Astrocytic Stratification Across Vertebrate Lineages

A Conserved Yet Progressively Diversified Taxonomy

Clustering of astrocyte cells across the five major vertebrate lineages revealed a conserved yet progressively diversified set of subpopulations. Unsupervised integration produced 10 major astrocyte subclusters, each supported by robust cross-species transcriptional signatures, and a resolution-0.5 reclustering additionally resolves one marginal subcluster. Cross-species comparisons show that astrocytes progressively expanded and specialized during vertebrate brain evolution, reaching their greatest molecular complexity in amniotes and mammals.

From Progenitor State to Mature Astrocyte Regulators

The astrocyte subclusters are arranged along a clear developmental and regulatory continuum across the vertebrate phylogeny. Ancient clusters predominant in fishes and amphibians retain a progenitor-like state; reptile- and bird-enriched clusters acquire regulators associated with regional specialization within pallial and subpallial domains; and mammal-restricted subclusters show the highest regulatory divergence, with strong induction of transcription factors linked to mature astrocyte differentiation, metabolic support and synaptic modulation.

SOX2 / PAX6 → HES5 / NKX2.2 / LHX2 → NFIA / SOX9

SOX9 and NFIA-NFIB Define the Amniote Trajectory

At the cluster level, two regulators anchor the amniote-derived astrocyte program. SOX9 scores highest in two distinct contexts: an ependymal-like cluster and a broad progenitor-like cluster. In parallel, NFIB is co-upregulated with NFIA in a mammal-enriched cluster, alongside co-high expression of the astrocyte homeostatic markers SLC1A2 and AQP4.

Here z denotes the cross-cluster z-score of mean normalized expression across the astrocyte subclusters; the two values below |z| = 1.96 are reported descriptively.

FOXJ1 Co-Scoring in Teleost Astrocytes

In teleost fishes, astrocyte-high cells co-scored for the ependymal marker FOXJ1, consistent with a close association between the astrocytic and ependymal programs in this lineage. A node-level AST-minus-FOXJ1 sequence test was negative, meaning this overlap reflects the shared regulatory architecture of these cells rather than a distinct sequence-defined subtype. Avian- and reptile-specific clusters in turn expressed unique regulators such as TBR1 and EMX2, further underscoring the lineage-specific tuning of the glial program.

Subcluster Enrichment Across the Tree

At resolution 0.5 the astrocyte subclusters (named Ast_00 through Ast_10 by their first-ranked transcription factor) show pronounced, lineage-biased enrichment (Figure 3G). Representative enrichment values illustrate the sharp partitioning: fish-dominated subclusters reach nearly 100 percent of astrocyte cells, while amniote-marked subclusters show lower but clearly lineage-restricted abundance.

• Ast_10_YBX1 fishes 100%
• Ast_09_RFX3 fishes 99.9%
• Ast_00_ZHX3 mammals 99.6%
• Ast_03_TRPS1 birds 4.29x
• Ast_08_ETV1 mammals 1.88x

Dot size reflects subcluster cell fraction and color reflects BH-significant enrichment fold change; grey marks non-significant differences. Bar width is proportional to fold change only; the 99.6-100 percent lineage shares are shown at full scale.

An Ancient, Lineage-Independent Program

A transcriptome age index computed from pseudobulk expression places the astrocyte program among the most ancient cell types in the atlas: Ependymal 1.0019 < Astrocytes 1.0024 < OPC 1.0028 < Oligodendrocytes ≈ Neurons 1.0029 < Microglia/CAMs 1.0042 < Endothelial 1.0047. Given the per-mille effect sizes and the use of log-normalized expression weights in place of TPM, this ranking is treated as descriptive.