Macrocycle synthesis strategy based on step-wise “adding and reacting” three components enables screening of large combinatorial libraries

Macrocycles provide an attractive modality for drug development, but generating ligands for new targets is hampered by the limited availability of large macrocycle libraries. We have established a solution-phase macrocycle synthesis strategy in which three building blocks are coupled sequentially in efficient alkylation reactions that eliminate the need for product purification. We demonstrate the power of the approach by combinatorially reacting 15 bromoacetamide-activated tripeptides, 42 amines, and 6 bis-electrophile cyclization linkers to generate a 3780-compound library with minimal effort. Screening against thrombin yielded a potent and selective inhibitor (Ki = 4.2 ± 0.8 nM) that efficiently blocked blood coagulation in human plasma. Structure–activity relationship and X-ray crystallography analysis revealed that two of the three building blocks acted synergistically and underscored the importance of combinatorial screening in macrocycle development. The three-component library synthesis approach is general and offers a promising avenue to generate macrocycle ligands to other targets.


Materials and Methods S3
Supplementary Results Library synthesis and characterization of top hits S10 Overall structure of human α-thrombin in complex with 57 S10 Comparison of thrombin-bound macrocycles 7 and 57 S11 Supplementary Table   Table S1: Statistics on X-ray structure data collection and refinement S12 Table S2: Intra-molecular interactions between atoms of 57 S13 Table S3: Inter-molecular interactions between atoms of 57 and thrombin S14 Table S4: Solvent excluded volume and buried surface of 57 and thrombin S15

Supplementary Figures
Figure S1: Side products in the thiol-to-amine macrocyclization reaction S16 Figure S2: LC-MS analysis thiol-to-amine macrocyclization reactions S17 Figure

Materials
All Fmoc-amino acids, coupling reagents, primary amines, and chemical linkers are commercially available.

LC-MS analysis
A Shimadzu-2020 single quadrupole LC-MS system was used for sample analysis. Samples were analyzed on a reverse phase C18 column (Phenomenex Kinetex®, 2.6 m, 100 Å, 50 × 2.1 mm) using a linear gradient of solvent B (acetonitrile, 0.05% formic acid) over solvent A (H2O, 0.05% formic acid), typically from 0 to 40% in 6 min at a flowrate of 1 ml/min, and by mass analysis in positive or negative mode.

Synthesis of N-bromoacetyl peptides
Tripeptides were synthesized on an automated peptide synthesizer (Intavis, MultiPep RSi) at a 50 mol scale on Rink amide MBHA resin (0.31 mmol/g) using Fmoc-chemistry. All amino acids were coupled twice at a 5-fold molar excess. The N-terminus was bromoacetylated on solid phase by addition of bromoacetic

N-alkylation of peptide library
Lyophilized N-bromoacetyl peptides were dissolved in H2O to generate 10 mM stocks. Amine reagents were prepared as 160 mM DMSO stocks. The peptides and amines were combinatorially reacted in v-

S4
bottom 96-well plates by transferring 10 l of peptide and 10 l of amine, the plates closed with a lid, and incubated at 37°C for 2 hrs. The completion of randomly picked reactions was monitored by LC-MS.

Removal of cysteine side chain protecting group (-SO3H)
To wells of 384 deep-well plates, 94 l of a solution of 60 mM NH4HCO3 buffer pH 8 and 425 M TCEP was transferred using an automated dispenser (Multidrop TM Dispenser, Thermo Scientific). To these wells, 2 l of 5 mM alkylated peptides were transferred by a Biomek® FXP laboratory automation workstation using a 96 pipetting head and 80 l disposable plastic tips. The plates were covered and incubated at 37°C for 1 hr. The quantitative removal of the sulfonate protecting group was confirmed by LC-MS for sample reactions. The final concentration of the peptides was 104 M. Each peptide was transferred to seven wells: to be cyclized with six linkers with one remaining linear.

S-to N cyclization
The peptides in the 384 deep-well plates (96 l, 104 M) were cyclized in a combinatorial fashion by adding the linkers 1-6 (4 l, 20 mM, in DMSO) using the Biomek® FXP laboratory automation workstation and 80 l disposable plastic tips. The final concentrations of peptide, amine, TCEP, and linker were 100 µM, 1.6 mM, 400 µM, and 800 µM respectively. One well per peptide was left linear as a control.
The protease activity was measured by monitoring the change in florescence intensity. The fluorescence intensity was measured with a Tecan Infinite F500 fluorescence plate reader (excitation at 360 nm, emission at 465 nm) at 25°C for a period of 30 min with a read every 3 min. The hits were identified by calculating percentage (%) of residual protease activity compared to protease activity without macrocycle.

