Cas9-independent tracrRNA cytotoxicity in Lacticaseibacillus paracasei

Abstract

CRISPR-Cas9 systems are widely used for bacterial genome editing, yet their heterologous expression has been associated with cytotoxicity. The Cas9 nuclease from Streptococcus pyogenes (SpyCas9) has been one common source, with reports of cytotoxicity with the nuclease alone or in combination with a single-guide RNA observed in some bacteria. However, the potential cytotoxic effects of other components of the CRISPR-Cas9 system remain unknown. Here, we report that expression of the short isoform of the trans-activating CRISPR RNA (tracr-S) from the S. pyogenes CRISPR-Cas locus is cytotoxic in Lacticaseibacillus paracasei, even in the absence of SpyCas9. Deleting a putative transcription regulator in L. paracasei alleviates tracr-S cytotoxicity and leads to expression of the long isoform of the trans-activating CRISPR RNA (tracr-L). Furthermore, cytotoxicity was specific to the tracr-S sequence and was linked to direct interactions with host RNAs. This work thus reveals that additional CRISPR components beyond Cas9 can interfere with the use of heterologous CRISPR-Cas systems in bacteria, with potential implications for the evolution of CRISPR immunity.

Introduction

CRISPR-Cas systems are prokaryotic adaptive immune systems that protect against foreign genetic elements (Barrangou and Marraffini 2014, Makarova et al. 2020). Type II CRISPR-Cas systems are widely recognized as the source of RNA-guided Cas9 effector nucleases used for genome editing and gene regulation (Jinek et al. 2012, Jiang et al. 2013, Adli 2018, Pacesa et al. 2024). Immune defense through DNA cleavage relies on two RNA molecules: the CRISPR RNA (crRNA), which guides Cas9 to specific DNA sequences flanked by a protospacer-adjacent motif (PAM), and the trans-activating CRISPR RNA (tracrRNA). The tracrRNA is essential for the maturation of crRNA, as it forms a duplex with pre-crRNA, allowing processing by RNase III to process it into mature crRNA (Deltcheva et al. 2011). The resulting RNA duplex then guides Cas9 to target DNA sites, which Cas9 cleaves at precise locations (Barrangou et al. 2007, Gasiunas et al. 2012, Jinek et al. 2012).

The precision and adaptability of Cas9 have made it a powerful genome-editing tool across a wide range of bacteria, typically achieved by heterologous expression (Jiang et al. 2013, Arroyo-Olarte et al. 2021). However, heterologous expression of CRISPR-Cas9 systems, particularly from Streptococcus pyogenes, has posed some challenges in various bacterial species due to cytotoxic effects (Vento et al. 2019). For example, stable plasmid-based Cas9 expression in Synechococcus elongatus (Wendt et al. 2016) and Corynebacterium glutamicum (Jiang et al. 2017, Liu et al. 2017) was found to be lethal, though employing transient expression strategies in Synechococcus or fine-tuning promoter strength and transcription termination in C. glutamicum allowed successful introduction of Cas9. In Escherichia coli, deactivated Cas9 (dCas9) can silence essential genes with as few as four nucleotides of complementarity at the PAM-proximal sequence (Rostain et al. 2023), and high dCas9 expression without a guide RNA induces transcriptional changes and cell morphology alterations (Cho et al. 2018). To date, little is known about the potential cytotoxic effects of other components in the CRISPR-Cas9 system.

Here, we show that plasmid-expressed Cas9 and tracrRNA from S. pyogenes are cytotoxic in Lacticaseibacillus paracasei when co-expressed without a crRNA. Surprisingly, even tracrRNA alone, without the S. pyogenes Cas9 (SpyCas9), exhibits cytotoxicity. We traced the cytotoxicity to the short isoform of tracrRNA (tracr-S). Cytotoxicity was specific to the tracrRNA sequence and likely arose through direct binding to host RNAs. Our findings reveal that the tracrRNA can be a source of cytotoxicity separately from Cas9, adding another factor to consider when utilizing CRISPR-Cas9 in non-native bacterial hosts.

Materials and methods

Bacterial strains and culture conditions

All strains are listed in Table S1. Lacticaseibacillus paracasei strain B (Siedler et al. 2020) was used as the primary L. paracasei strain in this study. All plasmids derived from the E. coli-Lactobacilli shuttle plasmid were propagated and maintained in E. coli EC135 (Zhang et al. 2012). Escherichia coli EC135 cells were grown in Luria Bertani medium (5 g/l NaCl, 5 g/l yeast extract, 10 g/l tryptone) at 37°C with shaking at 250 rpm. To select and maintain shuttle vectors containing an erythromycin resistant gene, erythromycin was added at 300 µg/ml. To select and maintain the shuttle vector containing E. coli ampicillin resistant gene with Lactobacilli chloramphenicol resistant gene in E. coli, ampicillin or carbenicillin was added at 100 µg/ml.

Lacticaseibacillus paracasei strain B, Lacticaseibacillus paracasei ATCC 334, Lacticaseibacillus paracasei ATCC 25302, Lacticaseibacillus casei ATCC 393, Lactiplantibacillus plantarum WCFS1, and Lacticaseibacillus rhamnosus GG cells were grown in deMan Rogosa Sharpe (MRS) medium at 37°C, with lower temperature referring to 29°C. Erythromycin and chloramphenicol were used as selection markers and added at both 10 µg/ml, unless mentioned otherwise. The experiments were mostly performed under aerobic condition, agar plates were sealed with parafilm, and liquid cultures were grown in screwed cap falcon tubes without shaking to minimize oxygen and evaporation. When indicated, anaerobic condition was performed in an anaerobic chamber.

Bacterial transformations

Transformation and growth curve analysis of L. paracasei

Lacticaseibacillus paracasei was transformed via electroporation (Siedler et al. 2020), an overnight culture was diluted 1:25 and grown to OD₆₀₀ 0.6, then cells were placed on ice for 10 min before spinning down via centrifugation for 10 min at 4000 rpm and 4°C. The cell pellet washed twice with 50 ml of ice-cold 10% glycerol, then resuspended in 0.5 ml of 10% glycerol. For electroporation, 100 µl of competent cells were added with 1 µg of pure plasmid DNA in a 2-mm electroporation cuvette kept on ice. The electroporator settings were as follows: 2.0 kV, 25 µF capacitance, 200 ohms. Immediately after electroporation, cells were recovered in 900 µl of MRS media supplemented with 500 mM sucrose, 20 mM MgCl₂, and 2 mM CaCl₂ for 3–4 h (MMRS). Then the cells were used for serial dilution and plated on an MRS plate containing antibiotics. The rest of the transformant was centrifuged for 3 min at 4000 rpm to remove excess media and spread on MRS agar with appropriate antibiotics. Plates were incubated at 37°C, and colony images were taken after 3 days. Colony numbers were counted, and CFU per µg of plasmid DNA was calculated and plotted using GraphPad Prism.

