Impaired episodic LH secretion in female mice with GFP in GnRH neurons

Abstract

The ability to assess the activity of gonadotropin-releasing hormone (GnRH) neurons has been greatly enhanced by transgenic animal models with targeted expression of green fluorescent protein (GFP). However, it has yet to be demonstrated that the GnRH system continues to exhibit a full range of normal physiological functions in the presence of such genetic manipulation. Accordingly, we have used repetitive blood sampling via indwelling venous catheters to define LH secretory patterns in normal and transgenic mice. Transgenic females proved to be reproductively competent as defined by fecundity, appropriate cyclic changes in vaginal cytology in intact adult females, and spontaneous LH surges as well as surges in response to steroid or mating stimuli. The expression of c-fos following such steroid treatment and mating in ovariectomized transgenics was similar to the expression previously reported in nontransgenic mice. Likewise, the percentage of retrogradely labeled GnRH neurons was similar to that reported in nontransgenic mice. However, episodic LH secretion, an index of GnRH pulse generator activity, was dramatically compromised in ovariectomized female transgenics compared with C57BL6 controls of both sexes and castrated transgenic males. Taken together, these findings suggest that the GnRH pulse generator is selectively impaired in ovariectomized females in which GnRH neurons express GFP.

hypothalamic gonadotropin-releasing hormone (GnRH) neurons are part of a neurosecretory system that is requisite for reproductive function. These neurons are few in number and diffusely distributed (29), making it difficult to identify GnRH neurons in slices of hypothalamic tissue. The development of transgenic mice in which expression of the reporter gene green fluorescent protein (GFP) is regulated by a portion of the GnRH gene promoter has circumvented some of these anatomical constraints. As originally demonstrated by Spergel et al. (30), a 3.47-kB fragment of the GnRH gene promoter directs GFP to hypothalamic GnRH neurons, resulting in GnRH neurons in which GFP is expressed specifically, conveying fluorescence to the cells and thereby permitting visualization of living neurons (30, 31).

The usefulness of such animals as a tool for studies of living GnRH neurons is based on the assumption of normal physiological function(s) after incorporation of GFP. Thus, we determined the functional status of three known modes of GnRH-related secretion in female mice: episodic LH secretion, the estradiol-induced LH surge, and the mating-induced LH surge. In transgenic females, we examined c-fos expression and retrograde labeling with fluorogold in the GnRH neurons as well as reproductive characteristics. In castrated transgenic males, we examined episodic LH secretion and fluorogold labeling.

MATERIALS AND METHODS

Animals.

Animals used in these studies were obtained from Spergel et al. (30) and had been bred to homozygosity. We focused on activity in homozygous animals after many generations of breeding, since this is the status of the animals presently used for experiments. Control C57BL6 animals were purchased from Charles River Laboratories (Wilmington, DE). Animals from the C57BL6 background were used as controls so that this initial characterization of hormone secretion could be performed without possibility of potential confounding by the presence of any transgenic proteins. Repetitive sampling experiments were performed at Georgia State University. Other components were performed at Emory University. All studies were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and Institutional Animal Care and Use Committee approval was obtained at the above-mentioned institutions. Animals were maintained on a 12:12-h light-dark cycle (lights out at 1900) with ad libitum access to water and food.

Assessing pulsatile GnRH release.

In rats, a strong temporal relationship exists between the onset of a GnRH pulse in hypothalamic perfusates and the secretion of LH into the plasma (19, 20), indicating that release of LH is a reasonable index of episodic GnRH release. Accordingly, we used repetitive sampling of venous blood in castrated mice to determine the pattern of LH secretion.

At the time of repetitive sampling, animals ranged from 8 to 13 mo of age. We used animals of these ages to maximize the volume of blood available for sequential samples. Moreover, these ages correspond to the age of maximum estrous cycle frequency reported for C57BL6 females (25). At the time of sequential sampling, the average age (mean ± SE) of transgenic females (n = 8) was 11.7 (±0.60) mo, and control animals (n = 6) were an average of 12.1 (±0.24) mo of age. Transgenic males (n = 14) were studied at 10.6 (±0.8) mo of age and control males (n = 7) at 11.2 (±0.16) mo of age. Transgenic and control females had been ovariectomized 1.8 (±0.20) or 1.9 (±0.30) mo, respectively. Male transgenics had been castrated for 2.0 (±0.58) mo, and male controls had been castrated for an interval of 2.6 (±0.60) mo.

