Abstract
Early-life stressors can affect reproductive development and change responses to adult stress. We tested if resource scarcity in the form of limited bedding and nesting (LBN) from postnatal days (PND) 4 to 11 delayed sexual maturation in male and female mice and/or altered the response to an acute, layered, psychosocial stress (ALPS) in adulthood. Contrary to the hypotheses, age and mass at puberty were unaffected by the present application of LBN. Under basal conditions and after ALPS, corticosterone concentrations in males, diestrous females, and proestrous females reared in standard (STD) or LBN environments were similar. ALPS disrupts the luteinizing hormone (LH) surge in most mice when applied on the morning of proestrus; this effect was not changed by resource scarcity. In this study, the paucity of effects in the offspring may relate to a milder response of CBA dams to the paradigm. While LBN dams exited the nest more often and their offspring were smaller than STD-reared offspring on PND11, dam corticosterone concentrations were similar on PND11. To test if ALPS disrupts the LH surge by blunting the increase in excitatory GABAergic input to gonadotropin-releasing hormone (GnRH) neurons on the afternoon of proestrus, we conducted whole-cell voltage-clamp recordings. The frequency of GABAergic postsynaptic currents in GnRH neurons was not altered by LBN, ALPS, or their interaction. It remains possible that ALPS acts at afferents of GnRH neurons, changes response of GnRH neurons to input, and/or alters pituitary responsiveness to GnRH and that a more pronounced resource scarcity would affect the parameters studied.
Significance Statement
The stress and reproductive neuroendocrine systems interact, and early-life stress has reproductive consequences in humans. This study in mice rejected the hypotheses that an early-life stress, limited bedding and nesting (LBN), would delay sexual maturation and alter the response to an acute, layered, psychosocial stress (ALPS) in adulthood. ALPS disrupts the proestrous luteinizing hormone surge, which is critical for ovulation; this disruption is not altered by LBN. To assess a possible mechanism for this disruption, we conducted electrophysiological recording of gonadotropin-releasing hormone neurons to test if ALPS reduces excitatory GABAergic input to these cells. The frequency of GABAergic input was similar among groups, suggesting that LBN and ALPS act elsewhere in the broader neuroendocrine network controlling reproduction.
Introduction
The neuroendocrine systems regulating stress and reproduction are important for organisms to respond to their environments and ensure the continuation of the species. Gonadotropin-releasing hormone (GnRH) neurons in the hypothalamus integrate many inputs and serve as the final common central output to the downstream reproductive axis. GnRH is released in a pulsatile manner and acts on the pituitary to stimulate the release of luteinizing hormone (LH) and follicle-stimulating hormone (Wildt et al., 1981), which activate gametogenesis and steroidogenesis. In males and during most of the female reproductive cycle, sex steroids exert negative feedback to reduce overall GnRH and LH concentrations. Sustained elevated estradiol concentrations in the preovulatory period (proestrus in rodents) exert positive feedback to induce prolonged surges of GnRH and LH release (Sarkar et al., 1976; Moenter et al., 1992; Pau et al., 1993; Evans et al., 1994). The LH surge triggers ovulation (Fevold et al., 1931). The organization of the neuroendocrine stress axis is similar, with corticotropin-releasing hormone release from hypothalamic neurons stimulating secretion of adrenocorticotropic hormone from the pituitary and the adrenal cortex producing glucocorticoids that provide negative feedback at the brain and pituitary (Smith and Vale, 2006).
Neuroendocrine axes interact with one another. In humans, stress has clinical effects on the reproductive system with social consequences (Chrousos et al., 1998). Perceived stress in adulthood can reduce likelihood of both natural conception and pregnancy via assisted reproductive technology (Klonoff-Cohen et al., 2001; Matthiesen et al., 2011; Akhter et al., 2016). Early-life stress can disrupt reproductive development, though whether stress delays or advances puberty appears to depend on the type and timing of stressor (Foster et al., 2008; Mendle et al., 2011, 2016; Boynton-Jarrett and Harville, 2012). In humans, it is challenging to disentangle the independent effects of early-life and adult stress on reproduction from each other and from other factors and to understand how a history of early-life stress affects the response to adult stress. These questions can be better addressed using animal models.
A common model for early-life stress in rodents is the limited bedding and nesting (LBN) paradigm in which dams and pups are moved to a low-resource environment for several days; this alters the way the dam interacts with the pup without ongoing investigator interference (Rice et al., 2008; Walker et al., 2017). Effects of LBN on reproductive outcomes have been mixed in both mice and rats. LBN delayed the age at vaginal opening, an external indicator of puberty, in some studies (Manzano Nieves et al., 2019; Knop et al., 2020), advanced it in another (Davis et al., 2020), and had no effect in others (Knop et al., 2019; Eck et al., 2020). Some studies found a delay in preputial separation in males (Knop et al., 2019, 2020; Davis et al., 2020). There are many models of adult stress including psychosocial (e.g., restraint), metabolic (e.g., hypoglycemia), and immune (e.g., endotoxin; Toufexis et al., 2014; McCosh et al., 2019; Phumsatitpong et al., 2021). In adult mice, exposure to an acute, layered, psychosocial stress (ALPS) paradigm on the morning of proestrus disrupts the preovulatory (proestrous) LH surge in most mice (Wagenmaker and Moenter, 2017). The mechanisms by which ALPS leads to this disruption are not known, but do not include disrupting the preovulatory estradiol rise. With regard to possible mechanisms, the rate of GABAergic transmission to GnRH neurons (which is excitatory in these cells; DeFazio et al., 2002) increases on the afternoon of proestrus (Adams et al., 2018) and around the onset of the estradiol-induced LH surge (Christian and Moenter, 2007). The increase in GABAergic postsynaptic currents (PSCs) could account for the increased activity in GnRH neurons during the surge. Stressors alter GABAergic signaling within other hypothalamic regions. For example, restraint stress reduced the frequency of GABA transmission in parvocellular neurons in the paraventricular nucleus (Verkuyl et al., 2005). A reduction in GABAergic input to GnRH neurons following ALPS could help explain the disruption to the LH surge by this psychosocial stressor.
Early-life stress can affect the response to adult stressors, though whether animals are more susceptible or resilient depends on the type of stressors and the outcomes measured (Peña et al., 2019). In the present study, we investigated the effect of LBN on reproductive maturation and adult response to stress in male and female mice. We also tested the hypothesis that ALPS disrupts the LH surge by reducing the excitatory input to GnRH neurons on the afternoon of proestrus and determined if LBN altered the ability of ALPS to disrupt the LH surge.
Materials and Methods
Animals
All animal procedures were approved by the University of Michigan Institutional Animal Care and Use Committee. Mice had ad libitum access to water and food; breeders were provided Teklad 2919 chow (Inotiv) through weaning. After weaning, mice were maintained on Teklad 2916 (Inotiv). The vivarium followed a 14/10 h light/dark cycle (lights on at 3 A.M. EST); at times, dim red light (<10 lux) was used overnight. A male GnRH-GFP [Tg(Gnrh1-EGFP)51Sumo MGI:6158457, C57Bl6/J background] mouse expressing GFP under the control of the GnRH promoter (Suter et al., 2000) was housed with one or two female CBA mice (strain 000656; The Jackson Laboratory) in cages containing 550–650 ml of corn cob bedding (Bed-o’Cobs ¼, The Andersons). The female body mass was measured on the day of pairing with the male. Females were examined for vaginal plugs for up to 5 d after pairing, and their body mass was measured again around hypothetical gestational days 12–14. An increased body mass (>110% of the initial mass) was used as an initial confirmation of pregnancy, at which point females were moved to individual cages with 550–650 ml bedding and a Nestlet (Ancare) and plastic igloo (Bio-Serv) for enrichment.