To assess growth dynamics, OD₆₀₀ was measured over 56 h at 37°C using a Synergy H1 plate reader (BioTek), with reads taken every 15 and 2 min of orbital shaking before each measurement. Following transformation and recovery, 10 µl of culture was diluted into 500 µl of MRS medium supplemented with 10 ng/µl erythromycin. From this dilution, 100 µl was transferred into a 96-well flat-bottom plate. Each condition was prepared in triplicate, and media-only controls were included as blanks. OD₆₀₀ values were normalized to the blank and plotted using GraphPad Prism.

Transformation of L. paracasei ATCC 334 and ATCC 25302

The competent cells were prepared based on a previously described protocol with some modification (Broadbent et al. 2014). Briefly, a 5-ml overnight culture of L. paracasei ATCC 334 and ATCC 25302 was grown in MRS medium at 37°C. The culture was diluted to a starting OD₆₀₀ of 0.1 in 200 ml of pre-warmed MRS medium supplemented with 1% glycine and grown to an OD₆₀₀ of 0.6. Cells were placed on ice for 10 min, then harvested by centrifugation at 4000 rpm for 10 min at 4°C. The pellet was washed twice with 200 ml of ice-cold, sterile distilled water and resuspended in 10% PEG-8000 to a final volume of 0.5 ml.

For electroporation, 100 µl of competent cells were mixed with 1 µg of plasmid DNA in a 2-mm electroporation cuvette that was pre-chilled on ice. The electroporation was performed at 2.5 kV, 25 µF, and 200 Ω. Immediately after electroporation, cells were recovered in 900 µl of MMRS and incubated at 37°C for 3–4 h. Following recovery, 5 µl of the culture was serially diluted and plated on selective agar to estimate transformation efficiency. Images of the plates were taken after 3 days of incubation at 37°C.

Transformation of L. casei ATCC 393

Competent L. casei ATCC 393 cells and transformation were prepared based on previously described protocol with some modifications (Song et al. 2014). Briefly, a 5-ml overnight culture of L. casei ATCC 393 was grown in MRS medium at 37°C. The overnight culture was then back diluted to a starting OD₆₀₀ 0.1 in 100-ml of MRS medium at 37°C for around 4.5 h to reach OD₆₀₀ of 0.6. The cells were placed in an ice bath for 30 min, pelleted at 4°C and 4000 rpm, and then resuspended in 40 ml of ice-cold EPWB buffer (0.6 mmol/liter NaH₂PO₄ and 0.1 mmol/liter MgCl₂ [pH 7.4]). The cells were pelleted, then washed two times with EPWB buffer and once with EPB buffer (EPWB plus 0.3 M sucrose). Finally, the cells were resuspended in 1 ml EPB buffer. For transformation, 500 ng of plasmid DNA was added to a 200-µl aliquot of competent cells on ice and transformed by electroporation in a cold 22-mm electroporation cuvette at 2.5 kV, 25 µF, and 200 Ω. Cells recovered in MMRS overnight. Following recovery, 5 µl of the culture was serially diluted and plated on selective agar to estimate transformation efficiency. Images of the plates were taken after 3 days of incubation at 37°C.

Transformation of L. plantarum WCFS1

The transformation protocol used for L. plantarum WCFS1 was based on Leenay et al. (2019). Briefly, an overnight culture grown in MRS was diluted 1:25 in fresh MRS supplemented with 2.5% glycine and was grown at 37°C without shaking in a 50 ml falcon tube until reaching OD₆₀₀ of 0.6–0.8. Then, 5 ml ice-cold MgCl2 (10 mM) was used to wash the cells twice, followed by two times washing with 5 ml ice-cold SacGly solution. Cells were resuspended in 500 µl ice cold SacGly and aliquoted at 60 µl to be used immediately. For electroporation, 250 ng of plasmids was used to transform 60 µl of electrocompetent cells, which then added to a pre-cooled 1-mm electroporation cuvette and electroporated with the following conditions: 1.8 kV, 200 U resistance, and 25 mF capacitance. Following electroporation, cells were recovered in MRS broth for 3 h at 37°C and then plated on MRS agar containing appropriate antibiotics and incubated for 2 days. Erythromycin concentration was 10 mg/ml in MRS liquid and solid medium.

Plasmid construction

Primers were synthesized by Integrated DNA Technologies. Plasmids and Benchling links of all the constructs used in this study are listed in Table S1. Gibson assembly, Q5 mutagenesis, and restriction ligation techniques were used for cloning. NEBuilder HiFi DNA Assembly Master Mix (New England Biolabs, #E2621) was used for plasmid construction by Gibson assembly. Q5 site-directed mutagenesis kit (New England Biolabs, #E0554S) was used for simple deletions, insertions, and substitutions. Restriction enzyme digestion with PvuI-HF (New England Biolabs, #R3150) and NotI-HF (New England Biolabs, #R3189) followed by ligation (New England Biolabs, #M0370) of the digested backbone with phosphorylated (New England Biolabs, #M0201) annealed oligos were used to insert different spacers in the plasmids containing repeat-spacer-repeat (RSR). Constructed plasmids were transformed to chemically competent E. coli EC135. Colonies were picked and inoculated in the liquid cultures for overnight growth, followed by plasmid extraction and verification by whole plasmid sequencing or Sanger sequencing.

Deletion of peg.72 8 in L. paracasei

To generate a clean deletion of peg.728 gene in L. paracasei, pRT_KO_peg.728 containing homology arms flanking peg.728 gene was first transformed into L. paracasei. Then, a single colony containing this plasmid was picked and made competent again to be transformed with pCRISPR-Cas9_RSR_peg.728 for counter selection. Surviving colonies were screened with colony PCR and sent for Sanger sequencing to confirm the expected deletion. Then both plasmids were cured from the mutant strain through multiple rounds of non-selective growth. Plasmid curing was confirmed by streaking mutant strains onto non-selective MRS agar, then inoculating colonies into MRS media with chloramphenicol (pRT_KO_peg.728) or erythromycin (pCRISPR-Cas9_RSR_peg.728). If growth occurred on either selective medium, the process was repeated with non-selective plating and inoculation until a colony showed sensitivity to both antibiotics.

