Natural variation in neuroendocrine traits is poorly understood, despite the importance of variation in brain function and evolution. Most rodents in the temperate zones inhibit reproduction and other nonessential functions in short winter photoperiods, but some have little or no reproductive response. We tested whether genetic variability in reproductive seasonality is related to individual differences in the neuronal function of the gonadotropin-releasing hormone network, as assessed by the number and location of mature gonadotropin-releasing hormone-secreting neurons under inhibitory and excitatory photoperiods. The experiments used lines of Peromyscus leucopus previously developed by selection from a wild population. One line contained individuals reproductively inhibited by short photoperiod, and the other line contained individuals nonresponsive to short photoperiod. Expression of mature gonadotropin-releasing hormone (GnRH) immunoreactivity in the brain was detected using SMI-41 antibody in the single-labeled avidin-biotin-peroxidase-complex method. Nonresponsive mice had 50% more immunoreactive GnRH neurons than reproductively inhibited mice in both short- and long-day photoperiods. The greatest differences were in the anterior hypothalamus and preoptic areas. In contrast, we detected no significant within-lines differences in the number or location of immunoreactive GnRH neurons between photoperiod treatments. Our data indicate that high levels of genetic variation in a single wild population for a specific neuronal trait are related to phenotypic variation in a life history trait, i.e., winter reproduction. Variation in GnRH neuronal activity may underlie some of the natural reproductive and life history variation observed in wild populations of P. leucopus. Similar genetic variation in neuronal traits may be present in humans and other species.
- genetic variation
- artificial selection
- evolutionary physiology
- brain variation
- gonadotropin-releasing hormone
within-population genetic variation regulating the abundance, location, and connections of neurons must contribute to evolution of brain function. Similarly, natural genetic variation in neuronal traits is presumably responsible for some proportion of intraspecific functional variation in vertebrates. Because neural circuitry regulates reproductive physiology and behavior, neural variation is likely to explain some unknown proportion of life history variation within as well as among species. At present, almost nothing is known about natural levels of genetically based neuroendocrine physiological variation related to life history variation within species of mammals. The physiological link between genes and life history patterns is important because genes must act on life history traits through physiological mechanisms, and thus physiological variation may shape or constrain life history evolution (16). Understanding natural neuroendocrine variation related to life history traits might help us learn how rapidly brains adapt to current challenges, including natural and anthropogenic change in biotic and abiotic factors.
The photoneuroendocrine pathway, a major neuroendocrine pathway regulating reproduction and other responses, is a complex neural circuit through which information about photoperiod causes changes in winter reproduction and other seasonal responses. This pathway is relatively well described physiologically in rodents (9, 34), and there are natural populations that contain genetic variation for winter reproductive phenotype (33). Because of this combination of qualities, this system has become a model for the study of physiological variation (16, 34). Individuals of many temperate-zone species limit energy expenditure during the winter months by inhibiting functions that are not essential for immediate survival, such as reproduction. This represents a useful adaptation for endotherms, since they must cope with the increased metabolic demands imposed by the harsh, cold climate, the relative lack of available food, and the increased risk of predation associated with extended foraging time (5, 7). Changes in the duration of day and night are the major cues triggering reproductive inhibition (7, 9). The signal for light is passed from a unique class of retinal photoreceptors through the retinohypothalamic tract to the suprachiasmatic nuclei, a major component of the circadian clock necessary for timekeeping and assessment of the duration of day (15). Neuronal signals are passed from the suprachiasmatic nuclei to the paraventricular nuclei of the hypothalamus, to the superior cervical ganglia, and then to the pineal gland via adrenergic neurons of the sympathetic nervous system (34). The pineal gland releases the hormone melatonin only during the dark phase, providing a physiological signal proportional to the duration of night and day (3, 15, 34).
The duration of the melatonin signal presumably alters the response of hypothalamic neurons that secrete gonadotropin-releasing hormone (GnRH), a decapeptide that is the master hormonal regulator of mammalian reproduction and the final common pathway centrally controlling reproduction in vertebrates (9, 36). GnRH is released in pulses and is transported via the hypothalamic hypophyseal portal vessels to the anterior pituitary, where it controls the release of luteinizing hormone (LH) and follicle-stimulating hormone (FSH). LH and FSH, in turn, regulate gonadal development, reproductive status, and sex-steroid production and release. Positive and negative feedback loops affect LH and FSH secretion, with testosterone from the testis inhibiting LH and FSH release, at least partially through inhibition of GnRH neurons (6, 11, 22, 31, 37).
Wild populations of Peromyscus leucopus are made up of individuals with widely different abilities to respond to photoperiod (17, 18, 27). Some members of any given population respond strongly to the short-day photoperiod typical of the winter months by exhibiting gonadal regression or significantly delayed reproductive development. Others, the nonresponsive (NR) phenotype, seem to be capable of reproducing at all times of the year. Most individuals express an intermediate response, showing some but not all of the traits of either extreme phenotype. This suggests that there is widespread, genetically based variability in the photoneuroendocrine pathway that regulates reproduction (17, 18, 33).
