“It finally got accepted!”, followed by “It’s finally out!” about a month later. I am certain this ‘finally’ feeling about their paper is very familiar to those well-acquainted with the peer review process, and it was no different for our recently published Resource article. The ‘biotagging paper’, as we call it within the Sauka-Spengler lab, is the culmination of several years’ of hard (and often frustrating) work that eventually paid off in more (unexpected) ways than one. Tatjana spearheaded the initial work for biotagging while still at Caltech, by transferring components and approaches she developed in the chicken system into the zebrafish. She worked together with Le and Tatiana, then postdoctoral fellows at Caltech, before the rest of us joined in for the lengthy optimisation, submission and review stage.
Part I: What is “biotagging”?
Biotagging is essentially an encompassing term for our Do-It-Yourself (DIY) in vivo biotinylation system for zebrafish researchers, which can be utilised in creative ways to suit specific biological needs. In vivo biotinylation was first employed in mouse (de Boer et al., 2003) by John Strouboulis when he was in Frank Grosveld’s lab and then applied for use in nuclei isolation from Arabidopsis thaliana by Roger Deal and Steven Henikoff (Deal and Henikoff, 2010). The technique was also applied to the nematode worm at around the same time (Ooi et al. 2009). The core of the technique lies in the ability of bacterial biotin ligase (BirA) to biotinylate an Avi-tagged protein-of-interest. In our binary biotagging system, the researcher decides where BirA will be expressed, which protein is Avi-tagged, and then generates transgenic lines that express these components. Crossing BirA-driver and Avi-effector heterozygous lines will give rise to ~25% of double-alleled offspring, where biotinylation of the Avi-tagged protein product only occurs in cells that also express BirA. The sky is the limit when it comes to the combinations of BirA/Avi that one can use. In the paper, we present a ‘starter’ toolkit consisting of multiple tissue-(neural crest, heart, blood) and cellular compartment-specific (ribosomes, nuclei) transgenic lines, as well as constructs to make your own lines.
Part II: Trials and Tribulations
The deconstruction (and reconstruction) of biotagging
The elegance of in vivo biotinylation means that we are not the only group to perform this method in vertebrates. For example, Michael Housley from Stainier lab (Housley et al. 2014) utilised in vivo biotinylation in zebrafish to apply the TRAP (Translating Ribosome Affinity Purification) method developed by Myriam Heiman and colleagues (Heiman et al., 2008). In vivo biotinylation experiments are not ‘difficult’ per se, but we found that obtaining a clear difference between nuclear and polyribosomal data required a remarkable amount of troubleshooting and optimisation. Our patience paid off, as this was rewarded by a wealth of information provided by a high resolution view into the migratory neural crest nascent (nuclear) and polyribosomal transcriptomes from ~200k cells.
In fact, the entire optimisation process came about by accident. In the paper, we described our surprising results when comparing the nuclear transcriptome of Sox10-positive cells at 16-18ss (migratory neural crest) to a ubiquitous control. By looking at both non-poly and polyadenylated transcripts (whole nuclear transcriptomes), our data did not yield any statistically significant neural crest-specific signature, which is what one would expect, as the enriched transcripts should be neural crest-specific. On the other hand, analysis of polyadenylated nuclear transcripts at 24hpf yielded a neural crest-specific signature. This led to further pain-staking deconstruction of our technique where, months later, we eventually came to the surprisingly simple but crucial element for the protocol to be as consistent as it is today – ensuring the complete lysis of cells (by using hypotonic buffer in excess) to release subcellular compartments into the lysate and minimise the presence of intact cell surface membranes. It is also worth noting, that a key element to the success of our protocol was the usage of an Avi-tagged chicken nuclear envelope protein, RanGAP, to label nuclei. Weirdly enough, chicken RanGAP expressed in zebrafish localised to the nuclei, but zebrafish RanGAP did not.
Having reconstructed the method, we were now eager to repeat the previous 16-18ss neural crest experiment. Imagine our initial dismay when the results were…strikingly similar. However, this was soon replaced by curiosity that drove us to carefully re-examine our results and try to figure out what IS actually going on…
Biotagging of migratory neural crest nuclei transcriptome reveals…what?
The brainstorming sessions were remarkably memorable. They were always long, often ‘lively’ as we picked at each other’s brains, and at times quite outrageous as frustrations ran high. It didn’t take us very long to notice that bidirectional transcription at non-coding regions was enriched in neural crest nuclei. However, it was a long journey after that, as we tried to quantify the phenomenon genome-wide, reproduce what we saw, believe in what we saw, and build our findings into a coherent story. Ultimately, we needed to drive home our main message – that bidirectional transcription at non-coding regions is tissue-specific, thus introducing a new method to detect active regulatory elements. These elements form the molecular signature of neural crest cells, which is traditionally based on the expression of protein-coding genes that are mainly transcription factors. We were also excited to find developmentally regulated long non-coding RNAs and transposable elements.
In short, we are proud of what we have managed to achieve with biotagging. The journey may have been long and arduous, but we have learned a lot from this project. We hope that we have provided a cool new system that includes a fully optimised tool (plasmids on Addgene) with clean protocols (available on the Resources page of our lab website), handy transgenic lines to get started with, as well as analysis pipelines tailored to biotagging datasets. Having worked out the technical intricacies of this system, this toolkit allows the zebrafish community (including us!) to study specific cellular populations in vivo on the systems level, tackling biological questions that could be important to development and disease.
Le A. Trinh, Vanessa Chong-Morrison, Daria Gavriouchkina, Tatiana Hochgreb-Hägele, Upeka Senanayake, Scott E. Fraser, Tatjana Sauka-Spengler. 2017. Biotagging of Specific Cell Populations in Zebrafish Reveals Gene Regulatory Logic Encoded in the Nuclear Transcriptome. Cell Reports Volume 19, Issue 2, p425–440