Spatiotemporal transcriptomic atlas of mouse organogenesis using DNA nanoball-patterned arrays

In the paper

“For example, the epithelial cell cluster ( Epcam + , Krt8 + ) could be grouped into specialized epithelial cell subtypes including hair follicle cells ( Krt5 + , Krt17 + ), alveolar type II cells ( Foxp2 + , Nkx2-1 + ), club cells ( Scgb3a2 + , S100a6 + ), and ciliated cells ( Dynlrb2 + , Foxj1 + ) from the lung”

In the paper

“There was substantial colocalization of the dye and the aggregated transcripts detected by Stereo-seq (Figures 2A and 2B), illustrating that cells can be segmented in an image-dependent manner.”

In the paper

“After filtering out the low-capture cells, we obtained 50,140 segmented cells with an average of 1,910 UMI and 792 genes per cell (Figure S1F).”

In the paper

“For E9.5-E15.5 stages, four to six sections were included from different embryos. For E16.5, 18 sections were profiled, with a series of 13 sections from one single embryo and five from another to assess reproducibility, allowing coverage of all major tissues and organs (Table S1).”

In the paper

“Among the E16.5 regulons, ARNT2 was enriched in the brain, FOXC2 in the meninges, GATA4 in the heart, SOX9 in the bone, and MEF2C in the muscle (Figure 3E).”

In the paper

“Importantly, Stereo-seq has a larger number of spots per 100 m m 2 , with smaller spot size and center-to-center distance, than any other published method (Figure 1B).”

In the paper

“For example, at E9.5, we identified 31 modules containing a total of 498 regulons, and at E16.5, we identified 35 modules containing a total of 469 regulons that correspond to different tissues (Figures 3D, S4F, and S4G).”

In the paper

“For example, in the E13.5 spinal cord, we identified the ventricular zone ( Hopx + ), marginal zone ( Slc5a7 + ), basal plate ( Vsnl1 + ), ventral ( Fut9 + ) and lateral parts of the spinal alar plate ( Hoxb8 + ), and superficial stratum of spinal basal plate ( Pdyn + ); in the E16.5 kidney, we localized artery ( Myl9 + ), cortex ( Lhx1 + ), medulla ( Nrxn3 + ), capsule ( Robo2 + ), and glomeruli”

In the paper

“Here, we report the development of a DNB-based genome-wide technology, Stereo-seq, that combines single-cell resolution, high sensitivity, and a large field of view”

In the paper

“Theaveragenumberofcapturedgenes per bin ranged from 1,770 at E15.5 to 3,900 at E10.5 and of UMI from 4,357 at E15.5 to 13,789 at E9.5 (Figure S2A; Table S1).”

In the paper

“In addition, since Stereo-seq captures high density of signals, we performed spatial reclustering of bins from areas corresponding to selected tissues, which revealed detailed subregions.”

In the paper

“Reclustering of specific cell populations could further identify cell subtypes or cell states.”

In the paper

“After filtering out the low-capture cells, we obtained 281,377 segmented cells with an average of 1107 UMI and 529 genes per cell (Figures 4A and S5A).”

In the paper

“These cell types were either located in specific anatomic regions (e.g., cardiomyocytes, hepatocytes, and different types of neurons) or represented common cell types spread across multiple sites (e.g., chondrocytes, epithelial cells, myoblasts, and endothelial cells).”

In the paper

“Thedistribution of genes and UMI perbin varied between sections from different time points (Figure S2B).”

In the paper

“So far, we have used Stereo-seq chips of up to an effective area of 13.2 cm 3 13.2 cm for different tissue sizes, subdivided into 50, 100, and 200 mm 2 chips for profiling sections from the mouse olfactory bulb ( 10.5 mm 2 ), a mouse hemibrain ( 24.2 mm 2 ) and whole mouse embryos ( 7.1 to 76.1 mm 2 ), respectively (Figure S1A).”

In the paper

“The distribution of specific markers ( Pcp4 and Slc17a7 ) captured by Stereo-seq exhibited the expected similarity to in situ hybridization (ISH) data taken from Allen Brain Atlas (ABA) (Lein et al., 2007) and clearer patterns than HDST (Vickovic et al., 2019) or Slide-seqV2-resolved (Stickels et al., 2021) expression data (Figure S1E).”

