Changes in the Bile Acid Pool and Timing of Female Puberty: Potential Novel Role of Hypothalamic TGR5

Abstract

Context

The regulation of pubertal timing and reproductive axis maturation is influenced by a myriad of physiologic and environmental inputs yet remains incompletely understood.

Objective

To contrast differences in bile acid isoform profiles across defined stages of reproductive maturity in humans and a rat model of puberty and to characterize the role of bile acid signaling via hypothalamic expression of bile acid receptor populations in the rodent model.

Methods

Secondary analysis and pilot studies of clinical cohorts, rodent models, ex vivo analyses of rodent hypothalamic tissues. Bile acid concentrations is the main outcome measure.

Results

Lower circulatory conjugated:deconjugated bile acid concentrations and higher total secondary bile acids were observed in postmenarcheal vs pre–/early pubertal adolescents, with similar shifts observed in infantile (postnatal day [PN]14) vs early juvenile (PN21) rats alongside increased tgr5 receptor mRNA expression within the mediobasal hypothalamus of female rats. 16S rRNA gene sequencing of the rodent gut microbiome across postnatal life revealed changes in the gut microbial composition predicted to have bile salt hydrolase activity, which was observed in parallel with the increased deconjugated and increased concentrations of secondary bile acids. We show that TGR5-stimulated GnRH release from hypothalamic explants is mediated through kisspeptin receptors and that early overexpression of human-TGR5 within the arcuate nucleus accelerates pubertal onset in female rats.

Conclusion

Bile acid isoform shifts along stages of reproductive maturation are conserved across rodents and humans, with preclinical models providing mechanistic insight for the neuroendocrine-hepatic-gut microbiome axis as a potential moderator of pubertal timing in females.

Pubertal timing is influenced by a myriad of physiologic and environmental inputs that integrate the (re)awakening of the neuroendocrine axis in adolescence. Whereas obesity is associated with earlier pubertal timing (1), conditions such as inflammatory bowel disease are associated with delayed reproductive maturity (2). Body weight and or body composition is central to signaling the initiation and regulation of reproductive maturation. However, how metabolic cues are integrated by the neuroendocrine system to sense the metabolic environment and initiate (or inhibit) reproductive maturation is not fully understood. Previous formative studies have identified anthropometric benchmarks that are believed to reflect a metabolic economy sufficient for reproductive maturation such as a minimum body weight (3) or critical threshold of body fat percent (4). Mechanistically, several signaling pathways have been proposed to underlie these benchmarks, including permissive cues from essential fatty acids (5), insulin-like growth factor 1 (IGF-1) (6-8), leptin (9-15), and inhibitory cues from ghrelin (12, 16) and epigenetic/transcriptional regulation of upstream controllers of gonadotropin-releasing hormone (GnRH) neuron pulsatility (17-21). These known peripheral metabolic hormones which influence the timing of puberty are mediated largely by arcuate nucleus (ARC) kisspeptin, neurokinin B, and dynorphin (KNDy) neurons. Although shifts in neuroendocrine activity during the pubertal transition are highly correlated with systemic metabolic status (22, 23), the discrete metabolic mechanisms within the neuroendocrine axis that control neuroendocrine maturation are not completely understood. As a result, this continued uncertainty limits effective therapeutic strategies to manage altered timing of reproductive transitions due to metabolic influences such as obesity or undernutrition.

Evidence has emerged over the last decade suggesting a physiologically important fraction of bile acid isoforms can enter systemic circulation where they serve an endocrine function at sites distal to the gut. Primary bile acids, cholic acid (CA) and chenodeoxycholic acid (CDCA), are synthesized in the liver and conjugated with glycine (G) or taurine (T) to produce T(G)CA and T(G)CDCA, requisite for lipid digestion (24, 25). These molecules are deconjugated by bile salt hydrolases (BSH) of mainly gram-positive gut microbiota, which is an essential first step in the synthesis of secondary bile acids such as lithocholic acid (LCA) and deoxycholic acid (DCA) (26, 27). Consequently, species diversity in the gut microbiome regulates bile acid metabolism and isoform composition in humans (25, 26). These bile acids exert their actions via the nuclear receptor farnesoid X receptor (FXR) and the G-protein coupled bile acid receptor, Gpbar-1/TGR5 (28, 29). Bile acid/TGR5 signaling provides a “sensing mechanism” for the metabolic state (fasted or fed state) in cells and tissues expressing bile acid receptors (reviewed recently elsewhere) (30). Several groups have established a role for bile acid/TG5R in thermogenesis, insulin signaling, incretin regulation, inflammation, and lipid and glucose homeostasis (31-33) and more recently as a possible mechanism contributing to the pathophysiology of polycystic ovary syndrome (PCOS) (34-36). Therefore, hepatic and gut microbial regulation of bile acid metabolism may represent a mechanism by which environmental factors and the gut microbiome influence host metabolic and reproductive health. However, the role of bile acid metabolism and tissue-specific expression of bile acid receptor populations during the pubertal transition are not clear.

Increases in both the total bile acid pool and shifts in bile acid isoform diversity concurrent with advancing reproductive maturation have been reported in rodent (37) and human studies (38-40). Children with intestinal resection and/or bile acid malabsorption were only at risk for lithogenic bile production, and subsequent gall stone formation, after puberty (39), which suggests that there may be essential changes in hepatic physiology and bile acid metabolism that occur during pubertal maturation. Collectively, these studies describe dynamic changes in the gut microbiome and bile acid isoforms from infancy through reproductive maturity, which have been recently reviewed (41). Whether dynamic changes in bile acid isoforms are consequence of reproductive maturation or play a direct role in facilitating the reawakening and maturation of the reproductive axis is unknown. We hypothesize that hepatic and gut microbial regulation of bile acid metabolism and isoform composition in circulation represents a mechanism through which environmental factors influence reproductive maturation during the pubertal transition.

The studies described herein reveal marked changes in the composition of the circulating bile acid pool in pre- or early puberty vs postmenarcheal adolescent females and in a rat model of pubertal maturation. We report parallel changes in species composition within the rodent gut microbiome associated with BSH activity, the primary role of which is deconjugation leading to increased secondary bile acids in circulation. These shifts precede a significant increase in tgr5 receptor mRNA expression within the mediobasal hypothalamus in female rats. We show that the mediobasal hypothalamus can respond directly to TGR5 stimulation to control GnRH release from hypothalamic explants in culture and that early overexpression of human TGR5 within the arcuate nucleus accelerates pubertal onset in female rats. Collectively, our studies provide novel evidence for the neuroendocrine-hepatic-gut microbiome axis as a potential moderator of pubertal timing in females.

Materials and Methods

Human Samples

Prepubertal and early pubertal samples were obtained from a Biospecimen Repository Precision for Medicine (Norton, MA; IRB Exempt, Cornell University, Ithaca, NY). Inclusion criteria for selection were ≤ 9 years of age, female, healthy. Although menarche or pubertal status was not available, age of selection was intended to select pre- or early pubertal females (42, 43). Early postmenarcheal samples were obtained from fasting adolescents ≤ 13 months postmenarche (range, 0.45-1.02 years postmenarche) who were participating in a prospective longitudinal study (Children's Mercy Kansas City, Kansas City, MO and Cornell University, Ithaca, NY; IRB# 00000779; Clinicaltrials.gov NCT04424576). Postmenarcheal samples were collected in the follicular phase (within 10 days of menses onset). Nonfasting adult samples were obtained from healthy participants >18 years of age (range, 19-38), with regular menstrual cycles (21-35 days) in the early follicular phase (1-9 days after menses). Adult participants were enrolled in a prospective study at Cornell University (IRB# 0908000633; Clinicaltrials.gov NCT01927432) between 2009 and 2010. Nonfasting plasma samples from a second, independent study were obtained from a recently completed pilot study of contrasting psychological symptoms in late pubertal/premenarcheal vs early postmenarcheal adolescents in order to further replicate our findings using more well-defined and comparable cohorts. Study and participant descriptions are available in Supplementary Methods (44). Across all cohorts, participants were healthy and not taking any medications known or suspected to interfere with reproductive or metabolic function. Sera (pilot study) and plasma (second cohort) were stored at −80 °C before being shipped on dry ice to the Columbia Biomarkers Core Laboratory (Irving Institute for Clinical and Translational Research, Columbia University Irving Medical Center, New York, New York) for analyses.

