The behavioral data of the animals whose brains were anatomically analyzed in this study have already been published (Chen et al. 2018). These behavioral data will be presented here as correlations to the anatomical findings from the current study. The current study examined correlations between three ante mortem values – wheel running, food consumption and body weight - and the levels and locations of NR2B in relation to excitatory synapses in layer 1 of the prelimbic area of mPFC. NR2B was labeled by the PEG method.
Areal Density of Excitatory Synapses
We compared the areal density of excitatory synapses belonging to pyramidal neurons and GABA-IN in layer 1 (mostly layer 1a, within 50 µm from pial surface) of the mPFC. Layer 1 of the 30 mg/kg cohort exhibited a significantly lower density of axo-spinous excitatory synapses (presumably of pyramidal neurons (White and Keller 1989)) than the 3 mg/kg cohort (Fig. 4a, mean ± SEM: 0.3700 ± 0.0143 for the 30 mg/kg cohort, 0.4656 ± 0.0170 for the 3 mg/kg cohort, p = 0.0007). The probability of encounter with randomly positioned objects would be lower for smaller objects (Sterio 1984). However, there was no significant difference in the average spine area between the two cohorts, calculated by averaging 50 spine areas of each animal (mean ± SEM in units of synapses per µm2: 226.2 ± 58.0 for the 30 mg/kg cohort; 341.1 ± 86.3, for the 3 mg/kg cohort, p = 0.484). This suggests that the lower areal density of axo-spinous excitatory synapses seen in the 30 mg/kg cohort did not result from decrease of spine sizes.
The number of axo-shaft excitatory synapses (presumably of GABA-IN (White and Keller 1989)) did not differ significantly between the two cohorts (Fig. 4B, mean ± SEM in units of synapses per µm2: 0.0751 ± 0.0132 for the 30 mg/kg cohort; 0.0847 ± 0.00845, for the 3 mg/kg cohort, p = 0.5503).
The Proportion of Excitatory Synapses Immunolabeled for NR2B Postsynaptically
Control ultrathin sections that were incubated in buffer lacking the primary antibody showed no PEG labeling at or near any of the 200 excitatory synapses that were sampled. This observation indicated that excitatory synapses with one or more PEG particles could be considered immunolabeled.
The proportion of excitatory synapses immunolabeled by NR2B was not different across cohorts, based on tallies of PEG particles across all synaptic locations - namely at and near the PSD, on extrasynaptic portions of the plasma membrane, or in the synaptic cleft or cytoplasm of spines forming excitatory synapses (p = 0.0956; mean ± SEM of 0.5968 ± 0.0643 for the 30 mg/kg cohort; 0.4553 ± 0.04619 for the 3 mg/kg cohort) or at excitatory synapses on dendritic shafts of GABA-IN (p = 0.0643; 0.5968 ± 0.0643 for the 30 mg/kg cohort; 0.4553 ± 0.04619 for the 3 mg/kg cohort ). Also, no group difference was found, when the PEG particles were tallied for membranous locations (at the extrasynaptic membrane, at or near the PSD). However, closer inspection revealed that the 30 mg/kg cohort exhibited a significantly higher proportion of spines immunolabeled for NR2B at the cleft and in the cytoplasm of spines (Fig. 5b, mean ± SEM at spines’ excitatory synaptic clefts: 0.0485 ± 0.0088 for the 30 mg/kg cohort, 0.0194 ± 0.0047 for the 3 mg/kg cohort, p = 0.0114; Fig. 5c, mean ± SEM of spine heads with PEG particles in cytoplasm: 0.254 ± 0.028 for the 30 mg/kg cohort, 0.186 ± 0.012 for the 3 mg/kg cohort, p = 0.0410).
At excitatory synapses of dendritic shafts of GABA-IN, no differences in the proportion immunolabeled for postsynaptic NR2B were found between the two cohorts at synaptic clefts (Fig. 5e, mean ± SEM: 0.0450 ± 0.0109 for the 30 mg/kg cohort, 0.0349 ± 0.0136 for the 3 mg/kg cohort, p = 0.571), in the cytoplasm near excitatory synapses (Fig. 5f, mean ± SEM: 0.432 ± 0.0569 for the 30 mg/kg cohort, 0.367 ± 0.0370 for the 3 mg/kg cohort, p = 0.353) or on the plasma membrane of the pre- or postsynaptic sides (not shown).
