Progressive increase in signal background

[EHaasDFCI]

Hi Everyone,

This is a great thread and ongoing issue here.

Adeeb, that seems to make sense to me. I believe there may be multiple causes to this increase in signal over time. How long were the samples acquired for in your report? Also, were they intracellularly stained or just surface? Were the samples, especially the without water rinse samples, which were run second, sitting at RT or in the fridge? Also, were they barcoded?

I have a hypothesis that the combination of barcoding, intracellular stain, and long acquisition/sitting in solution over the course of a day may be causing some of the same issues seen with the wash solution case. In the attached pic, you can see the concatenation of 4 files, each run in succession, without wash solution running in between with exactly the same pattern as seen above. This sample was barcoded, intracellularly stained, and run in CAS over the course of the 4 two hour runs. I believe that the amount of fixing and perm that is happening with the aforementioned steps is causing the degradation of the cells and over the day, even with the sample sitting in the fridge, sitting in CAS still causes some degradation. As you can see, the signal is slightly better at the beginning of each successive run than the end of the previous run, however, the starting signal gets progressively worse over the course of the day.

strange signal shift-5-8-19.JPG (35.28 KiB) Viewed 11496 times

Has anyone seen anything of the sort?

Eric

[OneJuan]

Eric,

Very interesting. It also happens in 141Pr in your case. Do you see it in other channels to the same or similar extent as well?

What normalization method are you using?

The samples from our group were normalized with the Fluidigm CyTOF 6.7

Thanks,
Juan

[mleipold]

Hi Eric,

Could you give a detailed protocol of your staining procedure? Especially, including any and all fix and perm steps and reagents. If you're using PFA, when did you initially open the glass ampule and when did you make the PFA dilution?

Also: when you see this increase in CD45-Pr141 signal, is it happening as a streak? In other words: you start the sample, there's no streak (just discrete Events), and over the course of the 2hr aliquot, the 141 channel gets progressively streakier?

Or is there no streak?


Mike

[EThrash]

Hi Juan, Hi Mike,
This is Emily at DFCI CIO lab. This is a sample we ran with Eric so I can fill in on protocol questions:

- It is present in other channels including: 144Nd, 154Nd, 164Nd, 166Di (slightly), 167Er (slightly). All other channels look fine.

- Protocol is 103Rh and CD45 staining (15min, 37deg water bath), then Surface Stain + Fc Block (45min ice), next sample barcoding (using maxpar Cell-ID 20 plex, fix I 10min RT, barocode incubation 30min RT), then we fix/perm (BD Cytofix/Cytoperm, 20min RT), then intracellular stain (30min ice), then fresh fix (2% FA made fresh from 16% + PBS, 10min RT), then DNA stain (191/193Ir diluted 1:5000 in fresh 2%PFA, overnight, 4 deg). Samples are acquired in CAS+beads and WB injector protocol. They are left at 4 deg until acquisition.

- The PFA dilution is made just prior to fresh fix.

Thanks for the input,
Emily

[EHaasDFCI]

Hi Mike,

Typically the samples run very clean. I can't remember off hand if there was any noticeable increase in signal from observing the rain plot. There very well could have been, that we didn't catch. This also affected several different samples, but all run in the same manner. We will see what we can do to minimize the time the samples are sitting in CAS prior to acquisition and see if that helps.

Thanks!
Eric

[AdeebR]

Wanted to add a note that like Juan and Eric, we're also see the greatest evidence of this in the 141Pr channel, but also it in other channels, particularly 145Nd and 146Nd. Relating to my original post, this seems to be most pronounced when there is a highly prevalent + high abundance marker in these channels, e.g. when we have CD3 in the 141Pr channel (I also note CD45 in Eric's case and CD7 in Juan's case, both of which are similarly high abundance markers).

