The following primer pairs were annealed with Q5: FW_proto8/RV_proto8, FW_proto8-5′/RV_proto8, FW_proto20/RV_proto20, FW_proto20-5′/RV_proto20, FW_proto22/RV_proto22, and FW_proto22-5′/RV_proto22 to yield the three protospacers with and without their PAM candidate. They were then inserted by HiFi in a SacII/SalI-digested pMTL82151 backbone to create pMTL-MPD1 to 6. Colony PCRs and Sanger sequencing were performed with primers ColE1+tra-F2 and pBP1-R1.

The MCS from pMTL82151 was inserted after the leader sequence in pMTL-MPD21 by digestion with SacI/NheI and ligation with T4 DNA Ligase, creating pMTL-MPD15 (the full sequence of the vector pMTL-MPD15 is available at www.plasmidvectors.com (RRID: SCR_023475), where it may be sourced). The FAST gene and the catP gene were inserted in the MCS after the leader sequence for characterization of the leader sequence by RT-qPCR and to function as a negative control, respectively. The catP and FAST genes were amplified with primer pairs FW_catP_SacI/RV_catP_NheI and FW_FAST_SacI/RV_FAST_NheI, respectively. The pMTL-MPD15 backbone and both genes were digested by SacI and NheI. The catP and FAST genes were ligated in the backbone with T4 DNA ligase (NEB), yielding pMTL-MPD30 and pMTL-MPD29, respectively. All plasmids were confirmed by Sanger sequencing with primers ColE1+tra-F2 and pCD6-R1.

The editing template consisted of two homology arms of 471 bp (LHA) and 491 bp (RHA). The spacer was designed to be 36 bp and at a distance of 29 bp from the RHA. Two Gram-positive replicons were selected, pBP1 and pCD6.

The cas9 gene, together with its P _fdxE promoter, was removed from pMTL8215_pRECas1_MCS (Canadas et al., 2019) by XbaI/NotI digestion, blunting, and re-ligation, with T4 DNA polymerase and T4 DNA Ligase, respectively, yielding the vector pMTL8215_P1339_MCS. The editing cassette was assembled in this plasmid. HAs were amplified with primer pairs FW_LHA_pyrE/RV_LHA_pyrE+BM4 and FW_BM4+1+RHA_pyrE/RV_RHA_pyrE from A. woodii genomic DNA, assembled by SOEing PCR (Splicing by Overlap-Extension PCR), and ligated into the vector after AscI/AatII digestion, creating pMTL-MPD7.

A. woodii leader sequence was amplified with primers FW_leader/RV_leader; the synthetic array was created by annealing primers FW_pyrE_endo2 and RV_pyrE_endo2. Both the leader sequence and synthetic array were assembled by HiFi into pMTL-MPD7 digested by SacII/AatII, yielding pMTL-MPD8.

The editing cassette from pMTL-MPD8, amplified with primers FW_leader and RV_RHA_pyrE+MCS, pMTL82151, and pMTL84151, were digested by AatII/NheI, SacII/AatII, or SacII/NheI and ligated together to obtain constructs with HAs alone, expressed spacer alone, or the whole editing cassette, respectively, with either Gram-positive replicon, yielding plasmids pMTL-MPD9 to 14.

Colony PCRs were performed with primers ColE1+tra-F2/pBP1-R1 or ColE1+tra-F2/pCD6-R1 and plasmids confirmed by Sanger sequencing. The same primers were used to check for plasmid presence in A. woodii, and the pyrE locus was analyzed with primers AW_CRISPR_pyrE_armF and AW_pyrEcomp_RHAR.

The spacer was designed to be 35 bp and at a distance of 26 bp from the RHA. Homology arms of 0.5, 1.0, and 1.5 kb were designed. The resulting editing cassettes were inserted into a backbone with the pCD6 Gram-positive replicon.