The level, rather than the proportion, of NR2B immunolabeled per excitatory synapse on dendritic spines and dendritic shafts yielded similar results. Specifically, the level of NR2B labeled at synapses of dendritic spines were significantly greater for the 30 mg/kg cohort, compared to the 3 mg/kg cohort at synaptic clefts (mean ± SEM: 0.0508 ± 0.0097 for the 30 mg/kg cohort, 0.0202 ± 0.0053 for the 3 mg/kg cohort, p = 0.0149) and in the cytoplasm (mean ± SEM: 0.410 ± 0.064 for the 30 mg/kg cohort, 0.261 ± 0.021 for the 3 mg/kg cohort, p = 0.0425) but not at the postsynaptic plasma membranes. There were no differences in the level of NR2B levels between the two cohorts at clefts of excitatory synapses on dendritic shafts (mean ± SEM: 0.0507 ± 0.012 for the 30 mg/kg cohort, 0.0395 ± 0.018 for the 3 mg/kg cohort, p = 0.610), in cytoplasm (0.916 ± 0.22 for the 30 mg/kg cohort, 0.534 ± 0.065 for the 3 mg/kg cohort, p = 0.122) or elsewhere, either.
Correlational Analyses of the Proportion of Excitatory Synapses Immunolabeled for NR2B with Food Consumption
Within the mPFC, the activation of pyramidal neurons projecting to dorsal raphe (DR) have been shown to promote food consumption of animals undergoing ABA, while GABAergic inhibition of this population of mPFC pyramidal neurons has been shown to underlie suppression of food consumption (Du et al. 2022). This finding prompted us to analyze the possible link between ketamine-mediated increase in food consumption of animals that had experienced ABA and the excitatory synapses formed on pyramidal neurons and GABA-IN of the mPFC. Compared to the control group with vehicle injection (not included in this anatomical analysis), the two ketamine cohorts ate significantly more during the food-restricted period of ABA1, FR2, corresponding to the day that ketamine was injected. Additionally, the 30 mg/kg cohort, but not the 3 mg/kg cohort also ate more on the day following the single ketamine injection, namely FR3 of ABA1. During ABA2, 14 days after ketamine injection, only the 30 mg/kg ketamine cohort ate significantly more than the control group during the food-restricted period of ABA2, spanning FR2 through FR4 (Chen et al. 2018). Prompted by these differences in food consumption behaviors across the two ketamine cohorts, correlation analyses between food consumption and the proportion of dendritic spines and shafts immunolabeled for postsynaptic NR2B at excitatory synapses during ABA1 FR, ABA2 FR, and recovery phases were conducted to determine whether individual differences in NR2B expression at excitatory synapses could have contributed to individual differences in the responsiveness to ketamine in terms of food consumption.
Correlations at axo-spinous excitatory synapses of pyramidal neurons. Fig. 6 shows the R-values of correlations between the proportion of excitatory synapses with postsynaptic NR2B levels and food consumption recorded for each of the days spanning ABA1 and ABA2. There were clear differences in the patterns between the two ketamine cohorts (Fig. 6a and b). The two time points that showed strong divergence between the cohorts were ABA2 R3 and ABA1 R4, which were examined more closely (Fig. 7).
Closer inspection of ABA2 R3 revealed that the 30 mg/kg cohort showed significantly higher food consumption than the 3 mg/kg cohort (mean ± SEM in units of kcal: 21.9 ± 0.894 for the 30 mg/kg cohort, 15.8 ± 0.962 for the 3 mg/kg cohort, p = 0.0005 by unpaired t-test, Fig. 7c). The 30 mg/kg cohort showed no correlation between NR2B immunoreactivity at dendritic spines and food consumption on any single day of recovery or average consumption across the days of recovery. In sharp contrast to the 30 mg/kg cohort, the 3 mg/kg cohort showed a significantly negative correlation between food consumption and the proportion of spines immunolabeled for postsynaptic NR2B at all locations except for near PSD (Pearson’s R = 0.2117, p = 0.6486 for the 30 mg/kg cohort; Pearson’s R=-0.7892, p = 0.020 for the 3 mg/kg cohort, Fig. 7d). The negative correlation indicates that those individuals consuming the most were those that expressed NR2B minimally at excitatory synapses on pyramidal neurons. These findings indicate, together, that enhanced excitability of pyramidal neurons dictated suppression of food consumption of the 3 mg/kg cohort. Conversely, the 30 mg/kg cohort was likely to have enhanced food consumption through weakening of food-suppressing excitatory pathway(s) that remained active in the brains of the 3 mg/kg cohort (discussion expanded with accompanying figure in the Discussion section).
