Unveiling Hypothalamic Molecular Signatures via Retrograde Viral Tracing and Single-Cell Transcriptomics

Abstract

Despite the importance of hypothalamic neurocircuits in regulating homeostatic and survival-related behaviors, our understanding of the intrinsic molecular identities of neural components involved in these complex multi-synaptic interactions remains limited. In this study, we constructed a Cre recombinase-dependent pseudorabies virus (PRVs) capable of crossing synapses, coupled with transcriptome analysis of single upstream neurons post-infection. By utilizing this retrograde nuclear Connect-seq (nuConnect-seq) approach, we generated a single nuclei RNA-seq (snRNA-seq) dataset of 1,533 cells derived from the hypothalamus of CRH-IRES-Cre (CRH-Cre) mice. To ensure the technical validity of our nuConnect-seq dataset, we employed a label transfer technique against an integrated reference dataset of postnatal mouse hypothalamus comprising 152,524 QC-passed cells. The uniqueness of our approach lies in the integration of diverse datasets for validation, providing a more nuanced diversity of hypothalamic cell types. The presented validated dataset may deepen our understanding of hypothalamic neurocircuits and underscore the essential role of comprehensive integrated transcriptomic data for technical validity.

Fig. 1

Generation of a nuConnect-seq-derived dataset. (a) A retrograde nuConnect-seq approach was used to generate a single nuclei RNA-seq (snRNA-seq) dataset from CRH-IRES-Cre (CRH-Cre) mice. (b) PRVB180, a pseudorabies virus expressing TK (thymidine kinase)-GFP (green fluorescence protein) in the Cre+ neurons and their upstream. (c) The flow of nuConnect-seq method. (d) UMAP plot showing the manually curated cell types of the presented dataset. (e) Violin plot showing the relative expression of neuronal and non-neuronal gene markers.

Fig. 2

Schematic representation of the generation of an integrated reference dataset of postnatal mouse hypothalamus for technical validation of nuConnect-seq-derived dataset. Ten independent datasets were integrated and batch-corrected in an anchor-based manner. Cell-level metadata was manually curated and standardized for all the datasets. A uniform informatic pipeline was used for clustering and cell type annotation.

Fig. 3

Cell type annotation of the integrated reference dataset of young adult mouse hypothalamus. (a) UMAP plots showing neuronal or non-neuronal cells (left) classified by combined expression of pan-neuronal markers (right). (b) Relative expression of pan-neuronal and non-neuronal marker genes is shown. (c) Final UMAP plot showing subtypes of neuronal and non-neuronal cells.

Fig. 4

Mapping and transferring labels for technical validation of nuConnect-seq-derived dataset. (a) Schematic representation showing identified anchors, which facilitate the transfer of discrete labels between integrated (reference) and lab-generated nuConnect-seq (query) datasets. (b) UMAP plots show the actual and predicted cell types on the nuConnect-seq dataset by projecting cells of the nuConnect-seq dataset onto UMAP embeddings. (c) Histogram plot showing the distribution of predicted cell type scores across the predicted cell types in the validation dataset. (d) UMAP plot showing neuronal and non-neuronal cells of the final merged (integrated and nuConnect-seq dataset).