In the proof of concept phase, we focused on a RNA-RNA complex that we previously characterized in great detail 262728. AlphaScreen^® was used to quantify the interaction between the trans-activating responsive (TAR) RNA element of HIV-1 29 and a RNA aptamer, R06 identified from a random library of oligonucleotides (Figure 2). These two oligoribonucleotides adopt a stem-loop structure, display complementary sequences in the apical loop (Figure 2C) and were demonstrated to give rise to loop-loop (also called kissing) interactions. Increasing concentrations (0-40 nM) of digoxigenin-coupled R06 (dig-R06) were incubated in the presence of increasing concentrations of biotin-coupled TAR (biot-TAR), and constant amounts of streptavidin-D and anti-Dig-A beads. Such titration experiments were carried out at various biot-TAR concentrations ranging from 0 to 40 nM (Figure 2B). This revealed typical bell-shaped AlphaScreen^® signals showing a maximum for 40 nM biot-TAR and 10 nM dig-R06. Indeed as the anti-Dig-A-bead amount remains constant, addition of dig-R06 reaches a point at which it is no longer captured; consequently the excess of free dig-R06 competes with immobilized R06 for interacting with TAR captured on D-beads. This results in a decrease in fluorescence signal.

To demonstrate the specificity of the interaction, a competition assay was carried out in which increasing concentrations of unconjugated (free) R06 (fR06) were incubated in the presence of 10 nM dig-R06 and 40 nM biot-TAR (Figure 3A). As expected AlphaScreen^® signal decrease correlated with increasing concentrations of fR06 added to the reaction, leading to an apparent IC50 of 16.7 nM ± 1.7 (Figure 3B), a value reflecting an affinity in the same order of magnitude as that calculated using surface plasmon resonance (SPR) 2628.

Using thermal denaturation which was monitored by UV absorption spectroscopy, SPR and gel-shift assays, we previously showed that magnesium ions stabilized the TAR-R06 complex 26. In addition, we also used the E. coli protein ROP, which is involved in the control of the ColE1 plasmid copy number and specifically recognizes loop-loop complexes. We previously demonstrated that ROP was able to bind to the TAR-R06 complex 3031. In the AlphaScreen^® assay described above, increasing concentrations of ROP were added to biot-TAR-dig-R06 complexes in the absence or in the presence of 3 mM MgCl₂. The addition of 1 mM ROP resulted in increased Alphascreen^® signal indicating the formation of a highly stable ROP-R06-TAR kissing complex (Figure 3C). A ~10 fold higher AlphaScreen^® signal was observed for this ternary complex in the presence of 3 mM MgCl₂ compared to no magnesium (not shown). These experiments demonstrate that this methodology provides signals correlated to the affinity of the complex. Indeed, no signal was detected for R06 variants that do not complex with TAR, whereas conditions known to increase the interaction (addition of magnesium or ROP protein) resulted in increased fluorescence signals. Collectively these results demonstrate that AlphaScreen^® is suitable for monitoring the interaction between an aptamer and its target.

Our objective was then to adapt this approach to the screening of large pools of SELEX-derived sequences. To this end, it was necessary to capture every selected candidate on the A beads. The above methodology in which biotinylated candidates are used would no longer be easily adapted to such a goal. We rather considered the use of a biotinylated anchor complementary to a pre-determined sequence on the candidates. Indeed every candidate contains fixed sequences flanking the random region, used in the SELEX process for the amplification of the selected candidates. An oligomer complementary to one of the flank efficiently allows for the capture of every candidate, at least those for which this region is not involved in a strong intra-molecular interaction. In order to validate this anchor-based approach a biotinylated oligod(T₃CT₂₁) (biot-dT) (Figure 2C) was assayed in the presence of 10 nM dig-R06 and 40 nM TAR bearing an oligor(A₂₁GA₃) tail (rA-TAR) (Figure 2C). Increasing concentrations of biot-dT (Figure 4A) led to a maximal AlphaScreen^® signal for rA-TAR concentrations ranging from 40 to 80 nM, in accordance with a stoichiometric association between TAR and the biot-dT anchor and the respective Donor and Acceptor bead capture capacities under the current assay conditions. Therefore the anchor-based methodology allows for monitoring aptamer-target interactions.

