There is no "the best" location. If your gene could have different isoforms with exons missing (frequently not always) at the 5' end, it would be better to look at a region further downstream towards the 3' end.

Depending on the RT-activity of your enzyme, longer transcripts could be reverse-transcribed only partly, so the 5' region could be under-represented. To be on the safe side, you should try more than one in order to avoid a bias due to some peculiarity of a single site. You might want to chose an amplicon where two exons are joined so you don't detect non-spliced pre-mRNA.

There is no "the best" location. If your gene could have different isoforms with exons missing (frequently not always) at the 5' end, it would be better to look at a region further downstream towards the 3' end.

Depending on the RT-activity of your enzyme, longer transcripts could be reverse-transcribed only partly, so the 5' region could be under-represented. To be on the safe side, you should try more than one in order to avoid a bias due to some peculiarity of a single site. You might want to chose an amplicon where two exons are joined so you don't detect non-spliced pre-mRNA.

Thanks for your answer Hüseyin Besir. I counldn't think of a better wording for "the best".

"... You might want to chose an amplicon where two exons are joined so you don't detect non-spliced pre-mRNA." Should I not include pre-mRNA in the analysis if I am interested in the expression profile?

I agree with Hüseyins comments on the primer design.

Yes, you can detect pre-mRNA as well. However, for the pre-mRNA detection, you should consider to include a minus RT control in your experiments (meaning non-reverse transcribed RNA). This control is used to judge the unintentional detection of genomic DNA contaminations in your reaction instead of the intended pre-mRNA (with the identical sequence).