1. Introduction
In recent years, solventogenic Clostridia have garnered significant attention in the post-genomic era, primarily owing to the comprehensive sequencing and annotation of their genomes [
1,
2]. This wealth of genomic information has provided valuable insights into the metabolism of these industrially important strains, thereby catalyzing new approaches to genetic analysis, functional genomics, and metabolic engineering for the development of industrial strains geared toward biofuel and bulk chemical production.
To facilitate these endeavors, various reverse genetic tools have been devised for solventogenic Clostridia. These tools include markerless gene inactivation systems, employing methods such as homologous recombination with non-replicative [
3,
4,
5] and replicative plasmids [
6,
7,
8,
9], as well as the insertion of group II introns [
10,
11,
12]. For all homologous recombination-based methods involving two crossing-over, the use of a counterselection technique is imperative. This may include employing CRISPR-Cas9 [
13,
14,
15,
16] or a counter-selectable marker, which have been constructed using the codon-optimized
mazF toxin gene from
Escherichia coli (under the control of a lactose-inducible promoter) [
7], the
pyrE [
8] gene (encoding an orotate phosphoribosyl transferase, leading to 5-fluoroorotate (5FOA) toxicity), the
upp gene (encoding an uracil phosphoribosyl transferase, leading to 5-fluorouracil (5FU) toxicity) [
5,
9], or the
codA gene [
4,
17] (encoding a cytosine deaminase that converts 5-fluorocytosine to 5FU, which is further transformed into a toxic compound by the product of the
upp gene).
It is worth noting that while strategies relying on 5FC/5FU selection are highly effective, they should be employed cautiously. 5FU is a well-known anticancer drug recognized for its mutagenic properties in human cancers [
18]. These mutagenic attributes have been demonstrated in various organisms, including
Caenorhabditis elegans,
E. coli, and
Mycobacterium tuberculosis [
19,
20,
21]. Moreover, the discharge of 5FU in the environment can impact aquatic organisms because of its mutagenic power [
22].
While working on the metabolic engineering of
C. acetotutylicum, we identified that the
upp/5FU selection method could no longer be used to further engineer the
CAB1060 strain [
23] as it became resistant to high concentrations of 5FU. In this study, we will demonstrate that this phenotype was due to a mutation in the
pyrR gene encoding a potential repressor of the pyrimidine operon. To fix this problem, we created a corrective replicative plasmid expressing the
pyrR gene, building upon the work of Bermejo et al. [
24]. Furthermore, in order to avoid the occurrence of the problem observed with the
CAB1060 strain, a preventive suicide plasmid was also developed, inspired by Foulquier et al. [
5], featuring the introduction of a synthetic codon-optimized
pyrR gene, which is referred to as
pyrR* with low nucleotide sequence homology to
pyrR. Additionally, a synthetic defined medium was optimized allowing a reduction in the concentration of 5FU required for the counterselection by a factor of 200 (from 1 mM to 5 µM), minimizing then the mutagenic effect of 5FU.
2. Materials and Methods
2.1. Bacterial Strains, Plasmids and Oligonucleotides
The bacterial strains and plasmids used in this study are referenced in
Table 1. The oligonucleotides used for PCR amplification that were synthesized and provided by Eurogentec (Seraing, Belgium) are listed in
Table 2.
