Response of lab-grown nerve cells from people with treatment-resistant depression to hydroxynorketamine and reelin

Distinct limitations exist to using animal models for depression, particularly because these are often homogeneous, which is in direct contrast to the heterogeneity of the human experience of depression. Most current rodent models focus on testing response to standard antidepressants, but there is a noted lack of well-validated experimental and translational models specifically for treatment-resistant depression (TRD) [1, 2]. Although the definition varies, TRD is broadly defined by inadequate response to two or more classes of antidepressant regimens despite adequate dose, duration, and adherence to treatment. Individuals with TRD who do not respond to multiple classes of antidepressants, or show only a partial response, often face additional challenges, including functional impairment, lower quality of life, high relapse rates, and increased risk of suicidality—all core factors that are not well replicated by animal models [1 –3]. In addition, the field's inability to study brain tissue from living humans limits the molecular changes that can be observed in depression or in response to antidepressant treatments. In this context, a clear need exists to create translational models to help transition novel therapeutics from bench to bedside for the most treatment refractory patients.

The development of human inducible pluripotent stem cells (iPSCs) by reprogramming human fibroblasts, keratinocytes, and lymphocytes has allowed investigation into the cellular mechanisms of various disorders and therapeutics in humans [4, 5]. The complex polygenic landscape of depression makes iPSC models particularly valuable, as they capture the full genetic architecture of individual patients' neurons—a capability previously unavailable to researchers beyond post-mortem samples. iPSC modeling has been successfully implemented to uncover key molecular insights across psychiatric disorders, including bipolar disorder, schizophrenia, autism spectrum disorders, and major depressive disorder (MDD) [6, 7]. To our knowledge, however, only two previous reports have examined iPSC modeling in individuals with TRD; both focused on the serotonergic system [8, 9].

The N-methyl-D-aspartate receptor (NMDAR) antagonist, glutamate modulator, and rapid-acting antidepressant ketamine is an effective treatment for TRD [10]. Although its precise mechanisms of action are varied and remain elusive, its primary therapeutic effects are thought to be mediated via activation of mechanistic target of rapamycin complex 1 (mTORC1) and related downstream signaling in the prefrontal cortex (PFC) and hippocampus [10, 11]. In addition, upregulation of proteins such as phosphorylated-mTOR (p-mTOR) and postsynaptic density-95 protein (PSD-95) as well as surface insertion of the glutamate A1 (GluA1) subunit of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) often accompanies ketamine's antidepressant-like effects [12]. Despite its antidepressant efficacy, ketamine's dissociative side effects and abuse potential limit its widespread use, underscoring the need to develop additional effective therapeutics for TRD.

In the context of developing novel treatments for TRD, one of ketamine's major metabolites, (2 R,6 R)-hydroxynorketamine ((2 R,6 R)-HNK), has also shown promising rapid-acting antidepressant effects. Preclinical studies suggest that (2 R,6 R)-HNK retains antidepressant-like behavioral efficacy and exhibits strong synaptic potentiation effects at excitatory synapses while lacking the dissociative and abuse potential associated with ketamine and its enantiomers [13, 14]. A recent Phase I clinical trial found that (2 R,6 R)-HNK had a favourable safety and tolerability profile in healthy volunteers, with no evidence of anesthetic or dissociative effects. In addition, its pharmacokinetic profile revealed dose-proportional increases in exposure as well as increases in gamma power at low and medium doses [15]. Our laboratory recently initiated the first Phase II study to evaluate its efficacy in participants with TRD (NCT06511908).

The search for novel rapid-acting antidepressants has also led researchers to examine endogenous proteins that may share mechanistic overlap with the putative therapeutic effects of ketamine and (2 R,6 R)-HNK, and reelin has emerged as a promising candidate. Because reelin signaling may mediate some of the antidepressant effects of ketamine and (2 R,6 R)-HNK [16], exogenous reelin administration could help identify downstream therapeutic targets that could induce an antidepressant response. Reelin is a large extracellular matrix glycoprotein that is decreased in the hippocampus of individuals with mood disorders [17]. Previous research with in vitro and in vivo chronic stress animal models found that ketamine and reelin demonstrate similar effects (e.g., upregulation in long-term potentiation) [18 –20], although reelin has not yet been investigated in individuals with TRD. Both reelin and ketamine have also been proposed as mediators of subchronic inflammation [21, 22], which is hypothesized to contribute to depressive symptomatology [23]. Reelin may also have dichotomous effects on inflammation; specifically, high levels of reelin expression within the blood may be pro-inflammatory and associated with disorders such as Alzheimer's disease and multiple sclerosis [24], but reelin overexpression within the brain may have potential therapeutic effects in animal models for these disorders [22]. Studies also suggest that reelin increases synaptogenesis, synaptic potentiation, mTORC1 activation, and dendritogenesis, all of which have been implicated in ketamine's antidepressant effects [25] and could thus prove effective at targeting TRD symptoms.

