When I was a graduate student, my PI asked me how many μL of a particular antibody I used per million cells, and I told him my staining conditions were “per 100 μL” not “per million cells”. This triggered a heated debate about antibody amount – he argued that the amount of antibody needed was heavily dependent on cell number and I argued that it was more important to add antibody based on the staining volume than the cell number. I honestly don’t remember who “won” that argument that day, but given that he has superior debate skills and I lacked concrete evidence at the time I probably didn’t convince him of his error. Now that I’ve been working in a flow cytometry core facility for a few years, I’ve noticed that questions surrounding antibody concentration and cell number come up regularly, usually in the form of “how many cells should I stain?”. This question seems to be considered most often when researchers have a limited number of cells to study or they’re trying to scale up their staining protocol to sort a larger number of cells. Thinking back about my debate with my graduate school PI, I wondered how well researchers understand how cell staining conditions affect their results in order to properly interpret their results. In this post I’m going to discuss the impact of cell staining conditions on results and how to apply those concepts to determine how many cells are needed for an experiment.
Factors that Impact Cell Staining
Many factors during the cell staining process can change the resulting fluorophore intensity – including incubation time, incubation temperature, cell number, staining volume, and antibody concentration. In an ideal scenario, all factors should be kept exactly the same for all samples and controls. Realistically, this can be a challenge and sometimes these factors vary. So what happens when you vary the staining conditions? Of the five factors I mentioned, I’m going to assume that the incubation time and temperature will remain the same because these are the easiest parameters to keep constant and therefore should always be the same. Antibody concentration is completely dependent on staining volume, so I tend to lump those two together and always stain in 100 μL. Of course with large numbers of cells 100 μL may not be the best choice, but I’ll discuss that later. Knowing that it should be quite easy to keep incubation time, incubation temperature, and staining volume consistent between tubes and experiments, we are left with cell number and antibody concentration. These two factors can be the hardest to keep consistent, and I’d like to go into further detail about how changing these factors impacts the results.
I will tell you now that antibody concentration is the most important factor in the cell staining protocol. If the antibody concentration changes, the difference in staining pattern can be drastic. Because of this the I strongly recommend titrating all antibodies and making a master mix of all antibodies to stain all samples. Titration will determine the optimal amount of antibody to use and a master mix will ensure that all samples get the same amounts of antibodies and reduce pipetting error. For reliable results I don’t recommend pipetting less than 2 μL. If small volumes are required, make a dilution of the antibody first! Figure 1 demonstrates how antibody concentration can affect the resolution of the positive population. In the figure, the best resolution is determined by the highest staining index.
Since antibody concentration is so critical, another tip is to keep track of the amount of antibody in mg/mL as opposed to the volume of antibody per volume of staining buffer. The reason for this has to do with simplifying titrations if the fluorophore is changed for a particular antibody clone. If I know that I’ve titrated a PE antibody to 0.1 μg in 100 μL and I decide to switch the fluorophore to FITC (same antibody clone), then I know that I easily calculate and use 0.1 μg in 100 μL even though the PE antibody has a concentration of 0.2 mg/mL and the FITC antibody has a concentration of 0.5 mg/mL. If I keep track of my antibody titrations as “1:100” or “2 μL”, then it is more work to switch fluorophores on antibodies. To be clear, the best practice is always to titrate every single antibody, but using the same mg/mL (or μg per 100 μL) often works and the approach of titrating each antibody clone regardless of fluorophore is better than not titrating at all.
Some people may be surprised to find out that cell number can have little effect on the data. In the example below I stained between 1-10 million cells with the same concentration of anti-CD4 antibody (Figure 2). Except for cell number all other staining conditions were kept the same. Even with a tenfold increase of cells there is no significant impact on the frequency or MFI of the positive population. Note that this may not be true for every single antibody because some antibodies are more sensitive to cell number than others, but in general cell number is not as critical as other staining conditions. In practice this means that a single antibody concentration can work for a range of cell concentrations.
How to apply this information
Now that we have good understanding of how cell number and antibody concentration impact results, we can address the question of how many cells should be stained for an experiment. Except that we still need some more information because the answer depends on the specifics of each individual experiment. When researchers ask me how many cells are needed for their experiment I typically need to know two things: how many cells can the researcher obtain from each sample and how rare is the smallest population of interest. When cell number is limited, researchers need to think more carefully about how many cells to use for controls and how changing the cell number impacts the staining of each marker in the panel. To address the question of “how many cells should be stained” I’ll break it down into three separate different questions:
- How many cells should be stained for the actual sample?
- How many cells should be stained for the controls?
- How many cells should be stained when titrating an antibody?
