Fluorescent Multiplex Staining of RNA and Protein Targets on the BOND RX Research Stainer

Hi, thanks. And thank you to Leica for inviting me to speak today.

I work in the Experimental Histopathology Science Technology Platform at the Francis Crick Institute, which means that we cover everything from tissue dissection, preparation, processing, through sectioning, staining, and imaging and analysis. We have quite a wide range of things we perform for researchers across the Crick. We offer advice and training in all these aspects of histology as well.

Today I'm going to talk about particularly some of the fluorescence work that we've been doing. It's a little bit of a mix of technical and some research projects. None of the research projects are mine, so they do lack some technical details from unpublished work, but hopefully the message will still be carried across.

We first got our BOND RX back in 2018, and we primarily bought it for doing RNA scope assays, automated, and we wanted to develop multiplex immunofluorescence as well. We didn't get started on any of the immunofluorescence and IHC work until after the pandemic. We're relatively new to it, but learning quickly. We do use our BOND RX quite a lot for routine immunohistochemistry. Find it very quickly and simple to use. We can run 30 slides in 2 1/2 hours, which is a lot better than some other slow stainers.

It's very simple to use. One of the things we've like about it as a research lab is the ability to just add whatever reagents that we want that are off brand. We've used different blocking reagents like fab fragments, things like that. We've had to optimize all different combinations of secondaries because we're a research lab working with things that are raised in rats and goats and hamsters and all sorts of odd combinations of antibodies. We've used different detection reagents, and even doing some staining enhancement with our DAB staining. It's very quick and easy to use and very flexible.

The antibody stripping capabilities of the bond are very good. We've not had any antibodies that have failed to strip nicely. By using just rounds of sequential heat induced epitope retrieval, we're able to then detect multiple antibodies in the same tissue, as many people have talked about today. This includes using antibodies with the same host species, which a lot of our researchers are quite amazed to be able to suddenly use two rubber antibodies in the same tissue.

As we've discussed before, some people like chromogenic detection and some people like fluorescence. With the chromogenic, so we're using DAB, fast red, HRP green, and HRP blues. One of the issues that we've come across is co-localing signals where the darker antigen can mask the lighter antigen. If you're looking at true co-localization of markers within a cell, it can sometimes give misleading results. There's some caveats with which order you can do chromogens in and there and ability through multiple rounds of heating.

Fluorescence also has its downsides. Although you could co-localize markers very easily because you image them independently, you do have a limited number of channels with standard imaging filters within microscopes due to the overlapping light spectra of the fluorophores. And, when especially if you're working in FFPE tissues, all tone fluorescence is the bane of your life.

Having said that, the way that we've majority planned to move forward is with fluorescence, and I'm going to talk a bit about some of the ways that we've got around some of those issues with some pharmacy machines. So probably teaching people that already know, but using immunofluorescence, some people like to use directly conjugated primary antibodies where your fluorophore is conjugated to your primary, and this allows high multiplexibility but low amplification of your signal. People get around that by using secondary antibodies that are conjugated, which allows you to multiplex when your primary antibodies are raised in different host species.

Then there's another technology called Tyramide signal amplification, whereby use an HRP conjugated secondary antibody that detects the primary antibody, and then use to add your TSA or tyramide signal amplification, and it deposits the fluorophore or whatever marker you're using into the tyrosine side chains of the protein and the surrounding tissue. You get a very high, I think it's something like 2 to 4 logs amplification of your signal using tyramide signal amplification.

I just want to do that brief introduction on that. Oh, and then you can strip your antibody complex using the heat-mediated epitope retrieval, but the conjugated heresy inside chains with the fluorophore or marker remain in situ, so it allows you to apply another round of primary and secondary antibodies, even if there's other species.

We've recently acquired a Vectra Polaris from Akoya Biosciences, which has now been rebranded as the Pheno Imager HT. And this is a multi-spectral side scanner, which allows you to identify spectral signatures to separate channels. This allows, so rather than just having the channels, the spectra to be very distinct, it allows you to identify what the whole spectrum is of the emission of the fluorophore, which allows you to put them a lot closer together.

This machine can identify 9 different colors, which includes 8 fluorophores and DAPI, so you can have a nine-colored image. It also features autofluorescence removal, so it can record the specific emission spectrum of the autofluorescence and can then digitally subtract that from your images, giving a much better signal-to-noise ratio within your images. And this technology also uses the opal fluorophores, which are TSAs, as I just described, which you can see, although the spectra are very close together, the machine is able to extract and separate these very close signatures. And this is just a little pretty example, which probably looks better on the other screen, of some multiplex staining that we've done.

To show you the auto reference removal that's possible in this machine, this is a quite low power picture of a human juvenile thymus. You can see in this picture on the left that there's quite a lot of tissue background, particularly in the sort of the teal channel, which is around the 480, and you can see here there's quite a large collection of red blood cells as well in some of the tissues. With the autofluorescence removal, it's labeled to separate that signature of the auto fluorescent blood cells and the tissue background from the real signal from the antibody that's placed in the 480 channel. It's a helpful tool to clean up your images post-processing.

