Glycine attenuates impairments of stimulus-evoked gamma oscillations in the ketamine model of schizophrenia

Although a substantial number of studies suggests some clinical benefit concerning negative symptoms in schizophrenia through the modulation of NMDA-receptor function, none of these approaches achieved clinical approval. Given the large body of evidence concerning glutamatergic dysfunction in a subgroup of patients, biomarkers to identify those with a relevant clinical benefit through glutamatergic modulation are urgently needed. A similar reduction of the early auditory evoked gamma-band response (aeGBR) as found in schizophrenia patients can be observed in healthy subjects following the application of an NMDA-receptor antagonist in the ketamine-model, which addresses the excitation / inhibition (E/I) imbalance of the disease. Moreover, this oscillatory change can be related to the emergence of negative symptoms. Accordingly, this study investigated whether glycine-related increases of the aeGBR, through NMDA-receptor co-agonism, accompany an improvement concerning negative symptoms in the ketamine-model. The impact of subanesthetic ketamine doses and the pretreatment with glycine was examined in twenty-four healthy male participants while performing a cognitively demanding aeGBR paradigm with 64-channel electroencephalography. Negative Symptoms were assessed through the PANSS. S-Ketamine alone caused a reduction of the aeGBR amplitude associated with more pronounced negative symptoms compared to placebo. Pretreatment with glycine attenuated both, the ketamine-induced alterations of the aeGBR amplitude and the increased PANSS negative scores in glycine-responders, classified based on relative aeGBR increase. Thus, we propose that the aeGBR represents a possible biomarker for negative symptoms in schizophrenia related to insufficient glutamatergic neurotransmission. This would allow to identify patients with negative symptoms, who might benefit from glutamatergic treatment.