Previous experiments conducted on an unselected colony of P. leucopus (12) and an unselected colony of P. maniculatus (24, 25) suggest differences in GnRH neuronal activity between males of responsive and NR phenotype under an inhibitory photoperiod. The antibodies used in the immunocytochemistry (ICC) of these studies, GnRH-BDB (12) and LR1-GnRH (24, 25), are known to preferentially bind to an epitope of pro-GnRH. It has been suggested that responsive males, when placed in inhibitory photoperiods, sequester prohormone in the neurons that synthesize GnRH and fail to secrete mature peptide (13, 23, 25). This implies that phenotypic differences are due to differences in responses of GnRH neurons to photoperiod. In these studies, the majority of the immunoreactive (IR) GnRH neurons of P. maniculatus were found within seven regions in and around the hypothalamus. Significant differences in neuronal location and density between responsive and NR mice could be traced to two of those: the lateral hypothalamus and the preoptic areas (24, 25).
We tested for genetic variation in GnRH neuronal characteristics by comparing two lines of wild-derived P. leucopus developed by artificial selection on reproductive responsiveness to short photoperiod. We controlled for both line and photoperiod to test for differences in response to seasonal changes in photoperiod, as well as differences independent of photoperiod. In addition, we controlled for effects of acute differences in hormone level by castration with testosterone replacement.
First, we proposed that P. leucopus belonging to a line artificially selected for strong reproductive inhibition in short photoperiod [reproductively inhibited (RI)] would produce and secrete lower levels of mature GnRH than animals from a line selected to be reproductively active in short photoperiod (NR). Therefore, we predicted that RI mice would exhibit significantly lower total numbers of IR-GnRH cells after processing with an antibody specific to mature GnRH, because GnRH neurons with prohormone but little or no mature GnRH would remain unlabeled.
Second, we tested a hypothesis that arose from results obtained in the first experiment. We proposed that higher total numbers of GnRH neurons in NR mice would permit reproductive activity even if some suppression of GnRH secretion occurs in short days. On the basis of this hypothesis, we proposed that NR animals in a permissive, long-day photoperiod might secrete higher levels of mature GnRH than RI animals in long days. Thus NR mice in long days would exhibit significantly higher total numbers of IR-GnRH cells after processing with an antibody specific to mature GnRH than would RI mice in long days. Furthermore, we predicted that significant differences in the number of IR-GnRH cells between selected lines and photoperiod treatments would be specific to the anterior/lateral hypothalamus and the preoptic areas.
MATERIALS AND METHODS
Mice were obtained from a laboratory colony at the Population and Endocrinology Laboratory of The College of William and Mary. The wild founders of the population were captured at latitude 37° 16′N, in the vicinity of Williamsburg, VA (18). Wild-caught animals were paired in a long-day photoperiod [16 h light, 8 h dark (LD)], yielding a parental generation to serve as stock for selection experiments. To establish short-day RI and photoperiod-NR lines, animals from the parental generation were transferred to short-day photoperiod [8 h light, 16 h dark (SD)] at birth and raised in SD. Mice were examined at 70 days of age and assigned a reproductive index based on testis size or the size of the ovaries, uterine diameter, and presence or absence of visible corpora lutea (18). Females with ovaries ≤2 mm in length, lacking visible corpora lutea, and uterine diameter of ≤0.5 mm were classified as RI. Females with large ovaries, large visible follicles or corpora lutea, and uterine diameter >1 mm were classified as NR. Males with a testis index of <24 mm2 were classified as RI, and those with a testis index of >32 mm2 were classified as NR (17, 18). Responsive males and females were paired to produce a photoperiod-responsive line. Nonresponsive males and females were paired to produce a NR line (18).
Experiment 1 was performed on young adult P. leucopus from the F4–F6 generations of the RI line (selected to be inhibited in SD) and the NR line. Dams and their pups were transferred from LD photoperiod to SD within 3 days of birth of the litter. Male offspring were weaned at 21–23 days and singly housed until 70 ± 3 days of age. RI animals chosen for this study (n = 8) had a testis index of <24 mm2, which was typical of suppressed males. Of RI males produced in the colony at the time of the study, 95% fit the selection criteria. NR chosen for this study (n = 9) had a testis index of >40 mm2, which was typical of reproductive males in LD. Of the NR males produced at the time of this study, 50% fit the selection criteria.
