Transfer hydrogenations catalyzed by streptavidin-hosted secondary amine organocatalysts

Here, the streptavidin–biotin technology was applied to enable organocatalytic transfer hydrogenation. By introducing a biotin-tethered pyrrolidine (1) to the tetrameric streptavidin (T-Sav), the resulting hybrid catalyst was able to mediate hydride transfer from dihydro-benzylnicotinamide (BNAH) to α,β-unsaturated aldehydes. Hydrogenation of cinnamaldehyde and some of its aryl-substituted analogues was found to be nearly quantitative. Kinetic measurements revealed that the T-Sav:1 assembly possesses enzyme-like behavior, whereas isotope effect analysis, performed by QM/MM simulations, illustrated that the step of hydride transfer is at least partially rate-limiting. These results have proven the concept that T-Sav can be used to host secondary amine-catalyzed transfer hydrogenations.

The biotinylated organocatalysts 1 and 2 were synthesised as previously reported. 1 T-Sav (Streptavidin Streptomyces avidinii recombinant, tetramer, MW ≈ 52 kDa, "core" Streptavidin with amino acids 13-139) was obtained commercially from ProSpec (PRO-791) as lyophilized powder in 10 mM KP i pH 6.5 and stored at −23 °C upon receipt until further use. According to the supplier T-Sav has the following amino acid sequence:
Size exclusion chromatography was performed using a ÄKTA Purifier workstation (GE Healthcare) or a Bio-Rad system with the respective column mentioned in the detailed procedure. `6

Expression and Purification
Monomeric streptavidin (M-Sav) was expressed using an E. coli expression system with the following protocol. Plasmid pRSET-mSA (see 2.2) containing the desired M-Sav gene (Fig.  S1) in a pRSET-A vector was transformed into calcium competent BL21 AI cells and grown for 16 h on LB agar plates containing 100 μg/mL ampicillin. A single colony from the plate was picked to inoculate a 5 mL LB culture for overnight. The starter culture was diluted into 500 mL of LB medium containing 100 μg/mL ampicillin. The culture was grown at 37 °C and 225 rpm to an OD 600 of 0.8−1.0 and induced with a final concentration of 0.5 % w/v L-arabinose. The culture was grown overnight at 20 °C. The pellet was harvested by centrifugation at 4000 rpm at 4 °C for 25 min, resuspended in 10 mL of wash buffer (50 mM Tris-HCl, 100 mM NaCl, pH 8.0) and lysed via sonication (7 min, 5 sec on, 10 sec off). 10 mL of lysate buffer (50 mM Tris-HCl, 100 mM NaCl, and 6 M GdnHCl, pH 8) were added to the suspension and left to incubate for 3 h at 4 °C. The insoluble fraction was removed by centrifugation at 4000 rpm at 4 °C for 25 min, and the supernatant was mixed with 3 mL of Ni-NTA affinity resin. After incubation at 25 °C for 1.5 h with occasional stirring, the resin was washed twice with 1.5 volumes of resin wash buffer (50 mM Tris-HCl, 100 mM NaCl, 6 M GdnHCl, and 10 mM imidazole, pH 7.5). M-Sav was eluted with elution buffer (3× 3.5 mL: 50 mM Tris-HCl, 150 mM NaCl, 6 M GdnHCl, 300 mM imidazole, pH 8.0). Samples of the wash and elution fractions were collected and analysed by SDS-PAGE (15% w/v). The elution fractions were added drop by drop to 40 mL of ice-cold refolding buffer (50 mM Tris-HCl, 150 mM NaCl, 0.3 mg/mL catalyst 1 or 2, 0.2 mg/mL oxidized glutathione, and 1 mg/mL reduced glutathione) under rapid stirring to refold the protein. The precipitates were removed by centrifugation at 4000 rpm at 4 °C for 25 min. The refolded protein solution was concentrated to 5 mL using Amicon ultra centrifugation with a 10 kDa cut-off. The concentrated protein solution was applied to size exclusion chromatography (Hi-Load TM column, Superdex 200 pg, 50 mM Tris-HCl, 150 mM NaCl, pH 8.0). Fractions containing protein (analysis by following 215, 255 and 280 nm UV traces) were collected and the samples loaded on SDS-PAGE (15% w/v) to check the purity of the protein. Fractions containing M-Sav were pooled, transferred to a centrifugal concentrator with a 10 kDa cut-off and the buffer was exchanged to 10 mM KP i , pH 7.0. `7  µmol, 1.0 eq.) and 78.27 µL MeOH were added (Scheme S1). The Eppendorf tube was placed in a Falcon tube, which was placed inside an incubating shaker (300 rpm, 25 °C).

Experimental Details for the Activity Screening of Catalysts 1 and 2 and T-Sav in the Transfer Hydrogenation from BNAH to Cinnamaldehyde
After 24 h, DCM (500 µL) and water (500 µL) were added, and the phases were separated. The aqueous phase was extracted (2 × 500 µL DCM), and the organic phases were combined. The volatiles were removed under reduced pressure, the residue was taken up in 700 µL CDCl 3 , and transferred to an NMR tube. This was directly subjected to 1 H NMR analysis (See section 3.2). Each T-Sav:1/2 or M-Sav:1/2 catalysis were run in triplicates. This procedure was also applied for the aldehydes (8)(9)(10)(11)(12)(13) screening. For the reactions where 5 equivalents of BNAH and 25% methanol were employed, adjustment in the calculation of BNAH and T-Sav stock solutions were made.