Animal Care and Use

All rat and mouse studies were conducted at Oregon Health and Sciences University in compliance with the Oregon Health and Sciences University Institutional Animal Care and Use Committee.

Rats

We used female Sprague Dawley rats obtained from Charles River Laboratories International, Inc. (Hollister, CA), randomly assigned to different experimental groups, and housed in a room with controlled photoperiod (12/12 hours light/dark cycle) and temperature (23-25 °C), with ad libitum access to tap water and pelleted rat chow (Purina Cat# 5012). At birth, different litters were mixed and distributed 12 pups per dam. All animals were weaned at the same age (postnatal day [PN]21) to avoid any potential effect of the dietary change on plasmatic bile acid profiles and pubertal development.

Mice

Female mice expressing enhanced green fluorescent protein (eGFP) under the control of the 5′-flanking region of the mouse Kiss1 gene (45) were bred and housed at the Oregon Health and Sciences University. Mice were housed 4 to 5 per cage under controlled conditions of temperature (23-25 °C) and photoperiod (12/12 light/dark cycle with lights on between 0600 and 1800 h), with standard pelleted food (Purina Cat# 5018) and water provided ad libitum.

Evaluation of Sexual Maturation

To determine the global changes in hypothalamic gene expression that occur at the time of female puberty, rats were euthanized at 4 different stages: infantile (Inf, 14 days [14d] of age); early juvenile (EJ, 21d of age), late juvenile (LJ, 28d of age) and late proestrous (LP, 32d-37d of age). According to criteria previously established (46, 47), 14-day-old animals are infantile while 21-day-old animals are in the early juvenile phase of prepubertal development. During this period, the vagina is not yet patent, and the uterine weight is 60 mg or less, with no accumulation of intrauterine fluid. At 28 days of age, the rats are in the late juvenile (LJ) phase of prepubertal development; their vagina is closed and there are no signs of intrauterine fluid accumulation. Animals in this phase exhibit a diurnal change in pulsatile plasma luteinizing hormone (LH) levels, with the LH pulses becoming more pronounced in the afternoons (48). Like in primates, this is the first hormonal manifestation of the increase in central drive that initiates puberty (49). Finally, rats still exhibiting a closed vagina but showing a uterus ballooned with fluid and a uterine weight of at least 200 mg are considered to be in late proestrus, that is, the phase of puberty when the first preovulatory surge of GnRH and gonadotropins takes place. They are considered to be in mid-puberty because ovulation has not occurred yet. The first ovulation occurs the following day. All animals were euthanized between 1600 and 1700 hours, and the medial basal hypothalamus (MBH) was immediately dissected and frozen on dry ice.

The animals subjected to intrahypothalamic administration of lentiviral particles were inspected every afternoon for vaginal opening, starting 2 days after intrahypothalamic administration of lentivirus. Once vaginal opening occurred, vaginal lavages were performed daily to identify the occurrence of the first estrus, which in rodents is manifested by a predominance of cornified cells. Although ovulation normally occurs on the day of estrous, detection of cornified cells does not indicate that ovulation has occurred, unless vaginal cornification is followed by the appearance of a predominance of leukocytes. The presence of these cells defines the diestrous phase of the estrous cycle and indicates that a functional corpus luteum was formed after ovulation. For these reasons, the age at first ovulation was considered to have occurred only when the cornified cells were followed by at least 2 days of lavages containing mostly leukocytes.

Tissue Dissection and Fecal Collection

Rats were euthanized by CO₂ inhalation. The MBH of female rats was dissected by a rostral cut along the posterior border of the optic chiasm, a caudal cut immediately in front of the mammillary bodies, and 2 lateral cuts half-way between the medial eminence and the hypothalamic sulci. The thickness of the tissue fragment was about 2 mm. This fragment includes the entire arcuate nucleus (ARC) and the median eminence (ME) and excludes the preoptic area of the hypothalamus to only study the ARC KNDy to GnRH terminal interaction and avoid the confounding effect of preoptic area kisspeptin neurons as done by us previously (18).

Fecal samples were retrieved from the colon and immediately frozen on dry ice.

Incubation of ARC-ME Explants and Measurement of GnRH Release

ARC-ME explants from 28-day-old rats were micro dissected and placed in small plastic vials precoated with 0.1% gelatin (one fragment per vial). Thereafter, the tissue fragments were preincubated for 30 minutes in Krebs-Ringer bicarbonate (KRB) buffer, pH 7.4, containing 4.5 mg/mL D-dextrose at 37 °C under an atmosphere of 95% O₂, 5% CO₂ with shaking (60 cycles per minute), as described previously (50, 51). At the end of this period, the medium was discarded and replaced with fresh medium for 1 hour (Basal). Thereafter, the medium was collected and replaced with medium containing 100 nM kisspeptin 110-119 (Phoenix Pharmaceuticals, Burlingame, CA), 10 μM INT777 (a TGR5-specific agonist), with or without the presence of 1μM KP234 (a GPR54 antagonist when used at 1μM or lower concentration) (52), and the incubation was continued for another hour. GnRH released to the incubation medium was determined as described below.

Measurement of GnRH

Assays for GnRH were performed by the Endocrine Technology and Support Lab, Oregon National Primate Research Center (Beaverton, OR). Briefly, the GnRH decapeptide (Sigma L-7134, St. Louis, MO) was radio-iodinated with I-125 and free I-125 was separated from I-125-GnRH by Sephadex QAE-25 column. GnRH standards were prepared in Krebs-Ringer phosphate buffer (KRP) used for samples in the study. The radioimmunoassay (RIA) used has been described in detail elsewhere (53). The final dilution of the anti-GnRH, (Oline Ronnekleiv; Oregon Health and Science University Cat# EL-14, RRID: AB_2715535), was 1:144 000. The sensitivity of this assay was 0.02 pg/tube and the range was 100 pg/tube. The intra- and interassay variation were 9% and 12%, respectively.

RNA Extraction and Quantitative Reverse Transcription–Polymerase Chain Reaction

Total RNA was extracted from ARC-ME using the RNeasy mini kit (Qiagen, Valencia, CA) following the manufacturer's instructions. RNA concentrations were determined by spectrophotometric trace (Nanodrop, ThermoScientific, Wilmington, DE). Total RNA (500 ng) was transcribed into cDNA in a volume of 20 μL using 4 U Omniscript reverse transcriptase (Qiagen). To determine the relative abundance of the mRNAs of interest, we used the SYBR GreenER™ qPCR SuperMix system (Invitrogen, Carlsbad, CA). Primers for amplification were designed using the PrimerSelect tool of DNASTAR 14 software (Madison, WI) or the NCBI online Primer-Blast program. Polymerase chain reaction (PCR) reactions were performed in a total volume of 10 μL containing 1 μL of diluted cDNA or a reference cDNA sample (see below), 5 μL of SYBR GreenER qPCR SuperMix and 4 μL of primers mix (1 µM of each gene specific primer). The PCR conditions used were 95 °C for 5 minutes, followed by 40 cycles of 15 seconds at 95 °C and 60 seconds at 60 °C. To confirm the formation of a single SYBR Green-labeled PCR amplicon, the PCR reaction was followed by a 3-step melting curve analysis consisting of 15 seconds at 95 °C, 1 minute at 60 °C, ramping up to 95 °C at 0.5 °C per second, detecting every 0.5 second and finishing for 15 seconds at 95 °C, as recommended by the manufacturer. All qPCR reactions were performed using a QuantStudio 12 K Real-Time PCR system; threshold cycles (CTs) were detected by QuantStudio 12 K Flex software. Relative standard curves were constructed from serial dilutions (1/2 to 1/1000) of a pool of cDNAs generated by mixing equal amounts of cDNA from each sample. The CTs from each sample were referred to the relative standard curve to estimate the mRNA content per sample; the values obtained were normalized for procedural losses using peptidylprolyl isomerase A (Ppia) as the normalizing unit. Primers used:

