Hormonal contraceptives during adolescence impact the female brain and behavior in a rat model.

Timed pregnant (gestational day 15) Sprague-Dawley rats (Inotiv) were singly housed upon arrival. Cages were checked daily, and the day pups were born was designated as postnatal day (PND) 0. Female offspring were weaned on PND21 into groups of 2–3 littermates/cage. Beginning on PND30, rats were monitored daily for vaginal opening (VO), a proxy measure for pubertal onset in female rats ( Lewis et al., 2002 ; Vidal, 2017 ). HC administration did not begin unless vaginal openings appeared by PND35 to be translationally relevant, as HCs are not often used in humans until after puberty ( Todd and Black, 2020 ). No rats were excluded from the experiment due to delayed pubertal onset. Colony conditions were maintained at a temperature of 23 ± 1 °C and humidity of 30–50 % with a 12:12 light:dark cycle (lights on 0600). Rats had access to food and water ad libitum. All procedures were performed in accordance with The Ohio State University Institutional Animal Care and Use Committee and NIH guidelines. Fig. 1 shows the experimental timeline and endpoints for each cohort. HCs most often used by adolescents typically contain ethinyl estradiol (EE), a synthetic form of 17β-estradiol, paired with the synthetic progestin, levonorgestrel (LNG) ( Braverman et al., 2014 ; DeLeo et al., 2016 ). Rats were randomly assigned to receive either vehicle (sesame oil, Sigma #8008-74-0) or a combination of 10 μg EE (Cayman #10006486) and 20 μg LNG (Cayman #10006318) (EE + LNG) suspended in vehicle via a 0.2 ml subcutaneous injection. Animals housed in the same cage received the same treatment to prevent cross-contamination. This HC dosage (not adjusted for body weight), formulation, and route of administration was based on previous research in rats ( Simone et al., 2015 ; Lacasse et al., 2022 ). Rats were weighed and injections administered daily (0900–1100 h) from PND35–56. This time period encompasses the majority of adolescence in female Sprague-Dawley rats ( Lewis et al., 2002 ; Vidal, 2017 ). Rats were not given HC injections on the day of euthanasia. In cohorts that underwent behavioral testing, injections continued beyond PND56 throughout testing and were administered each day upon task completion (1500–1600 h). Beginning on PND40, estrous cyclicity was monitored by daily vaginal lavage in all rats. Vaginal samples were collected via a cotton swab saturated in sterile 0.9 % sodium chloride, applied to slides, and stained with 1 % toluidine blue ( Everett, 1989 ). Estrous cycle stage was determined by an experimenter blind to treatment group at 20× using a Nikon Eclipse E200 light microscope (Tokyo, Japan) based on the predominant cell type, as previously described ( Ajayi and Akhigbe, 2020 ). Rats were “cycling” if they had a 4–6 day continuous estrous cycle (progression through diestrus 1 and 2, proestrus, and estrus, in that order). Rats were “acyclic” if samples were in diestrus 1 or 2 for at least 7 continuous days by the end of the HC administration period. Rats were “irregularly cycling” if neither condition was met by the end of the HC administration period. On PND57, rats in cohort 1 ( n = 5/group) received an intraperitoneal (i.p.) injection of Euthasol (Virbac) and were transcardially perfused with 0.1 M phosphate buffered saline (PBS). Cardiac blood was collected prior to perfusion and allowed to clot at room temperature for at least 30 min. Blood was centrifuged at 1000 g for 10 min at 4 °C. Serum was collected, aliquoted, and stored at −70 °C until assay. Brains were collected for future analyses. Luteinizing hormone (LH) was measured in serum samples via an ELISA kit (Abnova KA2332, sensitivity: 0.5 ng/ml; limit of detection: 0.5 ng/ml) according to manufacturer’s instructions. Samples were run in triplicate. LH levels below the limit of detection were assigned a value of “0.” Intra-assay CV was calculated using only samples with detectable LH. Intra-assay CV was <11 %. On PND57, rats (n = 5–7/group) in cohort 2 received an i.p. injection of Euthasol (Virbac) and were transcardially perfused with 0.1 M PBS. Cardiac blood was collected prior to perfusion and allowed to clot at room temperature for at least 30 min. Blood was centrifuged at 1000 g for 10 min at 4 °C. The brain was hemisected and one hemisphere used for LC-MS/MS analysis of allopregnanolone (ALLO). The other hemisphere was used for all other analytes. Hemispheres designated for each analysis were counterbalanced. Brains and plasma were stored at −70 °C until shipped on dry ice to the Endocrine Technologies Core within the Oregon National Primate Research Center at the Oregon Health and Science University for LC-MS/MS analyses. Simultaneous analysis of EE, LNG, 17β-estradiol, progesterone, and testosterone in serum was performed using ultra-high performance liquid chromatography-heated electrospray ionization-tandem triple quadrupole mass spectrometry (LC-MS/MS) on a Shimadzu Nexera-LCMS-8050 instrument (Kyoto, Japan) as previously described ( Park et al., 2017 ; Blue et al., 2018 ). Modifications to previously published protocols were made, as described, to accommodate analysis of ALLO. Briefly, 150 μl of serum was mixed with 100 μl ultrapure water (Milli-Q, EMD Millipore, Billerica, MA) containing isotope-labeled internal standards for each steroid and added to a 400 μl SLE+ extraction plate (Biotage, Charlotte, NC). Steroids were eluted with 3 × 600 μl dichloromethane (Sigma, St. Louis, MO), dried at 40 °C with forced air and reconstituted in 50 μl of 25 % (v:v) methanol:ultrapure water. For calibration curves, charcoal-stripped human serum (Golden West Biologicals) was spiked with unlabeled steroid standards in methanol and diluted serially to final concentrations between 0.009 and 10 ng/ml in a 12-point curve, except for ALLO, which had final concentrations between 0.098 and 100 ng/ml in a 12-point curve. Next, spiked standards were subjected to the SLE+ extraction procedure. Using a Shimadzu SIL-30CAMP autosampler, 15 μl of each sample was injected onto a Raptor Biphenyl column (100 mm × 2.1 mm × 2.7 μm particle size; Restek, Bellefonte, PA, USA) with guard cartridge (5 mm × 2.1 mm × 2.7 μm particle