Retraction of synapses and dendritic spines induced by off-target effects of RNA interference

Abstract

RNA interference (RNAi), which allows selective gene silencing, has been proposed for functional genomic analysis and for the treatment of human disease. However, induction of RNAi in mammalian cells by expression of double-stranded RNA can activate innate antiviral response pathways that perturb off-target gene expression. The activation and functional consequences of these effects in neurons are unknown. We find that expression of subsets of short hairpin RNAs (shRNAs) in rat hippocampal pyramidal neurons can have off-target effects that reduce the complexity of dendritic arbors and trigger the loss of dendritic spines. Morphological changes are accompanied by electrophysiological perturbations in passive membrane properties and a decrease in the number and strength of excitatory and inhibitory synapses. These perturbations depend on the shRNA sequence and are independent of the identity of the targeted protein. Our results indicate that off-target effects of RNAi severely perturb neuronal structure and function and may lead to the functional withdrawal of affected cells from the brain circuitry.

Figure 1.

shRNA expression induces retraction of dendrites and spines. A, Hippocampal pyramidal neurons in organotypic slice cultures transfected with control EGFP vector (left) or shLUCI (right) imaged at 7 DPT. B, Representative images of apical dendrites from neurons expressing EGFP (left) or shLUCI (right) vectors. C, Average number of dendritic branches for EGFP (black) and shLUCI (gray) neurons at 7 DPT. D, Cumulative distribution of spine density at 4 DPT for neurons transfected with EGFP (filled circles) and shLUCI (open circles) (n = 14–16). E, Cumulative distribution of lengths of apical dendritic spines at 4 DPT for neurons transfected with EGFP (black) and shLUCI (gray) (n = 2107 and 1582 spines, respectively). F, Summary data of average spine densities at 4, 7, and 14 DPT for EGFP (black bars) and shLUCI (gray bars) neurons (n = 7 and 16). G, Summary data of average spine length at 4, 7, and 14 DPT for EGFP (black bars) and shLUCI (gray bars) neurons (n = 7 and 16). H, Spine density relative to EGFP controls for neurons expressing shLUCI or coexpressing a dominant-negative mutant of PKR (PKR-DN) with shLUCI or EGFP at 7 DPT (n = 8–15). The shaded gray region shows 2 × SEM for control spine density. ∗p < 0.05 compared with control.

Figure 2.

Frequency and amplitude of mEPSCs and mIPSCs are reduced by shLUCI. A, B, Representative traces from control (top) and shLUCI-transfected (bottom) pyramidal neurons displaying mEPSCs (A) and mIPSCs (B). Calibration: A, 10 pA and 500 ms; B, 20 pA and 200 ms. C, Summary of the amplitude (left) and frequency (right) of mEPSCs and mIPSCs in control (black bars) and shLUCI-expressing (gray bars) neurons (n = 11–15). ∗p < 0.05 compared with control.

Figure 3.

Sequence dependence of the morphological perturbations and the activation of an interferon response. A, Representative whole-cell (left) and apical dendrite (right) images of neurons transfected with shMORA, shMORB, and shEGFP at 7 DPT. B, Top, Summary of the average number of dendritic branches in neurons transfected with the indicated plasmids at 7 DPT (n = 5–12). Middle, Average spine density of neurons transfected with the indicated plasmids and imaged at 7 DPT. Bottom, Relative levels of luciferase activity in dissociated hippocampal neurons at 7 DIV and 24 h after cotransfection with the interferon-inducible reporter (mx2::luciferase) and EGFP (control), shMORA, shMORB, or shEGFP (n = 3–4 independent repetitions). ∗p < 0.05 compared with control.