A Cas12a toolkit for fast and flexible gene editing and gene control in Group B Streptococcus

Figure 1 diagrams the design and potential uses of the two plasmids that underpin our suite of Cas12a‐based GBS genetic tools. pGBSedit is designed for genome editing, while pGBScrispri is for CRISPRi. Both plasmids are modified versions of the Cas12a‐based shuttle plasmid pJC005, which was previously developed for E. faecium genetic modification (Chua and Collins 2022). pGBSedit and pGBScrispri both encode conserved gram‐negative and gram‐positive origins of replication so that CRISPR RNA (crRNA) and homology arm cloning can be performed in Escherichia coli followed by electroporation transformation of purified plasmid into GBS.

pGBSedit encodes wild‐type (WT) Acidaminococcus Cas12a whereas pGBScrispri encodes a modified, catalytically inactive (E993A) dCas12a that can be used for inducible, targeted gene repression. In both versions of the plasmid, the cas12a/dcas12a gene is downstream of a Pxyl/tet inducible promoter bioengineered to maximize the dynamic range in response to its anhydrotetracycline (aTC) inducer, with minimal expression in the uninduced state and strong expression upon aTC exposure (Helle et al. 2011). TetR, the promoter's repressor, is constitutively expressed on the plasmid.

The crRNA cassette is identical in pGBSedit and pGBScrispri. Driven by a constitutive P3 small RNA‐optimized promoter, the crRNA region harbors a pair of apposed Esp3I restriction enzyme cut sites that, when digested, yield incompatible sticky ends into which a custom dsDNA spacer sequence can be ligated to encode a functional crRNA.

Once an active crRNA is established, with genomic target site complementarity adjacent to a Cas12a PAM, aTC induction of WT cas12a from pGBSedit leads to reliable selection against all GBS cells bearing the target sequence‐PAM combination. Viable CFUs are enriched for variants that escaped selection by mutation of the target site.

One experimentally useful mechanism of chromosomal repair is homologous recombination with an editing template that can be cloned onto the pGBSedit plasmid at a XhoI restriction enzyme recognition site downstream of the crRNA cassette. To demonstrate the effectiveness of this approach, we created an editing template in which the eGFP gene and a P23 constitutive promoter are flanked by approximately 500‐nt homology arms that direct it to an intergenic region conserved in CNCTC 10/84 and A909 GBS strains.

After cloning this editing template into pGBSedit, along with a targeting crRNA directed to the intended genomic insertion site (generating pGBSedit:eGFP1, Figure 2A), we transformed WT CNCTC 10/84 and A909 with the targeting plasmid. In the uninduced state without aTC, a liquid culture of the transformed GBS strain grew as an unselected lawn on solid agar media (Figure 2B,C, left side). However, when exposed to aTC added atop the agar media, there is dramatic and reliable selection against the WT background that comprises most of the lawn (Figure 2B,C, right side). In eight independent A909 biological replicates quantified for selection efficiency, the mean rate of aTC resistance was 9.4 × 10⁻⁵ (Figure 2D) Among the aTC‐resistant colonies, a substantial fraction carried the intended eGFP insertion as confirmed by colony PCR (Figure 2E) and expressed eGFP fluorescence (Figure 2F,G).

An anticipated use of the Cas12a tools in GBS is markerless gene (or other genomic region) deletion. To demonstrate the feasibility of this application and to quantify various possible Cas12a selection endpoints, we created pGBSedit:ΔcovR1 to create a markerless deletion of the covR gene. CovR is part of the two‐component signaling system that regulates numerous GBS genes (Mazzuoli et al. 2021; Claverie et al. 2024), including the cyl operon responsible for the biosynthetic pathway of β‐hemolysin/cytolysin (βHC), a pigmented, hemolytic GBS toxin (Armistead et al. 2020; Randis et al. 2014). We chose this target gene because its deletion leads to cyl de‐repression and an easily assayed hyperpigmented, hyperhemolytic phenotype (Lupo et al. 2014).

pGBSedit:ΔcovR1 has an anti‐covR targeting crRNA directed to the target sequence next to the 3′‐TTTC PAM at nucleotide position 436 in the covR gene. In the pGBSedit XhoI site, we cloned an approximately 1000‐bp editing template composed of fused 500‐bp covR upstream and downstream homology arms (Figure 3A).