Whole-genome sequencing of L. paracasei strains

To sequence the wild-type L. paracasei strain and three L. paracasei escaper colonies (E1, E2, and E3) harboring pCRISPR-Cas9, Nextera DNA Flex Library Prep kit (Illumina, #20 018 704) was used to prepare L. paracasei genomic DNA for whole-genome sequencing. Then an Illumina iSeq was used to perform whole-genome sequencing using 150 bp paired-end reads and Illumina adapters. The sequencing reads were processed and mapped to a reference L. paracasei genome using Geneious software.

Phenol-chloroform RNA extraction of L. paracasei for Northern blot

The transformed cells were plated on agar plates with erythromycin, and after three days, grown colonies were inoculated in 5 ml of MRS with erythromycin and grown overnight as pre-cultures. The pre-cultures then back diluted to OD₆₀₀ 0.01 and harvested at OD₆₀₀ 0.1. The cell culture volume corresponding to an overall OD₆₀₀ of 2 (20 ml of OD₆₀₀ 0.1) was mixed with 20% cold EtOH/phenol stop solution (95% of 100% w/v EtOH with 5% phenol (Roti-Aqua phenol #A980.1)) in a 15 ml falcon tube, inverted once, snap frozen in liquid nitrogen, and stored at −80°C.

For the RNA extraction, the cell suspension was thawed on ice and centrifuged at 4000 rpm for 20 min at 4°C. The cell pellet was resuspended in 600 µl of 10 mg/ml lysozyme TE pH 8.0 solution, transferred to an RNase-free 2-ml Eppendorf tube, and mixed with 60 µl of 10% w/v SDS. The tube was placed into a water bath set to 64°C for 2 min until the sample turned less turbid and mixed by inversion with 66 µl of 3 M NaOAc, pH 5.2. For the hot phenol extraction, ∼750 µl of phenol (Roti-Aqua phenol, #A980.3) was added, and the tube was incubated in the water bath at 64°C for 6 min with shortly vortexing the tube every 30 s. The sample was then chilled on ice for 1 min and centrifuged at 13 000 rpm for 15 min at 4°C. For the chloroform extraction, the upper aqueous layer was transferred into a 2 ml Phase Lock Gel tube (VWR International, #733–2478) and mixed by inversion with ∼750 µl of chloroform (Roth, #Y015.2). The tube was centrifuged at maximum speed for 10 min at 4°C, and the aqueous layer was transferred into a fresh RNase-free Eppendorf tube to be further purified by ethanol precipitation.

For the subsequent Ethanol precipitation, the aqueous layer was placed in a 2 ml microcentrifuge tube, and double the volume of a 30:1 mixture (EtOH:3 M NaOac, pH 6.5) was added. The tubes were then stored at −80°C for at least 2 h or overnight. The frozen samples were centrifuged for 30 min at 13 000 rpm at 4°C, the ethanol was removed, and the pellet was carefully washed with 300 µl 75% v/v ethanol to remove any remaining impurities. The sample was centrifuged, and the ethanol was removed carefully to keep the pellet intact. The pellet was washed again with absolute ethanol. After another centrifugation at 4°C for 10 min at 13 000 rpm, the ethanol was removed, and the pellet was air dried. After drying, 50 µl H₂O was added, and the pellet was resuspended for 5 min at 65°C and 1000 rpm on the heating block. Finally, the RNA was quantified and the quality controlled using a NanoDrop spectrophotometer.

Northern blotting analysis

For Northern blot analysis, 5 µg of each RNA sample was resolved on an 8% polyacrylamide gel containing 7 M urea at 300 V for 2 h and 25 min using a gel transfer system (Doppel-Gelsystem Twin L, PerfectBlue). Using an Electroblotter with an applied voltage of 50 V for 1 h at 4°C (Tank-Elektroblotter Web M, PerfectBlue), the RNA was transferred onto Hybond-XL membranes (GE Healthcare, #RPN203S), crosslinked with UV-light for a total of 0.12 Joules (UV-lamp T8C; 254 nm, 8 W), and hybridized overnight in 15 ml of Roti-Hybri-Quick buffer at 42°C with 5 µl γ-32P-ATP end-labeled oligodeoxyribonucleotides (5 pmol/µl) (Table S1). The labeled RNA was visualized with a Phosphorimager (Typhoon FLA 7000, GE Healthcare).

MS2-affinity purification coupled with RNA sequencing (MAPS)

Affinity purification assays were performed as previously described (Lalaouna and Masse 2015, Correia Santos et al. 2021) with some modifications. Briefly, L. paracasei strains were grown to an OD_{600 nm} of 0.1 in 25 ml culture. Cells were then chilled for 10 min on ice, centrifuged to remove the excess media, and the pellet was resuspended in 600 µl of buffer A (20 mM Tris–HCl pH 8.0, 150 mM KCl, 1 mM MgCl2, 1 mM DTT) in the 2 ml tube. A volume of 750 µl of 0.1-mm glass beads was added to lyse the cells using a Retsch instrument (10 min, 30 Hz; adaptors were pre-chilled at −20°C). The lysate was cleared by centrifugation for 10 min at 13 000 rpm at 4°C and collected in a fresh reaction tube. Approximately 30 µl of the lysates was collected in a different tube for RNA extraction and Northern blotting of the input samples as a quality control step.

While the lysates were being prepared, affinity purification columns were set up in a 4°C room. ∼70 µl of amylose (New England Biolabs) were added to 2 ml Bio-Spin disposable chromatography columns (Bio-Rad). Amylose beads were washed three times with 2 ml of Buffer A. Then, 1 ml of Buffer A with 250 pmol of MS2-MBP coat protein was added to the closed column, followed by incubation with rotation at 4°C. After 5 min, the column was open, and the MS2-coat protein was allowed to run through the column and collected in a separate tube. This incubation step was repeated one more time, and then the column was washed once with 1 ml of Buffer A and ready to be used for affinity chromatography. All the following steps were performed at 4°C. Lysate was added to the closed column and incubated for 5 min with rotation. The flow through was collected, and the incubation step was repeated once. Next, the column was washed 6 times with each 2 ml of Buffer A. Bound RNA was eluted using 600 µl of Elution Buffer (Buffer A + 15 mM maltose). This step was repeated one more time. The RNA from the eluted affinity purification as well as the lysate input was extracted with phenol-chloroform by adding 600 μl of P/C/I (CarlRoth, #X985.1) and mixed by vigorous shaking followed by centrifugation at 13 000 rpm for 15 min. Approximately 500 µl of the aqueous phase was collected in a new tube, followed by ethanol precipitation as described above and resuspended in 20 µl of pre-warmed nuclease-free water. RNA samples were treated with DNAse I using RapidOut DNA Removal Kit (ThermoFisher, #K2981) according to the manufacturer’s protocol. As a quality control, 5 µg of the input RNA and 250 ng of the pulldown RNA were used for Northern blotting analysis using the above method.

