A molecular assembler that produces polymers

Molecular nanotechnology is a rapidly developing field, and tremendous progress has been made in developing synthetic molecular machines. One long-sought after nanotechnology is systems able to achieve the assembly-line like production of molecules. Here we report the discovery of a rudimentary synthetic molecular assembler that produces polymers. The molecular assembler is a supramolecular aggregate of bifunctional surfactants produced by the reaction of two phase-separated reactants. Initially self-reproduction of the bifunctional surfactants is observed, but once it reaches a critical concentration the assembler starts to produce polymers instead of supramolecular aggregates. The polymer size can be controlled by adjusting temperature, reaction time, or introducing a capping agent. There has been considerable debate about molecular assemblers in the context of nanotechnology, our demonstration that primitive assemblers may arise from simple phase separated reactants may provide a new direction for the design of functional supramolecular systems.


Polymerization experiments
This section describes the conditions used in experiments involving the molecular assembler. The standard conditions used for figure 1 are described in the general experiment. For the experiments of figure 4, the same procedure was used, only changing the amount of compound 1 at the start of the experiment. Concentrations of compound 3 were determined by sampling the reaction mixture (see Sampling of the solution) and consequent analysis of these samples using UPLC. Experiments studying the growth of polymers (figure 3) were carried out in an identical way, now sampling the polymers instead of the aqeous layer (see Sampling and analysis of the polymer product). Polymer were analysed using both GPC and NMR.
General experiment: In a typical experiment, a total of 80 mg compound 1 (0.2 mmol), 4 mL, 0.5M TRIS buffer (pH 8.00) and a small stirring magnet (0.5 cm) were added to a 7mL flat vial. The solutions were stirred at 200 rpm (unless stated otherwise) and the vial was heated to the required temperature (usually 40 ˚C). After compound 1 had completely dissolved, 27 µl compound 2 (23 mg, 0.09 mmol) was carefully added on top of the water layer as an oil (by heating it to ~40˚C), and stirring was continued.
Sampling of the solution: Samples were taking regularly, by carefully extracting 20 µL of the aqeous layer. The extracted solution was immediately added to 1 mL of a 62 mM aqeous solution of maleimide (containing 0.16 mM 3-methyl-2-nitrobenzoic acid as an internal reference) to quench all remaining thiols. Samples were analyzed as described in the UPLC section.
Sampling and analysis of the polymer product: Formed polymers were present as a light yellow solid that floated on top of the water layer upon precipitation. Some solid was carefully extracted using a spatula and subsequently dried using paper tissue. The average chainlength was determined based on the thiol to disulfide ratio, that could be accurately determined using 1 NMR spectroscopy (see NMR section and Supplementary figure 1).
Supplementary figure 1 Effect of the temperature on polymer growth. Evolution of the average chain length of disulfide polymers, formed from the biphasic reaction between compounds 1 (25 mM at pH 8) and 2 at various temperatures.
Additional analysis of the polymer chainlength using mass spectrometry (using either ESI with QM detection, ESI with TOF detection or MALDI) proved unsuccesful. Further analysis and confirmation of NMR data was therefore achieved using gel permeation chromatography (GPC), using polystyrene as a reference.

Depolymerization:
A total of 17 mg polymer and 88 mg racemic dithiothreitol (8 eq) was dissolved in 4 mL THF. The clear solution was stirred for 72 hours, after which the solvent was reduced. The mixture was run over a short silica column (100% hexanes) to yield the depolymerized product (1,12-dodecanedithiol) as a colourless solid (15 mg, 88%). The resultant monomer was analyzed using both NMR and GPC.

Seeding experiments
Experiments were performed as described in section 1, but now the reaction was seeded with compound 3 before addition of thiol 2. Addition of small amounts of 3 (<0.5 mM) resulted in elimination of the lag period, while moving towards the same equilibrium concentration. Addition of higher concentrations of 3 resulted in a decrease of 3, although a period of constant [3] was never observed (see supplementary figure 1). The autocatalytic nature of the reaction was further established by following the initial rate of formation of compound 4 as a function of seed concentration

pH depedent experiments
We aimed to determine how the molecular assembler responds to changes in its environment, specifically how the concentration of 3 is influenced by various reaction conditions. Whereas the temperature simply sped up the lifecycle of [3] (but did not affect its concentration profile), the initial concentration of 1 is a key parameter (see Fig 4 of main manuscript). In addition the effect of pH on [3] was investigated (Supplementary figure 4).