Ppia-F: GGCAAATGCTGGACCAAACACAA and Ppia-R: GGTAAAATGCCCGCAAGTCAAAGA. Fxr-F: AGGAAGTGCAGAGAGATGGGA and Fxr-R: GCTTGGTCGTGGAGGTCACT. Lepr-F: CCCCTTCCGAGCAAATTAAAAGCAA and Lepr-R: TCTGACACGTCATCTTACTTCAGAG. Tgr5-F: CTGTGTTGGGGGCCCTATGT and Tgr5-R: CATGGCCACAGGCACAACTG. Expression data were normalized by dividing each individual value by the average of the Inf/PN14 group, expressed as arbitrary units (AU).

Cloning LV-TGR5-GFP

An expression vector encoding the human TGR5 cDNA fused to green fluorescent protein (GFP) was generously provided by Professor Vernena Keitel. The TGR5-GFP cassette was removed with NdeI and NotI (NEB, Ipswich, MA) and blunted on the 3′ site by incubation with T4 DNA Polymerase (NEB, Ipswich, MA). Our third-generation lentiviral (LV) vector plasmid (18) was digested with NdeI and SmaI followed by ligation with the TGR5-GPF cassette over night at 4 °C with T4 DNA ligase (NEB, Ipswich, MA). Infective lentiviral particles were produced by the Oregon National Primate Center Molecular Virology Core Service as previously described (18).

Validation of LV Constructs by Western Blots

To verify the ability of the LV-TGR5-GFP vectors to produce TGR5-GFP, we infected neuro-2a cells with the lentiviral particles described above (LV-GFP or LV-TGR5-GFP) and lysed the cells 7 days later with 500 µL RIPA buffer (Thermo, Waltham, MA) containing 10 µg/mL leupeptin and Pepstatin A, 10 µg/mL aprotinin, and 100 µg/mL PMSF). Protein concentrations were estimated using the Pierce 660 nm Protein Assay (Pierce, Rockford, IL). Samples were boiled for 5 minutes in reducing sample buffer and then fractionated in a 8% to 16% precast SDS-PAGE gel (Invitrogen, Carlsbad, CA). After electrophoresis at 130 V for 2 hours, the proteins were transferred for 4 hours at 4 °C onto a polyvinylidene difluoride membrane (Millipore, Billerica, MA). The membrane was blocked in 5% nonfat milk for 1 hour. Antibodies against the eGFP (Abcam Cat# ab290, RRID: AB_303395, Cambridge, MA) were used at a 1:1000 dilution in TBST (overnight at 4 °C) followed by a species-specific anti-IgG-HRP antibody (1 hour at room temperature, 1:50 000; Thermo Fisher Scientific Cat# ALI3404, RRID: AB_3095562). The signal was developed by enhanced chemiluminescence using the Western lightning chemiluminescence substrate (Pierce).

Stereotaxic Delivery of LV Particles

We used LV particles carrying a TGR5 overexpression cassette fused to GFP or a control lentivirus expressing GFP alone. A group of 11 rats (19-day-old animals) were anesthetized with a cocktail of ketamine (25 mg/mL), xylazine (5 mg/mL), and acepromazine (1 mg/mL), administered intraperitoneally (0.15 mL per 100 g of body weight). Thereafter, the rats were positioned on a stereotaxic instrument (David Kopf Instruments, Tujunga, CA) with the incisor bar set at +5 mm. A total volume of 1 μL of the virus (LV-TGR5-GFP, n = 6 or LV-GFP, n = 5) was injected bilaterally into the ARC at the rate of 250 nL/min, using a 10-μL Hamilton micro-syringe connected to a Stoelting Stereotaxic microinjector (Stoelting, Wood Dale, IL). The coordinates used were: 0.3 mm lateral from midline, 0.2 mm anterior from bregma, and 9.6 mm vertical from the surface of the skull. The surgical procedure lasted about 15 minutes. Following surgery, the animals were placed in a clean cage on a heating pad until returned to their home cages. Thereafter, they were treated for 3 days with an analgesic (Carprofen, 5 mg/kg) and an antibiotic (Baytril 10 mg/kg), administered subcutaneously. Five animals were injected with LV-GFP and 6 with LV-TGR5-GFP. After evaluation of sexual maturation, animals were evaluated for correct delivery of viral particles into the ARC by immunohistofluorescence. Two LV-TGR5-GFP animals showed no GFP signal in the ARC and thus were added as controls into the LV-GFP group.

Detection of Hypothalamic LV Infection

Cells transduced with lentiviral constructs were identified in 35 μm brain sections by immunohistofluorescence using a goat polyclonal antibody against eGFP (1:2000, Abcam Cat# ab290, RRID: AB_303395, Cambridge, MA). Following an overnight incubation with this antibody, the sections were incubated for 1 hour at room temperature with an Alexa 488 donkey anti-goat IgG (1:500; Thermo Fisher Scientific Cat# R37118, RRID: AB_2556546), followed by 1-minute incubation with Hoechst 33258 reagent (1:10 000; Thermo Fisher) to stain cell nuclei.

Detection of Tgr5 Gene Expression in GFP-Expressing Kisspeptin Neurons by Single-Cell Reverse Transcription–PCR

Tissue collection

To better visualize kisspeptin neurons of the ARC, we used MBH from adult female mice (1 ovary intact and 2 animals 1 week following ovariectomy) expressing GFP in kisspeptin neurons. Each mouse was given 15 mg ketamine intraperitoneally before collecting the MBH. The tissue blocks were dissected and mounted on a cutting platform in a vibratome well filled with ice-cold, oxygenated (95% O₂, 5% CO₂) high sucrose artificial cerebrospinal fluid (aCSF). Four coronal slices (240 μm) were cut through the hypothalamus. The slices were transferred to an auxiliary chamber containing oxygenated aCSF (in mM: 124 NaCl, 5 KCl, 1.44 NaH₂PO₄, 5 HEPES, 10 dextrose, 26 NaHCO₃, 2 CaCl₂) and kept there until cell harvesting.

Neuronal harvesting and reverse transcription

Single cells were harvested according to established procedures (54-56) with minor modifications. In brief, the ARC was dissected and exposed to papain and gentle trituration to disperse the neurons. The dispersed cells were visualized using a Leitz inverted microscope, patched and then harvested with gentle suction using the XenoWorks microinjector system (Sutter Instruments, Navato, CA). The contents of the pipette were expelled into a test tube containing a solution with 1×Superscript III buffer (Invitrogen), 15 U of RNasin (Promega, Madison, WI) and 10 mM of dithiothreitol in 5 µL total volume. Each harvested cell was reverse transcribed as described previously (54), with modifications in which both random primers (100 ng/cell, Promega) and anchored oligo(dT)₂₀ primer (400 ng/cell, Invitrogen), and Superscript III reverse transcriptase (100 U/cell) were used. In addition, aCSF harvested in the vicinity of the dispersed cells was subjected to reverse transcription (RT) and used as a negative control. Cells and tissue RNA used as additional negative controls were processed as described above, but without reverse transcription (−RT).

