Mechanically Tunable DNA Hydrogels as Prospective Biosensing Modules

Abstract

Sequence-programmable DNA building blocks offer high degree of freedom in designing arbitrarily complex networks of tunable viscoelastic properties. Yet, the deployment of DNA-based functional materials remains limited due to insufficient control over the emerging structures and their mechanics. In an ongoing effort to place structure-property relations in stimuli-responsive DNA materials on a firm foundation, here a systematic rheological study of self-assembling DNA networks is presented, comprised of short DNA nanomotifs, namely trivalent nanostars and bivalent linkers, where the latter differ in their composition on a single base-pair level. Notably, we found through combining conventional bulk rheology with diffusing wave spectroscopy (DWS-based) passive microrheology a relationship between the melting temperature of a DNA hydrogel and its DNA sequence composition. By providing a use case, we demonstrated how the determination of such empirical relations could impact the areas of biosensing and mechanical computing, where control over the system state and target identification are key.

Keywords: DNA materials; biosensing; melting temperature; microrheology; sequence‐programmability.

Figure 1

Schematic illustration of the formation of trivalent Y‐shapes and bivalent linkers with the DNA sequences shown in Table 1 of the Experimental Section. Combining the two building blocks at a well‐defined stoichiometry and low temperatures (below the melting temperature of the sticky ends in red) results in the formation of a DNA hydrogel. The latter is held together by hydrogen‐bonding interactions and can reversibly associate and dissociate by cycling the temperature on both sides of the melting transition.

Figure 2

Passive microrheology using diffusing wave spectroscopy on 1.5 wt% YLS12: temperature ramp from 70°C to 40°C. We plot both the intensity autocorrelation functions g⁽²⁾‐1 and the mean‐squared displacements above, around and below the melting temperature of the system. The increasingly long decorrelation times and appearance of intermediate‐time plateau in the MSD both signify gelation.

Figure 3

Fourier transforms allow conversion of the particle dynamics in Figure 2 into elastic, G′(ω) (filled symbols) and viscous, G″(ω) (empty symbols) moduli. At 70°C, all DNA strands dissociated and we found a viscosity similar to that of water (red data). At 40°C, YLS12 was found to be firmly in the gel state (violet data). The sol‐gel transition region for YLS12 was estimated to be around 58°C, where both moduli had a similar magnitude and nearly identical scaling of ω^0.5 (green data). The DWS results are complemented with schematic depictions of the fully connected gel phase (violet, left), partially connected sol‐gel transition phase (green, middle) and totally dissociated liquid phase (red, right). The Newtonian linear scaling of G″(ω) at 70°C for the tracer particles in water is included as a reference (black dashed line).

Figure 4

Passive microrheology using diffusing wave spectroscopy on 1.5 wt% YLS12, YLS11, YLS10, and YLS9 DNA hydrogels. To facilitate comparison between all hydrogels, we plot the raw intensity autocorrelation functions g⁽²⁾‐1 at fixed temperatures of 70°C (fully dissociated phase in all cases), 58°C (close to the sol‐gel transition temperature of YLS12) and 40°C (deep into the gel phase of YLS12). Comparison of the autocorrelation functions across all hydrogels shows that the curves decay more rapidly for shorter linkers. This decrease in decorrelation times from YLS12 to YLS9 implies a shift in melting temperature.

Figure 5

Arrhenius kinetics in 1.5 wt% YLS11 and YLS10 DNA hydrogels, taking into consideration the half‐decay time points of the intensity autocorrelation functions above the sol‐gel transition, where we expected a thermally activated network formation. We compare the activation free energy given by the extracted slopes from the logarithmic plots with the theoretical estimates by the nearest‐neighbor model of Allawi and Santalucia. We find reasonably good agreement (1–2 standard deviations) for systems bonded by sufficiently long sticky ends.

Figure 6

Temperature ramps in oscillatory bulk rheology on 1.5 wt% DNA hydrogels with linkers of different sticky‐end length. All ramps were performed at a fixed strain amplitude of 1% and frequency of 1 Hz. Red filled circles mark the elastic moduli G′(ω), while blue empty circles denote the loss moduli, G″(ω). The sol‐gel transition is indicated by an arrow in each case. The results were confirmed by performing multiple ramps at identical experimental conditions. The summary plot indicates how the melting temperature of the system changes upon removing bases from the sticky ends of the linkers. We found an approximately linear trend, where the melting temperature reduces by ca. 6°C per deleted base.

Figure 7

Extracted A,B) phase angles, C) relative stiffness, D) plateau elastic modulus and average mesh size in oscillatory bulk rheology on DNA hydrogels with linkers of different sticky‐end length. The phase angles in A) were extracted from frequency ramps at 40°C for YLS12, YLS11, YLS10 and YLS9, and in B) at 5°C below the corresponding melting temperature (T ≈ T _m − 5°C) of YLS8, YLS7, and YLS6. The strain in each case was fixed to 1%. A loss tangent of 1 marks the approximate transition point, where G′(ω) ≈ G″(ω). The insets in A) and B) show the relaxation times extracted from the corresponding frequency sweeps (cf. Figure S8, Supporting Information). All hydrogels were found to exhibit shear‐thinning behavior, with a reduction in viscosity on increasing frequency. C) The relative stiffness, defined as the ratio of G′(ω) and G″(ω), plotted for all hydrogels at a temperature of 40°C, strain of 1% and frequency of 1 Hz. We found an approximately linear decrease in relative stiffness with increasing the number of deleted bases. D) Approximately constant values for the plateau of the elastic modulus (filled squares) and associated average mesh size (empty squares) at low temperatures (T < T _m ) and fixed frequency of 10 Hz. The plateau values of the elastic modulus were obtained at temperatures differently spaced from the respective sol‐gel transitions, accounting for the large variations.

Figure 8

Linker‐exchange experiments: frequency sweeps in oscillatory bulk rheology at 15°C on 1.5 wt% YLS6 DNA hydrogels with Y:LS6 ratio of 2:3, ensuring that approximately all Y‐shapes are connected to LS6 toehold linkers. Adding A) LS9, B) LS10, or C) LS11 target linkers was then found to cause an increase in G′(ω) and G″(ω) in both concentration‐dependent and linker‐specific manner. Filled symbols signify the elastic modulus, G′(ω), whereas empty symbols mark the viscous modulus, G″(ω). Black symbols represent starting configurations (YLS6 hydrogels), red symbols mark the addition of target LS9, LS10, and LS11 at half the concentration of LS6 toehold linkers, and blue symbols denote further addition of target sequences, such that we obtain equal concentrations of toehold and target. We also show oxView schematic representations of the linker‐exchange process, where LS6 is replaced by LS9, LS10, or LS11 target linkers, respectively.

Figure 9

A) Increase in relative stiffness, G′(ω)/G″(ω) upon addition of LS9, LS10 and LS11 target linkers to YLS6 toehold hydrogel. The measurements were conducted at 10°C, 1% strain, and angular frequency of 0.3 s⁻¹. Filled circles represent the volume of added LS9 linker, empty triangles ‐ LS10 linker, and filled stars ‐ LS11 linker. The differential increase in relative stiffness allows discrimination between targets, where the method appears to be more sensitive at lower volumes (< 50 µL) due to the counteracting dilution effect. B) A schematic representation of an example use case utilising a lateral flow test for the real‐time detection of a target gene within a few drops of a liquid (e.g. blood) sample. The lateral flow image was generated with the help of a language model (Dall‐E 2, OpenAI) and subsequently annotated.