Longbo
Li
a,
Bianca
Matsuo
a,
Guillaume
Levitre
a,
Edward J.
McClain‡
bd,
Eric A.
Voight
b,
Erika A.
Crane‡
*c and
Gary A.
Molander
*a
aDepartment of Chemistry, Roy and Diana Vagelos Laboratories, University of Pennsylvania, 231 South 34th Street, Philadelphia, Pennsylvania 19104-6323, USA. E-mail: gmolandr@sas.upenn.edu
bDrug Discovery Science & Technology, Discovery Research & Development, AbbVie, Inc., 1 North Waukegan Rd, North Chicago, Illinois 60064-1802, USA
cDrug Hunter, Inc., 13203 SE 172nd Ave, Suite 166 PMB 2019, Happy Valley, Oregon 97086, USA. E-mail: erika@drughunter.com
dDepartment of Chemistry, University of Wisconsin−Madison, Madison, Wisconsin 53706, USA
First published on 9th February 2023
DNA-encoded library (DEL) screens have significantly impacted new lead compound identification efforts within drug discovery. An advantage of DELs compared to traditional screening methods is that an exponentially broader chemical space can be effectively screened using only nmol quantities of billions of DNA-tagged, drug-like molecules. The synthesis of DELs containing diverse, sp3-rich spirocycles, an important class of molecules in drug discovery, has not been previously reported. Herein, we demonstrate the synthesis of complex and novel spirocyclic cores via an on-DNA, visible light-mediated intermolecular [2 + 2] cycloaddition of olefins with heterocycles, including indoles, azaindoles, benzofurans, and coumarins. The DNA-tagged exo-methylenecyclobutane substrates were prepared from easily accessible alkyl iodides and styrene derivatives. Broad reactivity with many other DNA-conjugated alkene substrates was observed, including unactivated and activated alkenes, and the process is tolerant of various heterocycles. The cycloaddition was successfully scaled from 10 to 100 nmol without diminished yield, indicative of this reaction's suitability for DNA-encoded library production. Evaluation of DNA compatibility with the developed reaction in a mock-library format showed that the DNA barcode was maintained with high fidelity, with <1% mutated sequences and >99% amplifiable DNA from quantitative polymerase chain reaction (PCR) and next generation sequencing (NGS).
DNA-encoded library (DEL) technology has emerged as a powerful tool for hit identification in modern drug discovery.13–21 DELs can be readily assembled to afford exceptionally large combinational libraries that can be efficiently screened in a cost-efficient manner, as only nanomolar quantities of both drug-like molecules and the biomolecular targets are required.21 Furthermore, prior successes in the identification of valuable hits for challenging therapeutic targets via DEL chemistry have attracted increasing attention in academia and industry.22,23 Solution-based chemical syntheses on-DNA must be carried out within strict reaction parameters, as the transformations must be tolerant of aqueous conditions and the basic DNA backbone.24,25 To this end, visible light-mediated transformations have attracted significant attention in DEL chemistry for the ability to generate novel C–C bonds26–41 while expanding chemical space under extremely mild conditions.42–44 For example, Kölmel and coworkers demonstrated that this mild reactivity paradigm could be utilized for the synthesis of strained cyclobutane scaffolds on-DNA via visible light-mediated [2 + 2] photocycloaddition between cinnamate substrates and styrene derivatives (Fig. 1b).41 Additional relevant transformations have been performed off-DNA, including a highly diastereoselective dearomative [2 + 2] photocycloaddition of indoles and diverse alkenes to afford cyclobutene-fused indolines, also requiring inert conditions.45 New heteroarene dearomative functionalization strategies have recently demonstrated great value, particularly those enabled by visible light, enhancing the molecular complexity and three-dimensionality of these venerable products.46 Additionally, BMS in collaboration with the Knowles laboratory reported a different [2 + 2] photocycloaddition to access polysubstituted (hetero)spiro[3.3]heptane products.12
A recent collaborative study from our groups focused on unlocking the radical reactivity of 1,3-disubstituted [1.1.1]bicyclopentyl (BCP) halides for on-DNA reactions.33 Enabled by visible-light photochemistry, the method allowed the preparation of diverse, sp3-rich chemical matter on-DNA. As part of our continued study of BCP reactivity, we sought to develop a method that utilized the exo-methylenecyclobutanol products derived from the strain-driven hydrolytic ring opening of DNA-conjugated BCP halides (Fig. 1c).47,48 With well-documented, mild, DNA-compatible reactivity, we envisioned that visible-light photochemistry could be leveraged to engage these and other olefins in a triplet-sensitized [2 + 2]-cycloaddition with indoles and other heterocycles, providing access to novel spirocyclic substructures on-DNA.12,45,49 As an extension of this chemistry, we also explored additional DNA-conjugated olefin substrates, including exo-methylenespiro[2.3]hexanes,50 unactivated alkenes, styrene derivatives, and acrylamides. The latter three have revealed significant reactivity toward indoles and other heterocycles to form fused ring systems.45 Finally, we focused on important and abundant scaffolds in drug discovery as the [2 + 2] coupling partners, including indoles, azaindoles, benzofurans, and coumarins, generating complex spiro- and fused polycyclic ring systems.