After transformation of purified pGBSedit:ΔcovR1 into WT A909, we performed 14 independent aTC selection experiments during which we quantified the unselected background CFU concentration and the aTC selection output. The mean rate of selection escape across our 14 biological replicates was 1.1 × 10⁻⁴ (range 1 × 10⁻⁵—3.5 × 10⁻⁴, Figure 3B). We performed colony PCR on colonies from 8 independent biological replicates of the transformation step. Among the 55 PCR reactions that generated an electrophoresis gel band, 15 (27%) were of the expected size for the in‐frame gene deletion, while the remaining were consistent with wild type (Figure 3C). Inspection of the 15 suspected knockouts showed that all had the expected hyperpigmented appearance when grown in liquid culture (Figure 3D). To confirm that the knockout strain was hyperhemolytic, we performed three biological replicates of hemolysis assays with one of the genotypic knockouts from our experiment, which confirmed the expected phenotype (Figure 3E).

To understand the influence of editing template homology arm length on Cas12a markerless chromosomal deletion rates, we performed an experiment in which we created a partial deletion of the eGFP gene in the A909eGFP strain created as described above, using two different lengths of homology arms, 500 and 1000 bp. The crRNA targeting site in this experiment was located downstream of eGFP, outside of the open reading frame, so that alternative end‐joining events (see below section) would not result in a phenotypic knockout. Only template‐based editing, from the homology arms whose union deletes the promoter and first approximately 20% of eGFP, would result in phenotypic loss of fluorescence. This design allowed us to examine phenotype to surmise genotype. It revealed that the 1000‐bp homology arms were more efficient than the 500‐bp homology arms, leading to a 65% versus 39% success rate (Figure S1). This difference must be weighed against diminishing transformation rates with pGBSedit bearing longer homology arms, and the potentially increased cost and time requirements of bioengineering longer homology arm editing cassettes.

To examine whether Cas12a‐mediated mutations at the genomic target site occur in the absence of homologous recombination with an editing cassette, we generated covR and eGFP targeting versions of pGBSedit without homology‐based editing templates. pGBSedit:covR2 and pGBSedit:eGFP2 target the covR and eGFP coding sequences, respectively; pGBSedit:P23 targets the P23 promoter upstream of eGFP (Figure 4A,B). We transformed WT A909 with pGBSedit:covR2 and A909 with chromosomal eGFP (A909eGFP) with pGBSedit:eGFP2 or pGBSedit:P23. After selection with aTC, we examined the surviving colonies for phenotypic changes, noting that the majority of the WT A909 + pGBSedit:covR2 colonies were hyperpigmented and hyperhemolytic, and the A909eGFP + pGBSedit:eGFP2 and A909eGFP + pGBSedit:P23 colonies were not fluorescent. We performed colony PCR of the genomic regions flanking the three Cas12a target sites, then Sanger sequenced the resulting bands. This analysis identified two genomic deletions surrounding the Cas12a targeted sites in the covR targeted colonies (Figure 4C). One deletion was 2 bp in size and the other was 60 bp. Sanger sequencing revealed one genomic deletion each for the P23 promoter and eGFP targeted strains (Figure 4D). Predictably, mutations that delete significant portions of the P23 promoter and coding sequence frameshift mutations lead to loss of gene function in phenotypic testing (Figure 4E,F).

While the mutations that emerge from this template‐less CRISPR editing approach are unpredictably sized, we find that the speed and ease with which they can be generated make this an attractive protocol for rapid creation of mutants in regions of interest. In our benchmarking experiments, the full pipeline of crRNA cloning, GBS transformation and aTC selection, and PCR/Sanger sequence confirmation could all be completed in a week. Promising genomic changes made through this approach can then be more rigorously verified by markerless deletion or allelic exchange with an editing template.