We then used such a strategy for evaluating the evolution of populations derived from 7 rounds of RNA SELEX previously carried out in our laboratory against the domain II (DII) of the Hepatitis C Virus mRNA internal ribosome entry site. This domain that folds as a hairpin with a 7 nt loop, was used as the target. In vitro selection against DII was previously demonstrated to produce high affinity aptamers 32. To monitor the evolution of the RNA pool binding properties, candidates from the library (T0) and from rounds one to seven (T1-T7) were trapped on the acceptor beads using a digoxygenin-conjugated oligonucleotide, dig-primer (Figure 2C) complementary to their common 5' end. AlphaScreen^® signals obtained with D-beads carrying a 19 nucleotide long hairpin corresponding to the top part of DII were detected from round 4 and further increased at subsequent rounds, thus indicating the progressive enrichment of the population in strong binders (Figure 4B) in agreement with band shift assays 32 and SPR experiments carried out with bimolecular complexes formed between the immobilized HCV target and the SELEX pools (Figure 4C). These results demonstrate that the AlphaScreen^®-based approach (HAPIscreen) could be undertaken for screening large collections of selected candidates using a unique anchor oligonucleotide.

We then used HAPIscreen for evaluating the outcome of a SELEX experiment carried out using a RNA library against pre-microRNAs (premiRs). PremiRs display more or less perfect hairpin structures that are matured into functional miRs 3334. Aptamers raised against premiRs might consequently modulate miR interaction with proteins involved in the maturation process and impact on their regulatory function. Indeed oligonucleotides complementary to premiRs were shown to modulate the activity of miRs maturing enzymes 3536. SELEX was performed against two mixtures M1 and M2 of three human premiRs each (a, b, c and x, y, z, respectively) as described in Materials and Methods. At the end of seven SELEX rounds, the selected candidates were cloned and produced. Using HAPIscreen, candidates were trapped on the acceptor beads using digoxygenin-conjugated oligod(T₃CT₂₁) (dig-dT) (Figure 2C) complementary to their identical 3' end and individually assayed against each individual biotin-tagged target. One hundred ninety-two clones derived from the 7th round-enriched populations against either M1 or M2 were screened in duplicate blindly against the mixture of the three baits a, b and c or against the premiR × alone, respectively. It should be pointed out that the second experiment aims at identifying the partner actually targeted by a given aptamer selected against the mixture; repeating this experiment with immobilized y or z would allow the complete assignment of aptamers and targets. These experiments were carried out in 384-well plates and led to the identification of hits within both RNA populations (Figures 5A, B). Candidates from each selection were picked according to their high AlphaScreen^® signal (Figures 5A, B, red and grey dots, respectively) and tested individually by Surface Plasmon Resonance (Figure 5C). In the latter experiments, biotinylated pre-miRs a, b, c and x were individually immobilized on their respective sensor chip flow cells on which candidates were injected. Nine out of 12 (SELEX to × from M2) and 7 out of 12 (SELEX to a, b or c from M1) displayed evaluated K_D values lower than or equal to 30 nM for unique-based or mixture-based target assays, respectively. There was a fair agreement between both SPR and the Alphascreen^® analyses even though the order of the signals was not rigorously the same with the two techniques (Figures 5A and 4C). We also tested the behaviour of 5 candidates that generated a low Alphascreen^® signal using SPR. This revealed that 4 out of the 5 candidates gave a weak or no resonance signal (not shown).

High affinity aptamers identified by Alphascreen^® were cloned and sequenced. As usual for in vitro selection we picked sequences displaying a high degree of similarity and identical motifs. Sequence differences likely account for the slight variation observed in Alphascreen^® or SPR signals. Finally, using MFold 37, aptamers from SELEX against M2 were predicted to adopt a consensus hairpin structure with a loop containing several nucleotides complementary to pre-miR loops. This suggested the formation of loop-loop aptamer/premiR complexes as previously described for TAR aptamers 26. These aptamers and the aptamer-premiRs complexes are presently being characterized and will be described in a forthcoming manuscript. Using a similar approach, aptamers derived from SELEX against M1 also showed sequences complementary to premiR target loop, allowing for the formation of 8 potential adjacent base pairs but were not predicted to fold as hairpins.

