Cation-Responsive and Photocleavable Hydrogels from Noncanonical Amphiphilic DNA Nanostructures.

Thanks to its biocompatibility, versatility, and programmable interactions, DNA has been proposed as a building block for functional, stimuli-responsive frameworks with applications in biosensing, tissue engineering, and drug delivery. Of particular importance for in vivo applications is the possibility of making such nanomaterials responsive to physiological stimuli. Here, we demonstrate how combining noncanonical DNA G-quadruplex (G4) structures with amphiphilic DNA constructs yields nanostructures, which we termed "Quad-Stars", capable of assembling into responsive hydrogel particles via a straightforward, enzyme-free, one-pot reaction. The embedded G4 structures allow one to trigger and control the assembly/disassembly in a reversible fashion by adding or removing K+ ions. Furthermore, the hydrogel aggregates can be photo-disassembled upon near-UV irradiation in the presence of a porphyrin photosensitizer. The combined reversibility of assembly, responsiveness, and cargo-loading capabilities of the hydrophobic moieties make Quad-Stars a promising candidate for biosensors and responsive drug delivery carriers.

Nucleic acids have proven to be valuable building materials for the self-assembly of 2D and 3D architectures, nanomachines, and functional nanomaterials.[1−3] This success is not surprising, as nucleic acids benefit from superior biocompatibility, programmable interactions, and a decreasing production cost.[4,5]

DNA-based hydrogels, in particular, have shown promise in applications as varied as biosensing,[6,7] drug delivery,[8−10] synthetic biology,[11] and artificial cells.[12] Key characteristics sought in DNA-based hydrogels include fast, scalable, and cost-effective production, and stimuli-responsiveness, particularly in the form of triggered assembly/disassembly.

Hybrid DNA–polymer hydrogels, usually constructed from poly(acrylamide) (PAAm) or poly(ethylene glycol) (PEG) cross-linked via DNA side chains, have proven highly successful as biosensors and carriers, as they can respond to a wide range of stimuli, including thermal denaturation,[6] strand displacement,[13] small molecules opening DNA–aptamer bridges,[14] and metal ions triggering cleavage through DNAzymes.[15] Similar hydrogels have been engineered to undergo location-addressable swelling via reconfigurable DNA cross-links, applicable in soft robotics.[16] Replacing DNA duplexes with noncanonical i-motifs[17] and G-quadruplexes (G4s)[18−20] further enables responsiveness to pH and cation concentration, respectively. A combination of double-stranded DNA (dsDNA) linkers and i-motifs has been used to produce shape-memory DNA hydrogels, with potential applications in tissue scaffolding and engineering.[21] However, the hybrid approach suffers from potentially toxic breakdown products, particularly in the case of acrylamide-based hydrogels,[22] and it requires an additional production step to covalently link DNA and polymer.

Conversely, all-DNA hydrogels could feature higher biocompatibility and biodegradability,[23] while simplifying the production route, and have been applied to controlled drug release[24] and cell-free protein production.[25,26] The inclusion of i-motifs[27] and triplexes[28] enables pH-triggered, reversible sol/gel transitions,[29] while DNA–aptamer bridges allow for disassembly upon binding small molecule targets.[30,31] Currently, though, pH-triggered assembly/disassembly is the only example of a reversible structural response driven in all-DNA hydrogels by a physiologically relevant stimulus, limiting biomedical applicability.

Recently, we introduced a novel class of amphiphilic DNA nanostructures, dubbed C-Stars,[32−34] that combine the programmability of DNA with the robust self-assembly of amphiphiles. C-Stars consist of branched DNA junctions (nanostars) featuring a cholesterol moiety at the end of each arm, which enables the formation of crystalline frameworks or amorphous hydrogels sustained by hydrophobic interactions. DNA-based frameworks have been proposed as a general strategy to program designer nanomaterials, such as supramolecular assemblies of nanoparticles,[35,36] quantum dots,[37,38] and proteins.[39] Compared to these approaches, which are fully reliant on base-pairing or stacking interactions, the hydrophobicity-mediated self-assembly principle of C-Stars warrants enhanced robustness against design variations and changes in environmental conditions, making it possible to tune porosity, capture/release proteins, and trigger disassembly through toehold-mediated strand displacement. The previously demonstrated disassembly response, however, requires nonbiologically available single-stranded DNA as input and is thus not directly applicable to in vivo scenarios.

