Construction of a restriction-less, marker-less mutant useful for functional genomic and metabolic engineering of the biofuel producer Clostridium acetobutylicum

Background Clostridium acetobutylicum is a gram-positive, spore-forming, anaerobic bacterium capable of converting various sugars and polysaccharides into solvents (acetone, butanol, and ethanol). The sequencing of its genome has prompted new approaches to genetic analysis, functional genomics, and metabolic engineering to develop industrial strains for the production of biofuels and bulk chemicals. Results The method used in this paper to knock-out or knock-in genes in C. acetobutylicum combines the use of an antibiotic-resistance gene for the deletion or replacement of the target gene, the subsequent elimination of the antibiotic-resistance gene with the flippase recombinase system from Saccharomyces cerevisiae, and a C. acetobutylicum strain that lacks upp, which encodes uracil phosphoribosyl-transferase, for subsequent use as a counter-selectable marker. A replicative vector containing (1) a pIMP13 origin of replication from Bacillus subtilis that is functional in Clostridia, (2) a replacement cassette consisting of an antibiotic resistance gene (MLSR) flanked by two FRT sequences, and (3) two sequences homologous to selected regions around target DNA sequence was first constructed. This vector was successfully used to consecutively delete the Cac824I restriction endonuclease encoding gene (CA_C1502) and the upp gene (CA_C2879) in the C. acetobutylicum ATCC824 chromosome. The resulting C. acetobutylicum Δcac1502Δupp strain is marker-less, readily transformable without any previous plasmid methylation and can serve as the host for the “marker-less” genetic exchange system. The third gene, CA_C3535, shown in this study to encode for a type II restriction enzyme (Cac824II) that recognizes the CTGAAG sequence, was deleted using an upp/5-FU counter-selection strategy to improve the efficiency of the method. The restriction-less marker-less strain and the method was successfully used to delete two genes (ctfAB) on the pSOL1 megaplasmid and one gene (ldhA) on the chromosome to get strains no longer producing acetone or l-lactate. Conclusions The restriction-less, marker-less strain described in this study, as well as the maker-less genetic exchange coupled with positive selection, will be useful for functional genomic studies and for the development of industrial strains for the production of biofuels and bulk chemicals.


Background
In recent years, Clostridium acetobutylicum ATCC824 has been of interest in the postgenomic era due to the complete sequencing and annotation of its genome [1], supplying a wealth of information regarding its protein machinery. This global knowledge has prompted new approaches to genetic analysis, functional genomics, and metabolic engineering in order to develop industrial strains for the production of biofuels and bulk chemicals.
To this end, several reverse genetic tools have been developed for C. acetobutylicum ATCC 824, including a gene inactivation system based on non-replicative [2,3] and replicative plasmids [4][5][6][7] and the group II intron gene inactivation system [8,9]. Among these methods, only the method developed by Al-Hinai et al. [5] allows for in frame deletions and/or the introduction of genes at their normal chromosomal context without an antibiotic marker remaining. This system is made of two parts. The first part is a replicative vector containing (1) a pIMP13 origin of replication from Bacillus subtilis functional in Clostridia, (2) a replacement cassette consisting of an antibiotic resistance gene (Th R ) flanked by two FRT sequences, (3) two sequences homologous to the selected regions around the target DNA sequence, and (4) a codon-optimized mazF toxin gene from Escherichia coli under the control of a lactose-inducible promoter from Clostridium perfringens to allow for the positive selection of double-crossover allelic exchange mutants. The second part is a plasmid system with inducible segregational instability, enabling efficient deployment of the FLP-FRT system to generate marker-less deletion or integration mutants.
In 2006, our group patented a marker-less, in-frame deletion method [10] similar to the two-part method published by Al-Hinai et al. [5] in 2012. The first part of our method is based on the same replicative plasmid and the same replacement cassette, but it uses the uracil PRTase upp/5-fluorouracil (5-FU) system as a counterselection strategy. The second part is based on a plasmid carrying (1) the FLP-FRT system to generate markerless deletion and (2) the uracil PRTase upp/5-FU system to select for the plasmid loss after marker excision. This method was successfully used by the Metabolic Explorer Company to develop and patent an industrial recombinant strain of C. acetobutylicum for n-butanol production. As this method has not been described in detail and to make it available to and usable by the scientific community, we report how this method was developed and its use to create a restriction-less, marker-less strain of C. acetobutylicum. We show that this strain lacking upp (CA_C2879, encoding the uracil-phosphoribosyltransferase), CA_C1502 encoding Cac824I and CA_ C3535 encoding Cac824II (the second type II restriction enzyme) can be transformed by non-methylated DNA at very high efficiency and can be used for rapid gene knock-in and knock-out using the upp/5-FU counterselectable system for both functional genomic and metabolic engineering of C. acetobutylicum. This strain and the method were further used to delete three genes ctfAB and ldhA to create strains no longer producing acetone and lactate, respectively.

