The steps to a full proof of concept can be split into two phases: the first is to demonstrate that magnetogenetics can be used to control microbial gene expression, while the second is to combine different magnetic signals, biological pathways, and magnetic nanoparticles, to enable fine spatiotemporal control.
Phase 1: Microbial Magnetogenetics
A: Engineering of pathways that respond to signals which could be imposed by a magnetic field
Success:
In S. cerevisiae, we were able to induce beta-galactosidase expression downstream of the cell wall integrity pathway. This was shown to be possible by subjecting the cells to either heat stress (incubation at 37℃) or mechanical stress (vortexing). For more details, please see our design and results pages.
Other progress:
- We designed and assembled a system of plasmids that should switch on fluorescent protein expression in E. coli under heat shock, controlled by either the TlpA or TcI transcription factors.
- We also designed a fusion protein of EPG (electromagnetic perceptive gene), which directly responds to magnetic fields through a conformational change, with NanoLuc, a high-intensity luciferase. For more details on the design, please refer to our design page here
- Unfortunately, colony PCRs and sequencing of these constructs revealed assembly errors. Therefore, while we developed protocols for testing the activity of these constructs in solid and liquid media, we were unable to demonstrate successful activation.
B: Construction of hardware capable of generating the magnetic fields to activate these signals
Success - Heating:
After 7 iterations of high frequency coil design, we were able to demonstrate effective heating of our MNP suspensions and to characterise the heating efficacy of our coil designs. From this we were able to extrapolate suitable coil designs and MNP concentrations for spatial targeted heating. For more information on how this was achieved, see our hardware page here.
Importantly, we were also able to show that cells are able to grow on agar plates that contained these MNPs. For more information see our results page.
Success - Mechanical:
Relatively large (250 nm) magnetic nanoparticles were visualised under a light microscope, and seen to display Brownian motion. When placed in a magnetic field, the net movement became directional and each nanoparticle appeared to undergo a biased random walk.
We were also able to visualise the attachment of these nanoparticles to yeast cells. Since these MNPs are coated with Ni+-NTA and the cells used contain a His-tagged mechanosensor, we believe that the Ni-His interaction is what allows attachment to cells.
C: Magnetic activation of pathways
Success - Wsc1:
These Ni-NTA magnetic nanoparticles were added to colonies of yeast cells expressing the beta-galactosidase reporter system, as well as the elongated His-tagged Wsc1 sensor. These plates were then placed on a low frequency, high strength electromagnet for 1 hour. 3 shows the result 3 days after activation, which allowed for maximum maturation of the system. Qualitatively, the plate that was subjected to magnetic activation can be seen to have a much stronger blue pigment than the negative control plate. For more details on this experiment, refer to our results page.
Phase 2: Spatial Targeting
While a full spatial targeting system was outside the scope of what we could achieve with two and a half months in the lab, we did build hardware, design biological constructs, and model interactions to work towards this goal. These efforts, along with how we imagine they could be combined for a proof of concept, are detailed here. We hope that future teams wishing to build on our work will find this useful.
D: Magnetic activation of pathway in narrow region of space
Test-Wsc1
A grid of colonies was plated, and only left two rows were placed inside a coil. As seen in Figure 4, we saw activation of the whole plate, rather than the narrow region we were hoping for. This could mean that we require a lower field strength than the 20mT we used, or that we need to modify the design to allow a sharper peak. Alternatively, adjustments could be made on the biological side of things to increase the threshold at which the pathway is activated, as well as the sharpness of the switch. One method to make the switch shaper would be to introduce a positive feedback loop, such that the Rlm1-LexA transcription factor also activates its own expression.
Model-Thermal
Thermal diffusion equations were used to combine the known magnetic field strength of our coils and calculated specific absorption of our ferrofluid to model spatial distribution of temperature over time.
This gave us values for temperature gradient (celsius per second) at different points across a petri dish. Incorporating this into our models of yeast temperature-sensitive fluorescent protein expression allowed us to see how long it would take different regions of the plate to express mature fluorescent proteins. We faced a similar issue to what we saw in our Wsc1 test, with the change in signal not being particularly large over the range of conditions imposed by the magnetic field. This could be fixed on the hardware side by optimising field parameters, or on the biological side by creating a feedback loop that reinforces the signal.
E: Integration of two pathways (activated by distinct signals) into an AND gate
Design - E. coli
For E. coli, we designed a system that relies on a temperature-sensitive transcription factor for the expression of GFP fused to a quencher. Since the fusion point also contains a TEV cut site, fluorescence can be observed if there is active TEV protease in the cell. This is controlled by a separate magnetic signal, with the N- and C-terminal TEV domains each fused to a different terminus of EPG. When EPG folds under a magnetic field, the TEV termini are brought together. This allows the separation of GFP from the quencher, resulting in fluorescence.
Design - S. cerevisiae
In yeast, we designed an AND gate based on the split-GAL4 system. GAL4 is a transcription factor composed of two domains - the DNA binding domain that binds the promoter, and the activation domain that recruits transcriptional machinery. If each domain is put under the control of a different promoter, GAL4-controlled transcription will only occur when both promoters are active.
F: Development of hardware system able to generate two different magnetic fields, at controllable positions
Our final hardware system, MagentaBOX, has motorised stages for two magnets as seen below. For more details on the design, please refer to our hardware page.
G: Tight region of activation of AND gate via magnetic fields
If D is successful, E and F should allow you to activate the AND gate at an intersection between two axes. It would also be important to confirm the two signals are orthogonal, e.g., by performing transcriptomic experiments that quantify downstream effects of each signal on both pathways.