Mix barcoding agent - Pd and CD45-Pt

[ssaintsoon]

Hello CyTOF community,

I'm new to CyTOF barcoding and have some questions about using Pd and CD45-Pt barcodes together.

Initially, we didn't consider barcoding options for our panel, so we have two antibodies assigned to 106 and 110 Cd, which are also used by the 20-plex Pd barcoding kit.

Thus, I'm considering mixed barcoding agents and referring to the Gadalla et al. protocol. (https://doi.org/10.1016/j.xpro.2022.101643). Briefly, they barcoded 40 samples using:

CD45-194Pt with 20-plex Pd
CD45-198Pt with 20-plex Pd
Total: 40 samples
My question is: Is it possible to create an 8-choose-3 barcode using CD45-Pt and Pd together (e.g., 102, 104, 105, 108 Pd, and 194, 195, 196, 198 Pt)? Or should I tag CD45 only once for each sample?

Any suggestions or comments are much appreciated!


Best,
Kwon

[mleipold]

Hi Kwon,

A few things to discuss:

1. The 20plex Pd barcode system (or the Tellurium barcoding that SBio has in beta: viewtopic.php?f=7&t=5070) is cell-type agnostic, whereas CD45 only works properly with cells that express CD45. In many cases, that's fine, as you're studying T cells or something, but can be an issue for other sample types (bone marrow) or cell types (CD34+) which don't express CD45 as much. Even CD298 or b2m barcoding can slip up depending on marker expression.

2. The Pd barcode system is 6-choose-3 (102, 104, 105, 106, 108, 110). If I'm understanding you correctly, you're proposing to add a 4-choose-1 (194, 195, 196, 198) scheme of CD45-Pt on top of this, for a final result of 10-choose-4 (3 Pd, 1 CD45). This should work out, as each sample will only receive 1 CD45 (and 3 Pd).
- you have to be careful when doing affinity-reagent-based barcoding like CD45, as the individual CD45 probes are generally the same clone, which means they're in competition with each other. So, 1 CD45 would 100% of CD45, 2 CD45 would be 50/50 in each channel, 3 CD45 would be 33/33/33/ in each channel, and so forth.
- you can see this competition in Figure 1 here: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4323739/ ; as a result, I probably wouldn't recommend doing more than 3x CD45 in a *single sample*.

3. 8-choose-3 would give you 56 possible barcodes ( https://www.ncbi.nlm.nih.gov/pmc/articl ... gure/F223/ ): therefore, if 56 BCs are sufficient for your experiment, it would be possible to do this using no Pd 20plex reagents at all, only combinations of 8 CD45 channels in groups of 3. SBio sells several CD45 reagents (including Cd and Pt), so it would be possible to do it with just catalog reagents if you choose.


One final issue (that I've seen a number of people inadvertantly make): If you *do* use any CD45 barcode reagent, remember this is also probably in competition with any CD45 lineage reagent you might be using (eg, same clone as CD45-89Y in MDIPA, for example). So, if you *are* using a CD45 lineage reagent for gating purposes, either make sure you're using a different lineage clone, or take that into account with your experimental set-up, including order of steps. For example, if you stain CD45 BC first and surface lineage second, your lineage staining is going to be a lot lower than if you did it first. However, if you do lineage first and CD45 BC staining second, your BC staining will probably be dimmer.

Formally, you can dodge this a bit if you decide that anything that is properly CD45-barcoded is therefore inherently CD45+, and therefore anything that makes it through debarcoding is CD45+ (in other words, debarcoding has kind of gated on CD45+ for you).


Mike

[rhannan]

A followup on Mike's comments -

re: CD45 - if you're working with human cells, do note the loss of staining intensity in the CD45 clone HI30 after fixation. (See figure 1 here https://onlinelibrary.wiley.com/doi/ful ... to.a.24317). If you're working with other clones, I can't vouch for their lability - some might have more robust binding to fixed antigen. Anecdotally, I can confirm that fixed PBMCs (paraformaldehyde or SMART TUBE) barcode VERY POORLY with HI30. To fully implement your strategy with fixation labile CD45, you'd need a 'live barcoding' CD45 step, and a subsequent 'fixed barcoding' Pd barcode step.

I'll second Mike's suggestion and say if you're already considering making a partly CD45 barcode - consider having the whole barcode be CD45. Depending on how you set up your workflow, CD45 barcoding can allow for pooling of barcoded sample for combined surface staining prior to fixation. We do this with an 8-pick-4 scheme, and it works well.

EDIT: a quick thought - if your 106 and 110 channels are required for your staining panel, you won't be able to use any of the pd barcode as they come premixed and you (AFAIK) cannot buy individual Pd barcode masses. You probably wouldn't want to be mixing staining channels with barcode channels.

[mleipold]

Hi all,

Riley: good catch on the 106/110 issue, somehow I missed that from Kwon's first post.

I agree, if you're using 106 and 110 as a probe/marker antibody, you should *not* use 106 or 110-containing BCs. Formally there would still be some of the 20plex premixed codes that you could use, but that would drastically reduce your options. But that would still leave Kwon several of the Cd-CD45 and all the Pt-CD45 options.

I would also note: it's tricky to use cisplatin when you're using Pt-conjugated antibodies. It *can* be done if you're careful: for example, right now, I have a project where I have 198Pt-CD3 in-house conjugate and natural-abundance cisplatin as live-dead. In my case, 198Pt is relatively "far" from 195Pt (the highest channel in nat abund cisplatin at 34%, and throwing out anything 195Pt-high), but it probably would have been even better if I had used 194Pt-cisplatin instead.

