Setting up scatter
A common problem when you're new to cytometry or even just new to a specific cytometer is figuring out where your cells are! The default settings for forward and side scatter (FSC and SSC) aren't going to be correct for your cells. They're usually set up for the quality control (QC) beads. The newer spectral machines also often have huge dynamic ranges for scatter, which is great, but makes finding your particles of interest difficult sometimes. So, today, let's go over a simple workflow for setting up the instrument to get your cells on the screen reproducibly, and to get them in the same place every time so you can compare across experiments.
We're going to start by setting up the template. We're going to do what's sometimes referred to as "air gating", where we draw the gates without any data present. Rather than moving the gates to the cells, we're going to adjust the instrument settings to position the cells in the gates. That way, we put the cells where we want them, and get them in the same place every time.
Here's my template. I'm going to use human peripheral blood mononuclear cells (PBMCs) for this example, because these have quite characteristic scatter distributions that will be recognisable. I've got a gate for the lymphocytes, one for monocytes, and one for compensation beads (optional, but handy if you're using those). I've done this for both side-scatter channels. On the second side scatter channel (SSC-B), you can see I've scaled the data differently, so I can include a gate for granulocytes. That way, if I wanted to, I could easily see the boundaries of my lymphocyte population (on the first SSC channel), while having the granulocytes also on screen (on SSC-B). If the cells are positioned at the top of the plot, going off screen, they will not be accurately recorded for side scatter (on most instruments), and will simply get the maximum value assigned. Changing the axis scaling later will not allow you to separate out different populations above the maximum recorded value. So, having multiple scatter channels helps with diverse cell populations.
Once we've got a template, we put some cells on and hit acquire. We're just going to acquire (view the cells), not record (save the data).
What do we see here? The default scatter CytekAssaySettings, here on the Cytek Aurora, are not set up for mammalian cells. That big population on the left is debris or small particles present even in the buffer (oftentimes). Let's look at the Gain tab (I am probably going to refer to gains and voltages interchangeably, which is not really correct. Apologies in advance.)
Here we can see that FSC is set to 458. On most of the Aurora machines I've used, FSC gain needs to be near 100 for mouse or human immune cells. Let's try that.
Okay, progress. the small blue population that is appearing now along the bottom, sort of to right of centre is the lymphocytes. How do I know? Experience, but there's also a trick to figuring this out: backgating. If we are looking for a specific cell type, we can bring a sample stained with a marker that identifies that cell type. In this case, if I want find the lymphocytes, to use them as a guide for positioning everything, I could use an anti-CD3 or CD4 single-stained control. We usually want an unstained as well.
We need to know a channel where we expect fluorescence from our marker. I'll use BUV395, because that's a nice clean fluorophore in a region with limited autofluorescence interference, so we can look at UV2.
We want to plot the fluorescence channel where we expect signal against the scatter channel(s) we want to adjust.
We want to look at the single-stained control fluorescence channel as we adjust the gain on the scatter channel, using the positive population as a guide to tell us when we have the gain correct.
Okay, back to the workflow. At a gain of 100, the lymphocytes were still to the right of the "lymphocyte" gate I'd draw. I don't move the gate, I adjust the gain/voltage furhter until the cells are positioned in line with the gate along the FSC axis. We don't need to worry yet about them being below the target gate.
Once we get the FSC position correct, we adjust the SSC gain to bring the cells up in to the gates.
Then do the same for the other SSC channel(s).
Great! Now I have the cells in the positions I wanted. Done, right? Not quite.
If we look at that last FSC/SSC plot, we still have a large population of stuff in the lower left, well outside our gates. We've identified our cells of interest using the backgating trick or from experience, so we know that the stuff in the lower left is smaller than cells, i.e., probably junk (if we're working with cells). While we could just proceed with these settings, that's not a good idea for a number of reasons, at least on an analyser.
That population appears to be about 90% of the events. It will be 90% of the data we collect, so it will make our data analysis 10x slower, at least in the initial steps, and will fill up our computer storage 10x faster than needed.
Cytometers can only process a certain number of events per second. The rates are pretty high nowadays, but if you force it to process all that junk, your data quality may suffer. The faster the events go in, the more the cytometer has to "abort" events, meaning more data is lost.
Importantly, thresholds are best used on analysers, not sorters. On a sorter, you need to know just how much junk you're pouring into the machine to have any idea of whether the droplets being generated by the sort are actually pure. Also, a threshold allows you to basically ignore the debris and some dead cells, but this isn't a substitute for working out the problems with the sample preparation.
Since we're working on an analyser here, we should set a threshold. This allows us to say: "I don't care about anything below this point". The simplest is to set a threshold on FSC, which is typically the parameter that triggers acquisition, so this is the earliest you can discard the unwanted events. There is always a threshold automatically set on FSC, so what we are doing is adjusting the value (position of the threshold). The default on the Aurora is 50000. Let's try 500,000.
That cuts out the big population of debris, and the automated density colour scheme now shows us our data a bit more clearly.
What value should I set for the threshold? We want it to be high enough to cut out most of the junk, while being low enough to never risk losing data we care about. My rule of thumb is to use a value halfway between the lower bound of the (FSC-smallest) population of interest and the origin (0). So, if I look at my small lymphocytes, the lefthand edge is at about 1.2-1.4e6 as I've set it up, so I'll use 600,000 here.
We can add thresholds on other parameters, either separately or in combination. I like to use two thresholds combined, one on FSC and one on SSC. This is primarily because problems in the acquisition tend to manifest as low SSC events, so having an SSC threshold helps automatically exclude those.
What happens if we turn on the SSC, by adding it?
Our debris come back? Why? The default in SpectroFlo software is to use an "Or" combination. We don't want that. We are trying to say, "I want to keep events that are above the FSC threshold AND the SSC threshold." So, we need to use the logical "And" button.
And that's basically it. Now we just save the settings, which we can then pull up the next time we want to run a similar experiment. Should be stable for a few months, at least, provided your protocol doesn't change. See more about settings on the Cytek Aurora here.
If you prefer, here's a short video of the same workflow.
Side note: BD (Waters) now has Detector-Independent Settings on the A7, A8 and S8. This handles the scaling problem not only for scatter, but also for the fluorescence channels. This sort of thing is possible if you have all the information from the QC about detector linearity. If you're using one of those instruments, check out the linked white paper above.
Thanks to Laura Ferrer-Font for alerting to me to Detector-Independent Settings, and to Daniel Vocelle and Idun Dale Rein for discussion of the fluorescence channel assist for figuring out where the cells are.
Comments