Ghrelin Gene Deletion Alters Pulsatile Growth Hormone Secretion in Adult Female Mice

Using preproghrelin-deficient mice (Ghrl-/-), we previously observed that preproghrelin modulates pulsatile growth hormone (GH) secretion in post-pubertal male mice. However, the role of ghrelin and its derived peptides in the regulation of growth parameters or feeding in females is unknown. We measured pulsatile GH secretion, growth, metabolic parameters and feeding behavior in adult Ghrl-/- and Ghrl+/+ male and female mice. We also assessed GH release from pituitary explants and hypothalamic growth hormone-releasing hormone (GHRH) expression and immunoreactivity. Body weight and body fat mass, linear growth, spontaneous food intake and food intake following a 48-h fast, GH pituitary contents and GH release from pituitary explants ex vivo, fasting glucose and glucose tolerance were not different among adult Ghrl-/- and Ghrl+/+ male or female mice. In vivo, pulsatile GH secretion was decreased, while approximate entropy, that quantified orderliness of secretion, was increased in adult Ghrl-/- females only, defining more irregular GH pattern. The number of neurons immunoreactive for GHRH visualized in the hypothalamic arcuate nucleus was increased in adult Ghrl-/- females, as compared to Ghrl+/+ females, whereas the expression of GHRH was not different amongst groups. Thus, these results point to sex-specific effects of preproghrelin gene deletion on pulsatile GH secretion, but not feeding, growth or metabolic parameters, in adult mice.

The vast majority of the research investigating Ghrl-/-mice has been conducted in males. Specifically, some of the above referred studies did not include females at all (24,25,30). Other studies merely assessed body weight or food intake whereas the assessment of plasma hormone levels was exclusively performed in males (17,23). Notably, several studies suggest a modulatory role of sex on ghrelin actions (31)(32)(33)(34), but sex differences in Ghrl-/-mice, and specifically on GH secretion, have not been investigated thoroughly. In GHSR-deficient mice, only females fed on standard chow diet had lower body weight and adiposity, as compared to wild-type littermates (35). Furthermore, Ghsr-/female but not male mice fed on a HFD exhibited reduced taste responsiveness to linoleic acid compared to wild-type littermates (36). Finally, adult Ghsr-/-female but not male mice had reduced pulsatile GH secretion (37). We thus questioned a possible role of ghrelin gene in regulating GH secretion and its physiological consequences in female mice during adulthood, when the changes in pulsatile GH secretion and pattern associated with the rapid linear growth and pubertal maturation has ended (38,39). Specifically, we investigated if ghrelin gene deletion induces sex-specific dimorphic effects on GH secretion, feeding behavior or body and metabolic parameters in adult 20-40-week-old mice.

Assessment of Pulsatile GH Secretion
GH concentration time series were analyzed using an automated deconvolution method following established parameters (41). Measures include the number of pulses (over the 6-h sampling period) and mean pulse mass (i.e., mean of the summed pulses) as well as basal (i.e., non-pulsatile), pulsatile (i.e., sum of individual GH pulses) and total GH secretion (i.e., sum of basal plus pulsatile). The orderliness of GH secretion was calculated by Jack-Knife Approximate Entropy (JkApEn) as described earlier (42,43)

GHRH Immunostaining and Quantification
Brains of perfused mice (3 Ghrl+/+ and 5 Ghrl-/-males; 5 Ghrl+/+ and 8 Ghrl-/-females) were post-fixed, frozen and coronally cut at 40 µm into four equal series. Chromogenic immunohistochemistry against GHRH was performed as described before (46), using a previously validated rabbit anti-GHRH antibody (47) (immunoserum L0851, 1:10000) for 48-h at 4°C. Then, sections were sequentially incubated with a biotinylated anti-rabbit antibody and the avidin-biotin-peroxidase complex. A visible black/purple signal was developed with a diaminobenzidine/nickel solution (See Supplementary File for details). Quantifications were performed in the ArcN between bregma -1.58 and -2.06 mm, using the anatomical limits, according to the Paxinos mouse brain atlas (48). Total GHRH-immunoreactive cells were quantified, and data were expressed as positive cells (GHRH+) per section. Blind quantitative analysis was performed independently by two observers.

