Organocatalytic Control over a Fuel‐Driven Transient‐Esterification Network

Abstract Signal transduction in living systems is the conversion of information into a chemical change, and is the principal process by which cells communicate. In nature, these functions are encoded in non‐equilibrium (bio)chemical reaction networks (CRNs) controlled by enzymes. However, man‐made catalytically controlled networks are rare. We incorporated catalysis into an artificial fuel‐driven out‐of‐equilibrium CRN, where the forward (ester formation) and backward (ester hydrolysis) reactions are controlled by varying the ratio of two organocatalysts: pyridine and imidazole. This catalytic regulation enables full control over ester yield and lifetime. This fuel‐driven strategy was expanded to a responsive polymer system, where transient polymer conformation and aggregation are controlled through fuel and catalyst levels. Altogether, we show that organocatalysis can be used to control a man‐made fuel‐driven system and induce a change in a macromolecular superstructure, as in natural non‐equilibrium systems.

1 Experimental details

General materials and methods
Chemicals were purchased in the highest available purity and used without further purification unless reported otherwise. p-Nitrophenol 1, acetic anhydride 2, pyridine anhydrous, imidazole, isopropenyl acetate, vinyl acetate, poly(acrylic acid) (PAA ~130 kDa), N,N-dimethyl formamide (DMF) anhydrous, dimethyl sulfoxide (DMSO), N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride (EDC), acetonitrile, N,N-diisopropylethylamine (DIPEA), were purchased from Sigma-Aldrich. p-Nitrophenyl acetate 3a was from TCI Europe. 3-nitro-L-tyrosine ethyl ester (NY-ethyl ester) was from Chem Impex International. 1-Hydroxybenzotriazole hydrate (HOBt) was from Acros Organics. Solid salts were used for the preparation of aqueous buffers: sodium tetraborate decahydrate (Borax), boric acid, sodium hydroxide were from Sigma Aldrich and 3-(N-Morpholino)propanesulfonic acid (MOPS) from Alfa Aesar. Unless stated otherwise, all preparations and analyses were performed at room temperature (RT) (~21 ○ C) and atmospheric pressure. Nuclear Magnetic Resonance (NMR) experiments were performed using Agilent-400 MR DD2 (400 MHz for 1 H and 100.5 MHz for 13 C) at 25 ○ C using residual deuterated solvent signals as internal standard. To suppress the water peak, PRESAT configuration (suppress one highest peak) was used. Prior to every diffusion-ordered spectroscopy (DOSY) NMR measurement, scouting experiments were performed to obtain the longest t1 relaxation and get an optimal diffusion gradient and delay. DOSY spectra were analysed with MestReNova software. UV-Vis spectroscopic experiments were carried out using Analytik Jena Specord 250 spectrophotometer; quartz cuvette with a 1 cm path length, volume of 3 mL, at RT. Liquid Chromatography-Mass Spectrometry (LC-MS) was performed on a Shimadzu Liquid Chromatograph Mass Spectrometer 2010, LC-8A pump with a diode array detector SPD-M20. Negative and/ or positive mode Electro Spray Ionization Mass Spectrometry (ESI-MS) was used for the peak assignment. Fourier Transform Infrared Spectroscopy (FTIR) spectroscopy was performed with NicoletTM 6700 FT-IR Spectrometer from Thermo Electron Corporation equipped with OMNIC Software using the ATR method. Spectra were recorded at wavenumber range 4000-400 cm -1 with 4 cm -1 resolution. Prior to each experiment a background of the ZnSe crystal was measured. Dynamic light scattering (DLS) measurements were recorded with a Malvern Zetasizer Nano equipped with a 633 nm laser, collecting the optical data with a 90 ○ scattering angle; quartz cuvette with 1 cm path length, volume of 3 mL, at 20 °C. Cryo-EM (electron cryo-microscopy) images were obtained with a JEOL JEM3200-FSC, operated at 300kV with Gatan camera K2-Summit operated in counting mode with 10s exposure time, dose fractionated 0.2s aligned by SerialEM. Zero-Loss filtered with a 20eV slit. Samples were prepared with a Leica plunger EM GP, 21 °C, 98% RH, 8 seconds blotting time. The final images were analysed with ImageJ software. Polymer viscosity change was measured using a rheometer (AR G2, TA instruments) equipped with a steel plate-and-plate geometry of 40 mm in diameter, using a with hexadecane filled water trap.

UV-Vis assay
Stock solutions were prepared in MOPS buffer (pH 7.5, 100 mM), borate buffer (pH 8.0, 200 mM) or acetonitrile (for acetic anhydride 2 only -to avoid significant hydrolysis in the stock solution). Unless stated otherwise, the fuel cycle was performed with 0.1 mM p-nitrophenol 1, 0.5 mM acetic anhydride 2, 0-1 mM of pyridine, and 0-1 mM of imidazole in MOPs buffer (pH 7.5, 100 mM), in quartz cuvettes, path length of 1 cm (total reaction volume of 3 mL) at RT. The stock solutions of the reactants were always added in the following order: p-nitrophenol 1, pyridine, imidazole and acetic anhydride 2. Teflon caps were used to close the cuvette. The cuvette was turned upside down to mix the solution. The reactant peak was followed using slow time scan, measuring wavelength 400 nm. The pH was measured before and after the reaction (or followed during the reaction; see SI: pH monitoring of fuel-driven esterification network). The conversion was calculated with the extinction coefficients (see SI for calibration lines) and Lambert-Beer law: = , where A is the absorbance, ε the extinction coefficient, l the path length of the cuvette and C the concentration. The experiments with PAANY were performed similarly, only following the decrease/ increase at 420 nm from NY. Because we anticipated problems with concentrated sulfonate buffers with macromolecules (such as precipitation), we used borate buffer (pH 8.0, 200 mM). Experiments were performed at pH 8.0 to have a higher percentage of phenolate (negative charge) compared to phenol and co-solvent was avoided, since it could affect the polyelectrolyte behaviour in solution.

