Thermophilic site-specific recombination system for rapid insertion of heterologous DNA into the Clostridium thermocellum chromosome

Abstract

Clostridium thermocellum is an anaerobic thermophile capable of producing ethanol and other commodity chemicals from lignocellulosic biomass. The insertion of heterologous DNA into the C. thermocellum chromosome is currently achieved via a time-consuming homologous recombination process, where a single stable insertion can take 2–4 weeks or more to construct. In this work, we developed a thermostable version of the Serine recombinase Assisted Genome Engineering (tSAGE) approach for gene insertion in C. thermocellum utilizing a site-specific recombinase from Geobacillus sp. Y412MC61, enabling quick and easy insertion of DNA into the chromosome for accelerated genetic tool screening and heterologous gene expression. Using tSAGE, chromosomal insertion of plasmid DNA occurred at a maximum transformation efficiency of 5 × 10³ CFU/µg, which is comparable to the transformation efficiency of a replicating control plasmid in C. thermocellum. Using tSAGE, we chromosomally integrated and characterized 17 reporter genes, 15 homologous and 31 heterologous constitutive promoters of varying strengths, 4 inducible promoters, and 5 riboswitches in C. thermocellum. We also determined that a 6–7 nucleotide gap between the ribosome binding site (RBS) and the start codon is optimal for high expression by employing a library of superfolder green fluorescent protein expression constructs driven by our strongest tested promoter (P_{clo1313_1194}) with different distances between the RBS and start codon. The tools developed here will aid in accelerating C. thermocellum strain engineering for producing sustainable fuels and chemicals directly from plant biomass.

One-Sentence Summary: A highly efficient site-specific recombination system was created for Clostridium thermocellum, which enabled the rapid characterization of a large collection of genetic parts for controlled gene expression.

Graphical Abstract

Graphical Abstract.

Introduction

Clostridium thermocellum (Viljoen et al., 1926) (also known as ​​Acetivibrio thermocellus, Hungateiclostridium thermocellum, and Ruminiclostridium thermocellum) is an anaerobic thermophile (McBee, 1954) capable of deconstructing lignocellulosic biomass and fermenting the resulting cello-oligomers (Akinosho et al., 2014; Blumer-Schuette et al., 2008, 2014; Hasunuma et al., 2013; Mazzoli & Olson, 2020). Its native metabolic pathways produce a mixture of chemicals as fermentation products, including ethanol, acetate, lactate, formate, H₂, isobutanol, 2,3-butanediol, and free amino acids (Carere et al., 2008; Freier et al., 1988; Gheshlaghi et al., 2009; Holwerda et al., 2014; Levin et al., 2006; Mazzoli & Olson, 2020; McBee, 1954; Sai Ram et al., 1991; Sparling et al., 2006). Research in the past decade has focused on metabolic engineering of C. thermocellum toward industrial applications, with a primary focus on ethanol production (Argyros et al., 2011; Biswas et al., 2014; Brown et al., 2011; He et al., 2011; Hörmeyer et al., 1988; Lo et al., 2017; Papanek et al., 2015; Tian et al., 2017, 2019; Yu et al., 1985). One of the bottlenecks when exploring genetic modifications to improve production in this organism is a lack of efficient genetic tools needed to accelerate throughput of strain construction (Guss et al., 2012; Holwerda et al., 2020; Klapatch et al., 1996; Kwon et al., 2020; Olson & Lynd, 2012; Riley et al., 2019; Tripathi et al., 2010; Tyurin et al., 2004).

Because C. thermocellum transformation efficiency is too low to directly select for homologous recombination of a non-replicating plasmid into the chromosome during transformation, stable insertion of heterologous DNA into the C. thermocellum chromosome has primarily relied on a multi-step process that utilizes a temperature-sensitive replicating plasmid based on pNW33N. One method utilizes three regions of homology, as well as one selectable and two counter-selectable markers, as previously described (Olson & Lynd, 2012). Alternatively, a temperature-sensitive replicating plasmid with the gene(s) of interest flanked by DNA homologous to the chromosomal target can first be transformed into C. thermocellum at the permissive temperature, selecting for thiamphenicol resistance encoded by the cat gene. Then, incubation at a non-permissive temperature (60°C) allows for selection of genomic integration via homologous recombination. Finally, a counter-selectable marker gene like hpt or tdk present on the plasmid (now genomically integrated) allows for selection of the second recombination event in the presence of corresponding anti-metabolites (8-azahypoxanthine or 5-fluoro-2′-deoxyuridine, respectively). These protocols typically take about 4 weeks to complete in wild type C. thermocellum and even longer in slower-growing mutant strains, with each colony having a 50% chance of being the strain of interest in the final step. The use of thermophilic Lambda Red recombineering homologs to improve homologous recombination efficiencies followed by CRISPR-based counterselection utilizing a thermophilic Cas9 from Geobacillus stearothermophilus (Harrington et al., 2017) or the native C. thermocellum Type IB CRISPR system as a counterselection in a two-step method could decrease this time to 2 weeks with high efficiency (Walker et al., 2019), but this method still suffers from low transformation efficiency due to a lack of inducible promoters and the need for multiple steps. Thus, the speed of gene integration remains a significant bottleneck for screening chromosomally integrated genetic parts (e.g. promoters, reporters, ribosome binding sites, RBS) and heterologous pathways for metabolic engineering.