So, if you're using cisplatin as a live-dead *and* Pt-conjugated antibodies (commercial or in-house), I would recommend using isotopically enriched cisplain (eg, 194Pt-cisplatin), and then using the *other* 3 Pt channels for antibodies.


Mike

[rhannan]

Yes, you'd be limited to barcodes without 106 or 110 - meaning barcodes 1, 2, 6, and 12 from the 20-plex pd barcode.

My unsolicted advice: Kwon, if the ab on 106 and 110 is what is causing all this strife, it'll most likely be cheaper and easier to re-home 106/110 antibodies onto other mass channels, and be able to use the 20plex Pd barcode kit.

As much fun as it can be to make your own barcode, implementing/QC of homebrew barcode is non-trivial expense+labor that might not be the best use of your resources.

[ssaintsoon]

Dear Mike and Riley,

Thank you for all these comments! I apologized for insufficient explanations about my experiment;
- My sample is human PBMC so I think CD45 barcode should be fine.
- As Riley pointed out, 106/110 is already occupied by ab. Thus only four Pd premix will be availabe in my panel. Other Cd channels (112, 114, 116) are also assigned to ab.
- The number of samples will be ~60 with 4 groups. (Each group has 15 samples)
- The instrument is Helios.
- Considering CD45-Pt as barcode, viability marker will be 103-Rh intercalator.

Mike:

I have CD45-89Y in my panel (HI30) thus it will compete indeed.
- As you suggested, debarcoded cells might be CD45+. Does that means 89Y-CD45 could be employed as barcode? For example I have 194, 195, 196, 198Pt - CD45 and four Pd premix, resulted in 16 combinations, but if 89Y-CD45 is included, it will be 20. Theoretically it looks fine with me.
- I wonder there are no differences between 89Y and Pt signal intensities for CD45. If they are quite different, it may yield different numbers of CD45+ for each sample.
- Thanks for your clarification about 194Pt abundance! I wasn't aware of that issue.

Riley:

I totally agree with you regarding CD45 loss after fix. Thanks for your reference! What I planned was barcoding after staining. It cannot reduce batch effects at staining but better than signal loss.
Whole barcode be CD45 is much beneficial as it's "live" barcoding, but unfortunately I have no available Cd channels for now.
I will consider re-home 106 and 110. Might be simpler way. (You reminds me "Gordian knot")

Thanks for deeply considerated comments!

Best,
Kwon

[mleipold]

Hi Kwon,

Yes, you can use 89Y CD45 as a BC. There will be some signal differences between 89Y and the Pt CD45. However, the important part in the debarcoding is the *resolution* ("gap") between positive and negative in a given BC channel.

See Fig 4 here: http://dx.doi.org/10.1038/nprot.2015.020
And Fig 4 here (Slightly updated version of other paper): http://dx.doi.org/10.1142/9789813207813_0054

So, even if the 89Y and say, 196Pt absolute intensities are different, as long as there's good resolution between 89Yneg and 89Ypos, or 196Ptneg and 196Ptpos, the algorithm will debarcode properly.


Regarding your antibody panel vs open channel for CD45 BC: If I understand you correctly, you have the following channels open for potential CD45 BC: 89Y, 194, 195, 196, 198Pt, which would give you 5 channels and a max of 10 codes. It also looks like you also have 111Cd and 113Cd open: if so, that would give you 7 channels, for a max of 35 codes.


Mike

[rhannan]

Dang, I got scooped by Mike - yes, CD45 89Y is useful on a barcode. 89Y has a lower dynamic range / staining index but, as Mike said, so long as you can resolve + and - populations the zunder debarcoder is very robust. You may see slightly reduced yields/QC (bc sep/maha dist) on barcodes containing 89Y. Anecdotally, we used a higher amt of CD45 89Y than the Cadmium conjugates in our CD45 barcode (89Y,106/110/111/112/113/114/116Cd).

Kwon, I want to clarify here -

For example I have 194, 195, 196, 198Pt - CD45 and four Pd premix, resulted in 16 combinations, but if 89Y-CD45 is included, it will be 20

If what you are describing is including the four SBT Palladium barcodes which aren't obviated by your requirement for 110 and 106 - those will have to be applied at a second step, post CD45 barcode staining and regular surface panel staining. e.g. you have a 5-pick-3 (89,194/195/196/198) of CD45 barcode, giving you 10 unique barcodes. You can make four sets of these 10. Subsequent to surface staining+fixing, each set of 10 will be given one of the four valid SBT palladium barcodes. This gives you a max plex of 40. I think this is the highest-plex you can squeeze out of this arrangement, but I may have my maths wrong somewhere.

[ssaintsoon]

Dear Mike and Riley,

I'm glad to hear that 89Y-CD45 could be used as barcode! So signal differences are not problematic in bc as long as the debarcoder can distinguish bc +/-.
And Mike, yes I have open 111Cd for additional CD45 in my panel. I don't know why how can I forgot about it. Thanks for reminding me.
Thus, if I apply 6-choose-3 on 89, 111, 194, 195, 196, 198, it will generates 20 barcodes. AND combine with four SBT Pd barcodes, it looks like 80 barcodes are maxium numbers.

Huge thanks for your help and comments! This forum and your kindnesses give me a lot of insights.
Thank you, as always.

Best,
Kwon