Paper: A role for BCL2L13 and autophagy in germline purifying selection of mtDNA

Back in September I wrote about a joint project with Laura Kremer on the role of autophagy in purifying mtDNA selection. In summary, Laura designed a gene knockout experiment in order to determine whether targeted mitophagy contributes to germline selection against m.5024C>T in mice. I was responsible for analysing the resulting data. Ultimately we showed that BCL2L13, a homolog of the better-known yeast mitophagy receptor ATG32, is involved in mtDNA selection.

Working with heteroplasmy data, especially in mouse litters, is tricky. We do not fully trust the pyrosequencing measurements we rely on, pups born to different mothers are not directly comparable, and measurement errors in the mothers (often barely five or six per group, even in a large experiment such as ours) can propagate and skew results. A lot of my work was about trying to be very critical of the data and to ensure the results we report are robust and reproducible. Laura and colleagues were really supportive in this, and the final manuscript included a detailed description of the various subsampling experiments I performed to ensure reproducibility, as well as a discussion on why simpler out-of-the-box methods can be misleading. We hope our work not only elucidates the role of mitophagy in mtDNA selection, but also provides other researchers with a new, more analytical way of thinking about heteroplasmy data.

Our paper was accepted in PLOS Genetics just before Christmas, and it can now be seen online here. I will not rehash the main text, but I want to give a shout-out to our peer reviewers who were really engaged and helped us improve our work. They encouraged us to highlight the inter- and intra-mother variability in our dataset, and also to investigate the potential effect of the C57BL/6J genetic background on our results (we confirmed there was none).

Peer review often gets a bad rap, and being anonymous on the internet definitely can bring out the worst in people. But when it works, it is a pretty great part of the process.