Serine recombinase Assisted Genome Engineering (SAGE) is a method that has been recently developed for rapid and simple genomic integration of genetic cassettes into model and non-model organisms like Escherichia coli (Riley et al., 2023), Pseudomonas fluorescens SBW25, Pseudomonas putida Gpo1, Pseudomonas frederiksbergensis TBS10, Rhodopseudomonas palustris CGA009, and Rhodococcus jostii RHA1 (Elmore et al., 2023). This technique uses a large serine recombinase that facilitates a site-specific recombination event between two non-identical, approximately 50–100 base pair DNA sequences called attB and attP sites. Recombination between these attB and attP sites results in the formation of new attL and attR sites, leaving genetic “scars” that are not substrates for further recombination, making the recombination reaction irreversible and stable (Baltz, 2012). This is unlike the FLP-frt and the CRE-lox tyrosine recombinase-mediated systems, which are reversible and can result in strain instability issues when used multiple times for genome engineering (Baltz, 2012; Elmore et al., 2023; Riley et al., 2023).

SAGE requires the presence of an attB site that is specific for the serine recombinase being used, either present naturally or engineered into the chromosome of the organism. Then, a plasmid expressing the recombinase (here called the helper plasmid) and a plasmid containing the cognate attP site and the DNA insert of interest (called the cargo plasmid) can be co-transformed into the strain. Transient expression of the recombinase from the helper plasmid catalyzes recombination between the chromosomal attB site and the attP site on the cargo vector. This inserts the entire cargo vector into the organism’s chromosome (Fig. 1). SAGE recombinases have been shown to function well at mesophilic temperatures, but none have been tested at thermophilic temperatures (Elmore et al., 2023).

Fig. 1.

Thermostable Serine recombinase Assisted Genome Engineering (tSAGE) and backbone excision. Shown is strain AG10995 containing the Y412MC61 attB site, and analogous steps occur for strain AG8235 that contains the poly-attB cassette. (a) One or more attB sites, for instance from Geobacillus Y412MC61, was genomically integrated using homologous recombination. This parent strain is used for all tSAGE integrations. (b) The integrating cargo plasmid contains a chloramphenicol acetyl transferase (cat) gene flanked by Bxb1 attP and attB sites, the genetic cargo of interest, and a Geobacillus Y412MC61 attP site. The helper plasmid expresses the Geobacillus Y412MC61 recombinase. The cargo and the helper plasmids are co-transformed into competent cells of the parent strain using electroporation. (c) Upon expression of the Y412MC61 recombinase, the cargo plasmid (d) integrates stably into the C. thermocellum chromosome and creates Y412MC61 attL and Y412MC61 attR sites. (e) The Bxb1 recombinase-expressing helper plasmid is then transformed into the strain from (d). (f) Expression of the Bxb1 recombinase results in removal of cat, resulting in the strain in (g) with a Bxb1 attL site left behind. tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

In 2014, Yang et al. computationally predicted new serine recombinases and their cognate attP and attB sites in various prokaryotes, including one from Geobacillus sp. Y412MC61 and Parageobacillus thermoglucosidasius C56-YS93, each sharing 94% identity (Yang et al., 2014). Geobacillus strains can grow at thermophilic temperatures, and recently, the P. thermoglucosidasius C56-YS93 recombinase was used for plasmid integration in P. thermoglucosidasius into a native attB site (Styles et al., 2021). Together, this suggests that they could be adapted for use in other thermophiles such as C. thermocellum.

In this work, we explore mesophilic and Geobacillus sp. Y412MC61 recombinases to develop a highly efficient method for inserting DNA into the chromosome of C. thermocellum, which we call thermophilic Serine recombinase Assisted Genome Engineering (tSAGE) (Fig. 1). We then apply tSAGE to the development and characterization of a suite of genetic parts for use in metabolic engineering of C. thermocellum.

Materials and Methods

Strains and Culturing Conditions

LL1299, a C. thermocellum strain in which a gene encoding a restriction enzyme (clo1313_0478) was deleted (Tian et al., 2019), was the parent strain used throughout this study. AG8235 is a derivative of LL1299 in which a poly-attB landing pad with Bxb1, ΦBT1, ΦK38, ΦC1, RV, TG1, R4, BL3, A118, MR11, SPβ, and Φ370, and Geobacillus Y412MC61 attB sites replaces the clo1313_2366 locus. AG10995 is similar to AG8235 but has only the Geobacillus Y412MC61 attB site at this locus. C. thermocellum strains were grown inside a Coy Labs anaerobic chamber with a gas mixture of 85% N₂, 10% CO₂, and 5% H₂ on rich CTFUD medium (Olson & Lynd, 2012) without shaking, supplemented with 15 µg/mL thiamphenicol (Tm15) when needed to select for plasmids and 1.5% agar when needed for solid medium. Plasmids were constructed and maintained in E. coli TOP10 ∆dcm (AG583) (Guss et al., 2012) grown in LB (Miller) supplemented with 25 µg/mL chloramphenicol or 25 µg/mL neomycin when needed for plasmid maintenance with shaking at 37°C.

Plasmid Construction and Sequencing Confirmations

The plasmids used in this study are listed in Supplemental Table 1, and the plasmid sequences are available in Supplemental File 1. The plasmids were synthesized by Genscript Inc. Supplemental Table 1 also lists the parts of each plasmid, its function, and the library it was a part of. All sequencing was confirmed by Sanger sequencing (Genscript Inc.) or Oxford Nanopore Technologies sequencing (Plasmidsaurus).

