Transcranial magnetic stimulation to the dorsolateral prefrontal cortex modulates single-neuron activity in humans

This study demonstrates that transcranial magnetic stimulation to the dorsolateral prefrontal cortex directly modulates single-neuron activity in humans by facilitating inhibitory firing in executive control networks while suppressing excitatory drive in limbic circuits, thereby revealing the cellular mechanism underlying its therapeutic effects on distributed brain networks.

The Gist

The Big Picture: Fixing a Stuck Car Engine

Imagine your brain is a massive, complex city with different neighborhoods. One neighborhood, the Prefrontal Cortex (the "Boss" or "CEO"), is in charge of logic, planning, and keeping emotions in check. Another neighborhood, the Limbic System (the "Emotional Engine"), is where feelings like fear, sadness, and anxiety live.

In people with depression, the "Emotional Engine" often revs too high, while the "Boss" loses control. A common treatment called TMS (Transcranial Magnetic Stimulation) uses a magnetic "whack" on the head to try to fix this. We know it works, but for a long time, we didn't know how it worked inside the brain. Did it just calm the engine directly? Or did it send a message to a different part of the city to take control?

This study is the first time scientists have been able to "listen" to individual brain cells (neurons) in humans while giving them this magnetic whack. They discovered a fascinating two-step dance.

---

The Experiment: Listening to the Brain's Tiny Sparks

The researchers worked with four patients who were already in the hospital with tiny wires implanted in their brains to find the source of their seizures. These wires were like tiny microphones placed deep inside the brain's neighborhoods.

Instead of just listening to the "noise" of the whole city (which is what normal brain scans do), they listened to individual neurons firing their electrical sparks. They gave the patients a single magnetic pulse to the "Boss" area (the left side of the forehead) and watched what happened in the deep, hidden parts of the brain.

The Challenge: The magnetic pulse creates a huge electrical "static" noise that usually drowns out the tiny sparks of the neurons. The team invented a new way to clean up this static, allowing them to hear the neurons firing as quickly as 8 milliseconds after the pulse. That's faster than a blink of an eye!

---

The Discovery: The "Good Cop, Bad Cop" Routine

When they zapped the "Boss" area, they saw a very specific pattern of activity that looked like a well-rehearsed routine:

1. The Fast-Track "Good Cop" (Striato-Thalamic Network)

• What happened: Within milliseconds, the magnetic pulse woke up a specific group of neurons in the striatum and thalamus (deep brain hubs that act like traffic control centers).
• The Analogy: Imagine the "Boss" shouting, "Hey, Traffic Control! Wake up and take charge!" immediately.
• The Cells: These were mostly inhibitory neurons (the "brakes" of the brain). They started firing rapidly, peaking around 80–100 milliseconds.
• The Result: This is the brain's executive control system turning on. It's like the traffic cop stepping out to direct the flow of cars.

2. The Slow "Bad Cop" (Limbic Network)

• What happened: A little later (around 300 milliseconds), the neurons in the limbic system (the emotional engine) started to quiet down.
• The Analogy: Once the Traffic Control center (the Good Cop) is fully awake and directing traffic, it starts to slow down the reckless drivers in the Emotional Engine.
• The Cells: These were mostly pyramidal cells (the "gas pedal" or excitatory neurons). They stopped firing as much.
• The Result: The emotional over-reaction is suppressed, but only after the control system has been activated first.

The Key Takeaway: It's Not a Direct Hit

For a long time, people thought the magnetic pulse might be like a direct slap on the emotional engine to calm it down. This study shows that's not what's happening.

Instead, it's more like a relay race:

1. Step 1: The magnetic pulse wakes up the Executive Control Network (the traffic cops) first.
2. Step 2: These traffic cops then send a signal to slow down the Emotional Engine.

The study found that the "Traffic Cops" (striato-thalamic neurons) and the "Emotional Engine" (limbic neurons) are actually anti-correlated. When the cops get busy, the engine slows down. When the engine revs, the cops are quiet. The magnetic pulse flips the switch to get the cops working, which naturally quiets the engine.

Why This Matters

This explains why TMS is effective for depression. It doesn't just "turn down the volume" on sadness directly. Instead, it reboots the brain's management system. By waking up the deep brain circuits responsible for control and decision-making, the brain naturally gains the ability to regulate its own emotions.

In a nutshell: The magnetic pulse doesn't just calm the storm; it calls in the lifeguards (the deep brain control centers) who then calm the storm for you. This gives doctors a much clearer map of how to treat depression and other mood disorders in the future.

Technical Summary

1. Problem Statement

Transcranial Magnetic Stimulation (TMS) applied to the dorsolateral prefrontal cortex (dlPFC) is an FDA-cleared treatment for Major Depressive Disorder (MDD). However, the precise cellular mechanisms by which cortical stimulation influences deep brain circuits remain unknown.

• The Gap: While animal studies show TMS modulates local cortical neurons, and human fMRI/iEEG studies show network-level changes, no study has previously recorded single-neuron spiking in humans in response to dlPFC TMS.
• Competing Hypotheses:
  1. Direct Limbic Inhibition: TMS directly dampens hyperactive limbic structures (e.g., subgenual cingulate, amygdala) via cortico-limbic pathways.
  2. Top-Down Control (Striato-Thalamic): TMS engages cortico-striato-thalamic loops to restore executive control over affective systems, potentially gating limbic activity indirectly.
• Technical Challenge: Resolving single-neuron spikes immediately following TMS is difficult due to massive electromagnetic artifacts that typically obscure neural signals for tens of milliseconds.

