(Roberts and Szostak 1997) B. mRNA display
FIGuRE 3.2 Ribosome display (A) and mRNA display (B). A simple affinity selection is pictured. (A) In ribosome display, a library of DNA constructs is transcribed in vitro. The unmodified mRNA, which contains no stop codon, is translated to form the selection particle, a stalled ribosomal complex that consists of mRNA, ribosome, and protein. These ternary complexes are allowed to bind to immobilized target molecules. Unbound or nonspecifi-cally bound complexes are washed away, and mRNA from remaining complexes is eluted by removing Mg2+ from the buffer. The released mRNA is isolated and reverse transcribed to form an mRNA/cDNA duplex. PCR is then used, with or without diversification, to amplify the selected members of the library. (B) mRNA display begins in the same manner as ribo-some display, with in vitro transcription of a DNA library. The resulting mRNA must be modified, however, to include a DNA-puromycin linker at the 3' end. When this modified mRNA is translated, the ribosome stalls on the DNA, giving the conjugated puromycin an opportunity to enter the ribosomal A site and accept the nascent polypeptide. The mRNA-protein fusions are then purified, and the mRNA is reverse transcribed prior to selection. The selection step is conducted as in ribosome display, except that the biochemical conditions may be harsher, since the genotype-phenotype linkage is covalent in mRNA display. After washing, the cDNA is released by hydrolysis and amplified for another round.
performed in situ, without purifying the mRNA (He and Taussig 2007). When dsDNA has been recovered, one full round of ribosome display is complete. The entire process may be repeated as many times as necessary, and the stringency of the selection step may be increased with each round. Detailed protocols have been published using both prokaryotic (Zahnd et al. 2007a) and eukaryotic (He and Taussig 2007) extracts.
History, Advances, and Applications
"Polysome display" using short peptides was first described by Mattheakis et al. (1994). Shortly thereafter, Hanes and Pluckthun (1997) demonstrated a similar technique for proteins, which they called "ribosome display." Although polysomes can still form in the method that Hanes and Pluckthun reported, only ribosomes near the 3' end of the transcript should contain functional, native protein. Additionally, when transcription and translation are decoupled, the ratio of purified mRNA to ribosomes in the extract can be more carefully controlled in order to minimize the likelihood of forming polysomes, which wastefully employ multiple ribosomes on the same transcript and may complicate affinity selections due to avidity effects.
Ribosome display was originally used to improve the binding properties of a single-chain variable fragment (scFv) of an antibody (Hanes and Pluckthun 1997; Schaffitzel et al. 1999) and was then applied to select and evolve high affinity binders from synthetic naive scFv libraries (Hanes et al. 2000). Phage display, in a direct comparison with ribosome display, yielded less diverse, lower affinity scFvs (Groves et al. 2006). More recently, scaffolds other than scFvs have also been evolved successfully in ribosome display. Notably, designed ankyrin repeat protein (DARPin) libraries, which are stable, soluble, and well expressed in an S30 translation system, have been used in selections for high-affinity binders, initially to model ligands such as maltose binding protein (Binz et al. 2004) and subsequently to clinically relevant targets such as HER2 (Zahnd et al. 2007b), which is expressed on 20-30% of breast tumors.
Many properties other than binding affinity can also be evolved, provided that a selection pressure can be effectively applied. For example, scFvs that are stable and functional in the reducing cytoplasm (i.e., in the absence of disulfide bonds) have been evolved using decreasing redox potentials with each round (Jermutus et al. 2001). Ribosome display can also be used to evolve catalytic activity, provided that a mechanism-based inhibitor can be designed. Mechanism-based inhibitors, or "suicide inhibitors," bind to and inactivate an enzyme. This strategy was used in a proof-of-principle experiment to enrich a catalytically active ^-lactamase over an inactive point mutant (Amstutz et al. 2002). Additionally, ribosome display has been used to evolve folding properties by selecting against proteins that are sensitive to proteases or have relatively large hydrophobic areas exposed (Matsuura and Pluckthun 2003).
