AAV Serotype Comparison: Tropism, Receptors, and How to Choose

Selecting the correct AAV serotype is one of the most consequential decisions in any gene therapy or gene delivery experiment. The wrong choice means inadequate transduction of your target tissue, off-target expression elsewhere, or complete experimental failure. With more than 13 natural serotypes and dozens of engineered variants now available, understanding their distinct tropisms, primary receptors, and key limitations is essential before you design your experiment.

What Is an AAV Serotype? Receptors and Tropism Explained

An AAV serotype is defined by the amino acid sequence of its capsid proteins (VP1, VP2, VP3), which determine both antigenic identity and, critically, which host-cell receptors the virus binds. The primary receptor dictates tissue tropism — the spectrum of cell types and organs that each serotype can efficiently transduce. Because capsid proteins are the sole interface between the vector and the host cell, swapping the capsid (pseudotyping) is sufficient to redirect tropism entirely: the AAV2 genome packaged in an AAV9 capsid behaves like AAV9 in vivo. Tropism is also influenced by secondary receptors (co-receptors), the route of administration, promoter activity in the target cell, and host immune history.

Natural AAV Serotypes 1–13: Tropism and Primary Receptors

At least 13 naturally occurring AAV serotypes have been identified in primates and other species. The table below summarises their primary attachment receptors, main tissue tropisms, and notable research or clinical features. All serotypes listed are available in the BioHippo AAV catalog (4,000+ vectors from Vector Biolabs).

Engineered AAV Capsids: PHP.eB, AAV-DJ, Anc80L65, LK03, and rAAV2-retro

Directed evolution, rational design, and DNA shuffling have produced engineered capsids that outperform natural serotypes in specific niches. Here are the most widely used variants in research and clinical development.

AAV-PHP.eB — Unmatched CNS Transduction in C57BL/6 Mice (With a Critical Caveat)

PHP.eB was evolved from AAV9 using the CREATE (Cre-Recombination-based AAV Targeted Evolution) selection platform (Deverman et al., Nat Methods 2016). After a single intravenous injection in mice, PHP.eB achieves near pan-neuronal transduction of the entire CNS — orders of magnitude greater than AAV9 given systemically. However, its efficiency depends entirely on binding to LY6A (Sca-1), a GPI-anchored glycoprotein expressed on brain-vascular endothelium. LY6A expression varies with mouse genetic background.

Strain restriction — this is critical: PHP.eB works efficiently only in mouse strains that express LY6A on brain endothelium: C57BL/6 and FVB. It does not efficiently cross the BBB in CD-1, BALB/c, outbred ICR mice, non-human primates, or humans, which lack this LY6A haplotype on their brain vasculature. Hordeaux et al. (Mol Ther, 2019; DOI: 10.1016/j.ymthe.2019.02.013) demonstrated that the PHP.B/PHP.eB phenotype maps to a specific Ly6a haplotype and is completely absent in non-LY6A-expressing strains. Labs that have attempted to use PHP.eB in CD-1 or BALB/c mice or in NHP models have encountered complete loss of CNS-tropism advantage. If your model is not C57BL/6 or FVB, use AAV9 instead.

AAV-DJ — Enhanced In Vitro and Liver Transduction

AAV-DJ is a DNA-shuffled hybrid of AAV2, AAV8, and AAV9 capsids (Grimm et al., J Virol 2008). It exhibits enhanced transduction of a wide variety of cell lines in vitro compared to any single parental serotype, and efficient liver transduction in vivo. The modification of HSPG-binding residues reduces sequestration in the liver's sinusoids relative to AAV2, improving distribution. AAV-DJ/8 is a derivative that further reduces heparin binding, pushing distribution toward deeper hepatocyte access.

Anc80L65 — Ancestral Capsid for Retina and Inner Ear

Anc80L65 is a reconstructed ancestral AAV capsid derived computationally from phylogenetic analysis of known AAV sequences (Zinn et al., Cell Rep 2015). It efficiently transduces liver, retina (including difficult-to-reach Müller glia and photoreceptors), and inner ear hair cells — making it a promising vector for hearing loss gene therapy. Unlike AAV2, it does not rely on HSPG binding and may have lower pre-existing immunity in certain populations.

