Milan N. Stojanovic

Molecular robots guided by prescriptive landscapes

Traditional robots rely for their function on computing, to store internal representations of their goals and environment and to coordinate sensing and any actuation of components required in response. Moving robotics to the single-molecule level is possible in principle, but requires facing the limited ability of individual molecules to store complex information and programs. One strategy to overcome this problem is to use systems that can obtain complex behaviour from the interaction of simple robots with their environment. A first step in this direction was the development of DNA walkers, which have developed from being non-autonomous to being capable of directed but brief motion on one-dimensional tracks. Here we demonstrate that previously developed random walkers—so-called molecular spiders that comprise a streptavidin molecule as an inert ‘body’ and three deoxyribozymes as catalytic ‘legs’—show elementary robotic behaviour when interacting with a precisely defined environment. Single-molecule microscopy observations confirm that such walkers achieve directional movement by sensing and modifying tracks of substrate molecules laid out on a two-dimensional DNA origami landscape. When using appropriately designed DNA origami, the molecular spiders autonomously carry out sequences of actions such as ‘start’, ‘follow’, ‘turn’ and ‘stop’. We anticipate that this strategy will result in more complex robotic behaviour at the molecular level if additional control mechanisms are incorporated. One example might be interactions between multiple molecular robots leading to collective behaviour; another might be the ability to read and transform secondary cues on the DNA origami landscape as a means of implementing Turing-universal algorithmic behaviour.

A deoxyribozyme-based molecular automaton.

We describe a molecular automaton, called MAYA, which encodes a version of the game of tic-tac-toe and interactively competes against a human opponent. The automaton is a Boolean network of deoxyribozymes that incorporates 23 molecular-scale logic gates and one constitutively active deoxyribozyme arrayed in nine wells (3×3) corresponding to the game board. To make a move, MAYA carries out an analysis of the input oligonucleotide keyed to a particular move by the human opponent and indicates a move by fluorescence signaling in a response well. The cycle of human player input and automaton response continues until there is a draw or a victory for the automaton. The automaton cannot be defeated because it implements a perfect strategy.

