On post-surgery days (PSD) 10–30, 40 mg/kg CORT (C2505, Sigma-Aldrich, Milwaukee, WI) was subcutaneously administered, except in the control group which received 5% DMSO/saline vehicle. Previous studies have demonstrated that this dose and duration induces depressive-like symptoms, such as increases in anhedonia, anxiety, and despair, decreases in evoked grooming behavior, impairments in declarative memory, and weight deficits (Sterner and Kalynchuk, 2010). Animals received the CORT dose after LFP recordings, treatment administrations, and behavioral tasks were completed for the day.

On PSD 24–30, 15 mg/kg ketamine hydrochloride (K2753, Sigma-Aldrich, Milwaukee, WI) were intraperitoneally administered to the CORT+KET and CORT+DBS+KET groups. This dose and duration was chosen based on previous studies (Garcia et al., 2008), which show a rescue of several depressive-like symptoms listed in the previous section. Saline vehicle was administered to the control, CORT, and CORT+DBS groups.

On PSD 24–30 the mPFC electrode of animals in the CORT+DBS and CORT+DBS+KET groups were connected to an isolated pulse stimulator (STG 4008, Warner Instruments LLC, Hamden, CT) via a twisted wire cable. Continuous electrical stimulation was delivered using the following stimulation parameters: 100 microsecond square wave pulses at 130 Hz and current of 80 μA for 30 min. These parameters were selected based on previous literature (Hamani and Nóbrega, 2010; Hamani et al., 2012), which demonstrated a rescue of several depressive-like symptoms listed in the CORT section. Animals in the CORT+DBS+KET group were administered the ketamine dose immediately before stimulation began.

Non-specific changes in locomotor activity after corticosterone, ketamine, and stimulation (Karatsoreos et al., 2010; Parise et al., 2013; Papp et al., 2022) can confound behavioral and electrophysiological results, therefore we measured average velocity in the Open Field Test on PSD 31. Animals were placed in an open, square arena (50 × 50 cm) and allowed to explore for 10 min. The average speed of locomotion was calculated using the video tracking software TopScan Lite (Clever Sys Inc., Reston, VA). Similarly, non-specific changes in grooming behavior can confound the Groom Test (Planchez et al., 2019), therefore the number of seconds spent spontaneously grooming during the OFT were evaluated.

Cognitive dysfunction, including deficits in memory, often occur in depression (American Psychiatric Association, 2013). To measure these, we employed the Novel Object Recognition Test (NORT). In this test, a higher proportion of time spent with a novel object, compared to a familiar object, is interpreted as representing functional object recognition memory (Antunes and Biala, 2012). The methodology employed is in the Supplementary materials. In short, we calculated the average percent change in the Discrimination Index (DI) (difference in time exploring the novel versus familiar object, divided by the total time exploring both objects) between a familiarization day and two test days.

As healthy rodents groom themselves thoroughly in response to soiling of their coats, reduced evoked grooming in the Groom Test (GT) has been interpreted to represent apathy, a core symptom of depression (Kennedy, 2008; Planchez et al., 2019). The animals performed the GT on PSD 35, their dorsoposterior coats were sprayed with a 10% sucrose solution, and were observed for 10 min in the OFT/NORT arena. Video recordings were evaluated for the number of seconds spent grooming.

Anxiety is highly comorbid with depression (American Psychiatric Association, 2013), and has been shown to be decreased after ketamine treatment, though whether it relates to its therapeutic efficacy for depression is unclear (Hartland et al., 2023). Therefore, on PSD 36, the Elevated Plus Maze (EPM) was used to measure anxiety, where a higher proportion of time in the innately fear-inducing “open” arms is interpreted as less baseline anxiety (Planchez et al., 2019). The maze and methodology are described in the Supplementary materials.

