Using a second, overlap extension PCR, you can make whatever changes you want because your primers only need to anneal at the 3' end in the first PCR reaction.

In short, 3 PCR reactions are used to first generate your sequence in two halves and then to join the two products together. The first step involves mixing-matching your 4 primers into 2 reactions to amplify your halves. The primers consist of outside primers which would amplify the entire region of interest and mutagenesis primers that anneal within your sequence of interest but on either side of the site to be mutated. You can create whatever sequence you want between the sites that these mutagenesis primers anneal to your template as long as the two, mutagenesis, 5'-ends are the reverse complement of one another. These 2 "first reactions" are spiked into a third reaction, given a few cycles without added primer to extend the full length sequence, then outside primers are added back into the mix to amplify your full, mutant sequence (gel purify here for sure). This piece now has to be cloned back into your vector (use the infusion kit, and this'll be trivial).

I believe I mutated 6 bp in one quikchange reaction a long time ago, but they were right next to each other. Before learning about overlap extention, if I were mutating sites that weren't close together (within 6bp) I'd design primers to create the final product in steps. By overlap extension, you can zip together several "first" PCR products in one pot to make just about anything.

At PrimerGenesis, we have been successful at generating 8 point mutations with a single primer pair. The primer pair we used are 90 nucleotides long. Hope this helps :) Check our software for primer design here: http://primergenesis.com 

Happy mutagenesis :)

Bruno