DNA folds can cause multiple problems during PCR, specifically what we refer to as "copy-choice replication". Three dimensional structural scaffolds seem to give predictable patterns of PCRs and sometimes not so. But the common root of the problem lies in the fact that these regions are repeat rich and can form stable stem loop structurs either intramolecularly or even intermolecularly leading to complex products whose sequence doesnt match that of the original template. Once such example. albeit extreme was published below. 

Solutions - Several companies come up with reagents for complicated PCR products - eg. Q Solution from Qiagen, helps in GC rich templates and those that form secondary structures. 

Article Local repeat sequence organization of an intergenic spacer i...

DNA folds can cause multiple problems during PCR, specifically what we refer to as "copy-choice replication". Three dimensional structural scaffolds seem to give predictable patterns of PCRs and sometimes not so. But the common root of the problem lies in the fact that these regions are repeat rich and can form stable stem loop structurs either intramolecularly or even intermolecularly leading to complex products whose sequence doesnt match that of the original template. Once such example. albeit extreme was published below. 

Solutions - Several companies come up with reagents for complicated PCR products - eg. Q Solution from Qiagen, helps in GC rich templates and those that form secondary structures. 

Article Local repeat sequence organization of an intergenic spacer i...

Different additives for the PCR reaction can help to amplify sequencies with a complicated 3D structures, like betaine or DMSO.

I myself expressed the whole predicted premiRNA within cells( a mammalian expression vector expressing Genomic loci of pre-miR under a mammalian promoter) the RNA was made to cDNA using poly A addition method(mature miRNA does not have PolyA) & miRNA specific RT primers. the dominant product is supposed to be your mature miRNA processed within the cell. It is absolutely gold method to go for northern blot. Anyway I cloned my PCR product in TA vector and I sequenced lots of them!!!! the prediction might be precisely correct about pre-miR but it is not 100% perfect about mature sequence maybe using a loose method like what I explained here enables you to detect multiple variant of pre-miR processed products. that is mostly found common between clones is the mature one probably.

We have followed stem-loop RT protocol. Stem-loop primer was specific to the last six bp of mature miRNA and interestingly this 6 bp is always missing in the sequencing results. We have sequenced it for almost 10 times and none of them are similar except first 16 bp, which we got for every cases. For your information, we got expected size band by semi-Q PCR using cDNA; real-time PCR and solution hybridization has also been done and got positive result. What could be the possible reason?

Thank you everyone

In your qPCR are you sure that  you have amplified only one product? melt curve is nice and completely sharp? there are some possible answers first it is really caused because sequencing condition is not optimal and sth has happened in sequencing(low vector purification affects badly sequencing). Second, as i replied previously the real mature sequence might not be the same 24bp predicted. miRNAs size could be 16bp-24bp. Try other bioinformatics methods that i have used in my article. In case of miRNA amplification by PCR and qPCR u should use a detailed marker for seeing the products on acrylamide 12% gel. Dude, nothing else comes to  my mind right now. This whole procedure of finding and proving a miRNA is not sth easy and standardized and sometimes it sucks.I wish you best

Thank you very much Sepideh Parsi.

I know miRNA size may vary but we designed the stem loop primer, specific to the last 6 bp of predicted mature miRNA (24 bp long) and as i mentioned before, we got right size band by semi-Q PCR. Melt curve wasnt that sharp and nice but still they were good. Now suppose, they are amplifying another sequence rather than my desired sequence, we should get same sequencing result. But unfortunately none of the results were similar except first 16bp(what i mentioned earlier). We didnt sent purified plasmid for sequencing rather we sent gel extracted pcr product which was amplified by plasmid specific primer. I would like to mention that in this PCR we got desired size band(insert+ plasmid region)

I strongly suggest you to clone your pcr product obtained after qPCR.  when you send positive plasmids having ur seq  for sequencing generally sequencing result is far more better since you are using the universal M13 primers on the vector. And you definitely will loose sequencing signal on the first and end of sequence which is not important but u will have the nice complete sequence product which is ur miR. hope this be helpful

Md. Tariqul : I am suffering from this problem too . I designed the stem loop primer for specific sequence containing last 6 nucleotide of predicted mature miR but i am not finding the last 6 base pair in my cloned product. Moreover , in some of my sequencing results , the sequence of mature mir is 1-2 nucleotide change. I cloned them in PGEMT vector (TA cloning). I got desired results in blue/white screening , colony PCR but when sent for sequencing last 6 nucleotide is missing or the sequence get 1/2 nucleotide differ from mature mir.

What should i do ? will it be helpful to characterise the sequence of only pri-mir for novel mir or should i change the PCR condition or cDNA synthesis method of mature miR ?

Kindly help.