Estrogen Regulation of the Molecular Phenotype and Active Translatome of AVPV Kisspeptin Neurons

Abstract

In females, ovarian estradiol (E₂) exerts both negative and positive feedback regulation on the neural circuits governing reproductive hormone secretion, but the cellular and molecular mechanisms underlying this remain poorly understood. In rodents, estrogen receptor α–expressing kisspeptin neurons in the hypothalamic anteroventral periventricular region (AVPV) are prime candidates to mediate E₂ positive feedback induction of preovulatory gonadotropin-releasing hormone and luteinizing hormone (LH) surges. E₂ stimulates AVPV Kiss1 expression, but the full extent of estrogen effects in these neurons is unknown; whether E₂ stimulates or inhibits other genes in AVPV Kiss1 cells has not been determined. Indeed, understanding of the function(s) of AVPV kisspeptin cells is limited, in part, by minimal knowledge of their overall molecular phenotype, as only a few genes are currently known to be co-expressed in AVPV Kiss1 cells. To provide a more detailed profiling of co-expressed genes in AVPV Kiss1 cells, including receptors and other signaling factors, and test how these genes respond to E₂, we selectively isolated actively translated mRNAs from AVPV Kiss1 cells of female mice and performed RNA sequencing (RNA-seq). This identified >13 000 mRNAs co-expressed in AVPV Kiss1 cells, including multiple receptor and ligand transcripts positively or negatively regulated by E₂. We also performed RNAscope to validate co-expression of several transcripts identified by RNA-seq, including Pdyn (prodynorphin), Penk (proenkephalin), Vgf (VGF), and Cartpt (CART), in female AVPV Kiss1 cells. Given the important role of AVPV kisspeptin cells in positive feedback, E₂ effects on identified genes may relate to the LH surge mechanism and/or other physiological processes involving these cells.

A fundamental tenet of reproductive axis regulation is sex steroid feedback. In females, sex steroids secreted from the ovaries act in the brain to exert both negative and positive regulation on the neural circuits that govern reproductive hormone secretion. In female rodents, estradiol (E₂) and progesterone (P4) act during diestrus and estrous to dampen and control the frequency and amplitude of pulsatile gonadotropin-releasing hormone (GnRH) secretion from forebrain neurons (negative feedback), thereby governing the downstream pulsatile secretion of luteinizing hormone (LH) from the pituitary (1-3). Conversely, on proestrus, higher levels of circulating E₂ act in concert with P4 to stimulate a large “surge” secretion of GnRH (positive feedback), which itself stimulates the LH surge that drives ovulation (4-6). Despite these known effects, the cellular and molecular mechanisms by which sex steroids act in the brain to inhibit or stimulate GnRH secretion still remain poorly understood.

Because GnRH neurons lack estrogen receptor α (ERα), which mediates E₂ positive and negative feedback, other “upstream” ERα-expressing cells have been proposed as neural loci of E₂ feedback actions. Kisspeptin neurons are prime candidates to directly receive E₂ signals and mediate ERα-induced effects on GnRH secretion (7). Encoded by Kiss1, kisspeptin is a potent activator of GnRH neurons and downstream LH secretion (8-14), and mutations in Kiss1 or Kiss1r (the kisspeptin receptor) induce striking reproductive hormone deficits and infertility (15-17). In the rodent hypothalamus, kisspeptin neurons are detected primarily in 2 regions, the continuum in the anterior hypothalamus comprised of the anteroventral periventricular nucleus (AVPV) and neighboring periventricular nucleus (PeN) (for simplicity, referred to here as the AVPV; sometimes also called the RP3V) and the arcuate nucleus (ARC) (11,18,19). Both the AVPV and ARC kisspeptin populations express sex steroid receptors, including ERα (20,21), and directly stimulate GnRH neurons, with AVPV kisspeptin neurons projecting to GnRH soma in the preoptic area (POA) and ARC kisspeptin neurons projecting to GnRH neurons’ distal projections near the median eminence (22,23).

Although ARC and AVPV Kiss1 cells both express ERα, E₂ regulates the Kiss1 gene oppositely in the 2 regions, inhibiting Kiss1 expression in the ARC but stimulating Kiss1 expression in the AVPV (19,20,24,25). Correspondingly, in the absence of circulating E₂ (eg, after gonadectomy), Kiss1 messenger RNA (mRNA) levels increase in the ARC but decrease in the AVPV (19,20,24,26). The differential effects of E₂ on ARC and AVPV Kiss1 neurons likely relate to sex steroid feedback control of GnRH secretion. The ARC region is implicated in sex steroid negative feedback on GnRH pulse secretion (27-30), and ARC Kiss1 neurons likely participate in this process (31). Indeed, evidence now suggests that E₂-responsive ARC Kiss1 neurons, which also coexpress the neuropeptides neurokinin B and dynorphin, are key components of the GnRH pulse generator circuitry (32-37). In contrast, Kiss1 neurons in the AVPV mediate E₂ positive feedback induction of the sexually dimorphic GnRH surge that only occurs in females (5,38,39). Supporting this, Kiss1 levels in the AVPV are greater in females than males (19,40,41), and AVPV kisspeptin neurons in females but not males exhibit cellular activation during LH surges (21,39).

Although ERα is expressed in AVPV Kiss1 neurons, the extent and nature of estrogen effects in these neurons is not fully known. Besides stimulating Kiss1 expression, E₂ may also have important stimulatory or inhibitory effects on other genes in AVPV Kiss1 cells, but this has not yet been determined. Given the important role of AVPV kisspeptin neurons in E₂ positive feedback (5,42-44), E₂ effects on other genes in these neurons may affect the LH surge mechanism, although other unknown physiological processes that AVPV kisspeptin neurons might participate in may also be affected. In fact, our understanding of the various function(s) of AVPV kisspeptin neurons is currently limited, in part, by not knowing the extensive molecular phenotype of these neurons. Because the AVPV region also contains many nonkisspeptin cell populations, phenotyping the kisspeptin neurons has thus far been restricted to histological double-label analyses. Consequently, only a small handful of genes (or their protein products) besides ERα are currently known to be co-expressed in AVPV kisspeptin cells, including galanin (Gal), tyrosine hydroxylase (TH), met-enkephalin (encoded by the proenkephalin gene, Penk), vasopressin receptor 1a (Avpr1a), and P4 receptor (Pgr) (45-49). In some case, this has proven useful for further elucidating the function and regulation of these kisspeptin cells, for example, identifying the important roles of P4 and vasopressin in the activation of AVPV kisspeptin neurons during the LH surge (44,45). Yet, the functional significance of other known co-expressed factors, like TH (46) or galanin, is still unknown.

A more detailed and extensive knowledge of the co-expressed genes in AVPV kisspeptin neurons, including steroid and peptide receptors, and signaling factors, and how these genes respond to different regulatory signals, like E₂, is needed to fully understand the regulation and function(s) of these neurons. To address this gap in knowledge, the present study leveraged the recently developed RiboTag technique (50-52) to selectively isolate actively translated mRNAs from just the Kiss1 cells of the AVPV. The RiboTag approach has an added advantage of analyzing only mRNAs that are associated with polyribosomes and actively being translated into proteins (the “active translatome”), regardless of their transcriptional status. Combining this Ribotag isolation methodology with RNA sequencing (RNA-seq), we were able to (1) identify >13 300 mRNAs that are significantly co-expressed and actively being translated in AVPV Kiss1 neurons and (2) quantitatively compare the expression levels of these Kiss1 neuron-specific mRNAs in females under different E₂ conditions to determine how estrogens alter the molecular profile of AVPV kisspeptin neurons.

