You can PCR/mutate your region of interest, and then use the PCR product as a primer set for a mutagenesis polymerization (circular polymerization around the entire plasmid). This would be the Qikchange (sorry, I never remember how they spell it) type reaction, and you can check their protocols, and calculate what concentration of your PCR product would be equal to a minimal 0.2 or 0.3 uM primer concentration. Dpn-I enzyme can then destroy the (methlyated) plasmid template, leaving only the new product.
If you don't want an entire plasmid, you can still use a plasmid template (methlyated by the bacteria) and PCR/ mutate your 405bp, then use the PCR product as a primer, along with an external primer at the end of the 2.6kb, in separate reactions to each end of the 2.6kb, gel purify the two overlapping products, then combine them along with only the two external primers to PCR the 2.6kb.
Can you understand?
Another alternative would be to do mutagenesis (the Quikchange style can work again) to add silent mutations to create restriction enzyme sites closer to where you want them.
3d method: PCR 3 segments, mutated, and two non mutated end sections. Then add klenow polymerase with only one dNTP so it runs backward until it reaches that dNTP base, and stops. This creates overhangs. Your PCR product segments are designed so they overlap for the overhang length, so the 3 segments can be ligated together. If the onverhangs are not palindromic (as they usually are) they can not self ligate so it should be easier to ligate 3 inserts together at once than with normal palindromic recognition sequence overhangs. Do you think this can work?
4: PCR two end segments, and ligate into a vector. You only need to find a 3nt 1/2 of a blunt end RE enzyme site at future the junction with your PCR/mutated segment. Much easier than finding an entire unique 6bp where you need it. The two ends are ligated into the vector in such a way that they create a new, or regenerate the same, RE site at the junction, where you can in the next step cut and ligate in the mutated piece. Cohesive overhang RE ends can be used at the ends of the 2.6 kb to ligate into vector. If you can add or mutate some nt.s it should be easy to figure out. If not, you shop for junctions of the two "wing" segments to regenerate a RE site when ligated together that when cut at the next stage provides a site to insert the mutated segment.
I'd better stop.