Despair, the second core symptom of depression beside apathy (Kennedy, 2008), is typically measured in rodents via the Forced Swim Test (FST), where increased immobile time in the test trial putatively represents despair (Planchez et al., 2019). On PSD 38, the animals were habituated to the FST testing chamber (glass cylindrical tub 50 cm tall, 25 cm diameter, filled with 21°C water 40 cm high) for 10 min. On PSD 39, the animals were tested for 5 min. A video recording of these five minutes was analyzed to determine the number of seconds spent immobile. After each trial the animals were placed in a heated chamber, dried, and monitored for full recovery.

Weight deficits are another common symptom of depression (American Psychiatric Association, 2013). Furthermore, similar to locomotor activity, non-specific changes in weight can confound behavioral results. Therefore, weights were measured at the “pre” CORT/treatment time point PSD 8–10, as well as the “post” time points PSD 30–32 and 37–39. Relative weights were calculated by dividing an animal’s weight at the “post” time points by the “pre” time point. The average of three days was used due to high variability in weight caused by the SPT fasts.

Specific behavioral differences between groups were explored through post-hoc tests, as shown in Figure 2. During the GT (Figure 2A), the CORT group spent significantly less time grooming than the control group (Supplementary Table S1, Row 1), indicating apathy-like symptoms had been induced. The groups that received DBS or ketamine treatment separately spent significantly more time grooming than the CORT group, representing a rescue of apathy-like symptoms. In contrast, the CORT+DBS+KET group did not significantly differ from the CORT group, suggesting apathy-like symptoms were not rescued. In addition, non-specific alterations in grooming activity can confound the GT, however, spontaneous grooming in the OFT did not differ between groups (Figure 2B).

The CORT group spent more time immobile in the FST (Figure 2C), compared to the control group, indicating despair-like symptoms had been induced (Supplementary Table S1, Row 2). This behavior was rescued in all three treatment groups, as they spent less time immobile than the CORT group.

In the NORT (Figure 2D), the percent change in DI was greater in the CORT+DBS group than the CORT group (Supplementary Table S1, Row 3). This indicates an improvement in object memory, however, as there was no deficit in the CORT group compared to the control group, this does not represent a “rescue” of memory function. There were no differences in open arm time percentage in the EPM between the CORT and control or treatment groups, indicating that, as has been previously reported after chronic corticosterone (Bertholomey et al., 2022), comorbid anxiety-like symptoms were not induced (Figure 2E).

No groups differed from the CORT group in average velocity in the OFT (Figure 2F), though the CORT+DBS+KET group had lower average velocities than the control group (Supplementary Table S1, Row 4). This indicated that while chronic CORT, alone or in combination with repeated ketamine or mPFC DBS, did not induce a deficit in locomotion, a combination of all three reduced average velocity.

The average relative weight in the CORT group was less than in the control on PSD 31 (Figure 2G) and PSD 38 (Figure 2H), indicating depression-like weight deficits had been induced. The three treatment groups did not differ from the CORT group, and had lower relative weights than the control group on PSD 31 (Supplementary Table S1, Row 5) and PSD 38 (Supplementary Table S1, Row 6). Therefore none of the treatment strategies tested rescued these deficits in weight.

Theta power decreased relative to pre-treatment baseline on PSD 24+29 significantly in the CORT, CORT+KET, and CORT+DBS groups (Figure 3B; Supplementary Table S2, Row 1). For the CORT group, this differed significantly from the control and CORT+DBS+KET groups, which did not change. Meanwhile, theta peak frequency significantly increased relative to baseline in the control group, which differed trendwise from the CORT group which did not change.

In the control group, low gamma peak frequency increased significantly relative to baseline and trendwise relative to the CORT group (Supplementary Figure S2A). Meanwhile, low gamma peak width increased trendwise relative to the CORT group in the CORT+KET and CORT+DBS groups.

The offset significantly increased in the CORT+KET group immediately after treatment on PSD 24+29 relative to both baseline and the CORT group, which decreased significantly relative to the pretreatment baseline (Figure 3C; Supplementary Table S2, Row 2). In contrast, the offset decreased trendwise in the CORT+DBS+KET group relative to baseline.