One particular aspect of the ribosome display protocol that investigators have experimented with is the mechanism for stalling. In the original protocol, mRNA lacking a stop codon was used, and the translation mix was supplemented with anti-ssrA oligonucleotides to prevent release of stalled ribosomes by ssrA in the extract (Hanes and Pluckthun 1997). This method works relatively well at low temperatures and high Mg2+ concentrations, as the complexes may be used as long as 20 days at 4°C (Pluckthun et al. 2000), but the efficiency of ribosome display can still be improved. cDNA yields after a selection round can be increased by using the PURE system during translation, as ribonucleases (RNases), proteases, and ssrA are not naturally present (Villemagne et al. 2006). Introducing a pseudoknot into the transcript can also increase cDNA yield, as the ribosome is less likely to read to the end of the mRNA strand, where release can more readily occur (Kim et al. 2007). To stabilize the complexes at elevated temperatures, some have exploited peptide-mediated stalling by inserting the stalling sequence of E. coli secretion monitor (SecM) at the C-terminal end of the tether (Evans et al. 2005). This sequence interacts with the ribosome exit tunnel to cause elongation arrest (Nakatogawa and Ito 2002). This tactic has been used for in vivo ribosome display to evolve scFvs that fold and function efficiently in the cytoplasm of bacteria (Contreras-Martinez and DeLisa 2007). Yet another stalling technique, ribosome-inactivation display system (RIDS), employs the ricin A chain, which is capable of stabilizing eukaryotic ribosomal complexes even in the presence of a stop codon (Zhou et al. 2002). The template used in RIDS encodes a cDNA library, a short linker, the ricin A chain, and a C-terminal spacer. Once the ricin A chain has fully emerged from the ribosome, it catalyzes a single hydrolysis event in the large ribosomal subunit, which alters the binding site for elongation factors and stalls the complex.
Ribosome display was the first fully in vitro display technique used for the directed evolution of proteins, and it is the most widely reported. It is straightforward to perform, as it involves relatively few processing steps, and it can handle large library sizes of ~1012-1014 (Leemhuis et al. 2005). Also, the ribosome may increase the solubility of proteins that, if displayed by other methods, might aggregate. For example, a protein that tends to form amyloid-like fibrils remains soluble when expressed on the ribosome, presumably because the ribosome sterically hinders aggregation (Matsuura and Pluckthun 2003). This phenomenon has also been observed in the expression of the ligand-binding domain of the mammalian Nogo receptor (Schimmele et al. 2005).
One concern with ribosome display is that the ribosome or mRNA, rather than the expressed protein, might have some intrinsic affinity for the target molecule, which would complicate the selection process. Functional RNA molecules, or RNA aptam-ers, can actually be evolved through a similar evolutionary approach. In fact, this method, called SELEX (systematic evolution of ligands by exponential enrichment), predated ribosome display (Tuerk and Gold 1990). It is also possible that the ribo-some or mRNA may sterically hinder binding of the displayed protein to the target.
Another potential concern is that RNase contamination can degrade the mRNA and reduce the number of clones that can be recovered. RNase-free pipette tips, reagents, and vessels should be used whenever possible, and gloves must be worn at all times. No matter what types of precautions are taken, however, there will always be a low level of RNase activity present in the extract, as all cells naturally produce nucleases. In order to minimize this problem in a prokaryotic system, the extract should be made using a strain of E. coli that is RNase-deficient, such as MRE600 (Wade and Robinson 1966). Additionally, stem-loop secondary structure at the 5' and 3' ends of the mRNA increases resistance to exonucleases (Hanes and Pluckthun
1997). In a eukaryotic system, the rabbit reticulocyte lysate or wheat germ extract must be treated with RNase inhibitors. Regardless of the translation system used, however, all selection, washing, and elution steps should be performed in a cold room to minimize both RNase and protease activity. The only way to completely prevent RNase activity is to avoid using cell extracts.
The PURE system has been used to effectively eliminate the presence of RNases. Also, in the PURE system, release factors may be omitted from the translation mix to enhance the stability of the mRNA-ribosome-protein complex, enabling selections at temperatures up to 50°C (Matsuura et al. 2007). Traditional ribosome display has more limited uses, as it requires low temperatures (~4°C); however, all forms of ribosome display require high Mg2+ concentrations and a relatively gentle biochemical environment free of denaturants. The stability of the ribosomal complex is always a concern, partially because the genotype-phenotype linkage is noncovalent.
mRNA Display Technical Description mRNA display links genotype (mRNA/cDNA) and phenotype (protein) covalently through a DNA-puromycin linker. The protocol for generating these mRNA-protein selection particles begins in much the same way as the ribosome display protocol, except that the 3' end of the mRNA template must be covalently linked to a single-stranded DNA (ssDNA) linker bonded to puromycin (Figure 3.2B). The covalent linkage between the mRNA and ssDNA linker is generally accomplished through template-directed ligation, a process in which a DNA oligonucleotide complementary to the 3' end of the mRNA and the 5' end of the ssDNA linker is used as a splint to direct ligation between them. Puromycin is an antibiotic that structurally mimics the aminoacyl moiety of tRNA. It acts as a translation inhibitor by entering the ribosomal A site and forming a stable amide linkage to the polypeptide, causing it to release from the ribosome. In the context of mRNA display, puromycin allows the modified template mRNA to become covalently attached to the translated protein. The purpose of the DNA linker, which contains a stretch of ~21-27 adenine bases, is to pause translation long enough to allow puromycin to work (Keefe 2001). Without this linker, the yield of mRNA-protein fusions is drastically reduced (Nemoto et al. 1997).
