Given the proximity of astrocyte processes to synaptic terminals (38) and that AAVs readily transduce astrocytes, we hypothesized that AAVs can undergo anterograde transport also to astrocytes. To test this hypothesis, we used AAV1, which was previously shown to exhibit robust anterograde trans-synaptic spread properties (35, 36) in the vibrissa thalamocortical system of mice. The ventral posterior medial nucleus (VPM) of the vibrissa somatosensory system innervates primarily L4 and L6a and sends less dense projections to other layers including L2/3 of BX (39).
First, we tested whether AAV1 injections in VPM can lead to transduction of BX astrocytes and neurons (Fig. 1A). We coinjected AAV1 delivering the Cre gene under the cytomegalovirus promoter (CMV) AAV1-CMV-Cre and AAV1delivering the gene of the red chromophore TurboRFP under the human synapsin promoter (hSyn) AAV1-hSyn-TurboRFP in VPM. If a small number of AAVs are transported anterogradely, then we would expect ubiquitous, CMV promoter–driven expression of Cre-recombinase in BX astrocytes and neurons that can in turn engage the flip-excision (FLEx) switch to drive conditional gene expression (40). AAV1-hSyn-TurboRFP was used to mark the thalamocortical axon projections.
To test whether Cre was present in BX astrocytes, we additionally injected AAV5-GFaABC1D-FLEx-lck-GCaMP6f in BX to induce Cre-dependent expression of the GECI GCaMP6f selectively in astrocyte plasma membranes (Fig. 1B). This intersectional approach led to sparse labeling (~4 × 10−6 cells/μm3) of L2/3 BX astrocytes 3 weeks after AAV injections (Fig. 1B and fig. S1, A to C). We observed the highest density of astrocytes in L4 (19 × 10−6 cells/μm3) (Fig. 1B and fig. S1, A to C), the principal projection layer of the VPM as indicated by the density of TurboRFP-labeled thalamocortical axons (fig. S1, A, B, D, and E). The GFaABC1D promoter induced astrocyte-specific gene expression, the FLEx system allowed conditional expression in a few astrocytes expressing Cre recombinase, and the lck tag led to membrane tagging of GCaMP6f, revealing the cloud-like morphology of astrocytic nanoscopic processes (Fig. 1B and figs. S1A and S2A) (18). TurboRFP labeled thalamocortical axon projections to reveal their characteristic barrel-like projection patterns (39) expected in L4 of BX (Fig. 1, B and C, middle). All AAV injections (in BX and VPM) were performed during one surgery. We then tested whether Cre was also present in BX neurons. To do so, we instead injected AAV9-hSyn-FLEx-GCaMP6f in BX. We used the neuron-specific hSyn and the FLEx system to drive neuron-specific and Cre-dependent GCaMP6f expression. This intersectional strategy led to BX neuron labeling (Fig. 1C and fig. S1, D and E). We found the highest density of neurons in L4 (~11 × 10−6 cells/μm3) and L5/6 (~9 × 10−6 cells/μm3) (fig. S1F). In both preparations, no obvious TurboRFP+ cell bodies were detected in BX. So, the intersectional approaches resulted in sparse but bright labeling of L2/3 astrocytes and neurons (~3 × 10−6 cells/μm3) of the BX (fig. S1).
To confirm that the labeled cells were astrocytes or neurons, we used antibody labeling against S100β (Ca2+-binding protein concentrated in astrocytes; fig. S2A) and NeuN (neuronal nuclear antigen; fig. S2C), respectively. 94% of GCaMP6f+ BX cells were S100β+ in brains injected with AAV5-GFaABC1D-FLEx-lck-GCaMP6f (fig. S2B). Similarly, 97% of GCaMP6f+ BX cells were NeuN+ in brains injected with AAV9-hSyn-FLEx-GCaMP6f (fig. S2D).
Next, we investigated whether AAVs injected only in VPM can transduce astrocytes and neurons in BX. Because we suspected that only a small number of AAVs transfer from VPM to BX cells, we used vectors carrying genes encoding fluorescent proteins under control of the strong, ubiquitous CMV early enhancer/chicken β-actin (CAG) promoter (41). We injected (i) AAV1-CAG-GCaMP6f and AAV1-hSyn-TurboRFP (fig. S1A), (ii) AAV1-CMV-Cre and AAV1-CAG-FLEx-eGFP (enhanced green fluorescent protein) (fig. S1B), (iii) AAV1-CMV-Cre and AAV1-CAG-FLEx.tdTomato (Fig. 2, A to E), or (iv) AAV1-CMV-Cre, AAV1-CAG-FLEx-GCaMP6f, and AAV1-hSyn-TurboRFP (fig. S6, A and B) in VPM. These injections led to both neuron and astrocyte labeling (assessed by morphology) in BX. Therefore, a second cortical injection (intersectional strategy), the Cre-FLEx system, or a specific fluorescent protein being expressed is not required. However, a single-injection strategy relies on the use of very strong promoters, like the CAG promoter, and cannot be used in its current form for cell type–specific labeling. Therefore, we focused on intersectional approaches for the functional probing of astrocytes in this study because we could specifically label astrocytes while having a lower risk of toxicity induced by gene overexpression. Together, these experiments provide indirect evidence that a small number of AAVs transfer anterogradely from VPM thalamocortical axons to BX astrocytes and neurons.
To show direct evidence of anterograde AAV1 transfer to BX cells, we antibody-tagged AAV capsids (anti-VP1): First, we coinjected AAV1-CMV-Cre and AAV1-CAG-FLEx-tdTomato only in VPM because this strategy led to tdTomato-labeled astrocytes and neurons in BX. We perfused the animals 1 hour, 24 hours, or 12 days after the injection and then labeled brain slices with anti-VP1 (Fig. 2, A and B). One hour after AAV1 injection, AAV capsids dispersed around the injection site but did not enter the cell nuclei (Fig. 2C). Within 24 hours after injection, AAVs permeated the…