This is a good brief overview of a topic that’s becoming more important all the time: analysis on the single-cell level. And as the authors mention, it’s partly a case of wanting to do this, and partly a case of there being no other choice. Larger pooled tissue samples just don’t have the level of detail needed: you average out the numbers for the analytes you can detect, and follow that up with a much greater chance of missing the rare ones (or the values present in rare cell types, which get swamped out). And it’s just those variations that we so often need to focus on.
But single-cell work has (and most certainly still does) push the limits of analytical chemistry. The math is not encouraging. If you have an analytical technique that can detect things down to one femtomole, that’s pretty good, right? But a typical mammalian cell is a one picoliter container, which means that your one-femtomole sensitivity will limit you to millimolar concentrations inside that one cell – in other words, just the major stuff. Then there’s the question of spatial resolution and intracellular compartments. Lysing said cell and blending it up just gives you a smaller-scale version of the averaging problem you were trying to avoid by moving to single-cell analysis in the first place: what if a given species is much more important in the nucleus or mitochondrion than out in the cytosol? What do you do about membrane-bound components? And finally, there are the problems of finding and isolating particular cells of interest. If you’re looking (as you may well be) for rare cell types, like ones that have just flipped a switch to become cancerous, how do you find them? Many of the techniques used to identify cell types involve staining with small-molecule or immunologic reagents that aren’t compatible with the analysis you’re setting up for. Even with the more normal cell varieties, you have to be concerned with where they are in the cell cycle and their metabolic state, while being aware that you might be altering that state just through the isolation process. No, it ain’t easy.
Fortunately, it’s not impossible, either. The problems vary, though, with what you’re trying to measure. Single-cell genetic sequencing is probably the most robust measurement, thanks to advances in sequencing in general, and of course the amplification possible through PCR. Single-cell proteomics and small-molecule metabolomics, etc., are another thing entirely. Every technique available has sensitivity limits and blind spots even while working within those, and it’s important to keep both of those in mind. As the review details, a good number of these techniques are mass-spec-plus-something-else, as you might well imagine, since that’s the technique that combines the overall versatility, sensitivity, and throughput needed to make such experiments work at all. For physical separation of analytes, microscale capillary electrophoresis seems to be the leading technique.
Even now, though, there are some interesting results showing up. Bulk homogenate mass spec measurements, for one, tend to pick up a lot more glutamate and glutathione than you see in single-cell measurements. It could be that these components are sitting in the extracellular matrix and lost during single-cell preparation, or that they’re increased under the stress of that preparation, or that they’re broken out and released from other conjugated species more under some kinds of handling than others. (These aren’t mutually exclusive explanations, either).
One very difficult but very promising area combines imaging with chemical analysis, as in surface-mass-spec scanning techniques. Doing this at the resolution of cellular structures is really pushing up to the edge of what’s even possible at the moment, but things are continuing to improve, and we’re not bumping up against any laws of physics yet (or not quite). There are a lot of different ionization techniques out there, with many more to try, and already you can see real improvements in both the range of ions kicked off the sample, the amount of depth (and depth information) you can get from the beams themselves, and the number of chemical species that can be detected and identified. (That identification is nontrivial, too – as these methods improve, you find many examples of interesting molecular ions being detected that no one is quite sure how to assign yet). This really is the absolute forefront of analytic instrumentation, with combinations of mass spec and advanced imaging and labeling techniques pushing the field forward. The high spatial resolution requirements, unusual ion beam characteristics, and added simultaneous instruments (such as Raman probes or atomic force microscope tips) mean that these setups are bespoke custom rigs at the moment, the wires-hanging-everywhere look that tells you that you’re seeing one-of-a-kind instrumentation.
I’m very happy to see it. The ability to track both endogenous species and outside ones (such as our drug molecules!) inside single cells is going to be crucial to the overall program of understanding what the heck is going on in a real living system. Honestly, I can’t see any other way to do it. These techniques are simultaneously a reductionist approach (down to individual chemical species) and an integrative one (since you’re observing the real living system in all its complexity), and that’s just the kind of blend that offers the most information. Figuring out how to obtain more such data and (even more) figuring out how to interpret all of it will occupy a lot of people for a long time, but to very good effect.