Attempts to use a thermostable kanamycin resistance gene previously used for selection of transformants in Thermoanaerobacterium species at 60°C [36], [37] to select transformants in C. bescii was complicated by the fact that C. bescii (that grows optimally at 75°C) grows very poorly at or below 70°C. In fact, growth at 60°C increased the spontaneous mutation frequency significantly, from 10⁻⁷ to 10⁻⁵, making the detection of transformants over this background of spontaneous drug resistance problematic. Attempts to use a hygromycin phosphotransferase (hph) gene from E. coli, that had been selected for function at 85°C in Sulfolobus solfataricus [38] were compromised by the level of natural resistance to hygromycin in C. bescii and the fact that, in our hands, this hygromycin resistance gene is not reliable for selection above 70°C. To generate a mutant strain for nutritional selection of transformants, C. bescii cells were plated on 5-fluoroorotic acid (5-FOA). OMP decarboxylase, encoded by the pyrF gene in bacteria (ura3 in yeast), converts the pyrimidine analog 5-fluoroorotodine monophosphate to 5-fluorouridine monophosphate which is ultimately converted to fluorodeoxyuridine by the uracil biosynthetic pathway, a toxic product that kills growing cells that are synthesizing uracil [17]. Mutants of pyrF are, therefore, uracil auxotrophs resistant to 5-FOA. Spontaneous resistance to 5-FOA (8 mM) was observed at a frequency of approximately 10⁻⁵ at 60°C. One such mutant contained a deletion that included part of the carboxy terminus of pyrF (Cbes1377) open reading frame, the entire pyrC (Cbes1376) open reading frame and the amino terminus of pyrB (Cbes1375) open reading frame, diagrammed in Fig 1A, and was used for further analysis. The extent of the deletion was defined by PCR amplification of the pyrBCF region in the mutant (Fig 1C) and subsequent sequencing of the PCR product. Since mutations in pyrE also lead to uracil auxotrophy and 5FOA resistance, the region around the pyrE locus (Cbes1382) was amplified from this strain and sequenced to ensure that it was wild type. pyrE is located 3 ORFs and 2.4 kb downstream of pyrA and 5.6 kb downstream of pyrF and is unlikely to be transcriptionally coupled to pyrBCF or pyrA. While the deletion would be expected to affect only the pyrBCF genes, qPCR analysis was performed to monitor expression of the pyrA gene (Cbes1378) as well as the Cbes1374 open reading frame predicted to encode a uracil xanthine permease. Expression of pyrA and Cbes1374 in the deletion mutant was indistinguishable from the wild type suggesting that the deletion within the pyrBCF locus did not affect expression of surrounding genes. The ΔpyrBCF strain, JWCB002, (Table 1) was a tight uracil auxotroph and because it contained a deletion, reversion to uracil prototrophy was not a concern making prototrophic selection possible no matter how low the frequency of transformation. Growth of this mutant supplemented with uracil (20 µM) was indistinguishable from that of the wild type, reaching a cell density of ∼2×10⁸ in 20 hours. To assay transformation, a non-replicating plasmid was constructed with the wild type copy of the pyrBCF locus but containing an engineered restriction site within the pyrC open reading frame to distinguish it from the chromosomal wild type allele. This plasmid was used to transform the pyrBCF deletion strain selecting marker replacement events that repaired the deletion (strategy diagrammed in Fig 1A).

We were unable to transform C. bescii in many attempts using this strategy with DNA isolated from E. coli. We used and modified methods known to work well for other Gram-positive bacteria including electroporation [36], [39], [40], artificially induced competence [41], natural competence [19], [42], and methods that altered membrane permeability [43]. Mating with E. coli, a method of DNA transfer that works well for similar bacteria [15], [44], did not work for C. bescii or the other Caldicellulosiruptor species we tested using the same approach.