Methods

Animals

To identify which gene transcripts are co-expressed and actively translated specifically in AVPV Kiss1 neurons, we used the Ribotag mouse model (50-52). Ribotag (Rpl22^HA+) mice have loxP sites flanking exon 4 of the ribosomal protein L22 (RPL22) gene, followed by an identical coding sequence for exon 4 that has 3 hemagglutinin (HA) epitope sequences prior to the gene’s STOP codon (51,52). When Ribotag mice are crossed with a Cre mouse line, Cre-mediated recombination results in the expression of HA tags on ribosomes in cells expressing Cre. HA-tagged ribosomes can be isolated by immunoprecipitation and the accompanying RNA obtained, permitting isolation of actively translated mRNAs specifically from Cre expressing cells. Here, we crossed the Rpl22^HA+ mice with a well-validated Kiss1Cre line developed by Elias (53) to obtain Kiss1Cre+/Rpl22^HA+/+ progeny to be used for experimentation (n = 28). In these Kiss1Cre+/Rpl22^HA+/+ mice, recombination of the RPL22 gene only occurs in Kiss1 cells, and hence, the HA-tagged ribosomal protein will be incorporated into the ribosomes of only Kiss1 cells. As an internal control for the specificity of the Ribotag-mediated isolation of mRNAs, we used a small cohort of Kiss1Cre−/Rpl22^HA+/+ control mice (n = 8), which lack the HA tag and therefore should not permit HA pull-down of transcripts. All Kiss1Cre/Rpl22^HA+/+ mice were genotyped by polymerase chain reaction (PCR) of tail DNA (referred to hence as Kiss1Cre+/Ribotag or Kiss1Cre−/Ribotag mice); these mice were also genotyped for unwanted germline recombination, and such mice were excluded from the study. In the final experiment, adult (10-12 weeks of age) female C57BL6 mice (n = 3) were used for histological validation of the RNA-seq expression data via in situ hybridization (ISH) analysis. All mice were housed 2 to 3 mice/cage in a 12:12 light:dark cycle (lights off at 6 pm), with ad libitum access to food and water. All animal procedures were approved by the local IACUC committees at the University of California, San Diego (Kiss1Cre/Ribotag mice) or Albany Medical College (C57BL6 mice).

Hormone Treatment and Tissue Collection

Expression of the Kiss1 gene in the AVPV is well-known to be stimulated by E₂. Therefore, at 8 weeks of age, all female Kiss1Cre/Ribotag mice (Kiss1Cre+/Rpl22^HA+/+ and Kiss1Cre−/Rpl22^HA+/+ controls) were ovariectomized (OVX) under isoflurane anesthesia and briefly exposed to high E₂ for 4 days, via a subcutaneous silastic capsule, to stimulate the AVPV Kiss1 gene. This dose of E₂ has previously been shown in mice by our lab and others to elevate serum E₂ levels, increase AVPV Kiss1 expression, and properly suppress morning LH levels for E2 negative feedback (18,20,25,40,47,54,55). We theorized that this brief high E₂ pre-treatment would ensure strong Cre expression to fully drive recombination and incorporation of the HA-tagged ribosomes in AVPV Kiss1 cells similarly in all animals. After this 4-day pretreatment, all E₂ silastic capsules were removed, and the mice (all already OVX) were given 1 week to allow any residual circulating E₂ to be depleted. Following this E₂ washout period, some mice were implanted with another similar high E₂ silastic capsule to serve as the E₂ group (n = 12 Kiss1Cre+; n = 4 Kiss1Cre− controls), while the remaining females received no additional hormone treatment to serve as the OVX (no E₂) group (n = 16 Kiss1Cre+; n = 4 Kiss1Cre− controls). Five days later, all mice were sacrificed midday between 11 am and 2 pm, and blood and brains were collected. Blood serum collected at sacrifice was assayed to confirm elevated basal LH levels in the OVX group (due to lack of E₂ negative feedback) and low LH levels in the E₂ group (due to proper E₂ negative feedback provided by the implant). Serum LH was measured in singlet via a sensitive mouse LH RIA (lower detection limit: 0.04 ng/mL; average reportable range: 0.04-75 ng/mL) performed by the University of Virginia’s Ligand Assay and Analysis Core. As expected, OVX females without E₂ implants had elevated mean LH levels, whereas E₂-treated OVX females showed reduced mean LH levels (3.04 ± 0.22 ng/mL vs 0.22 ± 0.04 ng/mL, respectively; P < 0.05).

Brains from Kiss1Cre/Ribotag females were immediately collected on dry ice and stored at −80^oC. The anterior hypothalamic-POA region, including the AVPV, was microdissected and 2 consecutive 400 um thick coronal slices spanning the entire AVPV region were micropunched (2 mm diameter) per animal, as in previous studies (40). Micropunched brain samples were pooled within E₂ and OVX treatments (n = 4 mice/pooled sample) to allow for sufficient yield of isolated mRNA using the Ribotag immunoprecipitation method. There were a total of 4 pooled Cre+ OVX samples, 3 pooled Cre+ E₂ treatment samples, and 1 pooled sample each from OVX and E₂ Cre− controls. Pooled micropunched AVPV samples were stored in 1.7 mL Eppendorf tubes at −80^oC until immunoprecipitation was performed.

For the triple-label ISH co-expression experiment, ovary-intact adult C57Bl6 mice (n = 3) were used. For these C57Bl6 females, vaginal lavage and cytology were performed daily to track estrous cycle stage, and once in diestrus, female mice were sacrificed at midday (12-1 pm), and their brains collected onto dry ice and stored at −80°C. Frozen brains from these diestrus C57Bl6 females were sectioned on a cryostat into 20 um coronal sections, with sections encompassing the AVPV region and corresponding to approximately 0.01 mm anterior to bregma. These AVPV-containing brain sections were mounted onto SuperFrost Plus slides (Fisher Scientific) and stored at −80^oC until the triple-label ISH assay.

Immunoprecipitation and RNA Extraction

Immunoprecipitation was performed on pooled Kiss1Cre/Ribotag AVPV samples following the recent protocols used by Sanz and colleagues (51,52) with a few modifications to optimize the procedure for AVPV Kiss1 cells. Pooled tissue samples were homogenized in a buffer solution at 3% weight by volume (homogenization buffer solution: 72% H₂O, 9.6% NP-40, 9.6% 2M KCl, 3.2% 1.5M Tris-pH 7.4, 1.2% 1M MgCl_2, 2% cyclohexamide 5 mg/mL, 1% protease inhibitors, 1% heparin 100 mg/mL, 0.5% RNAsin, and 0.1% dithiothreitol). Samples were then centrifuged for 10 min (10K) at 4^oC and 10% of the sample lysate was transferred to a new tube, serving as the input (IN) sample. This Input sample contains all cell types found in the AVPV region, including but not restricted to Kiss1 neurons. An amount of 350 uL of lysis buffer (Qiagen Kit #74034) was added to the Input sample, vortexed, flash frozen, and stored at −80^oC. Biolegend Purified anti-HA, 11 Epitope Tag Antibody (#901501; 0.25 uL of antibody/100 uL lysate) was added to the remaining lysate and incubated at 4^oC for 2 h on a sample rotator. After the 2-h incubation, the sample was transferred to a tube of washed magnetic beads (Pierce Protein A/G #88803; 25 uL beads/100 uL sample). Magnetic beads were washed with 800 uL of the homogenization buffer solution as previously described prior to adding the sample to the beads. The antibody/bead sample was incubated for 1 h at 4^oC on a sample rotator. After 1 h, the sample tubes were placed on a magnetic rack, the lysate was removed, and 800 uL of a high salt buffer (53.4% H₂O, 30% 2M KCl, 10% NP-40, 3.3% 1.5M Tris-pH 7.4, 1.2% 1M MgCl_2, 2% cyclohexamide 5 mg/mL, and 0.05% dithiothreitol) was added to wash the beads. The samples were then incubated for 10 min at 4°C on a sample rotator before the beads were washed again using the high salt buffer. This procedure was repeated 3 times, after which 350 uL of lysis buffer (Qiagen Kit #74034) was immediately added to each sample and vortexed for 30 sec. The samples were then placed in the magnetic rack and the lysate (termed the “IP sample”) was removed and stored at −80^oC until RNA purification. The IP samples contain only RNA that was associated with HA-tagged ribosomes (ie, RNA specific to Kiss1 neurons in Cre+ mice). RNA was extracted from both the IN and IP samples using the Qiagen RNeasy Plus Micro Kit (#74034) per the kit instructions and then stored in aliquots at −80°C until reverse transcription PCR (RT-PCR) or RNA-seq.

RT-PCR Confirmation of Kiss1 Cell-specific mRNAs

To validate that the Kiss1Cre/Ribotag method effectively isolated mRNA from AVPV Kiss1 neurons and was also selective (ie, did not isolate mRNA from other non-Kiss1 cells in the micropunched tissue), we performed RT-PCR for 2 genes known to be co-expressed in AVPV Kiss1 cells (Kiss1 and TH) and a gene known to not be co-expressed in Kiss1 neurons (GnRH, which is expressed in some scattered cells residing in the greater AVPV/POA region). RNA (10 ng) from the input and IP samples was used to make complimentary DNA (cDNA) using the iScript cDNA Synthesis Kit (Bio-rad #1708891), per kit instructions. The cDNA was then used to test for the presence of Kiss1, TH, and GnRH in each sample, using RNA-specific primers for each transcript (Kiss1 Forward: CTGTGTCGCCACCTATGGGG; Kiss1 Reverse: GGCCTCTACAATCCACCTGC; TH Forward: CCCCCACCTGGAGTACTTTG; TH Reverse: ATCACGGGCAGACAGTAGACC; GnRH Forward: ATGGCCGGCATTCTACTGC; GnRH Reverse: CTGCTGGGTATAAAAACGC). cDNA (1 uL) was added to Jumpstart RedTaq mix (Sigma #P0982), primers, and H₂O, and RT-PCR was run with the following conditions: 94 × 15’ (94 × 30”, 57 × 30”, 72 × 60”), repeat × 30 times; 72 × 5’, and 4 × 5’, with the GnRH PCR run with an annealing temperature of 52^oC.