Experimental design
This study assessed the independent effects of and interactions between early-life and adult stress on reproduction. Pilot studies were conducted to examine timing of LBN treatment and effects of litter size. Based on these, offspring from the first litter of a dam were studied, and the litter size was normalized to 5–8 pups per mouse as small litters had early vaginal opening, whereas large litters were delayed as has been observed (Caron et al., 2012). Dams and litters were assigned to either standard (STD) housing or LBN treatment as an early-life stress. The body mass of individual offspring was monitored after early-life treatment through adulthood. To determine if LBN affected external markers of reproductive maturation, we monitored the offspring for preputial separation (male) or vaginal opening and first appearance of estrus, as indicated by cornification of the vaginal epithelium (female; Korenbrot et al., 1977). Estrous cycles of female offspring in adulthood were monitored to test if cycles were disrupted by LBN.
To evaluate if LBN alters the response to an adult stressor, adult offspring were assigned to either control (CON) or ALPS treatment, resulting in four experimental groups (STD-CON, STD-ALPS, LBN-CON, LBN-ALPS). Block assignment within litters and minimization strategies were used to balance adult treatments. Serum corticosterone concentration was assessed in males and in diestrous and proestrous females to determine if cycle stage altered baseline corticosterone concentration or the corticosterone rise in response to stress. In proestrous females, we also assessed if a history of LBN affected the disruptions of the LH surge by ALPS. To investigate the mechanisms by which ALPS disrupts the surge, we monitored the GABAergic transmission to GnRH neurons using whole-cell voltage-clamp around the time of the typical LH surge.
LBN paradigm
LBN was used to model early-life stress. STD cages contained 550–650 ml of bedding and one full 5 × 5 cm square of Nestlet material. LBN cages contained ∼100 ml of bedding, enough to cover the floor of the cage with a single layer, and a 2.5 × 5 cm piece of Nestlet. LBN cages were fitted with a wire mesh platform ∼1.5 cm above the cage floor (allFENZ 23 Gauge Hardware Cloth, Home Depot). Food was provided in a small container placed on the floor of the cage to prevent pups crawling into the food hopper from the raised platform and to permit unobstructed video monitoring of dam behavior. Food was replenished daily.
Females were monitored for births daily before lights off beginning 19 d after observation of vaginal plug or after pairing with a male if no plug was noted. The day of birth was designated postnatal day (PND) 0. Pups (CBB6/F1 hybrids) from litters born within 1 d of each other were cross-fostered if needed to standardize litter sizes to 5–8 pups by PND2. The STD or LBN treatment period began on PND4. From PND4 to 11, animals were undisturbed in their respective treatment cages apart from transferring for video monitoring and daily replenishment of food and water. On the morning of PND11, a tail blood sample was collected from the dam for assessment of serum corticosterone concentration, and then all dams and litters were transferred to clean cages containing 550–650 ml of bedding and a 5 × 5 cm Nestlet until weaning at PND21. All offspring were weaned with same-sex littermates into STD cages with the addition of a plastic igloo; the igloo was removed on PND28.
Dam behavior monitoring
Continuous video monitoring occurred for either 24 h from the morning of PND5 to the morning of PND6, 48 h from the morning of PND4 to 6, or for the duration of the early-life treatment from PND4 to 11. The number of exits made by the dam from the nest was manually counted for 1 h periods beginning at zeitgeber times (ZT) 1 (4 A.M. EST), ZT15 (6 P.M. EST), and ZT19 (10 P.M. EST). The average number of exits and amount of time off the nest per hour was calculated for each dam. Recordings were conducted using FFmpeg (version 4.3.1) on the following computers: MacBook Air (Mid 2009, running OSX El Capitan, Version 10.11.6, with a 2.13 GHz Intel Core 2 Duo), MacBook Pro (Mid 2012, running macOS Catalina, Version 10.15.7, with a 2.5 GHz Dual-Core Intel Core i5), MacBook Pro (Early 2015, running macOS Catalina, Version 10.15.7, with a 2.7 GHz Dual-Core Intel Core i5), Mac mini (2018, running macOS Sonoma, Version 14.0, with a 3 GHz 6-Core Intel Core i5), and a MacBook Pro (Late 2013, running macOS Mojave, Version 10.14.6, with a 2.3 GHz Intel Core i7).
Body mass
Dam mass was recorded on PND4 (start of paradigm), PND11 (end of paradigm), and PND21 (pup weaning). Before placing pups into the new cage at the start of the paradigm, the average mass for each litter was recorded to identify any potential outliers. At the end of the paradigm, on the morning of PND11, pups were ear-marked and identified with Sharpie markings on the tail to permit individual tracking of each pup. Pup mass was recorded daily through PND24 and then weekly through PND70 and as indicated below.
Offspring maturation
Beginning at weaning, offspring were checked daily for preputial separation or vaginal opening (Ojeda et al., 1986; J. E. Nelson et al., 1990). Body mass was recorded on the day of preputial separation/vaginal opening. After vaginal opening, female mice underwent daily vaginal lavage to identify the day of first estrus based on cornification of vaginal epithelium; body mass was recorded on the day of first estrus. Anogenital distance (AGD) was measured with digital calipers (Marathon) for 3 consecutive days during the 10th postnatal week and averaged for each animal.
Adult female estrous cycles
To study the effect of LBN on estrous cycles, we monitored the vaginal lavages daily from PND70 to 90 (Goldman et al., 2007; Caligioni, 2009; Byers et al., 2012). The number of cycles (defined as the number of days in proestrus preceded by diestrus or estrus per 21 d monitoring period), the mean cycle length (days between successive proestrous stages), and the percentage of days in diestrus, estrus, and proestrus are reported.
ALPS paradigm
Animals exposed to adult stress were habituated for at least 2 weeks beforehand to tail and general handling. Female mice were studied either on the first day of diestrus or on proestrus, as determined by vaginal cytology and confirmed by uterine mass (diestrus, <100 mg; proestrus, >125 mg) measured the same day. An ALPS paradigm that disrupts the proestrous LH surge in most females was used (Wagenmaker and Moenter, 2017); both male (>PND84) and female (≥PND90) mice were studied. At 0 h (6.5 h after lights on), a tail blood sample (∼30 μl) was collected, and the serum was stored for assessment of corticosterone. In females, an additional 6 μl of whole blood was obtained for LH measurement. CON mice were only removed from their home cage for tail blood sampling at times equivalent to sampling in stressed mice and remained in the vivarium. ALPS mice were placed individually into a new cage and transferred to a new room. At 1 h, the mice were placed in a restraint tube (Braintree Scientific, flat-bottom restrainer small or Tailveiner-150 restrainer). At 3 h, restrained mice were exposed to a component of red fox (Vulpes vulpes) urine as a predator odor (2,3,5-trimethyl-3-thiazoline; ∼6 nmol; Contech Enterprises). Tail blood samples were collected at the end of the paradigm (5 h, 2:30 P.M. EST, 2.5 h before lights out) for corticosterone and LH in females. For females in diestrus, an additional tail blood sample for LH was collected at the time of lights out (5 P.M. EST). For those in proestrus, additional tail blood samples to monitor the effect of LBN and/or ALPS on the LH surge were taken hourly from 3 to 7 P.M. EST, unless noted. Animals were considered to exhibit an LH surge if any PM value was >3.8 ng/ml. This threshold of 3.8 ng/ml was determined from the mean + 3SD of LH concentrations measured on the morning of proestrus at 0 h.
