Overcoming restriction as a barrier to DNA transformation in Caldicellulosiruptor species results in efficient marker replacement

Biomass recalcitrance represents the greatest obstacle to the efficient conversion of lignocellulosic biomass to commodity chemicals and biofuels [1–3]. For this reason, thermophilic cellulolytic bacteria that are capable of degrading and utilizing plant biomass are of special interest. Members of the Caldicellulosiruptor genus are able to utilize several plant-derived substrates efficiently, including unpretreated switchgrass, and are the most thermophilic of the cellulolytic bacteria (optimum growth temperature near 80°C) [4–6]. These species accomplish plant biomass degradation by producing an arsenal of extracellular carbohydrate degrading enzymes [4, 7, 8] that include cellulases with multiple catalytic enzyme modules in a single multi-domain enzyme. This is distinct from, but somewhat similar to, membrane-bound cellulosomes exemplified by Clostridium thermocellum and other anaerobes [5, 8–11]. Recent growth experiments on crystalline cellulose (Avicel) revealed a significant disparity in plant cell wall deconstruction capability among eight sequenced Caldicellulosiruptor species [4]. These distinctive features provide a unique opportunity for the identification of enzymes that facilitate plant biomass decomposition, as well as the basis for a better understanding of the mechanisms of crystalline cellulose degradation.

The development of Caldicellulosiruptor species for “consolidated bioprocessing” (CBP) [12] has been limited by the lack of genetic tools required to create stable strains with high yields of desired biofuels and/or bioproducts. Recently, we reported methods for efficient DNA transformation of C. bescii and C. hydrothermalis, including transformation of a shuttle vector [13] and the ability to direct marker replacement between non-replicating plasmids and chromosomal genes [14]. Restriction by CbeI was shown to be an absolute barrier to DNA transformation [15], but could be overcome by in vitro methylation of DNA by a cognate methyltransferase, M.CbeI [14].

Restriction-modification (R-M) systems were initially identified in Escherichia coli nearly 6 decades ago [16, 17] and are now known to be wide spread in bacteria and archaea. Almost 90% of bacterial genomes contain R-M systems and 43% contain four or more according to “The Restriction Enzyme Database” (REBASE) [18]. R-M systems comprise pairs of distinctive enzymatic activities, a restriction endonuclease and a DNA methyltransferase. R-M systems are classified as type I, type II, type IIS, type III and type IV according to enzyme composition, cofactor requirements, recognition sequence symmetry, location of DNA cleavage relative to the recognition site, and mode of action [19]. They provide the best-characterized defense mechanism in prokaryotes - a “self-nonself discrimination”, against invasion of foreign DNA that includes phages or conjugative plasmids [20, 21]. The methyltransferase subunits of R-M systems methylate specific sites in the host DNA (“self”) thus preventing cleavage by the cognate restriction endonuclease. Nonmethylated foreign DNA (“nonself”) is cleaved by the restriction endonuclease [22]. R-M systems also constitute a formidable barrier to efficient DNA transformation for genetic manipulation, especially DNA from other genera, most notably, E. coli. Eliminating restriction endonuclease activities in a number of host organisms, including Bacillus subtilis[23], Thermosynechococcus elongates[24], Borrelia burgdorferi[25], and Clostridium acetobutylicum[26], improved transformation efficiency and simplified the transformation protocols by removing the time-consuming laborious DNA modification steps.

Here we show that deletion of the gene encoding the CbeI (Cbes2438) restriction endonuclease resulted in a strain that is easily transformable with unmethylated DNA from E. coli, eliminating the need for in vitro methylation by M.CbeI (Cbes2437). CbeI is a type II restriction endonuclease that recognizes the sequence 5′-GGCC-3′ [15]. We further extend the current study to the analysis of chromosomal DNA modification in other species of Caldicellulosiruptor, showing that the pattern of DNA modification is quite diverse across the genus. Both GenBank [27] and REBASE [18] predicted that all 8 sequenced Caldicellulosiruptor species [28–33] contain a large number of R-M systems. While the isolation or construction of restriction-deficient strains for all members of this genus is impractical at this time, plasmid DNA from C. hydrothermalis, which appears to have a similar R-M system to C. bescii, is transformable to C. bescii without additional modification. We present a strategy for transformation and genetic manipulation of the other species within the Caldicellulosiruptor genus.

