Imaging structural and functional relationships between cells and ECM

Eunice Kennedy Shriver National Institute Of Child Health & Human Development

Investigators

Abstract

This project has been initiated based on studies of lunag pathology in a mouse model of osteogenesis imperfecta (OI). By combining histological and immunofluorescence imaging of ECM with fluorescence in-situ hybridization of messanger RNA (mRNA-FISH) in the same tissue section we were able to identify that neonatal lung pathology (lethal in homozygous animals) is caused by deficient lung inflation during fetal breathing movements. We are now working on expanding the study to include confocal Raman spectroscopic imaging for quantifying the amount and organization of collagen in fibers in the same sections (since the deficient lung inflation appears to be caused deficient collagen deposition in the ECM). Additionally we are working on integration of these imaging modalities with ultrasound backscatter microscopy (UBM) of lungs in live E18.5 embryos and with quantitative spatial transcriptomic imaging based on the new Xenium and Visium HD assays from 10X Genomics. Acute respitary distress in neonates is a common feature of OI and other skeletal dysplasias, which is primarily responsible for neonatal lethality in these disorders. Traditionally, it has been attributed to rib cage fractures and deformities affecting the babies after birth but not to lung hypoplasia associated with deficient fetal breathing movements (FBMs) before the birth. Having discovered severe lung hypoplasia in newborn mice, we realized that collagen deficiency and rib cage deformities and fractures may have even more pronounced effect on lungs in utero. If confirmed, these findings may dramatically affect the treatment of newborns with skeletal dysplasias by suggesting that lung hypoplasia should be expected, immediately evaluated by neonatal pulmonologists, and treated to improve the survival likelihood and reduce secondary lung pathology later in life. The wealth of information revealed by combining histological, spectroscopic, RNA, and functional imaging of tissue sections in this project suggested the importance of further advancing and developing the technology and methods for such multimodal imaging. This project has been initiated only two years ago. Since its initiation, we have developed methodology for evaluating lung inflation during FBMs based on UBM imaging in transversal orientation. Our preliminary observations of deficient lung inflation match histological and transcriptomic findings in the same embryos. Based on these very encouraging results, we initiated a collaboration with Dr. Deborah Krakow at UCLA who routinely performs general ultrasound examination of fetal well being in pregnancies with suspected skeletal dysplasia. The standard ultrasound protocol involves evaluation of the presence of FBMs and their frequency. The goal of the collaboration is to determine whether lung inflation during FBMs can be measured as well based on methodology similar to the one we developed for mouse embryo UBM. If successful, this approach to noninvasive in utero diagnostics of developing lung hypoplasia may enable early intervention and prevention of acute respiratory distress in newborn babies. Another key development during the first year of the project has been optimization of the Visium HD spatial transcriptomics assay (10X Genomics) for integration into the proposed multimodal tissue imaging approach. We have successfully developed and tested tissue fixation, embedding, and sectioning approaches necessary for such imaging. We have also successfully developed initial computer algorithms necessary for integrating histological, immunofluorescence, and spectroscopic imaging with the high resolution spatial transcriptome analysis. This work is still in its very early stages, yet our testing of samples from mice with osteoblast-specific knockout (cKO) of SEC24D has provided very encouraging results. For instance, the multimodal imaging approach indicates that this cKO disrupts activation of osteoblast progenitors on trabecular bone surfaces, suggesting that such progenitors are mostly former, dedifferentiated osteoblasts. During last year, the second full year of the project, we have focused on improving statistical analysis of spatial transcriptomics data. Testing of the improved technology revealed that commonly used methods for analysis of both single cell and spatial transcriptomics data produced severe artifacts, which we were able to trace to unjustified model assumptions built into these methods. Our findings were in line with commonly acknowledged false findings in single cell and spatial RNASeq. In collaboration with Dr. G. Margolin, we developed a new, weighted averaging approach for data analysis without assuming anything besides randomness of technical noise. This approach is closely related to prior work on statistics of cluster-randomized experiments. We showed that weighing transcript counts based on measured noise variances and utilizing weighted rather than standard unweighted tests reduced both false positive and false negative findings. Our approach eliminates the need for parametrizing data distributions and/or rescaling transcript counts, which may cause artifacts by distorting and biasing the data. The resulting analysis is less complex and produces more consistent differential gene expression. The first paper describing our findings is currently under review. While this two-year-old project has not generated any publications yet, it has already produced findings that may completely change how single cell and spatial RNASeq data are analyzed. Since gene up(down)regulation analysis based on single cell and spatial RNASeq underlies a large fraction of all biology-related studies these days, we consider the newly developed analysis procedures to be a major breakthrough. Overall, we anticipate that this project it will have a major impact on studies of tissue pathology in our and other laboratories similar to the lung findings that prompted its initiation.