Here, we expand upon the responsiveness of amphiphilic DNA nanostars by combining their robustness and versatility with a cation-dependent DNA G4 motif. The resulting nanostructures, which we term Quad-Stars, form nanoporous hydrogels responsive to biologically relevant K+ cations. The materials assemble in a simple, enzyme-free, one-pot reaction when exposed to K+ ions, the chelation of which leads to disassembly. The amphiphilicity of the networks allows for loading of small hydrophobic molecules, which can then be released upon disassembly. Besides K+ chelation, the material can also be disassembled upon addition of a G4-binding porphyrin photosensitizer and exposure to long-wave UV light, the combination of which disrupts the quadruplex through guanine photo-oxidation.[40−44] This additional mechanism is irreversible but enables precise spatial control over the response. Owing to the combined assembly reversibility, multistimuli-responsiveness, cargo-loading capabilities, and stability in model physiological conditions, we believe that Quad-Star hydrogels hold significant potential for applications such as smart biosensors and drug delivery.

Quad-Stars comprise four identical duplex “arms” bearing a cholesterol end-modification which, under suitable conditions, can join via a tetramolecular parallel G4 motif, as shown in Figure a,b. G-quadruplexes are noncanonical secondary structures that can form in guanine-rich nucleic acid strands unimolecularly, bimolecularly, or, as in Quad-Stars, tetramolecularly, i.e., between four distinct strands.[45] Depending on the strand orientation, G4s can be parallel, antiparallel, or hybrid. They comprise a series of stacked G-quartets interacting via Hoogsteen hydrogen bonding, as depicted in Figure a (top). Alkali metal ions, particularly K+, coordinate guanines, templating and stabilizing G-quartets and their stacking into G4s. This grants G4s, and in turn Quad-Stars, a degree of responsiveness with respect to cationic environments.

Figure 1

Design and characterization of amphiphilic DNA Quad-Stars. (a) Schematic of a G-quartet (top) and a parallel tetramolecular G4 (bottom). (b) Quad-Stars assemble from four duplex arms, each made of three strands. During the annealing process, single-stranded DNA components hybridize into duplex arms, but G4 formation and assembly of the designed Quad-Star occurs only in the presence of K+ ions. Two design variations, labeled as 5G and 6G (5 or 6 guanines in the G-run of the G4 forming sequence), are investigated. (c) Schematic of C-Stars[32] with arm length l = 28 base pairs (same as Quad-Stars). C-Stars and Quad-Stars share the same cholesterol-functionalized strands. (d) Agarose gel electrophoresis confirms that formation of individual, noncholesterolized Quad-Stars is K+-dependent. Slower bands for both 5G (green) and 6G (blue) only appear when motifs are annealed in KCl (300 mM), and their mobility is comparable to that of four-arm nanostars with an arm length of 28 bp (orange). (e) Intensity profiles corresponding to the gel in panel d. (f) Volume-based size distribution obtained via dynamic light scattering (DLS) further substantiating the formation of Quad-Stars in KCl, with a net shift from the size distribution in LiCl. As expected from panel d, the size of assembled motifs in KCl is comparable that of four-arm nanostars in 300 mM NaCl. (g) Circular dichroism (CD) proving that self-assembly relies on the formation of tetramolecular parallel G4s. Overlaid CD spectra in LiCl (faint colors) and KCl (saturated colors) clearly display the K+-induced shift (highlighted via arrows) in maximum and minimum toward ≈263 and 240 nm, respectively, as typical of parallel G4 structures. Samples for all reported measurements were prepared at [oligonucleotide] = 20 μM.

As shown in Figure b, each Quad-Star arm assembles from three DNA strands, one bearing a cholesterol moiety, connected via a triethylene glycol (TEG) spacer, another acting as linker, and a third terminating with a 5 or 6 nucleotide (nt)-long guanine run flanked by thymines (T GGGGG T or T GGGGGG T; see zoomed inset in Figure b), a sequence capable of assembling into tetramolecular parallel G4s under physiological conditions.[46] In the remainder of the text we refer to these designs as 5G and 6G, respectively.

In the assembled arm, a poly-T spacer is present between the G-run and the duplex portion to increase flexibility at the G4 site. Thymine-spacer lengths of 3, 5, 7, and 9 nucleotides were tested, and minimal differences were observed in the Quad-Star self-assembly behavior. In the remainder of the main text we report results for 9T spacers, while analogous findings for different lengths are featured in the SI.

