Prefrontal Cortical and Behavioral Adaptations to Surgical Delivery Mediated by Metabolic Principles

Abstract

We previously observed an association between mode of delivery and brain mitochondrial mechanisms in pups. We also showed that mitochondrial processes impact adult behavior. However, no experimental data is available to causally connect mode of delivery with cellular processes of neurons in the cerebral cortex and adult behavior. Here we show that surgical delivery of pups alters mitochondrial dynamics and spine synapses of layer 3 pyramidal neurons of the prefrontal cortex compared to the values of mice delivered vaginally. These alterations in ultrastructure seen in adult mice delivered surgically were associated with the development of behavioral phenotypes resembling those characteristic of animal models of psychiatric illness. This included impaired performance in prepulse inhibition as well as hyperlocomotion in the open field and elevated plus maze tests. Knocking out a mitochondria-related gene, UCP-2, blocked cellular and behavioral adaptations induced by surgical delivery. These results highlight a crucial role for brain mitochondrial adaptations in the process of birth to affect neuronal circuitry in support of normal and altered adult behaviors. Further, these findings were supported with neuroimaging data from human neonates delivered vaginally and surgically, suggesting that the murine findings have human clinical relevance.

Introduction

The transition between fetal and extrauterine life is complex and demanding. The fetus must rapidly achieve active breathing, enteral digestion, and independent thermogenesis rather than passively acquiring these needs through the placenta. The process of labor, which involves oxygen deprivation, application of mechanical forces, and hormonal surges, prompts this extreme change in fetal physiology. As of 2014, 32.2% of all births in the United States are deprived of these natural triggers in the form of surgical delivery (SD) (Hamilton et al. 2015). These changes could impact the fetus’ ability to cope with extrauterine life. If so, there may be long-term consequences that persist after the initial transition.

Perinatal life represents an important time in cortical development. Alterations in prefrontal cortical development have been implicated in the pathogenesis of neurodevelopmental diseases (Selemon et al. 1995, 1998). The cortex remains largely undifferentiated at the time of birth and is particularly vulnerable to hypoxic injury due to its high metabolic demands. Hypoxia, normally experienced during natural delivery, could affect the development of this vulnerable area of the brain, leading to lifelong alterations in neuronal architecture and related behavior.

We have previously found that perinatal hypoxia as well as vaginal delivery influence mitochondrial processes, and that mitochondrial processes affect adult behavior (Simon-Areces et al. 2012; Varela et al. 2016). In the present study we explored the putative involvement and role of mitochondrial adaptations in the prefrontal cortex (PFC) of the mouse in relation to vaginal versus surgical way of delivery and show the clinical relevance of these findings using neuroimaging from human neonates.

The PFC has been implicated in many neurodevelopmental disease states, including schizophrenia, attention deficit hyperactivity disorder (ADHD), and autism spectrum disorders (Harrison 1995, 1997; Ross and Pearlson 1996; Raedler et al. 1998; Weinberger and Marenco 2003). During the first few days of life, the PFC in rodents undergoes a critical time of development. When mature, the PFC is composed of distinct layers with differing inputs and functions, but at birth these layers have not yet fully formed and many cells remain undifferentiated (Van Eden and Uylings 1985). Alterations in the perinatal environment could influence the initial organization of this vulnerable region as well as alter the functioning and gene expression of newly differentiating neurons.

Materials and Methods

C-Section

B6 GCE Rosa UCP2^–/– and GCE Rosa UCP-2^+/+ mice were used for all experiments. The generation of UCP2^−/− (on a C57B6 background) mice has been previously described (Hua et al. 2008). Mice were maintained in temperature and humidity controlled rooms, in a 12/12 h light/dark cycle (lights on from 7:00 a.m.–7:00 p.m.). Food and water were provided ad libitum. All procedures were approved by the Institutional Animal Care and Use Committee of Yale University.

Four female mice were placed with 1 male each. We checked for vaginal plugs each morning. Once a plug was seen, the female was removed from the male. This was considered day 1 of pregnancy. Females were grouped according to the day on which they conceived. Once mice that plugged on the same day began to deliver naturally, it was assumed that others were full term. This was on day 19 or 20.