MGCΔcac1502 strain, a C. acetobutylicum strain that is transformable without previous in vivo plasmid methylation
Cac824I, the type II restriction endonuclease encoded by CA_C1502, is a major barrier to the electrotransformation of C. acetobutylicum with E. coli-C. acetobutylicum shuttle vectors [11]. The Cac824I restriction endonuclease recognition sequence 5′-GCNGC3′, where N can be any nucleotide, occurs infrequently in C. acetobutylicum DNA because of the high A + T DNA content (72 % A + T), but the sequence occurs frequently in E. coli plasmids. No methyltransferase that can be used in vitro to protect DNA from restriction by Cac824I is commercially available. Prior to the transformation of C. acetobutylicum, shuttle plasmids have to be methylated in vivo by transformation into E.coli ER2275 (pAN1) expressing the Bacillus subtilis phage φ3TI methyltransferase, which protects the shuttle plasmids from digestion by the clostridial endonuclease Cac824I [11]. This step is time consuming and may be a drawback if the genes to be transferred to C. acetobutylicum are toxic when expressed in E. coli. Therefore, a C. acetobutylicum strain deficient for this particular restriction system would be valuable for efficient electrotransformation without previous treatment of the plasmid to be transformed.
To delete the Cac824I encoding gene, the first step is the construction of a shuttle vector carrying the replacement cassette. The CA_C1502 replacement cassette was cloned into the BamHI site of the pCons2-1 and pCIP2-1 to generate the pREPcac15 and pCIPcac15 plasmids, respectively. The difference between these two plasmids is the origin of replication. The pREPcac15 contains a pIMP13 origin of replication from B. subtilis (rolling circle mechanism) functional in Clostridia, whereas pCIP-cac15 contains the origin of replication of the pSOL1 megaplasmid (θ replication mechanism). The pREPcac15 and pCIPcac15 plasmids were methylated in vivo in E.coli ER2275 (pAN1) and were used to transform C. acetobutylicum ATCC824 by electroporation. After selection on plates for clones resistant to erythromycin at 40 µg/ml, one colony of each transformant was cultured for 24 h in liquid SM with erythromycin and was then subcultured four times in liquid 2YTG medium without antibiotic ( Fig. 1a). To select integrants that lost the pREPcac15 or pCIPcac15 plasmids, 10 3 erythromycin resistant clones were replica plated on both RCA with erythromycin and RCA with thiamphenicol at 50 µg/ml. Whereas several colonies resistant to erythromycin and sensitive to thiamphenicol were obtained with pREPcac15 transformants, no such colonies were obtained with the pCIPcac15 transformants, which indicates that the θ replication mechanism of pCIPcac15 is less favorable for promoting double-crossover in C. acetobutylicum than a rolling circle mechanism. The genotype of clones with the desired phenotype was checked by PCR (polymerase chain reaction) analysis (Fig. 2a). The Δcac1502::mls R strain, which had lost the pREPcac15, was isolated. This strain was transformed with the pCLF1 plasmid expressing the FLP1 gene of S. cerevisiae encoding for the FLP recombinase. The expression of FLP1 was under the control of the promoter and RBS (ribosome binding site) from the thiolase gene from C. acetobutylicum. After transformation and selection for resistance to thiamphenicol at 50 µg/ml, one colony was cultured in liquid SM with thiamphenicol. One hundred thiamphenicol resistant clones were replica plated on both RCA with erythromycin and RCA with thiamphenicol. The genotype of the clones with erythromycin sensitivity and thiamphenicol resistance was checked by PCR analysis with primers CAC 0 and CAC 5 (Fig. 2a). Two successive 24-h liquid cultures of the Δcac1502 strain were conducted in the absence of antibiotics to remove pCLF1. The Δcac1502 strain that lost pCLF1 was isolated according to its sensitivity to both erythromycin and thiamphenicol. This strain was called MGCΔcac1502.
The efficiency of transformation of this strain with methylated and unmethylated pCons2.1 plasmid was evaluated and compared to the wild type strain. Both strains can be transformed with methylated pCons2.1 with similar efficiency, but only MGCΔcac1502 can be transformed efficiently with unmethylated DNA ( Table 1).
The following deletions described in this manuscript were conducted in this strain without previous in vivo plasmid methylation.

