Localization of a certain protein in the cytoplasm vs nucleus via ImageJ, Fiji

I have 3 cell lines which has possibly different localization profile of a certain protein.
I have shot pictures of the cells (from paraffinized samples) stained with DAPI (for nucleus), the Protein of interest and another protein to discern the cell body( ie the cytoplasm and the DAPI etc).
Please can you help me to figure out a way to use the Fiji and Image J to calculate the localization pattern objectively. Basically, I want to have the system in such detail that when I have a 2nd protein of interest to examine, I can just use this template with minor revision and accomplish it.
Thanks in advance.


Let’s break this down a bit…

Can you share those original image files with us here? That will give us a better idea of the data you are working with…

What do you mean by this exactly? In regards to “localization pattern”… How do you want to define that? For example… you might be able to:

  1. Segment your cells using the cell body stain (this depends on your images - whether you can delineate individual cells or not) <side note: do you even want to segment individual cells? or can you look at the entire population as a whole?> and
  2. detect your protein of interest within the cell body (using the above segmentation as a mask) and measure the areas…

In general - might be worth it to you to check out the Principles page of the Image J wiki. That will give you some tips/tricks for quantitative image analysis.

A lot depends on your images. There is only so much - and certain types of - information that you can ‘pull out’ of an image… most everything depends on the sample-prep and acquisition side of things. So seeing your images will help.

For some other helpful links in getting started with ImageJ in particular:

1 Like

what is the best way to upload the images that I shot

By localilazation I mean to ask, where is the protein, how much of it is in the cytoplasm vs how much in the nucleus.


When you are writing out a message here… you can just upload the files using this icon:


Or share via a link to a file sharing site - such as Dropbox.


Ok. First - we need to see your images.

Also - just think a bit more specifically on what you could measure that would answer your biological question.

  • Is this a simple ‘yes’ or ‘no’ presence of your protein in the cell bodies comparing various conditions?
  • Do you need to have individual cell bodies delineated and can you do that given your images?
  • If you can use the population instead… do you also want to take into account an estimate of cell # using nuclei count? If so - are you able to count the nuclei easily?


Be as specific as you can when formulating what you want to measure and why/how.


This is one set of picture of 1 particular cell line, E-R is the name of the cell.

The different pictures are DAPI for nucleus, AR 488 for the protein of interest, panCyto for the cell body including nucleus, merge is when all the DAPI< pan Cyto and AR is merged and what I want to accomplish is ascertain quantitatively how much AR is present
in the nuclues vs in the cytoplasm. The pictures were shot in a Nikon microscope and is set as ND2 file, requiring a NIS element to open it.

This is just 1 example, I have approximately 5 pictures from each of the 3 lines each one complete with DAPI, 488, merged and pan cyto.

If this works, meaning you can see them and realize what I am trying to do here, I will dropbox all of them, they are named and labelled well.

I need you help badly!



All the images here



Now that you have the images, I am responding to your earlier question.
From literature survey AR is a cytoplasmic protein which needs to be in the nucleus to do its job, good and bad. I am stuck with the bad cells, which means it has to localize to the nucleus to do the bad job of keeping cancer cells alive. Now, there are drugs to kill it, but Cancer cells grow resistant. When the cancer cells are resistant, do I see more AR in the nuclues? Thats the question I need to find answer of , compared to the WT , what happens in the B-R and E-R cells, the two lines I derived in the lab, where R = resistant. It is very helpful for a tad-bit-technophobe- me to have a very detailed protocol when AI is involved as with the Weka training in ImageJ
Thanks again.

I will take a look at your images and try to piece together some guided help for you tomorrow… but you need to know we are not here to build your workflows from scratch. We can assist you along the way - but you have to the majority of the heavy-lifting. Saying that - go through those links that I posted above in my first reply - at the bottom. That is a great place to start for getting into such things in ImageJ.

Tomorrow - I will try to look at your pics and come up with a rough skeleton for a workflow … at least one option.

Whatever way you can help will be the best help.

I have some interesting cell lines that i derived in the lab, and a rough quantification localization of the two major Proteins in these three cell lines itself can be a publication material for a short paper. We can definitely think about an authorship
for you if you are willing to put substantial time, for this particular AR, protein, i have another one waiting to be stained, will take pictures on friday the 18th.

First of all… I cannot collaborate without going through my boss. So I will not be taking you up on that offer - perhaps someone else will here? I will try to help once you are further along and if you have more questions in working on your analysis workflow… but you have to start by teaching yourself a bit more and go through those links I provided.

Too - just looked at your images… I’m not sure you will be able to distinguish areas of your cells that are the cell body that exclude the nucleus. Looking at your DAPI staining versus the panCyto channel - they seem to almost completely overlap. Do these cells have particularly small bodies - is that normal?

To highlight this issue - I did a quick Segmentation (using the tools in the links above) of each channel (applying auto-threshold ‘Triangle’), converted both to binary masks, and then XOR the two (using ‘Image Calculator’) to get the cell body regions… and you can see the cell body region is minuscule relative to the nucleus:

But this would be a quick-and-dirty start on how to segment the ‘cell body regions’ at least - to then measure your protein of interest within… but you’ll need to take the total areas (cell body versus nuclei) into account because of the huge differences.

1 Like

Thanks. I will put more effort in understanding the links you sent me, pls bear with me.

I agree the cells not being normal.

Total cell body needs to be taken into account is correct statement.

AR is distributed over total cell, more so in the nucleus.

What is the difference between Wt vs BR, and ER is the study question.

Took a quick look and those cells look small enough to be some sort of lymphocyte (or similar), meaning you would need very high resolution images to get decent cytoplasmic measurements, and most likely a better nuclear marker than DAPI (Hoechst at least, better to have an antibody marker), which doesn’t look like it is staining only the nucleus here. In addition, the cytoplasm is spread out both above and below the nucleus, meaning that for any cell where your depth of focus is not entirely within the nucleus, you will pick up background from cytoplasmic staining “above” and “below.” Even worse, if you have cytospun your cells, they are now squished into pancakes meaning that even with very high resolution confocal methods, it will be difficult to eliminate the out of focus contribution from the cytoplasm above and below the nucleus. And if these are widefield images, you will definitely get cytoplasmic contribution to the nuclear signal.

@etadobson’s analysis looks like a good start, but I would be concerned about the accuracy of the measurements if the difference in protein localisation is not large. You may want to consider nuclear/cytoplasmic fractionation and a western blot.


Great advice @Research_Associate !!! :slight_smile:

1 Like