Because GnRH secretion is sensitive to inhibition by steroid negative feedback and NR presumably differ from RI mice in sex steroid secretion in SD, we chose to control for acute individual differences in sex steroid production in this experiment. Mice chosen for this study were castrated and given a subcutaneous silastic implant (1.02-mm inner diameter, 5 mm in length) filled with evenly mixed testosterone (Sigma, St. Louis, MO) in silicone adhesive in a 1:3 ratio by mass. This treatment causes seminal vesicles to develop to a size typical of reproductively mature males (P. D. Heideman, unpublished data). The objective of this treatment was to ensure that both selected lines had equivalent amounts of testosterone, and at concentrations that, in terms of physiological response as assessed by seminal vesicle mass, were in the high physiological or pharmacological range. The wet mass of the paired testis was assessed and recorded immediately after surgical removal. Mice were allowed to acclimate to the hormone treatment for 2 wk before perfusion. Food (Agway Prolab Rat/Mouse/Hamster 3000, Syracuse, NY) and water were provided ad libitum. During the experimental period, the relative humidity of the animal rooms averaged 60 ± 20%, and room temperature was 23 ± 3°C.
Experiment 2 was performed on two groups of young adult P. leucopus from the F6–F9 generations of each of the two selected lines described above, resulting in four different treatment groups of animals. RI (n = 14) and NR raised in SD photoperiod (n = 13) were produced by transferring mothers and their pups from LD photoperiod to SD photoperiod within 3 days of birth of the litter. Male offspring were weaned at 21–23 days, singly housed until 70 ± 3 days of age when their testis index was assessed, and singly housed again until perfusion. RI (n = 14) and NR in LD (n = 14) were born, raised, weaned at 21–23 days, assessed at age 70 ± 3 days for their testis index, and singly housed until perfusion under LD photoperiod. Mice were chosen at random from each of the selected lines, but no sibling pairs were included. Thus mice whose testis index at age 70 ± 3 days did not meet the selection criteria for their line were still included in this experiment. Only intact mice were used in this experiment.
Perfusions and Sectioning
All perfusions were conducted in mice aged 70–100 days. Mice were weighed, euthanized with an overdose of isoflurane (Abbot Laboratories, North Chicago, IL), and allowed to enter respiratory arrest before perfusion. Mice were perfused through the left ventricle at ∼4 ml/min using a perfusion pump and bled via the right atrium. Perfusion of 5 ml of 0.1 M PBS at a pH of 7.4 was followed by perfusion of 50 ml of fresh, cold (5°C) 4% paraformaldehyde (Fisher Scientific, Fair Lawn, NJ) and saturated picric acid (Sigma) in PBS. Brains were removed and postfixed overnight at 4°C in 0.1 M PBS with 30% sucrose for cryoprotection. After perfusion, the mass of the paired seminal vesicles was assessed in animals from experiments 1 and 2. All brains were sliced within 4 days of perfusion. Frozen coronal sections (30 μm) were cut on a freezing sliding microtome and separated into four wells, each containing every fourth section. Wells were filled with brain antifreeze [37.5% sucrose, 37.5% ethylene glycol, and 10 g PVP-40 in 500 ml 0.02 M Tris-buffered saline (TBS)]. Brains were stored at −20°C until ICC.
Expression of mature GnRH immunoreactivity in the brain was detected using a single-labeled avidin-biotin-peroxidase-complex method. A total of three independent runs were carried out in experiment 1, and a total of seven independent ICC runs were carried out in experiment 2. Each independent run was balanced across treatments. Brain slices were rinsed five times for 6 min each in cold, 0.02 M TBS followed by incubation in cold (4°C) 1% sodium borohydride (Sigma) for 30 min. All subsequent treatments were conducted with gentle agitation at room temperature unless otherwise noted. Tissue was rinsed three times for 10 min each in cold TBS, followed by overnight incubation at room temperature with SMI-41 monoclonal antibody (Sternberger Monoclonals, Lutherville, MA) at a dilution of 1:20,000 in PBS with 0.25% lambda-carrageenan (Sigma), 1% bovine serum albumin (Sigma), and 0.3% Triton X-100 (Fisher Scientific) in TBS with 0.1% sodium azide at a pH of 7.8 (Fisher Scientific). SMI-41 is a mouse monoclonal IgG1 antibody reactive with the five amino acids adjacent to the C-terminus of the GnRH peptide and the amidation site. Therefore, only mature hormone is likely to be recognized by SMI-41 antiserum (38). Sections were given six 10-min rinses in TBS and incubated in biotinylated horse anti-mouse IgG at a dilution of 1:500 in TBS with 0.25% lambda carrageenan, 1% bovine serum albumin, and 0.3% Triton X-100 for 60 min at room temperature. After three more rinses in 0.02 M TBS, sections were incubated in avidin-biotin-peroxidase (Vector Laboratories Elite ABC-Peroxidase kit) in TBS for 60 min. Sections were given three rinses in TBS and placed in 1.5 ml of a solution of diaminobenzidine (0.2 mg/ml), NiSO4 (24 mg/ml), and diluted H2O2 (4.8 μl/ml of freshly prepared solution from 35 μl of 30% H2O2 in 965 μl of distilled H2O) (Sigma), in 0.02 M TBS. The color reaction was allowed to proceed for ∼12 min. After five 10-min rinses in 0.02 M TBS, sections were mounted on gelatin-coated slides and air dried, dehydrated in xylene, and coverslipped with Permount (Fisher Scientific).