MAPS cDNA library preparation and sequencing

DNAse I treated RNA samples were sent to Vertis Biotechnologie AG. Briefly, the RNA samples were first fragmented using ultrasound (4 pulses of 30 s each at 4°C). Then, an oligonucleotide adapter was ligated to the 3′ end of the RNA molecules. First-strand cDNA synthesis was performed using M-MLV reverse transcriptase and the 3′ adapter as primer. The first strand cDNA was purified, and the 5′ Illumina TruSeq sequencing adapter was ligated to the 3′ end of the antisense cDNA. The resulting cDNA was PCR-amplified with 13 cycles to about 10-20 ng/µl using a high fidelity DNA polymerase. The TruSeq barcode sequences, which are part of the 5′ and 3′ TruSeq sequencing adapters, are included in Table S1. The cDNA was purified using the Agencourt AMPure XP kit (Beckman Coulter Genomics) and was analyzed by capillary electrophoresis. For Illumina NextSeq sequencing, the samples were pooled in approximately equimolar amounts. The cDNA pool was size fractionated in the size range of 200–600 bp using a preparative agarose gel. An aliquot of the size fractionated pool was analyzed by capillary electrophoresis. Pooled libraries were single-read sequenced on an Illumina NextSeq 500 system using 75-bp read length.

MAPS data processing

Sequencing data preprocessing and mapping were conducted using standard bioinformatics tools. Raw reads were quality-trimmed with BBDuk (BBMap suite) (Bushnell et al. 2017) to remove adapter sequences and low-quality bases (parameters: trimq=10, k = 23, mink=11). The quality of the trimmed reads was evaluated using FastQC (Andrews 2010). High-quality reads were mapped to the reference genome L. paracasei strain B (Siedler et al. 2020) with BBMap, and the resulting SAM files were converted into sorted, indexed BAM files using SAMtools (Danecek et al. 2021). For visualization, the aligned BAM files were converted to BigWig format using bamCoverage (v3.5.1.0.0) (Ramirez et al. 2016) and loaded into the Integrated Genome Browser (IGB) (Freese et al. 2016). Gene and feature quantification was performed using featureCounts (Subread package) (Liao et al. 2014), with annotations derived from the corresponding GFF file. Differential expression analysis (DEG) was carried out using DESeq2 (v2.11.40.7) (Love et al. 2014) on the Galaxy (galaxy.eu) platform. The analysis compared RNA pulldown samples from L. paracasei transformed with plasmids containing the RNA bait (ptrRNA01_MS2) to three control groups (EV, ptrRNA01, pMS2). The DESeq2 tool was executed with default settings. Since the experimental condition had no biological replicates, dispersion estimates were derived from the three control samples. Genes with an adjusted P-value (FDR) < 0.05 and a log2 fold change (log2FC) > 1.5 were considered significantly enriched. A volcano plot was generated in GraphPad Prism (version 10.1.0) to visualize differentially expressed genes, with −log10(adjusted P-value) on the y-axis and log2 fold change (log2FC) on the x-axis. Principal component analysis (PCA) plots and sample-to-sample distance heatmaps were generated as quality control to assess clustering patterns. Potential base-pairing between the cellular RNAs and tracr-S was analyzed for the RNAs above log2FC threshold by using IntaRNA (Mann et al. 2017), an RNA–RNA interaction prediction tool. All intaRNA queries and target input sequences along with some main outputs are available in Table S2.

Sequence comparison and identification

We used OrthoANI to measure the overall similarity between two genome sequences (Lee et al. 2016). BLASTn was used to compare the nucleotide sequence and get the identity value of genes to the Core nucleotide nr database, and filter was used when comparing genes in particular strains (Altschul et al. 1990). Protein–protein BLAST program was used to retrieve homologs of the Peg.728 and RadC in the NCBI non redundant database. Alternatively, Domain Enhanced Lookup Time Accelerated BLAST (Boratyn et al. 2012) was also used to compare the sequence to the UniProt/SwissProt and PDB databases to obtain more characterized homologs.

Results

The S. pyogenes Cas9 and tracrRNA are cytotoxic when expressed from a plasmid in L. paracasei

Genome engineering of L. paracasei has gained interest due to the importance of L. paracasei in both food and health applications (De Filippis et al. 2020). Several SpyCas9-based genome editing tools have been developed for the family of Lactobacilli, including L. paracasei (Song et al. 2017, Huang et al. 2019, Leenay et al. 2019, Zhou et al. 2019, Goh and Barrangou 2021). As part of our previous efforts to implement Cas9 in L. paracasei strain B (herein referred to as L. paracasei) (Siedler et al. 2020), we transformed pCRISPR-Cas9, an E. coli–Lactobacilli shuttle plasmid with the low copy number pAMß1 origin-of-replication (∼10 copies per cell) carrying Cas9 and its tracrRNA (Le Chatelier et al. 1994, Leenay et al. 2019) into the strain and evaluated its transformation efficiency (Fig. 1A). However, this transformation yielded at most a few colonies despite having no crRNA on the shuttle plasmid, as opposed to the many colonies obtained in the transformation of the empty vector (EV) (Fig. 1B). The same plasmid achieved high transformation efficiencies in other Lactobacilli strains (Lactiplantibacillus plantarum WCFS1 and Lacticaseibacillus casei ATCC 393) but resulted in similarly few colonies in closely related L. paracasei strains (L. paracasei ATCC 25302 and ATCC 334) (Fig. S1), suggesting that the observed cytotoxicity is present not in every Lactobacilli but also is not limited to a single strain.

Figure 1.

Plasmid-expressed tracrRNA from Streptococcus pyogenes Cas9 is cytotoxic in L. paracasei. (A) Overview of the CRISPR-Cas9 cassette from Streptococcus pyogenes integrated into an E. coli-Lactobacilli shuttle vector pJP005-ΔNisRK backbone for application in L. paracasei. The diagram depicts pCRISPR-Cas9 plasmid, which is a plasmid containing SpyCas9 and tracrRNA insert. (B) Illustrations of the plasmid constructs used in L. paracasei transformations, followed by a representative dilution plate and a CFU/µg graph representing the transformation efficiency. Darker shade represents normal size colonies, and lighter shade represents either smaller size colonies or mix sizes colonies indicating cytotoxicity. Individual dots for the transformations indicate a single biological replicate (n = 4).

