There were some requests for me to post my responses regarding the use of
Modfit (to stimulate discussion, as I am sure it will. here they are:
1. If all you need to do is estimate the apoptotic fraction, then just
use CELLQuest and a dot plot region stats (DNA vs side scatter). I use
MulticycleAV (DOS version on a Pentium) for cell cycle and apoptosis
testing. It fits the debris AND apoptotic peak. The demo of Modfit was WAY
to slow on my Quadra 650 for even simple cellcycle data. I can send you
jpeg dot plots if you want to see what they look like.
2. Unless you absolutely need some means of determining statistics or
percentages of apoptotic cells in your population I would avoid using
Modfit. I used it for a while and quickly realized that for sample to
sample apoptosis assessment it was not necessary and very frustrating.
However, the program defaults to looking for aneuploid cells from a
clinical sample. When you are setting up your sample turn off the
aneuploid selector and click on Fresh/Frozen sample. Unless you know the
sample has a high apoptotic population, if you click on Apoptosis detector
the program, (in the automatic mode) will look for the first major peak.
Often that will be the remaining G1/G0. You have to set your markers for
each sample. Unfortunately for late stage apoptosis it wants to read it as
strictly debris. Unfortunately if your main goal is apoptosis detection
from cell lines you really have to know what your sample will look like
before you run ModFit, and make the custom adjustments to fit.
What I have to say now I will assume you don't have a lot of experience
with apoptosis detection by flow, so if you do.... my apologies. As a
better means of sample to sample data set up your histograms just like the
Modfit would acquire the data FL-2W vs. FL-2A (Dot Plot) and FL-2A
histogram. Now adjust your FL-2 histogram to a fluorescence signal (G1/G0)
peak is around 300. The apoptotic population will be at <200. If you have
a good +control your peak will be a tight peak at 180-200. As a sensitive
comparison, albeit not widely published point of detection, set up a FL-2
Height histogram right next to your FL-2 A histogram. What you will see is
the whole G1/G0>S>G2M peaks are crammed on the far left of the histogram.
The apoptotic population will skew off just to the left. Small debris will
continue to skew off to the left. I found for the best info on apoptosis
Z. Darzynkiewicz (I hope I haven't butchered his name) is the best.
Good Luck
3. Susan:
Modfit was design with clinical samples in mind. It was an impossible task
to use it as an evaluator for apoptosis until the version 2.0 came out
having the "synchronization wizard" as an option. This option is under
analysis in your menu bar and allows you to mark your G1, and G2 peaks and
also allows to have
alternative selections for the S phase distribution, in each individual
sample, but again your apoptotic population will appeard under the debris
alternative.
This is how you do it: you just need to select your sample, then go to the
syncronization wizard, define your peaks and S alternative, check the
debris to be shown and display your results, that's all, and forget about
any other manipulations except to gate them at your double discriminator
making your gate large enough from apoptotic to polyploid area. In samples
coming from tissue culture and gated properly the debris number is really
your apoptotic population defined as subG1 or VERY LATE apoptosis. (Still
there is a very possible misinterpretation, since necrotic populations can
also be found in this area).
Now, as a personal opinion I don't like to define apoptosis by subG1
population since in tissue culture and in conditions induced by
experimental drugs or other stress inducers, a big number of already
apoptotic cells coming from G2/M and S phases are displaced to S and G1
respectively, and not yet present as
sub G1. Still if an investigator want to have a quick look by using just PI
to decide if apoptosis is an alternative he or she want to pursue further,
I have found that batch analysis using cell quest with appropriate markers,
gives reliable information , is faster to obtain and I believe it is even
better than Modfit because using the algoritms to better define the overall
distribution of cells at different phase of the cycle may be largely
misinterpreted as arrest, when they are already apoptotic cells transiently
passing by. Feel free to ask me any more questions if you need any more
details.
GOOD LUCK.
4. Susan,
we do some apoptosis work under the same conditions you use. The fact that
you see the apoptotic population sometimes in the S phase, sometimes in G1
or on the debris side is normal. What happens is the following. As the
cells progress into apoptosis they cleave the DNA into smaller and smaller
pieces. At the beginning only few small DNA fragments (large fragments stay
in the nucleus) got washed out of the cells when permeabilized for PI
staining. The channels on the histogram - that represent a certain
brightness which is proportional to the DNA content of the cell - migrate
toward the left end of the graph because cells lose DNA (therefore bind
less PI therefore are dimmer). If this process start in a G2 or late S
phase cell it will show up the left (toward the G1 peak) and if you would
analyze the same cell later it would be first in G1 then in the debris
area. The later you analyze the apoptotic cells the more you see in the
debris phase. Earlier the G2 peak goes flatter, the S phase gets higher and
skewed toward the left. G1 increases and debris increases with the time.
This is our observation on cell lines.