2. Methodology

The study utilized a unique "opportunistic" recording setup in four neurosurgical patients with medication-resistant epilepsy who were undergoing intracranial monitoring (stereo-EEG).

• Participants: 4 patients (3 female, ages 35–56) implanted with depth electrodes containing microwire bundles (8 platinum-iridium microwires per bundle) in deep cortical and subcortical structures.
• Stimulation:
  • Target: Left dlPFC (using the modified Beam F3 coordinate).
  • Protocol: Single-pulse TMS (0.33–0.4 Hz) delivered via a MagVenture Cool-B65 coil.
  • Control: Sham TMS (coil flipped 180° to generate acoustic click without brain stimulation).
• Recording & Signal Processing:
  • Hardware: Simultaneous recording of macroelectrodes (field potentials) and microwires (single units).
  • Artifact Removal: A custom multi-stage algorithm adapted from scalp EEG-TMS studies was applied. It involved detecting saturation bounds, extending windows based on high-frequency residuals, and replacing data via autoregressive extrapolation.
  • Spike Sorting: Used Higher Order Spectral Decomposition (HOSD) for feature extraction and Gaussian Mixture Modeling (GMM) for clustering.
  • Timeline: Spikes were resolved as early as ~8 ms post-stimulation (previously impossible in humans).
• Analysis:
  • Cell Classification: Units were classified as putative Pyramidal cells (PY) or Interneurons (IN) based on waveform features (trough-to-peak duration, burst index, firing rate).
  • Network Correlation: Trial-by-trial correlations were computed between single-unit firing rates and Intracranial TMS-Evoked Potentials (iTEPs) recorded from macroelectrodes in cortico-striato-thalamic and limbic networks.

3. Key Contributions

1. First Human Single-Neuron Data: This is the first study to resolve single-unit activity in deep human brain structures in response to dlPFC TMS.
2. Artifact Mitigation: Demonstrated a robust method to remove TMS artifacts, allowing neural signal resolution within 8 ms of stimulation.
3. Cell-Type Specificity: Identified distinct, opposing responses in putative interneurons vs. pyramidal cells across distributed networks.
4. Mechanistic Insight: Provided direct cellular evidence supporting the "Top-Down Control" hypothesis over the "Direct Limbic Inhibition" hypothesis.

4. Key Results

A. Temporal Dynamics and Anatomical Distribution

• Facilitation: TMS elicited rapid firing facilitation in 46% of all modulated neurons.
  • Onset: As early as 8 ms.
  • Peak: ~80–100 ms.
  • Duration: Lasted up to ~600 ms.
  • Location: Primarily in bilateral striato-thalamic regions (putamen, thalamus). 66% of striato-thalamic units were facilitated vs. only 5% of limbic units.
• Suppression: TMS elicited firing suppression in 21% of modulated neurons.
  • Onset: Delayed relative to facilitation.
  • Peak: ~300 ms.
  • Duration: Lasted up to ~1000 ms.
  • Location: Primarily in bilateral limbic regions (hippocampus, amygdala, subgenual cingulate). 16% of limbic units were suppressed vs. 6% of striato-thalamic units.

B. Cell-Type Specificity

• Interneurons (INs): Significantly more likely to be facilitated (recruited), particularly in striato-thalamic regions.
• Pyramidal Cells (PYs): Significantly more likely to be suppressed, particularly in limbic regions.
• Conclusion: TMS recruits inhibitory interneurons in executive control circuits while suppressing excitatory drive in limbic circuits.

C. Network Correlations

• Trial-by-Trial Coupling: Responsive single-unit modulations were:
  • Positively correlated with iTEP amplitude in the cortico-striato-thalamic network.
  • Anti-correlated with iTEP amplitude in the limbic network.
• Directionality: The facilitation of striato-thalamic units preceded the suppression of limbic units, suggesting a causal chain where striato-thalamic activation gates limbic activity.

5. Significance and Implications

• Mechanism of Action: The findings support a model where dlPFC TMS does not directly inhibit limbic structures. Instead, it rapidly recruits striato-thalamic inhibitory interneurons, which subsequently exert top-down inhibitory gating on limbic excitatory pyramidal cells.
• Therapeutic Relevance: This cellular mechanism explains why dlPFC stimulation sites that are anti-correlated with the subgenual cingulate (a key depression node) are clinically effective. It suggests that therapeutic TMS works by rebalancing large-scale control networks rather than simply dampening local pathology.
• Clinical Translation: The rapid engagement of bilateral networks (even with unilateral stimulation) suggests that the specific targeting of TMS (e.g., Beam F3 vs. 5.5 cm rule) may be less critical than previously thought, as the stimulation engages widespread, distributed circuits.
• Future Directions: Establishes a baseline for understanding how repetitive protocols (like iTBS) induce plasticity in these specific cell types and networks.

In summary, this study provides the first direct cellular evidence that dlPFC TMS modulates depression by engaging a rapid, bilateral, inhibitory control loop (striato-thalamic) that suppresses excitatory limbic drive, resolving a long-standing debate regarding the neural mechanisms of TMS.