Assessment of photoswitching of a thioindigo bismethacrylate crosslinker
Our approach for the thioindigo linker was designed such that the reagent contains methacrylate function that can participate in conventional free-radical polymerization. Thus, we developed a facile protocol for gram-scale preparation of a bismethacrylate thioindigo (trans-1, Fig. 2a) with an overall yield of 15.4%, and with simple purification procedure that does not require column chromatography (refer to SI, Section 2.1 for the synthesis procedure and characterisation of trans-1). Despite having the methacrylate and ester substitution on the aromatic rings, thioindigo 1 is insoluble in all polar organic solvents including methanol, ethanol, dimethyl sulfoxide (DMSO), and dimethylformamide (DMF). When screening conventional organic solvents, we found that chloroform is the only solvent that dissolves 1 at concentrations up to 5 mM (3 mg·mL− 1), allowing for UV-Vis absorption and routine NMR analysis. The UV-Vis absorption spectrum of 1 displays a peak at 540 nm, similar to that of the previously reported thioindigos (Fig. 2b). Irradiation of solution 1 at λmax = 540 nm results in the decrease of the 540 nm absorbance and the appearance of the blue-shifted peak at 490 nm, with an isosbestic point at 503 nm, indicating the conversion to the cis-isomer. Irradiation of the cis-isomer solution by λmax = 490 nm resulted in the reversion of the absorbance to the trans- species. The UV-Vis absorbances of trans- and cis-thioindigos are better resolved than isomers of azobenzene,42 for the reasons explained above.24, 32 The reversible trans-/cis-photoswitching can be repeated for more than 10 cycles without any observable changes in the absorbance spectra of both isomers (Fig. 2c), indicating that no significant photodegradation of thioindigo 1 occurs in CHCl3 by alternating irradiation of green (540 nm) and blue (490 nm) wavelengths.
To determine the optimum wavelength for the photoisomerization that affords the strongest enrichment of each isomer, we employed a nanosecond pulsed, tuneable optical parametric oscillator (OPO) laser coupled with a UV-Vis spectrometer to screen the activation wavelength. By tracking the in situ absorbance measurements during irradiation, we were able to determine the rate of isomerisation, as well as the photostationary cis-/trans-ratio at discrete wavelengths in a photochemical action plot.41 We start by comparing the wavelength dependent switching efficiency to the absorption spectrum. The action plots of trans- to cis- and cis- to trans-isomerisation are depicted in Fig. 2d as green circles and blue squares, respectively. As one would intuitively expect, the highest switching rate for cis- to trans-isomerisation is observed in the wavelength regions where the absorption of the cis-1 is at a maximum. This is, however, not the case for the trans- to cis-isomerisation, where the fastest switching rate is observed at 500 nm, which is 30 nm blue-shifted from the absorption maximum. Photoinduced switching was found to persist up to λmax = 600 nm (orange colour), a region where thioindigo 1 shows minimal absorbance. These results alone, however, are not sufficient to inform the best wavelengths for photoisomerization. It is important to also consider the cis-/trans-ratio presented in Fig. 2e. For trans- to cis-isomerisation, the fastest switching rate was observed at 500 nm, yet this wavelength only produces a cis-/trans- ratio of 30% (blue circles, Fig. 2e). A significantly higher ratio of 75% can be achieved with longer wavelengths close to 550 nm (15 nm red-shifted from the absorption maximum), without significant reduction in switching rate. Notably, almost no trans- to cis-isomerization occurred under in the mild ultraviolet regime (350–400 nm). For cis- to trans-isomerisation, high trans-/cis- ratios (green squares, Fig. 2e) are observed in the wavelength region between 440–470 nm, which is blue-shifted relative to the absorption maximum. Taking into account both switching rates and isomer ratios, it was determined that the ideal wavelengths for trans- to cis-isomerisation and cis- to trans-isomerisation are 540–550 nm and 450–470 nm, respectively.
The cis-form of thioindigo 1 was observed to spontaneously revert to trans-isomer at ambient temperature (ca. 24 ºC), with a thermal recovery rate of 4.7·10− 4 s− 1 (inset Fig. 2b), similar to those of other previously reported thioindigos in aprotic solvents.31, 40 The relatively fast thermal isomerisation prevented us from determining the isomer ratios by 1H NMR analysis and obtaining a clear spectrum of the cis- isomer, since a 10–20 min interval is required before an NMR spectrum can be recorded post-irradiation. In addition, the concentration required for routine NMR measurements is much higher than that of UV-Vis absorbance measurements (2 mg·mL− 1 versus 0.2 mg·mL− 1), leading to a higher thermal cis- to trans- isomerization rate.28 We also observed precipitation during green light irradiation of trans-1 due to the strong aggregation effects of the cis- and trans- isomers in chloroform. Consequently, we were only able to obtain an NMR spectrum of a cis-1/trans-1 ratio of 3/7, by integration of the corresponding aromatic resonances (Fig. S9).