Entry | Deviations from standard conditions | Yield (%) |
---|---|---|
a Standard conditions: DNA (2 mM in H2O, 10 nmol, 1 equiv.), 1a (100 mM in DMSO, 100 equiv.), [Ir{dF(CF3)ppy}2(dtbbpy)]PF6 (2 mM in DMSO, 1 equiv.), glycerol (2 μL in 8 μL of DMSO), rt, 10 min, blue Kessil. Work-up: ethanol precipitation. n.d. = product not detected. rsm = recovered starting material. | ||
1 | None | >95 |
2 | No photocatalyst | n.d. (>95 rsm) |
3 | No light | n.d. (>95 rsm) |
4 | No glycerol | Degradation |
5 | 5 min instead of 10 min | 75 (25 rsm) |
6 | 25 equiv. of indole | 88 |
7 | 50 equiv. of indole | 93 |
8 | 200 equiv. of indole | 67 (19 rsm) |
9 | MeOH instead of DMSO | 15 (68 rsm) |
10 | DMA instead of DMSO | 44 |
11 | Ir(ppy)2(dtbbpy)]PF6 | n.d. (71 rsm) |
12 | Ir(ppy)3 | n.d. (70 rsm) |
13 | [Ir{dF(CF3)ppy}2(bpy)]PF6 | 83 |
14 | HP-A instead of HP-1 | n.d. (71 rsm) |
![]() |
To probe the regio- and stereoselectivity of the reaction on-DNA, off-DNA reactions were conducted between 1a and ethyl 3-methylenecyclobutane-1-carboxylate (see ESI Section 3.1† for details). Only one regioisomer was observed, which is consistent with previous off-DNA reports.45 However, four stereoisomers (two sets of diastereomeric enantiomers) were generated, in a 1.3:
1 ratio by chiral supercritical fluid chromatography (SFC). Despite the low stereoselectivity for this cycloaddition, the unprecedented novelty of these products covers uncharted chemical space, and methods to deconvolute DEL hits as stereoisomeric mixtures of comparable complexity are well established.52
We evaluated the scope of cycloaddition partners, reacting many diverse indoles and azaindoles with HP-1 (Fig. 2). Indoles with various functional groups at different positions about the aromatic ring resulted in good to excellent yields (2b–2i). Functional groups including halogens (2b–2c), a nitrile (2d), and CF3 (2e) were well tolerated with yields >80%. Functional handles at the 5-, 6-, 7-, and 8-positions of the indole ring including F and Cl substituents, as well as methyl and ester groups, also performed well under the standard conditions (2f–2i). Azaindoles and pyrrolo[2,3-d]pyrimidines with additional functional groups (Cl, Br, or ether substituents) gave good to excellent yields (2j–2p). Various electron-withdrawing groups at the 2-position of the indole were tolerated in this reaction. Indole-2-carboxamides provided moderate to good yields of cycloaddition products (2q, 2r). A 2-cyano-substituted indole provided a lower yield of 28% (as compared to the corresponding ester 2s). Even further, N-acetyl-protected indoles afforded moderate yields of products (2t, 2u).
Commercially available Boc-3-iodo-L-alanine methyl ester was used to prepare methylenecyclobutanol substrate HP-2 utilizing the same strategy as used with HP-1. Diverse indoles 3a–3c and azaindoles 3d–3f performed well in reactions with this headpiece, indicating that the cycloaddition is not limited to DNA-conjugated exo-methylenecyclobutanol HP-1.
