Scientists have long relied on investigating messenger RNA (mRNA) to explore the molecular basis of brain development. As an intermediary carrying instructions from genes to the protein-making machinery of the cell, mRNA offers a glimpse into cellular activity and is relatively accessible to measure. But mRNA levels do not always predict how much protein a cell makes. This distinction may be critical for understanding both typical brain development and neurodevelopmental conditions such as autism.

Recent studies measuring both RNA and proteins across human tissues found that this disparity is most pronounced in the cerebral cortex^1,2, but those studies were largely conducted on bulk tissue—a process that averages across cells, washing out individual variation³. As such, the mismatch could be even greater when examined cell by cell, particularly in the developing brain, where cells are rapidly changing identity and function.

Now, a new study published in Nature Biotechnology confronts this problem directly. The SFARI-supported researchers, led by Jingjing Li of the University of California, San Francisco, developed a method to measure proteins in single cells at an unprecedented scale and applied it to the developing human cerebral cortex. In doing so, they uncovered biology that RNA-based approaches had missed, including new insights into genes associated with autism.

Fetal brain cells are extraordinarily small, making protein measurement in these cells a considerable challenge. A developing neuron’s cell body can be as small as 7 to 10 micrometers across, roughly one-tenth the width of a human hair, and contains only a minuscule amount of protein⁴. Previous technologies for single-cell protein measurement worked best with much larger cell types, such as oocytes and cardiomyocytes^5,6, or could only detect a small fraction of the proteins present⁷.

To overcome this limitation, the researchers developed a refined approach using high-sensitivity mass spectrometry (a technique that identifies molecules by their mass) and precise sample handling to reliably detect around 800 proteins per cell across a range of brain cell types. This is comparable to the number of genes typically captured in single-cell RNA sequencing studies. In total, the team profiled over 2,300 cells sampled at three different points in fetal brain development.

Comparing protein and mRNA profiles from the same tissue samples revealed broad discordance. Even in excitatory neurons, where these levels were most aligned, the correlation was modest at best. In some cases, genes with high mRNA levels produced very little protein; in others, abundant protein was made from genes with low mRNA levels.

Importantly, the genes showing the greatest mRNA–protein discrepancy were not random. Those with high mRNA but unexpectedly low protein levels were significantly enriched for known autism risk genes, including those catalogued by SFARI. They also showed a higher rate of de novo mutations, new mutations not inherited from either parent, in people with autism compared to unaffected siblings. The authors suggest that for this gene group, protein levels are under tight regulatory control, and disruptions to this control may contribute to autism risk.

The researchers went further by reconstructing a key developmental trajectory at the protein level, tracing the progression from radial glia, the stem-like cells that generate most of the brain’s neurons, through an intermediate progenitor stage and finally to excitatory neurons. Along this path, they identified a cluster of proteins that became active during the shift from intermediate progenitors to neurons, which was enriched for genes carrying disruptive mutations in people with autism.

This transition appears to be a pivotal point in brain development, when a cell must simultaneously reorganize its internal scaffolding, overhaul its energy metabolism, reshape how its genes are processed and lay the groundwork for forming connections with other neurons, making it genetically vulnerable. The same research team had previously identified this window as a time of heightened autism risk using an entirely different approach⁸, lending additional weight to the finding.

More broadly, for neurodevelopmental conditions such as autism, gene expression alone may tell an incomplete story. Key information may lie in how precisely a cell controls what gets translated into protein. As proteins show substantially greater cell-type specificity than their corresponding mRNAs, they more faithfully reflect individual cell physiology. Measuring proteins directly may therefore be essential for identifying the molecular drivers of these conditions.

1. Liu Y. et al. Cell 165, 535–550 (2016) PubMed
2. Jiang L. et al. Cell 183, 269–283.e19 (2020) PubMed
3. Wang D. et al. Mol. Syst. Biol. 15, e8503 (2019) PubMed
4. Bubis J.A. et al. Nat. Methods 22, 510–519 (2025) PubMed
5. Virant-Klun I. et al. Mol. Cell. Proteomics 15, 2616–2627 (2016) PubMed
6. Ai L. et al. Mol. Cell. Proteomics 24, 100910 (2025) PubMed
7. Tracey L.J. et al. Curr. Protoc. 1, e118 (2021) PubMed
8. Wang L. et al. Nature 647, 169–178 (2025) PubMed