Surgical procedures.

All surgeries (bilateral ovariectomy, bilateral orchidectomy, and catheterization) were performed aseptically using isoflurane anesthesia. The tip of a catheter (a 35-mm segment of Silastic tubing: inner diameter 0.3 mm, outer diameter 0.64 mm) was positioned at the junction of the right atrium via the external jugular vein. The other end of the catheter was tunneled subcutaneously before being exited dorsally between the scapulae. Catheters were filled and flushed daily with heparinized saline (100 U heparin/ml saline).

Repetitive sampling.

Three to five days after catheterization, animals were placed on a tethering system (22) to minimize disruption during sampling. The following day, sequential blood samples were taken between 0800 and 1200, an interval during which episodic LH secretion has been demonstrated in mice (3, 4, 10, 18). Blood samples were collected in 1-ml syringes at 12-min intervals for 3 h. Following withdrawal of each 50-ul sample, blood volume was replaced with Plasmamate (Dublin Medical, San Diego, CA) supplemented with K⁺ (5.2 meq/l), glucose, and saline-washed red blood cells from C57BL6 donor animals. Plasma was obtained by centrifugation in 70-μl capillary tubes (Fisher Scientific, Pittsburgh, PA) in which hematocrits could be monitored. Plasma was subsequently dispensed into PCR tubes and frozen. At the termination of a sampling sequence the intravenous catheter was flushed, filled with heparinized saline, and plugged with a sterile gold pin (World Precision Instruments, Sarasota, FL). In addition to the untreated castrates described above, three female transgenics that had been ovariectomized 2 wk earlier were treated with the GnRH antagonist Antide in a solution containing 300 ng/ml in 20% propylene glycol and normal saline (Sigma Chemicals, St. Louis, MO), using a protocol previously demonstrated to suppress LH secretion following ovariectomy in mice (23).

Assessing the steroid-induced GnRH surge.

We employed a protocol previously demonstrated to result in LH (and presumably GnRH) surges in female mice. Four days following ovariectomy, animals received a 10-mm subcutaneous priming capsule containing estradiol benzoate (1 μg/capsule; Sigma) as described (7, 8). Seven days later, each animal recieved two additional 20-mm capsules at 0900 for induction of an LH surge. On the afternoon of the following day, animals were euthanized at 1200 (n = 5), 1400 (n = 4), 1600 (n = 4), 1950 (n = 5), or 2200 (n = 4). For reference, circulating levels of LH during spontaneous surges were determined via vaginal lavage, as detalied below. In this group of virgin females (n = 4), trunk blood was collected between 1900 and 2000 subsequent to a proestrous smear (see below). Trunk blood was obtained for LH assay, at which time brains were collected and placed in 4% paraformaldehyde in 0.1 M phosphate-buffered saline (PBS). Animals were 6.7 (±1.30) mo of age at the time of the experiments and had been part of the breeding colony. Each female had produced at least three litters.

Assessing the mating-induced GnRH surge.

Females were ovariectomized and 12 days later injected subcutaneously (sc) with 2 μg of estradiol benzoate at 0900. On day 14, postovariectomy animals were injected sc with progesterone (500 μg at 0900). Four hours later, females either were caged with a proven stud male for a 2-h mating test or served as unmated controls (26, 35, 36). Following the mating test interval, both groups of females were promptly euthanized, and trunk blood and brains were collected. Animals were 6.2 (±1.70) mo of age at the time of experiments and had been part of the breeding colony where each animal had produced at least three litters.

LH assay.

Plasma samples (∼20 μl) obtained during repetitive sampling were assayed as singles (14). All samples from females were analyzed in a single LH assay (minimum detectable dose of 0.4 ng/ml and an intra-assay coefficient of variation of 2.4%). All samples from male animals were analyzed in a second assay (minimum detectable dose of 0.42 ng/ml and an intra-assay coefficient of variation of 3.2%). Plasma samples obtained during the estradiol-induced and mating-induced LH surge experiments were assayed at a volume of 100 μl.

