Transmutable nanoparticles with reconfigurable surface ligands

See allHide authors and affiliations

Science  05 Feb 2016:
Vol. 351, Issue 6273, pp. 579-582
DOI: 10.1126/science.aad2212

Controlled colloid bonding using DNA

Colloidal particles can act as analogs of atoms for studying crystallization and packing behavior, but they don't naturally bond together the way atoms do. Short strands of DNA are one versatile way to link together colloidal particles (see the Perspective by Tao). Kim et al. designed a series of gold colloids with DNA ligands that reversibly bound to or released neighboring particles via DNA strands that opened or closed hairpin loops. Liu et al. devised a set of DNA strands that pack into origami structures. Inside each structure were strands that cage a gold nanoparticle. These were further linked to other uncaged nanoparticles to assemble a diamond-like structure. Changing the strand design yielded a wide range of sparsely packed colloidal crystals.

Science, this issue p. 561, p. 579; see also p. 582


Unlike conventional inorganic materials, biological systems are exquisitely adapted to respond to their surroundings. Proteins and other biological molecules can process a complex set of chemical binding events as informational inputs and respond accordingly via a change in structure and function. We applied this principle to the design and synthesis of inorganic materials by preparing nanoparticles with reconfigurable surface ligands, where interparticle bonding can be programmed in response to specific chemical cues in a dynamic manner. As a result, a nascent set of “transmutable nanoparticles” can be driven to crystallize along multiple thermodynamic trajectories, resulting in rational control over the phase and time evolution of nanoparticle-based matter.

The configurations of valence electrons that largely dictate the structural arrangement of atoms in crystals are more or less immutable. In contrast, the surface ligands that dictate analogous “bonds” in nanoparticle-based crystals are molecular in nature (18) and are therefore sensitive to the presence of various chemical signals. As a result, nanoparticle building blocks offer the possibility for dynamic and rational reconfiguration of the surface ligands in response to external stimuli (9, 10). Consequently, the preference for different nanoparticle components to bond to one another can be switched reversibly, allowing for a set of nascent particles to be driven down specific thermodynamic pathways via chemical binding events. To realize this goal, surface ligands must be capable of processing molecular binding events as inputs while producing changes in the nanoparticle bonding mode as outputs, all in a deterministic and reproducible manner. We show that DNA-based hairpin ligands, which undergo well-defined conformational changes upon binding of effector oligonucleotides, can be used to alter the bonding properties of the nanoparticles to which they are anchored. Inspired by pluripotent stem cells, which are capable of differentiating into multiple biological tissues, we introduce the concept of a transmutable nanoparticle—a building block with different possible binding characteristics that can be selectively activated and deactivated (with the appropriate chemical cues) and then used to generate discrete forms of complex crystalline matter.

Transmutable nanoparticle constructs were created by functionalizing citrate-stabilized gold nanoparticles with a dense monolayer of DNA (1, 1113). Free DNA “linkers” containing a sequence complementary to the nanoparticle-bound strands were then hybridized to these constructs, transforming them into nanoscale programmable atom equivalents (PAEs) (14). Bonding interactions between PAEs have been examined for nanoparticle constructs of various sizes (11, 15), shapes (16, 17), and compositions (1820) and can be determined a priori using a set of well-established design rules (11). Individual bonds between particles are formed via the hybridization of short complementary “sticky ends” that are located at the terminus of each DNA linker (6, 11, 12). Therefore, the bonding identity of a particle is dictated by its sticky ends—the base sequence, the number of strands attached to a particle, and the location of the sticky ends relative to the particle surface.