Assembled Quad-Stars are topologically analogous to previously introduced C-Stars (Figure c), with the key difference being that Quad-Star arms are held together by a stimuli-responsive G4 motif rather than an irreversible four-way junction. Similarly to C-Stars,[32,33] we therefore expect Quad-Stars to self-assemble into nanoporous networks owing to the clustering of cholesterol moieties into micelle-like structures, coordinated and linked by the tetravalent DNA motifs. Unlike C-Stars, however, we expect Quad-Stars and Quad-Star networks to respond to changes in ionic environment, self-assembling at sufficiently high concentration of G4-promoting K+ ions (and Na+ to a lower extent), while being destabilized by cations with a lower affinity for the G4 microenvironment, such as Li+ or Mg2+ (Figure b).

To verify the self-assembly of Quad-Stars into the sought branched architectures at thermodynamic equilibrium and under suitable ionic conditions, samples of both the 5G and 6G designs were slowly annealed in TE buffer containing 300 mM LiCl/NaCl/KCl or 2 mM MgCl2 (see Methods, SI). Network formation was here prevented by replacing cholesterol-tagged strands with nonfunctionalized ones having identical sequence.

For both 5G and 6G samples annealed in LiCl, agarose gel electrophoresis (AGE) shows a single band corresponding to disjoined nanostar arms. Samples annealed in KCl, instead, feature a second, slower band with mobility comparable to that of four-arm junctions (nanostars) with 28 bp arm length, which can thus be ascribed to correctly formed Quad-Stars (Figure d,e). Correct assembly is confirmed by dynamic light scattering (DLS) (Figure f), which demonstrates a cation-dependent size shift for both 5G and 6G designs, with samples in KCl showing the same average hydrodynamic diameter as four-arm nanostars with the same arm length. As shown by AGE and DLS, samples annealed in 2 mM MgCl2 (Figures S1 and S2, SI) and 300 mM NaCl (Figures S3 and S4, SI) exhibit a lower degree of Quad-Star assembly for the 6G design and no assembly for 5G, hence pointing at a degree of quadruplex stabilization intermediate between Li+ and K+, in agreement with literature.[47] When reducing the concentration of monovalent cations to 100 mM, assembly was only observed for 6G designs in K+-rich buffers (Figures S3 and S4, SI).

Circular dichroism (CD) spectra were used to confirm the presence, and determine the topology, of the G4s (Figure g). Compared to those prepared in LiCl, samples annealed in KCl present a clear shift in peak and valley positions toward 263 and 240 nm, respectively, accompanied by an amplitude increase, indicating the formation of parallel G4s, as intended.

Results for 5G and 6G Quad-Stars with different poly-T spacer lengths are collated in Figures S5–S7 (SI). Quad-Star designs with shorter G-runs (4G) were also tested and found incapable of assembling even in KCl buffers, producing results similar to those of negative controls where guanines are replaced by thymines (see Figures S5–S7, SI).

To explore the ability of Quad-Stars to isothermally respond to changes in ionic conditions, noncholesterolized Quad-Stars were annealed in 300 mM LiCl buffers, hence resulting in dispersed duplex arms, before triggering assembly via rapid injection of KCl (see Methods, SI). The assembly transient was then tracked by time-resolved fluorescence using N-methyl mesoporphyrin IX (NMM), whose quantum yield dramatically increases upon preferential complexation with parallel G4s.[48−50]

The obtained fluorescence intensity profiles monotonically increase after KCl injection, demonstrating G4 formation and isothermal assembly of Quad-Stars, and could be fitted to a simple model[46] (see Methods, SI) to extract the assembly half-life, t1/2, as exemplified in Figure a. The values obtained at the target conditions of 20 μM oligonucleotide concentration ([Quad-Star] = 5 μM) and 300 mM KCl clearly indicate that longer G-runs translate into faster assembly, with t1/2 ∼ 5 min for 6G and t1/2 ∼ 15 h for 5G. This trend is consistent with previous literature reports[46,51,52] and can be rationalized as the result of a nucleation-zipping assembly pathway.[53,54] Within this framework, longer G-runs afford increased stability to transient dimers and trimers hypothesized as intermediates for G4 assembly,[55,56] hence boosting folding kinetics.

