The Medicinal Chemistry in the Era of Machines and Automation: Recent Advances in Continuous Flow Technology

Abstract

Medicinal chemistry plays a fundamental and underlying role in chemical biology, pharmacology, and medicine to discover safe and efficacious drugs. Small molecule medicinal chemistry relies on iterative learning cycles composed of compound design, synthesis, testing, and data analysis to provide new chemical probes and lead compounds for novel and druggable targets. Using traditional approaches, the time from hypothesis to obtaining the results can be protracted, thus limiting the number of compounds that can be advanced into clinical studies. This challenge can be tackled with the recourse of enabling technologies that are showing great potential in improving the drug discovery process. In this Perspective, we highlight recent developments toward innovative medicinal chemistry strategies based on continuous flow systems coupled with automation and bioassays. After a discussion of the aims and concepts, we describe equipment and representative examples of automated flow systems and end-to-end prototypes realized to expedite medicinal chemistry discovery cycles.

Figure 1

(A) Cost and timing in early stages of drug discovery. Adapted from ref (6). Copyright 2010 Nature Publishing Group. (B) Iterative learning cycles of medicinal chemistry based on diverse discipline activities with examples of key approaches used before 1980 (blue), up to 2000 (orange), and nowadays (red).

Figure 2

Integrated fluidic workflow for the automated molecular design–synthesis–screening–analysis–optimization for iterative medicinal chemistry discovery cycles.

Figure 3

Examples of flow chemistry equipment.

Figure 4

Representative examples of available PAT for continuous flow processes.

Figure 5

Schematic representation (A) and case-studies (B,C) of the LeyLab.

Figure 6

Schematic representation of the OpenFlowChem platform.

Figure 7

Schematic representation of Chemputer used for the synthesis of diphenhydramine hydrochloride (Nytol, 11), sildenafil (Viagra, 12), and rufinamide (Banzel, 13). Reproduced with permission from ref (80). Copyright 2019, American Association for the Advancement of Science.

Figure 8

Representation of an homogeneous continuous flow assay using fluorescence resonance energy transfer (FRET). The enzyme solution and the carrier buffer, delivered at 25 μL min^–1 for each syringe pump, were mixed in coil reactor A. A superloop was placed between syringe pump 1 and coil reactor A, while an autoinjector was placed between syringe pump 2 and coil reactor A. The enzyme/inhibitor mixture from coil reactor A was mixed with HIV protease substrate 1 delivered by pump 3, which was connected to an inverted Y-piece to reduce the flow rate at 50 μL min^–1. Excitation and emission wavelengths for the fluorescence detector were 340 and 490 nm, respectively.

Figure 9

Development of a microfluidic confocal fluorescence detection assay for the identification of acetylcholine binding protein inhibitors. The bioassay carrier solution containing AChBP and DAHBA (16) was delivered at 5 μL min^–1 and mixed in a microfluidic chip (4 μL) with the nano-LC effluent containing the potential protein ligand pumped at 0.4 μL min^–1. When DAHBA (16) is displaced by eluting protein ligands, a bubble cell capillary in the reaction chamber of the miniaturized chip detects fluorescence variation by means of a photomultiplier tube.

Figure 10

Integration of electrochemical reaction cell with a continuous flow bioaffinity assay and LC-HRMS analysis. Different potentials (0, 0.4, 0.8, 1.2, and 1.5 V) and operative pH (3.5, 5.0, 7.0, and 10.0) were evaluated for the electrochemical conversion of each substrate.

Figure 11

Microfluidic continuous-flow injection titration assay (CFITA) for monitoring inhibition of thrombin peptidase activity. Enzyme inhibitor solutions were prepared and tested by dissolving the screened compounds at concentration ranging from 75 nM to 25 μM using a buffer (50 mM HEPES pH 8.0 containing 1 mM calcium chloride, 142 mM sodium chloride, 0.05% CHAPS, and 0.1% trehalose) containing 500 nM of Cy5 dye as the internal standard.

Figure 12

Schematic diagram of the SWIFT system from AbbVie.

Figure 13

Automated flow platforms for Suzuki–Miyaura reaction screening, optimization, and validation. The system was composed by a 192 well plate autosampler connected with a HPLC system for reaction segment injection on a predefined sequence, a pump that generates the flow stream, a reactor coil, and a 12-port valve for solvent selection. The reaction outcome is directed by the six-port switching valve into the UPLC-MS instrument for reaction analysis. The excess sample is directed through a diode array detector (DAD) for product collection or waste. Reproduced with permission from ref (102). Copyright 2018 American Association for the Advancement of Science.

Figure 14

Automated flow synthesis and purification of imidazo[1,2-a]-pyridine tested by frontal affinity chromatography (FAC) assay. Products were prepared by acid-catalyzed condensation between ethyl glyoxylate and acetophenone analogues, followed by reaction with aminopyridines. Ketoimine intermediates thus formed were submitted to thermal cyclization and finally diversified by amidation or hydrolysis of the ester moiety. For each compound, three concentrations (7.81, 31.25, and 62.5 μM) were prepared by diluting the corresponding stock solutions (125 μM in PBS) and injected in triplicate into the HPLC. Determination of binding constants (K_d) was performed at 254 and 262 nm in the presence of 20 nmol of HSA immobilized on the column.