Body mass was recorded at the start of ALPS and just before euthanasia. Males were transported to the laboratory at the end of the ALPS paradigm (2:30 P.M. EST). The mass of the adrenal glands, testes, and seminal vesicles was recorded and normalized to PM body mass. Diestrous females were transported to the laboratory after the LH sample at lights out, and proestrous females used for LH surge sampling were transported after the last sample 2 h after lights out. Adrenal and uterine masses were recorded in females.
Corticosterone administration
To mimic serum corticosterone concentrations induced by ALPS, STD-reared males from our main colony were fed corticosterone in Nutella, a design based on ongoing studies in females in which 2 mg/kg oral corticosterone at 0, 1, and 3 h (times corresponding to the ALPS transitions) mimics the serum corticosterone pattern observed during the ALPS paradigm (Gibson, 2023). To habituate animals to this feeding paradigm, we transferred the cage mates to a holding cage, and one mouse was left in the home cage with Nutella on a Petri dish for up to 5 min. Mice were habituated daily for at least a week prior to the experiment. On the day of the experiment, corticosterone (2 mg/kg) or 36% DMSO vehicle in Nutella (60 mg Nutella mixture for a 30 g mouse; range, 56.2–75.2 mg) was administered at 0 h (9:30 A.M. EST, 6.5 h after lights on), 1 h, and 3 h. Tail blood samples were collected as described above at 0, 1, 2, 3, 4, and 5 h after initial Nutella administration. Only one mouse per cage was sampled on a given day, as preliminary experiments determined that transfer to the holding cage while a cage mate consumed Nutella increased serum corticosterone concentration. Body and tissue masses were recorded for these mice as described for the ALPS paradigm.
Corticosterone enzyme immunoassay
Serum corticosterone concentrations were determined in duplicate samples diluted 1:100 by enzyme immunoassay (Arbor Assays, DetectX Corticosterone Kit, K014). Standard curves from 78.1 to 5,000 pg/ml or 39.0 to 10,000 pg/ml were run on each plate. Intra-assay coefficients of variation (%CVs) for standardss ranged from 3 to 6%; functional sensitivity, defined as the lowest standard with a CV <20%, was 39.0 pg/ml. The reportable range was 3.9–1,000 ng/ml.
Ultrasensitive LH assay
At the time of tail blood collection, 6 µl of whole blood was mixed with 54 µl of assay buffer (0.2% BSA—0.05% Tween 20—PBS), pH 7.5, and immediately placed on ice for up to 3 h and then stored at −20°C. Samples were assayed by Center for Research in Reproduction at the University of Virginia with the Ultrasensitive Mouse and Rat LH ELISA method (Steyn et al., 2013). The capture monoclonal antibody (anti-bovine LH beta subunit, 518B7) was provided by Janet Roser, University of California. The detection polyclonal antibody (rabbit LH antiserum, AFP240580Rb) was provided by the National Hormone and Peptide Program (NHPP). HRP-conjugated polyclonal antibody (goat anti-rabbit) was purchased from DakoCytomation (D048701-2). Mouse LH reference prep (AFP5306A; NHPP) was used as the assay standard. The limit of quantitation (functional sensitivity) is defined as the lowest concentration that demonstrates accuracy within 20% of expected values and intra-assay %CV <20% and was determined by serial dilutions of a defined sample pool. Intra-assay %CV is 2.2%. Inter-assay %CVs were 7.3% (low QC, 0.13 ng/ml), 5.0% (medium QC, 0.8 ng/ml), and 6.5% (high QC, 2.3 ng/ml). Functional sensitivity was 0.016 ng/ml, and the reportable range is 0.016–4.0 ng/ml. Samples were diluted 1:10, making the reportable range 0.16–40 ng/ml.
Electrophysiology
A subset of the adult proestrous females was used to characterize the effect of LBN and/or ALPS on GABAergic transmission to GnRH neurons. The mouse was transported to the laboratory between 3:00 and 3:30 P.M. EST (1.5–2 h before lights out), and the body mass was recorded. All solutions were bubbled with 95% O2/5% CO2 for at least 15 min before tissue exposure and throughout the procedures; chemicals were purchased from Sigma-Aldrich unless noted. The brain was rapidly removed and placed in ice-cold sucrose saline solution containing the following (in mM): 250 sucrose, 3.5 KCl, 26 NaHCO3, 10 D-glucose, 1.25 Na2HPO4, 1.2 MgSO4, and 3.8 MgCl2. Coronal brain slices (300 µM) containing the preoptic area and GnRH neurons were prepared in the sucrose saline solution with a vibratome (VT1200S, Leica Biosystems). Slices were incubated at room temperature for 30 min in a 50–50% mixture of the sucrose saline and artificial cerebrospinal fluid (ACSF) containing the following (in mM): 135 NaCl, 3.5 KCl, 26 NaHCO3, 10 D-glucose, 1.25 Na2HPO4, 1.2 MgSO4, and 2.5 CaCl2, pH 7.4. Slices were then incubated for 0.5–5 h in 100% ACSF before being transferred to the recording chamber mounted to an Olympus BX51WI upright fluorescent microscope.
Slices in the recording chamber were perfused (3–5 ml/min) with ACSF via a MINIPULS 3 peristaltic pump (Gilson). GABAergic PSCs were isolated by blocking ionotropic glutamate receptors with CNQX (10 µM) and D-APV (20 µM). Solution temperature was maintained between 29 and 32°C with an inline heating system (Warner Instruments). Individual GFP-positive GnRH neurons were visualized using infrared differential interference contrast and brief illumination with fluorescence microscopy. The recording pipette was filled with a high-chloride internal solution containing (in mM): 140 KCl, 10 HEPES, 5 EGTA, 0.1 CaCl2, 4 MgATP, and 0.4 NaGTP. A high-resistance (>1 GΩ) seal was made between the cell membrane and the pipette, and then the whole-cell configuration was achieved. The cell was held at −65 mV in voltage-clamp mode, and the recording quality was monitored by averaging the response to 16 hyperpolarizing voltage steps (5 mV, 20 ms, acquisition 100 kHz, filter 10 kHz). GABAergic PSCs were recorded during 2–3 min series (acquisition 10 kHz, filter 5 kHz).
Custom routines in Igor Pro (WaveMetrics) were used to detect PSCs, which were manually confirmed. The frequency of PSCs (number of events/recording duration) was determined for each cell. For each event, the interevent interval was calculated, defined as the backward interval from the time of that event's peak to the time of peak for the previous event. The cumulative probability distributions of interevent interval for each treatment group were also calculated. The true interevent interval for the first event in a recording is unknown and thus not included in calculations of interevent interval. For events with an interevent interval of at least 200 ms, the amplitude (the absolute value of the difference between the peak and baseline) was determined, and cumulative probabilities of amplitudes were calculated for each group; this analysis included first events preceded by >200 ms of the recording time. Isolated events (>200 ms interval in both the forward and backward direction between adjacent event peaks) were selected and averaged by cell. These averaged traces were used to estimate the decay time from 80 to 20% of the peak for each cell.
Statistical analysis
Statistical analysis was conducted in R (R Core Team, 2023); statistical packages used included rstatix (Kassambara, 2021), lme4 (Bates et al., 2015), lmerTest (Kuznetsova et al., 2017), lspline (Bojanowski, 2017), afex (Singmann et al., 2023), emmeans (Lenth et al., 2023), and nparLD (Noguchi et al., 2012). Plots were made with the packages ggplot2 (Wickham, 2016) and cowplot (Wilke, 2023). Tables were made with the flextable package (Gohel and Skintzos, 2023).