2.2. Growth Conditions
E. coli strains were grown in Luria–Bertani (LB) medium.
C. acetobutylicum strains were maintained as spores in synthetic medium (SM) at −20 °C as previously described or, for non-sporulating strains, directly in degassed and sterile serum bottles at −80 °C [
25,
26]. Spores were activated by heat shock at 80 °C for 15 min. Strains were grown under anaerobic conditions at 37 °C in Clostridial Growth Medium (CGM) supplemented each time with 30 gL
−1 of glucose [
27] or in CGM supplemented with 20 gL
−1 MES hydrate (Sigma-Aldrich
®, St. Louis, MI, USA), synthetic medium (SM) or in SM supplemented with 20 gL
−1 MES hydrate or in Reinforced Clostridial Medium (RCM) (Millipore, Burlington, MA, USA). The pH of CGM was adjusted at 6.0 or 5.2 with hydrochloric acid. The pH of RCM was adjusted at 5.8 with hydrochloric acid. The SM used for
C. acetobutylicum contained the following per liter of deionized water: glucose, 30 g; KH
2PO
4, 0.50 g; K
2HPO
4, 0.50 g; MgSO
4.7H
2O, 0.22 g; acetic acid, 2.3 mL; FeSO
4.7H
2O 10 mg; para amino benzoic acid, 8 mg; biotin, 0.08 mg. For
C. acetobutylicum liquid cultures, SM was complemented with NiCl
2, 3 mg; ZnCl
2, 60 mg; and nitriloacetic acid, 0.2 g. The pH of the medium was adjusted to 6.0 with ammonia. For solid media preparation, 1.5% agar was added to liquid media. The media were supplemented as needed with the appropriate antibiotic at the following concentrations: for
C. acetobutylicum, erythromycin (Ery) at 40 µg/mL, clarithromycin (Clari) at 40 µg/mL, and thiamphenicol (Tm) at 10 µg/mL; for
E. coli, carbenicillin (Cb) at 100 µg/mL and chloramphenicol (Cm) at 30 µg/mL. Stocks of 5-fluorouracil (5FU) and uracil (Sigma-Aldrich
®, St. Louis, MI, USA) were prepared at 0.1 M in dimethyl sulfoxide (DMSO) (Sigma-Aldrich
®, St. Louis, MI, USA) and stored at −20 °C.
2.3. DNA Manipulation
Genomic DNA was extracted from C. acetobutylicum strains using GenEluteTM Bacterial Genomic DNA Kits (Sigma-Aldrich®, St. Louis, MI, USA). Plasmid DNA was extracted from E. coli using NucleoSpin® Plasmid or NucleoBond® Xtra Midi kits (Macherey-Nagel, Düren, Germany). Phusion DNA Polymerase (New England Biolabs (NEB, Ipswich, MA, USA)) was used to generate PCR products according to the supplier’s standard protocols. OneTaq® 2X Master Mix with Standard Buffer (NEB, Ipswich, MA, USA) was used to screen colonies by PCR according to the supplier’s standard protocols. Restriction enzymes, antartic phosphatase, and T4 DNA ligase (NEB, Ipswich, MA, USA) were used according to the manufacturer’s instructions. DNA fragments were purified from agarose gel using a ZymocleanTM Large Fragment DNA Recovery Kit (Zymo Research, Irvine, CA, USA). DNA PCR fragments were purified using NucleoSpin® Gel and PCR Clean-up (Macherey-Nagel, Düren, Germany). Plasmid DNA and DNA PCR fragments were sequenced using the Sanger method (Eurofins Genomics, Ebersberg, Germany). PCR were conducted on a T100TM Thermal Cycler (Bio-rad, Hercules, CA, USA). DNA recombinations were performed using the GeneArtTM Seamless Cloning and Assembly Kit (Invitrogen, Thermofisher Scientific, Emeryville, CA, USA). Synthetic genes were synthesized by Geneart (Thermofisher Scientific, CA, USA).
2.4. Design of pyrR*
The nucleotide sequence of the
pyrR gene (
CA_C2113) was codon-optimized to create a synthetic
pyrR gene, named
pyrR*, with low nucleotide sequence identity to the wild-type
pyrR gene but in which substitutions would be as silent as possible and would not affect protein folding and function [
28,
29]. To do that, based on the codon usage table of
C. acetobutylicum downloaded at
https://gcua.schoedl.de/ (accessed on 7 December 2020), synonymous codon substitutions were introduced at all positions where it was possible, i.e., at all positions where the substitution did not replace a frequent codon by a rare one or conversely. The synthetic gene was synthesized by Geneart (Thermofisher Scientific, CA, USA). The sequence of the wild-type
pyrR gene and the synthetic
pyrR* are described in
Table 3.