By comparison, for both cohorts, correlations between food consumption and NR2B-immunoreactivity at dendritic spines were much weaker during recovery from ABA1 (Fig. 7b). For example, on ABA1 R4, when the 30 mg/kg cohort had significantly higher food consumption than the 3 mg/kg cohort (mean ± SEM in units of kcal: 20.3 ± 1.50 for the 30 mg/kg cohort, 13.81 ± 0.87 or the 3 mg/kg cohort, p = 0.002 by unpaired t-test, Fig. 7a), the 30 mg/kg showed only a weakly positive trend between food consumption and the proportion of spines immunolabeled for postsynaptic NR2B at all locations excluding the non-membranous near-PSD category, while the 3mg/kg cohort showed no correlation or trend (Pearson’s R = 0.5046, p = 0.2483 for the 30 mg/kg cohort, Pearson’s R = 0.0070, p = 0.9868 for the 3 mg/kg cohort, Fig. 7b). Similarly, on ABA1 R6, the 30 mg/kg cohort had significantly higher food consumption than the 3 mg/kg cohort (mean difference between cohorts ± SEM in units of kcal of 3.18 ± 0.928, p = 0.0045 on ABA1 R6 unpaired t-test), but the 30 mg/kg cohort showed only a weakly positive trend while the 3 mg/kg cohort exhibited no correlation or trend (on ABA1 R6 Pearson’s R = 0.4756, p = 0.2807 for the 30 mg/kg cohort, and Pearson’s R = 0.1146, p = 0.7870 for the 3 mg/kg cohort for all locations excluding the non-membranous near-PSD category). The lack of correlations during ABA1 indicates that pyramidal neurons either became disengaged in the regulation of food consumption and/or did not contribute in any uniform way towards the enhanced food consumption elicited by 30 mg/kg of ketamine.
Correlations at axo-shaft excitatory synapses of GABA-IN. Unlike the correlations found in pyramidal neurons, the correlations at excitatory synapses of GABA-IN appeared during recovery days after both ABA1 (Fig. 7e) and ABA2 (Fig. 7f) and only for the 30 mg/kg cohort. Specifically, on ABA1 R6 and ABA2 R2, a significantly positive correlation was found in the 30mg/kg cohort between food consumption and the proportion dendritic shafts immunolabeled for postsynaptic NR2B at the cleft, at the PSD, and near the PSD (Pearson’s R = 0.8058, p = 0.0288 for the 30 mg/kg cohort on ABA1 R6, Fig. 7e; Pearson’s R = 0.8231, p = 0.0229 for the 30 mg/kg cohort on ABA2 R2, Fig. 7f), indicating that those individuals with the greatest capacity to consume food expressed NR2B at excitatory synapses on GABA-IN most robustly, presumably suppressing the mPFC’s excitatory outflows that mediate suppression of food consumption. In contrast, for the 3 mg/kg cohort, only a weakly positive correlation trend was found on ABA1 R6 (Pearson’s R = 0.5888, p = 0.1246 for the 3 mg/kg cohort, Fig. 7e), and no trend was found on ABA2 R2 (Pearson’s R = 0.3901, p = 0.3394 for the 3 mg/kg cohort, Fig. 7f). This lack of trend for the 3 mg/kg cohort supports the interpretation that the inefficacious dose could not trigger the increased expression of NR2B at excitatory synapses of GABA-IN for suppressing the food consumption-suppressive excitatory outflow. The appearance of correlations for the NR2B-immunoreactivity during ABA1 for axo-shaft excitatory synapses of GABA-IN and not for the excitatory synapses of pyramidal neurons’ dendritic spines indicates that excitatory synapses on GABA-IN responded more quickly to the 30 mg/kg ketamine injection than the pyramidal neurons.
