oxford_logo Dunn_School

Research Vignettes on DNA Replication, DNA Computing, and Cancer Immunotherapy


DNA Replication (with Conrad Nieduszynski)
Life as we know it exists because of DNA's ability to replicate. While evolutionary pressure has made the probability of a replication error low, errors in replication can have catastrophic consequences: cancer, genetic disorders, and many other diseases have their roots in gene mutation. Studying these rare error events is important. However, current methods of studying DNA replication rely on ensemble averages, where rare but important events can be masked by the ensemble's average behaviour. We are using "big data" methods and new sequencing techniques to study these events at the single molecule level. This project is just getting off the ground - more to come soon!



DNA Computing (with Andrew J. Turberfield and Luca Cardelli)
Computing relies on a mechanism that can hold the value 0 (False) or 1 (True). In traditional silicon-based computing, these mechanisms are transistors. In recent decades, several groups have found that DNA molecules can play the role of transistors to produce biocompatible nanocomputing devices. Instead of electrical signals, one way in which DNA molecules can switch is via toehold-mediated strand displacement.

toehold mediated strand displacement
Toehold-mediated strand displacement. An invader strand replaces an incumbent strand in a DNA duplex by binding to a short toehold.

A DNA walker can step down a track of anchorages via the repeated action of a nicking enzyme followed by toehold-mediated strand displacement. In a system where multiple tracks intersect one another, a mechanism can be designed where a DNA walker can block the track for another walker where the two tracks intersect. These blocking operations make it straightforward to implement a NOR gate, which is functionally complete. Therefore, these "DNA walker circuits" can evaluate any formula posed in propositional logic.

burnt bridges walker
Burnt bridges DNA walker. A single-stranded DNA walker steps down a track of anchorages by using the action of a nicking enzyme followed by toehold-mediated strand displacement. (a) A DNA walker completes the restriction site for a nicking enzyme when it is hybridised to an anchorage. (b) The nicking enzyme cuts the anchorage, exposing the toehold on the walker. (c) The toehold on the walker hybridises to the next anchorage in the track. (d) The walker steps onto the next anchorage in the track via toehold-mediated strand displacement.

NOR is functionally complete, so any other gate can be constructed from the composition of NOR gates. However, simpler gate designs are often available that use fewer tracks. This suggests that there may be rules for simplifying DNA walker circuits. To determine the rules for simplification, we are creating a logic system for DNA walker circuits that can be proven both sound and complete.

logic gates
Logic gates that operate via the interaction of DNA walkers. Each track is represented as an arrow, and track intersections are represented by red crosses. Each walker is assumed to only step on its own track. Walkers labelled by x and y will only step if their corresponding propositions are true, while walkers labelled by 1 will always step. The gate evaluates to True if a walker makes it to END without being blocked. Otherwise, it evaluates to False.

DNA walkers can also be designed, built, and measured experimentally. When DNA origami is used to template the construction of the track, the systems can be built via self-assembly. The movement of a DNA walker can be measured by modifying the walker with a dark quencher and tethering fluorophores to the start and end of a track. The quencher on the walker reduces the emission from the fluorophore when the walker is nearby, and the fluorophores' signals can be detected with fluorescent spectrophotometry.



Cancer Immunotherapy (with Helen Byrne)
A clever new method uses an immune response to perform the targeted delivery of cancer therapy. When a tumour is subjected to radiotherapy, the tumour cells release a chemoattractant that draws macrophages into the tumour. These macrophages can be genetically modified to release an oncolytic virus when they encounter hypoxic conditions within the tumour, which in turn infects and kills the malignant cells. While this treatment is effective in mouse models, it also prompts a number of questions: how should this treatment be dosed, and what concentrations of modified macrophages are necessary to send a tumour into remission?

burnt bridges walker
The fraction of a tumour that consists of macrophages, uninfected tumour cells, oncolytic virus, and tumour cells that have been infected by the virus. The in vitro tumour spheroid is assumed to have radial symmetry.

We are answering these questions with a mathematical model that aims to understand the underlying physics of the system. This method allows us to make predictions about the movement of macrophages and oncolytic viruses within the tumour, which suggests how the tumour may respond to the therapy. The flexibility of a mathematical approach allows us to try different dosing schemes that would be otherwise difficult and costly to investigate experimentally. For example, the figure below shows the effect of a second dose of radiotherapy at different time points.

burnt bridges walker
Effect of radiotherapy timing on tumour radius when administered in combination with immunotherapy. Radiotherapy (RT) is administered at 7 days, followed by macrophages loaded with oncolytic virus (φ) introduced at 9 days. After the macrophages are introduced, a second dose of radiotherapy is administered after various time periods.

Modelling such complex biological systems is difficult, as there are a myriad of factors that can affect a tumour's growth. However, it is a mountain worth climbing. Mathematical modelling can suggest questions that guide experiments, reducing both cost and the need for animal experimentation. It also provides a mechanistic explanation for the processes that happen between data points (which are often far apart in in vivo cancer experiments). Finally, it distils out the processes and mechanics that are most important in a complex system, identifying what needs to be "peppered in" to the model to reproduce the behaviour we see in experiments or in the clinic.


CC-BY-3.0