Enabling synthesis in fragment-based drug discovery by reactivity mapping: photoredox-mediated cross-dehydrogenative heteroarylation of cyclic amines

A nanogram-to-gram workflow has been established for the identification and development of synthetic transformations which are enabling in Fragment-Based Drug Discovery (FBDD). In this study, we disclose a method for the synthesis of privileged sp2–sp3 architectures via direct cross-dehydrogenative coupling of heterocycles.


General Information
Nanomolar scale reactions in 1536 well plates were performed without exclusion of air or moisture. Micro-and millimolar scale reactions carried out in glass vials were performed with exclusion of air. This was achieved by using solvent that had been sparged with N2 for 15 mins and by purging the headspace of the reaction vial with a positive pressure of N2 and an outlet needle. Commercial solvents and reagents were used without further purification.
Analytical TLC was performed on Macherey-Nagel Alugram® Sil G/UV254 TLC plates visualized using UV (254 nm) then basic KMnO4 solution. Flash column chromatography was performed on a Biotage SP1 system; normal-phase chromatography performed with silica SNAP columns (32−63 µm particle size, KP-Sil, 60 Å pore size) and the stated solvent system (n.b. Petrol refers to Petroleum Ether bp 40-60 o C) and reverse-phase chromatography performed with Biotage C18 Ultra cartridges with an MeCN in H2O gradient, containing 0.1% HCO2H.
Photochemical reactions performed in 1536 microtiter plates (MTP) were irradiated with a GLW 50W IP65 RGB LED floodlight set to blue at a distance of 5 cm from the top of the plate.
Photochemical flow reactions were performed using a Vapourtec E-series platform equipped with the UV-150 module. This module consists of a temperature-controlled irradiation chamber where a transparent fluorinated ethylene polymer (FEP) reactor (1 mm i.d., 10 mL, S4 PN: 50-1287) is coiled around a blue LED assembly (emitting at 420 nm with a total output power of 17 W, . A 75 psi back pressure regulator was used (Kinesis, P-786) and reactions were monitored with in-line infrared (IR) monitoring using a Mettler Toledo ReactIR FD.
NMR spectra were recorded on a Bruker AV400 (Avance 400 MHz) spectrometer. Chemical shifts for 1 H and 13 C NMR spectra are reported as δ in units of parts per million (ppm) and quoted to the nearest 0.01 ppm relative to the residual protons in CDCl3 (7.26 ppm,77.16 ppm), d6-DMSO (2.50 ppm, 39.52 ppm) or CD3OD (3.31 ppm, 49.00 ppm). Coupling constants (J) are quoted in Hertz (Hz) and are reported to the nearest 0.1 Hz, with multiplicity reported according to the following convention: s = singlet, d = doublet, t = triplet, q = quartet, m = multiplet, br = broad and associated combinations e.g. dd = double of doublets. DEPT 135 and 2-dimensional experiments (COSY, HMBC, HSQC and ROESY) were used to support assignments and are reported where appropriate.

