AAV Tracing: How Anterograde and Retrograde Labelling Work

AAV tracing uses adeno-associated viral vectors to deliver a fluorescent protein gene into a chosen population of neurons, so that their cell bodies and axons become visible and their connections can be mapped. The decision that shapes every AAV tracing experiment is direction — whether the label travels from cell body to axon terminal, or from terminal back to cell body — and that is set by the capsid, not by the cargo. This guide covers how the method works, which serotype answers which question, the five-step workflow, and the interpretive limits that separate a defensible figure from an over-read one.

What is AAV tracing, and why the vector is not a dye

Adeno-associated virus is a small, non-enveloped virus with a single-stranded DNA genome. In the laboratory it is used as a vector: the viral genes are removed and replaced with a gene of interest, leaving only the protein shell (the capsid) and two short flanking inverted terminal repeats (ITRs).

The vector therefore carries no viral genes and cannot replicate. After it enters a neuron, the delivered DNA remains largely outside the chromosomes as a stable circular episome — one reason AAV expression is long-lasting and generally well tolerated in post-mitotic neurons.

In AAV tracing the delivered gene is usually a fluorescent protein such as GFP or mCherry. The neuron transcribes and translates it continuously, and because the protein fills the whole cell — including thin distal axons — the anatomy of the labelled population becomes visible under a microscope. This is the practical difference from a classical dye: the signal is manufactured by the cell rather than deposited into it, so it is renewed rather than diluted, and the same delivery logic can carry a sensor, an opsin, or a recombinase instead of a marker. Vectors across serotypes are listed in AAV Vector Research and the wider Adeno-Associated Viruses (AAVs) catalog.

Anterograde vs retrograde tracing: the capsid decides the direction

This is the question that decides the whole experiment, and it has three practical answers.

Anterograde labelling — mapping outputs

The vector transduces neurons at the injection site, and the fluorescent protein fills their axons, so the terminal fields reveal where that population projects. This is the standard approach to where does region A send its axons?

Most standard serotypes work in this direction. AAV1, AAV5, AAV8 and AAV9 are the usual choices. AAV2 is used when the labelled area must stay tightly restricted, because it spreads less from the needle tip; AAV5 is chosen when wider local coverage is wanted. A construct is written as serotype, promoter, then cargo — for example AAV5-hSyn-EGFP.

Retrograde labelling — mapping inputs with rAAV2-retro

Many brain regions contain several projection populations mixed together in the same space: neurons sending axons to one target sit next to neurons sending axons somewhere else, and neither position nor genetic markers reliably separate them. Injecting at the target and labelling backwards solves this — only the neurons that project to that target are labelled. This defines a population by its destination, and delivers sensors or effectors to exactly that pathway.

Natural AAV serotypes are taken up by axon terminals only weakly. The variant rAAV2-retro was produced by directed evolution specifically to solve this; it is efficiently internalised by axons and gives retrograde access to projection neurons with an efficiency the authors describe as comparable to classical synthetic retrograde tracers (Tervo et al., Neuron 2016). It answers which regions send axons into region A?

One published design. In the study that introduced rAAV2-retro, three vectors carrying three different fluorescent proteins were injected into three neighbouring striatal compartments of the same mouse:

Labelled neurons appeared in cortex, basolateral amygdala, and intralaminar thalamus. Because the three colours were in one brain, the authors could see directly whether the upstream neurons were separate populations or the same cells branching to several targets. No anterograde experiment answers that question, and no genetic marker separates these intermingled cells.

The same logic with a functional cargo. Replace the fluorescent protein with Cre. Inject rAAV2-retro-Cre into the target region, then inject a Cre-dependent vector such as AAV-DIO-hM4Di-mCherry into the source region. Only neurons that project to the target now carry the silencing receptor, and the fluorophore reports which cells those are — so the pathway can be switched off during behaviour.

Anterograde transsynaptic labelling — one step downstream

AAV1 and AAV9 were found to spread anterogradely across a synapse (Zingg et al., Neuron 2017). In the usual design, AAV1 carrying Cre is injected upstream and a Cre-dependent reporter is supplied downstream, either by a transgenic reporter line or by a second injected vector. Only postsynaptic neurons that receive input from the injected region are labelled, giving an input-defined population that can then be recorded from or manipulated.

AAV serotype tropism: choosing a capsid for AAV tracing

The table below is a starting point for reading the literature, not a specification. Tropism and spread depend on species, brain region, promoter, titre, and injection volume, so every combination should be validated in your own model.

The AAV tracing workflow, step by step

Five steps run from vector choice to imaging, and each one influences how the final image should be read.

1. Choose the vector — serotype, promoter, cargo

Three choices are made together: the serotype, which decides tropism and transport direction; the promoter, which decides which cell types express the gene; and the cargo, the fluorescent protein or effector.