Pulse detection.

Statistically significant episodes of hormone secretion were defined using Pulsar (24). The following G values were used: G(1) = 3.98, G(2) = 2.2, G(3) = 1.5, G(4) = 1.24, and G(5) = 0.93. The LH secretory episodes (peaks) were confirmed using cluster 8 (34).

Reproductive characteristics.

Time of vaginal opening and age at first proestrous were determined in transgenic females derived from three litters (n = 17). Additionally, vaginal cytology was examined in three cohorts of virgin females at 60–80 days of age (n = 8), 5–6 mo of age (n = 8), and 8–9 mo of age (n = 8). Animals were housed in group cages (4–5 mice/cage), and vaginal cytology was monitored daily for 14 days. Cells were collected using vaginal lavage between 0900 and 1300 and stained with hematoxylin and eosin (Sigma). Cycles were classified according to parameters desribed previously for the C57BL6 background strain (25). Four females from the two younger cohorts were also randomly assigned for repetitive blood sampling at a later date. A review of breeding records over 13 mo of colony existence provided data to determine average litter size.

Anatomical studies: c-fos immunocytochemistry.

Hypothalami from female animals used to test for steroid- and mating-induced LH surges were sectioned on a vibrating microtome at a thickness of 50 μm (from rostral forebrain to optic chiasm). The anatomical location of the hypothalamic sections corresponded to plates 17–30 of Paxinos and Watson's Atlas for the mouse brain (6). Sections were washed several times in PBS and then washed for 30 min in 0.5% hydrogen peroxide followed by an additional 30-min wash in 1% normal goat serum in a 0.1% solution of Triton X-100 (Sigma). Sections were incubated in fos primary antibody (Calbiochem, San Diego, CA) at a concentration of 0.1 ug/ml for 48 h at 4°C. Sections were then incubated for 60 min in the secondary antibody (biotinylated goat anti-rabbit-γ globulin, diluted 1:400; Vector Laboratories, Burlingame, CA) followed by a 60-min incubation with avidin-biotin-horseradish peroxidase (1:400; Vector Laboratories). Nuclear fos was visualized with 3,3′-diaminobenzidine tetrahydrochloride as described (35, 36), but without nickel enhancement, which frequently obscured GFP visualization. GnRH neurons labeled with GFP were counted and visually classified as either expressing or not expressing c-fos. Neurons were counted beginning with the section that contained the most rostral extent of the corpus callosum together with the anterior portion of the anterior commissure but caudal to the section containing the indentation indicating the rhinal fissure. Sections included in the analysis extended to the level of the brain where the anterior commissure forms a complete horizontal “ban” across the brain.

Retrograde labeling of GnRH-GFP neurons.

Intact animals (n = 9) were administered sc with 500 μl of fluorogold solution (2 mg/ml) in sterile water (Fluorochrome, Denver, CO). Each animal received two such injections 24 h apart. Fluorogold does not permeate the blood-brain barrier (27) and is well suited for retrograde labeling of GnRH neurons (28). Two weeks after fluorogold injection, the animals were anesthetized with halothane and decapitated. Hypothalami were sectioned from the rostral forebrain to the arcuate nucleus (50-μm sections) and mounted on charged slides. GnRH neurons tagged with GFP were counted and classified as either fluorogold-containing or non-fluorogold-containing neurons, as determined by fluorescence microscopy using an Olympus BX-51 microscope and the fluorescein isothiocyanate filter cube (excitation maximum 488 nm absorption maximum of 495 nm, emission efficiency 87%). Images were obtained using Magnafire Software (using 10 bits/color channel and 5-s exposure times; Optonics, Goleta, CA). Images were saved as 48-bit TIFF files and rendered in Adobe Illustrator without any color or contrast enhancement.

Statistics.