After translation, a dilution buffer is added and the mRNA-protein particles are purified. A two-step purification process, which may not be necessary for all applications, is required to yield particles that contain both mRNA and a complete protein translated in-frame. First, the stopped translation is incubated with immobilized DNA oligonucleotides containing a stretch of thymine bases (T), which are complementary to the stretch of adenines (A) in the DNA linker. All mRNA present in the translation mixture is bound in this step: mRNA-protein fusions, free mRNA template, and any mRNA containing a poly(A) sequence that may have been present in the cell extract used for translation. Unbound and nonspecifically bound material is washed away, and the remaining species are eluted from the solid support. Next, mRNA-protein fusions are separated from free mRNA using a functionalized solid support that is able to bind to the expressed protein. The expressed protein may be engineered to contain an epitope tag [FLAG or (His)6, for example] to facilitate this process. A C-terminal tag is preferable because only mRNA-protein fusions containing fully translated, in-frame proteins can bind. Free mRNA and mRNA-protein fusions containing frameshifts (deletions or insertions) are washed away. If for some reason it is detrimental to protein function to introduce a C-terminal tag, an N-terminal tag may be used, although most frame-shifted proteins will not be eliminated this way. In the absence of tags, disulfide bonding between immobilized sulfhydryl groups and cysteines present in the expressed protein can be used (Roberts and Szostak 1997).
Once the desired mRNA-protein fusions have been adequately purified, most protocols call for reverse transcription to form an mRNA/cDNA duplex. This eliminates any mRNA secondary structures that might interfere with selection and provides a cDNA template for PCR upon recovery of genotype. The mRNA-protein fusions are then ready to be used in selections. Selections are performed at 4°C, as in ribo-some display, although the buffer composition is more flexible, for example, Mg2+ need not be present, and strong denaturants are permissible. mRNA-protein fusions are allowed to bind to an immobilized target and, after washing, the bound mRNA-protein fusions are eluted by adding proteinase K or, if the affinity of the interaction is low enough, an excess of soluble target molecule. The cDNA of the mRNA/cDNA duplex is then purified and amplified by PCR. At this point, as in all in vitro techniques, the dsDNA may be used for further rounds of selection. Detailed protocols for mRNA display are available (Keefe 2001).
History, Advances, and Applications
An mRNA-protein fusion (or "in vitro virus") was first reported by Nemoto et al. (1997). It was Roberts and Szostak, however, who carried out a proof-of-principle experiment for peptide evolution by mRNA display (1997). In this experiment, mRNA-peptide fusions containing a myc epitope were enriched 20- to 40-fold relative to mRNA-pep-tide fusions containing a random sequence.
The first protein scaffold to be used for directed evolution with mRNA display was the tenth fibronectin type III domain (10Fn3) (Xu et al. 2002). This scaffold has an immunoglobulin-like fold with three solvent-accessible loops, which are structurally analogous to the complementarity-determining regions of antibodies. It was chosen for its thermostability, solubility, high expression level in E. coli, and lack of cysteines (which can complicate folding). In the first reported instance of directed evolution on this scaffold using mRNA display, the loops were completely randomized, and variants binding to TNF-a with high affinity were isolated (Xu et al. 2002). More recently, 10Fn3 was used to evolve dual-specificity antagonists to human and mouse vascular endothelial growth factor (VEGF) receptor-2 (Getmanova et al. 2006). This latter study was the first to report the biological activity (inhibition of VEGF-dependent proliferation) of binding molecules developed on the 10Fn3 scaffold.
The second protein scaffold to be used in conjunction with mRNA display was the human retinoid-X-receptor (hRXRa) (Cho and Szostak 2006). This scaffold was chosen because it is small, folds stably, and has two recognition loops in close proximity. The loops were randomized, and mRNA display was used to isolate variants that specifically recognized adenosine triphosphate (ATP) (Cho and Szostak 2006). Additionally, scFvs have been affinity-matured successfully by mRNA display (Fukuda et al. 2006). In the first model system reported, error-prone PCR and StEP were used to diversify an anti-fluorescein scFv. Affinity was improved ~30-fold after four rounds of off-rate selection (Fukuda et al. 2006).
mRNA display has also been used to construct peptide libraries containing one or more unnatural amino acids (Li et al. 2002; Muranaka et al. 2006). In one report, four unnatural amino acids were introduced using three tRNAs containing different four-base anticodons and one tRNA containing an amber anticodon. From the mRNA-displayed library, a novel streptavidin-binding unnatural peptide was evolved, demonstrating the utility of this approach. Another method for generating peptides with unnatural amino acids is to reassign sense codons (Josephson et al. 2005). This requires more control over the tRNA species in the translation mix, so a well-defined chemical composition (e.g., using the PURE system) may be necessary.
