Calculating localization of protein with very different distributions

I’m trying to measure (in CP) the localization of a protein to the Golgi. The protein undergoes a very large distribution change (from ER to Golgi) upon treatment, and I’m finding it difficult to get a Pipeline that can deal with this large morphological change.

er golgi

In the two attached files, the top one shows ER localization of the GFP protein of interest. The red/orange dots are a dsRed Golgi marker. The second image shows localization of the GFP protein to the Golgi.

I can easily segment the red Golgi objects. I can also segment the GFP localized to the Golgi, as well as the disperse GFP, but only in two separate modules (otherwise I get a lot of false objects).

So my first thought was to segment the Golgi, and then measure the GFP intensity inside the Golgi objects. Across various test treatments, this gives me a measureable difference, but with a very intra-well range:
image

Is there some way to improve the localization calculation? Or perhaps a change in the logical approach to measure the difference between the two conditions? At this point we don’t have a cell body dye, but could add one if necessary to help improve the segmentation (esp when the GFP signal is disperse).

Calculating the mean intensity of your protein of interest within the golgi area is not ideal. What if your imaging conditions change slightly?

I would suggest using a ratiometric measurement:

Integrated intensity of your protein of interest within the Golgi area / integrated intensity of your protein of interest in the whole cell.

1 Like