FAQ  •  Register  •  Login

Stability of Maxpar conjugated antibodies

Forum rules
It is fine to promote your company's reagents. Just make sure they are relevant to CyTOF, and do so in moderation and style :-)
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Feb 06, 2018 6:02 pm

Re: Stability of Maxpar conjugated antibodies

Hi Tomas,

The "no metal" baseline/threshold is something that, honestly, is still being worked out.

Most of the time, I use enough beads that I can see a tiny pellet in my well or tube, and therefore know that I *have* beads. And then assume that if I don't see signal, that the antibody didn't bind. I know from doing no-antibody control bead samples that the beads themselves don't have any signal in the Ln region.

Spherotech is developing some Eu-core beads. Since Eu is fluorescent, they can be used in flow and not photobleach. It's just a happenstance that it also makes them useful for CyTOF. I've gotten advanced samples of variable-Eu no-capture beads, and also constant-Eu variable-capture-density beads. Both sets are visible in the CyTOF. Unfortunately, I don't think they're in official release yet.

The main things you might want to play around with in these experiments is your Lower Convolution Threshold, and your Lower Event Length parameters. The default settings for both of those assume a decent amount of metal associated with your cell (usually driven by your Ir staining). Therefore, make sure you save your IMD file and try reprocessing it with different LCT and LEL values (usually much lower) to make sure you're actually "seeing" all your events.


The ideal bead for these experiments would be loaded with constant Cs133 (since I'm not using it for antibodies) to drive ion cloud size and signal intensities, and then have no-capture (baseline) through variable-capture beads in a ladder going up. But Fluidigm and Spherotech seem slow to capitalize on the idea.....


Mike
<<

GregBehbehani

Master

Posts: 85

Joined: Tue Apr 12, 2016 10:17 pm

Location: The Ohio State University, Columbus, Ohio

Post Tue Feb 06, 2018 6:43 pm

Re: Stability of Maxpar conjugated antibodies

Hi Mike and Tomas,

Wow, I'm surprised at the interest in this.

Mike, answers to your questions:

-Q: "To confirm: the idea would be that the binding to the IgG capture beads shouldn't change significantly even if antigen-binding activity goes down?
Or, in other words, if the beads retain their signal but the cell signal starts going down, then it's the antibody activity (antigen-binding) that's decreasing; but if the beads start losing their signal, it's metal loss?"

-A: Yes, exactly. I can't say we've definitively proven this (though I'm not exactly sure how we would; heat shock perhaps?), but that's been our experience.


-Q: "Side note: how much variation in signal intensity do you get with the antibodies on the capture beads under saturating conditions? The reason I ask, when I've tried it with BD's kappa capture beads, I've seen signals from ~150 Dual to ~1400 Dual, depending on the metal+clone."

-A: Each metal has a different sensitivity in the mass range, and each antibody has different number of cysteines available for conjugation, so it does vary quite a bit. However, if you label the same antibody on the same metal (which we often do) it's fairly consistent. Also when we've gotten and antibody that's had an unexpected low index, it has been dim when titrated. Relabeling it (provided the index goes back up to expected) results in an antibody that works as expected. It's reliable enough that I'll give my technicians a hard time if their labeling index comes back too low (but they know I love them).


-Q: "In the above comparison for activity vs metal, you would of course be comparing Day 0 to Day X for a given antibody prep, not between clones."

-A: Yes, exactly.



-"The main things you might want to play around with in these experiments is your Lower Convolution Threshold, and your Lower Event Length parameters. The default settings for both of those assume a decent amount of metal associated with your cell (usually driven by your Ir staining). Therefore, make sure you save your IMD file and try reprocessing it with different LCT and LEL values (usually much lower) to make sure you're actually "seeing" all your events."

-We keep our machine pretty clean, so we normally have our lower convolution threshold and event length set fairly low. We can detect the beads just fine.


Tomas,

-on capture beads: they are quite diverse, so be careful - most are polyclonal, so expect signal variation on staining time, concentration and temperature (and cost to impossible to reach saturation for all clones in that polyclonal mixture); some are monoclonal capture (anti mouse) these shall perform more reproducibly. Stating the obvious for a sake of completeness: you saturate the capture, so ideally always with the same conc of the tested Ab, irrelevant of the optimal titer of your Ab in your assay.