To test if the poor transformation efficiency is due to the Cas9 nuclease as previously described in other bacteria (Wendt et al. 2016, Jiang et al. 2017), we removed either cas9 (ptrRNA01) or tracrRNA along with each promoter from the plasmid (pCas9). Interestingly, we observed that both plasmids yielded many colonies although the colonies were smaller than those transformed with the EV (Fig. 1B). These results indicate that tracrRNA and Cas9 separately cause cytotoxicity, albeit to a lesser extent than when combined. The observed cytotoxicity from tracrRNA was intriguing given that Cas9-independent tracrRNA cytotoxicity has never been reported before to our knowledge, so we sought to better understand how the tracrRNA could drive such a phenotypic effect. We also measured growth curves of L. paracasei transformed with pCRISPR-Cas9, pCas9, ptrRNA01, and pEV. All constructs exhibiting reduced CFUs or smaller colonies yielded impaired growth in liquid culture. However, the distinction between no colonies and few small colonies seen on plates was not visible in the 56-h liquid growth curves (Fig. S2). Therefore, we relied on dilution plating as a more reliable indicator of cytotoxicity.

Cas9-independent cytotoxicity is caused by tracr-S

In S. pyogenes, the tracrRNA is transcribed from two promoters (P_tr-L and P_tr-S) located upstream of cas9 in a divergent orientation relative to the cas9 promoter (P_Cas) (Deltcheva et al. 2011). These promoters each produce a distinct isoform: a long isoform (tracr-L) and a short isoform (tracr-S), with tracr-S and its promoter embedded within tracr-L (Figs 1A and 2A) (Workman et al. 2021). Our original construct lacking cas9 still contained cas9’s transcriptional start site due to overlap between the −35 regions of P_cas and P_tr-L. To focus more on tracrRNA expression, we removed the upstream region of cas9 in ptrRNA01 and retained only tracrRNA and its promoters up to the −59 position of P_tr-L (ptrRNA02) (Fig. 2B). Transformation of ptrRNA02 yielded no colonies in L. paracasei, further indicating that the tracrRNA region in the shuttle plasmid caused the observed cytotoxicity (Fig. 2B).

Figure 2.

tracrRNA cytotoxicity is caused by the short isoform of tracrRNA. (A) The diagram depicts ptrRNA02 plasmid, which contains only tracrRNA and its promoters up to the −59 position. It highlights the locations of the P_tr-L and P_tr-S promoters and the location of the mutations introduced in *P_tr-L. The expected size of the tracr-S and tracr-L transcripts are indicated. (B) Illustrations of the plasmid constructs, followed by a representative of dilution plates and a CFU/µg graph representing the transformation efficiency of L. paracasei transformed with tracrRNA-related constructs. Darker shade represents normal size colonies, and lighter shade represents either smaller size colonies or mix size colonies indicating cytotoxicity. Each dot indicates a single biological replicate (n = 4). Data for ptrRNA01 and the EV transformations are identical as in Fig. 1B, as the experiments were performed in parallel.

We next explored the roles of tracr-L and tracr-S in the observed cytotoxicity. We deleted P_tr-L from ptrRNA02 while retaining the native P_tr-S and tracr-S sequence (ptrRNA03), which resulted in no colonies (Fig. 2B). We next deactivated the tracr-S promoter (P_tr-S) in ptrRNA02 by introducing two T-to-C mutations at the −10 element of P_tr-S (referred to as *P_tr-S), which was previously shown to abolish tracr-S expression yet retain tracr-L expression (Workman et al. 2021) (Fig. 2A and B). Transformation of ptrRNA02_*P_tr-S in L. paracasei yielded a similar number of colonies and colony sizes as the non-cytotoxic EV control (Fig. 2B), indicating that this plasmid no longer caused cytotoxicity. These results indicate that the observed Cas9-independent cytotoxicity is due to tracr-S and not tracr-L.

Deletion of a putative transcriptional regulator alleviates Cas9 and tracrRNA-linked cytotoxicity

Despite the cytotoxicity, we occasionally obtained a few colonies when transforming the pCRISPR-Cas9 plasmid encoding cas9 and tracrRNA. As these colonies may contain mutations that relieve cytotoxicity, we performed whole genome sequencing (WGS) on three colonies obtained from the transformation of pCRISPR-Cas9 (Fig. 3A). Sequencing analysis revealed that one escape colony (E1) contained a transposon insertion in the Cas9-tracrRNA promoter region, potentially disrupting transcription and enabling survival, while another colony (E2) had a large deletion removing the entire cas9-tracrRNA locus and the ColE1 ori as well as a frameshift mutation in the chromosomal permease gene peg.1892, associated with N-acetyl-D-glucosamine ABC transport (Table S3). Separately, the third colony (E3) had a transposon insertion in the pAM beta ori of the plasmid as well as in the chromosomal peg.728 gene encoding a putative transcriptional regulator with an XRE-family helix-turn-helix (HTH) domain. By comparing mapped reads for E1 and E3, we inferred that E3 carried plasmids with ~10 times fewer copies (Table S3), suggesting reduced Cas9-tracrRNA expression. However, the 1.6-kb transposon insertion in peg.728 raised the possibility that this gene contributes to Cas9 and tracrRNA cytotoxicity.

Figure 3.

Removal of the putative transcription regulator Peg.728 alleviates Cas9 and tracrRNA cytotoxicity. (A) Schematic illustration of the WGS process used to identify escaper colonies (E1, E2, E3) that survive the cytotoxicity of pCRISPR-Cas9, followed by deletion of peg.728 to generate Δpeg.728 mutant of L. paracasei to validate its importance in CRISPR-Cas9 cytotoxicity observed in L. paracasei. (B) Illustrations of the plasmid constructs, followed by a representative of dilution plates and a CFU/µg graph representing the transformation efficiency of L. paracasei wild-type and Δpeg.728 transformed with Cas9 and tracrRNA constructs as indicated. For each dilution plate image, the top bar shows the transformation of wild-type L. paracasei, and the bottom bar shows the transformation of Δpeg.728. Darker shade represents normal size colonies and lighter shade represents either smaller size colonies or mix size colonies indicating growth defects (n = 4). Data for wildtype transformations are identical as in Figs 1B and 2B, as the experiments were performed in parallel.