The spacer sequence was created by annealing primers FW_pheA_spacer and RV_pheA_spacer by PCR and digestion by SacI and AatII; the leader sequence was created by amplification with FW_leader/RV_leader and SacII and SacI digestion; both fragments were assembled in pMTL84151 digested by SacII/AatII. Homology arms of 0.5 kb, 1.0 kb, and 1.5 kb were designed. Left homology arms were amplified with primers FW_L0.5/L1.0/L1.5 and RV_LHA_pheA; right homology arms were amplified with primers FW_RHA_pheA and RV_R0.5/R1.0/R1.5. The LHA and RHA were assembled by SOEing PCR with primer pairs FW_L0.5/RV_R0.5, FW_L1.0/RV_R1.0, and FW_L1.5/RV_R1.5. Assembled HAs were ligated in both pMTL84151 and pMTL84151 containing the assembled leader and spacer after AatII/NheI digestion, yielding three plasmids containing the HAs alone (pMTL-MPD16 to 18) and three editing plasmids containing the expressed spacer and the HAs (pMTL-MPD19 to 21).

Colony PCRs were performed with primers ColE1+tra-F2/pCD6-R1 and plasmids confirmed by Sanger sequencing. A. woodii colonies were analyzed by colony PCR both for plasmid presence with primers ColE1+tra-F2 and pCD6-R1 and for the pheA locus with primers FW_pheAHA1.5_out and RV_pheAHA1.5_out.

Homology arms of 1,455 bp (LHA) and 1,489 bp (RHA) and a 36 bp spacer starting 40 bp from the LHA were designed.

Primers FW_hsdR1_spacer and RV_hsdR1_spacer were annealed by PCR and inserted in the pMTL84151 vector with the assembled leader and spacer used for pheA KO by SacI and AatII digestion and ligation with T4 DNA Ligase, resulting in the vector pMTL8415_leader_hsdR1_spacer. The left and right homology arms were amplified with the primers FW_LHA_hsdR1/RV_LHA_hsdR1 and FW_RHA_hsdR1/RV_RHA_hsdR1, respectively, and assembled by SOEing PCR with the FW_LHA_hsdR1/RV_RHA_hsdR1 primer pair. The vector containing the spacer and the assembled HAs were digested by AatII and NheI and ligated together with T4 DNA Ligase, resulting in the final editing vector for hsdR1 KO, pMTL-MPD22.

Colony PCRs were performed with primers ColE1+tra-F2/pCD6-R1 and plasmids confirmed by Sanger sequencing. Cell densities and the amount of DNA were investigated as potential factors to increase editing efficiency as follows: 100 µL of cells and 1 µg of DNA (T1); 100 µL of high-density cells and 1 µg of DNA (T2); 200 µL of high-density cells and 4 µg of DNA (T3). A. woodii colonies were analyzed by PCR for plasmid presence with primers ColE1+tra-F2/pCD6-R1 and the hsdR1 locus with primers FW_hsdR1_screen_1.5/RV_hsdR1_screen_1.5 and FW_hsdR1_del_screen/RV_hsdR1_del_screen.

The spacer is 36 bp, and homology arms are 1,496 bp and 1,465 bp and were designed for insertion of the cargo in place of the PAM. The cargo consists of the FAST reporter gene flanked by the thiolase (thl) promoter (P _thl ) and pyrE-hydA (orotate phosphoribosyltransferase and hydrogenase I) terminator from C. acetobutylicum and was inserted between the homology arms.

Primers FW_AWO_pheA_CCG_KI_spacer and RV_AWO_pheA_CCG_KI_spacer were annealed by PCR and inserted in the pMTL84151 vector with the assembled leader and spacer used for pheA KO by SacI and AatII digestion and ligation with T4 DNA Ligase. The left and right homology arms were amplified with primers FW_AWO_LHA_pheA_cargo/RV_AWO_LHA_pheA_CCG_cargo_MCS and FW_AWO_RHA_pheA_CCG_cargo_MCS/RV_AWO_RHA_pheA_cargo, respectively, and assembled by SOEing-PCR with the FW_AWO_LHA_pheA_cargo/RV_AWO_RHA_pheA_cargo primer pair. The assembled HAs were inserted in the vector containing the spacer by AatII and NheI digestion and ligation by T4 DNA Ligase, resulting in a KI vector (pMTL-MPD23) containing an MCS with BamHI, NotI, NdeI, and XbaI restrictions sites between the two homology arms.

A P _thl _FAST_term cassette was amplified with the primer pair FW_BamHI_FAST_cassette/RV_FAST_NotI_AscI from plasmid pMTL8415X_P _thl _FAST with the FAST gene flanked by the thl promoter and pyrE-hydA terminator. The cassette was then inserted into the MCS of the pMTL-MPD23 vector by XbaI and NotI digestion and ligation with T4 DNA Ligase, yielding pMTL-MPD24.