S5
Prior to the screen, we tested if non-reacted bis-electrophile reagents interfere with the thrombin activity assay, and found that this was not the case. We therefore decided to not add a reagent such as cysteine for quenching non-reacted electrophilic reagents prior to the assay.

Inhibition activity of crude reaction dilutions
The inhibitory activity of macrocycle reactions showing <15% residual thrombin activity in the primary screen was measured again (identical reactions as in primary screen) along with eleven 2-fold dilutions of the reactions. The dilutions were made by sequentially transferring 2 l of the reaction or dilution to 2 l of assay buffer to obtain peptide concentrations ranging from 100 M to 50 nM in 384-well plates. To these wells, a volume of 8 l of thrombin (4 nM) in assay buffer was added and incubated for 15 min. A volume of 5 l of the fluorogenic substrate Z-Gly-Gly-Arg-AMC (150 M) in assay buffer containing 3% DMSO was added, and the fluorescence intensity was measured for 30 min using an Infinite M200Pro plate reader (Tecan; excitation at 368 nm, emission at 467 nm) with a read every minute. Sigmoidal curves were fitted to the data to determine the apparent IC50 values using the software and equation described below.

Macrocycle synthesis
Linear peptides were synthesized at a 0.05 mmol scale by standard Fmoc solid-phase chemistry as described above. N-terminal peptide acylation was carried out by treating with 528 l bromoacetic acid (0.5 M in DMF, around 5 eq.) followed by the addition of 528 l of DIC (0.5 M in DMF, around 5 eq.). The reaction mixture was shaken at 400 rpm for 30 min. The reaction was repeated twice. The resin was washed three times with 2 ml of DMF. SN2 displacement reactions were performed by adding 1 ml of primary amine (0.5 M in DMF, 10 eq.), followed by shaking for 1 hr at RT. Peptides were washed, cleaved, and purified as the N-bromoacetyl peptides described above. Around 4 mg of purified and lyophilized peptide was dissolved in 80 ml of 60 mM NH4HCO3 buffer (pH 8) to reach a concentration of 125 M. Next, 20 ml of 2 mM linker reagent in ACN was added to reach a final concentration of 100 M peptide and 400 M linker reagent.
The reaction was incubated for 2 hrs at 37°C. The reaction was lyophilized, the products dissolved 10 ml of H2O containing 0.1% TFA, and purified by preparative RP-HPLC.

Analytical HPLC
The purity of macrocycles was assessed by analytical RP-HPLC (UV 220 nm) using a C18 column (Agilent Zorbax 300SB, 5 m, 4.6 mm × 250 mm) and a linear gradient of solvent B (acetonitrile, 0.1% TFA) over solvent A (H2O, 0.1% TFA) from 0−50% in 15 min at a flow rate of 1 ml/min. Typically, 100 nmol of macrocycle (around 5 g) were injected. The mass of the purified macrocycles was confirmed by ESI-MS.

Quantification of reaction yields
The yields of the various macrocycle synthesis steps were estimated based on the area under the peaks of the absorption spectra, either recorded by analytical HPLC or the LC-MS instrument, and are not isolated yields. The yield of macrocycle was calculated by dividing the area of the macrocycle peak by the sum of the area of all peaks corresponding to peptide-containing species (macrocycle, unmodified linear peptide, modified linear peptide, disulfide-linked peptide, etc).

Determination of the inhibitory constants of macrocycles
The IC50s  Sigmoidal curves were fitted to the data using Prism 5 (GraphPad software) and the following doseresponse equation: wherein x is macrocycle concentration, y is % protease activity, and p is Hill slope. IC50 values were derived from the fitted curve The inhibitory constants (Ki) were calculated using the following equation of Cheng and Prusoff: wherein IC50 is the functional strength of the inhibitor, [S]0 is the total substrate concentration, and Km is the Michaelis-Menten constant. The Km for thrombin and the substrate Z-Gly-Gly-Arg-AMC was determined to be 168 M.

Determination of target specificity
The specificity was profiled by testing A volume of 50 l of the substrate in assay buffer containing 3% DMSO was added. Fluorescence intensity was measured for 30 min using an Infinite M200Pro plate reader (Tecan; excitation at 368 nm, emission at 467 nm) with a read every minute. For APC protease, absorption was measured at 405 nm with the same plate reader for a period of 30 min with a read every minute. The reactions were performed at 25°C. The inhibitory constants were calculated as described above.

