Because RACE does not work very well.  RACE is not very robust technique.

Your best option is to find a lab, where they do it routinely and travel there to process your samples. I had a colleague who spent a year to make the RACE work....

Agreed, it is very sensitive technic when you expect to clone sequences expressed at very low level (>35 cycles in qPCR, starting from 50ng equivalent mRNA).

Critical steps are:

- quality of your cDNA

- length of your expected amplicon (easiest when < to 500 bp)

- In my hands, specific RT worked well for mRNA expressed at a reasonable level. Therefore for low expressed mRNA, I prefer to mix RN6 and pdT to generate full-length mRNA. I use this cDNA sample for RACE and for cloning after sequencing of mRNA extremities.

I would suggest you to:

- Do not use more than 2 µg mRNA for 25µl reaction;

- Use highly processing reverse transcriptase and checked whether it adds 5' nucleotide stretch (CCC or GGG, I can not remember.., I used Primescript);

- Add Betain is the RT mixture can also improve the quality of RT at mRNA extremities (Don't know the basis, but it worked in my hands);

- Purify your cDNA sample after reverse transcription (I simply make phenol/chloroform/AIA purification prior to precipitate by 10% AcNa+ and 2.5V EtOH 100);

- performed Nested-amplification whether PCR products are invisible or expressed at low level;

- use highly processive polymerase such as Phusion or Q5, but keep in mind that it does not add 5' adenyl required for TA cloning. If you have no idea about the amplicon size, try several polymerization times;

- About PCR program, I did the top-down method in order to start with low efficiency, high specificity amplification and then I increase the yield by decreasing annealing temperature:

with Phusion:

15 x :   98°C(5s) / 68°C (15s) / 72°C(30s/kbp)

20 x : 98°C(5s) / 64°C (15s) / 72°C(30s/kbp)

Note that primers I used have fusion temperature around 65-70.

Hope it will help you.

Good luck,

Gabriel

   I think the quality of your RNA prep and cDNA is the most likely issue.  Have you checked your RNA by UV.  If your RNA and/or cDNA were good, your control PCR should be working.

    I used the 5' RACE kit from Invitrogen and it puts a dC/dI stretch on the 5'-end that can be used as an anchor primer for subsequent PCR.  I would use a set of primers against a known gene as control primers to check the quality of your cDNA.  The kit I used came with primers specific for (S15) yeast ribosomal subunit.  Although these PCR primers worked very well, they worked too well.  I always got a PCR product, but testing other  validated PCR primers for genes, our lab is interested in, gave no product.  I don't know if this will help, but here is the paper:

Article Molecular characterization of the AdeI mutant of Chinese ham...

First thing, RACE is a very difficult technique...so even from kits it will be hard, and each different message may not work equally well.

From what I can tell about the theory of 5'RACE and the procedure you used, your first step with the random hexamers in my opinion will complicate everything else downstream.  I would stay with your GSP in your first strand RT reaction and NOT use the random hexamers so that the addition of your polyA with TdT will be minimized to only DNA that has been synthesized from your GSP (and you need to make sure that your primer is clear of the reaction or you will extend your primer as well).  Everything else you mentioned should be okay and all the other things above are also relevant (ie. good starting RNA, PCR conditions, etc etc).  And lastly, it's probably okay, but funny things have happened, make sure that the sequence you are using with your internal GSP actually exists...before you go ahead and do the RACE, make sure that your RNA of interest is actually there by doing a quick RT-PCR with relevant primer sets.  If it's not there to begin with (at least the downstream of your putative start site), it will never work.