Addition of a carboxy terminal tail to the normally tailless gonadotropin-releasing hormone receptor impairs fertility in female mice

Gonadotropin-releasing hormone (GnRH) is the primary neuropeptide controlling reproduction in vertebrates. GnRH stimulates follicle-stimulating hormone (FSH) and luteinizing hormone (LH) synthesis via a G protein-coupled receptor, GnRHR, in the pituitary gland. In mammals, GnRHR lacks a C-terminal cytosolic tail (Ctail) and does not exhibit homologous desensitization. This might be an evolutionary adaptation that enables LH surge generation and ovulation. To test this idea, we fused the chicken GnRHR Ctail to the endogenous murine GnRHR in a transgenic model. The LH surge was blunted, but not blocked in these mice. In contrast, they showed reductions in FSH production, ovarian follicle development, and fertility. Addition of the Ctail altered the nature of agonist-induced calcium signaling required for normal FSH production. The loss of the GnRHR Ctail during mammalian evolution is unlikely to have conferred a selective advantage by enabling the LH surge. The adaptive significance of this specialization remains to be determined.


INTRODUCTION 45
The propagation and survival of all species depends on reproduction. In vertebrates, the 46 process is controlled by hormones in the hypothalamic-pituitary-gonadal axis. Arguably, the hypothalamic decapeptide gonadotropin-releasing hormone (GnRH) is the most important brain 10 The data presented here challenge this idea. Addition of the chicken GnRHR C-tail to the 261 endogenous murine GnRHR blunted but did not block the LH surge. With few exceptions, females 262 expressing the mouse-chicken chimeric GnRHR (GnRHR-Ctail) were fertile, but with smaller 263 litter sizes compared to wild-type mice. Reductions in FSH rather than perturbations of the LH 264 surge likely explain their subfertility. The FSH impairment appears to derive from alterations in 265 GnRH-induced calcium signaling. 266 267

Effects of the chicken C-tail on gonadotropin synthesis and secretion 268
Serum FSH and pituitary Fshb mRNA levels are lower in GnRHR-Ctail than wild-type 269 mice. In males, this is associated with small, but significant decreases in testis mass and 270 spermatogenesis. There is a direct relationship between Sertoli cell number and spermatogenic 271 potential (56). Sertoli cell number is regulated by FSH during early postnatal development in 272 rodents (24,57). Though we did not quantify Sertoli cells in GnRHR-Ctail males or their FSH 273 levels prior to weaning, it seems likely that the FSH deficiency observed in adulthood also occurs 274 earlier in life in these animals. Indeed, depleting FSH in young but not adult mice reduces testis 275 size and sperm counts (24,58). In females, reduced FSH levels are associated with decreased 276 numbers of preovulatory follicles. Because most GnRHR-Ctail females exhibit LH surges and/or 277 corpora lutea, it is clear that the majority could and did ovulate. Therefore, the most parsimonious 278 explanation for the subfertility in these females is impaired follicle development secondary to FSH 279 deficiency. 280 Though present, LH surges are altered in most GnRHR-Ctail females. Unfortunately, we 281 were unable to fully characterize the nature of the changes, as we had difficulty capturing surges 282 in these animals when sampled on presumptive proestrus. Therefore, we could not measure the 283 dynamics (the precise time of onset, maximum amplitude, or duration) of their LH surges relative 284 to those of wild-type mice. Nevertheless, with our modified sampling protocol, we did observe LH 285 surges in GnRHR-Ctail females, which were reduced in amplitude. It is unlikely that this 286 contributed to their subfertility, however, as there are several mouse models with reduced LH surge 287 amplitudes that do not exhibit fertility defects [e.g., (42, 59)]. Moreover, the amplitude of the surge 288 varies dramatically between mice within a given strain [(60) and our unpublished observations]. 