Hi Felix,

there is the possibility of undetected 3'-UTR and the possibility of an annotated tanscript on the other strand. In your case you should consider alternative polyadenylation. An example is the BDNF gene: http://www.ncbi.nlm.nih.gov/pubmed/20733072

From my experience with 5'UTR'Ss there are many annotated alternative 5'ends, especially for low abundance transcripts. Experimentally, you may get an idea with an 3'RACE approach.

Felix, since you may work with a single exon gene, I assume your designed primer pairs is not specific to only mRNA, but also to genomic DNA if they are present in your extracted RNA. If your RT was done with random hexamers, you would see the impact of the contaminated genomic DNA. If RT was done with oligo(dT) primers, this would less be an issue since you would only amply the mRNAs. One choice is to treat your extracted RNA through a DNase I purification to get rid of any potential DNA contamination. Then repeat your realtime time PCR experiment. Hope this will help.

Thank you both Uwe an Yutao. My RT reaction was performed on DNAse1 treated RNA extracted with the Qiacube RNeasy mini protocol, and using oligo(dT) primers. I also performed 3'RACE and indeed got different lengths on different primers. Alternative polyadenylation certainly seems to fit. Thanks for the link as well!

Just to add to the answers you already have. This is a paper from the Darnell lab by my colleague Donny Licatalosi  that shows a genome-wide analysis of  alternative use of polyadenilation site and its  regulation by the splicing factor NOVA.  It will  give  a broader biological context to your question.

http://www.nature.com/nature/journal/v456/n7221/full/nature07488.html