Plasmids were confirmed by Sanger sequencing with ColE1+tra-F2/pCD6-R1/FW_pheA_cargo_seq/RV_pheA_cargo_seq.

WT A. woodii was transformed with the final editing vector, pMTL-MPD24, following the protocol named T3 in the previous section, with 200 µL of high-density cells and 4 µg of DNA. A. woodii colonies were analyzed by colony PCR for plasmid presence with primers ColE1+tra-F2/pCD6-R1 and the pheA locus with primers FW_pheA_KI_seq and RV_pheA_KI_seq.

Homology arms of 868 bp and 797 bp were used, and the synthetic array was designed to mimic C. autoethanogenum native array 4, with the native leader sequence controlling the expression of a 35 bp spacer flanked by direct repeats. The editing cassette was inserted into a pMTL83151 backbone.

pMTL83151 and CRISPR array leader amplified with FW_leader4_CLAU/RV_leader4_CLAU were digested by SacII/SacI and ligated together to yield pMTL-MPD25 (the full sequence of the vector pMTL-MPD25 is available at www.plasmidvectors.com (RRID: SCR_023475), where it may be sourced). Primers FW_CLAUpyrE_spacer1 and RV_CLAUpyrE_spacer1 were annealed by PCR and inserted into pMTL-MPD25 by SacI and BamHI digestion and ligation with T4 DNA Ligase, yielding pMTL-MPD27. This plasmid and the homology arms amplified with FW_CLAUpyrE_HA/RV_CLAUpyrE_HA from plasmid vFS67 (provided by Dr. Francois Seys) were digested by BamHI/NheI and ligated together to yield the final editing vector, pMTL-MPD28. An intermediate vector (pMTL-MPD26) containing the HAs alone was obtained by digesting pMTL83151 and the HAs by BamHI/NheI and ligation.

Colony PCRs were performed with primers ColE1+tra-F2 and pCB102-R1 and plasmids confirmed by Sanger sequencing. C. autoethanogenum colonies were analyzed by PCR for plasmid presence with ColE1+tra-F2 and pCB102-R1 and the pyrE locus analyzed with FW_pyrECLAU_screening and RV_pyrECLAU_screening.

A. woodii cells were collected by centrifugation at 7,000 g for 5 min at 4°C, and the pellet was resuspended in 2 mL of RNAprotect (QIAGEN). After a 5 min incubation at room temperature, the cells were collected again, and the pellet was stored at −80°C.

RNA extraction was carried out using a FastRNA Pro Blue kit (MP Bio). Cells were homogenized at 6,400 rpm for 45 s. All the centrifugation steps were carried out for 15 min at 4°C. The RNA was precipitated for 2 h at −80°C, and the RNA pellet was dried for 30 min at room temperature and resuspended in 50 µL of DEPC-H₂O.

To reduce genomic DNA contamination, samples were digested by TURBO DNase (Ambion). The resulting RNA was purified using the RNeasy Mini kit (QIAGEN). 350 μL of RLT buffer containing β-mercaptoethanol and 200 µL of absolute ethanol were added to the digested samples before purification according to the manufacturer’s instructions.

To check for genomic DNA contamination, a PCR (35 cycles, 50°C, and 2:00 min extension) with primers gDNA_FW and gDNA_RV (Supplementary Table S8), amplifying a 1,997 bp fragment in the A. woodii pyrE region, was carried out with Taq DNA polymerase (NEB). When genomic DNA was present, the samples were treated with TURBO DNase and purified with the RNeasy Mini kit a second time. The RNA samples were stored at −80°C.

The RNA concentration was assessed using a Nanodrop Lite spectrophotometer (Thermo Fisher). The quality was evaluated using an Agilent RNA 6000 Nano kit on an Agilent 2100 Bioanalyzer, and the RNA integrity number (RIN) was recorded for all the samples.

Complementary DNA was synthesized from 1 µg of RNA using the Omniscript RT Kit (QIAGEN), random hexamer (Thermo Fisher), and RNase inhibitor (NEB) following the manufacturer’s instructions.

