Quantification of nuclear import of pDNA

I’m using CellProfiler to quantify nuclear import of pDNA. Therefore, my cells are incubated with labeled pDNA, afterwards the nuclei are stained and images are taken with the confocal microscope. The images are divided into the green channel for the nuclei staining and the red channel for the pDNA.

Image analysis is done like that:

  • Images are loaded into CellProfiler and converted into grayscale images.
  • Nuclei are recognized in the green picture with the “Identify primary automatic”-module, are then shrunken to be sure to “examine” only nuclearly accumulated pDNA and not pDNA at the outside of the nuclei.
  • Then the intensity of the identified and shrunken objects are measured in the pDNA-picture
    What I analyse is the integrated intensity.

I’m now wondering if my image-analysis makes sense, respectively if it is correct, because I read in a paper that direct quantification of fluorescence intensity is not possible, they speak about saturation of fluorescence intensity. I think my images are not saturated.

As I’m the only one in my working group who is working with this program, I would be glad if you could help me.

Hi Annette,

It is most certainly possible to quantify fluorescence intensity provided that your images are not saturated; this could be what the paper was referring to. If your images are fine, then you’re good to go. Would you mind sharing what paper you read where it mentioned this?

As an example, if you have access, you can look at our paper in Genome Biology (in particular Fig. 1) where we go into some of the details of quantifying integrated nuclear intensity and some of the possible pitfalls. You can use imaging to measure cell ploidy using integrated intensity to obtain plots similar to what you would get with flow cytometry.

Regards,
-Mark

Thank you very much.
I already read the CellProfiler paper you mentioned, but I think I will have a closer look.
The paper I mentioned before has the PubMed ID: 15006612 - the paper is from 2004.

Thanks for passing the reference along. Yes, it seems that a worthwhile item to check is the relationship between intensity and area, as the authors of the Nature paper did. If it holds, then you have a surrogate for the values you want, whether the image is saturated or not.

Also keep in mind that the integrated intensity is merely the sum of the pixel intensities for each object, so (a) it is effectively a function of area as well, but (b) will be inaccurate if the image is saturated.

Cheers,
-Mark

Hi!

I checked the intensity of my images with ImageJ, and it analyses intenities >255 for my 8bit images. I’m quite unhappy because I spent lots of time at the confocal microscope to adjust everything the way that no pixels were saturated (using the RangeIndicator). But if you have a look at the images later on, they really seem oversaturated, so I’m sure I can’t use the images for quantification.
How can I measure the intensities of the objects in the images with CellProfiler? The images I want to analyse consist of very very small objects (labeled plasmid DNA inside of cells). I tried the module “Measure Image Intensity” but I’m afraid it calculates the sum of all pixel intensities for the whole image?
I attached an examples image.

Do you think it is necessary to apply an illumination correction? I read in the paper you gave me that this is suggested when quantification should be done.

Thanks for your help,
Yours,
Annette

It could be that part of the problem is that you are saving images as 8-bit and thereby either saturating at >255 since the actual intensities are greater (really bad) or compressing the dynamic range of your pixel intensities more than you should (somewhat bad). Check your software and see if you have the option of saving images as 16-bit; this may solve your problem.

As an aside, I notice that the image you uploaded was color (red) when it should be grayscale, assuming that this is a recording from a single wavelength. You may also want to check whether you can create grayscale images rather merged/color images.

[quote=“Annette”]How can I measure the intensities of the objects in the images with CellProfiler? The images I want to analyse consist of very very small objects (labeled plasmid DNA inside of cells). I tried the module “Measure Image Intensity” but I’m afraid it calculates the sum of all pixel intensities for the whole image?
I attached an examples image.[/quote]

I’m attaching a pipeline that would do would you want. Note that I needed to add a ColorToGray module, which could be omitted if you can get grayscale images as noted above. Basically, it identifies the spots and measures the intensity of each spot. You can then export these measurements but I haven’t added these modules since I don’t know what want exactly, e.g., do you want a per-image average intensity, a per-cell intensity, etc.

It doesn’t seem like it in this case. Your images don’t seem to have any obvious illumination issues that need correcting.

Regards,
-Mark
2010_09_24.cp (4.02 KB)