Electroconvulsive treatment prevents chronic restraint stress‐induced atrophy of the hippocampal formation—A stereological study

Abstract Introduction Electroconvulsive therapy (ECT) is one of the most efficient treatments of major depressive disorder (MDD), although the underlying neurobiology remains poorly understood. There is evidence that ECT and MDD exert opposing effects on the hippocampal formation with respect to volume and number of neurons. However, there has been a paucity of quantitative data in experimental models of ECT and MDD. Methods Using design‐based stereology, we have measured the effects of a stress‐induced depression model (chronic restraint stress, CRS) and ECS on the morphology of the hippocampus by estimating the volume and total number of neurons in the hilus, CA1, and CA2/3, as well as in the entire hippocampus. Results We find that CRS induces a significant decrease in volume exclusively of the hilus and that ECS (CRS + ECS) blocks this reduction. Furthermore, ECS alone does not change the volume or total number of neurons in the entire hippocampus or any hippocampal subdivision in our rat model.


| INTRODUC TI ON
Major depression (MDD) is a highly debilitating psychiatric disorder with a lifetime prevalence of more than 16% worldwide (World Health Organization, 2017). Although its underlying neurobiology remains elusive, a large body of research implicates dysfunction of the hippocampal formation in the etiology of MDD. The hippocampus has two major anatomic divisions: the hippocampus proper, consisting of three distinct Cornu Ammonis (CA1-3) regions and strata (oriens, pyramidale, and lacunosum moleculare), and the dentate gyrus, which is comprised of the molecular layer, the granule cell layer, and the hilus, also known as CA4. Previous human postmortem and imaging studies of the hippocampal formation report decreased volumes of the entire hippocampus as well as the dentate gyrus in patients suffering from MDD (Cole, Costafreda, McGuffin, & Fu, 2011;Cole et al., 2010;Gbyl & Videbech, 2018;Huang et al., 2013;McKinnon, Yucel, Nazarov, & MacQueen, 2009;Videbech & Ravnkilde, 2004).
Interestingly, these changes are reversed following treatment using antidepressant drugs and electroconvulsive therapy (ECT) (Boldrini et al., 2009(Boldrini et al., , 2013Huang et al., 2013;Nordanskog et al., 2010), indicating plasticity of the hippocampus. A recent longitudinal MRI study of severely depressed patients showed normalization of hippocampal volume following ECT, an increase sustained for four weeks posttreatment (Jorgensen et al., 2015). Additionally, a quantitative study of the total number of neurons in the hippocampus showed fewer granule cells in postmortem brain of unmedicated MDD patients compared to patients treated with selective serotonin reuptake inhibitors (Boldrini et al., 2013). Supporting these findings in humans is the extensive literature in rodent models of depression-like behavior and antidepressant treatment. Accordingly, rodents treated with a range of antidepressant drugs or electroconvulsive stimulation (ECS, the animal model of ECT) show increased hippocampal volume, neurogenesis, angiogenesis, and synaptic differentiation (Chen, Madsen, Wegener, & Nyengaard, 2009Hellsten et al., 2005;Kaae, Chen, Wegener, Madsen, & Nyengaard, 2012;Madsen et al., 2000;Olesen, Wörtwein, Folke, & Pakkenberg, 2017;Olesen, Wortwein, & Pakkenberg, 2015;Santarelli et al., 2003). The molecular mechanisms underlying the beneficial effects of antidepressant medication or ECT/ECS are thought to involve alterations in neurotransmitter systems, growth factors, and neurotropic factors (Lee & Son, 2009;Malberg, Eisch, Nestler, & Duman, 2000;Samuels & Hen, 2011;Segi-Nishida, 2011). In contrast, some previous studies showed "negative" structural effects such as neuronal loss, increased apoptosis, and retraction of dendritic networks in the hippocampus following ECS (Cardoso et al., 2008;Ito et al., 2010;Lukoyanov, Saa, Madeira, & Paula-Barbosa, 2004). Consequently, it remains uncertain how ECS, in interaction with a depressed state, affects the neuronal population and volume of the hippocampus, and its subregions.
The present study is a direct continuation of two previous stereological studies from our laboratory subjecting rats to three weeks of chronic restraint stress (CRS) and/or a clinical relevant schedule of ECS (Olesen et al., 2015(Olesen et al., , 2017. Data from these studies showed that CRS induces a depressive-like phenotype blocked by ECS (CRS + ECS). We also found increased neurogenesis in the subgranular zone (SGZ) that persisted up to 12 months after ECS alone or ECS in conjunction with the CRS model. However, we saw no changes in the volume or the total number of neurons in the granule cell layer following ECS or CRS + ECS, nor did CRS alone influence the hippocampal granule cell layer. In this study, we aimed to quantify the total number of neurons and the volume of the CA1-3 and hilus as well as the volume of the entire hippocampus using design-based stereology in the rat CRS + ECS model.

