Deletion of glyceraldehyde‐3‐phosphate dehydrogenase (gapN) in Clostridium saccharoperbutylacetonicum N1‐4(HMT) using CLEAVE™ increases the ATP pool and accelerates solvent production

Summary The development and advent of mutagenesis tools for solventogenic clostridial species in recent years has allowed for the increased refinement of industrially relevant strains. In this study we have utilised CLEAVE™, a CRISPR/Cas genome editing system developed by Green Biologics Ltd., to engineer a strain of Clostridium saccharoperbutylacetonicum N1‐4(HMT) with potentially useful solvents titres and energy metabolism. As one of two enzymes responsible for the conversion of glyceraldehyde‐3‐phosphate (GAP) to 3‐phosphoglyceric acid in glycolysis, it was hypothesised that deletion of gapN would increase ATP and NADH production that could in turn improve solvent production. Herein, whole genome sequencing has been used to evaluate CLEAVE™ and the successful knockout of gapN, demonstrating a clean knockout with no other detectable variations from the wild type sequence. Elevated solvent levels were detected during the first 24 h of batch fermentation, indicating an earlier shift to solventogenesis. A 2.4‐fold increase in ATP concentration was observed, and quantitation of NAD(P)H derivatives revealed a more reducing cytoplasm for the gapN strain. These findings expand our understanding of clostridium carbon metabolism and report a new approach to optimising biofuel production.


Introduction
ABE fermentation using solventogenic clostridial species has been used for the industrial production of acetone, butanol and ethanol for over a century (for a review see Green, 2011). In the century that has passed since the discovery of the original Weizmann strain, species such as Clostridium beijerinckii (Jones and Keis, 1995) and Clostridium saccharoperbutylacetonicum (Hongo et al., 1969) have surpassed the original Weizmann strain as the preferred choice for ABE fermentation when grown on carbohydrate material. Unlike the well-characterised strain Clostridium acetobutylicum, C. saccharoperbutylacetonicum N1-4(HMT) is a relative newcomer to the world of industrial ABE production. First discovered in 1969 (Hongo et al., 1969) the full genome sequence of the strain was not published until 2014 (Poehlein et al., 2014). Over the years, fermentation experiments have been performed using a wide variety of feedstocks that have confirmed this strain as a butanol hyperproducer (Tashiro et al., 2007;Al-Shorgani et al., 2012;Noguchi et al., 2013). Early work demonstrated that unlike C. acetobutylicum strains, loss of solvent production in C. saccharoperbutylacetonicum N1-4(HMT) did not result from loss of the sol genes but more so from the dysregulation of the sol operon and upstream genes encoding other pathway enzymes (Kostan et al., 2010). Further to this, the mega plasmid found in C. saccharoperbutylacetonicum N1-4(HMT) has also been linked to ester production of butyl acetate and butyl butyrate (Gu et al., 2019). To better understand butanol production in C. saccharoperbutylacetonicum N1-4(HMT), the key biosynthetic genes (either endogenous or exogenous) including the sol operon (bld-ctfA-ctfB-adc), adhE1, adhE1 D485G , thl, thlA1 V5A , thlA V5A and the expression cassette EC (thl-hbd-crt-bcd) were overexpressed in the strain (Wang et al., 2017b). In summary, overexpression of the sol operon resulted in a 400% increase in the production of ethanol with the highest increase in butanol (13.7%) seen in the strain with the over expression of the EC cassette. In an attempt to better understand and elevate the process of carbon catabolite repression (CCR), the sucrose metabolic pathway was shut down via inactivation of the gene scrO, resulting in a decrease in sucrose consumption by 28.9% and a decrease in ABE production by 44.1% using sucrose as the main carbon source. Additionally, deletion of the scrR gene alleviated CCR in the glucose/sucrose mixed fermentation, and overexpression of the endogenous sucrose pathway resulted in increased ABE production (Zhang et al., 2018).
