Brain Glutamate Dynamics Predict Positive Agency in Healthy Women

Contributions of brain glutamate to conscious emotion are not well understood. Here we evaluate the relationship of experimentally-induced change in neocortical glutamate (ΔGlu) and subjective states in well individuals. Drug challenge with d-amphetamine (AMP; 20 mg oral), methamphetamine (MA; Desoxyn®, 20 mg oral), and placebo (PBO) was conducted on three separate test days in a within-subjects double blind design. Proton magnetic resonance spectroscopy (MRS) quantified neurometabolites in the right dorsal anterior cingulate cortex (dACC) 140–150 m post-drug and PBO. Subjective states were assessed at half hour intervals over 5.5-hours on each session, yielding 3,792 responses per participant (91,008 responses overall, N=24 participants). Self-reports were reduced by principal components analysis to a single factor score of AMP- and MA-induced Positive Agency (ΔPA) in each participant. We found drug-induced ΔGlu related positively with ΔPA (ΔGluMA r=+.44, p<.05, N=21), with large effects in females (ΔGluMA r=+.52, p<.05; ΔGluAMP r=+.61, p<.05, N=11). States related to ΔGlu in females included rise in subjective stimulation, vigor, friendliness, elation, positive mood, positive affect (r’s=+.51 to +.74, p<.05), and alleviation of anxiety (r=−.61, p<.05, N=11). Self-reports correlated with DGlu to the extent they loaded on ΔPA (r=.95 AMP, p=5×10−10; r=.63 MA, p=.0015, N=11), indicating coherence of ΔGlu effects. Timing data indicated Glu shaped emotion both concurrently and prospectively, with no relationship to pre-MRS emotion (ΔGluAMP r=+.59 to +.65, p’s<.05; ΔGluMA r=+.53, p<.05, N=11). Together these findings indicate substantive, mechanistic contributions of neocortical Glu to positive agentic states in healthy individuals, most readily observed in women.


INTRODUCTION
Glutamate (Glu) is an ancient compound that likely shapes conscious experience, well-being, and agency in everyday life. It is well established that Glu mediates excitatory neurotransmission, learning, memory, motor activity, is under homeostatic control and is excitotoxic when dysregulated (Magi et al, 2019). This scope would seem to argue against additional major, undiscovered roles for Glu in the brain. The evolutionary history of Glu, however, reveals rich diversity. Phylogenetically, Glu occurs at high concentrations in species that are extraordinarily distant -such as bacteria and humans -evidencing biologic roles predating divergence of prokaryotes and eukaryotes 2.7 billion years ago (Commichau et al, 2008;Cooper, 2000). Glu-like receptors occur in plants and animals, indicating common ancestry ~ 1.6 billion years ago (Chiu et al, 1999;Meyerowitz, 1999;Wang et al, 1999). Glu is thus ancient and ubiquitous, providing ample time and material for the evolution of multiple, overlapping functions in living systems. This history suggests additional undiscovered roles for Glu in healthy organisms, where rapid homeostasis readily obscures functional processes in vivo. Indeed, science over the last nine decades has repeatedly underestimated the functional diversity of Glu (Danbolt, 2021;Watkins and Jane, 2006). Glu thus very likely has further, unanticipated roles in health and well-being.
Emerging work in clinical neuroscience indicates neocortical Glu shapes the etiology and treatment of disorders of mood, motivation, and behavior. Glx, a combination of Glu and glutamate (Gln), is reduced in frontal brain in major depressive disorder (MDD) and is elevated in bipolar disorder (BPD) (  . Low-doses of ketamine, a noncompetitive NMDA receptor antagonist, increase Gln in the ACC and reports of time distortion, dissociation, emotional blunting, cognitive disruption, excitement, and somatic activation (Coull et al, 2011;Krystal et al, 2005;Rowland et al, 2005). Conversely, the anesthetic propofol reduces Glu in motor cortex, sensory cortex, and thalamus alongside its effects on sedation (Zhang et al, 2009). These data suggest mechanistic contribution of neocortical Glu to visceral sensations, subjective states, and feelings of connection, attachment, and engagement. Motivated by the above data and history, we here investigate contributions of neocortical glutamate to conscious experience using drug challenge in healthy volunteers. This method provides experimental manipulation of neocortical Glu in individuals who are medically and psychiatrically well (White and Gonsalves, 2020;White et al, 2018). Toward this end, d-amphetamine (AMP), methamphetamine (MA) and placebo (PBO) were administered on three separate test days to healthy participants using a withinsubjects, counterbalanced, double-blinded design, with each participant serving as their own control.
Using this approach, we nd AMP and MA produce phasic rise in dACC Glu, an effect most apparent in females   (Fig. 1). This approach provides detailed information on conscious states, mood, emotion, metacognition, and visceral/somatic sensations in each participant, suitable for evaluation with phasic change in dACC Glu.
Our hypotheses were two-fold. First, we expected drug-induced change in neocortical Glu to predict the magnitude of drug-induced change in conscious states with motivational component, i.e., perceived and reported states of subjective stimulation, excitement, enthusiasm, and vigor. Second, we expected these effects to be more readily observed in female participants, due to their heightened glutamatergic response and increased vulnerability to stimulant dependence compared to males . The study thus provides new information on experimentally-induced change in neocortical Glu

MRS Quality Control and Analysis
Quality control entailed four steps. (a) MRS spectra from each session in each participant was t using LCModel (Provencher, 1993), and visually inspected for quality ( Fig. 1) Figure S1). This factor (Factor I) summarizes participants' subjective response to each drug, interpreted based on loadings greater than |.35|.