DNA primers and the biotinylated DNA anchor biot-dT, purchased either from Sigma or MWG Biotech, were purified by HPLC. All RNA targets and the digoxygenin 2'-O-methyl-LNA anchor (dig-primer) were chemically synthesized on an Expedite 8908 synthesizer (Applied Biosystems, USA) and purified by electrophoresis on denaturing 20% polyacrylamide, 7M urea gels. RNA candidates were synthesized by in vitro transcription using T7 RNA polymerase.

The AlphaScreen^® technology was used to assess the interaction between candidate oligonucleotides derived from SELEX experiments, and biotinylated target. Binding assays were performed using white 384-well Optiplates (Perkin Elmer) in a total volume of 25 μl. The AlphaScreen^® reagents (anti-dig-coated Acceptor beads and streptavidin-coated Donor beads) were obtained from PerkinElmer. biot-TAR and dig-R06 (see figure 2C for oligonucleotide sequences) were prepared in a 10 mM sodium phosphate buffer, pH 7.2 at 20°C, containing 140 mM potassium chloride, 20 mM sodium chloride and 3 mM magnesium chloride. Prior to the experiments the RNA samples were heated in this buffer at 95°C for 1 min and 30 s and cooled down on ice for 10 min. The protein ROP, purified as previously described 30, was prepared in this buffer with or without magnesium chloride. For the analysis of SELEX populations, denaturation and refolding of the candidate aptamers and targets prior to reaction with the anchor (biot-dT) was performed in water at 65°C for 3 min or 80°C for 1 min, respectively. After denaturation, candidate aptamer and target were quickly cooled down to 4°C for 3 min and then equilibrated at room temperature (RT) for 5 min before adding the selection buffer (20 mM sodium acetate, 140 mM potassium acetate, 3 mM magnesium acetate, 20 mM HEPES; pH 7.4). Equal volumes (5 μl) of each partner, candidate and target, were incubated for 45 min at room temperature (RT), at final concentrations of 0.2 μM and 0.625 μM, respectively. In parallel Acceptor beads (20 μg/ml) were incubated with dig-primer (0.625 μM) for 1 h at room temperature in the selection buffer. Then, 10 μl of each of the interacting partners were added to the plate, after 45-min incubation at RT, 5 μl of Donor beads at a 20 μg/ml concentration were added to the mixture. All manipulations involving AlphaScreen^® beads were performed under subdued lighting. The plates were allowed to incubate either 1 h or overnight in the dark at room temperature. Light signal was detected by using an EnVision^® multilabel plate reader from PerkinElmer.

The RNA library used for the selections was obtained by transcription of the DNA library (5'-GTGTGACCGACCGTGGTGC-N30-GCAGTGAAGGCTGGTAACC-3') as previously described 38. Two different primers P20 5'-GTGTGACCGACCGTGGTGC-3' and 3'SL containing the T7 transcription promoter (underlined) 5'-

TAATACGACTCACTATA

In order to generate candidates for high throughput screening, we set up a second automated workstation (Tecan Freedom EVO 200) equipped with a thermal cycler, an orbital shaker, a magnetic particle separation module and a vacuum separation module. Three hundred and eighty four clones (192 clones from either M1 or M2 populations) from round 7 were produced blindly on this second workstation. Candidates were directly amplified from colonies with a 5' end oligod(T₂₁CT₃) (underlined) lengthened P20 5'-

TTTTTTTTTTTTTTTTTTTTTCTTT

TAATACGACTCACTATA

SPR experiments were performed at 23°C with a BIAcore™ 3000 apparatus. The biotinylated premiRs were immobilised at 50 nM (300 to 400 RU) on SA (BIAcore, GE Heathcare Life Sciences; Sweden) or SAD200 m sensor chips (XanTech Bioanalytics; Germany) coated with streptavidin. Aptamers were injected at 500 nM in the SELEX buffer at a 20 μl/min flow rate. After each injection of the candidates, the target-surface was regenerated with a 1 min pulse of a mixture containing 40% formamide, 30 mM EDTA and 3.6 M urea prepared in milli-Q water. The sensorgrams were analysed with the BIAeval software 4.1 as previously described 30. Sensorgrams were double-referenced 39.