Parkinson’s disease is a partially complex genetic disease with roughly 15% of the cases being familial and early onset. A half dozen or so genes have been linked to the familial form of PD and collectively explains a third of the heritability in this category, which, by the way, is a pretty good number comparing to other degenerative disorders.
Parkin is the most common disease genes found in familial PD and not surprisingly has been the focus of the field for quite some time. A central mechanism regarding parkin function after a decade or so work is that, parkin, when activated by its upstream regulator PINK1, a Ser/Thr kinase, is translocated to mitochondria, where it is believed to instruct mitophagy and alleviate cellular stress.
It reminds me of my first real scientific activity, a threesome project, which involved biochemical interactions between three PD proteins: Pink, Parkin and Parl. Parl is a mitochondrial-resident protease. Its action on PINK is required for PINK1 localization and activity.
Since we know very little about these proteins, people have thrown virtually any type of high throughput analyses onto them, genetic screens in flies, proteomic networks, enzymatic assays searching for substrates and consensus binding motifs, you name it. This paper from Rick Youle’s group is yet another study fell in this category: a genome-wide RNAi screen.
Rick Youle has always been the leader in the parkin field and this particular RNAi screen assay is just another excellent example of how one should do a genome-wide study. The assay is very well designed, targeting specifically to the step of parkin translocation to mitochondria. The algorithm for automated detection of the translocation phenotype is clever; the controls are clean and complete; it has by far the most thorough and well-explained statistical analysis I’ve ever seen in a benchwork-dominant biological research paper.
And perhaps what I think it’s of particular value from this paper is that it’s not one RNAi screen. It’s actually two: one pooled siRNA library (a 4-clone pool per transcript) and one single siRNA library (3 individual clones per transcript) from two sources. RNAi has long been suffered from its off-target effect and non-predictable potency. Pooled RNAi minimizes the off-target effect, while individual siRNA gives confidence and replication when searching for a true candidate. And different algorithms in designing siRNA from the two companies sort of complement each other too.
However, despite such comprehensive and expensive effort, they virtually yield no result. Most hits from the screen turned out to affect parkin translocation due to their effect on PINK1 expression and/or activity. ~17% of the siRNA hits bind to the 3′ UTR of PINK1 and potentially downregulate PINK1 expression. This is a non-heard-of quasi off-target effect. And the proteins they found from the screen in fact regulate PINK1 expression, localization and activity for the most part.
Moral of the story? Big data sets can have little information. And a lot of times in biology, when finishing a project, you find yourself miles away from what you set out to find. Ain’t that great?