A persistent concern with CRISPR‐based genomic editing is the potential for off‐target mutagenesis. Cas12a editing systems have been shown to have promiscuous nickase activity (Murugan et al. 2020) that could drive higher than baseline rates of mutation if erroneous repair mechanisms were widely activated. To investigate this possibility, we performed Illumina whole‐genome sequencing of mutant strains recovered and tested in the above experiments. After unbiased de novo alignment and contig scaffolding against a reference GBS genome, we identified between 1 and 5 SNPs across our sequenced mutant strain genomes (Figure S2). Additionally, a prophage resident in the A909 genome was observed to have excised in one of our sequenced strains, which we have encountered in previous studies unrelated to Cas12a mutagenesis (Gopalakrishna et al. 2023). Given estimated secular mutation rates in GBS of approximately 10⁻⁹ mutations per bp per generation (Drake et al. 1998; Cuny et al. 2021; Rosinski‐Chupin et al. 2013; Sheppard et al. 2018), or approximately 0.002 mutations per 2 M‐bp genome per generation, and roughly 20 generations per laboratory outgrowth, 1–5 fixed SNPs in a strain that is 5–10 outgrowth steps from its progenitor roughly matches expectations and does not indicate widespread or biased off‐site mutagenesis activity by Cas12a.

To explore the potential of using dCas12a as a tool for inducible gene silencing, we used crRNA targeting βHC biosynthetic enzymes encoded by the cyl operon and crRNA targeting covR (Figure 5A). Previous studies of dCas12a‐based gene silencing have demonstrated strand bias in which crRNA targeting the template strand of the gene has a stronger silencing effect than crRNA targeting the non‐template strand (Zhang et al. 2017). This is the opposite of dCas9 gene silencing, which works best when the non‐template strand is targeted by the guide RNA (dCas9 template strand targeting can actually increase gene expression in some cases) (Gopalakrishna et al. 2023; Qi et al. 2013). To examine this aspect in our system, we used both template and non‐template targeting crRNA.

We transformed CNCTC 10/84 with pGBScrispri:cylx42 (template targeting) and pGBScrispri:cylx143 (non‐template targeting). CNCTC 10/84 is a naturally hyperpigmented/hyperhemolytic GBS strain, due to a SNP in the covR promoter that results in constitutively low expression (Zhu et al. 2020), so suppressing cyl in this background is expected to reduce pigmentation and cytotoxicity. By qRT‐PCR (Figure 5B) and in vitro hemolysis assays (Figure 5C), we observed cyl downregulation in the presence of the aTC inducer and a targeting crRNA. Addition of aTC without a targeting spacer was noted to cause a small but consistent upregulation of cyl expression, both by qRT‐PCR and phenotypic assay. As previously described, the template strand targeting pGBScrispri:cylx42 plasmid led to greater gene repression than the non‐template targeting pGBScrispri:cylx143 variant.

Silencing covR in A909, which is not a naturally hyperhemolytic strain, is expected to cause cyl de‐repression and overexpression of the βHC toxin. Again, this expected result was observed by qRT‐PCR (Figure 5D) and hemolysis assays (Figure 5E), only in the presence of the aTC inducer. (In contrast to CNCTC 10/84, aTC did not measurably increase βHC expression in A909 when no covR targeting crRNA was present, see Figure 5E leftmost two columns). Together, these results demonstrate the utility of inducible dCas12a as a rapid and reliable tool for targeted gene expression knockdown in GBS.

Whole blood collection by phlebotomy of healthy adult volunteers was conducted in accordance with an approved University of Pittsburgh IRB protocol (CR19110106‐001). Following the use of blood in GBS hemolysis assays, samples were discarded and no further volunteer‐level data were collected.

