A disulfide-based linker for thiol–norbornene conjugation: formation and cleavage of hydrogels by the use of light

Photolabile groups are the key components of photo-responsive polymers, dynamically tunable materials with multiple applications in materials and life sciences. They usually consist of a chromophore and a labile bond and are inherently light sensitive. An exception are disulfides, simple reversible linkages, which become photocleavable upon addition of a photoinitiator. Despite their practical features, disulfides are rarely utilized due to their impractical formation. Here, we report a disulfide-based linker series bearing norbornene terminals for facile crosslinking of thiol-functionalized macromers via light-triggered thiol–ene conjugation (TEC). Besides finding a highly reactive lead compound, we also identify an unexpected TEC-retardation, strongly dependent on the molecular linker structure and affecting hydrogel stability. Finally, we present a useful method for localized disulfide cleavage by two-photon irradiation permitting micropatterning of disulfide-crosslinked networks.


MATERIALS
All photosensitive steps were performed under light protection. Unless otherwise stated, chemicals were purchased from Sigma-Aldrich, TCI Europe or ABCR GmbH and used without further purification. Thiolterminated 8-armed PEG (8armPEG20k-SH, Mw ~20 kDa, tripentaerythritol core) was purchased from

INSTRUMENTATION
NMR spectra were recorded at 600 MHz for 1 H and 150 MHz for 13 C on a Bruker Avance III HD spectrometer and at 400 MHz for 1H and 100 MHz for 13  Column chromatography was performed with a Büchi MPLC-system equipped with the control unit C-620, fraction collector C-660, and UV-photometer C-635 (BÜCHI Labortechnik AG, Flawil, Switzerland).
Thereafter, trimethylamine (6.25 mL, 48 mmol, 4 eq) was added slowly and the reaction was stirred at room temperature. After 14 h of stirring, 1 H-NMR showed 80% consumption of starting materials. Hence, the reaction mixture was refluxed for 3 h to increase the conversion. The solvent was removed in vacuum and the white residue was dissolved in sat. aq. NaHCO3 solution (50 mL) to give a "clear" solution. The product was precipitated using conc. HCl giving a sticky white solid. The aqueous phase was extracted with diethyl ether (3 x 200 mL) and the combined organic layers were washed with brine.
The organic phase was dried over Na2SO4 and the solvent was removed in vacuum yielding a whitish oil. The raw product was purified by column chromatography (110 g silica) using DEE: ethyl acetate = 50% as eluent. To remove residual monosubstituted derivative, it was recrystallized from toluene twice yielding a white solid (4.23 g, 79%).

General procedure for the preparation of (homo)cystine-based linkers
The general procedure is roughly based on a modified synthetic procedure. 5 Cystine or Homocystine (1 eq) and finely ground NaOH (4 eq) were added to a one-neck flask and purged with Ar. Dry MeOH was added while stirring until a clear solution was obtained. Thereafter, the stirred reaction mixture was cooled with an ice bath and freshly distilled norbornene acid chloride (2 eq  For conversion into the Na salt, Nor-L-Cys (1425 mg, 3.0 mmol, 1 eq) was dissolved in MeOH (20 mL) and aqueous NaHCO3 solution (1 M, 6.0 mL, 6.0 mmol, 2 eq) was added. The mixture was stirred for 15 min and then dried in vacuum yielding 1555 mg (quant.) of a beige powder.    For conversion into the Na salt, Nor-D,L-Cys (240 mg, 0.50 mmol, 1 eq) was dissolved in MeOH (2 mL) and aqueous NaHCO3 solution (1 M, 1.00 mL, 2 eq) was added. The mixture was stirred for 15 min and then dried in vacuum yielding 234 mg (89%) of a beige powder.
MeOH was removed under reduced pressure. The obtained residue was dissolved in deionized H2O (100 mL) and washed with DCM (2 x 50 mL). The aqueous layer was acidified with HCl (1 M, 7.5 mL) causing precipitation of a white solid. The aqueous phase was extracted with ethyl acetate (4 x 50 mL).
The organic phase was washed with brine (50 mL) and dried with Na2SO4. Solvents were removed in vacuum and the residue was dissolved in MeOH and filtered through a nylon syringe. After drying in high vacuum, Nor-L-HCys (225 mg, 60%) was obtained as a white foam. For conversion into the Na salt, Nor-L-HCys (222 mg) was dissolved in MeOH (3 mL) and aqueous NaHCO3 solution (1 M, 873 µL, 2 eq) was added. It was stirred for 10 min before the solvents were removed in vacuum yielding a white foam (243 mg, quant.).