Nanomolar Scale Automated Chemistry Experiments in 1536 well Microtiter Plates (MTP)
Nanomolar scale screening reactions (125 nmol) were performed in Corning® 1536-well MTP (Corning 1536 COC White, Cat. No. 4570, Cyclic Olefin-Copolymer COC, 12.5 μL-wells, flat bottom, white). Greiner® 384-well MTPs (Cat No. 651201, Polypropylene, 120 µL, V-bottom, translucent) were used as the reagent source plate and the analysis plates. The source plate (384-well MTP) was dosed using an Andrew 1000G liquid handler (Andrew Alliance, Switzerland) equipped with Gilson Pipetman Pipettes (see: pages S8 & S83). Subsequently, this 384-well source MTP were used to dose the 1536-well reaction MTP using a Mosquito® HTS liquid handling robot (TTP Labtech, UK), using the appropriate Mosquito protocol (see: pages S8 & S88). On completion of dosing, the 1536-well reaction MTP were heat sealed with a 4titude Clear Heat Seal (4ti-0541) and irradiated with a GLW 50W IP65 LED floodlight set to blue light or white light at full power. At the end of the reaction time, the 1536-well reaction MTP were mirrored (2.0 µL) into pre-dosed 384-well analysis MTP (see: page S110). The 384well analysis MTP were pre-dosed by hand using multichannel pipettes (E1 ClipTip Equalizer Supplementary Information -S5 384 12-channel); the wells dosed with a 0.510 mM solution of internal standard (IS) 4bromobiphenyl in DMSO (0.050 µmol, 50 mol %, 98 µL per well). Dosing of the crude reaction mixture (2 µL) into the analysis plates was achieved using a mix aspirate, mix dispense cycle on the Mosquito® HTS liquid handling robot (see: Analysis plate dosing S110) and heat sealed with adhesive free aluminium foil seals, agitated on a plate shaker for 5 mins then subjected to LCMS analysis. Reactions were performed in glass vials equipped with stir bars (as detailed in the table above). Solid reagents were weighed by hand, liquid handling of stock solutions was performed by hand with Gilson Pipetman pipettes. Stock solutions were degassed for 10 minutes prior to dispensing into reagent vials equipped with septa and purged with nitrogen.

Micro-/Millimolar Scale Chemistry Experiments in Batch
Reaction vials were stirred and irradiated in an EvoluChem TM PhotoRedOx box (mono: HCK1006-01-016, duo: HCK1006-01-023) equipped with either one or two Evoluchem TM 455 nm 18W LEDs (HCK1012-01-002) as stated accordingly in the relevant optimization table, the appropriate vial holder (as detailed in the table above) and cooling by internal fan component.
Assay yields are reported based on LC-MS conversion with product confirmation by 1 H NMR.
Unfortunately yields could not be accurately determined by 1 H NMR as most products exhibit line broadening as a result of interconverting mixtures of rotamers.

Supplementary Information -S6
Scale up for isolation For 0.5 mmol scale up reactions, the crude reaction mixture was checked after 16 hours via LC-MS analysis, following this the product was isolated via aqueous extraction then subsequent purification according to the procedure detailed in the experimental section.

Continuous Flow Chemistry Experiments
Optimization Solid reagents were weighed into a 10-20 mL vial (Biotage, 354833), degassed DMSO added and the reaction mixture sonicated until all solid had dissolved. The stock solution was degassed by N2 sparging (10-15 mins) then volatile liquids (N-Boc pyrrolidine) added. The reaction mixture was pumped at the stated flow rate as a slug of reaction mixture pushed with DMSO through the reactor coil (10 mL reactor coil) irradiated by 420 nm LEDS (17 W total output power). The entirety of the crude reaction mixture was collected, and an aliquot analyzed directly by LC-MS to determine conversion based on coupled product and unreacted heteroarene.

Gram scale reaction
Solid reagents were weighed into a 250 mL pear shaped flask, degassed DMSO added and the reaction mixture sonicated until all solid had dissolved. The stock solution was degassed by N2 sparging (10-15 mins) then volatile liquids (N-Boc pyrrolidine) added. The reaction mixture was pumped at 1.00 mL.min -1 as a slug of reaction mixture pushed with DMSO through the reactor coil (10 mL reactor coil) irradiated by 420 nm LEDS (17 W total output power). The reaction was monitored by inline IR monitoring and aliquots collected every 30 minutes and analyzed by LC-MS to monitor conversion once the system had reached steady-state. The crude reaction mixture was collected, and the product was isolated via aqueous extraction then subsequent purification according to the procedure detailed in the experimental section. 2.0 eq B1 = 1.25 equiv B1 = 3.0 equiv B1 = 5.0 equiv D1 = D1 = D1 = D1 = D1 = D1 = 1.0 eq 2.0 eq 1.0 eq 2.0 eq 1.0 eq Supplementary Information -S10