Single-cell RT-PCR

The PCR was performed using 3 μL of cDNA template from each RT reaction in a 30 μL PCR mix. Fifty cycles of amplification were performed using a Bio-Rad C1000 Thermal Cycler (Bio-Rad, Hercules, CA) according to established protocols (25). PCR products were visualized with ethidium bromide on a 2% agarose gel.

Primers

Primer pairs for single-cell RT-PCR were based on the mRNA sequences of the genes of interest reported by the National Center for Biotechnology Information and were selected using the Clone Manager software (Sci Ed Software, Cary, NC). The primers were designed to cross introns to preclude genomic DNA amplification, and the PCR products from single cells were sequenced to confirm the identity of the product. The mouse primers are: Tgr5 (Gpbar1) accession number NM_174985; 97 bp product. Forward primer 480-499 nt; 5′ TGCCTCCTTCTCCACTTGAC 3′ and reverse primer 557-576 NT; 5′ GCTGCAACACTGCCATGTAG 3′.

Kiss1 accession number NM_178260; 120 bp product. Forward primer: 224-240 nt; 5′ TGCTGCTTCTCCTCTGT 3′ and reverse primer 327-343 nt; 5′ ACCGCGATTCCTTTTCC 3′. The proportion of kisspeptin neurons expressing TGR5 mRNA was not different comparing cells derived from ovary intact and ovariectomized adult female mice.

Assessment of the Bile Acid Pool Using Liquid Chromatography Mass Spectrometry

Nonfasted serum samples from female rats (PND 14, 21/EJ and 28/LJ) and fasted and nonfasting samples from humans were submitted to the Columbia University Metabolomics Core facility for analyses by liquid chromatography mass spectrometry (LCMS) per instructions from the core. Bile acid isoforms analyzed included in this panel were CA, DCA, ursodeoxycholic acid (UDCA), hyodeoxycholic (HDCA), LCA, tauro-(T)LCA, TCDCA, glyco-(G)CA, TDCA, GDCA, TCA, GCDCA, GUDCA, TUDCA, (α + Щ)-muricholic acid (MCA), β-MCA, and T(α + β)-MCA. This core facility is supported by a National Institutes of Health UL1 Clinical and Translational Science Award (grant number: UL1TR001873).

Microbiota Phylogenetic Profiling

DNA extraction, sequencing, and data preprocessing are described as follows. Microbial genomic DNA was extracted from about 20 mg fecal material using QIAGEN DNeasy PowerSoil kits following the manufacturer's instructions. The extracted DNA was cleaned using AMPure XP beads (Beckman Coulter). Following standard Earth Microbiome Project protocols (57), the V4 region of the 16S rRNA gene was amplified in triplicate using 515F (58) and 806R (59) primers, and the Platinum Hot Start PCR Master Mix (Thermo Fisher). PCR products were cleaned using AMPure XP beads and pooled for each sample. The amplicon pools were quantified with Quant-iT PicoGreen dsDNA Reagent (Invitrogen), and 100 ng of amplicons per sample were pooled prior to submission for sequencing. Paired-end sequencing (2 × 250 bp) was performed on the Illumina MiSeq platform at the Cornell Institute of Biotechnology. After de-multiplexing by the Genomics Facility according to standard protocols, fastq sequences were further analyzed per sample using the DADA2 pipeline (v. 1.6) rendering amplicon sequence variants (ASVs) (60). In brief, after inspecting quality, sequences were trimmed to remove the primers and the first 10 bases after the primer, keeping only 200 bases and 160 bases for the R1 and R2 files, respectively. After merging paired sequences and removing chimeras, taxonomy was assigned using the formatted RDP training set “rdp_train_set_16” up to genus level. Additional species level information was added through the silva database using the file silva_species_assignment_v128.fa.gz. This ASV matrix was pruned, removing ASVs present in the sequencing run but not the current dataset, rendering a sample-species matrix containing a total of 475 ASVs. Amplicon sequence variants that could be classified at species level were named using the species information (eg, sp932_Helicobacter_apodemus), while the rest were named using the closest known taxonomic level (signified by a “uc”—for unclassified—prefix, followed by the taxonomic level: g-genus, f-family, c-class, o-order, k-kingdom). To obtain microbiome profiles of equal sequencing depth (as used in the alpha diversity analyses) the sample-species matrix was subsampled to the minimum number of reads in a sample (10 347) and normalized. For relative microbiome profiles (as used in all other analyses), the sample-ASV matrix was normalized and limited to ASVs with a relative abundance of > 0.001 in the dataset.

Microbiome Analyses

All analyses were performed in R, version 4.0.4, using base R and packages dada2 (60), phyloseq (61), plyr (62), reshape2 (63) and ggplot2 (61). Beta diversity analyses were based on Euclidian distances calculated based on the centered log ratio transformed relative abundance ASV matrix. Gene family abundances per sample were estimated using picrust2 (64) with default parameters. Orthologue gene groups involved in secondary bile acid biosynthesis were based on KEGG database pathway ko00121.

Statistical Analysis

Kruskal-Wallis or Wilcoxon 2-sample statistics were conducted to determine whether overall differences in participant characteristics and bile acid concentrations across groups were detected. Analyses were conducted to contrast both individual bile acids, primary, secondary, conjugated, deconjugated, and relevant ratios. When overall significance was detected (Kruskal-Wallis), post hoc analyses of significant groups effects were conducted using Steel-Dwass tests. Data are reported as median (25th and 75th percentiles). The lower limit of quantitation defined by the Columbia Metabolomics Core was included when the measured bile acid values fell below the lowest detectable limit (in the pilot study, 9% of measured bile acid isoform values across human samples (13% in prepubertal, 6% in postmenarcheal, and 9% in adults). Significance was detected at P < .05. Statistical analyses were performed in JMP Pro 16 and 17 (Copyright © SAS).

All other statistical analyses were performed using Prism 10 software (GraphPad, San Diego, CA). The data were first subjected to a normality and an equal variance test. Data that passed these 2 tests were then analyzed by either ANOVA followed by the Student-Newman-Keuls (SNK) to compare multiple groups; the Dunnett's test to compare several groups to a single control group; or the Student t test to compare 2 groups. When comparing percentages, groups were subjected to an arc-sine transformation before statistical analysis to convert the values from a binomial to a normal distribution (65). The sample size was selected based on power analyses performed using the standard deviations that we normally observe when measuring the parameters examined in this study and an n = 6/group. These analyses provide at least 80% (type II error = 0.124) power to detect 2 effect sizes using either ANOVA or 2-sided 2-sample t test with a significance level of 0.05. The investigator was blinded to the group allocation in all physiological and molecular determinations.

Results

Higher Concentrations of Deconjugated and Secondary Bile Acids Were Detected in Postmenarcheal Adolescent Females Vs a Younger Cohort

To test the hypothesis that bile acid isoform shifts occur in conjunction with discrete stages of reproductive maturation in humans, we first conducted a pilot, cross-sectional analysis analyzing serum from 3 distinct cohorts of female human subject participants: (i) pre-/early pubertal, derived from the Precision for Medicine Biospecimen Repository; (ii) early postmenarcheal adolescents participating in a longitudinal cohort study (Clinicaltrials.gov NCT04424576); and (iii) healthy, regularly menstruating adult women who participated in a longitudinal study with banked sera (Clinicaltrials.gov NCT01927432). Samples were obtained during the early follicular phase in postmenarcheal adolescent and adult cohorts if regularly cycling. Samples obtained from the prepubertal and adult cohorts were nonfasting and the early postmenarcheal cohort were fasted. Participant clinical features are presented in Table 1. By design, the age of participants significantly increased across each group with the median ages being 7.0, 12.0, and 30.5 years, respectively. Pubertal stage was not available for the pre-/early pubertal cohort as the samples were derived from a biorepository where pubertal information was not available. Therefore, a critical assumption we assert in this study was that the biorepository samples were obtained from females likely to be pre- or early pubertal based on the normative range of thelarche (achieving Tanner breast stage 2) and age at menarche (42, 43).

