Nucleic acid nanotechnology is an incredibly versatile platform for constructing rationally-designed molecular systems. Much of the current attention is focussed on self-assembling structures that are designed to reach a static equilibrium, or computational architectures that compute a single function by relaxing to equilibrium.

One of the signature features of living systems, however, is that they operate continuously rather than relaxing to equilibrium. The circuits built by synthetic biologists from re-purposed cellular components reflect this principle. Components are continuously produced and degraded, and the system is dynamically responsive to a changing environment. Feedback control architectures, which reject external perturbations to maintain certain properties of key molecular species, are a particularly elegant example.

Other researchers seek to mimic specific functions of living systems –- such as replication, or cellular division –- using de novo chemical designs. In doing so, they grapple with the challenges of designing and controlling far-from equilibrium systems with innovative chemical motifs.

In this Introduction, Dr Ouldridge goes into these ideas in more detail and gives an overview of how his group is addressing them.

Programming cell-free synthetic biology diagnostics with RNA/DNA nanotechnologies

Poor water quality affects over two billion people across the globe. While we can’t often see or taste water contaminants, molecular machinery can. Here I will present our latest research on developing a ‘pregnancy test for water’ - a cheap, fast and reliable approach that allows anyone, anywhere to detect if their water is contaminated. Our approach builds off of advances in cell-free synthetic biology – extracting the machinery of natural organisms to perform their functions but in test tube reactions instead of living cells. Using the principles of synthetic biology, we can ‘rewire’ natural biosensors to produce visible signals when specific contaminants are present in a water sample. The addition of synthetic RNA genetic circuits optimizes these reactions to detect contaminants with high degrees of sensitivity and specificity. We have also expanded this platform by creating an interface with toe-hold mediated strand displacement circuits, a dynamic DNA nanotechnology that enables molecular computation through programmable interactions between nucleic acid strands. This work is creating a new application area for cell free synthetic biology and nucleic acid molecular programming that promises to increase the scale at which we can monitor the health of ourselves and our environment.

Sequence design for gene expression control and DNA computing with machine learning

Machine learning models that accurately predict function from sequence are becoming a powerful tool for molecular design. In this talk, I will first present our work on building models that can predict gene function (e.g. translation rate) from DNA sequence. What sets our approach apart from earlier work is that computational models are trained on very large synthetic gene libraries with millions of members. The massive size of the training data allows us to improve upon models trained exclusively on genomic data. We then use these predictors together with novel design algorithms to identify regulatory elements that result in precisely defined levels of gene expression. In the second part of the talk, I will briefly introduce a technique for executing similarity search over a DNA-based database of 1.6 million images. Queries are implemented as hybridization probes, and a key step in our approach is to learn an image-to-sequence encoding ensuring that queries preferentially bind to targets representing visually similar images.

Mitigation of context-dependence in mammalian synthetic biology

Engineering biology has tremendous potential to impact a number of applications, from energy, to environment, to health. As the sophistication of engineered biological networks increases, the ability to predict system behavior becomes more limited. In fact, while a system’s component may be well characterized in isolation, the salient properties of this component often change in rather surprising ways once it interacts with other components in the cell or when the intra-cellular environment changes. This context-dependence of biological circuits makes it difficult to perform rational design and often leads to lengthy, combinatorial, design procedures where each component is re-designed ad hoc when other parts are added to a system. In this talk, I will overview some causes of context-dependence, focusing mostly on mammalian genetic circuits, and will demonstrate how the problem of insulating a genetic module from context can be mathematically formulated as a control theoretic problem of disturbance attenuation. I will show two solutions in mammalian cells: feedforward control and feedback control. In particular, the feedback control architecture uses a covalent modification cycle as a key process to reach quasi-integral control, thus enabling attenuation of various disturbances arising from changes in intra-cellular context. These solutions support rational and modular design of sophisticated genetic circuits and can serve for engineering mammalian biological circuits that are more robust and predictable.

