A very big thank you to all authors on this study, but especially to @jensucre.bsky.social, @jkropski.bsky.social, @davisjmcc.bsky.social, @premiepulmdoc.bsky.social, and many more people for making this work possible! Excited for what's to come 🫁🧬
(17/17)

Our dataset is publicly available on GEO (GSE297945), and we have a spatial cell atlas to help visualize these data: lungcells.app.vumc.org/public/sucre...
(16/17)

Overall, this work establishes new frameworks for studying disease pathology in neonatal lung tissue, and can allow us to eventually identify targetable therapeutics to help premature infants bolster proper lung development!
(15/17)

Important to note - the above results are between cell types in the epithelial, endothelial, and mesenchymal lineages, excluding the immune lineage! Read the preprint to find out why!
(14/17)

When we intersected the results from all three terms (gestational age, life span, and DX score), we found 5 gene - cell type pairs overlapped between gestational age and life span; 2 are shared between gestational age and DX score; and 1 between life span and DX score.
(13/17)

Here, we found 18 gene expression changes associated with gestational age, 25 associated with life span, and 10 genes associated with disease severity (FDR < 0.1).
(12/17)

This regression model included terms for GA & LS. We then worked with Dr. Gail Deutsch at Seattle Children’s Hospital to assign a disease severity (DX) score for each tissue (represents the overall degree of damage observed) and used a similar regression framework with a term for DX score.
(11/17)

So, we established a framework that would account for spectrum of injury instead of using a ‘control vs. disease’ grouping schematic with linear regression to identify gene expression changes with gestational age (GA) and life span (LS) (LS = proxy for the degree of injury).
(10/17)

We then wanted to study the gene expression changes associated with both development and injury. However, every infant born prematurely experiences different levels of neonatal lung injury depending on a variety of factors (exposure to oxygen, mechanical ventilation, steroid treatments, etc).
(9/17)

One representative example is TNiche 1, which represents the alveolar niche and is composed of the key alveolar cell types, including AT1, AT2, and capillaries, and we can see that this niche increases in abundance across gestation, consistent with alveologenesis throughout lung development!
(8/17)

Then, we computationally identified 11 multicellular spatial “niches” using both Seurat v5 and an transcript-based approach developed by Dr. Ruqian Lyu in @davisjmcc.bsky.social’s group (www.biorxiv.org/content/10.1...). Both approaches found generally concordant results.
(7/17)

As an example, we find capillaries and alveolar type 1 (AT1) cells are unlikely to occur next to one another in samples from the early/late canalicular periods; but in samples from the alveolar stage of development and the adult BPD sample, they are more likely to be close to one another.
(6/17)

We then defined the cell-cell interactions across development, and to do this, we used cell proximity to quantify the likelihood that cell type pairs were in close proximity to one another within each developmental stage. This creates a map of cell interactions across the developing lung.
(5/17)

We were able to identify 40 unique cell populations across the 4 lineages, using canonical hallmarks established in previous studies. We noticed differences in cell type proportions as gestational age increased, reflecting different cell-specific patterns as development progressed.
(4/17)

With @10xgenomics.bsky.social, we profiled the expression of 343 genes from 17 human infant lung samples across a range of gestational ages and disease pathologies. We generated a ton of data (76 million transcripts in 1.2 million cells) - the largest spatial human neonatal lung dataset yet!
(3/17)

Human lung development requires carefully coordinated cellular communication to create sufficient surface area that can support gas exchange, and the exact molecular mechanisms and spatiotemporal relationships of various developmental pathways during lung development is unknown.
(2/17)