289 Though we only detected LH surges in ~50% of GnRHR-Ctail mice, it is unlikely that they were 290 truly blocked or absent in half of the animals. In the fertility trial, only 1 of 8 animals was sterile. 291 Similarly, in only 1 of 6 GnRHR-Ctail mice did we fail to observe corpora lutea in their ovaries. 292 Thus, the complete absence of LH surges appears to be a rare event in these mice, most likely 293 explained by inadequate FSH-stimulated follicle development and estrogen positive feedback. The 294 cause of the variable (and low) penetrance of the infertility phenotype is presently unclear, but the 295 animals were notably on a mixed genetic background. 296 The blunted LH surges in GnRHR-Ctail mice may derive, at least in part, from homologous 297 receptor desensitization. The effects of adding C-tails to mammalian GnRHRs have been 298 thoroughly investigated in vitro. In most cases, these manipulations are associated with agonist-299 induced receptor phosphorylation, arrestin recruitment, and receptor internalization (34, 36, 55, 300 61, 62). We similarly observed that the mouse-chicken chimeric GnRHR used here acquired the 301 ability to recruit arrestin in response to GnRH. It is therefore likely that GnRHR-Ctail is rapidly 302 internalized in response to agonist, but we did not assess this directly. Though arrestin recruitment 303 to the chimera's C-tail did not appear to explain the altered calcium signaling or retained ERK 304 activation in HEK 293 cells, we acknowledge that multiple GPCR/arrestin conformations,305 including differences between the receptor core and/or C-tail, can have distinct functions (63)(64)(65)(66). 306 Regardless, if we were able to measure the duration of LH surges in GnRHR-Ctail females, we 307 predict that it would be shorter than in wild-type mice. However, as most of these mice ovulated, 308 the amplitude and duration of these surges were clearly sufficient. We recently reported that 309 kisspeptin-54 induces surge-like LH release in juvenile mice. Although the duration of the LH 310 increase is shorter than natural surges, these mice still ovulate efficiently (67). Thus, both the 311 amplitude and duration of natural LH surges are greater than actually needed to induce ovulation 312 in mice. 313 314

Effects of the chicken C-tail on GnRH signaling 315
The reductions in gonadotropin production in GnRHR-Ctail mice indicate that the addition 316 of the chicken C-tail altered GnRH signaling. In heterologous HEK 293 cells, GnRH stimulation 317 of calcium mobilization was greatly impaired downstream of in turn,appeared 318 to be explained by attenuated activation of Gαq and reduced agonist stimulated inositol phosphate 319 production. As GnRH induction of Fshb mRNA expression in homologous LβT2 cells is calcium 320 dependent (68), it is possible that FSH deficiency in GnRHR-Ctail mice may result from alterations 321 in calcium signaling. In contrast, GnRH induction of ERK1/2 phosphorylation is intact 322 downstream of GnRHR-Ctail in HEK 293 cells. In gonadotropes, GnRH promotes ERK1/2 323 signaling via PKC, which in turn depends on diacylglycerol (DAG) more so than calcium (69). 324 Though GnRH induction of DAG production was attenuated downstream of GnRHR-Ctail in HEK 325 293 cells, it was sufficient to activate PKC-ERK1/2 signaling. As GnRH regulation of Lhb 326 expression is ERK1/2-dependent (46), this may help explain how LH production was relatively 327 unperturbed in gonad-intact GnRHR-Ctail mice. 328 The Gαq activation and, in particular, calcium mobilization impairments, downstream of 329 GnRHR-Ctail in heterologous cells do not fully recapitulate changes in GnRH signaling in 330 gonadotropes in GnRHR-Ctail mice. However, in both HEK 293 cells and gonadotropes, the 331 GnRHR-Ctail induced a more sustained calcium profile. Gonadotropes possess L-type calcium 332 channels, which are absent in HEK 293 cells (70,71), though the latter do have endogenous 333 calcium currents (72). GnRH-induced calcium oscillations in gonadotropes reflect both 334 mobilization from ER stores and influx via voltage-dependent L-type channels. The calcium 335 signaling (and defects therein) that we examined in