What the sources say

• Not checked yetcites Lein et al., 2007 source
• Not checked yetcites Vickovic et al., 2019 source
• Not checked yetcites Stickels et al., 2021

In the paper

“The resulting gene programs showed well-defined spatial patterns that, to a large extent, match anatomic regions and contained sets of genes that could be validated by ISH (e.g., Kitl colocalizing with Shh in the E9.5 notochord and Pantr1 with Sox2 in the E11.5 brain) (Figures S4A-S4E).”

What the sources say

• Not checked yetcites DeTomaso and Yosef, 2021

In the paper

“Standard DNB chips have spots with approximately 220 nm diameter and a center-to-center distance of 500 or 715 nm (Figure 1A, step 1), providing up to 400 spots for tissue RNA capture per 100 m m 2 .”

In the paper

“We observed high Wnt5a expression in mesenchymal cells and fibroblasts from the maxilla, whereas in the limb, it was located mainly in mesenchymal cells (Figures 6C and 6D).”

In the paper

“HES6 was highly active in neuroblasts, and its target genes included Ascl1 and Hes5 (Figures 5L and S7G).”

In the paper

“We also observed unknown potential regulators ( Ttc3 and Celf4 ) of interneuron migration that could be interesting for further exploration (Figure 4F).”

In the paper

“In addition, we identified highly specific expressions of extrinsic cues (ligands) in glioblasts (e.g., Rspo1/2/3 , Wnt1 , and Wnt9a ) and neuroblasts ( Fgf15, Wnt7b, and Tnc ) (Figures 5G-5I and MOSTA website), providing potential mechanistic insight for the further specification of these progenitors.”

In the paper

“Moreover, Stereo-seq is an affordable technology.”

In the paper

“At E14.5, the radial glia cells disappeared from the rostral part and remained in the more caudal domains of the ventricular zone, revealing the spatiotemporal development of this brain region.”

In the paper

“We observed that Wnt5a localized in the craniofacial region, limb, genital ridge, and brain (floor plate and neocortex) from multiple developmental stages (Figure 6B), consistent with the disease phenotype (Menezes et al., 2010).”

What the sources say

• Not checked yetcites Menezes et al., 2010

In the paper

“The resulting developmental trajectories revealed two branches arising from radial glia cells, one to neuroblasts andthe other to glioblasts (Figure 5D).”

In the paper

“The high resolution of Stereo-seq allows efficient image-based cell segmentation, but in some cases, where multiple cell types are close to each other, particularly smaller cell types like immune cells, cell segmentation may be imperfect.”

In the paper

“For example, HES1 was highly active in glioblasts, and its target genes included known genes specific to the glia lineage such as Aldh1l1 and Nfib .”

In the paper

“Radial glia cells ( Mki67 + , Dbx1 + ), which are the source of both neuroblasts ( Npas3 + , Adgrb3 + ) and glioblasts ( Aldh1l1 + , Efna5 + ), were the most abundant progenitors at E12.5, with widespread distribution.”

In the paper

“Similarly, the MSX1 regulon was more enriched in mesenchymal cells and fibroblasts in the maxilla, and mesenchymal cells in the limb (Figures 6C and 6D).”

In the paper

“Msx1 was expressed at different levels in multiple cell types of the maxilla including mesenchymal cells, but in the limb, it was rather restricted to the latter cell type only.”

In the paper

“The current DNB capture chip costs 220 RMB/mm 2 ($35/ mm 2 ) or 22,000 RMB/cm 2 ($3,500/cm 2 ).”

In the paper

“Similarly, although many genes were mainly enriched in one tissue (in some cases, they were confined to a specific region) or system, others were more widely distributed.”

In the paper

“Our E16.5 Stereo-seq map clearly identified the expression of interneuron progenitor markers ( Dlx1 + , Dlx2 + , Lhx6 + , Nkx2-1 + ) (Lim et al., 2018) within the medial ganglionic eminence (MGE) and the migratory trajectory of interneurons into the cerebral cortex (Figures 4D and S6D).”

What the sources say

• Not checked yetcites Lim et al., 2018 source

In the paper

“In this regard, over 10% of the Stereo-seq captured transcripts contained intronic reads (Figure S6E), comparable with the proportion captured by scRNA-seq (La Manno et al., 2018; Qiu et al., 2022b).”

