|Home | About | Journals | Submit | Contact Us | Français|
A build/couple/pair (B/C/P) strategy was employed to generate a library of 7936 stereochemically diverse 12-membered macrolactams. All 8 stereoisomers of a common linear amine precursor were elaborated to form the corresponding 8 stereoisomers of two regioisomeric macrocyclic scaffolds via head-to-tail cyclization. Subsequently, these 16 scaffolds were further diversified via capping of two amine functionalities on SynPhase Lanterns. Reagents used for solid-phase diversification were selected using a sparse matrix design strategy with the aim of maximizing coverage of chemical space while adhering to a preset range of physiochemical properties.
We have a long-standing interest in the design and synthesis of small-molecule libraries that combine the structural complexity of natural products and the efficiency of high-throughput synthesis.1,2,3,4 Recent evidence has shown the importance of structural complexity as measured by sp3 content and the presence of stereocenters with the successful transition from discovery to the clinic.5 As such, macrocycles are an important class of organic molecules showing a variety of impressive biological activities.6,7 There are numerous macrocycles currently in development or approved by the FDA for clinical use, including natural products (1, Figure 1)8 and their derivatives (2),9 as well as some designed and synthesized de novo (3).10 The large occurrence of macrocycles in nature and their impressive biological activities may be derived from the fact that they often exhibit both pre-organization as well as flexibility, both of which can aid binding to biological targets.11,12 Based off of these observations, we have utilized macrocyclic compounds in the past as a source of diversity and structural complexity within our screening collection. In this work, we present a continuation of this effort to build a small-molecule library that contains macrocyclic compounds but occupies a distinct area of chemical space.8–10
Utilizing a three phase build/couple/pair (B/C/P) synthetic strategy,13 we have previously reported a route to medium-sized rings and macrocycles (Figure 2) starting from 1,2-amino alcohol 4 and aldol-derived γ-amino acid 5.8,14 A key feature of the B/C/P strategy is the ability to generate the complete matrix of stereoisomers lending to the study of stereo/structure-activity relationships (SSAR) upon biological screening.8,15 Thus all eight stereoisomers of amine 6 were prepared and utilized in downstream pairing reactions including nucleophillic aromatic substitution (SNAr), Huisgen macrocyclization and ring-closing metathesis (RCM) to yield scaffolds 7–11. In the current study we have expanded this chemistry to include a pairing reaction which cyclizes the linear template from ‘head-to-tail’ (HtoT) leading to macrocycles 12 and 13. This pairing strategy increases the number of stereogenic centers contained within the ring as compared to the previously prepared scaffolds, as well as decreases the number of rotatable bonds present in the final compounds. As evident by a molecular shape-based diversity analysis derived from normalized principal moments of inertia (PMI) ratios1,16 macrolactams 12 and 13 occupy a distinct region of 3-dimensional space as compared to scaffolds 7–11 (Figure 3). Surprisingly, the HtoT scaffolds are much more rod-like than the RCM-derived macrolactams.
At the onset of this project, we envisioned two potential pairing strategies to effect HtoT macrocyclization: (1) an intramolecular SNAr reaction or (2) macrolactamization (Figure 4). Due to the overall success of the intramolecular SNAr reaction for the formation of the medium sized rings 7 and 8 in our initial study,8 we first set our sights on the application of this method to access macrolactam 12. The synthesis of the SNAr substrate begins with allyl carbamate formation accomplished by treatment of amine 6 with allyl chloroformate under Schotten-Baumann conditions (Scheme 1). This was followed by TBS removal with TBAF and methylation of the resultant secondary alcohol with methyl iodide. This synthetic sequence led to orthogonally protected diamine 16 in 54–80% yield. The choice of DMF as a solvent and the use of excess MeI was imperative to prevent intramolecular carbamate formation. The terminal PMB and Boc protecting groups were removed simultaneously upon treatment with TFA in the presence of anisole as a cation scavenger. Amidation of the resulting crude amino alcohol was then accomplished resulting in SNAr substrate 14 in yields of 56–70% for the two steps.