DLS measurements
Stock solutions were prepared in borate buffer (pH 8.0, 200 mM) and filtered with syringe filters (0.2 µm) before use. Unless stated otherwise, the fuel cycle was performed with 0.30 mM PAANY (0.24 mg/mL), 6 mM acetic anhydride 2, 0-30 mM of pyridine, and 0-0.75 mM of imidazole in borate buffer (pH 8.0, 200 mM), in quartz cuvettes, 1 cm path length (3 mL reaction volume) at 20 °C. The stock solutions of the reactants were always added in the following order: PAANY, pyridine, imidazole and acetic anhydride 2. Cuvettes were closed with Teflon caps and turned upside down to mix the solution. A fuel cycle was measured in continuous mode with 1000 measurements consisting of 11 runs each 3 min. An equilibration time of 2 min was applied for each measurement. Calculated size changes (%) are based on z-average diameters (nm). A DLS size calibration was performed to find the polymer concentration with minimal size fluctuation (see SI: DLS size calibration). The pH was always measured before and after the reaction.

Viscosity rheology measurements
For the blank experiments, 0.7 mL of the sample (PAANY (0.3 mM) in borate buffer 200 mM pH 8.0, borate buffer 200 mM pH 8.0 alone or borate buffer with 6 mM acetic anhydride fuel 2) was directly positioned on the rheometer plate. For the PAANY acetylation experiment, the stock solutions of the reactants were added into a glass vial in the following order: PAANY, imidazole, pyridine and acetic anhydride 2. The vial was turned upside down to mix the solution. Then, the 0.7 mL sample (0.3 mM PAANY, 0.75 mM imidazole, 0.3 mM pyridine, 6 mM acetic anhydride 2) was positioned on the rheometer plate. Time sweep measurements were performed at fixed strain (γ = 0.05 %) and frequency (ω =6.28 rad/s = 1 Hz). Flow step measurements were performed after time sweep measurements on the same sample.   Vinyl acetate 1000 molar eq. of fuel (100 mM) not sufficient

Kinetic model for fuel-driven esterification CRN
In this section the development of the numerical model for the reaction kinetics of the esterification CRN written in Matlab 2018b is discussed. First we deal with the kinetics of the blank reaction (uncatalysed), then the pyridine catalysis, followed by imidazole catalysis and eventually all individual reaction schemes are combined to model the entire fuel cycle (Scheme S1) with varying catalysts and fuel concentrations. In all cases, first an overview of the reaction pathway is given, followed by the rate equations (system of ODEs) and the experimental data fitting, showing the concentration profiles of the different species over time for the modelled and experimental data. We end with a note on how the model was optimized for the various experimental conditions.

Rate equations
When only the blank reaction takes place, the following equations apply:

Rate equations
Next, when only the blank reaction takes place and the pyridine catalysis, the following equations apply:

Experimental data fitting
The previous k-values from the blank reaction were again used to determine k cat2 35.0 mM -2 min -1 , k cat5 0.002 mM -1 min -1 and k cat6 1.525 mM -1 min -1 .     Then, the previous k-values were again used to determine k cat3 0.124 mM -2 min -1 and k cat4 0.0565 mM -1 min -1 .

Rate equations
When the blank reaction, pyridine and imidazole catalysis take place the following equations apply:

Experimental data fitting
All the simulated and experimentally determined k-values were again used to fit the esterification cycle with different pyridine, imidazole and acetic anhydride concentrations. In the next figures the fits are provided together with the specific k-values (optimized if needed).

Explanation for model optimization and deviations
The k-values were allowed to be optimised by Matlab using the constrained optimization function (fmincon) with a least squared cost function (opposing lower and upper bounds for the k-values: k min =0.5ˑk 0 and k max =2ˑk 0 ). Yet even with the optimization, these k-values show some deviation (Table S3): The k-values with most variation are related to the anhydride species and a deviation is anticipated, since the acetic anhydride hydrolysis is very rapid [3] and already occurs before the first sample has been measured. Besides, the mechanism for imidazole and pyridine catalysis is in fact more complicated than was assumed in this simplified model and deals with a pre-equilibrium of reactive intermediate formation (acetyl-pyridinium and acetyl-imidazole), acetate ion inhibition and contributions from nucleophilic and general acid/ base catalysis [1b, 6] . Finally, the reactive intermediates for imidazole and pyridine catalysis can also be interconverted, complicating the model.

DOSY diffusion coefficient and size calculation
By measuring the DOSY spectrum of the polymer with NMR and extracting the diffusion coefficient, we can confirm the particle size via the Stokes-Einstein equation: , where D is the diffusion coefficient, K B the Boltzmann constant, T the temperature, µ the viscosity of the bulk medium and d the solute diameter (polymer).