AAV capsids that cross into the brain

The research record

This is the project's file for this landmark, as written and checked. Sources are listed at the end; anything not yet confirmed is marked "Not yet verified".

Who

Benjamin Deverman, Viviana Gradinaru and colleagues at Caltech (2016, 2017); Qin Huang, Ken Chan, Deverman and colleagues at the Broad Institute (2024). Deverman co-founded Apertura Gene Therapy, which the Broad release says is developing these capsids. Funding statements of the capsid papers were not read.

What happened

Gene therapy for brain disease normally needs surgery to inject the vector (the delivery vehicle) into the target spot. Natural AAV9 crosses the barrier between blood and brain poorly. In 2016 Deverman et al. introduced CREATE, a Cre-recombinase-based method for evolving AAV capsids (the protein shell) in living mice. Their lead variant, AAV-PHP.B, delivered genes to the adult mouse brain at least 40 times more efficiently than AAV9, reaching most neurons and astrocytes (Nature Biotechnology 34:204-209). In 2017 Chan et al. made PHP.eB, which transduced 69% versus 49% of cortical neurons (NeuN-positive cells) compared with PHP.B at the same dose (Nature Neuroscience 20:1172-1179).

Then came the catch. In 2018 a marmoset study found no gain over AAV9 (Matsuzaki et al., Neurosci Lett 665:182-188), and a second paper showed the effect is limited to certain mouse strains such as C57BL/6J (Hordeaux et al., Mol Ther 26:664-668). In 2019 two groups showed PHP.B enters the brain through a mouse receptor, LY6A (Hordeaux et al., Mol Ther 27:912-921; Huang et al., PLoS One 14:e0225206). In 2023 the Gradinaru lab reported AAV.CAP-Mac, which crosses into the brains of marmosets, rhesus macaques and green monkeys after a vein injection (Chuapoco et al., Nature Nanotechnology). In 2024 Huang, Chan et al. described BI-hTFR1, a capsid reprogrammed to bind the human transferrin receptor. In mice carrying the human receptor it gave 40 to 50 times more reporter expression in the brain and spinal cord than AAV9, and about 30 times more vector genomes reached the brain with a GBA1 payload. When used to deliver GBA1 (the gene behind the lysosomal enzyme GCase), it raised enzyme activity in brain and spinal fluid above AAV9 (Science 384:1220-1227, 14 June 2024).

Why it moved us toward a cure

Many cure strategies, such as replacing GBA1, switching off SNCA (landmark 117) or delivering growth factors (landmark 39), need genes to reach many neurons, not one injection site. PHP.B showed that capsids can be engineered to do this, and the species failure taught the field to design for primate and human receptors.

Intravenous AAV9 had already been shown to reach the brain in 2009 (landmark 304).

Outcome

Advanced as a research tool; unproven as medicine. PHP.eB is a mouse tool. BI-hTFR1 has only been shown in mice, including human-receptor mice; the preprint reported no animal primate data and said the authors did not expect enhanced brain delivery in monkeys, because macaque and marmoset receptors did not support it in cells. CAP-Mac is a research tool in monkeys. No Parkinson's-specific clinical programme using these capsids was found. No correction or retraction was found for the 2016, 2017 or 2024 papers (Crossref).

Sources

• Deverman et al. 2016, Nat Biotechnol 34(2):204-209, DOI 10.1038/nbt.3440, PMID 26829320 (abstract via Europe PMC; also Caltech meeting abstract: 40- to 92-fold)
• Chan et al. 2017, Nat Neurosci 20:1172-1179, DOI 10.1038/nn.4593 (the 69% vs 49% figure is in the full text)
• Matsuzaki et al. 2018, Neurosci Lett, DOI 10.1016/j.neulet.2017.11.049; Hordeaux et al. 2018, DOI 10.1016/j.ymthe.2018.01.018; Hordeaux et al. 2019, DOI 10.1016/j.ymthe.2019.02.013 (bibliographic details via Crossref; papers not read)
• Addgene blog on LY6A and primate non-translation (2019); Pietersz et al. 2021 review
• Chuapoco et al., Nat Nanotechnol 2023, DOI 10.1038/s41565-023-01419-x (abstract via Europe PMC)
• Huang et al., Science 2024, DOI 10.1126/science.adm8386 (abstract via Europe PMC; preprint for the 30-fold genome figure and the primate cell data); Broad Institute release
• Not yet verified: the Matsuzaki and Hordeaux papers' own text; funding statements; any primate data for BI-hTFR1; whether the preprint's genome-count figure is unchanged in the published paper.