Before we get started, it's worth pointing out that while overnight staining is new to flow cytometry, it is pretty commonly used in immunofluorescence. Similarly, ELISAs are often left to incubate overnight for the binding of the analyte to the primary antibody. Perhaps it's also worth considering what the appropriate incubation time might be for an antibody in vivo where antibodies have half-lives measured in days.
So, in order to understand how the specificity of antibody binding is affected by overnight staining, we first need to clarify what we mean by specificity and how that works. Specificity is the signal we want to see (binding of the antibody variable domains to the desired antigen) as opposed to the signals we don't want, which is the antibody binding to pretty much anything else.
Specific binding of antibodies to the target antigen generally follows a sigmoidal curve. What this means is that there's a range where adding more antibody will give us a linear increase in signal, but after a while, that stops being the case. Increasing the concentration further won’t increase the specific signal, and instead will actually increase the non-specific signal, which is following a different curve. Since the curve is sigmoidal, in the plateau region on the upper right, moving from right to left means orders of magnitude reduction in antibody concentration with minimal change in specific binding.
What about non-specific binding? Well, it's probably best to think about that as the sum of many different interactions. The antibody can bind specifically to Fc receptors such as CD16, CD32 and CD64 via the Fc portion. Glycosylation on the antibody can also cause specific, low affinity interactions with many cellular receptors. Fluorophores can be charged or non-polar, facilitating weak interactions with cellular components, and the sheer size of antibody:fluorophore conjugates means they can get "stuck" to the cell or inside it.
Each of these non-specific interactions will have its own affinity (or how likely it is to stay stuck), all of which are generally a lot lower than the antibody's affinity for its antigen. Since there are many ways in which an antibody can bind non-specifically, each with its own binding rate and different pool of binding targets, I tend to think of non-specific interactions as increasing more or less in a linear relationship with antibody concentration and incubation time.
What this means in practice is that there's a range of antibody concentrations where we have good specific signal over the background noise. That range may be narrower or broader depending on how good our antibody is (high affinity or low affinity), the expression level of the target and how well we've blocked the non-specific interactions (a topic for another day!).
None of these considerations has to change with overnight staining. We still need to test the antibodies to determine the optimal dilution to use. That optimal dilution will probably have shifted. As a result, we'll be able to test concentrations of antibody that, while fine for short staining regimens, are vastly excessive for an overnight staining. This will of course generate non-specific staining, the same way you might observe non-specific staining if you incubated your cells in undiluted antibody for a standard time frame.
If you choose to use permeabilized cells, this does open the door to many more opportunities for non-specific binding. The interior of the cell is vastly more complex than the surface, and there's a greater chance of the conjugate getting stuck in intracellular compartments. In most cases, though, this can be managed in the same way, by titrating the antibody.
Let's have a look at how to adapt from a short staining to an overnight protocol. In this case, I'm using mouse splenocytes treated with the eBioscience Foxp3 Fix/Perm kit, so the interior of the cell is accessible. I should point out here that I always titrate the antibodies on my cells of interest in the panel. There’s really no point in titrating your T cell activation marker on splenocytes or PBMCs if you can’t differentiate the real staining from non-specific binding to monocytes or macrophages. So here we are looking at CD4 T cells. Foxp3 is being reported by the expression of Thy1.1 on the cell surface.
First, start with the reference staining pattern that you’re used to from a short surface staining. The same concentration usually isn’t appropriate for a longer incubation, and here with PD-1 we’re getting a lot of non-specific binding.
As we reduce the concentration, we start to see the correct staining pattern emerge.
And eventually we get good staining again:
To reassure you that specificity doesn’t have to be a big concern, here are some examples of overnight intracellular staining with negative controls. Rather than using an isotype or trying to block with unlabeled antibody, here we're looking for staining with the same antibody, but in the absence of the target antigen. All that's left should be non-specific binding.
