If this is the first time establishing this in your lab, expect to have some troubleshooting to do. In situ hybridization is time consuming to troubleshoot as there are so  many steps where things can go wrong.

It may be possible, especially with human-derived tissues, that losses of RNA have happened due to delays in fixation causing some necrotic change, over- fixation, or RNAse contamination in transit. We have more control in animal studies. The positive control may just be so much more abundant that you get some signal but your test signal is below the threshold.

So, what to try? Here's a few starter ideas:

SALTS:

Vary the concentrations in the washes. Use a little more salt and you may get more background, but also more signal. Once you have a signal you can work on clean up by titrating down the salts.

TEMPERATURE:

Lower temperatures (seems you've tried this). Try adjusting prehy, hyb and wash temperatures.

PRETREATMENTS:

Experiment with heat retrieval, Triton X-100, and proteinase K time/temperatures after deparaffinization to get the probes accessing their target. This can vary, and human derived pathology samples are often over-fixed and need more robust pretreatments to break the heavy cross links and get the tagged probe in there.

A NOTE ON PROTEINASE K: each brand uses a different number of IU/mg. The units of activity are what matter, not the mg/ml. Thus if you are copying a paper based on mg/ml concentration, your brand may be less active and thus need more time ,hotter temp or a higher concentration. 

HYBRIDIZATION BUFFERS:

A) Formamide-poorly deioinized formamide is a number one cause of poor hybridization in my experience. Even though it claims to be pre-deionized we always re-deionized it just before use in the hyb buffer. We make small batches and didn't store it long. ALSO-different GC contents of probes may need less/more percentage of formamide in the hyb buffer for success. We always used 50% formamide but some people claim success with varying this.

B) Salmon sperm DNA or other nucleic acid blocker-must be FRESHLY denatured and quenched on ice the day of use.

ENHANCEMENT:

You could try Tyramide signal amplification if your signals are weak.

RNASE HYGIENE:

Make sure any washes touching the sample AFTER deparaffinization and before hybridization are totally RNASE free. DEPC treated buffers, clean pipettors and designated glassware that has been RNASE treated or baked at 450C overnight are mandatory at this stage. don't let anyone handle bacteria or RNASes near your bench.

I hope this gives you some ideas about where to start. GOOD LUCK!

Just to follow up: We ended up lowering our formamide concentration based on the relationship between formamide and GC content in setting the optimum hybridization temperature (given the equation, which I haven't been able to find since as the web page was taken down, we assumed hybridization temperature to be sort of a proxy for "probe stability"--when will it be more stable given more heat?).

We also, ironically, drastically lowered our hybridization temperature. This worked!

We found the optimum conditions for our poly(A) probe to be 10% formamide in the buffer (up to 25% also works fine), with the probe denatured immediately before hybridization at 80*C for 5 minutes, then hybridized at 30-35*C, all else being standard (can provide a full buffer recipe upon request--note we did not add salmon sperm).

This protocol worked very well in paraffin-embedded tissue, but we never got it to work in thick free-floating tissue.

Thanks again Karen for your helpful and detailed response!