RNA-seq of Cre+ Samples and Bioinformatic Analysis

RNA from the Kiss1Cre+/Ribotag IP samples of the E₂ and OVX groups was prepared and sequenced by the Genomics Center at University of California San Diego’s Institute for Genomic Medicine, which has extensive experience running RNA-seq. The quality of all RNA samples was determined using the Agilent High Sensitivity RNA ScreenTape System. Only samples with an RNA integrity number > 8.0 were used for library preparation. Unstranded mRNA library kits from Illumina that used polyA enrichment were used to create the library. RNA-seq was performed using Illumina’s HiSeq4000 platform, using the SR75 run type. Quality control analysis (Fastqc v0.11.8) was performed on the raw RNA-seq data, followed by read trimming (Trimmomatic 0.38), alignment (STAR v2.6.0a), and quantification (RSEMv1.3.0) of reads using GRChm38.p6/mm10 and Mus_musculus.GRCm38.98.gtf.

All RNA-seq analysis and statistics were completed by the Center for Computational Biology & Bioinformatics at University of California at San Diego. The per-gene/per-sample count data from the quality control and count preparation was used in conjunction with the per-sample metadata RNA-seq data, to integrate and annotate these inputs in preparation for data exploration and preprocessing using the edgeR (56) Bioconductor (57) package written in R (57). This analysis was limited to known protein coding genes. The annotated data were used with same edgeR Bioconductor packages written in R and limma (58) to explore and preprocess these data in preparation for differential expression testing. This data exploration file was used to identify transcripts produced by AVPV Kiss1 cells. Gene transcripts with counts per million (CPM) greater than 1 for at least 3 samples (the smallest group size in this experiment) were retained for further analysis, while all others were discarded. The data set is deposited to the Gene Expression Omnibus data repository (59). Using the same limma and edgeR Bioconductor packages written in R, we then used the voom technique (60) to test for differential expression of identified transcripts between the E₂ and OVX (no E₂) groups. The overall and differential expression data were then used with the WebGestalt (61) and R package to test for the presence of certain genes and differential expression of genes between E₂ and OVX in annotated functions, pathways, and diseases. We also used the PathView (62) tool render and visualize Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway maps (63) by differential expression findings, with very low expressing transcripts (mean log CPM <1) in both E₂ and OVX groups excluded from differential expression and KEGG pathway analyses. The KEGG database links gene catalogs from completely sequenced genomes to higher-level systemic cell and organismal functions and displays them in >500 KEGG pathway maps (https://www.genome.jp/kegg/pathway.html) representing current knowledge of the molecular interaction, reaction, and relation networks for different biological categories (metabolism, cellular processes, organismal systems, etc). The top KEGG pathways represented in our AVPV Kiss1 cell RNA-seq data set were determined, and for differential gene expression and pathway analyses comparing E₂ and OVX conditions, statistical significance was set at P < 0.05.

Validation of Gene Co-expression Using Triple-Label In Situ Hybridization

To validate that several highly expressed genes identified in our Kiss1Cre/Ribotag RNA-seq experiment are in fact co-expressed in AVPV Kiss1 neurons, we performed 2 triple-label ISH assays using RNAscope. The first assay measured co-expression of Penk (proenkephalin) and Pdyn (prodynorphin) in AVPV Kiss1 neurons, 2 genes previously reported to be expressed in the AVPV region of rats (64). The second assay studied co-expression of Cartpt [cocaine- and amphetamine-regulated transcript prepropeptide (CART)] and Vgf (VGF nerve growth factor inducible), 2 genes highly expressed in the RNA-seq data that are also known to be expressed in the AVPV region (65,66) but have not previously been examined specifically in AVPV Kiss1 cells. We used the RNAscope fluorescent Multiplex kit with the following RNAscope catalogue riboprobes: Mm-Kiss1-C1—Mus musculus KiSS-1 metastasis-suppressor (Kiss1) mRNA, Mm-Pdyn-C2—Mus musculus prodynorphin (Pdyn) mRNA, Mm-Penk-C3–Mus musculus preproenkephalin (Penk) mRNA, Mm-Vgf-C2–Mus musculus VGF nerve growth factor inducible (Vgf) mRNA, and Mm-Cartpt-C3–Mus musculus CART prepropeptide (Cartpt) transcript variant 2 mRNA. The manufacturer’s recommended ISH protocol was followed. Briefly, brain tissue slices containing the AVPV were fixed in 4% paraformaldehyde at 4^oC for 15 min. The AVPV tissue was then dehydrated in 50%, 70%, and 100% ethanol washes for 5 to 10 min, a hydrophobic barrier was created around the tissue, and 5 drops of Protease IV was added to each tissue section. The tissue was incubated for 30 min and the Protease IV was washed off twice with 1× phosphate-buffered saline solution. Excess liquid was removed and a premixed riboprobe solution (115.4 uL Kiss1, 2.3ul Penk, and 2.3ul Pdyn per slide) was added to each slide. The probe hybridized to the tissue for 2 hours at 40^oC. The slides were then washed in 1× wash buffer (supplied with the RNAscope kit) twice for 2 min each. There were then 4 amplification steps alternating between 30 and 15 min per step. Between each amplification step, slides were washed twice in 1× wash buffer for 2 min. On the last amplification step, the Alt-C color module was used such that the fluorophores were Atto 550 for Kiss1, Atto 647 for Pdyn or Vgf, and Alexa 488 for Penk or Cartpt. Upon completion of the fourth amplification step and a final wash in 1× wash buffer, 4 drops of 4′,6-diamidino-2-phenylindole were added to each slide, which were incubated for 30 sec; excess DAPI was removed, and then the slides were immediately cover-slipped using ProLong Gold Antifade Mountant. Slides were stored at 4^oC in the dark.

Images of mRNA fluorescent staining were obtained for the middle AVPV/ PeN region of each female using a Zeiss LSM 880 at 40× (oil) magnification. Images were analyzed using ZEISS Zen Blue software. For each female, the number of identifiable Kiss1-expressing neurons in the AVPV were counted, with a minimum of 90 Kiss1-expressing cells counted per female (all in diestrus stage). The number of Penk- and Pdyn-expressing cells or Cartpt- and Vgf-expressing cells that were co-localized with Kiss1-expressing cells was also counted. Using the RNAscope manufacturer’s counting criteria as a guideline, an AVPV cell was considered to express Kiss1 (or the other 2 genes) if there were ≥15 clustered dots, which represent individual mRNA copies for that specific gene. The percentage of AVPV Kiss1 neurons co-expressing Penk and/or Pdyn, or Cartpt- and/or Vgf, was determined for each female and then mean co-expression levels calculated.

Results

Validation of Ribotag Immunoprecipitation and Specificity of AVPV Kiss1 Cell Isolation

To confirm successful and selective immunoprecipitation pulldown of mRNA from AVPV Kiss1 cells, RT-PCR was performed using cDNA from both the IN samples (mRNA from all cells in the micropunched AVPV region) and IP samples (mRNA isolated from only Kiss1 cell ribosomes). As expected, both the IN and IP samples from Kiss1Cre+/Ribotag females expressed Kiss1 (Fig. 1), indicating successful isolation of mRNA from AVPV cells that produce Kiss1. Given prior histological reports of TH mRNA (and protein) co-expression in AVPV Kiss1 neurons (47,67,68), the TH gene was used as another positive control; as expected, TH was expressed in both the Kiss1Cre+ IN and IP samples (Fig. 1). To test whether mRNA isolation was specific to AVPV Kiss1 cells, RT-PCR was performed for GnRH, which is known to be expressed in the greater AVPV/POA region but not in Kiss1 cells. GnRH was detected in the IN sample of all Kiss1Cre+ mice, signifying its presence in the micropunches as expected but was not detected in any of the corresponding IP samples of the same mice, indicating that the mRNA isolation and pulldown procedure was specific to just the Kiss1 cells (Fig. 1). In further support of the pulldown specificity, none of the IP samples from Kiss1Cre− control mice, which lack HA-tagged ribosomes, expressed Kiss1, TH, or GnRH (Fig. 1), even though all 3 mRNAs were expressed in the IN samples of these Cre-control mice (Fig. 1). Collectively, these findings support that the isolation procedure selectively targeted mRNA from Kiss1 cells, but not other AVPV cells.