The location and number of mature GnRH-secreting neurons in experiment 1 were assessed in independent counts by MA and by PDH, carried out blind with respect to treatment using an Olympus CH2 compound light microscope. The numbers and locations of neurons identified in independent counts by MA and PDH in experiment 1 were very similar, the overall results were qualitatively identical, and both data sets gave identical statistical results; the data presented here are those measured by MA. The location and number of mature GnRH-secreting neurons in experiment 2 were assessed in a single count by MA, carried out blind with respect to treatment using the same microscope. All brain structures and nuclei in this paper are referred to using abbreviations and nomenclature consistent with those given by Paxinos and Watson (32).
Brain areas were first estimated relative to major landmarks using a stereotaxic coordinate atlas for the rat brain (32). The regions where IR-GnRH cells were scored using the brain atlas for the rat were later compared with those identified in a stereotaxic coordinate atlas for the deermouse, P. maniculatus (10). Because the deer mouse atlas lacked detail in much of the hypothalamus, the rat brain atlas was used to estimate boundaries of many of the brain areas, and terminology from this latter source is used here. P. leucopus have much larger eyes and optic nerves than the similarly sized laboratory mouse, and in our judgment, the hypothalamic structures and landmarks are more similar to those of the rat than those of the laboratory mouse. However, because brain areas in the rat may differ from brain areas in P. leucopus, areas to which we assigned the same name may not be homologous.
IR-GnRH cells in experiment 1 were found in the region limited rostrally by the horizontal limb of the diagonal band of Broca and caudally by the arcuate hypothalamic nucleus and medial tuberal nucleus. Because coronal sections in animals from experiment 2 were collected farther anterior and posterior than in animals from experiment 1, IR-GnRH cells in experiment 2 were found in the region limited rostrally by the tenia tecta and infralimbic cortex and caudally by the lateral and medial mammillary nuclei.
Data were analyzed using Statview SE + Graphics software and SuperANOVA software (Abacus Concepts, Berkeley, CA) running on a Macintosh computer. In experiment 1, we compared mean IR-GnRH neuron numbers between selected lines using Student's t-tests, with P < 0.05 as the level of significance. Comparisons between selected lines were carried out for the total number of neurons, for each individual brain structure where IR cells were scored, and for the larger, arbitrarily defined brain regions used for analysis of IR-cell distribution (see Table 1).
In experiment 2, we compared mean IR-GnRH neuron numbers using a two-way ANOVA with P < 0.05 as the level of significance. Comparisons across selected lines and photoperiod treatments were carried out for the total complement of neurons, for neuron numbers in each individual brain structure where IR-cells were scored, and for the larger brain regions used for analysis of IR-cell distribution. Bonferroni corrections were carried out on the nine individual brain structures having an average of four or more IR neurons per brain. Brain structures with fewer than four IR-neurons were considered likely to be of relatively low relevance and were excluded from analysis as single units. After Bonferroni correction, the criterion for statistical significance for these brain areas was P < 0.006; because Bonferroni-corrected criteria for significance are α/n, where α = 0.05 and n = the number of brain structures having biologically relevant complements of IR neurons. Furthermore, because the size of the GnRH neuronal complement of RI and NR mice may be related to life history traits in addition to selection line and photoperiod, we analyzed our results from experiment 2 using analysis of covariance, with body mass, paired testis mass, or paired seminal vesicle mass as cofactors.
Because the mean number of IR-GnRH neurons might vary across ICC runs, we decided to account for any potential differences in the consistency of the procedure. Results from experiment 2 were further analyzed using a three-way ANOVA, with ICC run as a factor along with selection line and photoperiod treatment. A one-way ANOVA comparing the seven ICC runs was also carried out to identify any significant differences that might be due to inconsistencies in our methods. Neither test result approached significance (P > 0.509).
We detected significantly (t = 2.322; P = 0.035) lower total numbers of immunoreactive GnRH neurons in the RI line than in the NR line (Figs. 1A and 2). In addition, we found significant differences in paired testis mass (t = 6.84; P < 0.001; Fig. 3A) as well as significant differences in paired seminal vesicle mass (t = 6.13; P < 0.001; Fig. 3B) between selected lines. We did not detect statistically significant differences in body mass between selected lines (t = 1.75; P = 0.101; Fig. 3C).
To understand whether overall differences could be attributed to GnRH neurons in particular areas of the brain, individual structures and nuclei were combined into three regions for further analysis (Table 1). The more anterior region (Anterior in Table 1) includes all brain structures posterior to the fusion of the two halves of the corpus callosum but anterior to the preoptic areas, such as the vertical limb and most of the horizontal limb of the diagonal band of Broca, the medial septal nucleus, and most of the lateral septal nucleus [plates 12 through 17; Paxinos and Watson (32)]. We detected a trend (t = 1.757; P = 0.099) for a difference between lines in the anterior region, suggesting that NR may contain a larger IR-neuron complement than RI in these areas, although results did not reach significance (Fig. 1B).
