Key tools for studying neural circuits combine Cre recombinase dependent expression of either: markers (GFP), optogenetic and chemogenetic actuators (e.g. channelrhodopsin and DREADDs) or biosensors (GCaMPs). A common design is to use AAVs to deliver the gene-of-interest that is made Cre dependent by flanking its inverted open-reading frame with specific Cre recognition sites (loxP/lox2272; DIO or FLEX design). Recent studies have sounded the alarm on leaky, non-Cre dependent expression by injecting these AAVs into wild-type C57BL/6 mice. Using this approach we find 6% off-target expression of FLEX- tdTomato (Addgene 28306). This level of off-target expression of FLEX/DIO AAVs leads to confounds in the interpretation of results, and coupled with their wide-spread use (>2,000 papers), highlights the need for improved Cre-dependent designs. Fixing the off-target leak begins with understanding two of the main sources of the problem. Problem #1 is that current FLEX sequences recombine during cloning, plasmid amplification, and AAV production (Fischer et al., 2019). To reduce recombination we replaced the lox sites with mutant sites that form shorter hairpins (lox71/66). In vitro testing of this FLEX2 design indicates spontaneous recombination is reduced 4-fold and expression after Cre recombination is improved 8-fold. Problem #2 is that the AAV ITR sequences have promotor activity (Haberman et al., 2000), which leads to the expression of the actuator even though it is oriented 3’-5’ relative to the neuronal promoter. Our solution was to combine a split intron and a split open-reading frame of the actuator into two fragments and then invert one of them. The split intron serves to remove residual lox sites from the final coding sequence. Preliminary data shows the effectiveness of this approach to deliver diphtheria toxin A chain to ablate hippocampal interneurons in VGAT-Cre mice. Approx. 50-70% of GABAergic neurons in the CA3 stratum radiatum and dentate hilus were ablated. These mice developed spontaneous recurring seizures (SRS) in the absence of any other precipitating cause. This supports the hypothesis that loss of GABAergic inhibition can cause epilepsy. The goal of this grant is apply the FLEX2 design to the most commonly used Cre-dependent AAV tools. In collaboration with Addgene, these plasmids will be used to produce AAV particles. Both the FLEX2 plasmids and AAV will be directly distributed by Addgene to the research community.