Published: 22 May 2024
Welcome to a world where molecules wield extraordinary power, capable of healing, rejuvenating, and transforming lives. Concealed within each life-changing drug lies a story of perseverance, dedication, and chemical innovation. Today, we invite you to embark on a journey, led by our experienced chemistry team at BioAscent, as we describe a recently adopted approach to the rapid optimisation of metal-catalysed reactions with the ambition of expediting the delivery of tomorrow's life-changing medicines.
Over the past three decades, metal-catalysed reactions have played a pivotal role in enabling the successful explorations of numerous chemotypes, leading to the discovery of countless candidate molecules and groundbreaking drugs. However, these processes occasionally present challenges, leaving vast reaction spaces to be explored due to the intricate interplay of numerous factors, such as precatalyst, ligand, base, solvent, and more. Drawing inspiration from the work of scientists at Boehringer Ingelheim,1,2 scientists at BioAscent have developed a ‘Pool and Split’ chemistry platform capable of the rapid optimisation of a variety of metal-catalysed coupling reactions.
The ‘Pool and Split’ approach to the combinatorial optimisation of metal-catalysed reactions has been described by several researchers over the years including Wieland and Breit3 in 2010 and Moran and coworkers4 in 2015. However, researchers at Boehringer Ingelheim have most recently demonstrated how the approach can be effectively employed to rapidly and efficiently explore expansive reaction spaces for C–N bond-forming processes, such as the copper-catalysed Ullmann reaction (Scheme 1).
Scheme 1. Synthesis of 3 as presented in the ‘Pool and Split’ paper by Boehringer Ingelheim.1
The optimisation process for the copper-catalysed system involves the use of 24 ligands, 4 copper sources, 3 bases, and 6 solvents in a combinatorial manner, enabling the exploration of 1728 reaction combinations within just 24 reaction tubes (Figure 1). Each reaction tube is equipped with 4 copper sources and 3 bases. The 24 ligands are categorised into 4 groups, denoted as ligand-groups 1 to 4 (LG1-LG4, Figure 1). All 6 ligands of LG1 are introduced into each of the 6 tubes in the first row, while all LG2 ligands are added to each tube in the second row, and so forth. Additionally, 6 distinct solvents are distributed across the 6 columns of the array before heating the reactions as required.
Following the initial reactions, LCMS analysis identifies the reaction tube with the most favourable conditions. The copper salts and ligands from the successful reaction tube can then be deconvoluted by carrying out a further 24 reactions in a subsequent step, employing the optimal solvent and the original mixture of 3 bases. The bases can be deconvoluted in a final step, revealing the optimal conditions.
Figure 1a. Representation of the ‘Pool and Split’ workflow presented in the Boehringer Ingelheim paper1 detailing the optimisation of the synthesis of compound 3 (Scheme 1).
Figure 1b. Representation of the ligand groups as presented in the Boehringer Ingelheim paper .1
Inspired by this approach the chemistry team at BioAscent has constructed a ‘Pool and Split’ platform which has been successfully applied to a number of templates to improve the yields of key project compounds. We have curated a series of catalyst kits capable of optimising reactions for C–N, C–C, C–S, and C–O metal-catalysed coupling reactions, as well as implementing informatics-based solutions to expedite analytical data collection and analysis. Below are a number of examples that demonstrate just how effective this new approach can be.
Scheme 2. Optimised Ullmann reaction to prepare compound 6.
The initial conditions used to prepare compound 6 (CuI, trans-1,2-diaminocyclohexane, K3PO4, DMF, 100 °C) resulted in a modest 15% yield. The ability of the ‘Pool and Split’ technique to cover an expansive reaction space demonstrated how seemingly small but synergistic changes to the ligand, base, and solvent can have a major impact on the final yield (88%).
Scheme 3. Optimised Suzuki reaction to prepare compound 9.