The first question is quite simple. By default, we suggest staining 1×106 cells. However, the number of total cells is best determined by the frequency of the rarest population of interest. Let’s say we want to examine a blood sample that contains 50% neutrophils, 10% monocytes, 30% lymphocytes, and 0.5% regulatory T cells (Tregs). If the focus of the experiment is on the monocytes as the rarest population of interest, we don’t need a large number of cells. Knowing that the monocytes are about 10% of the total cells, we can calculate that collecting 50,000 events will give us about 5,000 monocytes. If the sample is not limited, I recommend staining 0.5×106-1×106 cells. However, if the sample is limited, I can calculate that I would be able to stain 1×105 cells and still be able to analyze an appropriate number of cells. Keep in mind that cells are lost in the staining process and pushing sample through the cytometer, so don’t stain 1×105 cells and expect to be able to analyze exactly 1×105 cells!
Now let’s change the focus of the experiment – let’s say the goal is to examine the Tregs. Based on the knowledge that the Tregs are only 0.5% of the total cells, only 250 Tregs could be analyzed in a total of 50,000 cells. It is recommended to have at least 100-500 events in the rarest population, depending on how distinct the markers are on the population. In this scenario, it is possible that 50,000 total cells could be used to analyze Tregs, but more cells would certainly be a good idea.
The second and third questions regarding controls and titrations are a bit more difficult to answer and boils down to how many cells are available. If the sample amount is not limited, then the same number of cells should be used to stain the controls and the sample. If the sample amount is limited, then may be possible to stain fewer cells for the controls to maximize the number of cells in the fully stained sample. The abundance of populations in the sample should also be taken into consideration for fluorescence minus one (FMO) controls. If the FMO is needed for a rare population of cells, then the same number of cells should be stained and analyzed for the sample and FMO control. If the marker is higher up on the gating tree and therefore used for a larger percentage of cells, it may be possible to get by staining fewer cells – assuming that the same staining intensity is achieved with a lower concentration of cells. If it is difficult to obtain enough cells for both the sample and the controls, one solution is to determine if compensation beads can be used instead of single stained cells. However, compensation beads are not always as good as cells and cells will always be needed for FMO controls.
If it is determined that fewer cells can be used for controls compared to the sample, the staining conditions should be considered. Often times one concentration of antibody could be used for a range of cell concentrations, so the same antibody amount and staining volume may be acceptable. However, if there is a large difference in cell number in the sample and the control or if an antibody is particularly sensitive to cell number, then better results can be achieved if the antibody concentration and cell concentration are the same. For example, if the control is stained with 0.25×106cells and 1 μL of antibody in 100 μL we can keep the antibody and cell concentrations the same by staining 5×106cells and 20 μL of antibody in 2 mL. However 20 μL is a lot of antibody, so it certainly cheaper to consider staining 5×106cells in 10 μL of antibody in 1 mL or even 5 μL of antibody in 0.5 mL. The important thing is to keep the antibody concentration exactly the same and then keep the cell concentration similar between the controls and samples, knowing that a range of cell concentrations can work for a single antibody concentration. When scaling up a staining protocol, the best practice would be to stain two tubes—the normal and scaled up number of cells—in the same experiment and determine if the staining patterns are comparable.
Another common scenario where cell number becomes a critical discussion point is transitioning from a benchtop analyzer to a sorting experiment. For example, after analyzing 1×106 cells on a benchtop analyzer and finding a really interesting population, the next step may be to sort that population. But in order to get enough cells for the experiment, let’s say that 10×106 cells are needed based on the known frequency of the interesting population and the number of cells needed for the post-sort experiment. The staining protocol be scaled up for the sorting experiment similar to what I described above. It is unlikely that ten times as much antibody needed to stain ten times as many cells! I typically make an assumption that titrated antibodies work for up to 5×106 cells, so as an example I would recommend scaling up the original conditions of 0.1 mg of antibody and 1×106 cells in 100 mL to 0.2 mg of antibody and 10×106 cells in 200 mL. However, this is a generalization and scaling up a staining protocol in this way probably won’t work for every antibody and every cell type. The best way to figure it out is to test it and compare to the original protocol!
The key to any good science experiment is consistency – use the same reagents, protocol, and instrument and real results will be reproducible. But there are times when scientists aren’t able to be as consistent as they want to be. As I discussed, there may not be enough cells for all of the controls or cost of reagents may become a limiting factor. In these scenarios, a good flow cytometrist will understand their staining protocol and how the cell concentration and staining conditions may be altered without compromising the results. And a good graduate student will prove their PI wrong with solid data. (I may be a few years late and he might not read this, but I think an “I told you so” is in order!)