Competent Cell Preparation and Transformation

For E. coli, an overnight-grown culture of E. coli AG583 was used to inoculate 50 mL of LB in a 250-mL flask to an OD₆₀₀ of ∼ 0.05. The culture was then grown for ∼2–3 hr in a shaking incubator at 37°C to a final optical density at 600 nm (OD₆₀₀) between 0.5 and 0.7. The cells were centrifuged at 7000 X g at 4°C, washed three times with ice-cold 10% glycerol, and the pellet was resuspended in 500 µL of 10% glycerol. 30-µL aliquots of AG583 electro-competent cells were either stored at −80°C for later use or immediately used to transform plasmids. Plasmids were electroporated into competent E. coli AG583 using a Bio-Rad Gene Pulser Xcell set for exponential decay, 25 µF, 200 ohm, and 1800 volts in a 1-mm cuvette. After the electrical pulsing, the cells in the cuvette were resuspended in 950 µL of SOC medium in 1.5-mL microfuge tubes, incubated at 30°C for 3–4 hr, and plated on LB with appropriate antibiotics (chloramphenicol or neomycin) and incubated at 30°C overnight. The colonies were confirmed by PCR for successful plasmid transformation. Plasmids were extracted using a ZymoResearch midi prep kit from 50-mL liquid cultures (LB with appropriate antibiotics) using the low copy number plasmid extraction protocol.

Clostridium thermocellum competent cells were made as previously described (Guss et al., 2012; Tyurin et al., 2004). Briefly, a frozen stock of C. thermocellum stored at −80°C was used to inoculate 5 mL of CTFUD (with neomycin for tSAGE backbone excision), which was grown at 50°C inside a Coy anaerobic chamber. The 5-mL seed culture was used to inoculate 500-mL CTFUD (with neomycin for tSAGE backbone excision) and grown overnight at 50°C in the anaerobic chamber to an OD₆₀₀ between 0.5 and 0.7. The culture was then chilled on ice, centrifuged aerobically at 7000 g for 15 min on a benchtop centrifuge set at 19°C, and washed with ice-cold electroporation buffer (250-mM sucrose and 10% glycerol) three times without resuspending the pellet each time. Then, the washed pellet was resuspended in an additional 100 µL of fresh, sterile electroporation buffer, resulting in a total volume of approximately 500 µL of competent cells. The cells were either used immediately for transformation or were stored at −80°C for use later. 30 µL of competent cells and 1 µg of each plasmid were combined in an electroporation cuvette, which was transformed using a square wave pulse (1000 V, 1.5 ms, 1-mm cuvette) in a Bio-Rad Gene Pulser Xcell using plasmids extracted from E. coli AG583. After the electrical pulsing, the cells were resuspended in 1 mL CTFUD (no antibiotic) and recovered at 50°C overnight inside the anaerobic chamber. The recovered culture was plated on CTFUD-agar with appropriate antibiotics. The resulting colonies were picked into CTFUD with appropriate antibiotics, grown at 50°C, and then confirmed by PCR. The primers used in this work for PCR confirmation of integration are listed in Supplemental Table 2.

Strain Construction

The attB landing pads were genomically integrated into LL1299, replacing clo1313_2366, using the homologous recombination method detailed below and elsewhere (Groom et al., 2018; Tripathi et al., 2010; Yayo et al., 2021). The reporter, promoter, riboswitch, and RBS libraries were inserted into the chromosome using tSAGE. The strains generated in this study are listed in Supplemental Table 3.

Homologous recombination for genomic integration of genetic cassettes

For insertion of the attB sites into the chromosome, after plasmid transformation, colonies typically appeared after 3–4 days and were picked into 5-mL liquid CTFUD + Tm15. The cultures were incubated until turbid (1–3 days) at 50°C and were then subcultured into 5 mL of fresh CTFUD + Tm15 and incubated at 60°C to select for genomic integration of the replicating plasmid. Once turbid, cultures were streak plated on CTFUD + Tm15 agar plates and incubated at 60°C. Single colonies were picked into 5-mL liquid CTFUD + Tm15, grown at 60°C, and genomic integration of the plasmid was confirmed using PCR. Two isolates with genomic integration at the downstream homologous arm and two with genomic integration at the upstream homologous arm were chosen for the next steps. The cultures were sub-cultured in fresh CTFUD (no antibiotic) and grown at 60°C to allow time for a second recombination event and to lose the plasmid. The cultures were then streaked onto CTFUD-agar with 10 µg/mL FUDR and incubated at 60°C. Individual colonies were picked into liquid CTFUD with 10 µg/mL FUDR, grown at 60°C, and PCR screened to check for the presence of the genetic cassette. Typically, ∼50% of strains tested reverted to WT, while ∼50% of strains contained the genetic cassette from the plasmid transformed, as expected.

tSAGE, screening mesophilic recombinases, and backbone excision using mesophilic recombinases

DNA insertion into C. thermocellum into the corresponding attB site via tSAGE was accomplished by co-transforming 1 µg of the integrating cargo plasmid and 1 µg of the helper plasmid pNA42 via electroporation into competent cells of C. thermocellum AG8235 or AG10995. After an overnight anaerobic recovery at 50°C, the cells were plated on CTFUD + TM15 agar plates. Colonies typically appear in 2–3 days after incubation at 50°C. Colonies were picked into liquid CTFUD + TM15, and subsequently, confirmed by PCR using primers #13 and #14, listed in Supplemental Table 2.