Changes in the exponent were very similar. The exponent increased in the control (trendwise) and CORT+KET group (significant) relative to the pretreatment baseline and the CORT group. In contrast, the exponent of the CORT+DBS+KET group decreased significantly compared to baseline and trendwise compared the CORT group.

Sample entropy significantly decreased in the control and CORT+KET groups relative to their baseline and the CORT group immediately after treatment on PSD 24+29 (Figure 3D; Supplementary Table S2, Row 3). Meanwhile, the CORT and CORT+DBS+KET group both significantly increased in sample entropy relative to their pretreatment baseline. The CORT and CORT+DBS+KET groups increased relative to baseline across all parameter combinations tested (Supplementary Figure S1), while the control group increased in 7 of 12 pairings. Furthermore, the control and CORT groups significantly differed in every pairing, while the CORT and CORT+KET groups differed in all but one.

Theta power by PSD 38 decreased significantly in the CORT, CORT+KET, and CORT+DBS+KET groups relative to baseline (Figure 4B; Supplementary Table S2, Row 6). Theta power decreased trendwise more in the CORT+KET group compared to the CORT group. Furthermore, in control, CORT, and the CORT+KET group, theta peak frequency decreased significantly relative to baseline. Like power, theta peak frequency decreased trendwise more in the CORT+KET group compared to the CORT group.

Low gamma power (trendwise), peak frequency (significant), and peak width (significant) decreased relative to baseline by PSD 38 in the CORT group (Supplementary Figure S2C; Supplementary Table S3, Row 6). Similarly, low gamma peak frequency decreased trendwise by PSD 38 in the CORT+KET group.

The exponent significantly decreased by PSD 38 in the CORT and CORT+DBS groups (Figure 4C; Supplementary Table S2, Row 7) relative to pre-CORT baseline, while it increased trendwise in the CORT+KET group. For the CORT group, this differed significantly from the CORT+KET group, and trendwise from the control and CORT+DBS+KET groups. Furthermore, the offset decreased trendwise in the CORT group, which differed significantly from the CORT+KET group and trendwise from the CORT+DBS+KET group.

By PSD 38, sample entropy was significantly lower in the control, CORT+KET, and CORT+DBS groups compared to the CORT group (Figure 4D; Supplementary Table S2, Row 8), which had increased relative to its baseline. The CORT+KET group decreased trendwise from its baseline. It should be noted that while the control and CORT+KET groups differed from the CORT group in all 12 parameter combinations, the CORT and CORT+DBS groups only differed in one parameter combination (Supplementary Figure S4). The CORT group increased significantly from baseline in seven parameter combinations.

We first explored the relationship between the depression-related measures, the Groom Test (GT) on PSD 35 and the Forced Swim Test (FST) on PSD 39, and the long-term electrophysiological changes on PSD 38. Electrophysiological correlates of behavior at this time parallel clinical biomarkers for steady-state depression or remission, given the extended time since the last CORT or treatment administration (eight days). Furthermore, as this is the time point closest to the GT and FST, these electrophysiological correlates may provide evidence of the mechanism underlying the long-term depression-related behavioral effects of CORT and the treatment strategies. For the control group pool, healthy performance in the Groom Test correlated with the increase in theta peak frequency in the control group (Figure 5A; Supplementary Table S3, Row 1). In the CORT+DBS group pool, remission-like performance in the Groom Test correlated with the increase in low gamma peak width in the treated group compared to the CORT group (Figure 5B). In the CORT+KET group pool, the Groom Test correlated with the increase in the exponent (Figure 5C) and offset, as well as the decrease in sample entropy (Figure 5D), in the CORT+KET group relative to the CORT group and its baseline. In addition, remission-like performance in the Forced Swim Test correlated with the increase in offset compared to the CORT group (Figure 5E).