Figure 2

Kinetics of G4-dependent formation in soluble Quad-Stars via NMM fluorimetry. (a) Kinetic profiles of G4-folding at [Oligo] = 20 μM (Oligo = oligonucleotide) and [KCl] = 300 mM. As G4s form, NMM stacks on G-quartets and undergoes fluorescence emission enhancement. Normalized fluorimetry data are presented as mean (data points, one marker every three points shown) ± standard deviation (shaded region) and fits (solid lines) to theoretical tetramolecular kinetics.[46] Half-life values are reported as mean (vertical lines) ± standard deviation (shaded intervals). Fainter colored profiles at the bottom are negative controls (no KCl injection), reported as average data points only. (b) Half-life of formation Vs (i) [Oligo] at fixed [KCl] = 300 mM and (ii) [KCl] at fixed [Oligo] = 20 μM. (i) Data have been linearly fitted (solid lines) in log–log scale according to power law t1/2 = A[Oligo]1– to extract the reaction order from the fitting slope, resulting in n = 3.605 ± 0.740 for 5G (R2 = 0.911) and n = 3.711 ± 0.127 (R2 = 0.996) for 6G. Both values closely match the expected stoichiometric value of 4. Note that, owing to the prohibitively slow formation rate, assembly of the 5G motifs could not be tested at the lowest oligonucleotide and KCl concentrations used for 6G. Data in panel b are presented as mean (data points) ± standard deviation (error bars). Negative controls in panel a are averages of independent duplicates. All other data have been averaged across (at least) triplicates. See Figure S8 (SI) for kinetic profiles for panel b.

In Figure b(i) we explore the dependency of t1/2 on oligonucleotide concentration and, expectedly, we observe a decreasing power-law trend that allows us to extract the reaction order of the assembly process as n = 3.605 ± 0.740 for 5G and n = 3.711 ± 0.127 for 6G, closely matching the theoretical stoichiometry of 4.

Figure b(ii) proves that Quad-Stars fold significantly faster in the presence of higher potassium concentrations, with t1/2 decreasing from ∼10 h to ∼5 min when [K+] is increased from 30 to 300 mM for 6G. A similarly strong dependence was found for 5G.

Overall, the behaviors displayed in Figure b demonstrate the ability of Quad-Stars to isothermally assemble in a cation- and concentration-dependent fashion and, importantly, that assembly rates dramatically depend on both potassium and oligonucleotide concentrations.

We thus proceeded to investigate the ability of cholesterolized Quad-Stars to form 3D networks driven by hydrophobic interactions, as previously observed for C-Stars[32−34] and sketched in Figure a–d.

Figure 3

Reversible, isothermal assembly of Quad-Stars via NMM fluorimetry and epifluorescence microscopy. (a) Schematic representation of cholesterol-induced micelles and individual double-stranded DNA arms formed after annealing in 300 mM LiCl and 2 mM MgCl2, respectively (see Figure S9, SI). Presented data refer to samples annealed in 2 mM MgCl2. (b) Kinetic profiles of G4-folding in cholesterolized Quad-Stars via NMM fluorimetry. Normalized fluorimetry data are presented as mean (data points, one marker every three points shown) ± standard deviation (shaded region) and fits (solid lines) to theoretical tetramolecular kinetics.[46] Half-life values are reported as mean (vertical lines) ± standard deviation (shaded intervals). See Figure S11 (SI) for examples of negative controls for 6G Quad-Stars. (c) Area fraction profiles extracted from epifluorescence microscopy time lapses depict the self-assembly kinetics of hydrogel aggregates. See control experiments in Figure S12 (SI). Unprocessed videos (bright-field and epifluorescence) are provided in the SI (Videos S1, S2, S3, and S4). (d) Schematic representation of the 3D percolating network leading to hydrogel aggregate formation in cholesterolized Quad-Stars upon addition of K+ ions. (e, f) Epifluorescence micrographs of 5G and 6G Quad-Star aggregates, respectively, formed after 2.5 h from KCl injection and imaged via an Alexa488 fluorophore linked to dsDNA arms. Scale bars are 50 μm. (g) Reversibility of assembly probed via NMM fluorimetry. Addition of K+ ions (black vertical line) leads to assembly, while their chelation by cryptand injection (red vertical line) leads to fast disassembly. Depicted sample is 6G. (h) Three consecutive assembly/disassembly cycles of 6G Quad-Stars demonstrate the reversibility of hydrogel aggregate formation. See Figure S13 (SI) for zoomed-in detail of disassembly profiles. Cropped micrographs in insets (I–VI) depict sample status at assembly stages indicated by arrows. Scale bar is 50 μm. All insets show fields of view of the same size. Data in panels c, g, and h are shown as mean (solid lines) ± standard deviation (shaded regions) of at least three repeats. Area fraction profiles in panels c and h were obtained via image segmentation in FIJI (see Methods, SI). Micrographs in panels and insets were background corrected via rolling ball algorithm in FIJI (radius = 100 pixels). See Figure S14 (SI) for raw micrographs of panels e and f.

