One-step generation of error-prone PCR libraries using Gateway® technology

Antoine Gruet 0 Sonia Longhi 0 Christophe Bignon 0 0 Architecture et Fonction des Macromolecules Biologiques (AFMB), UMR 7257 CNRS and Aix-Marseille University , 163, Avenue de Luminy, Case 932, 13288 Marseille, Cedex 09 , France Background: Error-prone PCR (epPCR) libraries are one of the tools used in directed evolution. The Gateway technology allows constructing epPCR libraries virtually devoid of any background (i.e., of insert-free plasmid), but requires two steps: the BP and the LR reactions and the associated E. coli cell transformations and plasmid purifications. Results: We describe a method for making epPCR libraries in Gateway plasmids using an LR reaction without intermediate BP reaction. We also describe a BP-free and LR-free sub-cloning method for in-frame transferring the coding sequence of selected clones from the plasmid used to screen the library to another one devoid of tag used for screening (such as the green fluorescent protein). We report preliminary results of a directed evolution program using this method. Conclusions: The one-step method enables producing epPCR libraries of as high complexity and quality as does the regular, two-step, protocol for half the amount of work. In addition, it contributes to preserve the original complexity of the epPCR product. - Background Gateway is an appealing technology because its cloning efficiency is close to 100% [1]. This feature is particularly welcome when dealing with numerous target genes, for instance in Structural Genomics. Unfortunately, high throughput gene expression following gene cloning in Structural Genomics programs has also revealed that many recombinant proteins are insoluble in E. coli thereby precluding their crystallization and their study by X-ray crystallography. Among the different techniques used to overcome this insolubility problem, one is directed evolution. The use of directed evolution for improving recombinant protein solubility can be summarised as follows. A random library of mutants generated by error-prone PCR (epPCR) and/or DNA shuffling [2] is screened for variant proteins more soluble than the wild-type (wt) protein. To that end, the mutated DNA sequences may be expressed as fusion proteins with a C-terminal solubility reporter such as the green fluorescent protein (GFP) [3]. To assess the solubility gain provided by the mutations, the mutated coding sequences are then sub-cloned from the solubility reporter expression plasmid to a GFP-free expression plasmid and the solubility of the tag-free variant is compared to that of the tag-free wt protein expressed under the same conditions. Although the Gateway technology is less used in directed evolution than in Structural Genomics programs, it has been nevertheless successfully applied in a directed evolution study that made use of both epPCR and DNA shuffling [4]. The evolved Tobacco Etch Virus (TEV) protease exhibited significantly higher solubility than the wtTEV protease. Incidentally, this study also revealed a few weak points that seemed to be specifically associated with the use of the Gateway technology rather than with the screening process or the protein to evolve. In particular, (i) the number of expression clones was found to be relatively small, as also reported in another study [5]; (ii) the generation of epPCR and DNA shuffling libraries was labor intensive because of the need for BP and LR reactions to be carried out, and of the corollary transformations and intermediate plasmid medium preparations [6]; (iii) the subcloning of the coding sequence of selected mutants from the reporter expression plasmid to a non-reporter expression plasmid was also time-consuming because of the same requirements. While the first of the drawbacks listed above can be easily addressed by transforming expression cells by electroporation, addressing the other two requires devising a novel cloning and sub-cloning strategy. With the specific purpose of overcoming these limitations while maintaining the obvious advantages of the Gateway technology, we devised a method that allows eliminating the BP step from the generation of the library and both the BP and LR steps from the sub-cloning process. We applied this method to generate a diversity library of the intrinsically disordered C-terminal domain of the measles virus nucleoprotein (NTAIL) [7,8] as a first step towards the dissection of the molecular mechanisms underlying its interaction with the C-terminal X domain (XD, aa 459-507) of the viral phosphoprotein [9-17]. A split-GFP reassembly assay [18-20] was used to screen the library and to identify clones with novel binding properties. Results 1) Generation of an epPCR library The conventional procedure for generating epPCR libraries using the Gateway technology comprises two recombination reactions (BP and LR) [4]. We first addressed the question as to whether each recombination reaction and associated E. coli cell transformation decreased the complexity of a given library. A typical Gateway recombination reaction can be described as the transfer of an insert from a donor to a non-recombined acceptor to yield a recombined acceptor. Therefore, the library complexity loss can be evaluated by comparing the number of colonies provided by: (i) a theoretical experiment made of a 100% efficient LR reaction (i.e. a reaction where all the non-recombined acceptor (i.e., Gateway plasmid before LR reaction) molecules are used to yield recombined acceptors (i.e., Gateway plasmids after LR reaction)) followed by a 100% efficient cell transformation (i.e. a transformation where all recombined acceptor molecules are uptaken by cells and where each cell uptakes one recombined acceptor molecule); (ii) an actual cell transformation by a recombined acceptor; (iii) an actual cell transformation by an actual LR reaction using the same donor construct (i.e., the other substrate of the LR reaction) and the same non-recombined acceptor as in the previous two instances. The results of this comparison are reported in Table 1. Since 25 fmoles of acceptor correspond to 1.55 1010 molecules, if the LR reaction and cell transformation were each 100% efficient, then one Table 1 Assessment of the efficiency of E.coli transformation by different DNA species The number of colonies obtained after E. coli transformation by a recombined acceptor1, or by an LR reaction2 is reported. T7pRos E. coli cells were either electroporated as described in Methods, or transformed by heat-shock. Transformed cells were selected on ACplates. The results are the mean value and standard deviation of three independent transformations using the same LR reaction 125 fmoles of pNGG-NTAIL (Table 2). 2using 25 fmoles of non recombined acceptor plasmid (pNGG, Table 1) and 100 fmoles of linear NTAIL coding sequence flanked by attL recombination sites (Figure 1, stage 1, right panel) should obtain 1.55 1010 colonies per 25 fmoles of input acceptor. However, transforming E. coli ce (...truncated)