Light-activated proteins (Collaboration with Karel Svoboda)

The goal of this project is to develop methods to target light-activated proteins, such as Channel Rhodopsin II (ChR2) and Halorhodopsin (NpHR), to specific subcellular compartments in neurons. Both ChR2 and NpHR generate electrical currents in response to light: ChR2, an ion channel, passes depolarizing cation current in response to blue light, whereas NpHR, a pump, generates hyperpolarizing Cl- currents in response to yellow light. Neurons expressing these proteins can be efficiently excited or inhibited with light. ChR2 and NpHR can thus be expressed in specific neuronal populations to determine if activity in these cells is sufficient and necessary to drive a function, for example a behavior. ChR2 has been used to map synaptic circuits in brain slices, by combining patch clamp recording of postsynaptic cells with stimulation of presynaptic neurons expressing ChR2. ChR2 can also be used to stimulate neurons in a spatial pattern in vivo, for example to determine motor maps.

A major limitation to the use of light-activated proteins in circuit mapping is their tendency to localize nonspecifically to different neuronal compartments. ChR2 and NpHR appear to be expressed equally well in axons and dendrites. In most neural tissues, dendrites and local and long-range axons are intermingled. The presence of these proteins in axons means that photostimulation can have non-local effects. Thus, it is virtually impossible to stimulate

Fig. 1 Circuit mapping in brain slices with ChR2
(A) Glutamate uncaging mapping.  The schematic shows a brain slice in which synaptic responses are recorded in a single neuron (red).Neurons are excited by photolysis of caged glutamate, typically using a UV laser that is scanned over the brain slice (blue line). If glutamate is photoreleased near the soma (but not on distal dendrites or axons) it evokes action potentials. Postsynaptic whole-cell currents (or potentials) recorded in the red neuron are used to generate a map in a computer. This so-called ‘synaptic input map’ is a quantitative representation of the spatial distribution of synaptic input to the recorded neuron.

(B) ChR2-assisted circuit mapping. A specific sub-population of neurons is targeted for expression of ChR2 (green). Only ChR2-positive neurons and axons (2, 3) are excited by a blue laser that is scanned over the brain slice (blue lines), whereas ChR2-negative neurons are not perturbed (1). Since severed axons can be excited (3) connectivity between distal brain regions can be studied even in a brain slice. However, because of axonal excitation, the spatial map contains little information about the positions of ChR2 even in a brain slice. However, because of axonal excitation, the spatial map contains little information about the positions of ChR2 positive neurons. (C) Circuit mapping with ChR2 targeted to the soma and / or dendrites. The map now provides a representation of the spatial distribution of ChR2 neurons making synapses with the recorded cell.

dendrites from one ChR2-positive cell without also stimulating neighboring ChR2-positive axons that can arise from distant and functionally unrelated neurons (Fig. 1). Similarly, action potential propagation can be blocked by photostimulation of NpHR-positive axons. It is therefore of great interest to generate versions of NpHR and ChR2 that can be excluded from axons, by targeting them to dendrites and somata. In other applications it is advantageous to specifically target light-activated proteins to axons.

Here we propose to generate peptides encoding signals that target light-activated proteins to specific subcellular compartments allowing neurons to be activated or inhibited for neural circuit analysis. Our goal is to target proteins to (1) the soma and dendrites, (2) axons or (3) either somatic or postsynaptic sites. We will accomplish (1) and (2) by forcing light-activated proteins to interact with Myosin motors, whereas (3) will be accomplished by fusing light-activated proteins to domains that either anchor transmembrane proteins to the cytoskeleton or restrict their diffusion to a confined area. Our recent work indicates that these strategies are likely to be successful: Myosin motors can target arbitrary transmembrane proteins to polarized compartments with great efficiency, and using mRNA display we have generated novel proteins that bind with high affinity and specificity to PSD-95, and that can be used to target heterologous proteins to postsynaptic sites.