Mapping astrocyte heterogeneity in the human cerebellum

1. Overview

This project aims to define the molecular, developmental, and spatial logic of astrocyte heterogeneity in the human cerebellum through integrative analysis of single-cell and spatial transcriptomic datasets. Astrocytes are increasingly recognized as active regulators of neural circuit formation, function, and homeostasis, yet their diversity and developmental programs remain poorly understood in the human brain. Building on our recent work in the mouse cerebellum, which established a robust framework to resolve astrocyte identities, maturation trajectories, and spatial organization, this study will systematically assess the conservation and divergence of astrocyte programs, regulatory networks, and cell–cell interactions between mouse and human. By focusing on the cerebellum as a highly stereotyped model system, the project addresses a key gap in our understanding of human astrocyte biology with direct relevance to neurodevelopmental and neurological disorders.
The project relies exclusively on publicly available human datasets, and previously collected and fully authorized human fetal tissue samples. Ethical approval for the use of human fetal cerebellar material for transcriptomic analyses and immunohistochemistry has already been obtained (protocol no. 0067260; Ethics Committee of A.O.U. Città della Salute e della Scienza di Torino), and no new human sample collection is planned. All required facilities, expertise, and preliminary data are in place to ensure feasibility.
The proposed work is expected to be completed within six months, with the following milestones: data curation and preprocessing (months 1-2), integrative and cross-species analyses including trajectory and spatial mapping (months 3-4), validation and synthesis of results (month 5), and dissemination and data deposition (month 6).
In line with open science principles, all generated data, analysis pipelines, and outputs will be shared through public repositories and open-access platforms. Raw and processed datasets will be deposited in established repositories, code will be released via GitHub, and all project outputs will be archived with persistent identifiers. This commitment ensures transparency, reproducibility, and broad community access to the results.

2. Introduction

Background and Rationale
For decades, neuroscience has largely focused on neurons, treating glial cells as secondary support structures. This neuron-centric perspective has generated fundamental insights, yet it provides only a partial understanding of brain dynamics. A critical piece of this puzzle lies in astrocytes, which are now increasingly recognized as pivotal orchestrators of neuronal activity and cerebral homeostasis. Their roles extend far beyond mere support, positioning them as central players in brain physiology and pathology ¹.
Recent advances in single-cell and spatial omics have revealed that astrocytes are not a homogeneous population but instead display remarkable molecular, functional, and spatial diversity^2–4 . This emerging view challenges the traditional concept of a single “astrocyte identity” and suggests that astrocyte heterogeneity is a key determinant of how neural circuits are formed, maintained, and remodeled.
Despite these advances, a major hurdle remains: in most brain regions, astrocyte diversity manifests as a continuous transcriptomic and morphological gradient rather than as discrete, readily identifiable populations⁵. This intrinsic complexity hampers efforts to link molecular profiles to defined astrocyte identities, developmental programs, and specific functional roles.

The Cerebellum as a Model to Decode Astrocyte Diversity
The cerebellum offers a unique solution to this challenge. Owing to its highly stereotyped architecture, and to the clear spatial organization and gross morphologies of its major astrocyte types^6,7, the cerebellum represents an ideal model system in which molecular profiles can be directly linked to specific astrocyte identities and functions.
Our recent work in the mouse model (Cerrato et al., bioRxiv 2025, under revision)⁸ leveraged this cerebellar framework to uncover an unprecedented level of astrocyte transcriptional diversity. We identified previously unrecognized astrocyte profiles with distinct spatial distributions and functional signatures, reconstructed their maturation trajectories, inferred gene regulatory networks, and traced their embryonic clonal relationships. Unexpectedly, these analyses revealed lineage relationships between specific astrocyte populations and other neural cell types, suggesting shared developmental programs. Moreover, our findings indicate that cerebellar astrocyte heterogeneity emerges from a complex interplay between intrinsic genetic programs and extrinsic environmental cues. Together, these results provide a conceptual and methodological foundation to investigate how distinct astrocyte populations interact with neurons, other glial cells, and non-neural elements to establish and maintain cerebellar architecture and function.