size; Restek) held at 45 °C. Analytes were eluted (rate = 0.4 ml/min) with an eluent gradient initially consisting of 35 % mobile phase A (0.15 mM ammonium fluoride in water) and 65 % mobile phase B (pure methanol) that was linearly changed to 87 % B over 5.5 min and then to 100 % B over 0.1 min. The column was then washed with 100 % B for 1.8 min before being returned to 65 % B over 0.1 min and allowed to equilibrate for 3.4 min, for a total method time of 10.9 min. Retention times were: 17β-estradiol, 1.97 min; EE, 1.95 min; LNG 4.24 min; testosterone, 3.54 min; ALLO, 4.41 min; progesterone, 5.78 min. Steroids were detected via heated electrospray ionization in positive (LNG, testosterone, ALLO, progesterone) and negative (17β-estradiol, EE) ionization modes utilizing ultra-fast polarity switching and scheduled multiple reaction monitoring (MRM). The MRM transitions for ALLO were: 301.25 > 283.10 (quant), 301.25 > 135.15 (qual). MRM transitions for other steroids were as previously described ( Park et al., 2017 ; Blue et al., 2018 ). The assay range for ALLO was 0.200–100 ng/ml; all other assay ranges were as described previously ( Park et al., 2017 ; Blue et al., 2018 ). Data processing and analysis were performed using Lab-Solutions Software V5.97 (Shimadzu). Intra-assay variation was <7.1 % and assay accuracies were between 87.0 %–110.3 %. For LC-MS/MS analyses of brain tissue, frozen brains were weighed and homogenized in 1 ml PBS containing azide and isotope-labeled internal standards for steroids. Following homogenization, 3 ml ethyl ether was added to each tube. Tubes were mixed by shaking, centrifuged at 2000 × g for 5 min at 4 °C and snap-frozen in dry ice. Tubes were decanted, dried under forced air, and extracted steroids were reconstituted in 250 μl of 25 % (v:v) methanol:ultrapure water. Samples were then subjected to SLE+ extraction and LC-MS/MS analysis as described above. No brain or serum samples from the vehicle group had detectable amounts of EE and/or LNG. Thus, no samples from the vehicle-treated group were excluded due to HC contamination. Samples in which the analyte of interest was not detected were assigned a value of “0.” In samples where the analyte of interest was detectable but not quantifiable, the value used for that sample was the value for the limit of detection of the analyte of interest/√2, as per previous published work ( Handelsman and Ly, 2019 ). When >20 % of the samples in a group were detectable but not quantifiable, data are presented but were not statistically analyzed ( Jalabert et al., 2022 ). On PND57, rats ( n = 12/group) in cohort 3 received an i.p. injection of Euthasol and were transcardially perfused with 0.1 M PBS. Brains, adrenals, and ovaries were removed. Excess fat was trimmed from the adrenals and ovaries and all organs were weighed. Weight of one brain from the EE + LNG-treated group and one set of adrenals from the vehicle-treated group were not collected due to experimenter error. The hypothalamus and mPFC were microdissected and stored at −70 °C until RNA extraction. Additional brain regions were collected for future analyses. Due to space limitations on NanoString panels, mPFC and hypothalamus samples from a randomly selected subset of rats ( n = 6/group) in cohort 3 underwent RNA extraction. mPFC and hypothalamus samples were analyzed from the same rats. RNA was extracted using Qiazol and the Qiagen RNeasy Mini Kit (Hilden, Germany; #74106). RNA quality and NanoString nCounter (NanoString Technologies, Seattle, WA) analyses were conducted by the OSU Comprehensive Cancer Center Genomics Shared Resource Facility. RNA integrity number, equivalent for each sample, was 9.0–9.6. NanoString nCounter analysis was completed with a custom kit built for rats, which included 350 neuroendocrine, neuroimmune, neurotransmitter and neuro-plasticity related genes ( Suppl Table 1 ). Of the eight potential housekeeping genes assessed, Aars was selected since its expression significantly correlated to overall gene expression across samples. Aars expression did not differ across treatment groups. Behavioral testing was conducted after three weeks of HC treatment starting on PND57 rather than earlier during the treatment period, to assess how behaviors in late adolescence/early adulthood were shaped by exposure to HCs during the neurodevelopmental window of adolescence. Behavioral testing was run in two separate cohorts of rats. Different cohorts were used for behavioral tests to minimize time between the adolescent period and completion of behavioral assessments, and best match timing of physiological and transcriptional assessments. Rats ( n = 11–13/group) in cohort 4 performed three behavioral tasks on consecutive days in the following sequence: social preference test, novelty-induced hypophagia, forced swim test. Rats completed one behavioral assay per day. Behavioral assays took place between 0900 and 1500 h. Except for the novelty-induced hypophagia home cage assessment, sessions were video-recorded for later analysis. All behavior testing and analyses were done blind to condition. The social preference test was completed as previously described on PND57 or PND60 ( Breach et al., 2024 ). Briefly, rats were habituated to the testing room for 10 min, then placed into the center of a three-chamber apparatus (91.5 cm × 61 cm × 13.5 cm) and allowed to explore the center compartment for 5 min. Following habituation, the rat was placed into a transport cage while an age-, sex-, and strain-matched stimulus rat was secured underneath a wire cage in one of the side chambers. An empty wire cage was secured to the opposite side chamber. Chamber doors were then removed, and the experimental rat was returned to the center of the chamber and allowed to explore for 10 min. The location and identity of the stimulus rat was counterbalanced across groups. Time spent interacting (sniffing, touching) with the empty wire cage versus the cage containing the stimulus rat (rat cage) was quantified. A discrimination ratio [(time interacting with rat cage − time interacting with empty cage)/(time interacting with rat cage + time interacting with empty cage)] was calculated. Novelty-induced hypophagia was completed as in previous studies ( Bechtholt et al., 2007 ; Dulawa, 2009 ). On days 1–3 (PND55–57 or PND58–60), rats were given cereal rewards overnight in the home cage to prevent neophobia. On day 4, which occurred on PND58 or PND61, the home cage assessment was conducted in the vivarium to determine baseline ability and motivation to consume the cereal. On day 5 (PND59 or PND62), the test was conducted to determine if the stress of a novel, aversive environment diminished drive for the cereal reward. For the test, a plastic grid was placed on the bottom of an empty, clean cage under bright white light (700 lx) in a novel room. Latency to sample the cereal was recorded for the home cage assessment and test. Two vehicle-treated rats were accidentally run twice in the home cage assessment and were removed from all analyses. The forced swim test was used to assess stress-coping ( Haim et al., 2014 ). Plexiglas cylinders (diameter: 30.5 cm, height: 49 cm) were filled with 25 ± 0.5 °C water to a height of 33 cm. On PND60 or PND63, rats were habituated to the testing room for 10 min under bright white light, then placed into the cylinder and allowed to swim for 15 min while observed by an experimenter. The following day (PND 61 or PND64), the same procedure was repeated, but the rat was allowed to swim for 5 min. The time that the rat spent immobile (complete immobility or partial immobility with 1–2 legs moving underwater), duration to first immobility, time spent climbing (two front paws breaking the water surface), and time spent swimming (3–4 legs moving underwater) were quantified as were the number of dives (swimming straight down) and fecal boli. The first 5 min of the training day (Day 1) and the entirety of the 5-min test day (Day 2) were analyzed. On each day, water was changed between each rat. Rats ( n = 19/group) in cohort 5 were tested on the open field test and elevated plus maze on consecutive days. The open field test was completed on PND57 to assess exploratory and anxiety-like behavior ( Snyder et al., 2021 ). The open field consisted of an opaque Plexiglas arena (60 cm × 60 cm × 40 cm), with the floor divided by gridlines, resulting in 36 squares, 10 cm × 10 cm. Rats habituated to the testing room for 10 min under dim red light and then were placed into a corner of the open field. The rat was allowed to explore the open field for 10 min. Time spent in the center and periphery of the open field were quantified, as were entries into the center and gridlines crossed. Anxiety indices ( Cohen et al., 2013 ; Pestana et al., 2023 ) were calculated via the following formula: 1-{[(Time in center/duration of test) + entries into center/total gridlines crossed)]/2}. The elevated plus maze was completed on PND58 to assess exploratory and anxiety-like behavior ( Sabihi et al., 2017 ; Snyder et al., 2021 ). The elevated plus maze consisted of a plus sign-shaped maze (arms 50 cm long, 10 cm wide) placed on a 50 cm tall platform. Two arms opposite each other were enclosed by walls (closed arms), while the other two had no walls (open arms). Rats habituated to the testing room for 10 min under dim red light and were then placed into the center of the EPM facing an open arm. The rat was then allowed to explore the elevated plus maze for 10 min. Time spent, as well as entries into, the open and closed arms were quantified. Anxiety indices ( Cohen et al., 2013 ; Pestana et al., 2023 ) were calculated via the following formula: 1-{[(Time in open arms/duration of test) + entries into open arms/total arm entries)]/2}. Stretch attend postures (majority of body in one arm but stretching body forward/bending around corner) and head dips (downward movement of rat’s head toward the floor from the open arms) were counted as measures of risk assessment ( Macrì et al., 2002 ). One vehicle-treated and one EE + LNG-treated rat were excluded after falling off the elevated plus maze. Group data are reported as the mean ± SEM. Analyses were conducted using GraphPad Prism software version 10.1.0 (La Jolla, CA) with significance set at p ≤ 0.05, with exceptions noted below. Outliers were identified using the Grubb’s test, with alpha = 0.05. For assessment of estrous cycling data, a Chi-square test was used. Body weight, social preference test, forced swim test, and novelty-induced hypophagia data were analyzed by two-way ANOVA or mixed-effects analyses with repeated measures. Treatment and day (body weight, forced swim test, novelty-induced hypophagia) or cage (social preference test) were used as independent factors. When a significant interaction effect was found, planned post-hoc comparisons were calculated using Sidak’s multiple comparisons test (body weight) or Fisher’s LSD test (novelty-induced hypophagia). Normality was confirmed in R (2-way ANOVA or mixed effects model) or determined via the Shapiro-Wilk test ( t -tests). Normally-distributed data were analyzed using two-tailed unpaired t-tests, except when there was an a priori hypothesis, in which cases a one-tailed t-test was used. Welch’s correction was applied when group variances were unequal. For non-normal data, variances were first analyzed with Levene’s test ( https://www.statskingdom.com/230var_levenes.html ). If the homogeneity of variance assumption was met, non-normal data were analyzed using the Mann Whitney test. If variances were unequal, non-normal data were log transformed and analyzed with a two-tailed unpaired t-test with Welch’s correction. For all significant effects, effect size was calculated using online resources (Cohen’s d, https://www.socscistatistics.com/effectsize/default3.aspx ) or R (partial η 2 ). NanoString data were normalized and differential expression testing was performed using DESeq2 in R ( Love et al., 2014 ). In brief, EstimateSizeFactors in DeSeq2 was used for normalization to the housekeeping gene, Aars . The DeSeq design parameter controlled for experimental group (four levels: PFC VEH, HYP VEH, PFC EE + LNG, HYP EE + LNG). Results tables were generated for pairwise comparisons of interest (e.g., PFC EE + LNG vs PFC VEH, HYP EE + LNG vs HYP VEH). Data are expressed as log2FC of vehicle-treated rats and uncorrected p -values were reported ( Jung and Sohn, 2014 ; Witcher et al., 2021 ; Tapp et al., 2022 ).