What the sources say

• Not checked yetcites La Manno et al., 2018 source
• Not checked yetcites Qiu et al., 2022b

In the paper

“Similarly, Stereo-seq has a genome-wide coverage, but due to limitations in capture, relevant genes with low expression may be missed.”

In the paper

“Some genes were expressed in most or all embryonic stages, whereas others had a restricted window.”

In the paper

“At this stage, there was only occasional appearance of sparsely distributed neuroblasts, whereas glioblasts localized mostly in the central and caudal parts.”

In the paper

“Although the function of many of the transcription factors controlling these regulons has been previously reported, we have identified the spatial domains in which they act.”

In the paper

“Importantly, Stereo-seq also captures intronic transcripts, making it possible to calculate spatial RNA velocity with singlecell resolution, which as in scRNA-seq studies (La Manno et al., 2018; Qiu et al., 2022b), can facilitate the study of cell fate transitions.”

What the sources say

• Not checked yetcites La Manno et al., 2018 source
• Not checked yetcites Qiu et al., 2022b

In the paper

“As informative examples, NKX2-1 , which is mutated in brainlung-thyroid syndrome (Shetty et al., 2014), was mainly enriched in brain (hippocampus and subpallium) and lung and Lmod3 , which is associated with nemaline myopathy (Berkenstadt et al., 2018), was enriched in the skeletal muscle (Figure 6B).”

What the sources say

• Not checked yetcites Shetty et al., 2014
• Not checked yetcites Berkenstadt et al., 2018

In the paper

“The high-density RNA capture with DNB chip requires a sequencing depth of around 150 gigabases/cm 2 , which costs 5,000 RMB ( $800) based on current short-read sequencing platforms.”

In the paper

“At E16.5, radial glia cells further decreased and concentrated in the caudal part, with neuroblasts showing a more restricted distribution and glioblasts expanding in number in the original locations.”

In the paper

“(F) Heatmap showing the regulons with significant spatial autocorrelation grouped into different modules based on pairwise spatial correlations of an E9.5 section (E1S1).”

In the paper

“(A) Stereo-seq chips of different sizes ranging from 50 mm 2 to 174.24 cm 2 .”

In the paper

“(A and B) Heatmaps showing the genes with significant spatial autocorrelation grouped into different gene modules based on pairwise spatial correlations of E9.5 (left, E1S1) and E16.5 (right, E1S1) sections.”

In the paper

“Importantly, we have profiled a total of 53 sagittal sections corresponding to an overall area of 23.4 cm 2 , but more sections may still be necessary to provide adequate coverage of all regional tissue types.”

In the paper

“Hence, Stereo-seq spatially characterizes the transcriptomic organization and individual cell-type composition of complex tissues in an unbiased manner with high sensitivity and large field of view.”

In the paper

“Recent cell atlases using high-throughput single-cell RNA-sequencing (scRNA-seq) and single-cell assay for transposase accessible chromatin sequencing (scATAC-seq) technologies have provided an increasingly detailed view of mouse developmental gene expression and gene regulatory dynamics”

What the sources say

• Not checked yetcites Cao et al., 2019 source
• Not checked yetcites He et al., 2020
• Not checked yetcites Pijuan-Sala et al., 2020

In the paper

“Therefore, Stereo-seq data can be used to both identify functional anatomic regions in large tissues through bin-based clustering, and cell types based on image-guided cell segmentation.”

In the paper

“The colocalization of Pantr1 with Sox2 is consistent with the regulatory function of this long noncoding RNA in early neuronal differentiation (Goff et al., 2015).”

What the sources say

• Not checked yetcites Goff et al., 2015

In the paper

“These data can be extracted from scRNA-seq datasets using algorithms such as single-cell regulatory network inference and clustering (SCENIC) (Aibar et al., 2017), but the lack of spatial information is likely to confound important associations.”

What the sources say

• Not checked yetcites Aibar et al., 2017 source

In the paper

“Kitl is a pleiotropic factor that acts in the ontogeny of different cell types (Broudy, 1997).”

What the sources say

• Not checked yetcites Broudy, 1997

In the paper

“Early mouse embryos (Lohoff et al., 2022; Peng et al., 2019), specific late-stage-embryonic tissues such as the cerebral cortex (Di Bella et al., 2021; Stickels et al., 2021), and whole midgestation embryos (Liu et al., 2020; Srivatsan et al., 2021; Yao et al., 2020) have been analyzed using these approaches.”

