Designer indicators for two-photon recording of subthreshold voltage dynamics

The authors developed and validated two novel genetically encoded voltage indicators, JEDI3sub and JEDI3hyp, which overcome previous sensitivity limitations to enable high-resolution, two-photon recording of millivolt-scale subthreshold voltage dynamics across diverse neuronal structures in awake, behaving mice.

The Gist

Imagine the brain as a bustling, high-tech city. For a long time, scientists could only listen to the "loud noises" of this city: the loud, all-or-nothing shouts of neurons firing (called action potentials or spikes). These are like the city's fire alarms or siren blares—easy to hear from far away.

But the real magic of the city happens in the quiet whispers and subtle murmurs between the sirens. These are the subthreshold voltage dynamics: tiny, millivolt-scale changes in a neuron's electrical state that happen before it decides to shout. These whispers determine how neurons integrate information, learn, and process the world. Until now, trying to hear these whispers deep inside the brain was like trying to hear a single person whisper in a crowded stadium while wearing noise-canceling headphones.

This paper introduces a new set of "super-earphones" called JEDI3 indicators (specifically JEDI3sub and JEDI3hyp) that finally allow scientists to hear those whispers clearly, even deep inside the brain of a living, moving mouse.

Here is a simple breakdown of how they did it and what they found:

1. The Problem: The Old Microphone Was Too Clunky

Scientists previously used tools called GEVIs (Genetically Encoded Voltage Indicators). Think of these as tiny microphones attached to neurons that glow brighter or dimmer when the electricity changes.

• The Issue: The previous version, JEDI-2P, was great at hearing the loud "sirens" (spikes), but it was too insensitive to hear the "whispers" (subthreshold changes). It was like a microphone that only picked up shouting but missed the conversation happening right next to it.
• The Goal: They needed a microphone sensitive enough to detect changes as small as a few millivolts, which is the size of the "whispers" that happen when neurons receive input from other cells.

2. The Solution: Evolution in a Test Tube

The team didn't just guess how to fix the microphone; they used high-throughput screening, which is like a massive, automated "survival of the fittest" contest for proteins.

• The Lab: They created nearly 100 different versions of the sensor by making tiny mutations (like changing a single letter in a word) in the protein's DNA.
• The Test: They put these sensors into cells and gave them tiny, weak electrical shocks (simulating the "whispers").
• The Winners: Two sensors emerged as champions: JEDI3sub and JEDI3hyp.
  • JEDI3sub is the "All-Rounder." It's incredibly sensitive to both the whispers and the shouts. It's the best choice for most situations.
  • JEDI3hyp is the "Specialist." It is hyper-sensitive to "hyperpolarization" (when a neuron gets quieter or more negative). It's like a specialized ear that can hear the silence between the notes, perfect for studying specific types of brain cells that operate at very low voltages.

3. The Results: Hearing the Brain's Secret Language

The team tested these new sensors in awake, behaving mice (mice that were running on a wheel, not sedated). Here is what they discovered:

• The "Whispers" are Everywhere: They could finally see the tiny electrical fluctuations in neurons that happen before a spike. This is like finally seeing the ripples in a pond before the splash.
• The Brain's "Mood" (Brain State): They found that a neuron's electrical "mood" changes depending on what the mouse is doing.
  • When a mouse's pupil dilates (gets bigger, usually meaning it's alert or excited), the neurons in the deep layers of the brain change their electrical rhythm.
  • Some neurons got "quieter" (hyperpolarized), while others got "louder" (depolarized). This suggests the brain has different subgroups of cells that react differently to the same alert signal.
• The "Memory Replay" (Sharp-Wave Ripples): In the hippocampus (the brain's memory center), they watched neurons during "sharp-wave ripples." These are moments when the brain replays memories, often while sleeping or resting.
  • Using JEDI3hyp, they saw that specific inhibitory neurons (the "brakes" of the brain) get a tiny electrical "jolt" right before the ripple starts, then get "quieted down" immediately after. This gives scientists a new map of how the brain controls memory replay.
• Seeing the Whole Tree: They didn't just look at the neuron's body (the trunk); they looked at the dendrites (the branches). They could see electrical signals traveling up and down the branches, showing how information flows from the tips of the dendrites to the cell body in real-time.