How do we go about building these panels of multiple antibodies? We start by, as with everyone, optimizing antibodies and validating that they are indeed specific. We've talked quite a lot about the need for antibody validation, and that still applies here before you start anything multiplexing. All our antibodies are first optimized for the heat, for the antigen retrieval and the dilution that we use on the BOND RX, using DAB staining in single colors.

Once we're happy with the staining from our antibody, we then test the antibodies in different positions in the protocol. We'll test it with one heat strip, with three heat strips, with five heat strips, and with seven heat strips. And some antigens, you'll get improved heat staining over the course of the rounds of heating. For example, this one here, you can see the signal is much stronger and more intense after five rounds of retrieval. But some antibodies don't survive through that much heat. You have to decide and compromise which antibodies you're going to put first and which ones you're going to put second.

Once you've done all this pre-work, you can then decide to put them all together in one slide. But at this point, you must, again, make lots of choices and trade-offs. When you're pairing your different antibodies with the fluorophores, firstly, you must consider the fluorophore brightness. The opals are all different intensities. There's no point putting your weakest marker with the lowest intensity fluorescence. There's a bit of a trade-off here as to which ones you pair with which fluorophores. It does also mean that you can, one that you're struggling to see in DAB, if you put it with one of the strongest fluorophores, sometimes the TSA amplification will give you a boost in the signal versus even the DAB signaling staining.

It's a good idea to avoid co-localizing markers in very closely overlapping spectra because it is a machine learning It looks at the spectra, tries to work out what's what, and there's no point putting two stains that co-localize in all the same cells in very close spectrum. You need to separate those further apart in the spectrum. It's also important to consider what overlaps with the auto fluorescent spectra, because you're going to remove that.

Here's an example where we put a KR67, which I think you can see in the raw image here, in the 480 channel, which overlaps a lot with the autofluorescence. You can see here in the raw image there's quite a lot of nuclear staining all over because this is an immunized spleen. Then we found when we unmixed the sample and unmixed the fluorophores and removed the auto present, we ended up digging into our 480 signal because it was a nuclear stain and looked a lot like red blood cells. It's good to avoid anything that looks like red blood cells. And you can see here, this is a pseudo pathology review, which is just done in brown and blue to make the pathologists happy.

If you look at the individual nuclear, you can see that they've almost been like chipped away the actual real antibody signaling instead of getting a nice blanket nuclear signaling that we get. It's important. what you choose to put near the autofluorescence spectra as well as you're going to be removing it. And Opal 780 must be the last in the sequence because it has a slightly different chemistry that is a two-step amplification.

Once you've put together, considering all these factors with all your different antibodies, you can then go ahead and test it. And then after that, there tends to be a phase of reviewing and sometimes certain antibodies just don't look great with certain opals. You might want to switch them around. Sometimes you find that you need to change the titration of the opals as well. And at this point, you don't tend to change the antibody conditions, you tend to change the opals. And we've used everything from one in 100 to one in 2000 dilution of the opals. There's quite a range over which they work with different ones. You must optimize it for your epitope of interest. 

I'm going to talk about a couple of the projects that we've worked on with these panels. This is a project from human juvenile thymus, where the researcher has identified through single cell RNA sequencing and various other methods, several novel populations of stem cells within the human thymus. And they essentially needed to identify these cells for co-localizing markers to definitively say that that is the correct cell. And this also includes some of their novel RNA seq markers, which is why I've not labelled what the antibodies are there. This is still unpublished. And through using this multiplex system, they were able to then identify the populations of their stem cells, both along the subcapsular region of the thymus and within the medullary region within the same tissue using this and they were able to quantify the different populations. That was the nail in the coffin for their in vitro work that we've been able to do as well.

The second project I'm going to talk about is from the high lab at the Crick. They're looking at radiotherapy treatments on mouse models of tumors, and directly translating that into the radiotherapy treatments that they're giving in the clinic. It doesn't look amazing on this screen. It looks a bit better on the other screen if you can see it there. This is a GFP marker of the tumor with a CD8 marking T-cell infiltration. You can see that with radiotherapy alone, there are very few T-cells at the population, and they also seem to be stalled at the border of the tumor here. Whereas when they combine their radiotherapy treatment with drug treatment, they can see a massive influx of T-cells and a breaking down of this barrier at the edge of the tumor. And they were able to observe this just by looking at CD8 in single IHC stains.

What they wanted to do was be able to see what's happening with those cells. What are they interacting with? What's happening there? We built up a panel for them. This is only a five-plex panel, which included collagen, podoplanin, smooth muscle actin, and then still the GFP and the CD8 to identify the tumor and the T-cells. And then they were able to see the interaction between the yellow podoplanin fibroblasts here with trapping the CD8 positive cells at the border. Whereas when they combined with this drug treatment, they were able to see a complete breakdown of that barrier at the level of the fibroblasts. And they're doing quite a lot of analysis to understand the interaction between the different cell types and how this affects the treatment.