Figure 1.

RT-PCR verification of successful isolation of mRNA specifically from AVPV Kiss1 cells in adult Kiss1Cre+/Ribotag female mice. The number above each lane denotes the pooled sample number (n = 4 mice/pooled sample). Input (IN) samples contain mRNA from all cell types, including Kiss1 cells, in the micropunched tissue of the greater AVPV region (prior to the selective HA immunoprecipitation step). IP samples contain transcripts selectively isolated with the HA immunoprecipitation process, which is dependent on Cre and should only occur in Kiss1 cells of Cre+ mice. Thus, the IP lanes for Cre+ samples (IP#1 to IP#7) signify mRNA identified in AVPV Kiss1 cells, including Kiss1 and Th as predicted. By contrast, GnRH mRNA is present in all IN samples as expected but not in the IP of any Cre+ samples (IP#1 to IP#7), supporting that the pulldown is selective to mRNA from Kiss1 cells. As an additional validation, Cre− controls, which do not have HA-tagged ribosomes regardless of hormone condition, do not contain mRNA in their IP samples (IP#8 and IP#9), even though Kiss1, Th, and GnRH are all expressed in the Cre− IN samples (IN#8, IN#9) as expected. cDNA and PCR negative controls are located on the right of the gel, and 100 bp ladder is shown on both sides of the gel.

RNA-Seq Quality Control and Gene Transcript Expression in AVPV Kiss1 Cells

The RNA integrity value of all Kiss1Cre+ IP samples was 9.4 or greater, indicating RNA quality sufficient for sequencing via RNA seq. These samples all exceeded the minimum desired sequencing parameters in the number of total reads (≥20 million), number of aligned reads (≥10 million), and fraction of uniquely aligned reads (≥80%), providing confidence in the quality of the RNA. The mean per-position Phred scores were high (above 30) at all positions, further indicating high sequence quality. Levels of overrepresented sequences and Adapter Content were low, suggesting that the library prep was also of high quality.

RNA-seq of the Kiss1Cre+/Ribotag samples indicated that Kiss1 cells in the AVPV of adult females produce approximately 13 300 different mRNA transcripts, including Kiss1 and genes previously known to co-expressed with AVPV kisspeptin, like TH, Gal, Pgr, and Esr1 (Fig. 2). We first determined what genes were most highly expressed in AVPV Kiss1 cells, regardless of hormonal status. Secretogranin II (Scg2), a member of the secretogranin family of neuroendocrine secretory proteins, was the transcript with the highest expression. Other very highly expressed transcripts included several genes related to intracellular signaling, such as Gprasp1, Gna, Atp1a3, Calm1, and Ywhag, genes important for protein synthesis and regulation, including Cpe, Hspa8, Hsp90ab1, Eef1a1, Ubb, and Ubc, genes involved in essential cellular processes, like Gaa, Pkm, Kif1a, and Ckb, and genes involved with secretion and synapses, like App, Syp, and Rtn1. The 50 transcripts with the highest overall mean expression (ie, the mean expression of all OVX and E₂ samples) in AVPV Kiss1 cells are listed in Table 1.

Figure 2.

A glimma plot demonstrating the expression of >13 300 transcripts selectively isolated from adult female AVPV Kiss1 cells and subjected to RNA-seq. Each dot represents a single gene transcript. The x axis represents the overall mean expression levels of the transcripts, with higher x axis values indicating higher mRNA expression. The y axis represents the expression difference between females treated with E₂vs OVX females, with greater positive values indicating higher mRNA levels in E₂ females and greater negative values signifying higher mRNA levels in OVX females. Green dots are gene transcripts that are expressed significantly more (P < 0.05) in the E₂vs OVX condition, grey dots are transcripts that are expressed similarly between OVX and E₂ conditions, and orange dots are transcripts that are expressed significantly more (P < 0.05) in the OVX vs E₂ condition. Several genes previously identified in AVPV Kiss1 cells using traditional histological assays are noted by arrows.

Table 1.

The top 50 genes with the greatest expression in female AVPV Kiss1 cells

ENSEMBL ID	ENTREZ ID	Gene	Total mean expression	Mean E2 expression	Mean OVX expression	Adjusted P-value
ENSMUSG00000050711	20 254	Scg2	12.20	12.64	11.87	< 0.01
ENSMUSG00000027523	14 683	Gnas	11.69	11.66	11.71	0.76
ENSMUSG00000015656	15 481	Hspa8	11.67	11.75	11.61	0.26
ENSMUSG00000025579	14 387	Gaa	11.66	11.70	11.63	0.61
ENSMUSG00000019505	22 187	Ubb	11.47	11.51	11.44	0.77
ENSMUSG00000040907	23 2975	Atp1a3	11.44	11.51	11.40	0.32
ENSMUSG00000037742	13 627	Eef1a1	11.41	11.54	11.31	0.06
ENSMUSG00000043388	243 339	Tmem130	11.40	11.48	11.35	0.34
ENSMUSG00000022892	11 820	App	11.36	11.59	11.19	< 0.01
ENSMUSG00000025151	94 275	Maged1	11.22	11.84	10.75	< 0.01
ENSMUSG00000023944	15 516	Hsp90ab1	11.21	11.24	11.18	0.58
ENSMUSG00000043384	67 298	Gprasp1	11.10	11.22	11.02	0.25
ENSMUSG00000006930	15 114	Hap1	11.10	11.19	11.04	0.19
ENSMUSG00000055430	58 243	Nap1l5	11.09	10.86	11.26	0.04
ENSMUSG00000008348	22 190	Ubc	11.05	11.03	11.05	0.90
ENSMUSG00000026223	64 294	Itm2c	11.03	11.45	10.71	< 0.01
ENSMUSG00000004207	19 156	Psap	11.01	11.16	10.90	0.02
ENSMUSG00000021270	15 519	Hsp90aa1	10.94	10.95	10.93	0.89
ENSMUSG00000019923	52 696	Zwint	10.93	10.93	10.94	0.99
ENSMUSG00000006651	11 803	Aplp1	10.93	10.95	10.91	0.83
ENSMUSG00000032294	18 746	Pkm	10.89	11.03	10.79	0.06
ENSMUSG00000058297	94 214	Spock2	10.88	11.08	10.72	< 0.01
ENSMUSG00000001175	12 313	Calm1	10.85	11.23	10.57	< 0.01
ENSMUSG00000030695	11 674	Aldoa	10.77	10.90	10.68	0.14
ENSMUSG00000020315	20 742	Sptbn1	10.76	10.84	10.70	0.84
ENSMUSG00000002265	18 616	Peg3	10.75	10.72	10.78	0.82
ENSMUSG00000070802	43 4128	Pnmal2	10.73	10.65	10.80	0.27
ENSMUSG00000023004	22 143	Tuba1b	10.65	10.64	10.66	0.95
ENSMUSG00000039278	30 052	Pcsk1n	10.62	10.54	10.69	0.51
ENSMUSG00000072235	22 142	Tuba1a	10.60	10.57	10.62	0.75
ENSMUSG00000014602	16 560	Kif1a	10.60	10.65	10.55	0.51
ENSMUSG00000062825	11 465	Actg1	10.60	10.65	10.55	0.43
ENSMUSG00000019986	52 906	Ahi1	10.59	10.63	10.56	0.68
ENSMUSG00000033585	17 984	Ndn	10.56	10.39	10.69	0.19
ENSMUSG00000051391	22 628	Ywhag	10.56	10.67	10.48	0.10
ENSMUSG00000031144	20 977	Syp	10.51	10.64	10.42	0.14
ENSMUSG00000034994	13 629	Eef2	10.48	10.60	10.39	0.06
ENSMUSG00000041607	17 196	Mbp	10.44	10.22	10.61	0.40
ENSMUSG00000057738	20 740	Sptan1	10.44	10.47	10.42	0.93
ENSMUSG00000037852	12 876	Cpe	10.44	10.43	10.44	0.93
ENSMUSG00000018451	103 712	6330403K07Rik	10.38	10.30	10.44	0.46
ENSMUSG00000066357	83 669	Wdr6	10.38	10.39	10.37	0.88
ENSMUSG00000021087	104 001	Rtn1	10.34	10.43	10.28	0.25
ENSMUSG00000026204	19 275	Ptprn	10.32	10.61	10.10	< 0.01
ENSMUSG00000025855	19 085	Prkar1b	10.30	10.40	10.22	0.188
ENSMUSG00000001270	12 709	Ckb	10.29	10.52	10.12	0.01
ENSMUSG00000026576	11 931	Atp1b1	10.27	10.39	10.18	0.09
ENSMUSG00000025203	20 250	Scd2	10.26	10.12	10.37	0.29
ENSMUSG00000074657	16 572	Kif5a	10.26	10.35	10.19	0.28
ENSMUSG00000036699	72 693	Zcchc12	10.25	9.71	10.66	< 0.01

Transcripts are listed in descending order of total mean expression (mean log CPM, regardless of hormone condition). P < 0.05 signifies significant differential expression between e2 and ovx conditions, denoted in bold type.