CbeI, a potent restriction endonuclease in C. bescii, recognizes and cleaves the same sequence as HaeIII, unmethylated DNA at the sequence 5′-GG/CC-3′ [24]. We anticipated that methylation by HaeIII methyltransferase (M.HaeIII) might protect DNA from E. coli from cleavage by CbeI and overcome it as a potential barrier to DNA transformation. This, in fact, was shown to be true for transformation of the archaeon Sulfolobus acidocaldarius using M.HaeIII to avoid restriction by SuaI [45]. Plasmid DNA treated with M.HaeIII, in vitro, was partially protected from cleavage by both HaeIII and CbeI in vitro (Fig 2), but no transformants were obtained when this DNA was used in electroporation experiments or added to cells that had been subjected to a procedure to induce natural competence in Mycobacterium and Thermoanaerobaterium species [41], [42]. In addition, various strains of E. coli containing combinations of methyltranferases were used to prepare DNA for transformation (Table 2), a method that was successful for transforming Clostridium thermocellum using a dam⁺dcm⁻ E. coli strain [32]. No transformants of C. bescii were detected using DNA from these strains. In total we performed more than 1000 electroporation experiments varying conditions for cell growth, transformation conditions, and assay conditions as well as using DNA from different strains of E. coli or methylated with M.HaeIII.

As shown in Fig 3A, the region of the chromosome that contains CbeI also contains an open reading frame, Cbes2437, predicted to encode an adenine specific methyltransferase [46]. This open reading frame was cloned into an E. coli expression vector, pDCW73 (Fig 3B) that placed a His-Tag at the carboxy terminus of the protein allowing purification on a Ni-NTA column. E. coli cells containing this plasmid were viable at 23°C but not 37°C suggesting that expression of M.CbeI was toxic to growing cells. Expression of this methyltransferase was, therefore, performed at 23°C to avoid problems related to toxicity and in E. coli BL21-CodonPlus(DE3)-RIPL to alleviate problems arising from the significant differences in codon usage between M.CbeI and E. coli. Purified M.CbeI from E. coli was the size predicted from the open reading frame, 33 kDa (Fig 3C). No cleavage of DNA was detected by purified CbeI at 75°C when DNA from E. coli was methylated in vitro by the purified methyltransferase (Fig 3D) and we named this enzyme M.CbeI. To determine the optimal temperature for M.CbeI methyltransferase activity, we performed the in vitro methylation reactions with purified M.CbeI at temperatures ranging from 25°C to 100°C and tested the modified DNA for restriction by CbeI. Reactions performed between 65°C and 85°C, the growth temperature range of C. bescii, resulted in the best protection against cleavage by CbeI.

Even though CbeI is an isoschizomer of HaeIII and M.CbeI would be expected to methylate the same sequence as M.HaeIII, methyltransferases vary in the sites of methylation and specific or cognate methylation may be required for full protection. The pattern of DNA methylation by M.CbeI was compared to that by M.HaeIII using a method [47], [48] that relies on the fact that the extent of incorporation of fluorescently labeled dideoxynucleotides during DNA sequencing is influenced by methylated bases in the template DNA. pUC18 DNA was methylated in vitro by either M.CbeI or M.HaeIII and direct sequencing of the DNA revealed that DNA methylated with M.CbeI showed a higher degree of incorporation of dideoxyguanosine in the 5′-GGCC-3′ recognition sequence than DNA methylated with M.HaeIII. N4-methylcytosine results in an increase in the complementary G (GGCC) signal and this signature (Fig 3E) indicates that M.CbeI methylated DNA contains N4-methylcytosine (m4C). M.HaeIII methylates the C5 position of cytosine (m5C).

Plasmid DNA from E. coli (dam⁺dcm⁺) methylated by M.CbeI in vitro readily transformed the C. bescii ΔpyrBCF strain resulting in marker replacement of the deletion with the wild type allele containing the engineered KpnI site (Fig 1). Amplification of the pyrBCF region from wild type C. bescii resulted in a 3.2 kb product while the product generated from the deletion strain was 1.63 kb. Amplification of this region in the transformant generated a wild type size product. Digestion with KpnI resulted in no cleavage of the product generated from the wild type or the ΔpyrBCF mutant. The product generated from transformant was digested with KpnI showing that the transformant contained the allele from the plasmid and its presence in the C. bescii chromosome resulted from marker replacement (Fig 1C). Transformation efficiencies were routinely on the order of 50 transformants per microgram of non-replicating plasmid DNA (Table 2, Fig 1 B). This extremely low transformation efficiency may be an underestimate of the actual efficiency as the plating efficiency of C. bescii on selective solid medium is less than 10⁻⁴ (plating 10⁶ cells as determined by cell count resulted in fewer than 100 colonies).

