Emergence of Function and Selection from Recursively Programmed Polymerisation Reactions in Mineral Environments†

Abstract Living systems are characterised by an ability to sustain chemical reaction networks far‐from‐equilibrium. It is likely that life first arose through a process of continual disruption of equilibrium states in recursive reaction networks, driven by periodic environmental changes. Herein, we report the emergence of proto‐enzymatic function from recursive polymerisation reactions using amino acids and glycolic acid. Reactions were kept out of equilibrium by diluting products 9:1 in fresh starting solution at the end of each recursive cycle, and the development of complex high molecular weight species is explored using a new metric, the Mass Index, which allows the complexity of the system to be explored as a function of cycle. This process was carried out on a range of different mineral environments. We explored the hypothesis that disrupting equilibrium via recursive cycling imposes a selection pressure and subsequent boundary conditions on products. After just four reaction cycles, product mixtures from recursive reactions exhibit greater catalytic activity and truncation of product space towards higher‐molecular‐weight species compared to non‐recursive controls.

, 30mM L-glutamic acid (Sigma,, 30 mM L-lysine (Sigma, CAS: 56-87-1) and 100 mM glycolic acid (Sigma, CAS: 79-14-1) in HPLC-grade H2O were adjusted to pH 2.5 using H3PO4 and heated at 90 °C for 15 hours in open cap glass vials on a 75 vial insert heating slab ( Figure S12). For recursive mineral reactions, 150 mg of solid mineral matrix was added to the vials at the beginning of the first reaction cycle. After 15 hours of heating, the reaction solutions had completely evaporated, and products were re-suspended in 10 ml HPLC-grade H2O. Re-suspended product was vortexed, and 1 ml of product solution was transferred to 9 ml fresh reagent solution and 135 mg fresh solid matrix in a fresh vial. Solutions were re-adjusted to pH 2.5, and the process was repeated for each subsequent cycle. For non-recursive control reactions, the first cycle was carried out as in the recursive reactions; however, products were not transferred to fresh feedstocks and no further solid matrix material was added to the mixture. Montmorillonite, gypsum, quartz, calcite, chalcopyrite, opal and kernite solid matrix materials were sourced from Richard Tayler minerals. Solids were crushed and passed through a sieve with a 3 μm cut-off prior to addition to the reaction. Crushed glass was generated using the same process from the same glass vials that were used for all reactions.

Mass Spectrometry Data Acquisition
Product mixtures were diluted 1:100 in HPLC-MS grade H2O, filtered through a nylon membrane with 0.2 μm pore size into glass vials (#2-SVW8-CPK, ThermoScientific) and loaded on to an autosampler (#WPS-3000TRS, ThermoScientific) hooked up to a quaternary pump (#LPG-3400RS, ThermoScientific). Samples were injected in 10 μl aliquots into a Bruker Maxis Impact II in a 1 ml min -1 flow of HPLC-MS grade H2O + 0.1% formic acid (Sigma,. Measurements were taken in positive ion mode, with the instrument calibrated to a range of 50-1200 m/z using sodium formate calibrant solution. Voltage of the capillary tip was set to 4800 V, end plate offset at -500 V, funnel 1 RF and funnel 2 RF at 400 Vpp, hexapole RF at 100 Vpp, ion energy at 5.0 eV, collision energy at 5 eV, collision cell RF at 200 Vpp, transfer time at 100.0 μs and pre-pulse storage time at 1.0 μs.