The mesophilic serine recombinases Bxb1, φBT1, φC1, RV1, TG1, R4, BL3, A118, MR11, and φ370 (Elmore et al., 2023) were tested in C. thermocellum at 42°C, 48°C, and 50°C. Competent cells of AG8235 were co-transformed with 1 µg of the poly-attP integrating plasmid (pHS96) and 1 µg of each of the ten serine recombinase-expressing helper plasmids (Supplemental Table 1). Following electroporation, the cells were resuspended in CTFUD and recovered overnight at 42°C, 48°C, or 50°C. The cells were then plated on CTFUD + TM15 agar and incubated at 50°C in the anaerobic chamber. The resulting colonies were picked into liquid CTFUD + TM15 and tested for genomic integration of pHS96 at the clo1313_2366 locus using primers #13 and #14, listed in Supplemental Table 2.

Plasmid pNA122 is a modified version of pNA28, with Bxb1 attB and attP sites flanking the E. coli origin of replication and cat marker. To test plasmid backbone removal, competent cells of AG11004 (strain AG10995 with pNA122 genomically integrated using tSAGE) were transformed using electroporation with 1 µg of pNA56G. The cells were resuspended in CTFUD (no antibiotics) and recovered overnight at 50°C. The recovered cells were then plated on CTFUD with 300 µg/mL neomycin. Colonies typically appeared after 3–4 days and were picked into CTFUD with 300 µg/mL of neomycin. After PCR confirmation for backbone excision, these strains were grown in CTFUD without antibiotics.

Inducible Gene expression, FbFp, and Superfolder GFP Assays

To assay the fluorescence of superfolder GFP (sfGFP) and each flavin-based fluorescent protein (FbFp), overnight grown cultures were moved to aerobic conditions, washed once with 1X PBS, resuspended in 3 mL of 1X PBS, and then incubated at room temperature overnight in the dark to enable green fluorescent protein (GFP) maturation. Where appropriate, PCR-confirmed strains were grown overnight at 50°C inside an anaerobic chamber with inducers at the following concentrations. L(+)arabinose (Fricke et al., 2020; Guzman et al., 1995) (1.56, 3.125, 6.25, 12.5, 25, and 50 mM; EMD Millipore Corp., 178680-100GM), D-(+)-Xylose (Müh et al., 2019; Williams-Rhaesa et al., 2018) (3.125, 6.25, 12.5, 25, and 50 mM; Sigma-Aldrich, X1500-500 G), Cumate (Ji et al., 2019) (7.8, 15.6, 31.25, 62.5 125, 250, and 500 µM; System Biosciences, QM150A-1), 2-aminopurine (Marcano-Velazquez et al., 2019) (0.0625, 0.125, 0.25, 0.5, 1, and 2 mM; Thermo Fisher Scientific, J64919-MD), and sodium fluoride (Baker et al., 2012; Ma et al., 2021; Ren et al., 2012; Yadav et al., 2022) (0.031, 0.0625, 0.125, 0.25, 0.5, and 1 mM; Sigma, S6776). For sfGFP, the fluorescence was measured with excitation at 488 nm and emission at 510 nm in a BioTek fluorescence plate reader. For mScarlet and its variants, the fluorescence was measured with excitation at 540 nm and emission at 620 nm. For the FbFps, five different excitation and emission wavelengths were tested (Ex:450/9, Em: 480/9; Ex: 464/9, Em: 486/9; Ex: 455/9, Em: 483/9; Ex: 455/9, Em: 491/9; Ex: 465/9, Em: 493/9). The excitation and emission at 465/9 and 493/9 found as the best and used throughout assays. The fluorescence was normalized to an OD₆₀₀ of 1.0.

Codon Optimization

Codon optimization was performed using the GenSmart Codon optimization tool (www.genscript.com/tools/gensmart-codon-optimization) (Koblan et al., 2018), with the expression host system set to Clostridium acetobutylicum. The DNA coding for the protein’s amino acid sequence was then modified using this tool.

All data used to generate figures are provided in Supplemental File 2.