What the sources say

• Not checked yetcites Lohoff et al., 2022
• Not checked yetcites Di Bella et al., 2021
• Not checked yetcites Liu et al., 2020 source

In the paper

“Stereo-seq defines the spatiotemporal window of developmental disease vulnerability”

In the paper

“These results demonstrate the utility of MOSTA for defining the spatiotemporal windows of disease-related gene expression during development and for establishing potential vulnerabilities.”

In the paper

“Stereo-seq is a DNB-based spatially resolved transcriptomic technology with genome-wide coverage, cellular resolution, high sensitivity, and large field of view.”

In the paper

“These parameters are all critical for accurately profiling the transcriptomic heterogeneity of relatively large mid- and late-stage mammalian embryos.”

In the paper

“Due to its characteristics and applicability, Stereo-seq also has the potential to move into routine clinical practice as an extraordinary diagnostic tool complementary to medical imaging and histopathology data.”

In the paper

“MOSTA comprehensively characterizes cell-type and cellstate heterogeneity at whole embryo level, which is relevant for reconstructing cell differentiation processes involving regional and subregional specificities.”

In the paper

“We then investigated Msx1 , a homeobox gene encoding a transcription factor with known crosstalk with WNT5A (Lin et al., 2011).”

What the sources say

• Not checked yetcites Lin et al., 2011

In the paper

“Nowadays, scRNA-seq datasets are becoming widely used to predict the origin and mechanisms of genetic diseases including developmental disorders (Watanabe et al., 2019).”

What the sources say

• Not checked yetcites Watanabe et al., 2019

In the paper

“Besides the study of development, Stereo-seq and its future refinements have the ability to transform multiple other research fields.”

In the paper

“The dorsal midbrain contains complex multilayered structures like the cerebral cortex that form the superior and inferior colliculi (Arimura et al., 2019).”

What the sources say

• Not checked yetcites Arimura et al., 2019

In the paper

“Mice and humans differ in multiple aspects of normal physiology and disease; hence, there are caveats, but this approach clearly has the potential to uncover disease mechanisms.”

In the paper

“These findings suggest an asynchronous and spatially heterogeneous model of neurogenesis and gliogenesis, with glioblasts appearing early at specific anatomic locations where they remain until later differentiation.”

In the paper

“All relevant procedures involving animal experiments presented in this study are compliant with ethical regulations regarding animal research and were conducted under the approval of the Animal Care and Use committee of the Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences (license number IACUC2021002).”

In the paper

“Reads with MID containing either N bases or more than 2 bases with quality score lower than 10 were filtered out.”

In the paper

“UMI with the same CID and the same gene locus were collapsed, allowing 1 mismatch to correct for sequencing and PCR errors.”

In the paper

“The whole procedure was integrated into a publicly available pipeline SAW available at https://github.com/BGIResearch/SAW.”

In the paper

“All relevant procedures involving animal experiments presented in this study are compliant with ethical regulations regarding animal research and were conducted under the approval of the Animal Care and Use committee of the Guangzhou Institutes of Biomedicine and Health, Chinese Academy of Sciences (license number IACUC2021002).”

In the paper

“Finally, although we have developed algorithms including the SCC for the analyses in this study, others may be needed for a broader range of applications in various tissues from different species.”

In the paper

“For Figure 1C, to ensure that proper comparisons were made, the data of Stereo-seq were binned into bin 3 (3 3 3 DNB, 2 m m), bin 14 (14 3 14 DNB, 10 m m) or bin 140 (140 3 140 DNB, 100 m m).”

In the paper

“CID sequences on the first reads were first mapped to the designed coordinates of the in situ captured chip achieved from the first round of sequencing, allowing 1 base mismatch to correct for sequencing andPCRerrors.”

In the paper

“To validate the cell type identity characterized in our adult mouse hemibrain section, annotated scRNA-seq profiles of the adult mouse brain downloaded from Linnarsson Lab website (http://mousebrain.org/adolescent/) (Zeisel et al., 2018) were used for comparison.”

What the sources say

• Not checked yetcites Zeisel et al., 2018 source