Introduction
Schizophrenia places a substantial burden on people with this illness, more so as there is no satisfactory pharmacotherapy for some of its core symptoms, namely cognitive and negative symptoms ( Correll and Schooler, 2020 ). A prominent challenge in the pursuit of an effective pharmacotherapy is that patients respond heterogeneously to treatments ( Kumar et al., 2020 ;McCutcheon et al., 2020 ). Therefore, biomarkers to identify the subgroup of patients that would benefit from certain treatments are urgently needed.
A promising approach to new treatments is the modulation of aberrant neural glutamatergic activity in patients with schizophrenia. The glutamate hypothesis of schizophrenia presumes a hypofunction of the N-methyl-d-aspartate receptor (NMDAR) which is essential for glutamate neurotransmission. On this account, there were several at-(SST + ) -aminobutyric acid (GABA) interneurons ( Alherz et al., 2017 ). Based on the foundational hypotheses on the generation of fast cortical oscillations (cf. Buzsaki and Wang, 2012 ;Traub et al., 1996 ;Whittington and Traub, 2003 ), the micro-circuital interplay of the respective interneurons with glutamatergic pyramidal cells produce said frequency oscillations ( Lisman et al., 2008 ). Further, impaired generation of gamma oscillations due to NMDAR knockout on PV + interneurons can be linked to cognitive deficits in mouse models ( Carlen et al., 2012 ).
The auditory evoked gamma-band response (aeGBR) is among the sensory evoked GBOs which have been of special interest, given that they do not only reflect sensory processes but appear to be affected by attention and memory ( Cho et al., 2006 ;Herrmann et al., 2010 ). The aeGBR appears 25 to 100 ms after an auditory stimulus and exemplifies the top-down operations in sensory processing, as its magnitude is profoundly altered by task difficulty ( Mulert et al., 2007 ). Several studies have proven that aeGBRs are impaired in all stages of schizophrenia , high risk subjects ( Leicht et al., 2016 ), and even first-degree relatives of people with the illness ( Leicht et al., 2011 ). It is noteworthy, that these detriments are accompanied by a reduced activity of a network including the anterior cingulate cortex (ACC) , which has repeatedly been implicated in the pathophysiology of schizophrenia and its cognitive dysfunctions ( Reid et al., 2019 ;Takayanagi et al., 2017 ).
The ketamine model of schizophrenia offers the opportunity to study this illness without having to test patients. It utilizes the administration of subanesthetic doses of the NMDAR antagonist ketamine (or its eutomer S-Ketamine) to reduce NMDAR dependent glutamatergic neurotransmission, which causes the emergence of schizophrenialike positive, negative, and cognitive symptoms in healthy volunteers ( Krystal et al., 1994 ) or aggravating symptom severity in patients ( Lahti et al., 1995 ). This model particularly depicts aberrant GBOs since under physiological conditions ketamine demonstrates the highest affinity for the NMDARs expressing GluN2C and GluN2D subunits, which are most frequently expressed on the aforementioned PV + and SST + GABAergic interneurons ( Khlestova et al., 2016 ;Kotermanski and Johnson, 2009 ). The inhibition of the respective interneurons is supposed to result in a disruption of the local microcircuits ( Lisman et al., 2008 ) and leads to an impaired generation of GBO resembling deficiencies observed in patients with schizophrenia ( Jadi et al., 2016 ).
Glycine binds to an allosteric binding site of the NMDAR, thus enabling signal transduction following the engagement of glutamate as well as promoting and enhancing the binding of glutamate to the NM-DAR ( Leeson and Iversen, 1994 ). These characteristics allow glycine to attenuate neurophysiological impairments in patients suffering from schizophrenia ( Greenwood et al., 2018 ;Kantrowitz et al., 2018 ) and neurophysiological impairments related to (induced) NMDAR hypofunction as animal studies have proven .
Based on the strong interplay of schizophrenia symptoms, NMDAR dysfunction and receptor co-agonists, this study aimed to delineate the impact of an exogenously induced NMDAR dysfunction on the generation of the aeGBR and the effect of glycine thereon. This was achieved by means of a cognitively demanding auditory choice reaction task after a pretreatment with glycine and during the continuous infusion of S-ketamine. We hypothesized that glycine-pretreatment mitigates the disturbances of the aeGBR and the interrelated emergence of schizophrenia-like symptoms during S-ketamine administration in healthy volunteers.

Participants
Twenty-six healthy male participants were enrolled in this study. The general procedure was approved by the Ethics Committee of the Medical Association Hamburg and carried out in accordance with the latest version of the Declaration of Helsinki. Written informed consent was obtained from all participants after the nature of the procedures had been fully explained. One participant dropped out due to adverse effects caused by S-ketamine (dissociative effect/headache). Another participant had to be excluded due to poor EEG data quality. Accordingly, we included 24 healthy male participants in EEG data analysis. Volunteers were either students or medical staff of the University Medical Center Hamburg and received a monetary compensation for the EEG recording sessions.
The inclusion criteria included male sex, an age between 18 -40 years (mean 24 years, 19 -32 years, SD 3.7 years), right-handedness as assessed by means of the Edinburgh Handedness Scale (mean 77.2%, 48 -100%, SD 16.8%), normal or corrected visual acuity and German at native speaker level. All participants had normal IQ as tested with the German WST vocabulary test (mean 111.5 points, 99 -122 points, SD 6.3 points). and they had experienced an average of 16.5 educational years (13 -21 years, SD 2.4 years). The sample was limited to participants of male sex based on safety concerns regarding the application of S-ketamine in case of an undetected pregnancy.
Exclusion criteria were any acute or previous psychiatric disorders, a family history of schizophrenia or bipolar disorder, the use of illegal drugs, active medication or health conditions representing a contraindication to the administration of S-ketamine. All subjects were screened with pure tone audiometry for seven frequencies ranging from 120 to 8000 Hz (Esser Home Audiometer 2.0) in order to ensure normal hearing in both ears. Participants with an auditory threshold higher than 20 dB in any of the frequencies were excluded from the study.
The required sample size was calculated using G * Power 3.1 ( Faul et al., 2007 ). With respect to the reduction of the aeGBR amplitude through the application of S-ketamine and the potential attenuation of this effect via pretreatment with glycine, we expected a medium effect size in the range of p 2 = 0.1. This estimation was based on a previous study comparing the effect of the administration of S-ketamine on the aeGBR to placebo ( d = 0.6) ( Curic et al., 2019 ). We planned to apply regular statistical analyses and error probabilities (ANOVA, = 0.05; 1-= 0.95). Thus, the sample size analyses led to a total sample size of n = 21 required for a repeated measure, within factors ANOVA for one group and four measurements. To completely counterbalance the order of the four experimental conditions, we decided to include 24 participants.