Mice chosen for surgical delivery were euthanized via cervical dislocation. The abdomen was incised using sterile scissors and the 2 horns of the uterus were removed. Each pup was removed from the amniotic sac as quickly as possible using forceps to tear the sac, taking care not to harm the pup. Once the amniotic sac was ruptured the pup was squeezed out of the sac. Time from cervical dislocation to extraction of final pup was under 3 min. Pups were then kept warm under a heat lamp and faces, noses, and bodies were stimulated with a soft towel. Once all mice were removed, they were stimulated until they looked sufficiently pink and were spontaneously breathing. They then were rubbed with bedding from the foster mother’s cage to acquire her scent and placed with the foster mother.

The mice born via vaginal delivery were removed from the mother, rubbed with bedding from the foster mother’s cage, and placed with a foster as soon as they were identified. During the day the cages were checked every 2 h. During the dark cycle they were checked once, so as to not disturb other animals in the facility.

Foster mothers were CD1 females. They were ordered from the Charles River Laboratories and arrived at E19. The experimental pups were exchanged for the biological pups at the time of birth. This exchange happened within 48 h of the CD1 foster mother’s delivery. Litter size was kept constant between 6 and 8 pups.

Electron Microscopy

Six-month old male mice were deeply anesthetized and perfused with 0.9% saline containing heparin, followed by fixative solution. The fixative solution contained 4% paraformaldehyde, 0.1% gluteraldehyde, and 15% picric acid in phosphate buffer. Fixed brains were then stored in the fixative solution overnight at 4 °C. They were washed 3 times with 0.9% saline and replaced in 4 °C until cut into 50 μm sections on the vibratome.

Vibratome sections of the prefrontal cortex were cut into 50 μm slices and stored in sequence. Sections were then processed for electron microscopy. Cells with a visible nucleus were analyzed. A blinded investigator manually traced mitochondria profiles using ImageJ software. All cells were checked twice for consistency. Mitochondria cross-sectional area was calculated. Mitochondria density was estimated by dividing the number of mitochondria profiles by the cytosolic or cellular areas. Mitochondria coverage was estimated by dividing the total area of mitochondria by the cytosolic or cellular areas as specified. Differences in mitochondria density and coverage were tested using t-test. P ≤ 0.05 was considered statistically significant.

Spine synapses were counted by a blind investigator according to the rules of the dissector technique, and the volumetric density of spine synapses (synapse/μm³) was determined. Protocol followed as previously described in Diano et al. (2006).

Behavior

The open field test apparatus was a square, polyurethane arena (36.5 cm × 36.5 cm × 30 cm, Plexiglas). The animal was placed in the lower left corner of the apparatus. Locomotion speed, distance traveled, entries into the central zone, and time spent in contact with the outer walls, were recorded for 30 min. Behavioral testing took place from 1000 to 1400 h (i.e., in the light phase of the light–dark cycle). The apparatus was cleaned with 10% ethanol after each animal exposure. ANY-Maze Software™ (Stoelting Company, Wood Dale, IL) was used to record and analyze behavioral data.

Spatial memory was assessed using the two-trial Y-maze task. A single Y-maze was made of black Plexiglas and consisted of 3 arms with an angle of 120° between each of the two arms. Each arm was 8 cm × 30 cm × 15 cm (width × length × height). The 3 arms were randomly designated: start arm, in which the mouse started to explore (always open), novel arm, in which the mouse started to explore (always open), novel arm, which was blocked during the first trial but open during the second trial.

The maze was placed on a flat surface within the behavioral testing room. Proximal visual cues (pictures within the arms of the apparatus) and distal visual cues (the configuration of the room, curtain, wall art) remained constant throughout testing. The floor of the maze was covered with white chip bedding. Between each trial the apparatus was cleaned with 10% ethanol and new bedding was added. Behavioral testing took place from 10:00 to 14:00 h (i.e., in the light phase of the light–dark cycle).

The Y-maze test consisted of 2 trials separated by an inter-trial interval (ITI) of 90 min to assess spatial memory. The first trial had a 5-min duration and allowed the mouse to freely explore only 2 arms (start arm and other arm) while the third arm was blocked. After a 90 min ITI, the second trial also of 5 min duration was conducted during which all 3 arms were accessible and novelty vs. familiarity was compared in all 3 arms. ANY-Maze Software™ (Stoelting Company, Wood Dale, IL) was used to record and analyze behavioral data.

The elevated put maze consisted of a 4 16 cm × 5 cm arm apparatus elevated 50 cm above the floor. Two of the arms are enclosed by 12 cm walls and the other 2 arms are open. The maze is illuminated equally in all 4 arms. Mice were placed in a closed arm and allowed to explore for 5 min. The maze was cleaned with 10% ethanol between trials. Behavior was recorded by overhead camera equipment and Any-Maze™ Software.