4. Why This Matters

Think of the brain as a complex orchestra.

• Before: Scientists could only hear the conductor's loud baton hits (the spikes).
• Now: With JEDI3, they can hear the individual instruments tuning up, the subtle harmonies, and the quiet cues that tell the orchestra when to speed up or slow down.

This technology opens the door to understanding:

• How we learn: By seeing the tiny electrical changes that happen before a memory is formed.
• Diseases: Many neurological disorders (like epilepsy, Alzheimer's, or autism) might start with these "whispers" going wrong long before the "sirens" (seizures or cell death) start.
• Deep Brain Function: It allows scientists to look deep inside the brain (where the "quiet" layers are) without needing to stick a needle into the brain, which is painful and damaging.

In summary: The authors built a pair of "super-sensors" that act like high-fidelity microphones for the brain's quietest conversations. By using these, they have unlocked a new layer of understanding about how our brains process information, remember things, and react to the world, all without ever having to stop the mouse from running.

Technical Summary

1. Problem Statement

While genetically encoded calcium indicators (GECIs) have revolutionized the study of neuronal spiking activity, they fail to report subthreshold voltage dynamics (graded changes in membrane potential below the action potential threshold). These subthreshold events (EPSPs, IPSPs, and low-frequency oscillations) are critical for understanding synaptic integration, dendritic computation, and network synchronization.

Existing tools for subthreshold recording face significant limitations:

• Patch-clamp electrophysiology: Offers high temporal resolution but is technically demanding, low-throughput, invasive, and difficult to perform in deep brain layers or on small dendrites in awake, behaving animals.
• Current Genetically Encoded Voltage Indicators (GEVIs): Previous two-photon (2P) optimized GEVIs (e.g., JEDI-2P) were primarily tuned for action potential detection. They lacked the sensitivity to reliably detect millivolt-scale subthreshold changes (typically <10 mV) required to study phenomena like inhibitory postsynaptic potentials (IPSPs) or brain-state-dependent fluctuations.

2. Methodology

The authors developed a multiparametric two-photon high-throughput screening platform to engineer and select novel GEVIs specifically optimized for subthreshold sensitivity.

• Screening Strategy:
  • Base Sensor: Started with JEDI-2P, a proven 2P GEVI for deep-tissue spike recording.
  • Library Construction: Generated nearly 100 libraries via saturation mutagenesis (targeting the voltage-sensing domain and the GFP moiety) and combinatorial mutation.
  • Assay Conditions: Screened variants in HEK293-Kir2.1 cells held at a resting potential of -82.6 mV. Cells were subjected to brief (~10 ms) electrical field stimulations of varying strengths.
  • Selection Criteria: Prioritized variants showing robust fluorescence responses to weak stimulations (simulating subthreshold depolarizations) while maintaining brightness, photostability, and fast kinetics.
• Characterization:
  • In Vitro: Used whole-cell voltage clamp combined with 2P imaging to measure excitation spectra, brightness, photostability, and response kinetics (rise/decay times) across various voltage steps (-100 mV to +30 mV).
  • In Vivo: Validated performance in awake, behaving mice using multiple imaging modalities:
    • ULoVE (Ultrafast Local Volume Excitation): For high-speed, single-cell recording in cortical Layer 2/3.
    • FACED (Free-space Angular-Chirp-Enhanced Delay): For large-population imaging (hundreds of cells) in the visual cortex.
    • Resonant Scanning 2P Microscopy: For recording hippocampal interneurons and deep-layer (Layer 5) somas/dendrites.

3. Key Contributions: JEDI3sub and JEDI3hyp

The screening yielded two novel indicators, JEDI3sub and JEDI3hyp, which differ from JEDI-2P by 7–8 specific mutations.