The Knowledge Gap: From Mouse to Human
While these discoveries in the mouse cerebellum represent a major advance, a critical question remains unresolved: to what extent are these principles conserved in humans? Addressing this gap is essential both for fundamental neuroscience and for translational relevance.
Direct experimental access to developing and adult human brain tissue is inherently limited. However, a few high-quality transcriptomic datasets from the developing and adult human cerebellum are now publicly available. Although these resources were not originally generated with a specific focus on astrocyte biology, they constitute an exceptionally powerful and largely unexploited source of information. Systematic reanalysis of these datasets, guided by insights and conceptual frameworks derived from the mouse, provides a unique opportunity to interrogate astrocyte heterogeneity, developmental dynamics, and cell–cell interactions in the human cerebellum. Importantly, this effort will be complemented by a newly generated human fetal cerebellar dataset recently generated in our lab, enabling direct integration of publicly available resources with internally generated data across key developmental stages.

Research Questions
Guided by our data obtained in the mouse model and by the availability of large-scale human single-cell transcriptomic resources, this project is centered on three core questions:
Q1. To what extent is cerebellar astrocyte heterogeneity conserved between mouse and human, and which molecular and functional features are species-specific?
Q2. Do the developmental trajectories and regulatory programs that shape astrocyte diversification in the mouse cerebellum also operate in humans?
Q3. How do interactions between distinct astrocyte profiles and their neuronal and non-neuronal neighbors contribute to cerebellar development, organization, and function?

Hypotheses
Based on existing literature and our preliminary findings, we advance the following hypotheses:
H1. Core astrocyte identities and regulatory programs are conserved between mouse and human cerebellum, while species-specific adaptations give rise to human-enriched astrocyte states.
H2. Human cerebellar astrocytes follow developmental trajectories analogous to those described in mouse, governed by conserved gene regulatory networks but characterized by extended temporal dynamics and increased complexity.
H3. Distinct astrocyte populations engage in specific, spatially constrained interactions with neurons and other cell types, relying on conserved signaling pathways across species.

Importance and Impact
Understanding the molecular and developmental logic of astrocyte heterogeneity in the human cerebellum is essential for achieving a comprehensive view of cerebellar function and dysfunction. The cerebellum is increasingly implicated not only in motor control but also in cognition, emotion, and a wide range of neurodevelopmental and neurodegenerative disorders. Astrocytes are emerging as key contributors to these conditions, however, their specific roles remain poorly defined.
In the long term, this project will lay the groundwork for identifying astrocyte-specific vulnerabilities and adaptive responses in cerebellar disorders. By moving beyond a one-size-fits-all view of astrocytes, it opens the door to precision, astrocyte-targeted therapeutic strategies and represents a paradigm shift in how glial cells are conceptualized in brain health and disease.

3. Methods

Participants/Sample

This project is based on the integrative analysis of human and mouse transcriptomic and spatial datasets and does not involve the recruitment of living participants.
The human material analyzed will include:

Publicly available single-nucleus RNA sequencing (snRNA-seq) datasets from adult human cerebellum ^9–11.
Publicly available snRNA-seq datasets from the developing human fetal cerebellum¹⁰.
A human fetal cerebellar transcriptomic dataset recently generated in the lab, comprising samples collected between 19 and 21 gestational weeks (unpublished).

The newly generated fetal dataset was designed to achieve comprehensive tissue coverage while preserving spatial information. Sampling was performed across the antero-posterior, medio-lateral, and ventro-dorsal axes of the cerebellum. A similar strategy was proven critical to enrich for spatially restricted astrocyte populations in our previous analyses in mice⁸. In addition, the laboratory has access to fixed-frozen human fetal cerebellar tissue sections spanning 18–40 gestational weeks. These samples will be used for immunostaining-based validation of astrocyte markers and spatial localization inferred from transcriptomic analyses.
Mouse data used for cross-species comparison will derive from our previously generated sc/snRNAseq datasets of the adult and developing mouse cerebellum⁸.