Linear mixed models (LMMs) were used to analyze these data as these models can account for the dependencies among data attributable to experimental design and permit missing data (Singmann and Kellen, 2019). For example, litter was included as a random effect for outcomes measured in offspring, as mice within a litter are not fully independent from one another, violating assumptions of more traditional tests such as ANOVAs. The type of test for each outcome measure and associated figure are in Table 1. Type 3 tests with effects coding were used, as recommended by Singmann and Kellen (2019). The Kenward–Roger approximation was used for estimation of degrees of freedom for LMMs. Post hoc comparisons were made using pairwise tests of estimated marginal means (emmeans). For multiple pairwise comparisons, p values were adjusted using Holm's method; confidence intervals could only be adjusted using the more conservative Bonferroni’s method with this package in R. The pairwise comparisons for all tests are in Table 2.
Residual and Q–Q (quantile–quantile) plots were used to check the assumptions of models; alternative models were selected when available if assumptions were not met, as reported in Table 1. Nonparametric longitudinal analysis has been used to assess mouse behavioral data across time (Noguchi et al., 2012; Pernold et al., 2019) and was used to analyze the number of exits that the dams made from the nest over time. Because the ultrasensitive LH assay can report a maximum concentration of 40 ng/ml and many of our samples in proestrous mice exceeded this concentration, data for this parameter were not normally distributed, and an accurate estimate of the effect of stress on absolute LH concentrations was precluded. We instead focused on the binary outcome of whether or not the mouse exhibited an LH surge (at least one measurement ≥3.8 ng/ml) and fit these data with a generalized linear mixed-effect model for the binomial logistic regression family. The sample sizes and the low variance in the adult CON groups, because of the high likelihood of observing a surge in those groups, precluded the model from adequately estimating the interaction term of early-life treatment (LBN) and adult treatment (ALPS). Thus, the formula was simplified to consider only the independent main effects of each stressor.
The number of PSCs was fit with a generalized linear mixed-effect model for the negative binomial (NB) family to handle count data with meaningful zeros (no observed PSCs during recording period). To assess if early-life or adult treatment affected PSC properties (interval and amplitude), we compared the distributions from all four treatment groups with the Anderson–Darling test using the kSamples package in R (Scholz and Zhu, 2023). If the test comparing all four distributions was significant, post hoc Anderson–Darling tests comparing the (1) STD-CON and STD-ALPS, (2) LBN-CON and LBN-ALPS, (3) STD-CON and LBN-CON, and (4) STD-ALPS and LBN-ALPS groups were conducted, and p values were adjusted for multiple comparisons. To help with interpretation of the comparison of these distributions, a bootstrapping approach was used to estimate the 95% confidence interval of the mean of each treatment group and the difference in means for the four comparisons described just above (1–4). This approach, inspired by Ho et al. (2019), was necessitated by the non-normality of the distributions and adapted using the custom code to account for the experimental design. Briefly, the dataset was resampled 5,000 times; for each resampling iteration, the mean of each treatment group was calculated, as was the difference in the means for the comparisons described above (1–4). The average of these group means and differences in means were calculated for the 5,000 iterations. The boundaries for the 95% confidence intervals, or percentile intervals, were found by ranking the group's mean estimates for all 5,000 iterations and selecting the 2.5th percentile and the 97.5th percentile as the lower and upper bounds of the interval, respectively; the same process was applied to obtain a confidence interval for the differences in means.
Code accessibility
The PSC detection and analysis code used is freely available online at https://gitlab.com/um-mip/coding-project/. The FFmpeg and R analysis code is freely available online at https://github.com/gibson-amandag/LBN. These scripts are also provided as Extended Data 1. Analyses were conducted on a Lenovo Yoga 9, 11th Gen Intel Core i7, running Windows 11 Home.
Extended Data
Zip file of custom code for PSC detection and analysis, ffmpeg recording of dam behavior, and R analysis. Download Extended Data, ZIP file.
Results
LBN dams exited the nest more frequently
LBN was applied from PND4 to 11 (Fig. 1A; numbers in Extended Data Table 1-1). LBN dams had a higher body mass than STD dams at PND4, 11, and 21 (Fig. 1B; Extended Data Table 1-2; STD, n = 25; LBN, n = 24; Table 2, Row 1, p = 0.033). Regardless of treatment, dams gained body mass during the paradigm (Table 2, Row 2, p < 0.001) and decreased body mass between PND11 and 21 (Table 2, Row 4, p = 0.007). There were no differences in morning serum corticosterone concentration between STD and LBN dams on PND11, indicating LBN did not chronically elevate this hormone in dams [Fig. 1C; STD, n = 24; LBN, n = 24; t46 = −1.67; p = 0.102; difference (STD-LBN) = −9.08; 95% CI = [−20.04, 1.88]; Cohen's d = −0.48]. This may indicate the LBN phenotype is milder in CBA dams than in other strains. Dam behavior was captured on video during the paradigm. The number of exits that each dam made from the nest was scored for 1 h periods beginning at ZT1, 15, and 19 and then averaged for each PND (STD, n = 19–25; LBN, n = 19–24). LBN dams had more exits than STD dams throughout the paradigm (Fig. 1D; warmer colors indicate dams with more exits over time; Extended Data Table 1-2; p < 0.001). Interestingly, there was no difference in the percentage of time that STD and LBN dams spent on the nest (Fig. 1E; Extended Data Table 1-2; p = 0.156). Together, these observations suggest that LBN dams have more fragmented interactions with the pups.
Table 1-1
Number of dams in each group for studies in Figure 1. Dam behavior includes values for number of nest exits per hour and percentage of time spent off the nest. Download Table 1-1, DOCX file.
Table 1-2
Statistics for tests of dam mass and dam behavior over time. Linear mixed model and pairwise comparisons of the dam mass on PND4, 11, and 21 was fit with the equation mass ∼ early-life treatment * PND@ + (1 | dam). Postnatal day (PND) was treated as a factor variable. Dam behavior parameters were analyzed with a nonparametric longitudinal model using the nparLD package in R, with the F1 LD F1 Model. The subject variable was each dam, early-life treatment (STD or LBN cage) was the between-subject factor (‘whole-plot’ factor), and PND was the within-subject factor (‘sub-plot’ repeated factor). Download Table 1-2, DOCX file.
LBN affected pup mass
Prior to treatment on PND4, there were no differences in offspring body mass between the litters that would receive STD (mean ± SEM, 2.8 g ± 0.06, 25 litters) and LBN (2.9 g ± 0.06, 24 litters) treatment [Fig. 2A; t(47) = −1.18; p = 0.245; difference (STD-LBN) = −0.10; 95% CI = [−0.27, 0.07]; Cohen's d = −0.34]. After treatment on PND11, LBN offspring were smaller than STD offspring (Fig. 2B; Extended Data Table 2-1; Table 2, Row 5, p = 0.006; STD, 74 females and 80 males; LBN, 74 females and 58 males). Offspring sex did not affect mass at PND11 (p = 0.616) or interact with treatment (p = 0.124); the overall litter averages are thus displayed in Figure 2B. This demonstrates that mass gain during the treatment window was slower in LBN offspring, consistent with prior studies of this early-life stressor.
Table 2-1
Linear mixed models of offspring mass on PND11 and anogenital distance. Equations were outcome ∼ early-life treatment * sex + (1 | dam). Early-life treatment is STD vs LBN rearing. Sex is males vs females. Download Table 2-1, DOCX file.
Table 2-2
Linear mixed model of the offspring mass from PND11-72 fit with the equation mass ∼ early-life treatment * PND@ + (1 | dam) + (1 | mouse). Early-life treatment is STD vs LBN rearing. PND@: Linear splines at PND21 and 35 allow the model to change the slope of the line for the segments between PND11-21, from 21-35, and from 35-72. Male and female offspring were fit with separate models. Download Table 2-2, DOCX file.
