🧬 AAV Vector Capsid Engineering

The AAV Capsid — A 60-Subunit Icosahedral Shell With Serotype-Specific Surfaces

Adeno-associated virus is a small, non-enveloped parvovirus whose protein shell — the capsid — is assembled from 60 copies of three overlapping viral proteins (VP1, VP2, VP3) in a T=1 icosahedral lattice. Enclosed within is a ~4.7 kb single-stranded DNA genome. Because AAV cannot replicate without a helper virus and integrates only rarely, recombinant AAV (rAAV) vectors — genome replaced with a therapeutic transgene — have become a leading gene therapy delivery platform. Critically, the outward-facing capsid surface differs substantially between naturally occurring serotypes, and this surface variation is the raw material capsid engineers work with.

• 60: Capsid subunits (VP1:VP2:VP3 ≈ 1:1:10 ratio)
• ~4.7 kb: Genome size (single-stranded DNA)
• 12+: Known natural serotypes (AAV1–AAV9, rh10, others)
• ~25 nm: Capsid diameter (one of the smallest viral vectors)

Capsid architecture and the origin of serotype diversity

Structural organization: • VP1 (largest, ~87 kDa): contains a phospholipase A2 domain needed for endosomal escape, buried internally until triggered • VP2 (~72 kDa): nuclear localization contribution • VP3 (~62 kDa): forms the bulk of the outer shell and carries essentially all receptor-binding surface loops • The three proteins share a common C-terminal sequence — VP1 and VP2 are simply N-terminal extensions of VP3

Surface topology: • The capsid surface has raised protrusions at three-fold symmetry axes, depressions ("canyons") at two-fold axes, and pore-like channels at five-fold axes • Nine hypervariable regions (HVR-I through HVR-IX) sit on these surface loops and account for most of the sequence divergence between serotypes • These hypervariable loops are the primary determinants of both receptor engagement (tropism) and antibody epitopes (immunogenicity) — the same real estate governs both properties

Why natural diversity exists: • Serotypes were isolated from different human and non-human primate tissues (AAV2 from human tissue, AAV8 and AAV9 have primate/human origins with distinct tropism) • Millions of years of co-evolution with host immune systems and receptor availability shaped each serotype toward a different balance of infectivity and immune escape • This natural library of surface variants is the starting point capsid engineers now use as raw material, and the reason no single natural serotype is optimal for every therapeutic application.

Tissue Tropism — How Surface Loops Decide Which Cells the Vector Enters

"Tropism" describes which cell types and tissues a vector preferentially transduces. For AAV, tropism is set almost entirely by the capsid surface: primary attachment receptors (often glycans such as heparan sulfate proteoglycan, sialic acid, or galactose) and secondary co-receptors (proteinaceous, e.g. AAVR) are engaged by the hypervariable surface loops. Because each natural serotype presents a different loop sequence, each one defaults to a different tissue preference — a property gene therapy developers must either exploit or overcome depending on the target organ.

• CNS / heart / muscle: AAV9 natural tropism (crosses blood-brain barrier)
• Liver (hepatocytes): AAV8 natural tropism (high default liver uptake)
• Retina / CNS (local): AAV2 natural tropism (first FDA-approved AAV product)
• Required: Universal AAVR co-receptor (nearly all serotypes use it)

Receptor engagement and the mechanics of tissue selectivity

Two-step entry model: • Primary attachment: low-affinity, high-avidity binding to abundant cell-surface glycans concentrates virions at the cell membrane (e.g., AAV2 binds heparan sulfate proteoglycan; AAV1/6 bind sialic acid; AAV9 binds terminal galactose) • Co-receptor engagement: the universal AAV receptor (AAVR, a multi-domain transmembrane protein) is required by nearly all serotypes for productive internalization via clathrin-mediated endocytosis • The identity and abundance of these glycan and protein receptors differs across tissues, so a serotype optimized for one glycan (e.g., abundant in liver sinusoids) will default to hepatic uptake regardless of the intended target

Why tropism matters clinically: • Systemic (IV) administration exposes the entire vasculature — the liver is disproportionately exposed due to its filtering role and fenestrated sinusoidal endothelium, so most natural serotypes accumulate there by default • Off-target liver uptake wastes vector dose, raises hepatotoxicity risk, and can trigger dose-limiting immune responses • A serotype naturally well suited to muscle (e.g., AAV1/AAV6) may be a poor choice for CNS-directed therapy, and vice versa • This is precisely the constraint capsid engineering aims to break — decoupling "which tissue gets transduced" from "which serotype happens to exist in nature."