None of the endpoints assessed were confounded by differences in pubertal onset, as day of VO was similar between groups (Mean Vehicle = 33.37 ± 0.18, Mean EE+LNG = 33.46 ± 0.14, p > 0.05). Several validation measures confirmed the effectiveness of the HC treatment paradigm ( Fig. 2A , B ). Vaginal cytology assessment of the estrous cycle demonstrated that most vehicle-treated rats were cycling, whereas most rats treated with EE + LNG were not cycling (χ 2 (2) = 93.0, p < 0.001) ( Fig. 2A ). As HCs can suppress the HPG axis, concentrations of LH were assessed using ELISA. In contrast to vehicle treated rats, serum LH levels in all EE + LNG-treated rats were undetectable (t(3) = 6.66, p < 0.01, Cohen’s d = 4.70) ( Fig. 2B ). Body weight and weights for several endocrine organs were assessed to determine if HCs had systemic effects ( Fig. 2C – F ). For body weight, there was a main effect of treatment (F(1, 106) = 72.93, p < 0.0001, partial η 2 = 0.42) and a main effect of time (F(1.98, 209.2) = 1658, p < 0.001, partial η 2 = 0.87) as well a treatment × time interaction (F(21, 2224) = 159.0, p < 0.001, Cohen’s d = 1.36) ( Fig. 2C ). Post-hoc analyses showed that EE + LNG-treated rats gained less weight during adolescence compared to vehicle-treated rats beginning on PND38 ( p ’s < 0.01). Since EE + LNG treatment significantly decreased body weight, endocrine organ weights were corrected for body weight (relative organ weight = organ weight/body weight). Relative ovarian weights of EE + LNG-treated rats were reduced (t(14.66) = 2.10, p ≤ 0.05, Cohen’s d = 0.87) ( Fig. 2D ), while relative brain (t(21) = 2.32, p < 0.05, Cohen’s d = 0.96) ( Fig. 2E ) and adrenal (t(20) = 3.75, p < 0.05, Cohen’s d = 1.61) ( Fig. 2F ) weights were increased in EE + LNG-treated rats. LC-MS/MS was used to measure HCs in serum and the brain following adolescent HC administration, as well as to determine the effects of HCs on serum and brain 17β-estradiol, progesterone, testosterone, and ALLO concentrations ( Fig. 3A – L ). EE (t(4) = 8.87, p < 0.001, Cohen’s d = 5.61) ( Fig. 3A ) and LNG (t(4) = 9.81, p 20 % of vehicle-treated rats had levels that were detectable but not quantifiable ( Fig. 3C ). HC treatment did not significantly affect serum progesterone ( p > 0.05) ( Fig. 3D ). However, HCs significantly decreased ALLO (t(6.96) = 3.52, p < 0.01, Cohen’s d = 1.90) ( Fig. 3E ) and testosterone (U = 0, Median Vehicle = 0.042, Median EE+LNG = 0.006 p < 0.01, Cohen’s d = 2.92) ( Fig. 3F ) in serum. In the brain, EE (t(3) = 10.28, p ≤ 0.001, Cohen’s d = 7.27) ( Fig. 3G ) and LNG (U = 0, Median Vehicle = 0, Median EE+LNG = 0.65, p ≤ 0.001, Cohen’s d = 5.39) ( Fig. 3H ), were detectable only in HC-treated rats. As in serum, we were unable to detect if HC treatment significantly affected brain 17β-estradiol, as >20 % of vehicle-treated rats had levels that were detectable but not quantifiable ( Fig. 3I ). HC treatment did not affect progesterone ( p > 0.05) ( Fig. 3J ). However, ALLO (t(6.80) = 2.00, p < 0.05, Cohen’s d = 1.08) ( Fig. 3K ) and testosterone (t(9) = 2.14, p < 0.05, Cohen’s d = 1.41) ( Fig. 3L ) were decreased in the brains of rats treated with HCs. NanoString nCounter analysis was completed to assess transcriptional changes associated with adolescent HC administration ( Fig. 4 ; Suppl Table 1 ). Analysis was performed in the hypothalamus and mPFC, as they are brain regions that are sensitive to changes in ovarian hormones and that mature during adolescence ( Sisk and Foster, 2004 ; Herbison, 2016 ; Juraska, 2024 ). In the hypothalamus, fold change expression of Cldn5, Asb7, Htr1a, Gabra4 , and Cxcl1 were significantly increased in EE + LNG-treated rats, while fold change of expression of Ocln, Cspg4, Bax, Ccl1, Itgax, Ifna2, Mapk10, Il18, Nos3, and Esr2 were decreased ( p < 0.05) ( Fig. 4A , B ; Suppl Table 2 ). In the mPFC, fold change expression of Gphn, Homer1, Drd1, Ccr8, Tgfbr2, Gba , and Tlr11 were significantly increased in EE + LNG-treated rats, while fold change of expression of Igf1r, Tnfrsf12a, Cd69, Ldha, Oxtr, S100b , and Aldh1a1 were decreased ( p < 0.05) ( Fig. 4C , D ; Suppl Table 3 ). A cohort of rats completed the social preference test, novelty-induced hypophagia, and forced swim test to assess functional effects of adolescent HC administration. Though there was a social preference in the social preference test, as indicated by significantly more time spent with a stimulus rat compared to an empty cage (main effect of stimulus: F(1,22) = 45.28, p 0.05, Suppl Fig. 1B ). Stress coping behavior in the forced swim test, including time spent immobile and time spent swimming, was not impacted by HC administration ( Suppl Table 4 ). There was a main effect of day on select behaviors within the forced swim test ( Suppl Table 4 ). Specifically, time immobile increased, while time climbing and number of dives decreased, from Day 1 to Day 2. These changes across days indicate that overall, the FST procedure was effective in inducing a passive coping response, but the lack of differences across treatment groups indicate that HCs do not influence despair-like behavior. In the novelty-induced hypophagia task ( Fig. 5 ), there were main effects of treatment (F (1,20) = 28.72, p < 0.001, partial η 2 = 0.32) and testing environment (F (1, 19) = 20.11, p < 0.001, partial η 2 = 0.19) as well as a significant interaction between treatment and testing environment (F(1,19) = 21.28, p < 0.001, Cohen’s d = 1.81). In vehicle-treated rats, the latency to sample a reward in the novel cage was longer than in the home cage ( p 0.05). The latency to sample the reward in the home cage was similar for vehicle and HC-treated rats ( p > 0.05). A separate cohort of rats was tested on the open field test and elevated