I'm not sure where people get the idea from that you can't quantify with Northern but you can with RT-PCR. That is not correct. Both allow quantification (relative one or absolute one, depends if you use a house keeping gene or spike the sample with the synthesized version). Both have different advantages and disadvantages.
For Northern you see what you see, if the probe is specific enough you know the melting temperature is reasonable and comparable to you control gene/template - you have very solid data with little fear of interference from whatever else is in the sample. And you can see the length of your construct is correct in total, not artificial from amplification. So you see processing or degradation etc very well. But Northern needs a certain amount of template and of course if you want the highest sensitivity - radioactivity has a certain messiness to it.
RT-PCR needs because of the amplification way less material, so it is more sensitive, but you detect indirectly and that makes it more tricky to set up correctly. Designing correct probes and primers, test them, determine their efficacy in dilution controls, specificity in the sample and the potential inhibition of the RT-PCR from your sample takes time and some effort and is often not done. If you use Sybr Green you need to show the denaturation curve or gel to prove your bands are actually your target and not simply unspecific for example. You won't see processing for a while as in RNAi or such if you don't target the cut sites. And you can have a false signal from genomic DNA if your primers don't stretch over exon junctions/splice sites.
I am not in the right position to answer the question but Northern blot can tell us the amplified product is the right one we want or just a mixture of multiple products.
Northern can't give you quantitative measurement of your expression gene but it is very useful to locate your gene within different tissue. also reverse transcriptase is semiquantitative and the best for qualitative and quantitative is the Real time as Bjorn said.
'Reliability' of a technique for gene expression is actually compounded by several factors.....Assuming if every step of analysis is followed properly..then RT-PCR is as reliable as northern but can give much more information than a northern.....in less time and using less amount of sample.....
You will definitely need both information. RT-PCR can, with several precautions, give quantitative results, Northern blots may tell you a lot about integrity of RNA and in some cases also about accumulation of processing intermediates.
Northerns can be quantified using the right software and in fact are way faster that RT_PCR because you do not have to play around with primer dimer issues, melting curves, etc. Both are good techniques and frankly it depends on what you have in your lab, available funds, and whether you are careful enough (northerns can be tricky).
Low copy number genes might not show up on Northerns unless you have a pristine clean technique that allows you to do overnight exposures without increasing your background (this is not easy to do).
I always start out with northerns, because once you have your gene of interest cloned, you basically get results in 3 days. In 3 days you might get your RT-PCR primers but won't have all the issues worked out.
A very tricky question. The only clear cut answer is that qRT-PCR is more sensitive and is reliable over a greater dynamic range of RNA concentrations than Northern blots.
Northern blot has its advantages which include greater specificity and the fact that you can visualize bands and detect potential splice variants.
I'm not sure where people get the idea from that you can't quantify with Northern but you can with RT-PCR. That is not correct. Both allow quantification (relative one or absolute one, depends if you use a house keeping gene or spike the sample with the synthesized version). Both have different advantages and disadvantages.
For Northern you see what you see, if the probe is specific enough you know the melting temperature is reasonable and comparable to you control gene/template - you have very solid data with little fear of interference from whatever else is in the sample. And you can see the length of your construct is correct in total, not artificial from amplification. So you see processing or degradation etc very well. But Northern needs a certain amount of template and of course if you want the highest sensitivity - radioactivity has a certain messiness to it.
RT-PCR needs because of the amplification way less material, so it is more sensitive, but you detect indirectly and that makes it more tricky to set up correctly. Designing correct probes and primers, test them, determine their efficacy in dilution controls, specificity in the sample and the potential inhibition of the RT-PCR from your sample takes time and some effort and is often not done. If you use Sybr Green you need to show the denaturation curve or gel to prove your bands are actually your target and not simply unspecific for example. You won't see processing for a while as in RNAi or such if you don't target the cut sites. And you can have a false signal from genomic DNA if your primers don't stretch over exon junctions/splice sites.
I agree completely with Ulrike Jung. qRT-PCR can't be a very tricky technique. Essentially because of its sensitivity it can amplify signals as well as errors. In addition, since this technique is widely accepted as the golden standard for quantitative gene expression, people have the tendency of trust any result uncritically. All in all, although qRT-PCR it's easier to perform than, and more widely largely accepted of northern blot, I personally think that northern blot is more reliable. Of course, when gene expression is low and/or or starting material is scarce, qRT-PCR is the only possible option.
Northern is reliable in that you know exactly what you're getting when detecting your RNA of interest because you have standard ladder. You can also quantitate the intensity of the bands in the Northern to give you relative abundance in a kinetic study. However, qPCR is more sensitive and can be used to determine absolute levels of RNA/cDNA even if the gene expression is relatively low. However because you have an RT step before qPCR step, the outcome can be biased. To remove bias you can use different quantitative methods to normalize your data. I think both methods are useful so designing experiments using both will only make your paper look better!
These two methods are used for different reasons. If you are looking for abnormalities, you will need to look at each exon and Northern does not allow you to do that. Most labs will use a Northern linked to their RT-PCR method to confirm the identity of their PCR product. For tissue distribution studies, Northerns are more suitable but realtime PCR would also be appropriate.
I am one of the "old guys" that have done a lot of northerns during their carreers. When having trouble since I was working with genes from a family with up to 95% homology at the mRNA level, whent to RNAse-protection assays and now I am working mainly with RT-qPCT. The big advantage I find for Northern blot is that you can size the transcript. See if there are splice variants, or even degradation intermediates, which is very difficult with the other techniques. It works fine with intermediate and abundant RNAs. RNase protection is very sensitive but does not allow to size transcripts.... it is also a lot of work to set it up (although I remember the time I was doing this as a lot of fun, maybe because I was younger). For most purposes the RT-qPCR works fine, but you should be aware that you do not know if you are detecting an splicing variant different form the RNA coding for a protein of interest or even a degradation intermediate.
I think Jose has a point about splice variant stuff, but in my experience we can certainly design a primer set which could detect only a variant of our interest. Nevertheless, for several purposes, Northern blotting still has upperhand over real-time PCR. e.g. in microRNA field, one can look for the expression of microRNAs using real-time PCR but as you know, miRNA are very small RNAs and the quantifications using real-time PCR at times is misleading. Therefore, to confirm qPCR data, we generally perform Northern-blot and / or in situ hybridization.
All in all, both the techniques are reliable as well as questionable under certain conditions. So at least once, we should validate our data with both the methods to proceed further.