Study design
This study followed a double-blind (regarding pretreatment), randomized, placebo-controlled crossover design. All participants underwent four EEG recording sessions, which differed regarding the pretreatment and continuous infusions. A median nine-day and minimum five-day interval separated each of the four test sessions.
The glycine-pretreatment was administered at a dosage of 200 mg/kg bodyweight ( Greenwood et al., 2018 ) as an intravenous infusion in 500 ml 0.9% sodium chloride (NaCl) solution over one hour. Placebo was administered analogously as a NaCl infusion. Both pretreatments were prepared by an unblinded third person prior to the recording sessions and the ready-to-use infusions were indistinguishable. Subsequently, during the S-ketamine sessions a subanesthetic dose of Sketamine hydrochloride (Ketanest® S, Pfizer) was administered intravenously in a 0.9% NaCl solution for a duration of 75 min. The infusion was started with an initial bolus of 10 mg over 5 min followed by a maintenance infusion of 0.006 mg/kg/min, reduced by 10% every 10 min ( Curic et al., 2019 ). Placebo was administered analogously as NaCl in-fusions. Heart rate, blood pressure and oxygen saturation were continuously monitored during all sessions. The clinical rater was blinded with respect to the pretreatment but not regarding the continuous infusion condition due to the clinical effects of S-ketamine.

Psychometric assessment
Immediately after each EEG-recording session, the psychiatric symptomatology was always assessed by the same experienced psychiatrist (S.C.) using the Positive and Negative Syndrome Scale (PANSS) ( Kay et al., 1987 ). Based on our previous studies the PANSS scores were evaluated using the five-factor model ( van der Gaag et al., 2006 ).

Stimuli
An attentionally demanding auditory reaction task used in previous studies of our group ( Curic et al., 2019 ;Leicht et al., 2015 ;Mulert et al., 2007 ) was employed 10 min after the start of the maintenance infusion. Three tones of different pitch (800, 1000 and 1200 Hz; 40 repetitions of each) were generated and presented to participants through electrostatic headphones in pseudorandomized sequences and interstimulus intervals (ISI: 2.5-7.5 s, mean: 5 s), resulting in a task duration of 10 min. The duration of each tone was 250 ms, and the sound pressure level was set to 85 dB. Participants were instructed to respond to these stimuli as quickly and accurately as possible, by pressing a button with the left index finger when the low tone was presented, and by pressing another button whenever the high tone occurred (target tones). The middle tone was not a target for the pressing of a button. Reaction times (the amount of time from stimulus onset to the pressing of a button) and errors (pressing no button or pressing the wrong button within a timeframe of 150 and 2000 ms after onset of the stimulus) were registered. Only trials with correct responses to target tones were considered for further analyses.

EEG recording
In a sound-attenuated and electrically shielded room subjects were seated in a reclined chair with a head rest with their eyes open and asked to look at a fixation cross presented at a monitor 1 m in front of them. The EEG was recorded at a sampling rate of 1000 Hz including an analog band-pass filter (0.1 -1000 Hz), with 64 active electrodes mounted on an elastic cap (actiCap, Brain Products, Gilching, Germany) in an extended 10/20 system ( Curic et al., 2019 ), using the Brain Vision Recorder software Version 1.21 (Brain Products). Eye movements were recorded through four EOG channels. An electrode at the FCz position was used as reference, the electrode at position AFz served as ground. Impedances were kept below 5 k Ω.