Differences in LH pulse frequency and pulse amplitude were compared using a one-way analysis of variance (P < 0.001). In the case of steroid-induced and mating-induced responses, differences in plasma LH concentrations were also analyzed by one-way analysis of variance (P < 0.001). The difference in percentages of c-fos expression were determined using the Kruskal-Wallis one-way analysis of variance and a Mann-Whitney U-test (P < 0.01).

RESULTS

Figure 1A details plasma LH concentrations in four of eight transgenic females spanning the age range of animals used in the present study (Fig. 1A, graphs 1–4) and in two of six C57BL6 controls (Fig. 1A, graphs 5 and 6). Episodic LH secretion was robustly expressed in control females but was significantly limited in transgenic females. The mean plasma LH concentrations, LH pulse amplitudes, and LH pulse frequencies measured in female mice are shown in Fig. 1B. LH concentrations in serum from one transgenic and one control female animal were undetectable, and, since these animals had been sampled on the same day, data from these two animals were excluded from the numerical analyses. LH pulse frequency was reduced in transgenic females relative to controls, although mean levels of LH and LH pulse amplitudes did not differ between these two groups. This difference was striking in that transgenic females exhibited only one pulse during the 3-h sampling period, whereas control animals produced 3–5 pulses (4.1 ± 0.5). Although not characterized by a robust episodic secretory pattern, circulating LH in transgenic mice appeared to retain dependence on GnRH secretion, as indicated by the greatly reduced plasma LH concentrations in three transgenic females that had been ovariectomized and treated with the GnRH antagonist Antide. Such treatment reduced plasma LH concentrations from 2.91 (±0.47) ng/ml at 2 wk postovariectomy to undetectable (e.g., <0.4 ng/ml) following treatment (data not shown).

Fig. 1.

A: representative patterns of episodic LH secretion in ovariectomized female mice of GnRHhGFP2 (graphs 1–4) or ovariectomized nontransgenic C57BL6 lineage (graphs 5 and 6). A total of 8 transgenic females and 6 control females were sampled. *Secretory episodes as identified by Pulsar. B: mean LH pulse amplitude and pulse frequency in ovariectomized gonadotropin-releasing hormone (GnRH)-green fluorescent protein (GFP) mice (n = 7 from total of 8 sampled transgenic females; black bars) or ovariectomized nontransgenic animals of the C57BL6 lineage (n = 5 from a total of 6 sampled female controls; open bars). Data from 2 females (1 transgenic and 1 control) that were sampled on the same day had undetectable LH levels and were excluded from the pulse analyses. Pulse frequency was significantly higher in control animals relative to transgenics (P < 0.05).

Robust LH pulses were observed in both castrated transgenic and control castrated males (Fig. 2A). All males exhibited at least two LH pulses during the 3 h of sampling (range: 2–6 pulses). Transgenic males (Fig. 2A, graphs 1–4) had an average of 2.8 (±0.7) pulses per 3 h and control males (Fig. 2A, graphs 5–6) had an average of 2.3 (±0.9) pulses per 3 h. Mean LH levels, pulse amplitude, and pulse frequency did not differ between transgenic and control males (Fig. 2B).

Fig. 2.

A: representative patterns of episodic LH secretion in castrated male mice of GnRHhGFP2 (graphs 1–3) or castrated nontransgenic C57BL6 lineage (graphs 4–6). A total of 14 transgenic males and 7 control males were sampled. *Secretory episodes as identified by Pulsar. B: mean LH, pulse amplitude, and pulse frequency in castrated GnRH-GFP male mice (n = 14; black bars) or castrated nontransgenic animals of the C57BL6 lineage (n = 7; open bars). Parameters of hormone secretion did not differ between controls and transgenics (P < 0.05).