Prepulse Inhibition

PPI experiments were performed using the SOF-825 Startle Reflex System (MED Associates). Mice were acclimated to a startle chamber for 5 min with background noise (62 dB) at least one day before the day of PPI analysis. Mice were presented with 5 min of background noise, followed by 3 consecutive blocks of stimuli. Block 1 consisted of 10 trials of a 38 ms startle stimuli (120 dB) separated by a 10–20 second interval to determine baseline startle response. Block 2 preceded by an 18 ms prepulse of 67, 70, 73, or 76 dB 100 ms prior to the startle (4 trials for each prepulse level); or an 18 ms prepulse stimulus only (1 for each prepulse level). Block 3 was a repetition of Block 1 to ensure mice did not habituate to startle stimulus. Prepulse inhibition for each dB level was calculated as 1–[(avg. of prepulse+startle)/(avg. startle only)]. Background only and prepulse only trials were observed to ensure lack of response with no startle stimulus.

Human participants

This study was approved by the Yale University Human Investigation Committee. All participants were born preterm (≤ 28 weeks gestation) and were scanned without sedation using a feed-and-wrap protocol in a 3 T Siemens Verio Clinical system as part of routine clinical care (a 3D T2 weighted image, slice thickness = 0.906 mm, matrix size = 256×256, FoV = 256 mm, TR = 4000 ms, TE = 263 ms, Flip Angle = 120°, Bandwidth = 299 Hz/pixel).

Participants were excluded for all major brain injuries with the exception of low grade (1 or 2) intraventricular hemorrhage (IVH) or cerebellar hemorrhage. Table 1 displays the participants’ demographic information. Overall, 14 participants were born via vaginal birth and 39 participants were born via C-section. There were no significant differences in sex, postmenstrual age at birth, postmenstrual age at scan, and proportion of participants with cerebellar hemorrhage. The vagina birth group had a significantly greater number of participants with low grade IVH.

Table 1.

Participants characteristics

	Neonates born via vaginal birth (n = 14)	Neonates born via C-section (n = 39)	P-value
Sex (M/F)	9/5	24/15	0.86
Postmenstrual age at birth (weeks)	25.74 ± 1.32	26.33 ± 1.02	0.10
Postmenstrual age at scan (weeks)	36.89 ± 2.16	38.52 ± 3.02	0.07
Low grade IVH (Y/N)	9/5	9/30	0.008
Cerebellar hemorrhage (Y/N)	3/11	4/39	0.35

Tensor-based morphometry (TBM)

For TBM analysis, anatomical images were linearly aligned to an infant anatomical template created from an independent (Spann et al. 2018) using a 12 parameter affine registration by maximizing the normalized mutual information between images. Next, anatomical images were non-linearly registered to an evolving group average template in an iterative fashion using a previously validated algorithm (Scheinost et al. 2017). This algorithm iterates between estimating a local transformation to align individual brains to a group average template and creating a new group average template based on the previous transformations. The local transformation was modeled using a free-form deformation (FFD) parameterized by cubic B-splines. This transformation deforms an object by manipulating an underlying mesh of control points. The deformation for voxels in between control points was interpolated using B-splines to form a continuous deformation field. Positions of control points were optimized using a conjugate gradient descent to maximize the normalized mutual information between the template and individual brains. After each iteration, the quality of the local transformation was improved by increasing the number of control points and decreasing the spacing between control points to capture a more precise alignment. A total of 4 iterations were performed with decreasing control point spacings of 15 mm, 10 mm, 5a mm, and 2.5 mm. To help prevent local minimums during optimization, a multi-resolution approach was used with 3 resolution levels at each iteration. The determinant of the Jacobian of the deformation field was used to quantify local volume differences between the registered images and the template. This metric provided an estimate of voxel-wise volume changes for all transformed images with respect to the group averaged template and was used for further analysis.

Statistical Analysis

Data were analyzed using SigmaPlot. Data with more than 2 groups were analyzed using a One-Way ANOVA. Post hoc testing between groups was performed using the Fisher LSD test with a P-value of < 0.05 being considered significant. Groups of 2 were analyzed using a student’s t-test. Again, a P-value of <0.05 was considered significant.