Despite significant understanding of the Clostridium genus and its genome it is notoriously difficult to engineer. As a result, tools available for genetic engineering of Clostridium have lagged behind those for Gramnegative species such as E. coli. Moreover, the development of tools for creation of robust strain development in C. saccharoperbutylacetonicum N1-4(HMT) has lagged behind that of other Clostridium strains such as C. acetobutylicum and C. beijerinckii, where the ClosTron system developed by Heap et al. (2007) has been widely used (Underwood et al., 2009;Cartman et al., 2010;Cooksley et al., 2010Cooksley et al., , 2012Antunes et al., 2011;Wietzke and Bahl, 2012;Xu et al., 2015). However, CRISPR/Cas technology, a powerful counter-selection method, has recently been used to screen against homologous recombination events in Clostridium (Bruder et al., 2016;Huang et al., 2016;Nagaraju et al., 2016;Wasels et al., 2017). Moreover, CLEAVE TM technology developed by Green Biologics has previously been shown to successfully generate strains carrying desired SNPs as well as those with deletions and insertions (Atmadjaja et al., 2019), although detailed whole genome sequencing (WGS) analyses have not previously been performed on the engineered strains to screen for unwanted mutations. Hence, this method was chosen for chromosomal deletions in the current study and CLEAVE TM was assessed using WGS.
The enzyme of interest in the current study is GapN, a cytosolic non-phosphorylating NADP-dependent glyceraldehyde-3-phosphate dehydrogenase (GAPDH; EC.1.2.1.9) that catalyses the irreversible oxidation of glyceraldehyde-3-phosphate (G3P) to 3-phospholglycerate ( Fig. 1; Iddar et al., 2002). However, three classes of GAPDH exist that are known to be involved in the central carbon metabolism pathway, with the other two being; (i) a NAD + -dependent glycolytic enzyme (EC.1.2.1.12, GapA in C. saccharoperbutylacetonicum) found in the cytoplasm of all organisms that plays a key role in the Embden-Meyerhoff pathway (Fothergill-Gilmore and Michels, 1993); and (ii) a NADP + -dependent GAPDH found in photosynthetic organisms (EC. 1.2.1.13,absent in C. saccharoperbutylacetonicum), that is a key component of the reductive pentose-phosphate cycle (Brinkmann et al., 1989). Originally discovered in photosynthetic eukaryotes (Mateos and Serrano, 1992), GapN was subsequently identified in Streptococcus mutans and S. salivarius (Boyd et al., 1995). Recombinant GapN from Streptococcus mutans has been expressed in Corynebacterium glutamicum as a route for NADPH generation to facilitate L-lysine production (Takeno et al., 2010), and recombinant GapN from C. acetobutylicum has been expressed in E. coli and has been shown to have an absolute specificity for NADPH (Iddar et al., 2002). GapN activity is likely to diminish the production of NADH via GAPDH (Fig. 1), decreasing this valuable source of reducing power that is obligately required solvent production via ABE fermentation. Furthermore, GapN activity also diminishes the production of ATP by phosphoglycerate kinase (PGK; Fig. 1), which is likely to influence cell growth and potentially impact upon ABE yield. Hence, it was hypothesised that deletion of gapN would enhance ABE yield, presumably via increasing the production of NADH and ATP. The study herein describes the deletion of gapN from C. saccharoperbutylacetonicum using CLEAVE TM , assessment of this mutagenesis tool using WGS, and assessment of the impact of gapN deletion upon growth, acid/solvent titre, redox poise and ATP yield.

Results
Deletion of gapN in C. saccharoperbutylacetonicum N1-4 (HMT) using CLEAVE TM Deletion of gapN was carried out using the proprietary CRISPR/Cas technology CLEAVE TM developed by Green Biologics Limited (Atmadjaja et al., 2019). After successful construction of the pMTL82154-gapN-HR vector ( Fig. S1 and Table S1) and transformation into E. coli, this construct was isolated and was used to transform C. saccharoperbutylacetonicum N1-4(HMT). Successful transformants were subjected to several rounds of subculturing intended to promote a double recombination event between pMTL82154-gapN-HR and the chromosome, yielding a clean deletion of the entire gapN gene.
