Figure 2. Validating Intersectional Constructs for Cell-Type-Specific Tracing In Vivo.

(A) Coronal section through the barrel cortex of a Vgat-Cre/PV-Flp mouse after injection of RV-EGFP, without prior injection of helper AAVs. No transduced cells were detected, demonstrating that RV-EGFP transduction is TVA-dependent (cell counts: n = 2 mice, 16 sections).

(B–D) Coronal sections of wild-type (WT; B), PV-Flp (C), and Vgat-Cre (D) mice after the injection of AAV-TVA-mCh and AAV-oG into barrel cortex and subsequent RV-EGFP. Even in the absence of all or one recombinase, TVA was expressed, allowing RV-EGFP to enter cells at the injection site. oG was not expressed, preventing RV-EGFP from spreading transsynaptically to cells outside the injection site (cell counts: WT, n = 4 mice, 24 sections; PV-Flp, n = 4 mice, 22 sections; Vgat-Cre, n = 5 mice, 30 sections).

(E) Injection of AAV-TVA-mCh alone (no AAV-oG that enables transsynaptic spread) into a Vgat-Cre/ PV-Flp animal followed by RV-EGFP injection. Cells were solely counted on sections with large numbers of double-labeled cells. Injections yielded mostly double-labeled, PV-positive cells but also a few mCherry-negative, EGFP-positive cells. These exclusively green cells made up on average 16% of all EGFP-positive cells and are a result of direct RV-EGFP entry, because of low-level expression of TVA in the absence of Cre/Flp. Because these cells do not have a mCherry signal, we termed this phenomenon ‘‘invisible TVA’’ (cell counts: n = 4 mice, 10 sections; scale bar inset, 20 μm).

(F) Injection of AAV-TVA-mCh and AAV-oG into a Vgat-Cre/PV-Flp animal followed by RV-EGFP injection. This injection reflects the experimental conditions. Cells were counted on all sections containing starter cells (cell counts: n = 12 mice, 237 sections).

Scale bars: overview, 1,000 μm; inset, 200 μm. Cell counts are mean of cells per section.