TBM data were analyzed with voxel-wise general linear modeling. Postmenstrual age at birth, IVH status, cerebellar hemorrhage status, and sex were included as covariates. Imaging results are shown at a cluster-level threshold of P < 0.05 family-wise error (FWE) correction as determined by AFNI’s 3dClustSim program (version 16.0.09) using a cluster-forming threshold of P = 0.001, 10 000 iterations, a PFC gray-matter mask, and a smoothness estimated from the residuals using 3dFWHMx.

Results

To determine whether mode of delivery has a long-term impact on the neurons located in the PFC, we analyzed pyramidal neurons in layer 3 of the PFC at 5–6 months of life. Due to our previous findings that perinatal hypoxia exposure resulted in long-term mitochondrial adaptations, we looked to see if there were any differences in the mitochondrial parameters found in mice born by surgical delivery when compared to vaginal delivery. We found that WT mice born via SD had a significantly lower mitochondrial cytoplasmic density (M = 0.451 ± 0.0285/μm²) than the VB group (M = 0.857 ± 0.169/μm²), t (19) = 2.483, P = 0.023 (Fig. 1). This demonstrates that mode of delivery can impact the development of the neuron and cause mitochondrial adaptations that persist into adult life.

Figure 1.

Mitochondrial and synaptic density in layer 3 of the PFC of UCP2^+/+ mice delivered either vaginally or surgically. (A) Representative electron micrograph of pyramidal cell bodies of layer 3 of the prefrontal cortex. Asterisks represent mitochondria. (B) Cytoplasmic mitochondrial density of UCP2^+/+ VB vs SD mice in layer 3 of the PFC (n = 10, n = 11, respectively). (C) Representative electron micrograph of spine synapses in layer 3 of the PFC in UCP2^+/+ mice delivered either vaginally or surgically. Arrowheads represent synapses. (D) Synaptic density of UCP2^+/+ VB vs SD mice in layer 3 of the PFC (n = 30, n = 30, respectively). Significance with respect to VB (* = P < 0.05, ** = P < 0.01). Error bars represent SEM.

Perinatal life represents a crucial time for proper synapse formation in the PFC. We analyzed specifically layer III of the PFC, which contains excitatory glutamatergic pyramidal neurons (Fonnum 1984). These neurons receive corticocortical input and integrate a large magnitude of sensory input (Fuster 1985, 1997; Pandya and Yeterian, 1996; Rolls 1998; Cavada et al. 2000; Petrides 2000). They are tonically active and thought to be responsible for gating pertinent information relevant for higher cognitive functioning (Fuster and Alexander 1971; Kubota and Niki 1971; Fuster 1973; Funahashi et al. 1989; Miller et al. 1996; Leung et al. 2002). Previous studies have demonstrated that alterations in this region of the brain have been associated with mental illness and impairments in mood and executive function. In diseases such as schizophrenia for example, spine density has been found to be lower (Hao et al. 2009) whereas some studies have demonstrated hyperconnectivity in this region in models of autism spectrum disorders (Testa-Silva et al. 2012).

We found that surgically delivered mice had elevated spine synapse density in this layer of the PFC (Fig. 1). We found that WT mice delivered surgically (16.200 ± 1.232/6 405 μm³) had a significantly higher synapse density than the VB group (M = 12.233 ± 1.154/6 405 μm³), t(58) = −2.349, P = 0.022. Changes we observed here may represent altered connectivity to other areas of the brain, and therefore suggest that integration of information could be perturbed.

In order to evaluate whether the changes seen in the mitochondria and synapses were associated with changes in behavioral phenotypes, we performed multiple behavior tests at various ages. At 9 weeks of age, male mice were tested in an open field for 30 min. We measured total distance traveled, average speed, and time spent in the corners of the maze. We use the activity in open field primarily to measure the animal’s baseline level of locomotion.

We found a significant effect of mode of delivery in wild type mice on the distance traveled in the open field t (37) = −2.789, P = 0.008 (Fig. 2). The WT VB mice (M = 43.240 ± 3.673 m) were found to travel a decreased distance compared to the SD group (M = 57.974 ± 3.799 m).

Figure 2.