Fluorescent Sensors Based on Aptamer Self-Assembly

In vitro selection of oligonucleotides has made possible the isolation of aptamers with high binding affinity for small molecules, proteins, or even whole cells.1 Recently, aptamers were demonstrated to rival antibodies in analytical methods based on heterogeneous assays.2 Aptamers often undergo a conformational change upon ligand binding and, if so, can be converted to fluorescent sensors by either modification with fluorescent oligonucleotide analogs3 or double-end-labeling with donor and acceptor fluorophores.4 We describe herein a simple and potentially general approach to the rational construction of sensors5 based on fluorophore-labeled heterodimeric aptamers that assemble as a function of ligand concentration. Oligonucleotide aptamers generally consist of a single oligonucleotide chain comprised of two partially complementary domains. These domains are connected through loops, and very often these loops are not essential for ligand binding. We hypothesized that removal of such loop regions would convert the aptamer into subunits that could, nonetheless, reassemble to form the ligand-binding pocket. We were encouraged by a precedent in which an anti-adenosine ribonucleotide triphosphate rATP aptamer consisting of two subunits was still able to bind to an rATP affinity column.6 We further assumed that, by adjusting the complementarity, we would be able to tune the system to an equilibrium that favors individual subunits over assembled heterodimer but that shifts toward the latter upon the binding of ligand. If each subunit were labeled with a fluorophore, then proximity-dependent communication of the fluorophores could be used to assess the position of the equilibrium and hence the concentration of the ligand. Ligand-dependent self-assembly of aptamers is analogous to the assembly of heavy and light chains of an Fv antibody fragment through ligand binding that was used to develop an “open sandwich fluoroimmunoassay”.7 However, in the case of aptamers, the subunit interactions can be adjusted easily by modifying the length of complementary regions, and therefore the subunit affinity required for ligand-dependent assembly can be attained more predictably. In the course of constructing sensors capable of reporting the activity of anticocaine catalytic antibodies,8 we developed oligodeoxynucleotide aptamers for cocaine (1) by standard methods.1 To construct a hetorodimeric aptamer, we began with a cocainebinding 39-mer with Kd ) 5 μM at c(Mg2+) ) 1 mM and, based on the suggested secondary structure, separated it at a predicted loop into two subunits9 (apparent Kd of the self-assembled aptamer, ∼200 μM by equilibrium gel filtration).10 We labeled one subunit with a 5′-6-carboxy-fluorescein fluorophore (6-FAM, F in F-C1) and the other with a 3′-dabcyl quencher (C2-D) (Figure 1). Dabcyl is a universal, nonfluorescent, non-Förster quencher that is used in molecular beacons to follow the intramolecular hybridization of the 5′and 3′-ends.11 Dabcyl is characterized by an efficient π-overlap quenching in the hybridized state and little energy transfer quenching in the nonhybridized state. As predicted, in the selection buffer (c(TRIS) ) 20 mM, pH ) 7.4, c(NaCl) ) 140 mM) at the optimum chain concentrations of c(F-C1) ) 10 nM and c(C2-D) ) 60 nM and c(Mg2+) between 0 and 4 mM, the two subunits behaved as a self-assembling fluorescent cocaine sensor. The sensor reliably reported concentrations of cocaine in the range from 10 to 1250 μM with fluorescein emission at 518 nm (λex ) 472) that was quenched to 65% of the initial value (Figure 2). The concentration range can be shifted to 1 μM by employing higher concentrations of Mg2+ or by prolonging incubation times. This sensor showed excellent selectivity for cocaine over its metabolites benzoyl ecgonine (2) and ecgonine methyl ester (3) (Figure 2). Two important negative controls support a mechanism of action based on ligand-driven association of subunits. First, when C2 without dabcyl was substituted for C2-D, no fluorescence quenching was observed. Second, when C2-D was combined with an analogue of F-C1 incapable of binding cocaine due to substitution of the N12-N20 segment d(ATGAAGTGG) with d(AAAAAAAAA), no cocaine-dependent quenching was (1) Hermann, T.; Patel, D. J. Science 2000, 287 (5454), 820-825 and references therein. (2) (a) Jayasena S. D. Clin. Chem. 1999, 45 (9), 1628-1650. For homogeneous competitive assays based on aptamers, see: (b) Wang, Y.; Killian, J.; Hamasaki, K.; Rando, R. R. Biochemistry 1996, 35 (38) 1233812346. (c) Jayasena, S.; Gold, G. PCT Int. Appl. WO 9931276 A1, CAN: 131: 54725, 1999. (3) Jhaveri, S. D.; Kirby, R.; Conrad, R.; Maglott, E. J.; Bowser, M.; Kennedy, R. T.; Glick, G.; Ellington, A. D. J. Am. Chem. Soc. 2000, 122, 2469-2473. (4) Stojanovic, M. N.; de Prada, P.; Landry, D. W., manuscript in preparation. (5) De Silva, A. P.; Gunaratne, H. Q. N.; Gunnlaugsson, T.; Huxley, A. J. M.; McCoy, C. P.; Rademacher, J. T.; Rice, T. E. Chem. ReV. 1997, 97 (5), 1515-1566 and references therein. (6) Huizenga, D. E.; Szostak, J. W. Biochemistry 1995, 34, 656-665. (7) (a) Ueda, H.; Kubota, K.; Wang, Y.; Tsumoto, K.; Mahoney, W.; Kumagai, I.; Nagamune, T. BioTechniques 1999, 27, 738-742. (b) Ueda, H.; Tsumoto, K.; Kubota, K.; Suzuki, E.; Nagamune, T.; Nishimura, H.; Schueler, P. A.; Winter, G.; Kumagai, I.; Mahoney, W. C. Nat. Biotechnol. 1996, 14, 1714-1718. (8) Yang, G.; Arakawa-Uramoto, H.; Wang, X.; Gawinowicz, M. A.; Zhao, K; Landry, D. W. J Am. Chem. Soc. 1996, 118, 5881-5890. (9) All oligonucleotides were custom-made and purified by Integrated DNA Technologies (Coralville, IA) and were used as received. (10) Isolation and full characterization of cocaine-binding aptamers will be reported separately: Stojanovic, M. N.; de Prada, P.; Landry, D. W., manuscript in preparation. (11) Tyagi, S.; Bratu, D. P.; Kramer, F. R. Nat. Biotechnol. 1998, 16 (1), 49-58. Figure 1. Self-assembly of two aptamer subunits F-C1 and C2-D in the presence of cocaine (1): D, dabcyl quencher; F, 6-FAM; smaller font F quenched 6-FAM. The secondary structure of the aptamer is based on mutational analysis.1