The key SNAr macrocyclization was first attempted utilizing conditions developed in our previous studies.8,11 Treatment of 14 with NaH in THF at 0 °C led solely to the formation of oxazolidinone 18, which could not be avoided even with careful exclusion of water and the use of KH. Heating alcohol 14 in the presence of excess CsF was also attempted, which showed no consumption of starting material by LC/MS at room temperature but ultimately led to decomposition after extended heating (> 1 day). The use of DBU in DMF17 was also unsuccessful even after extended heating. Finally, desired product 12 could be obtained using TBAF in THF/DMF (in the prescence of 4 Å molecular sieves)18 however the yields were generally low (25–45%) and highly variable. In light of the problems encountered promoting this transformation we decided to shift our efforts to a macrolactamization approach via intermediate 15 (Figure 4).
The synthesis of 15 (Scheme 2) commenced with PMB deprotection of intermediate 6 via treatment with DDQ in buffered CH2Cl2. The primary alcohol was subsequently subjected to an intermolecular SNAr reaction with aryl fluoride 20, in the presence of TBAF and 4 Å molecular sieves in THF/DMF resulting in ether 21 in good yield (55–75%). Bis-deprotection of the Boc and t-butyl ester groups provided crude amino acid 15 which was gratifyingly converted to macrocyle 12 in good yield (83%) upon treatment with BOP-Cl in the presence of DIEA.
In order to access the meta-alkoxynitro macrocycle (13), we next decided to investigate a modified synthetic route in which the intermolecular SNAr reaction was replaced with a Mitsunobu reaction19 to install the nitrobenzoic acid. As shown in Scheme 3, primary alcohol 19 was treated with phenol 22 in the presence of DIAD and PPh3 to give phenyl ether 23 in 75–85% yield. The Boc and t-butyl ester groups were removed with TFA and the resulting amino acid (24) cyclized in the presence of BOP-Cl to afford macrocycle 13 in 78% yield over two steps.
With established routes to both the para- and meta-alkoxynitro scaffolds, all eight isomers of macrocycles 12 and 13 were synthesized in 5-g quantities in preparation for solid-phase library production. The key macrocyclization reaction proceeded in high yield for all eight isomers of the linear amine scaffold (Table 1). The stereochemistry of the precursors 12 and 13, along with the regiochemistry of the aryl nitro substitutent, did not appear to influence the cyclization efficiency with combined yields for the double deprotection and lactamization ranging from 75–88%.
In order to load the scaffolds onto solid-phase for library production, SynPhase™ Lanterns20 functionalized with a monomethoxy Backbone Amide Linker (BAL)21,22,23 were selected. Use of the BAL linker allows for loading via an amine (in this case an aniline) without sacrificing a diversity site. Immobilization would be achieved via reductive alkylation, while subsequent N-capping with sulfonyl chlorides, acid chlorides and isocyanates would afford the corresponding sulfonamides, amides and ureas upon cleavage with TFA. Thus, chemoselective reduction of the nitro functionality was achieved with SnCl2 to afford anilines 25 and 26 (Scheme 4) for loading onto solid-support. Of note, reduction of the meta-alkoxynitroaryl group proved more sluggish than the para regioisomer requiring heating at 60 °C for full conversion. Nitro reduction proceeded smoothly for all isomers in yields ranging from 80–89%.