HEK 293 cells is limited to mobilization from 336 internal stores. Nevertheless, it is evident that GnRH-induced calcium oscillations also differ 337 between gonadotropes of wild-type and GnRHR-Ctail mice. In wild-type pituitaries, we observe 338 the previously reported heterogeneity of responses: oscillatory, biphasic, and transient (73). In 339 contrast, GnRH stimulates a more homogenous calcium response in gonadotropes of  Ctail mice and one that is not observed in wild-type animals. Relative to wild-type, gonadotropes 341 of GnRHR-Ctail mice show sustained calcium oscillations, which extend well after the GnRH 342 pulse. The mechanisms underlying this sustained activity are not clear but depend to some extent 343 on influx of calcium via L-type channels. Regardless, the changes in calcium signaling from 344 primarily transient intracellular release of calcium to a sustained influx of extracellular calcium 345 may contribute to the observed reductions in FSH synthesis in GnRHR-Ctail mice. 346 In contrast, pulsatile LH secretion, which depends upon GnRH-induced calcium 347 mobilization (74), appears to be intact in GnRHR-Ctail males [note that we did not measure 348 pulsatile LH secretion in females because of the high variability between estrous cycle stages (60) 349 and the estrous cycle irregularities in GnRHR-Ctail mice]. This 'normal' LH secretion may be 350 more apparent than real, however. Exogenous GnRH stimulates less LH secretion in male GnRHR-351 Ctail than wild-type mice, despite their equivalent pituitary LH contents. GnRH is similarly less 352 effective in stimulating LH release in GnRHR-Ctail females, but they also show marked decreases 353 13 in pituitary LH content relative to wild-type, precluding a definitive interpretation of the results. 354 LH secretion is blunted in both sexes following gonadectomy and at the time of the LH surge in 355 females. Therefore, the alterations in GnRH stimulated calcium signaling may also affect LH 356 secretion, which is most evident when GnRH pulse frequency or amplitude is enhanced. 357 It is possible that the phenotypes of GnRHR-Ctail mice are explained by reduced receptor 358 expression rather than (or in additional to) altered receptor function. Indeed, Gnrhr mRNA levels 359 are reduced in gonad-intact GnRHR-Ctail relative wild-type mice. We do not know if this 360 translates into differences in GnRHR protein expression. Unfortunately, we were unable to identify 361 reliable antibodies for measurement of GnRHR protein in the pituitary. We also could not validate 362 GnRHR ELISAs used by others (75) (data not shown). In vitro ligand binding assays in pituitaries 363 from the two genotypes do not provide a viable alternative means for receptor protein 364 quantification, as Gnrhr mRNA levels decrease dramatically in cultured cells relative to in vivo 365 and the genotype difference in Gnrhr expression does not persist in culture (data not shown). 366 Regardless, we hypothesize that the reduced Gnrhr mRNA levels in GnRHR-Ctail mice are 367 themselves a consequence rather than a cause of altered GnRH signaling. Not only does GnRH 368 positively regulate the expression of its own receptor (76, 77), but the wild-type and Ctail forms 369 of the murine GnRHR are expressed at equivalent levels when transfected in heterologous cells. 370 Thus, there does not appear to be any inherent difference in the stability of wild-type and Ctail 371 forms of the receptor. 372 373 Evolutionary significance of the loss of the C-tail 374 Finally, in light of all of the results, it is tempting to speculate that the loss of the C-tail 375 from the mammalian GnRHR may have conferred a selective advantage by augmenting G protein 376 coupling, leading to enhanced calcium mobilization, FSH production, folliculogenesis, and 377 fertility. However, more recent phylogenetic analyses suggest that the loss of the C-tail may be an 378 ancient event in jawed vertebrates, predating mammalian evolution (40). It is unclear what 379 advantage this may have conferred when it first emerged and why it has only been retained in 380 mammals and