I think the most elegant and flexible solution would be to insert a pair of opportune single restriction sites flanking the GFP gene. Use only silent mutations, if you can, otherwise introduce as few mutations as possible and verify the correct expression of the plasmid before proceeding. You can use standard fusion PCR or something like the Quikchange kit (Agilent) to introduce your restriction sites.

One you have them in, you can easily insert/remove any gene you want by using primers containing the necessary restriction sites.

Good luck

Excising an unwanted gene from a genome or vector can be done by recombinases. But the problem at hand here is not mere excision but replacing the older gene with a new one, which ofcourse can be done by recombinanses again. But the cheaper and elegant technique would be heterostaggered PCR strategy. This technique requires some four sets of primers designed meticulously. I think this article  can help you in this regard.http://www.ncbi.nlm.nih.gov/pubmed/22917676. All the best.

I agree with Andrea, just design on the computer what you would like to have, and build it with your own PCR primers . You may not even need restriction sites if you use PCR assembly with 4 primers. Whenever you have plasmids encoding fusion proteins that don't have unique restriction sites to cut and replace portions, it is easier start from scratch and build exactly what you want. If you need help with the primer design, just ask.

How about using SOE (splice overlap extension) PCR?

You can also use cloning by homologous recombination in S. cerevisiae.

So the thing is I don't have any restriction site at the end of GFP. How am I going to insert any construct or Restriction site ? Will this QuickChange kit introduce a restriction site onto either end of GFP ? so afterwards, I can clone RFP in place of GFP easily?

One thing I forgot to mention, my GFP in the plasmid has one restriction site at 3' end 

Hi Naushad,

Using SOE PCR eliminates the need for a restriction site between the target gene and the tag sequence, although you might need to consider a flexible linker between the two.

Do you have a suitable 5' site adjacent to your gene of interest as well?

yes i have but it doesn't have any restriction site. I guess i would need to study more on SOE PCR

i read abt SOE PCR but If i m right, it is used to anneal two sequences together which are not homologous or do not have any restriction sites. In my case, i need to remove GFP and get RFP running into the plasmid. 

As you mentioned that GFP in the plasmid has one restriction site at 3' end.  I would like to use the Quikchange kit (Agilent) to introduce a proper restriction sites at 5' end near GFP.  This is a simple and easy way to dissolve your problems when you want to replace it with YFP or CFP or RFP or whatever else you want.