Despite the absence of normal pulsatile hormone secretion in ovariectomized animals, readily discernable 4- or 5-day ovarian cycles were observed in intact transgenic females between 5 and 9 mo of age. This group included three females in which repetitive blood samples were subsequently obtained (data from 2 of the 3 animals are shown in Fig. 1A, graphs 3 and 4), and no more than one pulse was observed during any of the 3-h sampling intervals. Vaginal opening occurred in transgenic females at 27.3 (±3.2) days of age. The onset of ovarian cycles as determined by vaginal cytology (first proestrous smear) was 57.4 (±4.1) days of age. These parameters were comparable with those reported for the second line of GnRH-GFP transgenic females (29) in which vaginal opening occurred at 28 (±2.0) days of age and the onset of fertile cycles in homozygotic females was 60.8 (±6.4) days of age. In the youngest group of females (60–90 days of age), changes in vaginal cytology were apparent but often did not present a clearly cyclical pattern. In older females (i.e., those between 5 and 6 mo and between 8 and 9 mo of age), cycles were comparable with those previously reported for the background strain (23). Litter size in the breeding colony ranged from seven to 10 pups, with an average of 9.0 (±1.3) pups/female. On average, females produced litters 30.1 (±2.7) days after being placed with a proven male.

To determine whether the differences in hormone secretion were due to a failure of GnRH neurons to access the portal vasculature, we used retrograde labeling with fluorogold. A total of 1,208 GnRH neurons from females (n = 5) and 1,310 GnRH neurons from males (n = 4) were examined for the presence of fluorogold. Fluorogold could be identified by the punctuate “flecks” of yellow-gold superimposed on the emerald background of GFP-florescence (Fig. 3C, inset). The average number of fluorogold-containing neurons per animal was 112 (±37) in females and 99 (±20.2) in males. Overall, fluorogold was detected in 41.9 (±5.7) and 38% (±7.2) of the neurons in females and males, respectively.

Fig. 3.

A–C: representative sections of brain tissue containing fluorogold-containing neurons and GFP-labeled GnRH neurons (at arrows). Sections correspond to plates 26, 28, and 29 in Ref. 6. C, inset: a fluorogold-positive GnRH-GFP neuron indicated by the punctuate yellow granules (at arrows in the inset). The anatomical location of the fluorogold-positive and 2 fluorogold-negative GnRH neurons is indicted by the white rectangle in C.

To determine whether other modes of hormone secretion were compromised in transgenic females, we examined the steroid- and mating-induced GnRH release and the spontaneous preovulatory surge again using LH secretion as our end point. Figure 4A shows the time course of estradiol-induced surge in transgenic females. Circulating levels of LH were low at 1200. The first significant increase in plasma LH secretion was detected at 1600 and persisted until at least 2200 (P < 0.05). Maximum LH concentrations occurred at 1950, shortly after lights were turned off in the animal room during the steroid-induced surge. Levels of LH were also elevated at this time during a spontaneous surge whose timing was determined on the basis of vaginal lavage (Fig. 4A, gray bar). Steroid-induced surges had significantly higher levels of LH compared with hormone levels during the spontaneous surge (P < 0.05). Activation occurred in 70% of GnRH neurons during the steroid-induced surge, as indicated by the presence of c-fos (Fig. 4B). Neurons positive for c-fos were distributed through the rostral-caudal extent of the GFP-expressing population (Fig. 4C).

Fig. 4.

A: time course of steroid-evoked LH surge in transgenic females. Filled bars indicate samples obtained after lights out (1900); black bars indicate LH levels in steroid-treated animals, and gray bar indicates LH levels during a spontaneous surge in a separate group of females. One animal in the steroid-treated group had circulating levels of LH that were greater than the limit of assay detection (20 ng/ml) at the volume assayed. Since there was insufficient serum to reassay this sample, it was assigned a value of 20 ng/ml. Steroid-induced surges had significantly higher levels of LH compared with hormone levels during the spontaneous surge (P < 0.05). B: %GnRH neurons expressing c-fos during the steroid-induced surge. C: distribution of c-fos expression in a GFP-identified GnRH neuron during the steroid-induced surge. Templates 17, 24, and 29 for mouse hypothalamic anatomy have been modified from Ref. 6 and are reproduced with permission, containing data from the present study. AC, anterior commissure; DBB, diagonal band of Broca. *Significant increase in LH levels over measurements at both 1200 and 1400.

Figure 5A shows the mating-induced LH surge in transgenic females. Circulating levels of LH were low in unmated females. However, following a 2-h mating test, plasma LH was significantly increased in females observed to experience at least one intromission. Activation occurred in ∼60% of GnRH neurons in mated females, as indicated by the presence of c-fos (Fig. 5B). Neurons positive for c-fos were distributed through the rostral-caudal extent of the GFP-expressing population (Fig. 5C).