All data reported above were analyzed based on values of the proportion of dendritic spines or shafts immunolabeled for NR2B at excitatory synapses. Correlation analyses performed by counting every single NR2B immunolabel and normalizing the sum by the number of axo-spinous or axo-shaft excitatory synapses (i.e., NR2B levels) also yielded significance for the same days of food consumption.
Together, these data indicate that the long-lasting gain of ABA resilience by the 30 mg/kg dose, measured as an increase of food consumption more than 20 days after the single injection, correlated with increases of NR2B-immunoreactivity at excitatory synapses of GABA-IN and with the loss of correlation of NR2B-immunoreactivity at axo-spinous synapses. Conversely, we hypothesize that the lack of an ameliorative effect of the 3 mg/kg dose during ABA2 is likely to have been due, in part, to the failure of the mPFC of these animals to de-couple the mPFC synaptic circuitry linking axo-spinous synapses with suppression of food consumption and/or enhance NR2B expression at excitatory synapses on GABA-IN that suppress the excitatory outflow from mPFC that suppress food consumption (discussion is expanded in the Discussion section, with accompanying figure).
Correlations Between Wheel Activity and the Proportion of Excitatory Synapses Immunolabeled for NR2B Postsynaptically
Earlier work from this lab showed that the mPFC pyramidal neurons projecting to dorsal medial striatum positively modulate wheel running of animals experiencing ABA while inhibition by GABAergic interneurons in mPFC suppresses wheel running (Santiago et al. 2021). We also reported earlier that both ketamine cohorts ran significantly less than the control group during ABA2 but not during ABA1(Chen et al. 2018). These observations prompted us to assess correlations between wheel activity during ABA2 and the proportion of dendritic spines and shafts immunolabeled for postsynaptic NR2B at excitatory synapses in the mPFC of the two ketamine cohorts.
Correlations at axo-spinous excitatory synapses of pyramidal neurons. Figures 8a and 8b show the day-by-day values of R-values of Pearson correlation between wheel running during different time zones with NR2B immunoreactivity. Similar to the pattern observed for food consumption, there were clear differences in the correlation patterns between the two cohorts with regard to wheel activity. Note that wheels were absent during recovery phases, thereby precluding correlation analyses during the recovery days.The 30 mg/kg cohort exhibited similar dynamics of R-values across the different NR2B locations. During FAA, the 30 mg/kg cohort, but not the 3mg/kg cohort, exhibited a dip of R-values for most NR2B locations except for the location at-PSD (Fig. 8b). This difference across the cohorts prompted us to look more closely at the correlations between wheel running and NR2B labeling during FAA (Fig. 9).
During the ABA2 FR period, both doses yielded significant reductions in averaged wheel running per day compared to the vehicle cohort (Chen et al. 2018). Although wheel activity between the two ketamine cohorts were not significantly different (mean ± SEM in units of km: 6.106 ± 1.69 for the 30 mg/kg cohort; 1.656 ± 1.74 for the 3 mg/kg cohort, p = 0.7938 by the unpaired t-test, Fig. 9a), correlation analysis revealed differences across the cohorts. Specifically, at excitatory synapses of dendritic spines of pyramidal neurons of the 30 mg/kg cohort, a negative trend was found between wheel activity and the proportion of spines labeled for postsynaptic NR2B at all postsynaptic locations except for near-PSD (Pearson’s R=-0.6274, p = 0.0959 for the 30 mg/kg cohort, Fig. 9b), indicating that those individuals with minimal FAA were those that had reduced NR2B expression at excitatory synapses of pyramidal neurons the most. No such trend was found in the 3 mg/kg cohort (Pearson’s R = 0.3822, p = 0.3500 for the 3 mg/kg cohort, Fig. 9b).
