Kisspeptin deficiency leads to abnormal adrenal glands and excess steroid hormone secretion

Abstract

Knockout mice for the kisspeptin receptor, Kiss1r (Kiss1r^−/−) and its ligand kisspeptin, Kiss1 (Kiss1^−/−) replicate the phenotype of isolated hypogonadotropic hypogonadism (IHH) associated with variants of these genes in humans. A recent report suggests that kisspeptin may be involved in human fetal adrenocortical development and function. Herein, we characterized the adrenal function and morphology in Kiss1^−/− mice that do not go through normal puberty. Two fetal markers were expressed in eosinophilic cells potentially derived from the X-zone that should disappear at puberty in male mice and during the first pregnancy in female animals. Although the hypercorticosteronism observed in Kiss1^−/− females corrected overtime, hyperaldosteronism persisted at 14 months and correlated with the overexpression of Star. To determine if KISS1 and KISS1R genes are involved in the development of primary aldosteronism (PA) and hypercortisolism [Cushing’s syndrome (CS)] in humans, we sequenced these 2 genes in 65 patients with PA and/or CS. Interestingly, a patient with CS presented with a germline KISS1 variant (p.H90D, rs201073751). We also found three rare variants in the KISS1R gene in three patients with PA: p.C95W (rs141767649), p.A189T (rs73507527) and p.R229R (rs115335009). The two missense variants have been previously associated with IHH. Our findings suggest that KISS1 may play a role in adrenal function in mice and possibly adrenocortical steroid hormone secretion in humans, beyond its recently described role in human fetal adrenocortical development.

Introduction

Almost two decades ago, inactivating variants in the kisspeptin receptor (KISS1R) gene were found in patients with idiopathic hypogonadotropic hypogonadism (IHH) (1). The essential role of KISS1R in the initiation of gonadotropin-releasing hormone (GnRH) signaling was shown in Kiss1r knockout (KO) mice (Kiss1r^−/−): indeed, Kiss1r^−/− mice developed IHH similar to what is observed in patients carrying KISS1R inactivating variants, and like the human patients, responded to exogenous GnRH treatment (1). KO mice for the KISS1R ligand, kisspeptin (Kiss1^−/−) developed a phenotype similar to that of Kiss1r^−/− mice, confirming that kisspeptin is the physiological ligand of KISS1R and that both molecules are needed for GnRH release (2,3).

Recently, it was described that kisspeptin treatment induced dehydroepiandrosterone sulphate (DHEAS) secretion by the human adrenocortical carcinoma cell line, H295R, and by cultured second-trimester human fetal adrenal (HFA) cells (4). Additional expression studies showed that kisspeptin is expressed widely in fetal adrenal cortex (4). These results supported a potential role of kisspeptin in the development of the adrenal and/or the regulation of adrenocortical function, at least during fetal development.

Following up on these reports, we studied adrenal pathology and secretion in Kiss1 KO mice: we demonstrate here that the inactivation of Kiss1 leads to persistence of the fetal X-zone in mice and that this was associated with hypersecretion of corticosterone and aldosterone. Although corticosterone levels normalized in older animals, hyperaldosteronism persisted. We then screened human patients with hypercortisolism (HC) or hyperaldosteronism caused by adrenal tumors. Interestingly, we identified one missense KISS1 and three KISS1R variants (two missense and one synonymous) among the patients with steroid hormone hypersecretion; all had been previously described in patients with IHH or Kallmann syndrome.

Results

In Kiss1^−/− mice, the X-zone persists

We analyzed the histology of adrenals from Kiss1^+/+ and Kiss1^−/− females at 5 and 12 months (Fig. 1A and D). At 5 months of age, Kiss1^−/−mice had eosinophilic cells that accumulated at the border between the cortex and the medulla (Fig. 1A, females; and Supplementary Material, Fig. S1A, males). Similar cells were not observed in the adrenal glands of wild-type (WT) Kiss1^+/+animals.

Figure 1.