plus maze ( Fig. 6 ). In the open field test, time in the periphery was increased in EE + LNG-treated rats compared to those treated with vehicle (t(26.51) = 2.13, p < 0.05, Cohen’s d = 0.70) ( Fig. 6A ), while time in the center was decreased (t(26.39) = 2.23, p < 0.05, Cohen’s d = 0.74) ( Fig. 6B ). Additionally, rats treated with EE + LNG entered the center significantly fewer times (t(36) = 4.76, p < 0.001, Cohen’s d = 1.54) ( Fig. 6C ) and crossed fewer gridlines compared to vehicle-treated rats (t(36) = 5.55, p < 0.001, Cohen’s d = 1.80) ( Fig. 6D ). EE + LNG-treated rats had a significantly higher anxiety index (t(27.97) = 2.25, p < 0.05, Cohen’s d = 0.74) ( Fig. 6E ). In the elevated plus maze, rats treated with EE + LNG spent more time in the open arms (t(34) = 2.40, p < 0.05, Cohen’s d = 0.80) ( Fig. 6F ), though there was no significant difference in time spent in the closed arms ( p = 0.07) ( Fig. 6H ). EE + LNG treatment tended to diminish entries into the open arms ( p = 0.07) ( Fig. 6G ) and decreased entries to the closed arms (t(34) = 5.29, p < 0.001, Cohen’s d = 1.76) ( Fig. 6I ). EE + LNG-treated rats had a significantly lower anxiety index (t(34) = 2.56, p < 0.05, Cohen’s d = 0.85) ( Fig. 6J ). Rats treated with EE + LNG also showed significantly fewer stretch-attend postures compared to vehicle-treated rats (t(32.87) = 3.72, p 0.05) ( Fig. 6L ).

These studies are the first to validate a rat model of adolescent HC exposure, in addition to showing that this regimen of HC administration to intact, adolescent female rats affects the brain, body, and behavior. The neurodevelopmental effects of HC administration during this critical period of development are largely unknown ( Cahill, 2018 ). Thus, an animal model of adolescent HC administration will help define the short- and long-term effects of HC use on the brain and behavior as well as facilitate investigation of the molecular underpinnings contributing to such changes ( Lacasse et al., 2022 ; Hampson, 2023 ). Ultimately, increased knowledge about how HCs impact the adolescent brain may lead to more informed use and provide people with agency over their care ( Petersen et al., 2023 ).

These studies are the first to validate a rodent model of adolescent HC administration and assess its effects on the brain and behavior. Chronic treatment with EE and LNG, two synthetic steroids commonly used in HCs prescribed to adolescents ( Braverman et al., 2014 ; De Leo et al., 2016 ), disrupted HPG axis function and both HCs were detectable in the serum and brain. Adolescent HC administration also affected expression of a variety of genes in the hypothalamus and mPFC, indicating transcriptional effects in the brain. Finally, adolescent HC administration altered behavior in the novelty-induced hypophagia paradigm, open field test, and elevated plus maze. Together, these data point to effects of HCs on the brain and behavior during the critical neurodevelopmental period of adolescence. Evidence in humans indicates that HC use during adolescence is widespread (Daniels et al., 2020; Abma and Martinez, 2023 ) and may differentially affect the brain and behavior compared to use during adulthood ( Skovlund et al., 2016 , 2018 ; Sharma et al., 2020a , 2020b ). To date, rodent studies examining the effects of HCs on the brain and behavior have exclusively used adults ( Porcu et al., 2019 ; Lacasse et al., 2022 ; Tronson and Schuh, 2022 ). Thus, the first objective of this paper was to validate a HC administration paradigm in adolescent rats, allowing for the study of the potential effects of adolescent HC exposure on the developing brain and behavior. We confirmed that adolescent HC treatment disrupted estrous cycling. We also found that HC administration reduced peripheral LH levels and decreased relative ovarian weight. Reduced ovarian weight is typically associated with diminished ovarian hormone output and HPG axis activity ( Simone et al., 2015 ; Allaway et al., 2021 ; Lacasse et al., 2022 ). Thus, our paradigm of HC administration during adolescence effectively interrupted HPG axis activity, which is consistent with what is frequently observed in HC users ( Armstrong and Kennedy, 1972 ; Ishikawa, 1992 ; Paccola et al., 2013 ; Hampson, 2020 ; Gavina et al., 2023 ). However, we cannot eliminate the possibility that, because we administered HCs at a time when the HPG axis is still maturing in rats, the timing of our HC regimen may have affected its development ( Vetter-O’Hagen and Spear, 2012 ; Picut et al., 2015 ). As an additional way to validate our HC administration paradigm, LC-MS/MS was used to assess concentrations of HCs. We showed that EE and LNG were detectable in serum and the brain following adolescent HC administration. Estradiol and progesterone receptors are expressed in the brain and, given that synthetic hormones in HCs have a high binding affinity for their cognate receptors ( Brinton et al., 2008 ; Morissette et al., 2008 ; Weiser et al., 2008 ; Lacasse et al., 2022 ), our data suggest that HCs may have direct central actions to affect brain function and behavior. We also used LC-MS/MS to assess serum levels of endogenous sex steroid hormones as well as the neuroactive progesterone metabolite, ALLO, all of which have been shown to be suppressed by HC administration in adult rats ( Porcu et al., 2012 ; Santoru et al., 2014 ) and in many human HC users ( Zimmerman et al., 2014 ; Hampson et al., 2020; Pletzer et al., 2023 ). These analyses revealed that adolescent HCs decreased serum and brain ALLO and testosterone, but not progesterone. That HCs did not reduce endogenous progesterone was unexpected, as the synthetic progestin used in our studies, LNG, inhibits basal and LH-stimulated progesterone synthesis, as well as the