Following sub-culturing, the cells were transformed with the killing vector pMTL83251_Ldr_DR_Sp_DR (Fig. S2) that is capable of producing crRNA that targets the PAM protospacer site derived from the middle of the gapN gene. The crRNA encoded by the killing vector was designed to recognize the PAM/protospacer within cells that have been unable to undergo the double recombination event and still contain the wild type gapN locus, resulting in killing of only these wild type cells. Following successful transformation of the killing vector, colonies were screened using the HR_F1 and HR_R2 primers that anneal upstream and downstream of gapN and were used in the construction of the HR vector ( Fig. S1A and Table S1). Colonies that lacked gapN resulted in a PCR product of 1256 bp compared to the wild type at~3000 bp containing the gapN gene. Successful colonies were sent for Sanger sequencing to confirm successful deletion of gapN. Fig. 1. GapN is the monophosphorylating NADP-dependant GAPDH found in C. saccharoperbutylacetonicum N1-4(HMT). Along with GapA, GapN is responsible for the conversion of glyceraldehyde-3-phosphate to 3-phosphoglycerate. With the GapN reaction producing NADPH rather than NADH and also circumventing ATP generation by PGK, it is hypothesised that deletion of gapN will result in an increase in the concentration of ATP and elevate the NADH:NAD + ratio. Pathways for acidogenesis, solventogenesis and generation of PMF are shown to highlight their involvement in consumption/production of NADH and ATP. To determine whether the genome editing via CLEAVE TM introduced any unwanted mutations, WGS was carried out on both wild type and ΔgapN C. saccharoperbutylacetonicum N1-4(HMT). Genome libraries were prepared using the Nextera XT v2 protocol and sequenced using an Illumina MiSeq benchtop sequencer according to the manufacturer's instructions (Illumina, San Diego, CA, USA). Raw reads have been uploaded to the European Nucleotide Archive and are available under accession numbers ERS6580404 (wild type C. saccharoperbutylacetonicum N1-4(HMT); 'C1') and ERS6580405 (ΔgapN C. saccharoperbutylacetonicum N1-4(HMT); 'C2'). The resulting short paired-end reads were assessed using FastQC before and after trimming of the reads using Trimmomatic (Bolger et al., 2014). Trimmomatic removes adaptor sequences inserted during library preparation, as well as low quality reads and bases from each dataset. The final trimmed reads from the wild type and the ΔgapN strain were mapped to the reference C. saccharoperbutylacetonicum N1-4(HMT) genome obtained from Genbank (CP004121.1). The mapping was carried out using BWA and Samtools (Li, 2013), and then Qualimap (Garc ıa-Alcalde et al., 2012) was used to assess the quality of genome mapping of both strains (Table 1). Qualimap revealed that the coverage depth varied, with 68.6 AE 33.2 seen for the ΔgapN strain compared to 18.8 AE 11.3 for the wild type strain. Although coverage varied, 97.12% of the total 596 655 reads for the wild type strain were mapped and 95.21% of the total 2 093 890 reads from the ΔgapN data were successfully mapped to the reference genome.
The mapped Illumina reads for the ΔgapN strain were analysed for variations compared to the reference genome using Snippy (https://github.com/tseemann/ snippy). All 17 variations that were identified in the ΔgapN WGS (Table 2) were manually compared to our wild type WGS using the Integrated Genome Viewer (Robinson et al., 2011), and were found to be present in both our wild type parent strain and the ΔgapN strain. This indicates that these variations are not due to the CLEAVE TM genome editing process and merely reflect minor differences in our genome sequences compared to the reference strain (CP004121.1).

Solvent quantitation reveals an earlier entry to solventogenesis for the ΔgapN strain
The ΔgapN and wild type C. saccharoperbutylacetonicum N1-4(HMT) strains were cultured via batch fermentation in YETM medium containing 40 g l À1 glucose and supplemented with 0.1 M MES for pH control. Both strains reached similar cell densities, with maximal OD 600 values of 7.2 AE 0.2 for the wild type and 7.4 AE 0.3 for the ΔgapN strain ( Fig. 2A). Between 9 h and 21 h in the fermentation, the media pH for ΔgapN strain was less acidic throughout compared to the wild type C. saccharoperbutylacetonicum N1-4(HMT) (Fig. 2B). The redox potential of the growth media for both strains behaved in a predictable manner, both dropping to~À300 mV vs. NHE in the first few hours (Fig. 2C), and the rate of glucose consumption was greater between 10 h and 20 h the ΔgapN strain (Fig. 2D).
