How are single cells sorted in an scRNA-Seq protocol by FACS?

Question

I am computational guy trying to understand how FACS sorting works in an scRNA-Seq protocol. About sorting single cells in each well of a 384-well plate, I have a question that would be grateful if you could answer it.

The protocol reads that cells are sorted in the wells based on their marker levels. For example it says:

NK single cells were collected from the CD19-/TCR-β- events by gating for NK-1.1+ events in NK-1.1 vs. Gr1.

The part that I don't understand is that how it is possible to know positive/negative events or even to gate on events beforehand? I know that FACS first examines markers and then sorts them into positive, negative or waste collections. Here is my question:

But how do we know what an expression level is acceptable for calling a cell positive, and also in more complicated case, how do we to gate on a subpopulation of cells (like top right quadrant) when no events have yet crossed the laser beam because we don't know the baseline of CD markers' expression levels yet?

For instance, It could be possible that in my panel setup a CD marker can never reach a value more than X, but in your setup it can never reach a value more than X/2.

N.B. As protocol says, subsampling of the main sample for FACS is not possible to infer these values beforehand because it is very likely that the populations of different cell types is not conserved in the subsample.

Answer 1

I will append the previous sentence in your protocol, and put what might help you in bold:

To obtain B cells, NK cells and monocytes, a splenocyte suspension was stained with PE-Cy7-conjugated CD19, eFluor 450-conjugated NK-1.1, PerCP Cy5.5 Gr1, FITC TCR-β, APC CD11b and PE B220 (CD45R). B220+ and B220neg (germinal center) B cells were collected by gating for CD19+(TCR-βneg) cells and then by B220 against theCD19 marker. NK single cells were collected from the CD19neg/TCR-βneg events by gating for NK-1.1 positive events in NK-1.1 vs. Gr1.

The cell types in question have unique properties that can be stained for using (immuno)histochemistry, i.e. chemicals and/or antibodies. The antibodies themselves tend to be conjugated, i.e. they are covalently bound to fluorophores typically which are fluorescent. That means a cell type can be labeled by fluorescence, and the FACS machine can sort fluorescent cells from non-fluorescent cells.

They stained the different cell types with different markers. The markers are sufficiently different from each other to be able to use them together in some cases. For instance, whatever FITC is could be used simultaneously with PE-Cy7-conjugated antibody.

1. First, they collect two subsets of B cells (the B220 positive and negative cells) by using the CD19 marker (PE-Cy7-conjugated antibody that binds CD19).
2. Then, from one of the subsets, the B220 positive one, they sort another subset which is positive for eFluor 450 (which is conjugated with an NK-1.1 antibody). This collected subset fulfills their criteria for being NK cells.
3. This subset is used for single cell capture on a plate (e.g. for SMART-seq2) or using microfluidics (e.g. for 10X Chromium).

For a FACS machine, you can select the threshold level of e.g. florescence that will be used as a cutoff between positive and negative events (cells). This is arbitrary and is usually subject to trials.

Answer 2

we don't know the baseline of CD markers' expression levels yet?

There are standard settings that can be used between runs that will give consistent results from one sample to the next. There are also controls (fluorescent beads) that can be run beforehand to set the machine to ensure that those standard settings will work, and that allow you to make the relatively minor adjustments to bring them into a consistent range.

But you also have another misapprehension:

subsampling of the main sample for FACS is not possible to infer these values beforehand because it is very likely that the populations of different cell types is not conserved in the subsample.

Just because the positive cells are not in the subsample doesn't mean you can't set normal values.

The key is that a positive cell isn't a subtle thing. In a typical flow cytometry stain, your positively stained cells will be much brighter than your negatives; at worst tens of times, often thousands of times brighter. Setting a gate fairly close to the upper end of your negatives (in your subsample) will in practical terms guarantee that your positives will be outside the gate.

I grabbed a random example from a Google search; this is a pretty typical looking gating:

Notice that in each gate, the positive population is (eyeballing) a thousand times, to several thousand times, brighter than the negative (the X axis is a log scale). When you have that much room to work with, delicate subtle effects in baselines settings can be safely ignored.