We studied four factors that are important in dictating the bonding interactions between PAEs: (i) the type of recognition sequence that defines the type of bonding, (ii) the effective stoichiometry of PAEs, (iii) the density of bonding elements on individual PAEs, and (iv) the hydrodynamic size of PAEs. Oligonucleotide-based ligands represent an ideal means to control these bonding properties, as a variety of DNA motifs have been designed that enable structural reconfigurability in both nanoparticle-based and DNA origami–based materials (6, 2131). In this work, dynamic control of particle bonding is enabled by the introduction of multiple “protecting groups” placed adjacent to the sticky ends in the linker sequences; these protecting groups consist of sequences that form hairpin structures (Fig. 1) (2830). When these hairpins are in a “protected” state (where opposite ends of the hairpin sequence hybridize to one another), the sticky ends are buried within the dense DNA monolayer, sterically inhibiting the interparticle hybridization events necessary to mediate particle association. This hairpin structure can undergo a transition to an activated state via the introduction of a short “effector” oligonucleotide that is complementary to the full hairpin sequence. The resulting conformational change pushes the sticky ends to the periphery of the particles, allowing constructs with complementary sticky ends to bind to one another. These effector strands contain a short overhang sequence that does not hybridize to the hairpin (32, 33); this allows the hairpin structures to be reprotected by adding a second oligonucleotide that is fully complementary to the effector strand, resulting in refolding of the hairpin and steric inhibition of the sticky end (fig. S1).

Fig. 1 Scheme of transmutable nanoparticles and their activation pathways.

DNA hairpin–based “protecting groups” disguise terminal sticky end sequences (blue DNA segments) that mediate particle bonding, allowing PAEs to be selectively and reversibly transformed from an unactivated, “transmutable” state (left) to an “activated” state (right) through the binding of sequence-specific effector oligonucleotides (purple strands, right image).

The transmutable nature of these particles is most simply demonstrated by controlling the type of bonding elements (i.e., the sequence of the sticky ends) to be either self-complementary or non–self-complementary. Unary systems with self-complementary sticky ends typically form face-centered cubic (fcc) lattices, whereas binary systems with particles of equal size but complementary sticky ends typically form body-centered cubic (bcc) lattices (1113). Therefore, in principle, the bonding specificity of a particle could be switched between the self-complementary and non–self-complementary [i.e., fcc-favoring or bcc–favoring] states by functionalizing a particle with both types of sticky ends, each with separately addressable protecting groups (Fig. 2A and tables S1 to S9).

When both types of sticky ends were protected and hidden, the transmutable nanoparticles remained in their nascent state; small-angle x-ray scattering (SAXS) data confirmed that there were no interparticle hybridization events occurring and that all particles remained dispersed (Fig. 2A, center). When the particles were exposed to DNA strands that opened the hairpin protecting groups and activated the self-complementary sticky ends, the particles immediately assembled and subsequently formed an fcc lattice upon brief annealing (Fig. 2A, bottom, and table S10). Conversely, when the protecting groups adjacent to the non–self-complementary sticky ends were activated, the particles formed a bcc lattice (Fig. 2A, top). No evidence of a bcc lattice was found when the self-complementary sticky ends were deprotected, and vice versa, indicating that the separate hairpin structures were opened in a completely orthogonal manner. In both the fcc- and bcc-forming systems, the introduction of a short oligonucleotide that caused the hairpins to reform resulted in a complete loss of structure, with the particles returning to their nascent state. The particles were able to undergo at least three rounds of deprotection/reprotection to transition to either an fcc or bcc lattice, independent of their previously assembled states; beyond three cycles, extended x-ray beam exposure caused sample degradation. The bonding modes of the particles could even be directly switched in a one-step process, without having to drive the particles to their nascent state, by adding two different oligonucleotides simultaneously—one that deprotected the sticky ends not engaged in interparticle bonding, and one that reprotected the sticky ends already engaged in particle bonding. These transitions were exacted through three cycles, where addition of the oligonucleotides caused near-instantaneous transitions of the bonding specificity of the particles, thereby allowing particles to reversibly switch between bonding modes in a rapid and robust manner (Fig. 2B).

Fig. 2 Programming the type of bonding of transmutable nanoparticles.

(A) Transmutable particles can be driven from an unactivated state (center) to crystallize down distinct thermodynamic pathways by selectively deprotecting self-complementary (green) or non–self-complementary (blue and red) sticky ends, which favor fcc (bottom) or bcc (top) lattice symmetries, respectively. (B) SAXS data for transmutable particles cycled between fcc and bcc phases with no observable structural hysteresis.

