Modular Design of Programmable Mechanofluorescent DNA Hydrogels

Mechanosensing systems are ubiquitous in nature and control many functions from cell spreading to wound healing. Biologic systems typically rely on supramolecular transformations and secondary reporter systems to sense weak forces. By contrast, synthetic mechanosensitive materials often use covalent transformations of chromophores, serving both as force sensor and reporter, which hinders orthogonal engineering of their sensitivity, response and modularity. Here, we introduce FRET-based, rationally tunable DNA tension probes into macroscopic 3D all-DNA hydrogels to prepare mechanofluorescent materials with programmable sacrificial bonds and stress relaxation. This design addresses current limitations of mechanochromic system by offering spatiotemporal resolution, as well as quantitative and modular force sensing in soft hydrogels. The programmable force probe design further grants temporal control over the recovery of the mechanofluorescence during stress relaxation, enabling reversible and irreversible strain sensing. We show proof-of-concept applications to study strain fields in composites and to visualize freezing-induced strain patterns in homogeneous hydrogels.


Supplementary Note 2: Impact of module functionalization on rheological properties
We report the impact of strain and frequency on the measured rheological properties of the DNA hydrogel before and after functionalization with module D1 (Supplementary Figure 2a,b). The hydrogel functionalization was analyzed in-situ by first analyzing the pregel and then functionalization of it with the D1 module in the rheometer.
First, 40 μL of 0.7 wt% pregel (in TE buffer containing 12 mM of MgAc2 and 100 mM of NaCl) were loaded into the rheometer and subsequently set into a solid gel by running a temperature ramp from 10 to 80 and to 10 °C with 5 min stay at 80 °C and at 3 °C·min -1 heating and cooling rate. Subsequently, we analyzed this unfunctionalized gel more deeply, by strain sweep experiments from 0.1 to 300% at 0.1 Hz (Supplementary Figure 2a) and via a frequency sweep from 0.01 up to 100 Hz at 1% strain (Supplementary Figure 2b). To allow for further functionalization with the D1 module, we removed any strain or stress-dependent properties in the hydrogel by re-annealing using a temperature ramp (heating to 80 °C then cooling back to 10 °C) while recording the variation of the mechanical properties with temperature upon cooling (Supplementary Figure 2c).
After removing the rheometer upper plate, we loaded 345 µL D1 module (20 µM, equimolar to the functionalization sites A1, A4, B1 and B4) in TE buffer containing 12 mM of MgAc2 and 100 mM of NaCl onto the hydrogel, and left the module to diffuse and hybridize into the DNA hydrogel overnight at room temperature. After ca. 10 h, we placed back the upper rheometer plate. Since fluorescent and PAGE-purified strands are too expensive to run experiments at such scale, we used non-fluorescent strands with standard purification to assemble the modules used for rheology characterization. Hybridization of the modules in the DNA hydrogel induces an increase of the volume of about 50%, as seen by the gap distance increase of the rheometer plates from 0.2 mm to 0.3 mm to adapt to the new thickness.
We then ran similar experiments as above for the non-functional gel, i.e. a strain sweep from 0.1 to 300% strain at 0.1 Hz (Supplementary Figure 2a), as well as a new frequency sweep from 0.01 up to 100 Hz at 1% strain (Supplementary Figure 2b). Finally, we performed the same temperature ramp as before (to remove strain or stressinduced damages) while recording variations of modulus upon cooling (Supplementary Figure 2c).
Initially surprising, the loss and storage modulus of the hydrogels remained very similar before and after functionalization with module D1. However, it needs to be considered that the volume increase by 50% (strength decrease) compensates the increase of crosslink density (strength increase) associated with the hybridization of the module inside the hydrogel.
Strain sweeps show an earlier decrease of storage modulus for D1-functionalized hydrogels compared to the pristine DNA hydrogel. Here, the swelling of the gel associated with module hybridization stretches the initial network and results in network failure at lower strain. This decrease of storage modulus corresponds to an increase of loss modulus because strain-induced reorganization of the network of DNA bonds (sacrificial bonds) allows energy dissipation. A similar yet less pronounced increase of loss modulus upon failure is also visible for the pristine DNA hydrogel. The frequency sweeps are nearly identical before and after functionalization indicating that crosslinking with module D1 does not change the network temporal response, while again, one has to bear in mind that the hydrogel is diluted to 50 % and has an increased amount of crosslinks in total. Finally, the most dramatic differences associated to the hybridization of D1 module are visible during temperature ramps, as displayed in the cooling curves after removing strain/stress-induced damage (Supplementary Figure 2c). The pristine DNA hydrogel shows a single increase of loss and storage modulus near 70 °C, corresponding to X/X* hybridization. However, after functionalization, multiple transitions for the loss modulus are observed, which indicates a rich variety of transient supramolecular bond formation across the network. These new transient networks further induce an increase of storage modulus between 45 and 65 °C compared to the pristine hydrogel. While, this peculiar response is relevant and reproducible, we suggest that the in-depth evaluation of these phenomena is subject to further rheological characterization beyond the conceptual scope of this manuscript to clearly identify and explain these variations on a consistent structural basis.

