DNA-encoded library versus RNA-encoded library selection enables design of an oncogenic noncoding RNA inhibitor

Significance Drug discovery generally investigates one target at a time, in sharp contrast to living organisms, which mold ligands and targets by evolution of highly complex molecular interaction networks. We recapitulate this modality of discovery by encoding drug structures in DNA, allowing the entire DNA-encoded library to interact with thousands of RNA fold targets, and then decoding both drug and target by sequencing. This information serves as a filter to identify human RNAs aberrantly produced in cancer that are also binding partners of the discovered ligand, leading to a precision medicine candidate that selectively ablates an oncogenic noncoding RNA, reversing a disease-associated phenotype in cells.

TTCCCTAT ATAGGGAA Red = Unmatched S10     (2) was completed on the RNA 3D folds with the top 0.5% of Zobs score for each compound, and each nucleotide preference in the randomized region is reported as bits.
DiffLOGO analysis (3) was also completed on diastereomer pairs and are shown beneath the corresponding LOGOS. The secondary structures of the RNA 3D folds with the four highest Zobs scores are also shown.

Preparative HPLC. HPLC analyses were conducted on a system composed of Waters 2487
Dual Absorbance Detector, Waters 1525 Binary HPLC Pump and Waters Fraction Collector III.
The system was equipped with a Sunfire PREP C18 19 ´ 150 mm. Analyses were conducted with a flow rate of 5 mL/min with a gradient of 0-100% MeOH (+ 0.1% TFA) in water (+ 0.1% TFA) over 40 min followed by 5 min at 100% MeOH (+ 0.1% TFA).  Synthesis resin (amino-functionalized, 10 µm dia., 0.29 mmol/g, 300 mg, Rapp-Polymere) and the aforementioned 160-µm QC resin (30 mg) were transferred to a syringe (6 mL resin was prepared and characterized as previously described. (5) DNA-encoded library resin barcoding. All library synthesis and library handling was performed in a UV-free room. A general protocol for DNA-encoded solid-phase synthesis (DESPS) has previously been described. (5) Oligonucleotides are indicated in bold with the "≈" designation.

SOLID-PHASE DEL SYNTHESIS & SCREENING
Numeric identifiers were described previously. (5)  Hit amplification and preparation for NGS. Samples for NGS analysis were prepared as previously described. (5,6) The qPCR matrix was prepared containing Taq DNA polymerase NGS data processing. Sequence trimming, pattern matching, and UMI aggregation proceeded as previously described. (5,6) Sequences were ranked by UMI mean string distance and UMI count. Beads with mean string distance < 5 and UMI count < 10 were not considered. Compound replicates were calculated as the sum of remaining sequences having identical structure-encoding regions but distinct bead-specific barcodes. (7) Cheminformatic analysis. All in silico combinatorial library and hit cheminformatic analysis was performed using open-access software (DataWarrior v 4.7.2).(8) Hits were clustered by chemical similarity (T > 0.75).

S49
Synthesis of the hit compounds: Hit compounds emerging from the FACS screen were synthesized in parallel on Rink Amide Polystyrene resin (50 mg, 27.5 µmol). The resin was swollen in mL DMF for 10 min, filtered, and deprotected with 1 mL of 20% piperidine in DMF twice for 5 min each. The resin was washed (5 x DMF) followed by bromoacetylation with 1 mL of a cocktail containing 20% DIC and 80% 1.2M bromoacetic acid in DMF for 30 min. After washing (5 × DMF), bromide displacement was accomplished by treatment with 1 mL of 1 M propargylamine dissolved in DMF for 1 h, followed by additional washing (5 × DMF). Compounds were solubilized and purified by preparative HPLC as described above. Fractions containing the expected mass were pooled and evaporated to dryness under vacuum prior to full characterization.

S50
After HPLC purification 5. After 3 h, the slides were washed with Nanopure water twice and allowed to dry completely on the benchtop.
2DCS selection: The 3´3 ILL was 5'-end labeled with 32 P and purified as previously described. (9) The RNA library was folded in 1´ Binding Buffer (BB1; 8 mM Na2HPO4, pH 7.0, 185 mM NaCl, 1 mM EDTA) by heating at 60 °C for 10 min followed by cooling to room temperature on the bench top. All competitor oligos (C1-C8), each in an amount equivalent to the number of total compound delivered to the array surface, were folded separately in 1´ AB1 as described for 3´3 ILL. The folded oligos were mixed together with 5'-32 P labeled 3´3 ILL S53 followed by addition of MgCl2 (1 mM) and bovine serum albumin (BSA, 120 µg/mL) in a total volume of 600 µL. The array surface was preequilibrated with 1´ BB2 (1´ BB1 supplemented with 1 mM MgCl2 and 120 µg/mL BSA) for 5 min, after which the excess buffer was removed. The mixture of 3´3 ILL and competitor oligonucleotides was then applied to the surface, and the array was incubated for 20 min at room temperature. The glass slide was then washed with 1´ AB2 three times and dried for 1 h. The array was imaged by using Molecular Dynamics Typhoon variable mode phosphorimager.
Reverse transcription and PCR amplification to install barcodes to encode each compound were performed as previously described. (10) The DNA thus obtained was purified using native For competitive binding assays, both WT and mutant pri-miR-27a RNAs were transcribed as previously described (see Table S6 for sequences). (11) The RNA was purified by gel electrophoresis and folded in 1´ Binding Buffer at final concentration of 750 nM by heating at 60 °C for 5 min and then slowly cooling to room temperature. This RNA was aliquoted (10 µL) and diluted with an equal volume of 1´ Binding Buffer. To each pri-miR-27a sample was added 10 µL of folded Cy5-labeled 5'GAG/3'CCC loop displayed in pri-miR-27a's Drosha site (folded as described above; 5 nM final concentration). To each tube was then added 10 µL of 750 nM of 9 (final concentration of 100 nM), prepared in1´ Binding Buffer containing 0.05% (v/v) Tween-20.
The samples were incubated at room temperature for 20 min, followed by thermophoresis