Other applications of mRNA display include the study of protein-protein interactions (Huang and Liu 2007) and protein-DNA interactions (Tateyama et al. 2006). One study used a random peptide library and, in parallel, a natural library derived from human tissues to select peptides that bound to Ca2+/calmodulin (Huang and Liu 2007). The purpose of using two different libraries in this study was to allow for comparisons between the selected peptides from each library. Indeed, it was found that the Ca2+/calmodulin-binding peptides selected from the random peptide library correlated well with peptides selected from the natural library, which validated the results and demonstrated the usefulness of mRNA display for studying protein-protein interactions.
The main advantage of mRNA display compared to ribosome display is that the covalent linkage between genotype and phenotype can withstand harsh biochemical treatments that would cause dissociation of the stalled ribosomal complex. For example, mRNA display has been used to evolve proteins that can bind ATP in the presence of 3 M guanidine hydrochloride, a strong denaturant (Chaput and Szostak 2004). Another notable advantage of mRNA display is that the size of the selection particle is minimized (i.e., there is no ribosome or bulky fusion protein that might have some unanticipated interaction with the target molecule). As far as library size, mRNA display is among the best of in vitro methods, similar to ribosome display and certain forms of covalent DNA display. The number of mRNA-protein fusions that can be generated is limited only by translation volume, translation efficiency, and the efficiency of puromycin bonding to the nascent polypeptide. As such, libraries containing greater than 1014 molecules have been made (Leemhuis et al. 2005).
As in ribosome display, care must be taken to avoid RNase contamination. While mRNA-protein fusions are generally stable at 4°C, they are not stable at room temperature in the presence of RNases. In contrast, ssDNA and dsDNA fusions are much more robust (Kurz et al. 2001). One disadvantage of mRNA display compared to ribosome display is that the "genotype" portion of the selection particle cannot be recovered simply by removing Mg2+ from the buffer. This may be problematic when trying to recover the genotype of very tight binders by competitive elution.
In these cases, proteinase K digestion might be necessary, which is less specific. However, there is an alternative method that uses a photocleavable 2-nitrobenzyl linker between the mRNA and protein (Doi et al. 2007). In this system, the desired nucleic acid sequences may be recovered by brief (15 minute) UV irradiation.
As suggested by its name, covalent DNA display tethers a displayed protein to its encoding cDNA by a covalent linkage. There exist multiple strategies for accomplishing this feat, as described in this section and as shown in Figure 3.3. Because these strategies are all relatively new and not widespread, the technical descriptions, applications, and unique advantages and disadvantages are described separately for each technique. Finally, common advantages and disadvantages among all covalent DNA display methods are summarized.
One version of covalent DNA display is actually an adaptation of mRNA display (Tabuchi et al. 2001). In this protocol, the 3' end of unmodified template mRNA is hybridized to a DNA primer/linker carrying puromycin at its 3' end. During translation, the ribosome stalls at the mRNA/DNA duplex and the puromycin forms an amide linkage with the polypeptide chain. Reverse transcription then yields a stable cDNA-protein fusion (Figure 3.3A, top), which is used for selections. This strategy eliminates the need for modifying the mRNA template.
Alternatively, a photo-crosslinking method may be used (Kurz et al. 2001). In one version of this protocol, a modified mRNA template (mRNA-ssDNA-puromycin; see the section titled "mRNA Display, Technical Description" in this chapter) is translated to create an mRNA-protein fusion. Then, a DNA primer containing psoralen, a molecule that reacts to UV light, is hybridized over the junction between mRNA and ssDNA. UV irradiation allows the DNA primer to become covalently attached to the ssDNA portion of the modified mRNA template. Next, the mRNA is reverse transcribed to make cDNA, and RNase H treatment is used to degrade the mRNA. Finally, a DNA primer complementary to the 3' end of the cDNA is hybridized and elongated to make dsDNA. The resulting DNA-protein fusion (Figure 3.3A, middle) is used for subsequent selections. In another version of the photo-crosslinking protocol, a DNA primer containing a branched phosphoramidite linked to puromycin is hybridized to the 3' end of an unmodified mRNA template and covalently attached through psoralen. After in vitro translation and appropriate purification measures, reverse transcription, RNase H treatment, and second-strand synthesis are performed to create the final selection particle (Figure 3.3A, bottom).
Both ssDNA- and dsDNA-protein fusions, as described previously, offer valid alternatives to mRNA display, although their use has not been widely reported. It has been proposed that such fusions might be valuable for applications in which RNase activity cannot be effectively controlled (Kurz et al. 2001). Theoretically, any protein that can be displayed by mRNA display can be displayed by the previously mentioned methods. The two major disadvantages of mRNA display-based covalent
Selection Particles for Covalent DNA Display
A. mRNA display-based mRNA ^ Puromycin ^ \
DNA linker " t
Protein of interest (Tabuchi et al. 2001)
Was this article helpful?