| Animals and experimental groups
All the animal procedures were performed in accordance with guidelines from the Danish Animal Experimentation Inspectorate and approved by the local ethical committee for experimental animals (License no. 2003/561-644). Forty-eight male Wistar rats (Taconic M&B, Denmark), 7-8 weeks old, weighing 220-250 g were used in the experiments. Pairs of rats subjected to the same treatment were housed together in each cage, with a 12-hr light/dark cycle and ad libitum access to standard rodent chow and tap water. Each treatment group consisted of 12 rats; group one received sham-ECS (control), group two was subjected to chronic restraint stress (CRS), group three was treated with ECS alone, and group four received a combination of CRS and ECS (CRS + ECS). Before experimentation, the rats acclimatized in their home cage for one week. Three rats died during experimentation. There were no significant differences in the mean body weights between the groups at day 0 of the study (Olesen et al., 2015).

| Chronic restraint
CRS can provoke depression-like behavior in rats (Chiba et al., 2012;Watanabe, Gould, & McEwen, 1992). In brief, we administered CRS by confining the rats in a flexible wire mesh restrainer secured in both ends. This limits their freedom to move but allows free breathing and avoids overheating of the animal. In their home cage together with their cage mate, pairs of rats were placed in separate restrainers for 6 hr each day (10 a.m. to 4 p.m.) for 21 consecutive days. During the restraint sessions, the rats were removed from the colony to ensure that rats in the control and ECS groups were not affected by ultrasonic vocalizations from the CRS animals.

| Electroconvulsive stimulation
ECS was administered transcranially using a PSCC-10 pulse stimulator (DCM Electronics, Denmark) as previously described (Olesen et al., 2015), with control animals subjected to the same procedure, but without passage of a current (sham-ECS). A properly administered ECS causes a generalized tonic-clonic seizure lasting 20-30 s (Larsen et al., 2005). Both sham-ECS and ECS treatments were given three days per week (Monday, Wednesday, and Friday, 8-9 a.m.) for 3 weeks, approximating the clinical practice of ECT in humans. To minimize postseizure effects, ECS was performed 1 hr prior to the restraint stress sessions. After sham-ECS or ECS treatments, we returned the rats to their home cage.

| Forced swim test
The procedure of the forced swim test was as described previously (Olesen et al., 2015). In brief, on experimental day 22, the rats were individually placed in a Plexiglas cylinder (50 cm in height and 19 cm diameter) containing 28 cm of water (25 ± 0.5°C) and exposed to a 15-min swim session (pretest). Twenty-four hours later, the rats were re-exposed to a five-min swim session (test).
Both sessions were recorded on video, and the total duration of immobility during the test and the final 5 min of the pretest were recorded. As sham-ECS, CRS, and ECS treatments preceded the behavioral testing, we calculate the difference in % immobility between the test and pretest to correct for variability of the behavioral responsiveness in individual rats (Hageman, Nielsen, Wortwein, Diemer, & Jorgensen, 2009). The rats were considered immobile when floating in an upright position and making only small movements to keep their heads above water (Stogner & Holmes, 2000).

| Tissue processing
Twenty-four hours after the behavioral assessment, the rats were anesthetized using hypnorm/midazolam and then transcranially perfused using PBS followed by 4% paraformaldehyde (PFA). The brains were then carefully removed from the skull, weighed, and postfixated overnight in 4% PFA. The following day brains were transferred to a 30% sucrose (w/v) solution in 0.1 M phosphate buffer for three days at 4°C. Finally, the brains were frozen on dry ice and kept at −80°C until sectioning. Prior to sectioning, the two hemispheres were separated, and one was chosen randomly (with the following hemispheres chosen systematically left/right) before being mounted on an ULTRApro 5000 Cryostat (Bright Instruments, UK). Then, the complete hippocampus was sectioned (Bregma −1.80 to −7.04; Paxinos & Watson, 1986) into 80-µm-thick coronal sections, and finally, every 5th section was subsampled with a random start between sections one and five for final cell counting and structure volume estimation.