The RNA-seq analysis revealed genes for many receptors, neuropeptides, and neurotransmitter synthesis that are co-expressed in female AVPV Kiss1 cells. Along with confirming the few previously known co-expressed ligands (Gal, Th, and Penk), we identified a number of new ligand transcripts in AVPV Kiss1 cells (Table 2). For example, we found that AVPV Kiss1 cells, like ARC Kiss1 cells, express abundant levels of Pdyn under both E₂ and OVX conditions, although significantly higher in OVX (Table 2). Vgf, Pnoc, and Cartpt were some additional newly-described ligand transcripts (Table 2). Genes relating to gamma-aminobutyric acid (GABA) synthesis or transport, like Vgat, Gad1, and Gad2 were also expressed at very high levels, with the glutamate-related gene Vglut2 also present at moderate levels. Along with confirming known co-expressed receptors ERα (Esr1), Pgr, vasopressin receptor 1a (Avpr1a), and prolactin receptor (Prlr), AVPV Kiss1 cells were also found to express a number of other receptors important for neuroendocrine or neural function, such as thyroid hormone receptor (Thra), CRH receptor (Crhr1), glucocorticoid receptor (Nr3c1), androgen receptor (AR), kappa opioid receptor (Oprl1), and multiple variants of the neuropeptide Y receptor (Npy1r, Npy2r, Npy4r, Npy5r) (Table 2). Interestingly, while Esr1 was very highly expressed, Esr2 (estrogen receptor beta) was only present at very low levels (Table 2).

Table 2.

Genes related to a number of ligands (top) and receptors (bottom) expressed to different degrees in AVPV Kiss1 cells

LIGANDSENSEMBL ID	ENTREZ ID	Gene	Overall Mean expression	Mean E2 expression	Mean OVX expression	Log FC	Adjusted p value
ENSMUSG00000045573	18 619	Penk	10.164	10.405	9.984	-0.422	0.057
ENSMUSG00000037428	381 677	Vgf	9.928	10.307	9.644	-0.664	0.001
ENSMUSG00000026787	14 417	Gad2	9.757	10.01	9.566	-0.445	0.003
ENSMUSG00000070880	14 415	Gad1	9.460	9.519	9.417	0.102	0.351
ENSMUSG00000037771	22 348	Vgat	8.640	8.673	8.716	-0.076	0.655
ENSMUSG00000000214	21 823	Th	8.655	9.177	8.263	-0.915	<0.001
ENSMUSG00000116158	280 287	Kiss1	7.192	8.759	6.016	-2.734	<0.001
ENSMUSG00000024907	14 419	Gal	8.287	8.686	7.987	-0.701	0.020
ENSMUSG00000021647	27 220	Cartpt	7.372	8.104	6.823	1.278	<0.001
ENSMUSG00000029361	18 125	Nos1	7.314	7.538	7.145	-0.393	0.010
ENSMUSG00000027400	18 610	Pdyn	7.207	6.824	7.494	0.669	0.022
ENSMUSG00000045731	18 155	Pnoc	6.652	6.970	6.414	-0.556	0.025
ENSMUSG00000031980	11 606	Agt	6.311	5.922	6.603	0.687	0.059
ENSMUSG00000030500	140 919	Vglut2	5.762	5.840	5.704	-0.140	0.537
ENSMUSG00000005892	22 044	Trh	5.285	5.220	5.333	0.106	0.838
ENSMUSG00000049796	12 918	Crh	5.075	5.023	5.115	0.086	0.905
ENSMUSG00000061762	21 333	Tac1	4.554	4.313	4.735	0.432	0.229
ENSMUSG00000024256	11 516	Adcyap1	4.147	4.459	3.913	-0.557	0.152
ENSMUSG00000029236	56 183	Nmu	4.132	4.145	4.121	-0.032	0.926
ENSMUSG00000032532	12 424	Cck	3.495	4.005	3.112	-0.902	0.157
RECEPTORSENSEMBL ID	ENTREZ ID	Gene	OverallMean Expression	Mean E 2 expression	Mean OVXexpression	Log FC	Adjusted P value
ENSMUSG00000058756	21 833	Thra	9.075	8.895	9.210	0.314	0.143
ENSMUSG00000029778	11 517	Pacapr1	8.478	8.358	8.569	0.211	0.163
ENSMUSG00000019768	13 982	Esr1	7.884	7.672	8.044	0.372	0.105
ENSMUSG00000027584	18 389	Oprl1	7.176	7.086	7.244	0.157	0.303
ENSMUSG00000024431	14 815	Nr3c1	6.542	6.296	6.726	0.430	0.056
ENSMUSG00000005268	19 116	Prlr	6.476	6.883	6.171	-0.714	0.011
ENSMUSG00000044288	12 801	Cnr1	6.408	6.092	6.645	0.554	0.023
ENSMUSG00000046532	11 835	Ar	6.279	6.394	6.192	-0.205	0.366
ENSMUSG00000039059	99 296	Hrh3	6.180	6.350	6.052	-0.301	0.243
ENSMUSG00000025905	18 387	Oprk1	6.022	5.874	6.133	0.254	0.215
ENSMUSG00000028004	18 167	Npy2r	5.908	4.206	7.185	2.976	<0.001
ENSMUSG00000031870	18 667	Pgr	5.759	6.936	4.875	-2.064	<0.001
ENSMUSG00000021779	21 834	Thrb	4.223	4.296	4.169	-0.132	0.655
ENSMUSG00000055737	146 00	Ghr	4.046	4.068	4.030	-0.044	0.895
ENSMUSG00000036437	18 166	Npy1r	4.031	4.360	3.785	-0.595	0.117
ENSMUSG00000030043	21 336	Tacr1	3.847	3.731	3.934	0.201	0.485
ENSMUSG00000044014	18 168	Npy5r	3.524	2.967	3.942	0.964	0.029
ENSMUSG00000047259	17 202	Mc4r	3.363	3.537	3.232	-.0318	0.496
ENSMUSG00000018634	12 921	Crhr1	3.067	2.796	3.270	0.480	0.289
ENSMUSG00000011171	22 355	Vipr2	3.032	2.912	3.122	0.208	0.640
ENSMUSG00000020123	54 140	Avpr1a	1.919	1.440	2.277	0.845	0.136
ENSMUSG00000021055	13 983	Esr2	1.379	1.358	1.395	0.033	0.971
ENSMUSG00000048337	19 065	Npy4r	0.130	2.620	-1.740	-4.360	< 0.001

Transcripts are listed in descending order of overall mean expression (mean logCPM from all samples, regardless of hormonal condition). P-values < 0.05 denote significantly different expression between OVX and E₂ conditions.

Differential Expression of Gene Transcripts in AVPV Kiss1 Cells

Kiss1 cells in the AVPV are regulated by E₂, with higher AVPV Kiss1 expression when E₂ is elevated and lower Kiss1 expression in OVX (no E₂) females. One of our goals was to identify what other specific transcripts are significantly increased or decreased in AVPV Kiss1 cells under conditions of elevated E₂vs absent E₂. Of the >13 300 transcripts in AVPV Kiss1 cells, 683 gene transcripts were significantly differentially expressed based on E₂ status, with 484 transcripts, including Kiss1, being significantly higher in the E₂ condition (Fig. 2, green dots; P < 0.05) and 199 transcripts being significantly higher in the OVX condition (Fig. 2, orange dots; P < 0.05). The remaining >12 600 genes did not show significantly altered expression levels with or without E₂ (Fig. 2, gray dots). Fig. 3 depicts a heat map showing the expressions levels of the top 90 genes with the greatest differential expression between E₂ and OVX conditions; interestingly, the vast majority (~90%) of these top 90 differentially expressed genes, including Kiss1, were higher in the E₂ condition.

Figure 3.

A heat map depicting the top 90 (based on t value) AVPV Kiss1 cell transcripts that are most differentially expressed (DE) between E₂ and OVX conditions. Of these 90 gene transcripts, 79 were more highly expressed in E₂-treated females, whereas 11 were expressed more in the OVX condition. Row Z-scores indicate the relationship between the transcript expression level for that sample and mean expression level of that transcript, with red representing high gene expression and blue representing low gene expression. The individual sample numbers, 4 for the OVX condition and 3 for E₂ condition, are provided at the top of the heat map underneath the appropriate hormonal treatment.