Statistical Tests of Hypotheses
Glu effects on Positive Agency. Relationship of change in Glu and subjective states were evaluated using a correlation approach, with delta (Δ) scores of drug-induced changes in Glu entered as the predictor and Factor I score (z-score) of drug-induced change in Positive Agency entered as the dependent measure.
Contributions of biological sex. Sex differences in the direction and magnitude of relationships were assessed by Fisher r-to-z transformation.
Follow-up Analyses were conducted to determine speci city, phenomenology, coherence, and timing of signi cant effects, as described below. △ △ △ △ Speci city. Relationships with change in Glx, Gln and Factor I response was evaluated jointly and separately by sex, to provide information on speci city of relationships with neurometabolites.
Phenomenology. Relationship of ΔGlu and ΔAUC values (18 measures loading >|.35| on AMP and MA Factor I, Table 1) were evaluated using a correlation approach, providing information on speci c subjective states related to Glu. This analysis was restricted to females, who accounted for signi cant effects at the group level.

Manipulation and Validity Checks
Four sets of manipulation and validity checks were conducted to verify the e cacy of the study drugs, validity of participants' self-reports, and validity of the follow-up timing bins calculations and analysis.
(1) Drug effects on scored self-reports were assessed by within-subjects, repeated-measures ANOVAs with two levels of drug (drug, PBO) and eight levels of time (TP1-8, Fig. 1). The analysis provides data on magnitude, direction, and timing of drug effects on self-report measures, with attention to loadings > |.35| (Table 1).

Power and Effect Size Estimation
Power analyses were conducted in G*Power 3.1.9 using an alpha of .05 (Cohen, 1988;Faul et al, 2009). Effect sizes (Cohen's d) were calculated using the formula ((mean-0)/SD), with Cohen's d values of .2 interpreted as small effects, .5 as medium effects, and .8 as large effects. Pearson correlations of .1 were interpreted as small effects, .3 as medium effects, and .5 as large effects (Cohen, 1988(Cohen, , 1992.

PCA Results
Factor I: Positive Agency. Principal components analysis (PCA) of AUC AMP and AUC MA produced a primary factor of response to AMP and MA (Factor I; eigenvalue > 6.0, Figure S1), and interpreted based on loadings > |.35| (  Fig. 3, Table S1). Findings in males were not signi cant (Table 2, Figure S2). AMP. ΔGlu AMP related positively to ΔPA AMP in females, a large effect ( Table 2, Fig. 3).
ΔGlu AMP and ΔGlx AMP ndings in males and combined sample were not signi cant (Table 2, Table S1, Figure S2). These data indicate large effects of ΔGlu on ΔPA in females, accounting for effects at the sample level.  Bin 1= timepoints prior to MRS (timepoints 1-3, Figure 1).
Coherence. Measures' strength of relationship to Factor I (i.e., loadings on Factor I) predicted their strength of relationship to Glu, a large effect ( Glu AMP r = 0.95, p = 5x10 − 10 (1-tailed); Glu MA r = 0.63, p = .0015 (1-tailed)). Self-report measures thus related to ΔGlu to the extent they loaded on PA (Fig. 3F-G), evidence of coherence of Glu effects across self-report instruments.