The products of acidogenesis and solventogenesis were then analysed for the ΔgapN and wild type strains for the growth experiments described in Fig. 2. The ΔgapN strain exhibited reduced concentrations of acids throughout the fermentation ( Fig. 3A and B). In addition, depletion of acid concentrations in the ΔgapN strain also occurred at a faster rate in the ΔgapN strain compared to the wild type ( Fig. 3C and D). The largest observed discrepancies between the strains for acetic acid and butyric acid measurements were observed at 15 h: 0.057 g l À1 of butyric acid and 0.073 g l À1 of acetic acid were measured for the ΔgapN strain, compared to 0.170 g l À1 of butyric acid and 0.29 g l À1 of acetic acid for the wild type strain. The concentrations of acetone and butanol produced by the ΔgapN strain were higher during the first 30 h of the fermentation, and the largest observed discrepancies between the strains for acetone and butanol measurements occurred at 15 h: the concentration of acetone produced by the ΔgapN strain was 3.2 g l À1 compared with only 1.1 g l À1 in the wild type strain; [butanol] produced by the ΔgapN strain was 6.87 g l À1 compared to 3.5 g l À1 in the wild type strain. Peak concentrations of butanol in the ΔgapN, however, were similar to those of the wild type cultures at~13 g l À1 .
Since the maximal concentrations of butanol and acetone were similar for ΔgapN and wild type strains, it was hypothesised that solvent toxicity was the limiting factor for these measurements. To test this hypothesis, a butanol toxicity test was carried out that measured growth rates in the presence of varying concentrations of solvents. These data indicate that deletion of gapN does not affect the ability of C. saccharoperbutylacetonicum to tolerate solvent toxicity (Fig. S3).
Deletion of gapN results in a reducing cytoplasm with a higher ATP pool Herein, it was hypothesised that loss of gapN may increase the NADH:NAD + ratio, which could enhance the reducing power required for ABE fermentation (Fig. 1).
del, deletion; ins; insertion; snp; single nucleotide polymorphism. Comparison was performed using Snippy. The ΔgapN deletion is not shown. Hence, it was of interest to quantify the NAD(P)H derivatives during fermentation. Due to the labile nature of NAD (P)H derivatives, cells were grown in serum bottles and assayed at 4 and 24 h to provide concentrations during early exponential phase (acidogenesis) as well as in late exponential/stationary phase (solventogenesis). Deletion of gapN did not appear to affect the concentrations of NADH and NADPH although a marked reduction in the concentrations of NAD + and NADP + during both acidogensis and solventogenesis was observed (Table S2), which is indicative of a more reducing environment in the ΔgapN strain compared to the wild type (Fig. 4). Deletion of gapN was also hypothesised to elevate ATP levels, so the concentrations of ATP were also monitored during fermentation in bespoke bioreactors (Monaghan et al., 2021) and were found to be higher in the ΔgapN strain compared to wild type for the majority of fermentation period (Fig. 5). The greatest difference in ATP concentrations between the ΔgapN and wild type strains was observed at 15 h, where 11.2 µmol g À1 cell dry mass was recorded for the ΔgapN strain compared to 4.7 µmol g À1 cell dry mass for the wild type.

Discussion
The gapN gene was successfully deleted in C. saccharoperbutylacetonicum and no unwanted mutations were detected in the ΔgapN strain compared to the isogenic wild type parent strain, as confirmed by WGS. The ΔgapN strain grew at a similar rate to the wild type, with a slightly more rapid consumption of glucose during exponential phase (Fig. 2). However, the most striking difference in the ΔgapN strain was that acetic acid and butyric acid were converted to acetone and butanol much earlier in the fermentation (Fig. 3), reflecting an earlier shift into solventogenesis and higher solvent titres in exponential phase compared to the wild type. Ultimately, the final titre of butanol was the same as the wild type strain, probably reflecting butanol toxicity as the limiting factor. As there was no difference in butanol tolerance between the wild type and mutant strains (Fig. S3), it would be interesting to engineer butanol tolerance into ΔgapN and wild type strains to investigate the limits of butanol production during the early solventogenic shift.