Table 2 compares the expression levels between OVX and E₂ conditions for a number of relevant ligands (or their synthesizing proteins) and receptors found in AVPV Kiss1 cells. As expected, differential expression analysis revealed that the known E₂-responsive gene Pgr was, like Kiss1, significantly higher in E₂vs OVX (no E₂) females (Figs. 2 and 3; Table 2). Several other ligand or receptor genes, such as TH, Gal, Gad2, Prlr, and Npy4r, also exhibited significantly greater expression levels with E₂ treatment compared to OVX (P < 0.05), whereas fewer genes, like Npy2r, Npy5r, Pdyn, and Cnr1, were expressed higher in OVX than E₂ conditions (P < 0.05; Fig. 3; Table 2). Many other ligand and receptor genes, like Esr1, Penk, Gad1, Vgat, and Vglut2 were expressed at similar levels between OVX and E₂ conditions (Table 2).

Biological Pathways Regulated by Gene Transcripts Expressed in AVPV Kiss1 Neurons

KEGG pathway analysis (63) was performed to determine how RNA transcripts expressed by AVPV Kiss1 cells cluster within specific known biological pathways and whether the expression of these clustered transcripts tends to activate or inhibit these biological processes. Of the top 10 KEGG pathways with the lowest false discovery rate (pGFdr) represented by AVPV Kiss1 cell RNA transcripts, regardless of hormone status, there were several signaling pathways, such as the mitogen-activated protein kinase and calcium signaling pathways, as well as pathways involved in amphetamine addiction and alcoholism (Table 3). Comparing specifically between the E₂ and OVX (no E₂) groups, there were only 3 KEGG pathways that were significantly different between the 2 hormone conditions (P < 0.05 for each), of which the neuroactive ligand-receptor interaction (NLRI) pathway contained the most genes and was the most significant (based on the lowest false discovery rate and the lowest P-value; Table 3). This particular KEGG pathway, shown in Figure 4, includes many neuropeptides and receptors known to regulate neural and endocrine function and/or reproduction. The specific gene groups (containing 1 or more genes) in the NLRI pathway found to be significantly expressed in AVPV Kiss1 cells are depicted as nonwhite boxes in Figure 4 (ie, all gray, green, and orange boxes); gene groups in the NLRI pathway that were not expressed in our data set are depicted as white boxes. Overall, our AVPV Kiss1 cell data set had 86 gene groups—comprising 175 individual gene transcripts—that were expressed in this NLRI pathway.

Table 3.

KEGG biological pathways represented by gene transcripts produced by AVPV Kiss1 cells

Pathway name	ID	Pathway genes expressed in AVPV Kiss1 cells (n)	Differentially expressed genes (n)	False discovery rate	P-value	Status
Top 10 KEGG pathways (regardless of hormonal status)
Mitogen-activated protein kinase signaling pathway	04010	219		0.00022
Amphetamine addiction	05031	60		0.00037
Regulation of actin cytoskeleton	04810	167		0.00063
Calcium signaling pathway	04020	143		0.00063
Alcoholism	05034	104		0.00088
Basal cell carcinoma	05217	42		0.00092
Herpes simplex infection	05168	127		0.00108
Cytokine-cytokine receptor interaction	04060	86		0.00108
Neurotrophin signaling pathway	04722	111		0.00899
RNA transport	03013	143		0.01451
KEGG pathways that differ significantly between E2 and OVX (no E2) conditions
Neuroactive ligand-receptor interaction	04080	175	32	0.00022	<0.001	Activated
Amphetamine addiction	05031	60	12	0.00574	<0.05	Activated
Jak-STAT signaling pathway	04630	89	13	0.01656	<0.05	Inhibited

The top 10 biological pathways (identified and sorted by the lowest false discovery rate), represented by transcripts expressed by AVPV Kiss1 cells, regardless of hormonal status. The only 3 KEGG pathways that differed significantly (P < 0.05) between E₂ and OVX females are listed at the bottom, with the neuroactive ligand-receptor interaction pathway containing the most differentially expressed transcripts. The status column refers to an estimated overall effect of all the identified genes, relative to the E₂ condition (ie, “activated” or “inhibited” more in the E₂ group).

Figure 4.

The neuroactive ligand-receptor interaction (NLRI) KEGG pathway and gene groups that are significantly expressed by AVPV Kiss1 cells in this pathway. Grey shaded boxes are genes or gene groups that are expressed but not significantly different between OVX and E₂ females. Green shaded boxes are genes or gene groups that are expressed at significantly greater levels (P < 0.05) in E2 than OVX females, whereas orange boxes depict greater expression (P < 0.05) in OVX females. White boxes are genes in this NLRI pathway that were not expressed in our Kiss1-Ribotag RNA-seq dataset.

Gene groups in the NLRI pathway that were differentially expressed (P < 0.05) between E₂ and OVX (no E₂) conditions are depicted in Figure 4 as green or orange boxes, respectively. Of the 175 individual transcripts in the NLRI pathway found to be expressed in AVPV Kiss1 cells, only 32 transcripts (6 for peptides and 26 for receptors), were differentially expressed between E₂ and OVX conditions, and these are shown in a heat map in Figure 5. Interestingly, 5 of the 6 differentially expressed peptide transcripts, including Kiss1, were more highly expressed with E₂ treatment, whereas Pdyn was the only protein ligand in the NLRI pathway whose RNA was expressed at higher levels in the OVX condition (Fig. 5). Of the 26 receptors in the NLRI pathway that were differentially expressed between E₂ and OVX, approximately half [such as cholecystokinin receptor A (Cckar), NPY receptor 4 (Npy4r), and prolactin receptor (Prlr)] had higher expression with E₂ while the other half [including NPY receptor 2 (Npy2r), calcitonin receptor (Calcr), and some GABA receptor subunits (Garbra3, Gabra4)] had greater expression in the OVX condition (Fig. 5).

Figure 5.

Heat map of the 32 individual AVPV Kiss1 cell transcripts in the NLRI KEGG pathway (Fig. 4) that are significantly differentially expressed (P < 0.05) between E₂ and OVX conditions. A total of 6 peptide transcripts and 26 receptor transcripts from AVPV Kiss1 cells were differentially expressed in the NLRI pathway and are listed alphabetically in the heat map. Row Z-scores indicate the relationship between the transcript expression level for that sample and mean expression level of that transcript, with red denoting high expression and blue denoting low expression. The individual sample numbers are provided at the top of the heat map underneath the appropriate hormonal treatment.

Validation of RNA-seq data using triple label ISH

Prior studies relied on co-labeling histology experiments to identify genes (or their protein products) co-expressed in AVPV Kiss1 cells; to date, such studies have demonstrated about a dozen such co-expressed genes or proteins (21,45-49,69,70). Our present RNA-seq data identified thousands of mRNA transcripts expressed in female AVPV Kiss1 cells, including many for signaling factors or their receptors. To use a second method to confirm that transcripts identified in our Kiss1Cre+/Ribotag experiment are in fact expressed in the AVPV cells that actively express Kiss1 in adulthood, we performed several ISH assays in AVPV sections collected from adult diestrus females. In the first assay, in addition to Kiss1 expression, we examined Penk because it was very highly expressed in our RNA-seq data (Table 2) and met-enkephalin, a protein products of Penk, was previously reported to be moderately (~33%) co-expressed with kisspeptin protein in the AVPV (assessed with IHC) (48). We also examined Pdyn because it was highly expressed in our RNA-seq data and was the only neuropeptide ligand differentially expressed at higher levels in the OVX group (Table 2, Fig. 5), which may have interesting functional implications. In addition, Pdyn is known to be co-expressed in ARC Kiss1 cells but has not yet been shown to be expressed in other Kiss1 populations. In a triple-label ISH assay, we found that Kiss1, Penk, and Pdyn are all highly expressed in the AVPV region of diestrus females and show considerable co-expression with each other (Fig. 6). More than 50% of AVPV Kiss1 cells also co-express Penk (Fig. 6C and 6E), and similarly, ~50% of AVPV Kiss1 cells co-express Pdyn (Fig. 6C and 6E). Moreover, there were many identified AVPV cells where all 3 genes were co-expressed (Fig. 6C and 6D); specifically, ~32% of AVPV Kiss1 cells co-expressed both Penk and Pdyn (Fig. 6E) in diestrus females.

Figure 6.