All strains used in this study are listed in Table 1. All growth of Caldicellulosiruptor species was preformed under anaerobic conditions in in modified DSMZ 516 medium [24] at a final pH 6.8. Liquid cultures were inoculated with a 1–2% inoculum or with a single colony and then incubated at 75°C overnight in anaerobic culture bottles or Hungate tubes degassed with at least three cycles of vacuum and argon. A solid medium was prepared by mixing an equal volume of liquid medium at a 2X concentration with 1% (wt/vol) Phytagel (Sigma) previously autoclaved to solubilize. Both solutions were maintained at 95°C and poured into petri dishes immediately after mixing. Initial plating of C. bescii in soft agar overlays allowed the cells to grow but they did not form discrete colonies because of the soft and liquid nature of the agar matrix. Increasing the agar concentration from 0.3% to 1.5% in the overlay allowed both abundant growth and the isolation of discrete colonies. Cells from overnight cultures were pelleted, washed in 1X base salts [24] three times and resuspended in 300 – 500 µl of 1x base salts. Cell suspension (100 μl) was mixed with 4 ml of soft top agar (1.5%) and poured across the top of a solid medium plates were incubated in anaerobic jars degassed with at least three cycles of vacuum and argon at 75°C for 3 to 5 days. E. coli strains, DH5α (dam⁺dcm⁺), BL21 (dam⁺dcm⁻), or ET12567 (dam⁻ dcm⁻) were used to prepare pDCW70 DNA. Cells were grown in L broth supplemented with apramycin (50 µg/ml) and plasmid DNA was isolated using a Qiagen Mini-prep Kit. Chromosomal DNA from C. bescii DSM 6725 was extracted using the Quick-gDNA™ MiniPrep (Zymo) according to the manufacturer's instructions.

C. bescii DSM 6725 was inoculated into 10 ml of modified DSMZ 516 medium and grown anaerobically at 60°C for 24 h. Cells were harvested at 18,000 ×g for 5 min, washed twice with 1 x base salts [24], resuspended in 1 ml of 1 x base salts and plated by mixing 100 µl of cell suspension with 4 ml of 0.3% agar and overlaying onto defined modified DSMZ 516 medium (no yeast extract or casein) supplemented with 20 µM uracil and 8 mM 5-FOA (US Biologicals, Swampscott, MA). The plates were incubated anaerobically at 60°C for three days and 5-FOA resistant colonies were transferred to 10 ml of defined modified DSMZ 516 medium with 20 µM uracil and 8 mM 5-FOA and incubated overnight at 75°C anaerobically. To test for uracil auxotrophy, cells were subcultured in defined modified DSMZ 516 medium with or without uracil (20 µM). Cell number was measured in a Petroff Houser counting chamber using a phase-contrast microscope with 40× magnification.

Primers used in these constructions are listed in Table 3. All PCR amplifications were performed using Pfu Turbo DNA polymerase (Agilent Technologies). A 1.858 kb fragment containing the pSC101 replication origin was amplified from pDCW68 [24] using primers DC081 and DC230, which contain KpnI and AatII sites at each 5′end, respectively. A 4.343 kb fragment containing the apramycin resistance and pyrBCF cassettes was amplified from pDCW68 using primers DC084 and DC232 to which an AatII and KpnI site had been added to each 5′ end. An additional fragment (1.801 kb) containing DNA sequences used in other work not relevant to the experiments described here was amplified using primers DC212 and DC213. These three DNA fragments were cut by restriction enzymes, KpnI and AatII, and then ligated to yield pDCW69 (8.014 kb). pDCW70 was constructed by introducing a single nucleotide change (an A to C transversion resulting in a silent mutation) in the +978 amino acid of pyrC (Cbes 1376) ORF using “PCR based Site Directed Mutagenesis”, using DC 214 and DC 215 primers, to create the KpnI site (GGTAC/C), in pDCW 69. To construct pDCW73, the 0.837 kb M.CbeI (Cbes 2437) open reading frame was amplified by PCR using primers DC238 and DC239 using C. bescii genomic DNA as template. The PCR product was digested with BamHI and XhoI and ligated to pET24d [56], which had also been digested with BamHI and XhoI. This vector contains a His-tag sequence that is added to the C-terminus of the expressed protein. All plasmids used in this study were sequenced to confirm their structure.

