Going (peri)Nuclear

Localisation analysis is all about asking where proteins, organelles, cells or anything else are. This is, of course fairly meaningless unless you have some point of reference. In today’s post, we’re going to be looking at quantifying perinuclear localisation.

The problem with subjectivity

To the seasoned Cell Biologist / Microscopist, it’s easy to look at an image and be able to say: “that’s clearly a (reticular|endosomal|mitochondrial|nuclear|cytosolic) distribution”. Job done? Not quite (…but any excuse to pull out regex).

Don't write in, it's just for fun.

Don’t write in, it’s just for fun.

For one thing, that opinion is based on years of experience which has a way of setting you into a groove. Also, it’s unlikely that any journal will accept that as a form of quantitation. We need objectivity, or at least predictable and repetitive subjectivity.

Quantifying Fluorescence Intensity – Think Ratios!

The problem with trying to quantify fluorescence intensity is that there are about fifty things that can affect the final intensity (but that’s another post). A great way to get around these problems is to normalise through the use of ratios.

To quantify perinuclear localisation, we’re going to use the integrated intensity (that is the sum of each pixel intensity value) in different regions of our cell image to calculate a fraction of the signal that is perinuclear.

First things first, start up our trusty Image Analysis tool Fiji and let’s get going.

Setting Measurement parameters

Fiji will measure exactly what you tell it to. So we need to tell it what we want measured. Run [Analyze > Set Measurements…] and make sure that Area, Integrated Density and Mean Grey value are selected. It’s always nice to have the Label selected too so you have a record of which image you’re measuring. Hit OK.

Background subtraction

Use the square selector tool (first on the toolbar) to select a background region. Try to avoid any localised bright spots that are clearly due to antibody clumping or debris:

2015-03-19-nuclear02Hit ‘m’ (or select [Analyze > Measure]). The results table will show a value for each of the categories selected above. The Mean value is the mean Background intensity for the image.


Click anywhere in the image to deselect the region (or Ctrl + Shift + A, to Select None). Subtract the mean background intensity from every pixel using [Process > Math > Subtract…]. Input the mean value (in this case ‘1946’) and hit OK.

To clear the results table, either close it (don’t save), or select the data and hit backspace.

Defining an anchor region

Again, to avoid subjectivity in defining where our nuclear (and thus perinuclear) region is, we need to use a different channel to make a selection. Enter DAPI (or Hoescht, DRAQ5 or your favourite nuclear marker).

Double click on the Wand Tool (8th from left on toolbar), to bring up the tool options. For a 16bit image, I found that the 8-connected mode with a tolerance of ~5000 works quite well.


In the nuclear channel, click on the brightest spot in the nucleus until a good estimate of the outline is selected (if you have problems, go back and adjust the tolerance).

2015-03-19-nuclear05“But my selection is in the wrong image!” I hear you cry (although I should probably get my hearing checked). Here’s a pro tip that you never realised that you could live without:

Switch to the other image and run [Edit > Selection > Restore Selection], also available as Ctrl+Shift+e. This will copy your last selection onto the current window.


Making Measurements

You don’t need the nuclear channel open any more so close that down. To get started, hit ‘m’ to measure the nuclear region (NUC). Here’s where the magic starts. Use [Edit > Selection > Enlarge] and enter a value that fits your experiment (ideally using a positive control of something you know is peri-nuclear). Below, I’ve used a value of 15 pixels.

Measure again, this value minus your previous measurement will constitute the perinuclear (PN) measurement.

Finally, using the polygonal selection tool (3rd from left), select the outer boundaries of the cell. Try to get as close as you can whilst still including all of the cell. Because the background has been subtracted it doesn’t matter if you select some background as it will be effectively zero. Measure again, this it your TOTAL intensity.


ANOTHER PROTIP: For ultimate reproducibility, it’s worth keeping a record of the selections, so you can come back to them later if need be. For each selection, hit ‘t’ or [Edit > Selection > Add to Manager] from where you can save or recall the ROIs.

Crunching the Numbers

Under [Edit > Options > Input/Output] make sure “Copy Column Headers” is selected, so you’re not stuck trying to remember what each of the values is.

2015-03-19-nuclear08Copy the data out into Excel or your favourite spreadsheet application

NOTE: You will have two values for integrated density which are the same for uncalibrated images. If in doubt use the Raw Integrated Density which is the sum of pixel intensity values.

As the background has been subtracted already, the analysis is very simple:

To calculate the nuclear fraction divide the NUCLEAR Integrated Density by the TOTAL Integrated Density. 2015-03-19-nuclear09

To calculate the perinuclear fraction, subtract the NUCLEAR Intensity from the PERINUCLEAR intensity and divide this value by the TOTAL intensity.


To calculate the remainder, subtract the former two fractions from 1 to get the fraction that is neither in the nucleus nor the perinuclear region.


Repeat for all of your cells, then calculate the mean nuclear, perinuclear and other fractions as well as the standard deviations if it so pleases you.

Plot it!

You can plot the data any way you please, however two suggestions that come to mind would be a stacked column (below left) or a clustered column. Here the clustered data are plotted +/- standard deviation of the mean.



2 thoughts on “Going (peri)Nuclear

    1. Dave Mason Post author

      Hi Ann, thanks for your comment. I’ve certainly taught a few people this technique, so the answer is maybe but not specifically!

      The nice thing is that you can use the same or similar idea to do all sorts of things. If you follow the link below, there’s an example where I’ve used radial distributions in spheroids. Same basic idea.




Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.