Sorting kills cells?

Joanne Lannigan jl7fj at
Thu Jun 17 16:46:11 EST 2004

Hi Ray:
There are a number of things to consider, as you mention, one of them being
the health of the cells prior to entering the sorter world. I find that FSC
and SSC plots are not enough to discriminate compromised cells, as can be
easily demonstrated by using a viability dye in combination with your
scatter plots and you will see PI positive cells within your "live" gate. So
even though your cells appear in your "live" scatter gate presort, they may
already be somewhat compromised and by time they come out the other end (and
sit) they may have been pushed over the edge and may have altered light
scatter properties. If you use a viability dye and only sort cells whose
membranes are intact you have a better chance of retrieving a healthier
population in the end.
Another thing to consider is your threshold. If you set your threshold to
exclude cellular debris and small stuff then the instrument is blind to
these events (even though they are still in your sample). If the instrument
doesn't see them then it cannot make a sort decision to exclude them. I try
and keep my threshold low enough to just exclude electronic noise and use my
light scatter gate to exclude unwanted cellular debris. a little slower but
definitely cleaner.
Another trick is to use FSC width as a sort parameter to exclude doublets,
not only whole cell doublets but also aggregates of cellular debris which
can sometimes be large enough to fall in your "live" scatter gate and do not
stain with PI because there is no DNA. These appear as low FSC width (I know
this because I sorted the low width population to see what it was!). You
will also find using GFP vs. FSC-width as a sort criteria really improves
purity as it will get rid of any negative cells that may have been stuck to
a GFP positive cell when it went through the laser and hence was sorted as a
positive but when it sits in your collection vessels it "unsticks" itself
and now becomes an individual negative event.
I often see what you are seeing when customers insist on not using a
viability dye in their preps and I am sorting on scatter and gfp only. In
those cases I calculate my purity on the "live" gate only as a previously
gfp positive cell that has a leaky membrane (and now falls outside of the
"live" gate) will most likely have lost its gfp.
Just a few thoughts, hope it helps.

Joanne Lannigan, MS
Director, Flow Cytometry Core Facility
University of Virginia
P.O. Box 800734
Charlottesville, VA 22908-0734
Office: 434-924-0274
Lab: 434-243-2695
Fax: 434-982-1071
email: joannelannigan at

> From: Ray Hester <rhester at>
> Date: Mon, 14 Jun 2004 13:50:24 -0600
> To: Cytometry Mailing List <cytometry at>
> Subject: Sorting kills cells?
> Hi,
> We recently sorted GFP-positive cells for an investigator and
> re-analysis showed what appeared to be, by light scatter, a substantial
> number (8 %) of dead cells among the sorted cells.
> My question, is it possible that mechanics of the sort (this was a
> relatively slow sort - 1,000 cells/sec - through a 70 u nozzle tip)
> could result in this number of dead cells that would be displayed as
> dead cells by light scatter  after being sorted  (re-analysis was done
> immediately after the sort)?	We have a FACSVantage SE and use 200 mW of
> 488 light to excite the GFP.
> In the accompanying attachment, the upper dot plot (061004.003)
> represents the unsorted sample, R1 = the sort gate, and R2, what I would
> consider viable cells by light scatter.  Cells to the left of the R2
> gate, I would usually consider non-viable.
> The lower dot plot represents re-analysis of the sorted cells. If one
> could consider only the cells within the R2 gate, the sort purity would
> approach 97% (8903/9147).  However, if you consider all of the cells in
> the dot plot, the sort purity is, at best, 89% (8903/10,000) - and
> perhaps less, since there are additional events off scale to the left.,
> and some of these could be cells.
> Obviously there are a number of factors to consider, e.g., cell type,
> condition of cells prior to being sorted, etc. The investigator reports
> the sorted cells, when re-plated, are growing well so it doesn't seem as
> if the dead cells (again, by light scatter) seen in the
> analysis-after-sorting dot plot are indicative of more dead cells to
> follow at some time post- sort.
> Nevertheless, I wondered if this type of result  (5 to 10% post-sort
> dead cells) was a common occurence.
> Thanks for any thoughts.
> Ray Hester
> Univ. of South Alabama
> rhester at
> This attachment - 'GFPSort 001.jpg' -  163.75 KBytes - can be viewed at
> pg 

More information about the Cytometry mailing list