In solventogenic Clostridia, acid production facilitates the generation of the majority of cellular ATP as a result of kinase activity of the enzymes involved in acetic and butyric acid production (Grupe and Gottschalk, 1992;D€ urre and Hollergschwandner, 2004; Fig. 1). The associated ATP generation in acidogenesis is important as it enables the cells to produce adequate concentrations of ATP for vegetative growth, as well as to establish an ion gradient via V-type ATPases that will enhance the electrical component of the proton motive force (PMF) that drives NADH production via the RnF complex with concomitant oxidation of reduced ferredoxin (Poehlein et al., 2017). This ion gradient enables ADP + Pi recycling when the cells shift into solventogensis via the use of F 1 F O ATPases (Jones and Woods, 1986), as illustrated in Fig. 1. The ΔgapN strain was shown to have elevated concentrations of ATP (compared to the wild type strain) for a sustained period during batch fermentation (Fig. 5).
Between 15 and 24 h there is an observed drop in the measured ATP concentration of the ΔgapN strain over the wild type. This observed decrease in ATP corresponds to the time points with the lowest acid titres and the most rapid rate of solvent production in the ΔgapN strain (Fig. 3). These data suggest that the deletion of ΔgapN increases the ATP concentration such that the ΔgapN strain is able to diminish acid production and allow for an earlier shift into solventogenesis. As well as an increase in ATP production, there is an increase in the ratio of NADH:NAD + in the ΔgapN strain. ATP production, ΔpH generated by acid production and reducing conditions within the cell play a key role in the shift to solventogenesis Wietzke and Bahl, 2012;Zhang et al., 2014;Liu et al., 2018). Previous work on C. acetobutylicum reported that overexpression of both 6-phosphofructokinase (PfkA) and pyruvate kinase (PykA) resulted in an increase in intracellular ATP and NADH, which was also accompanied by elevated butanol and ethanol production (Ventura et al., 2013). It has been shown that in cultures with lower ATP levels acidogenesis is a major route for energy generation, whereas cultures with elevated ATP levels produce more solvents (Meyer and Papoutsakis, 1989). Enhanced butanol production has also been observed as a result of blocking NAD(P)H consumption in Clostridium beijerinckii NCIMB 8052 where insertional inactivation of a NADH-quinone oxidoreductase (nuoG) resulted in increased NAD(P)H and ATP as well as elevated butanol production (Liu et al., 2016). Overall, it can be seen that increases in both ATP and NAD(P)H in solventogenic Clostridium aids in maximising the solvent producing potential of the bacteria.
Previous metabolic flux analysis in in C. acetobutylicum (Yoo et al., 2015) has revealed that GapN is poorly expressed during solventogenesis, with 0.56 mRNA molecules per cell in comparison to 66 mRNA molecules  per cell of GAPDH. This low expression results in only 3500 molecules of GapN protein per cell compared to 190 000 molecules of GAPDH. This study estimated that GapN would be responsible for only 5% of flux through the glyceraldehyde-3-phosphate oxidation pathway, so it was initially surprising that in the current study deletion of gapN in C. saccharoperbutyacetonicum elicited such significant changes in solvent production, NADH:NAD + ratio and ATP yield. However, there are significant metabolic differences between C. acetobutylicum and C. saccharoperbutyacetonicum, including the absence of the Rnf complex in the former, that could account for variations between species.
In conclusion, we have engineered a C. saccharoperbutylaconicum strain that switches earlier to solventogenesis and have provided new insights into the role of GapN in controlling redox poise and ATP yield.

Bacterial strains, plasmids and culture conditions
The clostridium strain used in this study was Clostridium saccharoperbutylacetonicum N1-4(HMT) and plasmids used were based upon pMTL82154 or pMTL83251 (Heap et al., 2007). Liquid cultures were recovered from 15% glycerol stocks and grown at 32°C in reinforced clostridium media (RCM) (Sigma) that had been autoclaved in sealed serum bottles. E. coli strains were grown aerobically at 37°C in Luria-Bertani media supplemented with the appropriate antibiotic when required.

