It depends what you want to do. Normally, you don't need to isolate the organelles and then the DNA. 'Standard' protocol would be to isolate total DNA, then use a CsCl gradient to separate the organellar from the nuclear DNA. But then again, it depends on your organisms as plastid containing organisms will have the additional plastid genome. If you are 'only' after mitochondrial genes, then I would use PCR. So, if you provide a bit more info and what your aim is, I might be able to give a better answer.
thank you ..Actually I want to do sequencing of mtDNA. But PCR based method has a problem... There is some report that human genome has mtDNA insert.So may be I will get mixed up target enrichment...is there any suggestion to avoid that?
As the human mitochondrial genome is rather small, you could use PCR to amplify it and subsequently sequence the PCR product. Because there are so many mitochondria in most cells, using whole cells for PCR might already work without the hassle of isolating DNA first. As I don't work with human mitochondria, I am not sure what this mtDNA insert is? How big is it? As far as I know, human mtDNA has always been reported as ~15 kb. Not sure why you want to sequence human mtDNA but is that insert not informative for your work as well?
Is the insert one piece? If so, you could design outward facing primers that amplify the rest of the mtDNA but not the insert. You need long-range PCR though but should be possible. Obviously, there are errors involved in PCR so use high-fidelity enzymes. Of course, sequencing is not error-proof either. Also, it all depends on number of samples but if you are interested in inter-individual changes than I assume that is more than one sample anyway. Not sure if this has been helpful but this seems the easiest approach for me. Good luck.