• JEDI3sub: Optimized for general subthreshold detection with enhanced sensitivity to depolarizations.
• JEDI3hyp: Optimized for hyperpolarization sensitivity and performance at more polarized potentials (below -80 mV), particularly at physiological temperatures.
• Mutations: Key mutations include:
  • Voltage Sensing Domain (VSD): S144V, M395T, R88S, V399F.
  • GFP Domain: N150R (restores brightness), K162R, T209A, C295S.
• Performance Metrics:
  • Sensitivity: JEDI3 indicators produced 2.7x to 3.6x larger fluorescence responses to 10-mV subthreshold steps compared to JEDI-2P.
  • Dynamic Range: JEDI3sub achieved a dynamic range of ~60.6% in vivo, a 1.2-fold improvement over JEDI-2P.
  • Kinetics: Maintained fast response times suitable for tracking spikes (though slightly slower than JEDI-2P for subthreshold steps, they remain physiologically relevant).
  • Photostability: Comparable to JEDI-2P, allowing for extended recordings (>30 mins).

4. Key Results

A. Enhanced Subthreshold Detection in Awake Mice

• ULoVE Recordings: In cortical Layer 2/3, JEDI3sub demonstrated a 1.7-fold larger subthreshold response than JEDI-2P. It outperformed the recently reported ASAP5 indicator in subthreshold amplitude (1.8-fold improvement).
• Population Imaging (FACED): JEDI3sub successfully tracked subthreshold activity in 95 neurons simultaneously in the visual cortex during drifting grating stimuli.

  • 92% of cells showed visually evoked subthreshold activity (compared to ~49% for calcium-based spike detection).

  • Revealed population-level correlations and orientation selectivity at the subthreshold level.

B. Hippocampal Sharp-Wave Ripples (SPW-R)

• Using JEDI3hyp in PV+ interneurons in the hippocampus (CA1), the authors recorded behaviorally relevant subthreshold dynamics during SPW-Rs.
• Findings: PV interneurons exhibited a stereotypical depolarization during the ripple (peaking 21 ms after onset) followed by a hyperpolarization (trough at 116 ms). This confirms the role of feedforward inhibition in ripple termination, a mechanism previously difficult to observe optically in interneurons.

C. Brain-State-Dependent Dynamics and Dendritic Imaging

• Pupil-Linked Oscillations: JEDI3hyp detected millivolt-scale (2–3 mV) subthreshold voltage changes in Layer 5 pyramidal neurons and interneurons correlated with pupil dilation/constriction (a proxy for brain state).
  • Layer 5 IT-PNs: Revealed two functional subclasses: some hyperpolarized and others depolarized during pupil dilation.
  • VIP vs. SOM Interneurons: Confirmed VIP depolarization and variable SOM responses during state changes.
• Somatodendritic Imaging: Using a dual-recombinase approach for sparse labeling, the authors simultaneously imaged Layer 5 somas and their apical dendrites in Layer 1.
  • Demonstrated that low-frequency (2–10 Hz) subthreshold oscillations are synchronized between soma and dendrite during quiet wakefulness, overcoming the technical limitations of patch-clamping dendrites.

5. Significance and Impact

• Bridging the Gap: JEDI3 indicators fill a critical void in the optical toolbox, enabling the study of millivolt-scale subthreshold dynamics in deep brain tissue using two-photon microscopy.
• Scalability: The sensors enable the transition from single-cell patch-clamp studies to population-level and subcellular (dendritic) optical recordings in awake, behaving animals.
• New Biological Insights: The technology has already uncovered:
  • Previously uncharacterized functional subclasses of Layer 5 neurons.
  • Detailed subthreshold mechanisms of hippocampal ripple generation and termination.
  • Synchronized dendritic-somatic voltage dynamics linked to brain states.
• Future Applications: These tools open avenues for investigating how subthreshold integration is altered in neurological disorders, how learning modifies synaptic strength, and the interplay between neuromodulators and circuit dynamics across multiple brain layers.

In summary, this work represents a major advancement in neurophotonics, providing robust, genetically encoded tools to visualize the "hidden" subthreshold computations that drive neuronal information processing in vivo.