LB liquid broth mix, LB agar, TS liquid broth mix, TS agar, erythromycin (Erm), anhydrotetracycline, M17 liquid medium mix, glycine, Esp3I, T4 ligase, T4 polynucleotide kinase, XhoI, Gibson assembly master mix, DpnI, Q5 Taq polymerase, Qiagen QIAprep Spin Miniprep Kit, M17 broth, PEG6000.

GBS strains A909 (serotype Ia, sequence type 7) and CNCTC 10/84 (serotype V, sequence type 26) (Zhu et al. 2020; Nizet et al. 1996; Lancefield et al. 1975) and their derivatives were grown at 37°C under stationary conditions in tryptic soy (TS) medium (Fisher Scientific cat. # DF0370‐17‐d). GBS was supplemented with 5 μg/mL Erm as needed for selection. E. coli strain DH5α was grown at 37°C with shaking in Luria‐Bertani (LB) medium (Fisher Scientific cat. # DF9446‐07‐5) supplemented with 300 μg/mL Erm as needed for selection.

Statistical analyses were performed with GraphPad Prism v. 10.2.3 for macOS.

Plasmid backbone pJC005 (Chua and Collins 2022) was used as a backbone for pGBSedit and pGBScrispri. A modular crRNA expression cassette with a small RNA promoter (P3) driving the expression of the repeats and Esp3I editable sham spacer was added to both vectors. Digestion with Esp3I allows for complete removal of the sham spacer and ligation of a new spacer as annealed DNA oligos with matching sticky ends (protospacer). A unique XhoI cut site was also added at the end of this cassette for linearization of the plasmid in use with Gibson Assembly addition of homology arms. This modular cloning cassette was designed and ordered as a gene fragment from Twist Bioscience, then added to both vectors via Gibson assembly. For plasmid‐based complementation or overexpression, we used the pGBScomp vectors (High/Low expression; Addgene, see Supplemental Protocol Data S1 "Complementing KO strains with pGBScomp"), which accept ORFs by SalI linearization and Gibson assembly. The expression difference is from the promoter sequence. pGBScomp High has a P23 promoter and pGBScomp Low has a P_spac promoter upstream of the gene insertion site.

Cas12a‐compatible GBS genomic target sites with TTTV PAM sequences were identified with the aid of the CRISPick server hosted by the Broad Institute (Pedrozo et al. 2022). For dCas12a CRISPRi, spacer sequences complementary to the plus‐strand should be selected; for Cas12a mutagenesis, plus‐ or minus‐strand spacer sequences are equally effective. Also see Figure S4.

After selecting a spacer, the 23‐nt sequence can be used to design custom ssDNA oligonucleotides for annealing to make a dsDNA protospacer, which is then cloned into Esp3I‐digested and gel‐purified pGBSedit or pGBScrispri.

Editing templates included homology arms with approximately 500–1000 bp of overlap (on each end of the intended deletion or insertion). The Cas12a target site, determined with the aid of CRISPick, was positioned roughly equidistant from the homology arms. Editing constructs were ordered as synthesized gene fragments but could alternatively be generated through recombinant techniques depending on the intended end use. We cloned the editing construct into linearized XhoI‐digested pGBSedit using Gibson assembly and used PCR and Sanger or whole‐plasmid sequencing techniques to verify the insert.

Once cloning products had been sequence‐confirmed in E. coli, plasmids were purified using the Qiagen QIAprep Spin Miniprep Kit according to manufacturer instructions. Electrocompetent GBS was prepared and transformed as previously described (Hooven et al. 2019).

Following GBS transformation with an editing plasmid, a single colony was resuspended in 20 μL phosphate buffered saline (PBS). 2 μL of this resuspension was used as a template for a colony PCR reaction to confirm the presence of the intended plasmid. Simultaneously, the remaining 18 μL of the PBS‐resuspended sample was seeded into a 10 mL TS + Erm culture, which was allowed to grow stationary at 37°C for 6–8 h.

