I suspect today’s post will be widely useful. I considered writing it up as a proper publication, but I can’t be certain how generalizable the findings are without running this on many, many panels, so I guess it’s probably better to crowdsource it and get feedback. Anyway, here goes: yes, you can store master mixes of antibodies, even for extremely large panels. Please use this carefully and put in the legwork of doing the validation yourself with your panel(s).
This can work with Brilliant dyes without creating artefacts due to interaction, but there are certain caveats, and some extra testing is required. See below.
First, I'm going to point out that this has been done to a limited extent previously (to the best of my knowledge). Thomas Liechti working with Mario Roederer has validated storage of master mixes for 48-72hrs in this publication (see supplementary figure S2), although it would have been nice to see the dot plots for the BV interactions. Bio-Rad has shown that their StarBright dyes can be pre-mixed for at least a year with no issues.
Here’s how:
Titrate down your antibodies as detailed here, here and here.
Prepare separate mixes for surface or intracellular staining.
If you are using overnight staining, move as many of the Brilliant dyes to the overnight mix as possible. This helps because it reduces the concentration of antibody required (see post here).
Use Brilliant Stain Buffer at 20-30% of the total volume. In my hands, Brilliant Stain Buffer works slightly better than Brilliant Stain Buffer Plus, but you can use either one (adjust the Plus buffer down ~4x).
Check for any bad actors with the Brilliant conjugates. As mentioned here, certain Brilliant conjugates cause interactions much, much more than others like the Ly-6G in my example), and these often can’t be blocked with Brilliant Stain Buffer at any concentration. If you see this, you’ll generally see it in short staining times as well as longer ones or in stored mixes. Swap these out for others (sometimes even the same dye but conjugated to a different clone or from a different supplier will fix the issue). This isn’t terribly common (I had three such problems in the set of all the Brilliant dyes in my four large panels this year).
Use the tandem stabilizer. This prevents breakdown of the tandem dyes over time.
Prepare the mix in a dark tube (e.g., black Eppendorf or foil-wrapped Falcon tube).
Don’t add fixable viability dyes. There are two reasons for this: one, they’ll react with stuff in the mix, creating background, and two, they’ll lose reactivity for dead cell staining. Add these at the point of staining to the portion of the mix you plan to use on the day.
Store cold (4-6C).
You can add extra sodium azide if you want, but if you’re using a lot of antibodies, the concentration will already be adequate to prevent microbial growth. I suggest using clean buffers and you may wish to prepare them sterile for best results.
Don’t add hybridoma supernatant. This will contain enzymes and other stuff that can change your fluorophores.
Note that testing for stability of your master mixes is going to require that you've stabilized everything else (instrument settings & QC, staining protocol, cell preparation, controls, unmixing) in order to leave the master mix as the biggest remaining source of error. We'll look at some of that later, but the Liechti paper covers a lot of the ground there.
How long are these mixes good for? I’ve tested out to a month with no detectable problems (with big panels, though, it's always possible to miss something somewhere). In comparisons with freshly prepared mixes, samples stained with the stabilized master mix exhibited less variation, which is what we’d expect if I’m getting slight changes in antibody concentration due to pipetting error. I suspect that once the mix has sat overnight, anything that’s going to change has already changed unless you get microbial contamination or light exposure. Given that I’m already leaving my mixes overnight for the staining, I’m probably already weeding out potentially problematic conjugates during the testing and titration process.
Here's a ~50-colour panel on human PBMCs. I've stained a single donor's frozen cells with a freshly prepared mix versus a mix that sat in the fridge for a month (this is the longest time point for this set).
Dimensionality reductions provide a nice way of visualizing all the data, and they're generally good for picking up variation in staining. For instance, batch effects usually cause segregation (at least within islands/blobs). Looking at the individual markers can be more convincing, though. Here's a couple of pairs where we might expect problems.
A quick note on tandems here: Using CCR2 with a labile tandem like PE-Cy7 is not good panel design. You want to avoid putting labile tandems on highly expressed markers (the broken signal is proportional to original signal), and you definitely want to avoid using labile tandems on markers expressed on both myeloid and lymphoid cells. Myeloid cells catalyze tandem breakdown, so you get more on them than on the lymphocytes, meaning you can't properly unmix the marker for all cells in the data.
Here's a plot of all the markers, tested with a freshly prepared mix (fresh) versus a stored master mix (MM7) every week for a month. The labels are small, so FYI the rows alternate fresh/MM going down. I've highlighted two markers that were variable and had to be swapped out.
Now, as said above, I’m not the first to try this. I’m aware that scientists at Bio-Rad have done this with the StarBright dyes and, I believe, some Alexa Fluors out to multiple months. I suspect that several of you readers will have done this as well. I’ve also tried this previously, about seven years ago, and had disastrous results due to Brilliant interactions, tandem breakdown and other factors. I think the key thing here is that I’m now pretty confident that this can work for those problematic dye families. The main caveat is that I’ve only tried this for a few of my panels, and I know it won't work for all panels. In fact, it probably won't work for most people's panels using SIRIGEN polymers unless you're using them at very low concentrations, as I do. Tips on that below.
Base buffers I've used:
PBS + 2.5% FCS + 2mM EDTA (note: 0.5% BSA might actually be better for the mixes, but I have not tested this.)