The X-zone does not regress in Kiss1^−/− adrenals. (A) Histology analysis showed an accumulation of cells at the border of the cortex and the medulla that may be fetal X-zone cells in the adrenal glands of female Kiss1^−/− mice that were not observed in Kiss1^+/+ animals at 5 months of age. C: Cortex, M: Medulla. A white dashed line is indicating the border between the cortex and the medulla. Arrowhead points to the fetal cells composing the X-zone. (B) 20αHSD staining confirmed the fetal-like differentiation of the cells. The white arrowhead points to the fetal cells composing the X-zone. (C) The expression of X-zone markers, Akr1c18, Pik3c, Inha and Cyp17a1 was increased in the adrenal glands of female Kiss1^−/− mice (compared with Kiss1^+/+) both at 5 and 12 months of age, by RT qPCR. Bars represent the induction fold of at least four individual adrenals ± standard deviation. P-value was calculated using Student’s t-test. (D–E) Histology by hematoxylin and eosin (D) and 20αHSD staining (E) demonstrate that fetal cells were still observed in the adrenal glands ofKiss1^+/+and Kiss1^−/− animals at 12 months.

To characterize these cells, we used an antibody for the fetal marker, 20α-hydroxysteroid dehydrogenase (20αHSD). As shown in Figure 1B, 20αHSD-positive cells were present in the X-zone of Kiss1^−/− mice but not in WT Kiss1^+/+ animals (Fig. 1B, females; and Supplementary Material, Fig. S1B, males). Accordingly, mRNA of several X-zone markers, such as Akr1c18 (encoding for 20αHSD), Pik3c2g, Inhα and Cyp17a1 (5–8) was significantly increased in the adrenal glands ofKiss1^−/−female compared with WT Kiss1^+/+ animals at 5 months of age (Fig. 1C). In males, only Akr1c18 expression was significantly increased in the adrenal glands of Kiss1^−/− animals (Supplementary Material, Fig. S1C) consistent with the immunostaining results (Supplementary Material, Fig. S1B). The higher expression of Akr1c18 and Pik3c2g mRNA observed in the heterozygote, Kiss1^+/− animals was not statistically significant (there was significant heterogeneity between the animals studied).

We did not quantify the number of 20αHSD-positive cells per adrenal examined, because these cells were only seen in the adrenals of KO animals at 5 and 12 months of age. For example, we did not observe any 20αHSD-positive cells in the 5-month-old WT males, whereas the four KO males studied had numerous cells positive for 20αHSD. Similarly, none of the five parous WT females at 5 and 12 months of age that we studied had any 20αHSD-positive cells, whereas all the six KO females at 5 months and four out of 5 at 12 months of age had 20αHSD-positive cells (P < 0.005).

Altogether, these results demonstrated that cells that retained the expression of the fetal marker 20αHSD persisted at the cortico-medullary junction of the adrenal glands of Kiss1^−/− animals at 5 months of age. As previously described, cells retain fetal characteristics form the mouse X-zone (also called fetal zone), which is expected to degenerate after the first pregnancy in female mice and after puberty in male animals (9,10). Thus, the abnormal puberty and sterility previously described in Kiss1^−/− mice (2) could explain, at least in part, the persistence of the X-zone observed in Kiss1^−/− mice.

We therefore also studied the expression of X-zone markers in Kiss1^−/−females and compared it with that in nulliparous female Kiss1^+/− and WT Kiss1^+/+ animals, in which the X-zone did not degenerate. The expression of three of the four markers studied, Akr1c18, Pik3c2g and Cyp17a1 (Supplementary Material, Fig. S2), was significantly increased in the KO Kiss1^−/− mice compared with nulliparous Kiss1^+/+ females, suggesting that the X-zone was expanded in the adrenal glands of the KO Kiss1^−/− adrenals, even when compared with the X-zone found in age-matched WT adrenal glands with expected persistence of the X-zone. The persistence of the X-zone by histology (Fig. 1D) and 20αHSD staining (Fig. 1E) persisted too in the 12-month-old Kiss1^−/−female mice.