If you look at the whole tumor, you can see here that the tumor that was treated with radiotherapy alone, you can see the CD8 cells stalled at the boundary here, whereas in the double treated example, they have a breakdown of the tumor already and getting a lot less clear boundaries of the tumor. And that correlates with clinical aspects that they've seen of how patients are responding to the radiotherapy in the clinic as well.

Next I'm going to move on to talk about RNA scope. We've already heard quite a bit about it today, so I won't go into the technical aspects. But running it on the BOND RX is very simple. In our lab, we've done brown and red single chromogenic assays, one, two, three, and four plex. Fluorescent RNA scope which uses the Akoya opal fluorophores base scope with red chromogenic staining. And we've basically been applying this in all sorts of different research things. We've gone from everything from mouse FFPE through to human cells that are grown in culture and then sectioned. Anything that you can get on the slide, you can do the staining on the machine. We've had cytospins, human, mouse, frozen, FFD, you name it, we've probably done it. And all of these have individual requirements for the pre-treatments that you give for the RNA spoke. So basically, that is a different optimization for each of these projects.

There's a lot of researchers that are interested in combining RNA scope with antibody staining. And there's just a couple of examples here. We had research that was looking in the brain at GAD positive glutamatergic, sorry, GABA-rhagic neurons and V-GLUT2 positive glutamatergic neurons in the brain. But when they used antibodies against those targets, they were so highly abundantly expressed, they couldn't find the cell bodies of the neurons. They switched to using RNA scope, which discreetly labelled the individual cells nicely of the inhibitory and excitatory neurons within their sample. And then they were able to combine this with an RFP antibody in their transgenic lines that allowed them to see what populations of cells they were marking with their RFP tracer.

This second image here is not done on the vectrophi, so you can see the difference. And this was an example where the researcher wanted to combine a protein and the transcript of the same target because they were interested in looking at the regulation of the transcription and translation of this target here. And finally, we've also had a play around with this chromogenic as well. We've put our antibody here, which is an amyloid, so you can see a nice little amyloid plaque in the brain. Combining that with the HRP green chlorophore with a brown RNA scope, and we use our red counterstain that we did by hand because at the time, it wasn't a red counterstain on the blonde, but there is now.

We found through our experience that many antibodies that have come our way seem to work sequentially after RNA scope. We've had a little play with the co-detection assay that we talked about earlier. But in our experience, most antibodies seem to be fine after the RNA scope. The pre-treatments are quite mild in terms of antigen damage. Most of our projects, we haven't had massive problems with that.

We've also then been able to extend this into imaging on the Vectra Polaris, so using multispectral imaging on a joint RNA and protein staining. This is an example from a lab that we're looking at different hormone receptors across the whole brain in a mouse. And they were able to use four different RNA scope targets combined with an additional 5th antibody to identify neurons. This is a NeuN standing in red here. And they were able to build up using serial frozen mass sections. They built up an atlas of the whole brain of how these different receptors were expressed across brains of different conditions for their experiments. It can be done in quite large scale as well with the unmixing and the scanning.

To finish, just going to talk a little bit briefly about some of what we're planning to build from here on. Obviously, we want to get going with more multiplex panels. They take quite a lot of time to build up with all of those different individual steps, but we're getting more requests for those panels to be built.

We're starting a project trying to combine RNA scope with multiple antibodies. The theory is that if one antibody works after RNA scope, then why can't you use four? We're now starting projects tentatively with two antibodies where we're use the heat stripping of the antibodies in between after. In theory, we can end up with up to 8 mixed targets of RNA and protein that we can again image on the Vector Polaris.

We have some ongoing projects adapting the four-plex RNA scope to be used on an imaging mass cytometry platform, which I can't talk a lot about the details of because it's still in the early stages. But this is exciting. They're then planning to be able to image with heavy metal conjugated antibodies. They have a panel of 30 antibodies already optimized, and they want to put some RNAscope on top of that. We've found that we can combine four RNAscope targets, and then have up to a panel of 30 antibodies. And of those 30, we only had two antibodies that struggled after the RNAscope pre-treatment, and we were just able to pick a different clone and get around those issues. We're playing with the adaptability of the system.

We’re future looking, hoping to think about trying to do bleaching of fluorophores either chemically or with light to be able to further increase the multiplexability capabilities of our immunofluorescence. Both by using conjugated primary antibodies and potentially also using amplified signal as well and seeing if we can manage to bleach the section so that we can re-stain with more antibodies and build up our multiplexability even more.

That's all I was going to say about what we're doing. Just a quick thanks to everyone that's in my team and our collaborators. All the pictures you've shown are from these researcher’s projects. And then we've also worked closely with Leica, Akoya, and ACD in getting all these assays up and running. Thank you.