Similarly, the second set of correlations examined the relationships between the depression-related measures (FST, GT), and short-term (PSD 31) electrophysiological and cognitive changes. In the control group pool, correlates at this time point parallel clinical biomarkers that track the early stages of depression after exposure to prolonged stress. In this pool, the increased exponent in the control group compared to the CORT group correlated with the GT (Figure 5F; Supplementary Table S3, Row 2). In contrast, in the treatment group pools, correlates at this time point parallel clinical biomarkers measured one day after treatment offset that may predict long-term remission. For the CORT+KET group pool, the increase in the exponent in the treated group also correlated with the GT (Figure 5F), while in the CORT+DBS group pool the improvement in the object memory task, the NORT, in the treated group correlated with the GT (Figure 5G). In addition, the decreased offset of the CORT+DBS group correlated with the FST (Figure 5H).

We then related the immediate electrophysiological changes post-treatment (PSD 24+29) with the depression-related GT and FST behavioral tasks. Here, the control group pool correlates are intended to represent biomarkers of the chronically-stressed mPFC dysfunctionally responding to acute stress, in this case the mild stress of restraint, intraperitoneal saline injection, and sham stimulation. In this group pool, we see that the increase of theta power in the control, compared to the decrease of the CORT group, correlated with their GT performance (Figure 6A; Supplementary Table S3; Row 3). Meanwhile, in the treatment group pools, these correlates represent biomarkers measured immediately after treatment administration that may predict long-term remission. In the CORT+DBS group pool, the increase in low gamma peak width (Figure 6B) and offset (Figure 6C) in the treated group, compared to the CORT group, correlated with the GT. Similarly, in the CORT+KET group pool, the increase in the offset relative to the CORT group and the baseline in the treated group also correlated with the GT (Figure 6D). In addition, the FST correlated with both the exponent (Figure 6E) and the offset (Figure 6F). Finally, in all four group pools, sample entropy immediately after treatment correlated with the GT (Figure 6G).

We then investigated whether the electrophysiological and behavioral biomarkers of depression-related behavior were correlated across the experiment, as this could hint at potential neural mechanisms of action for CORT and the treatment strategies. For each group pool, we first correlated the electrophysiological biomarkers from the long-term time point PSD 38, which have the potential to be causally related to performance on the GT and FST, with electrophysiological changes at the short-term post-treatment time point PSD 31. Those correlated PSD 31 measures were then correlated with the measures taken immediately after treatment on PSD 24+29. In the control group pool, the increase in theta peak frequency in the control group on PSD 38 correlated with the increase in the exponent in the control group on PSD 31 (Supplementary Figure S5A; Supplementary Table S3, Row 4). In turn, the increase in the exponent of PSD 31 correlated with the increase in theta power immediately after sham treatment on PSD 24+29 (Supplementary Figure S5B; Supplementary Table S3, Row 5). In the CORT+KET group pool, the decreased sample entropy in the treated group compared to the CORT group on PSD 38 correlated with the decrease in theta peak frequency on PSD 31 (Supplementary Figure S5C). In turn, the decrease in theta peak frequency on PSD 31 correlated with the decrease in sample entropy (Supplementary Figure S5D) and increase in offset and low gamma peak width immediately after treatment on PSD 24+29. For the same group pool, the decreased exponent (Supplementary Figure S5E) and offset, as well as the increased sample entropy (Supplementary Figure S5F) on PSD 38 in the CORT+KET group correlated with the increased exponent on PSD 31. In turn, the increased exponent on PSD 31 correlated with the decreased sample entropy immediately after treatment on PSD 24+29 (Supplementary Figure S5G). Finally, in the CORT+DBS group pool, improved performance in the NORT in the treated group correlated with the immediate increase in low gamma peak width immediately after treatment (Supplementary Figure S5).

To investigate the relationship between mPFC electrophysiological measures and the weight deficits observed in the CORT group, we performed similar analyses as in the first two sections for the control group pool. Decreased relative weight at the long-term time point on PSD 38 correlated with a decreased exponent on PSD 38 in the CORT group (Supplementary Figure S6A; Supplementary Table S3, Row 6). This decrease on PSD 38 correlated with a decrease in the exponent at the short-term time point (Supplementary Figure S6B). In turn, as described in section 4.4.2, the increase in the exponent on PSD 31 correlated with the increase in theta power immediately after sham treatment. Meanwhile, weight deficits on PSD 31 correlated with the decrease in the exponent (Supplementary Figure S6C) and increase in sample entropy (Supplementary Figure S6D) immediately after treatment.