Chromosomal mutagenesis
The gapN gene was deleted in C. saccharoperbutylacetonicum N1-4(HMT) using CLEAVE as previously described (Atmadjaja et al., 2019), and all oligonucleotides are listed in Table S1. Firstly, a homologous recombination vector was generated with a deletion cassette that contains homology arms that are able to replace all of or part of the intended site of mutation. To construct the gapN deletion cassette, colony PCR of C. saccharoperbutylacetonicum N1-4(HMT) was carried out (Fig. S1A) to amplify 2 9 1 Kbp regions upstream and downstream of gapN, containing complementary sequences such that it was possible to generate a seamless in-frame deletion cassette fragment that lacked the gapN gene. Following successful PCR reactions, fragment isolation and Sanger sequencing to confirm the correct sequence of the deletion cassette, it was successfully blunt-end cloned into the StuI restriction site of pMTL82154 (Fig. S1B). The new pMTL82154_gapN_HR vector was first transformed into E. coli. The HR vector pMTL82154 is a lineage of the previously described pMTL80000 shuttle vector system (Heap et al., 2007).
The killing vector, pMTL83251_Ldr_DR_Sp_DR, is also derived from the pMTL8000 shuttle vector system (Heap et al., 2007). The parent pMTL83251 vector contains a Gram-positive replicon pCB102, ermB antibiotic marker, ColE1 + tra Gram-negative replicon and a multiple cloning site (MCS). Pre-engineered into the MCS of pMTL83251 is the native leader sequence (Ldr), a 181 bp sequence that is found downstream of the Cas2 machinery in C. saccharoperbutylacetonicum N1-4(HMT) (Atmadjaja et al., 2019). With the leader sequence preengineered into pMTL83251, it allows for the construction of CRISPR/Cas clusters on the plasmid that target the cleavage of specific chromosomal loci. The CRISPR/ Cas targeting system seen in C. saccharoperbutylacetonicum N1-4(HMT) is comprised of a target-specific spacer flanked by direct repeats (DR_Sp_DR) that is downstream of the Cas2 sequence. For the successful deletion of gapN, a DR_Sp_DR cluster was designed (Fig. S2A), synthesised, and successfully cloned into pMTL83125 (Fig. S2B). The 35 bp spacer sequence that was used in the DR_Sp_DR cluster is present within the native gapN gene.

Whole genome sequencing
WGS was carried out using the Illumina MiSeq system. Genome libraries for both the wild type and ΔgapN strains were prepared using the Nextera â XT DNA Library Prep Kit, which is optimised for small genomes. Once the libraries were prepared, they were sequenced using an Illumina MiSeq benchtop sequencer. Following Illumina â MiSeq, the read quality was assessed with FastQC (http://www.bioinformatics.babraham.ac.uk/projects/ fastqc). To improve read quality, reads were trimmed using Trimmomatic (Bolger et al., 2014). Trimmomatic improves read quality of the Illumina reads by removing sequence adapters, trimming bases at the beginning and end of the reads if they fall below the stipulated threshold, followed by eliminating any of the reads that fall below the minimum read length (default setting of 36 bp). Following this, reads were mapped using BWA (Li, 2013) to the reference genome for C. saccharoperbutylacetonicum N1-4(HMT) obtained from Genbank (CP004121.1). The quality of the genome maps were assessed using Qualimap (Garc ıa-Alcalde et al., 2012). Functional annotation was performed using PROKKA (Seemann, 2014) and Roary (Page et al., 2015) was used for analysis of gene content and the pangenome.