Purification of M.CbeI was similar to the method described by Chung et al. [24]. BL21-CodonPlus(DE3)-RILP cells (Agilent Technologies), containing pDCW73, were used for M.CbeI protein expression. Cells were grown at 23^°C in L broth supplemented with kanamycin (25 µg/ml) and chloramphenicol (50 µg/ml) to OD₆₀₀ 0.7 and induced by addition of 0.5 mM isopropyl β-D-1-thiogalactopyranoside (IPTG) at 23°C overnight. His-tagged (carboxy terminus) M.CbeI was purified as described previously [24] except for the use of a His-Spin Protein Miniprep™ (Zymo Research). Protein concentration was determined by the Bio-Rad protein assay using bovine serum albumin (BSA) as the standard. Purified protein was displayed using sodium dodecyl sulfate (SDS) polyacrylamide gel electrophoresis (PAGE) and stained with Coomassie brilliant blue G-250 as described [57]. Protein purity was determined to be >98%.

For in vitro methylation, DNA isolated from E. coli DH5α (dam⁺dcm⁺) was treated with either M.CbeI or M.HaeIII methyltransferase (NEB). 5 microgram of purified M.CbeI was incubated with 50 mM Tris-HCl, 50 mM NaCl, 80 μM S-adenosylmethionine (SAM), 10 mM Dithiothreitol (DTT) at pH 8.5 and 20 µg of DNA substrate in 400 µl reaction, and incubate for 2 hours at 78°C. The M.HaeIII methylation reaction was performed according supplier's instructions. To allow complete methylation, an additional 10 units of M.HaeIII and 80 µM SAM was added to the reaction every 4 h of incubation at 37°C for a total of 12 h. Methylated DNAs were purified and concentrated by Phenol/Chloroform extraction and ethanol precipitation. The extent of protection was determined by cleavage using HaeIII and NotI (NEB) restriction enzymes according to the supplier's instructions.

To identify the site of methylation by M.CbeI, unmodified DNA was compared to that after methylation and the changes were determined by direct visualization in automated DNA sequencing chromatograms [47], [48]. In vitro methylation of pUC18 DNA isolated from E. coli DH5α (dam⁺dcm⁺) was carried out using M.HaeIII (NEB) and purified M.CbeI. The efficiency of methylation was determined by cleavage of the methylated and unmethylated DNA with HaeIII (NEB), purified CbeI, and C. bescii cell free extracts (CFE). Digested DNA was displayed by agarose gel electrophoresis and visualization using ethidium bromide staining. Automatic sequencing was performed using primers M13F(−20) and M13R(−20) in an ABI automated PRISM big-dye-terminator system (Macrogen, Inc, Maryland). Sequences were analyzed using the Chromas Lite v2.01 (Technelysium Pty Ltd.) and ABI chromatograms were compared by aligning the Sequencing traces and using SeqDoc [58].

To prepare cells for transformation, 2.5 milliliter of a freshly grown JWCB002 (ΔpyrBCF) culture was inoculated into 500 ml of fresh medium, and incubated at 78°C to mid-log phase (OD₆₈₀ – 0.1, approximately 10⁸ cells/ml). The cultures were cooled to room temperature for 1 hour, harvested by centrifugation (5000 × g, 15 min) at 25°C and washed twice with 250 ml of pre-chilled 10% sucrose. After the final wash, the cell pellets were resuspended in a total volume of 1 ml of pre-chilled 10% sucrose and frozen in microcentrifuge tubes in a dry ice/ethanol bath in 50 μl aliquots for storage at −80°C. Plasmid DNA (0.5–1.0 µg) was added to cells, gently mixed and incubated in 10% sucrose for 15 minutes at room temperature. Electrotransformation of the cell/DNA mixture was performed via single electric pulse (1.8 kV, 600Ω, and 25 microF) in a pre-chilled 1 mm cuvette using a Bio-Rad gene Pulser. After pulsing, cells were incubated overnight at 75°C in 10 ml modified DSMZ 516 medium supplemented with 20µM of uracil, harvested by centrifugation (at 5000×g for 20min) and resuspended in 1 ml of 1 x base salts. A cell suspension (100 microliters) was plated onto defined medium without uracil. Plates were incubated in anaerobic jars at 75°C for three to four days. To confirm marker replacement of the pyrBCF region in the transformants, DNA from uracil prototrophic transformants was used to amplify the chromosomal region using primers DC163 and DC188 which anneal outside the regions of the pyrBCF fragment contained on pDCW70 (Fig 1). PCR products of this locus amplified from the wild type, the deletion mutant and the transformants were digested with KpnI and sequenced.