Real-Time Quantitative PCR Measurement
Hypothalami were quickly dissected, frozen in liquid nitrogen and stored at -80°C. The mRNAs were extracted with Trizol reagent and cDNAs were obtained from the reverse transcription of total RNA (ThermoFisher, Waltham, MA, USA). The mRNA levels of Ghrh and Ghsr were quantified relative to the housekeeping genes Ppia and Gapdh. Relative quantification (RQ) was calculated relative to the Ghrl+/+ males (See Supplementary File for details).

Statistical Analysis
Data are expressed as mean ± SEM, and statistical analyses were performed using Statview software (SAS institute) or GraphPad Prism (GraphPad Software). Differences across sex and genotype were identified by 2-way ANOVA, followed by multiple comparisons, using Bonferroni post-hoc analysis. Differences were considered significant when p<0.05.

RESULTS
Body Weight, Organ Weights, Body Fat Partitioning, Metabolic Parameters, Locomotor Activity, Spontaneous Food Intake and Fasting-Induced Food Intake Are Unaltered in Adult Ghrl-/-Mice As shown in Table 1, body weight, naso-anal distance and organ weights were not different in adult Ghrl-/-as compared to Ghrl+/+ males and females, respectively. Moreover, fed morning, fed evening and 24-h fasted blood glucose levels as well as glucose tolerance were not different in adult Ghrl-/-as compared to Ghrl+/+ males and females. Spontaneous diurnal and nocturnal food intake, fasting-induced food intake ( Figure 1) and home cage ambulatory activity (not shown) were not different in adult Ghrl-/-as compared to Ghrl+/+ males and females. In addition, meal number, mean meal size and mean meal duration in ad libitum fed mice were not different amongst genotypes (not shown). Finally, parameters of bones architecture measured by micro-CT were not different in adult Ghrl-/-as compared to Ghrl+/+ males and females (not shown), although a tendency toward reduction of bone volume fraction was observed (volume of mineral bone per unit volume of the sample, BV/TV; ANOVA, genotype effect: p=0.0894, sex effect: p<0.0001, sex x genotype effect: p=0.9014) in 80-week-old mice only.
The Ultradian Pattern of GH Secretion Is Altered in Adult Ghrl-/-Females, but Not in Ghrl-/-Males GH pituitary contents were not different in adult Ghrl-/-, as compared to Ghrl+/+ males and females ( Table 1). As reported before (37), we observed a sexually dimorphic GH secretion pattern. Indeed, mean pulse mass was higher in males than in females (p<0.05 in both Ghrl+/+ and Ghrl-/-mice, post-hoc test), whereas basal GH secretion and JkApEn values were lower in males than in females (p<0.05 in both Ghrl+/+ and Ghrl-/-mice for basal GH, p<0.01 in Ghrl+/+ mice and p<0.0001 in Ghrl-/mice for JkApEn, post-hoc test) ( Figure 2). An increase in the number of GH pulses in females as compared to males was observed in Ghrl-/-mice only (p<0.01, post-hoc test). Furthermore, we observed that the regularity of GH secretion was decreased in Ghrl-/-female mice when compared to Ghrl+/+ female mice ( Figure 2). More precisely, pulsatile GH secretion and mean pulse mass were reduced by 43% and 56% respectively in Ghrl-/-females as compared to Ghrl+/+ females (p<0.05 and p<0.01 for pulsatile GH and mean pulse mass, respectively, posthoc test) while total and basal GH secretion were not significantly affected by ghrelin deletion. The maximal pulse amplitude was also reduced by 50% in Ghrl-/-compared to Ghrl+/+ females (26 ± 4 versus 13 ± 3 ng/mL, p<0.05, post-hoc test). In contrast, JkApEn values increased by 27% in Ghrl-/females as compared to Ghrl+/+ females (p<0.01, post-hoc test). In males, we did not find genotype differences for any of the GH secretion parameters, however, the inter-individual variability of pulsatile GH, maximal pulse amplitude and mean pulse mass was higher than in females (see representative GH profiles).