To ensure that differences in weight and locomotion did not confound our measures of depression-like behavior, we correlated each pool’s relative weights on PSD 38, and their average velocity in the OFT, with evoked groom and immobility time. While locomotion correlated with the GT in the control group pool (Supplementary Table S3, Row 7), there was no significant difference in locomotion between these two groups (Figure 2H). Therefore, this factor could not have driven the differences between groups in the GT. For the CORT+DBS+KET group pool, we also correlated locomotion with electrophysiological measures, however there were no correlations that explained the decreased locomotion in the treatment group. There were no significant correlations between weight and behavior. Overall, these findings are evidence that the deficits observed in the GT and FST were not due to non-specific motor or metabolic deficits.

As the neural underpinnings of sample entropy are not well understood, we tested whether sample entropy significantly correlated with spectral parameters. In general, sample entropy correlated consistently with the exponent and offset, and at times with theta power and low gamma peak width (Supplementary Figure S6, Row 2; Supplementary Table S3, Row 8/9).

Eight days after chronic CORT administration, we found that a long-term increase in theta peak frequency in the CORT group, compared to the control group, correlated with apathy-like behavior in the GT. At the same time point, a decrease in the exponent correlated with concurrent weight deficits, while earlier, at the short-term, one day post-CORT time point, a decreased exponent in the CORT group correlated with the GT. Together, these periodic and aperiodic markers represent potential biomarkers for early and steady-state depression symptoms after chronic stress. Further, when considering the neural underpinnings of these electrophysiological measures, our findings may point toward the potential long-term mechanisms of corticosterone in inducing depression-like behavioral and weight symptoms. The current understanding of steady-state depression etiology is the hyperactivity and excitability of the mPFC, which drives excessive long-range inhibitory signals to regions involved in motivation, cognition, reward, and emotion (Pizzagalli and Roberts, 2022). Therefore, since a decreased exponent represents decreased inhibition (Lombardi et al., 2017), and chronic stress decreases inhibitory GABAergic transmission in male rodents (Ghosal et al., 2020), these findings support the theory that corticosterone recapitulates the stress-based depression-inducing hyperexcitation of the mPFC (Sterner and Kalynchuk, 2010; Wilber et al., 2011; Arnsten et al., 2023). They also agree with in-silico findings regarding decreased exponent values in a inhibitory interneuron knock-down depression model (Guet-McCreight et al., 2024). While the potential role of aberrantly increased theta peak frequency in the depressive symptoms is unclear, we speculate that since hyperactivation of the hippocampal to mpFC pathway, which operates at the theta frequency (Padilla-Coreano et al., 2019), has been correlated with depression severity in humans (De Kwaasteniet et al., 2013; Sambataro et al., 2014) and rodent models (Airan et al., 2007), and the hippocampus operates at a higher theta frequency than the mPFC (Biskamp et al., 2017), perhaps the increase in peak frequency represents the pathway’s depression-related hyperconnectivity. However, more work will need to be done to understand the meaning of this potential biomarker.