I'm also starting to reject my earlier hypothesis that this relates to wash solution accumulation in the sample tubing. We did test the pH of the liquid that was accumulating in between the inner and outer layers of the sample and did find that it was very low, which does suggest that wash solution could be infiltrating this space. However, I've since assembled some sample lines where I've siliconed the inner and outer layers together, thereby eliminating the void space and ensuring good alignment of the tubing ends at the grounding nut, and yet we're still seeing evidence of this happening in some samples.

So I'm inclined to agree with both Eric and Juan that while exposure to wash solution may reproduce this phenomenon, it seems like there's something else causing the problem when we see it occur during routine acquisitions.

[dgeanon01]

Hi all,

We've successfully troubleshooted this artifact @ HIMC Mount Sinai, and I've attached our findings. This drift can be caused by residual acid in the CyTOF lines if you acquire cells immediately after washing, however it is not what has been driving the issues we've had (and many others have had). To sum up briefly, this signal drift issue results from poorly titrating CyTOF antibodies (overstaining) and not properly fixing cells prior to acquisition -- this leads to the shearing off of antibody/metal labeled polymer during acquisition (caused by pressure from PSI) w/ subsequent increase in background signal on negative populations. It sounds trivial, but titrating reagents can become difficult when staining and fixing large/massively barcoded cell pools (not always simple to titrate reagents properly when working with tens of millions of cells).

In sum, we've adopted 4%PFA fixation as our standard CyTOF fixation, and make sure to scale volumes appropriate (500uL/5M cells). Whenever this signal drift occurs, it's a sign that the CyTOF antibody is being used in excess and must be titrated down (even if the positive signal appears dim). The combination of these two changes has all but eliminated this artifact from our CyTOF experiments.

Best,
Daniel Geanon

[mleipold]

Hi Daniel,

Thanks for the valuable followup. I'd say your (much more detailed) experience is consistent with a few more informal observations I've made when I've used more PFA, or done a really tight titration (most often in the development of a new marker, rather than validation of a new prep of a previously-used marker).

One followup question I do have: I think it's relatively common to have a bit "extra" antibody around as a safety margin for when you encounter a poor-quality sample (dead cells and debris that "soak up" reagent and throw off the effective titer). Therefore, if you're titrating your antibodies very tightly, you wouldn't have this margin. Have you noticed any effect when you do run into a poor-quality sample?


Mike

[TomerWeizmann]

Having encountered similar phenomena (due to wash residues, clogs, high pressure of PSI, re-tuning, neb replacement etc.) we installed an easily changeable disposable 35u filter on the sample line (of course in addition to filtering via 35u before acquisition etc.), thus circumventing most of the subsequent clogs-related faults.
Please see attached.

We did not observe carryover (we replace two fresh DIW/CAS tubes between samples until ev/s = 0 and raindrop is clean), reduction in system performance or any changes in our SOPs.
We usually use wash/TS only on startup and shutdown, and replace the filter every day.

P.S. In the mentioned ‘signal drift’ issues: Pd barcodes, Ir, Cisplatin, beads signal etc. may not be affected whilst antibodies will, as they are not covalently bound, + we also recommend using 4% PFA when possible (filter before fixing,, and correct titration (I agree with Mike on that) barcoding, and to minimize sample time, aim for up to two hrs, in DIW at RT to reduce degradation, especially if not running in CAS). Also, make sure your suspension is kept homogenous (mix), to avoid acquisition of cells with different densities in a gradient.

Always verify markers (and EQbeads) MMI stability vs Time (within and between samples), and gate out (FlowAI and Cut may help) faulty acquisition regions if present. And please share your data.

Good luck!
Tomer

[mleipold]

Hi Tomer,

Thanks, that's great info!

Obviously, the more challenging a sample (tissue disaggregates, etc), the more likely there would be clogging material. Do you find that you need to replace the disposable 35um sample line filter very often during a run on an average day running PBMCs or WB, etc?

Have you noticed any background impurities when looking at the rain plot in TOF mode? I was just wondering whether the filters were "clean".


Mike