The second region of brain structures (Preoptic in Table 1) is limited rostrally by the medial preoptic area and includes all the preoptic areas, the most rostrally located parts of the anterior hypothalamic area, and the parts of the lateral hypothalamus that are anterior to the tuber cinereum (plates 18 through 24; Ref. 32). We detected a trend (t = 1.983; P = 0.066) for a difference between lines in the preoptic region, suggesting that NR may contain a larger IR-neuron complement than RI in these areas, but results did not reach significance (Fig. 1C).
The more posterior region (Posterior in Table 1) includes all IR cells that appear rostrally along with the tuber cinereum, and extends through the median eminence, most of the arcuate nucleus, and the entire ventromedial hypothalamus (VMH; plates 25 through 32; Ref. 32). We did not find significant differences or trends suggesting potential differences in IR-GnRH neuron number between selected lines (t = 0.511; P = 0.617) in the posterior region (Fig. 1D).
Significant differences in IR cell numbers between selected lines (t = 2.309; P = 0.036) could be traced to the combination of the two most anterior groups (Anterior + Preoptic, Table 1). Differences in IR-GnRH neuron numbers were not significant when major groups were combined in any other way (Table 1). No individual brain nucleus or structure showed statistically significant differences in IR-GnRH cell numbers between selected lines.
We detected significantly (F = 37.339; P < 0.001) lower total numbers of IR-GnRH neurons in the RI line than in the NR line. We did not detect significant differences between mice raised in LD photoperiod and mice raised in SD photoperiod (F = 0.650; P = 0.424; Fig. 4A). There were significant differences in paired testis mass between selected lines (F = 56.94; P < 0.001) and between photoperiod treatments (F = 72.06; P < 0.001; Fig. 5A). In addition, we found significant differences in paired seminal vesicle mass between selected lines (F = 32.42; P < 0.001) as well as between photoperiod treatments (F = 73.84; P < 0.001; Fig. 5B). We did not detect statistically significant differences in body mass between selected lines (F = 2.66; P = 0.109) or between photoperiods (F = 2.47; P = 0.122; Fig. 5C).
Because of potential relationships between body mass, testis mass, or seminal vesicle mass and the size of the GnRH neuronal complement, we analyzed our initial results using analysis of covariance. The number of IR-GnRH neurons was not related to paired testis mass (F = 0.20; P = 0.660), paired seminal vesicle mass (F = 1.05; P = 0.310), or body mass (F = 0.48; P = 0.492). In each analysis of covariance, we found statistically significant differences between selected lines (P < 0.001 in all), but there were no differences due to photoperiod treatment (P ≥ 0.20 in all) or interactions between line and photoperiod (P > 0.50 in all). Thus the effect of artificial selection on the mean GnRH complement of RI and NR appears to be independent of differences in paired testis mass, seminal vesicle mass, and body mass.
IR-GnRH cells were scored in more than 100 different brain regions and nuclei (Table 2). Statistically significant differences (P < 0.05) between selected lines were found in 15 of these brain structures, whereas only one showed statistically significant differences between photoperiods (Table 2). Given the large number of independent statistical tests performed simultaneously and a significance level of P < 0.05, approximately five of the significant differences found in individual brain regions could result from chance alone. To detect potentially spurious positives and provide greater statistical confidence in the results, Bonferroni corrections to a lower P value (<0.006) were carried out on the nine brain structures containing an average of at least four IR cells per brain (Table 3).
Of the nine brain regions averaging at least four IR-GnRH neurons, five had significant P values after Bonferroni corrections: horizontal limb of the diagonal band of Broca, vertical limb of the diagonal band of Broca, medial and anteroventral preoptic areas, lateral preoptic area, and olfactory tubercle and islands of Calleja. No single brain nucleus or small region showed statistically significant differences in IR-GnRH cell numbers between photoperiod treatments (Table 3).
To understand whether overall differences could be attributed to GnRH neurons in major spatial regions of the brain, individual structures and nuclei were combined into five regions representing spatial locations for further analysis. The two additional regions in this experiment, relative to experiment 1, include IR cells located in brain nuclei more rostral or more caudal, respectively, than those included in the three groups defined in experiment 1.
The most anterior region (rostral; Fig. 4B) includes all brain structures anterior to the fusion of the two halves of the corpus callosum (plates 8 through 11; Ref. 32). We detected significant differences across selected lines (F = 8.489; P = 0.006) but not between photoperiods (F = 0.240; P = 0.627) in this region of brain structures.