To interrogate this gene’s contribution to cytotoxicity, we deleted peg.728 from L. paracasei (Δpeg.728) and assessed the transformation of plasmids containing cas9 and tracrRNA (Fig. 3B). For all plasmids exhibiting cytotoxicity, deleting peg.728 resulted in an increase in colony counts as well as increased colony sizes. For two plasmids that originally yielded no colonies (ptrRNA02, ptrRNA03), deleting peg.728 resulted in many colonies that were small, indicating that the deletion did not fully reverse cytotoxicity. Thus, peg.728 contributes to the cytotoxicity caused by tracr-S.

Deleting peg.728 de-represses tracr-L expression

As peg.728 encodes a putative transcriptional regulator (Table S4), we assessed tracrRNA expression in the wild-type and Δpeg.728 strains by Northern blotting analysis (Fig. 4A). In the Δpeg.728 strain transformed with ptrRNA01, both tracr-S and tracr-L were detected. We also observed tracr-S and tracr-L with a 3′ extension based on a downstream probe (Fig. 4A). In the wild-type strain, tracr-S was expressed similarly as in Δpeg.728, but tracr-L was not detected (Fig. 4B). Using ptrRNA02_*Ptr-S with the tracr-S promoter deactivated, we again observed expression of tracr-L in the Δpeg.728 strain but not the wild-type strain (Fig. 4B). These findings reveal that peg.728 contributes to tracr-L repression either directly or indirectly.

Figure 4.

tracr-L expression is repressed via peg.728 in L. paracasei. (A) Northern blotting results of the wild-type and Δpeg.728 L. paracasei transformed with initial tracrRNA construct ptrRNA01. Three oligo probes were used to detect tracrRNA isoforms. Probe 1: binds at the anti-repeat to detect both tracr-S and tracr-L. Probe 2: binds at the promoter of tracr-S (at −1 up to −20 position of P_tr-S), to detect only tracr-L isoform. Probe 3: binds at the sequence after the poly-U tail stretch of tracrRNA intrinsic terminator to detect isoforms with 3′ end extension. The 5S rRNA was used as loading control. (B) Northern blotting results of the wild-type and Δpeg.728 L. paracasei transformed with tracrRNA construct with removed 5′ UTR of Cas9, leaving only upstream of P_tr-L up to −59 position and deactivated tracr-S promoter (ptrRNA02_*Ptr-S). Three oligo probes were used to detect tracrRNA isoforms. Probe 1: binds at the anti-repeat to detect tracr-S and tracr-L. Probe 4: binds at the 3′ end of the tracrRNA, which is the stem loop 3 region of tracr-S. Probe 5: binds at the tracr-L region. The 5S rRNA was used as loading control.

The cytotoxicity of tracr-S is sequence-dependent and involves the anti-repeat and nexus regions

To explore whether the sequence of tracr-S is important in cytotoxicity, we replaced the tracr-S sequence in ptrRNA03 with an alternative tracrRNA sequence from Lacticaseibacillus rhamnosus GG Cas9 (LrhCas9) (Fig. 5A). Transformation of the resulting ptrRNA03Δtr-S:: Lrh construct resulted in colony numbers and sizes comparable to the EV in both the wild-type and Δpeg.728 strains (Fig. 5B). The LrhCas9 tracrRNA was also highly expressed under P_tr-S (Fig. S3), ruling out major changes in RNA expression when swapping tracrRNAs. Therefore, tracr-S cytotoxicity is linked to its sequence.

Figure 5.

The cytotoxicity of tracr-S is sequence-dependent, involving the anti-repeat and nexus regions. (A) Schematic representation of the S. pyogenes tracr-S secondary structure, highlighting its major domains: anti-repeat (purple), nexus (brown), linker (blue), stem-loop 2 (orange), and stem-loop 3 (green). Below, sequence alignments show wildtype tracr-S, a chimeric version with the anti-repeat and nexus replaced with tracrRNA sequence from L. rhamnosus (tr-SΔ1,2:: Lrh), and a version where the entire tracr-S is replaced by the L. rhamnosus tracrRNA while retaining the native promoter (Δtr-S:: Lrh). Sequences shared across all three variants are marked with asterix. (B) Schematic illustration of tracr-S and the deletions or chimeric modifications introduced into its sequence. Shown alongside are representative dilution plates and a CFU/µg DNA graph indicating transformation efficiency in wild-type and Δpeg.728 L. paracasei strains transformed with the indicated plasmids. For each dilution plate image, the top bar shows the transformation of wild-type L. paracasei, and the bottom bar shows the transformation of Δpeg.728. Darker shade represents normal size colonies and lighter shade represents either smaller size colonies or mix size colonies indicating growth defects (n = 4). Data for ptrRNA03 and EV transformations are identical as in Fig. 3B, as the experiments were performed in parallel.

We next determined which regions of tracr-S contribute to its toxic effects. In the duplex form where tracrRNA interacts with crRNA, the tracrRNA can be divided into several domains based on its structure: the anti-repeat, nexus, linker, two additional stem-loops forming the tracrRNA tail, and a poly-U tail (Briner et al. 2014) (Fig. 5A). In the wild-type strain, deleting the anti-repeat resulted in no colonies, whereas deleting the nexus led to high colony counts with tiny colonies. Deleting both the nexus and anti-repeat together resulted in high colony counts with larger colonies, comparable to those observed with ptrRNA03 transformation in Δpeg.728, although the colonies were smaller than those observed in the EV transformation (Fig. 5B). Further deletion of the remaining domains resulted in no colonies in the wild-type strain and small colonies in the Δpeg.728 strain, indicating that a truncated tracr-S was more cytotoxic than full-length tracr-S likely due to elimination of the termination loop structure, leading to aberrant transcription (Fig. S4). These results indicate that the anti-repeat and nexus regions are key contributors to tracr-S cytotoxicity, whether through direct action or inducing misfolding of the tracrRNA that reverses cytotoxicity.

As deleting the anti-repeat and nexus regions removes cytotoxicity, we tested the impact of replacing these regions with the same regions from LrhCas9 tracrRNA to maintain the original length (ptrRNA03Δ1,2:: Lrh) (Fig. 5A and B). This modification also resulted in high colony counts and small colonies, similar to the anti-repeat and nexus deletion variants. While we still cannot exclude potential effects on RNA folding or contributions from other regions in the tracr-S, the comparable phenotype suggests that the anti-repeat together with nexus sequences of SpyCas9 tracrRNA contributes to the observed cytotoxicity (Fig. 5B).