Kiss1^−/−mice had hypersecretion of glucocorticoids

We next analyzed the effect of Kiss1 inactivation on glucocorticoid secretion. In contrast to human adrenal cortex, adult mouse adrenocortical cells do not secrete cortisol due to the absence of Cyp17a1, a steroidogenic enzyme that is essential for the biosynthesis of both androgens and cortisol (8). As Cyp17a1 may be present in fetal adrenocortical cells and was slightly overexpressed in the adrenal glands of Kiss1^−/−animals at 5 months of age, we measured serum cortisol levels in these mice: cortisol was undetectable in all genotypes including, as expected, in WT animals. Serum corticosterone levels, on the other hand, were significantly increased in Kiss1^−/−females at 5 months of age, concomitantly with a decrease in ACTH levels (Fig. 2A), pointing to the adrenal origin of this hypersecretion. Like in other forms of mouse models of adrenocortical Cushing’s syndrome (CS), this biochemical phenotype was limited to the female sex, and Kiss1^−/− male mice had normal corticosterone and ACTH levels (Supplementary Material, Fig. S3A).

Figure 2.

Female Kiss1^−/− animals have primary hyperaldosteronism. (A) Serum corticosterone and ACTH concentration were up- and downregulated, respectively, in Kiss1^−/−female compared with Kiss1^+/+ animals at 5 months of age. Bars represent the mean of either corticosterone or ACTH level of at least six individual adrenals ± standard deviation. (B) RTqPCR for mc2r, Star, Cyp11a1, Hsd3b, Cyp21, Cyp11b1 and Cyp11b2 in Kiss1^+/+, Kiss1^+/− and Kiss1^−/− from the adrenal glands of 5-month-old female mice. Bars represent the mean of the hormone level of at least six individual glands ± standard deviation. P-value was calculated using Student’s t-test. (C) Immunofluorescence staining of Akr1b7 on Kiss1^+/+ and Kiss1^−/− adrenals at 5 months. (D) Serum aldosterone concentration was significantly increased in Kiss1^−/− females at both 5 and 12 months. Bars represent the mean of aldosterone level of at least six individual samples ± standard deviation.

Increased corticosterone levels were associated with the upregulation of three members of the steroidogenic pathway, Star, Cyp21 and Cyp11b1 (Fig. 2B). However, by histology, zona fasciculata looked normal with the expected expression of Akr1b7 at this stage (Fig. 2C). In the absence of any hyperplasia or tumors, both corticosterone and ACTH levels normalized in Kiss1^−/− females by 12 months of age (Supplementary Material, Fig.S4).

Kiss1  ^−/− mice developed hyperaldosteronism

Aldosterone levels in Kiss1^−/− mice were significantly elevated in both female and male animals at 5 months of age (Fig. 2D; females and Supplementary Material, Fig. S3C; males), and they remained increased at 12 months of age (Fig. 2D). The higher aldosterone levels observed in Kiss1^−/− mice were not associated with an increased expression of aldosterone synthase (Cyp11b2), neither at 5 nor at 12 months of age, and there was no hyperplasia or tumors histologically at either age (Fig. 1D).

At 5 months of age, the overexpression of Star and Cyp21 involved in the synthesis of both corticosterone and aldosterone (Fig. 2B) could, at least in part, explain the elevation of aldosterone in Kiss1^−/− female mice; however, it does not explain hyperaldosteronism in male mice and both Star and Cyp21 expression normalized, along with all other molecules of the steroidogenic pathway that we tested by 12 months of age in animals of both sexes (Supplementary Material, Fig. S3).