Depsipeptide Library Screening and Mass Index
The complete mass list of 6363 depsipeptide products was compiled in Python 3.6. All possible compositions arising from leucine (L), glutamic acid (E), lysine (K) and glycolic acid (g) were calculated from a dictionary of monomer masses read from a .json file. Branched and cyclic products were screened for by removing one water mass (18.01056 amu) for every two proposed branching points. Cationic adducts of each branched, cyclic and linear product were accounted for by adding the following masses: 1.007276 (H + ), 22.989 (Na + ), 38.963 (K + ), 18.034 (NH4 + ). The mass list was compiled as a Python dictionary, with keys corresponding to a string of monomer units plus dehydrations and adducts. Dehydrations were added in the string format "-<n> H2O" and adducts were added in the string format "+ <i>", where n and i correspond to number of dehydrations and cationic adduct, respectively. For example, "LEK -1 H2O + Na" corresponds to a trimer of one L, one E and one K monomer with one extra dehydration (a potential additional cycling or branching point) plus a sodium adduct in +1 charge state.
Mass lists in .csv format were read by a script written in R 2.7 and run in R-studio. Extracted ion chromatograms (EICs) for each mass were extracted from MS-1 data stored in 32-bit mzml files, which were generated from raw data using Proteowizard MS Convert. Masses were extracted with an absolute error threshold of + / -0.01. Total intensities for each EIC were stored in csv files. The first column of each csv file contained a list of depsipeptide composition strings, each subsequent column contained corresponding intensities for depsipeptide products for each sample. A small example data set is presented in Table S1. A noise threshold of 4.55% of maximum intensity was applied to this data. Noise filtered data was used for the Mass Index measurement. The Mass Index was used as a metric for assessing the effect of recursion on the reaction carried out in nine mineral environments ( Figure S3). 2. For each composition product, addition of the intensities for each adduct.
3. For each composition string, calculation of the relative ratio of each monomer string ("L", "E", "K", "g") as a decimal fraction, excluding extra string characters for dehydration and whitespace.
4. Multiplication of each of these fractions by the total intensity measured for the composition product.
5. Addition of total monomer intensities over the entire library of 6363 compositions.
6. For each total monomer intensity, division by the total product intensity and multiplication by 100 to obtain % monomer intensity.
An example of csv data input into the monomer composition script is presented in Table S2.

Circular Dichroism
Circular dichroism (CD) was used to screen for secondary structures in recursive product mixtures. CD spectra of product mixtures were taken after cycles 1, 3 and 5, and compared to unreacted starting material.
Product mixtures were diluted 1:1 in pH 8.0 sodium phosphate buffer (93 mM Na2HPO4, 7 mM NaH2PO4) in HPLC-grade H2O, vortexed and sonicated at 45 °C for approximately 15 minutes, before chilling at 4 °C overnight. Immediately prior to CD measurements, buffered product mixtures were diluted 1:20 in fresh pH 8 buffer solution. CD measurements were taken at room temperature using a Jasco J-810 spectropolarimeter in a 0.2 cm path with a data pitch of 0.1 nm, in continuous scanning mode at a scan speed of 100 nm min -1 , 2 second response time, accumulation of 1 and bandwidth of 1 nm. Spectra were recorded from 190-400 nm.
Data in Supplementary Fig. 10 show some form of secondary structure -possibly β-sheets, which are common for lysine-rich sequences and also hydrophobic peptides.

Fourier Transform Infrared Spectroscopy
FTIR measurements ( Supplementary Fig. 11) were taken to confirm the presence of secondary structures measured by CD.

Minerals
Minerals were sourced from Richard Tayler minerals. Composition, structure and water solubility are given below in Table S3.

Wet-Dry Cycling Reactions
Wet-dry cycling reactions were carried out in 15 ml glass vials placed on to 75-well heating slabs ( Supplementary Fig. 12). The temperature of the heating slabs was set to and maintained at 90 °C via a custom PID controller designed and built in-house, the ThermoShield. Fine power control of resistive loads such as the silicone heating mat employed in this project is achieved via phase angle control utilizing a TRIAC. An EPCOS B57560G104F NTC Thermistor submerged in a vial filled with mineral oil (CAS: 8042- Gerber files, Bill of Materials, Arduino sketch and documentation can be found on github: http://datalore.chem.gla.ac.uk/Origins/ThermoShield.git Figure S12: Temperature Control System for Wet-Dry Cycling Reactions. Thermistor was placed in a vial of mineral oil in the centre of the heating slab. Temperature was continuously monitored throughout the 15 hour heating cycles, through Termite and manually using a standard analogue thermometer inserted into the vial of mineral oil.