Table 1.

Characteristics of the human study participants

	Prepubertal (n = 10)			< 1 yr Postmenarche (n = 10)			Adults (n = 10)
	Median	25th %ile	75th %ile	Median	25th %ile	75th %ile	Median	25th %ile	75th %ile	P Value
Age at menarche (yr)				12.63	12.02	13.20	12.50	11.00	13.00	NS
Age (y)	7.0a	4.00	8.25	12.5b	12.00	13.00	30.5c	23.75	36.50	<.0001
Gynecological age (y)				0.54a	0.46	0.85	19.00b	11.25	23.50	.0002
Height (m)	1.17a	1.03	1.36	1.59b	1.53	1.66	1.62b	1.59	1.67	<.0001
Weight (kg)	24.04a	18.59	35.49	48.90b	45.60	53.38	61.10c	55.58	70.40	<.0001
BMI (kg/m2)	17.66a	16.01	20.04	18.63a	17.44	20.81	23.32b	22.27	24.71	.0003

^a,b,cDifferences between groups are denoted with different superscripts after post hoc analyses. Group effects were determined with Kruskal-Wallis Tests. Post hoc comparisons were conducted using Steel-Dwass tests. Post hoc significance between groups at P < .05.

Abbreviations: BMI, body mass index; NS, not significant.

Shifts in bile acid profiles across reproductive stage are depicted in Fig. 1. Overall, there was no difference in the total bile acid pool across cohorts (Fig. 1A). The ratio of conjugated to unconjugated bile acids was significantly lower in the peripubertal cohort vs the prepubertal cohort (Fig. 1B). Similarly, the concentration of secondary bile acids was lowest in the pre-/early pubertal cohort and was significantly greater in the peripubertal and adult cohorts vs the pre-/early pubertal cohort (Fig. 1C). Differences in the individual bile acid isoforms across the 3 independent cohorts are presented in Supplementary Table S1 (44).

Figure 1.

Bile acid conjugation status is altered in the perimenarche window vs the pre–/early pubertal window in adolescent females. Serum samples obtained from pre–/early pubertal (n = 10), early postmenarche (n = 10), and adult women undergoing normal menstrual cycles were analyzed by mass spectrometry for bile acid isoforms (see Table 1). A, Total bile acid pool was not altered over life stage. B, Ratio of conjugated to unconjugated bile acids was reduced during the perimenarche stage then returned to prepubertal levels in adult women. C, Total secondary bile acids increased during perimenarche and remained elevated in adult women. Bars with differing letters (a and b) differ (P < .05).

Comparisons were repeated in an independent, secondary cohort of pre- and postmenarcheal adolescents (described in Supplementary Methods) (44). Briefly, premenarcheal adolescents were at least Tanner 3 by breast or pubic hair, based on parent or physician report and postmenarcheal adolescents were within 2 years of their very first menses. On average, the postmenarcheal adolescents were older (12 vs 11 years, P = .04) and had higher concentrations of LH vs their premenarcheal counterparts (3.65 vs 1.23 mIU/mL, P = .002; Supplementary Table S2) (44). Differences in individual bile acids are reported in Supplementary Table S3 (44); GCA and GCDCA, conjugated primary bile acids, were significantly higher in the premenarcheal vs postmenarcheal cohort (429.79 vs 210.52, P = .04; 1949.32 vs 1341.39, P = .02; Supplementary Table S4) (44). Consistent with individual bile acid data, total primary conjugated bile acids were higher in the premenarcheal vs postmenarcheal cohort (2516.93 vs 1803.92, P = 0.02; 72% vs 55%, P = .01; Supplementary Table S4) (44). No differences were detected in total bile acid pool and total secondary bile acids (Supplementary Table S3) (44).

Deconjugation and Increases in Secondary Bile Acid Concentrations Occurred During Rat Prepubertal Maturation

Differences in bile acid isoforms across pubertal stages in human subjects were further explored in a more carefully controlled manner in female rats during the infantile to pubertal transition. Pubertal stages were defined following Sergio Ojeda's classification (46) and expanded further in our subsequent publications (18-21). Bile acid isoform shifts obtained from serum of nonfasted female rats collected at infantile (PN14), early juvenile (EJ; PN21), and late juvenile (LJ; PN28) are presented in Fig. 2. The total bile acid pool significantly increased from PN14 through PN21 and remained elevated through PN28 (Fig. 2A). Consistent with the human data, the ratio of conjugated to unconjugated bile acids dropped from PN14 to PN21 (Fig. 2B). Accordingly, the concentration of secondary bile acids also significantly increased from PN14 to PN21 and PN28 (Fig. 2C).

Figure 2.

Bile acid conjugation status is altered during the postnatal period in juvenile and adolescent female rats. Serum samples obtained from postnatal day (PN)14 (n = 5), PN21 (n = 5) and PN28 (n = 5) were analyzed by mass spectrometry for bile acid isoforms. A, Total bile acid pool was increased at PN21 and PN28 compared to PN14. B, Ratio of conjugated to unconjugated bile acids was reduced was reduced on PN21 and PN28 compared to PN14. C, Total secondary bile acids were increased during PN21 and PN28 compared to PN14. Bars with differing letters (a and b) differ (P < .05).

Shifts in the Gut Microbiome Favor BSH Activity During Peripubertal Development in Rats

Conjugated bile acids are deconjugated in the gut microbiome via bacterial bile salt hydrolases (BSH) and converted to secondary bile acids (DCA, LCA) by microbial hydroxylation, implicating the gut microbiome as a modulator of bile acid activity on host physiology. Therefore, we assessed whether gut microbial species diversity shifted concurrent with dramatic changes in bile acid isoforms, favoring BSH activity and therefore implicating the gut microbiome as the mediator linking bile acid isoforms to reproductive maturity.

Compositional changes in the microbiome were assessed via 16S sequencing from PN14 to PN21 (Fig. 3). Overall, the relative abundance shifted dramatically from PN14 to PN21; the microbiome in the PN14 rats was dominated by Escherichia coli, whereas the PN21 rats show a more diverse pattern including Bacteroidetes and Firmicutes species. Microbiome maturation in humans follows a similar trajectory with high Proteobacterial abundances in the first years of life, evolving to an adult-like microbiome dominated by Bacteroidetes and Firmicutes after the introduction of solid food, similar to the PN14 to 21 transition in female rats. Concomitantly, the amount of different species in the older rats was about 3-fold higher (paired Wilcoxon test, 2-sided, P = .005), while bacterial abundances were more evenly distributed (Pielou evenness index, paired Wilcoxon test, 2-sided, P = .0022, Fig. 3B-3D). Alpha diversity, as evaluated by the Shannon diversity index, a measure summarizing species richness and species evenness, was significantly higher in the older rats (paired Wilcoxon test, 2-sided, P = .0022) (Fig. 3B-3D).Where community composition is highly similar within PN14 rats, PN21 rats presented markedly different communities, as evidenced by our beta diversity analyses (Fig. 3E and 3F). Next to E. coli, several ASVs were differentially abundant between the 2 life stage groups (Supplementary Table S5) (44). These findings collectively suggest a significant diversification of the gut microbiota during this time period, transitioning from an E. coli-dominated community at 14 days to a more diverse community by day 21.

Figure 3.