| Tissue staining
The sections were mounted on SuperFrost + slides and allowed to air-dry for 40 min at room temperature. Then, the slides were rehydrated in dH 2 O for 10 min, followed by staining in 0.02% cresyl violet (#C5042, Sigma-Aldrich, Switzerland) for 15 min. This process was repeated before the slides were dehydrated in 96% ethanol for 5 min and 99% for 2 × 2 min. Finally, slides were placed in xylene for 2 × 15 min before being coverslipped using Pertex mounting medium (#00801, Histolab, Sweden). Note: The sections were BrdU immunostained prior to cresyl violet staining, results of which we report in our previous study of neurogenesis (Olesen et al., 2015).

| Stereology
We estimated the total number of neurons in the pyramidal cell layer of the CA1 and CA2/3 as well as the total number of neurons in the hilus using the fractionator design and optical disectors (Gundersen, 1986;West, Slomianka, & Gundersen, 1991). In brief, the fractionator design states that the total number of neurons (N) is estimated by multiplying the total number of cells counted (∑Q -) with the reciprocal sampling fractions: where ssf is the section sampling fraction, asf is the area sampling fraction, t Q − is the number-weighted mean section thickness, h is the height of the disector, and q i is the number of neurons  Table 1). We used morphological criteria to identify the neurons, which are characterized by a clearly defined nucleus with colorless cytoplasm and dark nucleolus (for representative micrographs, see Figure 1). The same sections were used to determine the volumes of the CA1 and CA2/3 pyramidal cell layer, the hilus, and the entire hippocampus, all of which were estimated using Cavalieri's estimator (Gundersen & Jensen, 1987). Finally, the unilateral cell numbers and volumes were multiplied by two to obtain the total bilateral cell numbers and volumes. The sub-regions used for cell counting and volume estimation were based on delineations found in the literature (Paxinos & Watson, 1986;West et al., 1991).

| Statistics
One-way ANOVA followed by Tukey's multiple comparisons test was used to analyze the data. Shapiro-Wilk normality test was used to examine the normal distribution and the Brown-Forsythe test together with Bartlett's test to test for equal variance between groups. Values exceeding mean ± 2 × SD were considered as outliers. The coefficient of variance (CV) and the coefficient of error (CE, Gundersen, Jensen, Kiêu, & Nielsen, 1999) were calculated to estimate the precision of the estimates and are shown in Table 2.
The statistics and graphical presentation were performed using GraphPad Prism 7 (version 7.02, GraphPad Software Inc., La Jolla, CA, USA), and data are presented as dot plots with horizontal bars indicating the mean. The level of significance was set to p < 0.05.  decrease in the volume of the hilus in the CRS alone group (p = 0.049, Figure 2a), whereas the volume of the CA1 was statistically significantly increased by 18% in rats treated with CRS + ECS compared to ECS alone (p = 0.026, Figure 2b). We saw no volumetric alterations following ECS alone, nor did we find any changes of the volume of the CA2/3 (F 3,39 = 0.322, p = 0.809, Figure 2c) or the entire hippocampus in the ECS and CRS treatment groups (F 3,37 = 1.903, p = 0.146, Figure 2d).