As noted above, molecular research into depression—including TRD—has been limited by our inability to directly investigate living human neurons. To address this gap, the current exploratory study sought to examine whether iPSC-derived neurons from participants with TRD, which model individual genetic variability, could be used to assess novel rapid-acting therapeutics in a translational manner. The impact of (2 R,6 R)-HNK and reelin on iPSC-derived neurons from participants with TRD was examined to determine molecular parallels between this novel in vitro method and previous in vivo research. The following proteins were measured: mTOR, extracellular-related signaling kinase (ERK), PSD-95, GluA1, Synapsin I, disabled adaptor protein (Dab1), tyrosine kinase B (TrkB), and NMDA-receptor subunit-2B (NR2B); all of these proteins are robustly implicated in the effects of rapid-acting antidepressants across multiple therapeutics [12, 26]. In addition, gene set enrichment analyses on bulk RNA-sequencing were conducted to determine changes in hallmark signaling pathways. The main objective of this series of experiments was to identify potential shared rapid antidepressant mechanisms between reelin and (2 R,6 R)-HNK. The study hypothesis was that broad changes would be observed after treatment with reelin and (2 R,6 R)-HNK in proteins and signaling pathways previously implicated in the pathophysiology of TRD and in antidepressant response to ketamine, including glutamatergic synaptic signaling and inflammatory responses.

To determine the effects of putative rapid-acting antidepressants, reelin and (2 R,6 R)-HNK were applied to mature iPSC-derived cortical neurons from participants with TRD (n = 5). After one and 24 h of treatment, cells were collected to assess time-dependent changes in synaptic protein and gene expression.

This exploratory study investigated the effects of 1µᴍ (2 R,6 R)-HNK and varying concentrations of reelin on iPSC-derived neurons obtained from individuals with TRD. The parallels found in our iPSC-derived neurons, animal models, and clinical research demonstrate that iPSC-derived neurons can not only provide insights into polygenic psychiatric illnesses and the mechanisms of established treatments but also serve as an in vitro platform for screening, translating, and eventually personalizing promising novel therapeutics. Using iPSC-derived neurons from different patient populations allows for the unique opportunity to research changes in human neuronal tissue after exposure to established and novel therapeutics.

Notably, reelin had time- and concentration-dependent impacts that paralleled the effects of (2 R,6 R)-HNK on cell lines derived from participants with TRD. Proteins integral to synaptic plasticity such as PSD-95, Synapsin I, Dab1, and p-ERK were all significantly increased at one hour in response to both (2 R,6 R)-HNK and higher concentrations of reelin, underscoring that the two treatments had similar effects. These effects were reversed at the 24-h mark, suggesting either compensatory downregulation after induction of structural plasticity or that prolonged exposure to reelin or (2 R,6 R)-HNK might have deleterious effects on cell cultures. In addition, the increase in Synapsin I suggests that both reelin and (2 R,6 R)-HNK have effects at the pre-synaptic level and may increase neurotransmitter release, which supports previous research on ketamine's pre-synaptic mechanisms [35]. p-ERK expression was upregulated by both agents at the one-hour timepoint, echoing previous work demonstrating its rescue by ketamine in an animal model of depression [36]. Previous research also found that ERK was downregulated in the PFC of individuals with MDD [37] and noted a correlation between ketamine's antidepressant effects and ERK signaling [38]. Recent research has implicated cAMP signaling in the antidepressant actions of ketamine and (2 R,6 R)-HNK [39, 40]. cAMP signaling is a known mediator of ERK signaling [41] and may be a key driver of these antidepressant-like responses. Future research should determine whether cAMP signaling is necessary for the observed effects of reelin and (2 R,6 R)-HNK within this model. Interestingly, immunocytochemical analyses (see Supplement) revealed time- and dose-dependent increases in proxy TrkB and NR2B expression, paralleling Western blot results; the highest levels of expression were found at one hour after treatment with 50nᴍ of reelin and (2 R,6 R)-HNK. However, further research is needed to establish the most appropriate methods for ICC analysis. These findings, which used punctal analyses as a proxy for protein expression, should be interpreted with caution.

Given the similarities between the molecular actions of high-dose reelin and (2 R,6 R)-HNK, this study also assessed common differentially expressed genes in all treatment groups; results were particularly interesting for miR-27b and GPR52. miR-27b was previously found to be impacted by chronic stress exposure in a rodent model [42] and has been associated with reward-related brain activation [43]. Clinically, plasma concentrations of miR-27b were found to correlate with the efficacy of alprazolam in patients with comorbid anxiety and alcohol use disorders [44]. The decrease in GPR52 expression observed in this study after treatment with all doses of reelin and (2 R,6 R)-HNK further supports the role of therapeutically targeting G protein-coupled receptors in MDD and TRD [45]. GPR52 encodes an orphan Gs-coupled receptor that is highly expressed in the brain, particularly in striatal regions where it colocalizes with D2 dopamine receptors [46]. This finding aligns with emerging evidence that GPR52-dependent regulation of the nucleus accumbens-ventral pallidum-ventral tegmental area circuit may be crucial for maintaining proper reward processing and motivation [47]. In addition, GPR52 knockout mice displayed reduced anxiety-like symptoms [48], and a GPR52 inverse agonist was recently found to ameliorate chronic stress-induced deficits in dopaminergic activity and reward motivation [47]. An orally available GPR52 agonist (HTL0048149) has recently advanced into Phase I human clinical studies with future potential in schizophrenia [49], providing direct translational potential for this marker. Finally, ARHGAP8, a Rho GTPase involved in synaptic structure and AMPAR-mediated transmission, was found to be decreased with the highest concentration of reelin. ARHGAP8 increases have previously been linked to neurodevelopmental and neuropsychiatric disorders [50 –52], and decreasing its expression may be another method by which reelin exerts antidepressant-like effects.