Power SYBR Green (Thermo Fisher) was used for the qPCR, and primer pairs targeting the gyrA gyrase gene in the A. woodii genome, orfB of the plasmid backbones, and the FAST gene employed are listed in Supplementary Table S8. Standard curves allowed for efficiencies of primer pairs to be calculated. Gene expression was calculated for each sample using the Pfaffl method without a calibrator gene: E x p r e s s i o n = E t a r g e t − C p t a r g e t E r e f e r e n c e − C p r e f e r e n c e .

A Python script was developed to collect the host organism genome identifier, along with the information found on the CRISPRFinder database (Grissa et al., 2007b; Couvin et al., 2018), and compile all the information collected on each match for each spacer. A schematic overview is presented in Supplementary Figure S1, and a complete diagram is available in Supplementary Figure S2. Briefly, each spacer is run through BLAST and then filtered to reduce the number of false-positives and retain biologically relevant hits.

False-positives were eliminated by removing hits with more than 20% mismatch between spacers and hits, and only one mismatch was allowed in the seven nucleotides seed region adjacent to the PAM. Biologically relevant hits are then used to compile the PAM consensus sequence, based on filtering capabilities for mobile genetic element (MGE) sequences. Hits are kept based on three elements: the title of the hit containing the keywords “phage” or “plasmid;” their presence in a phage sequence predicted by PHASTER; and the annotation of the hit containing the keywords “transposase,” “phage,” “integrase,” or “terminase.”

The performance of the script was assessed using data available in the literature on Type I-B systems from Clostridium spp. that have been successfully used for genome editing. The results are presented in Supplementary Table S9. The script successfully identified the majority of the functional PAMs. For C. tyrobutyricum, it failed to identify the TCG alternative (Zhang et al., 2018), and for C. thermocellum, it identified only TTA/TTG/TCA instead of TTN and TNA, which have been experimentally validated as functional (Walker et al., 2020). For a majority of those organisms, the script predicted PAMs that had not been identified by manual analysis in the literature (marked with asterisks in Supplementary Table S9). In vivo tests are necessary to determine if those PAMs allow for interference and to conclude on the performance of the script.

Clostridium spp. of interest containing a Type I-B system, Clostridium butyricum, Clostridium limosum, and Clostridium novyi were analyzed using the script. The results are presented in Supplementary Table S10. The results obtained indicate that PAMs containing the nucleotide pairs TC or TT are common in Type I-B CRISPR/Cas systems. The number of candidate PAMs for each organism is quite high, indicating that those CRISPR/Cas systems are likely to recognize multiple PAMs (Fischer et al., 2012).

The CRISPRFinder database (Grissa et al., 2007b; Couvin et al., 2018) identified a Type I-B system in A. woodii, as presented in Figure 1A. This system contains a Csx8, a Cas8a homolog, confirmed by protein BLAST with homology to multiple Cas8a proteins from Clostridium spp. Cas8a is specific to Type I-A systems, but the gene organization of the A. woodii Cas locus is characteristic to Type I-B systems (Makarova and Koonin, 2015). Only one CRISPR array, named array 2, with an evidence level of 4 is identified in the database. It is located directly downstream of the Cas locus; it contains 47 spacers and its direct repeat is “ATT​TAC​ATT​CCA​ATA​TGG​ATC​TAC​TCA​AAT.” This CRISPR array was analyzed using the Python script, and the results are compiled in Supplementary Table S11.

The script identified multiple matches with invading elements but none with less than 17% mismatch. Although the script computed no clear consensus sequence for the 5′ PAM, three out of the eight matches contain two consecutive cytosines. Interestingly, a match with a 5′ CCA PAM in a phage gene was found, and this PAM was shown to be functional for the C. difficile Type I-B system (Boudry et al., 2015; Maikova et al., 2019). Thus, it is the main candidate for harnessing the endogenous CRISPR system in A. woodii.