-Yes, we try to make sure to saturate the beads, which hasn't been too hard to do.

best,

Greg
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Feb 06, 2018 7:19 pm

Re: Stability of Maxpar conjugated antibodies

Hi Greg,

My main interest is in the overall sensitivity of the instruments. The HIMC has 3 instruments, and as far as I've been able to test, they're comparable in terms of both absolute signal and in mass sensitivity profile. But as the Tricot et al paper demonstrated (and as I've shown in talks for 5 Stanford instruments), that's not always the case.

So, I'm interested in anything that will help me better understand the instruments in my lab, as well as between labs (both regular research projects, as well as the Multicenter paper we just published). Especially if we're going to be combining data between sites, I think the CyTOF community as a whole needs a better understanding of "does my instrument look like yours?" and "How low can you go?".....especially if we start to comp data at the low end. The normalization beads definitely help, but they're not a 100% fix.


Mike
<<

GregBehbehani

Master

Posts: 85

Joined: Tue Apr 12, 2016 10:17 pm

Location: The Ohio State University, Columbus, Ohio

Post Tue Feb 06, 2018 8:29 pm

Re: Stability of Maxpar conjugated antibodies

Hi Mike,

I think this a great consideration, but I'm not terribly impressed by the Trichot et al. findings.

As you know, the CyTOF 1 machines were a pretty mixed group; "Jimmy" (CyTOF#2) in Garry's lab was clearly different from "Amy" (CyTOF#6?), and there were several different cohorts of machines after that; they were all fairly unique. I'm certainly not an expert in ion optics, but my understanding is that the mass filter design and settings were subtly different between the different instruments, and this has a big effect on ion transmission efficiency. Additionally, it doesn't look as though Tichot et al. took into consideration several important effects in their observed vs. expected calculations; specifically: 1. Differences in the manual tuning of each CyTOF 1 (which can have a huge effect on the oxidation, and +1 spillovers), 2. The expected differences in ionization efficiency and oxidation between the different isotopes, 3. the fact that a significant portion of signal in each channel would come from M+1 and M+16 spillover (it would have been better to run each metal separately, as Fluidigm has done), 4. the fact that the exact isotopic distribution of a given element can vary slightly depending on where on earth it was mined. 5. the detector/amplifier suppression effect that occurs of many ns after a "bright" signal (only on CyTOF1 machines). Thus, I'm not sure I would worry too much about the findings of this study.

From my conversations with them, Fluidigm has been working extensively on standardization of these parameters in the newer machines. Between these design changes, the improved ion optics, and the auto-tuning, the CyTOF2 and Helios machines behave much more consistently. There are still some strange effects (we're working on a paper about one of them), but they seem to be fairly subtle. I think even Fluidigm may be getting pretty tired of supporting all of the CyTOF1 "unicorns" out there and I've heard of them giving incentives for people to upgrade to Helios machines.

I think the best way to account for differences is exactly the way you recently published; run a standard reference cell with the same staining panel. I think this will be more accurate and relevant than anything you might do with IgG beads.

best,

Greg
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Feb 06, 2018 8:55 pm

Re: Stability of Maxpar conjugated antibodies

Hi Greg,

I'll definitely agree that CyTOFv1 instruments had a lot more variation between them. Even between the two that HIMC had (#4 and #22), there was a difference.

However, after Tricot et al came out, I did basically the same experiment on 5 CyTOFs at Stanford (3 v2 and 2 Helios), and saw differences with the same ICP-MS standard that Tricot used. Not just in ion transmission efficiency (which, since it's a standard solution, can be calculated explicitly from Found/Expected), but also had one machine that was an outlier in terms of its peak mass sensitivity (one v2 was shifted relative to the other two v2 instruments, which were more like the Helios instruments).

Since we used the same standard solution bottle on all 5 instruments, isotopic distributions would have been the same. We made sure to keep the signal intensities within a factor of 2 of the Tuning Solution, to make sure we could measure everything at the low ends but not blow the top off the detector in the peak areas.


I've attached a PDF of the Multielement data, as well as some prestained cell data between instruments (was done during the development work for the Multicenter paper).


Mike
Attachments
Leipold-Multielement and Stained cells-multiple machines.pdf
(745.77 KiB) Downloaded 397 times
<<

GregBehbehani

Master

Posts: 85

Joined: Tue Apr 12, 2016 10:17 pm

Location: The Ohio State University, Columbus, Ohio

Post Tue Feb 06, 2018 9:34 pm

Re: Stability of Maxpar conjugated antibodies

Hi Mike,

Thanks for sharing this; very cool.

