FAQ  •  Register  •  Login

Inconsistent cell counts and distribution during acquisition

Forum rules
Please be polite and civil. We know that troubleshooting is vexing...
<<

TomerWeizmann

Contributor

Posts: 43

Joined: Mon Apr 07, 2014 11:58 am

Location: Weizmann Institute of Science - Israel

Post Mon Oct 20, 2014 10:26 am

Inconsistent cell counts and distribution during acquisition

Dear friends,

When following run time vs event length (or other parameter), often it can be observed that as the run progresses the number of events acquired (either cells or beads) drops. Moreover, the cells are not distributed evenly – large ones focus at the beginning of the run and gradually “lost”.

Attached please find images for illustration:

https://drive.google.com/folderview?id= ... sp=sharing

This phenomenon is more pronounced when using a 3ml loop for example, but very evident with the 0.5ml loop or with the autosampler, regardless of re-suspension and agitation prior to injection into the loop.

Accordingly, sample acquisition is not homogenous (inconsistent) and the results may be biased (heavily depending on the duration the sample has been left in the loop and the critical time of acquisition start, delay, and the washed out leftover at the end).

I think this happens due to gravimetric settling of the cells which sinks to the leading front of the loop.

I would appreciate your insight regarding this matter and its possible effects on the data acquired.

Sincerely,

Dr. Tomer Meir Salame
Flow Cytometry Unit
Biological Services Department
Weizmann Institute of Science
Wolfson building, basement 1, room 21
P.O. Box 26, Rehovot 7610001, Israel
Tel: ++972-8-934-2721 / 2235
Mobile: ++972-57-5489307
Fax: ++972-8-934-4194
E-mail: tomer-meir.salame@weizmann.ac.il
http://wws.weizmann.ac.il/Biological_Se ... etry-about
Dr. Tomer-Meir Salame
Head, Mass Cytometry Unit
Life Sciences Core Facilities
Weizmann Institute of Science
E-mail: tomer-meir.salame@weizmann.ac.il
https://www.weizmann.ac.il/LS_CoreFacil ... etry/about
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Tue Oct 21, 2014 4:02 pm

Re: Inconsistent cell counts and distribution during acquisi

Hi Tomer,

I have noticed the same thing and have assumed it was due to progressive mixing and dilution of the sample by the carrier solution at the trailing edge of the injected sample bolus.

One thing to note regarding your observation about cell "size". It is my understanding that "event length" does not actually correlate to cell size. Every particle results in an ion cloud of approximately the same size (which is more a factor of gas expansion kinetics). Longer event length instead signifies poor separation between ion clouds/multiple ion clouds merging together, which is more a function of staining intensity and acquisition rate. So the reason that event length drops over time is because the event acquisition rate drops over time so the number of merging ion clouds/doublets drops.
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Oct 21, 2014 4:46 pm

Re: Inconsistent cell counts and distribution during acquisi

Hi Tomer,

Adeeb answered before I could. :)

Yes, each injection starts out as a square bolus of sample. However, due to diffusional broadening and interactions with the walls of the tubing, the bolus tails off over time. So, you get a square start to each injection, which then trails off toward the end.


The cell length parameter is *not* really a good way to separate out big cells (granulocytes, etc) from small cells (platelets, monocytes, etc). The size of the ion cloud after the plasma is more a function of total metal content (and therefore how much the cloud expands due to like charge repulsion once all the chelators and other counter charges are destroyed) than the size of the initial starting cell. Since Ir intercalator signal is often the brightest signal in your staining, that also somewhat normalizes "cell length" or "event length" for initial cell size.

Longer cell length has more to do with resolution between cells. When the ion clouds overlap (or, minimally, aren't resolved well) because the sample is very concentrated, the likelihood that two cells will be counted as one cell increases. You can test this yourself: Take a standard sample, take half and inject it. Take the other half, dilute it 3-5-fold, and run that diluted sample. The diluted sample should have barely any events above 50 length units or so, because it would have few if any doublets.



I have added a PDF of some of my data. In this case, we did two serial injections in the same file acquisition. Each was ~600 sec long. You can see two squarish starts, with trailing off toward the end of each injection.

I made several gates, corresponding to hi vs lo cell length (CL), at the start and end of each of the two injections (total of 8 gates). I then gated out dead cells, and then did CD3 vs CD20 to look at B-T doublets. As you can see, CL-hi always has more B-T doublets than CL-lo. Also, gates from the beginning of each injection (more concentrated part) always have more B-T doublets than gates from the end of each injection (when the sample is more dilute).


Mike
Attachments
Cell length-doublets.pdf
(824.87 KiB) Downloaded 411 times
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Tue Oct 21, 2014 8:57 pm

Re: Inconsistent cell counts and distribution during acquisi

So I continued to think about Tomer's question and did a little analysis to address this issue. I've unfortunately found some concerning results. Please see attached PDF.

I used data acquired from a whole blood sample that was introduced over the course of 4 sample injections (2 in loop 1 and 2 in loop 2). The phenomenon was very consistent across both pairs of injections.

