If this is (can be) a problem depends on the size of the product that is amplified from the pre-mRNA. In qPCR the amplified sequences should be rather short (~100bp), and residing introns often will produce considerably longer products (several 100 or 1000 bp). The amplification of such long products can effectively suppressed by the typically very short cycling times (0s denaturation, 0s annealing, 2s extension). So it is likely that it is not a problem.

But you can check it. You can measure the amount of pre-mRNA using primers that would work only on unspliced RNA. So you can at least get a hint if this would generate some signal that could interfere with your quantification.

I am not sure if oligo-dT priming will favour spliced mRNA, because the poly-A tail is added already during transcription, and splicing starts later. Usually, sequence-specific primin for RT is the best option w.r.t to sensitivity and performance. Random priming should be used only when it is not clear which sequences should be amplified or if the cDNA is used to amplify a lot of different sequences.

Contaminating gDNA can only be a problem when the mRNA expression of the gene is very low (otherwise the few gDNA copies won't be any significant amount). You can apply a DNAse digestion, and you can do no-RT controls to check for the impact of contaminating DNA.

It is depending on the size of intron. If the intron is long enough, the PCR condition will be able to preferentially amplify only the mature mRNA instead of amplifying the long pre-mRNA.  The simplest solution we usually do is designing primers on the exon-exon  junctions discriminating between two splicing variant; but carefully check on the sequence similarity to make sure that the primer distinguish two amplicon.

Totally agree with the above atatements about amplicon size.

I would also recommend using anchored oligodT, i.e. oligos that have something but not T at their 3' end. This will decrease the amount of polyA you reverse-transcribe since the primer will preferentially bind to the beginning of the polyA.

Short cycles are good, but I think a 0/0/2 s cycle as above is too short.

Finally, if you are not sure about pre-mRNA you could design pre-mRNA-specific primers and measure the amount of that. Actually, you might be lucky needing to design just one primer binding at an intron or exon/intron junction.

Anyway, I never heard that someone had problems with pre-mRNA

Thanks for the responses, they all contain useful information.

Just to clarify, the primers amplifying the shorter variant would yield a product the exact same size as they would if they were incorrectly amplifying the pre-mRNA of the full-length variant. So unfortunately I can't rely in differences on amplicon size. Perhaps the best I can do is design pre-mRNA-specific primers and hope for no amplification.

Still, if anyone has any other suggestions to approach this problem, I would be happy to hear them.

Our laboratory has always had trouble with gDNA contamination. You have to be fairly brutal with the DNase I and always run -RT controls. It's next to impossible to get a completely DNA free sample of total RNA (or whatever you're isolating) but at least with a -RT control you'll know how much of your signal is DNA-amplified. 

And yes, you can detect specific splice variants and even pre-mRNA using the right primer pairs. We regularly use RT-qPCR to quantify levels of HIV splice variants. I initially try both poly-T and random hexamers if I'm not using a specific set of primers when running the RT. 

it depends on which target do you use, cDNA or RNA (two-step or one-step RT). If you use cDNA, the fully spliced mRNA will amplified.

Marziyeh, how is that possible? I thought that if zou use one-step, then RNA is transcribed to cDNA in the same *tube*, and then the PCR occurs on that template. So there is, technically, no difference in 1- or 2-step PCR. Is it?

I agree with Tristan; No RT controls for DNase I treated samples are essential for determining gDNA contamination. As a general rule, check where the no RT and no template control curves come up in the amplificaiton graph; for me it tends to be @ around 32-35 cycles. As for mRNA splice variants, the cDNA synthesis process does not really discriminate between them. What you can do is design primers that span exons to guarantee that you get amplification of a particular splice variant as some have mentioned here. As you said, using your primers you cannot trust that you're amplifying unspliced mRNA, especially if you have gDNA contamination. Using NCBI Primer Blast you can ask the software to design a primer for you that spans exons, this way gDNA shouldn't be able to create amplicons that can mask your low-expressed gene of interest, and if the curves come up a the same time as your no RT control using your current primers then spanning exon junctions may be necessary to improve the sensitivity and specificity of your assay. Unless you can prove that the no RT control amplifies much later than your gene of interest you can be sure a reviewer will question the specificity of your assay.

In addition to the DNase I treatment and no RT controls, I was thinking that  if the short form of your target of interest only contains part of intron 1, then maybe a good control would be to sit a forward primer in exon 1 and a reverse primer in intron 1 (upstream of the transcriptional start site of the short form).  This primer set will only amplify contaminating genomic DNA and unspliced mRNA.  This would give you the "background" Cq levels for your samples.  This, of course, after DNase I treatment and in addition to the no RT controls.  It is possible that your "background" will be undetectable.

On another note, it is important in assays like this to make sure the sizes of your amplicons for all primer sets are as close as possible and that your primer set amplification efficiencies are the same.  If your RNA quality is high, then using oligo dT priming in the RT rxn should be fine for a 1 kb transcript.  If your RNA quality is low, for instance if you are isolating RNA from FFPE samples, you may want to try random priming, and additionally keep your amplicons as small as possible (we often use 60-70 bp amplicons).

Melting analysis (assuming you are using Sybr green) should be a sensitive way to determine if there is any problem with amplification of the primary RNA from the longer transcript when you use the short-specific primer pair.  Your description is a bit ambiguous with respect to whether there is an intron segment between the short product first exon and the second exon.  If there is, you can also design a primer which bridges the junction between the end of the first exon of the short mRNA and the second exon.  This primer would have only a 3' match of 4 to 6  bases with the 5' sequence of the second exon and the 5' part of the primer would be from the 3' end of the short promoter first exon.  This would avoid amplification of both genomic DNA and the primary transcript from the longer product promoter.  In fact, you could design such a primer for the longer promoter product as well--and still use the same reverse primer for both amplifications.