BioSMART: BIOreactor Spatial Mapping and Actuation in Real Time

Many drugs and chemicals we have today are made not in chemical plants, but in bioreactors: vessels containing microbes or other cells which can create the chemical we want. We do this because using living cells to create complex chemicals can be much cheaper in terms of energy, raw materials and the cost of making the plant itself (which does not need to operate at high temperature and pressure as some chemical plants do). Nevertheless, bioreactors are harder to use than their chemical plant cousins, because living cells are sensitive to their environment in a range of complex and difficult-to-model ways. The only sensible way to make sure that the bioreactor is working at maximum capacity is to watch what is going on during the reaction.

In this project we want to, for the first time, monitor the conditions of cells throughout the bioreactor by using the cells themselves to tell us what is what is going on. We can do this by genetically modifying the cells to change their physical properties based on their local environment. We call these genetically-modified cells biosensors, and in our case they report their condition by fluorescence: making a protein which glows when you shine light on it.

While studies have demonstrated the development of biosensors and their benefits in bioprocessing, so far the implementation of biosensors in industrial processes has been hampered by a lack of infrastructure for their use. This is because most analytical techniques to date have not been living, but rather based on chemical or physical development of signals. In order to really capitalise on the enhanced sensitivity and specificity of biosensors, development of hardware and data analysis tools for integrating them into industrial bioreactors is needed. This proposal seeks to fill this gap.

Monitoring the fluorescent glow is a challenge; the bioreactor itself isn't nice and transparent, but murky and turbid. We can't just look through it, because light from the outside will scatter (or bounce) multiple times before it gets out again. Fortunately, there is a technique called Fluorescence Diffuse Optical Tomography (fDOT) which can account for scattered light. It cannot resolve as small features as a microscope, but in a bioreactor this isn't important, as centimeter-scale resolution is enough. We will build a system that can monitor the whole bioreactor using fDOT, by shining a laser at different points on the reactor surface and watching the resulting glow from the cells on all sides; by taking measurements from lots of different locations and using a suitable computer algorithm, we can get a 3D model of how the glowing cells are distributed. With this information, we can then use modelling to predict the cell behaviour and to automatically control the bioreactor conditions to improve production.

As a demonstration, we will focus on monitoring the buildup of lactic acid which is a byproduct of anaerobic (oxygen-poor) reaction conditions; excess lactic acid is toxic, and can limit the performance of (or even kill) the cells that produce it. By engineering the cells to glow based on how much lactic acid there is nearby, we can monitor the reaction and either increase the amount of oxygen added to the reaction, stir the tank, or even redesign the reactor itself to avoid local differences in the reaction conditions. The case of lactic acid is just an example; future cells might report changes in temperature, shear stress, oxygenation or any other parameter, with a different-coloured glow for each.

Overall, this project represents the first step towards a new frontier where the cells in a bioreactor not only produce the chemicals we want, but tell us what is going wrong in the reaction and how to fix it. The result is a reactor that can make complex chemicals much more cheaply, and given how much of our modern world relies on these chemicals, that can have subtle but pronounced benefits throughout the global economy.