While the short- and long-term time points represent steady-state depression-like conditions, electrophysiological markers at the immediate time point on PSD 24+29 may represent the depression-inducing, aberrant response of the mPFC to stress (restraint and sham treatment). At this time point, an immediate decrease in theta power and increase in sample entropy in the CORT group correlated with apathy-like performance in the GT, and each other. Sample entropy is a measure of signal irregularity and complexity, and has been shown to correlate with a region’s degree of functional activity, processing, and connectivity (Wang et al., 2018). It has also been demonstrated to increase in frontal EEG electrodes in depression patients (Čukić et al., 2020; Lin et al., 2020). Meanwhile, aperiodic-adjusted theta power has been shown to decrease in the mPFC when rats are in a perceived “safe” environment (Adhikari et al., 2010). Therefore, similar to the steady-state decrease in mPFC inhibition discussed previously, we hypothesize that the increase in functional activity and decrease in safety-related theta power during mild stress was indicative of corticosterone’s depression-inducing hyperactivation of the mPFC (Sterner and Kalynchuk, 2010; Wilber et al., 2011; Arnsten et al., 2023). Furthermore, we found that the exponent and theta peak frequency on PSD 38 correlated with the exponent on PSD 31, which in turn correlated with the exponent immediately after treatment on PSD 24+29. This consistent and interrelated thread of electrophysiological indicators within the mPFC hints at potential relationships between them, however, causality can not be inferred from the present study.

After ketamine treatment alone, we found that a long-term decrease in sample entropy, and an increase in the exponent and offset, significantly correlated with remission of apathy-like behavior in the GT, and despair-like behavior in the FST. At the short-term post-treatment time point PSD 31, one day after the last ketamine treatment, the GT again correlated with an increase in the exponent relative to the CORT group. This was on par with the increase in the control group, which also correlated with the GT. Alone, these findings indicate these three measures may function as biomarkers for long-term treatment efficacy after ketamine. However, when coupled with our growing understanding of the neural dynamics they represent, these measures may also point toward potential mechanisms of ketamine’s therapeutic action. As previously discussed, the aperiodic exponent and offset have been correlated with inhibition, and sample entropy has been correlated with functional activity and connectivity. Meanwhile, excessive functional activity and connectivity of the mPFC is central to theories regarding the etiology of depression and CORT-based depression models. Therefore, we hypothesize that the increase in inhibition and decrease in function, in line with the control group, potentially represent ketamine exerting its long-term therapeutic effects by “normalizing” mPFC hyperactivity. Indeed, ketamine increases the inhibitory tone of the mPFC of rats 24 h post-treatment (Yin et al., 2021; Mingardi et al., 2023), and reduces the functional connectivity of the human prefrontal cortex for up to two weeks post-treatment (Chen et al., 2019; Siegel et al., 2021). Therefore, the modulation of aperiodic parameters and sample entropy after ketamine are perhaps indicative of this sustained cellular and functional normalization. Interestingly, our finding of an increased exponent in the mPFC after ketamine parallels findings after human electroconvulsive and magnetic seizure therapy (Smith et al., 2023). Furthermore, immediately after treatment, the GT also correlated with acutely decreased sample entropy and increased offset, while the FST correlated with increased offset and exponent. These findings indicate that, in addition to their potential as short- and long-term treatment efficacy biomarkers, these measures also have potential as immediate biomarkers. Similar to our interpretation of the sustained post-treatment electrophysiological correlates, we speculate that these immediate correlates represent processes critical to a rapid mPFC “normalization” mechanism. Rodent studies have found ketamine normalizes mPFC dopaminergic (Wu et al., 2021) and GABAergic signaling (Ghosal et al., 2020), the latter in particular potentially driving the observed normalization of the aperiodic parameters. Meanwhile human MRI signal in the medial frontal cortex is normalized via reductions in functional activity and connectivity immediately after ketamine during an emotionally-valenced stimulus task (Reed et al., 2018; Morris et al., 2020), potentially paralleling our finding of decreased sample entropy.