Photoswitching of thioindigo-containing polymer in conventional organic solvents and water
Following the determination of the optimal wavelength for photoisomerization of crosslinker 1, we proceeded to integrate the crosslinker into a polymer structure. Surprisingly, the methacrylate moieties on thioindigo 1 are highly resilient against free radical activation, as no change in 1H NMR spectrum was observed even when a solution of 1 and azobisisobutyronitrile (AIBN, 1 mol%) in chloroform was heated under reflux at 70 ºC for 48 h. Attempts to use the methacrylate group as an acceptor in Michael thiol addition was also unsuccessful, since conventional base catalysts, including pyridine and triethylamine, did not initiate the thiol-methacrylate reaction; stronger catalysts such as 1,8-diazabicyclo(5.4.0)undec-7-ene (DBU) and dimethylphenylphosphine led to the decomposition of thioindigo scaffold and discoloration. The breakthrough reported herein was ultimately achieved by free radical initiated thiol-ene chemistry of the methacrylates of 1 with thiols using AIBN and at 65 ºC in CHCl3. For polymerization, we selected poly(ethylene glycol) (PEG) as the water-soluble precursor and modified the end-group to confer the thiol function, PEG44-(SH)2 (Mn = 2176 g·mol− 1, Ð = 1.03, refer to the SI, Section 2.2 for the synthesis procedure). Thioindigo-containing PEG was subsequently prepared by free radical thiol-ene step-growth polymerization (Fig. 3a), forming polymer P1 with a molar mass of 13,058 g·mol− 1 (Ð = 2.73). Importantly, P1 is soluble in a wide range of polar solvents, enabling the study of photoswitching by UV-Vis absorbance measurements. Based on our action plot study, we selected the LED light sources with emissions centred at 540 and 470 nm for the reversible photoisomerization of P1 (Fig. S4). Compared to the chloroform solution of small molecule thioindigo bismethacrylate trans-1, P1 displayed similar absorption spectrum in all investigated solvents, and green light (λmax = 540 nm) irradiation on P1 solutions led to the decrease of the 540 nm peak (Fig. 3b-3e). However, the accompanying absorbances of the cis- isomer are not well resolved, suggesting lower trans- to cis-conversion ratios compared to the photoisomerization of small molecule trans-1.
Inspection of Fig. 3 reveals that the absorbance at 490 nm decreases with increasing solvent dielectric constant after green light irradiation. Specifically, in non-polar aprotic solvents such as dioxane (ε = 2.25) and chloroform (ε = 7.58), the cis-/trans- absorbance ratios at the photostationary state decrease from 0.57 to 0.43, respectively. The cis- to trans-conversion can also be induced by blue light at λmax = 470 nm, reaching a photostationary with higher trans-isomer content compared to green light treatment. The absorbance spectra in of P1 in polar solvents (those with ε values > 30) indicate complex solute–solute interactions – especially in polar protic solvents such as methanol and water, as can be seen from the broadening of the absorbance (Fig. 3e and 3h). These interactions inevitably limit the assessment of the cis-/trans- ratios at the photostationary states in highly polar solvents. However, the decrease and increase of the 540 nm peak, characteristic of the thioindigo photoisomerization, can be induced by green and blue light irradiation, respectively, suggesting photoswitching of the solutes. Overall, our investigation offers a critical answer for the utility of thioindigo in soft matter materials: thioindigo moieties – when covalently incorporated within polymer main chain – display appreciable photoisomerization in both aprotic and protic polar solvents.
Interestingly, solutions of P1 in water at c ≥ 5 wt% display thermoresponsive gelation behaviour that can be photo-modulated. Specifically, an aqueous solution of P1 at c = 5 wt% formed a solid gel at 37 ºC, reaching a storage modulus of ca. 1000 Pa. Rheological assessment of the hydrogel showed a decrease in the storage modulus to ca. 100 Pa under green light (λmax = 540 nm) irradiation (Fig. 4a). Blue light (λmax = 470 nm) irradiation of the mixture resulted in the recovery of the G’ value to approximately 60% of the initial value before green light irradiation. The photoswitching of the G’ value can be induced for several cycles, corresponding to the reversible trans-/cis- isomerization of the thioindigo units. Furthermore, the G’ value can recuperate spontaneously at 37 ºC, to a value similar to those recorded under blue light irradiation (Fig. 4b).