Interference assays are used to assess PAM functionality by transformation of the WT strain with plasmids containing a protospacer and different candidate PAMs. Functional PAMs trigger interference and result in lower transformation efficiencies (Pyne et al., 2016; Walker et al., 2020). Three candidate 5′ PAMs identified by the Python script were tested experimentally in A. woodii, as shown in Figure 1Bi. Mismatches adjacent to the PAM reduce the likelihood of effective interference. Therefore, protospacers with no mismatch in the first position and with only one mismatch in the first 10 nucleotides were selected for testing. Each protospacer was assembled into a plasmid with the 3′ sequence from the direct repeat “ATTTA” and either the 5′ sequence of the direct repeat “CAAAT” or its corresponding PAM candidate. WT A. woodii was separately transformed with the six plasmids, pMTL-MPD1 to 6; transformation efficiencies were expressed as a percentage of the transformation efficiency obtained with the pMTL82151 positive control and normalized to the highest and lowest values.

Although high variations were observed between replicates (Figure 1Bii), the reduction in transformation efficiency in the presence of the GCC and CCA PAMs was shown to be statistically significant (p-value <0.033) compared to the presence of the protospacers alone. Sanger sequencing of colony PCR amplicons showed a high prevalence of mutations in the CCA PAM, with one of the cytosines missing. Those colonies with mutated PAMs are thus escape mutants of the endogenous CRISPR/Cas system. Taken together, these results strongly suggest that the double cytosine is essential to the interference phase of the studied Type I-B CRISPR/Cas system. The CCA 5′ PAM was selected for subsequent genome editing experiments in A. woodii.

For efficient targeting of invading DNA, the CRISPR array needs to be transcribed into pre-crRNA before maturation and Cascade assembly. In order to harness the A. woodii CRISPR system, a better characterization of the leader sequence responsible for the transcription of the array is required. For that purpose, the plasmid pMTL-MPD30 was built with the FAST reporter gene placed downstream of the leader of the aforementioned array 2. Control plasmids were built with either the catP reporter gene replacing the FAST reporter gene downstream of the leader sequence or with the thiolase promoter replacing the leader sequence for expression of the FAST-encoding reporter gene. These controls confirm primer specificity and efficiency, respectively. The transcript levels of the FAST gene were expressed (Figure 1C) with orfB as the reference gene to take into account the plasmid copy number. Transcript levels of the controls confirmed adequate annealing and specificity of the primers. The leader sequence allows for a low level of transcription in all cell growth stages with a clear reduction in the RNA level in the late stationary phase. This experiment confirms the functionality of the leader sequence for expression of CRISPR arrays for targeting DNA.

To establish the potential of the A. woodii Type I-B CRISPR/Cas system for genome editing, the pyrE gene was targeted for inactivation through the creation of a 300 bp deletion known to lead to uracil auxotrophy (Baker et al., 2022). Six plasmids were constructed (Figure 2A) with HAs alone, spacer alone, or the whole editing cassette and either the pCD6 or pBP1 Gram-positive replicon. The vectors with HAs and spacer alone function as controls to assess the impact of homologous recombination and self-targeting by Cas3, respectively. Only the whole editing cassette with HAs and expressed spacer allows for homologous recombination and selection against WT by self-targeting. WT A. woodii was separately transformed with the aforementioned plasmids in biological duplicates and plated after 5 h and 7.5 h of recovery.

The presence of the expressed spacer targeting the pyrE gene, with either replicon pCD6 or pBP1, with or without HAs, significantly reduced transformation efficiencies to between 0.06% and 2.17% of the transformation efficiency obtained with the corresponding plasmid containing HAs alone (data not shown). This shows the efficient self-targeting by the endogenous CRISPR/Cas system of the pyrE gene, leading to cell death. Colony numbers obtained varied significantly between the different replicates; colony PCR results are presented in Figure 2B. Successful gene knock-out was obtained with the full editing cassette, with either Gram-positive replicon pBP1 or pCD6. The latter was found to lead to 100% plasmid loss after one subculture without selection (Baker et al., 2022); it was, therefore, chosen for subsequent editing vectors.

Analysis of the escape mutants revealed single crossover events and frequent deletion of the spacer. The deletion of the spacer can be explained by the presence of the identical direct repeats flanking the spacer, facilitating recombination under Cas3 selective pressure.

The impact of homology arm length on editing efficiency was studied, tested with 0.5 kb, 1.0 kb, and 1.5 kb, for a 354 bp in-frame deletion in the pheA gene, previously shown to lead to phenylalanine auxotrophy (Baker et al., 2022).