Gut microbial composition diversity is markedly different in female rats at PN14 compared to PN21, the latter presenting a more diversified gut microbiota with a higher genetic potential for conjugation and secondary bile acid biosynthesis. Fecal pellets from the colon were obtained following exsanguination from female rats at PN14 and PN21 (n = 6/life stage). Samples were subjected to 16S sequencing and analyzed per Methods. A, Stacked bar plots correspond to relative genera abundances, based on amplicon sequence variant relative abundances merged by genus. The top 20 genera are colored according to the legend (right). Complete bacterial abundance profiles can be found in the supplementary information (44). B-D, Diversity is higher for PN21 compared to PN14 rats across all metrics, from Shannon diversity index (B), observed richness (C), to Pielou's evenness index (D) (paired Wilcoxon test, 2-sided, P = .031). E, Beta diversity within and between life stage groups. F, Principal coordinate analyses separates both life stage groups on the horizontal axis, and individuals within the PN21 group on the vertical axis. G, Microbial genetic potential for conjugation status and secondary bile acid biosynthesis is higher in PN21 compared to PN14 rats. Reactions for 3 orthologue gene families involved in microbial secondary bile acid biosynthesis pathways are shown. Molecular conversions associated with the detected functions are depicted on top of the differential abundance graphs. Multiple arrows are indicative of multi-step reactions. In that case, the enzymes of the detected gene groups perform only part of the reaction. Molecular structure illustrations correspond to the compounds printed in italic. Gene family abundances, estimated using picrust2, are plotted for 14- and 21-day-old rats for bile salt hydrolase (P = .0013) (left), 7- 7 α-hydroxy-steroid dehydrogenase (P = .0069) (middle), and 3-dehydro-bile acid Delta 4,6-reductase (P-6.07 × 10-6) (right) are shown. P values were calculated using a 2-sided paired t test. Box plots represent the first and third quartiles of the distribution and the median line. Whiskers extend from the quartiles to the last data point within 1.5×IQR, with outliers beyond. Significance levels: ***: 0.001, **: 0.01, *: 0.05.

We further evaluated whether bile acid deconjugation might be higher in PN21 vs PN14 rats by evaluating the relative abundance of genes which expressing BSH (Fig. 3G). The abundance of genes encoding BSH, which catalyzes the deconjugation reaction resulting in the removal of taurine or glycine from chenodeoxycholate or cholate, was significantly increased in the fecal pellets obtained from PN21 vs PN14 rats (paired t test, 2-sided, P = .0013; left side Fig. 3G). Two additional gene families involved in subsequent bile acid conversions, 7-alpha-hydroxy-steroid dehydrogenase and 3-dehydro-bile-acid delta 4.6 reductase, were significantly lower (paired t test, 2-sided, P = 6.07 × 10⁻⁶) and higher (paired t test, 2-sided, P = .0069) abundance in PN21 vs PN14 samples, respectively.

Tgr5 Expression Increases in the Hypothalamus During Early and Late Puberty in Female Rats

As a follow-on from assessment of circulating bile acid pool composition in our human and rat studies, we next sought to examine the possibility that bile acid receptor (nuclear receptor farnesoid X receptor (fxr) and the G-protein coupled bile acid receptor, (Gpbar-1/tgr5) expression profiles may be regulated in the ARC during the infantile to pubertal transition in female rats consistent with alterations in the composition of the ligands for these receptors. Rat ARC Tgr5 and fxr mRNA were below threshold in our previously published RNAseq dataset GSE94080 (including rats aged PN7-PN33), possibly due to the inherent low copy number of G-protein coupled receptors. ARC fxr mRNA expression was undetectable by qPCR, while easily detectable within liver (Fig. 4A). In contrast, ARC expression of tgr5 mRNA remained relatively low at the Inf (PN14) stage, followed by a numerical 2- to 3-fold increase in mRNA expression at the early and late juvenile (EJ, LJ) phases. Relative tgr5 mRNA content increased dramatically in late puberty rats (PN30-34) (Fig. 4B), corresponding with increased estrogen levels as determined by a significant increase in uterine weight (Fig. 4A, inset). The relative expression of leptin receptor mRNA remained unchanged during the pubertal transition (Fig. 4B), confirming previous results by others (9) and demonstrating a lack of interaction between tgr5 increased expression and leptin signaling.

Figure 4.

Tgr5 mRNA is expressed in the rodent hypothalamus and is regulated during the pubertal transition in female rats. A, qPCR standard curve for FXR using cDNA from rat ARC (blue circles) or liver tissue (red circles) demonstrating the lack of FXR expression in the rat ARC. B, Hypothalamic tissue from female rats were processed for qPCR for tgr5 and leptin receptor (lepr) mRNA levels at postnatal day (PN) 14 (Infantile or Inf), PN21(early juvenile or EJ), PN28 (late juvenile or LJ), and during late puberty (LP) (n = 5-11/life stage). Uterine weights (shown in inset) were used to identify individual stages during the pubertal transition. Bars with differing letters (a, b) differ (P < .05). C, Single cell neurons were isolated from the arcuate nucleus of female mice expressing enhanced green fluorescent protein (GFP) under the control of the 5′-flanking region of the mouse Kiss1 gene (45) and assessed by qPCR for tgr5 mRNA. Abbreviations: AU, arbitrary units; MM, molecular marker; −RT, no reverse transcriptase control; Tissue Control, cDNA from mouse hypothalamus.

In mammals the reawakening of the reproductive axis during puberty is manifested by a diurnal increase in GnRH and hence LH release. This is largely controlled by ARC KNDy neurons (66), also known for their integrative role in the metabolic control of puberty maturation (67). To study the possible role of tgr5 in KNDy neuron activity, we carried out single-cell PCR studies using eGFP-labeled kisspeptin neurons (45) from adult female mice. Approximately two-thirds of the single ARC kisspeptin neurons were positive for tgr5 mRNA (Fig. 4C).

TGR5 Activation Increases GnRH Secretion in a Kisspeptin-Receptor–Dependent Manner in Hypothalamic Explants

To assess the participation of bile acid signaling through TGR5 within the hypothalamus, we utilized medial basal hypothalamic (MBH) fragments from PN28 rats in vitro, containing the ME-ARC region. First, we evaluated the ability of these explants to respond to a physiologically relevant neuroendocrine stimulus, by significantly stimulating GnRH release under kisspeptin exposure. The co-incubation with the kisspeptin receptor blocker KP234 significantly reduced the kisspeptin-induced GnRH release. Moreover, the incubation of the MBH fragments with a TGR5 agonist (INT777) induced a significant increase in GnRH release that was also blocked by co-incubation with KP234. This demonstrates that bile acid–dependent GnRH release depends on kisspeptin release from ARC KNDy neurons (Fig. 5).

Figure 5.

Stimulation of TGR5 within ARC/median eminence fragments results in increased GnRH secretion ex vivo. Hypothalamic fragments containing the ARC nucleus and median eminence were placed in culture and stimulated with kisspeptin (KP), or INT777 (a TGR5 specific agonist) in the absence or presence of the Kisspeptin Receptor/GPR54 antagonist, KP234. GnRH secretion was then assessed by radioimmunoassay. Bars are means and vertical lines represent standard error of the mean. Two-way repeated measures ANOVA, n = 6 rats/group with different letters showing significance, a vs b P < .001 and a vs c P < .05.

