Maybe you try the protocols published in:

http://www.ncbi.nlm.nih.gov/pubmed/18314717

http://www.jstor.org/stable/3283741?seq=1#page_scan_tab_contents

Thank you! I'm waiting on the top article to arrive via interlibrary loan, and have tried the miniprep isolation, but never quantified my results (I just recently designed a qPCR to quantify the ratio of genomic DNA : mtDNA). 

Why do you want to isolate it first, just use specific primers on total DNA  - or afraid of nuclear pseudogenes ?? I have designed a method long time ago, try this. Good luck, RW

PCR is your best bet. mtDNA isolated from mitochondria contains up to 50% nDNA, even after  treatment of isolated mitochondria with DNaseI.  Therefore, mtDNA isolation from mitochondria is normally followed by CsCl gradient purification of supercoiled mtDNA, or at least by exonuclease I digestion.  However, it will be difficult to procure enough worms to sustain the scale required for CsCl gradient,  perhaps, more difficult than to optimize PCR.  You don't have to ampify your mtDNA as a single fragment.  We typically work with two overlapping fragments, and went to as many as four for difficult species.

Hi Caroline,

I'm not sure this is appropriate for your project, but I think that, rather struggling to obtain pure mtDNA, you might want to think about an approach of target enrichment through hybridization capture and ultra-deep sequencing. 

You can look at Maricic et al. (PLOS ONE, 2010; http://journals.plos.org/plosone/article?id=10.1371/journal.pone.0014004) for a description of the approach for mtDNA. The approach is simple and robust and it has been applied successfully to ancient DNA and to museum samples.

The utility of this approach will depend on  the scale of your project and on the fact that you have at least some long PCR products to capture the entire mitochondrial genome. 

In practice, you start with a couple of long PCR, ideally spanning the entire mitochondrial genome and then you fragment the amplicons. Subsequently the fragmented PCR products are biotinylated and immobilized on streptavidin-coated to produce home-made immobilized capture probes (baits). The baits can be used to capture mitochondrial DNA from total DNA extractions of different specimens and the captured DNA can be used for next generation sequencing. Typically the captured DNA of each individual is removed from baits, indexed with a specific barcode and different samples/specimens are multiplexed in a single Illumina lane.

The drawback of the approach is that, if using different species, hybridization capture can work badly in the species that are distantly related from the ones from which you obtained the probes (but see http://www.ncbi.nlm.nih.gov/pmc/articles/PMC3202286/).

 I hope this helps.

 Cheers

 Lorenzo

I guess I should preface this question with the fact that I've been trying to optimize long PCR on my samples for almost 3 years... If anyone has any insight into designing primers better, I would appreciate it. I have sequenced anchor genes (cox1, cox2, several NAD genes, ATPase, etc) and attempted to amplify the genome using primers designed based on those. PCR either fails or produces non complete amplicons. I have optimized long PCR for 2 species, but it only works about 20% of the time (rough estimate). It's hard to do this without any reference, as the gene order varies between species so I don't have a good gauge of the expected amplicon sizes, thus struggle with anneal/extension times. I have tried BD advantage long polymerase, long amp taq, and Takara, and only had success with Takara polymerase.

Hi Caroline,

I do not have lot of experience with long PCR, besides being advised by collegues that Takara is the polymerase of choice.

Anyway, if you can obtain long PCR products from two species with 20% success, possibly this is enough to use those products for capture the mtDNA from the other species with the streptavidin coated beads bait apporach and then sequence with Illumina.

Gene order would not be a problem since PCR product used as probes are fragmented to 150-800 bp before use them as probes.And it works with ancient DNA!

By the way, if you do not want to be involved in the target capture protocols, you should be able to recover contigs coverging the entire mitochondrial genome by simply directly sequencing total DNA with Illumina platforms and enough coverage.

See this for instance: http://www.biomedcentral.com/1471-2164/13/114

Ciao.

Lorenzo

Thank you all for your help!