Fig. 5.

A: plasma LH concentrations during the mating-induced surge in transgenic females. B: %GnRH neurons expressing c-fos following mating. C: distribution of c-fos expression in GFP-identified GnRH neurons following mating with intromission. Templates 17, 24, and 29 for mouse hypothalamic anatomy have been modified from Ref. 6 and are reproduced with permission, containing data from the present study.

DISCUSSION

Several mouse models have been developed in which neurons express GFP (1, 5, 33, 37). Some attempts have been made to validate GFP for use in neuronal identification by demonstrating that single channel currents in cultured cells or oocytes were not significantly altered by the expression of GFP (2, 21).

Clear indications of appropriate physiological function are particularly important if transgenic animals expressing GFP are to be used for electrophysiological studies of GnRH neurons. Recordings from unlabeled neurons, which would help define any untoward effects of GFP expression, may be feasible in some neuroendocrine systems (1, 37) but are virtually impossible to obtain in GnRH neurons. Since a majority of GnRH neurons express GFP in the available transgenic mouse models (30, 31), we opted to monitor the functionality of GnRH neurons as a population by assessing intermittent LH secretion in both male and female transgenic mice.

The findings of the present study suggest that the frequency of episodic GnRH release is significantly reduced in female mice expressing GFP in GnRH neurons. Pulse frequencies in control females were within the ranges reported for agonadal mice (3, 4, 10, 18), but a majority of female transgenic mice expressed, at a maximum, only one LH pulse in 3 h. The only other study of LH pulses in nontransgenic ovariectomized mice reported an interpulse interval of ∼80 min (10). It is unlikely that the reduction in episodic hormone secretion in the present study is an artifact of blood sampling since the C57BL6 control females exhibited a relatively normal pulse frequency, although we used a slightly longer sampling interval than the earlier study in females (i.e., 12 vs. 10 min). In addition, the attenuated pulse frequency in transgenic females likely does not reflect increased age since C57BL6 females exhibit peak cycle frequency between 5 and 12 mo of age (25) and, in the present study, control and transgenic groups were of comparable age. Moreover, reduced pulsatility probably cannot be accounted for by failure of the GnRH nerve terminals to reach or access the portal vasculature. In the present study, the percentage of GFP-GnRH neurons in transgenic females labeled with fluorogold was similar to that reported for nontransgenic mice (17, 28) and similar to transgenic males. It should be noted that retrograde labeling of some GnRH neurons could be a consequence of uptake at the level of the organum vasculosum of the lamina terminalis, which is outside the blood-brain barrier. However, as previously reported, the median eminence is densely innervated by GFP-containing fibers in these mice (30, 31). Therefore, it is likely that GFP-labeled GnRH neurons have access to the portal vasculature. Finally, although the pertinent data are limited to three females, LH secretion was undetectable following treatment with a GnRH antagonist. Therefore, it is unlikely that the secretory profiles of LH in transgenics are independent of GnRH secretion. Given the robust relationship between episodic GnRH and LH secretion observed in gonadectomized rats (19, 20), our findings support the conclusion that episodic GnRH secretion is reduced in female mice in which GnRH neurons are expressing GFP.

The spontaneous and estradiol-induced surge was normal in GnRH-GFP transgenic females. Although this functionality might be assumed (based on fecundity), it should be noted that healthy litters are born to female hpg mice following transplantation of GnRH neurons from donors (9), although such females fail to express spontaneous LH (and presumably GnRH) surges. Moreover, only 30% of hpg females receiving donor GnRH neurons exhibit an LH surge in response to steroid treatment (11). Instead, ovulation is driven by a mating-induced reflex release of hormone (10), a response that appears to be controlled by neural circuitry distinct from a classical pulse or surge generator, at least in males (4). Therefore, successful breeding does not support the conclusion that the activity of GnRH-GFP neurons and/or the neural circuitry functions normally in transgenic animals.