Table 2-3
Linear mixed models for maturation with the equation maturation feature ∼ early-life treatment + (1 | dam). Early-life treatment is STD vs LBN rearing. Models were fit for age and for mass at vaginal opening, first estrus, and preputial separation. Download Table 2-3, DOCX file.
We continued recording offspring mass into adulthood to test if there was an effect of rearing conditions on subsequent growth, creating separate linear models of growth for female and male offspring. LBN treatment did not alter body mass growth in females (Fig. 2C; Extended Data Table 2-2; STD, postweaning, 24 litters and 74 mice; LBN, postweaning, 22 litters and 73 mice). In contrast in males, LBN treatment altered the overall pattern of growth (Fig. 2C; Extended Data Table 2-2; p < 0.001; STD, postweaning, 19 litters and 63 mice; LBN, postweaning, 14 litters and 42 mice). To understand how LBN changed the trajectory of growth, we conducted post hoc comparisons of body mass at discrete days throughout development (PND11, 21, 35, 56, and 72). The masses were not different through PND35 (PND11, Table 2, Row 6, p = 0.129; PND 21, Table 2, Row 7, p = 0.129; PND 35, Table 2, Row 8, p = 0.121). As males transitioned from adolescence to adulthood, the LBN mice appeared to gain mass more slowly, and by PND56, the LBN mice were ∼1 g smaller than STD mice (Table 2, Row 9, p = 0.045) with the difference persisting into the end of the observation period at PND72 (Table 2, Row 10, p = 0.024).
LBN did not alter reproductive maturation or estrous cycles
To assess if LBN altered external markers of reproductive maturation, the age and mass at vaginal opening and first estrus were monitored in females (STD, 24 litters and 74 mice; LBN, 22 litters and 73 mice, unless otherwise noted below); age and mass at preputial separation were monitored in males (STD, 19 litters and 63 mice; LBN, 14 litters and 41 mice). LBN did not affect age at vaginal opening (Fig. 2D; Extended Data Table 2-3; p = 0.217) or first estrus (p = 0.221), and there were no differences in body mass at these milestones (Fig. 2E; Extended Data Table 2-3; vaginal opening, p = 0.754; first estrus, p = 0.758; STD, 23 litters and 70 mice; LBN, 22 litters and 73 mice). Similarly, the age (Fig. 2D; Extended Data Table 2-3; p = 0.177) and mass (Fig. 2E; Extended Data Table 2-3; p = 0.846) at preputial separation were not affected by LBN. Adult AGD was not affected by LBN in either sex [model mean ± SEM (mm), female STD, 6.4 ± 0.08; female LBN, 6.3 ± 0.09; male STD, 16.8 ± 0.09; male LBN, 16.6 ± 0.1; Extended Data Table 2-1; p = 0.257], but the typical increased AGD in males versus females was observed (Extended Data Table 2-1; Table 2, Row 11, p < 0.001). To test if LBN altered estrous cycles, daily vaginal lavages were obtained from PND70 to 90 (STD, 23 litters and 73 mice; LBN, 22 litters and 73 mice). Figure 3A shows representative estrous cycles from both groups. LBN had no effect on the number of estrous cycles (Fig. 3B; Extended Data Table 3-1; p = 0.359) or features of the cycle including the length (Fig. 3C; Extended Data Table 3-1; p = 0.457) or the percentage of days spent in each stage (Fig. 3D; Extended Data Table 3-1; p = 0.865). Together, these results indicate that LBN did not disrupt reproductive maturation or estrous cyclicity.
Table 3-1
Statistics for estrous cycles from postnatal days 70-90. The number of cycles was fit with the linear mixed model equation # of cycles ∼ early-life treatment + (1 | dam). The log of the mean cycle length in days was fit with equation log10(cycle length) ∼ early-life treatment + (1 | dam). Early-life treatment is STD vs LBN rearing. The number of days spent in each cycle stage was assessed with a Chi-squared test (n = 3066). Download Table 3-1, DOCX file.
Early-life stress did not alter the corticosterone response to adult stress
To determine if early-life stress alters the serum corticosterone response to adult stress, STD and LBN mice were exposed to an ALPS paradigm (Wagenmaker and Moenter, 2017) or remained in nonstressed, home cage CON conditions (numbers in Extended Data Table 4-1). There were no effects of early-life stress at any point in either sex or either cycle stage in females (Extended Data Table 4-2); thus, results are combined in Figure 4. Baseline corticosterone levels were the same in CON and ALPS males (Table 2, Row 12, p = 0.883) and diestrous (Table 2, Row 16, p = 0.826) and proestrous (Table 2, Row 18, p = 0.387) females. Baseline corticosterone levels were elevated in proestrous relative to diestrous mice that received ALPS (Table 2, Row 25, p < 0.001). In all three groups, CON mice exhibited the typical diurnal increase in corticosterone (male, Table 2, Row 14, p < 0.001; diestrus, Table 2, Row 20, p < 0.001; proestrus, Table 2, Row 22, p < 0.001). Similarly, all three groups had a similar response to ALPS treatment, postparadigm corticosterone concentrations being 2–3-fold greater in ALPS than in CON mice (male, Table 2, Row 13, p < 0.001; diestrus, Table 2, Row 17, p < 0.001; proestrus, Table 2, Row 19, p < 0.001). These results indicate that early-life stress in the form of LBN treatment from PND4 to 11 did not alter this neuroendocrine response to a series of psychosocial stressors in adulthood in either males or females.
Figure 4-1
The ALPS paradigm caused small changes in tissue mass in males. Individual values and model mean ± SEM for A. morning body mass; B. percent change in body mass after adult treatment; C. adrenal mass; D. normalized adrenal mass; E. seminal vesicle mass; F. normalized seminal vesicle mass; G. testicular mass; and H. normalized testicular mass. Some error bars obscured by mean lines. * p < 0.05, ** p < 0.01, *** p < 0.001. Numbers are in Table 4-3. Results from the full statistical models are in Table 4-4. Abbreviations: STD, standard-reared; LBN, limited bedding and nesting; CON, adult control; ALPS, acute, layered, psychosocial stress in adulthood. Download Figure 4-1, TIF file.
Figure 4-2
LBN and ALPS cause limited changes in tissue masses in females. Individual values and model mean ± SEM for A. morning body mass; B. percent change in body mass after adult treatment. C. adrenal mass; D. normalized adrenal mass; E. uterine mass; F. normalized uterine mass in diestrous (left) and proestrous (right) females. Some error bars obscured by mean lines. * p < 0.05, ** p < 0.01, *** p < 0.001. Numbers are in Table 4-7. Results for the full statistical models are in Table 4-8. Abbreviations: STD, standard-reared; LBN, limited bedding and nesting; CON, adult control; ALPS, acute, layered, psychosocial stress in adulthood. Download Figure 4-2, TIF file.
Figure 4-3
Acute elevation of serum corticosterone decreases testicular mass in males. Individual and model mean ± SEM for A. morning body mass; B. serum corticosterone concentrations; comparisons between 0 and 2 mg/kg treatment at each hour; C. percent change in body mass after adult treatment. D. adrenal mass; E. normalized adrenal mass; F. seminal vesicle mass; G. normalized seminal vesicle mass; H. testicular mass; and I. normalized testicular mass. Some error bars are obscured by mean line. Vehicle (0 mg/kg): 11 litters and 19 mice, except for adrenal mass with 18 mice; corticosterone (2 mg/kg): 11 litters and 17 mice. * p < 0.05, ** p < 0.01, *** p < 0.001. Results for the full statistical models are in Tables 4-5 and 4-6. Download Figure 4-3, TIF file.
