Let’s head off to the outer limits of current imaging technology. Two recent papers do exactly that, and I’m going to propose combining them for even more instrumental craziness. The first covers a class of cellular structure that I had no idea even existed, even though it’s been studied for many years: bacterial gas vesicles. There are species of aquatic bacteria (and Archaea) that produce hollow protein structures (some 200 nM long) whose walls are so hydrophobic that they exclude water and are too tight to let larger hydrophobic compounds in. What does that leave them to be filled with? Gases.
The bacteria produce these to regulate their density and maintain buoyancy; they’re basically nanoscale swim bladders. But as this review shows, there are people taking advantage of these structures for imaging. The contrast between the gas phase and the cytosol makes these structures available for ultrasound detection, and I have to salute the first person to whom that unusual thought occurred. Micron-scale bubbles (stabilized by lipid or protein) are used as an ultrasound contrast agent, so it makes sense, and in 2014 a Berkeley/Toronto team showed that it works quite well. In fact, different sorts of vesicles from different species can be distinguished by their ultrasound behavior.
If you could get cells to produce such things for you on demand, you’d have a shot at an ultrasound reporter gene construct (of all things). It’s not easy. The gas vesicles actually need several genes to produce the mix of proteins in their walls, and getting everything to express and assemble properly seems to have taken some experimentation, but a report earlier this year shows that conventional disease bacteria can be made to express them and they can be detected in turn to localize the organisms in vivo.
But these vesicles can also be exploited for another detection method: NMR. Air-filled regions have high contrast in T2 magnetic resonance imaging, and with sufficient resolution, you could pick up these gas vesicles as well. The review paper mentions another possibility – allowing these to fill with hyperpolarized xenon – that has also been realized by the same group (now at CalTech). This gives huge gains in sensitivity, as hyperpolarized species do (normally only a tiny fraction of the potential atoms can be observed in a regular NMR experiment; this makes everything signal).
Mentioning NMR, though, brings me to another recent paper. I wrote a while back about an exotic NMR possibility, using spin-defect diamond crystals as detectors, that could possibly lead to NMR detection in very small volumes (such as a single cell or even smaller). Here’s a paper from a Harvard/MIT collaboration that’s taking this much closer to reality. The problem with using a single spin defect center is that while you can detect NMR signals in a very small volume, the resolution is poor (100 Hz or so under the best conditions). NMR-using chemists will recognize that as very fuzzy indeed – you can get vastly better spectral resolution on a 1970s electromagnet NMR, although you’re not going to be picking up any single-cell-volume signals with it, to be fair.
This new paper uses several of these unpaired-electron defects in an array, along with a much fancier overall method (coherently averaged synchronized readout, CASR, which I do not understand this morning well enough to describe). What is immediately apparent, though, is that they’re getting real NMR spectra of small test molecules (ethyl formate, xylene, glycerol, trimethyl phosphate) in picoliter-sized volumes of sample. There are many more refinements that are possible to the technique, and the authors seems hopeful (well, as authors generally are, admittedly) that the method will indeed be able to realize single-cell NMR experiments, multiplexed experiments across parallel detectors for bulk samples, and other completely new fields of research.
You’d think that these two techniques would work very well together – the extreme spatial resolution of the diamond-defect detectors with the gas vesicles inside single cells. (I haven’t even mentioned another NMR possibility that’s explored in that review article I started with, the idea of using aquaporins as NMR reporters). If we really do get down to that kind of resolution, all sorts of other exotic ideas will surely swim into view. This gets to Freeman Dyson’s point about “tool-driven revolutions” rather than concept-driven ones. Build a new instrument, that does new things and gives you new ways of looking at the world, and new concepts will follow. Let’s see if it works here!