It is also interesting to note that the global metabolic regulation through ERK observed here aligns with our RNA sequencing findings of altered mTORC1 signaling, as both pathways are known to coordinate cellular energy status and growth signaling. Here, mTORC1-related signaling was significantly upregulated by both (2 R,6 R)-HNK and reelin (50 nM) one hour after treatment exposure, consistent with previous reports [14]. While mTOR signaling dynamics can be challenging to capture via Western Blot, the transcriptional signature of this pathway across our RNA sequencing data underscores that these multiple complementary methodologies detected clinically relevant signaling. Interestingly, upregulation of the androgen response was also observed after treatment with 50 nM of reelin.

This study also found that the highest concentration of reelin (50 nM) significantly increased TNF-α signaling via NF-κβ. Though generally considered pro-inflammatory, increases in TNF-α signaling are important for neuronal plasticity, cognition, and CNS development [53]. TNF-α plays a vital role in synaptic scaling and plasticity by modulating both excitatory and inhibitory transmission via regulation of AMPARs and gamma aminobutyric acid (GABA) receptors [54, 55]. In MDD, chronic upregulation of TNF-α has been associated with worsened symptomatology and poorer clinical outcomes [56]. However, animal models found transient increases in TNF-α signaling after acute administration of ketamine that were associated with antidepressant-like effects [57], which parallels changes observed in our RNA sequencing results. The exact role of reelin in mediating inflammatory response, particularly within the brain, is still under discussion. High levels of circulating reelin have been shown to promote leukocyte infiltration and thrombosis as well as positively correlate with increased concentrations of cytokines [24, 58]. While high levels of reelin have been associated with inflammatory conditions, homeostatic balance may be necessary for proper neuronal functioning of reelin. Mice with reduced reelin signaling are more susceptible to chronic stress-induced behavioral and biological deficits, and exogenous reelin supplementation was found to rescue these deficits [59, 60]. As with ketamine, a transient increase in inflammation could be beneficial to rapid-acting antidepressant response. Future research should ascertain whether similar responses are found within cell lines derived from healthy volunteers.

Several reviews have discussed both the limitations and merits of iPSC modeling for neuropsychiatric diseases, emphasizing the need for thorough model characterization [4 –6]. Modeling depression using iPSCs presents unique challenges due to the moderate genetic heritability of the disorder (~30 - 50%) as well as its genetic heterogeneity [61]. Because the bulk RNA-sequencing results within our data only capture cellular changes at one hour, not 24 h, future research is needed to determine the longer-term transcriptomic effects of (2 R,6 R)-HNK and reelin. Additionally, our sample size of exclusively female participants limits conclusions that can be drawn to TRD participants more broadly. Previous studies using iPSC-derived neurons nevertheless support the feasibility of this approach [6], revealing important distinctions in response patterns to conventional therapies. For instance, previous iPSC studies found that non-responders to selective serotonin reuptake inhibitors (SSRIs) showed upregulated excitatory serotonergic receptor activity, serotonin-induced hyperactivity, and altered neurite growth and morphology, patterns notably different from those observed in healthy volunteers and treatment responders [8, 9]. These findings suggest that iPSC models can capture clinically relevant differences in treatment response. Healthy control cell lines, as well as cell lines from responders and non-responders to ketamine, should be included in future research studies to determine molecular signalers involved in clinical response to rapid-acting antidepressants, as our small sample size was underpowered to conduct these analyses. Given that individuals with TRD show particularly low response rates to SSRIs, it is possible that molecular alterations induced by ketamine and other rapid-acting agents may be even more readily detectable in iPSC models, in concert with superior clinical outcomes. This enhanced signal-to-noise ratio could make iPSC-derived neurons even more valuable for screening novel rapid-acting antidepressants and understanding their mechanisms of action.

Despite these intriguing preliminary findings, the field of iPSC-derived neuronal modeling for depression is still in its infancy [62]. Our results underscore the potential of this method for therapeutic screening disease modeling and translating relevant discoveries into first-in-human drug development, but further optimization is needed. Future research priorities include: 1) replicating key markers identified here to establish optimal molecular readouts; 2) investigating both earlier and long-term timepoints to assess the durability of treatment effects; and 3) comparing patient-derived cell lines with those from healthy volunteers to identify differential responses, similar to previous studies that found bipolar disorder-specific responses to lithium compared to healthy volunteer cells [63]. Despite their preliminary nature, these findings nevertheless advance the characterization of iPSC-derived neurons from TRD participants and demonstrate their value for translational screening of novel rapid-acting therapeutics.

Supplemental Material