The effects of kisspeptin on β‐cell function, serum metabolites and appetite in humans

Aims To investigate the effect of kisspeptin on glucose‐stimulated insulin secretion and appetite in humans. Materials and methods In 15 healthy men (age: 25.2 ± 1.1 years; BMI: 22.3 ± 0.5 kg m−2), we compared the effects of 1 nmol kg−1 h−1 kisspeptin versus vehicle administration on glucose‐stimulated insulin secretion, metabolites, gut hormones, appetite and food intake. In addition, we assessed the effect of kisspeptin on glucose‐stimulated insulin secretion in vitro in human pancreatic islets and a human β‐cell line (EndoC‐βH1 cells). Results Kisspeptin administration to healthy men enhanced insulin secretion following an intravenous glucose load, and modulated serum metabolites. In keeping with this, kisspeptin increased glucose‐stimulated insulin secretion from human islets and a human pancreatic cell line in vitro. In addition, kisspeptin administration did not alter gut hormones, appetite or food intake in healthy men. Conclusions Collectively, these data demonstrate for the first time a beneficial role for kisspeptin in insulin secretion in humans in vivo. This has important implications for our understanding of the links between reproduction and metabolism in humans, as well as for the ongoing translational development of kisspeptin‐based therapies for reproductive and potentially metabolic conditions.


| INTRODUCTION
Metabolism and reproduction are fundamental aspects of mammalian physiology which are intricately linked. However, our understanding of the hormonal links between these two biological systems is limited.
Recent evidence in animals suggests that the recently discovered hormone kisspeptin may link reproduction and metabolism; however, until now human in vivo studies have not been performed.
Kisspeptin sits at the apex of the hypothalamo-pituitary-gonadal axis, controlling downstream reproductive hormone secretion. Kisspeptin has pivotal roles in fertility, 1 which are being utilised by kisspeptin-based therapies now in development for common reproductive disorders. [2][3][4][5] Kisspeptin (kiss1) and its receptor (kiss1r) are expressed in the hypothalamus, pancreatic β-cells, liver and adipose tissue, [6][7][8][9] suggesting roles in key autocrine and paracrine metabolic processes. Kisspeptin has recently been reported to increase glucosestimulated insulin secretion (GSIS) in rats 10 and monkeys, 11 and this is supported by in vitro islet studies which demonstrate enhancement of GSIS by kisspeptin. 8,10,12,13 Interestingly, it has also been reported that kisspeptin can inhibit GSIS at lower kisspeptin concentrations 13 and low glucose concentrations, 14 but stimulate GSIS at high kisspeptin concentrations. 13 There are neuroanatomical and functional connections between kisspeptin and important hypothalamic appetite-regulating neuropeptides. [15][16][17] Intracerebroventricular administration of kisspeptin has been reported to alter food intake in mice 18 and chicks, 19 while other studies report no effect of kisspeptin administration on appetite in rats. 20 In addition, impaired kisspeptin signalling disrupts metabolism and promotes glucose intolerance and obesity in mice. 21 Thus, data in animals strongly suggests that kisspeptin plays a role in metabolism, but the effects of kisspeptin on metabolic parameters in vivo in humans are currently unknown. There is therefore an important clinical need to elucidate the metabolic effects of kisspeptin in humans, to further our understanding of the physiology of human metabolism, as well as to inform the ongoing development of kisspeptin-based therapies.
In this study, we investigate the effects of kisspeptin on β-cell function, metabolites and appetite in humans. We demonstrate that in

