Gestational stress decreases postpartum mitochondrial respiration in the prefrontal cortex of female rats

Abstract

Postpartum depression (PPD) is a major psychiatric complication of childbirth, affecting up to 20% of mothers, yet remains understudied. Mitochondria, dynamic organelles crucial for cell homeostasis and energy production, share links with many of the proposed mechanisms underlying PPD pathology. Brain mitochondrial function is affected by stress, a major risk factor for development of PPD, and is linked to anxiety-like and social behaviors. Considering the importance of mitochondria in regulating brain function and behavior, we hypothesized that mitochondrial dysfunction is associated with behavioral alterations in a chronic stress-induced rat model of PPD. Using a validated and translationally relevant chronic mild unpredictable stress paradigm during late gestation, we induced PPD-relevant behaviors in adult postpartum Wistar rats. In the mid-postpartum, we measured mitochondrial function in the prefrontal cortex (PFC) and nucleus accumbens (NAc) using high-resolution respirometry. We then measured protein expression of mitochondrial complex proteins and 4-hydroxynonenal (a marker of oxidative stress), and Th1/Th2 cytokine levels in PFC and plasma. We report novel findings that gestational stress decreased mitochondrial function in the PFC, but not the NAc of postpartum dams. However, in groups controlling for the effects of either stress or parity alone, no differences in mitochondrial respiration measured in either brain regions were observed compared to nulliparous controls. This decrease in PFC mitochondrial function in stressed dams was accompanied by negative behavioral consequences in the postpartum, complex-I specific deficits in protein expression, and increased Tumor Necrosis Factor alpha cytokine levels in plasma and PFC. Overall, we report an association between PFC mitochondrial respiration, PPD-relevant behaviors, and inflammation following gestational stress, highlighting a potential role for mitochondrial function in postpartum health.

Keywords: Chronic unpredictable stress; Mitochondria; Postpartum; Prefrontal cortex; Pregnancy; Susceptibility.

Graphical abstract

Fig. 1

Experimental Timeline. Nulliparous and primiparous female rats were weighed and handled upon arrival at the vivarium. Females were weight-matched (within parity groups) and assigned to stress or non-stress conditions. Chronic mild unpredictable stress (CMUS) began on the day corresponding to gestational day (GD) 10 in primiparous rats and continued for 10 days through GD 19. Following parturition on GD 21/postpartum day (PD) 0, a series of behavioral tests were performed during the postpartum period. The sucrose preference test (SPT) was performed from PD 0–4, the elevated plus maze (EPM) was performed on PD 4, maternal care was assessed during the light cycle and at the onset of the dark cycle on PD 2, 3, 5, and 6, and the one-session forced swim test (FST) was performed on PD 10 or 11. All rats that underwent behavior were euthanized on the days corresponding to PD 11 or 12 and mitochondrial respiration was measured in the prefrontal cortex and nucleus accumbens. Other molecular analyses (protein and cytokine measurements) are representative of this same euthanasia timepoint.

Fig. 2

CMUS successfully induced PPD-relevant behaviors in gestationally stressed dams. (A) P + S dams had significantly lower body weight gain during GD 10–19 (stress period) than P dams (p = 0.04), while S females did not significantly differ in body weight gain from C females, as evidenced by significant main effects of both stress (p = 0.006) and parity (p < 0.0001), but no significant interaction between stress and parity (p = 0.4). n = 8–11/group. (B) There was a significant main effect of parity on sucrose preference (p = 0.004), and trends for a main effect of stress (p = 0.08), where P + S dams exhibited significantly lower preference for a 1% sucrose solution compared to primiparous controls (p = 0.04). There was no statistically significant interaction between stress and parity in 1% sucrose preference (p = 0.1). n = 5–6/group. (C) There was no significant interaction between stress and parity on percent time spent in open arms (p = 0.3), and no main effects of either parity (p = 0.5) or stress (0.98). n = 9–11/group. P + S dams had, on average, significantly reduced bouts of (D) nursing (p = 0.01) and (E) licking/grooming (p = 0.009) compared to P dams. n = 9/group. (F) There was a significant main effect of stress on both nulliparous and primiparous females in latency to immobility in the forced swim test (p < 0.0001), but no significant interaction (p = 0.3) or main effect of parity (p = 0.9). Both S and P + S groups had significantly shorter latencies to become immobile compared to respective controls (C vs. S: p = 0.0001, P vs. P + S: p = 0.02). n = 9–10/group. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Fig. 3

