Diversity-oriented synthesis encoded by deoxyoligonucleotides

Abstract

Diversity-oriented synthesis (DOS) is a powerful strategy to prepare molecules with underrepresented features in commercial screening collections, resulting in the elucidation of novel biological mechanisms. In parallel to the development of DOS, DNA-encoded libraries (DELs) have emerged as an effective, efficient screening strategy to identify protein binders. Despite recent advancements in this field, most DEL syntheses are limited by the presence of sensitive DNA-based constructs. Here, we describe the design, synthesis, and validation experiments performed for a 3.7 million-member DEL, generated using diverse skeleton architectures with varying exit vectors and derived from DOS, to achieve structural diversity beyond what is possible by varying appendages alone. We also show screening results for three diverse protein targets. We will make this DEL available to the academic scientific community to increase access to novel structural features and accelerate early-phase drug discovery.

Fig. 1. Conceptual comparison of DNA-encoded library synthesis strategies.

Shapes with solid coloring represent multifunctional skeletons with defined exit vectors, and shapes with graduated coloring represent collections of building blocks with diverse pharmacophoric features. a A single fixed skeleton library wherein appendage diversity results from fixed exit vectors of the central skeleton. b A skeleton-free library constructed by direct linkage of diverse building blocks to each other. c A library comprising multiple skeletons bearing well-defined but variable exit vectors to collections of building blocks with diverse chemical features.

Fig. 2. Overview of skeletons selected for inclusion in the DOSEDO library.

Skeletons shown are indicative of the compounds loaded onto DNA. Isomer counts report the number of isomers (stereogenic and positional) for the indicated skeletons. Red spheres indicate two distinct isomers at each stereogenic center – in no cases were stereochemical mixtures or skeletons of unknown absolute configuration used in the library construction. Asterisks indicate skeleton series having alternate DNA attachment points.

Fig. 3. Summary of building block validations.

a Summary of amine-capping validation experiments, including a general synthetic scheme for the preparation and use of a proline-based model construct (Pro-DNA), and charts showing the performance of acylation, reductive alkylation, and sulfonylation. Orange points show %AUC (area-under-the curve) corresponding to desired product species; blue points show %AUC of unknown DNA species as assessed by MaxEnt1 deconvoluted mass spectra. “BB (building block) num” is a ranking of compounds in two phases: (i) decreasing %AUC product with increasing %AUC unknowns up to 10, then (ii) decreasing %AUC product with >10 %AUC unknowns. The highlighted gray region of each chart denotes building blocks that met our inclusion criteria of >70%AUC product with <10%AUC unknowns. b Effect of buffer on performance of Suzuki couplings comparing a subset of skeletons [with variably capped amines] bearing ArI (aryliodide) versus ArBr (arylbromide). The skeleton series, amine capping, and model Suzuki-coupling building blocks are shown, alongside a heat-map representation of the difference in %AUC product and unknown species present when analogous reactions were performed at pH 10 vs. pH 8. Also displayed is a stacked bar chart comparing the performance of ArBr and ArI skeletons, on aggregate, with respect to product formation (bars without borders) and unknown species (bars with black borders) in different buffers. c Summary of Suzuki-coupling building-block validations including the constructs and conditions used and a comparative analysis of performance for a model ArI and ArBr (with the chart labeled “Arylbromide (II)” making use of higher catalyst and boronic acid/ester loading). Additional reaction time is also noted. Orange points show %AUC corresponding to desired product species; blue points show %AUC of unknown DNA species as assessed by MaxEnt1 deconvoluted mass spectra. Source data are provided as a Source Data file.

Fig. 4. Assessment of reactivity in building block and skeleton validation experiments.

a Combined performance of ArI (aryliodide) and ArBr (arylbromide) Suzuki-coupling reactions, where each common boronic acid/ester moiety was replaced with a methyl group in silico, then visualized by Sammon mapping of a Morgan2 fingerprint distance matrix to two dimensions, coloring and sizing by preference for inclusion in DNA-encoded libraries (DELs) (4 (orange) = highest), lowest preference building blocks (BBs) are layered above higher preference BBs for visualization. b Performance of the same BBs represented in a specific to the ArI model system, with point size still indicating consensus preference category but colored by adjusted %AUC (area-under-the curve) product (%P), with orange = highest. c As for b showing data specific to the ArBr model system. d Comparison of amine-capping performance by Sammon mapping of individual building blocks (with no in silico treatment of common reactive moieties), sized by %AUC product, blue = aldehydes, green = carboxylic acids, red = sulfonylchlorides. e Heat-map representation of skeleton validation, showing data for five different amine-capping reactions followed by a single Suzuki coupling with (4-sulfamoylphenyl)boronic acid. Each row represents reactions of a specific skeleton. Column labeling indicates: i – adjusted %AUC for the desired amine-capped product; ii – adjusted %AUC of the desired Suzuki coupled product; iii – adjusted combined %AUC for the desired Suzuki coupled product as well as the Suzuki product of non-N-capped starting material. Black cells indicate either that the chromatographic data obtained was insufficient to resolve starting material and product peaks, or that no DNA species were present based on chromatography (potentially due to loss of the DNA pellet during ethanol precipitation or instrumentation error). f Synthetic plan for DOSEDO library production using a single ArI and ArBr skeleton to represent the wider collection, with numbers indicating how many building blocks were used for each step. Source data are provided as a Source Data file.

Fig. 5. Assessment of input sequences.

a The number of barcodes observed at a given number of counts for the input DOSEDO_v1 library, showing the relative distribution of Br- and I-derived sub-libraries (orange and purple, respectively). b As in a for DOSEDO_v2. c Unobserved sequences in DOSEDO_v2, with two sets of features highlighted in blue for being unobserved with unusual incidence as SAR (structure OR sequence activity relationship). Chart is graduated in blue in line with Cy2 identity to better perceive depth. Source data have been provided as a Source Data file.

Fig. 6. Analysis of validation screening exercises.

a Carbonic anhydrase IX (CAIX) screen output summarized by sum of lower bound (lb) calculated enrichments for all Cy2-Cy3 disynthons with all primary sulfonamide-containing building blocks included in the DOSEDO library shown (summed lower bound enrichments <25 was excluded from the plot for clarity). b Chemical space map generated through UMAP multidimensional scaling of a Tanimoto distance matrix derived from Morgan fingerprints of enumerated encoded compounds. A random sample of 50,000 compounds from the enumerated DOSEDO library was used as background chemical space, along with 44 selected compounds for resynthesis. These compounds are colored by their binding target. c Calculated enrichments of selected resynthesis target compounds from CAIX- and isocitrate dehydrogenase 1 (IDH1) R132H-validation screens. Color scheme as for b. d Structure of 3 alongside differential scanning fluorimetry (DSF) results with N-His-GST (glutathione S-transferase) tagged USP7 as well as His-tagged GST (5 µM) in 50 mM Tris–HCl, pH 7.4. Two peaks belonging to GST (T_m 54.6 °C) and ubiquitin specific peptidase 7 (USP7) (T_m 44.9 °C) were observed in the melt curves of GST-tagged USP7. No thermal stabilization of USP7 by 3 was observed by tracking the USP7 peak; GST stabilization by 3 was observed by tracking the GST peak. 3 induces thermal stabilization of His-tagged GST in a dose-dependent manner, up to 3.3 °C at 100 µM. N = 3 technical replicates. Source data are provided as a Source Data file.