Leptospires concentrations in the liver, kidney, and blood were evaluated independently in R (version 4.2.1) by simple linear regression. “Strain” (LAD-1, LR131) and “mutation status” (control, LipL32 KO) were fit as fixed effects along with a strain × mutation status interaction and evaluated by ANOVA. Specific contrasts of interest were determined between least square means. Error bars represent SE, and significance was determined when P-values ≤ 0.05.

Endpoints of animals inoculated with mutant strains were compared with those of animals infected with Leptospira cells containing empty pMaOriPE plasmid by the log-rank test, and a P-value below 0.05 was considered statistically significant.

Plasmid curing to recover marker-less mutants

Mutant cells cultured from the kidney of infected animals were brought to 10³ leptospires/mL, and 100 µL of each suspension was used to seed HAN agar plates for individual colony formation, at either 29°C or 37°C. Individual colonies were selected and transferred to 100 µL of HAN media, and 1 µL of each suspension was used as a template for PCR with primers targeting the PEgRNA and the lipL32 gene (Table S1) as a positive control. Cells were further inoculated in HAN media with and without spectinomycin, and those that grew exclusively in media without antibiotics were considered plasmid free. The absence of plasmids and consequently expression of Cas9n-RT were also evaluated by immunoblotting with anti-Cas9 antibodies.

ETHICS APPROVAL

All animal experimentation was conducted in accordance with protocols as reviewed and approved by the Animal Care and Use Committee at the National Animal Disease Center and as approved by U.S. Department of Agriculture institutional guidelines.

SUPPLEMENTAL MATERIAL

The following material is available online at https://doi.org/10.1128/mbio.01516-24.

Legends to Supplemental Figures. mbio.01516-24-s0001.docx.

Legends for Fig. S1 to S3.

Figure S1. mbio.01516-24-s0002.tif.

Beta-galactosidase activity in retrieved colonies.

Figure S2. mbio.01516-24-s0003.tif.

Replating and validation of KO mutants.

Figure S3. mbio.01516-24-s0004.tif.

Plasmid stability by immunoblotting.

Supplemental Tables. mbio.01516-24-s0005.docx.

Primer sequences and constructions.

ASM does not own the copyrights to Supplemental Material that may be linked to, or accessed through, an article. The authors have granted ASM a non-exclusive, world-wide license to publish the Supplemental Material files. Please contact the corresponding author directly for reuse.