9 years, 2 months, 16 Days to Go: Progress on Implementing Synthetic Cells

On 15 June 2010, I gave a talk at the 2nd International Workshop on Bio-Design Automation in which I declared that I wanted to see the creation of a fully programmed nanoscale machine, with all components specified and designed by engineers (no magic components). It should be a ``general purpose" machine with capabilities that include sensing of environment, import/export of chemicals, communications (with other machines), locomotion, and energy harvesting/metabolism. In this talk I will give an update on various elements of creating such a machine, including our progress and frustrations with various subsystems. Depending on time, interest, and how early in the morning I have to get up, I’ll touch on metabolism, input/export machinery, computational modeling and design tools, and preliminary ideas about how to implement locomotion. This work is done jointly with the Build-A-Cell consortium in the US, which is an open collaboration supporting the science and engineering of building synthetic cells.

Molecular networks as tools to optimize enzymes

Artificial molecular networks are rationally designed to process information at the molecular level. They can convert molecular signal, amplify them, apply specific functions (e.g. non-linear thresholding) or combine multiple signals. Although our ability to construct these networks mostly relies on nucleic acid chemistry and biochemistry, DNA-based molecular computers can be connected upstream and downstream to other chemistries or molecular processes. Additionally microfluidic partitioning allows to use a single network to perform millions or billions of parallel operations in independent compartments.

We have built PEN DNA networks to connect an enzymatic activity (upstream) and the amplification of the gene encoding that activity (downstream). Because evolution is based on a loop from phenotype to genotype and back, these networks, together with a protein expression process, allow us to program enzyme evolution, toward targets dictated by an artificial program. 

DNA Nanoscience on the Surface of Cells

The field of DNA nanoscience has achieved many astonishing capabilities, such as moderate molecular-scale computation and complex 3D self-assembled nanostructures. Biological environments present a very challenging, but also potentially promising domain of application of DNA nanoscience. While in vivo degradation is an obstacle for DNA nanodevices, it is not insurmountable, and the benefits of DNA computing can still be applied to in vitro assays. We describe two aspects of our current work on DNA-based cancer identification and targeted drug delivery.

First, DNA hairpins implementing a cascading chain reaction circuit bind to the membrane surfaces of particular cancer cell lines using a set of distinct aptamers, each for targeting a specific membrane protein. The selection of distinct aptamers attached one-to-one to each hairpin of the cascade implements a multi-input logical AND to identify a specific cell line matching the corresponding membrane protein profile. Thus, the circuit only generates an output if all membrane proteins, and therefore all hairpins, are present on the cell surface. Output can then be transduced into fluorescent indicators for in vitro assays, or flow downstream to bind and trigger DNA capsule drug delivery vehicles to implement in vivo therapeutics. Second, we have designed peptoid-coated, well-sealed DNA origami capsules for safe and effective drug delivery. DNA capsules loaded with therapeutic drugs can then be programmed to recognize a general signal produced by the cell identification cascade circuit, rather than tailoring its own limited mechanisms per each specific target cell. This effectively increases the fan-in logic and processing power of this DNA-based drug delivery system. 

21 molecular algorithms using reprogrammable DNA self-assembly

In recent work we designed a reprogrammable set of 355 DNA strands, or DNA tiles, capable of implementing a wide variety of algorithms. These tiles undergo algorithmic self-assembly: a form of molecular computation where molecules attach to a growing nanostructure and where each attachment executes a logical instruction, or step, of the computation.

Writing a new DNA algorithm is easy: just choose a subset of our DNA tiles. We implemented a total of 21 6-bit algorithms, including bit-copying, sorting, recognizing palindromes and multiples of 3, random walking, obtaining an unbiased choice from a biased random source, electing a leader, simulating cellular automata, generating deterministic and randomised patterns, and running a period-63 counter. The average per-tile error rate over the 21 different programs was less than 1 in 3000.