Synthesis of disodium 1,1'-dimethyl-N,N'-bis(bicyclo[2.2.1]hept-5-ene-2-ylcarbonyl)-L-cystine ester, Nor-L-CysMe
The synthesis of Nor-L-CysMe is roughly based on a modified synthetic procedure. 6 L-Cystine dimethyl ester dihydrochloride (1195 mg, 3.5 mmol, 1.0 eq) and triethylamine (2.0 mL, 14.7 mmol, 4.2 eq) were suspended in dry DCM (30 mL) in a round bottom flask equipped with a septum. The mixture was cooled to 0 °C and NorCOCl (1.1 mL, 7.7 mmol, 2.2 eq) was added dropwise using a syringe. After complete addition, the ice bath was removed and the reaction mixture was stirred at room temperature. The progress of the reaction was monitored by 1 H-NMR. After 5.5 h the reaction mixture was diluted with ethyl acetate (100 mL) and extracted with saturated sodium bicarbonate solution (50 mL), water (50 mL) and brine (50 mL). The organic phase was dried with Na2SO4 and the solvent was removed in vacuum. The raw product (solid foam) was purified by column chromatography using light pertol:ethyl acetate (50-60%) as eluent. Two fractions were received of which one contained both exo-and endo-norbornene based products (514 mg, 29%) and the other only the endo-norbornene product (1004 mg, 56%). The white liquids were dried at HV while being carefully heated with warm air.
A solution of carbic anhydride (1220 mg, 7.4 mmol, 2 eq) in dry THF (15 mL) was added to the stirred solution. Thereafter, trimethylamine (1.55 mL, 11.1 mmol, 3 eq) was added slowly and the reaction was stirred at room temperature overnight. After 14 h 1 H-NMR showed 80% consumption of starting materials. Hence, the reaction mixture was refluxed for 3 h to increase the conversion. Then the reaction mixture was dried in vacuum before it was dissolved in sat. aq. NaHCO3 solution (50 mL) to give a "clear" solution. The product was precipitated using conc. HCl giving a sticky white solid. Ethyl acetate (100 mL) was added to dissolve the product. After phase separation the aqueous layer was extracted with more ethyl acetate (3x 50 mL) and the combined organic layers were washed with brine, dried over  In experiments conducted in D2O, the signal of the proton at the α-carbon (CH-NH, 2H) served as internal reference for calibration, whereas in DMSO-d6 the amide-proton (NH, 2H) was used.

Linker reactivity estimation by 1 H-NMR in DMSO-d6.
Experiments were conducted similar to the measurements in D2O. Here, linkers were reacted with 2-mercaptoethanol at a functional group ratio of 1:1 in DMSO-d6 by irradiation at 400-500 nm (20 mW cm -2 ) in presence of TPO-L (ethyl (2,4,6-trimethylbenzoyl)phenylphosphinate). between measuring tool and glass plate utilizing capillary force, before the tool was lowered into the measuring position at a gap of 0.8 mm. The formulations were sheered with a strain of 1% and a frequency of 1 Hz. After 1 min of equilibration, photocuring of the hydrogel samples was induced by UVirradiation projected via a waveguide from the underside of the glass plate using an OmniCure® LX400 LED UV Spot Curing System with a 385 nm LED head (Excelitas Technologies) with a specific power of ~6 mW cm -2 at the measuring platform. 7 All measurements were performed at least in triplicates, except for experiments in which a high amount of photoinitiator was used or the thermal or mechanical stability of the hydrogel was investigated. Error bars are illustrating the standard deviation in graphs.

Light-induced hydrogel formation was investigated with selected linkers
The formed hydrogels disks were utilized for subsequent swelling tests in PBS.

Hydrogel swelling experiments.
For swelling experiments, the hydrogel disks, produced in the course of photorheology measurements, were carefully removed from the photorheometer measuring platform and placed in individual tared petri dishes. After initial weighing of the samples, they were swollen in PBS (~2 mL). For weighing, the liquid was pipetted off, and the hydrogel disks were carefully patted dry with paper towel. Thereafter, the disks were submersed in fresh PBS again. All swelling experiments were performed in triplicates.