Cytotoxicity of tracr-S in L. paracasei is linked to interactions with cellular RNAs

Given the importance of the tracr-S anti-repeat and nexus regions, we hypothesized that tracr-S could be hybridizing to cellular RNAs in L. paracasei, causing interference and cytotoxicity even in the absence of Cas9. Previous studies have shown that tracrRNA can interact with cellular RNAs through the anti-repeat region without crRNA (Jiao et al. 2021). To identify interacting RNA candidates, we performed MS2 aptamer affinity purification coupled with RNA sequencing (MAPS) (Lalaouna and Masse 2015, Correia Santos et al. 2021). As part of this approach, stem loops 2 and 3 of the tracrRNA are each replaced with an MS2 aptamer (Dong et al. 2018). The expressed tracrRNA-MS2 then serves as an RNA bait, and interacting RNAs were co-purified and identified through next-generation sequencing (Fig. 6A). As a quality control, we confirmed that the cytotoxicity and transcriptional profiles of this MS2-tagged construct (ptrRNA01_MS2) were comparable to those observed with the tracrRNA plasmid (ptrRNA01) (Fig. 6B and S5). Several host RNAs were enriched in the pulldown sample containing tracrRNA-MS2 compared to controls expressing tracrRNA without the aptamer, the aptamer alone, or an EV (Fig. 6C and Table S5). The top five enriched transcripts mapped to radC (peg.1635), a hypothetical protein (peg.148), a protein tyrosine phosphatase (peg.727), a DegV family protein (peg.1813), and a glycopeptide antibiotic resistance protein (peg.1758) (Fig. 6C and Table S5).

Figure 6.

The tracrRNA interacts with cellular RNAs in L. paracasei. (A) The experimental setup of MS2-aptamer affinity purification coupled with sequencing (MAPS) to obtain tracrRNA interacting cellular RNAs. (B) Quality control of MAPS samples by Northern blotting analysis of wild-type L. paracasei carrying EV, plasmid containing tracrRNA (ptrRNA01), plasmid containing ptrRNA01_MS2 where MS2-aptamer was fused to tracrRNA by replacing the stem-loop 2 and 3, and plasmid containing MS2 aptamer under tracrRNA promoter. “I” indicates the input samples before pulldown, and “P” indicates samples after pulldown. Probe 1 binds at the anti-repeat, probe 6 binds at the MS2 aptamer region, and 5S rRNA probe was used to detect the loading control. (C) Volcano plot showing the differential abundance of RNAs in the MAPS experiment compared to controls. Each point represents an RNA, with the x-axis indicating log2 fold-change (FC) and the y-axis showing −log10 (adjusted P-value). The threshold for enriched RNAs is set at FC > 1.5 (dotted line). RNAs significantly enriched in the pulldown appear in the upper right, while depleted RNAs are in the upper left. Non-significant RNAs cluster near the center. Turquoise points represent RNAs above the FC threshold, labeled with SEED gene identifiers, highlighting those most abundant in the pulldown sample and potentially interacting with the bait tracrRNA-MS2. Gene descriptions corresponding to these identifiers are provided in Table S5. (D) Genome browser (IGB) view of mapped reads from the tracrRNA-MS2 pulldown experiment across five enriched target RNAs (radC, peg.148, peg.727, peg.1813, peg.1758), shown alongside IntaRNA-predicted interaction profiles. Read coverage tracks are displayed for four conditions: EV, tracrRNA (ptrRNA01), MS2-tagged tracrRNA (ptrRNA01_MS2), and MS2 aptamer alone (pMS2). Below each coverage plot, heatmaps represent interaction energies (minimal free energy, MFE in kcal/mol) reflecting the sum of hybridization energy and unfolding energies in both RNAs between tracr-S and each target transcript. The horizontal axis shows the position along each target RNA (including 250 bp upstream and downstream of the coding sequence), and the vertical axis shows positions along tracr-S (nucleotides 1–89). Stronger predicted interactions (lower MFE) are shown in darker shades. At the bottom is the top-ranked IntaRNA-predicted interaction for each target, including where it localized, base-pairing models, and the corresponding MFE. Interacting regions of tracr-S are highlighted. Gene models are annotated with gene identifiers.

To predict possible base-pairing interactions between tracr-S and these RNAs, we used IntaRNA (Mann et al. 2017) with target sequences that included the coding sequence (CDS) as well as 250 bp upstream (US) and downstream (DS) of each CDS (Fig. 6D and Table S4). Interestingly, interactions with radC, peg.148, and peg.1758 mRNAs were mostly localized near the 5′ end of the transcripts, a region where RNA–RNA interactions are more likely to interfere with translation initiation. Among them, radC stood out due to its high abundance, and the predicted binding sites spanned primarily the anti-repeat and nexus regions of tracr-S. Although RadC is poorly characterized, it has been implicated in DNA repair and oxidative stress response (Saveson and Lovett 1999, Attaiech et al. 2008) and is upregulated during oxidative stress (Yuan et al. 2021). Notably, we observed that tracr-S cytotoxicity was alleviated under anaerobic growth (Fig. S6), consistent with the possibility that radC is a relevant target.

To assess whether the interaction with radC was responsible for cytotoxicity, we introduced mutations into tracr-S, which disrupted its top three predicted interactions with radC (Fig. S7). Some of these mutants no longer have strong predicted interactions between the anti-repeat and nexus regions of tracr-S and radC. However, all variants remained cytotoxic (Figs. S7 and S8). Further analysis revealed that, despite the loss of predicted interaction with radC, these mutants still retained predicted interactions of the tracrRNA with peg.148 and peg.1758 near their 5′ ends. This suggests that these interactions may also contribute to the cytotoxic phenotype (Fig. S8).

To better understand which interactions contribute to cytotoxicity, we expanded our IntaRNA analysis to include other tracr-S variants with altered sequences. For the cytotoxic variants, we consistently observed predicted interactions with radC, peg.148, and peg.1758 at the 5′ UTR or at the beginning of the coding sequence. In contrast, non-cytotoxic variants lacked strong interactions with these targets (Fig. S8). These results suggest that cytotoxicity is not solely due to the predicted interaction between tracr-S and radC mRNA and instead likely arises from interactions with other RNAs.

Finally, we analyzed the status of the five top interacting RNAs in the related Lactobacilli strains in which the tracrRNA construct was cytotoxic (L. paracasei ATCC 334 and ATCC 25302) or not cytotoxic (L. casei ATCC 393, L. plantarum WCFS1) (Fig. S1). All five RNAs were highly conserved among L. paracasei strains, while L. casei contained radC, peg.1813, peg.727, and peg.1758 and L. plantarum contained radC and peg.1813. Notably, the radC and peg.1813 in L. casei and L. plantarum exhibited lower similarity to those in L. paracasei (Table S6), which were associated with reduced predicted interactions with the S. pyogenes tracrRNA (Fig. S9). These genomic differences could explain why the S. pyogenes tracrRNA construct is cytotoxic in L. paracasei but not in L. casei or in L. plantarum.

