Cell-Specific Targeting of Genetically Encoded Tools for Neuroscience

Abstract

Genetically encoded tools for visualizing and manipulating neurons in vivo have led to significant advances in neuroscience, in large part because of the ability to target expression to specific cell populations of interest. Current methods enable targeting based on marker gene expression, development, anatomical projection pattern, synaptic connectivity, and recent activity as well as combinations of these factors. Here, we review these methods, focusing on issues of practical implementation as well as areas for future improvement.

Keywords: actuator; chemogenetics; effector; optogenetics; reporter; transgene expression; viral vector.

Figure 1. The relationship between “cell type” and gene expression is complex

A) Neocortical GABAergic interneurons illustrate the hierarchical organization of neuronal identity. Due to this organization, the number of “cell types” can be correctly asserted to be arbitrarily large or small, highlighting the heuristic nature of the very concept of “cell type.” Figure reproduced from Rudy et al. 2011 {Rudy:2011gr}. B) Single-cell RNAseq reveals that common marker genes are expressed across multiple subclasses of interneurons. The results suggest novel candidate marker genes that may allow more precise targeting of diverse neuronal subpopulations. Figure reproduced from Zeisel et al. 2015.

Figure 2. Targeting based on expression of endogenous genes

A) Knock-in lines use site-specific integration to insert the transgene of interest directly into the marker gene locus, causing the transgene to be expressed wherever the marker gene is expressed. Use of a self-cleaving 2A sequence reduces alterations in expression of the endogenous gene. B) Bacterial artificial chromosome (BAC) transgenic lines insert the transgene of interest into the marker gene in a large genomic fragment maintained in a BAC. The BAC is then inserted into a random genomic location. C) Viral vectors can be targeted using short promoter sequences that replicate the endogenous gene expression pattern. D) Two-component driver/responder systems allow versatility through combinatorial use of a selectively-expressed driver, which activates transgene expression, and a responder/reporter, which contains the transgene of interest. Prototypical examples include the GAL4-UAS system, the Cre-Lox system, and the tTA-TRE/tetO system. Both the driver and responder/reporter can be either genomic or encoded in a viral vector. E) One of the most popular responder viral vector strategies is a Cre-activated “FLEX” or “DIO” switch. The trans-gene of interest is initially inactive in an inverted orientation, and the presence of Cre irreversibly flips the transgene into the correct orientation, activating expression. This configuration has been used with great success in AAV but is less effective in some other systems, possibly because efficient inversion of the transgene may rely on intermolecular recombination between multiple copies of the viral genome.

Figure 3. Targeting based on development

A) In in utero electroporation, naked plasmid DNA is injected into the brain of a mouse pup at a specific day of embryonic development, and a series of electrical pulses is applied to facilitate plasmid uptake. Since plasmid is not retained in actively mitotic cells, this method targets transgene expression to neurons born on the day of electroporation. B) Another method to target cells at specific developmental timepoints is through the use of mouse lines expressing CreERT2 under the control of developmentally regulated genes. Since CreERT2 is inactive until induced with a pulse of tamoxifen, this method targets neurons expressing maximal levels of CreERT2 at the time of tamoxifen induction, with a possible bias for actively dividing cells.

Figure 4. Targeting based on anatomy

A) A common method for targeting optogenetic activation to a population of cells projecting from one area to another. A viral vector encoding ChR2 is injected into the first area, and an optical fiber is implanted above the second area. Illumination of the axon terminals in the second area will evoke action potentials selectively in the cells projecting from the first area. B) Another method to achieve the same goal is to inject a retrograde viral vector encoding ChR2 into the second area and implant an optical fiber into the first area. A common variation of this approach uses a retrograde viral vector encoding Cre recombinase. C) An approach for monosynaptic retrograde targeting involves first injecting AAVs or plasmids encoding the TVA receptor, rabies glycoprotein G, and mCherry. These infect a population of starter cells (red). Later, EnvA-pseudotyped rabies or VSV vector encoding GFP (green) is injected, infecting TVA-positive starter cells and traveling retrogradely over one synapse. D) For monosynaptic anterograde targeting, AAVs encoding the TVA receptor, VSV glycoprotein G, and mCherry are injected. The area is later infected with EnvA-pseudotyped rabies/VSV encoding GFP (green), infecting TVA-positive starter cells and traveling anterogradely over one synapse.

Figure 5. Targeting based on recent activity

A) Activity-dependent transgene expression relies on inserting a chemically inducible driver gene such as tTA or CreERT2 into an immediate early gene, which is rapidly upregulated upon increased neuronal firing. B) In a typical experiment, the animal is exposed to a context or stimulus paired with doxycycline withdrawal or a tamoxifen pulse. This expresses ChR2 preferentially in the cells most active during conditioning. After sufficient ChR2 protein has accrued, blue light illumination enables activation of the ensemble of neurons that were active during conditioning.