Why it matters: CAG or CMV promoters drive strong, broad expression. A neuron-specific promoter such as hSyn restricts expression to neurons; CaMKIIa favours excitatory forebrain neurons and GFAP favours astrocytes. Promoter specificity is a strong bias, not an absolute guarantee, and should be confirmed in your own tissue.

2. Inject at a defined coordinate

A small volume — typically tens to a few hundred nanolitres — is delivered by stereotaxic injection. Volume, titre, and injection speed together determine how far the vector spreads from the needle tip.

Common pitfall: a larger volume does not simply give a brighter signal. It enlarges the effective injection site, which can make a projection look broader than it really is.

3. Entry and genome conversion

The capsid binds cell-surface receptors, the particle is internalised, and the single-stranded genome is converted to double-stranded DNA before transcription can start. This conversion step is one reason expression is not immediate.

4. Wait for expression

Signal usually becomes detectable after about one to two weeks. Most laboratories wait roughly three to four weeks before perfusion so that distal axons are fully filled; published AAV1 transsynaptic work commonly uses a four-week survival time (Zingg et al., Neuron 2017).

Practical note: a short survival time is a frequent cause of apparently weak results. Terminal fields fill later than cell bodies, so an early time point can under-report long-range projections.

5. Fix, section, and image

Native fluorescence is often sufficient. Many groups still amplify the signal by immunostaining against GFP or mCherry, particularly for thin axons, low-expressing cells, or thick cleared tissue — see the antibody catalog for anti-GFP and anti-RFP reagents.

Where AAV tracing is used in circuit neuroscience

Mapping the output of a defined population

A single anterograde injection shows the full terminal field of a region. Combined with a Cre driver line, the map becomes cell-type specific: only the genetically defined subpopulation is labelled, and only its axons are followed.

Identifying the sources of input

A retrograde vector injected into a target region labels the cell bodies that project into it. Injecting different fluorescent proteins into neighbouring targets, in the same animal, shows whether the upstream populations are separate or overlapping (Tervo et al. 2016).

Reaching input-defined populations

The AAV1 anterograde transsynaptic strategy labels downstream neurons by the input they receive, rather than by genetic identity or location alone. Those neurons can then be traced further, recorded, or manipulated (Zingg et al., Curr Protoc 2022).

Combining anatomy with function

Because the cargo is a gene, the same anatomical logic delivers functional tools: a calcium indicator for imaging, an opsin for optogenetics, or a chemogenetic receptor for reversible silencing. This is the main reason AAV replaced dye-based tracing in many laboratories — anatomy and causal manipulation come from one injection strategy.

Brain-wide labelling without stereotaxic surgery

Systemic delivery of blood–brain-barrier-crossing capsids labels neurons throughout the CNS. This is useful for whole-brain expression of a sensor or reporter, but it is not a projection-mapping tool, because the injection site no longer defines the labelled population.

What AAV tracing cannot tell you

These limits are the difference between a defensible figure and an over-interpreted one, and they are worth reading before the first injection.

• It is not a measure of synaptic strength. A dense terminal field shows that axons are present. It does not show how strong, or even how functional, those synapses are.
• Standard AAV tracing is not polysynaptic. AAV1 anterograde spread is described as reaching first-order downstream neurons (Zingg et al. 2020). For multi-step circuit tracing, other systems are used.
• Direction is rarely absolute. Several serotypes show some uptake in the non-intended direction. Anatomical controls, including injections into regions known not to project to the target, are needed to exclude this.
• Titre and volume shape the result. A high titre enlarges the effective injection site and can increase unintended uptake. Titre should be reported, and kept constant across compared groups.
• Recombinases are not inert. AAV-mediated Cre expression at commonly used titres has been reported to cause neuronal loss and behavioural changes in the injected region, independently of the intended floxed target. Cre-only and reporter-only control groups are therefore necessary (Rezai Amin et al., J Neurochem 2019).
• Results do not transfer between models automatically. The LY6A dependence of the PHP.B capsid family is the clearest example: a capsid can work well in one mouse strain and poorly in another.
• Cargo size is limited. Roughly 4.7 kb, including the ITRs, constrains what can be packaged in a single vector.

AAV tracing vs rabies, herpesvirus, and classical tracers

The methods are complementary, and many published circuit studies combine an AAV strategy with one of the others.

AAV vectors and constructs for tracing experiments

Ready-made constructs cover the standard designs described above — retrograde tracers, retrograde Cre, and Cre-dependent reporters and effectors for the second injection.

Confirm serotype, titre, promoter, and intended-use statements on each product page before purchase. If you would like help matching a serotype and construct to your model and pathway, request a quote with your species, target region, and projection of interest.

