For many years I used S35 radioactive or digoxigenin labeled probes but in the past 6 months have been using the RNAscope technology that Justin Smith mentioned above.  This is not an inexpensive method but it is undoubtedly the very best in terms of specificity and simultaneous localization of multiple RNA targets in the same cell. I do not think that I will ever use S35 or digoxigenin again.

We are working with S35 labeled as well as Dig-labeled probes to detect mRNA levels of interested genes. Radioactive route is really sensitive and I personally prefer this one although it is not very easy to 'handle', regarding the whole process involves lots of procedures.

I haven't used radioactive labeling, but I have seen perfectly good and clean staining with Dig-AP staining in the POA. The advantage of fluorescence is that it becomes more and more specific with each conjugate. I recommend doing them on cryo-sectioned samples rather than whole mount and then sectioning due to penetration issues. And as a rule of thumb, use ISH only for visual data to see expression patterns and use qPCR for quantitative measurements. I have seen a lot of criticism over data because of statements about a sample just "looking" like there may be more or less expression due to what looks like up/down-regulation. Like I said, I haven't done radioactive labeling, but make sure it provides explicitly accurate signal to transcript quantification. Lastly, it takes time to get a good ISH protocol working when the lab hasn't established a working protocol already. Good luck!

Very detailed and interesting analysis of Sarma. We had used only ISH with radioactive probes. The advantage is to quantify the signal obtained both by film or by emulsion. You can keep your autoradiograms and your labeled sections. The autoradiographic film give the advantage to screen the brain regions where your target of interest is present. After, you can label your cells (by revealing the signal present in cells). 

Our lab has used both DIG and radioactive in situ methods.  Recently, we have invested in the RNAscope technology from Advanced Cell Diagnostics (see Smith et al, 2014 Physiology and Behavior for example).  The specificity of probe amplification is unparalleled allowing for quantification as one "dot" of fluorescence corresponds to one strand of labeled RNA.  In fact, RNAscope is used in clinical diagnostic settings as well as in laboratory research.  Also, we have been able to combine fluorescent RNA labeling with fluorescent IHC in free-floating PFA-fixed mouse brain sections which is perfect for anatomical studies.  Let me know if you have any specific questions.  Good luck! 

Thanks for all of your comments, much appreciated! Having spoken to colleagues, I think it likely that we will go now the radioactive route.

For many years I used S35 radioactive or digoxigenin labeled probes but in the past 6 months have been using the RNAscope technology that Justin Smith mentioned above.  This is not an inexpensive method but it is undoubtedly the very best in terms of specificity and simultaneous localization of multiple RNA targets in the same cell. I do not think that I will ever use S35 or digoxigenin again.

We always rely on non-radioactive ISH (single- or double ISH, either colorimetric or fluorescent). Our experience with non-radioactive ISH is summarized in the enclosed manuscript. If you have any further question, please feel free to contact me at any time.

In my hands S35 is more sensitive than FISH or DIG ISH. If your mRNA is low expressed, radiactive is the way to go.

You can improve resolution and store for years dipping in emulsion your slides.

I tried RNAscope and works for some probes but got bad results on others.

I have worked with 35S, 33P, DIG, and Biotin labeled riboprobes and now have been trying the new branched DNA probe kits (like RNAscope) in formalin fixed and fresh frozen rodent brain/spinal cord tissues.  If you are just starting to set up the lab for in situ hybridization histochemistry, I would go the Advanced Cell Diagnostics (ACD) RNAscope or Affymetrix/Panomics QuantiGene ViewRNA route.  Even though I have had the best results with 35S and 33P in sensitivity and specificity, the reagents for radioactive in situ are getting more expensive and more difficult to obtain.  There is also the issue of radiation safety, decay of probes, length of time to expose slides (ranging from weeks to months) to emulsion, and troubleshooting many steps.

If you do not have much experience with ISH, the branched DNA (bDNA) probe kits are easy to use and are quick in getting results.  The issue of non-radioactive lack of sensitivity has improved with bDNA technology due to designing ~20 oligonucleotide pairs to your target RNA which ensures specific binding of what mRNA you're labeling.  When the oligo pairs bind, the signal is produced by a series of preamplifiers, amplifiers, and label (which can be chromagen or fluorescent).  The benefits of using the kits are being able to buy the probes from the company (no need to clone and transcribe probes yourself) and to probe for up to 3 or 4 RNAs in cells.  Furthermore, all reagents are provided for the pretreatment, hybridization, detection steps in the kit and results are obtained within a week (no need to wait for weeks or months in emulsion).  The problem with the kits are not being able to design your own probes, although the companies will provide technical support in designing new probes for a price.  Other problems I've had are inconsistent results.  Some probes worked great in my tissue and other probes have not worked.  I haven't been able to troubleshoot why yet, but it may be due to pretreatment of tissue or probe design.  So, you need to work with the company to troubleshoot.

Sorry for a long winded answer.  Hope this info helps!

We have some experience in non-radioactive ISH (either colorimetric or fluorescent). Please go through the enclosed paper to get your own idea. Best, Jose L. Lanciego