In Vitro Release of GH Is Unaltered in Pituitary Explants From Adult Male and Female Ghrl-/-and Ghrl+/+ Mice
To test if the alteration of the ultradian GH pattern in adult Ghrl-/-females was due to modifications at the pituitary level, we assessed basal and stimulated GH secretion from pituitary gland explants ( Figure 3A). As reported earlier (26), basal and GHRHinduced GH release were significantly higher in males than in females (ANOVA, sex effect: p<0.0001). However, basal GH release or GH release stimulated by GHRH or KCl was not different in pituitary explants from Ghrl+/+ and Ghrl-/-mice in either sex.

DISCUSSION
As compared to wild-type littermates, Ghrl-/-mice of both sexes lack major alterations in body growth, metabolic parameters or feeding behavior at adulthood. Adult Ghrl-/-females display, however, decreased amplitude of GH secretion and increased irregularity of secretion accompanied by increased immunoreactive signal for GHRH in the ArcN without change in the pituitary GH content or GH secretion from pituitary explants.

Ghrl Deletion Affects Pulsatile GH Secretion and Regularity of Ultradian GH Pattern in Adult Female, but Not Male, Mice
Among the proghrelin-derived peptides, ghrelin seems to play the most powerful control on GH secretion. Indeed, ghrelin treatment potently increases GH secretion in humans and rodents. However, the role of endogenous ghrelin on GH secretion has been more difficult to demonstrate. Endogenous ghrelin seems to regulate the amplitude of GH secretion in healthy men since ghrelin levels and GH pulse amplitude strongly correlate (49). Pharmacological blockage of GHSR in male rats decreased the amplitude of GH release but did not affect the pattern of GH secretion (50). Also, no correlation was found between acyl ghrelin or total ghrelin and GH plasma peaks in freely behaving male rats (1,51). In male rats, we previously showed that ghrelin displays an ultradian pattern of secretion through the light and dark periods with pulses occurring with a frequency similar to GH (1). In mice, however, due to the small volume of blood and the absence of ghrelin assays with sufficient sensitivity, determining the ultradian pattern of ghrelin secretion and a possible sexual dimorphic pattern of secretion are challenging. To our knowledge, few studies have investigated the link between endogenous ghrelin and GH secretion in women or female rodents. Transgenic female, not male, rats with reduced hypothalamic GHSR expression display reduced pulsatile GH secretion (52). In contrast to adult males, adult Ghsr-/-female mice showed a reduction of total, basal and pulsatile GH secretion (37). Furthermore, Ghrl-/-and Goat-/male mice lack alterations of pulsatile GH secretion at adulthood, 1 | Growth and body parameters, body fat partitioning and metabolic parameters in male and female Ghrl-/-mice.
Here, we observed that adult Ghrl-/-males lack significant alterations in GH secretion whereas adult Ghrl-/-females display a reduction of pulsatile GH secretion, maximal pulse amplitude and regularity of GH secretory pattern (i.e., higher ApEn values). ApEn is considered a barometer of GH negative feedback mechanism on GH release. As observed in all species investigated so far (39), we confirmed that females show higher ApEn values and reduced pulsatile GH secretion as compared to males. The increased ApEn values in Ghrl-/-females are in line with the reduced pulsatile GH secretion, whereas the unchanged ApEn values in Ghrl-/-males point to unchanged pulsatile GH secretion, as observed herein. Of note, the higher interindividual variability observed in males for several parameters of GH secretion may have prevented from unmasking statistical genotype differences in this sex. In any case, our data demonstrate a significant effect of endogenous proghrelinderived peptides on GH secretion in adult female mice. Since both Ghsr-/-females and Ghrl-/-females show a reduction of the pulsatile GH secretion, it seems reasonable to hypothesize that endogenous ghrelin-evoked GHSR signaling increases GH secretion in adult female mice mainly. It is important to highlight that stress is known to reduce plasma GH levels (54). The blood sampling procedure used in the current study induces only a modest rise in plasma corticosterone levels that are within the physiological range of endogenous diurnal variations (53). Furthermore, Ghrl-/-mice displayed similar plasma corticosterone levels than Ghrl+/+ mice in non-stressed conditions and in response to restrain-stress (unpublished observations). Thus, it seems unlikely that differential responses to the experimental blood sampling procedure may have contributed to reduce GH secretion in Ghrl-/-females.

