I am attempting to construct a deletion cassette of 1.497 kbp using three individual amplicons of sizes 279 bp, 1035 bp, and 243 bp (in that order). Fragment 1 (279 bp) and fragment 2 (1035 bp) have an overlap region of 30 bp, while fragment 2 (1035 bp) and fragment 3 (243 bp) have an overlap region of 30 bp as well. In the first round of overlap extension PCR, I am adding 0.1 pmol of each fragment (in 25 uL reaction volume). In the second round of overlap extension PCR, I am adding the terminal primers (forward primer of fragment 1 and reverse primer of fragment 3). Along with the desired 1.497 kbp band, I am getting many other bands, primary of which is a band at ~1.3 kbp, which is so close to the desired band, that I am unable to extract the desired band only from the gel. I am unable to understand where that ~1.3 kbp band could be coming from. In fact, at times, the ~1.3 kbp band is brighter than the desired ~1.5 kbp band. At first I thought it could be due to presence of either the wild type genome or primer dimers. But even after gel extraction of the original amplicons, I am getting that undesired band. Changing annealing temperatures does not seem to help either. How do I resolve this issue? I have attached a picture to explain the situation better.