Correlations at axo-shaft excitatory synapses of GABA-IN. Axo-shaft excitatory synapses of the 30 mg/kg cohort exhibited higher correlations between NR2B and wheel running than at axo-spinous synapses, especially during the dark hours of FR3 (21:00–7:00) (Fig. 8d, 9b). There was no significant group difference in the wheel activity between the two ketamine cohorts on ABA2 FR3’s dark hours of 21:00–7:00 (mean ± SEM in units of km: 12.73 ± 2.18 for the 30 mg/kg cohort, 9.87 ± 2.46 for the 3 mg/kg cohort, p = 0.4086 by unpaired t-test, Fig. 9c). Yet, for the 30 mg/kg cohort, there was a highly significant positive correlation between wheel activity on ABA2 FR3 during dark phase 21:00–7:00 and the proportion of dendritic shafts of GABA-IN immunolabeled for postsynaptic NR2B at all postsynaptic locations other than near the PSD (Fig. 9d, Pearson’s R = 0.9437, p = 0.0004 for the 30 mg/kg cohort). No such correlation was found in the 3 mg/kg cohort (Fig. 9d, Pearson’s R = 0.1652, p = 0.6959 for the 3 mg/kg cohort).
Data reported were also analyzed by correlating wheel running and the levels of dendritic spines or shafts immunolabeled for NR2B. Correlation analyses counting every single NR2B immunolabel and normalizing the sum by the number of spines or shafts also revealed significance for the same sets of data.
Together, these data indicate that the dose of 30 mg/kg was more efficacious in promoting ABA resilience measured as reduced wheel running through gains of NR2B at axo-spinous excitatory synapses of pyramidal neurons and decreased NR2B-immunoreactivity at axo-shaft excitatory synapses of GABA-IN. These observations are the opposite of the previous finding pertaining to the mPFC of animals without ketamine treatment (Santiago et al. 2021). The current data suggest that ketamine may mediate plasticity of the mPFC synaptic circuitry by unmasking mPFC pathways supporting adaptive behaviors, such as to suppress the hunger-evoked hyperactivity. Moreover, ketamine promoted synaptic changes in opposite directions depending on the cell type (pyramidal neuron vs GABA-IN), as was described above for the ketamine-induced changes supporting food consumption.
Correlations Between Body Weights and the Proportions of Excitatory Synapses Immunolabeled for NR2B
Body weight would be expected to be affected jointly by food consumption and wheel activity, each of which are regulated by specific synaptic circuits within mPFC (Santiago et al. 2021; Du et al. 2022). Pearson’s correlation analyses revealed correlations between body weight and the levels of postsynaptic NR2B at excitatory synapses of dendritic spines, presumably of pyramidal neurons and dendritic shafts, presumably of GABA-IN. Similar to correlations between NR2B immunoreactivity and food consumption (Fig. 6), correlation patterns of the two cohorts also differed for body weights (Fig. 10). R-values of the 3 mg/kg cohort were low during the two recovery phases and higher during FR periods. Also, the R-value dynamics of postsynaptic NR2B of the 3 mg/kg cohort were uniform across all synaptic locations (Fig. 10a). Oppositely, the 30 mg/kg cohort exhibited high R-values during the two recovery phases and lower R-values during FR periods. Within the 30 mg/kg cohort, postsynaptic NR2B near the PSD had a different R-value dynamic than NR2B at all other locations, as those near the PSD exhibited negative R-values during most of the course of ABA (Fig. 10b). Overall, ABA1 R3 and ABA2 R3 stood out as the days showing the strongest correlation between body weight and NR2B at excitatory synapses of pyramidal neurons’ spines and GABA-IN’s dendritic shafts. These correlations are shown in greater detail in Fig. 11.