Catalytic Molecular Beacons

We have constructed catalytic molecular beacons from a hammerhead‐type deoxyribozyme by a modular design. The deoxyribozyme was engineered to contain a molecular beacon stem–loop module that, when closed, inhibits the deoxyribozyme module and is complementary to a target oligonucleotide. Binding of target oligonucleotides opens the beacon stem–loop and allosterically activates the deoxyribozyme module, which amplifies the recognition event through cleavage of a doubly labeled fluorescent substrate. The customized modular design of catalytic molecular beacons allows for any two single‐stranded oligonucleotide sequences to be distinguished in homogenous solution in a single step. Our constructs demonstrate that antisense conformational triggers based on molecular beacons can be used to initiate catalytic events. The selectivity of the system is sufficient for analytical applications and has potential for the construction of deoxyribozyme‐based drug delivery tools specifically activated in cells containing somatic mutations.

Exercises in molecular computing.

Conspectus The successes of electronic digital logic have transformed every aspect of human life over the last half-century. The word “computer” now signifies a ubiquitous electronic device, rather than a human occupation. Yet evidently humans, large assemblies of molecules, can compute, and it has been a thrilling challenge to develop smaller, simpler, synthetic assemblies of molecules that can do useful computation. When we say that molecules compute, what we usually mean is that such molecules respond to certain inputs, for example, the presence or absence of other molecules, in a precisely defined but potentially complex fashion. The simplest way for a chemist to think about computing molecules is as sensors that can integrate the presence or absence of multiple analytes into a change in a single reporting property. Here we review several forms of molecular computing developed in our laboratories. When we began our work, combinatorial approaches to using DNA for computing were used to search for solutions to constraint satisfaction problems. We chose to work instead on logic circuits, building bottom-up from units based on catalytic nucleic acids, focusing on DNA secondary structures in the design of individual circuit elements, and reserving the combinatorial opportunities of DNA for the representation of multiple signals propagating in a large circuit. Such circuit design directly corresponds to the intuition about sensors transforming the detection of analytes into reporting properties. While this approach was unusual at the time, it has been adopted since by other groups working on biomolecular computing with different nucleic acid chemistries. We created logic gates by modularly combining deoxyribozymes (DNA-based enzymes cleaving or combining other oligonucleotides), in the role of reporting elements, with stem–loops as input detection elements. For instance, a deoxyribozyme that normally exhibits an oligonucleotide substrate recognition region is modified such that a stem–loop closes onto the substrate recognition region, making it unavailable for the substrate and thus rendering the deoxyribozyme inactive. But a conformational change can then be induced by an input oligonucleotide, complementary to the loop, to open the stem, allow the substrate to bind, and allow its cleavage to proceed, which is eventually reported via fluorescence. In this Account, several designs of this form are reviewed, along with their application in the construction of large circuits that exhibited complex logical and temporal relationships between the inputs and the outputs. Intelligent (in the sense of being capable of nontrivial information processing) theranostic (therapy + diagnostic) applications have always been the ultimate motivation for developing computing (i.e., decision-making) circuits, and we review our experiments with logic-gate elements bound to cell surfaces that evaluate the proximal presence of multiple markers on lymphocytes.

Autonomous molecular cascades for evaluation of cell surfaces

Molecular automata are mixtures of molecules that undergo precisely defined structural changes in response to sequential interactions with inputs1–4. Previously studied nucleic acid-based-automata include game-playing molecular devices (MAYA automata3,5) and finite-state automata for analysis of nucleic acids6 with the latter inspiring circuits for the analysis of RNA species inside cells7,8. Here, we describe automata based on strand-displacement9,10 cascades directed by antibodies that can analyze cells by using their surface markers as inputs. The final output of a molecular automaton that successfully completes its analysis is the presence of a unique molecular tag on the cell surface of a specific subpopulation of lymphocytes within human blood cells.