Ring tensiometry measurements
Ring tensiometry measurements were used to calculate the critical micelle concentration (CMC) of compound 3. Surface tension is plotted against ln [3], whereby the point from which the surface tension no longer decreases corresponds with the CMC. This can be calculated using the intercept of the two drawn lines (

GPC measurements
GPC analysis of polydisulfides was performed on a Shimadzu LC-20AD instrument, equipped with a Refractive Index (RI) detector and two PSS SDV 5 μm linear M columns. HPLC grade THF was used as the eluent at 1.0 mL/min at 30 °C. Samples were passed through 0.2 μm PTFE filters prior to analysis. Monodisperse polystyrene standards were used for calibration. Number average molar mass (Mn), weight average molar mass (Mw), and dispersity (Ɖ) were calculated using Shimadzu LabSolutions GPC analysis program. Pure compound 2 (monomer) and three samples obtained at consecutive time points in the polymerization experiment were analysed using both 1 H NMR and GPC. In addition, a polymer sample of the final time point was depolymerized and subsequently analyzed using NMR and GPC to show conversion to the monomer. Comparison between the two methods, as well as the unmodified GPC data, are given in supplementary table 1. In all cases, good agreement between both analysis methods was obtained.

Dynamic Light Scattering measurements
Analyses were performed using a Malvern Zetasizer Nano ZEN5600 model system recording particle and molecule size. Instrument control and data processing were performed using Zetasizer software. Disposable plastic cuvettes were used with 1.0 mL of sample solution. Measurements were repeated thrice for every concentration. Measurements were done using an equilibrated heating probe at 60 °C, setting the appropriate parameters for water. All samples were prepared in a 0.5 mM TRIS buffer (pH 8.00) and were filtered using a microfilter (poresize 0.22 µm) prior to measurement.
Measurements on solution containing compound 3 gave a maximum intensity for a particle size of around 7 nm at Ph 8.0, regardless of the concentration.

Interferometric scattering microscopy (iSCAT) measurements
The iSCAT experimental set-up is similar to that described by Young et al, 1 with a 532 nm diode laser used as the incident light source. Frames were recorded at 1 kHz with an exposure time of 0.98 ms, using a CMOS camera. Focus in the z axis is maintained using an autofocus system relying on the total internal reflection (TIRF) of a 638 nm beam. Instrument control was performed using the custom software written in LabView.
Data processing. Data processing was performed using the custom software written in Python, as described elsewhere. 1 In brief, differential imaging was achieved by subtracting sets of images temporally offset by a time Δt. The signal-to-noise ratio was then improved by spatially (3 x 3 binning) and temporally averaging the differential images (50 images).
Particle detection was performed as described by Young et al. 16 Briefly, diffraction-limited spots were identified by the software, and fitted to the 2D Gaussian function to give the ratiometric contrast value.
Coverslips and sample preparation for iSCAT analysis. Samples for iSCAT analysis were prepared in a TRIS buffer (pH 8), using a 0.004 mM concentration of compound 3.
Glass coverslips (no. 1.5, 24 x 50 mm, VWR; and 24 x 24 mm, VWR) were cleaned by sequential sonication in MilliQ water, isopropanol and MilliQ water (5 min each), and dried under stream of nitrogen. Chambers were prepared by attaching a (24 x 24) coverslip on top of a (24 x 50 mm) one with double-sided tape. All coverslips and chambers were prepared on the day of the analysis or the day before.
Contrast-to-mass (C2M) calibration was performed in the corresponding buffer solution, since the C2M conversion may change slightly as a result of buffer content. The calibration protocol included measurement of a protein oligomer solution, with masses of 90, 180, 360 and 540 kDa. Each calibration experiment was analysed using same software as described above. The mean peak contrast was determined in the software using Gaussian fitting. The mean contrast values from the calibration protein solution were then plotted (Supplementary figure 10) and fitted to a line, ‫,ݔܾ=ݕ‬ with ‫-ݕ‬contrast , ‫-ݔ‬mass and ܾ-C2M calibration factor.
Using the result of this calibration (b = -43674) and the average ratiometric contrast (-0.00461) an average particle mass of 201 kDa was calculated. Based on this number and the molar mass of compound 3 (629 Dalton), an aggregation number of 320 can be calculated. In addition, by assuming spherical particles with the same density as water, a diameter of 8.6 nm could be calculated, in good agreement with DLS data.
Several iSCAT images of the molecular aggregates are given on the next pages.

