Single-cell RNA-seq and RAR expression in fetal germ cells

Method Details

Mouse urogenital ridges, testes, ovaries and adrenal glands collection

CD-1 female mice were bred with heterozygous Tg(Nr5a1-GFP)transgenic male mice. Adult females were time-mated and checked for the presence of vaginal plugs the next morning (E0.5). E10.5 (8±2 tail somites (ts)), E11.5 (19±4 ts), E12.5, E13.5, E16.5 and E18.5 embryos were collected and the presence of the Nr5a1-GFP transgene was assessed under UV light. Sexing of E10.5 and E11.5 embryos was performed by PCR with a modified protocol from (McFarlane et al., 2013). Urogenital ridges from each sex, XY adrenal glands, testes or ovaries were pooled for tissue dissociation.

Single cell suspension and library preparations

Urogenital ridges and adrenal glands were enzymatically dissociated at 37ºC for 20 and 40 minutes, respectively, using the Papain dissociation system (Worthington #LK003150). Cells were resuspended in DMEM 2%FBS, filtered through a 70 μm cell strainer and stained with the dead cell marker Draq7™ (Beckman Coulter, #B25595). Viable single cells were collected on a BD FACS Aria II by excluding debris (side scatter vs. forward scatter), dead cells (side scatter vs. Draq7 staining), and doublets (height vs. width). Testes and ovaries (from E12.5 to E16.5) were enzymatically dissociated at 37ºC during 15 minutes in Trypsin-EDTA 0.05% (Gibco #25300054), resuspended in DMEM 2%FBS and filtered through a 70 μm cell strainer. After counting, 3000 to 7000 single cells were loaded on a 10x Chromium instrument (10x Genomics). Single-cell RNA-Seq libraries were prepared using the Chromium Single Cell 3′ v2 Reagent Kit (10x Genomics) according to manufacturer’s protocol. Each condition (organ, sex and developmental stage) was performed in two biological independent replicates.

Sequencing

Library quantification was performed using the Qubit fluorometric assay with dsDNA HS Assay Kit (Invitrogen). Library quality assessment was performed using a Bioanalyzer Agilent 2100 with a High Sensitivity DNA chip (Agilent Genomics). Libraries were diluted, pooled and sequenced on an Illumina HiSeq4000 using paired-end 26 + 98 bp as the sequencing mode. Libraries were sequenced at a targeted depth of 100 000 to 150 000 total reads per cell. Sequencing was performed at the Health 2030 Genome Center, Geneva.

Bioinformatic Analysis

Data processing with the Cell Ranger package, cell selection and quality controls

Computations were performed at the Vital-IT Center for high-performance computing of the SIB (Swiss Institute of Bioinformatics) (http://www.vital-it.ch). Demultiplexing, alignment, barcode filtering and UMI counting were performed with the Cell Ranger v2.1 pipeline (10x Genomics). Algorithms and versions used are listed in the key resources table of . Data were mapped to the mouse reference genome GRCm38.p5 in which the eGFP (NC_011521.1) combined with the bovine GH 3′-splice/polyadenylation signals (Stallings et al., 2002) (NM_180996.1) sequences have been added.

Data processing with the Cell Ranger package, cell selection and in-house quality controls

For cell-associated barcode selection, we computed the knee point and the inflection point of the ranked barcodes distribution plot. The local minimum between these points on the density curve (density base R function, bw=500, n=4096) of the UMI number per barcode was detected using the quantmod package. This local minimum was used as a threshold for cell-associated barcode selections. When no local minimum could be detected between the two points, the nearest local minimum was used. Quality controls regarding abnormal mitochondrial or ribosomal content, UMI number, detected gene numbers, unmapped reads and putative doublet identification (Scrublet) were performed, but no data filtering was applied as no important anomalies were detected.

Generation of a single-cell transcriptional atlas of germ cells sex determination and differentiation

The transcriptomes of the individual cells were sequenced at the depth of ~150,000 reads/cell. After barcode filtering based on the unique molecular identifiers (UMI) distribution, we obtained 92,267 cells. It included 14,904 cells from E10.5, 16,581 cells from E11.5, 19,551 cells from E12.5, 25,012 cells from E13.5, and 16,219 cells from E16.5. Among the 52,463 XY cells and the 39,804 XX cells, the median number of UMIs was 17,493 and 17,655 and the median number of detected genes was 4,802 and 4,658, respectively.

Gene expression normalization

UMI counts per gene per cell were divided by the total UMI detected in the cell, multiplied by a scale factor of 10,000 and log transformed.

Germ cells selection

After barcode filtering based on the unique molecular identifiers (UMI) distribution, we obtained 92,267 cells. It included 14,904 cells from E10.5, 16,581 cells from E11.5, 19,551 cells from E12.5, 25,012 cells from E13.5, and 16,219 cells from E16.5. Among the 52,463 XY cells and the 39,804 XX cells, the median number of UMIs was 17,493 and 17,655 and the median number of detected genes was 4,802 and 4,658, respectively. To determine which cells were germ cells, we selected all genes detected in more than 50 cells and performed ICA on log normalized values.  To assess for batch effect, we built a nearest neighbor graph using BBKNN function (BBKNN package). Clustering was performed using Scanpy Louvain method with resolution 1 and UMAP were generated using Scanpy UMAP method with default parameters. We selected clusters with a strong expression of 10 well-known germ cells markers.

Category