Construction of the MGCΔcac1502Δupp strain: the first marker-free C. acetobutylicum strain with two deleted genes
To develop a positive screening of integrants, we used the "upp/5-FU as counter selection marker" system. The C. acetobutylicum upp gene (CA_C2879) encodes uracil phosphoribosyl-transferase (UPRTase), which catalyzes the conversion of uracil into UMP, thus allowing the cell to use exogenous uracil [12]. The pyrimidine analog 5-fluoro uracil (5-FU) can be converted by UPRTase into 5-fluoro-UMP, which is metabolized into 5-fluoro-dUMP, an inhibitor of thymidylate synthetase, toxic for the cell. The use of the upp expression cassette as a counter-selection marker is linked to the construction of a C. acetobutylicum strain deleted for the upp gene, thus resistant to 5-FU.
To delete upp, the upp replacement cassette was cloned into the BamHI site of pCons2-1 to generate the plasmid pREPupp. The plasmid pREPupp was used to transform the MGCΔcac1502 strain by electroporation without previous in vivo methylation. After selection on plates for clones resistant to erythromycin at 40 µg/ml, one colony was cultured for 24 h in liquid SM with erythromycin and was then subcultured in liquid 2YTG medium without antibiotic (Fig. 1a). To select integrants having lost the pREPupp plasmid, 10 3 erythromycin resistant clones were replica plated on both RCA with erythromycin and RCA with thiamphenicol. The genotype of the clones resistant to erythromycin and sensitive to thiamphenicol was determined by PCR analysis (Fig. 2b). The MGCΔcac1502Δupp::mls R strain that lost pREPupp was isolated. When the resistance to 5-FU was analyzed, we showed that this strain was resistant to up to 1 mM 5-FU compared to 50 µM for the MGCΔcac1502 strain. This strain was then transformed with the pCLF1 plasmid, and selection of MGC Δcac1502Δupp strain with sensitivity to both erythromycin and thiamphenicol was performed, as previously described for the MGCΔcac1502 strain (Fig. 2b).

