Real-time Spatial Sequencing
I have more things to say about spatial sequencing. But today I’ve been playing with a fun idea. What a platform one step beyond current spatial approaches look like?
What if you could sequence the transcriptome of live tissue in real time?
That is obtain a full time course of RNA as it’s transcribed and cleared from the cell. The average lifetime of mRNA in the cell is measured in minutes. So this 2D spatial+high resolution time course information would be far richer the single time points we usually look at.
Much of the material we analyze is also degraded. So there’s at least the potential for improving data quality dramatically from live cells.
The whole idea is probably just a short leap into science fiction territory. But it seems like it might be fun to explore the idea anyway!
How?
In the diagram above I propose the basic approach. Live tissue is prepared and mounted on a surface. This surface contains a porous array. Then RNA needs to be released from the cells (we’ll discuss specifics below). The RNA passes through the porous array and down to a sensor array for sequencing. This sequencing array could be an array of nanopores or it could be some other real time sequencing approach.
What’s important is that it gives a continuous readout.
Problems
We’ve got several problems to address:
How do we release RNA from cells without destroying the cells.
How dense can we make the porous array?
How will sequencing work in this context?
Releasing RNA
There could be multiple approaches here. We could have some kind of mechanical pore forming process, essentially a micro or nano pipette. This could create a channel through which material would diffuse down to the porous array.
You could use electroporation. Basically blast the cells with voltage which temporarily creates pores in the cell though which material could pass. This would have the advantage of being able to induce pores, sample material and then give cells some time to recover.
You could use biological nanopores which would insert into the cell surface and release material at some rate. Again, you may be able to apply a bias voltage here to turn the sampling process “on” or “off”. Or it may be possible to use other chemical means to open and close the pores to release RNA.
Array Density
Ideally you’d want to sample cells at ~1 micron resolution. This would require a 1 micron pore density. But this doesn’t necessarily need to be the case.
The “sampling pores” could be on the order of 1 to 100 nanometers in size. This would ensure that they are sampling from a single cell. But they could be spaced at 100 microns or more.
This would mean you’re still getting 2D spatial information, and it’s still coming from single cells. But you’re not capturing every cell. This could be sufficient for many applications.
Spacing out pores could be an advantage because the fundamental sensing approach may require sensors >100 microns (e.g. currently available nanopore membranes).
Alternatively perhaps each of the “sampling pores” can be constructed in such a way that it labels material. For example perhaps some kind of tailing or other enzyme could be used to label the RNA such that you label material before it reaches the sensor.
Those enzymes could be surface attached or even coupled to a pore through which the RNA translocates near the sampling pore aperture.
Sequencing
Sequencing would require a real time approach. However a PacBio-style or nanopore approach might work here. Ideally sensors would be packed at ~1 micron density. But as described above, they don’t strictly need to be.
A concern with protein nanopore sequencing would be the bilayer stability in the presence of all this junk that’s being thrown out of the cell. This maybe “fine”, but you could incorporate filtering into the porous array above the nanopore.
You would also need to figure out if you can add the necessary adapters or work directly from polyA tailed material.
Potentially this might work better on a PacBio-style chip. As I understand it PacBio currently bind polymerases are part of the prep. Here you’d probably need to have the polymerases in the ZMWs and could potentially sequence the RNA directly.
Summary
It seems like a fun idea, and perhaps might be worth perusing. I suspect you could probably at least build something in a research context that would look cool.
Under the US patent rules I have a 1 year grace period to file a provisional based on the above. So hey, if you think this might be fun give me a shout! (new@sgenomics.org).