Table 4-1
Number of litters and mice with serum corticosterone measurements before and after adult treatment. Download Table 4-1, DOCX file.
Table 4-2
Statistics for serum corticosterone in male and female offspring. Data from males were fit with the linear mixed model equation log10(cort) ∼ early-life treatment * adult treatment * time + (1 | mouse) + (1 | dam). Data from females were fit with the linear mixed model equation log10(cort) ∼ cycle stage * early-life treatment * adult treatment * time + (1 | mouse) + (1 | dam). Cycle stage is diestrus vs proestrus; early-life treatment is STD vs LBN rearing; adult treatment is CON vs ALPS; time is pre (0 h) vs post (5 h). Download Table 4-2, DOCX file.
Table 4-3
Number of litters and male mice with mass measurements on the day of adult treatment. Lower numbers for some tissue masses are attributable to loss of or damage to tissue at dissection. Download Table 4-3, DOCX file.
Table 4-4
Statistics from linear mixed models of male masses on day of adult treatment. Data were fit with the formula feature ∼ early-life treatment * adult treatment + (1 | dam). Early-life treatment is STD vs LBN rearing; adult treatment is CON vs ALPS. Download Table 4-4, DOCX file.
Table 4-5
Number of litters and female mice with mass measurements on the day of adult treatment. Adrenal masses were not collected from females used for electrophysiology studies (Figures 6-7). Lower numbers for some tissue masses are attributable to loss of or damage to tissue at dissection. Download Table 4-5, DOCX file.
Table 4-6
Statistics from linear mixed models of female masses on day of adult treatment. Data were fit with the formula feature ∼ early-life treatment * adult treatment * cycle stage + (1 | dam). Early-life treatment is STD vs LBN rearing; adult treatment is CON vs ALPS; cycle stage is diestrus vs proestrus. Download Table 4-6, DOCX file.
Table 4-7
Statistics from linear mixed models of male masses on day of vehicle (0 mg/kg) or corticosterone (2 mg/kg) treatment. Data were fit with the formula feature ∼ dosage + (1 | dam). Download Table 4-7, DOCX file.
Table 4-8
Statistics for serum corticosterone in males with vehicle or corticosterone administration. Data were fit with the linear mixed model equation log10(cort) ∼ dosage * time + (1 | mouse) + (1 | dam). Dosage is 0 mg/kg vs 2 mg/kg; time compares 0 h, 1 h, 2 h, 3 h, 4 h, and 5 h. Download Table 4-8, DOCX file.
Body mass was monitored before and after treatment (males, Extended Data Fig. 4-1; Extended Data Tables 4-3, 4-4; females, Extended Data Fig. 4-2; Extended Data Tables 4-5, 4-6). Consistent with the weekly monitoring of body mass in early adulthood, LBN males were smaller than STD males at the start of the experiment (Extended Data Fig. 4-1A; Extended Data Table 4-4; Table 2, Row 28, p = 0.037), and LBN did not affect the initial body mass in females (Extended Data Fig. 4-2A; Extended Data Table 4-6). ALPS animals of both sexes lost a greater percentage of body mass during treatment (males, Extended Data Fig. 4-1B; Extended Data Table 4-4; Table 2, Row 30, p < 0.001; females, Extended Data Fig. 4-2B; Extended Data Table 4-6; Table 2, Row 39, p < 0.001), likely attributable in part to no access to food or water during the last 4 h of stress treatment. By chance, morning body mass of ALPS males was greater than CON males (Extended Data Fig. 4-1A; Extended Data Table 4-4; Table 2, Row 29, p = 0.008), complicating interpretation of these observations, but males had apparent changes following ALPS in the normalized mass of seminal vesicles (Extended Data Fig. 4-1F; Extended Data Table 4-4; Table 2, Row 34, p = 0.048) and mass of the testes (absolute, Extended Data Fig. 4-1G; Extended Data Table 4-4; Table 2, Row 36, p = 0.003; normalized, Extended Data Fig. 4-1H; Extended Data Table 4-4; Table 2, Row 37, p = 0.021). Females did not exhibit changes in either uterine or adrenal mass following ALPS (Extended Data Fig. 4-2; Extended Data Table 4-6). Small changes in organ masses in males and females associated with LBN treatment are in Extended Data Tables 4-4 and 4-6 and Table 2 (Rows 31, 33, 35, and 40).
To test if corticosterone could reproduce the effects of ALPS on organ masses in males, additional mice were fed corticosterone or vehicle (Extended Data Fig. 4-3; Extended Data Tables 4-7, 4-8). Corticosterone decreased testicular mass (Extended Data Table 4-7; absolute, Extended Data Fig. 4-3H; Table 2, Row 51, p = 0.010; normalized, Extended Data Fig. 4-3I; Table 2, Row 52, p = 0.085), suggesting this parameter may be sensitive to stress.
ALPS decreased afternoon LH in proestrous mice; LBN had no additional effect
Samples for LH collected at the end of the ALPS paradigm and at lights out from diestrous mice were averaged (STD-CON, nine litters and 10 mice; STD-ALPS, eight litters and 8 mice; LBN-CON, seven litters and 9 mice; LBN-ALPS, nine litters and 9 mice). Neither early-life stress (p = 0.687) nor adult stress (p = 0.067) had an effect on mean PM LH concentrations on diestrus (Fig. 5A; Extended Data Table 5-1), though the effect of adult stress approached the level set for significance. We tested if LBN alters the ability of ALPS to disrupt the proestrous LH surge (Wagenmaker and Moenter, 2017; STD-CON, 7 litters and 8 mice; STD-ALPS, 11 litters and 16 mice; LBN-CON, 8 litters and 8 mice; LBN-ALPS, 14 litters and 19 mice). The maximum observed LH, the proportion of mice with an LH surge, and individual LH concentration profiles are in Figure 5B–D, respectively. Adult stress decreased the proportion of proestrous mice with an LH surge (logistic regression, χ2 = 26.12; p < 0.001), but exposure to early-life stress did not change the likelihood of observing an LH surge (χ2 < 0.01; p = 0.991; Fig. 5C).
Table 5-1
Statistics from linear mixed models of average LH in diestrous mice on day of adult treatment. Data were fit with the formula average LH ∼ early-life treatment * adult treatment + (1 | dam). Early-life treatment is STD vs LBN treatment; adult treatment is CON vs ALPS treatment. Download Table 5-1, DOCX file.
Neither LBN nor ALPS reduced the frequency of GABA PSCs in GnRH neurons
The frequency of GABA PSCs in GnRH neurons increases around the time of the LH surge on proestrus (Adams et al., 2018). We thus tested the hypothesis that ALPS decreases the frequency of these PSCs (STD-CON, 14 cells; STD-ALPS, 15 cells; LBN-CON, 15 cells; LBN-ALPS, 14 cells; five litters and six mice in all groups). There were no differences in the passive properties or recording quality among treatment groups (Fig. 6A–D; Extended Data Table 6-1). Representative PSC recordings from neurons in each group are in Figure 7A; the average PSC from each group is in Figure 7B. Neither LBN nor ALPS altered the frequency of GABA PSCs (Fig. 7C; Extended Data Table 7-1). Shifts in PSC patterns can occur within datasets with the same mean; however, when averaged by cell, the interevent interval of GABA PSCs was also similar among groups (Fig. 7D; Extended Data Table 6-1; n as above except LBN-ALPS n = 13 as one cell did not have PSCs). There were no differences in mean amplitude (Fig. 7E) or decay time (Fig. 7F; Extended Data Table 6-1). In contrast, the cumulative distribution of interevent intervals for all events is shifted toward longer intervals in the ALPS groups compared with those in the CON groups (Fig. 7G; Extended Data Table 7-2; pairwise AD tests; STD-CON vs STD-ALPS, p < 0.001; LBN-CON vs LBN-ALPS, p < 0.001).