Deletion of the CA_C3535 gene in the MGCΔcac1502Δupp strain using the upp/5-FU system as a counter-selectable marker for the loss of plasmid
The CA_C3535 gene encodes Cac824II, a potentially bi-functional enzyme carrying both a type II restriction endonuclease and methylase activities. To delete CA_C3535, the CA_C3535 replacement cassette was cloned into the BamHI site of the pCons::upp to generate the plasmid pREPcac3535::upp. The plasmid pREPcac3535::upp was used to transform the C. acetobutylicum MGCΔcac1502Δupp strain by electroporation without previous in vivo methylation.
After plate selection for clones resistant to erythromycin at 40 µg/ml, 100 transformants were replica plated on RCA with erythromycin, RCA with thiamphenicol and RCA with 5-FU at 400 µM (Fig. 1b). All transformants were resistant to erythromycin and thiamphenicol and were sensitive to 5-FU compared to the parental strain, which was resistant to 5-FU. This result demonstrates that the expression of the upp gene carried by pREPcac::upp confers sensitivity to 5-FU.
To select for Δcac3535::Em R integrants that lost the pREPcac3535::upp plasmid, erythromycin-and 5-FU-resistant clones were selected on RCA plates containing erythromycin and 5-FU from 100 µl of a liquid culture of the MGCΔcac1502Δupp (pREPcac3535::upp) strain. Approximately 500 colonies were obtained, and 100 of them were replica plated on both RCA with erythromycin and RCA with thiamphenicol. Most of the clones (95 %) were resistant to erythromycin and sensitive to thiamphenicol. Four clones were checked by PCR analysis (Fig. 2c) All four clones had the correct phenotype, and one of the clones was selected as the MGC Δcac1502ΔuppΔcac3535::mls R strain. This strain was then transformed with pCLF::upp, a derivative of the pCLF1 plasmid that also carries the upp gene, in order for the positive selection of plasmid loss after the excision of the mls R marker. After the first selection of clones resistant to thiamphenicol and sensitive to erythromycin, a second selection of clones resistant to 5-FU and sensitive to thiamphenicol was performed to obtain the MGC Δcac1502ΔuppΔcac3535 strain that was control by PCR ( Fig. 2c) for the presence of all the marker-less deletions. Finally, when compared to C. acetobutylicum ATCC824 wild type, the growth of the restriction-less marker-less strain in MS medium at pH 4.5 ( Fig. 3) was shown to be unaffected by the different deletions.
The unmethylated pCons2.1 plasmid was used to evaluate the transformation efficiency of the MGCΔcac1502 and the MGCΔcac1502Δupp Δcac3535 strains. The transformation efficiency of MGCΔcac1502ΔuppΔcac3535 for unmethylated pCons2.1 was ~eightfold higher than that of MGCΔcac1502 (Table 2). Fig. 1 General diagram representing gene replacement via allelic exchange at the Y locus, and excision of the MLS R marker by the FLP recombinase to create an unmarked Y deletion mutant. The boxed regions of X and Z genes represent approximatively the regions of homology incorporated into the replicative plasmid used for the double-crossover event (~ 1 kbp each). a Initial strategy used for the construction of the MGCΔcac1502 and MGCΔcac1502Δupp strains, b counter-selection strategy with the 5-FU/upp system used for the construction of the MGCΔcac1502ΔuppΔcac3535 strain

Determination of the recognition sequence of Cac824II encoded by CA_C3535
CA_C3535 encoded a 993 amino acid protein with a calculated molecular mass of 116,842 Da. The amino acid sequence analysis revealed high similarities with two restriction endonucleases: AcuI from Acinetobacter calcoaceticus SRW4 [13] and Eco57I from E. coli RFL57 [14] with 44 and 46 % identity, respectively. Both enzymes belong to the IIg family of restriction enzymes and possess both a restriction and methylase activity.
To heterologously express the AcuI-encoding gene in E. coli [13], it was necessary to first express the AcuIM methylase-encoding gene because the methylase activity of AcuI was not sufficient to protect DNA against its restriction activity. We applied the same strategy for the expression of CA_C3535-encoding Cac824II: we cloned into the pSOS2K2 gene and expressed in E. coli the CA_ C3534 gene that encodes a putative methylase and that is located immediately downstream of CA_C3535 gene in the C. acetobutylicum chromosome. The pSC-CAC3534 plasmid expressing CA_C3534 has three AcuI recognition sites, but when we tried to digest it with AcuI, it was completely protected from the activity of this enzyme. To express, purify and determine the recognition sequences of Cac824II, we cloned CA_C3535in the pPAL vector using the E. coli BL21-AI cells containing the pSC-CAC3534 plasmid as host. The Cac824II endonuclease was purified, and its activity towards unmethylated pCons2.1 in the presence of SAM was determined. AcuI recognizes the 5′-CTGAAG-3′ sequence and cuts the pCons2.1 plasmid two times, resulting in two fragments of 2411-and 882-bp. Figure 4 shows that Cac824II gives the same restriction pattern as AcuI. To confirm that the AcuI and Cac824II recognition sequences were identical, pCons2.1 was digested by 50 μg of Cac824II in the presence of 1 U of AcuI. Figure 4 shows that the restriction pattern was unchanged, which definitively confirms that AcuI and Cac824II are isoschizomers.