Results and Discussion

Development of Thermophilic SAGE Recombination System for C. Thermocellum

A strain of C. thermocellum was constructed that contains a “landing pad” of 13 orthogonal attB sites inserted into the clo1313_2366 locus of strain LL1299, resulting in strain AG8235 (Fig. 1a illustrates strain AG10995 with a single attB site, and a similar scenario applies to strain AG8235 with multiple attB sites). We first tested the efficiency (number of transformed colonies per μg of DNA) and accuracy (percent of colonies that have the plasmid inserted at the attB site) of 10 mesophilic recombinases (Bxb1, φBT1, φC1, RV1, TG1, R4, BL3, A118, MR11, and 370) and the Geobacillus sp. Y412MC61 recombinase. This was accomplished by co-transforming strain AG8235 with two plasmids: (1) the “integration-testing plasmid” pHS96, which contains corresponding attP sites for 12 of the recombinases, a cat gene that confers thiamphenicol resistance driven by the GAPDH promoter, P_GapD, a second cat gene that is an artifact of a previous cloning strategy, and the E. coli p15A origin of replication, and (2) the “helper plasmid”, which consists of plasmids pNA56-G, pNA58-G through pNA66-G, and pNA42 expressing the Bxb1, RV1, φBT1, φC1, A118, MR11, φ370, TG1, R4, BL3, and Y412MC61 recombinases, respectively, from the clo1313_2638 promoter (Fig. 1b demonstrates this concept using a cargo plasmid with a Y412MC61 attP site and a helper plasmid that expresses Y412MC61 recombinase; see Supplemental Table 1). The transformation efficiency using Y412MC61 was equivalent to that of transformation with a replicating control plasmid: 2.86 × 10³ CFU/µg for integration of pHS96 and 3.32 × 10³ CFU/µg for the replicating plasmid pAMG216 (Fig. 2). Of the mesophilic recombinases, only TG1 and Bxb1 enabled recombination between their cognate attP and attB sites, with a transformation efficiency of 6 and 18.7 CFU/µg of integration vector, respectively, at a recovery temperature of 50°C. PCR testing of pHS96 at the Y412MC61 attB site of AG8235 (6 colonies) and the Bxb1 and TG1 attB sites of AG8235 (4 colonies each) confirmed integration with 100% accuracy (Supplemental Fig. S1). None of the other eight mesophilic recombinases tested in this work resulted in plasmid integration. These results confirm that the Y412MC61 recombinase is a thermostable enzyme capable of catalyzing efficient insertion of DNA into the chromosome in thermophilic organisms like C. thermocellum and that Bxb1 and TG1 can additionally function under thermophilic conditions, albeit less well than Y412MC61.

Fig. 2.

Transformation efficiency of plasmid pHS96 integrated into the chromosome of the poly-attB landing pad containing C. thermocellum strain AG8235 using TG1, BXB1, and Y412MC61 recombinases, with replicating plasmid pAMG216 as positive control. Data represent the average and standard deviation of three biological replicates.

Because the attB and attP sites are not identical, recombination results in two new sites, called attL and attR, that each consist of a portion of the attB and a portion of the attP site (Fig. 1c–d illustrates this using only Y412MC61 attB and attP sites). The location of the crossover event dictates the exact sequence of the attL and attR, and this location has not been experimentally validated for Y412MC61. Therefore, we PCR-amplified the region around clo1313_2366 and sequenced the resulting PCR product. The sequencing results showed integration of pHS96 occurred at the Y412MC61 attB site via recombination at the central (TG), consistent with bioinformatic prediction (Yang et al., 2014) (Supplemental Fig. S2).

Plasmid Backbone Excision Using Bxb1 and TG1 Recombinases

As described above, tSAGE integrates the complete cargo plasmid irreversibly into the chromosome of C. thermocellum via a single recombination event, leaving the antibiotic resistance gene on the chromosome. Removal of the E. coli origin of replication and the thiamphenicol resistance gene would allow further genetic modifications. With the discovery that TG1 and Bxb1 recombinases function in C. thermocellum, we tested whether they could be used to remove the backbone of the integrated plasmid (Fig. 1e–g depicts the concept using Bxb1 attB and attP sites). Since the landing pad of AG8235 has 13 attB sites, including for Bxb1 and TG1, homologous recombination was used to place the single Y412MC61 attB at the clo1313_2366 locus of strain LL1299, resulting in strain AG10995. We introduced a Bxb1 attB and attP sites on either side of the E. coli origin and the cat gene from pNA28 to make pNA122, which was inserted into the chromosome of AG10995 by tSAGE, resulting in strain AG11004. In order to test the efficiency of backbone excision, AG11004 was transformed with pNA56G, which expresses the Bxb1 recombinase. PCR screening of six resulting colonies showed that all six colonies had undergone backbone excision (Supplemental Fig. S3), and those strains were each sensitive to thiamphenicol, as expected. Thus, Bxb1 recombinase efficiently excised the backbone and resulted in a final strain with a small 50 bp recombination scar and the genetic cargo. Because the pNW33N origin of replication is unstable in C. thermocellum, growth in the absence of neomycin resulted in the loss of the Bxb1 recombinase plasmid, producing a strain that can be used for further strain engineering.

Reporter Screening Using tSAGE

Generally, efforts focused on characterizing gene expression are performed on replicating plasmids (Olson et al., 2015), but because of variability of plasmid copy number and/or instability, they are often not as reproducible and applicable to methods requiring stable chromosomal expression of genes and pathways. Thus, we next used tSAGE with the Y412MC61 recombinase to insert a series of gene expression parts into the C. thermocellum chromosome. Evaluation of these gene expression tools often involves use of a reporter gene, of which a wide variety are currently available (Streett et al., 2021). Previous studies testing promoters in C. thermocellum utilized beta-galactosidase (lacZ) as a reporter (Olson et al., 2015), but the assays can be variable due to the need to add reagents to measure activities. Fluorescent reporters are simpler for measuring activity directly, but many of the most widely used reporters are derived from mesophilic organisms and require O₂ to become fluorescent, such as GFP and its color, stability, and brightness variants. However, these proteins can be synthesized anaerobically and then exposed to O₂ to allow protein maturation to create the fluorophore (Scott et al., 1998). Anaerobically active FbFp have also been described, though they are typically not as bright as GFP (Mukherjee et al., 2012). We used tSAGE to integrate 17 reporters into the genome of C. thermocellum strain AG8235, including two variants of the FbFp gene from Chloroflexus aggregans (cagFbFP-V1 and cagFbFP-V2 (Nazarenko et al., 2019), the wild-type gene for YNP3-FbFp and its reportedly brighter variant YNP3Y116F-FbFp (Inskeep et al., 2013; Wingen et al., 2017), the FbFp from Meiothermus ruber (mrFbFp) (Wingen et al., 2017), and two versions each of mKATE (Pletnev et al., 2008), sfGFP, and mScarlet and its variants (Bindels et al., 2017; Gadella et al., 2023) with and without codon optimization. In all these plasmids, the reporter gene was driven by the P_{Clo1313_1194} promoter from C. thermocellum DSM1313. This promoter was previously shown to be a strong promoter in C. thermocellum (Olson et al., 2015). Cells were grown overnight to stationary phase in rich medium prior to being assayed.