Specific capture and temperature-mediated release of cells in an aptamer-based microfluidic device

Isolation of cells from heterogeneous mixtures is critically important in both basic cell biology studies and clinical diagnostics. Cell isolation can be realized based on physical properties such as size, density and electrical properties. Alternatively, affinity binding of target cells by surface-immobilized ligands, such as antibodies, can be used to achieve specific cell isolation. Microfluidics technology has recently been used in conjunction with antibody-based affinity isolation methods to capture, purify and isolate cells with higher yield rates, better efficiencies and lower costs. However, a method that allows easy release and collection of live cells from affinity surfaces for subsequent analysis and detection has yet to be developed. This paper presents a microfluidic device that not only achieves specific affinity capture and enrichment, but also enables non-destructive, temperature-mediated release and retrieval of cells. Specific cell capture is achieved using surface-immobilized aptamers in a microchamber. Release of the captured cells is realized by a moderate temperature change, effected via integrated heaters and a temperature sensor, to reversibly disrupt the cell-aptamer interaction. Experimental results with CCRF-CEM cells have demonstrated that the device is capable of specific capture and temperature-mediated release of cells, that the released cells remain viable and that the aptamer-functionalized surface is regenerable.

Recognition and sensing of low-epitope targets via ternary complexes with oligonucleotides and synthetic receptors.

Oligonucleotide-based receptors or aptamers can interact with small molecules, but the ability to achieve high-affinity and selectivity of these interactions depends strongly on functional groups or epitopes displayed by the binding targets. Some classes of targets are particularly challenging: for example, monosaccharides have scarce functionalities and no aptamers have been reported to recognize, let alone distinguish from each other, glucose and other hexoses. Here we report aptamers that differentiate low-epitope targets such as glucose, fructose, or galactose by forming ternary complexes with high-epitope organic receptors for monosaccharides. In a follow-up example, we expand this method to isolate high-affinity oligonucleotides against aromatic amino acids complexed in situ with a non-specific organometallic receptor. The method is general and enables broad clinical use of aptamers for detection of small molecules in mix-and-measure assays, as demonstrated by monitoring postprandial waves of phenylalanine in human subjects.

Molecular Computing with Deoxyribozymes

Publisher Summary This chapter discusses the molecular computing with deoxyribozymes. Silicomimetic molecular computing is based on an idea that individual molecules can perform basic logical operations and make simple decisions based on the presence or absence of multiple factors in solution. Using deoxyribozymes, nucleic acid catalysts made of DNA, and recognition regions for oligonucleotides—stem-loops—a so-called “full set of molecular logic gates,” was constructed in the study described in the chapter, which allowed for the combination of individual gates into more complex circuits. These gates and their circuits analyze the sets of oligonucleotides as inputs and produce changed substrate oligonucleotides as outputs. The chapter describes the initial efforts to integrate molecular computing devices with more traditional approaches to nanomedicine, such as those using nanoparticles for drug-delivery. This approach opens possibilities to increase the functional complexity of delivery systems. A three-layer cascade is described, in which microscopic particles coordinate their activity without any direct physical contact. The elementary unit of a network of microparticles is a single particle covered with a DNA computing or sensing element. Individual bead senses the presence of an input stimulus (or multiple stimuli) in solution, and, according to a set of rules encoded on this bead by computing elements, it releases an oligonucleotide signal as an output through a catalytic process.

Light-up properties of complexes between thiazole orange-small molecule conjugates and aptamers

The full understanding of dynamics of cellular processes hinges on the development of efficient and non-invasive labels for intracellular RNA species. Light-up aptamers binding fluorogenic ligands show promise as specific labels for RNA species containing those aptamers. Herein, we took advantage of existing, non-light-up aptamers against small molecules and demonstrated a new class of light-up probes in vitro. We synthesized two conjugates of thiazole orange dye to small molecules (GMP and AMP) and characterized in vitro their interactions with corresponding RNA aptamers. The conjugates preserved specific binding to aptamers while showing several 100-fold increase in fluorescence of the dye (the ‘light-up’ property). In the presence of free small molecules, conjugates can be displaced from aptamers serving also as fluorescent sensors. Our in vitro results provide the proof-of-concept that the small-molecule conjugates with light-up properties can serve as a general approach to label RNA sequences containing aptamers.