Nonlinear decision-making with enzymatic neural networks

Artificial neural networks have revolutionized electronic computing. Similarly, molecular networks with neuromorphic architectures may enable molecular decision-making on a level comparable to gene regulatory networks^1,2. Non-enzymatic networks could in principle support neuromorphic architectures, and seminal proofs-of-principle have been reported^3,4. However, leakages (that is, the unwanted release of species), as well as issues with sensitivity, speed, preparation and the lack of strong nonlinear responses, make the composition of layers delicate, and molecular classifications equivalent to a multilayer neural network remain elusive (for example, the partitioning of a concentration space into regions that cannot be linearly separated). Here we introduce DNA-encoded enzymatic neurons with tuneable weights and biases, and which are assembled in multilayer architectures to classify nonlinearly separable regions. We first leverage the sharp decision margin of a neuron to compute various majority functions on 10 bits. We then compose neurons into a two-layer network and synthetize a parametric family of rectangular functions on a microRNA input. Finally, we connect neural and logical computations into a hybrid circuit that recursively partitions a concentration plane according to a decision tree in cell-sized droplets. This computational power and extreme miniaturization open avenues to query and manage molecular systems with complex contents, such as liquid biopsies or DNA databases.

Microscopic agents programmed by DNA circuits

Information stored in synthetic nucleic acids sequences can be used in vitro to create complex reaction networks with precisely programmed chemical dynamics. Here, we scale up this approach to program networks of microscopic particles (agents) dispersed in an enzymatic solution. Agents may possess multiple stable states, thus maintaining a memory and communicate by emitting various orthogonal chemical signals, while also sensing the behaviour of neighbouring agents. Using this approach, we can produce collective behaviours involving thousands of agents, for example retrieving information over long distances or creating spatial patterns. Our systems recapitulate some fundamental mechanisms of distributed decision making and morphogenesis among living organisms and could find applications in cases where many individual clues need to be combined to reach a decision, for example in molecular diagnostics.

High-resolution mapping of bifurcations in nonlinear biochemical circuits

Analog molecular circuits can exploit the nonlinear nature of biochemical reaction networks to compute low-precision outputs with fewer resources than digital circuits. This analog computation is similar to that employed by gene-regulation networks. Although digital systems have a tractable link between structure and function, the nonlinear and continuous nature of analog circuits yields an intricate functional landscape, which makes their design counter-intuitive, their characterization laborious and their analysis delicate. Here, using droplet-based microfluidics, we map with high resolution and dimensionality the bifurcation diagrams of two synthetic, out-of-equilibrium and nonlinear programs: a bistable DNA switch and a predator-prey DNA oscillator. The diagrams delineate where function is optimal, dynamics bifurcates and models fail. Inverse problem solving on these large-scale data sets indicates interference from enzymatic coupling. Additionally, data mining exposes the presence of rare, stochastically bursting oscillators near deterministic bifurcations.

Biomimetic copper(I)--CO complexes: a structural and dynamic study of a cali

Four novel calix[6]arene-based cuprous complexes are described. They present a biomimetic tris(imidazole) coordination core associated with a hydrophobic cavity that wraps the apical binding site. Each differs from the other by the methyl or ethyl substituents present on the phenoxyl groups (OR1) and on the imidazole arms (NR2) of the calix[6]arene structure. In solution, stable CO complexes were obtained. We have investigated their geometrical and dynamic properties with respect to the steric demand. IR and NMR studies revealed that, in solution, these complexes adopted two distinct conformations. The preferred conformation was dictated only by the size of the OR1 group. When R1 was an ethyl group, the complex preferentially adopted a flattened C3-symmetrical structure. The corresponding helical enantiomers were in conformational equilibrium, which, however, was slow on the 1H NMR time scale at -80 degrees C. When R1 was a methyl group, the low-temperature NMR spectra revealed the partial inclusion of one tBu group. The complex wobbled between three dissymmetric but equivalent conformations. Hence, small differences in the steric demand of the calixarene's skeleton changed the geometry and dynamics of the system. Indeed, this supramolecular control was promoted by the strong conformational coupling between the metal center and the host structure. Interestingly, this was not only the result of a covalent preorganization, but also stemmed from weak interactions within the hydrophobic pocket. The vibrational spectra of the bound CO were revealed to be a sensitive gauge of this supramolecular behavior, similar to copper proteins in which allosteric effects are common.