WT A. woodii was separately transformed with six vectors (Figure 3A), three containing the HAs alone and three editing plasmids also containing the expressed spacer. The transformation efficiency obtained with each editing vector was normalized to the transformation efficiency obtained with the corresponding vector containing only the HAs. While low transformation efficiencies and high variations in colony numbers between each replicate were observed, the presence of the spacer reduced the transformation efficiency to between 0.053% and 0.32% of the efficiency obtained with the vectors carrying HAs alone (data not shown), showing efficient targeting of the A. woodii genome by the selected spacer. Characteristics of the colonies analyzed for the three replicates of this experiment are shown in Figures 3B, C, as well as the results of the colony PCRs of the pheA locus for the first replicate of this experiment.

At both 5 h and 7.5 h recovery, the 1.5 kb HAs resulted in more consistent knock-out generation with between 33.3% and 100% gene editing efficiency. After the plasmid loss protocol, 91.7% of colonies tested (72 in total) had successfully lost the plasmid. The phenotype of the strains was confirmed by plating on minimal media with and without phenylalanine supplementation (Figure 3D).

These results are the first report of successful gene editing using the A. woodii endogenous Type I-B CRISPR/Cas system and represent a promising basis for further engineering of this chassis. The knock-out of pheA followed by plasmid loss and confirmation of the phenotype highlights how straightforward and rapid the technique is, while illustrating the importance of the length of the editing template.

To test the robustness of the system for gene KO in A. woodii, the restriction subunit gene hsdR1 responsible for cleavage of the cas9-containing plasmids was targeted. Deletion of this gene has never previously been generated and at 3.2 kb represents a 10-fold increase in the size of the DNA deleted. WT A. woodii was transformed with the editing vector pMTL-MPD22 (Figure 4A), following protocols with increasing cell densities and DNA amounts referred to as T1, T2, and T3. Plating after 7.5 h of recovery yielded only WT colonies (data not shown). The generation of knock-out mutants was demonstrated (Figure 4B) with all three methods of transformation plated at 5 h of recovery, with 16.7%, 50%, and 75% efficiency for T1, T2, and T3, respectively. Sanger sequencing of the PCR-amplified plasmid sequences for T1 colonies 1 and 2; all T2 colonies; and T3 colonies 1, 2, and 4 revealed the absence of the spacer in the plasmid of colonies T1 colony 2 and T3 colony 4. As previously observed, recombination between the direct repeats flanking the spacer facilitates the excision of the spacer from the plasmid, rendering it non-functional. All ΔhsdR1 colony PCR amplicons were confirmed by Sanger sequencing. Plasmid loss was undertaken for the six confirmed ΔhsdR1 colonies, with 80.6% of those tested (36 in total) being shown to have successfully lost the plasmid.

To confirm the impact of the R-M site on transformation efficiency and the phenotype of the hsdR1 deletion, the R-M site 5′-TAAGN₅TCC-3′ was inserted in one, two, or three copies in the backbone pMTL82151. WT and ΔhsdR1 A. woodii strains were separately transformed with the four plasmids. For two of the replicates, the transformation efficiency of the ΔhsdR1 strain with the empty backbone was lower than the transformation efficiency of the WT, representing 20.59% and 52.94% of the WT transformation efficiency. More experiments are needed to determine if this is due to the hsdR1 deletion or to competent cell preparation variations.

As shown by the two-way ANOVA test presented in green in Figure 4C, the presence of the R-M site on plasmids has a significant impact on the transformation efficiency of WT A. woodii. Increasing numbers of recognition sites negatively impact the transformation efficiency. Deletion of the HsdR1 subunit of the R-M system significantly increased relative transformation efficiencies obtained with plasmids containing the R-M recognition site compared to WT, as shown by the two-way ANOVA test shown in blue. This confirms recognition of the 5′-TAAGN₅TCC-3′ motif by the HsdR1 subunit. Surprisingly, the presence of two or three R-M sites on the plasmids resulted in an increased relative transformation efficiency of A. woodii ΔhsdR1 compared to plasmids with none or one site.

