Thank you!!

Hi Sho! Thanks! The video was made by Phospho Biomedical Animation. They have an amazing team! www.phospho.co.uk

🙌 Kudos to Carsten from @zeiss-microscopy.bsky.social who helped fs-laser milling of the sample, and Manuel, Andrea and @normanrz.com from @webknossos.org for help with large data storage, viewing, tracing and segmentation.
It has been an honour for me to work with you all together on this project!

🙌 Kudos to X-ray physicists Anne from @psich.bsky.social, Peter and @apacureanu.bsky.social from @esrf.fr for beamtime support and Alfred and Jayde from @esrf.fr for developing XNH denoising.

🙌 Kudos to members of the schaefer lab @tobiasackels.bsky.social now PI at @unibonn.bsky.social who acquired pilot datasets, and Sina and Mihaly for help in modelling and analyzing odorant databases.

Massive kudos to the team! 🙌
To @carlesbosch.bsky.social who started the correlative workflow. To the supervisors of the project @carlesbosch.bsky.social, @apacureanu.bsky.social and @andreas-t-schaefer.bsky.social who saw the potential of synchrotron X-ray imaging and pushed for its development.

With 100 nm voxel size, we further utilised the resolving power of XNH to trace lateral dendrites of sister mitral/tufted cells. Intriguingly, both soma location and lateral dendrite architecture explain some of the observed diversity in tuning.

We simulated a toy model, and found that when sister cells show ‘balanced diversity’, they have high encoding capacity as well as high generalization ability, compared to when sisters are identical or completely uncorrelated.

However, some sister cells consistently differ in temporal dynamics to others across repeats. We named the observed similar but non-identical tuning ‘balanced diversity’. What could this imply in terms of olfactory processing?

The tuning correlation remains high even after we took away the strongest activating odours.

The correlative 2P-XNH workflow is also very robust. Here, we present results from 3 samples from 3 animals. A first in correlative structure-function studies of sub-mm3 brain volumes.
Across the 3 samples, we saw that sister cell odour tuning is by large very similar in anaesthetized mice.

After in vivo 2-photon, we imaged our samples with XNH, found the imaged cells and traced their apical dendrites.
Our best sample contained 289 sister cells from 50 glomeruli - that’s >2400 sister pairs. XNH gave the most unequivocal sister identification and we got the largest sister cell datasets.

Here, we used synchrotron X-ray nano-holotomography to identify sister cells. XNH is an emerging structural imaging method. It’s fast, non-destructive and intrinsically volumetric - so no slicing involved and no worries over registering images!

This long-standing question had been studied with pioneering approaches by
Tan 2010 doi.org/10.1016/j.ne...
Dhawale 2010 doi.org/10.1038/nn.2...
Kikuta 2013 doi.org/10.1016/j.ne...
Arneodo 2018 doi.org/10.1038/s414...
But sister cell identification remained a key challenge and the yield was low.

In mice, olfactory sensory neurons expressing the same olfactory receptor converge to ~2 glomeruli per bulb. The circuit then diverges, with a few 10s of sister mitral/tufted cells receiving inputs from each glomerulus.
Why does the circuit diverge?
Are sister cells redundant or functional?