a small number of other vertebrates. We are limited in what we can conclude or 381 interpret from the one mouse model we examined here. While adding the chicken C-tail decreased 382 FSH production, this may not have been the case if we had instead fused the Xenopus or Clarias 383 C-tails, which do not appear to perturb GnRH signaling in vitro. Therefore, the presence of a C-384 tail, in and of itself, does not necessarily impair or alter G protein coupling to the GnRHR. The 385 specific sequence of the tail is relevant. It could be informative to reconstruct ancestral GnRHRs 386 (78) and then examine the effects of removing their C-tails on signaling. Though challenging, this 387 may ultimately provide more, or at least parallel, insight into the potential adaptive significance of 388 the loss of the C-tail. Regardless, the data presented here demonstrate that LH surges are possible 389 in mammals in the presence of a GnRHR with a disruptive C-tail and suggest that FSH synthesis 390 is dependent upon the nature of GnRH-dependent calcium signaling in gonadotropes. 391 incorporating ClaI sites at both ends (Table 1), replacing the amino acids underlined in Figure 10 harboring ClaI sites downstream of adaptors added at both ends. These C-tails were PCR 444 amplified using primers complementary to the adaptor sequences (Table 1), digested with ClaI, 445 purified, and ligated into ClaI-digested dephosphorylated Flag-GnRHR-Ctail, from which the 446 chicken C-tail was excised. All clones were confirmed by Sanger sequencing at GenomeQuébec, 447

MATERIALS AND METHODS
Montreal, Canada. 448 The polycistronic Gαq (82) and DAG (83) biosensors, and β-arrestin-1-YFP (84)  into pGEM-T Easy. The stop codon in exon 3 was replaced with a ClaI restriction enzyme site by 458 site-directed mutagenesis. The ClaI-flanked C-tail coding sequence from the chicken Gnrhr (also 459 used for the Flag-GnRHR-Ctail construct described above) was inserted, and the correct 460 orientation was verified by sequencing. The whole DCA containing the chimeric exon 3 was 461 ligated between the XmaI and NotI sites in pKOII (86), 3' of the Frt flanked neomycin (neo) 462 selection cassette. We used a two-step process to generate the upstream chromosomal arm (UCA) 463 and the "floxed" exon 3 regions. First, a genomic DNA fragment starting 1 kb upstream of exon 3 464 and terminating immediately after the stop codon in exon 3 was amplified by PCR using a 5' 465 primer introducing a XmaI restriction site and a loxP site, and a 3' primer introducing a PmeI 466 restriction site (Table 1). This amplicon, along with a PmeI-XhoI fragment comprising the bovine 467 growth hormone (BGH) polyA sequence (obtained by PCR from the pcDNA3.0 expression vector) 468 were ligated in a 3-part ligation between the XmaI and XhoI restriction sites of pBluescript II KS. 469 To complete the UCA, a 3.6 kb fragment spanning exon 2 and terminating 1 kb upstream of exon 470 3 (the position of the upstream loxP site) was amplified by PCR using primers incorporating 5' 471 KpnI and 3' XmaI sites (Table 1)  then split into three separate plates. Two plates were frozen at -80°C after the addition of 10% 487 DMSO. Cells in the remaining plate were grown to confluence. Genomic DNA was extracted, 488 cleaned with a series of 75% ethanol washes and digested overnight with XmaI. Homologous 489 recombination events were screened by Southern blot using sequential hybridization with 5' and 490 3' probes external to the homology arms (see Table 1 for the primers used to generate the probes). Gnrhr Ctail/+ mice (genotyped using primers Exon3 and Exon3-Ctail in Table 1). Gnrhr Ctail/+ females 501 and males were then crossed to generate wild-type (WT; Gnrhr +/+ ) and Ctail (Gnrhr Ctail/Ctail ) mice. 502 Genotyping was verified by PCR using the gDNA Gnrhr primers (Table 1) Estrous cyclicity was assessed in 6-week-old mice for 21 consecutive days as described in 513 (89). At nine weeks of age, females were paired with WT C57BL/6 males (Charles River, 514 Senneville, QC, Canada) for a six-month period. Breeding cages were monitored daily and the 515 frequency of delivery and number of pups per litter were recorded. Pups were removed from cages 516 14 days after birth. 517