Overexpression of tgr5 Within the Arcuate Nucleus Results in Accelerated Timing of Vaginal Opening and First Ovulation in Female Rats

Thus far, we have demonstrated marked biochemical changes within the circulating bile acid pool during the pubertal transition in girls and in female rats. Consistent with these changes within the bile acid pool, we also showed expression and activity of tgr5 within the ARC nucleus kisspeptin neurons supporting the hypothesis that the kisspeptin neuronal population may be central to orchestrating responses to bile acid/TGR5 signaling leading to first ovulation in rats. To examine this hypothesis, we cloned the human TGR5 gene fused to the eGFP into a lentiviral construct (LV-TGR5-GFP). Western blot characterization of in vitro infected neuro-2a cells shows an expected 27 kDa band for the eGFP protein expressed by the LV-GFP vector and a 62-kDa band (eGFP: 27 kDa + TGR5: 35 kDa) when cells are infected with the LV-TGR5-GFP vector (Fig. 6A). Next, we used bilateral stereotaxic injections of the LV constructs into the ARC of female rats on PN19. Immunohistofluorescent analysis of the site of injection using antibodies against eGFP was used to identify the transduced cells (Fig. 6B). Animals with correctly placed LV-TGR5 injections in the ARC (n = 4) showed advancement in vaginal opening (VO) at PN 32.75 ± 0.25 compared to control animals with off-target LV-TGR5 or LV-GFP (n = 7) at PN 35.57 ± 0.78. On the other hand, the first estrus (FE, as assessed by the detection of cornified cells in vaginal smears followed by 2 consecutive days of leukocytes) was advanced in animals with correctly placed LV-TGR5 injections with an average first estrus at PN 32.75 ± 0.25 compared to control animals at PN 37.86 ± 0.46 (Fig. 6C). These results support the hypothesis that increased hypothalamic TGR5 signaling advances female pubertal tempo in a rat model of reproductive development.

Figure 6.

Gain of function of TGR5 within the ARC nucleus modulates the timing of puberty in rats. A, Western blot validation of lentiviral particles carrying green fluorescent protein (LV-GFP) or LV-TGR5-GFP in infected neuro-2A cells. Antibodies directed against GFP detect 27 kDa band (arrowhead) when using LV-GFP or a 62 kDa band (GFP + TGR5, arrow) when using LV-TGR-GFP. B, Stereotaxic-directed injections of LV-GFP or LV-TGR5-GFP were delivered into the ARC of PND19 female rats. C, Animals were allowed to recover and monitored for age at vaginal opening (VO) and first estrus (FE). Five animals were injected with LV-GFP and 6 with LV-TGR5-GFP. After evaluation of sexual maturation, animals were evaluated for correct delivery of viral particles into the ARC by immunohistofluorescence. Two LV-TGR5-GFP animals showed no GFP signal in the ARC and thus were added as controls into the LV-GFP group.

Discussion

The present studies provide insights into the timing of reproductive maturation integrated by the hepatic-gut microbiome-neuroendocrine axis. We find that timing of reproductive milestones during sexual maturation coincides with changes in bile acid isoforms that align with compositional changes in the gut microbiome which may favor bile acid isoform metabolism, and that hypothalamic bile acid signaling via TGR5 receptor could likely affect GnRH neuron response to kisspeptins. These findings support our central hypothesis, which is that bile acid/TGR5 signaling may provide a key metabolic signal capable of modulating the timing of reproductive maturation. First, we observed a pronounced increase in unconjugated and secondary bile acid concentrations in pre- vs postmenarcheal adolescent females, which was also observed in rats by PN21, approximately 2 weeks prior to vaginal opening and first ovulation. An important caveat to the present studies is that the timing of first ovulation in a rodent model and pre- vs postmenarche in humans are clearly not perfectly interchangeable. In adolescent girls, menarche is not a perfect predictor of first ovulation, and other important pubertal milestones, including adrenarche, thelarche, and peak growth velocity, are important determinants of reproductive maturity; however, these types of data were not available within our human cohort. Second, in female rats, changes in bile acid isoform composition in circulation occurred concurrently with changes in the microbiome alpha diversity and increased representation of gut bacterial species with predicted BSH activity. Tgr5 mRNA expression was confirmed within single ARC kisspeptin neurons in adult female mice and more broadly within the mediobasal hypothalamus in rats. Third, Tgr5 mRNA increased in a biphasic manner with an initial relatively small increase in expression during the early pubertal stages of development followed by a marked increase in tgr5 mRNA expression in late puberty, suggesting an estrogen dependent response. Future studies will be focused in characterizing the expression of Tgr5 in ARC KNDy and preoptic area kisspeptin neurons during pubertal maturation in both rodent models by double-labeling RNAscope. Specific pharmacological activation of TGR5 increased GnRH secretion in rat hypothalamic explants ex vivo, and this effect was dependent on kisspeptin receptor activity. Finally, early overexpression of human TGR5 within the ARC advanced age at vaginal opening and first ovulation in the rodent model. Taken together, our preliminary studies provide translational evidence of a possible hepatic-gut microbiome-neuroendocrine circuit that contributes to reproductive maturation via bile acid metabolism and central TGR5 receptor activity.

In the pilot study, we observed bile acid differences in the relative concentration of conjugated vs deconjugated bile acids and higher circulating secondary bile acids, consistent with a possible deconjugation event during the pubertal transition. While the differences in deconjugated and secondary bile acids did not reach statistical significance in the secondary dataset, this is somewhat expected as the premenarcheal cohort was at least Tanner III, considerably older than the prepubertal cohort of the pilot study (11 vs 7 years), and more similar in reproductive stage; Tanner staging and menarche are clinical approximations of reproductive maturity and therefore not precise. Future research using cohorts established a priori for the study of bile acid physiology and reproductive maturity leveraging more physiological tools to measure reproductive stage in humans are needed to objectively align the changes in bile acid isoforms with reproductive stage. Our finding of increased concentrations of deconjugated bile acids are supported by a recent study evaluating the role of bile acids in nonalcoholic fatty liver disease (NAFLD) in pre- and postpubertal children and adolescents with obesity (40). Montagnana and colleagues reported glycine-conjugated bile acids were higher in prepubertal vs pubertal subjects (40). However, in a departure from our findings, decreases in glycine-conjugated bile acids were not associated with age in females and the authors did not report significant differences in secondary bile acids between developmental stages. Similar to our secondary cohort, the prepubertal stage was on average older (median 10, IQR 9-11 years) vs the prepubertal cohort in our analysis (median 7, IQR 4-8 years). Therefore, it is possible the cohorts in the study by Montagnana and colleagues were too close in reproductive age and as a result, differences in bile acid profile shifts were undetected.

We did not detect differences in the total bile acid pool across our reproductive cohorts, which disagrees with some, but not all previous research (38-40). Differences in fasting status among our cohorts and unclear stage of reproductive maturity in our pre–/early pubertal cohort may have contributed to discrepancies in our study vs others. The pre–/early pubertal cohort was not fasted at the time of blood collection which may have resulted in a higher overall bile acid pool estimate in that cohort (68) and confounded our ability to detect differences in the total bile acid pool from the fasted early postmenarcheal cohort. Moreover, our pre–/early pubertal cohort ranged from 4 years old to 9 years old, therefore our cohort likely covered a spectrum from prepubertal through early pubertal, creating heterogeneity within that cohort itself. To address these limitations, these comparisons were also conducted in a controlled, experimental rodent model to reveal significant increases in the bile acid pool. Our future research will seek to replicate these human findings in a larger, well-phenotyped cohort across more precise and defined stages of reproductive maturity. In summary, while our total bile acid pool data are interpreted cautiously given the design limitations, we report robust and novel shifts in the bile acid isoform profile between pre–/early pubertal and postmenarcheal females, toward deconjugation and an increased circulating pool of secondary bile acids.

The changes in bile acid isoform composition we observed across reproductive stages implies a role for the gut microbiome in mediating deconjugation events and increased serum levels of secondary bile acids during reproductive maturation. Indeed, other investigators have hypothesized that the gut microbiome modulates the onset of menarche, although no known mechanism has been clearly identified (41). Unfortunately, human stool samples across all 3 cohorts examined in our study were not available for analysis. As a result, we turned to a rodent model to investigate the relationship between changes in the circulating bile acid pool and changes in the diversity of the gut microbiome during the pubertal transition.

