Could you check whether your gene of interest is expressed using an orthogonal method, like quantitative PCR? This could give you a good idea whether your RNA of interest is expressed at a high enough level to detect with in situ hybridisation. Seeing more intense staining in the corpus callosum isn't necessarily an adverse finding because some RNAs actually localise to axons - if you could confirm this another way (eg, RNA extraction from corpus callosum vs cortical tissue, followed by qRT-PCR) you might have found out something very cool about the subcellular localisation of your target.

As I understand it, the RNAse treatment is to establish background: you are literally digesting away ALL the RNA in your sample, so any probe that sticks under these conditions is binding non-specifically. If you see nothing after RNAse treatment, that is...exactly the expected result.

This sort of seems like the opposite of a problem, really.

ISH is always a balance of "cool and permissive enough to allow hybridization to occur" and "hot and disruptive enough that only specific hybridization can occur", and the optimal point will rarely be perfect: for weak signals you will usually have to accept that the background will be higher, and for strong signals you can be harsher, with the caveat that smaller, less intense sites of legitimate expression may be lost.

Formamide is just there to help the hybridization occur. Washing with formamide should reduce non-specific binding while preserving specific binding, while washing with RNAse A should ruin all specific binding.

Also, be really, really careful with RNAse A: if you're working with RNA routinely, you really don't want to accidentally contaminate a bunch of buffers with a potent RNAse.

what is the rationale for RNAse A treatment when the target signal is RNA? I don't understand why this step that seems detrimental to the detection of a signal

Thanks a lot for your replies.

To my understanding, RNase A should specifically degrade only single stranded RNA molecules. So, the rational is that, if my probe is binding specifically to its target, then RNase A should't affect it, but it should only affect the RNA probe that is not specifically bound to the target, so unspecific binding. This is a washing step that is typically used in In Situ Hybridization (to my knowledge).

In fact, with another probe that I also used, specific for a different RNA target, I see no difference in expression pattern between the two washing conditions (with and without RNase A).

Also, based on papers that I read, I noticed that some people also don't use RNase A but they do this after verifying that the expression pattern observed with and without RNaseA is the same. What happens in this case is that, as already mentioned, the pattern observed is the same, but the intensity of the signal is higher without RNase A digestion (because of the unspecific signal).

So, the problem is that, with the probe I mentioned, the pattern is completely lost after RNase A digestion, which makes me think that it is actually binding unsepcifically.

But I could be wrong, of course.

Anyway, I will try to validate the In Situ results with qPCR but if you have any comment on what I just wrote here, I would appreciate it.

This sounds a lot like one of those "if you have a stonking signal, you can get away with this" scenarios, where all the examples of it working are robust and optimised to the hilt.

The fact it's removed by RNAse, and it's specifcially the RNAse that does it, implies that whatever the interaction is you're seeing, it's RNA-based.

You could try "add RNAse A to a sample long enough to degrade all RNA, then add your probe": if you see no binding, then yeah: your interaction is RNA-based.

You might, for example, have some mismatches in your probe: RNAse A does, apparently, cleave at mismatches, so you wouldn't need many to remove the probe signal.

Another check would be to design a second probe to the same transcript (different region) and try that. If again you see signal in the same places, then...trust it.

I'm not doubting that the signal is RNA based, as I'm using a radioactive probe, the signal that I detect can only come from the RNA probe.

Pre-incubating the slices with RNase before adding the probe will for sure get rid of all the binding because, once I add the probe, it will be destroyed by the RNase as well. That's why I'm using this step only at the end, after the hybridization step.

So you're saying that the RNase A step is not a valid method to verify if the signal is specific? In other words, if the signal I see is the result of my probe binding its specific RNA target and not to the fact that the probe is binding unspecifically to other RNA molecules?

"RNA based" as in "RNA-RNA interaction". Your probe IS binding to RNA in your sample, and not being chelated by some other tissue component: that is what this is telling you.

I was not suggesting you were acquiring random radioactivity from elsewhere.

As noted, RNAse A can cleave mismatches, so if your probe isn't a PERFECT match to your target, it might be partially cleaved. This will lower the binding strength so it may dissociate more easily, and will then be cleaved again. Law of mass action will result swiftly in "no remaining probe".

You could check mismatch potential with PCR (amplify the target region from some rat brain cDNA, sequence it and see if you have any SNPs or similar).

To be honest, if your binding conditions are quite stringent, you might see signal degradation even in cases of a perfect match: RNA:probe complexes exist in equilibrium with free unbound probe, and at any given moment some probes will be dissociating and other associating. As soon as you remove all the unbound probe (via digesting it away), you're shifting the equilibrium in favour of dissociation. Depending on how stringent your conditions are, this might result in a downward spiral toward signal loss. You hinted at this yourself:

"some people also don't use RNase A but they do this after verifying that the expression pattern observed with and without RNaseA is the same. What happens in this case is that, as already mentioned, the pattern observed is the same, but the intensity of the signal is higher without RNase A digestion (because of the unspecific signal)"

The intensity is not dropping because of "unspecific signal", it's dropping because some actual legitimate interaction is being destroyed. The approach just relies on this happening at a slower rate than all the non-specific interactions being destroyed.

Honestly "adding a ton of RNAse to a protocol that explicitly depends on RNA interactions" just seems like an incredibly risky thing to do.

And you're potentially covering your entire lab in a thin veneer of RNAse, which is never good. If it's stable enough that you don't think you can inactivate it (as suggested by your response to my pre-incubation suggestion), then maybe it's not something you should be using so liberally.

Other options to consider: radiolabelled DNA probes (less susceptible to degradation), or FISH.

There are a ton of new developments in fluorescent ISH tech now, many of which can be multiplexed, and the sensitivity can be really, really good (single molecule level).