Table 6-1
Statistics from linear mixed models of electrophysiology properties on day of adult treatment. The mean value for each cell was calculated, and data were fit with the formula feature ∼ early-life treatment * adult treatment + (1 | dam) + (1 | mouse). Early-life treatment is STD vs LBN rearing; adult treatment is CON vs ALPS. Download Table 6-1, DOCX file.
Table 7-1
Statistics for number of postsynaptic current (PSC) events per 240 s in GnRH neurons on the day of adult treatment. As the frequency data were skewed right and included zeros, a generalized linear mixed effects negative binomial model was used. Data were fit with the model equation # events per 240 s ∼ early-life treatment * adult treatment + (1 | mouse) + (1 | dam). The joint_tests function of the emmeans package was used to obtain these p-value estimates from the model. Early-life treatment is STD vs LBN rearing; adult treatment is CON vs ALPS. Download Table 7-1, DOCX file.
Table 7-2
Pairwise comparisons of distributions of interevent interval and amplitude for PSCs recorded in GnRH neurons. The Anderson-Darling criterion (AD), the standardized test statistic (T AD), and asymptotic p-value were calculated with the kSamples package (Scholz & Zhu, 2023). Bootstrapping was used to estimate the mean difference. The confidence interval is not adjusted for multiple comparisons. The p-values for both tests were adjusted using Holm’s method for multiple comparisons. Download Table 7-2, DOCX file.
Because of the large number of events in each group, the Anderson–Darling test could be identifying significant differences in the distribution that are not biologically relevant. To determine the effect size of these changes, we used bootstrapping to estimate 95% confidence intervals for the differences in means between groups. The mean interevent interval for events from STD-ALPS mice was estimated to be 0.78 s (95% CI, [0.65, 0.92]) longer than for events from STD-CON mice; in LBN mice, ALPS lengthened mean interevent interval by ∼0.32 s (95% CI, [0.19, 0.46]). This suggests that ALPS alters the excitatory input to GnRH neurons on the evening of proestrus. We interpret these results with caution, however, as three CON cells with high PSC frequencies contribute a disproportionate number of short interevent intervals.
The amplitude distribution for the STD-ALPS groups was shifted toward larger events (Fig. 7H; Extended Data Table 7-2; pairwise AD tests; STD-CON vs STD-ALPS, p = 0.006; STD-ALPS vs LBN-ALPS, p < 0.001). In LBN mice, ALPS did not shift the amplitude distribution (Extended Data Table 7-2; pairwise AD tests; LBN-CON vs LBN-ALPS, p = 0.711). The mean amplitude of the STD-ALPS group was ∼3.6 pA larger (95% CI, [1.36, 5.81]) than the mean of the STD-CON group and ∼5.2 pA larger (95% CI, [3.09, 7.34]) than the mean of the LBN-ALPS group, as estimated with bootstrapping.
Discussion
The stress and reproductive neuroendocrine systems interact, and early-life stress has reproductive consequences in humans. We tested the hypotheses that early-life stress in the form of resource scarcity provided through LBN would delay sexual maturation and alter the response to subsequent stress exposure in adulthood in mice. LBN, as applied in the present study, did not delay external indicators of sexual maturation in males or females. Furthermore, the corticosterone response to adult psychosocial stress was not altered by LBN in either sex. On proestrus, adult stress disrupts the LH surge, but this is not affected by a history of LBN. In contrast to our hypothesis that ALPS disrupts the LH surge by decreasing the frequency of excitatory GABA PSCs in GnRH neurons, these currents were not appreciably altered by either adult psychosocial stress or LBN. This suggests that the disruption of the LH surge by adult stress is not attributable to changes in the GABAergic input to GnRH neurons.
These findings lead us to reject our hypotheses, but it is important to point out that this rejection could be for at least two reasons. First, the hypotheses could simply be wrong. Second, CBA dams were more resilient to effects of LBN treatment than strains previously used, as indicated by a lack of persistently elevated serum corticosterone concentrations on PND11. Such a milder maternal effect could have diminished disruptions by this early-life resource scarcity on pups’ neuroendocrine maturation.
The LBN paradigm was chosen as the model for early-life stress because animal behavior is minimally disrupted by ongoing investigator interference (Walker et al., 2017). Over its implementation in several labs, LBN effectiveness has been evaluated in three main ways: dam behavior, pup mass at the end of the paradigm, and/or dam stress parameters. In both rats (Brunson et al., 2005; Ivy et al., 2008; Molet et al., 2016) and mice (Rice et al., 2008; Gallo et al., 2019), maternal care is fragmented by LBN treatment, leading to more transitions between behavioral states and more exits from the nest. This fragmentation of maternal care was confirmed in CBA dams in the present study. On the last day of treatment, mice and rat offspring of LBN dams exhibit elevated basal corticosterone concentrations (Gilles et al., 1996; Avishai-Eliner et al., 2001; Brunson et al., 2005; Rice et al., 2008; Naninck et al., 2015) and increased adrenal masses in rat pups (Avishai-Eliner et al., 2001; Brunson et al., 2005). Because the aforementioned outcomes require terminal studies in neonates, monitoring of body mass is a common proxy measure of the impact of LBN treatment in offspring, with LBN pups being smaller than STD pups after the paradigm (Gilles et al., 1996; Avishai-Eliner et al., 2001; Brunson et al., 2005; McLaughlin et al., 2016; Moussaoui et al., 2016; Knop et al., 2019, 2020; Manzano Nieves et al., 2019; Eck et al., 2020). We similarly observed a lower pup mass at the end of the LBN paradigm in both male and female CBB6/F1 hybrid offspring, suggesting that the LBN paradigm was effectively implemented as an early-life stressor in our laboratory.
Strain can impact dam behavioral responses to LBN and the effect on offspring mass in mice (Pardo et al., 2023), and there are indications that the LBN treatment may induce a milder phenotype in CBA dams and their CBB6/F1 hybrid offspring in the present study. Morning plasma corticosterone concentrations were elevated in rat dams at the end of the paradigm on PND9 (Ivy et al., 2008); serum corticosterone was not, however, elevated on PND11 in CBA dams. Adaptation to psychosocial stressors resulting in decreased glucocorticoid output has been observed in mice (Wagenmaker and Moenter, 2017) and ewes (Wagenmaker et al., 2010). Perhaps such acclimation occurs more quickly in CBA dams than in other species or strains, which could contribute to milder outcomes in their pups. The PND11 difference in body mass resolved quickly in female offspring, and there were no further impacts of LBN treatment on body mass growth through adulthood of the females we studied. In contrast, LBN males had mildly slowed growth that was evident later in adulthood. This suggests there may also be a difference in susceptibility to LBN treatment between the female and male offspring.
The effects of LBN on reproductive maturation vary across studies (Knop et al., 2019, 2020; Manzano Nieves et al., 2019; Davis et al., 2020; Eck et al., 2020). In the present study, we did not observe differences in body mass near the time of puberty or changes in the age at reproductive maturation, in part because we normalized litter sizes to provide more consistent nutrition. Some of the variations in the literature may be related to age of LBN exposure and/or subtle protocol differences such as the type of bedding or nesting material, the wire platform material, or vivarium conditions, along with the species and strain of animals. The variability in body mass may also underlie some of the variability in vaginal opening outcomes, as a decrease in body mass is known to delay vaginal opening (R. E. Nelson and Robison, 1976; Kirkpatrick and Rutledge, 1987; Castellano et al., 2011; Caron et al., 2012; Wu et al., 2016). Together, these studies point to the importance of considering possible confounding and interacting factors when assessing the effect of early-life stress on body mass and reproductive maturation.
