Global greenhouse gas emissions have been rising at an unprecedented rate, with the associated climate instability now being recognized throughout the world by governments as a serious threat to ecosystems, human health, and national economies. To curtail this trend and limit the global temperature rise to 2 °C above pre-industrial levels will require a radical reduction of the use of primary fossil resources for the coming decades [1] and increase the use of low-carbon fuels and chemicals [2] derived from sustainable and waste sources. Gas fermentation offers an opportunity to recycle carbon and harness energy from synthesis gas (syngas) generated from any biomass (such as municipal solid waste, organic industrial waste, or agricultural waste) or industrial off-gases (e.g., from industrial sources like steel mills or processing plants) for the production of transportation fuels and chemical intermediates [3, 4]. The commercialization and at-scale deployment of gas fermentation technology is being actively pursued by several companies with the first commercial units currently under construction [4, 5]. At the heart of the technology are acetogenic bacteria that act as biocatalysts by fixing carbon from gases such as carbon monoxide and/or carbon dioxide in the presence of hydrogen [6]. The principle challenges in commercial exploitation of the vast potential of gas-fermenting acetogens are the relatively basic understanding of acetogens and, in particular, the limited availability of genetic tools and high-throughput genetic engineering platforms [2, 7].

Clostridium autoethanogenum is a model acetogen that is being pursued for fuel (ethanol) and chemical (2,3-butanediol) production at commercial scale [4, 5]. However, relatively few genetic tools have been reported for C. autoethanogenum [4, 5]. In C. autoethanogenum, key insights on the energetics and carbon flux balance have been gained by gene knockout studies using ClosTron, a group II intron-based retrohoming gene disruption tool [8, 9]. However, this intron insertion-based gene inactivation tool has its own limitations as it leaves a huge scar consisting of a fragment of the group II intron along with the antibiotic selection marker. Gene deletions by homologous recombination in C. autoethanogenum [10] are achievable but at a very low frequency leading to labor-intensive screening processes and lower efficiencies or by leaving a scar or marker in the genome. A more reliable and stable genetic modification tool that enables scarless genome modifications is preferable.

CRISPR/Cas9 system is an exciting breakthrough in DNA editing technology. Clustered Regularly Interspaced Short Palindromic Repeat, CRISPR, is a bacterial acquired immune system to combat phage infections that has been intelligently adapted for biotechnology purposes [11–13]. CRISPR/Cas9 from Streptococcus pyogenes relies on a 20-nucleotide information in its crRNA–tracrRNA chimeric RNA (single-guide RNA, sgRNA) to guide Cas9 endonuclease to the target DNA where it introduces double-stranded breaks (DSB). In most eukaryotes, the DSB are repaired by non-homologous end joining. However, in prokaryotes the repair is by homologous recombination and is mediated by a DNA repair template. CRISPR/Cas9-mediated genome modification has been shown in a diverse array of microbial systems including in a few Clostridia, recently [14–18].

Here we describe the applicability of Streptococcus pyogenes type II CRISPR/Cas9 system for genetic modification of C. autoethanogenum which already has a type-1B CRISPR [19]. We further show that the adaptation of the heterologous CRISPR/Cas9 system for use in C. autoethanogenum required constructing and screening a small library for stronger tetracycline-inducible promoter(s). For the exemplification of the CRISPR/Cas9 system, two genes, namely a NADPH-dependent primary:secondary alcohol dehydrogenase (adh; CAETHG_0553) and a 2,3-butanediol dehydrogenase (2,3-bdh; CAETHG_0385), were chosen. The rationale for targeting these genes is centered on their involvement in ethanol and 2,3-butanediol metabolism [20, 21] and the fact that both genes had been previously inactivated (using ClosTron methodology) without having an impact on growth [10, 22], thus making them predictable targets for genetic tool validation.