Supplementary Note 5: Melting transitions and fluorescence responses of the mechanosensing modules
We measured the melting transitions of the different modules by temperature-dependent fluorescence spectroscopy. The modules were assembled according to the previous protocol and diluted to 1 μM in TE buffer containing 12 mM of MgAc2 and 100 mM of NaCl. Subsequently, 50 μL were introduced in a 3 x 3 mm fluorescence microcuvette (Hellma Analytics) and covered with a drop of oil to prevent evaporation. We recorded the fluorescence using an Ocean Optics spectrometer equipped with a Peltier-heated cuvette holder. An exemplary heating profile for the D1 modules is presented Supplementary Figure 5a and consists of two symmetric heating/cooling cycles between 15 and 85 °C at 2 °C·min -1 with 2 min equilibration time at 85 °C and 15 °C. The fluorescence spectrums were recorded using 490 nm as excitation wavelength. The maximum of emission at 600 nm is plotted as a function of time for module D1 for two heating cycles (Supplementary Figure 5a). Supplementary Figure 5b compares the different modules by plotting the fluorescence response as a function of temperature (2 nd heating ramp). The same increase of fluorescence is observed around 70 °C for modules D1, HP and T, while the module D2 melts around 45 °C. The similar response for D1, HP and T relates to the meltings of the M (79 °C) or N (77 °C) arms which are predicted to happen at slightly lower temperature than D1 (84 °C) or HP (78 °C). In stark contrast, in D2, the breakage of the sacrificial bonds is clearly observed, corresponding to the expected melting transition (50 °C in Table 2, main MS). Note that the 5 °C discrepancy between observed and calculated values arises from the difference of concentrations used for Tm simulation (10 μM DNA) and for fluorescence measurements (1 μM DNA). Values at 10 μM are more realistic concerning the material and rheological behavior.

Supplementary Note 7: Temporal recovery experiments
Figure 3d of the main manuscript displays different mechanofluorescent modules that lead to different temporal recovery behavior. The mechanism for irreversible fluorescence recovery is presented in Supplementary Figure 7a. For the two modules with a simple sacrificial duplex (D1 and D2) there is a competition between self-dimer formation and reformation of the original hetero-duplex. Since each side of the mechanofluorescent module binds to one of the RCA products they remain in close proximity after cleavage of the sacrificial duplex which gives a kinetic advantage to self-dimer (Dx/Dx and Dx*/Dx*, x = 1, 2) formation, while the heterodimer (Dx/Dx*, x = 1, 2) is thermodynamically favored because it maximizes the number of base pairs. The difference between module D1 and D2 lies in the stability of the self-dimers. NUPACK (http://www.nupack.org) simulations show that the self-dimers of D1 are stable at RT ( In D1 the formation and relative stability of the kinetic products slows down the reformation of the original FRET pair and fluorescence decrease after stress release. On contrary, for module D2 the lack of stability of the self-dimers at RT yields a spontaneous recovery of the more stable original duplex D2/D2*. This recovery is nevertheless slow because the two arms need to find each other by diffusion while remaining attached to the hydrogel network. For the last one, HP, the two arms of the module are covalently linked, and they cannot move far apart. The module HP therefore quickly returns to its original conformation after stress release, as it is both kinetically and thermodynamically favored (Supplementary Figure 7h). This results macroscopically in a fast fluorescence decrease.