Incorporation of thioindigo into click-crosslinked PEG hydrogels for visible light induced photoswitching of physical properties
Having confirmed the photoswitching of the thioindigos within polymer chains in an aqueous environment, we subsequently designed a strategy to incorporate the photolabile function into hydrogels with robust mechanical strength, enabling photo-modulation of the hydrogels’ stiffness. The very low solubility of the thioindigo bismethacrylate trans-1 prevents direct mixing of the reagent in a resin formulation for crosslinking. Thus, we initially synthesised a polymer precursor P2 from 1 and PEG12-(SH)2 by free radical thiol-ene polymerization in CHCl3 (Fig. 5a) The intermediate product P2 has sufficient solubility in DMF at a concentration suitable for polymer crosslinker with a multi-arm linker (> 1 g·mL− 1). PEG12-(SH)2 was used in excess, ensuring the remaining thiol function can be employed in the subsequent crosslinking, which occurs via catalyst-free nucleophilic thiol-propiolate addition with a PEG448-(propiolate)4 in DMF (Fig. 5a).43 The resultant organogels were subsequently treated with excess water over 24 h to obtain fully swollen hydrogels. We were thus able to covalently embed various amounts of thioindigo within the network structure, forming a series of hydrogels with increasing weight ratio of thioindigo over PEG component (Fig. 5b), from 0.55 to 5.5 wt%. These materials (GelT1 to GelT5) allow us to systematically investigate the effect of thioindigo and its photoisomerization on hydrogels’ properties, compared to a PEG-based hydrogel (GelT0) with no thioindigo.
We observed a considerable change in the optical and physical properties even with the small amount of thioindigo incorporated into the network structures. Specifically, with 0.55 wt% incorporation of thioindigo, a clear red hydrogel was formed. Increasing the weight content of the thioindigo to ≥ 1.6 wt% resulted in the formation of opaque and deep red hydrogels, as compared to the clear and transparent hydrogel with no thioindigo GelT0 (Fig. 5c). Notably, we observed a decrease in the swelling ratio Qequilibrium (Fig. 5b) and the associated increase in hydrogels’ stiffness (Fig. S11) at the fully swollen state with increasing thioindigo content, despite the decrease in the thiol molar stoichiometry for thiol-propiolate crosslinking (lower crosslinking density in hydrogels with higher thioindigo content). These results indicate the significant physical aggregation of the thioindigo moieties, enhancing the mechanical properties of the networks. The hydrogel with the highest thioindigo content (5.5 wt%, GelT5) showed variations in stiffness in response to both temperature change and light irradiation (Fig. 6a). In particular, GelT5 displayed an increase in G’ value from 4.1 kPa to 4.7 kPa when the temperature was increased from 24 ºC to 37 ºC. When the hydrogel was subjected to green light irradiation at 37 ºC, the G’ value decreased to ca. 3 kPa, and reverted back to 3.8 kPa under blue light irradiation (Fig. 6b). The photo-modulation of the hydrogel’s stiffness by alternating green and blue light irradiation can be repeated for 3 cycles under rheological assessment, similar to the photo-responsiveness displayed by physically crosslinked hydrogel prepared from P1. Likewise, the green light-induced softened hydrogel slowly recovered its storage modulus in the dark. Other hydrogels GelT3 and GelT4 also displayed switching of the stiffness, by 3% and 8% variation in G’ value, respectively in response to the changes in green and blue light irradiation (Fig. S11). Hydrogels GelT0, GelT1 and GelT2 did not show any changes in the storage modulus when subjected to light irradiation, and only minor increase in stiffness at elevated temperature (Fig. S11).
Biocompatibility of hydrogels
Biocompatibility is an important parameter for materials intended for cell-related research and biomedical applications. In order to assess whether the hydrogel materials and photoswitching in the current study are biocompatible, Annexin V and Propidium Iodide (PI) staining were performed to evaluate apoptosis of cells exposed to hydrogels and trans-/cis- photoswitching process. GelT5, which contains the highest amount of thioindigo and displayed a significant change in stiffness under green light irradiation, was selected for biocompatibility assessments. Figure 7a indicates that HEK-293T cells incubated with hydrogels before and after irradiation with green LED light (λmax = 540 nm) maintained their polygonal morphology similar to the untreated control. In contrast, HEK-293T cells incubated with the positive control (DMSO) displayed changes in cytoplasmic morphology, appearing as clusters with a rounded shape, suggesting DMSO has a cytotoxic effect on the cells. Crucially, the percentages of live HEK-293T cells treated with hydrogels and the photoirradiation process remained above 90% after 24 h incubation (Fig. 7b-7c). Treatment with hydrogels did not result in significant differences in the percentage of apoptotic cells as compared to the untreated control (Fig. 7c). On the other hand, the cell viability obtained in cells exposed to hydrogels was higher than that of the cells exposed to DMSO.
We further assessed the cytotoxicity of hydrogels using human peripheral blood mononuclear cells (PBMCs) from healthy donors. The majority of PBMCs treated with hydrogels before and after irradiation were alive, achieving a viability of more than 80% (Fig. 7d-7e). As a positive control, treatment with DMSO led to low cell viability, evidence by the high percentage (> 90%) of late apoptotic and necrotic cells (Fig. 7e). Of note, hydrogels before and after irradiation with an LED green light showed similar cell viability in both HEK-293T cells and PBMCs (Fig. 7c and 7e). In summary, these findings indicate that thioindigo-containing hydrogels – and the photoswitching process – both have good biocompatibility with cells.