Neuroendocrine Basis by Which Ghrl Deletion Affects GH Pulse Amplitude and Regularity of Ultradian GH Pattern in Female Mice
Ghrelin-induced GH secretion involves pituitary and hypothalamic mechanisms (39). In vitro, ghrelin acts on somatotropic cells of the pituitary to release GH (55). Also, ghrelin enhances GHRH-induced GH release and impairs somatostatin-mediated inhibition of GH release. Notably, ghrelin treatment does not induce GH secretion in male mice lacking GHSR exclusively in somatotropic cells (56). Ghrelin also indirectly acts at the pituitary by stimulating GHRH neurons and inhibiting somatostatin neurons via multiple mechanisms. For instance, ghrelin directly excites GHRH neurons, and also indirectly increases GHRH neuron excitability by decreasing their inhibitory GABA inputs (57). Here, we found that Ghrl-/and Ghrl+/+ females show similar pituitary GH contents as well as a similar basal, KCl-induced and GHRH-induced secretion of GH in vitro. Conversely, a higher number of GHRH+ neurons was visualized in the ArcN of Ghrl-/-females, as compared to Ghrl+/+ females, and no changes in the GHRH levels in the median eminence (not shown) nor in the hypothalamic levels of Ghrh mRNA. Thus, hypothalamic, rather than pituitary dysfunctions are more likely associated to the reduction of the pulsatile GH secretion in Ghrl-/-females. Transgenic female rats with reduced GHSR expression in hypothalamic tyrosine hydroxylase-expressing neurons show reduced GH secretion and lower number of GHRH neurons (58), also suggesting hypothalamic basis for altered GH secretion induced by a deficit of ghrelin action in female rodents. Importantly, the GHRH neurons are poorly visualized in naive rodents because the GHRH peptide is rapidly transported to the neuron terminals for release (59). Thus, the above referred study in rats used colchicine treatment to estimate the number of GHRH neurons.
Here, however, the GHRH immunolabeling was performed in brain sections of naive mice undermining our capacity to estimate the number of GHSR neurons in each experimental group. Since Ghrl-/-females display a reduction of pulsatile GH secretion, it seems reasonable to hypothesize that GHRH neurons were more easily visualized in Ghrl-/-females because GHRH secretion was reduced. Such hypothesis could be further tested by assessing plasma GH levels in Ghrl+/+ and Ghrl-/-females following somatostatin withdrawal, which triggers GHRHdependent GH secretion in vivo (60). In this regard, it is possible that somatostatin neurons, which control the timing of GH secretory pulses and whose release activity is also under the control of ghrelin (4), are differentially affected in Ghrl-/-females and contribute to modulate GH secretion. Of note, expression of somatostatin receptor 2 on GHRH neurons displays sexual dimorphism in mice, suggesting that somatostatin interacts with GHRH neurons to control sexual dimorphic GH responses (47). Further studies are required to precisely determine the hypothalamic dysfunctions affecting the GH secretion in Ghrl-/females. Also, the reason why alterations of GH secretion only take place in females remain uncertain. Of note, the ghrelin system displays important dimorphic responses (34). Indeed, studies in women showed that estradiol potentiates ghrelin-stimulated pulsatile GH secretion and GHRH/ghrelin synergy (61)(62)(63). In mice, females were also more sensitive to the GH-releasing effects of ghrelin treatment than males (34). Strikingly, we found that Ghsr mRNA levels in the hypothalamus were not different among genotypes, although Ghsr mRNA levels tended to be lower in Ghrl-/-than Ghrl+/+ females. Previously, Ghsr mRNA levels were found similar in the ArcN of 8-week-old male and female rats (64) and slightly lower in the hypothalamus of 6-month-old female rats, as compared to males (65).

