PCR band moves lower than its size on agarose gel - (Aug/14/2012 )

Anybody experienced something like this before?

I have amplified a 3.5 kb band with PCR but on agarose gel it stops at nearly 3.1 kb. I gel extracted the band and ligated it to pJet and sent for sequencing. The sequencing results show this is the same 3.5 kb band that I was expecting.

But then again, when I digest the plasmid with RE enzymes to get the insert out, the band that appears is exactly at the same size of 3.1 kb. It's driving me mad. All other PCR products move to 3.5 kb, except this product.

Highly G/C rich would give some secondary structure which could result in running at a different size.

when I digest the plasmid to release the insert, I see another faint band at 2.5 kb too. That's why I thought maybe there is a secondary structure forming there. But if this band successfully ligates into my plasmid then it means that it is in linear form when it is replicating in TOP10. How come it forms secondary structure so quickly after digestion? I don't understand.

You can probably try to run it in denaturing conditions to test this, I think it will be the same as for RNA, formamide in the sample or something.
Also differences in other reaction components maybe causing shift, as seem commonly in the RE digest bands. We add appropriate restriction buffer to our DNA marker when checking restricted fragments. Doesn't happed in PCR though, but maybe there is something in your reaction that makes it more charged.