In mice with Cre-driven excision of IL-2, we don’t see any staining for IL-2:
And here’s an example of phospho-STAT5 in cells with and without IL-2 stimulation:
In the unstimulated control we don’t expect much pSTAT5 except maybe some in the Tregs due to capture of endogenous IL-2 via CD25. With the stimulation we get nice upregulation of pSTAT5.
Actually, in some cases overnight staining can reduce non-specific staining. With the antibody concentrations being so much lower, it’s likely that low affinity interactions, which drive non-specific staining, are much less likely to occur. The antibody is more likely to get “used up” binding to the intended target, rather than having a bunch of excess antibody free to bind other stuff.
In this example, non-specific interactions between brilliant violet conjugates are reduced with the overnight staining with a low concentration of antibody. And, in the presence of the Brilliant Stain Buffer, this non-specific interaction can actually be completely killed, unlike in the short staining condition.
Similarly, I get reduced non-specific binding to macrophages with lower concentrations of antibody in overnight staining. Here we’re looking at PE-Cy5-conjugated antibodies against NK1.1 and TCR-beta, neither of which is expressed by macrophages. However, PE-Cy5 and other PE tandems are notorious for non-specific binding to myeloid cells. These histograms show the PE-Cy5 signal on F4/80 macrophages. With the overnight staining, I can drop the PE-Cy5 antibody concentrations ten-fold, get brighter staining on the targets and reduce the non-specific binding to macrophages. This will improve my ability to gate on the target cells—I’ll have better separation—and also reduce the overall spreading in high parameter panels that would be created by the extra fluorophore noise on the myeloid cells.
In the interest of saving you all some time and money, here are some antibody:fluorophore conjugates that do not work well with overnight staining. Each of these antibodies works well in overnight staining in other colors; these particular conjugates are sticky.
Anti-mouse Ly-6G APC-Fire 750. Other conjugates, including APC-eFluor780, work fine.
CD19 Monoclonal Antibody (eBio1D3 (1D3)), APC-eFluor 780, eBioscience™
CD19 Monoclonal Antibody (eBio1D3 (1D3)), PerCP-Cyanine5.5, eBioscience™
CD28 Monoclonal Antibody (37.51), PerCP-Cyanine5.5, eBioscience™
CD44 Monoclonal Antibody (IM7), eFluor 506, eBioscience™
CD62L (L-Selectin) Monoclonal Antibody (MEL-14), APC-eFluor 780, eBioscience™
APC-R700 Hamster Anti-Mouse CD152 (UC10-4F10-11)
anti-human CD45 Monoclonal Antibody (HI30), Qdot 800
PE-Cy™5 Mouse Anti-Human CD152 (too bright)
NovaFluor conjugates
And here are some that are best used with surface staining (not permeabilized), whether it be short or long:
BD OptiBuild™ BV480 Rat Anti-Mouse CD192 (CCR2)
BD OptiBuild™ BV750 Rat Anti-Mouse CD192 (CCR2)
APC/Fire™ 750 anti-mouse CD192 (CCR2) Antibody
BD Horizon™ BB515 Mouse Anti-Mouse CD199 (CCR9)
CD199 (CCR9) Monoclonal Antibody (eBioCW-1.2 (CW-1.2)), PerCP-eFluor 710, eBioscience
anti-mouse CD69 (H1.2F3)
anti-mouse GL7
PE/Fire™ 810 anti-mouse CD366 (Tim-3) Antibody
LAP (TGF-beta)
That's out of >800 anti-mouse antibodies and >100 anti-human. As I mentioned in the last post, for most human surface markers, you'll be better off using cells that are not permeabilized.
Reagents used:
Really enjoying these posts, thank you! I am curious what your approach to staining with a large panel when you have many BV dyes. From your previous post, preservation of Tregs was greatly affected by the media used, and you recommend DMEM or IMDM with some additives. For the BV dyes, using the BV staining buffer resolves the issues with BV::BV interactions, but may not preserve the Treg populations. What are your thoughts on this?