Ghrl Deletion in Mice Does Not Significantly Affect Food Intake, Growth or Metabolism at Adulthood
As previously reported (12,23,25), we confirmed that adult Ghrl-/-micemale and female -lack alterations of spontaneous food intake, meal pattern or fast-refeeding hyperphagia. Also, adult Ghrl-/-mice show no major growth or metabolic deficits, as previously shown (26), despite adult Ghrl-/-females show alterations of their GH secretion pattern. Notably, previous studies also showed that young Ghrl-/-males display alterations in ultradian GH secretion patterns that are not associated with changes in growth or metabolic parameters (26). Similarly, young Goat-/-mice show reduced GH pulse amplitude but unaltered body weight or body growth, although an increase in plasma insulin-like growth factor I levels has been proposed to compensate the alterations in GH secretion in these mice (53). Thus, the lack of proghrelin-derived peptides seems to induce some sexually dimorphic alterations in the GH axis that are insufficient to affect eating behaviors, body growth and metabolism at adulthood. Still, altered GH secretion during adulthood could contribute to specific phenotypes associated with ageing or pathophysiological conditions. Indeed, we observed a tendency to reduced volume of mineralized bone and significant reductions in the mass of soleus muscle in older 80-week-old Ghrl-/-mice that may be linked to ghrelin since this hormone plays a protective effect on bones structure in older mice (66) and both acyl and des-acyl ghrelin enhance muscle anabolism (67) and prevent skeletal muscle atrophy (68).

The Absence of Evident Dysfunctions in Adult Mice Lacking Ghrl Gene Must Be Interpreted With Caution
The study of mice with genetic modifications of different elements of the ghrelin system, such as proghrelin, GOAT or GHSR, has been instrumental to reveal some aspects of their physiological role. For instance, the observation that Goat-/mice, in contrast to WT mice, suffer severe hypoglycemia and become moribund in a starvation protocol indicates that ghrelin plays essential functions under energy deficit (21). In this regard, it is interesting to stress that Ghrl-/-mice exhibit less evident alteration than mice lacking GHSR [as review in (12)]. For instance, GHSR-deficient mice, but not Ghrl-/-mice, show reduced hyperphagia after fasting (25). GHSR-deficient mice show some alterations, as compared to Ghrl-/-mice, even under ad libitum feeding conditions. In contrast to Ghrl-/-mice, adult Ghsr-/-mice fed on a regular chow displayed reduced body weight and linear growth, regardless of the sex (37). Furthermore, meal frequency was reduced in Ghsr-/-male mice (37). The exact molecular mechanisms underlying such distinct phenotypes in Ghsr-/-mice, as compared to Ghrl-/mice, are uncertain. On one hand, GHSR acts via several ligandindependent mechanisms that include its capacity to induce constitutive intracellular signaling or its ability to cross-talk with other receptors (69). Also, the deletion of Ghrl gene not only eliminates ghrelin but also obestatin and des-acyl-ghrelin, which seem to impair the effect of ghrelin. For instance, obestatin treatment antagonizes ghrelin-induced GH secretion in rats (70), although these observations were not confirmed by others (71,72). Des-acyl ghrelin appears to block some of the effects of ghrelin via either GHSR-dependent or GHSR-independent mechanisms (12,73,74). Finally, the liver-expressed antimicrobial peptide 2 (LEAP2) was recently recognized as another endogenous peptide ligand for GHSR (75). LEAP2, which is released by endocrine cells of the liver and the intestinal tract, abrogates ghrelin-evoked and constitutive GHSR activities (75,76). Notably, LEAP2 displays a binding affinity to GHSR similar to ghrelin, but its level in plasma is~10-fold higher than ghrelin level in ad libitum fed conditions (69). Thus, LEAP2 may play a more important role than ghrelin on GHSR under the tested conditions, masking some of the consequences of Ghrl gene deletion.
In conclusion, present data suggest that ghrelin plays a more prominent role in the regulation of pulsatile GH secretion in adult female than male mice. The mechanism of the sex-specific effect of preproghrelin deletion on GH pulsatility still needs to be refined. The physiological implications of the altered GH pulsatility in preproghrelin deleted mice would also need to be clarified since it is not associated with major growth, feeding or metabolic phenotypes.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The animal study was reviewed and approved by Animal Experimentation Committee CEEA.34 of University of Paris.

AUTHOR CONTRIBUTIONS
RH, GF, MP, JE, and VT designed experiments. RH, GF, NL, AL, OF, PZ, and OV performed experiments. FR performed jkApEn and deconvolution analysis of GH pulsatility parameters. CC performed all studies and analysis on bone parameters. CT provided the colony of preproghrelin deficient mice. RH, GF, NL, MP, and VT analyzed data. MP and VT wrote the manuscript. JE edited the manuscript. All authors contributed to the article and approved the submitted version.