LK03 — Superior Human Hepatocyte Transduction

LK03 is a chimeric capsid evolved by serial selection in a humanised mouse liver model (FRG mice engrafted with primary human hepatocytes). Based on PubMed-indexed research by Lisowski et al. (Nature, 2014; DOI: 10.1038/nature12875), LK03 transduces human primary hepatocytes at dramatically higher efficiency than AAV2 or AAV8 in vivo, while AAV8 — though excellent in rodent liver — transduces human hepatocytes approximately 20-fold less efficiently than mouse hepatocytes. LK03 has entered clinical programmes for haemophilia and metabolic liver disease where human hepatocyte transduction efficiency is the primary bottleneck.

rAAV2-retro — Retrograde Circuit Tracing and Functional Manipulation

rAAV2-retro (Tervo et al., Neuron 2016) is a rationally engineered capsid that enables robust retrograde axonal transport after injection into axon terminals or projection target sites. Unlike conventional AAVs that transduce cells at or near the injection site, rAAV2-retro is taken up at axon terminals and transported retrogradely to the soma — labelling and transducing neurons projecting to the injection site. This makes it invaluable for anterograde-retrograde circuit dissection, in combination with DREADD chemogenetics or channelrhodopsins. Note: rAAV2-retro uses AAV2 ITRs but an engineered capsid sequence; it should not be confused with conventional AAV2.

Capsid Engineering Strategies: How New Serotypes Are Created

The engineered variants above emerged from three main approaches, and next-generation methods are rapidly expanding the toolkit:

• Directed evolution (in vivo selection): Large libraries of capsid variants (created by random mutagenesis or shuffling) are injected into animals, and variants that transduce the target tissue are recovered and re-selected over multiple rounds. The CREATE platform (Deverman et al. 2016) added a Cre-recombinase gatekeeping step, ensuring that only capsids that successfully entered a defined Cre-expressing cell type in vivo could replicate — producing PHP.B and PHP.eB.
• Rational design — surface mutations: Specific surface-exposed tyrosine residues on the AAV2 capsid (e.g., Y444F) are sites of proteasomal ubiquitination. Mutating these residues reduces proteasomal degradation of incoming virions, increasing effective transduction without altering tropism. Additional point mutations can reduce binding by pre-existing neutralising antibodies.
• DNA shuffling / chimeric capsids: Sequences from multiple parental serotypes are fragmented and reassembled in vitro, creating chimeric capsids that can combine beneficial properties (e.g., AAV-DJ combines AAV2 in-vitro potency with AAV8/9 in-vivo breadth). LK03 is similarly a five-way chimeric capsid.
• Machine learning-guided design: Emerging approaches from companies such as Dyno Therapeutics and Capsigen use deep learning on capsid sequence-function datasets to design novel capsid proteins in silico, bypassing the need for large-scale in vivo screening libraries.

AAV Serotype Comparison: A Decision Guide for Your Experiment

Use the table below to match your research goal to the most appropriate serotype(s). Always confirm serotype availability and titer with your supplier before beginning, and consider pre-existing neutralising antibody status for any clinical or NHP work.

BioHippo AAV Products by Serotype

BioHippo's AAV collection (4,000+ vectors) from Vector Biolabs covers all major natural and engineered serotypes discussed in this guide. Ready-to-use vectors with common promoters (CMV, CAG, Synapsin, CaMKII, EF1α) and reporters (GFP, mRFP, Cre, Cre-GFP, shRNA) are available in AAV1, AAV2, AAV5, AAV6, AAV8, AAV9, AAV-DJ, AAV-PHP.eB, AAV-Retro, Anc80L65, and more. Each product lists its serotype, titer (vg/mL), and functional validation data.

• AAV-GFP (AAV Serotype 9) — CNS-wide reporter for C57BL/6 and outbred mice
• AAV-GFP (AAV Serotype 8) — liver-preferring reporter
• AAV-GFP (AAV Serotype 2) — retina, CNS neurons
• AAV-GFP (AAV Serotype 5) — lung, retina, low seroprevalence
• AAV-GFP (AAV Serotype 1) — skeletal muscle, anterograde CNS tracing
• AAV-GFP (AAV Serotype 6) — skeletal muscle, airway epithelium

For more than 2,000 expression constructs — including cell-type-specific promoters (Synapsin, CaMKII, GFAP, TH) combined with any of the above serotypes — browse the full AAV catalog. Also see our companion posts: AAV Basics — A Beginner's Guide and Detection of Titer and Purity of Recombinant AAV.