Discussion

SpyCas9 has been widely used as genome-editing tools due to its precision and adaptability (Doudna and Charpentier 2014, Arroyo-Olarte et al. 2021, Pacesa et al. 2024). However, heterologous expression of SpyCas9 in non-native bacterial hosts can pose several challenges, including cytotoxicity linked to Cas9 nuclease activity and off-target effects (Vento et al. 2019, Rostain et al. 2023). While Cas9-associated cytotoxicity is well-documented, the potential cytotoxic effects of other CRISPR-Cas9 components remain largely unknown. In this study, we found that plasmid-based expression of the S. pyogenes tracr-S is cytotoxic in L. paracasei. While less likely, we cannot rule out the possibility that tracr-S destabilizes the plasmid or inhibits expression of the antibiotic resistance marker, sensitizing the cells to the applied antibiotic.

While exploring the basis of cytotoxicity, we discovered that peg.728, which encodes a putative transcription regulator with a HTH domain (Aravind et al. 2005), represses tracr-L expression and contributes to cytotoxicity. Peg.728 modestly resembles to the CpG repressor from Lactobacillus phage φg1e (26% similarity) and the PBSX repressor from Bacillus subtilis (29% similarity) (Table S4), which are both known to regulate lysogeny and lytic growth transitions (Sauer et al. 1990, Kim and Little 1993, McDonnell and McConnell 1994, Kakikawa et al. 2000), suggesting that peg.728 functions as a repressor in wild-type L. paracasei that could be directly or indirectly blocking tracr-L expression. This highlights potential regulatory mismatches between the heterologous CRISPR-Cas9 system and the native transcriptional machinery of L. paracasei.

Our findings suggest that tracr-S cytotoxicity may be linked to sequence-dependent interactions with host RNAs (Jiao et al. 2021). Using MS2-affinity purification coupled with RNA sequencing (MAPS) (Lalaouna and Masse 2015, Correia Santos et al. 2021), we identified several co-purified host RNAs, with radC as the most enriched transcript in the pulldown (Fig. 6B). However, disrupting the predicted interaction between tracr-S and radC did not alleviate cytotoxicity, suggesting that cytotoxicity arises from interactions with other RNAs. Although radC, peg.148, and peg.1758 emerged as likely contributors based on enrichment and predicted interactions, we cannot rule out the involvement of other transcripts, including those less enriched in the pulldown but still capable of forming deleterious interactions.

The observed cytotoxicity following heterologous tracrRNA expression raises the possibility that off-target RNA interactions may contribute to the evolution of tracrRNA-dependent CRISPR-Cas systems. Given that tracrRNA is among the most abundant small RNAs in S. pyogenes (Deltcheva et al. 2011), horizontal transfer of the CRISPR-Cas locus into new genomic contexts or the evolution of the CRISPR-harboring strain could increase the likelihood of such interactions (Chylinski et al. 2013, Faure et al. 2019). These pressures may, in turn, contribute to sequence divergence in tracrRNAs, host transcripts, or regulatory elements to maintain compatibility (Umu et al. 2016).

Beyond the complete mechanism underlying tracr-S cytotoxicity, several questions remain from this study. First, it is still unclear whether Peg.728 directly represses tracr-L or contributes to its repression indirectly. Furthermore, the impact of tracr-L repression in tracr-S cytotoxicity remains unexplored. Given their closely spaced tandem arrangement, it is possible that tracr-L repression results in the overexpression of tracr-S (Chauhan et al. 2024), although we did not observe this effect directly (Fig. 4A). It also remains uncertain whether cytotoxicity is solely driven by RNA–RNA interactions or if other factors—such as host RNA-binding proteins (e.g. Hfq) or nucleases (e.g. RNase III)—are involved, given the structured nature of tracrRNA (Deltcheva et al. 2011, Otaka et al. 2011, Lalaouna et al. 2013). Further tests, including electrophoretic mobility shift assays to assess Peg.728 interactions, pulldown experiments to identify additional interacting proteins and RNAs (Waters et al. 2017, Melamed et al. 2018, Iosub et al. 2020), and targeted knockouts, could provide deeper insights into the mechanisms underlying tracr-S toxicity.

As part of this work, we also observed that Cas9 expression alone is cytotoxic. Although the underlying cause remains unexplored, our finding that tracr-L is repressed in L. paracasei suggests this absence may contribute to Cas9 cytotoxicity, as tracr-L has been shown to autoregulate Cas9 by forming a Cas9: tracr-L repressor complex targeting P_Cas (Workman et al. 2021). However, this cytotoxicity could also result from plasmid loss, which sensitizes the cells to the antibiotic (Silva et al. 2012).

Overall, our findings highlight the challenges in adapting heterologous CRISPR-Cas9 systems to non-native hosts, particularly due to unintended interactions. The observed cytotoxicity suggests that careful optimization of tracrRNA or sgRNA expression and the selection of regulatory elements tailored to the host organism are crucial for successful implementation (Wendt et al. 2016, Jiang et al. 2017, Liu et al. 2017). Using alternative CRISPR-Cas9 systems from closely related species may offer a more suitable approach for heterologous CRISPR-Cas9 genome editing, potentially reducing cytotoxicity.

Author contributions

Conceptualization: J.M.V., C.L.B., A.Q.A.; Data curation: A.A., J.M.V.; Formal analysis: A.A., J.M.V., I.K.; Investigation: A.A., J.M.V., I.K., T.A.; Methodology: J.M.V., A.A., T.A.; Writing – Original Draft: A.A.; Writing – Review & Editing: A.A. and C.L.B. with subsequent input from all authors; Visualization: A.A.; Supervision: C.L.B.; Funding acquisition: C.L.B.

Conflict of interest

C.L.B. is a co-founder and officer of Leopard Biosciences, co-founder and Scientific Advisor to Locus Biosciences, and Scientific Advisor to Benson Hill. The other authors declare no competing interests.

Funding

This work was supported by a grant from the DFG - Deutsche Forschungsgemeinschaft (BE 6703/1-2 to C.L.B.).

Data availability

Raw sequencing data from the MAPS experiment are available in the NCBI Sequence Read Archive (SRA) associated with BioProject PRJNA1214207. Plasmids generated in this study and all original data are available upon reasonable request.