KISS1 and KISS1R genes were sequenced in a cohort of patients with adrenal tumors

We sequenced the KISS1 and KISS1R genes in a cohort of 66 patients with adrenal steroid hypersecretion syndromes due to a variety of tumors, including 36 patients with primary aldosteronism (PA), 20 with combined aldosteronism and hypercorticolism, and 10 with ACTH-independent CS (Table 1). One KISS1 variant, pH90D (c.268C > G, rs201073751) that is relatively frequent in the African population (AF: 0.022 in gnomAD) was identified in a patient with combined PA and HC (Table 2). This variant has been previously seen in a patient with central precocious puberty (CPP), leading investigators to believe that it may be activating KISS1 signaling; in silico it is predicted to be likely benign and in vitro studies that have been performed elsewhere did not show an increase of KISS1 activity or stability (11). In our investigation, we saw some decrease of the KISS1 expression by immunohistochemistry in the tumor of the patient carrying the variant; however, KISS1 gene sequencing in tumor cells failed to identify additional genetic changes or loss of heterozygosity (LOH) (data not shown).

Table 1.

Cohort of patients sequenced at germ line level for KISS1 and KISS1R genes

Patients’ diagnosis	n = 66		Tumor location
	Female	Male	Bilateral	Unilateral	N/A
PA	21	15	13	13	10
PA/HC	18	2	11	4	15
CS	10	-	6	1	3

Patients were divided into three groups based on their hormonal phenotype: patients with PA alone or combined with HC (PA/HC) and those with CS. The number and gender of patients as well as the tumor location are shown.

Table 2.

Allele frequency of KISS1 and KISS1R sequence variants identified in this study versus gnomAD exomes controls

Gene	DNA change	Protein change	SNP identification	In silico prediction	Patients	Allele frequency controls	References
KISS1	c.268C > G	p.H90D	rs201073751	Likely benign	1 PA/HC	0.00137	Silveira et al., 2010 (11); Chan et al., 2011 (12); Huijbregts et al., 2012 (13)
KISS1R	c.565G > A	p.A189T	rs73507527	Benign	1 PA	0.00456	Sykiotis et al., 2010 (14); Miraoui et al., 2013 (16)
KISS1R	c.285C > G	p.C95W	rs141767649	Uncertain significance	1 PA	0.0000192	Francou et al., 2016 (15)
KISS1R	c.687C > T	p.R229R	rs115335009	Likely benign	1 PA	0.00239	-

Abbreviations: PA: primary aldosteronism; PA/HC: primary aldosteronism/hypercortisolism and CS: Cushing’s syndrome.

Three KISS1R variants, two missense and one synonymous, were identified specifically among patients with PA. The two missense variants, p.A189T (c.565G > A, rs141767649) that are relatively frequent in African population (AF: 0.065 in gnomAD) and p.C95W (c.285C > G, rs141767649), were previously published in IHH cohorts; they have been tested in vitro (15,16). Only the tumor harboring the p.A189T variant was available to us; analysis of the tumor DNA did not reveal any LOH or other somatic events in KISS1R defect but, again, by immunohistochemistry, there was some decrease of KISS1R expression in tumor cells (data not shown).

Beyond adrenocortical hyperplasia, three of the patients with sequence variants had multinodular thyroid disease and/or kidney cysts. There was no history of malignancies and all patients to date have not developed any other tumors and remain free of any cancers.

Discussion

In this work, we report that in Kiss1 KO mice the X-zone persists in both male and female animals, as shown by the continuing expression of the Akr1c18, Pik3c8, Inha and Cyp17a1 markers. Interestingly, not all these molecules showed identical profiles in both sexes and across ages, pointing perhaps to a model of molecular regulation in Kiss1 KO mice that is different from the other animals showing X-zone persistence.