Thermostability of 8arm-D,L-HCys 100 hydrogel
An aliquot (45 µL) of 8arm-D,L-HCys 100 prepolymer solution (containing 0.4 mM LITPO) was in situ photopolymerized at 20 °C on the photorheometer at the settings described in section 6.3.2.0, before the temperature was continuously increased up to 90°C at a heating rate of 1 °C/min. To prevent evaporation of water, the measuring system was sealed with a ring of paraffin.

Mechanostability of 8arm-D,L-HCys 100 hydrogel
For the amplitude sweep, after in situ photo-polymerization of 8arm-D,L-HCys 100 prepolymer solution, the amplitude of the applied strain was logarithmically increased from 0.1% to 1000% at a constant frequency of 1 Hz. For the dynamic strain amplitude cycling experiment, after in situ polymerization, a series of four strain cycles was applied varying strains between 350% and 1% in 60 s periods.

Stabilization of 8arm D,L-HCys 100 hydrogel by a non-cleavable background network
Several prepolymer solutions were prepared from stock solutions of 8armPEG20k-SH ( Figure S20 and S21 indicate that the presence of the disulfide is triggering the elimination reaction, which occurs most probably due to a neighboring group effect of the disulfide in combination with the free carboxylic acid in close proximity to the ester. 9       hydrogel was deforming and swelling heavily leading finally to disintegration within a few days. On the contrary, when the hydrogel was immersed in 10x PBS in the initial 20 h, samples were first losing mass due to osmosis. After changing to normal PBS, samples were swelling as well, but remained their shape and did not dissolve anymore. These samples were monitored for more than a month but were stable much longer.   were not stabilized, using 20 mol% excess of ene successfully preserved long time hydrogel integrity. Hence, an ene-excess of 20% appears to be the optimum for this formulation (8arm-D,L-HCys 120).
This experiment is of particular interest, since it shows, that Nor-D,L-HCys does not fully react when applied at an 1:1 thiol-ene ration. Consequently, an excess of ene is required here to consume unreacted thiols and even increase the crosslinking density.   or FITC500 for one week. In both cases microchannels could not be visualized due to insufficient cleavage for the FITC-dextran to enter. When FITC500 was used, the openings of channels micropatterned at laser powers above 40 mW were visible. At lower laser powers or when FITC2000 was used, a bulge could be observed at the edge of the scanning area indicating local swelling. Scale bars = 20 µm.

COMPUTATIONAL STUDY
To get a better insight into the cause of this TEC retardation and further evaluate the structure-property relationship, the reaction of the retarded linker Nor-L-Cys and the elongated, highly-reactive Nor-L-HCys with a 2-mercatoethanol thiyl radical was computationally modelled. The attack of the thiyl radical on the acids A and B (Scheme S1) was chosen as a simplified model for the investigations.
Scheme S1. The systems chosen for the computational studies of the reaction with the thiyl radical. The acids A and B reflect the systems Nor-L-Cys and Nor-L-HCys.
The thiyl radical can attack the double bond of the norbornene fragment yielding two different radical products p1 and p2. Moreover, a new chiral center appears simultaneously with the formation of the new C-S bond. Thus, for each of the systems A and B four possible radical products have to be taken into account: p1_R, p1_S, p2_R and p2_S. Scheme S2 illustrates the computed reaction pathways for system A. Figure S34 presents the computed free energy profiles for the aforementioned reaction pathways for both systems A and B in comparison.
Since the studied systems are highly flexible conformationally, the allowed conformational space was carefully explored. The free energies of the reactants, transition states, and products depicted in Figure   S34 correspond to the most favorable conformations. One can see that system B leads to the most stabilized thermodynamically product B_p2_R, with G = -4.2 kcal mol -1 . The energetically most favorable product in the case of system A is A_p2_S, with G = -1.7 kcal mol -1 . Thus, the thermodynamic driving force of the computationally modeled event is 2.5 times larger for system B as

Computational details
The conformational space of all flexible molecules has been initially searched using the OPLS_2005 force field 12 and the systematic Monte Carlo conformers search routine implemented in MACROMODEL 11.5. 13 Accordingly, the structures located at force field level have then been subjected to (U)B3LYP-D3/def2-TZVP 14-19 geometry optimization. The nature of all stationary points (minima and transition states) was verified through the computation of the vibrational frequencies. Gibbs free energies (G298) at 298.15 K were evaluated using a quasi-RRHO treatment of vibrational frequencies and GoodVibes program. 20,21 All energies are reported in kcal mol -1 .
The density-based solvation model SMD 22