Blood collection 535
Blood was collected by cardiac puncture two weeks post-operatively (on diestrus afternoon 536 for sham-operated females). Blood was allowed to clot for 30 min at room temperature and 537 centrifuged at 1000 g for 10 min to collect serum. Sera were stored at -20 o C until hormone assays 538 were performed. For LH pulsatility assessment in 10-week-old males, 2 µl of blood were collected 539 from the tail tip every 10 min over 6 hours, starting 3 hours before lights off. For the LH surge 540 onset and profile in 10-week-old females, 2 µl of blood were collected from the tail tip every 20 541 min over 8 hours on proestrus (as assessed by vaginal cytology). For LH surge amplitude 542 assessment, four blood samples (4 µl each) were collected from the tail tip over 11 consecutive 543 days: at 10:00 AM, and at 6:00, 7:00 and 8:00 PM. The surge amplitude was defined as the 544 maximal concentration of LH measured on days determined to be proestrus by vaginal smears. For 545 the GnRH-induced LH release experiment, 4 µl of blood were collected from the tail tip of 10-11-546 week-old females (diestrus afternoon) and males just prior to and 15-, 30-, and 60-minutes post-547 i.p. injection of either 0.25 ng (males only), 0.75 ng (females only), or 1.25 ng of GnRH per g of 548 body mass, diluted in 0.9% NaCl. Prior to all tail tip blood collection, animals were acclimatized 549 by massaging the tail daily for two weeks. Tail tip blood samples for LH analysis were immediately 550 diluted (1:30) in 1X PBS containing 0.05% Tween (PBS-T), gently vortexed, and placed on dry 551 ice. Diluted blood was stored at -80 o C until assayed. 552 553

Hormone analyses 554
Serum FSH and LH levels were determined in males at the Ligand Assay and Analysis Core 555 at the University of Virginia Center for Research in Reproduction using the mouse/rat LH/FSH 556 multiplex assay (detection limit: 2.4 to 300 ng/mL; intra-assay CV < 10%). In females, serum FSH 557 was measured by the MILLIPLEX kit (MPTMAG-49K, Millipore, Massachusetts, USA) 558 following the manufacturer's instructions (minimal detection limit: 9.5 pg/mL; intra-assay CV < 559 15%) and serum LH was measured using an in-house sandwich ELISA, as previously described 560 (92) (detection limit: 0.117 to 30 ng/mL; and intra-assay CV < 10%). Whole blood LH levels from 561 both males and females were also measured using the in-house sandwich LH ELISA. 562 563 Gonadotropin pituitary content assessment 564 Pituitaries were collected from 12-to 13-week-old female (randomly cycling) and male 565 mice, placed on dry ice, and manually homogenized in 300 μl cold 1X PBS. Homogenates were 566 centrifuged at 13,000 rpm for 15 minutes at 4°C. Total protein concentration was measured using 567 20 the Pierce BCA Protein Assay Kit (23225; ThermoFisher Scientific) following the manufacturer's 568

instructions. 569
For FSH content assessment, samples were diluted 1:50 and FSH levels were measured by 570 the MILLIPLEX kit (females) or by RIA (males) at the Ligand Assay and Analysis Core at the 571 University of Virginia Center for Research in Reproduction. For LH pituitary content, samples 572 were diluted 1:1,000,000 in PBS-T, and LH levels were measured using the LH ELISA indicated 573 above. FSH and LH values were normalized over total protein content per pituitary. 574 575

GnRH treatment of LβT2 cells 576
LβT2 cells were plated at 650,000 cells/well in 12-well plates and cultured overnight. The 577 next day, cells were starved for 16-18 hours in serum-free medium. Cells were then pre-treated for 578 20 minutes with BAPTA-AM (20 µM) and then stimulated with one pulse of 10 nM GnRH 579 (hereafter referred to as low GnRH pulse frequency). Two hours post-GnRH stimulation, media 580 was replaced with fresh media containing the BAPTA-AM and incubated for an additional 2 hours. 581 For high GnRH pulse frequency treatment, cells were stimulated with 10 nM GnRH for 5 min, 582 every 45 minutes for a total of 10 pulses in the presence of the BAPTA-AM (20 µM) or vehicle. 583 The latter were also included between GnRH pulses. 584 585

