Many people with depression do not get better with standard treatments like medication. One promising alternative is transcranial magnetic stimulation (TMS), a non-invasive procedure that uses magnetic pulses to stimulate specific brain regions. A particular pattern of TMS called continuous theta-burst stimulation (cTBS) is thought to reduce overactive brain activity in depression, but the investigators do not yet fully understand how it works at the level of brain cells and connections.

This study aims to determine the biological mechanism by which cTBS changes brain activity in people with depression. Specifically, the investigators are testing two competing ideas: (1) that cTBS works by weakening the connections between brain cells through a process called long-term depression (LTD), which is driven by a chemical messenger system called glutamate; or (2) that cTBS works by increasing the brain's natural "braking" system, driven by a different chemical messenger called GABA.

To test these ideas, participants with depression will receive cTBS along with one of four FDA-approved medications, or placebo, that either boost or block these chemical messenger systems. The investigators will measure changes in brain activity using electroencephalography (EEG) recorded simultaneously with TMS. Specific patterns in the EEG signal, called TMS-evoked potentials (TEPs), act as a window into how different brain cell types are responding to stimulation.

Each participant will complete four study visits, each testing a different drug-TMS combination in random order. One group of participants will test drugs targeting the glutamate system (d-cycloserine and memantine). A second group will test drugs targeting the GABA system (lorazepam and baclofen). All drugs are given as a single oral dose and are commonly used in clinical practice.

Understanding exactly how cTBS works at a biological level could open the door to more effective, personalized TMS treatments.

Major depressive disorder (MDD) affects an estimated 280 million people worldwide, and a substantial proportion do not respond adequately to first-line pharmacological treatments. Continuous theta-burst stimulation (cTBS) is a widely used form of repetitive transcranial magnetic stimulation (rTMS) that quiets targeted brain networks, and holds particular promise for patients who are not well-served by conventional excitatory TMS protocols due to safety or tolerability concerns. Despite its common clinical use, the synaptic mechanisms by which cTBS works remain poorly characterized, limiting systematic efforts to improve its efficacy.

Two mechanistic hypotheses have been proposed. The first is that cTBS induces LTD-like synaptic depression through NMDA receptor (NMDAR)-dependent mechanisms, analogous to the low-frequency stimulation protocols that produce AMPA receptor internalization and synapse weakening in animal models. Evidence for this comes from studies in the healthy motor cortex demonstrating that cTBS-induced corticomotor inhibition is blocked by NMDAR antagonists. The second hypothesis is that cTBS works through GABA receptor-mediated inhibition, either via GABA-A receptors reducing signal propagation through membrane hyperpolarization, or GABA-B receptors suppressing presynaptic neurotransmitter release. Whether GABAergic mechanisms contribute to cTBS effects has not been directly tested, and these hypotheses are not mutually exclusive.

Critically, all prior mechanistic evidence comes from the motor cortex in healthy volunteers. The dorsolateral prefrontal cortex (dlPFC), the clinical target for depression, differs substantially from motor cortex in anatomy, interindividual variability, and plasticity. Depression itself is associated with reduced synaptic plasticity, including reduced expression of NMDAR subunits and synapse-related genes in postmortem prefrontal tissue. Whether cTBS mechanism established in healthy motor cortex translates to the depressed dlPFC cannot be assumed, and has not been tested. Furthermore, whether cTBS-induced inhibition can be pharmacologically enhanced remains entirely unexplored.

This study uses a randomized, double-blind, placebo-controlled crossover design to directly test NMDAR- and GABA receptor-mediated contributions to cTBS-induced plasticity in the dlPFC of individuals with MDD. Participants are assigned to one of two parallel aims. Aim 1 tests glutamatergic mechanisms: participants complete four visits receiving cTBS paired with placebo, d-cycloserine (DCS), or memantine (MEM), with two placebo visits. Aim 2 tests GABAergic mechanisms: participants complete four visits receiving cTBS paired with placebo, lorazepam (LZP), or baclofen (BAC). All drugs are FDA-approved and administered as single oral doses timed to peak plasma concentration approximately two hours prior to cTBS.

