Differentiating ISH signals from background


I have a rather elementary question on how to interpret the Allen Brain mouse ISH data. Since there was no counterstain used in these experiments, any visible chromogenic signals should presumably represent target gene expression. Looking at the ISH data for Aif1 (Iba1, a well-known microglia- and macrophage-specific marker in the brain) as an example, I can see some dark puncta which appear to be the ISH signals based on their frequency/distribution in the tissue (blue arrows in the image below). In addition to these puncta, however, you can also see in this image (and in the majority of images available in the ISH data browser) that there is a very clear chromogenic signal highlighting every other cell in the tissue. I presume that these nuclear signals are simply background/artifact from the staining process, as the Allen Brain RNAseq data and other literature in the field unequivocally show that Aif1 is selectively expressed by immune cells in the brain. However, a colleague of mine with more background in ISH believes these signals may indicate widespread, low-level expression of the Aif1 mRNA.

Can someone please clarify the source of these nuclear signals? Am I correct in assuming that the dark puncta in the image below are true signals and lighter nuclear signals are simply background, or do they simply represent different degrees of Aif1 expression?

Hi there,

I asked a few of our team members for input on your question and this was their response:

I agree with their assessment and visualization of signal vs background in the ISH image. They are correct that diffuse pink signal is generally regarded as background signal from the staining process, although we cannot rule out the possibility of some low level expression. Generally we have more confidence that dark purple-black signal is true signal.

You might also consider reading this: http://help.brain-map.org/display/mousebrain/Documentation

In Supplementary Data: Allen Mouse Brain Atlas, the document ISH Platform Controls might be a good reference when interpreting this data.

Hopefully this helps!