AAV tracing: frequently asked questions

What is AAV tracing, in one sentence?

AAV tracing is the use of adeno-associated viral vectors to deliver a fluorescent protein gene into a selected group of neurons, so that their cell bodies and axons can be followed and their connections mapped.

Is AAV tracing anterograde or retrograde?

Both are possible, and the capsid decides. Most standard serotypes are used for anterograde labelling. For retrograde labelling, the engineered variant rAAV2-retro was developed specifically to give efficient access to projection neurons through their axon terminals (Tervo et al. 2016).

Does AAV cross synapses?

AAV1 and AAV9 have been shown to spread anterogradely to directly connected postsynaptic neurons. The effect is used deliberately with a recombinase and a Cre-dependent reporter. Spread is described as first-order rather than polysynaptic, and it depends on the pathway (Zingg et al. 2020).

How long should I wait before perfusion?

Expression usually becomes visible after one to two weeks. Three to four weeks is a common survival time so that distal axons are fully labelled, and published AAV1 transsynaptic experiments frequently use four weeks. The optimal time depends on serotype, promoter, and the distance of the projection.

Which AAV serotype should I use for retrograde tracing?

rAAV2-retro is the engineered capsid designed for this purpose, and natural serotypes are taken up by terminals only weakly by comparison. Efficiency is nonetheless pathway-dependent, so confirm labelling in your specific projection before treating an absent signal as an absent projection.

When should I choose rabies virus instead of AAV?

Choose a deletion-mutant rabies system when you need monosynaptic input mapping onto a defined starter population with high labelling efficiency. Choose AAV when you need long survival times, low toxicity, or functional tools such as sensors and opsins in the same neurons.

Do I need antibody staining if my vector already expresses GFP?

Not always. Native fluorescence is often sufficient for cell bodies and proximal processes. Immunostaining against GFP or mCherry is commonly used to recover thin distal axons, weakly expressing cells, or signal in thick tissue.

Will a capsid that works in my mouse line work in another strain or species?

Not necessarily. The clearest documented example is the PHP.B capsid family, whose enhanced entry into the brain depends on the endothelial protein LY6A. That protein differs between mouse strains and is not present in primate brain endothelium, so the advantage does not transfer (Hordeaux et al. 2018).

Plain-language glossary of AAV tracing terms

References

1. Tervo DGR, Hwang BY, Viswanathan S, et al. A designer AAV variant permits efficient retrograde access to projection neurons. Neuron 2016;92(2):372–82. PMID 27720486 · doi:10.1016/j.neuron.2016.09.021
2. Zingg B, Chou XL, Zhang ZG, et al. AAV-mediated anterograde transsynaptic tagging: mapping corticocollicular input-defined neural pathways for defense behaviors. Neuron 2017;93(1):33–47. PMID 27989459 · doi:10.1016/j.neuron.2016.11.045
3. Zingg B, Peng B, Huang J, Tao HW, Zhang LI. Synaptic specificity and application of anterograde trans-synaptic AAV for probing neural circuitry. J Neurosci 2020;40(16):3250–67. PMID 32198185 · doi:10.1523/JNEUROSCI.2158-19.2020
4. Hordeaux J, Yuan Y, Clark PM, et al. The GPI-linked protein LY6A drives AAV-PHP.B transport across the blood-brain barrier. Mol Ther 2019;27(5):912–21. PMID 30819613 · doi:10.1016/j.ymthe.2019.02.013
5. Hordeaux J, Wang Q, Katz N, Buza EL, Bell P, Wilson JM. The neurotropic properties of AAV-PHP.B are limited to C57BL/6J mice. Mol Ther 2018;26(3):664–8. PMID 29428298 · doi:10.1016/j.ymthe.2018.01.018
6. Zingg B, Dong HW, Tao HW, Zhang LI. Application of AAV1 for anterograde transsynaptic circuit mapping and input-dependent neuronal cataloging. Curr Protoc 2022;2(1):e339. PMID 35044725 · doi:10.1002/cpz1.339
7. Rezai Amin S, Gruszczynski C, Guiard BP, et al. Viral vector-mediated Cre recombinase expression in substantia nigra induces lesions of the nigrostriatal pathway associated with perturbations of dopamine-related behaviors and hallmarks of programmed cell death. J Neurochem 2019;150(3):330–40. PMID 30748001 · doi:10.1111/jnc.14684

This guide is an orientation to published methodology and does not replace the original protocols. Serotype behaviour, titre requirements, and survival times must be validated in your own model, species, and pathway. Serotype behaviour summarised here reflects rodent studies. Confirm construct details, serotype, titre, and intended-use statements on each product page before purchase. Research use only unless otherwise stated on the product page.