The cas9 and sgRNA derived from S. pyogenes CRISPR/Cas9 system were introduced into C. autoethanogenum on two different plasmids, sequentially. Except for in controls, the sgRNA plasmids contained the homology arms (HAs) that served as DNA editing template. While C. autoethanogenum maintained sgRNA plasmids, several attempts to introduce a plasmid carrying cas9 under the control of a native constitutive phosphotransacetylase–acetate kinase promoter [20] were not successful, likely due to toxicity caused by uncontrolled Cas9 protein expression. This was addressed by regulating the expression of cas9 by a tetracycline-inducible promoter, tet3no [23]. Two sgRNAs with unique binding sites to the target gene (Fig. 1a, b) were individually expressed using a native Wood–Ljungdahl cluster promoter [24].

The adh gene was targeted first (Fig. 1a). Following confirmation of the presence of cas9 by PCR (Fig. 1a, control-1), sgRNA plasmids with (psgRNA-adh-T1_HA and psgRNA-adh-T2_HA) and without HA (psgRNA-adh-T1) were then introduced. The cas9 expression in colonies transformed with cas9 and sgRNA plasmids was induced with 32 ng/ml anhydrotetracycline. The induced colonies were then screened for 891 bp deletion within adh by PCR using primers flanking the HA (Fig. 1a). In the absence of HA or DNA editing template, no deletion was detected (Fig. 1a, Control-2) and Sanger sequencing of these PCR products did not show insertions/deletions (INDELs). Four colonies were obtained on screening plates with psgRNA-adh-T1_HA and psgRNA-adh-T2_HA. Interestingly, amplicons of ~2.9 kbp instead of a ~2.5 kbp size were detected in two colonies with psgRNA-adh-T1_HA (Fig. 1a, T1, 1 and 2), implying a partial deletion in adh rather than the expected ~0.9-kbp deletion. From the remaining two colonies (Fig. 1a, T1, 3 and 4), no fragment was amplified implying a probable integration of the plasmid at the target locus. Sanger sequencing of ~2.5 kbp PCR amplicons confirmed the partial deletion in adh (Fig. 1a, ∆adh clone T1.1 and 1.2; Additional files 1, 2). The mutants with anticipated length of deletion were not generated. Two of the four colonies from psgRNA-adh-T2_HA amplified fragments corresponding to the wild type (Fig. 1a, T2, 3 and 4), and the remaining two, similar to psgRNA-adh-T1_HA, likely have the plasmid integrated at the targeted locus (Fig. 1a, T2, 1 and 2). This could be likely due to poor recognition of the target site by guide RNA adh-T2.

The partial deletion of adh only in the presence of all three components: cas9, sgRNA, and DNA editing template indicated the activity of the heterologous CRISPR/Cas9 system in C. autoethanogenum and scope for further improvement. To further optimize the CRISPR/Cas9 system for improved performance in C. autoethanogenum, two modifications were identified: (1) enhanced control of cas9 expression and (2) positioning one of the HAs close to Cas9 cleavage site.

In order to have an enhanced control over cas9 expression, a set of variants of tetracycline-inducible promoters was constructed based on a method described previously [25] whereby the ten least conserved bases in the −35 and −10 boxes of the rRNA consensus sequences of C. autoethanogenum were randomized. For inducible expression, the tet operator (tet3no) from the tetracycline-inducible promoter system [26] was inserted in between the randomized −35 and −10 boxes (Fig. 2a). Twelve variants from the inducible promoter library (IPL) were screened with chloramphenicol acetyltransferase gene catP as the reporter. Five of these variants could not be grown in liquid media, possibly due to the strength of these promoters. Of the seven remaining promoters, (IPL1 2, 3, 5, 8, 11, and 12), only IPL12 promoter showed significant activity. Even though the non-induced IPL12 promoter showed leaky activity that was higher than the original tet3no promoter, upon induction the activity of IPL12 promoter was approximately ninefold higher than that of tet3no (Fig. 2b). Therefore, the IPL12 promoter was chosen to drive cas9 expression (Fig. 1b).