UCP2+/+ mice delivered surgically and vaginally studied in the open field, elevated plus maze, and prepulse inhibition. (A) Total distance traveled by the UCP2+/+ mice delivered vaginally and surgically in the open field (n = 20, n = 19). (B) Average speed of the UCP2+/+ mice measured during the test (n = 20, n = 19, respectively). (C) Total time spent in the corner of the open field arena in the UCP2+/+ mice (n = 20, n = 19, respectively). (D) Total time UCP2+/+ mice spent in the open arm of the elevated plus maze (n = 16, n = 17, respectively). (E) Total distance UCP2+/+ mice traveled in the open arm of the elevated plus maze (n = 16, n = 17, respectively). (F) Average speed of the UCP2+/+ mice in the elevated plus maze (n = 16, n = 17, respectively). (G) Prepulse inhibition performed on UCP2+/+ mice delivered vaginally and surgically and fostered (n = 11, n = 5, respectively). Significance with respect to VB (* = P < 0.05, ** = P < 0.01). Error bars represent SEM.

In addition to distance traveled in the open field, we analyzed their average speed in the test. There was a significant effect of mode of delivery in wild type mice on the speed in the open field t (37) = −2 759, P = 0.009. In particular, the VB group (M = 0.0241 ± 0.00205 m/s) differed significantly from the SD group (M = 0.0322 ± 0.00209 m/s). It appears that surgical delivery leads to an increase in average speed in the open field in the wild type mice. It is of importance to note that hyperlocomotion and hyperactivity have previously been used as a marker for schizophrenia (Amann et al. 2010) and is also characteristic of models of ADHD (Schubert et al. 2015).

We also assessed the total time mice spent in the corner of the open field arena. Significant differences were found between the wild type mice, t(37) = 3.296, P = 0.002. The SD wild type mice (M = 214.342 ± 54.738 s) spent less time in the corner than the VB group (M = 434.225 ± 39.069 s).

At 10 weeks of age the mice were tested in a Y maze in order to evaluate their spatial orientation and novelty recognition. While this behavioral test is not specific for PFC functioning (O’Keefe and Dostrovsky 1971), we wanted to see if surgical delivery had a global impact on the brain causing indiscriminate changes in behavior. Time spent in the novel arm, entries into the novel arm, and latency to enter the novel arm were investigated and no differences were found based on mode of delivery. While locomotion and anxious behavior is affected by mode of delivery, hippocampus dependent performance is not affected. This would indicate that SD does not indiscriminately impact pups’ behavior but rather acts in a more specific manner. This correlates with previous data showing that ablation of the PFC results in hyperlocomotion, decreased PPI, but preserved novelty recognition (Yee 2000).

At 15 weeks of age, anxiety levels of mice and exploratory behavior were investigated using an elevated plus maze. This behavioral test is performed to evaluate their willingness to enter an open, elevated arm. Mice that spend more time on the elevated platform are thought to be less anxious and more explorative. We found that in the wild type group, mode of delivery did significantly affect the time spent in the open arm t (31) = −3.481, P = 0.002 (Fig. 2). In particular, the WT SD group (M = 54.650 ± 5.883 s) spent significantly more time in the open arm than the VB group (M = 29.175 ± 4.354 s).

Total distance traveled and average speed were both measured in the elevated plus maze. Mode of delivery did affect total distance traveled, similar to previous findings in the open field t (31) = −2.468, P = 0.019. Particularly the VB group (M = 6.833 ± 0.606 m) traveled a shorter distance than the SD group (M = 9.130 ± 0.700 m). Average speed in the elevated plus maze mimicked our previous findings in the open field. There was a significant difference between mode of delivery in the wild type animals t (31) = −2.674, P = 0.012. In the wild type animals, the VB group (M = 0.0228 ± 0.00199 m/s) traveled more slowly than the SD (M = 0.0318 ± 0.00268 m/s). Again, this demonstrates that in multiple arenas, the mice born via SD have an increase in locomotion. Hyperlocomotion together with apparent lower level of anxiety was also phenotypes associated with NMDAR hypomorph animals, an animal model with relevance to schizophrenia (Mohn et al. 1999; Barkus et al. 2012).

Prepulse inhibition is a test of auditory gating in which mice learn to dampen their response to a startle by being warned with a preceding tone. Mice that do not inhibit their startle response when warned are believed to have impairments in PFC function (Yee 2000). We tested mice delivered either by vaginal birth or surgery in the prepulse apparatus at 9–10 weeks of age. We found that the surgically delivered mice decreased their startle by only 36.5 ± 3.34% compared to the VB group which decreased their startle by 47.5 ± 2.30%. t(14) = 2.698, P = 0.017. These observations are in line with our data showing altered cellular and synaptic integrity PFC pyramidal cells.