Interestingly, in addition to being highly correlated with remission-like behavior and one other at the same time points, sample entropy, offset, and the exponent were also correlated with one another across time points. Tracking these correlations across time may help bridge the gaps in understanding between the relatively well studied immediate correlates of ketamine’s efficacy, and the less well understood long-term correlates. Specifically, we found that the three measures correlated with an increase in the exponent at the short-term timepoint, which in turn correlated with decreased sample entropy immediately after treatment. In addition, we found that the long-term decrease in sample entropy correlated with a short-term decrease in the theta peak frequency, which in turn correlated with the immediate decrease in sample entropy, as well as the increase in the offset and low gamma peak width. Interestingly, unlike the other correlates, the shift in theta peak frequency was in the opposite direction of the control group, decreasing even more than in the CORT group. As previously discussed, modulation of peak theta frequency may relate to modulation of the hippocampal-mPFC pathway, which has previously been demonstrated after ketamine in humans (Siegel et al., 2021) and rats (Gass et al., 2019). Meanwhile, resting-state low gamma activity has been shown to increase in the mPFC after ketamine in both rodents (Qin et al., 2023) and humans (Cornwell et al., 2012), which correlated with treatment efficacy in humans (Nugent et al., 2019). Taken together, our findings potentially indicate that the increase of low gamma activity and inhibition, and decrease in complexity, immediately after ketamine treatment drove a short-term increase in inhibition and decrease in theta activity, which culminated in the remission-related long-term increase in inhibition and decrease in complexity. However, as this interpretation is based on correlations, we can not draw definitive conclusions regarding causality.

After mPFC DBS alone, a long-term increase in the low gamma peak width and improvement in cognitive behavior in the NORT correlated with remission-like behavior in the GT. Furthermore, an increased offset at the short-term time point correlated with the FST. Finally, an immediate decrease in sample entropy, as well as the increase in the offset and low gamma peak width, all correlated with the GT. These correlations indicate these measures may serve as biomarkers for the long-term treatment efficacy after DBS, and when considering their neural origins and interrelationships, they may also hint at its mechanism of action. mPFC low gamma activity has previously been shown to increase to healthy levels after mPFC DBS in rodents (Jia et al., 2019) and humans (Scherer et al., 2023), and has been implicated in the cognitive functions of the mPFC, including memory in rodents (Zhang et al., 2019) and humans (Senkowski and Gallinat, 2015). Indeed, in the present study, improvements in object recognition memory in the NORT correlated with the immediate post-treatment increase in low gamma peak width. Furthermore, low gamma power has been shown to be generated via the interplay between glutamatergic and GABAergic neurons (Tort et al., 2013), and interestingly, cognitive deficits stemming from dysfunction of these cell types in the mPFC have been implicated in the etiology of apathy (Levy and Dubois, 2006). Particularly implicated are two components of recognition memory, working memory and episodic memory, which involve the mPFC (Ragozzino et al., 2002; Blumenfeld and Ranganath, 2007), and are measured by the NORT (Antunes and Biala, 2012). Furthermore, working memory performance has been previously correlated with apathy in human schizophrenia patients (Raffard et al., 2016). Therefore, our findings hint that immediate and long-term modulation of glutamatergic and GABAergic neuron interplay may have driven long-term improvements in cognition that then related to remission-like performance in the GT. The translatability of this finding is supported by work in humans showing that mPFC DBS increases gamma activity (Scherer et al., 2023) and improves long-term measures of memory (Runia et al., 2023). Again, we must be cautious in overinterpreting these correlations, more work will need to be done to demonstrate causal links, if any. Furthermore, it is unclear how the immediate and short-term modulation of complexity and inhibition could causally relate to the long-term depression-related behavioral measures given the lack of intermediate correlates and relatively weak modulation, in comparison to ketamine treatment.

Finally, we hypothesized that a novel combinatorial treatment of ketamine and mPFC DBS would be synergistically effective in its antidepressant-like effect. As in both treatments separately, this strategy rescued despair-like symptoms in the FST, however, it failed to rescue apathy-like symptoms in the GT (Figure 2). To search for an explanation for these counterintuitive findings, we looked for immediate, short- and long-term mPFC LFP correlates of the GT. The only hint we uncovered was a negative correlation with the increase in sample entropy immediately after treatment, on par with the CORT group, and in contrast to the control and separate treatment groups. Across the aforementioned biomarkers of the GT and FST, the combination group resembled a mixture of the other groups, which, coupled with a lack of correlations at the short- and long-term time points, hinders the identification of a potential mechanism across time. However, by investigating this interference in the modulation of mPFC activity further, future studies may uncover causal relationships between electrophysiological and behavioral changes.