Data analysis
Data preprocessing and analysis was carried out using the Brain Vision Analyzer (BVA) Version 2.1 (Brain Products). The preprocessing procedure was conducted in accordance with previous studies ( Curic et al., 2019 ;Leicht et al., 2010 ). After band-pass filtering (1-80 Hz) and down-sampling to 500 Hz, a topographic interpolation (spherical splines) of up to five channels was performed (the mean number of interpolated channels did not differ between conditions). Channels were selected for interpolation if more than 5% of data in the respective channel was affected by non-physiological artifacts (e.g., flat lines or electrode "pop "), or muscle artifacts exceeding amplitudes of ± 70 μV. An independent component analysis (ICA) was performed to identify and remove blinks, drifts, muscle artifacts and saccadic spike potentials based on their characteristic topographies, time courses, and frequency distributions (the mean number of removed ICA components did not differ between conditions: PlaPla 6.8 components, Pla-Ket 7.6 components, Gly-Pla 6.0 components, Gly-Ket 6.9 components). The continuous EEG was segmented into epochs of 3000 ms starting 1000 ms prior to the auditory stimulus. Segments including amplitudes exceeding ± 70 V within a 600 ms window starting 200 ms pre-stimulus in any channel were automatically rejected. After re-referencing to common average reference and baseline correction (using an interval of − 210 to − 10 ms pre-stimulus), averaged event-related potential (ERP) waveforms were computed. Only waveforms based on at least 35 segments were accepted and included in further analyses.

aeGBR amplitude and phase-locking factor (PLF)
The aeGBR amplitude and PLF were computed using a complex Morlet wavelet transformation (Morlet parameter c = 5, instantaneous amplitude (Gabor) normalization) as applied in several previous studies ( Curic et al., 2019 ;Leicht et al., 2010 ;Mulert et al., 2010 ). This wavelet transformation was performed on averaged ERPs to obtain the phaselocked evoked gamma amplitude. The analysis followed our previously published procedures and extracts the highest value within the timeframe of 30-100 ms post-stimulus at the electrode Cz of the wavelet layer with the central frequency of 40 Hz (frequency range 32-48 Hz) ( Curic et al., 2019 ;Leicht et al., 2016Leicht et al., , 2015Leicht et al., , 2011Leicht et al., , 2010Mulert et al., 2007 ;Roach and Mathalon, 2008 ;Spencer et al., 2008 ;). The PLF was calculated by performing a wavelet transformation at the single trial level and extracting complex-phase information with all vector lengths normalized to the unit circle before averaging the phase information ( Curic et al., 2019 ;Leicht et al., 2010 ). Gamma PLF peaks were defined as the highest value of the wavelet layer centered around 40 Hz within the timeframe of 30-100 ms post-stimulus at the electrode Cz.

Source analysis of the aeGBR
Source-space localization analyses were performed with the lowresolution brain electromagnetic tomography (LORETA) KEY software package v2017 ( Pascual-Marqui, 2002 ). The EEG source localization of the aeGBR across 30-50 Hz was executed for every subject within a time window of 30 to 70 ms post-stimulus.
The voxel-wise comparison of cortical activities between conditions was conducted using a one-tailed t -test for paired groups (statistical significance threshold p < 0.05) provided in the sLORETA software, which was based on the direction of the effects in our previous results ( Curic et al., 2019 ) and the corresponding hypothesis. A statistical nonparametric mapping randomization method was used to automatically adjust for multiple comparisons with a Fisher's random permutation test with 5000 randomizations.