I gated on 4 time segments during acquisition and looked at the relative ratios of T cells, monocytes and granulocytes (populations across a range of cell sizes).

First off, as we've all noted each injection started off more concentrated and then tapered off over time but this effect was much more pronounced in loop 2 than in loop 1. It also appears that this has significant effects on the data:

In loop 1 the relative ratio of granulocytes to T cells is fairly consistent over time but there is a a trend towards slight over-representation of T cells towards the end of the run. In the case of loop 2, the ratios start off comparably to loop 1, but they diverge dramatically over time. The relative frequency of T cells went from 15% - 45%, while the frequency of neutrophils dropped from 70% - 26%. In contrast, the ratio of CD4 to CD8 (comparably sized cells) remained fairly consistent over the course of the run in both loops.

This results do suggest that cell type specific transmission is quite inconsistent over time, particularly with loop 2. I have suspected for a while that the two loops behave quite differently and this confirms it.

Could others confirm whether they also see this with their CyTOF2?
Attachments
Ratio_over_time.pdf
(182.67 KiB) Downloaded 489 times
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

mleipold

Guru

Posts: 5792

Joined: Fri Nov 01, 2013 5:30 pm

Location: Stanford HIMC, CA, USA

Post Tue Oct 21, 2014 9:25 pm

Re: Inconsistent cell counts and distribution during acquisi

Hi Adeeb,

Do you find that the ratio re-sets itself at the beginning of each loop?

In other words, does Injection 3 (#2 on Loop 1) look like Injection 1 (#1 on Loop 1),including restoring the improved ratio as compared to Injection 2 (#1 on Loop 2)?


Or does the ratio continue to get worse?


Mike
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Wed Oct 22, 2014 4:13 pm

Re: Inconsistent cell counts and distribution during acquisi

Hi Mike,

The ratio resets itself at the beginning of the loop. I've attached a table of the cell frequencies over time for all 4 loop runs.

Adeeb
Attachments
Ratios over time_table.pdf
(103.67 KiB) Downloaded 397 times
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

emallen

Participant

Posts: 18

Joined: Thu Oct 23, 2014 6:10 pm

Post Wed Oct 29, 2014 2:47 pm

Re: Inconsistent cell counts and distribution during acquisi

Hello Adeeb, in your experiment which sample loops correspond to Loop 1 and Loop 2? I apologize if this information was posted elsewhere, I couldn't find it.

Elyse
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Tue Nov 04, 2014 3:16 pm

Re: Inconsistent cell counts and distribution during acquisi

Hi Elyse,

"Loop 1" is the loop that runs from position 8 to position 4 on the 8 port valve. "Loop 2" runs from 6 to 2.
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

AdeebR

Grand master

Posts: 169

Joined: Thu Mar 13, 2014 5:58 pm

Location: NYC

Post Fri Jul 24, 2015 11:25 pm

Re: Inconsistent cell counts and distribution during acquisi

Hello CyTOFers,

I thought I'd update this thread with some interesting information that came up during discussions at CYTO. Fluidigm has also done some in house testing on this and confirmed the discrepancies between the two loops. The reason for the drift over time is likely due to differential settling of cells in the loops (larger cells settle faster than smaller cells), and this problem is worse in in the case of loop 2 because it is horizontally oriented, as opposed to loop 1, which is vertically oriented.

One obvious workaround is to only use loop 1 (or whichever loop is vertically oriented on your system) for acquisition.

Another helpful piece of information is that when running samples on loop 1, if you allow a little air to enter the loop before injected your sample and then inject slightly less than a 500uL sample volume (I've been doing ~450uL) you will end up introducing a small air bubble that separates the sample from the carrier fluid. This prevents the mixing and progressive dilution of the sample with H20 at the trailing end of the sample, maintaining a more consistent event rate over time and also helping to reduce the cell settling issue. This approach only really works for loop 1; since loop 2 runs in the opposite orientation your air bubble would end up being at the front of your sample which wouldn't do much good.

Hope that's useful to folks,

Adeeb
Adeeb Rahman
Icahn School of Medicine at Mount Sinai, NYC
<<

ChristophS

Contributor

Posts: 21

Joined: Thu May 07, 2015 2:11 pm

Location: Switzerland

Post Tue Jul 28, 2015 8:18 am

Re: Inconsistent cell counts and distribution during acquisi

Hi everyone,

to add on to Adeeb's post about using an air bubble on loop 1:
I have done that now routinly for small samples e.g. 150ul and used a slightly different procedure in order to control the bubble size.
Since loop 1 simply drains into the waste by gravity if loop 2 is active and your remove the syringe, I have the sample in the syringe about 1mm retracted (so 2-3ul of air) and attach it while loop 1 is active.
Only then I switch manually to loop 2 for the injection into loop 1. Like this I get a nicely defined bubble in an otherwise full loop.

Best

Christoph
Christoph Schwärzler
Director Cytometry
Flow Cytometry Facility (https://www.cytometry.uzh.ch/en.html)
University Zürich
Next

Return to CyTOF troubleshooting

Who is online

Users browsing this forum: No registered users and 9 guests