Attempts to prepare compound 9 with typically robust conditions such as Pd(dppf)Cl2, aq. K2CO3, 1,4-dioxane, 100 °C proved ineffective, yielding around 20% over the course of several repeated reactions. ‘Pool and Split’ showed that ‘go to’ catalytic systems containing ligands such as the Buchwald biaryl monophosphines were also ineffective but as shown above, [Pd(crotyl)Cl]2 along with TTBP.HBF4 was successful in delivering a 78% yield. Interestingly, even the seemingly subtle substitution of Pd(OAc)2 in place of [Pd(crotyl)Cl]2 was enough to completely kill the reaction.
Scheme 4. Optimised Ullmann reaction to prepare compound 12.
Literature methods for the preparation of compound 12 proved ineffective in our hands yielding < 6% over repeated reactions (CuI, trans-1,2-diaminomethylcyclohexane, K2CO3, DMF, 110 °C). ‘Pool and Split’ allowed us to rapidly home in on improved conditions yielding a much improved 63% yield (Scheme 4).
Scheme 5. Optimised carbon-sulfur cross-coupling reaction to prepare compound 15.
C–S bond-formation has typically proven to be a reliable reaction in our hands but the preparation of compound 15 proved less successful, yielding only 25% using the standard Pd2(dba)3, Xantphos combination. ‘Pool and Split’ revealed a number of less obvious combinations involving monophosphine ligands, but in the end, we found DPEPhos (91% yield) and DPPF (95% yield, Scheme 5) to be most effective.
Scheme 6. Optimised Buchwald-Hartwig reaction to prepare compound 18 using a mixture of Pd salts and the 6 ligands of LG1.
Attempts to prepare compound 18 employing a variety of literature precedented conditions for the coupling of secondary amines delivered no more than 32% yield. ‘Pool and Split’ was again able to unveil successful conditions and revealed the deleterious effect of commonly used bases such as NaOtBu (Figure 2.). Preparation of compound 18 was expedited by deploying the initial pool conditions to deliver a much improved 89% yield.
Figure 2. Pie charts displaying the results of the reaction optimisation to prepare compound 18.
Since carrying out the initial validation experiments, the ‘Pool and Split’ approach has continued to gain traction at BioAscent with the team showing the utility of the approach in an ever-expanding list of metal-catalysed reactions. Ongoing work includes advancements in the areas of informatics and miniaturisation of the technology.
To find out more about our services or to enquire about working with BioAscent, contact a member of the team.
References
(1) Steimbach, R. R.; Kollmus, P.; Santagostino, M. A Validated “Pool and Split” Approach to Screening and Optimization of Copper-Catalyzed C−N Cross-Coupling Reactions. J. Org. Chem. 2021, 86, 1528−1539.
(2) Fordham, J. M.; Kollmus, P.; Cavegn, M.; Schneider, R.; Santagostino, M. A “Pool and Split” Approach to the Optimization of Challenging Pd-Catalyzed C-N Cross-Coupling Reactions. J. Org. Chem. 2022, 87, 4400–4414.
(3) Wieland, J.; Breit, B. A Combinatorial Approach to the Identification of Self-Assembled Ligands for Rhodium-Catalysed Asymmetric Hydrogenation. Nat. Chem. 2010, 2, 832−837.
(4) Wolf, E.; Richmond, E.; Moran, J. Identifying Lead Hits in Catalyst Discovery by Screening and Deconvoluting Complex Mixtures of Catalyst Components. Chem. Sci. 2015, 6, 2501−2505.
About the author
Michael Kiczun, a Team Leader in the Medicinal Chemistry Department at BioAscent, brings over 20 years of experience working across multiple therapeutic areas for various organisations. These include Organon, Schering-Plough, MSD, Prosidion, and the University of Dundee Drug Discovery Unit.
This blog post reflects the collaborative efforts of a talented team. I'd like to express my gratitude to Mounir Al Masri, Steven Bennett, Kirsten Deytrikh, Dan Fletcher, Mhairi Matheson, and Michael Mathieson for their invaluable contributions.