Exemplary 1 H NMR Spectrum after Extraction of the Biphasic Reaction Mixtures
The conversion was estimated by integration of the substrate (red box, doublet) and the product (yellow box, singlet) aldehydic peaks (Fig. S2, region between 9-10 ppm). Formation of the product was also confirmed by the presence of the H α and H β peaks (yellow box, region between 2.5-3.5 ppm). Exemplary of the crude of reaction for the reduction of cinnamaldehyde (3) to hydrocinnamaldehyde (5) in presence of BNAH (4) is shown in the green spectrum. The same methodology was applied to calculate the estimated conversion for all the other substrates (8)(9)(10)(11)(12)(13). Further information is listed in section 6. Nicotinamide (4.8 g, 1 eq., 40.00 mmol) was dissolved in 100 mL of 1,4-dioxane-methanol (4:1), and benzyl bromide (4.75 mL, 1 eq., 40.00 mmol) was added. The reaction mixture was refluxed at 80 °C for 16 h, after which time a precipitate was observed. This solution was cooled and 1,4-dioxane (50 mL) was added to further precipitate the final product. After filtration, the precipitate was washed with 1,4-dioxane (3x50 mL) and N-benzyl-3carbamoylpyridinium bromide was obtained (4a, 6 Under nitrogen atmosphere, the bromide salt of 1-benzyl-3-carbamoylpyridinium bromide (4a, 1.06 g, 5.00 mmol) was dissolved in distilled water (100 mL) and sodium hydrogen carbonate (2.10 g, 5 eq., 25.00 mmol) was added. Sodium dithionite (4.35 g, 5 eq., 25.00 mmol) was then added portion-wise and the reaction mixture was stirred at room temperature for 3 h in the dark. During this time, the solution turned from orange to yellow and a yellow product precipitated. The solid was filtered, washed with cold distilled water (3x50 mL) and dried over phosphorus pentoxide under vacuum to afford a bright tallow solid (4, 1.00 g, 93.43% yield).   The estimated conversion of cinnamaldehyde (3) into hydrocinnamaldehyde (5)  Formation of an unidentified side-product was observed around 9.25 ppm. The peak around 9.25 ppm resemble the characteristic doublet of an aldehyde, thus the peak was assumed to refer to a single H. Therefore, the estimated conversion of this reaction was made by a ratio between the aldehydic peaks of the starting material 3, product 13 and this side-product.     The estimated conversion of (E)-3-(4-methoxyphenyl) acrylaldehyde (11) into 3-(4methoxyphenyl) propanal (17)

Michaelis-Menten Kinetic for the Reaction Between 3 and 4 7.1 Procedure for the Kinetic Assessment
The pseudo-first order kinetic assessment for the transfer hydrogenation of BNAH (4) to cinnamaldehyde (3) was setup following the procedure reported in the section 3.1 of this supplementary information. Reaction were run in triplicate and stopped after 24 hours, and the estimated conversion was assessed by 1 H NMR spectroscopy. Cinnamaldehyde was used at the following concentrations: 0, 0.8, 1.6, 3.3, 4.4, 5.5, 6.6, 7.7 and 8.8 mM. BNAH concentration was maintained constant at 20 mM, meanwhile 0.076 mM of T-Sav:1 complex were employed (Fig. S3 and Table S1).

Computational details of the molecular models set up
The origin of atom coordinates of streptavidin with bound biotin was adapted from X-Ray structure as available in Protein Data Bank (PDB ID 1STP). 14 Biological assembly, missing atoms, protonation state of titratable amino acids, optimization and molecular dynamics (MD) simulations performed for the T-Sav with biotin-catalyst were described in our previous paper. 1 In brief, after adding missing hydrogen atoms, the system was neutralized by addition of 7 sodium counter ions and placed in the box of 87  86  76 Å 3 of TIP3P 15 water molecules. Then, optimization and molecular dynamics (MD) simulations were performed at molecular mechanistic (MM) level using NAMD software, 16 with the AMBER ff-03 parameters. 17 Missing atoms types, charges and parameters were generated using the Antechamber software package together with a general AMBER force field (GAFF) 18,19 available in Amber Tools. NPT MD of 10 ns with time step of 1 fs at 300 K were carried out after previous optimization, heating (from 0 to 300 K with 0.001 K temperature increment) and equilibration processes of 100ps. The constant temperature and pressure were controlled using the Langevin piston method. 20 Periodic boundary conditions (PBC) using the particle mesh Ewald method were applied. A Cut-off for nonbonding interactions was applied using a smooth switching function between 14.5 to 16 Å. Analysis of the time evolution of the RMSD of the backbone atoms confirmed that the system was equilibrated after series of MD equilibration, consisting in 3 steps: 1) 10ns of MD with Sav-biotin complex; 2) 4 ns MD of biotin linked with the substrate; and 3) 20ns of MD of biotin with substrate and BNAH (Fig. S4). Last structure from 10 ns MDs was used in order to explore the hydride transfer step from BNAH to cinnamaldehyde in the active site of T-Sav. After optimization of the system, those residues located 20 Å beyond any atoms of the substrate were kept frozen in the remaining calculations. As explained below, potential energy surfaces, free energy surfaces and spline corrections have been performed using fDynamo 21 library together with implemented AMBER force field. 22