| IVGTT
On arrival at the Clinical Research Facility (CRF), and after a period of acclimatisation, two intravenous cannulae (one in each antecubital fossa) were inserted (one for blood sampling and the other for intravenous infusion administration). Following baseline sampling, kisspeptin or vehicle infusion was started at T = 0 minutes and infused until T = 225 minutes.
Next, 0.3 g kg −1 of 20% dextrose (Hameln, Gloucester, UK) was administered intravenously over 120 seconds starting from T = 45 minutes (i.e. when kisspeptin levels had reached a steady state, Figure 1A,B). To obtain the glucose and insulin values required for calculation of the acute insulin response to glucose (AIRg) and minimal model insulin sensitivity index (S i ), an established frequent sampling protocol 22 was used ( Figure 1A). AIRg was calculated as the incremental AUC (using the trapezoid rule 23 ) of insulin from T = 45 minutes to T = 55 minutes (i.e. 0-10 minutes post-glucose load). S i was determined using the minimal model (MLAB software) 24

| Human β-cell line (EndoC-βH1 cells)
EndoC-βH1 cells were seeded onto ECM-coated 48-well plates at After incubation for 60 minutes, the supernatant was collected, placed onto ice and centrifuged for 5 minutes at 3000 rpm at 4 C.
The supernatant was then transferred into a fresh tube. Cells were lysed in 50 μL of cell lysis solution (TETG: 20 mM Tris PH 8.0, 1% Triton X-100, 10% glycerol, 137 mM NaCl, 2 mM EGTA). The lysate was then removed and placed onto ice, and centrifuged at 3000 rpm for 5 minutes at 4 C. Insulin concentration was measured using an ultrasensitive HTRF kit (Cisbio Bioassays), and secreted insulin was normalized as percentage of total insulin content.

| Statistical methods
Using STATA, an a priori power calculation was performed using a   Figure 1D).
The disposition index is a well-validated method for assessing β-cell function 32 as it is comprised of measures of insulin secretion, insulin sensitivity and prevailing glucose concentrations (which were similar between groups; Figure 1E).
Kisspeptin administration resulted in elevated circulating kisspeptin levels ( Figure 1B), which, as expected, resulted in elevated LH levels, confirming peptide bioactivity ( Figure S1A). Consistent with previous studies, 5,33 testosterone levels did not rise during the timeperiod of kisspeptin administration ( Figure S1B).

| Kisspeptin dose-dependently stimulates GSIS in vitro
To assess the effect of kisspeptin on GSIS in vitro, human islets from six donors were incubated with 3 and 17 mM glucose with a range of concentrations of kisspeptin (0, 2.7 and 1000 nM kisspeptin). In donor human islets, kisspeptin enhanced insulin secretion (Figures 2A and S2  which was reduced by kisspeptin administration and has previously been shown to be negatively correlated with insulin secretion. 36 All annotated metabolites that significantly changed during kisspeptin administration (but not vehicle administration) have been collated in Table S4.

| Kisspeptin does not affect appetite or food intake in humans
To assess the effect of kisspeptin on appetite and food intake in During kisspeptin administration, plasma kisspeptin levels were elevated ( Figure 4D), which resulted in elevated LH levels ( Figure S5C) and no significant change in testosterone ( Figure S5D), similar to those from IVGTTs.