Deletion of the ctfAB genes in the MGCΔcac1502ΔuppΔcac3535 to create a strain no longer producing acetone
The ctfAB genes (CA_P0163-CA_P0164) located on the pSOL1 megaplasmid encodes an acetoacetyl-CoA:acyl CoA-transferase involved in the first specific step of acetone formation [15]. To delete ctfAB, the ctfAB replacement cassette was cloned into the BamHI site of the pCons::upp to generate the plasmid pREPctfAB::upp. The plasmid pREPctfAB::upp was used to transform the C. acetobutylicum MGCΔcac1502ΔuppΔcac3535 strain by electroporation without previous in vivo methylation and cell containing the plasmid were selected on RCA plate with erythromycin at 40 µg/ml. To select for ΔctfAB::Em R   Most of the clones (90 %) were resistant to erythromycin and sensitive to thiamphenicol. Four clones were checked by PCR analysis (with primers CTF-0 and CTF-5 located outside of the ctfAB replacement cassette and primers CTF-D and CTF-R located inside of ctfAB). All four clones had the correct phenotype, and one of the clones was selected as the MGC Δcac1502ΔuppΔcac3535ΔctfA B::mls R strain. The fermentation profile of this strain was compared to the MGC Δcac1502ΔuppΔcac3535 control strain during batch fermentation at pH 4.5 (Fig. 5). The production of acetone was totally abolished but the production of acetic acid was increased more than sixfold while butyric acid was only slightly increased, proving that the acetoacetyl-CoA:acyl CoA-transferase is mainly involved in the consumption of acetic acid.
Deletion of the ldhA gene in the MGCΔcac1502ΔuppΔcac3535 to create a strain no longer producing lactate The ldhA genes (CA_C0267) encodes a lactate dehydrogenase involved in the last step of l-lactate formation [15]. To delete ldhA, the ldhA replacement cassette was cloned into the BamHI site of the pCons::upp to generate the plasmid pREPldhA::upp. The plasmid pREPldhA::upp was used to transform the C. acetobutylicum MGCΔcac1502ΔuppΔcac3535 strain by electroporation without previous in vivo methylation and cell containing the plasmid were selected on RCA plate with erythromycin at 40 µg/ml. To select for ΔldhA::Em R integrants that lost the pREPldhA::upp plasmid, erythromycin-and 5-FU-resistant clones were selected on RCA plates containing erythromycin and 5-FU from 100 μl of a liquid culture of the MGCcac1502ΔuppΔcac3535 (pREPldhA::upp) strain. Approximately 500 colonies were obtained, and 50 of them were replica plated on both RCA with erythromycin and RCA with thiamphenicol. Most of the clones (80 %) were resistant to erythromycin and sensitive to thiamphenicol. Four clones were checked by PCR analysis (with primers LDH-0 and LDH-5 located outside of the ldhA replacement cassette and primers LDH-D and LDH-R located inside of ldhA). All four clones had the correct phenotype, and one of the clones was selected as the MGC Δcac1502ΔuppΔcac3535ΔldhA::mls R strain.
The fermentation profile of this strain was compared to the MGC Δcac1502ΔuppΔcac3535 control strain during batch fermentation at pH 4.5 (Fig. 5). The production of l-lactate was totally abolished proving that ldhA encodes the main l-lactate dehydrogenase of C. acetobutylicum.

Discussion
We developed a simple and efficient method to create mutations in the Clostridium acetobutylicum chromosome. This method is based on the use of (1) a replicative plasmid, (2) a deletion cassette containing both DNA sequences with homology to the flanking region of the target gene (to delete it) and an antibiotic resistance gene surrounded by FRT sequences (as an excisable marker), and (3) the upp gene, which encodes the uracil-phosphoribosyl-transferase, as a counter-selectable marker. A plasmid that replicates via a rolling circle mechanism was more efficient in terms of double cross over frequency than a plasmid that replicates through a theta mechanism. This result is in agreement with previous findings in Bacillus subtilis showing that plasmid replication through a rolling circle mechanism favors recombination between homologous sequences [16,17].
The deletion cassette can be rapidly constructed through a three-step procedure using pre-constructed building blocks. After a fusion PCR and TOPO cloning of the product, a predesigned antibiotic resistance gene surrounded by two FRT sites in direct repeats is inserted. The upp gene is located on the plasmid outside of the deletion cassette. This allows the positive selection of clones that have lost the plasmid and integrated the deletion cassette by a double recombination event. We demonstrate here that this event occurs at a frequency of 10 −5 , which means that without the selection procedure, it would be much more difficult to isolate the correct deletion mutant by replica plating alone.