With ample quantities of all stereoisomers of anilines 25 and 26 in hand, a sparse matrix design strategy was implemented to select library members to be synthesized.24 A virtual library was constructed (for each scaffold) incorporating all possible building block combinations at R1 (aniline) and R2 (secondary amine) using a master list of reagents (R1 = sulfonyl chlorides, isocyanates and acid chlorides; R2 = sulfonyl chlorides, isocyanates, acids and aldehydes).25 This provided a virtual library of 7,845 compounds per scaffold. Physicochemical property filters were then applied to eliminate building block combinations that led to products with undesired physicochemical properties.26 The following property filters were applied to yield a total of 5363 compounds: MW ≤625, ALogP −1 to 5, H-bond acceptors and donors ≤10, rotatable bonds ≤10 and TPSA ≤140. In order to increase the percentage of ‘Lipinski compliant’ products, a ‘75/25’ rule was also implemented where 75% of all library members had MW <500. A total of 496 compounds per scaffold were selected from the remaining set using chemical similarity principles, maximizing diversity but retaining near neighbors for built-in SAR. The reagents selected for library production include 22 sulfonyl chlorides, 19 isocyanates and 26 acid chlorides for aniline capping (Chart 1) along with 2 sulfonyl chlorides, 3 isocyanates, 21 acids and 22 aldehydes for secondary amine capping (Chart 2). The same set of reagents was used for each stereo- and regioisomer, thereby maintaining the ability to generate SSAR for each building block combination. The property distribution for the selected products (7936 compounds total) is shown in Figure 5 and Table 2.
As shown in Scheme 6 scaffolds 25a–h and 26a–h were loaded onto the BAL functionalized SynPhase L-series Lanterns via treatment with an excess of NaBH3CN in 2% AcOH/DMF at 40 °C for 3 days. (Lanterns were equipped with radio frequency transponders to enable tracking and sorting of library members.) Scaffolds 25 and 26 contain two handles for the introduction of appendage diversity: the immobilized aniline and a protected secondary amine, both suitable for reaction with various electrophiles. The first diversity site, the aniline, was reacted with a total of 68 building blocks including isocyanates, sulfonyl chlorides and acid chlorides.27,28 The second diversity site was then revealed via removal of the Alloc group upon treatment with Pd(PPh3)4 in the presence of excess 1,3–dimethyl barbituric acid. This secondary amine was capped with 49 building blocks including isocyanates, sulfonyl chlorides, acids and aldehydes. This was followed by cleavage from the Lantern, which was achieved by treatment with a 1:1 solution of TFA:1,2-dichloroethane, to afford a total of 7936 products with an average yield of 8.5 μmol/Lantern. On average, yields were slightly higher for the para-aniline derived compounds than the meta-aniline compounds (9.1 μmol vs 8.0 μmol).29
All library products were analyzed by ultra-performance liquid chromatography, and compound purity was assessed by UV detection at 210 nm. An overview of library purity with respect to building blocks and stereochemistry is provided in Figure 6. The average purity of the library was 86% with 84% of the library being >75% pure. In general all building blocks performed well during library production other than a few which are worth discussion. Reagents used for capping diversity site 1 which resulted in lower purities include ethyl sulfonyl chloride (R1 = 2) and cyclopentyl isocyanate (R1 = 25), both of which affected products predominantly for the meta- cores across most diversity site 2 building blocks. Isoxazole acid chloride (R1 = 63) led to low purity final compounds for most of the 16 cores and diversity site 2 building blocks. Overall, building blocks used to diversify the secondary amine faired better in terms of final product purity, with only one building block (oxazole-4-carbaldehyde - R2 = 47) showing reduced purities across R1 building blocks.
A novel library of 7936 12-membered macrocycles was successfully prepared via a macrocyclization pairing reaction. 16 cores were synthesized from 8 isomers of linear amine scaffold 9 to establish “built-in” stereo-structure activity relationships (SSAR). Two regioisomers of aniline capping site were also included in our design to further increase skeletal diversity. This library was generated utilizing a traceless BAL linker to provide final compounds that contained less rotatable bonds and one less potential H-bond donor in an effort to provide more “drug like” compounds for our screening collection. A sparse matrix of building blocks was employed to further enhance the appendage diversity of our library, while including built-in structure activity relationships and honing in on a desirable range of physiochemical properties.
This work was funded in part by the NIGMS-sponsored Center of Excellence in Chemical Methodology and Library Development (Broad Institute CMLD; P50 GM069721), as well as the NIH Genomics Based Drug Discovery U54 grants Discovery Pipeline RL1CA133834 (administratively linked to NIH grants RL1HG004671, RL1GM084437, and UL1DE019585).