| DISCUSSION
This is the first study investigating the effect of kisspeptin on β-cell function and appetite in humans in vivo. We demonstrate that the reproductive hormone kisspeptin enhances GSIS in humans in vivo without affecting insulin sensitivity. Our in vitro data show a direct effect of kisspeptin on β-cells, which are known to possess abundant kisspeptin receptors. 8 Furthermore, kisspeptin modulates serum metabolites in humans, but does not influence appetite or food intake.
These data indicate that kisspeptin plays a role in glucose homeostasis in humans, and therefore is a hormonal mediator linking reproductive and metabolic systems.
Kisspeptin administration increased insulin secretion and disposition index during IVGTT but not during MMTT. The glucose concentrations were more markedly elevated (as expected) during the IVGTT (mean peak glucose 14.01 mM, Figure 1E) compared to the MMTT (mean peak glucose 7.48 mM, Figure S5B). Therefore, this suggests that kisspeptin increases insulin release only when ambient glucose concentrations are high in humans in vivo. In keeping with this, our data show that in human islets in vitro, kisspeptin increases insulin  also Table S4) secretion at 17 mM but not at 3 mM glucose (Figure 2A). This is also consistent with previous data in human islets in vitro, which show that kisspeptin stimulates insulin release at higher ambient glucose concentrations (20 vs. 3 mM). 8,12 However, there are differing reports of the effects of kisspeptin on insulin secretion from animal islets. Kisspeptin, at doses ranging from 0.1-1000 nM, has been reported to increase GSIS in mouse, rat and pig islets, when the prevailing glucose concentrations were ≥ 10 mM. 8,10,12,37 In contrast, other investigators have reported that kisspeptin, at doses ranging from 0.1-1000 nM, inhibits GSIS in a dosedependent manner in mouse and rat islets/perfused pancreata. 13,14,38 The lack of consensus in the literature may be due to methodological differences in the experiments and/or species differences, which are often observed between animal and human islet preparations. 39 Figure S1F). Overall, it appears that although kisspeptin enhanced insulin secretion, in healthy individuals this did not result in altered glucose levels, but this would be interesting to investigate in patients with abnormal glucose homeostasis such as in diabetes.
Testosterone has also been shown to increase insulin secretion from isolated islets. 41 However, the effects of kisspeptin administration on metabolism during our study were not confounded by altered circulating testosterone levels as serum testosterone did not rise during the time-period of this study, because more prolonged kisspeptin administration is required to produce elevations in testosterone levels. 33 The gut hormones, GLP-1, PYY and glucagon have key roles in glucose homeostasis. Additionally, in rodent studies, GLP-1 has been shown to alter hypothalamic kisspeptin expression and neuronal activity, 42,43 and in mice glucagon stimulates hepatic kisspeptin production to alter GSIS. 13 In our study, there was no difference in circulating GLP-1, PYY or glucagon levels following intravenous glucose during kisspeptin administration compared to vehicle ( Figure S1C-E). This is consistent with previous data which show that intravenous glucose administration alone does not activate the incretin response mediated by gut hormones. 44 Furthermore, the above studies suggest that GLP-1 and glucagon may act upstream of kisspeptin, whereas in our study we directly administered kisspeptin, which may act downstream of GLP-1 and glucagon. The participants were fasted during the IVGTTs, and fasting has been shown to reduce hypothalamic kisspeptin expression. 45 However, as shown in Figure 1B, kisspeptin administration significantly raised plasma kisspeptin levels, and therefore would be sufficient to overcome the effect of suppression of endogenous kisspeptin by fasting. Furthermore, kisspeptin administration has been shown to increase GSIS in both fasted and fed monkeys. 11 Metabonomics, the identification and analysis of metabolites in biological fluids, is an emerging field of study which provides a nonbiased methodology to identify novel pathways to guide further research. To provide further insights into the metabolic effects of kisspeptin, we compared the distribution of metabolites in serum samples taken pre-kisspeptin administration (T = −15 minutes) to those taken when plasma kisspeptin levels had reached steady state (T = 45 minutes) and equivalent vehicle timepoints. Importantly, samples were collected before the glucose loads to prevent this from confounding the results. We demonstrate for the first time that kisspeptin modulates serum lipids and small molecules in humans.
Several of these metabolites include classes of lipids (i.e. lysophosphatidylcholines, phosphocholines and sphingomyelins), which have been shown to be associated with insulin secretion. 36 This provides further evidence of kisspeptin's modulation of human metabolism, which (with further study) may provide important mechanistic data.