As for C-Stars, we expected the duplex arms to form cholesterol-induced micelles even prior to Quad-Star formation.[32] DLS in Figure S9 (SI) indeed confirms micelle formation in both 300 mM LiCl and 300 mM KCl, while no micelles are observed in 2 mM MgCl2, probably owing to the much reduced ionic strength. Consistently, while we observe no evidence of G4 formation in noncholesterolized Quad-Stars in Li+-rich buffers, cholesterolized designs showed a moderate tendency to aggregation in 300 mM LiCl, arguably due to colocalization of the G4-forming domains within micelles. In contrast, cholesterolized arms annealed in 2 mM MgCl2 show no aggregation and were thus chosen as the starting condition for K+-triggered network-assembly experiments.

Upon addition of K+, Quad-Star networks are expected to assemble due to G4-mediated formation of nanostar motifs, which bridge neighboring micelles. As confirmed by time-resolved NMM fluorimetry in Figure b, indeed, the addition of 300 mM KCl rapidly triggers quadruplex formation. Cholesterolized designs display substantially faster assembly compared to noncholesterolized samples, as summarized in Figure S11c (SI). The speed up is particularly noticeable for 5G samples, showing a 100-fold decrease in half-life (from 15 h to 10 min). We propose that this rate increase is due to colocalization of G4-forming strands within the micelles.

NMM fluorimetry assays in Figure b capture the isothermal folding of G4 motifs within cholesterolized Quad-Stars, rather than their aggregation into extended network phases. To verify the occurrence of the latter, and study its kinetics, isothermal assembly was monitored via epifluorescence microscopy. For these experiments, visualization was enabled by a covalently linked Alexa488 fluorophore to prevent NMM-induced photocleavage upon strong illumination (see Methods, SI).

Assembly kinetics were monitored by determining the time-dependent area fraction of the microscopy field of view (FOV) occupied by aggregates (see Methods, SI). Figure c demonstrates a faster assembly rate for 6G compared to 5G, consistent with the trends in G4-folding determined by fluorimetry (Figure b). Micrographs in Figure e,f feature round fluorescent aggregates after KCl injection for both 5G and 6G, with the latter displaying larger and more closely packed aggregates.

To disentangle the effect of increased ionic strength following KCl injection from that of G4 stabilization, in Figure S12 (SI), we monitored the response of samples annealed in 2 mM MgCl2 to the addition of 300 mM LiCl. A degree of aggregation was observed, particularly with 6G designs, but this was significantly less pronounced than the response observed with KCl, consistently with greater quadruplex-stabilization ability of K+ and the evidence that Li+ only leads to limited aggregation at equilibrium (Figure S10, SI). Consistent trends are observed in samples preannealed in 300 mM LiCl upon addition of further 300 mM LiCl or 300 mM KCl (Figure S12, SI).

Potassium-stabilized G4 motifs are known to structurally weaken[57] and rapidly unfold[58−62] upon K+ removal, which should lead to Quad-Star and network disassembly. We assessed this reversibility route via NMM fluorimetry by injecting the chelating agent [2.2.2] cryptand[58,62] after a potassium-induced assembly transient, while recording NMM fluorescent emission. Figure g shows that the disassembly is almost instantaneous upon cryptand addition, as reflected by the sudden drop in fluorescence, confirming the reversibility of Quad-Star assembly.

To further explore reversibility in Quad-Star network formation, we performed three subsequent rounds of isothermal assembly/disassembly by alternating KCl and cryptand injections into the same 6G sample. Figure h displays the associated time-dependent area-fraction profile that demonstrates formation of visible aggregates after each KCl addition and an abrupt disassembly at each K+ chelation. A slower assembly rate is observed for each subsequent KCl injection, likely due to the progressive dilution of the samples following multiple solution injections (see Methods, SI).