Gestational stress decreases mitochondrial respiration in the prefrontal cortex but not nucleus accumbens. (A) In prefrontal cortex (PFC), there was a significant interaction and main effect of stress, with no main effect of parity, on complex I (CI) (interaction: p = 0.01, stress: p = 0.02, parity: p = 0.15), complex I + II coupled (CI + CII) (interaction: p = 0.02, stress: p = 0.005, parity: p = 0.2), and maximal (ETS) (interaction: p = 0.02, stress: p = 0.007, parity: p = 0.14) respiration, where post hoc analyses demonstrated significantly decreased respiration in all measures in P + S dams compared to P dams ((CI) P vs. P + S: p = 0.004, (CI + CII) P vs. P + S: p = 0.002, (ETS) P vs. P + S: p = 0.003). n = 8–11/group. (B) In PFC, there was a trend for an interaction of stress and parity on complex II respiration (p = 0.06) and a significant main effect of stress (p = 0.04), but no main effect of parity (p = 0.21), although post hoc analyses revealed lower respiration in P + S compared to P dams (p = 0.02). n = 8–11/group. (C) There were no significant differences observed between groups in PFC mitochondrial content, as measured by CI + II/ETS-linked capacity (interaction: p = 0.62, stress: p = 0.63, parity: p = 0.72). n = 8–11/group. (D) In nucleus accumbens (NAc), there were no significant interactions or main effects of stress or parity on CI, CI + CII, or ETS respiration (see Supplementary Table 1 for statistical details). n = 9–11/group. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Fig. 4

Gestational stress decreases mitochondrial complex I protein but has no effect on other complexes or 4-HNE protein expression in prefrontal cortex. (A) Representative Western blots of complex protein. (B) In prefrontal cortex, we found a significant interaction between stress and parity on complex I protein expression (p = 0.006), with no significant main effects of parity or stress, such that P dams exhibited a trend for increased protein expression compared to C females (C vs. P: p = 0.07), and P + S dams had significantly reduced protein expression compared to P dams (P vs. P + S: p = 0.01). No significant differences were observed between groups in protein expression of (C) complex II, (D) complex III, (E) complex IV, or (F) complex V. Similarly, (G) the representative 4-HNE western blot image demonstrates (H) there were no significant effects of stress, parity, or interaction on 4-HNE (see Supplementary Table 1 for statistical details). n = 8–11/group. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed, 4-HNE = 4-hydroxynonenal. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.

Fig. 5

Gestational stress decreases tumor necrosis factor alpha (TNF-α) levels in PFC and plasma. (A) We found a significant main effect of stress on TNF-α in the prefrontal cortex (PFC), (p = 0.05), but no significant main effect of parity (p = 0.65), nor interaction (p = 0.86). Post hoc analyses demonstrated no significant differences between P and P + S dams. Note that order of groups on graph are different from other figures as to highlight the significant main effect of stress. n = 8–11/group. (B) In P (pink dots) and P + S (red dots) dams, CI + CII coupled respiration in the PFC was predictive of TNF-α in the PFC, where higher levels of the cytokine correlated with lower oxygen consumption. (Simple Linear Regression, n = 16). (C) In plasma, there was a significant main effect of parity on TNF-α (p = 0.02), while there was no significant interaction (p = 0.19) or main effect of stress (p = 0.8). However, post hoc analyses revealed no significant differences between groups. n = 7–11/group. (D) We observed that CI + CII coupled respiration in the PFC was predictive also of TNF-α levels in plasma, with inclusion of all groups, such that higher levels of plasma TNF-α were associated with lower levels of oxygen consumption. (Simple Linear Regression, n = 36). C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001. (For interpretation of the references to colour in this figure legend, the reader is referred to the Web version of this article.)