Determination of aPTT and PT
Coagulation times were determined in human plasma using a STAGO STart4 coagulation analyzer (Diagnostica). Human single donor plasma was used (Innovative Research). For the extrinsic coagulation, 50 l of plasma with inhibitor (0.5, 1, 2, 5, 10, 20, and 40 M) and without inhibitor was placed in the incubating chamber of the instrument for 1 min at 37°C. After incubation, 100 l of Innovin (recombinant human tissue factor, synthetic phospholipids, and calcium in stabilized HEPES buffer system; Dade Behring/Siemens) was added using the pipet connected to the instrument. Upon addition of this reagent, the electromagnetically induced movement of a steel ball in the plasma was monitored. The time until the ball stopped moving was recorded as coagulation time. For the intrinsic coagulation, 100 l of plasma with inhibitor (0.5, 1, 2, 5, 10, 20, and 40 M) and without inhibitor was incubated with 100 l of Pathromtin* SL (silicon dioxide particles, plant phospholipids in HEPES buffer system, Siemens) for 2 min at 37°C.
Subsequently, the coagulation was triggered by the addition of 100 l of CaCl2 solution (25 mM, Siemens).

Crystallization, data collection and structure determination
Human α-thrombin was purchased from Haematologic Technologies (Catalogue number: HCT-0020). The protein-stabilizing agent was removed by using a PD-10 desalting column (GE Healthcare) equilibrated with 20 mM Tris-HCl, 200 mM NaCl, pH 8.0. Buffer exchanged human α-thrombin was incubated with the macrocycle 57 (N14-PR4-A) at a molar ratio of 1:3 and subsequently concentrated to 7.5 mg/ml by using a 3000 MWCO Vivaspin ultrafiltration device (Sartorius-Stedim Biotech GmbH). Macrocycle 57 was added during the concentration to ensure that a 3-fold molar excess is preserved. Geometrical parameters of the model are as expected or better for this resolution. The solvent excluded volume and the corresponding buried surface were calculated using PISA software and a spherical probe of 1.5 Å radius (Table S4). Intra-molecular and inter-molecular hydrogen bond interactions were analysed S9 by PROFUNC, LIGPLOT+ and PYMOL software (Tables S2 and S3). The RMSD between 57 (N14-PR4-A; PDB 6T7H) and 7 (P2; PDB 6GWE) atoms was calculated using CLICK server. S10

Library synthesis and characterization of top hits
The reactions of the 15 peptides with 42 amines were performed in 96-microwell plates using robotic liquid handling and applying the conditions described for the model peptide, leading to a final peptide concentration of 5 mM. After alkylation and LC-MS confirmation of the product for sample reactions, we transferred two l of each of the 630 reactions to seven wells of 384-microwell plates, removed the sulfonate protecting groups using TCEP (425 M, 94 l), and cyclized the peptides (100 M) by adding 8-fold molar excess of reagents 1 to 6 (20 mM, 4 l). Uncyclized control reactions were performed in parallel. All steps were performed on a liquid handling platform using disposable tips and LC-MS analysis of sample reactions confirmed macrocycle formation as the main product.

Overall structure of human α-thrombin in complex with 57
Human -thrombin consists of two polypeptide chains of 259 (H-chain) and 37 amino acid residues (Lchain) covalently linked via a disulfide bridge (Cys122 of H-chain with Cys1 of L-chain). The electron density of the H-chain is clearly visible for all residues with the exception of the last two carboxyl-terminal residues (Gly246 and Glu247). The L-chain of human -thrombin can be traced unambiguously from Phe1G to -Gln244). Like other serine proteases, the human -thrombin has three disulfide bridges (Cys42 -Cys58, Cys168 -Cys182 and Cys191 -Cys220) and an active site containing the catalytic triad His57, Asp102 and Ser195 residues that are located at the junction of both barrels. The overall structure of human -S11 thrombin in complex with 57 does not show any striking rearrangements of the main backbone if compared to the structure of human  -thrombin in complex with 7. A structural alignment performed using GESAMT revealed a root mean square deviations (RMSD) of the C superimposed atoms of 0.36 Å. The major differences are between Arg73 and Arg77A probably due to crystal packing interactions.
The two chromatograms at the bottom of each page show the LC-MS analysis of control reactions in which peptide BrAc-Gly-Cys(SO3H)-R (R = Gly-Trp-NH2 appendix for better UV absorption) (a) and the library peptide BrAc-9-11-Cys(SO3H)-R (containing the UV-absorbing building block 9) (b) were synthesized on solid phase, purified by precipitation using diethylether (but not by HPLC), alkylated with benzylamine, and cyclized with linker 1 (model peptide) or linker 2 (library peptide). The model peptide was analyzed by analytical HPLC and the mass of molecules in peaks determined by peak collection and MS determination.
The library peptide was analyzed by LC-MS.