Cyp17a1 gene encodes for a steroidogenic enzyme essential for the production of cortisol and androgens in the adrenal cortex. Cyp17a1 expression is limited at the zona fasciculata and reticularis in the human cortex, but it is absent in the mouse definitive zone. Kiss1^−/− mice had Cyp17a1 overexpressed at the RNA level at 5 months of age, as one would expect from their persisting X-zone. It is noteworthy, however, that Cyp17a1 expression normalized at 12 months despite the X-zone still being present. This could be due to either the number of cells expressing Cyp17a1 being too low in the total RNA extracted from the whole adrenal gland (too low to be detected by RTqPCR at this stage), or that fetal cells of the type we observed elsewhere (see following text) may not be identical to those that express Cyp17a1 in Kiss1^−/− mice. In animals with adrenal cortex-specific complete inactivation of the main regulatory subunit type 1α of the PKA, Prkar1a (the AdKO mouse), fetal markers such as Akr1c18 and Cyp17a1 continue to be expressed highly across the mouse’s life span (7), but the phenotype of these mice is far more dramatic than the one we observed in the Kiss1^−/− mice: they developed CS due to an expansion of cells that were zona reticularis-like arising from the definitive cortex (17). In Kiss1^−/−mice, the increase in Cyp17a1 expression is not associated with hyperplasia, tumors or any form of cellular expansion, and consequently cortisol secretion, if any, remained below the detection level for any assay.

Our mouse data supported the observations made in human tissues and cells: KISS1R is widely expressed in the HFA cortex and may regulate adrenal steroid hormone secretion (4). Thus, we then investigated the possibility that KISS1 and KISS1R genetic variants may be present among human patients with various adrenocortical gluco- and mineralo-corticoid-secreting tumors. We identified three missense variants, one KISS1, two KISS1R and one KISS1R synonymous variant, among 66 patients with PA alone or combined with hypercortisolemia. Two of these variants, p.C95W (rs141767649) and p.A189T (rs73507527), have previously been described in patients with IHH (15,16). The functional consequences of the p.H90D variant remain unclear as it has previously been identified both in the homozygote state in a patient with CPP, as well as in IHH, in the heterozygote state (11), and it is relatively common in Africans, suggesting it may be a non-functional polymorphism. The carrier of this variant was also diagnosed with central hypogonadism at the age of 53, a year after undergoing adrenalectomy for his PA; at the time, he had neither CS nor hyperaldosteronism (data not shown).

These data overall pointed to the possibility that KISS1 and/or KISS1R are involved in the adrenocortical development and hormonal secretion. In older adrenal cortex, in both humans and mice, hyperaldosteronism may be the consequence of KISS1/KISS1R deficiency, although significantly stronger data need to be collected to conclude a definite causative effect. It should be noted that the first study that pointed to this possibility was by Nakamura et al. (18) when KISS1 was still known as ‘metastin’, a molecule with an effect in thyroid and other cancer metastases (19). It is noteworthy that out of four patients with KISS1/KISS1R variants, two also developed either thyroid multinodular disease (rs141767649) or kidney cysts (rs201073751), and a third one had both conditions (rs115335009). However, neither they, nor the mice we studied, have developed any other tumors to date, and their type of adrenocortical hyperplasia almost never develops into malignancy. Aging the Kiss1^−/− mice further in our laboratory did not lead to any differences in malignancies from their normal littermates (data not shown) despite the described associations between KISS1/KISS1R expression and their deregulation in cancer (20). Finally, these data are also supportive of KISS1/KISS1R signaling involvement in obesity, metabolic disorders and even gonadal steroid hormone perturbations (i.e. in polycystic ovarian syndrome) in a sexually dimorphic manner (21–23), as is the case in the Kiss1^−/− mice we studied here.

Materials and Methods

Patients and sequencing

All the patients signed consent forms and were admitted to the National Institute of Health (NIH) Clinical Center under research protocols approved by the Institutional Review Boards at NIH. Germ line and adrenal tumor DNA from patients were sequenced for KISS1 and KISS1R genes using the primers previously published (1).

