Massively parallel in vivo Perturb-seq reveals cell-type-specific transcriptional networks in cortical development

Abstract

Leveraging AAVs' versatile tropism and labeling capacity, we expanded the scale of in vivo CRISPR screening with single-cell transcriptomic phenotyping across embryonic to adult brains and peripheral nervous systems. Through extensive tests of 86 vectors across AAV serotypes combined with a transposon system, we substantially amplified labeling efficacy and accelerated in vivo gene delivery from weeks to days. Our proof-of-principle in utero screen identified the pleiotropic effects of Foxg1, highlighting its tight regulation of distinct networks essential for cell fate specification of Layer 6 corticothalamic neurons. Notably, our platform can label >6% of cerebral cells, surpassing the current state-of-the-art efficacy at <0.1% by lentivirus, to achieve analysis of over 30,000 cells in one experiment and enable massively parallel in vivo Perturb-seq. Compatible with various phenotypic measurements (single-cell or spatial multi-omics), it presents a flexible approach to interrogate gene function across cell types in vivo, translating gene variants to their causal function.

Keywords: AAV vectors; CRISPR screen; brain development; corticogenesis; in vivo Perturb-seq; single cell genomics.

Figure 1.. Barcoded AAV serotype screen in vivo identified AAV-SCH9 efficiently targeting developing brains.

(A) Schematics of AAV library administration in utero followed by Fluorescence-activated cell sorting (FACS) cell enrichment and barcode analysis. (B) Immunofluorescence analysis 48-hour after AAV library administration, co-stained with markers of newborn projection neurons (TBR1 and CTIP2) and intermediate progenitors (TBR2) in cortical laminar and ganglionic eminence (GE); quantification of percentage of GFP⁺ cells co-expressing key markers. (C) Heatmap of AAV vectors abundance proportion in initial library, 48-hour post-transduction in HT22 cells and embryonic brain; each row represents an AAV serotype. (D) Principal component analysis of the abundance of AAV library, HT22 cells and mouse brain 24- or 48-hour post-transduction. (E) AAV-SCH9 percentage abundance in initial AAV library, HT22 cells and mouse brain 24- or 48-hours post transduction. (F) Volcano plots of AAV serotype abundance changes in mouse brain or HT22 cells 48-hour post-transduction compared to initial AAV library. Scale bars indicate 250μm (left in B), 50μm (middle in B) and 10 μm (right in B and C). Error bars indicate standard error of the mean.

Figure 2.. AAV-SCH9 labeled newborn neurons spreading across brain regions in vivo across developmental stages.

(A) Schematics of a secondary serotype screen: 14 AAV serotypes were barcoded and introduced in pool followed by scRNA-seq 48-hour later. (B) UMAP visualization of 11 major cell populations identified from sorted (GFP⁺) and unsorted cells; cell types include: upper and deep layer projection neurons (ULPN, DLPN), migrating neurons, apical progenitors, intermediate progenitors (IP), interneurons derived from medial ganglionic eminence (IN-MGE), interneurons derived from non-medial ganglionic eminence (IN-non-MGE), Cajal-Retzius cells, fibroblast, mural cells, and microglia. (C) AAV serotype barcode expression in each cell type. Each AAV serotype is associated with 3 barcodes. (D) Immunofluorescence of AAV-SCH9-transduced brain sections co-stained with markers including CTIP2, TBR1 and TBR2. Arrows indicate representative cells with marker co-localizations. (E) Adult neurons can be targeted by AAV-SCH9 within 2 days. Brains were analyzed 2-, 7-, or 21-days post injection. (F) Whole brain imaging characterizes AAV-SCH9 tropism. Top: consecutive serial optical slabs across the whole brain, each image represents 1 mm maximal projection. Bottom: Zoomed images from a single Z plane showing the GFP-KASH expression in neocortex (left and middle) and cerebellum (right). (n=4 animals). (G) AAV-SCH9 or lentiviral reporter (GFP) resulted in diverse brain region labeling (n=2–3 animals/condition). Scale bars indicate 1mm (in F, top right in E), 500μm (left in D and top in G), 50μm (left in E, bottom in F and right in G), or 10μm (right in D). Error bars indicate standard error of the mean.

Figure 3.. hypPB transposon enhanced and stabilized expression in embryonic and adult brains and peripheral nervous systems.

(A) Schematics of molecular design to enhance transgene expression. (B) Co-transfection of hypPB increased GFP expression in vitro. (C-E) Transposon stabilized expression in vivo across embryonic brain, adult brain, and adult dorsal root ganglion. (F) AAV-SCH9 labeled more cortical neurons with hypPB (n=3 animals/condition). Asterisks indicate P-value<0.0001 with unpaired t-test. Each point represents a cortical column. (G) Increased expression intensity in cortical neurons with AAV9-PHP.eB-hypPB across cortical layers from ventricular zone (0) to pia (1) (n=3 animals/condition). (H-J) Whole genome sequencing analysis of AAV-SCH9-hypPB transduced cells in vivo showed genomic regions of integration events. (H) Each line indicates a unique hybrid read between mouse genome and transposon. (I) Percentage of integration events compared to the baseline in mouse genome. (J) Example reads aligned to the transposon sequence (in orange boxes) and mouse genome. Scale bars indicate 100μm (in B, right in C, right in D and in E) or 500μm (left in C and left in D). Error bars indicate standard error of the mean.

Figure 4.. In vivo Perturb-seq identified cell type-specific changes across perturbations of transcription factors.

(A) Schematics of screen design. (B) UMAP plot of filtered cells with a high-confidence single perturbation, with each cell colored by annotated cell type (left), gRNA identity (top right), or batch/channel (bottom right). (C) Top: schematics of 5’ and 3’ scRNA-seq capture mechanism. Bottom: percentage of cells with gRNA UMI greater than 5 across cell types in 5’ and 3’ scRNA-seq. VLMC: vascular and leptomeningeal cells; PN: projection neurons. (D) Percentage of cells assigned to one or more gRNAs. (E) Percentage of cells with insertion/deletion in Foxg1 gRNAs targeting loci by Foxg1 perturbation compared to NT2 controls. (F) Cell cluster proportion changes by each perturbation. (G) Dot plot of number of differentially expressed genes (DEGs) and cell number across perturbations. (H) Volcano plots of cell-type specific effects on DEGs of Foxg1-gRNA1 perturbation in L6-CT and L5-IT neurons. (I) Perturbation of Foxg1 led to a hybrid neuronal cell type. Left: UMAP plots of cell type and gRNA identity; dotted line highlights the hybrid subcluster (L6-CT-cluster 3). Middle: zoomed UMAP plots showing expression levels of mis-expressed TFs in the hybrid subcluster. Right: proportion of perturbation groups in each cluster.