Maybe someone knows why this happens. The situation is that after PCR purification of gel products (cut one band), on the next electrophoresis, instead of one band, two bands appeared, how could this happen?
I suspect that your amplimer contains a base mismatch. This mismatch forms a bubble in the dna so that the double stranded dna runs slowly through the gel because of the bubble in the structure. If you cut out this band and denature it and then renature it you will get 2 bands of apparent (but not real) different sizes. Actually they are the same size but run differently in the gel. On re annealing you get 3 products......normal/normal strand and mutant/mutant strand which both run at the same size and a heteroduplex dna of one noemal and one mutant strand ( or vise versa) and this heterodu[lex again has the mismatched base and bubble so runs slowly and once again you will have 2 bands on agarose gel separation. If you sequence both bands you should see the mismatched bases in your sequences
I agree with Dr. Paul above, this happens due to formation of heteroduplexes. Your original band contains more than one product, with no noticeable difference in mobility, but the slow moving band is a heteroduplex. On cutting the expected band and Re-PCR, all the three combinations are generated again. We observed this and resolved in our study on such observed heterogeneity during ribosomal DNA ITS region amplification in Asiatic Vigna species (see Saini et al., Genet. Res., Camb. (2008), 90, pp. 299–316.). We also proved the differences (indel lengths, 2 bp and above) among clones by doing heteroduplex analysis by mixing different clones, in that study.
If you are interested in getting the amplicons, instead of band purification go for cloning and sequence.