21/ 🎯 Bottom line: Net1 mRNA has to reach the midbody for cells to complete abscission efficiently, and Net1 protein is a new player in building the branched actin network that makes abscission, and faithful cell division, possible.

20/ We tested this by measuring Arp2/3 buildup at the midbody across our cell lines. Arp2/3 accumulation was lost in Net1 knockouts, restored by the full-length UTR rescue, and NOT restored by the ΔLE rescue, exactly what our model predicted.

14/ But what's the mechanism? Does having Net1 mRNA at the midbody actually build up more Net1 protein there? Using our knockout/rescue system: yes! Only Net1 protein made from RNA containing the LE showed up in meaningful amounts near the midbody.

12/ So where exactly does the cell cycle stall when Net1 RNA can't get to the midbody? Net1 knockout cells linger too long in telophase, the last stage of mitosis, pointing to a problem with abscission itself. The full-length UTR transgene fixed this. The ΔLE version didn't.

11/ To confirm this wasn't a fluke, we made a Net1 knockout line and rescued it with Net1 transgenes carrying either the full 3' UTR (traffics to the midbody) or a version missing the LE, "ΔLE" (doesn't traffic). Again, the LE was required for cells to divide efficiently.

10/ Now we finally knew what to target! We used antisense oligos (ASOs), short synthetic sequences that bind to and block a specific stretch of RNA, against the LE. They blocked Net1 RNA from reaching the midbody. And, unexpectedly, cell division slowed way down! ⏳

9/ One region in the middle of the UTR stood out. Oligos from it were enough on their own to send the reporter to the midbody. We called this the "localization element" (LE). Removing just this region also blocked transport, so it's both sufficient and necessary.

8/ Okay, but which sequences *within* that 3' UTR are doing the work? To find out, we used a massively parallel reporter assay (MPRA). We tiled ~500 short DNA pieces across the Net1 3' UTR, put each into a reporter RNA, and measured how well each one reached the midbody.

7/ Many "zip code" sequences that direct RNAs to specific places live in the 3' UTR, the tail end of an RNA that doesn't code for protein. We fused Net1's 3' UTR to a reporter RNA and tracked it by RT-qPCR. Net1's 3' UTR alone was enough to send the reporter to the midbody.

6/ Luckily, we can isolate the RNA contents of midbodies and compare them to whole cells. We and others had already found that specific RNAs are shuttled to the midbody. We started with one of the most enriched RNAs, Net1, and asked: how does it get there?

4/ Some background: the midbody recruits a series of proteins to carry out abscission. But are RNA molecules recruited too? And if so, does it actually matter? Does having RNA there help the midbody do its job?

Excited to share our newest work! We asked whether RNA molecules need to be in a specific place in the cell to do their job. We found that a single RNA must reach the midbody, the structure that pinches two dividing cells apart, for cell division to finish efficiently. 🧵⬇️

18/ With this pool, critical-element mutations stayed lethal, but we never saw mutations in a stem rescued by compensatory mutations. No clear evidence that these specific structures drive activity.

17/ To test the functionality of those structures, we built another pool. We mutated every stem, plus made structure-preserving "compensatory" mutations that should restore the fold.

16/ Elements this big made us wonder if structure matters. With Chase Weidmann, we solved the secondary structures of the 260 nt elements using SHAPE-MaP.

15/ Again, shuffling the critical element killed activity. But the support elements tolerated even wide shuffling windows. So they work through bulk nucleotide content, with exact nucleotide order mattering much less.

14/ Does the exact order of nucleotides matter, or just the overall content (how many adenosines, etc.)? To test this, we made thousands more variants where we kept the content of a window fixed but shuffled the order.

13/ Same story. Any mutations in the critical element killed activity. The support element tolerated small mutation windows but not larger ones.

12/ Could it be that deletions just throw off the spacing between key parts? To check, we left the length intact and instead slid a window of mutations across the element.

11/ These critical elements aren't just a quirk of our workhorse neuronal cell line. With Michael Kiebler's lab, single-molecule imaging showed they're also essential for shipping RNA to axons in primary hippocampal neurons.

9/ We call these "critical elements." Necessary, but not sufficient on their own, which fits the picture that the full, active element is just really big.

8/ Round 2: necessity. Charlie made thousands of small deletions across the 260mer. This exposed tiny regions that are absolutely essential. They won't tolerate losing even a few nucleotides.

7/ Surprisingly, only pieces keeping 200–250 nt of the element still worked. These localization elements are therefore HUGE, way bigger than the elements that control splicing or translation.

6/ To investigate this, Charlie built ~50,000 mutant versions of these elements and tested them with MPRAs. Round 1: chop the 260mers into thousands of smaller pieces and ask which ones still send RNA to neurites.

4/ A few years back we paired this with massively parallel reporter assays (MPRAs) and found ~260 nt sequence elements in the 3′ UTRs of several genes that are both necessary AND sufficient to send an RNA to neurites. academic.oup.com/nar/article/...

Maybe I’ll keep tidying my own dataframes a little while longer.

Announcing the 2026 edition of the EMBO workshop on RNA localization and local translation! This meeting will be held June 30 - July 4 near Porto, Portugal. Come for exciting updates in the field from both established investigators and trainees. See the link below for details!

How does one review his paper if he’s infallible?

If we replace wildtype TDP-43 with an ALS mutant TDP-43, we again see that the same RNAs become aberrantly neurite-enriched. So this could be happening in ALS patient cells, but whether it makes contributions to patient phenotypes is another question for another day.

Does any of this have anything to do with ALS? Maybe. We see the same RNA localization defects happening in primary mouse motor neurons and human iPS-derived motor neurons upon TDP-43 loss.