The second region (anterior; Fig. 4C) includes all brain structures posterior to the fusion of the two halves of the corpus callosum but anterior to the preoptic areas, such as the vertical limb and most of the horizontal limb of the diagonal band of Broca, the medial septal nucleus, and most of the lateral septal nucleus (plates 12 through 17; Ref. 32). Significant differences across selected lines (F = 21.507; P < 0.001) but not between photoperiods (F < 0.01; P = 0.987) were found in the anterior region.
The third region of brain structures (preoptic; Fig. 4D) is limited rostrally by the medial preoptic area and includes all the preoptic areas, the most rostrally located parts of the anterior hypothalamic area, and the parts of the lateral hypothalamus that are anterior to the tuber cinereum (plates 18 through 24; Ref. 32). Significant differences across selected lines (F = 19.566; P < 0.001) but not between photoperiods (F = 0.723; P = 0.399) were found in the preoptic region.
The fourth region (posterior; Fig. 4E) includes all IR cells that appear rostrally along with the tuber cinereum and extend through the median eminence, most of the arcuate nucleus, as well as the entire VMH (plates 25 through 32; Ref. 32). We did not find significant differences between selected lines (F = 3.591; P = 0.064) or photoperiod treatments (F = 1.004; P = 0.321) in the posterior region of brain structures.
The most posterior region (caudal; Fig. 4F) includes the more caudal portions of the lateral hypothalamus and parts of the mammillary nuclei but does not contain any portions of the ventromedial hypothalamus. We did not find significant differences across selected lines (F = 2.996; P = 0.091) or photoperiod treatments (F = 0.071; P = 0.792) in the caudal region (plates 33 through 38; Ref. 32).
Our results indicate that selection on a life history trait may alter a specific neuronal trait in a natural population. This is the first evidence for heritable and selectable variation in a specific neuronal trait in a wild population and the first to link such variation to phenotypes that would be expected to be important for fitness in a natural population. Genetic neuronal variation has been reported in laboratory mice that have been under relaxed selection or inbred over many generations (8, 21). However, experiments on domesticated rodents provide no insight into levels of natural variation in wild populations, making studies on natural populations such as this one critical.
Previous experiments by Heideman et al. (18) on a natural population of P. leucopus indicate that the gonadal response to photoperiod is a heritable, additive genetic trait. One generation of artificial selection resulted in a testis index heritability mean value of 0.74 for male offspring-on-father comparisons. Additive genetic variation is relevant here because it is the only variation on which natural selection can act. Because testis size and reproductive status are partially under the control of GnRH (9, 36) and artificial selection for differential testis size in SD resulted in line-specific changes in GnRH neurons, we suggest that the size or function of the GnRH neuronal complement is heritable in the population studied here.
Our data are consistent with the hypothesis that within-species, genetic variation in photoperiod responsiveness is due in part to variation in the number and function of GnRH neurons. P. leucopus selected to retain year-round reproductive competence possess a larger complement of mature GnRH-releasing cells in both SD and LD than mice selected to experience RI. Our results from experiment 1 are consistent with previous findings obtained in an unselected but phenotypically diverse colony of P. maniculatus tested in SD only (23–25). In addition, our results are consistent with and extend the conclusions from previous experiments on unselected female P. leucopus from a different locality, where variation in photoperiod responsiveness was found to correlate with variation in GnRH cell numbers and location in SD (12).
One major difference from previous immunocytochemical studies comparing GnRH neurons of photoperiod-responsive and photoperiod-NR Peromyscus spp. allows new conclusions to be reached. Our animals come from selected lines containing mostly SD RI or mostly photoperiod-NR mice. Therefore, because either artificial selection on a single trait (gonadal development in SD) or genetic drift led to changes in neuronal density, and because photoperiod responsiveness is a heritable, additive genetic trait (18), it is likely that the differences in IR-GnRH neurons observed between selected lines are genetic in origin. In addition, these genetically distinct lines of mice allow us to test for selected line-specific differences in the photoneuroendocrine pathway in excitatory or inhibitory photoperiods, whereas many previous experiments on the plasticity of the GnRH neuronal system (12, 13, 24, 25) required SD conditions to unmask photoperiod responsiveness and nonresponsiveness.
Differences between selected lines were not distributed evenly across brain regions. One region of ventromedial brain areas did not differ significantly between selected lines in either experiment 1 or 2. In contrast, highly significant differences in IR-GnRH cell abundance were found in the preoptic area and anterior hypothalamus (AH) in experiment 2, and differences in these areas approached significance in experiment 1, despite the lower sample sizes. This suggests that preoptic and anterior hypothalamic GnRH neurons play a role in seasonal regulation of reproduction and contribute to variation in reproductive responsiveness to photoperiod, whereas more posterior structures of the hypothalamus are not directly involved in the mediation of this process. Brain areas that have been implicated as likely sites of action of SD and melatonin on the reproductive axis are variable among mammals (34), but the AH and medial preoptic area have been implicated previously in P. leucopus. Glass and Lynch (14) and Glass and Knotts (13) demonstrated that melatonin-containing pellets implanted in the medial preoptic area, AH, and supra- and retrochiasmatic areas elicit gonadal regression during LD to the same extent as SD exposure. Because melatonin from the pellets was found to diffuse to a distance of only 0.2 mm, and similar doses of melatonin delivered subcutaneously could not trigger testicular regression, it is likely that these neural sites mediate the antigonadal action of melatonin (13, 14).