Another mechanism for controlling nanoparticle bond identity is to toggle the number of bonds that can be formed between particles (Fig. 3) (11, 34). This can be done either by changing the number of particles available to engage in bonding (Fig. 3A) or by changing the density of bonding elements on the individual particle constructs (Fig. 3B). Both methods can be achieved by using two different protecting groups simultaneously, which enables control over the number of bonds that are activated by the introduction of effector strands. In the first method, transmutable particles are functionalized with a single type of protecting group and are turned “on” or “off” in order to control the relative stoichiometry of particle types in solution. In the second method, transmutable particles possess both types of protecting groups, allowing for the formation of low and high valence states through the activation of one or both sets of protecting groups, respectively.

Fig. 3 Changing particle stoichiometry and linker density of transmutable nanoparticles.

Transmutable particles functionalized with two types of oligonucleotides that contain the same sticky end but different protecting groups can be used to control (A) effective PAE stoichiometry and (B) linker density per particle. (A) At a 1:1 ratio of activated transmutable particles (red) and nontransmutable (blue) particles, a bcc lattice was formed and unactivated particles (gray) remained in the supernatant. At a 5:1 ratio, an AlB2 lattice was formed. (B) Upon the introduction of complementary nontransmutable particles (blue), the low-linker density state favored a bcc lattice, whereas the high-linker density state favored an AlB2 lattice. In both cases, transmutable particles could be repeatedly cycled between the different states.

The first method was examined by designing a system where bcc and AlB2 lattices would be predicted to exhibit similar stabilities. In such a system, it has been shown that the molar ratio of particles influences the structure that is obtained; a 1:1 molar ratio of particles results in the formation of a bcc lattice, whereas a 5:1 molar ratio results in the formation of an AlB2 lattice (11). To test this capability, we synthesized two types of transmutable particles, each with a unique protecting group sequence but the same sticky end type; 20% contained one protecting group and 80% contained a second protecting group. This batch of particles was then combined in a 5:1 ratio with nontransmutable particles containing a complementary sticky end. When the first protecting group type was activated, complementary particles existed in a 1:1 ratio and a bcc lattice was synthesized; the particles that were not deprotected remained unbound in solution (Fig. 3A, left). Upon unfolding of the second protecting group, the activated and nontransmutable particles existed in a 5:1 ratio, which drove the lattice toward the formation of an AlB2-type structure (Fig. 3A, right) This system could also be driven between the two lattice types in a reversible manner by selectively protecting/deprotecting one or both sets of particles (fig. S2 and table S11).

The second method was examined by designing a system where changing the relative number of sticky ends on complementary particles would favor different phases. Because each transmutable particle in this system included two different protecting groups that could be addressed orthogonally, the particles could access “low-density” and “high-density” states. In the low-density state, the transmutable particles possessed fewer accessible sticky ends than their complements; in the high-density state, the transmutable particles possessed more accessible sticky ends than their complements (Fig. 3B). The former favored a bcc lattice, whereas the latter favored an AlB2 lattice. Note that these crystal structures matched those that would be predicted from a previously established phase diagram for conventional PAEs (11). This indicates that despite the large number of inactive linkers on the particle surface in the low valence state, these protected binding elements did not interfere with the hybridization of the activated sticky ends. As in each of the previous cases, the transitions between different bonding modes were fully reversible (fig. S3).

The final factor investigated in this work was the hydrodynamic size of the PAEs, because a particle’s coordination number is dictated by the relative sizes of the particles bonded to one another (2, 11). Hairpin structures can be used to modify the length of a DNA bond, as the folded state of a hairpin is significantly shorter (3 to 9 nm difference) than the unfolded state (25, 26). In this particular design, the sticky ends were able to form a bond when the hairpins were both folded and unfolded (Fig. 4A). This is because a short duplex region was used to increase the distance between the sticky ends and the hairpins; hence, the sticky ends were not sterically inhibited from bonding when the hairpin was in the folded state (25). As a result, the bonds between particles were not being formed and broken with the introduction of effector strands, but rather the length of the bonds and size of the PAE were being dynamically changed. In these transmutable particles, two hairpins were included within each linker strand to maximize the difference in length between the folded (PAE radius 35 nm) and unfolded (PAE radius 41 nm) states.