As a final test to the robustness of A. woodii Type I-B CRISPR/Cas system as a genetic engineering tool, its capacity to knock-in a reporter gene was tested. As no 5′ CCA trinucleotide could be found in the target intergenic region downstream of pheA, an alternative 5′ CCG PAM was chosen for spacer design. Its selection was based on the aforementioned interference assay, suggesting that the double cytosine alone is essential for efficient targeting. WT A. woodii was transformed with the editing vector pMTL-MPD24 (Figure 5A), which was designed to insert a 614 bp DNA fragment expressing the FAST reporter gene downstream of pheA in place of the targeted PAM.

Based on the PCR analysis of six colonies from plating at 5 h and 7.5 h, successful knock-ins were obtained with both recovery times (Figure 5B), with an overall efficiency of approximately 57%. Following plasmid loss, plating of the clones onto minimal media with and without supplementary phenylalanine confirmed that the insertion had not disrupted pheA (Supplementary Figure S3). Expression of FAST was assayed using flow cytometry (Figure 5C), with WT A. woodii and WT A. woodii with P _thl _FAST in a pMTL84151 plasmid being used as negative and positive controls, respectively. As expected, WT A. woodii displayed the same level of fluorescence with and without the ligand. In the presence of the ligand, the plasmid-based population fluorescence was heterogeneous with two clear cell populations; one population displayed a level of fluorescence similar to the level of fluorescence without a ligand and a smaller one displayed a higher level of fluorescence. The KI strain, on the contrary, displayed a homogeneous population with a level of fluorescence higher than that without the ligand. It is important to note that the fluorescence of the KI population (KI + ligand in Figure 5Civ) is lower than the fluorescence of the plasmid-based fluorescent population (plasmid + ligand (2) in Figure 5Civ). The auto-fluorescence of the plasmid-based population is also slightly higher than that of both the WT and KI strains.

In C. autoethanogenum, the CRISPRFinder database identified a Type I-B system between positions 1,491,001 and 1,499,580 in reverse orientation. The database did not predict a Cas8 gene, which is essential for a functional CRISPR system. A protein BLAST of the gene situated between the Cas7 and the Cas6 revealed homologies to proteins labeled Cas8b in other Clostridium spp. For instance, an alignment with the Cas8b protein of the Type I-B system of Clostridium tetani has an E-value of 1e-148. Overall, the gene organization is the same as the one described for A. woodii.

Four CRISPR arrays were identified, including three with an evidence level of 4; array 2 is directly downstream of the Cas locus with 21 spacers, while array 3 (42 spacers) and array 4 (33 spacers) are immediately upstream of the Cas locus. All these arrays have a similar direct repeat, with only one nucleotide varying: “ATTTAAATACATCT(C/T)ATGTTGAGGTTCAAC.” This suggests that they are related and should yield coherent results when run in the Python script. Hits with less than 15% mismatch are presented in Supplementary Table S12 (hits with mismatch between 15% and 20% are presented in Supplementary Table S13 for further information). Array 4 had the most hits with 15 in phage regions or phage-related genes.

Overall, the Python script predicted a TCH/TTR putative 5′ PAM. Most notably, the spacer in position 27 of this array is a perfect match to a phage-related gene in Clostridium ljungdahlii with, in position 5′, the sequence “CTTCA.” The 5′ TCA PAM was selected for harnessing the endogenous system as a genetic engineering tool in C. autoethanogenum.

Design of the pyrE KO editing vector was based both on a 561 bp in-frame pyrE deletion already performed in C. autoethanogenum in our laboratory with CRISPR/Cas9 and on its native CRISPR array 4. WT C. autoethanogenum was conjugated with three plasmids containing HAs alone, expressed spacer alone, and both HAs and expressed spacer, as described for A. woodii in Figure 2A. Colony PCR analysis of transformants (Figure 6A) demonstrated that the editing vector containing both the HAs and the expressed spacer resulted in an editing efficiency of 100%. In contrast, vectors harboring only one part of the editing cassette did not result in any ΔpyrE mutants. Plasmid loss was achieved; the additional subculture without selection as described in Materials and Methods increased the plasmid loss rate from 33.3% to 58.3% (out of 12 colonies tested in both conditions). The phenotype of the strains obtained after plasmid loss was confirmed by plating on minimal media with and without uracil supplementation (Figure 6B). This is the first report of the successful use of the C. autoethanogenum Type I-B endogenous system for genome editing and demonstrates the robustness of the developed workflow—from the automated PAM identification pipeline to gene knock-out.