Glass and Knotts (13) also showed that exogenous melatonin-induced testicular regression during LD is accompanied by changes in the GnRH neuronal system of P. leucopus. Melatonin-containing pellets stereotaxically implanted in the AH significantly increased the optical density of IR-GnRH perykarya and the total area covered by IR-GnRH fibers in the median eminence. The change in intensity and density of the staining was independent of the distance between IR-cells and the melatonin implant. Glass and Knotts (13) used GnRH-BDB for immunocytochemical processing, an antibody reactive to pro-GnRH. The fact that melatonin levels mimicking the effects of SD lead to enhanced immunoreactivity of GnRH perykarya in the AH and medial preoptic area (POA) suggests that this treatment increases the intracellular content of the prohormone and that the antigonadal action of melatonin involves suppression of GnRH release rather than its synthesis (13).
Previous experiments by Heideman et al. (19) demonstrate that the RI and NR selected lines also differ significantly in their ability to bind melatonin in SD. NR mice showed higher 2-[125I]iodomelatonin binding in the medial preoptic area and bed nucleus of the stria terminalis than RI mice, but no difference was apparent in the dorsomedial nucleus of the hypothalamus. This suggests that differences in the density or affinity of the melatonin receptors that ultimately affect mature GnRH secretion are likely to play a role in variability of photoperiod responsiveness (19).
The dorsomedial nucleus of the VMH, along with the ventromedial area of the dorsomedial hypothalamus, constitutes the mediobasal hypothalamus. The mediobasal hypothalamus has been identified as the neural site where melatonin mediates seasonal responses to photoperiod in the Syrian hamster (Mesocricetus auratus) (29, 30). Neurons in the VMH ultimately affect GnRH-secreting cells, possibly by responding to gonadal steroids and influencing sensitivity to steroid negative feedback (29). Lesions to the VMH of male Syrian hamsters have been shown to hasten testicular recrudescence during SD, and it has been suggested that the VMH-dorsomedial hypothalamus complex is critical for the attainment and maintenance of gonadal regression in that species (1). Interestingly, we found no differences in IR-GnRH neuron numbers between RI and NR in these regions in either experiment 1 or 2. Relative to our study, this result might be explained with two hypotheses. First, that interneurons expressing melatonin receptors in the VMH help mediate the effects of melatonin on GnRH neurons elsewhere. Alternatively, it is possible that testicular regression in P. leucopus is mediated by the AH and medial POA without input from the VMH or other posterior hypothalamic nuclei. To our knowledge, there are no published lesion studies of the anterior hypothalamic and preoptic areas of P. leucopus.
The differences we observed in IR-GnRH neuron complements between selected lines appear to be due to differences in responsiveness to photoperiod that are independent of short-term sex-steroid production, as the differences were apparent even when sex-steroid levels were controlled by castration and delivery of a constant (and thus nonphysiological) dose of testosterone in experiment 1. Because all animals in experiment 1 experienced the same steroid environment for 2 wk before perfusion, it is unlikely that acute or short-term chronic effects of steroid negative feedback are solely responsible for any differences in the observed numbers of IR-GnRH neurons between RI and NR mice. Our method for controlling circulating levels of sex steroids would not, however, control for selected line-specific differences in sensitivity to steroid-negative feedback. In other words, the observed differences might be due to variation in the sensitivity of the reproductive axis to steroid-negative feedback, producing differences between lines even when sex steroid levels are held constant. Support for this hypothesis comes from measurements of the paired seminal vesicle mass of mice from experiment 1 (Fig. 3B). After exposure to the same steroid dose, NR mice had larger seminal vesicles than RI mice. Similar differences in seminal vesicle mass were observed in intact NR and RI in LD animals from experiment 2 (Fig. 5B).