TMS-EEG provides the primary measurement approach. TMS-evoked potentials (TEPs) are scalp-recorded electrical responses to individual TMS pulses that reflect summated excitatory and inhibitory postsynaptic activity from stimulated neuronal populations. Characteristic peaks are named by polarity and latency: P30 and P60 reflect glutamatergic excitatory transmission, N45 reflects GABA-A-mediated inhibitory tone, and N100 reflects GABA-B-mediated inhibitory tone. Up to 200 single TMS pulses are delivered per TEP session at the individualized dlPFC target. TEPs are acquired at several timepoints each visit. cTBS consists of 600 pulses delivered at 80% of resting motor threshold.

TMS is delivered using the Nexstim NBS-6 system with integrated real-time neuronavigation, with targeting based on individual structural MRI acquired prior to experimental visits.

The central hypothesis is that cTBS reduces P30 amplitude through LTD-like synaptic depression driven by NMDAR activation and consequent AMPA receptor internalization. Under this hypothesis, DCS will enhance and memantine will block the post-cTBS reduction in P30, without corresponding changes in N45 or N100. If instead GABAergic mechanisms contribute, cTBS will increase N45 and/or N100 amplitude, with additive effects from lorazepam and baclofen respectively. The design allows estimation of relative contributions from each receptor system.

A functional measure of dlPFC network integration, the N-back working memory task, is administered before and after cTBS at each visit. Resting-state fMRI acquired at baseline is used for exploratory post-hoc correlations with TEP outcomes and EEG source localization. Peripheral blood is collected for drug plasma quantification, and saliva samples are collected for BDNF Val66Met genotyping.

Statistical analysis uses a linear mixed model with Drug Type, TMS Type (active vs. sham), and Time (pre-drug, post-drug/pre-cTBS, post-cTBS) as within-subject factors, including all two- and three-way interactions, with relevant demographic and clinical covariates. The study is powered to detect a moderate effect size (Hedges' g = 0.5) with 80% power at α = 0.05, requiring 31 completers per aim. Enrollment targets 35 completers per aim, with up to 80 participants enrolled to account for attrition.

Inclusion Criteria:

• Can safely receive TMS and study drugs
• Stable medication regimen for one month prior to study participation, and for the duration of the study
• Not currently receiving TMS, ECT, or ketamine
• No active safety concerns related to suicidality
• Moderate to severe Major Depressive Disorder as indicated by the Patient Health Questionnaire or Quick Inventory of Depressive Symptomatology

Exclusion Criteria:

• History of seizures or epilepsy
• History of intracranial pathology or lesions from any etiology
• History of traumatic brain injury including prolonged loss of consciousness more than 15 min
• Signs of increased intracranial pressure
• Any major neurological conditions (ex: recent stroke, tumor, neurodegenerative disorders, etc.)
• Major medical conditions that may cause a medical emergency in case of a provoked seizure (cardiac malformation, cardiac dysrhythmia, asthma, etc.)
• Severe migraines that may result in treatment intolerance.
• Inability to tolerate MRI.
• Pregnancy
• Known allergic reaction to d-cycloserine, baclofen, memantine, or lorazepam

De-identified individual participant data will be shared through the National Institute of Mental Health Data Archive (NDA) in accordance with NIMH data sharing expectations. Data to be shared will include demographic information, clinical scale scores, and TMS-EEG outcome data. Data will be submitted to the NDA following standard de-identification procedures and will be made available to qualified researchers through the NDA data access request process.

Data will be submitted to the NDA within 1 year of primary study completion or upon publication of primary results, whichever comes first.

Data will be accessible to qualified researchers through the NIMH Data Archive data access request process.

Participants will be aware of which Aim they are participating in, where they will receive all of the different arms in random order. They will not know which drug they are receiving or if the TMS is active or sham.