A lack of effect of LBN on the estrous cycle was a more consistent observation across studies in both mice (Manzano Nieves et al., 2019) and rats (Davis et al., 2020; Eck et al., 2020) and was confirmed in the present study. Estrous cycles also remain unchanged following maternal separation as an early-life stress (Rhees et al., 2001; Ferraz et al., 2023). Although the observation of typical adult estrous cyclicity does not preclude other underlying changes in reproductive physiology (Wang and Moenter, 2020), the ability of animals exposed to early-life stress to establish cyclicity is an indication that aspects of the reproductive system can recover from developmental perturbations caused by this treatment.
Our findings that LBN did not alter adult basal corticosterone concentrations in the morning or afternoon are consistent with the observations of others (Brunson et al., 2005; Knop et al., 2019, 2020; Eck et al., 2020). The original study of LBN in mice did, however, observe persistently elevated basal corticosterone concentrations in 4–7 month-old males (Rice et al., 2008). In one study, adult LBN rats responded similarly to STD-reared rats when exposed to a 1 h restraint stress (Eck et al., 2020). The latter is consistent with our finding that adult LBN mice exhibited similar corticosterone profiles to STD-reared mice in response to a 5-h, layered stress paradigm, ALPS, which included restraint. This suggests that any transient changes in the neuroendocrine stress response following perinatal LBN exposure were normalized by subsequent rearing and weaning into STD housing conditions.
The primary motivation for this work was to study the reproductive consequences of LBN, including how it affects responses to ALPS exposure. In diestrous mice, the ALPS effect to reduce mean LH concentrations approached the value accepted for significance. This may reflect stress suppression (Yang et al., 2017) of the pulsatile LH release typical of this stage (Czieselsky et al., 2016), but the infrequent sampling in the present study is not designed to assess pulse parameters. The ALPS paradigm was developed in the context of understanding the effects of acute stress exposure on the sustained preovulatory increase in LH concentration that occurs on the afternoon of proestrus. As reported (Wagenmaker and Moenter, 2017), ALPS initiated on the morning of proestrus disrupts the LH surge in most mice. We hypothesized LBN exposure would alter the effect of ALPS on the LH surge, but LBN had no additional effect, suggesting that the paradigm studied for early-life stress did not confer either resilience or susceptibility to the adult stress studied for this parameter. There are several potential explanations for this finding. First, LBN may be milder than other perinatal stressors, such as lipopolysaccharide exposure (Li et al., 2007), that have lasting effects on the reproductive consequences of adult stress. Second, the preweaning return to STD housing conditions may have also buffered the effects of early resource limitations. Third, the work by Peña et al. (2019) demonstrated the challenges of trying to predict if experiencing one stressor will lead to susceptibility or resilience to a subsequent stressor by comparing the impact of early-life stressors on the behavioral responses to 10 d of chronic social defeat. Most pertinent to the work presented here, maternal separation with reduced bedding from PND2 to 12 did not affect postdefeat behavior, whereas the same treatment from PND10 to 17 increased susceptibility (Peña et al., 2019), a difference which the authors attributed to the transition out of the stress hyporesponsive period for pups who experienced the paradigm later in development (Rincón-Cortés and Sullivan, 2014). The lack of effect of LBN from PND4 to 11 could thus be reflective of pups experiencing the paradigm during the stress hyporesponsive period.
The mechanisms by which ALPS disrupts the LH surge, both in terms of the components of the stress response and the site of action within the reproductive axis, remain unknown. GnRH neurons from proestrous mice receive a higher frequency of GABAergic input, which is excitatory in these cells (DeFazio et al., 2002), in the evening than in the morning, consistent with the switch from negative to positive feedback and the timing of the LH surge (Adams et al., 2018). We thus tested the hypothesis that GABA input to GnRH neurons is diminished by ALPS. The frequency of GABAergic PSCs in GnRH neurons was not altered by stress, nor was the mean interevent interval by cell. The distribution of all interevent intervals was shifted toward longer intervals in cells from ALPS mice, but this appears to be primarily related to three CON cells with a high frequency of PSCs. It is possible that one mechanism by which ALPS ultimately disrupts the LH surge is by reducing the incidence of GnRH neurons receiving a high frequency of GABAergic input, but the current study is not powered to assess this. The observation that the amplitude distribution of PSCs from STD-ALPS mice is shifted toward larger events runs counter to the hypothesis that this stressor reduces the efficacy of GABA input to GnRH neurons. The magnitudes of these observed changes in event distributions are small and near our limit of detection for differentiating signal and noise in these electrophysiological recordings. Amplitude did not differ when comparing the mean values from each cell; thus, the subtle shifts in the cumulative probability distributions of events may also reflect some bias toward the amplitude of PSCs from cells with more events included in the analysis.
One possible caveat to this work is that brain slice preparation could sever key neuronal networks that may be critical for the disruption of reproductive output following ALPS or that other in vivo changes attributable to early-life or adult stress do not persist in this configuration. In this regard, both acute and chronic stressors induced measurable changes in synaptic physiology of the hypothalamic paraventricular nucleus in brain slices (Levy and Tasker, 2012). The increase in GABA PSC frequency in GnRH neurons on the evening of proestrus occurs concurrently with the expected time of the LH surge, but the experimental design precludes the ability to directly correlate the properties of PSCs to the incidence of the LH surge in that animal; in this regard, uterine mass in animals used for PSC recordings were consistent with proestrus (Extended Data Fig. 4-2). The source of the increased GABAergic transmission to GnRH neurons during the LH surge is not known and has been postulated to be the suprachiasmatic nucleus (Christian and Moenter, 2007) or the anteroventral periventricular kisspeptin neurons (Wang and Moenter, 2020). Although GABAergic transmission was not altered by ALPS or LBN, it is possible that peptidergic modulators from these, or other, populations are altered in a manner that disrupts the LH surge.
While persistent effects of LBN were not observed into adulthood in this study, care must be taken not to overextrapolate these findings as demonstrating that early-life psychosocial and environmental manipulations are of no consequence. To allow rigorous control of conditions, this study considered a single type of early-life stress during 1 week of development, after which time animals were returned to STD housing conditions. Altering the timing, type, or duration of the stressor may lead to different outcomes. Indeed, a recent study found persistent reproductive effects following 3 weeks of postweaning social isolation, including on vaginal opening, estrous cycles, and activity of hypothalamic neurons (Agus et al., 2024), indicating that housing conditions during certain developmental periods can lead to changes that persist into adulthood. The observations of the present study direct future attention to the effects of both early-life and adult stress on the broader neuroendocrine network controlling reproduction, including upstream neuronal populations and pituitary gonadotropes.
Footnotes
The authors declare no competing financial interests.
We thank R. Anthony DeFazio, Elizabeth Wagenmaker, and Laura Burger for their expert technical assistance and Jenn Jaime for the assistance with animal work. We thank Michael Roberts for his input on postsynaptic current analysis approaches. We thank Christian Burgess, Kevin Jones, Audrey Seasholtz, and Joanna Spencer-Segal for their feedback on the project and manuscript draft. We also thank Chris Andrews of Consulting for Statistics, Computing, and Analytics Research (CSCAR) at the University of Michigan for statistical consulting. This work was supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development R01HD041469; A.G.G. was supported by the National Defense Science and Engineering Graduate Fellowship and Eunice Kennedy Shriver National Institute of Child Health and Human Development of the National Institutes of Health F31HD108872.
This is an open-access article distributed under the terms of the Creative Commons Attribution 4.0 International license, which permits unrestricted use, distribution and reproduction in any medium provided that the original work is properly attributed.