In this study, every fourth section was immunocytochemically stained, suggesting that the total number of IR-GnRH neurons, based on results from experiment 1, average ∼400 in RI and 580 in NR. Similar total GnRH neuron complements can be projected from our results in experiment 2, although slightly higher averages of ∼425 IR cells in RI and ∼700 IR cells in NR were scored. The maximum reported diameter of most GnRH neurons is 10–20 μm (36), which is consistent with our observation of IR-GnRH perykarya in P. leucopus. Our 30-μm coronal sections might have included counts of some cell bodies that were partially within adjacent sections. If so, then the total reported above may slightly overestimate the total complement of IR-GnRH neurons in this population. These numbers are in the range of the total expected for rodents of this size, and suggest that a large proportion of total GnRH neurons were immunoreactive in this study. Total GnRH cell numbers range from 300 to 400 in Djungarian hamsters (41), 650 to 750 in Syrian hamsters (20), and 500 to 1,300 in the rat (35, 40). Thus our results are consistent with the hypothesis that NR have higher total numbers of GnRH neurons, which results in a higher number of cells producing mature hormone in NR than in RI at all times, and permits reproductive activity even if some suppression of GnRH secretion occurs in SD. Variation between selected lines in the number of GnRH neurons producing mature GnRH may have functional significance and may be at least partially responsible for differences in photoperiod responsiveness among P. leucopus.
Previous results by Korytko et al. (24) indicate that the related species P. maniculatus exhibits a shift in the number of IR-GnRH neurons between animals in SD and animals in LD. That we were unable to detect within-lines differences in the density and location of IR-GnRH cells between photoperiods is remarkable in the context of the large differences observed in paired testis mass and paired seminal vesicle mass of RI and NR in LD and SD. Our results are consistent with at least two hypotheses. First, it is possible that the GnRH neuronal network mediates testicular regression and the temporal loss of reproductive capacity in RI mice under inhibitory photoperiods by means that are independent of the number of IR-GnRH cells synthesizing mature hormone. Studies on RI Siberian hamsters (4) indicate that mature GnRH peptide levels in brain tissue may increase without a corresponding shift in IR-neuron numbers after exposure to LD photoperiod. Therefore, the observed increase in protein synthesis appears to occur within a stable population of IR-GnRH-producing cells (4). It is also possible that our ICC labeled even minute amounts of mature neuropeptide. If all GnRH neurons contain at least some mature hormone at any given time, we might have stained nearly the entire neuronal complement in all animals in both SD and LD. We observed IR-neurons of various staining intensities, including some so faint as to be almost undetectable. Faint staining might be due to nonspecific binding or weak cross reactivity, or, alternatively, our methods may have been semi-quantitative in identifying only a subset of the total GnRH neuronal complement.
An individual's reproductive status during SD is related to its endocrine state. Photoresponsive P. maniculatus are known to exhibit lower concentrations of testosterone and LH in SD than in LD, whereas nonphotoresponsive animals in SD maintain circulating levels of both hormones at levels identical to those found during LD (23). In our study, the significantly higher numbers of mature GnRH neurons observed in NR raised in SD photoperiod suggest that these individuals are able to secrete a sufficient amount of neuropeptide, even under inhibitory photoperiods, to maintain steady gonadotropin pulsing and release into the bloodstream. The elevated levels of LH and FSH resulting from increased GnRH secretion would, in turn, mediate sufficient testosterone synthesis and spermatogenesis to permit the year-round mating behaviors and fertility that define photoperiod nonresponsiveness (33).
In summary, mammals can be highly variable within species in life history traits related to reproduction. The photoneuroendocrine pathway studied here is one of the few for which the physiological traits that produce life history variation are being identified systematically and tested for phenotypic and genetic variation within single populations. The implication from the still-limited data is that variation in this and possibly other pathways can be high and is likely to be due to multiple neuroendocrine causes. More studies making these connections are necessary to link variation in life history traits to neuroendocrine variation and ultimately to the genes that contribute to life history variation in natural populations.
One of the two major competing hypotheses in evolutionary physiology is that selection of complex physiological pathways can eliminate variation to provide a pathway that functions in some optimal fashion (26, 39). The alternative hypothesis is that selection on different functions of the multiple genes controlling such pathways may be too weak or too variable to produce a single optimal result, resulting in sustained high levels of variation (2, 26). In combination with the results of previous studies on this natural population (18, 19, 28), the presence of variation in IR-GnRH neurons is consistent with the hypothesis that complex physiological pathways may not be able to reach an optimum function, even if that optimal possibility exists. If these results extend to other vertebrates, then genetic variation in neuronal traits may also be common in humans and other species. Furthermore, genetic variation in numbers of specific types of neurons might contribute to disease risk associated with neuronal number or activity as well as to variation in life history phenotype and physiological function. High levels of natural neuroendocrine variation may enable rapid microevolutionary change of the brain and might have contributed to macroevolutionary trends in vertebrate brain evolution.
This research was supported by a grant from the National Science Foundation (no. 9875866) to P. D. Heideman and by the College of William and Mary.
We thank Dr. Emilie F. Rissman for advice on the experimental design and support for the experiments, Michelle Rightler for assistance with animal breeding and care, Dr. Diane C. Shakes for assistance with image capture and processing, Dr. Cheryl Jenkins for assistance with manuscript preparation, and Jessica Tate for assistance with perfusions and brain sectioning.
Present address of S. D. Sullivan: Box 800578, Health Sciences Center, University of Virginia, Charlottesville, VA 22908.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2005 the American Physiological Society