2.5. Construction of pCat-upp-pyrRmut
This plasmid was constructed, based on the pCat-
upp described by Foulquier et al. [
5], by introducing the
pyrRmut gene containing the mutation
g.344C>T encoding the PyrR A115V protein found in our mutant strain. The mutant
pyrR gene was PCR amplified with Phusion DNA polymerase using genomic DNA from
CAB1060 as the template and PSC 75 and PSC 76 primers containing
BamHI restriction sites. The PCR fragment and the pCat-
upp were digested by
BamHI for one hour at 37 °C. The plasmid was dephosphorylated with antarctic phosphatase for 30 min at 37 °C. The PCR fragment and the plasmid were purified with NucleoSpin
® Gel (Macherey-Nagel, Düren, Germany) and a PCR Clean-up kit (Macherey-Nagel, Düren, Germany). The PCR fragment was cloned by ligation into the plasmid with T4 DNA ligase overnight at 16 °C to obtain pCat-
upp-pyrRmut. The ligation was transformed in One shot
TM TOP 10 chemically competent
E. coli following the manufacturers’ instructions (Thermofisher Scientific, CA, USA).
2.6. Construction of pCat-upp-pyrR*
This plasmid was constructed from the pCat-
upp described by Foulquier et al. [
5] by introducing the synthetic
pyrR* gene under the control of the thiolase promoter. The entire pCat-
upp plasmid was amplified with Phusion DNA Polymerase using PSC 51 and PSC 52 for linearization and was purified with NucleoSpin
® Gel (Macherey-Nagel, Düren, Germany) and a PCR Clean-up kit (Macherey-Nagel, Düren, Germany). The
pyrR* gene was amplified with Phusion DNA Polymerase using the synthetic
pyrR* gene as the template. A first PCR was performed with PSC 61 and PSC 62 primers to amplify the
pyrR* gene with its native RBS and downstream homology arms. The first PCR fragment was purified with NucleoSpin
® Gel (Macherey-Nagel, Düren, Germany) and a PCR Clean-up kit (Macherey-Nagel, Düren, Germany). A second PCR was performed on the first PCR product with PSC 72 and PSC 62 to introduce the upstream homology arm. The final PCR fragment of
pyrR* was purified from agarose gel using the Zymoclean
TM Large Fragment DNA Recovery kit (Zymo Research, Irvine, CA, USA) and cloned into the linearized pCat-
upp plasmid by recombination using the GeneArt
TM Seamless Cloning and Assembly kit (Thermofisher Scientific, CA, USA). The plasmid was transformed in One shot
TM TOP 10 chemically competent
E. coli following the manufacturers’ instructions (Thermofisher Scientific, CA, USA).
2.7. Construction of pCat-upp-pyrR*-Δldh
This plasmid was constructed based on the
pCat-upp-pyrR* (this study) and the pCat-
upp-
Δldh described by Nguyen et al. [
23]. The
pCat-upp-pyrR* plasmid was linearized by digestion with the
BamHI restriction enzyme for one hour at 37 °C and dephosphorylated with antartic phosphatase for 30 min at 37 °C. The pCat-
upp-
Δldh plasmid was digested by
BamHI, and the fragment containing the
ldh homology arms was purified from agarose gel using a Zymoclean
TM Large Fragment DNA Recovery kit (Zymo Research, Irvine, CA, USA). The two fragments were ligated using T4 DNA Ligase overnight at 16 °C. The plasmid was transformed in One shot