in SM medium
Once the deletion cassette is integrated into the chromosome, the expression of the flp recombinase allows (1) the excision of the antibiotic marker for a clean in-frame deletion of the targeted gene (without polar effect) and (2) consecutive gene deletions. Such a strategy was previously applied to marker-less gene deletion in E. coli [18] and Mycobacterium smegmatis [19]. The plasmid expressing the FLP recombinase-encoding gene was further improved by coexpressing the upp gene to use it as a positive selection for the plasmid loss after excision of the MLS R marker. A similar tool was developed by Al-Hinai et al. [5] using a plasmid (1) that expresses the FLP recombinase-encoding gene and (2) that has an inducible segregational instability to promote the plasmid loss.
In this study Cac824II (encoded by CA_C3535), the second type II restriction enzyme of C. acetobutylicum predicted by REBASE [20], was biochemically characterized and it was demonstrated that it is an isoschizomer of AcuI [13] recognizing the 5′-CTGAAG-3′ sequence. It was also shown that Cac824II methylase (encoded by CA_C3534) protects DNA against restriction by Cac824II and AcuI by probably methylating one of the adenine in the 5′-CTGAAG-3′ sequence. Two Cac824II restriction sites are present (in the ampR gene and in the colE1 origin of replication) in most the shuttle vector use to transform C. acetobutylicum and it was then justified to construct a marker-less strain deleted from CA_C3535. The transformation efficiency of MGCΔcac1502ΔuppΔcac3535 for unmethylated pCons2.1 was much higher (~eightfold higher) than that of MGCΔcac1502 and it will be an interesting strain to develop new genetic tools based on suicide vectors [20].
The restriction-less marker-less strain and the method was successfully used to delete two genes (ctfAB) on the pSOL1 megaplasmid and one gene (ldhA) on the chromosome to get strains no longer producing acetone or l-lactate. This work demonstrate that (1) ctfAB encode an acetoacetyl-CoA:acetate CoA-transferase that coupled acetone formation to acetate consumption and (2) ldhA encodes the main lactate dehydrogenase in C. acetobutylicum although a second gene ldhB (CA_C3552) is also present [15]. A strain with clostron inactivated ctfAB genes was previously constructed [21]. From the physiological analysis of this mutant and with the help of a mathematical model [22], it was demonstrated that butyrate was mainly reconsumed by the phosphotransbutyrylase-buyrate-kinase pathway and not by the acetoacetyl-CoA:acetate CoA-transferase in agreement with the data presented in our study.

Conclusions
The restriction-less, marker-less strain and the genome modification method presented here become simple and convenient tools that are useful for research groups involved in functional genomic studies of C. acetobutylicum and for further metabolic engineering of this strain to produce bulk chemicals and biofuel. As a demonstration of the efficiency of the method, we constructed two strains unable to produce l-lactate or acetone. Furthermore, this method was successfully used by the Metabolic Explorer Company to develop and patent an industrial recombinant strain of C. acetobutylicum for n-butanol production [23] at high yield.

Bacterial strain, plasmids and oligonucleotides
The bacterial strain and plasmids used in this study are listed in Table 3. The specific oligonucleotides used for PCR amplification were synthesized by Eurogentec (Table 4).

Culture and growth conditions
C. acetobutylicum was maintained as spores in synthetic medium (SM) as previously described [24,25] Spores were activated by heat treatment at 80 °C for 15 min. All C. acetobutylicum strains were grown in anaerobic conditions at 37 °C in SM, in Clostridium growth medium (CGM) [26] in 2YTG [27] or in reinforced clostridial medium (RCM) (Fluka). Solid media were obtained by adding 1.5 % agar to the liquid media. Media were supplemented, when required, with the appropriate antibiotic in the following concentrations: for C. acetobutylicum, erythromycin at 40 µg/ml and thiamphenicol at 50 µg/ml; for E. coli, erythromycin at 200 µg/ml and chloramphenicol at 30 µg/ml. Transformations of C. acetobutylicum were conducted by electroporation, as previously described [11]. 5-FU was purchased from Sigma, and stock solutions were prepared in DMSO (dimethyl sulfoxide).