We also use log scale histogram to show the debris part better. Of course
you can not use the log scale for cell cycle assessment. Make sure you
exclude the doublets, the extremely small debris near the origo and use the
same method for permeabilization. That is either ethanol (preferably
overnight) or ethanol/detergent. Detergent removes the cell from the
nucleus and you analyze only the nuclei for DNA content. The DNA content
will look lower in that case (you lose more DNA fragments). Check the
cells out in a fluorescent microscope or if you have one even better in a
confocal microscope. You'll see what I was talking about. Please, let me
know if this was useful or I can be of further assistance to you.
5. Hi, Susan,
We do a lot of cell cycle analysis, including apoptosis assessment,
with Modfit in cell lines. I wish I could tell you that there is an easy
fix. Unfortunately, with a single color and an anueploid population
(usually a tetraploid that overlaps the G2/M peak--endoreduplication in
action), you cannot make measurements that work. I have spent a number of
hours with Verity both in person and on the phone to try and resolve this
very issue. This type of analysis asks too much with just one dye. Also
this particular measurement (hypodiploid peaks) is very inaccurate. When a
cell dies, by whatever mechanism, it fragments the DNA--from 1-thousands of
fragments. You can only say with some surety that the cells form
hypodiploid peaks. Omerod and others have documented this in several
places. See Ormerod, et al. Acta Oncologica, Vol. 32, #4, 417-424, 1993
and Darzynkiewicz, et al. in Human Cell, Vol. 11, #1, 3-11. You can
separate the diploid from the tetraploid with cyclin measurements (see
Current Protocols in Cytometry), but apoptosis in these populations might
be better measured with TDT or annexin V assays. I have managed to
confuse most of our scientists by reminding them that the number of cells
that any one measurement might make is less important than how and if cells
die. We start out by using some vital dye to determine if the cells die
and under what conditions. Then we look for the typical apoptotic
morphology under the scope. From there we might proceed with a TDT,
annexin V measurement, or caspase 3 assessment. We have also done
measurements of reactive oxygen species, or mitochondrial degradation, as
well as cell cycle determination with propidium iodide. We try to find the
best assessment(s) for the cell and treatment involved. Sorry for the bad
news.
6. We had the same problem a few years ago with unreliable/
questionable placement of the G0/G1 peak and sub-G0 (apoptosis?). We now
use a more reliable assay to measure apoptosis - an Apo-BrdU FITC kit from
Pharmingen (cat #6576 kk). It allows you to identify apoptotic cells by
staining with FITC-labeled anti-BrdU Ab. You can simultaneously stain with
PI to also measure cell cycle. This allows you to identify what portion of
the cell cycle is apoptotic, even if the cells are anueploid in nature. We
find the assay very useful for our study of mesothelioma cells, treated
with various chemotheraputic drugs, that create notorious amounts of
debris, anueploid peaks and apoptotic cells.
Hope this helps.
7. >The manual is not helpful.
Truer words were never spoken!
8. Dear mrs. Zunino,
I would like to draw your attention to annexin-V-Alexa 594 for measuring
the apoptotic (intact) cell population. You do not need to fixate (and
conseqently have less debris) and measure in the same channel as PI thus
you can still use you other markers. For further info you may contact our
website at : www.nexins.com
We have world wide patents on this application and work closely with dr.
Chris Reutelingsperger, University of Maastricht who is doing a lot of
research for NeXins (e-mail: c.reutelingsperger@bioch.unimaas.nl).
8. Hello, Dr. Zunino,
You have identified the prime difficulty in trying to identify Apoptotic
events in a single-parameter histogram. In large measure, Apoptotic events
are indistinguishable from debris using PI staining alone. Thus, secondary
markers of Apoptosis such as Annexin V are employed for 2-parameter
differentiation of "true" Apoptosis. ModFit LT uses the historical
assumption of a Apoptotic events appearing as a more-or-less Gaussian
distribution sub-G0G1, which is of course in the debris range also. In the
absence of a defined distribution over which the Apop range can be placed,
the Apop range "competes" with the debris range for events. This allows
subtle differences in the placement of the Apop
range and the debris (E1) range to effect major changes in the relative
results. Since attempts at gating out debris produce their own set of
assumptions and problems, assessing Apoptosis in single-parameter PI
distributions remains problematic for all DNA analysis software.
Please let me know if you may like a further discussion on this topic.
Best regards,
Mark
Mark E. Munson
Sales Manager
Verity Software House, Inc.
Phone: 207-729-6767
Fax: 207-729-5443
verity@vsh.com
http://www.vsh.com
Dr. Susan J. Zunino
Lehrstuhl fur Genetik
University of Erlangen
Staudtstr. 5
91058 Erlangen
Germany
Tele. 49 9131 8528784
FAX 49 9131 8528526
e-mail: szunino@biologie.uni-erlangen.de
This archive was generated by hypermail 2b29 : Wed Apr 03 2002 - 11:53:55 EST