Fermentation conditions
ABE fermentation was carried out as recently described using a bespoke anoxic bioreactor system (Monaghan et al., 2021). Briefly, 1 l Pyrex Quickfit culture vessels were used (SciLabware, Hartlepool, United Kingdom) with 500 ml culture volumes at 32°C and media consisted of yeast extract tryptone media (YETM) (40 g l À1 glucose, 2.5 g l À1 yeast extract, 2.5 g l À1 tryptone, 0.5 g l À1 ammonium sulphate and 0.025 g l À1 iron sulphate) at pH 6.2, supplemented with 0.1 M 2-(N-morpholino) ethanesulfonic acid (MES) free acid (Merck) for pH control. MES buffer was chosen as it has a pK a of 6 and has previously been shown to be the very effective at the pH range observed during ABE fermentation. Anaerobic conditions in the fermentation were generated by sparging filtered (0.2 µm pore size) oxygen-free nitrogen through the fermentation media for 20 min preinoculation and then 5 min post-inoculation. Seed cultures were established by growing recovered RCM grown cells, in 80 ml YETM in serum bottles overnight to an OD 600 of~4.0. The final inoculation was 10% (v/v). Fermentations were carried out in triplicate. Throughout the fermentations OD 600 , pH and redox poise (Inlab Redox Micro electrode from Mettler Toledo, Leicester, United Kingdom) of the fermentation media was measured. The redox electrode was calibrated against quinhydrone (87 mV at pH 7.0, 264 mV at pH 4.0, Normal Hydrogen Electrode (NHE) correction for a Ag/AgCl electrode in 3 M KCl = +210 mV). Following sampling, supernatant and cells were separated by centrifugation at 8000 g for 10 min. Supernatant and cell pellet were separated and frozen at À80°C for later use.

Acid and solvent quantitation
Acids and solvents from culture supernatants were quantitated using Gas Chromatography Mass Spectrometry (GCMS) using an Agilent 6890N instrument as previously described (Monaghan et al., 2021). The GC was equipped with a Phenomenex 7HG-6013-11 Zebron column. Helium (> 99.999%) was used as the carrier gas, with a constant flow rate of 1 ml min À1 . A 0.2 µl water sample was injected with a 100:1 split. Injection temperature was set to 150°C, the GCMS transfer line temperature was set to 280°C, ion source to 230°C, and quadrapole to 150°C. After injection, column temperature was held at 30°C for 5 min, after which this increased at a linear gradient to 150°C at the 20 min mark. Compounds were identified by comparing retention times of each of the compounds with retention times of reference compounds.

Quantitation of NADH and NADPH
NADH, NADPH and their oxidised derivatives were extracted essentially as previously described (Beri et al., 2016). Briefly, 2 ml of bacterial culture was added to 1 ml of 1 M HCl for NAD(P) + extraction and 1 ml of 1 M KCL for NAD(P)H extraction. Cells were incubated at 55°C for 10 min and then the pH was adjusted to pH 6.5 for acidic samples and pH 7.5 for basic samples using 1 M HCl and 1 M KOH dropwise with continual vortexing. Samples were centrifuged for 15 min at 4400 g, with the supernatant being saved for subsequent analysis. To determine the dry cell weight (DCW), cells were harvested were centrifuged at 12 000 rpm for 5 min, washed twice with distilled water, and dried to a constant weight at 80°C.
Concentrations of NADH and NADPH were measured using previously described methods (Nisselbaum and Green, 1969;Bernofsky and Swan, 1973;Baker, 2016). In brief, the reaction mixture for the NADH assay consisted of 100 µl 1 M tricine-NaOH (pH 8), 100 µl 40 mM EDTA, 100 µl 0.1 M NaCl, 100 µl 4.2 mM MTT, 100 µl 16.6 mM PES, 100 µl 100% EtOH. The reaction mixture for NADPH assay consisted of 100 µl 1 M tricine-NaOH (pH 8), 100 µl 40 mM EDTA, 100 µl 0.1 M NaCl, 100 µl 4.2 mM MTT, 100 µl 16.6 mM PES, 100 µl 10 mM Glucose-6-phosphate; 500 µl of each assay mixture was added to 100 µl of extracted sample, topped up to 900 µl with 0.1 mM NaCl and incubated at 37°C for 5 min. To generate standard curves, similar reactions were set up where the extracted sample was replaced with nucleotide solutions of known concentration; 100 µl of alcohol dehydrogenase (Sigma Aldrich, Burlington, MA, USA) and 100 µl glucose-6-phosphate dehydrogenase (Sigma Aldrich) were then added (10 U of each) per reaction. The reaction was incubated at 37°C for 1 h in the dark. 500 µl of 5 M NaCl was then added to stop the reaction and precipitate the MTT. Samples were centrifuged at 10 000 g and 4°C for 5 min. Supernatants were decanted, MTT pellets were resuspended in 1 ml of ethanol and absorbance was measured at 570 nm.

