19/ Remarkably, we turned off ANG4 expression in the middle colon. Using our Xenium panel, we showed the loss of the middle colon-adapted goblet cell population, demonstrating that ILC2s are essential for actively maintaining this adaptation.

18/ Intrigued by these results, we reached out to David Artis and his group @WeillCornell to test if ILC2s drive the middle colon’s microbiota adaptation. I drove down to NYC with my tools and together with the talented @hirothehero.bsky.social we turned the lights off on the mid colon.

17/ Together with my officemate of 4yrs and close friend Eric Brown we were then able to show ILC2 derived signals could induce the middle colon signature in ex vivo explants.

16/ We leveraged our sc data to compare all immune cell types across regions between SPF and GF mice and found only ILC2s showed a unique microbiota driven signature in the middle colon (B) which we could also validate with flow cytometry for a variety of markers.

14/ Using our data to design a customized 480 gene panel we then leveraged high resolution Xenium technology to map all our cell states in space and validated the regional association of cell states as well as the middle colon adapted cells.

13/ We found unique structural cells defined each region (enterocytes, fibroblasts, and goblet cells) and strikingly the middle colon associated region (B) had unique subsets only found in the presence of the microbiota.

12/ To uncover the cellular drivers of the gut’s spatial landscape and the middle colon adaptation we used scRNA-seq. Using a customized extraction protocol we profiled 99 cell states of epithelial, stromal and immune cells in one go from four distinct regions of the colon.

11/ We then tested if spatial gut circuits were resilient to inflammation using a spatiotemporal model of acute colonic inflammation. Both robust and adaptable landscapes were disrupted at peak inflammation but recovered over time, showcasing their resilience.

10/ We did find an exception tho: the middle colon adapts! Its identity depended on the microbiota, marked by genes like Ang4, Retnlb, Itln1, Pnliprp2, Aqp4, and Slc9a3. ANG4 protein staining showed exquisite specificity for the middle colon, serving as a biomarker in later exps.

9/ Ok, so how does the landscape respond to the microbiota? We found in germ-free mice, most genes maintained regional expression bias, even classical antimicrobial genes—suggesting a surprisingly robust and microbiota-independent landscape (the same applied to circadian rhythm).

8/ First surprise: distal gut regions shared unique expression patterns! Quantifying this revealed the duodenum shares many unique genes with the colon, hinting at unexpected redundancy in biological processes—highlighted again for SLC transporters.

7/ With these tools we were then able to map the continuous spatial transcriptome in both the X-axis (duodenum to colon) and Y-axis (serosa to epithelium) allowing us to interrogate the distribution of gene expression for any biological theme of interest (SLCs example).

6/ Next challenge - reconstruct the tissue to its native state and unify segments within and across mice to facilitate downstream analysis and visualization. Enter the room @lichenhao.bsky.social, a brilliant scientist that helped me supercharge the study along with my computational skills.

5/ The idea: use the gut swiss roll (before you started to see them everywhere) and ideally with fresh frozen tissue. With Toru Nakata (best hands I worked with), we figured it out and with Asa Segerstolpe, profiled 40+ rolls using 10xGenomics Visium. #SpatialBiology