Posted by
mmodel on
URL: http://confocal-microscopy-list.275.s1.nabble.com/detecting-apoptosis-in-one-cell-type-in-coculture-tp7584201p7584217.html
I only would like to add to that that apoptotic cells may not pick up some of live/dead stains, such as PI, because their membranes remain more or less intact for some time.
Mike Model
-----Original Message-----
From: Confocal Microscopy List [mailto:
[hidden email]] On Behalf Of Alfred Bahnson
Sent: Friday, September 11, 2015 9:48 AM
To:
[hidden email]
Subject: Re: detecting apoptosis in one cell type in coculture
*****
To join, leave or search the confocal microscopy listserv, go to:
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopyPost images on
http://www.imgur.com and include the link in your posting.
*****
Leoncio,
I would encourage you to pursue an approach based on time-lapse imaging and automated image analysis to detect the transient episodes of death of individual cells among your adherent target cell population over time. A time-lapse approach is important because in many cases, the adherent cells phagocytose the killed cells shortly after death. You will see this in the "videos." End-point assays only tell part of the story; they miss the journey that your culture took to get there.
You can readily detect cell death by uptake of propidium iodide (PI) in your TRITC channel. The excitation wavelength is less photo-toxic than with the FITC and DAPI channels, and exposures can be minimized in time and intensity because the PI signal is strong.
I would recommend setting up grids where you acquire sets of adjacent viewfields that can be stitched together after acquisition so that you have sufficient cell numbers for good statistics. Run duplicate wells and duplicate grids within wells to estimate precision. Working in multi-well plates with 15 minute scan intervals, you will have time to include a lot of viewfields.
Development of the automated image analysis method will be a challenge requiring the skills of at least one expert or dedicated person willing to become an expert. I don't think there is yet any assembled package that can be plugged in for this purpose (I'd like to hear otherwise!), but there are many places to start: ImageJ (FIJI), BioImageXD, and CellProfiler are free and open-source platforms that encourage users to share in development. I think I've said that correctly, more or less. We have used these programs successfully for various projects without ever altering the "source" code.
An attractive feature of the automated image analysis approach is that the method can evolve to become more efficient and more automated over time, and as improvements are made they can be tested on past image sets, effectively improving previous data (no harm in that, right?). So right off the bat it is important to emphasize and seek consistent quality in the image set: focus, exposure, contrast adjustment for bright-field or phase.
Using a small z-stack can be helpful. Phase is a problem in multi-well plates. I am mentioning non-fluorescent imaging parameters because your analysis method will probably need to incorporate some logical approach for discerning T-cells from target cells, and you will likely need to put together features from either phase-contrast or bright field images to help will this, in addition to the PI signal from your fluorescent images. You will also probably want to count live cell numbers in each viewfield for purposes of having a denominator upon which to quantify death rates.
However, it is possible that the size of your dying cells alone, based on fluorescence, could distinguish between T-cell death and target cell death in many cases.
A lot depends on where you are now in terms of familiarity with methods for segmentation (FIJI's "Trainable Weka Segmentation is very cool:
http://fiji.sc/Trainable_Weka_Segmentation) and putting together steps in an automated method for outputting counts from large image sets. Also, when it comes to optimizing non-fluorescent imaging of cells for automated segmentation (for live-cell counts and possibly for distinguishing target cells and T-cells), it's not easy. I would like to see a table of the many approaches to this goal, the pros and cons of each method. Unfortunately, each different type of target cell may require a somewhat different solution. Don't expect perfection. I think attention should be paid to statistical quality control (based on duplicate wells and duplicate grids within wells), as well as manual verification of segmentation in intermediate steps and manual verification of output numbers in randomly selected image sets. So the automated analysis method needs to incorporate manual verification at various stages for quality control / method validation.
Of course, if suitable fluorescent probes can be incorporated into this approach, all the better!
Al Bahnson, Partner
Kairos Instruments, LLC
Pittsburgh PA
412-735-9983
On Thu, Sep 10, 2015 at 11:31 AM, Leoncio Vergara <
[hidden email]>
wrote:
> *****
> To join, leave or search the confocal microscopy listserv, go to:
>
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> Post images on
http://www.imgur.com and include the link in your posting.
> *****
>
> Thanks for the responses so far. I am getting some interesting leads
> to follow but I am realizing the solution may not be straightforward.
> I have quite a bit of experience in several forms of optical
> microscopy but I am new apoptosis. I am quite excited with my
> involvement in image based high throughput screening for cancer research.
>
> We are trying to develop an image based assay moving from flow
> cytometry to imaging. Our first approach was to modify the Flow assay
> already in place for use in imaging and it was then that we found this new set of problems.
> Our goal is to setup an assay on fixed cells in large format multiwell
> plates. In the process we plan to do live cell imaging experiments our
> final goal is to setup a drug screening protocol.
>
> I was hoping to find a marker ( or combination of markers) to quantify
> cell toxicity which can be loaded into cells and then washed before
> adding T cells to the cultures. Non wash assays like CellEvent and
> Annexin V have the problem of interference from the Tcells. An option
> is to label the Tcells to separate them on analysis, but at high
> seeding ratios the Tcells cover most of the well area and at the low
> magnifications (and hence low
> NA) used for screening we don't have enough z-discrimination to
> separate the Tcells on top from the tumor cells underneath. We have 4
> channels available, the classical DAPI, FITC, TRITC and Cy5
> combination (405,
> 488,561 and 640 excitations).
>
> One of the goals is to work with multiple cell lines (more than a
> dozen) so any method based on expressing FP based indicators would
> make the assay too complicated.
>
> Leoncio
>
>
>
> On Thu, Sep 10, 2015 at 2:41 AM, Markus Rehm <
[hidden email]> wrote:
>
> > *****
> > To join, leave or search the confocal microscopy listserv, go to:
> >
http://lists.umn.edu/cgi-bin/wa?A0=confocalmicroscopy> > Post images on
http://www.imgur.com and include the link in your
> posting.
> > *****
> >
> > Dear Leoncio,
> >
> > Could you expand a bit on what your requirements for the experiment are?
> > Do you need time-lapse information of apoptosis kinetics and cell
> tracking
> > or are end point read outs sufficient? Depending on your needs, the
> > approach taken may be quite different.
> >
> > Best wishes
> > Markus
> >
> > Dr. rer. nat. Markus Rehm
> > Biomedical Research Lecturer
> > Dept. of Physiology & Medical Physics & Centre for Systems Medicine
> > Royal College of Surgeons in Ireland RCSI York House York Street
> > Dublin 2 Ireland
> >
> > phone: 00353 (0)1 4028563
> > email:
[hidden email]
> >
https://research1.rcsi.ie/pi/mrehm/> >
http://www.systemsmedicineireland.ie/> >
>