|Home | About | Journals | Submit | Contact Us | Français|
Sexual development is inhibited in Siberian hamsters (Phodopus sungorus) in short days (SD), and a small uterus is an obvious indicator of photo-inhibition. The small uterus in SD is presumably due to the delayed onset of estrous cycles. However, in an earlier study, the investigators reported that serum estradiol (E2) concentration was significantly higher in young females raised in SD than in long days (LD), with the highest concentrations measured in SD at 4 wk of age. These seemingly contradictory findings were investigated in the present study. First, uterine mass and body mass were measured in SD- and LD-reared hamsters from 1 to 12 wk of age. Uterine mass was significantly greater in LD than in SD by 3 wk of age and onward. Thereafter, our investigation focused on 4-wk-old hamsters. Serum E2 concentrations in LD and in SD were not significantly different and there were no significant LD-SD differences in uterine estrogen receptors (ER), as measured by immunohistochemistry and quantitative real-time RT-PCR. Therefore, alternative explanations for the photoperiodic difference in uterine size in young Siberian hamsters are considered.
Age at puberty can vary significantly in short-lived, temperate zone rodents depending on their season of birth. Females born before or near the summer solstice, i.e., when day length is either increasing or near its annual maximum, mature rapidly and produce litters during the year of their birth. Conversely, females born well after the summer solstice are very likely to delay puberty and their first breeding until the following spring (Negus et al. 1986, Butler et al. 2007). One of the hallmarks of delayed puberty in females is the stunted growth of the uterus. In Siberian hamsters, Phodopus sungorus, Ebling (1994) measured uterine mass at 5 and 10 wk of age in females held in either short days (SD) or long days (LD). Whereas uterine mass increased more than two-fold in long day (LD) females, no change in uterine mass was detected in SD females. At both 5 and 10 wk of age, Ebling (1994) determined uterine mass in SD hamsters was significantly less than in LD females. Subsequent studies also documented significant uterine size discrepancies in SD and LD hamsters at 10 and 13 wk of age (Place et al. 2004, Timonin et al. 2006).
Smaller uterine size in SD hamsters has been assumed to result from the indirect effects of the SD melatonin (MEL) signal, with MEL being secreted into the circulation from the pineal gland at night at higher concentrations and for a longer duration than in hamsters held in LD (Darrow and Goldman 1985, Hoffmann et al. 1986). Stunted growth of the uterus has been presumed to result from the suppression of gonadotropins in SD (Dodge and Badura 2002) and the associated reduction in the concentrations of estrogens (e.g., estradiol). To our knowledge, this hypothesis has never been formally tested and this conjecture has been called into question by a study that reported significantly higher serum E2 concentrations in SD than in LD hamsters between 4 and 11.4 wk of age (van den Hurk et al. 2002). In that study, E2 in SD females was as much as eight-fold higher than in LD hamsters (8 wk of age), and significantly higher concentrations were noted in SD at each of the ages studied (4, 8, and 11.4 wk). The highest absolute concentration (~2400 pmol/L or ~ 654 pg/mL) was measured in SD females at 4 wk of age. Given the growth promoting effects of E2 on the uterus (Scotti et al. 2007), we found the report of higher E2 in SD hamsters (van den Hurk et al. 2002) to be perplexing. Unfortunately, van den Hurk et al. (2002) did not report uterine mass data, thus the present study is the first to investigate the apparent mismatch between serum E2 concentrations and uterine growth in P. sungorus.
To that end, we collected uterine and body mass data at regular intervals from 1 to 12 wk of age, to more clearly elucidate the differences when females develop in LD or in SD. Earlier investigations that reported the effects of photoperiod on uterine mass in P. sungorus evaluated females at only one or two ages (Ebling 1994, Place et al. 2004, Timonin et al. 2006), and the youngest animals studied were 5 wk of age (Ebling 1994). When we determined the photoperiod-induced differences in uterine size were evident by 3 wk of age, we then focused our investigation on LD and SD hamsters at 4 wk of age for the following reasons: 1. The highest E2 concentration reported by van den Hurk et al. (2002) was in SD females at this age and the LD-SD difference in serum E2 concentration at 4 wk was substantial (approximately 5-fold), and 2. Neither LD nor SD females have matured by this age, as indicated by the lack of vaginal patency (Place et al. 2004) and the absence of any signs that ovulations had occurred (van den Hurk et al. 2002). Our investigations of 4-wk-old hamsters avoided potentially confounding variables, such as stage of estrous cycle in LD females or LD-SD differences in reproductive state.
Using an assay previously validated for the measurement of E2 in Siberian hamsters (Scotti et al. 2007), we did not to replicate the findings of van den Hurk et al. (2002), i.e., that serum E2 concentration is significantly higher in SD than in LD hamsters at 4 wk of age. However, because we found serum E2 concentrations to be nearly identical in 4-wk-old SD and LD females, we looked to the uterus for possible mechanisms to explain the photoperiod-induced difference in uterine size. To test the hypothesis that estrogen receptor (ER) abundance differs in uteri from SD and LD hamsters, we assessed ERs by immunohistochemistry for ERα and quantified mRNA levels by real-time RT-PCR for ERα (Esr1) and ERβ (Esr2).
Siberian hamsters from our colony (14 h of light per day, 14L) were transferred to LD (16L) or SD (10L) as breeding pairs to generate experimental LD and SD females. The time of lights-off was synchronized for all animals to 1700 Eastern Standard Time (EST). Animals were originally derived from wild-bred stock obtained from Dr. K. Wynne-Edwards, Queen’s University. Experimental females were weaned on postnatal day 18, placed in polypropylene cages (2 to 4 siblings/cage), and maintained in the photoperiod in which they were born. Food (Teklad 8626, Madison, WI, USA) and water were available ad libitum. Ambient temperature and relative humidity were held constant at 21 ± 5 °C and 50 ± 10 %, respectively.