conversion of pregnenolone to progesterone in cultured rat luteal cells ( Tellería et al., 1994 ). One possible explanation is that the doses of EE and LNG used here, as well as the EE:LNG ratio, were lower than what has been used in previous studies in adult female rats that have shown suppressed endogenous progesterone in the brain and plasma with HC treatment ( Porcu et al., 2012 ; Lacasse et al., 2022 ). The effects of HCs on 17β-estradiol in the serum and, for the first time, in the brain were also assessed. We were unable to run statistical comparisons on these data as some samples were not quantifiable, which is a technical limitation that has been previously documented in the literature ( Taves et al., 2011 ). However, low doses of HCs that we used here were previously shown to not alter serum estradiol levels in adult rats ( Simone et al., 2015 ) although higher doses were effective ( Porcu et al., 2019 ). In the brain, it is also worth considering that there are regional differences in the concentration of 17β-estradiol ( Taves et al., 2011 ). Thus, using an entire hemisphere in our analyses may have diluted concentrations of 17β-estradiol, thereby creating a floor effect in our data. Adolescent HC administration affected other physiological measures, including body weight, which was reduced by HCs. Reduced body weight has been observed in adult rats that have undergone a similar HC regimen and can be taken as additional evidence that HCs are exerting a physiological effect ( Lacasse et al., 2022 , 2023 ). Adolescent HC administration also increased relative adrenal weight, which occurs with hypothalamic-pituitary-adrenal (HPA) axis dysregulation in rodents ( Herman et al., 2005 ; Ulrich-Lai et al., 2006 ). Therefore, adrenal hypertrophy following adolescent HC administration may reflect perturbation of the HPA axis, an effect also seen in adult users of HCs as well as adult female rats administered HCs ( Tronson and Schuh, 2022 ). Lastly, adolescent HC administration increased relative brain weight. The brain undergoes protracted maturation during adolescence ( Fuhrmann et al., 2015 ; Juraska, 2024 ) and thus, gross weight changes may reflect shifts in typical adolescent development by HCs. As an initial step to elucidate potential neurodevelopmental processes HCs could be impacting, we used NanoString nCounter assays to investigate gene expression profiles in two brain regions, the hypothalamus and mPFC. The hypothalamus was analyzed since HCs can alter endocrine activity via the HPG axis ( Kjeld et al., 1976 ; Lacasse et al., 2022 ). In the hypothalamus, adolescent HC administration altered fold change expression of genes related to the blood brain barrier ( Cldn5 , Ocln ), in addition to monoamine ( Htr1a ), GABAergic ( Gabra4, Cspg4 ), cell-cell ( Cxcl1, Bax, Ccl1 ), neuroimmune ( Itgax, Ifna2, Mapk10, Il18, Nos3 ), and hormone ( Esr2 ) signaling. We also examined the mPFC, development of which continues over adolescence and is influenced by the same ovarian hormones HCs disrupt ( Delevich et al., 2021 ; Juraska, 2024 ). In the mPFC, fold change expression of genes associated with synapses ( Gphn, Homer1 ), monoamine ( Drd1 ) and neuroimmune ( Ccr8 , Tgfbr2 , Gba , Tlr11 , Igf1r, Tnfsf12a , Cd69 ) signaling, cell metabolism ( Ldha ), hormone signaling ( OXTR ), and glia ( Sb100, Aldh1a1 ), were differentially affected by HC administration. Thus, although both brain regions are hormone sensitive, there were regional differences in the genes impacted by HCs. It remains to be determined whether HCs act via different mechanisms (i.e. suppression of endogenous hormones or acting as hormones themselves by binding to hormone receptors) to exert effects dependent on brain region and gene of interest. Relatively few studies have examined transcriptional effects of HC administration in adult rodent models. Of the animal studies that have assessed the same transcripts that we found showed changes in expression after adolescent HC administration, adult female rats did not show HC-induced changes in in cortical or hippocampal expression of Gphn ( Sassoè-Pognetto et al., 2007 ) or protein levels of the α4 subunit of the GABA A receptor which is encoded by the gene, Gabra4 ( Porcu et al., 2012 ). Thus, increased expression of these genes observed in our study may be unique to adolescent HC treatment. Conversely, previous studies of adult HC administration in rodent models have found increased expression of other GABA receptor genes, such as Gabrg2 (which encodes the γ2 subunit of the GABA A receptor), in the cortex and hippocampus ( Follesa et al., 2002 ; Porcu et al., 2012 ), and we did not observe changes in this transcript in either the mPFC or the hypothalamus. This further supports the notion that adolescent versus adult HC administration elicits unique transcriptional effects in the brain. The extent to which the other genes up- or down-regulated by adolescent exposure to HCs also occur in adults remains to be determined, as most of the transcripts we found to be differentially expressed in our study have not been assessed after HC administration in adult models. Nonetheless, the observation that genes across biological categories were impacted in a region-specific manner suggests that adolescent HC treatment may differentially influence selective aspects of neurodevelopment of the hypothalamus and mPFC, including myelination, the maturation of excitatory-inhibitory balance, or neuromodulatory and/or neuroendocrine function in these brain regions ( Schulz and Sisk, 2016 ; Delevich et al., 2021 ; Juraska, 2024 ). To evaluate functional effects of adolescent HC exposure, we assessed behaviors known to be sensitive to HC exposure in adult rodents and/or dynamic over adolescence ( Macrì et al., 2002 ; Simone et al., 2015 ; Porcu et al., 2019 ; Schuh et al., 2024 ). Using the social preference test, we found no effects of adolescent HC treatment on sociability. Other studies have shown that HC administration impairs social behavior in adult rats ( Santoru et al., 2014 ) and decreases prosocial behavior in contraceptive pill users ( Strojny et al., 2021 ). While seemingly contradictory to the present results, these data could point to age-specific effects of HCs on social behavior, although additional behavioral tasks are needed to determine if adolescent HC administration affects other aspects of social behavior, particularly those requiring reciprocal social interactions not assessed in the social preference test. We also did not observe significant effects of HC treatment on behavior in the forced swim test. This finding is consistent with some ( Schuh et al., 2024 ) but not all ( Santoru et al., 2014 ) studies in adult rodents treated with HCs, highlighting the need for more work on HCs and behaviors of relevance to stress coping and depression. In humans, HC use has been linked to depression (i.e., higher incidence of depression diagnosis, antidepressant use and suicide), particularly in adolescents and users of progestin-only HCs ( Skovlund et al., 2016 , 2018 ). Thus, assessing adolescent administration of progestin-only HCs on depressive-like behaviors is an important next step for animal research. A behavioral test that showed sensitivity to adolescent HCs was novelty-induced hypophagia, where HC-treated rats rapidly consumed the cereal reward regardless of whether it was presented in a familiar or novel environment. This result likely does not reflect differences in hunger, because the HC-treated rats were similar to vehicle-treated rats in the latency to consume the cereal reward during a home cage test. Instead, these data could point to an effect of HCs increasing motivation for a reward, as the drive to overcome the novel, aversive environment to obtain a reward may be higher in rats treated with HCs versus those treated with vehicle. This explanation aligns with work in humans which found increased sensitivity to monetary reward in HC users ( Bonenberger et al., 2013 ). The novelty-induced hypophagia finding may also indicate that HCs reduce risk assessment and threat appraisal since a novel environment, which is typically perceived as risky or threatening, did not alter latency to pursue the reward in HC-treated rats. This interpretation is consistent with the reduction in stretch-attend postures displayed by HC-treated rats in the elevated plus maze, which is a common risk assessment behavior in rodents ( Macrì et al., 2002 ). Risk assessment and threat appraisal both develop over adolescence ( Macrì et al., 2002 ; Kredlow et al., 2022 ), thus it is possible that HC administration during adolescence may disrupt or shift the typical maturational trajectory of these behaviors. The reduced latency to sample the cereal treat in a novel environment in the novelty-induced hypophagia paradigm that we observed in HC-treated rats could also reflect decreased anxiety-like behavior ( Bechtholt et al., 2007 ). Thus, we completed additional tests of anxiety-like behavior and found that HCs had differential effects dependent on behavioral assay. In the open field test, HC-treated rats spent more time in the periphery, consistent with an increase in anxiety-like behavior. In the elevated plus maze, HC-treated rats spent more time in the open arms, indicating deceased anxiety-like behavior. In the elevated plus maze, both increases and decreases in anxiety-like behavior have been reported in adult rats treated with HCs dependent on dose, with the lower doses as used here having a similar anxiolytic effect ( Simone et al., 2015 ), suggesting this may be a feature of HC use that is independent of age. A similar consensus cannot be reached for the open field test, as recent work in adult mice found no effect of HCs on anxiety-like behavior ( Schuh et al., 2024 ). Of note, we showed that adolescent HC administration also decreased locomotion (i.e. fewer gridlines crossed in the open field test, fewer closed arm entries in the elevated plus maze). However, even after correcting for locomotor differences by calculating anxiety indices ( Cohen et al., 2013 ; Pestana et al., 2023 ), anxiety-like behavior remained increased in the open field test and decreased in the elevated plus maze. These behavioral assays assess different subdomains of anxiety, such that the elevated plus maze has been used as a measure of state anxiety, while the open field test has been used as a measure of trait anxiety ( Walf and Frye, 2007 ; Snyder et al., 2021 ). Thus, HCs in adolescence may increase trait anxiety while decreasing state anxiety. In humans, evidence regarding HCs and anxiety is limited, although some studies have shown that HC use is associated with the onset of anxiety symptoms ( Noachtar et al., 2023 ; Kowalczyk et al., 2024 ). Whether HCs differentially regulate subdomains of anxiety among human users of HCs warrants further study ( Labad et al., 2006 ; Vulink et al., 2006 ; Graham and Milad, 2013 ; Masama et al., 2022 ; Kowalczyk et al., 2024 ). Future work is needed to address several limitations of the current studies as well as follow up on and extend the current results. First, we employed a commonly used HC regimen for rodent studies ( Simone et al., 2015 ; Lacasse et al., 2023 ), yet HCs are a highly heterogeneous class of drugs ( Lacasse et al., 2022 ; Tronson and Schuh, 2022 ). Thus, assessments of other doses, different progestin types, and routes of administration are necessary to more fully elucidate the effects of adolescent HC administration ( Lacasse et al., 2024 ). It will also be important to determine the mechanisms through which adolescent HC administration affects the brain and behavior, including the relative contribution of estrogen and/or progesterone signaling as well as whether the neurobehavioral effects of adolescent HCs are via direct modulation by exogenous ligands (which we determined here are entering the brain) and/or via suppression of endogenous hormones. Additional studies are also needed to uncover the functional significance of the transcriptional changes we observed in response to adolescent HC administration, with the differentially expressed genes guiding particular attention toward certain processes and mechanisms of brain maturation that are known to occur in adolescence, such as neuroimmune mediated circuit refinement, myelination, and the maturation of excitatory-inhibitory balance. Lastly, future studies should assess whether the effects of adolescent HC administration are organizational, producing permanent effects, or activational and thus temporary. Toward this end, experiments assessing how adolescent HC administration affects endpoints later in adulthood following cessation of HC administration are needed, as in this study we only examined behavior in late adolescence/early adulthood while rats were still being treated with HCs. In humans, there is some evidence indicating that neurobehavioral effects of adolescent HC use may persist into adulthood, whether use is consistent or not ( Anderl et al., 2020 ). This limited data is suggestive of organizational effects, but future experiments are needed to determine if adolescent HC exposure exerts permanent effects on the developing brain, if adolescence is a critical period during which HC exposure may alter neurodevelopmental trajectories, and what the duration of such a critical period may be.

Hormonal contraceptives (HCs) have been available for over 60 years ( Connell, 1999 ; Tyrer, 1999 ) and are used by 150 million women worldwide ( Haakenstad et al., 2022 ). HCs are a popular contraceptive choice given that they are highly effective, reliable, and reversible ( Teal and Edelman, 2021 ). These qualities are due to their mechanism of action, as HCs are potent synthetic exogenous hormones that exert negative feedback on the hypothalamic-pituitary-gonadal (HPG) axis to prevent ovulation, diminish endogenous ovarian hormone production, and suppress hormonal fluctuations across the menstrual cycle ( Kjeld et al., 1976 ; Blue et al., 2018 ; Hampson, 2020 ; Lacasse et al., 2022 ). Aside from contraception, HCs are also prescribed for non-contraceptive, therapeutic reasons including to diminish symptoms of polycystic ovarian syndrome and endometriosis as well as to treat irregular cycles, heavy menstrual bleeding, menstrual period pain, and acne ( Maguire and Westhoff, 2011 ; Schrager et al., 2020 ). HC use is often initiated during adolescence and use among adolescents is widespread ( Daniels and Abma, 2020 ; Todd and Black, 2020 ). Indeed, approximately 25 % of all women aged 15–19 in the United States alone use HCs ( Daniels and Abma, 2020 ; Todd and Black, 2020 ; Abma and Martinez, 2023 ). Although HCs are considered safe for adolescents ( Braverman et al., 2014 ; De Leo et al., 2016 ), it is important to consider that adolescence is a critical period of ongoing brain and behavioral maturation mediated in part by the same ovarian hormones (e.g. estrogens and progesterone) that HCs disrupt ( Fuhrmann et al., 2015 ; Herting and Sowell, 2017 ; Koolschijn et al., 2014 ; Schulz and Sisk, 2016 ; Delevich et al., 2021 ; Juraska, 2024 ). Therefore, it stands to reason that HC use during adolescence may impact neurobehavioral development ( Pletzer and Kerschbaum, 2014 ; Cahill, 2018 ; Taylor et al., 2021 ; Beltz, 2022 ). A limited number of human studies suggest that HCs during adolescence may differentially affect the brain and behavior compared to use during adulthood ( Skovlund et al., 2016 , 2018 ; Sharma et al., 2020a , 2020b ). However, human studies are complicated by individual differences in a variety of factors such as experience, genetics, HC formulation and dose, and duration of use ( Heller et al., 2022 ; Tronson and Schuh, 2022 ; Hill and Mengelkoch, 2023 ). These variables are challenging to effectively or ethically control and manipulate in human participants, but these hypothesized sources of between-user heterogeneity can be manipulated and controlled in rodent models. Rodent models can also be leveraged to understand the effects of adolescent HC exposure on the brain at a molecular level of resolution not currently feasible in humans. While there are some differences between the rodent estrous cycle and the human menstrual cycle (e.g. length of cycle), the overall hormonal profile is similar and both are disrupted by HC exposure ( Hampson, 2020 ; Kundakovic and Rocks, 2022 ; Tronson and Schuh, 2022 ). Thus, rodent models are effective tools for studying the neurobehavioral effects of HC use. Notably, all studies to date investigating the brain and behavioral impact of HCs in rodent models have exclusively used adult females (for review see Concas et al., 2022 ; Hilz, 2022 ; Song et al., 2023 ; Lacasse et al., 2024 ), thus the field is lagging in knowledge about HC effects on the adolescent brain. Here we aimed to develop a rodent model of adolescent HC exposure and validate its efficacy via assessments of HPG axis functioning. We also determined the effects HC exposure during adolescence on a selection of genes both implicated in and affected by neurodevelopmental maturation, as well as a range of behaviors that develop over adolescence including sociality, emotionality, and risk assessment ( Macrì et al., 2002 ; Woodward et al., 2023 ; Juraska, 2024 ).