Therefore, we have demonstrated that Quad-Stars assemble into discrete hydrogel aggregates, with sizes ranging from a few to a few tens of micrometers, following an enzyme-free isothermal reaction. Aggregate formation is reversible and responds to changes in cation concentration, exemplified by addition and chelation of the K+ ions stabilizing the inner G4 motifs.

The amphiphilic nature of Quad-Stars, combined with their reversibility and responsiveness, makes them promising candidates as delivery vectors for hydrophobic therapeutic molecules, which present a unique challenge in drug delivery due to their low solubility and poor bioavailability.[63−66] Here, we demonstrate the ability of Quad-Stars to encapsulate and release rhodamine B, a model hydrophobic molecule which can be easily visualized via fluorescence microscopy.

Rhodamine B was added to samples prepared as for the isothermal assembly experiments, prior to KCl injection. Microscopy images in Figure a,b confirm that Quad-Star aggregates are able to encapsulate hydrophobic rhodamine B, as demonstrated by the higher fluorescence intensity detected within the aggregates as compared to the background: 27.2 ± 4.7% higher for 5G and 34.7 ± 3.4% higher for 6G.

Figure 4

Uptake and release of model hydrophobic cargo rhodamine B. (a, b) Epifluorescence micrographs of 5G and 6G Quad-Star aggregates, respectively, at the end of a 3.5 h assembly run in the presence of rhodamine B. Hydrogel aggregates appear brighter than the background, proving preferential segregation of rhodamine B within the hydrophobic cholesterol pockets. (c) Representative profiles illustrating aggregate disassembly (A.F. = area fraction) upon cryptand injection (red vertical line), thereby causing the release of the rhodamine B cargo (N.B.I. = normalized background intensity). Profiles are extracted from single FOVs of representative samples, as release timescales strongly depend on the proximity of the injection to the FOV (see Figure S15, SI). (d) Light-induced disassembly via NMM photocleavage. NMM excitation (395 nm) leads to light-induced damage to G4 motifs and disassembly of hydrogel aggregates. Insets are composite images (bright-field and NMM fluorescence) corresponding to time steps indicated by arrows. Data are presented as mean (circles) ± standard deviation (shaded interval) of three FOVs within the same sample (6G Quad-Stars). (e) Photoinduced engraving in a sample well, reading “IC” (Imperial College), obtained by automatically exposing the pattern to 395 nm LED, thereby triggering location-selective NMM-mediated photo-disassembly. Note that the underlying square-lattice pattern is an artifact of slightly uneven illumination and FOV stitching, rather than representing a physical characteristic of the sample. Scale bars are 50 μm, unless otherwise specified. Epifluorescence micrographs are unprocessed. Bright-field insets have been contrast-enhanced for ease of visualization. See Figure S16a,b (SI) for the raw images.

Disassembly of the cargo-loaded hydrogels through K+ chelation leads to release of rhodamine B into solution. As demonstrated in Figure c, we followed the kinetics of Quad-Star network disassembly and cargo release simultaneously by monitoring the aggregate area fraction and background fluorescence intensity, respectively. Here, the cryptand solution was injected in the sample well far from the imaged FOV, hence the time delay between injection and disassembly, which is then observed as a sharp drop in the area fraction. Simultaneously, the background intensity displays a sharp increase, as rhodamine is released from the aggregates within the FOV and nearby locations in the sample, before slowly decreasing as the excess dye diffuses away from the site of release.

To substantiate the applicability of Quad-Stars as drug delivery carriers, their stability in model physiological conditions was tested. Aggregates were preassembled in K+-rich buffer as previously described, before undergoing supernatant removal and exchange with either phosphate buffered saline (PBS) 1× , which better reflects typical cellular ionic conditions,[67] or cell culture medium (Dulbecco’s modified Eagle medium, DMEM) supplied with 10% or 30% fetal bovine serum (FBS), as previously done to probe the biostability of DNA nanostructures.[68−70] Samples were monitored over the course of 60 h at room temperature. Undisturbed Quad-Star aggregates in K+-rich buffers kept growing and coarsening into a layer nearly covering the bottom of the well (Figure S17a,b, Videos S9 and S10 for 5G/6G, respectively, in the SI). Aggregates in PBS 1× neither did visibly grow nor disassemble, but coarsened and coalesced into larger microgels, indicative of an enhanced fluidity (Figure S17c,d, Videos S11 and S12 for 5G/6G, respectively, in the SI). Finally, hydrogel particles in DMEM + 10% or 30% FBS fully retained their initial morphology (Figure S18a,b, Videos S13 and S14 for FBS 10% and Figure S18c,d, Videos S15 and S16 for FBS 30%, for 5G/6G, respectively, in the SI). These results confirm the stability of Quad-Star aggregates in model physiological conditions, and thus their potential as drug delivery carriers.

