It was in 2014 — I was a rather naïve student with limited lab experience—when I got interested in RNA biology. For my Master’s project, I wanted to broaden my experience in biochemistry, so I applied to the lab of Stefan Ameres at the IMBA in Vienna. On the first day, Stefan introduced me to Julius Brennecke with the idea that I would have two mentors and a project on small RNA biogenesis that was of interest for both groups. That day, I also met Rippei Hayashi, a post-doc in Julius’ Group who would become my supervisor in the lab.
The Piwi-interacting (pi) RNAs, play a crucial role in transposon silencing. The piRNA pathway was complicated to me, as it probably is to most people. But the thought that a large proportion of plant and animal genomes is made up of transposon sequences and that every organism has developed a strategy to deal with these potentially harmful genome intruders made the topic very exciting to me. What also helped at the beginning was that the aim of my project seemed rather simple, if not trivial, at first. My task was to identify an elusive exonuclease that makes up the 3' ends of piRNAs.
Previous work by Fabio Mohn and Dominik Handler in Julius’ lab had demonstrated that there must be an unidentified exonuclease in the fruit fly Drosophila that trims pre-piRNAs to their mature length at their 3' end (Mohn, Handler, Science, 2015). The fly genome encodes a few dozen predicted exonucleases and one of them—fittingly called Nibbler—was a strong candidate, although at that time I think only Julius was convinced about it. Nibbler was no stranger to Stefan, as during his work in Phil Zamore’s lab he identified it as a nuclease that trims some fly microRNAs to their mature length. And work from the Bonini and Liu labs further showed that loss of Nibbler also results in changes of piRNA length profiles. But the very modest length changes for only a subset of the many piRNAs did not spark a level of interest that would have sufficiently motivated a post-doc or a PhD student in the lab. For me, however, it was a perfect little project.
Drosophila ovaries generate hundreds of thousands of different ~23-30nt long piRNAs. As a first key experiment to understand the role of Nibbler in piRNA biogenesis we wanted to separately analyze piRNAs that reside in the different Argonaute proteins in normal flies and flies lacking Nibbler. Thanks to the incredible setup at the Vienna Biocenter campus with its diverse and high caliber scientific service units we generated Nibbler mutant flies by genome engineering within 2 months. Luckily, these were homozygous viable.
Having these flies in hand, I got started with experiments! In the beginning this meant dissecting large quantities of ovaries from flies, an initially challenging but then quickly a very dull task. But then, I finally could clone the small RNAs associated with the three Argonaute proteins Piwi, Aubergine and Ago3 and walked my libraries over to the deep-sequencing unit next door. It took two weeks before I got in return gigabytes of strings of As, Ts, Gs, and Cs. But these two weeks passed by quickly, as I delved into the computer. I wanted to learn bioinformatics. In the end it was my data, and I wanted to make sense out of the millions of sequences we got. I plowed through old scripts from the lab and started playing with them to generate my own versions of the various pipelines. When the sequencing of my piRNA libraries was finished, Rippei and I quickly realized that only piRNAs loaded into Ago3 showed a clear increase in their mean length in Nibbler mutants.
This was quite interesting as the previous work from Fabio and Dominik suggested that the endonuclease Zucchini, which also can generate piRNA 3' ends acts mostly on Piwi and Aubergine, but much less so on Ago3. Did Nibbler act downstream of Zucchini? Or could there be two separate ways for piRNA 3' end formation with Zucchini and Nibbler being the key enzymes for the two pathways? We set up a classical genetic epistasis experiment and generated flies lacking both Zucchini and Nibbler. If Nibbler would act downstream of Zucchini, loss of both factors should phenocopy the situation in Zucchini-only mutants. This was, however, not the case. Instead, many piRNAs entirely disappeared, in agreement with the two-pathway model. But in addition, something really surprising happened: The levels of many piRNAs were barely affected, at first arguing against our model. But when we carefully looked at these piRNAs we realized that they have drastically different length patterns. Mapping these piRNAs to the genome revealed that they were spaced directly next to each other, arguing that two cleavage events mediated by piRNAs themselves generate their 5' and their 3' ends. Now, it all made sense. Nibbler and Zucchini act in two genetically separate pathways, which can partially compensate for each other. And in the absence of both pathways, a third piRNA biogenesis system based on Argonaute proteins alone becomes unvieled.
Now, nearly two years after I stepped into IMBA, the impact of an initially small question becomes apparent: piRNA 3' end formation is a remarkably flexible process. Based on a true collaborative project with Rippei, we have provided a comprehensive roadmap for piRNA biogenesis. I anticipate that our work will guide the mechanistic dissection of this exciting area of small RNA biology.
See also story featured in the IMBA website