TM TOP 10 chemically competent
E. coli following manufacturers’ instructions (Thermofisher Scientific, CA, USA).
2.8. Construction of pCat-upp-pyrR*-Δldh::sadh-hydG
This plasmid was constructed based on the pCat-upp-pyrR*-Δldh plasmid (this study) by introducing an operon composed of sadh and hydG genes (GenBank: AF157307.2), with their own RBS, under the control of the ldh promoter. The pCat-upp-Δldh was digested by StuI for one hour at 37 °C, dephosphorylated with antartic phosphatase for 30 min at 37 °C and purified with NucleoSpin® Gel (Macherey-Nagel, Düren, Germany) and a PCR Clean-up kit (Macherey-Nagel, Düren, Germany). sadh and hydG were amplified with Phusion DNA Polymerase from a synthesized sadh_hydG gene as the template. A first PCR was performed with PSC 104 and PSC 105 to amplify sadh_hydG genes and introduce the ldh promoter region upstream of sadh and the ldh terminator downstream of hydG. After purification with NucleoSpin® Gel (Macherey-Nagel, Düren, Germany) and a PCR Clean-up kit (Macherey-Nagel, Düren, Germany), the first PCR product was amplified with Phusion DNA Polymerase using PSC 106 and PSC 107 to introduce upstream and downstream homology arms to recombine with the pCat-upp-pyrR*-Δldh plasmid. The final PCR fragment was purified from agarose gel using a ZymocleanTM Large Fragment DNA Recovery kit (Zymo Research, Irvine, CA, USA) and cloned into the pCat-upp-pyrR*-Δldh by recombination using the GeneArtTM Seamless Cloning and Assembly Kit (Thermofisher Scientific, CA, USA). The plasmid was transformed in One shotTM TOP 10 chemically competent E. coli following the manufacturers’ instructions (Thermofisher Scientific, CA, USA).
2.9. Construction of pSOS95-pyrR
This plasmid was constructed based on the pSOS95 plasmid described by Bermejo et al. [
24] by introducing the native
pyrR gene under the control of the thiolase promoter. The pSOS95 plasmid was digested by
BamHI and
SfoI and purified from agarose gel using the Zymoclean
TM Large Fragment DNA Recovery Kit (Zymo Research, Irvine, CA, USA). The
pyrR gene was amplified with Phusion DNA Polymerase using the genomic DNA from the
C. acetobutylicum strain
Δcac1502 as a template. The PSC 58 and PSC 46 used for this amplification introduced the
BamHI restriction site upstream of the
pyrR gene RBS and the
SfoI restriction site downstream of the
pyrR gene. The PCR fragment was digested by
BamHI and
SfoI and purified with NucleoSpin
® Gel (Macherey-Nagel, Düren, Germany) and PCR Clean-up kit (Macherey-Nagel, Düren, Germany). The PCR fragment was cloned in the pSOS95 plasmid by ligation using T4 DNA Ligase overnight at 16 °C. The plasmid was transformed in One shot
TM TOP 10 chemically competent
E. coli following the manufacturers’ instructions (Thermofisher Scientific, CA, USA).
2.10. Transformation Protocol
The transformation of C. acetobutylicum was performed by electroporation according to the following protocol. From a culture of C. acetobutylicum in CGM at A620 between 1 and 2, a new serum bottle with 50 mL of CGM was inoculated at A620 of 0.1. When the culture reached A620 between 0.6 and 0.8, the culture was placed on ice for 30 min and transferred under an anaerobic chamber (Jacomex, Dagneux, France), where all the following manipulations were performed. The cells were harvested by centrifugation at 7000× g for 15 min (Centrifuge 5430, Eppendorf, Framingham, MA, USA) and washed in 10 mL of ice-cold electroporation buffer (EB) composed of 270 mM sucrose and 10 mM MES hydrate at pH 6.0. Then, the pellet was resuspended in 500 µL of EB, and cells were transferred into a sterile electrotransformation vessel (0.40 cm electrode gap × 1.00 cm) with 5–100 µg plasmid DNA. A 1.8 kV discharge was applied to the suspension from a 25 µF capacitor and a 400 Ω resistance in parallel using the Gene Pulser (Bio-Rad, Hercules, CA, USA). Cells were transferred directly to 10 mL of warm CGM and incubated for 6 h at 37 °C before plating on RCM supplemented with the required antibiotics.
2.11. Microbiological Enumeration on Solid Media
C. acetobutylicum was cultivated in CGM until reaching an A620 of 0.55 (Libra S11, Biochrom, Cambridge, UK). Subsequently, the culture was transferred to an anaerobic chamber, and 100 µL of various dilutions (10−1 to 10−6) of the culture was plated onto CGM MES or SM MES agar supplemented with the necessary antibiotics, ranging from 0 to 200 µM for 5FU and from 0 to 50 µM for uracil. Following incubation at 37 °C for a period of 1 to 4 days, the resulting colonies were counted.