Redirecting Tropism — Directed Evolution and Rational Capsid Loop Engineering

Because tropism is encoded in a handful of surface loops, engineers can modify those loops directly rather than relying on a naturally occurring serotype. Two complementary strategies dominate: directed evolution, in which enormous randomized capsid libraries are put through iterative rounds of selection in vivo or in target cell types, and rational design, in which structural and receptor-binding knowledge is used to install specific peptide insertions or point mutations at defined loop positions.

• 10⁷–10⁹: Peptide-display library size (randomized 7-mer insertions)
• 3–5: Typical selection rounds (in vivo / target-cell panning)
• ~4×: Off-target (liver) reduction (illustrative, engineered vs natural)
• ~1.7×: Target transduction gain (illustrative, engineered vs natural)

Directed evolution and rational design workflows for tropism redirection

Directed evolution (AAV peptide-display / capsid shuffling): • A library of AAV capsid genes is built with randomized short peptides inserted at a permissive surface loop (commonly within VP3 near residue 587 in AAV2 numbering) or with chimeric shuffled sequences from multiple parental serotypes • The resulting virus pool is administered to the target tissue or applied to target cells in vitro; capsids that transduce the desired cells are recovered by PCR-rescuing the packaged genome • Iterative rounds progressively enrich variants with improved target-cell binding and reduced off-target retention; barcoded libraries and next-generation sequencing now let many candidate loops be screened in parallel within a single animal

Rational structure-guided design: • Cryo-EM capsid structures at near-atomic resolution reveal exactly which surface residues contact known receptors, enabling targeted point mutations that ablate an undesired interaction (e.g., mutating the heparan-binding footprint) while preserving capsid assembly and genome packaging • Receptor-specific targeting ligands (single-domain antibodies, designed binders) can be genetically fused into an exposed loop to redirect the entire capsid toward a chosen surface marker • Detargeting mutations at galactose- or heparan-sulfate-binding residues reduce default liver sequestration independently of what new tropism is being introduced

Trade-offs engineers must balance: • Inserted loops must not disrupt capsid self-assembly, genome packaging, or the conformational changes needed for endosomal escape • Gains in target-tissue transduction and reductions in off-target (liver) uptake are illustrative and vary widely by construct — every engineered variant requires its own empirical biodistribution study.

Pre-Existing Neutralizing Antibodies — A Major Treatment-Limiting Barrier

AAV is a common, largely non-pathogenic human infection, so a substantial fraction of the population has already been exposed to one or more natural serotypes through casual environmental contact. That prior exposure leaves behind circulating IgG neutralizing antibodies (NAbs) that recognize the same hypervariable surface loops responsible for tropism. When a patient with pre-existing NAbs receives an AAV vector built on that serotype, circulating antibody coats the capsid before it reaches target cells, blocking receptor engagement and cell entry — turning a molecularly sound therapy into a clinical failure for that patient.