Because a difference between vaginal and surgical birth was the effect on mitochondria of layer III PFC neurons, we sought to determine whether mechanisms associated with mitochondrial adaptations could play a role in the emergence of the observed cellular and behavioral alterations. Mitochondrial uncoupling protein 2 (UCP2) is an inner mitochondria membrane bound protein that was previously associated with mitochondrial adaptations and spine synapse formations and synaptic plasticity (Fagel et al. 2006; Tsai et al. 2013; Varela et al. 2016).

Ablation of UCP2 was preventive in perinatal hypoxia-induced changes in cellular and spine synapse integrity of principle cells in the cortex (Varela et al. 2016). To interrogate whether lack of UCP2 may affect cellular and behavioral attributes of surgical versus vaginal delivery, we evaluated the same parameters as described above regarding wild type mice in UCP2 knock out (UCP2^–/–) mice.

At 6 months of age we analyzed the ultrastructure of layer III PFC cells of mice delivered vaginally or surgically. We did not find a significant difference in mitochondrial size or density (Fig. 3). This suggests that mode of delivery acts in a UCP2 dependent manner to alter lifelong neuronal mitochondrial dynamics. Knocking down UCP2 effectively abolished altered mitochondrial adaptations induced by surgical delivery.

Figure 3.

Mitochondrial and synaptic density in layer 3 of the PFC of UCP2^–/– mice delivered either vaginally or surgically. (A) Representative electron micrograph of cell bodies of layer 3 of the prefrontal cortex. Red asterisks represent mitochondria. (B) Cytoplasmic mitochondrial density of UCP2^–/– VB vs SD mice in layer 3 of the PFC (n = 9, n = 9, respectively). (C) Representative electron micrograph of spine synapses in layer 3 of the PFC in UCP2 ^–/– mice delivered either vaginally or surgically. Arrowheads represent synapses. (D) Synaptic density of UCP2^–/– VB vs SD mice in layer 3 of the PFC (n = 30, n = 30, respectively). Significance with respect to VB (* = P < 0.05, ** = P < 0.01). Error bars represent SEM.

We next analyzed synaptic density in layer III of PFC in UCP2^–/– animals delivered vaginally and surgically. We found no significant difference in synaptic density (Fig. 3). This observation is consistent with the findings in a recent study showing that perinatal hypoxia-induced alterations in mitochondrial and synaptic density can be prevented by ablation of UCP2 (Varela et al. 2016). Our results also confirm earlier studies indicating that synaptic changes can be regulated by mitochondrial functions (Dietrich et al. 2008).

To explore whether the lack of changes in ultrastructure observed in the UCP2^–/– mice had behavioral correlates, we applied the behavioral tests described above on UCP2^–/– mice. We observed mice in the open field at 9 weeks of age and analyzed speed, distance traveled, and time spent in the corners of the apparatus. Interestingly we could not find any significant differences in any of the aforementioned parameters (Fig. 4). Mode of delivery seems to act in a UCP2 dependent manner to change locomotion. While SD was associated with hyperlocomotion in the WT animals, none of these differences were seen in the UCP2^–/– animals.

Figure 4.

UCP2^–/– mice delivered surgically and vaginally studied in the open field, elevated plus maze, and prepulse inhibition. (A) Total distance traveled by the UCP2^–/– mice delivered vaginally and surgically in the open field (n = 7, n = 15). (B) Average speed of the UCP2^–/– mice measured during the test (n = 7, n = 15). (C) Total time spent in the corner of the open field arena in the UCP2^–/– mice (n = 7, n = 15). (D) Total time UCP2^–/– mice spent in the open arm of the elevated plus maze (n = 4, n = 11, respectively). (E) Total distance UCP2^–/– mice traveled in the open arm of the elevated plus maze (n = 4, n = 11, respectively). (F) Average speed of the UCP2^–/– mice in the elevated plus maze (n = 4, n = 11, respectively). (G) Prepulse inhibition performed on UCP2+/+ and UCP2^–/– mice born vaginally and not fostered (n = 8, n = 9, respectively). Significance with respect to VB (* = P < 0.05, ** = P < 0.01). Error bars represent SEM.

UCP2 ^–/– mice were also studied in a Y maze apparatus at 10 weeks of age. No significant differences were found in the animals’ ability to recognize the novel arm. This parameter was not affected in the WT mice as well.

We placed the UCP2^–/– mice in an elevated plus maze. While SD was associated with increased exploration of the open arm in the WT animals, no significant difference was found between the UCP2^–/– animals. In terms of distance traveled and average speed, there was no significance found between the VB or SD deliveries in the UCP2^–/– animals as seen before in the open field.