2.12. 5FU Selection Protocol
C. acetobutylicum was cultivated in CGM until reaching an A
620 of 0.55 (Libra S11, Biochrom, Cambridge, UK). The spreading protocol was the same as described in
Section 2.11. After isolation, 50 colonies were picked and plated on a fresh plate with the same concentration of 5FU. The plates were incubated at 37 °C from 1 to 2 days. Once the colonies had grown, they were picked and patched onto plates with and without thiamphenicol to determine the percentage of double crossing-over events. Colonies showing a double crossover phenotype were screened by PCR to verify that genome editing occurred.
2.13. Locus Verification in C. acetobutylicum after Metabolic Engineering
After the genome edition of
C. acetobutylicum, the different loci were checked by PCR amplification. In order to check for the insertion of point mutations, the genome was amplified by PCR with the primers of the
Table 4, and the PCR fragment obtained was sent for sequencing.
2.14. Analytical Procedures
Viability percentages were calculated using the following formula:
with
x corresponding to the number of colonies growing after replicating on a fresh plate without antibiotics or 5FU.
The double-crossing over percentage was calculated through the Tm sensitivity using the following formula:
with
x corresponding to the number of colonies growing after replicating on a plate without antibiotics and
y corresponding to the number of colonies growing after replicate on a plate with antibiotics.
Culture growth was monitored by measuring optical density over time using a spectrophotometer at A620 (Libra S11, Biochrom, UK). For sample analysis, glucose, acetate, butyrate, acetone, isopropanol, ethanol, and butanol concentrations were measured using High-Performance Liquid Chromatography (HPLC) analysis (Agilent 1200 series, Les Ulis, France). Before injection, samples were centrifuged at 15,000× g for 5 min (Centrifuge 5424, Eppendorf, Framingham, MA, USA), and the supernatants were filtered through a 0.2 µm filter (Minisart® RC 4, Sartorius, Epsom, UK). The separations were performed on a Bio-rad Aminex HPX-87H column (300 mm × 7.8 mm) (Bio-Rad, Hercules, CA, USA), and detection was achieved using either a refractive index measurement or a UV absorbance measurement (210 nm). The operating conditions were as follows: temperature, 14 °C; mobile phase, H2SO4 (0.5 mM); and flow rate, 0.5 mL/min. Excel 2019 was used for statistical analyses.
4. Discussion
In the present study, we have shown that the overexposure of
C. acetobutylicum to 5FU can make it resistant to this drug. Spontaneous mutations are induced in the bacterial chromosome, notably in the
pyrR gene. As described in other publications, the
pyrR gene encodes for PyrR, which is the repressor of the pyrimidine operon [
20,
21,
30]. The presence of a mutation in this protein can lead to the cessation of its function, resulting in an overproduction of UMP. This overproduction of UMP can protect against the harmful effects of 5FUMP, which is a molecule that is toxic to bacteria. We observed the appearance of a mutation in the PyrR protein of
C. acetobutylicum that completely avoids the selection of the double crossing-over step when using the pCat
upp/5FU system. The principle of the
upp/5FU system was based on the use of a strain in which the
upp gene has been deleted and the use of 5FU as a counterselection agent. The
upp gene encodes for an uracil phosphoribosyl transferase (UPRTase) that could convert uracil to UMP and 5FU to 5FUMP. 5FUMP is a molecule that prevents cells from producing intermediates needed for DNA synthesis, thereby causing cell death (
Figure 5a) [
21]. According to Peters et al., 5FUMP and UMP compete for binding to thymidylate synthase. 5FUMP targets thymidylate synthase and inhibits its activity, which is used to convert UMP to TMP. The production of TMP is decreased, and subsequently, DNA production is decreased [
35].
