Increasing the library size in cDNA display by optimizing purification procedures.

BACKGROUND
The library size is critical for selection in evolutionary molecular engineering (directed evolution). Although cDNA display has become a promising in vitro display technology by overcoming the instability of mRNA display, it is hindered by low yields. In this study, we improved the yield of cDNA display molecules by carefully examining each step of the preparation process.


FINDINGS
We found that steric hindrance of ribosomes binding to the mRNA-protein fusion molecules was interfering with biotin-streptavidin binding. Additionally, reducing buffer exchange by performing RNase digestion in the His-tag-binding buffer to release the cDNA display molecules improved their His-tag purification.


CONCLUSION
Our optimized conditions have improved the yield of cDNA display molecules by more than 10 times over currently used methods, making cDNA display more practically available in evolutionary molecular engineering.


Findings
For nearly two decades, evolutionary molecular engineering (directed evolution) has played a role as a complementary partner to rational protein design in protein engineering [1]. In evolutionary molecular engineering, the size of the library is crucial for selection efficiency. Recently, the productivity and versatility of in vitro display technologies have increased the size of libraries by using a cell-free translation system. For example, ribosome [2] and mRNA displays [3,4] have large libraries because the amount of mRNA-peptide/protein (genotype-phenotype) complex with a ribosome or puromycin is proportional to the input of mRNA (~10 12 /ml) in a cell-free translation system. In the case of mRNA display, ligation methods have been developed to synthesize mRNA-puromycin-linker practically [5][6][7]. However, essentially the lability of mRNA in both ribosome and mRNA displays has restricted the experimental selection conditions. Thus, the cDNA display method was developed to improve the stability by converting mRNA to cDNA with a novel puromycin-linker [8]. This technology allows researchers to screen large combinatorial libraries against molecules on a cell surface (e.g. receptors) [9], and to use peptide libraries containing two or more disulfide bonds [10,11]. Although the cDNA display method was useful for in vitro peptide and protein selection, its productivity was hindered by the generation of mRNA/cDNA-protein fusion molecules; only around 0.1% of the initial mRNA was ligated to proteins with a puromycin-linker [8]. Recently, this efficiency has been improved to more than 1% by introduction of a novel puromycin-linker and minor modification of previous method [12,13]. However, the yield of cDNA display fusion molecules is still smaller than that of mRNA display fusion molecules (20-30%).
The aims of this study were to investigate which processes during the preparation of cDNA display fusion molecules shown in Figure 1 cause its low yield, and to increase the yield by overcoming these problems.
First, we determined the binding capacity of streptavidincoated magnetic beads (SA-beads) (Dynabeads MyOne streptavidin C1 streptavidin magnetic beads, Invitrogen, Carlsbad, CA, USA) sufficient to collect most of the mRNA-protein fusion molecules from the rabbit reticulocyte lysate. In our previous study, SA-beads with a binding capacity of 360 pmol of biotinylated DNA primers were used for purifying of 48 pmol of mRNA-linker conjugates   [8]. We thought this binding capacity should be sufficient for capturing this amount of biotinylated mRNA-protein fusion molecules, but the final yield of mRNA/cDNA-protein fusion molecule was much lower than expected. We speculated that ribosomes might strongly bind the mRNA-protein complex in the lysate before purification with the SA-beads. Indeed, we found that Ethylenediaminetetraacetic acid (EDTA) treatment for releasing ribosomes was effective to purify mRNA-protein fusion molecules from the translation mixture (Figure 2A, B). And in the present study, we found that SA-beads with a binding capacity 200 times the amount of biotinylated mRNA-protein fusion molecules were required to purify almost the all fusion molecules ( Figure 3A, B). Similarly, the final amount of purified cDNA-protein fusion molecules also increased with increased SA-beads ( Figure 3C, D). These results suggest that steric hindrance might interfere with biotin-streptavidin binding considerably. Second, we optimized the process from RNase T1 digestion to His-tag purification of cDNA display molecules shown in Figure 1A. In the cDNA display method, it is very important to separate mRNA/cDNA-protein fusion molecules from mRNA/cDNA-linker conjugates (which are not fused with its coding proteins) to reduce the background in the in vitro selection. Thus His-tag sequence (His X6) was incorporated to the C-terminal region of a coding protein.
In this study we examined whether RNase T1 can digest the guanine base of the SBP-linker in the His-tag-binding buffer containing imidazole. If possible, buffer exchange for His-tag purification after RNase T1 digestion would be eliminated and so there could be no loss of cDNA display molecules. We found that RNase T1 worked well in the His-tag-binding buffer, and cDNA-protein fusion molecules were efficiently purified by His-tag purification without any buffer exchanges ( Figure 4A, B). As a result, this modification increased the yield of mRNA/cDNA-protein fusion molecules by 1.5 times over the previous method. Moreover, this improvement allows us to save time and cost in the preparation of mRNA/cDNA-protein fusion molecules and to help make cDNA display technology easier to use.
One of the crucial problems with cDNA display is the low yield of cDNA-protein fusion molecules, which is less than 1% of input mRNA-linker conjugates [8,12]. In this paper, we identified that the mRNA-ribosome-protein complex may sterically hinder the biotin-streptavidin interaction between the puromycin-linker on the fusion molecules and the SA-beads. In addition, mRNA-protein fusion molecules without a ribosome could also interfere with the biotin-streptavidin interaction. Because of these reasons, more SA-beads than expected are required to purify most of the mRNA-protein fusion molecules. The addition of EDTA into the lysate after translation to remove the bound ribosome effectively increased the yield of cDNA display molecules. Furthermore, the simplification of His-tag purification after the release of cDNA display molecules from the SAbeads by performing RNase T1 digestion in the His-tagbinding buffer also increased the yield of cDNA display molecules. Finally we achieved 17% of final yield of cDNA display molecule based on the input mRNA-linker conjugates ( Figure 4B), which is more than 10 times higher than in our previous study [8,12]. Additionally, we recently also succeeded in releasing cDNA display molecules from SAbeads by using Endonuclease V instead of RNase T1 by designing a new puromycin-linker [14]. Thus, we believe that this new linker and our currently optimized conditions will make cDNA display more useful and practical for in vitro protein selection.