Triple-label ISH analysis of Kiss1, Penk, and Pdyn coexpression in the female mouse AVPV. (A) A representative image of cells in the AVPV of a diestrus female, with Kiss1 (red), Penk (green), and Pdyn (purple) expression depicted. (B and C) Higher magnification images of the white boxed area in (A), showing good Kiss1 coexpression with Penk and/or Pdyn in the AVPV. (D) Higher magnification image of the 3 cells from the yellow box in C, showing co-expression of Kiss1/Penk (green arrow), Kiss1/Pdyn (purple arrow), or Kiss1/Penk/Pdyn (white arrow) in the AVPV. (E) Mean + SE percentage coexpression between Kiss1, Penk, and Pdyn in AVPV cells (n = 3 diestrus females).

In another ISH assay, we validated 2 additional candidate genes, Cartpt and Vgf, that were highly expressed in the RNA-seq data (Table 2). Both Cartpt (CART) and Vgf (VGF nerve growth factor inducible) are known to be expressed in the AVPV region but have not been examined specifically in Kiss1 cells. Confirming our RNA-seq results, we found that both Cartpt and Vgf show significant co-expression with Kiss1 in the AVPV region of diestrus females (Fig. 7). Approximately 55% of AVPV Kiss1 cells co-express Cartpt (Fig. 7C and 7E), while ~40% of AVPV Kiss1 cells co-express Vgf (Fig. 7C and 7E). In these diestrus females, ~25% of identified AVPV Kiss1 cells co-expressed both Cartpt and Vgf (Fig. 7D and 7E).

Figure 7.

Triple-label ISH analysis of Kiss1, Cartpt, and Vgf co-expression in the AVPV of adult female diestrus mice. (A) A representative image of AVPV from a diestrus female, with Kiss1 (red), Cartpt (green), and Vgf (purple) expression depicted. (B and C) Higher magnification images of the white boxed area in (A), showing Kiss1 co-expression with Cartpt and/or Vgf. (D) Higher magnification image of cells from the yellow box in (C), showing co-expression of Kiss1/Cartpt (green arrow), Kiss1/Vgf (purple arrow), or Kiss1/Cartpt/Vgf (white arrow). (E) Mean + SE percentage coexpression between Kiss1, Cartpt, and Vgf in AVPV cells (n = 3 diestrus females).

Discussion

Kisspeptin neurons in the AVPV region are important components of the LH surge mechanism that triggers ovulation in females (5,42,71). Prior studies demonstrated that E₂ upregulates Kiss1 gene expression in the AVPV (19,20) and permits circadian activation of these neurons in synchrony with the LH surge (21,39). Yet, how these AVPV Kiss1 neurons are regulated, positively or negatively, by other hormones or upstream neurons in relation to the surge process is still not completely characterized, nor have other possible co-transmitters in these Kiss1 neurons been fully determined. In addition, whether or not AVPV Kiss1 neurons participate in other reproductive and/or nonreproductive processes still remains unclear, in part because the phenotype of these cells is incompletely understood. To begin to address these outstanding issues, the present study asked several key questions, including (1) what other genes are actively co-expressed in AVPV Kiss1 cells of adult females? (2) Are there other co-expressed genes in these AVPV cells besides Kiss1 that are also under E₂ regulation, and, if so, are these other genes stimulated or inhibited by E₂? and (3) Are there other specific ligands or receptors present in AVPV Kiss1 cells that could lend insight into possible functions or regulation of these cells? The heterogenous nature of the AVPV region has restricted prior investigations to double label ISH or IHC studies. While useful, such histology approaches are costly and time consuming for assessing many genes and typically rely on studying targeted candidates one by one. We overcame this limitation here by using the Ribotag technique to selectively isolate all actively translated mRNA from just AVPV Kiss1 neurons, permitting identification of thousands of co-expressed gene transcripts and identifying estrogen effects on both known and novel genes that may be critical for AVPV Kiss1 neuron function.

Our goal was to isolate mRNA transcripts from Kiss1 neurons to identify on a large scale which other transcripts are co-expressed in these cells, at what levels these transcripts are expressed (high, medium, low), whether E₂ exposure alters such expression levels, and, if so, in which direction (stimulatory, inhibitory). Before running the RNA-seq, we first used 4 complimentary assessments to validate that our Ribotag method properly isolated Kiss1 cell-specific transcripts. First, we performed PCR to confirm that Kiss1 mRNA was in fact present in the IP of Cre+ samples. Second, we used PCR to similarly confirm that Th, a gene previously shown with double ISH and IHC (for protein) to be highly coexpressed with AVPV Kiss1, was also expressed in the IP of Cre+ samples (Fig. 1). Third, we confirmed that GnRH, which is present in the greater AVPV/POA area but not in Kiss1 neurons, was expressed in the IN of all samples but not in the IP of any Cre+ samples (Fig. 1). This supports that the pulldown is selective and not isolating mRNA from non-Kiss1 cells (ie, GnRH cells) that are present in the IN. Fourth, we performed PCR on IN and IP samples from Cre− controls that do not have HA-tagged ribosomes in Kiss1 cells (or any other cells). The Cre− samples all expressed Kiss1, Th, and GnRH in their IN, as expected, but did not express any of these genes in their IP (Fig. 1), confirming that the pulldown method relies on Cre+ expression and does not appear to nonselectively isolate mRNA. Collectively, these validation steps showed that our Ribotag procedure isolated several mRNAs known to be in Kiss1 neurons, did not isolate mRNA known to be absent in Kiss1 neurons, and did not nonselectively isolate mRNA if Cre+ was absent. This gives confidence that the mRNAs identified with the subsequent RNA-seq are likely from AVPV Kiss1 neurons. Further supporting this, our RNA-seq data confirmed expression of the dozen or so known genes (or their protein products) previously reported to be co-expressed to varying degrees in AVPV Kiss1 cells, including Th (46, 47), Gal (48), Penk (48), Esr1 (20, 21), Esr2 (20, 69), Pgr (45), Avpr1a (49), Crh1 (70), Prlr (72), and Mc4R (53). Finally, as discussed more in the following text, triple-label ISH (Figs. 6 and 7) confirmed that Pdyn, Cartpt, and Vgf, genes newly identified in our RNA-seq data, were indeed co-expressed with Kiss1 mRNA in the AVPV of female mice.

The RNA-seq identified approximately 13 300 genes that were significantly expressed in AVPV Kiss1 neurons. Not surprisingly, many of the highest-expressing genes were for standard cell maintenance, intracellular signaling, protein synthesis and regulation, and neural signaling/secretion. This was echoed in some of the top KEGG biological pathways implicated by the gene transcripts, which contain genes important for critical cellular processes of neurons such as cellular organization, axonal growth, and organelle transfer (regulation of actin cytoskeleton), intracellular signal transduction (mitogen-activated protein kinase signaling pathway, calcium signaling pathway), neural firing, neurotransmitter signaling, and intercell interaction (cytokine-cytokine receptor interaction, calcium signaling pathway, amphetamine addiction), neural cell differentiation and survival (neurotrophin signaling pathway), and protein synthesis and regulation (RNA transport). Thus, many of the genes represented in some of these top pathways are likely found in many neurons, including but not limited to AVPV Kiss1 neurons. Interestingly, transcripts found in AVPV Kiss1 cells were highly represented in 2 of the 5 substance-dependence KEGG pathways, amphetamine addiction and alcoholism, perhaps because both of these KEGG pathways heavily involve dopamine and glutamate signaling (along with dynorphin), which AVPV Kiss1 neurons are known to co-express but whose function in these neurons—reproductive or other—is still unknown. The KEGG pathway that was most differentially expressed between E₂ and OVX was the NLRI pathway, which contains Kiss1 and many other ligands and receptors involved in neural communication and regulation. This KEGG pathway revealed a number of newly identified co-expressed factors, including some that were E₂ sensitive, which may either be co-released by AVPV kisspeptin neurons or underlie the regulation of these neurons by afferent circuits. The amphetamine addiction KEGG pathway, which has a number of neural signaling factors represented like dopamine, glutamate, and dynorphin, was also differentially expressed between E₂ and OVX females.

Our overall mean expression analysis, along with the differentially expressed KEGG analysis, identified a number of neuropeptide ligands, or in the case of neurotransmitters, related synthesis or transport proteins, as well as receptors that are likely expressed in AVPV Kiss1 cells. Some of these co-expressed ligands and receptors were previously identified with double-label ISH or IHC assays, as previously noted. Along with those known genes, our RNA-seq data identified, for the first time, additional ligands and receptors of interest present in AVPV Kiss1 cells, including genes for neuropeptides (dynorphin, VGF nerve growth factor, CART, and nociceptin/orphanin FQ), neurotransmitters (GABA and glutamate), hormone receptors (thyroid hormone receptor, glucocorticoid receptor, and androgen receptor), and neuropeptide/neurotransmitter receptors (kappa opioid receptor, NPY receptor subtypes, and GABA receptors). These findings identify new signaling factors that may be co-released along with kisspeptin. Actions of these secreted factors could relate to the known role of AVPV kisspeptin cells in governing GnRH neurons during the preovulatory LH surge (E₂ positive feedback) or yet-to-be identified functions of AVPV kisspeptin cells in other process.