Potential Energy Surfaces
Exploration of the Potential Energy Surfaces (PES) was carried out by choosing the appropriate combination of internal coordinates ( i ). Thus, a first attempt to explore a twodimension PES explicitly controlling the hydride transfer and the distances defining the rehybridization of the carbon atoms, confirms that the two processes are not coupled: first the internal electron transfer takes place, followed by the hydride transfer (Fig. S5). Then, the antisymmetric combination of distances defining the position of the transferring hydride from `38 the donor to the acceptor atoms was used as reaction coordinate. A harmonic constraint was used to maintain the proper interatomic distances along the reaction coordinate, and a series of conjugate gradient optimizations and L-BFGS-B optimization algorithms were applied to obtain the final potential energy of the minimized constrained geometry. The QM sub-set of atoms were described first by the RM1 semiempirical Hamiltonian 23 and later by the M06-2X functional with the standard 6-31+G(d,p) basis set using the Gaussian09 program. 24

Free Energy Surfaces
FESs were obtained, in terms of a Potential of Mean Force (1D-PMF), using the Umbrella Sampling approach 25 combined with the Weighted Histogram Analysis Method (WHAM). 26 Series of MD simulations were performed adding a constraint along the selected reaction coordinates with an umbrella force constant of 2500 kJ·mol -1 ·Å -2 . In every window QM/MM MD simulations were performed with a total of 5 ps of equilibration and 20 ps of production at 303 K using the Langevin-Verlet algorithm 27 with a time step of 1 fs. Structures obtained in previously computed PESs were used as starting points for the MD simulations in every window. `39

Spline Corrections
A correction term is interpolated to any value along the reaction coordinates, , in the FES. A continuous energy function is used to obtain the corrected PMFs: where S is the spline function and is the difference between the energies obtained at low-Δ level (LL) and high-level (HL) of theory of the QM part. The RM1 semiempirical Hamiltonian was used as LL method, while a density functional theory (DFT)-based method was selected for the HL energy calculation. In particular, HL energy calculations were performed by means of the hybrid M06-2X functional 28 using the standard 6-31+G(d,p) basis set. These calculations were carried out using the Gaussian09 program. The resulting Free Energy Surface is shown in Fig. S6 while evolution of key-interatomic distances and dihedral angle along the PMF of the hydride transfer step computed at M06-2X:RM1/MM level is shown in Fig. S7.

Quantum tunnelling corrections
Taking into account that the chemical step under study involves the transfer of a light particle (hydride transfer), the quantum tunelling contribution was evaluated under the framework of Ensemble Averaged Variational Transition State Theory (EA-VTST): [29][30][31][32][33] (S3)  32,[34][35][36] In the present work, calculation of the tunneling transmission coefficient, (T), were calculated with the small-curvature tunneling (SCT) approximation, which includes reaction-path curvature appropriate for enzymatic hydride transfers. 34,37,38 The final tunneling contribution (see main text) is obtained as the average over the reaction paths of 10 TS structures.

Isotope effects
Quasi-classical equilibrium (BIE and EIE) 39 and kinetic isotope effects (KIEs) have been computed for isotopic substitutions of key atoms from 100and 16 couples of stationary structures, comparing the hydride transfer TSs and reactant complex,pre-transfer and post-`41 transfer intermediates, Michaelis complex and BNAH in the water solvent at RM1/MM and M06-2X/MM levels of theory, respectively. Then, the ratio between the rate constants corresponding to the light atom "L" (protium) and the heavier isotope "H" (deuterium) can be computed using the Transition State Theory (TST), as: In eq. S5, the subscripts H and L refers to heavy (deuterium) and light (protium) isotopologs, respectively, the total partition function, Q, was computed as the product of the translational, rotational, and vibrational partition functions for the isotopologs in the two stationary structures under comparison, a and b. ZPE refers to the difference in the zero point energies between a and b. The isotopic effect of the tunneling is considered as the pre-exponential factor . The subset of atoms used to define the Hessian for these KIE calculations ( ( )) /( ( )) were those of the QM region, consistent with the "cut-off rule" and the local nature of isotope effects. 40 The Born−Oppenheimer, rigid-rotor, and harmonic oscillator approximations were considered to independently compute the different contributions. Two set of calculations were carried out, considering the ground state as the BNAH species in aqueous solution, or in the active site of the T-Sav. As previously observed, 41 the results listed in Table S3 show how RM1/MM method is giving lower values of QC KIE than those computed with M06-2X/MM. Finally, as observed in Table  S4, quantum tunnelling effects additionally increase the computed quasi-classical KIEs by 1.29 ± 0.27 folds.