Materials and Procedures

Publicly available datasets will be retrieved from established repositories, including the dataset from Sepp et al.¹⁰ (https://doi.org/10.11588/data/QDOC4E), the human brain atlas generated by Siletti et al.⁹ and available through the CELLxGENE platform (https://cellxgene.cziscience.com/collections/283d65eb-dd53-496d-adb7-7570c7caa443), and the dataset generated by Tejwani et al.¹¹, available in GEO under accession number GSE246183. When available, pre-processed data will be used; otherwise, raw sequencing data will be processed using standardized pipelines for genome annotation, quality control, and normalization.
Quality control and filtering of low-quality nuclei, potential doublets, and ambient RNA contamination will be performed separately for each dataset. Subsequently, datasets will be integrated using established methods such as Seurat canonical correlation analysis (CCA) and Harmony, selecting the approach that most effectively minimizes batch effects. Following integration, an initial low-resolution (“coarse”) clustering will be performed to identify the broad astrocyte population. Astrocytes will then be subsetted and re-integrated using a similar pipeline and subjected to high-resolution re-clustering to resolve subtle heterogeneity that may have been overlooked in the original studies.
Cross-species integration will be performed using label transfer and correlation-based approaches to map human astrocyte profiles onto mouse-defined identities. Developmental dynamics will be investigated using comparable analyses leveraging our mouse postnatal dataset⁸ and trajectory inference algorithms such as URD and CellRank. Spatial transcriptomics datasets of the human cerebellum (e.g. Visium 10X) will be analyzed to map astrocyte populations within defined anatomical domains. Finally, cell–cell communication between astrocytes and neuronal or non-neuronal populations will be inferred using NeuronChat to identify conserved and species-specific communication patterns.

Planned Analyses

Analyses will follow a hypothesis-driven framework designed to link astrocyte molecular identity, developmental dynamics, and cellular interactions across species.
To test H1, conservation and divergence of astrocyte identities between mouse and human will be assessed through cross-species integration based on orthologous gene expression and transcriptional similarity metrics. Human astrocyte populations will be systematically compared to mouse-defined reference identities to identify conserved core programs and species-specific adaptations. Statistical significance of cross-species correspondences will be evaluated using similarity and enrichment metrics with appropriate multiple-testing correction.
To test H2, human fetal astrocyte progenitor populations and their putative mature counterparts will be identified and ordered along developmental stages. Developmental trajectories and associated regulatory programs will be reconstructed and compared between human fetal and mouse postnatal datasets to assess conservation of maturation logic, temporal dynamics, and regulatory architecture. Differences in trajectory structure, timing, or gene regulatory features will be interpreted as evidence of species-specific developmental divergence.
To test H3, astrocyte populations will be mapped onto spatially resolved transcriptomic datasets to define their anatomical localization within the cerebellum. Cell–cell communication analyses will then be used to infer signaling interactions between astrocytes and neighboring neuronal and non-neuronal cell types. Conserved and species-specific interaction patterns will be identified.
Contingent analysis decisions will be guided by predefined criteria. If cross-species mapping yields ambiguous or low-confidence assignments, then additional similarity measures and marker-based validation will be applied. If rare astrocyte populations are underrepresented in individual datasets, then analyses will be strengthened through addition of further publicly available datasets. If spatial information is incomplete for specific populations, then inferred localization will be supported by immunostaining-based validation.

Ethics and Data Management

All analyses involving human material will be conducted in compliance with applicable ethical guidelines and regulations. The use of human fetal central nervous system samples for gene expression analyses and immunohistochemistry was approved by the Ethics Committee of A.O.U. Città della Salute e della Scienza di Torino – A.O. Ordine Mauriziano di Torino – ASL Città di Torino (protocol no. 0067260; approval date: June 4th, 2019). All samples were collected for non-profit research purposes and without pharmacological interventions. No new human samples will be collected for this project; all analyses will be conducted on previously collected and fully authorized material.
All human data used in this project are fully anonymized. No personal, clinical, or identifying information linking samples to individual donors is accessible to the investigators. Anonymization will be maintained throughout data handling, analysis, and sharing, in accordance with ethical standards and data protection regulations.
All datasets will be curated and annotated with standardized metadata to ensure traceability, reproducibility, and compliance with community standards. Upon completion of the project, raw and processed transcriptomic data will be deposited in publicly accessible repositories such as the NCBI Gene Expression Omnibus (GEO), with raw data released immediately or following a justified embargo period when necessary. Processed data will be made openly available at the time of publication. Analysis scripts, custom code, and computational pipelines will be shared via GitHub. Additionally, all project outputs will be archived on Zenodo and assigned DOIs. Publications will be disseminated through preprints and published Open Access, with the final versions deposited in the University of Turin Institutional Open Access Repository (AperTO).