The modifications discussed above were tested on a second gene, 2,3-bdh. The expression of cas9 was driven by IPL12 promoter (IPL12-cas9) and at least one of the HAs was designed to be within 80 bp from Cas9 cleavage site (Fig. 1b) unlike in the previous case where it was at a distance of 250 bp (Fig. 1a) as a measure to avoid potential partial deletion.

Following cas9 induction, colonies harboring pIPL12-cas9 and one of the two sgRNA plasmids, psgRNA-2,3bdh-T1_HA and psgRNA-2,3bdh-T2_HA, were screened by PCR for deletion in 2,3-bdh using primers flanking the HAs. While a ~3.5-kbp fragment was amplified from wild type and colonies carrying either pIPL12-cas9 or psgRNA-2,3bdh-T1_HA or psgRNA-2,3bdh-T2_HA only (Fig. 1b, control, lanes w, 1, 2, and 3), a deletion product of ~2.5 kbp was observed in colonies (Fig. 1b, T1 lanes 2-6 and T2 lanes 1-3, 5, and 8) carrying cas9, sgRNA, and DNA editing template. The desired modification was observed with both sgRNAs targeting 2,3-bdh. This ~1-kbp deletion within 2,3-bdh was further confirmed by sequencing the PCR products from sgRNA T1, clone 2 (Fig. 1b, ∆2,3-bdh clone T1.2; Additional file 3), and sgRNA T2 clone 1 (Fig. 1b, ∆2,3-bdh clone T2.1; Additional files 1, 4). The plasmids from positive clones were cured while maintaining the gene deletion by passaging the strains twice on non-selective agar plates. With five out of six and four out of eight colonies harboring cas9, DNA editing template, and either sgRNA T1 or T2 having desired deletion in 2,3-bdh, the efficiency of the re-designed system to introduce desired deletion was >50 % compared to the previous attempt.

In conclusion, the data reported herein demonstrate the workability of the CRISPR/Cas9 tool in C. autoethanogenum. In order to efficiently work in C. autoethanogenum, the CRISPR/Cas9 system requires the controlled expression of cas9 and the constitutive expression of sgRNA in the presence of DNA editing template. The new IPL12 tetracycline-inducible promoter significantly increased the efficiency of Cas9-mediated genome editing. Even with screening a relatively small library, promoters with a wide range of expression strengths ranging between the original tet3no to the strong IPL12 promoter were obtained. The developed promoter library has the added potential of expanding the prospective applications of this approach in the metabolic engineering of acetogens. With our modifications, we achieved >50 % efficiency in gene deletion, which is comparable to the efficiencies reported in other Clostridia. The efficiency of CRISPR/Cas9 system adapted for Clostridium beijerinckii is unclear [15, 27]. However, Li et al. reported editing efficiencies of up to 100 % in C. beijerinckii and Clostridium acetobutylicum with the use of nickase variant of cas9 [18]. In Clostridium cellulolyticum, gene deletions were only possible with an engineered nickase variant of cas9 as the wild-type cas9 could not be introduced in C. cellulolyticum [14]. The use of an inducible promoter to control the expression of wild-type cas9 would have probably been sufficient to overcome the problem of expressing wild-type cas9 in C. cellulolyticum. Likewise, combining an engineered nickase with the inducible promoter may add additional benefit. In Clostridium ljungdahlii, gene deletions with 50–100 % efficiency have been reported with a single-plasmid system comprising both cas9 and guide RNA expression cassettes with Pthl and ParaE constitutive promoters driving the expression of cas9 and guide RNA, respectively [17]. The expression of cas9 from a constitutive promoter could have been likely possible due to the absence of a CRISPR system in C. ljungdahlii [19]. The expression of guide RNA from C. acetobutylicum’s ParaE promoter, similar to that in C. ljungdahlii, and the use of nickase-only Cas9 variant may further improve the efficiency of the CRISPR/Cas9-based genome editing in C. autoethanogenum. The above CRISPR/Cas9-based genome editing strategy can be further adapted for gene insertions and to create multiple gene knockouts [11, 12].