DNA manipulation techniques
Total genomic DNA from C. acetobutylicum was isolated as previously described [27]. Plasmid DNA was extracted from E. coli with the QIAprep kit (Qiagen, France). Pfu DNA Polymerase (Roche) was used to generate PCR products for cloning, and Taq Polymerase (New England BioLabs) was used for screening colonies by PCR with standard PCR protocols employed for all reactions. DNA restriction and cloning were performed according to standard procedures [28]. Restriction enzymes and Quick T4 DNA ligase were obtained from New England BioLabs (Beverly, MA) and were used according to the manufacturer's instructions. DNA fragments were purified from agarose gels with the QIAquick gel purification kit (Qiagen, France).

Construction of pUC18-FRT-MLS2
Inverse PCR was performed using the pKD4 plasmid [18] as a template and oligonucleotides PKD4.1 and PKD4.2 as primers to amplify the plasmid region with the FRT sites but without the kanamycin resistance marker. This blunt end fragment was later ligated to the MLS r gene obtained after a HindIII digestion of the pETSPO plasmid [4] and Klenow treatment. The corresponding plasmid (pKD4-Ery1) was then used as a template to amplify by PCR the macrolide lincosamide streptogramin B resistance (MLS r ) gene, functional in Clostridia and flanked by two FRT sites and two StuI sites, using the oligonucleotides FRT-MLSR-F and FRT-MLSR-R as primers.
This fragment was directly cloned into the SmaI digested pUC18 to generate the pUC18-FRT-MLS2 plasmid.

Construction of pCons2.1
Inverse PCR was performed using the pETSPO plasmid [4] as a template and oligonucleotides PCONSAccI (mutating a BamHI site) and PCONSEcoRI as primers. The PCR product, containing a pIMP13 B. subtilis origin of replication functional in Clostridia (rolling circle mechanism of replication) and a catP gene conferring resistance to thiamphenicol was phosphorylated and ligated to yield the pCons0 plasmid. This plasmid was then digested with BamHI to remove the spoA cassette, and the DNA fragment was purified and ligated to generate the pCons2-1 plasmid.

Construction of pCIP2-1
The pIMP13 origin of replication from pCons2-1 was replaced by the origin of replication of the pSOL1 megaplasmid. The origin of replication of pSOL1 was amplified by PCR using C. acetobutylicum total DNA as a template and oligonucleotides ORI3-D and ORI3-R as primers. This PCR product was cloned into the pCR-BluntII-TOPO vector, and the resulting plasmid was digested by EcoRI to obtain the 2.2 kb EcoRI fragment containing the origin of replication of pSOL1. The pCons2-1 plasmid was digested by EcoRI, and the 2.4 kb fragment was ligated to the 2.2 kb EcoRI fragment to generate the plasmid pCIP2-1.

Construction of pREPcac15
Two DNA fragments surrounding cac1502 were amplified by PCR using C. acetobutylicum total DNA as the template and two pairs of oligonucleotides as primers.

Construction of pCIPcac15
The cac1502 replacement cassette above was cloned into the BamHI site of the pCIP2-1 to generate the plasmid pCIPcac15.

Construction of pCLF1
The FLP1 gene was amplified by PCR using the pCP20 plasmid [29] as a template and oligonucleotides FLP1-D and FLP1-R as primers. These primers introduced BamHI and SfoI restriction sites on the ends of the PCR product. After a BamHI-SfoI double digestion, the PCR product was cloned into the BamHI-SfoI sites of the pSOS95 expression vector to generate the pEX-FLP1 plasmid. The 1585-bp SalI fragment of pEX-FLP1 containing the FLP1 expression cassette was cloned into the SalI site of pCons2-1 to generate the pCLF1 plasmid.

Construction of pCons::upp
The upp gene with its own ribosome binding site (RBS) was amplified by PCR from C. acetobutylicum total DNA with the oligonucleotides REP-UPP-F and REP-UPP-R as primers. The 664-bp PCR product was digested by PvuII and was cloned into pCons2.1, digested by BcgI and treated with T4 DNA polymerase to generate the pCons::upp plasmid. In this way, the upp gene was located just downstream of the catP gene to construct an artificial operon with upp expressed under the control of the catP promoter.