Fig. 4 Programming the hydrodynamic size of transmutable particles.

(A) The selective folding or unfolding of hairpin segments in DNA linkers can be used to shorten or lengthen DNA bonding elements, and therefore the effective size of the PAEs. (B) SAXS data confirm that short DNA bonding elements (folded hairpins) favored the formation of AlB2 lattices (state i). Unfolding these hairpins resulted in an increase in interparticle distance and a loss of long-range order (state ii), but annealing caused the particles to reorganize into a Cs6C60 lattice (state iii). Upon refolding the hairpins, the particles followed the reverse trajectory with a decrease in interparticle distance and a loss of long-range order (state iv), but annealing the lattice restored the original AlB2 structure (state v). (C) Interparticle distances measured at each of the states shown in (B) indicated a rapid change in bond length relative to the time scale of crystal formation.

When these size-adjustable transmutable particles were combined with complementary particles that had a fixed radius of 14 nm, the particles’ coordination numbers were different for the folded and unfolded states: An AlB2 lattice was favored for the folded state, whereas a Cs6C60 lattice was favored for the unfolded state. This is because the coordination number of the larger particles is determined by how many smaller particles can physically fit around them (11). Therefore, when the radius of the transmutable particles was extended, the particles were able to achieve a larger coordination number.

This change could be monitored via in situ SAXS measurements, giving insight into how these transmutable particles transitioned between two different states (Fig. 4, B and C). Starting at the folded state (AlB2 lattice), when the hairpin structures were unfolded, the lattice retained the same basic structure, but with expanded lattice parameters and a substantial loss of long-range order (indicated by a shift to a smaller q values and broadening in the scattering peaks). After a brief annealing period, the lattices transformed into a Cs6C60 structure with long-range order. The reverse transition occurred in a similar manner, where the readjustment of the lattice parameters occurred first (with a corresponding loss of long-range order), followed by reorganization and change in particle coordination number (fig. S4). From this observation, it can be concluded that changes to the bonding nature of the particles occur on a faster time scale than the lattice formation and reorganization processes. Indeed, the bonding mode appears to change nearly instantaneously upon introduction of the appropriate chemical stimuli, whereas the crystallization process for these transmutable particles occurs on a similar time scale and manner as previous PAE constructs (35). Taken together, these data indicate that activated transmutable PAEs are nearly indistinguishable from nontransmutable PAEs, enhancing their utility in materials synthesis schemes.

We have used the programmable nature of DNA hairpins and the concept of nanoparticle-based PAEs to develop constructs with bonding behaviors that can be dynamically modulated in response to specific chemical stimuli. The resulting structural plasticity manifested in transmutable particles delineates the power and potential to control PAE architectures, and lays an important foundation for more complex and exotic forms of adaptive matter.

Supplementary Materials

Materials and Methods

Supplementary Text

Figs. S1 to S4

Tables S1 to S11

References and Notes

Acknowledgments: This material is based on work supported by Air Force Office of Scientific Research award FA9550-11-1-0275 (C.A.M., Y.K., R.J.M., M.R.J.) and by the Center for Bio-Inspired Energy Science (CBES), an Energy Frontier Research Center funded by the U.S. Department of Energy, Office of Science, Basic Energy Sciences under award DE-SC0000989-0002 (C.A.M., Y.K.). Y.K., R.J.M., and M.R.J. acknowledge the Ryan Fellowship at Northwestern University. M.R.J. acknowledges the NSF for a graduate research fellowship. Portions of this work were carried out at the DuPont-Northwestern-Dow Collaborative Access Team (DND-CAT) beamline located at Sector 5 of the Advanced Photon Source. DND-CAT is supported by E. I. DuPont de Nemours & Co., Dow Chemical Company, and the state of Illinois.
View Abstract


Navigate This Article