Supplementary Fig. 1

Representative total protein image of stain-free western blot (post-transfer). C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed.

Supplementary Fig. 2

Main effects of parity in water baseline and no effects of stress on 0.25% sucrose preference or total liquid consumption. (A) Water intake baseline during the week prior to parturition exhibited a significant main effect of parity (p = 0.001) but no significant interaction (p = 0.23) or main effect of stress (p = 0.95) such that primiparous females had significantly higher water intake compared to nulliparous groups. n = 6–11/group. (B) Preference for a weak (0.25%) sucrose solution exhibited trends for a main interaction between stress and parity (p = 0.08) and main effect of parity (p = 0.06), but no significant main effect of stress (p = 0.69). n = 5–6/group. (C) Total fluid intake during 1% sucrose measurements were not significantly different between groups, evidenced by no main effects of parity (p = 0.27), stress (p = 0.82), nor interaction (p = 0.55). n = 5–6/group. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01,***p ≤ 0.001.

Supplementary Fig. 3

No significant differences in open arm entries, distance, or velocity during the elevated plus maze. (A) Open arm entries were not different between groups as evidenced by no main effects of parity, stress, nor interaction n = 9–11/group. (B) All groups traveled similar distances during testing, with no significant main effects of parity, stress, nor interaction. n = 9–11/group. (C) Average velocity was similar between groups. n = 9–11/group. See Supplementary Table 2 for statistical details. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM.

Supplementary Fig. 4

No significant effects of stress on self-grooming or consummatory behaviors. On average, P + S dams did not exhibit differences in (A) self-grooming (p = 0.65) (B) or consummatory (eating/drinking) behavior (p = 0.6) during home cage maternal care observations. n = 9/group. P = primiparous controls and P + S = primiparous stressed. All data are represented as mean ± SEM..

Supplementary Fig. 5

No effects of stress or parity on mitochondrial content in PFC or NAc. (A) There were no significant differences between groups in expression of the mitochondrial outer membrane protein, TOM20 in the prefrontal cortex. n = 7–11/group. (B) There were no significant interactions or main effects of stress or parity on complex II respiration in nucleus accumbens (NAc). n = 9–11/group (C) There were no significant differences in CI + II/ETS-linked capacity in the NAc between groups. n = 9–11/group. See Supplementary Table 2 for statistical details. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM.

Supplementary Fig. 6

There were no effects of parity or gestational stress on other measured PFC cytokine levels. There were no significant main interactions, or effects of stress or parity on levels of (A) IL-1β, (B) IL-1α, (C) IL-6, (D) or IL-10 in the prefrontal cortex. n = 7–11/group. See Supplementary Table 2 for statistical details. C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM.

Supplementary Fig. 7

Significant effect of parity on additional plasma cytokine levels. (A) IL-1β levels exhibited a significant main effect of parity (p = 0.002), but post hoc analyses revealed no significant differences between groups. n = 8–11/group. (B) Analysis of IL-6 revealed a significant main effect of parity (p = 0.003). n = 8–11/group. (C) There was a trend for an interaction between parity and stress (p = 0.06) and a significant main effect of parity (p = 0.008), but no main effect of stress (p = 0.44) in IL-18 levels. Post hoc analyses revealed no significant differences between groups. n = 8–11/group. (D) Similarly, there was a trend for an interaction (p = 0.096) and a significant main effect of parity (p = 0.002) on IFN-γ levels. n = 8–11/group. (E) There was a significant main effect of parity (p = 0.01), but no effect of stress or interaction on VEGF levels. Post hoc analyses revealed no significant differences between groups. n = 8–11/group. See Supplementary Table 2 for statistical details. (F) TNF-α in prefrontal cortex does not correlate with TNF-α in plasma. (Simple linear regression, n = 37). C = nulliparous controls, S = nulliparous stressed, P = primiparous controls, and P + S = primiparous stressed. All data are represented as mean ± SEM. *p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001.