Wild-type YNP3 FbFp was the brightest among the flavin-based fluorescent proteins at 51.50 ± 5.41 relative fluorescent units normalized to optical density (RFU/OD₆₀₀), while the FbFp from Meiothermus ruber was found to be the dimmest (0.37 ± 0.78 RFU/OD₆₀₀). Expression of mKATE and mScarlet did not result in any detectable fluorescence in C. thermocellum. While the mScarlet variants with and without codon optimization demonstrated a wide range of brightness, the codon-optimized version of mScarlet-I3 was the brightest variant (337.1 ± 9.2 RFU/OD₆₀₀) in C. thermocellum. The two sfGFP variants were found to be the brightest reporters in C. thermocellum, being ∼60-fold brighter than the brightest FbFp (Fig. 3). Subsequently, sfGFP was the chosen reporter for screening promoters in C. thermocellum.

Fig. 3.

Reporter gene characterization in C. thermocellum using tSAGE. A library of tSAGE-integrated reporter genes expressed from the P_{Clo1313_1194} promoter were assayed for fluorescence. FBFP refers to flavin-based fluorescent protein, which does not need oxygen to become fluorescent. mrFBFP is from Meiothermus ruber. CagFBFP-V1 and CagFBFP-V2 are from Chloroflexus aggregans and the YNP3FBFPs are from the YNP metagenome project. mKATE, mSacrlet, and its variants are red fluorescent proteins, and sfGFP is a green fluorescent protein, which all require oxygen for the chromophore to mature. Data represents the average and standard deviation of three biological replicates. FBFP, flavin-based fluorescent protein; sfGFP, superfolder GFP; tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

Promoter Screening Using t-SAGE

Rational strain engineering typically requires an array of well-characterized low-, mid-, and high- strength constitutive promoters. Olson et al. previously tested a replicating plasmid-based library of homologous promoters using thermostable beta-galactosidase and alcohol dehydrogenase genes as reporters (Olson et al., 2015), and we aimed to extend this work by characterizing chromosomally inserted promoters with sfGFP as a reporter.

When choosing promoters, native/homologous promoters can have higher expression levels, while heterologous promoters can be more stable by not being targets for homologous recombination with the native locus. To identify thermophilic heterologous promoters to test in C. thermocellum, BLAST-P was used to identify homologs of the top three C. thermocellum genes in the closely related thermophilic anaerobe Acetivibrio clariflavus DSM 19732 and the more distantly related thermophilic, facultative anaerobe Geobacillus thermodenitrificans NG80-2, with the upstream region of the top hits being used as additional promoters in the library. Additionally, promoters of the lactate dehydrogenase gene from Geobacillus stearothermophilus, the ribosomal protein gene rplS from Geobacillus thermoglucosidasius (Reeve et al., 2016), and previously used promoters from Thermoanaerobacterium saccharolyticum JW/SL-YS485 genes tsac_0046 (a pyruvate ferredoxin oxidoreductase gene), tsac_0068 (an ABC transporter), tsac_0530 (a 30S ribosomal protein), and tsac_0915 (a second pyruvate ferredoxin oxidoreductase) were also selected for further testing (Walker et al., 2019).

We used tSAGE to integrate 15 native C. thermocellum promoters and 31 heterologous promoters fused to sfGFP into the Y412MC61 attB site in the AG8235 chromosome (Supplemental Table 1). Each of the integrating vectors used in this promoter library (Supplemental Table 1) have the same backbone with a Y412MC61 attP site, a colE1 origin for propagation in E. coli, and a thiamphenicol resistance gene driven by the C. thermocellum P_gapD promoter, and they were inserted using helper plasmid pNA42, as above.

Among the homologous promoters tested, the promoters of genes clo1313_1194 (AG9469), two versions of clo1313_3011 (211 and 387 bp upstream of the gene, creating AG9462 and AG9178, respectively), and clo1313_2638 (AG9176) had the highest sfGFP expression of 3426,1785, 1445 and 1655 RFU/OD₆₀₀, respectively (Fig. 4). In comparison, two length-variants (309 bp for AG9170 and 263 bp for AG9171) of the homologous P_gapD promoter from C. thermocellum, a commonly used promoter for strong expression, had sfGFP expression levels of 407 ± 28 and 326 ± 22 RFU/OD₆₀₀, respectively (Fig. 4).

Fig. 4.