Reverse transcription and quantitative PCR 586
Pituitaries were collected two weeks post-gonadectomy (on diestrus afternoon for sham-587 operated females), snap-frozen in liquid nitrogen, and stored at -80°C. Total RNA from pituitaries 588 and LβT2 cells was isolated with TRIzol following the manufacturer's instructions. Pituitaries 589 were first homogenized in 500 μl TRIzol using a Polytron PT10-35 homogenizer. RNA 590 concentration was measured by NanoDrop and 250 ng of RNA per sample were reverse transcribed 591 as in (93). Two μl of cDNA were used as a template in 20 μl reactions for quantitative-real time 592 PCR analysis on a Corbett Rotorgene 600 instrument (Corbett Life Science) using EvaGreen 593 reagent master mix. Relative gene expression was determined using the 2 -ΔΔCt method (94) with 594 the housekeeping gene ribosomal protein L19 (Rpl19) as reference (primers in Table 1). 595 596 21 Bioluminescence resonance energy transfer (BRET) assays 597 HEK 293 cells were plated at a density of 400,000 cells/well in 6-well plates. The next day, 598 cells were co-transfected with PEI with 1 μg GnRHR-WT or GnRHR-Ctail expression vector (or 599 empty vector as control) along with 1 μg of a polycistronic Gαq biosensor (95) or DAG biosensor 600 (83). Twenty-four hours post-transfection, cells were detached by manual pipetting, and plated on 601 poly-D-lysine-coated 96-well white plates at a density of 50,000 cells per well. The next day, cells 602 were washed twice with Tyrode's buffer (140 mM NaCl, 1 mM CaCl2, 2.7 mM KCl, 0.49 603 mM MgCl2, 0.37 mM NaH2PO4, 5.6 mM glucose, 12 mM NaHCO3, and 25 mM HEPES, pH 7.5). 604 Next, cells were loaded with 5 μM Coelenterazine 400a for 5 min in the dark at room temperature, 605 and signals were subsequently recorded by a Victor X light plate reader (Perkin Elmer Life 606 Sciences) starting 10 seconds before and continuing 30 seconds after 100 nM GnRH (or vehicle) 607 injection, at 0.33 millisecond intervals. Net BRET was calculated as the ratio of the acceptor signal 608 (GFP10 515/30 nm filter) over the donor signal (RLucII, 410/80 nm filter). ΔBRET was calculated 609 by subtracting the average of basal BRET signals from ligand-induced signals (96). Experiments 610 with the Gαq biosensor and Anolis, Xenopus, and Clarias chimeric receptors were conducted as 611 above, with the exception that a Synergy 2 Multi-Mode Microplate Reader (Bio Tek) was 612 employed. Acceptor and donor signals were read three times before and after 100 nM GnRH (or 613 vehicle) injection, at 16 s intervals. 614 615 Cell line protein extraction, immunoprecipitation, and western blotting 616 Cellular extracts from HEK 293 and LbT2 cell lines were isolated using RIPA lysis buffer 617 (50 mM Tris-HCl, 150 mM NaCl, 10 mM EDTA, 1% Triton X-100) as described in (97) 3 washes with TBS-T, membranes were further incubated with horseradish peroxidase (HRP)-628 conjugated goat anti-rabbit antibody (1:5000) in blocking solution for 2 hours at room temperature. 629 For assessment of total ERK1/2 expression, membranes were stripped with 0.3 M NaOH, washed 630 and incubated with anti-ERK1/2 (1:1000) following the same procedure as above for phospho-631 ERK1/2. Blots were incubated in Western Lightning ECL Pro reagent (Perkin Elmer) and then 632 exposed on HyBlot CL film (E3012, Denville Scientific) or with a digital GE Amersham Imager 633 600. Band intensities were measured in arbitrary units using Image J software (       Intracellular Ca 2+ was measured as relative luminescence emitted every 22 ms over 0.5 min. Data are presented as the ratio of total luminescence after GnRH over maximal luminescence (not shown) following Triton X-100 treatment from 3 independent experiments (mean ± SEM).            Comparisons of AUC from a (B) wild-type (275 cells; 11.5 ± 4.7 vs. 7.1 ± 2.8 a.u. for GnRH and GnRH/Nim, respectively; p < 0.0001) and a (E) Ctail mouse (127 cells; 8.15 ± 3.9 vs. 5.4 ± 2.0 a.u. for GnRH and GnRH/Nim, respectively; p < 0.0001).