Besides enabling reversible control over aggregate formation depending on ionic conditions, the bridging G4 element of Quad-Stars makes it possible to trigger disassembly through a different, orthogonal stimulus, namely UV irradiation, in the presence of NMM. While typically used as a fluorescent G4 probe (as in Figures and 3b), NMM has recently been reported to induce G4-photocleavage upon strong UV exposure.[40]

To test this response, we exposed assembled 6G Quad-Star hydrogels in the presence of NMM to long-wave UV (395 nm), and then monitored the aggregates over time in both fluorescence (NMM emission at 670 nm) and bright-field.

As shown in Figure d, we observe rapid disassembly of Quad-Star aggregates following injection of NMM under UV exposure, with complete dissolution occurring within 90 s. Aggregate disassembly is observed in both fluorescence and bright-field (see Figure d, insets), proving that the observed effect is not due to NMM photobleaching, as also confirmed by fluorescence imaging of Alexa488-labeled samples (Figure S16c, SI). Disassembly is found to be irreversible, in agreement with literature reporting that NMM induces selective photocleavage at G4 sites following guanine photo-oxidation.[40−44]

A key advantage of light-induced disassembly is the possibility of activating a spatially localized response. To demonstrate this feature, we used the motorized microscope stage to expose a portion of a large 6G sample shaped as the letters “IC”. Figure e shows selective disassembly of the Quad-Star aggregates in the exposed region and the emergence of the sought pattern.

In summary, Quad-Stars combine the robustness and versatility of amphiphilic DNA nanostars with the stimuli-responsiveness of noncanonical G4 motifs, thus allowing for controlled assembly/disassembly into hydrogel aggregates in response to biologically relevant K+ cations.

We can tune the folding rates of individual motifs as well as the overall rate at which Quad-Stars aggregate into hydrogel particles via multiple parameters, namely potassium and oligonucleotide concentration, and the number of stacked G-quartets in the bridging G4s. The formed phases are stable under model physiological conditions and can naturally load small hydrophobic molecules, releasing them upon changes in the ionic environment. Therefore, we believe this material could be useful as the basis of drug delivery systems, exploiting cation concentration gradients existing among intra- and extracellular compartments. Moreover, in combination with a photosensitizing porphyrin ligand, Quad-Star phases can be made to irreversibly disassemble upon localized UV exposure, unlocking a further control mechanism for cargo release.

Besides the direct applicability of these responsive nanostructured phases to encapsulation and controlled release of biomedically relevant cargoes, Quad-Star aggregates may find application as scaffolds for artificial, cell-like agents that many seek to construct in the context of bottom-up synthetic biology.[71−74] Here, the responsive hydrogels may act as membrane-less compartments, creating microenvironments which host active components whose action could couple with the responsiveness of the networks, hence achieving a unique synergy between structure and functionality.[75−77]

To this end, and to further augment the range of factors regulating Quad-Star assembly/disassembly, we envisage the possibility of exploiting helicase enzymes, either present in solution or tethered in the vicinity of the quadruplex-forming sites, or even potentially produced via cell-free protein expression systems, to selectively unfold G4s.[78−81] The range of available responses could be further expanded thanks to the vast array of synthetic chemical species developed to tag, stabilize, or destabilize G-quadruplexes for molecular biology assays or therapeutics.[82]

Finally, we note that while this contribution focused on the assembly/disassembly thermodynamics and kinetics of Quad-Star phases, the nanostructural properties of these materials remain to be discovered, particularly with respect to the ability of “regular” cholesterolized nanostars to form crystalline phases depending on design parameters such as flexibility and size.[33,34] Quad-Stars could display a similar propensity toward the formation of ordered phases, but the added complexity of the geometrical and steric constraints imposed by the G4 motif could lead to the emergence of novel crystal geometries.