Constitutive promoter characterization using tSAGE in C. thermocellum. A library of tSAGE-integrated promoters fused to superfolder GFP were assayed for fluorescence. Promoters were from C. thermocellum (clo1313), T. saccharolyticum (Tsac), Geobacillus thermodentrificans (GTNG), Caldicellulosiruptor bescii (Athe), Acetovibrio clariflavus (Clocl), G. thermoglucosidasius (Geoth), and Streptococcus pneumoniae (SPH), as detailed in the text. Homologous and heterologous promoters are denoted by dark gray and white shading, respectively. The parent strain AG8235 without a fluorescent marker is denoted by light gray shading. Data represent the average and standard deviation of three biological replicates. GBP, green fluorescent protein; tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

The promoter of the gene gtng_2506 from G. thermodenitrificans NG80-2, an ortholog to the clo1313_2638 gene from C. thermocellum, and the promoter of gene tsac_0046 were the strongest heterologous promoters, with sfGFP expression of 1374 and 1124 RFU/OD₆₀₀, respectively (Fig. 4). The promoter of the gene athe_2105 from another thermophilic biomass degrading bacterium, C. bescii, as well as the promoters of genes clocl_4203 and clocl_2515 from A. clariflavus, exhibited similar expression levels when compared to the strong native P_gapD promoter (Fig. 4). The promoter of gene tsac_0530, which has been previously used in C. thermocellum to drive recombineering machinery for CRISPR (Walker et al., 2019), was found to be a mid-level promoter with an sfGFP expression of 75.9 RFU/OD₆₀₀. The cat promoter, a promoter that drives the chloramphenicol acetyltransferase gene from different strains of S. pneumoniaedasius, and the promoter of tsac_0915, which have also previously been used in C. thermocellum (Walker et al., 2019), had very low expression in C. thermocellum (17.4 and 6.8 RFU/OD₆₀₀). Collectively, we have validated a library of promoters that contains strong, mid-level, and weak promoters from homologous and heterologous sources for use in strain engineering in C. thermocellum.

Inducible Promoter and Riboswitch Screening Using t-SAGE

Inducible promoters and riboswitches are essential tools in synthetic biology that enable time- and dose-dependent gene expression. Currently, the inducible gene expression systems available for C. thermocellum are limited to the laminaribiose-inducible promoter, which is based on its native celC operon (Mearls et al., 2015; Newcomb et al., 2007), and the pbuE riboswitches from Bacillus subtilis that respond to 2-aminopurine (Marcano-Velazquez et al., 2019). Laminaribiose is not ideal as an inducer because it can activate the native celC operon in C. thermocellum, it is metabolized by C. thermocellum, and it is too expensive for large-scale applications. Furthermore, both systems have only been tested on replicating plasmids and would need to be tuned for use at single copy on the chromosome. Therefore, we next explored alternate inducers and chromosomally encoded, regulated gene expression systems in C. thermocellum using tSAGE.

Xylose is a potentially useful inducer molecule because it is not natively metabolized by C. thermocellum (Tafur Rangel et al., 2020), and thus can be used as an orthogonal inducer. Moreover, C. thermocellum has also been engineered to utilize xylose, and so it could also be used for transient induction or specifically during growth on lignocellulose, which releases xylose during enzymatic deconstruction. Therefore, we first designed a xylose-inducible expression system (Fig. 5a) based on the system native to C. bescii (Williams-Rhaesa et al., 2018). In C. thermocellum, this promoter resulted in approximately 40-fold increased sfGFP expression when induced with 6.25 mM or higher concentrations of xylose (Fig. 5b). In the absence of xylose, no sfGFP expression was detected, indicating tight control of gene expression (Fig. 5b). We also tested the xylose-inducible promoter from Clostridioides difficile (Müh et al., 2019) (Fig. 5c), the cumate inducible promoter from P. putida (Ji et al., 2019) (Fig. 5d), and the arabinose inducible promoter from E. coli (Fricke et al., 2020; Guzman et al., 1995) (Fig. 5e), but all exhibited very low levels of expression and none were inducible in C. thermocellum.

Fig. 5.

Characterization of inducible promoters fused to sfGFP integrated via tSAGE into C. thermocellum. (a) Overview of the design of the C. bescii xylose inducible expression system. CatR, chloramphenicol resistance gene; xylR, xylose regulator from C. bescii; P_xylC, xylose-inducible promoter from C. bescii. Normalized fluorescence of strains with the chromosomally integrated (b) xylose inducible promoter from C. bescii, (c) xylose inducible promoter from C. difficile, (d) cumate-inducible promoter from P. putida, and (e) arabinose inducible promoter from E. coli, and grown under increasing concentrations of xylose, xylose, cumate, and arabinose, respectively. Data represent the average and standard deviation of three biological replicates. sfGFP, superfolder GFP; tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

Riboswitches are RNA domains that can control translation upon ligand binding during transcription or post-transcriptionally. A recent study demonstrated that the 2-aminopurine (2-AP)-inducible pbuE riboswitches identified in Bacillus subtilis function in C. thermocellum strains on replicating plasmids. These riboswitches exhibited background expression in the absence of the inducer 2-AP (Marcano-Velazquez et al., 2019). Mutant pbuE riboswitches were also generated by introducing rational mutations to the wild-type riboswitch included modifications to extend the P1 stem from 5 base pairs to 8 (P1 = 8) and 10 (P1 = 10) base pairs to improve the sensitivity and inducibility (Marcano-Velazquez et al., 2019). To increase the tightness of the riboswitches and test their activities as genomically integrated single-copy expression cassettes, we integrated the wild-type and the two mutated pbuE riboswitches fused to sfGFP into the chromosome of C. thermocellum using tSAGE. When driven by the strong P_{Clo1313_1194} promoter, all three pbuE riboswitches showed moderate inducibility (approximately 14-, 8-, and 3-fold induction, respectively) but also leakiness when uninduced (Fig. 6a). To reduce the leakiness of the pbuE riboswitches, we replaced the strong P_{Clo1313_1194} promoter with medium-strength promoters P_{Tsac_0068} and P_gapD. The P_{Tsac_0068} promoter did not decrease the leakiness, but it resulted in a lower maximum expression level (Fig. 6b). The P_gapD promoter improved both the leakiness and the inducibility in the wild-type and P1 = 8 pbuE riboswitches, with an approximately 50- and 30-fold increase in expression upon induction, respectively (Fig. 6c). The P1 = 10 riboswitch could not be tested under the control of the P_GAPD promoter due to multiple unsuccessful attempts at transformation of the plasmid into C. thermocellum. In addition to the pbuE riboswitches, we also tested two fluoride-inducible riboswitches from Bacillus cereus and Thermotoga petrophila (Baker et al., 2012; Ma et al., 2021; Ren et al., 2012; Yadav et al., 2022) but these constructs were not inducible in C. thermocellum (Fig. 6d–e).