Discrete changes in the rodent gut microbiome concurrent with the observed bile acid isoform profile shifts suggest a role for the gut microbiome in regulating the conjugation status of the bile acid isoform pool. Indeed, the sterolbiome (69) is an emerging field of research, acknowledging the endocrine actions of bile acids mediated through the gut microbiome (reviewed elsewhere) (69). In the present study, we provide mechanistic insights into precisely how the timing of reproductive milestones may be shifted by the hepatic-gut microbiome-neuroendocrine axis. We observed dramatic changes in microbial composition and diversity of the gut microbiome in our rodent model which favored an increase in microbiota with predicted BSH activity (27) and occurred parallel to significant bile acid isoform profile shifts consistent with our human cohort. Our study setup does not allow isolating cage effects from life stage effects as individuals were co-housed per life stage. Co-housing rodents typically reduces compositional variation among them due to coprophagy. Therefore co-housing might have dampened variation within and between groups. The physiological importance of deconjugation and increased secondary bile acids in circulation may reflect the relative affinity of these bile acid isoforms for TGR5 signaling. As reported by Kawamata and colleagues (70), the secondary bile acids TLCA, LCA, and DCA have the highest relative affinity for TGR5 binding compared to other bile acid isoforms (71, 72). Given the similarities of shifts in the bile acid isoform composition in adolescent girls and female rats, it is a plausible hypothesis that deconjugation and conversion to secondary bile acids via the gut microbiome led to the increased secondary bile acid concentrations detected in circulation in our discovery cohort of postmenarcheal adolescents. We emphasize the need to interpret the microbiome data conservatively, as our rodent model and human participants were at developmentally different stages of reproductive maturation. Unfortunately, we cannot disentangle cage effects and age effects from the observed microbiome effects, but this is the best experimental model we could use without reducing the number of animals per cage over time and generating the confounding factor of increasing food availability. It is currently not clear how the shifts in microbial diversity are mediated during this life stage; changes in alpha diversity and increased microbial populations capable of BSH enzyme activity and bile acid deconjugation may be related to weaning and nutritional status during this peri-adolescent period (73). Future research will test whether shifting gut microbiome composition toward (or away from) preferential BSH enzyme activity advances or delays pubertal timing.

Several metabolic signals have been postulated to influence the activation of the neuroendocrine brain at puberty, including IGF-1, leptin, and ghrelin (7, 8, 12, 22, 74, 75). These metabolic signals, and in particular IGF-1 and leptin, have been shown to play a permissive role in the pubertal process in humans and rodent models (9, 76, 77). In the present studies, we used leptin receptor expression as a positive control to contrast tgr5 expression within the hypothalamus. Unlike leptin receptor expression, which was continuously expressed across all PN stages, we observed dramatic upregulation of tgr5 mRNA within the mediobasal hypothalamus during the pubertal transition in the female rat. The timing of these events along with coincident changes in microbiome diversity and increased unconjugated/secondary bile acids suggests there may be coordination of metabolic activity within the hepatic-gut microbiome-neuroendocrine axis. Our studies advance the understanding of how metabolic cues such as bile acid/TGR5 signaling may influence reproductive maturation in female rodents by demonstrating that TGR5 receptor activation can increase GnRH secretion ex vivo and this effect required kisspeptin receptor activity. It may not be surprising that TGR5 receptor activation increased GnRH secretion within this model to a lesser extent compared to the effect of kisspeptin and may reflect the role of bile acid/TGR5 signaling as a modulator of the neuroendocrine axis rather than a requirement for attainment of puberty. Further, we selected PN28 in female rats for these studies since this was the timing of the initial small rise in TGR5 mRNA expression within the MBH. A major limitation of our study is the use of a TGR5 agonist at a single postnatal age; this approach demonstrates the presence of a functional TGR5 receptor in the hypothalamus and its action on GnRH release without providing further insight into the endogenous mediators of hypothalamic TGR5 activation. Future studies will focus on the combined actions of both kisspeptin and primary and secondary bile acids on TGR5-mediated stimulation of GnRH release between PN28 and later life stages, or the use of TGR5 blockers to inhibit endogenously mediated bile acid receptor activation. Our results suggest that these 2 signaling systems interact to control GnRH neuronal activity during the pubertal transition.

The timing of puberty (at least in the female rats) aligns closely with upregulation of tgr5 expression within the mediobasal hypothalamus. Perhaps our most compelling studies relate to acceleration of first ovulation following the overexpression of human TGR5 within the ARC prior to the endogenous increase in tgr5 mRNA expression during the pubertal transition. Our goal was to increase expression of TGR5 closer to the time when the endogenous shift in bile acid isoform composition was occurring. Further in vivo pharmacological testing will be required to identify the nature of the endogenous bile acid species and the exact window of action on the activation of the pubertal process. Because secondary bile acids are reabsorbed along the gut lumen and can re-enter the systemic circulation as ligands with high affinity for TGR5 (24, 25), we posit that bile acid isoform shifts toward deconjugation and secondary bile acid synthesis via the gut microbiome may serve as a metabolic sensor acting centrally via TGR5 to alter the timing of reproductive maturation.

Early onset of pubertal milestones, such as menarche, are associated with significant short- and long-term psychological and physical health burdens, such as depression and anxiety, increased risk of cardiovascular disease, type 2 diabetes, and reproductive complications such as infertility (78-83). Living with obesity in childhood is associated with earlier achievement of pubertal milestones (precocious puberty) in males and females (1), although obesity may disproportionally affect females (84-88). In the present manuscript, we describe a novel metabolic mechanism capable of modulating the timing of reproductive maturity in rodents and observe similar differences in the physiologically relevant bile acids in both the pilot and secondary cohort. Examining the potential role of the hepatic-gut microbiome-neuroendocrine axis under physiological conditions may increase our understanding of the link between altered metabolic and normal reproductive function and provide a clearer basis for nutritionally relevant mechanisms that underlie reproductive maturation.

Abbreviations

aCSF

artificial cerebrospinal fluid

ARC

arcuate nucleus

ASV

amplicon sequence variant

BA

bile acid

BSH

bile salt hydrolase

CA

cholic acid

CDCA

chenodeoxycholic acid

DCA

deoxycholic acid

eGFP

enhanced green fluorescent protein

FXR

farnesoid X receptor

G

glycine (glycol)

GFP

green fluorescent protein

GnRH

gonadotropin-releasing hormone

IGF-1

insulin-like growth factor 1

KNDy

kisspeptin, neurokinin B, and dynorphin

LCA

lithocholic acid

LH

luteinizing hormone

LV

lentiviral

MBH

medial basal hypothalamus

ME

median eminence

PCR

polymerase chain reaction

PN

postnatal day

RT

reverse transcription

TGR5

G-protein coupled bile acid receptor

Funding

These studies were funded by National Institute of Child Health and Human Development grants 5R01HD084542 and R21AG061141 to A.L. Portions of the human research reported in this publication were supported by the National Institute of Child Health and Human Development under Award Number R21HD095372 and a Weill Cornell-Cornell Multi-Investigator Seed Grant. Texas AgriLife Research funding to H.V. supported portions of the human data biochemical analyses. D.V. received funding from the Horizon 2020 and innovation programme under the Marie Sklodowska-Curie grant agreement No 895167. I.L.B. is funded by the Pew Charitable Trusts and the David and Lucile Packard Foundation. H.V. was concurrently funded by the Canadian Institutes of Health Research (postdoctoral fellowship) during portions of the analysis and writing of the manuscript. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Author Contributions

H.V.B.: statistical analysis, data curation, writing—original draft, review, and editing. M.S.R. and A.L.: hypothesis and experimental design, data collection, writing—original design, review and editing. D.V.: data curation, formal analysis, visualization; I.L.B.: resources, supervision—data curation and analysis, writing—review and editing. O.K.R.: performed research, analyzed data, data curation, writing—methods, review and editing.

Disclosures

H.V.B. has received travel support from the National Cattlemen's Beef Association. A.L. has nothing to disclose. M.S.R. has nothing to disclose. D.V. has nothing to disclose. I.L.B. has nothing to disclose. O.R. has nothing to disclose.

Data Availability

Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.