In the present study, both the endocrine-driven surge and the mating-induced GnRH surge are operant. The spontaneous LH surge also occurs. These observations are important because they indicate that the reduction in pulsatility does not reflect a global impairment of hormone secretion in females. The expression of c-fos in GFP-identified GnRH neurons during the surge is consistent with a hypothalamic (as opposed to pituitary) mechanism of surge generation in response to estradiol and the mating stimulus. Taken together, these data indicate that the classical neuroendocrine reflex(es) underlying the preovulatory and sensory-mediated mating-induced surge appears to be functional in females with GFP-expressing GnRH neurons despite the reduction in episodic hormone secretion in the ovariectomized animal. In this regard, a recent study demonstrated that relatively few GnRH neurons sustain aspects of reproductive function that are associated with episodic GnRH release (e.g., the onset of puberty), whereas a larger pool of GnRH neurons were required to generated the preovulatory GnRH surge (15). Transgenic females in the present study generated robust surges in response to three different paradigms (i.e., the endogenous estradiol signal, simulated hormone treatment, and mating). These findings suggest that it is unlikely that the limiting factor for the intermittent mode of hormone secretion in females in the present study is a reduction in the number of GnRH neurons.

A further reduction of pulse frequency probably occurs in intact females, but this could not be determined due to blood volume and hormone detection issues. It appears, however, that any additional attentuation resulting from steroid-dependent inhibition had no impact on other reproductive features such as ovarian cycles (as assessed by vaginal cytology) and fecundity (as assessed by litter size). Although this may seem curious, it must be noted that gonadal function in mice appears to be supported over a wide range of pulse generator outputs (3, 4, 10, 18). Moreover, a continuous (non pulsatile) GnRH infusion drives gonadal development and vaginal opening in female hpg mice that lack endogenous GnRH secretion due to a spontaneous deletion in the GnRH gene. Some of these hpg females exhibited changes in vaginal cytology, although ovulation could not be confirmed by histology (13). This, coupled with an observation of normal gondal function in intact males with long interpulse intervals (3, 4), suggests that a regular pattern of intermittent GnRH secretion may be less critical in mice than it is in rats, humans, and nonhuman primates.

The underlying cause of reduced pulsatility in transgenic females is not clear. In the present study, normal intermittent secretion in males suggests that this deficit may not be due to the presence of GFP in neurons per se. However, multiple internal and external factors regulate reproduction (16). Some of these internal and external cues may be more important in regulating pulsatile hormone secretion in females than males. A recent study revealed that the complexity of receptor phenotypes expressed GnRH neurons in GnRH-GFP female mice (32). It is possible that the function of receptors that are important in controlling pulsatile hormone in females is altered by GFP expression. Second, GnRH neurons likely require integration into a variety of hypothalamic circuits (which may differ between males and females) that relay these internal and external cues. The diffuse distribution of GnRH neurons is widely recognized (29), and the gross distribution of individual GnRH neurons appears similar between trangenic and nontransgenic mouse models. However, the required precision of the final anatomical positioning of GnRH neurons within the hypothalamus for proper reproductive function is unknown. Therefore, it is possible that subtle differences in anatomical positioning and, by extension, integration of individual GnRH neurons into hypothalamic circuitry are altered by expression of GFP. This speculation notwithstanding, Gibson et al. (13) urged caution with respect to inferences of pulsatile GnRH release in hpg mice on the basis of gonadal function following transplantation of donor GnRH neurons. Our findings indicate that similar caution is also warranted in the analysis of data from GnRH-GFP transgenic mouse models.

GRANTS

This study was supported by National Institute of Child Health and Human Development (NICHD) Grants HD-41380 and HD-45436 to K. J. Suter and by National Center for Research Resources Grant RR-00165 to Yerkes Primate Research Center. Stephanie Chen collected some of the anatomical data and was supported by a Minority Student Fellowship from the Howard Hughes Foundation (Grant No. 71100-529803). Cluster analysis was performed as part of the workshop Analysis of Hormone Pulsatility at the University of Virginia, Center for Biomathematical Technology. Assays to determine circulating levels of estradiol generated by implants were performed by the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core, NICHD (SCCPRR), U54-HD-28934.