Following this 6–8‐h outgrowth, 1:10 and 1:100 dilutions of the culture were prepared (diluted in TS + Erm). 150 μL of the following cultures were then plated on separate TS + Erm agar plates: undiluted, 1:10, and 1:100, undiluted + aTC, 1:10 + aTC, and 1:100 + aTC. For the aTC‐containing cultures for plating, we add 6 μL of a 2 mg/mL aTC stock to the 150 μL of sample being plated.

These plates were allowed to grow for 36 h at 37°C. A multiple‐log fold difference between colony counts on the aTC and non‐aTC plates after growth is indicative of successful Cas12a selection against wild type GBS. Candidate colonies from the aTC plates were selected and each resuspended in 20 μL PBS. 2 μL of this resuspension was used in a colony PCR reaction to assess genotype. Simultaneously, the remaining 18 μL of the PBS‐resuspended sample was plated for single colonies on a TS + Erm plate. Using an isolated colony from the plate struck for single colonies, we resuspended a colony in 20 μL 1× PBS and used 2 μL of this resuspension in a colony PCR of the target region. Once the mutation was confirmed by a band shift on the electrophoresis gel, we seeded the remaining 18 μL of colony resuspension into a 40 mL TS broth culture to begin plasmid curing.

40 mL TS broth cultures were inoculated using an isolated colony that has been genotypically confirmed and grown stationary at 37°C overnight. A second passage in 40 mL of TS broth was performed using 100 μL of the previous culture as inoculum. 1:10,000 and 1:100,000 dilutions of the grown second passage were plated on TS agar plates. Colonies were dual patched onto TS and TS + 5 μg/mL Erm agar plates to detect cured clones by antibiotic sensitivity profiling.

To use pGBSedit without an editing cassette to generate alternative end‐joining mutants, the same experimental procedures are followed as with homology‐driven mutagenesis. However, because free end joining mutants can be the result of indels too small for easy detection by comparing PCR bands on an electrophoresis gel, we first performed a colony PCR screen on candidate colonies, checking for the presence or absence of the wild type Cas12a target (the sequence complementary to the pGBSedit spacer). Colonies that screened negative were then further tested by PCR and sequencing to characterize indels.

Individual TS + Erm cultures were inoculated using GBS harboring a targeting pGBSedit plasmid and grown for 8 h. Serial dilutions and CFU plating on TS + Erm plates for each overnight culture were performed to access total viable CFU/mL. Simultaneously, 100 μL of each overnight culture was plated onto TS + Erm + 500 ng/mL aTC at both 1:10 or 1:100 dilutions. After 36 h of growth, colonies on aTC plates were counted and compared to the starting culture colony counts. Colonies from the aTC plates were selected and screened for genotypes by NEB OneTaq colony PCR (Cat. M0482L). Genotypes were counted based on 1% agarose gel electrophoresis PCR band sizes or sequencing results.

GBS harboring the pGBScrispri plasmid are grown in TS + Erm to maintain the plasmid. For phenotypic assays, induction of knockdown was achieved by the addition of 250 ng/mL aTC (stock 2 mg/mL in water) to cultures from isolated single colonies. Cultures were grown stationary at 37°C for 18 h.

10 mL TS broth cultures were inoculated using isolated colonies and grown overnight at 37°C shaking. Overnight cultures were washed twice with PBS before being OD₆₀₀ normalized to 1 in PBS. 10 mL of OD₆₀₀ normalized cultures were spun at 4000 RPM for 5 min, then resuspended in 1 mL of PBS to 10× concentrate. 200 μL of each sample was loaded in technical triplicates into black optical bottom 96‐well plates. GFP was excited at 488 nm, and emission was measured at 515 nm.