We expanded this reaction further to afford dispiro[2.1.3.1]nonane ring systems. The DNA-conjugated exo-methylene spiro[2.3]hexane substrates were prepared using a nickel-catalyzed cyclopropanation of [1.1.1]propellane and readily accessible styrenes or acrylates.50HP-3 demonstrated good reactivity with indoles containing various groups at the 2-position, including an ester (4a), amide (4c), and nitrile (4d). Azaindole 4b and N-Ac indole 4e also worked well with HP-3. Related HP-4, prepared from the corresponding vinyl pyridine, performed well with indoles 5a–5e in >90% yield. Indoles with 2-substituted amides gave moderate to excellent yields (5f, 5g, and 5h). 2-Cyano and N-Ac indoles afforded products 5i and 5j in 70% and 90% yields, respectively. HP-5 was prepared from commercially available 3-exo-methylenecyclobutanecarboxylic acid and performed well with various indoles (6a, 6c–6f) and azaindoles (6b). Even further, we successfully scaled this reaction from 10 nmol (6b) to 100 nmol (6b′) with no loss in yield, revealing excellent promise for subsequent functionalization and implementation in DNA-encoded library production.
The reactivity of unactivated alkene headpieces was also investigated (Fig. 2). For HP-6, 50 equivalents of indole gave higher yields (e.g., 7a, 79% yield) than 100 equivalents (e.g., 7a′, 46% yield + 39% starting material) after irradiation for 10 min. This is likely caused by the lower reactivity of HP-6 compared to exo-methylenecyclobutane headpieces HP-1–HP-4. The indole at a higher concentration competes with DNA-conjugated alkenes and reacts with the excited triplet 1,2-diradical indole species to form a dimer, which leads to a lower yield of product in the same reaction time. Similar products have been observed in analogous off-DNA [2 + 2] photocycloaddition studies.45 Various indoles (7a–7d) and azaindoles (7e and 7f) performed well under this set of conditions. We expect high regioselectivity of 7a–7f as indicated by a similar reaction off-DNA.45 Another headpiece, HP-7, was also investigated with indoles (8a–8d) and azaindoles (8e and 8f) (Fig. 2). The complex polycycles obtained are unique, with no reported precedent in the literature.
Activated alkenes, including styrene derivatives HP-8–HP-15 and acrylamides HP-16 and HP-17, reacted smoothly with azaindole 1k with yields >50% (Fig. 3). The regioselectivity of a similar product off-DNA (indole 1a and styrene) has been reported in the literature.45para-(HP-8), meta-(HP-9), and ortho-substituted styrenes (HP-10) performed well with bromoindole 1k in 77%, 91%, and >95% yields, respectively. Chloro-substituted HP-11, primed for post-functionalization, provided a 57% yield of product. Fluoride-substituted HP-12 reacted well in 78% yield, and vinyl pyridine headpieces (HP-13–HP-15) provided polycyclic products in yields >70%. Acrylamide HP-17 showed higher reactivity than methyl acrylamide HP-16, providing the desired cycloaddition products in 89% and 63% yields, respectively. We explored the reactivity of other indoles toward the vinyl pyridine-derived substrates further. Vinyl pyridine HP-13 performed well with Boc-protected indole-2-carboxamides (9k and 9l) and indole-2-carbonitrile (9m), as well as N-acylindole-2-carboxy ester (9n), with yields >50%. Acrylamide HP-17 reacted with indole-2-carbonitrile to afford 9o in a moderate 42% yield.
Given the prevalence of diverse heterocyclic compounds in the medicinal chemistry field, additional heterocyclic coupling partners were also explored, including benzofuran and coumarin derivatives. Methyl benzofuran-2-carboxylate (1y) reacted smoothly with olefin headpieces 10a–10h to afford products in good yield (Fig. 4). The off-DNA reaction of benzofuran and exo-methylenecyclobutanol proceeds in a ∼1:
1 dr and ∼3
:
1 regioisomeric ratio (see ESI Section 3.2† for details). This regioisomeric ratio is consistent with what has been reported off-DNA.45 A variety of benzofurans were reacted with HP-5. The 2-carboxamide afforded 10i in a moderate 46% yield and, surprisingly, 3-methylbenzofuran-2-carboxylate afforded 10j in a good 70% yield of the cycloaddition product. This substrate showed good reactivity with other DNA-conjugated olefin substrates, including exo-methylenecyclobutanes HP-1 (10k) and HP-3 (10l), as well as vinyl pyridine HP-13 (10m) to give >60% yield in each case. Benzofurans with a Cl and/or methyl ether substitution both proceeded in low yields (see ESI Section 7, Fig. S4†), as did the corresponding 2-nitrile substrate.