When the A115V PyrR mutation was discovered in
C. acetobutylicum, we compared it with other mutations in homologous proteins already described in the literature. A conserved protein sequence required in PRPP binding can be found in many species. Ghode and Singh described the G125V and R126C mutations
in M. tuberculosis as being in this conserved zone [
20,
21]. According to Ghode, a mutation in this region could block the production of 5FUMP. As the A115V PyrR mutation is situated close to this site, it could be one of the reasons why our strain is resistant to 5FU (
Figure 6). In addition,
pyrR encodes the regulatory protein of the pyrimidine operon. According to Ghode and Fields [
20,
36], after mutations in PyrR of
M. tuberculosis or
M. smegmatis, or when PyrR is completely deleted in
B. subtilis, the protein no longer performs its regulatory function, and the pyrimidine operon is overexpressed [
30]. This results in the overproduction of UMP, which protects the bacteria from the toxic effects of 5FUMP. This mechanism described in M
ycobacteria seems to work in the same way in
C. acetobutylicum as a single mutation occurrence in PyrR (A115V) caused a 5FU resistance of the strain. In a
C. acetobutylicum pyrRmut strain, the exact mechanism (affected PRPP binding or impaired regulatory function with a consequent pyrimidine operon overexpression or both) leading to 5FU resistance is not known yet. The two hypotheses have been summarized in
Figure 5b.
To overcome this problem, we had to revise the protocol previously described by our team. First, we realized that the composition of the media plays an essential role in the resistance of the strain to 5FU. It is preferable to use a synthetic media that is not supplemented with uracil. The yeast extract present in CGM brings uracil into the media and thus protects against the toxic effect of 5FUMP. This hypothesis was tested by adding low concentrations of uracil to the synthetic media. Bacterial growth was no longer affected by the presence of 5FU in the media. After optimizing the media for 5FU selection, we constructed two plasmids to restore the sensitivity of the strain to 5FU. The first plasmid is a replicative plasmid overexpressing a native version of
pyrR. It is used to overcome the problems of 5FU selection when a strain mutated in the
pyrR gene has already integrated a pCat-
upp. The second plasmid is a pCat-
upp containing a codon-optimized version of the
pyrR gene called
pyrR*. The 5FU selection problem for genome editing is directly bypassed by this method. With both of these strategies, the concentration of 5FU could be reduced from 1 mM to 5 µM, thus minimizing the risk of spontaneous mutation. Both the use of
pyrR* and the use of the SM media will be beneficial for all the counterselection methods involving
upp and 5FU as well as those utilizing
codA and 5FC [
4].
Once the new protocol was established, we demonstrated that it was possible to both delete and insert genes of interest in the
C. acetobutylicum ∆cac1502∆upp∆cac3535 pyrRmut strain in a single step using the pCat-
upp-pyrR*/5FU system. An isopropanol production pathway from
C. beijerinckii was inserted at the
ldh of
C. acetobutylicum ∆cac1502∆upp∆cac3535 pyrRmut∆ldh::sadh-hydG strain utilizing this technique. We decided to insert the
sadh and
hydG genes from
C. beijerinckii NRRL 593 following a publication by Dusséaux et al. [
32]. SADH is an NADPH-dependent primary-secondary alcohol dehydrogenase that catalyzes acetone reduction, and HydG is a putative electron transfer protein [
34,
39].
hydG was introduced into the
C. acetobutylicum ∆cac1502∆upp∆cac3535 pyrRmut strain genome at the same time as
sadh, since these two genes are located in the same operon in
C. beijerinkcii NRRL 593. It was assumed that the HydG activity would positively affect the SADH activity, allowing the strain to obtain better isopropanol production [
34]. The final production of our
C. acetobutylicum ∆cac1502∆upp∆cac3535 pyrRmut ∆ldh::sadh-hydG strain is lower than the one obtained by Dusséaux (up to 4.7 g·L
−1 of isopropanol produced in a culture of 30 h) with a lower molar ratio of isopropanol/acetone [
32]. This result can be explained by the fact that in this strain, both genes were overexpressed in a multi-copy replicative plasmid and under the control of the
ptb promoter, which is a stronger promoter than the
ldh promoter [
40].