Animal studies

Kiss1  ^−/− mice that have been previously described in the literature (1,2) have been kindly provided by Dr S. Seminara (Reproductive Endocrine Unit, Massachusetts General Hospital, Harvard Medical School, Boston, MA). Mice were maintained in a 14 h light (6 a.m.–8 p.m.)/10 h dark cycle. Both male and female mice at 5 and 12 months of age were sacrificed for the RNA, protein and blood collection as we have previously described (24). Procedures were approved by the Eunice Kennedy Shriver National Institute for Child Health and Human Development Institutional Animal Care and Use Committee. Mice were euthanized by slow replacement of air with CO₂ followed by cervical dislocation. Blood was collected by intracardiac puncture and placed in serum separator tubes (365956, BD microtainer). After centrifugation (4500 g, 3 min., 4°C), serum was stored at −80°C. Tissues were snap frozen on dry ice and stored at −80°C until use or fixed in 4% paraformaldehyde (PFA) for 24 h at 4°C for immunohistochemistry (see following).

Immunohistochemistry

Fixed tissues were dehydrated, paraffinized and sectioned by Histoserv Inc. (Germantown, MD, USA). Hematoxylin and eosin staining (H&E) was performed by Histoserv. After deparaffinization in Histoclear (HS-202, National diagnostics, USA), 5 μ thick sections were rehydrated using ethanol gradient before the epitope retrieval in Vector antigen retrieval solution (H3300, Vector Labs, USA) at 95°C. PBS/BSA 1% was used as blocking solution before overnight incubation at 4°C with primary antibody. 20αHSD (591009, Antibody Research) and Akr1b7 (sc-27763, Santa Cruz) primary antibodies were detected using the appropriate secondary antibody coupled to biotin (Jackson Immuno-Research Laboratory, USA). The signal was then amplified by incubation with a streptavidin-horseradish peroxidase (HRP) amplification (016-030-084, Jackson Immuno-Research Laboratory). The substrate, 3,3′-diaminobenzidine tetrahydrochloride (DAB) was incubated to detect the HRP activity (SK-4105, Vector Labs). The slides were counterstained with hematoxylin (K8008, Dako).

Reverse transcription quantitative polymerase chain reaction (PCR)

Five hundred nanograms of total RNA extracted from frozen adrenals using RNA kit (12183018A, Life Technologies, USA) were reverse transcribed using the superscript III first strand synthesis supermix (11752050, ThermoFischer, USA). Quantitative PCR reactions were performed on 2 μl of one-twentieth of dilution using either Taqman probes or specific primers (see following text). Each reaction was performed in duplicate with Taqman Fast Advanced Mastermix (4444963, Life Technologies) or SYBR green mix (4309155, Applied Biosystems, USA). Results were expressed as an induction fold calculated using ΔΔC_T method. All measurements were normalized to a housekeeper gene and represent the mean value of at least six adrenals per genotype ± standard deviation. Statistical analysis was performed with Student’s t-test when results followed a normal distribution. Taqman Gene Expression Assay Probes used in this study were as follows: Mm00490735_m1 (Cyp11a1), Mm01159156_g1 (hsd3b7), Mm00441558_m1 (Star), Mm99999915_g1 (gapdh), Mm01262510_m1 (mc2r), Mm00484040_m1 (Cyp17a1), Mm00439683_m1 (Inha) and Mm00506289_m1 (Akr1c18). Primers used in this study were previously described (25).

Biochemical measurements

Blood sampled by intracardiac puncture at 9 a.m. was placed in serum separator tubes (365967, BD Microtainer) and centrifuge at 7000 rpm. The supernatant was then collected and stored at −80°C. Serum concentrations were measured using commercially available assays: corticosterone (55-CORMS-E01, ALPCO Diagnostics), ACTH (KT-6010, KAMIYA) and aldosterone (ADI-900-173, Enzo Life Sciences).

Statistical analysis

Data were presented as mean ± standard deviation; relative changes were described as an induction fold. All data distributions were assessed for approximate normality. Continuous data were compared between two independent groups using t-tests using Graphpad Prism (Graphpad, San Diego). For the comparison between mice that had or did not have X-zone, Fisher’s exact test was used for the comparisons, given the small numbers.

Mouse protocol registration number

ASP#18-033.

Human clinical protocol

00-CH-0160.

Funding

This work was supported by the Intramural Research Program of the Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD).