Unwanted frameshift after overlap extension PCR
I did an overlap extension PCR to introduce a couple of point mutations in the middle of a gene sequence. The construct did not express any protein when transfected. Later, I sequenced the construct to find that there has been a frameshift, exactly at that point where one of the amino acids was mutated, thereby leading to truncation, etc.
I intend to do the whole thing all over again, but I'm confused what step to start with. I am confused whether the frameshift was introduced in the Overlap PCR step (where two initial products are thermocycled for a few cycles before adding the flanking primers) or in one of the earlier PCRs which amplified the two products to be spliced together. Is it the overlap extension step where there is a higher chance of a shift being introduced?
Since I haven't encountered this problem before, I am unsure if I should just do the overlap extension again (I have the two smaller PCR products), or start from scratch. The primers seem to be fine.