Frequently Asked Questions

What is an AAV serotype?

An AAV serotype is a variant of adeno-associated virus distinguished by the amino acid sequence of its capsid proteins. The capsid determines which cell-surface receptors the virus binds, and therefore which tissues it preferentially transduces (tissue tropism). Different serotypes also carry distinct antigenic profiles, meaning the immune system recognises each differently — relevant for pre-existing immunity and re-dosing strategies. At least 13 natural serotypes and more than 100 engineered variants have been described to date.

Which AAV serotype is best for the brain?

The optimal serotype depends on your mouse strain and injection route. AAV9 is the broadly applicable choice for CNS transduction: after systemic intravenous injection, AAV9 crosses the blood-brain barrier in all tested mouse strains, rats, NHPs, and humans, though efficiency is moderate and substantial liver and heart transduction co-occurs. AAV-PHP.eB achieves vastly superior pan-neuronal transduction after IV injection in C57BL/6 and FVB mice only — it relies on LY6A receptor expression on brain endothelium. For direct intracranial or intracerebroventricular injection into any species, AAV1, AAV5, AAV8, or AAV9 are all effective depending on the target cell type.

What is the difference between AAV9 and PHP.eB?

PHP.eB was evolved from AAV9 and contains specific capsid mutations that allow it to bind LY6A (Sca-1) on brain vascular endothelial cells. In C57BL/6 mice, this dramatically increases BBB transcytosis, achieving orders-of-magnitude higher CNS transduction than AAV9 after the same IV dose. The practical difference: PHP.eB in C57BL/6 ≈ whole-brain labelling from a tail-vein injection; AAV9 IV in C57BL/6 ≈ moderate CNS labelling plus substantial peripheral organ transduction. In any non-LY6A-expressing model (CD-1, BALB/c, NHP, human), PHP.eB offers no advantage over AAV9 for CNS delivery and may perform worse.

Can AAV cross the blood-brain barrier?

Yes, but with serotype-specific efficiency and species-specific constraints. AAV9 crosses the BBB after intravenous injection in mice, rats, non-human primates, and humans — this is the basis for Zolgensma (SMA; FDA 2019). PHP.eB crosses the BBB with much greater efficiency, but only in LY6A-expressing mouse strains (C57BL/6, FVB). AAV8 crosses the BBB to a limited extent. For intracranial stereotaxic injection, the BBB is bypassed entirely and serotype choice shifts to optimising cell-type targeting and expression level rather than transport across the vascular wall.

What AAV serotypes are used in approved gene therapies?

As of mid-2025, the following AAV-based gene therapies have received regulatory approval:

• Luxturna (voretigene neparvovec-rzyl) — AAV2, subretinal, RPE65 mutation-associated retinal dystrophy; FDA 2017
• Zolgensma (onasemnogene abeparvovec) — AAV9, IV, spinal muscular atrophy (SMA); FDA 2019
• Hemgenix (etranacogene dezaparvovec) — AAV5, IV, haemophilia B (Factor IX Padua); FDA November 2022. Confirmed: AAV5-based, not AAV8. (Sekayan et al., Expert Opin Biol Ther 2023)
• Roctavian (valoctocogene roxaparvovec) — AAV8, IV, haemophilia A (Factor VIII); EMA 2022
• Elevidys (delandistrogene moxeparvovec) — AAVrh74 (rhesus macaque-derived), IV, Duchenne muscular dystrophy (DMD ages 4–5); FDA June 2023
• Glybera (alipogene tiparvovec) — AAV1, IM, lipoprotein lipase deficiency; EMA 2012 (first gene therapy approved in Europe; withdrawn from market 2017 for commercial reasons — no longer available)

References

1. Hordeaux J et al. The GPI-Linked Protein LY6A Drives AAV-PHP.B Transport across the Blood-Brain Barrier. Mol Ther. 2019;27(5):912–921. DOI: 10.1016/j.ymthe.2019.02.013
2. Lisowski L et al. Selection and evaluation of clinically relevant AAV variants in a xenograft liver model. Nature. 2014;506(7488):382–386. DOI: 10.1038/nature12875
3. Sekayan T et al. Etranacogene dezaparvovec-drlb gene therapy for patients with hemophilia B. Expert Opin Biol Ther. 2023;23(12):1173–1184. DOI: 10.1080/14712598.2023.2282138