ATP Luminescence assay
ATP was quantified in cells growing in YETM medium using a luminescence assay according to the manufacturer's instructions (Abcam ab113849, Cambridge, United Kingdom). To determine the DCW, cells were harvested were centrifuged at 12 000 rpm for 5 min, washed twice with distilled water, and dried to a constant weight at 80°C.

Sugar quantitation
Sugars were quantitated essentially as previously described (Monaghan et al., 2021). Briefly, culture supernatants were homogenised and centrifuged at 13 400 g for 5 min; 200 µl of the sample was then added to 600 µl of HPLC grade water, achieving a 94 dilution and a total volume of 800 µl. Glucose concentrations were measured using cation exchange chromatography at 60°C using a Phenomenex Rezex ROA H+ column at 1 ml min À1 5 mM sulphuric acid using an Agilent 1100 series refractive index detector to monitor glucose elution. Concentrations of samples were determined by comparison to a standard curve for glucose with integrated peak areas used for the determination of glucose concentration.

Butanol toxicity
To test butanol toxicity, 30 ml serum bottles were filled with RCM, inoculated with 15% C. saccharoperbutylacetonicum glycerol stocks, and were grown overnight at 32°C. The overnights were then used to inoculate (10% v/v) YETM media containing 40 g l À1 glucose in serum bottles under anaerobic conditions. The cells were left to grown for 4 h at 32°C until an OD 600 of~1 was reached, when they were challenged with various concentrations of butanol ranging from 0.5% to 5% (v/v). Subsequent growth rates were then monitored to assess the impact of butanol toxicity.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Generation of a homologous recombination vector for gapN deletion. A) PCR approach to generate the deletion cassette: (i) Amplification of 1 kb fragments upstream and downstream of the gapN gene with 48 bp of complementary sequences; (ii) Two 1 kb PCR products with complementary ends. (iii) Product of overlap-extension PCR, ready to be blunt-end ligated into pMTL82154. B) Vector map of 'pMTL82154_gapN_HR' that contains the homologous recombination (HR) fragment (i.e. the deletion cassette) cloned into the StuI site of pMTL82154 (verified via StuI restriction digests and sequencing). pMTL82154_gapN_HR contains a pBP1 Gram-positive replicon, catP antibiotic maker, ColE1 +tra Gram-negative replicon and a catP reporter gene. Fig S2. Generation of a killing vector for elimination of transformants that do not contain the gapN deletion. A) Overview of the killing vector targeting cassette for endogenous CRISPR-Cas for genome editing. The native leader sequence (Ldr) is a 181 bp sequence found downstream of the Cas2 machinery in C. saccharoperbutylacetonicum N1-4 (HMT) (Atmadjaja et al., 2019). The CRISPR/Cas targeting system is comprised of a target-specific spacer (i.e. gapN spacer) flanked by direct repeats (DR_Sp_DR) that is downstream of the Cas2 sequence. B) Vector map of 'pMTL83251_Ldr_HR_Sp_HR' that contains the targeting cassette from panel A. Successful cloning was confirmed via colony PCR and sequencing. Fig. S3. Butanol toxicity test of wild type (black bars) and ΔgapN (grey bars) strains of C. saccharoperbutylacetonicum N1-4(HMT). Cells were grown to an OD 600 of 1 and were then challenged with varying [butanol]. Doubling times were calculated for the 48 h of growth that followed solvent addition. Table S1. Oligonucleotides used in this study. Table S2. Ratio of nucleotide cofactors in wild type (WT) and ΔgapN C. saccharoperbutylacetonicum N1-4(HMT).