Figure 15

Automated flow synthesis and testing of BRD9 bromodomain inhibitors by frontal affinity chromatography (FAC) assay. Desired products were generated by in flow Curtius rearrangement of the in situ generated acyl azide intermediate, followed by tert-butyloxycarbonyl (BOC) deprotection, triazole ring formation, and Suzuki cross-coupling reaction under conventional batch conditions. Detection was performed at 220 and 254 nm. The amount of active loaded protein and the affinity constants were calculated by injecting in duplicate a solution of bromosporine in DMSO at concentrations of 15, 7.5, 3.75, and 1.875 μM, both in PBS and in 100 mM ammonium acetate buffer starting from a 50 mM stock solution.

Figure 16

Microfluidic platform for the synthesis of imidazopyridines. Compounds were first screened by computer-based target prediction and best compounds were then tested for receptor activity. Desired products were prepared by acid-catalyzed Ugi three-component reaction between amine, aldehyde, and isocyanide and purified via preparative HPLC. Functional EC₅₀ values obtained for the test compounds were converted to K_i values using the Cheng–Prusoff equation.

Figure 17

Automated flow synthesis, purification, and analysis tetracyclic tetrahydroquinolines as a novel class of PXR agonists. Products were obtained by multicomponent Povarov reaction, cis/trans-adducts were separated by flash chromatography, and single enantiomers were isolated by chiral HPLC and characterized by NMR and circular dichroism. Activity at the PXR receptor was determined by AlphaScreen technology.

Figure 18

Integrated flow platform for the automated sulphonamide synthesis and in-line biological screening of T-cell tyrosine phosphatase (TCPTP) inhibitors. The synthesized compounds were tested using Caliper’s standard fluorogenic assay on Caliper 250 HTS system. The data were generated by monitoring the fluorescence of an on-chip incubation of TCPTP and 6,8-difluoro-4-methylumbelliferyl phosphate. Reproduced with permission from ref (124). Copyright 2005 Wiley.

Figure 19

Closed-loop CyclOps platform from Cyclofluidic Ltd. for fully integrated and fully automated synthesis, purification, and screening assisted by algorithm-based drug design of Abl kinase inhibitors. Ponatinib analogues were synthesized by Sonogashira cross-coupling reaction between aryl halides and alkynes and purified by in-line preparative HPLC before testing. The Omnia kinase activity assay technology was employed to monitor the real-time kinase activity. For each tested inhibitor, a 3-fold dilution series was generated by an integrated liquid handling robot. The enzyme and the substrate solution were added to each test solution to assess the residual enzyme activity by fluorescence (excitation 360 nm, emission 485 nm). The data for each assay was fitted by linear regression and processed by Matlab software.

Figure 20

Closed-loop CyclOps platform from Cyclofluidic Ltd. for fully integrated and fully automated synthesis, purification, and screening assisted by algorithm-based drug design of dipeptidyl peptidase 4 (DPP4) inhibitors. The library was generated by nucleophilic substitution between BOC-protected diamines and 8-bromo xanthine derivatives, followed by acid-promoted deprotection of tert-butyloxycarbonyl group. After in-line purification by preparative HPLC, biological assays were carried out in 384 well plates. The enzyme (0.82 mU mL^–1 for porcine DPP4 or 34 U mL^–1 for human DPP4) was added, and the residual enzyme activity was monitored by adding the substrate at a final concentration equivalent to the K_M^app.

Figure 21

Closed-loop CyclOps platform from Cyclofluidic Ltd. for fully integrated and fully automated synthesis, purification, and screening assisted by algorithm-based drug design of hepsin inhibitors. Tested compounds were obtained by condensation between silylated amino acids and sulfonyl chlorides, acyl chlorides, or isocyanates, followed by amidation with amidine. The CyclOps bioassay module consisted of a fraction collection station, a reagent station, liquid handling robotics, plate store, an integrated plate reader, a syringe drive, and a two-way, six-port injection valve fitted with a 200 μL loop. Data thus generated were processed by CyclOps software and analyzed with Matlab suite for determining IC₅₀ values.

Figure 22

Fully integrated and automated flow system for the generation of BACE1 inhibitors. The bioassay chip was primed for 2 min with streams of enzyme (90 nM), substrate (0.9 μM), and assay buffer, at 0.8 mL min^–1 for each pump. After in-line purification and analysis, each compound was injected into the dispersion capillary by a liquid handler directly from the preparative HPLC and dispersed by assay buffer into the chip. After 30 min of incubation time, using a gradient calibration, the enzyme activity was determined vs the corresponding fluoresce in concentrations, and the resulting dose–response curve was used to extrapolate the IC₅₀ values. Reproduced with permission from ref (128). Copyright 2014 Wiley.

Figure 23

Schematic representation of the first automated microfluidic platform for synthetic biology.

Figure 24

Microfluidic platform for ultrahigh-throughput hit deconvolution by sequencing of DNA-encoded compound beads. An integrated waveguide irradiates the droplet flow at 365, inducing the photochemical cleavage of compound from the bead into the droplet volume. Droplets dosed with compound (1–3 μM) are then incubated for 18 min within a Frenz-type delay line, at the end of which droplets are focused back into single file and the confocal laser-induced fluorescence detectors measure droplet fluorescence and the software analyze the data.

Figure 25

Integrated flow platform for the synthesis and identification of protein binders. Fluorescence spectra for each compound were recorded at concentration ranging from 0.4 to 2.4 nM, T = 298 K, pH = 7.4, and λ_emission = 280 nm.