4. Pilot Data

The feasibility of the proposed analytical framework is supported by our recent study in the mouse cerebellum⁸, currently under peer review. In this work, we established an integrative single-cell transcriptomic pipeline that resolved previously unrecognized astrocyte subpopulations, reconstructed maturation trajectories, inferred gene regulatory programs, and linked molecular identities to spatial organization within the cerebellum. These analyses demonstrated the robustness of the workflow across developmental stages and validated its ability to capture the extent of astrocyte heterogeneity with unprecedented resolution. Importantly, this pilot study provides proof of principle that the proposed cross-dataset integration, trajectory inference, and cell–cell interaction analyses are technically feasible and biologically informative. The current project builds directly on this validated framework, extending it to the human cerebellum and cross-species comparisons rather than reproducing the mouse analyses.

5. Budget

The requested budget will support computational analyses, limited experimental validation, and dissemination of results.

- Computational Resource (~$4000-5000)

Funds are requested to support the purchase of a dedicated computational workstation required for the analysis and integration of large-scale single-cell and spatial transcriptomic datasets. The workstation will be used for data preprocessing, integration, trajectory inference, and cross-species analyses that are computationally intensive and cannot be efficiently performed on standard desktop machines.

- Antibodies and Consumables (~$3,000)

Funds will cover the purchase of antibodies and basic consumables required for immunostaining of fixed-frozen human fetal cerebellar tissue sections. These experiments will be used to validate astrocyte markers and spatial localization inferred from transcriptomic analyses.

- Dissemination and Open Science (~$1,500-2,000)

Funds will be allocated for dissemination of results, including open-access publication fees and presentation of findings at relevant scientific meetings or workshops.

6. References

1. Verkhratsky, A. & Parpura, V. Recent advances in (patho)physiology of astroglia. Acta Pharmacol. Sin. 31, 1044–1054 (2010).

2. Batiuk, M. Y. et al. Identification of region-specific astrocyte subtypes at single cell resolution. 1–15 (2020). doi:10.1038/s41467-019-14198-8

3. Escartin, C. et al. Reactive astrocyte nomenclature, definitions, and future directions. Nat. Neurosci. 24, 312–325 (2021).

4. Endo, F. et al. Molecular basis of astrocyte diversity and morphology across the CNS in health and disease. Science (80-. ). 378, eadc9020 (2022).

5. Bayraktar, O. A. et al. Astrocyte layers in the mammalian cerebral cortex revealed by a single-cell in situ transcriptomic map. Nat. Neurosci. 1–10 (2020). doi:10.1038/s41593-020-0602-1

6. Cerrato, V. et al. Multiple origins and modularity in the spatiotemporal emergence of cerebellar astrocyte heterogeneity. PLoS Biology 16, (2018).

7. Turrini, G., Bianciotto, O. T., Cerrato, V. & Buffo, A. Ontogenesis and diversification of cerebellar glia: A developmental perspective. Current Topics in Developmental Biology (Elsevier, 2026). doi:10.1016/bs.ctdb.2026.01.008

8. Cerrato, V. et al. A single-cell transcriptomic atlas maps cerebellar astrocyte diversity and uncovers the transcriptional code underlying their maturation trajectories. bioRxiv 2025.07.17.665323 (2025).

9. Siletti, K. et al. Transcriptomic diversity of cell types across the adult human brain. Science (80-. ). 382, eadd7046 (2023).

10. Sepp, M. et al. Cellular development and evolution of the mammalian cerebellum. Nature 625, 788–796 (2024).

11. Tejwani, L. et al. Longitudinal single-cell transcriptional dynamics throughout neurodegeneration in SCA1. Neuron 112, 362-383.e15 (2024).

Mapping astrocyte heterogeneity in the human cerebellum

Mapping astrocyte heterogeneity in the human cerebellum

Funding Progress

Funders

Peer Reviews

Related Funding

Request for Proposals - Neuroscience

Nonprofit Recipient

Topics

DOI