The principal limitation to the metabolite analysis approach used in the present study is that the measurements taken for each chemical species do not yield absolute quantities. Rather, they provide values that are proportional to the concentration of chemicals within the biological fluid, which are only useful in relative comparison within the experiment rather than in comparison to established concentration ranges. We have accounted for this limitation through the use of linear mixed effect models which rely solely on the differences observed in the relative abundance of individual metabolites to generate patterns of statistical significance. This allows the determination of significance and magnitude of metabolic changes.
In light of emerging evidence for neuroanatomical and functional connections between kisspeptin and key appetite-regulating neurones in the hypothalamus, 15,16 and animal data which suggest a role for kisspeptin in energy homeostasis, 21 we investigated the effect of kisspeptin on appetite in healthy men. Our data demonstrate kisspeptin had no effect on appetite and food intake in men. This is in keeping with rodent data showing that male kiss1r knockout mice have unaltered food intake. 21 Our findings have pharmacological and potential therapeutic relevance as the plasma kisspeptin levels achieved in this study, which enhanced GSIS, are similar to those required to restore LH pulsatility in women with hypothalamic amenorrhea 3 and trigger oocyte maturation in in vitro fertilisation protocols. 2 During IVGTTs, pharmacological kisspeptin administration increased insulin secretion and increased disposition index (IVGTT-DI). IVGTT-DI quantifies the ability of the β-cell to counter insulin resistance, 32 with lower baseline IVGTT-DI values independently predicting conversion from normal glucose tolerance or impaired glucose tolerance to type 2 diabetes within 5 years. 32 Therefore, our finding that kisspeptin increases GSIS and IVGTT-DI shows metabolically the beneficial effects of kisspeptin. This is especially important as kisspeptin-based treatments are currently being developed to treat reproductive disorders 2-5 and such treatments may therefore have additional potentially beneficial metabolic therapeutic applications. Our finding that kisspeptin administration also improves GSIS suggests that kisspeptin could have a dual therapeutic role in patients with diabetes to improve hypogonadism 33 as well as enhance insulin release, specifically only when hyperglycaemia occurs (thereby potentially avoiding the risks of hypoglycaemia associated with other diabetes treatments). Thus, our study lays the foundation for future studies exploring the effect of kisspeptin in the treatment of diabetes. Furthermore, both kisspeptin-54 and the smaller fragment, kisspeptin-10 (which may be produced by enzymatic breakdown) have been shown to enhance GSIS in vivo in humans in this study and in vivo in monkeys, 11 respectively. Therefore, metabolically active breakdown products of kisspeptin may serve to potentiate the metabolically beneficial effects of administered kisspeptin.
Our data demonstrating that pharmacological elevation of circulating kisspeptin levels results in a significant increase in GSIS in humans in vivo may also have potential physiological relevance. The circulating kisspeptin levels achieved by pharmacological administration of kisspeptin during this study are also observed in humans physiologically during pregnancy due to placental kisspeptin production (1-10 nM). 46 Therefore, the higher circulating kisspeptin levels during normal pregnancy may play a physiological role to enhance insulin secretion to protect the mother and foetus from increasing glucose levels and the development of gestational diabetes mellitus. In keeping with this hypothesis, pregnant women with diabetes have lower kisspeptin levels than pregnant women without diabetes. 47 In addition, previous rodent work has shown that selective ablation of the kisspeptin receptor, kiss1r, from pancreatic islet β-cells in pregnant mice results in impaired glucose tolerance and reduced insulin secretion. 48 Together these data suggest that elevated kisspeptin during pregnancy may play a positive physiological role in glucose homeostasis to protect against the development of gestational diabetes mellitus.
In summary, reproduction and metabolism are fundamental and interdependent aspects of mammalian physiology. This comprehensive study demonstrates that administration of the reproductive hormone kisspeptin to humans significantly increases GSIS in vivo and in vitro via actions on pancreatic islet cells, with associated alterations in serum metabolites associated with insulin secretion. We also show that kisspeptin does not affect appetite or food intake in healthy men. This is the first human in vivo report of the effect of kisspeptin on β-cell function, metabolites and appetite, which is important for our understanding of the links between reproduction and metabolism in humans, as well as the ongoing development of kisspeptin-based treatments for common reproductive and potential metabolic disorders.