Effect of the chainlength of the alkylthiol
To investigate the effect of the chain length of the alkylthiol on the kinetics of the assembler's formation as well as the polymerization process, an additional alkyldithiol was synthesised (1,20eicosanedithiol). We found that the surfactant synthesised from this dithiol (3b) aggregated in the form of vesicles instead of micelles (see Supplementary figure 11-18). While this surfactant can be synthesised, no detectable concentration could ever be measured during the biphasic (assembler) experiments. Polymerization did proceed under these conditions, but at a much slower rate than for experiments involving the (12 carbon) assembler. Seeding the experiments with neatly synthesised surfactant 3b, resulted in fast disappearance of 3b, while having no effect on the polymerization speed. The polymerization process for this elongated dithiol was notably slow, requiring an increase in temperature of 25 degrees (from 40 to 65°C) to achieve a same rate of polymerization.

Synthesis of compounds
General: All commercial chemicals were purchased from Sigma-Aldrich and were used without further purification. NMR spectra were recorded on a Bruker Topspin 400 (400 MHz) spectrometer in CDCl3 (unless otherwise reported). Chemical shifts are given in ppm with respect to tetramethylsilane (TMS) as internal standard. Coupling constants are reported as J-values in Hz. Column chromatography was carried out using Acros silica gel (43-60 μm). Both 2 and 2a has been previously described, but were prepared using a different synthesis strategy, 2 while 3 had not been previously described. Synthesis procedures for the dithiols described here are based on the methods described by Hasegawa et al. 3 Obtained NMR spectra for previously characterized compounds were in agreement with literature.

UPLC measurements and calibration
The concentration of UV active components of the reaction were monitored using a Waters Acquity ultra performance liquid chromatography UPLC H-Class system with photodiode array (PDA) detector. Instrument control and data processing were performed using Empower software. An Acquity UPLC BEH C18 column (130 Å, 1.7 μm, 2.1mm× 50 mm) was used. A mixture of 2O:MeCN:5% TFA in H2O with a gradient of 93:2:5 → 0:95:5 over 5 min was used as mobile phase. Peak areas were integrated at a wavelength of 330 nm. Sample preparation as well as calibration for compounds 1 and 4 has been previously described by Morrow et al. 5 The concentration calibration for compound 3 is given in figure Supplementary figure 19.

Polymerization mechanism
This secton describes additional experiments, carried out to provide further support for the proposed polymerization mechanism in the main manuscript.
Activity of the assembler: To further establish the activity of the assembler (composed of 3) for the production of polymers, additional experiments were carried out. In these experiments, micelles of 3 were actively removed from to reaction mixture. If the micelles of 3 act as the active polymer assembler (and not some background reaction), this should reduce the degree of polymerization. To remove the micelles, after eight hours, the reaction mixture was cooled to 0˚C after which the apolar layer solidified. The aqeous layer (containing the micelles) was removed, and replaced with a fresh buffer solution containing 1. This action was repeated after 24 hours. After 32 hours, the apolar layer was analyzed using NMR, to determine the amount of conversion of thiol 2 to disulfide. This value was then compared to the reference experiments, to which the same cooling procedure was applied, but for which the aqeous layer (containing the micelles) was not replaced. The observed conversion to thiol was: Experiment: 18 ± 5 % Reference: 36 ± 2 % Removal of the micelles of 3 thus results in a lesser degree of polymerization. Note that during the periods inbetween aqeous layer replacement, the amount of 3 had been (almost) restored to its equilibrium value. This implies that even though the amount of polymer assembly by the micelles of 3 was significantly reduced, a sufficient (average) concentration remained to explain the 18% conversion in these experiments.
Polymer growth after depletion of compound 3: Upon depletion of thiol 2, no more surfactant 3 can be formed. Continuous destruction then results in a decrease in the concentration of 3, until no more 3 remains. At 40˚C and pH 8.0, it takes approximately 4-5 days before 3 has completely disappeared (main manuscript, Fig 2), but polymer growth continues even after this period (main manuscript , Fig 3a/b). The remainder of the thiol groups is then converted via a different mechanism, but at a slower rate. We assume that further alkyl thiol conversion then proceeds at the interface of the water and thiol/disulfide layers, since no significant amount of any surfactant can be detected by UPLC any longer.

NMR analysis of polymer size:
The average chain length of the produced polymers can be determined based on the thiol to disulfide ratio (assuming negligible cyclized products). Since the SCH2 proton signals of the thiol and disulfide are sufficiently separated in 1 H NMR, the ratio of these peaks was used to determine the average chainlength: Disulfide: δ 2.68 (t, J= 7.5 Hz) Thiol: δ 2.52 (q, J= 7.3 Hz) Supplementary figure 23 1 H NMR of disulfide polymers made from 2.