Manipulation & Validity Checks
Drug e cacy. Drug effects on self-reports and summary scores (i.e., scored measures, AUC values, Factor I scores) were highly signi cant. These data indicate e cacy of the study drugs and validity of summary score calculations, quality control and data reduction procedures (Supplemental Results,   Tables S3-S4).
Subjective responses & timing bins. AUC values differed by bin, with rise in value over time (Supplemental Results, Figure S3). AUC and PCA Factor I responses did not differ by sex (Supplemental Results, Table S5). These data indicate feasibility of time-dependent prediction of emotion by Glu, and an overall lack of sex differences in the subjective response to AMP and MA. Our main nding was a robust positive relationship of experimentally-induced DGlu and positive agentic emotion in females (Table 2, Figure 3). This effect was large in size and occurred for AMP and MA, indicating reproducibility across study drugs and test days ( Figure 3). Induced emotion was independent of Gln (Table S1) Table 2, Table S2), evidence of replicability. In addition DGlu predicted self-reports to the extent these measures loaded on the factor of agency (Factor I; r=.95, p=5x10 -10 for AMP; r=.63, p=.0015 for MA; Figure 3). Thus DGlu related to self-reports to the extent they involved an incentive motivational component (i.e., a positive agentic response). These data demonstrate coherence of Glu effects on subjective states, with ndings generalizable across study conditions (AMP, MA), measures (vigor, friendliness, elation, positive mood), and data reduction approaches (Factor 1, AUC; Figure 2).
Effect timing was informative, with drug-induced change in Glu shaping both concurrent and later reports of positive emotion (DGlu r=+.59 to +.65, p's<.05 with AMP; DGlu r=+.53, p<.05 with MA). Change in neocortical Glu preceded or co-occurred with self-reports (timing in Figure 1), indicating contribution of Glu to current and subsequent emotion. Positive emotions stayed higher throughout the period of testing, lasting ve hours post-drug and 2.5 hours post-Glu assessment ( Figure 1, Table 2, Figure S3). This duration of effects has therapeutic implications, as Glu may provide a marker to target and personalize interventions in MDD, substance use disorder, and to improve overall well-being during periods of health.
The ndings are also consistent with prior work indicating drug-induced Glu predicts  and to greater extent than males (Anker and Carroll, 2011). National epidemiologic data further indicate females' earlier chronological age of rst use of cocaine and amphetamine, and females' more rapid progression from initial use to drug dependence compared to males (Becker and Hu, 2008). Responses to psychostimulants are thus modulated by biological sex in ways that facilitate females' rapid acquisition and persistence of drug dependence. Our ndings indicate a role of neocortical glutamate in subjective experience after drug ingestion, with pronounced effects in females. These subjective effects may shape both the etiology and trajectory of stimulant dependence and MDD. Heightened glutamatemediated learning of contextual cues, drug-cue associations, and glutamate-mediated reward processing in females would contribute to more rapid acquisition and severity of drug dependence in females compared to males. In the context of MDD, our ndings advance Glu as a novel treatment target for medication and adjunctive treatment for positive emotion recovery (Cole et al, 2022).
The present study has both strengths and weaknesses. Strengths include use of a within-subjects, placebo-controlled crossover drug challenge design; assessment of subjective states through multiple self-report instruments at eight time points on three test sessions per participant; and rigorous procedures for data quality and data reduction. Use of within-subjects, repeated-measures assessment of states provide deep phenotyping of subjective states, emotion, visceral and somatic sensations, and metacognition suitable for analysis with experimentally-induced change in Glu.
Limitations included the modest sample size, low statistical power to detect sexual dimorphism in Glu effects on emotion, and relatively high CRLB uncertainty for Gln due to the PRESS acquisition. While relative CRLBs are common practice in reporting MRS data quality and 20% is a common threshold, this threshold is likely overly conservative for Glu and Gln (Kreis, 2016). As Gln is di cult to distinguish from Glu at the present TE at 3T, future studies should implement acquisition parameters that more effectively differentiate Gln from Glu. The present measures of Glx, Glu and Gln include both metabolic pools and neurotransmitter levels of Glu and Gln, as MRS supplies data on the total tissue metabolite within the voxel. Future work can utilize larger samples, assess ovarian, testicular, and adrenal hormones; epigenetics; sex-dependent gene expression; and social constructions of gender.
In summary, we here identify a robust positive relationship of acute rise in dACC glutamate and positive agentic subjective states in healthy females. Timing was concurrent and prospective, with no relationship to pre-MRS emotion. To our knowledge, this is the rst demonstration that acute change in glutamatergic compounds in human cortex alters a broad range of positive agentic states in well individuals. The study thus indicates a substantive, mechanistic contribution of neocortical Glu to positive agentic emotion that is readily observed in females.

Declarations
Funding and Disclosure This work was supported by the National Institute of Health Grant DA029189 (TLW); Hanlon Foundation (TLW); Zimmerman Fund for Scienti c Innovation Awards in Brain Science, Robert J. and Nancy D.
Carney Institute for Brain Science (TLW); and COBRE Center for Central Nervous System Function NIH P20 1P20GM130414-03 (EGW). Part of this research was conducted using computational resources and services at the Center for Computation and Visualization, Brown University, NIH grant S10 OD016366.
The views expressed in this article are those of the authors and do not necessarily re ect the position or policy of the National Institute of Health. TLW has served as scienti c advisor and consultant to Strategic Aid Partners, a 501c3 organization (San Francisco, CA). The authors report no con ict of interest.
Author contributions TLW conceived the study idea and initiated, designed, and directed the study. TLW and MAG wrote the original and updated drafts of the manuscript. TLW and EGW collected the MRI data. TLW, ADH, and MAG conducted the data quality control procedures and performed the statistical and data analyses. TLW, MAG, and HEJ created the tables and gures. TLW, MAG, EGW, and ADH provided input on data analysis and interpretation of results. TLW, MAG, ADH, EGW, and HEJ revised the manuscript. All authors read and approved the nal manuscript.