In vitro selection using a dual RNA library that allows primerless selection

Abstract

High affinity target-binding aptamers are identified from random oligonucleotide libraries by an in vitro selection process called Systematic Evolution of Ligands by EXponential enrichment (SELEX). Since the SELEX process includes a PCR amplification step the randomized region of the oligonucleotide libraries need to be flanked by two fixed primer binding sequences. These primer binding sites are often difficult to truncate because they may be necessary to maintain the structure of the aptamer or may even be part of the target binding motif. We designed a novel type of RNA library that carries fixed sequences which constrain the oligonucleotides into a partly double-stranded structure, thereby minimizing the risk that the primer binding sequences become part of the target-binding motif. Moreover, the specific design of the library including the use of tandem RNA Polymerase promoters allows the selection of oligonucleotides without any primer binding sequences. The library was used to select aptamers to the mirror-image peptide of ghrelin. Ghrelin is a potent stimulator of growth-hormone release and food intake. After selection, the identified aptamer sequences were directly synthesized in their mirror-image configuration. The final 44 nt-Spiegelmer, named NOX-B11-3, blocks ghrelin action in a cell culture assay displaying an IC50 of 4.5 nM at 37 degrees C.

Figure 1

Design of the shared PCR product that leads to the short (A) and full-length (B) RNA libraries. The PCR product serves as a shared template for generating both the short and the full-length library. Tandem promoter sequences for the T3 and T7 RNA polymerase, respectively, are located at the 5′ end of the forward strand (F) and are part of the forward primer at the same time. The integration of the reverse primer that contains two ribonucleotides, introduces the (necessary) cleavage sites for alkaline hydrolysis of the template for synthesis of the short library. (A) To employ the short RNA library for the selection step, the reverse strand of the PCR product is cleaved by alkaline hydrolysis at the positions of the ribonucleotides so that the main part of the reverse primer is removed from the transcription template. The use of the T7 RNA polymerase ensures that transcription starts directly after the T7 promoter to generate the short RNA library. An excess of guanosine monophosphate in the transcription reaction establishes a monophosphate at the 5′ end that is essential for the template-directed ligation of the forward primer after the selection step. (B) To employ the full-length RNA library for the selection step, the in vitro transcription is carried out utilizing T3 RNA polymerase. As a consequence, the library contains the full sequence of the T7 RNA promoter at the 5′ end. Both polymerases add to their respective transcripts partially non-templated single nucleotides (3′-microheterogeneities) so that the transcribed libraries result in N- and N+1 mixtures, indicated by an additional ‘N’. prom.: promoter; nt: nucleotides; fp-bs: forward primer binding site; rp-bs: reverse primer binding site.

Figure 2

The short (A) and full-length (B) library and the workflow for in vitro selection. Both variants of the dual RNA library differ in the absence [(A); short library] or presence [(B); full-length library] of the primer binding sites. The random region, depicted as a huge loop structure, is clamped by seven terminal base pairs. (A) In order to amplify the short library after the selection step, the primer binding sites are added by a template-directed ligation strategy. The two dangling nucleotides at the 5′ end improve the ligation yield; the 5′-terminal nucleotide is monophosphorylated. Subsequent to ligation the RNA is reverse transcribed and amplified by PCR. The PCR product is a shared template between both libraries and can be switched to (B). Ribonucleotides within the reverse primer permit the cleavage of one strand of the PCR product by alkaline hydrolysis. The following transcription with T7 RNA polymerase regenerates the (fully truncated) short library. (B) In order to amplify the full-length library, the isolated RNA sequences are reverse transcribed and amplified by PCR. As described under (A), the PCR product is a shared template that can be switched to (A). To transcribe the RNA for the next round of selection with the full-length RNA library, the dsDNA is transcribed using T3 RNA polymerase.

Figure 3

Template-directed ligation of the short RNA library. The cartoon (A and B) presents the structural designs of the short RNA library (N- and N+1-transcripts) and the double-stranded adapters which consist each of a ligate and a bridge. The randomized region (34 nt) is flanked by seven nucleotides that can form base pairs (see Figure 1), where upon the 5′ end carries two additional fixed nucleotides. Six nucleotides at both ends of the library serve as hybridization sites for the forward bridge and the two reverse bridges. The uridines (U) and the inosine (I) in the reverse bridge 1 (A) and 2 (B), respectively, introduce the alkaline fission sites into the primer binding site. The forward bridge, the reverse ligates 1 and 2 additionally contain a 3′-terminal 2′,3′-dideoxynucleotide (3′-H) to prevent mispriming during PCR. Up to 50% of all transcripts are N+1-transcripts (B) that contain a non-coded nucleotide at the 3′ end. In order to ligate primer binding sites to these molecules as well, the reverse adapter 2 was designed. It comprises the reverse ligate 2 i.e. 1 nt shorter at its 5′ end than ligate 1, and reverse bridge 2 that offers the universal base inosine (I) for hybridization to the non-coded nucleotide in the N+1 position of the transcript. The autoradiogram (C) exemplarily shows the results of the template-directed ligation. While ligation yield of the forward adapter to the 5′ end of the RNA molecules was acceptable (lane a), the ligation of the two reverse adapters 1 and 2 to the 3′ end of the library was with ∼33% quite insufficient (lane b). However, in the case that both forward and reverse adapters are present in equimolar amounts at the same time, the overall ligation yield was markedly improved (lane c). FA: forward adapter; RA-1: reverse adapter 1; RA-2: reverse adapter 2; forw: forward; rev: reverse; prom: promoter; nt: nucleotides.

Figure 4

Progress of manual selection rounds (10–18) using the short library. The histogram shows the progress of the manual in vitro selection carried out with the short library starting from round 10. The black circles indicate the peptide target concentration offered in the respective selection round. In each selection round equal amounts of radio-labelled RNA were incubated with the respective peptide concentrations and with pure selection matrix as a control in order to verify the specific enrichment of binding sequences versus unspecific binding. The fraction of RNA specifically binding to the target peptide (white columns) is compared to the fraction of RNA that unspecifically binds to the selection matrix (grey columns).

Figure 5

Ghrelin-binding sequences. After 16 selection rounds, 23 clones were sequenced. The sequences comprise the fixed nucleotides (nucleotides are underlined) and the former randomized region. Column (F) indicates the frequency of occurrence of each individual sequence. The obtained sequences were grouped into two families and compared to the already published ghrelin-binding sequence NOX-B11 (21). Three conserved motifs are indicated by the background colours (blue, red and yellow). All sequences comprise of a 25 nt motif (red), which was already described as a sequence involved in ghrelin-binding (21). Firstly, the new sequences were synthesized as aptamers and ranked in a competition assay for their ghrelin binding potency in comparison to aptamer NOX-B11. All clones showed similar or improved binding characteristics (data not shown). An assortment of clones were synthesized as Spiegelmers and their IC₅₀-values were determined in cell culture experiments. NOX-F1 turned out to be the strongest binder and this oligonucleotide was further truncated and modified with a hexaethylene glycol linker (L) to give the final sequence NOX-F1-trunc. n.t. = not tested.

Figure 6

Inhibition of ghrelin-mediated activation of GHS-R1a by Spiegelmers NOX-B11-3 (= NOX-F1-trunc) and NOX-B11. CHO cells expressing GHS-R1a were stimulated with 2 nM ghrelin at 37°C in the presence of the indicated concentrations of the respective Spiegelmers, and the resulting Ca²⁺-mobilization was determined. The ghrelin action is suppressed by both Spiegelmers in a dose-dependent manner, but the sequence of NOX-B11-3 shows an ∼5-fold better IC₅₀ than NOX-B11.