A number of technical challenges needed to be overcome to implement our large algorithmic tile set including: out-of-equilibrium growth from an input-providing seed structure, robustness to assembly errors, reconfiguration of the final DNA nanostructure for AFM imaging, and DNA sequence design of 355 DNA strands with stringent energetics requirements. Our reprogrammable architecture enabled programming while at the bench: we could come up with a new algorithm in the morning and implement it on the same day. Ideas from theoretical computer science were used to show what kinds of computations our system is capable of. The talk will showcase how the development of such multipurpose molecular machines, reprogrammable without detailed knowledge of the machine’s physics, could establish a creative space where high-level molecular programmers can flourish.

Dynamic self-assembly of encapsulated DNA nanotubes

Biological cells adapt, replicate, and self-repair in ways that are unmatched by man-made devices. These processes are enabled by the interplay of receptors, gene networks, and self-assembling cytoskeletal scaffolds. Taking inspiration from this architecture, we follow a reductionist approach to build synthetic materials by interconnecting nucleic acid components with the capacity to sense, compute, and self-assemble. Nucleic acids are versatile molecules whose interactions and kinetic behaviors can be rationally designed from their sequence content; further, they are relevant in a number of native and engineered cellular pathways, as well as in biomedical and nanotechnology applications. I will illustrate our work on self-assembling DNA scaffolds that can be programmed to respond to environmental inputs and to canonical molecular signal generators such as pulse generators and oscillators. I will discuss recent work on the encapsulation of these dynamic scaffolds inside emulsion droplets serving as a mimic of cellular compartments. I will stress how mathematical modeling and quantitative characterization can help identify design principles, guide experiments, and explain observed phenomena.

Tracking the first evolutionary steps in non-equilibrium settings

We have driven the first steps of molecular evolution with non-equilibrium settings, often temperature gradients, but also with a recent focus on air-water interfaces. These settings accumulate molecules, select them for length, enable strand separation and allow continuous feeding through bulk solutions. Using sequencing, we see diverse pathways in sequence space that are able to create structure from randomness. These experimental findings make us better understand what are the pitfalls towards open ended evolution. In many cases still, these studies are performed with the help of a protein, but we converge towards RNA only experiment, driven by hybridization, gravity, temperature and hopefully, autocatalysis.

The origin of chemical reactions

What constrains how a chemical reaction emerges, and what reactions were required for the emergence of life? This is important since discussions about the emergence of life often are built around narratives regarding the prebiotic plausibility of the formation of the chemicals required to form a cell. In addition, the goal to make life in the lab is stuck around poor definitions of life, and over engineering to address these definitions. A solution would be to identify a feature exclusively associated with all life and develop a detection system for that feature that could be used for making life in the lab and the search for alien life.

Thinking about molecular self-assembly that thinks

Thinking was once considered a unique privilege of humankind, but modern understanding has evolved to accept the depth and richness of animal cognition, and there are serious arguments about in what sense a single cell might be considered to "think". The next natural question would be to ask about a network of molecules, or even a single molecule – can it think? Is it just a matter of degree, as Darwin said, and not a matter of kind? Leaving that question aside for the philosophers, we can ask clearer questions about information processing, and attempt to characterize information processing in simple molecular systems. Remarkably, the perhaps simplest framework for molecular interactions – passive self-assembly – already exhibits rich information dynamics including the ability to simulate Turing machines, cellular automata, Boolean circuits, and neural-network-like pattern recognition – both theoretically and in experimental demonstrations using DNA nanotechnology.

Discussion of key questions

Guy hosts a discussion session on key questions.

1. What do we need to make DNA/RNA nanotech in living cells more predictable and usable?

2. In what ways should we be looking to further emulate natural molecular systems? In what ways should we be trying to do things differently?

3. What key applications/features can DNA/RNA nanotech in living cells enable?