Next, we studied prepulse inhibition in wild type and UCP2^–/– mice delivered vaginally. We found that the UCP2^–/– mice actually performed better, meaning that they had a decreased startle response when paired with the prepulse than the WT mice at 18 weeks (Fig. 4) (t (15) = −2.227, P = 0.042). The UCP2^–/– mice reduced their startle response by 41.3 ± 5.59% whereas the WT only decreased their response by 24.4 ± 5.06%. This indicates that alterations in UCP2 expression, such as during surgical delivery, could negatively impact performance and functioning of the PFC.

Finally, we investigated whether similar structural differences in the PFC would be observed in human neonates using anatomical MRI. Evaluating 53 prematurely born neonates of ≤28 weeks gestation and no major brain injury with T2-weight MRI collected as part of routine care at term equivalent age, we performed Tensor-based morphometry (TBM) to estimate local brain volumes on the voxel-level for the PFC. When compared to neonates born vaginally, neonates born via C-section exhibited significantly (P < 0.05 corrected) greater gray matter volume in the medial PFC (Fig. 5), consistent with the elevated spine synapse density observed in our wild type surgically delivered mice.

Figure 5.

Comparison of gray matter volume in the PFC for neonates born via C-section (n = 39) or vaginal birth (n = 14). Neonates born via C-section had significantly greater volume in the medial PFC compared to neonates born vaginally. All results shown at P < 0.05 corrected for multiple comparisons.

Discussion

Taken together our observations revealed that surgical delivery underlies alterations in structure and function of the prefrontal cortex and that these events are mediated, at least in part, by mitochondrial processes associated with UCP2. Surgically delivered mice demonstrated behaviors characteristic of animal models of psychiatric conditions including hyperlocomotion, lower anxiety and impaired performance of animals in auditory gating test and these changes were prevented by knocking out UCP2. Previsously we have found that the presence of UCP2 is important determinant of cortical oscillations (Hermes et al. 2016), which are likely involved in the phenotypes we describe in the current paper. Our findings in mice have significant clinical relevance as these results are consistent with changes in the PFC we detected in MRI’s of human neonates at term equivalent age with and without surgical delivery.

These findings are with potential clinical relevance in today’s era as cesarean sections are the most common surgical procedure. In the United Sates, the rate of cesarean section increased steadily over a decade from 21% in 1998 to 32% in 2007, accounting for 1.4 million births (Menacker and Hamilton 2010). This trend was seen for women in all age, racial and ethnic groups, and for infants of all gestational ages during this time period. It is even found worldwide in countries with lower gross domestic product (GDP) including Brazil, Republic of Korea, and Iran, which reported cesarean section delivery rates of 45.9%, 37.7%, and 41.9%, respectively in 2008 (Gibbons et al. 2012).

The extent to which women’s request for cesarean for non-medical reasons has contributed to these rates is a contentious issue (Goer 2001). Existing evidence from both retrospective and prospective studies is limited, utilizing different definitions of’maternal request’, and reporting rates of between 1% and 48% in public sector healthcare systems, and 60% in the private sector (Thomas et al. 2000; Sakala et al. 2002). In addition, a recent Cochrane report speculated that physician convenience, preferences and financial incentives contribute to high rates in the private sector. As women consider the benefits of SD which include reductions in perineal pain, uterovaginal prolapse and incontinence (Sultan et al. 1994; Sultan and Stanton 1996; Farrell et al. 2001) avoidance of pain during labor, and avoidance of emergency cesarean sections, it is crucial that they are provided with accurate and complete information regarding the consequences of C sections to empower them to make informed decisions. Current counseling includes discussion of reported maternal adverse effects of such as risk of infection, difficulties breastfeeding (Francome and Savage 1993), abnormal placentation, or uterine scar rupture in subsequent pregnancies (Hemminki and Merilainen 1996; Wen et al. 2004). Risks to the infant identified higher admissions to neonatal intensive care units (Treffers and Pel 1993), iatrogenic prematurity (Wagner 2000), fetal laceration (Smith et al. 1997), increased neonatal respiratory problems (Madar et al. 1999), and an increase in special educational needs in later life linked to the timing of the cesarean section (MacKay et al. 2010; Kapellou 2011). A paramount contribution to this discussion between providers and patients would include a review of the unclear impact cesarean section has on long-term neurodevelopment.

