Human-specific morphoregulatory signatures in basal radial glia characterise neocortex evolution

Basal radial glia as a lever on neocortex size

The question behind this paper is where human-specific gene regulation acts during cortical development. The authors start from a well-established link: the abundance and proliferative potential of basal radial glia (bRG), progenitors that divide in the subventricular zone away from the ventricle, track with neuron number and neocortex size across species. bRG morphology, lineage behaviour and metabolism have all been proposed as contributors to that link.

What remained uncharacterised was the genomic side: which cis-regulatory elements (CREs), the DNA sequences that control when and where genes are expressed, have gained activity in human bRG. Because foetal tissue from great apes is not accessible, the authors built cerebral organoids (CeOs) from gorilla, chimpanzee and human induced pluripotent stem cells and isolated aRG, bRG, basal intermediate progenitors and neurons from them. They profiled chromatin accessibility and transcription at week 8 of organoid development, benchmarking against human foetal cortex at gestational weeks 13–14, stages at which all four cell types are present.

599 elements more accessible in human bRG

The central measurement is a comparison, not an absolute count. A CRE was called a human bRG candidate element only if it was significantly more accessible in human bRG than in both chimpanzee and gorilla bRG, using a threshold of log2 fold change of at least 1.5 and p ≤ 0.05. That double comparison is what makes the set human-increased rather than merely primate-variable, and the regions were benchmarked against peaks in human foetal bRG.

This produced 599 human bRG candidate CREs. Their accessibility z-scores across replicates of the three species are shown in the paper, and k-means clustering grouped the 599 elements into three broad accessibility patterns across organoid samples and cell types. Two of the elements are human accelerated regions, sequence regions with unusually fast change on the human lineage. Linking the elements to genes through cell-type-specific Hi-C contact data from radial glia and basal intermediate progenitors yielded 589 unique target genes in physical contact with the 599 elements.

The target-gene set is where the biological signal first appears: Gene Ontology analysis found enrichment for forebrain development, regulation of cell–cell contacts and regulation of mature neural cell morphology. That is a statement about the genes contacted by human-increased elements, not yet about measured morphology.

Gene expression converges on the same morphology theme

The transcriptional arm of the study asks a separate question with a separate denominator. Comparing chimpanzee and human bRG, the authors found 939 differentially expressed genes at log2 fold change of at least 1 and adjusted p ≤ 0.05. The expression pattern of these genes in human organoid bRG correlated highly with human foetal bRG, which supports using the organoid as a stand-in for the foetal context at this stage.

Genes upregulated in human bRG were predominantly enriched for processes regulating neural cell morphology. Terms such as regulation of neuron or cell projections and axonogenesis appeared specifically for genes upregulated in human bRG and were not detected for aRG and basal intermediate progenitors, which the authors read as a bRG-specific phenotype. PALMD, previously implicated in basal progenitor morphological complexity and proliferative potential, was significantly higher in human than chimpanzee bRG but not in human aRG or intermediate progenitors.

To summarise this across gene sets, the authors built a morphoregulatory pathway activity score from three Gene Ontology groups: cell morphogenesis, cell projection morphogenesis and cell projection organisation. In chimpanzee organoids the score rose from aRG to bRG and intermediate progenitors and again to neurons. In human cells the same general rise occurred, but human bRG reached a score almost identical to human neurons, above human aRG and intermediate progenitors and above chimpanzee progenitors. The score is a weighted summary of expression, not a direct measurement of cell shape, so it indicates pathway-level activity rather than morphology itself.

Human bRG appear to reuse neuronal programs

Many genes driving the morphology enrichment and the pathway score are classically associated with neuronal morphogenesis. The authors tested whether ancestral neuronal programs are active in human bRG by projecting the 465 human bRG-upregulated genes onto published datasets of human foetal, macaque foetal and mouse embryonic cortex.

In human foetal cell types the genes split into two classes: one more highly expressed in basal intermediate progenitors and neurons, the other relatively higher in radial glia. Across human foetal developmental stages from gestational week 11 to 18, the genes appeared consistently in radial glia as well as excitatory and inhibitory neurons. In macaque foetal cortex the same gene set was primarily expressed in neural progenitor cells and excitatory neurons, with enrichment in bRG-like populations. The authors interpret this dual pattern as co-option: human bRG gained activity in morphoregulatory elements and genes that are embedded in ancestral neuronal programs, rather than inventing a wholly new program. That interpretation is the authors' reading of expression overlap, not a demonstrated causal sequence.