Statistical analyses
All statistical analyses were performed using IBM SPSS Statistics Version 24. Comparisons of reaction times and error rates, PANSS scores, aeGBR amplitude, and PLF between all conditions were conducted using 2 (ketamine) x 2 (glycine) repeated measure analyses of variance (RM-ANOVA). All follow-up simple main effect analyses and post-hoc contrasts were subject to a Bonferroni correction.
Based on our previous study, we specifically investigated the Pearson correlations between the PANSS negative scores and the relative changes of the aeGBR ( Curic et al., 2019 ). To calculate the percental changes, the Pla-Pla condition was defined as baseline and used as the reference for all contrasts.

Behavioral performance
We observed a significant main effect of ketamine on reaction times (F(1,23) = 28.4, p < 0.001), with increased reaction times occurring during the application of S-ketamine ( Fig. 1 A). Regarding error rates, a significant main effect of ketamine occurred (F(1,23) = 17.7, p < 0.001)  with S-ketamine increasing the error rate ( Fig. 1 B). Neither glycine pretreatment nor the interaction between glycine and ketamine affected the behavioral performances.

Auditory evoked GBR amplitude and PLF
Around 50 ms after stimulus presentation in all four conditions, the evoked gamma activity increased at electrode Cz ( Fig. 3 B). Regarding the peaks of the evoked gamma band amplitude, a significant interaction effect between ketamine and glycine occurred (F(1,23) ( Fig. 3 A).

Association between neurophysiological and psychopathological variables
To assess the association between neurophysiological (aeGBR) and psychopathological (PANSS negative score) changes following the application of S-ketamine, we compared the Pla-Ket and the Pla-Pla condition. The correlation analysis revealed that increases of the PANSS negative score were negatively correlated with the relative reduction of the aeGBR under the influence of S-ketamine (Pearson's r = − 0.53, p = 0.008) ( Fig. 5 A).
When contrasting the pretreatments of both ketamine sessions (Gly-Ket minus Pla-Ket) to assess the effect of glycine on this association, the reduction of the PANSS negative score correlated with the relative increase of the aeGBR after pretreatment with glycine (Pearson's r = − 0.67, p < 0.001) ( Fig. 5 B).
In an explorative post-hoc analysis, we divided all individuals into two groups with regard to their glycine-dependent reduction of the PANSS negative score comparing both ketamine sessions, using a clinically significant reduction (of > 5.5 points corresponding to the scale intervals of the Clinical Global Impressions-Severity (CGI-S) rating) as the cut-off ( Leucht et al., 2019 ). Employing binary logistic regression, the relative increase of the aeGBR ( ∆aeGBR) predicted a clinically relevant attenuation of the PANSS negative score comparing both ketamine conditions (Chi-Square = 13.91, df = 1, p < 0.001). This model determined a ∆aeGBR increase of 64,5% as the cut-off, which correctly predicted 18 out of 19 (94.7%) cases where there was no clinically significant attenuation of the PANSS negative score, and four out of five (80%) cases where there was a significant attenuation.