Construction of pREPctfAB::upp
Two DNA fragments upstream and downstream of ctfAB (CA_P0162-CA_P0163) were amplified by PCR using the total DNA from C. acetobutylicum as a template and two pairs of oligonucleotides as primers. With the primer pairs CTF 1 and CTF 2 or CTF 3 and CTF 4, 1144-and 1138-bp DNA fragments were obtained, respectively. Both primers CTF 1 and CTF 4 introduce a BamHI site, whereas primers CTF 2 and CTF 3 have 5′ extended sequences that introduce a StuI site. DNA fragments CTF 1-CTF 2 and CTF 3-CTF 4 were joined in a PCR fusion with primers CTF 1 and CTF 4, and the resulting fragment was cloned into the pCR4-TOPO-Blunt vector to generate pTOPO::ctfAB. At the unique StuI site of pTOPO::ctfAB, the 1372-bp StuI fragment of pUC18-FRT-MLS2 carrying the antibiotic resistance MLS r gene with FRT sequences on both sides was introduced. The ldhA replacement cassette obtained after BamHI digestion of the resulting plasmid was cloned into the BamHI site of the pCons::upp to generate the plasmid pREPctfAB::upp.

Construction of pREPldhA::upp
Two DNA fragments upstream and downstream of ldhA (CA_C0267) were amplified by PCR using the total DNA from C. acetobutylicum as a template and two pairs of oligonucleotides as primers. With the primer pairs LDH 1 and LDH 2 or LDH 3 and LDH 4, 1135-and 1161-bp DNA fragments were obtained, respectively. Both primers LDH 1 and LDH 4 introduce a BamHI site, whereas primers LDH 2 and LDH 3 have 5′ extended sequences that introduce a StuI site. DNA fragments LDH 1-LDH 2 and LDH 3-LDH 4 were joined in a PCR fusion with primers LDH 1 and LDH 4, and the resulting fragment was cloned into the pCR4-TOPO-Blunt vector to generate pTOPO::ldhA. At the unique StuI site of pTOPO::ldhA, the 1372-bp StuI fragment of pUC18-FRT-MLS2 carrying the antibiotic resistance MLS r gene with FRT sequences on both sides was introduced. The ldhA replacement cassette obtained after BamHI digestion of the resulting plasmid was cloned into the BamHI site of the pCons::upp to generate the plasmid pREPldhA::upp.

Construction of pCLF::upp
The 1585-bp SalI fragment of pEX-FLP1 containing the FLP1 expression cassette was cloned into the SalI site of pCons::upp to generate the pCLF::upp plasmid.

Cac3535 expression and purification
For the general cloning methods of restriction endonuclease genes in E. coli, the first step to clone and express the recombinant CA_C3535 gene into E. coli was to preprotect the host genomic DNA against the restriction activity of the Cac3535 bi-functional enzyme. The CA_ C3534 methylase-encoding gene was thus amplified by PCR with Phusion DNA polymerase using C. acetobutylicum ATCC824 total genomic DNA as the template and Cac3534-d-AgeI and Cac3534-R-PvuI as primers. After digestion with AgeI and PvuI, the resulting 1748-bp fragment was then cloned into pAH105 [30] a pSC101 derivative, that has been previously digested with AgeI and PacI, resulting in the pSC-CAC3534 plasmid. In this construct, the CA_C3534 gene expression was placed under the control of the pGI 1.6 promoter [31]. The E. coli BL21-AI strain (Invitrogen) was then transformed by the pSC-CAC3534 plasmid to give the BL21-AI-3534 strain. This strain, with host genomic DNA protected against the restriction activity of the Cac3535 bi-functional enzyme, was finally used as the host strain for the CA_C3535 gene over-expression using the T7-based expression system (see below). The Cac3535 protein was expressed in E. coli BL21 AI-3534 and was purified using the Profinity eXact Protein Purification System, following the recommendations of the manufacturer (Biorad). The CA_C3535 gene was amplified by PCR with Phusion DNA polymerase using C. acetobutylicum ATCC824 total gDNA as the template and Cac3535d-SpeI and Cac3535-R-BamHI as primers. The resulting 3002 bp fragment was cloned into the Zero Blunt TOPO vector (Invitrogen) to generate the TOPO-CAC3535 plasmid. After verification by DNA sequencing, the 2988-bp SpeI-BamHI fragment from the latter plasmid was then introduced into the pPAL7 vector previously digested with the same enzymes to give the final pPAl-3535-I_2.4 plasmid.