Correlations at axo-spinous excitatory synapses on pyramidal neurons. Both the 3 mg/kg and 30 mg/kg cohorts responded to the single ketamine injection with increased body weight, relative to the vehicle group at recovery following ABA2, and only the 30mg/kg cohort improved in weight gain at recovery following ABA1 and during ABA2 FR period (Chen et al. 2018). However, two sample mean t-tests showed no differences between the body weights of the 3 mg/kg and 30 mg/kg ketamine cohorts on any of the days of recovery, including ABA1 R3 or ABA2 R3 (mean ± SEM in units of grams: 17.8 ± 0.619 for 30 mg/kg cohort, 18.3 ± 0.618 g for the 3 mg/kg cohort, p = 0.8199 on ABA1 R3, Fig. 11a; mean ± SEM in units of grams: 19.2 ± 0.507 for the 30 mg/kg cohort; 20.12 ± 0.580 for the 3 mg/kg cohort, p > 0.05 for ABA2 R3, Fig. 11c). In spite of the lack of difference in body weights between the two cohorts, comparisons of the correlations revealed differences across the doses. On ABA1 R3, a significant positive correlation was found between body weights and the proportion of spines immunolabeled for all postsynaptic NR2B other than near-PSD of the 30 mg/kg cohort but not the 3 mg/kg cohort (Pearson’s R = 0.7667, p = 0.0443 for the 30 mg/kg cohort and Pearson’s R = 0.0048, p = 0.9911 for the 3 mg/kg cohort, Fig. 11b). Similarly, on ABA2 R3, a positive correlation trend was found for the 30 mg/kg cohort but not the 3 mg/kg cohort (Pearson’s R = 0.6316, p = 0.0930 for the 30 mg/kg cohort and Pearson’s R = 0.3655, p = 0.3732, Fig. 11d). A positive correlation for the 30 mg/kg cohort (Fig. 11d) was still present when body weight was averaged from ABA1 R2-R7 (not shown), suggesting ultrastructural redistribution of NR2B that occurred and stabilized shortly after the end of ABA1’s FR period.
Correlation analyses for the same days and ultrastructural location of NR2B revealed equally significant outcomes using NR2B levels at excitatory synapses instead of the proportions of NR2B-immunoreactive excitatory synapses.
Correlations at axo-shaft excitatory synapses on GABA-IN. On ABA1 R3, a negative trend was found for the correlation between body weight and the sum of the proportion of shafts immunolabeled for postsynaptic NR2B at and near the PSD of excitatory synapses of GABA-IN of the 3 mg/kg cohort. No correlation or trend was found for the 30 mg/kg cohort (Pearson’s R = 0.2033, p = 0.6620 for the 30 mg/kg cohort and Pearson’s R=-0.5531, p = 0.1551 for the 3 mg/kg cohort, Fig. 11e). On ABA2 R3, there was also a negative trend for the 3 mg/kg cohort between body weight and the proportion of dendritic shafts immunolabeled NR2B at the cleft, at PSD, and near PSD of GABA-IN (Pearson’s R=-0.5995, p = 0.1162 for the 3 mg/kg cohort, Fig. 11f). Oppositely, a positive trend was found in the 30 mg/kg cohort (Pearson’s R = 0.5946, p = 0.1200 for the 30 mg/kg cohort, Fig. 11f).
The same behavioral data set was correlated to NR2B levels, assessed by counting every single NR2B immunolabel and normalizing the sum by the number of dendritic shafts encountered. The outcome revealed greater significances of correlations (Fig. 11g, h). In contrast to the negative trend shown in Fig. 11e for the proportions of synapses immunolabeled, there was a significantly negative correlation between body weight and the level of dendritic shafts immunolabeled for NR2B for the 3mg/kg cohort on ABA1 R3, and no correlation or trend for the 30mg/kg cohort (Pearson’s R = 0.3562, p = 0.4330 for the 30mg/kg cohort and Pearson’s R=-0.7176, p = 0.0450 for the 3mg/kg cohort, Fig. 11g). Likewise, in contrast to the negative trend shown in Fig. 11f, there was a significant positive correlation between body weight and the level of dendritic shafts immunolabeled for NR2B for the 30mg/kg cohort but only a negative trend for the 3mg/kg cohort (Pearson’s R = 0.7428, p = 0.0348 for the 30mg/kg cohort and Pearson’s R=-0.6738, p = 0.0670 for the 3mg/kg cohort, Fig. 11h).
Overall, at dendritic spines of pyramidal neurons, the increase in the proportion and levels of NR2B had an ameliorative effect on body weight restoration and more so for the 30 mg/kg dose than for the 3 mg/kg dose (Fig. 11b). In sharp contrast to the changes at dendritic spines, the changes at excitatory synapses on GABA-IN reflected losses of NR2B of 3 mg/kg cohort, seen as ameliorative for weight retention, changing to a positive correlation by ABA2 and enhanced NR2B that contributed to 30 mg/kg cohort’s body weight gain.