Discussion
We have presented evidence that pretreatment with glycine mitigates ketamine-induced impairments of the aeGBR. Furthermore, we have demonstrated that the relative changes of the aeGBR correlate with the severity of schizophrenia-like symptoms associated with the administration of ketamine. While glycine-pretreatment did not affect behavioral performance and PANSS factor scores at the group level, the aeGBR amplitude allowed us to identify individuals with a relevant psychopathological benefit from the pretreatment with glycine.
Cortical and subcortical GBO both at rest or task-driven (e.g. aeGBR) are critically involved in cognitive functions such as working memory ( Howard et al., 2003 ;van Vugt et al., 2010 ) and are found to be altered in schizophrenia ( Dienel and Lewis, 2019 ;Uhlhaas and Singer, 2010 ). The early aeGBR is known to be reduced across all stages of schizophrenia  and corresponding impairments can be found in the ketamine model ( Curic et al., 2019 ). Moreover, the activity of the aeGBR generators within the dACC is reduced in patients with schizophrenia  and in healthy subjects following the acute infusion of S-ketamine ( Curic et al., 2019 ). Accordingly, we were able to replicate these findings in the present study. Based on these observations the aeGBR has been proposed as a correlate to disrupted glutamatergic neurotransmission in schizophrenia, since the generation of oscillations in the gamma frequency range depends on a feedback-loop encompassing pyramidal cells and GABAergic interneurons ( Lisman et al., 2008 ). This premise is further supported by our finding that the alleviation of ketamine-related aeGBR impairments follows the modulation of glutamatergic neurotransmission through the NMDAR co-agonist glycine. Yet, the effect of glycine on humans had neither been studied in the ketamine model of schizophrenia ( Haaf et al., 2018 ) nor in light of aberrant GBO. Regarding another neurophysiological measure, the acute administering of glycine normalized the reduced duration mismatch negativity (MMN) amplitudes in patients suffering from schizophrenia ( Greenwood et al., 2018 ). This was paralleled by findings of an improved frequency MMN after continuous treatment with D -serine (another NMDAR glycine-binding site agonist) ( Kantrowitz et al., 2018 ). Further, MMN-deficits can be observed in rodents following the application of ketamine and can be attenuated through the application of glycine . These findings support the translational application of electrophysiological biomarkers, which could facilitate drug development ( Javitt et al., 2020 ).
In the aforementioned feedback-loop, PV + and SST + interneurons significantly contribute to the model of NMDAR-dysfunction in schizophrenia, since these subtypes express high densities of NMDARs containing either GluN2C (PV + ) or GluN2D (SST + ) subunits for both of which ketamine demonstrates a higher affinity under physiological conditions compared to other subunits ( Bygrave et al., 2019 ). Conversely, both types of interneurons are reduced in schizophrenia according to postmortem studies ( Konradi et al., 2011 ). One might speculate that the disruption of inhibitory feedback might lead to an excessive yet uncoordinated downstream release of glutamate through pyramidal cells, causing an imbalance of glutamatergic excitation and GABAergic inhibition (E/I imbalance) ( Kehrer et al., 2008 ). This offers a feasible explanation for the counterintuitive observation that the acute administering of racemic ketamine increases extracellular glutamate levels in several brain regions assessed by means of 1H-MRS ( Stone et al., 2012 ), paralleled by findings in schizophrenia ( Merritt et al., 2016 ). Further, the reports of increased resting-state gamma band oscillations in schizophrenia Baradits et al., 2019 ) as well as corresponding findings in the ketamine model ( Bianciardi and Uhlhaas, 2021 ) -seemingly contradicting the notion of a hypo-glutamatergic state -might be attributed to a putative lack of synchronized neuronal activity. Hence, the concurrent inhibition of PV + and SST + GABAergic interneurons offers an explanation implementing both the incapacity to increase GBO following a cognitive-demanding stimulus and elevated GBO at rest in light of the E/I imbalance resembling a reduced signal/noise ratio.
Accordingly, animal studies found the ability to increase stimulusevoked gamma to be diminished ensuing global disruption of NMDAR signaling or the administering of NMDAR-antagonists or selective reduction of PV + -cell activity by means of either optogenetic methods ( Sohal et al., 2009 ) or gene knockout ( Bygrave et al., 2019 ). Whereas, the results of animal studies focusing on SST + interneurons remain limited and inconclusive ( Alherz et al., 2017 ). Nonetheless, a dysfunction restricted to a specific neuron subtype is likely not enough to explain the pathophysiology and complex neurophysiological changes seen in schizophrenia ( Bygrave et al., 2019 ).