10 mL TS + Erm cultures were inoculated using isolated colonies and grown to mid‐log (OD₆₀₀ = 0.6) before being split into two 5 mL cultures: untreated and 250 ng/mL aTC‐supplemented. Cultures were incubated at 37°C for 1 h post treatment before being spun down and resuspended in 1 mL of RNAprotect Bacterial Reagent (Qiagen Catalogue: 76506). Bacteria were incubated in RNAprotect at room temperature for 5 min before being pelleted in microcentrifuge tubes at 15,000 RPM for 2 min. Excess RNAprotect was removed, and pellets were stored at −80°C until RNA extraction. RNA protected pellets were thawed on ice, and RNA was extracted using the Qiagen RNeasy Kit (Cat. 74104) per manufacturer protocol. RNA was eluted with warmed (95°C) ultra‐pure water in place of elution buffer. DNA was removed with the Invitrogen DNA‐free Kit (Cat. AM1906) per manufacturer protocol. cDNA was synthesized with the AppliedBiosystems High‐Capacity cDNA Reverse Transcription Kit (Cat. 4374966) per manufacturer protocol. cDNA and reverse transcription negative controls were 1:10 diluted with ultra‐pure water before 1 μL was used as template in 10 μL qPCR reactions. Bio Rad SsoAdvanced Universal SYBR Green Supermix (Cat. 1725272) was used for all qPCR reactions. Target gene mRNA measurements were normalized to the housekeeping gene sortase A (srtA) mRNA.

10 mL of heparanized human whole blood was spun down for 10 min at 2000 RPM, then washed repeatedly in 5 mL of Hank's buffered saline solution (HBSS). Packed red blood cells (PRBC) were spun for 10 min at 2000 RPM, and the supernatant was removed with a serological pipette. 500 μL of PRBC were added to 50 mL of HBSS to achieve a 1% PRBC solution. 10 mL cultures of GBS harboring the pGBScrispri plasmid were grown for 18 h with or without 250 ng/mL of aTC, then pelleted and resuspended in 10 mL of PBS. 10 mL resuspensions were OD₆₀₀ normalized to 1.0, then 10 mL of OD₆₀₀ normalized A909 cultures were concentrated 10× by pelleting and resuspending in 1 mL of PBS. 10/84 cultures were 1:10 diluted in PBS after OD normalization. Hemolysis assays were performed with 100 μL of PRBC mixed with 100 μL of bacterial suspension and incubated at 37°C for either 15 min (10/84) or 1 h (A909). Hemolysis assays were performed in 96 well V‐bottom plates in technical triplicate. After hemolysis, V‐bottom plates were spun down at 2000 RPM for 2 min, and supernatants (100 μL) were collected for each sample, and absorbance at 415 nm was used to measure released hemoglobin in the supernatant. Percent hemolysis was calculated relative to a 1% triton control.

Genomic DNA samples were prepared for whole‐genome sequencing (WGS) using the Illumina DNA Prep tagmentation kit with IDT for Illumina Unique Dual Indexes. Sequencing was performed on an Illumina NextSeq2000 platform with a 300‐cycle flow cell to generate paired‐end reads (2 × 150 bp). A 1%–2% PhiX control library was included to ensure optimal base‐calling accuracy. Initial processing—including demultiplexing, read trimming, and analytics—was performed using Illumina DRAGEN software (version 3.10.12) (Behera et al. 2024). Quality control included FastQC metrics evaluation.

Further processing and genomic assembly utilized a custom bioinformatics pipeline. Reads underwent stringent quality filtering, trimming, and duplicate grouping using BBTools. Filtered reads were aligned to the A909 GBS strain reference genome to create a high‐confidence consensus sequence, requiring a minimum depth of 10 reads at consensus sites. Quality‐filtered reads were then de novo assembled into contigs using SPAdes (Bankevich et al. 2012) in careful mode. Contigs were scaffolded onto the consensus reference using RagTag (Alonge et al. 2022), followed by patching to resolve scaffold gaps and inconsistencies. The primary scaffold from each assembly was extracted and annotated using the Prokaryotic Genome Annotation Pipeline (PGAP) (Tatusova et al. 2016). SNPs were determined by aligning mutant strain reads to assembled parent scaffolds with Bowtie 2 (Langmead and Salzberg 2012) and counting variants in the alignment using Geneious Prime SNP annotation, set to a cutoff of ≥ 75% of reads showing the nucleotide change. Pipeline scripts and dependencies were managed within a custom conda environment and run locally.