| D ISCUSS I ON
The primary finding of the present stereological study is a regionspecific decrease in the volume of the hippocampal hilus in depression model rats, which normalized to control level in the group also receiving ECS. This CRS-induced hilar shrinkage did not result in a loss of neurons or other alterations in the entire hippocampus or its subregions. Furthermore, we demonstrate that ECS and CRS + ECS in combination do not significantly affect the volume or the total number of neurons resident in the CA1, CA2/3, hilus, or the complete hippocampus as compared to the control group.
In the exact same rats as used in this study, we have previously shown that CRS significantly increases depression-like behavior in the FST (Percent immobility: Control = 6.3%, CRS = 14.2%, p < 0.05) and that the elevated immobility is normalized following ECS (Percent immobility: CRS + ECS = 7.7%; Olesen et al., 2015).
Comparable results are reported in a number of studies in animal models of depression-like behavior (Kaae et al., 2012;Lee, Jarome, Li, Kim, & Helmstetter, 2009;Luo et al., 2014;Schoenfeld et al., 2017). We suppose that these discrepant results may be related to  (Cao et al., 2017;Cobb et al., 2013). Although we are unable to exclude the occurrence of possible changes in dendritic plasticity of the hippocampal CA1 and CA2/3, our data indicate that any such changes or decreases in the hilar volume must have minor impact on our estimations of the total hippocampal volume.
Interestingly, application of ECS to depression model rats (CRS + ECS) had no significant effect on the volume of the hilus compared to that in controls. This is in line with a previous stereological study demonstrating reduced hippocampal volume in a genetic rat model of depression, which normalized following chronic ECS (Kaae et al., 2012). Also, MRI studies and meta-analyses of se-  (Gbyl & Videbech, 2018;Jorgensen et al., 2015;Nordanskog et al., 2010). Consequently, our results suggest that ECS can normalize a reduced hilar volume possibly by altering various neurotransmitter and hormone systems; for example, glucocorticoid-induced changes in processes which are thought to be involved in major depression (Pariante & Miller, 2001), effects that are opposed using antidepressant drugs (Calfa et al., 2003;Raone et al., 2007), and ECT (Kunugi et al., 2006;Yuuki et al., 2005). In general, this normalization may also apply to other paradigms of rodent depression models and likewise in case of ECT treatment of patients suffering from MDD. This conjecture is further supported by our result of unaltered volume of the hippocampus or its subdivisions following ECS alone. However, a few previous quantitative studies have reported that ECS alone can increase the volume of the entire hippocampus, GCL, and hilus in naïve rats (Chen et al., 2009;Kaae et al., 2012). Whereas our study used a clinically relevant schedule of ECS (three times per week for three weeks), animals in the other studies received ECS for ten consecutive days. Thus, the observed volumetric discrepancies may at least in part be explained by differing experimental conditions. This speculation is further supported by a study by Lukoyanov et al. (2004) showing no changes following five consecutive days of ECS. To our surprise, we observe a difference in the mean volume of the CA1 between ECS-and ECS + CRStreated rat groups. However, neither volume differed significantly The present study shows no significant changes in the total number of neurons after CRS, ECS, and CRS + ECS in any hippocampal regions examined. Similarly, some previous stereological studies report no significant changes in the total number of neurons in the rat CA1 and CA2/3 following chronic imipramine (Chen, Madsen, Wegener, & Nyengaard, 2008) or ECS treatment (Lukoyanov et al., 2004). However, the latter study did report fewer in the hilus in rats with six consecutive ECS sessions, a result that we cannot confirm or reject due to our different experimental design. Even though our ECS procedure differs from several former studies using acute and repeated ECS administration (Ito et al., 2010;Kaae et al., 2012;Malberg et al., 2000), we consider that our design better emulates the clinical practice.
Three weeks of daily CRS did not alter the total number of neurons in any hippocampal subdivision, which stands in contrast to a previous study showing increased apoptosis in the CA3 (Sapolsky, 2000). The authors of that study hypothesized that stress would increase apoptosis in CA2/3 due to increased glucocorticoid or CRS levels. In accordance with that hypothesis, a more recent semiquan-  (Cobb et al., 2013;Cole et al., 2010Cole et al., , 2011McKinnon et al., 2009;Schmaal et al., 2016). The CRS paradigm used in our study better emulates first MDE than recurrent episodes, which may partly describe our finding of unaltered hippocampal volume. We suppose that subjecting animals to "blocks" of repeated CRS might better model the induction of recurrent depressive episodes and thus improve our understanding of the link between hippocampal changes and MDD pathophysiology.
In conclusion, this paper shows that CRS causes a decrease in the volume of the hippocampal hilus which is normalized to control levels after ECS treatment (CRS + ECS). The hilar atrophy In the context of our previous findings in the exact same animals (Olesen et al., 2015(Olesen et al., , 2017 that molecular mechanism other than apoptosis and neuronal loss may mediate CRS-induced depression-like behavior.

ACK N OWLED G M ENTS
The authors wish to thank Susanne Sørensen for providing excellent technical assistance. We also acknowledge professional edit- Doctor Sofus Carl Emil Friis and wife Dorit Friis's Trust.

CO N FLI C T O F I NTE R E S T
Nothing to declare.