Bearing all considerations outlined above in mind, the association between an impaired early aeGBR and the emergence of schizophrenialike negative symptoms as seen in this study leads to the presumption that glutamatergic dysfunction, aberrant neural oscillations, and negative symptoms are interdependent. Accordingly, our results contribute to a growing body of evidence that suggests an interrelation of impaired evoked GBO in a cognitive demanding task based on NMDAR dysfunction and negative symptoms as reported in both the ketamine model ( Curic et al., 2019 ) and schizophrenia itself . These assumptions are further substantiated by our discovery that the NM-DAR co-agonist glycine mitigates ketamine-related aeGBR impairments which in turn presents as a recovery of PANSS negative scores.
Remarkably, four out of five individuals showing a clinically relevant attenuation of the PANSS negative score ( Leucht et al., 2019 ) following glycine-pretreatment could be identified post-hoc by the relative aeGBRincrease with a high specificity and sensitivity ( Fig. 5 B). Hence, the aeGBR could serve as a putative biomarker for target-engagement of glutamatergic remedies in schizophrenia and could help to identify individuals who might benefit from the corresponding treatment. Insufficient target-engagement or competing target-engagement of co-administered medication (i.e. clozapine) might explain the varying results of glycine and D -serine treatments in schizophrenia ( Kantrowitz, 2017 ). Nevertheless, further verification of our observations is required given the heterogeneous response to the paradigm and considering we are the first to report a mitigating effect of a glutamatergic substance on psychopatho-logical and neurophysiological aberrations in the ketamine model of schizophrenia.
An extensive body of research supports the acute application of ketamine to healthy subjects as a pharmacological model of schizophrenia given that it produces changes in behavior, clinical symptoms, as well as neurophysiology that mimic changes seen in patients ( Frohlich and Van Horn, 2014 ;Moghaddam and Javitt, 2012 ). In line with these findings, we observed schizophrenia-like symptoms, assessed by means of the PANSS, accompanied by increases in reaction time and error rates following the application of S-ketamine. Still, our results were tested in healthy subjects using a disease model and require a replication in patients with schizophrenia. Glycine-pretreatment however, paralleled by unsuccessful clinical trials of glutamatergic agents ( Kantrowitz, 2017 ), did not reverse these effects at the group level, while clearly mitigating ketamine-associated aeGBR reductions. Nonetheless, glycinepretreatment reduced PANSS negative scores in the subset of individuals that also demonstrated a profound increase of the aeGBR, whilst no individuals with an increase of the PANSS negative factor by more than two points responded to glycine-pretreatment regarding the aeGBR.
Finally, our study was limited by several factors that merit consideration. First, the small sample size of the experiment represents an important limitation, possibly contributing to the fact that we did not observe a mitigating effect of glycine on behavioral results (e.g., error rate or reaction time) due to medium or small effect sizes. Moreover, we acknowledge that the results of our post-hoc binary logistic regression analysis demand replication with a larger sample size. Secondly, the homogenous group of participants (male sex and highly educated) limits the generalizability of our results and warrants replication with a larger and more heterogeneous subject collective. Thirdly, the psychotomimetic effects of S-ketamine could have led to the partial unblinding of participants. While blinding for the pretreatment was thoroughly realized, further studies could implement dopaminergic agonists or non-glutamatergic psychotomimetic agents in place of saline infusions as the control for ketamine.
In conclusion, this is the first study to report a normalizing effect of glycine, an NMDAR co-agonist, on ketamine-related decreases of the aeGBR peak amplitude and on the activity of its generator located in the ACC in healthy human subjects during the performance of a cognitively demanding auditory choice reaction task. Further, we found an association between glycine-related aeGBR increases and improvements in schizophrenia-like negative symptoms. Remarkably, an increase of the aeGBR predicted a clinically relevant psychopathological response to glycine-pretreatment. This points to the applicability of the aeGBR as a putative rapid biomarker for responders to a glutamatergic therapy for negative symptoms. To this end, our intriguing results call for the investigation of the transferability of this effect to patients suffering from schizophrenia. Prior to a clinical trial we suggest a proof-of-concept investigation of glycine-effects concerning gamma oscillations in a small to medium sample of patients with schizophrenia as being the next logical step.