FAM107A loss simplifies bRG shape

The functional test moves from correlation to perturbation. The authors knocked out FAM107A and CNGA3 in human cerebral organoids by CRISPR/Cas9 and analysed electroporated bRG identified as GFP-positive, SOX2-positive and EOMES-negative cells in the subventricular zone. Each cell was segmented through a Z-stack and assigned a morphotype, and process number was quantified.

Knockout of FAM107A significantly reduced the number of all processes and of primary processes growing directly from the cell body, without affecting process length, branching or arborisation. CNGA3 knockout did not significantly alter any measured morphological parameter. Both genes are also expressed in aRG, but their knockouts did not affect canonical aRG morphology, which supports a bRG-specific effect. The two genes therefore do not behave identically: FAM107A has a measurable role in process number in this assay, while CNGA3's contribution was not captured by these morphological readouts.

The authors also report that FAM107A knockout reduced HuCD-positive, KI67-negative neurons in the subventricular zone and cortical plate, and that overexpressing human FAM107A in chimpanzee organoids produced a phenotype similar to the human knockout. They interpret the latter as evidence that FAM107A levels must be tightly regulated in both directions, consistent with prior work on morphoregulatory proteins.

Organoid evidence and its developmental ceiling

The comparisons rest on organoids rather than on great ape foetal tissue, which the authors present as a necessity and a strength: organoids contain bRG and parallel aspects of first-trimester neural development, including cellular morphology, tissue architecture and overall gene expression, and their epigenomic landscape is broadly conserved. The week 8 organoid stage was chosen because all four cell types are present and because organoid and foetal brain development are highly similar at the cellular and transcriptomic level at that point.

The limitation the authors flag as most relevant to cortical evolution is that organoids show limited expansion and formation of the outer subventricular zone, the structure that expands during the second trimester in the human foetal brain. They suggest media improvements such as adding leukaemia inhibitory factor may increase bRG proliferation and support a structured outer subventricular zone. They also note that inter-species differences in developmental timing could influence when species-specific phenotypes appear, while arguing that mid-neurogenic milestones are largely conserved and that mid-neurogenic organoid development appears highly aligned across gorilla, chimpanzee and human.

A further boundary is the nature of the evidence for co-option: the overlap with neuronal expression is correlative across datasets, and the causal claims in the paper come from the knockout experiments on two genes, not from rewriting the candidate elements themselves.

Regulatory gain, not a new gene repertoire

The paper's distinctive claim is about the level at which human bRG differ. The strongest signal is not a set of human-only genes but a set of 599 CREs with increased accessibility in human bRG relative to both other great apes, plus 939 differentially expressed genes between chimpanzee and human bRG whose upregulated fraction is enriched for morphology terms. The authors place this alongside earlier examples of differentially expressed shared genes and human accelerated regions affecting radial glia behaviour, and describe their contribution as extending that concept to several hundred genomic regions with inter-species differences in CRE activity in great apes.

For readers weighing the result, the priority is the pairing of two independent lines: chromatin accessibility and transcription both point to cell-projection and morphology regulation, and the perturbation data give one gene, FAM107A, a measurable morphological role in human bRG. CNGA3 remains implicated by expression and enhancer linkage but not by the reported morphological readouts. The most defensible reading is that human-specific morphoregulatory signatures distinguish human bRG from other great apes and are a candidate mechanism for the morphological elaboration of these progenitors, with the causal weight concentrated in the FAM107A perturbation and the element-level causality still untested.

Reading the species comparison correctly

Two comparisons in this paper answer different questions and should not be merged. The 599-element set is defined against both chimpanzee and gorilla, so it captures changes specific to the human lineage among the three species sampled. The 939-gene set is a chimpanzee-versus-human contrast, and the authors checked direction by asking whether human-upregulated genes were also lower in gorilla bRG, concluding the pattern reflects human upregulation rather than chimpanzee downregulation. A gene that is human-increased relative to chimpanzee is therefore not automatically human-specific relative to gorilla unless the gorilla comparison supports it.

The practical consequence is that the element-level claim is the more stringent one, while the gene-level and pathway-score claims describe human–chimpanzee differences with gorilla used as a directional check. Readers comparing this work with other studies of human-gained enhancers should also note the cell-type and stage frame: the measurements are from bRG isolated at week 8 of organoid development, benchmarked against human foetal cortex at gestational weeks 13–14, not from adult tissue or from the second-trimester outer subventricular zone.

Sources used in this answer