But your data look really good, I don't see that much variation at all. Your normalized data on final slide look particularly great.

What's the problem?

I don't think you could do this experiment on 5 fluorescent flow cytometers and get any better alignment between machines (not to mention you'd be hard pressed to find 5 identically spec'ed flow machines).

Only the liquid data seems a bit off, but was this dual counts or raw intensity? If you're looking at raw intensity, I would expect more variation as the dual counting conversion plays quite a significant role in normalizing the data, and if not used, could account for a lot of the difference.

Regardless, I don't really care much about liquid data, it's the cells (and beads) that matter and their are some effects that might occur in liquid mode (esp. oxidation) that may be different and not all that relevant. It seems like your machines are very consistent with beads and cells. So I think you're doing great (not that I didn't know that already).

As for the Tichot paper, my issue with their found vs. expected calculation is just that the "expected" calculation seems to assume perfect ionization, transmission, and no oxidation; but no one would ever "expect" that. It makes it seem like the differences are due to some error in the machine "finding" the ions, rather than being entirely anticipated based on the relevant physics and chemistry.

best,

Greg
<<

mleipold

Guru

Posts: 5796

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Feb 06, 2018 10:12 pm

Re: Stability of Maxpar conjugated antibodies

Hi Greg,

The Stanford Liquid data was Dual, on freshly tuned instruments. For my calculations, I also did assume a 2% M+16 spill for all elements and their isotopes. BTW, in the last slide, the Orange and the Green are the same instrument, just before (Orange, v2) and after (Green, v2-to-Helios) Helios upgrade.

The Observed/Expected is basically calculating functional ion transmission (ion optic efficiency + ionization efficiency + increase or decrease at M or M+16 due to oxidation). My understanding from Fluidigm is that the ionization potentials of the lanthanides are basically the same, and effectively 100% under the plasma conditions (compared to Xe, which is only something like 10% ionized at each isotope). Also, the fact that the Eu isotopes (which barely oxidize) fall well on the same curve as La and Pr (in both Tricot and my data) suggests to me that oxidation isn't a major source of error in these calculations. (Note: if you want to repeat this at your site, the flow rate differene between Helios and v1/v2 *does* make a differene in Tricot Eq 1).


I guess my basic issue is, unless you *look* for differences with things like the Multielement solution or Prestained Cells, you won't see them, even if they're there. I definitely agree that solutions and discrete events like Beads or Cells aren't the same: the trends are usually the same but the magnitudes usually shrink.

But you also do have cases like the CD16-Sm149 on NK cells in my last slide: in the Raw data, the Pink (CyTOFv1) Median signal is >100 Dual higher than the others (probably due to the different mass sensitivity profile seen in the Bead data). The Fluidigm normalization does bring it down, but the Orange and the Pink are still a different bunch than the other 3 instruments. And in the MATLAB normalized data, the Pink is now bunched nicely with the 3 others, but the Orange is still 150 Dual higher (50% brighter, in this example).

Note that the outlier machine in the Liquid data was *not* part of the Bead or Prestained cell data (we could only "borrow" the machine for a bit :)

If the Multicenter paper was getting a few different clusters (Prestained vs Site-stained) just based on an increase in negative population *background* for 3 markers, I have to assume that a difference of 150 Dual in the positive population would also affect clustering.

Regarding flow instruments: you're probably right....people generally agreed that our Multicenter paper's CV's were within the range expected between flow instruments/experiments. Maybe Tomas can chime in on his EuroFlow experience regarding flow instrument consistency and standardization, to give us a better reference.....


Mike
<<

JamesW

Contributor

Posts: 37

Joined: Tue Nov 21, 2017 6:59 am

Post Fri Feb 09, 2018 4:31 am

Re: Stability of Maxpar conjugated antibodies

Dear All,

Thanks for the information about stability and the further in-depth discussion, it is very helpful to someone starting out like myself. I am mostly using the dimmer metals for CD45 based barcoding as per the paper Mike and others published in JI a few years back. I will keep an eye on 139La. I may also try beads to monitor the staining index of newly labelled Abs, particularly for less expressed targets.

Best,
James
Previous

Return to Reagents for CyTOF

Who is online

Users browsing this forum: No registered users and 13 guests