The novel identification of co-expressed receptors in AVPV kisspeptin cells could help identify how these cells are regulated by endocrine or neural factors. Like with the co-expressed ligands, the role of these expressed receptors could relate to the E₂-induced LH surge mechanism or to other reproductive or nonreproductive processes that are currently unknown. Interestingly, and perhaps not surprisingly, many of the newly identified ligands and receptors have been linked in one way or another to GnRH and LH secretion and/or reproduction, although whether any of those roles are due to these proteins specifically in AVPV kisspeptin cells is not yet known. For other gene transcripts not already linked to reproduction, future studies can test whether they are related to the LH surge or other reproductive processes or, conversely, have nonreproductive functions, as is the case with ARC kisspeptin cells, which have also recently been implicated in temperature control and metabolism/energy balance (73-79). Relatedly, it is also important to stress that the function of identified receptors may or may not be to regulate the Kiss1 gene (or kisspeptin protein) specifically, but rather may regulate other genes or processes in these cells. For example, AR signaling is known to have no effect on AVPV Kiss1 gene expression, suggesting AR’s role may be to affect other genes or processes in AVPV Kiss1 cells. Ideally, the present data can be used as a spring board for future hypothesis testing of possible functional roles of AVPV kisspeptin neurons.

Many prior studies demonstrated that AVPV Kiss1 expression is stimulated by E₂ and reduced in no E₂ (OVX) or low E₂ conditions. A present goal was to identify other transcripts in AVPV Kiss1 cells that are significantly increased or decreased by elevated E₂. This could help identify additional players in the LH surge mechanism of which AVPV kisspeptin cells are integral. A prior large-scale expression study identified a number of genes in the entire AVPV/POA region as a whole that change expression with E₂status, but it was not known in which specific cells, including Kiss1 cells, most changes were occurring (80). Here, we identified nearly 700 gene transcripts that appear to be differentially expressed in AVPV Kiss1 cells based on E₂ status, with almost 500 transcripts being stimulated by E₂, similar to the Kiss1 gene, and ~200 transcripts being reduced by E₂ (Figs. 2, 3, and 5; Tables 1 and 2). Note that it is currently unknown which of these gene effects are due to E₂ action directly in kisspeptin cells vs indirectly via other upstream neurons (or secondary effects). Supporting the possibility of some direct E₂ regulation, AVPV Kiss1 cells highly express Esr1 (Table 2) (20, 21). Esr2 was also expressed, although only at very low levels, which matches previous ISH findings (20) of much lower co-expression of Esr2 than Esr1 with Kiss1. Still, some of the differential expression of genes in our data set may possibly be indirect due to E₂ actions elsewhere. Regardless, it is likely that translational changes observed under the 2 E₂ conditions could ultimately relate to E₂-regulated processes, including most obviously the E₂-driven LH surge and perhaps other yet-to-be identified behaviors or physiological events modulated by estrogens.

The genes identified in this study give insight into possible functions and regulation of AVPV kisspeptin cells. We have tried to be conservative in our analyses, and the present RNA-seq data ultimately requires additional confirmation from other methods to provide converging evidence for Kiss1 cell coexpression. To this end, we performed several triple-label ISH assays to validate the finding of 4 highly expressed genes, Pdyn, Penk, Cartpt, and Vgf, in the RNA-seq data set. None of these 4 genes had previously been reported to be co-expressed with Kiss1 in the AVPV, although 1 of Penk’s protein products, met-enkephalin, had been reported to be expressed with ~1/3 of AVPV kisspeptin neurons in colchicine-treated mice (measured with IHC (48)). Pdyn is known to be co-expressed with ARC kisspeptin neurons but not yet shown for other kisspeptin populations (34,55,81-84). Supporting our RNA-seq findings, the RNAscope assays found that a large number—approximately 40% to 55%—of AVPV Kiss1 cells express either Pdyn, Penk, Cartpt, or Vgf. The percentage of Penk co-labeling with Kiss1 was slightly higher than in the prior met-enkephalin/kisspeptin IHC study (~50% vs ~33%), although this could be due to differences in techniques, RNA vs protein levels, or estrogen status (diestrus females in our study vs mixed cycle stage in the IHC study). Pdyn and dynorphin protein had previously been reported in the greater AVPV region of rats (64,85), but this is the first demonstration of Pdyn (or Cartpt and Vgf) specifically in AVPV Kiss1 cells. The functional role of dynorphin, or any of the other 3 co-expressed neuropeptides for that matter, in these Kiss1 cells is unknown, although it is interesting that Pdyn levels in AVPV Kiss1 cells increase in the absence of E₂, similar to Pdyn expression in the ARC region where it is also co-expressed with Kiss1 (55,81,83). Given that our ISH was in diestrus females, it is possible Pdyn/Kiss1 co-expression levels may be even higher than observed (50%) in other conditions where E₂ levels are lower or absent, such as OVX or reproductive senescence. Likewise, the degree of Kiss1 co-expression with Cartpt and Vgf, both higher with E₂ in our RNA-seq data, could change depending on sex steroid status.

Future studies can use the present RNA-seq findings to guide mechanistic interrogations of AVPV kisspeptin cell function and regulation. While our present ISH data validates the RNA-seq data for Pdyn, Penk, Cartpt, and Vgf genes, for thoroughness, additional co-labeling assays are needed in future studies to confirm co-expression of other genes of interest identified in this study. For such validation, it will be important to consider beforehand factors which may decrease or increase a given gene’s expression, such as sex steroid levels and female cycle stage, age, sex, circadian time of day, and metabolic status. For example, in our RNA-seq data, Npy4r was present at low to moderate levels under elevated E₂ but virtually absent in OVX samples, whereas Npy2r was expressed at much higher levels under OVX than E₂ conditions (Table 2). Indeed, our present data are limited to adult female mice with either elevated or absent circulating E₂. Whether similar genes are co-expressed in males or at different physiological states remains to be determined, as does assessment of gene expression at different circadian times, given that AVPV kisspeptin cells are under control of the suprachiasmatic nucleus clock (21,39). As a final consideration for future IHC validation studies, our RNA-seq measured actively translated mRNAs, which are likely to correlate with protein levels; yet, it is possible in some cases that posttranslational modifications could lead to differences between the mRNA levels identified here and protein levels.

There are several considerations when interpreting the present data. The first is that we isolated mRNA from Kiss1 neurons across the entire AVPV-PeN continuum. Currently, these Kiss1 cells are thought of as one population that extends from the AVPV into the PeN, and there have yet to be reports of differential co-expression (or functional differences) between AVPV and PeN Kiss1 neurons. Still, we do not know if specific RNA-seq genes are co-expressed with Kiss1 in just the AVPV nucleus, just the PeN nucleus, or both; future histological studies are needed to assess this. The second consideration is that the present method used Kiss1Cre-mediated induction of the HA ribosome tag, which remains present thereafter, regardless of subsequent Kiss1 gene expression levels. In female mice, the Kiss1 gene first turns on in the AVPV ~2 weeks after birth (47), increasing thereafter to reach adult levels by ~ 4 weeks old (86). Thus, at the time of brain collection in adulthood, some AVPV cells with HA-tagged ribosomes may no longer express Kiss1 at all or may not currently be expressing the Kiss1 gene at that moment. Thus, the RNA-seq–identified transcripts require further histological co-expression validation to see if these genes co-express with AVPV cells actively expressing Kiss1, as was done in our Kiss1/Pdyn/Penk and Kiss1/Cartpt/Vgf triple-label assays. Finally, although unlikely, we cannot exclude the small possibility that polyribosomes in axons projecting from arcuate Kiss1 neurons to the AVPV region may have been captured in the AVPV tissue punches. As with the other caveats, this can be addressed with double-label validation assays. Ultimately, we view the present work as a starting point and valuable resource to the field for testing new hypotheses of AVPV kisspeptin cell regulation and functions, but we emphasize that such studies should first validate and confirm the co-expression of identified genes of interest.

Additional Information

Disclosures: The authors have nothing to disclose.

Data Availability

Some data sets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author by reasonable request or are publicly available in the Gene Expression Omnibus data repository noted in the references.