We investigated the reactivities of coumarin derivatives with different DNA-conjugated alkenes, including exo-methylenecyclobutanes, unactivated alkenes, styrene derivatives, and acrylamides (Fig. 4). Visible light-mediated intermolecular [2 + 2] cycloadditions of coumarin-3-carboxylates and acrylamide analogs off-DNA have been previously reported.53 We conducted these analogous off-DNA reactions to confirm the regio- and diastereoselectivity of the [2 + 2] cycloaddition products (see ESI Section 3.3† for details). Initial on-DNA experiments revealed that a small percentage of lactone hydrolysis product was observed under standard conditions, along with the expected [2 + 2] cycloaddition adduct as the major product. We found that the [2 + 2] product could be funneled to hydrolyzed product in aqueous solution by heating at 70 °C for 20 min. Overall, 7-methylcoumarin (1z) showed good reactivity with methylenecyclobutanes HP-1, HP-3, HP-4, and HP-5, affording spiro[3.3]heptanes 11a–11d in good to excellent yields. Unactivated alkene substrate HP-6 resulted in a low 27% yield (11e), while HP-7 did not afford any desired product. For the activated alkene substrates, vinyl pyridine HP-13 and acrylamide HP-17 reacted to provide 11g and 11h in 73% and 38% yields, respectively. The reactivity of coumarin derivatives containing an ester, ketone, or amide was evaluated with exo-methylenecyclobutane HP-5. For coumarin-containing esters and ketones, hydrolysis followed by decarboxylation yielded spiro[3.3]heptane products 11j–11n in >45% yield. For a coumarin-containing amide, only the hydrolysis product without decarboxylation was observed, affording 11o in 94% yield. Furthermore, unactivated alkenes HP-6 and HP-7 reacted to provide 11p and 11q in good yields. Control experiments with a headpiece that lacked an alkene functional group were conducted under the standard reaction conditions with benzofuran and coumarin (HP-A, see Table 1 and ESI Section 6† for details). We observed benzofuran and coumarin adducts in 10% and 7% yields, respectively, which indicates the potential addition of benzofuran and coumarin to the DNA tag. Therefore, actual yields for 10a–10m and 11a–11n are slightly lower than reported.
To ensure that this method would be compatible with DNA-encoded library production, exo-methylenecyclobutane substrate HP-5 was prepared on an elongated DNA tag and then subjected to the reaction with indole 1a or other control conditions (Table 2). Upon isolation of the single constructs by ethanol precipitation and filtration, a series of ligations were performed, mimicking the library production process, followed by quantitative polymerase chain reaction (qPCR) and next-generation sequencing (NGS) of the full-length tag (see ESI Section 8† for details). The ligations for all six samples proceeded with good efficiencies (80–100%), as confirmed by gel electrophoresis. Amplification efficiency of the full-length sequence by qPCR was comparable for all six samples (93–98%) and the reaction maintained 99% amplifiable DNA, as compared to the no light, no reagents control (entry 6, Table 2). In contrast, some of the most often used non-photonic on-DNA chemistries often have only 30–50% amplifiable DNA remaining.54 Finally, NGS analysis revealed that all samples as compared to the control had <1% mutated sequences.
Entry | Conditions | Amplifiable DNA (%) | Mutated sequences (%) |
---|---|---|---|
1 | Standard reaction conditions | 99 | 0.5 |
2 | Standard conditions under N2 | 100 | 0.1 |
3 | Reagents with no irradiation control | 100 | 0.2 |
4 | Irradiation with no reagents control | 99 | 0.4 |
5 | Standard conditions, no BCP-I | 100 | 0.2 |
6 | No irradiation, no reagents control | 100 | 0 |
Footnotes |
† Electronic supplementary information (ESI) available: Experimental procedures and characterization. CCDC 2215502. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d3sc00144j |
‡ EAC and EJM are former employees of AbbVie. |
This journal is © The Royal Society of Chemistry 2023 |