Types of fluorophores that are okay in a stabilized master mix (as far as I can tell):
Star Bright dyes
Alexa Fluors
Conventional dyes: APC, PE
Conventional tandems: PerCP-Fire 806, PE-Dazzle 594, PE-eFluor610, PE-Fire 640, PE-Cy5, PE-Fire 700, PE-Fire 744, PE-Cy7, PE-Fire 810 (yes!), APC-Fire 750, APC-Fire 810
Brilliant UV
Brilliant Violet
Brilliant Blue (BB515, BB660, BB700)
Spark dyes
Real Blue dyes
KIRAVIA Blue 520
Small (?) molecules: Pacific Blue, eFluor660, R718
Stuff that doesn't work:
Fixable viability dyes. They react and become non-functional.
NovaFluors. Sorry, this won’t work for OMIP-102. That said, you could probably modify OMIP-102 to remove NovaFluors, swapping the CD4-NB580 to CD4 StarBright Blue 580 or Spark Blue 574 and moving the CD8-NB555 to KIRAVIA Blue 520, StarBright Blue 615 or StarBright UV795 (for the S8/A8).
Stuff I haven’t tested:
Adding mouse serum, rat serum, human serum or any other blocking reagents. This is on my "further testing required" list. I expect that purified IgG and Fc block will be fine. I've done some testing of this and saw no differences, but would strongly recommend testing this in your hands with your specific reagents (or leave these items out of the master mix).
Qdots (I'm guessing these will be fine, but you'll want to keep the concentration low to reduce aggregation).
PerCP, PerCP-eFluor710, PerCP-Fire 780, APC-eFluor780, APC-H7 (these should be fine).
PE-Cy5.5, PE-AF610, PE-AF700, APC-Cy5.5, APC-Cy7, PerCP-Cy5.5 (again, probably fine, but maybe re-evaluate your life choices if using these!)
Real Yellow dyes. These are newer, so I've had less time to test them for longitudinal stability. I will be testing RY610, RY775 and RY703 (I don't use RY586).
Tetramers. These may work in the surface mix, but I would be wary about including these due to the potential for the peptides to wash out of the MHC pocket in the presence of detergent (mild detergents in the Brilliant Stain Buffer and stronger stuff in the perm buffers).
Streptavidin: Again, it could work, but there’s biotin in serum, so I would expect you’d need a higher concentration to counteract that. Alternatively, use buffers containing only BSA, which will hopefully lack substantial biotin content.
Streptavidin:biotin tetramer reagents: I’m thinking of antigen baits for detecting antigen-specific B cells, but you might be doing something else like homemade cytokine capture reagents. These are on my “To Test” list and based on other experiments I’ve done with them, they should be fine. The streptavidin-biotin bound is strong enough that it basically doesn’t come undone even if you store the tetramer in the presence of free biotin for years. So, yeah, probably fine.
Non-antibody reagents
How to get the Brilliant dyes stabilized:
The BUV dyes generally give the highest rates of interaction, so focusing on these will get you the most benefit.
Reduce the concentrations of Brilliant dye conjugates using overnight staining as much as possible.
Avoid using BUV dyes on high expression markers. I'll have to prepare some examples to show what's going on here at some point, but it appears that if you have a high density of BUV labeling, this provides a "nest" for Brilliant dye interactions. These are lower affinity interactions, so having higher density of dye labeling will increase the avidity, thus increasing the unwanted effect.
Identify any pairs that create interactions in your testing. Then swap out one or the other (or both) of the interacting conjugates with a substitute. In many cases, a direct swap with a different clone or the same clone from a different vendor on the same fluorophore will fix or reduce the problem. I don't know why. Alternatively, swap to a dye from a different family for the problematic conjugate(s).
Alternatively, we can swap the CD45RA to StarBright UV510, which is spectrally similar to BUV496, with a bit less spillover in the UV and slightly more in the short end of the blue.
Since StarBright dyes don't interact with SIRIGEN polymers, this eliminates the Brilliant dye interaction entirely.
Some suggestions for the cautious:
The most likely failures come from Brilliant dye interactions, particularly the Brilliant UV dyes. If you don't go through the optimization steps above, you can prepare the master mixes without those and add them at the point of use. This is still going to be easier and more consistent between batches than preparing the whole mix fresh each time.
The next most likely source of failure is light exposure. Hence the black tubes. Cold storage is also going to help, so if you're repeatedly using a master mix, keep it on ice while it's out of the fridge and consider aliquoting it to minimize heat cycling and the potential for contamination (and other handling issues).
Tandem breakdown appears to be a non-issue with the tandem stabilizer added and the use of black tubes. This make sense because the storage conditions are very similar to the stock vial. Again, if you want to be extra careful, you could leave out the most labile tandems (PE-Fire 810, APC-Fire 810, PE-Cy7) from the mix. As mentioned above, be careful with your panel design and use these bright, labile tandems on low expression markers.
More Brilliant Stain Buffer probably isn’t going to fix your problems. It does not for me and causes a measurable albeit slight increase in background. Nor does the Brilliant Stain Buffer Plus work better in my hands.
Finally, I'll mention that there are alternatives to making your own stabilized master mixes. For instance, BioLegend (and I think BD) offers a service to prepare lyophilized aliquots of your panel. This can be particularly helpful if you won't have cold storage where you need to perform the staining.
Hope this helps, and good luck!
White-throated Swallow, South Africa