Vaginal delivery is a process which has been conserved over time throughout the mammalian species. While hypoxia is typically thought of as a pathologic condition, there is a physiologic level of hypoxia which mammals have evolved to tolerate during labor. This stress potentially acts as a trigger for the neonate to adapt to life outside of the uterus. When the brain experiences episodes of pathologic hypoxia such as during a seizure or a stroke, UCP-2 is known to be induced and have a neuroprotective role (Fagel et al. 2006; Tsai et al. 2013; Varela et al. 2016). Similarly, we believe it is involved with neuroprotection during delivery as well as developmental programming due to these early events’ impact on later neurodevelopment of the individual. We show that this impact on neurodevelopment translates to differences in behavioral phenotypes into adulthood.

We found that late into adulthood, mice delivered surgically show decreased mitochondrial density in layer III of the PFC. Mitochondria are dynamic organelles and they can change their shape, number, and position in response to different stressors or needs of the cell (Scarpulla 2002). Mitochondrial density and number have been correlated with the baseline activity of the neuron. For this reason, we believe our mitochondrial findings can provide insight into the metabolic environment of the cell. Maintenance of metabolism and energy production is crucial to meet the neuron’s synaptic needs (Jeanneteau and Arango-Lievano 2016). The neuron’s function is dependent on its ability to properly form neural networks and communicate with surrounding cells. Part of this communication is in the form of an oscillatory potential that synchronizes with surrounding cells. Therefore, the neuron is far from stagnant and maintenance of this constantly changing potential places a large burden on the mitochondria. Additionally, execution of the action potential is energetically demanding. These impressive energetic demands highlight the neuron’s dependence on mitochondrial mechanisms. The basic function of the neurons could be compromised making our findings clinically relevant.

In addition to the differences in mitochondria, we also see that surgical delivery results in differences in synapse formation in layer III of the PFC. Specifically, we find that those born via SD have more synapses compared to those born vaginally. While it has been previously demonstrated that pyramidal cells in layer III of the PFC of schizophrenic subjects have a decreased spine density, models of autistic mice, which also have PFC dependent behavioral abnormalities have been shown to have hyperconnectivity associated with low speed action potential transmission (Hao et al. 2009; Testa-Silva et al. 2012). One proposed mechanism of the development of schizophrenia and austism involves synaptic pruning, a process that occurs during adolescence (Zhan et al. 2014). Many synapses form in early life but are later eliminated depending on which connections are reinforced by external inputs and learning. Aberrations in this process lead to improper organization of this region resulting in neurodevelopmental disease. The alteration in synaptic density we observe may be a sign of abnormal initial formation of synapses or improper pruning of the synapses later in life. While mice delivered surgically may not be classically “schizophrenic” or “autistic”, they demonstrate dysfunction in an area of the brain that has previously been implicated in psychiatric and neurodevelopmental disorders.

However, when UCP-2 is knocked out, a maladaptive response to surgical delivery is eliminated. This reveals that UCP-2 has a causative role in the changes we see. Additionally, it indicates that the abnormal delivery itself is not harmful to the fetus but rather the way in which the brain reacts to it, meaning an individual’s genetic makeup determines its vulnerability to unnatural delivery. We believe these findings are clinically relevant and potentially related to the association between mode of delivery and psychiatric disease. This would support the epidemiologic studies which correlate obstetric complications to the development of psychologic disease. While many individuals may be exposed to perinatal stressors which predispose them to psychiatric disease, we show here how one’s genetic background could prevent certain events from impacting development.

Our observations regarding the long-term impact of surgical delivery in mice and humans are different than that of a recent study by Chiesa et al. (2018). In that study, the authors concluded that there were no long-term consequences of cesarean section on specific parameters of mouse behavior and neurobiology. Indeed these mice did not show differences in open field or social testing, nor did they express differences in glutamatergic and GABAergic activity. However, they did find transient changes in morphology of pyramidal cells in CA3 following surgical delivery. Additionally there were significant differences in vocalization patterns and grooming behaviors which have been used as a metric for obsessive compulsive like behaviors in rodents. These findings echo a point made by Chiesa et al., which is that there are many contradictory results, including those of epidemiological studies, regarding the long-term consequences of surgical delivery on phentotypes of the offspring. The contradictory nature of these studies is evidence of the complexity of this topic. Further work is needed to uncover the intricacies of the impact surgical delivery has on health and development of the offspring.

Funding

This study was supported by NIH grants AG052005, AG052986, AG051459, DK111178 and NKFI-126998 from the Hungarian National Research, Development and Innovation Office (T.L.H).