• 30–70%: Seroprevalence, common serotypes (population-dependent, illustrative)
• ~1:5 – 1:20: NAb titer exclusion threshold (protocol-dependent cutoff)
• Majority blocked: Transduction loss at high titer (illustrative, natural capsid)
• Seropositivity: Typical trial screen-failure driver (common exclusion reason)

How pre-existing antibodies neutralize AAV vectors and constrain patient eligibility

Mechanism of neutralization: • Circulating IgG antibodies bind the same exposed hypervariable loops that mediate receptor attachment, either sterically blocking receptor engagement or preventing the conformational changes needed for endosomal escape • Because NAb epitopes overlap heavily with tropism-determining loops, neutralization can occur even at low circulating titers — a small number of bound antibodies per capsid is often sufficient to abolish transduction • Unlike a one-time immune clearance event, this blockade happens before the vector ever reaches a target cell, so no amount of dose "getting through" the periphery compensates — it is an all-or-nothing filter at the point of cell entry

Clinical consequence — patient exclusion: • Most current AAV gene therapy trials screen for pre-existing NAb titers and exclude seropositive patients above a defined cutoff, since dosing them wastes vector, risks triggering an anamnestic immune response, and produces no therapeutic benefit • Seroprevalence varies by serotype, geography, and age, but commonly renders a meaningful share of otherwise-eligible patients ineligible for treatment with a given natural serotype • Because AAV cannot currently be re-dosed once an immune response is mounted (the first exposure itself boosts NAb titers), a single failed or excluded treatment attempt effectively closes the door on future AAV-based retreatment with that serotype • This constraint has motivated substantial investment in reducing capsid immunogenicity directly, rather than only screening it out.

Because tropism-determining loops and antibody-binding epitopes overlap on the same capsid surface, engineering one property can inadvertently affect the other — successful capsid engineering programs must optimize tropism and immunogenicity together, not in isolation.

Engineering Capsids to Evade Pre-Existing Antibodies and Expand Eligibility

Just as surface loops can be reshaped to redirect tropism, they can also be modified to reduce recognition by pre-existing neutralizing antibodies — without necessarily abandoning the tropism gains already engineered in. By mapping the specific epitopes bound by common human sera and then mutating, shielding, or swapping those residues (while preserving receptor-binding function elsewhere on the capsid), engineers aim to produce variants that slip past a patient's existing antibody repertoire. This can restore access to AAV gene therapy for patients who would otherwise be excluded by seropositivity.

• Cryo-EM + serum panels: NAb epitope mapping method (defines residues to modify)
• Polymer coating: PEGylation / shielding approach (sterically masks epitopes)
• Meaningful ↑: Illustrative eligible-pool gain (fewer seropositive exclusions)
• Reduced, not zero: Residual neutralization risk (partial antibody escape)

Strategies for reducing capsid immunogenicity while preserving function

Epitope-directed mutagenesis: • High-resolution cryo-EM structures of capsid–Fab (antibody fragment) complexes pinpoint the exact residues contacted by neutralizing antibodies from patient sera panels • Point mutations or short loop swaps at these specific epitopes can abolish antibody binding while leaving receptor-engaging residues elsewhere on the capsid intact — decoupling immune escape from tropism where the two footprints do not fully overlap • Chimeric capsids built by shuffling segments from multiple natural serotypes can combine favorable tropism from one parent with reduced cross-reactivity to common antibody repertoires from another

Chemical and physical shielding: • Covalent PEGylation (attaching polyethylene glycol polymers to surface lysines) sterically masks epitopes and reduces antibody accessibility, at some cost to receptor engagement that must be balanced empirically • Exosome- or lipid-associated encapsidation of AAV particles physically hides the capsid from circulating antibody until closer to the target cell

Clinical payoff — expanding the eligible population: • Directed-evolution or engineered capsids selected specifically against pooled human sera panels (rather than only against target-cell binding) can restore meaningful transduction in samples that fully neutralize the natural parent serotype • This directly translates into practical benefit: patients previously excluded by a positive NAb screen for a given serotype may become eligible for treatment with an immune-evasion-engineered variant • Reduction in immunogenicity is rarely complete — engineered variants typically lower but do not eliminate neutralization risk, so pre-treatment antibody screening remains part of clinical practice even with improved capsids.

The long-term goal of capsid engineering is a single design cycle that jointly optimizes three properties — target-tissue transduction, off-target (especially hepatic) avoidance, and resistance to pre-existing immunity — since each property is governed by overlapping regions of the same 60-subunit protein shell.