Analysis Plate Setup
After 16 hours, the 1536-well reaction plates were mirrored into two 384 well MTPs ( Figure   SI-3), each well primed with 98 µL of internal standard IS (4-bromo biphenyl, (50 mol %) in DMSO, 0.510 mM). The internal standard solution was pipetted by hand using a multi-channel pipette and 2 µL of the crude reaction mixture added using a mix-aspirate-mix-dispense function (TTPLabtech Mosquito® HTS). Due to the tip spacing on the Mosquito, the duplicate reactions which are positioned above and below each other in one column of the reaction plate ( Figure SI-2) now appear side-by-side in two separate columns in the analysis plate ( Figure SI-3). Details of the contents of each well can be found in workbook included the Supporting Information section on the journal website.   Supplementary Information -S13   Supplementary Information -S17

Photochemical reactions in flow
Optimization protocol: {Ir[dFCF3(ppy)2]dtbbpy}PF6 6a (0.02 equiv, 0.01 mmol), 5bromoisoquinoline 3a (0.5 mmol, 1.0 equiv), (NH4)2S2O8 7 (2.0 mmol, 4.0 equiv) and TsOH.H2O 8 (1.0 mmol, 2.0 equiv) were weighed into a 5 mL crimp top glass vial. The vial was sealed, degassed DMSO added (5 mL) and the reaction vial was sonicated to ensure all reagents were in solution. Following this, the solution was sparged with N2 for 10 mins then N-Boc pyrrolidine 4b added (1.5 equiv or 3.0 equiv). The clear, yellow solution was then pumped at the stated flow rate as a slug of reaction mixture pushed with DMSO through the reactor coil (10 mL reactor coil) irradiated by 420 nm LEDs (17 W total output power). Once the system had reached steady-state (as determined by in-line IR monitoring) 2 an aliquot was taken and analyzed by LC-MS to determine conversion. a Percent conversion approximation based on relative HPLC peak area integrations of starting material compared with product. b 4 mol % of photocatalyst 6a.
2 These peaks were chosen as they were strong signals observed in the IR spectra of the crude reaction mixture. These peaks have not been unequivocally assigned, but we hypothesize that they correspond to the ν(C=O) stretch of the Boc group on the starting material or product (1685 cm -1 ) and a ν(S-O-H) bend of the bisulfate byproduct from decomposition of 8 (1210 cm -1 ).
Supplementary Information -S18  We hypothesize that the reaction proceeds via Minisci-type addition of an α-amino radical into a protonated heteroarene. Generation of the α-amino radical could occur through hydrogen atom transfer (HAT) from a reactive oxygen species generated from decomposition of the persulfate 7 (Figure SI-9a). However, and alternative mechanism could exist where the α-amino radical is generated via oxidation of the amine lone pair by the Ir catalyst 6a or persulfate 7 followed by deprotonation ( Figure SI-9b).

Reactivity Map -amine scope for coupling of 3a
Table SI-5: Heat map for coupling of 3a with a variety of cyclic amines 3, 4, 5 3 Assay yield determined through HPLC conversion of starting material to product. 4 Site of reaction observed on fully characterized product denoted by ( ). 5 In the cases of unsymmetrical, substituted 5-and 6-membered amines (*), there is evidence of more than one product with the same desired mass, as observed by LC-MS. However, as these compounds were not isolated and fully characterized, we cannot confidently determine whether this additional product is a diastereomer or regioisomer, although the former is likely. PhotoRedOx box on a stirrer plate and irradiated for 16 hours with stirring at 500 rpm.

Supplementary Information -S34
Observation of the fully deprotected des-Boc-des-Ac-(±)5j (18%) was surprising as the Boc deprotection was performed with 4N HCl at room temperature and to the best of our knowledge, there are no reports of acetamide deprotection under these conditions. We hypothesize that des-Boc-des-Ac-(±)5j arose from deprotection of des-Boc-(±)5j-minor as only one amide tautomer was observed by 1 H & 13 C NMR it seems feasible that the other des-Boc-(±)5j-minor tautomer was unstable under the Boc deprotection conditions likely caused by neighbouring group participation of the adjacent isoquinoline nitrogen.