I wouldn’t be surprised if the subject of titrating antibodies is one of the least favorite topics among users. We all know someone who skips titrations and just uses the most convenient antibody amount. But as panels get larger, antibody titration becomes much more important. Recently I’ve learned that titrating viability dyes on the Aurora spectral cytometer is a bit different than typical antibody titrations.
I first became concerned with using Zombie NIR (Cytek’s recommended viability dye) when I started to play around with the aurora immediately after its installation in our facility.I accidentally stumbled across some alarming data: I found a population of CD3-TCR+ cells (Figure 1). But there’s no way CD3-TCR+ cells could exist, right? Unless I somehow discovered a new, undocumented population of cells … After some more tests, I figured out that the CD3-TCR+ population frequency varied depending on where I put my live cell gate and didn’t show up when I used Zombie aqua instead of Zombie NIR. Since my TCR antibody was conjugated to AF700, I decided that Zombie NIR caused some false positives in AF700 and I should just avoid using the two together in all future panels.
I did discuss this with Cytek, and they did give me a solution related to my concern with Zombie NIR: it’s critical to titrate Zombie NIR. If you just want the short answer to my problem, Cytek is 100% correct. However, I decided to keep digging for answers to know precisely why I found issues with Zombie NIR and AF700.
The first thing to note is how fixable viability dyes work. These viability dyes, like Zombie NIR, react with primary amine groups on proteins. That means that all cells will be stained with the viability dye because all cells express surface proteins. These dyes work as for viability because dead cells have permeable membranes, making the surface and intracellular proteins accessible to the viability dye and the fluorescence intensity much, much higher than live cells. Based on this information we can expect that if the manufacturer’s protocol is used (1 uL of Zombie NIR in 100 uL PBS), the Zombie NIR signature will be detected on all of the populations of cells (Figure 2).Now what does this have to do with my impossible CD3-TCR+ cells? At first, I thought it had to do with unmixing – when spectral unmixing is calculated, the negative population must be free of any fluorophore signature to avoid unmixing errors. In my case, some false positives in AF700. But this should be an easy fix if a universal negative is used, right? Unfortunately not. It turns out that Zombie NIR is much more sensitive to titration than the other Zombie dyes. To determine this, I tested four Zombie dyes (violet, aqua, yellow, and NIR) at four different concentrations (1, 0.3, 0.1, and 0.03 uL in 100 uL of PBS). Then for each Zombie dye I played around with unmixing the four files with the different concentrations of dyes as the reference control. Zombie violet, aqua, and yellow behaved as expected – if the highest concentration is used as the reference control, all four files are unmixed without errors (Figure 3). However, if one of the lower concentrations is used as the reference control, any file containing a higher concentration now contains unmixing errors because the reference control was not as bright or brighter than the “sample”. The results for Zombie NIR, on the other hand, were not quite as expected. When the reference control for Zombie NIR was a higher concentration, all “samples” spread into AF700 and APC-Cy7 and lower concentrations had unmixing errors (Figure 4-5). But if Zombie NIR was titrated properly, the unmixing worked well. So am I getting my impossible population of CD3-TCR+ cells by unmixing with a reference control that contains negative cells that aren’t truly negative? No. The data in Figures 3-5 were unmixed with a universal negative and the data looks exactly the same when I don’t use a universal negative.
Then why did I find an impossible population of CD3-TCR+ cells? The answer has to do with spreading error. We don’t tend to worry about spreading error with viability dyes because we gate away the positive cells and spreading error is only an issue with positive cells. Except that I just showed you in Figure 2 that if the viability dye isn’t titrated properly, the live cells will be stained. And that’s when you have to worry about a dye that stains all of your cells spreading into other fluorophores!! It makes sense that I only saw this with Zombie NIR and not the other three that peak on the violet laser because we know from the spread matrix that fluorophores with higher peak emission wavelengths tend to cause more spreading error. So after a year of trying to figure this out, I’m back to my initial conclusions: Zombie NIR spreads into AF700 (and APC-Cy7) and proper Zombie NIR titration is critical. But at least I understand it a lot better!
What is the best way to decide when Zombie NIR is titrated properly? Unlike titrating antibodies where the goal is to determine the lowest concentration to get a maximal staining index, the fixable viability dyes should be titrated down until the live cells are not showing the signature for the dye (Figure 6). Don’t forget that a mixture of both live and dead cells should be stained for a titration experiment! For my experiments I left a mouse spleen in the fridge overnight to obtain about 30% dead cells, but you should titrate the dye on your cells of interest. After the viability dye has been titrated, the reference control can be created by staining only dead cells and then spiking in unstained cells.
Nice work
Thanks for sharing!
Very nice work! Thank you!
Thank you so much. Been having a terrible time with Zombie NIR.
Will titrate further based on this info.
Do you include a viability dye in your antibody titrations? If viability needs to be titrated but also included in antibody titrations, then how should we approach that issue? Great blog, thanks for sharing.
Hi Kelly! Thank you much! I do not add the viability dye in titrations. In most cases, gating out the dead cells based on the scatter graph (FSCxSSC) will keep most of the dead cells out and the rest won’t impact the titration data much.
Nice work sir. I like the detective work you have done. I wish I found this earlier, would have saved me the time… Because I did the same thing lol