Here we report the construction of a deletion of cbeI (Cbes2438), which encodes a potent restriction endonuclease whose activity is apparently an absolute barrier to transformation of DNA from E. coli to C. bescii. We recently showed that in vitro methylation of DNA from E. coli with M.cbeI, a cognate methyltransferase from C. bescii, allowed transformation. Deletion of cbeI is the first targeted deletion in this genus and the resulting mutant, JWCB018 (ΔpyrFA ΔcbeI), is readily transformed by DNA isolated from E. coli without in vitro methylation. PCR amplification and sequencing suggested that this deletion left the adjacent methyltransferase (M.CbeI) intact and its function was confirmed by the fact that chromosomal DNA isolated from JWCB018 was protected from digestion by CbeI and HaeIII in vitro.

The construction of this mutant strain removes a substantial barrier to routine transformation and chromosomal modification and is substantially more efficient than other possible methods including modification of the vector prior to transformation, using engineered vectors containing no or fewer restriction sites recognized by restriction endonuclease in host [38, 39], conditional inactivation of the R-M systems [40, 41] or using group II intron insertion technology to disrupt a restriction system in Clostridum acetobutylicum[26]. The ability demonstrated here to make targeted gene deletions is a powerful and direct tool for the investigation of in vivo gene function and the deletion of this endonuclease resulted in a strain that will provide the basis for further genetic manipulation. The combined efficiencies of transformation and homologous recombination (with as few as 450 bp of homology) in C. bescii allow us to use non-replicating plasmids for genetic manipulation. This is fortuitous and a significant benefit for the development of Caldicellulosiruptor species as CBP organisms. The proven CBP microbe, Clostridium thermocellum, for example, is genetically tractable but the efficiency of transformation and/or recombination does not permit the use of non-replicating plasmids for marker replacement, significantly extending the time required for mutant construction [42].

Plasmid DNA isolated from C. hydrothermalis was able to transform C. bescii JWCB005 (ΔpyrFA) and C. bescii JWCB018 (ΔpyrFAΔcbeI), however total DNA isolated from the C. bescii transformant did not transform C. hydrothermalis. This apparent contradiction could be due to the low concentration of pDCW89 in total DNA isolated from the C. bescii transformant [13]. Alternatively, this may be due to yet another unidentified R-M system difference between these species. REBASE [18] predicts that the genome of C. bescii encodes as many as six methyltransferases in addition to M.CbeI and that C. hydrothermalis contains as many as six restriction endonucleases, one of which (Calhy0018) is a type IV (methyl-directed) restriction endonuclease not present in C. bescii. If this type IV restriction endonuclease is active and recognizes methylated DNA formed by a C. bescii methyltransferase not present in C. hydrothermalis, the DNA isolated directly from C. bescii would be cleaved. Somewhat surprisingly, the disruption of only one functional restriction enzyme was enough to overcome restriction as a barrier to transformation of C. bescii by E. coli DNA. This may be due to the fact that BamHI and BspEI are rare cutters in C. bescii genomic DNA and neither of these sites is present in pDCW88 used to make the deletion of cbeI. Only one BspEI site is present in the pDCW89 shuttle vector and that may be protected due to an overlapping dam site created by in vivo methylation in E. coli DH5α. It is also possible that the cognate restriction endonuclease of the BspEI methyltransferase is inactive or absent altogether.

Perhaps the most powerful conclusion to be drawn from these data is the direction forward in overcoming R-M systems as a barrier to transformation of other Caldicellulosiruptor species. As shown in C. bescii, a 4 base cutter such as CbeI (5′-GGCC-3′) is a formidable obstacle to transformation. In contrast, 6 base cutters can be easily avoided during plasmid construction (as with BamHI and BspEI here), as long as the functional R-M system is known. Based on our findings, all Caldicellulosiruptor species show at least three DNA methyltransferase activities (Table 1), though it is not known that all modification activities are paired with a cognate restriction activity. In vitro methylation of plasmid DNA with cell-free extracts may be useful going forward to identify restriction activities that present obstacles to transformation. In any case, these observations emphasize that successful transformation is largely empirical. It was perhaps fortunate that C. bescii was the first chosen for genetic manipulation, because only one methyltransferase activity was required for efficient transformation of DNA from E. coli.