Fig. 6.

Characterization of riboswitches using tSAGE in C. thermocellum. The pbuE riboswitches expressed using (a) P_{clo1313_1194}, (b) P_{tsac_0068}, and (c) P_gapDH fused to sfGFP were grown with increasing concentrations of 2- aminopurine and characterized for normalized fluorescence. AG8235 (no sf-GFP) (black bar), pbuE WT (white bar), pbuE P1 = 8 (light gray bar), and pbuE P1 = 10 (dark gray bar). The fluoride-inducible riboswitches from (d) B. cereus and (e) T. petrophila were fused to sfGFP and grown with increasing concentrations of sodium fluoride and characterized for normalized fluorescence. Data represent the average and standard deviation of three biological replicates. sfGFP, superfolder GFP; tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

RBS Spacer Library Screening Using t-SAGE

The number of bases between the RBS and start codon impacts translation efficiency, but optimal spacing has not yet been explored in C. thermocellum. The P_{Clo1313_1194} promoter, the most highly expressed homologous promoter in our library, has an RBS and spacer sequence reading “AGGGGGAAAAAAACT” before the ATG start codon (RBS underlined). To find the best distance between the RBS and the start codon in C. thermocellum, we created a library in which the distance between the RBS and the start codon was varied from 3 to 12 bases. We also tested the consensus RBS by changing the RBS of P_{Clo1313_1194} to “AGGAGGAAAAAAACT”. The resulting library was genomically integrated into the chromosome of C. thermocellum using tSAGE and the sfGFP fluorescence was measured (Fig. 7). The nucleotide distance of six to seven bases between the RBS and the start codon was found to result in the highest gene expression from P_{Clo1313_1194}. Also, the consensus RBS (AGGAGGA) performed similarly to the native RBS.

Fig. 7.

Characterization of RBS spacing using tSAGE in C. thermocellum. A library of RBS spacers with different spacing between the RBS and start codon of P_{clo1313_1194} were fused to sfGFP, inserted into the chromosome via tSAGE, and characterized for normalized fluorescence. The RBS is capitalized, the spacer is lowercase and underlined, and the subscript ATG is the start codon. Native P_{clo1313_1194} promoter (black bar), decreased distance (light gray bars), or increased distance (white bars) between the RBS and the start codon. The dark gray bar indicates when the RBS was changed to the consensus (AGGAGGA). Data represent the average and standard deviation of three biological replicates. RBS, ribosome binding site; sfGFP, superfolder GFP; tSAGE, thermostable Serine recombinase Assisted Genome Engineering.

Conclusion

The development of rapid and high-throughput genetic tools is critical for accelerating progress of rational metabolic engineering for industrial biotechnology. For strains that will be deployed commercially, heterologous pathways will generally need to be integrated into the chromosome rather than expressed on replicating plasmids the require antibiotic selection. Therefore, tools that simplify insertion of DNA into the chromosome are needed. Here, we show that tSAGE can integrate heterologous DNA into the C. thermocellum chromosome at similar efficiencies to transformation with autonomously replicating plasmids, and that use of a second recombinase can result in removal of the selectable marker to create a strain lacking antibiotic resistance genes, enabling further strain engineering. We used tSAGE to develop a suite of chromosomally integrated constitutive promoters, a tightly regulated promoter, a tightly controlled riboswitch, and an RBS spacer library. These tools will add value to the C. thermocellum metabolic engineering field, and will serve as the foundation for applying these tools to other thermophilic microorganisms.

Notes

Notice: This manuscript has been authored by UT-Battelle, LLC, under Contract No. DE-AC0500OR22725 with the U.S. Department of Energy. The U.S. Government retains and the publisher, by accepting the article for publication, acknowledges that the U.S. Government retains a non-exclusive, paid-up, irrevocable, world-wide license to publish or reproduce the published form of this manuscript, or allow others to do so, for the U.S. Government purposes. The Department of Energy will provide public access to these results of federally sponsored research in accordance with the DOE Public Access Plan (http://energy.gov/downloads/doe-public-access-plan).

Funding

This material is based upon work supported by the Center for Bioenergy Innovation (CBI), U.S. Department of Energy, Office of Science, Biological and Environmental Research Program under Award Number ERKP886.

Conflicts of interest

AMG, HSS, and NA have submitted a patent application related to use of the thermophilic recombinase. YK, NA, AMG, and CAE have submitted a patent application related to use of the xylose-inducible promoter.