Different sets of animals were used in Experiments 1 and 2, as well as for different determinations within Experiment 2. All experiments were carried out in accordance with the European Commission legislation on the protection of animals used for experiments (EC Directive 86/609/EEC).
Measurements of uterine and body mass were made on six to seven SD and LD females at each of the following ages: 1, 2, 3, 4, 6, 8, 10, and 12 wk. Sibling females were divided across age cohorts, thus siblings were never in the same age group. Body mass was measured just prior to death for each of the predetermined ages. Animals at 1 wk of age were killed by decapitation, whereas older animals were given an overdose of pentobarital sodium. Uteri were removed, dissected free of fat and connective tissue, and weighed on an analytical balance.
Because female hamsters weigh less than 20 g at 4 wk of age, serum samples from entire SD and LD litters had to be pooled (females only) to assure adequate sample volumes. Five SD and six LD litters (three to five female pups per litter) were given an intraperitoneal overdose of pentobarbital sodium and exsanguinated by retro-orbital bleed. Blood was clotted at room temperature for 1 h and centrifuged at 1000 g for 20 min in 4 °C. Drawn off serum was pooled by litter, frozen and maintained at −80 °C until assayed for E2. Serum samples were analyzed in duplicate by a radioimmunoassay (RIA) previously validated in Siberian hamsters (Scotti et al. 2007). Briefly, the RIA used was a solid-phase 125I kit (Diagnostic Products Corporation, [now Siemens], Los Angeles, CA, USA), modified by addition of a pre-assay ether extraction. Following the addition of 3H-estradiol (50 µL, ~1900 cpm) to determine extraction efficiencies, 300 µL serum samples were extracted in diethyl ether, dried under N2, and reconstituted in 335 µL of assay buffer. A 100 µL aliquot was counted on a scintillation counter to calculate recoveries for 3H-estradiol and separate 100 µL aliquots were added to Coat-A-Count® tubes in duplicate for RIA. The cross-reactivity of the highly specific antibody to other estrogens is less than 2%, save for estrone (10%). A standard curve (15.6–1000 pg/mL) was made by serially diluting an E2 stock solution in assay buffer. Tubes were incubated at room temperature (23 °C) for 3 h following the addition of 125I-estradiol tracer. Tubes were aspirated of their contents then counted in a gamma counter. Volume and percent extraction recovery specific to each sample were used to calculate concentrations interpolated from the standard curve. The E2 assays met all quality assurance criteria and internal controls were run at the beginning, middle, and end of each assay. The intra- and inter-assay coefficients of variation were <10%, and the minimum detectable limit (MDL) of the assay was 18 pg/mL. Samples with an estradiol concentration below the level of detection were assigned this value for statistics and graphing.
As a biological validation to our E2 assay, we also measured serum E2 concentration in older hamsters (16–20 wk age), when LD females (n = 12) are expected to be cycling and should have higher E2 concentrations than immature SD females (n = 5). As expected, serum E2 concentration was highly variable in LD females (18.0 – 125.2 pg/mL), but uniformly low in photo-inhibited SD females (all below the MDL, 18.0 pg/mL). We have found vaginal cytology does not reliably track the estrous cycle in P. sungorus, and no attempt was made to monitor the estrous cycle of LD females. Thus, the sampling likely represents LD females at different stages of the estrous cycle and SD females that were uniformly anestrus.
For analysis of SHBG, blood samples were collected from a separate cohort of LD and SD female hamsters at 4 wk of age (n = 12 and 8, respectively). Competition assays were used to determine the affinities of hamster SHBG for E2 and testosterone relative to 5α-dihydrotestosterone (DHT). The serum concentration of SHBG was determined by the steroid binding capacity assay, employing 3H-DHT as the labeled ligand and dextran-coated charcoal as the separation agent (Hammond and Lähteenmäki 1983).
A mid-portion of the uterine horn and an ovary from six SD and six LD females at 4 wk of age were embedded separately in paraffin, sectioned at 6 µm, and mounted on glass slides for standard histology (uterus and ovary) or on Superfrost® Excell slides (Thermo Fisher Scientific Inc., Waltham, MA, USA) for immunohistochemistry (uterus). Sections for standard histology were stained with hematoxylin and eosin (H&E).
For immunohistochemistry, sections from LD and SD uteri were alternately placed on each slide to control for potential staining variability between slides. Adjacent sections were mounted on separate slides for negative controls. After dewaxing and rehydration in a series of ethanols, slides were submerged in Antigen Unmasking Solution (1:100 v/v in H2O; H-3300; Vector, Burlingame, CA, USA) and microwaved on hi-power for two 15-min bouts. Endogenous peroxides were quenched in hydrogen peroxide (0.5 % in methanol) for 20 min. Sections were then incubated overnight at 4 °C in CleanVision™ Blocking solution (ImmunoVision Technologies, Norwell, MA, USA) plus 10 % goat serum to block nonspecific binding sites. Monoclonal mouse anti-human ERα antibody (M7047, Dako, Carpinteria, CA, USA) was diluted 1:40 in dilution buffer and incubated with sections for 48 h at room temperature. Sections were incubated with the secondary antibody, biotinylated goat anti-mouse IgG (1:200 in dilution buffer; BA-9200; Vector, Burlingame, CA, USA), for 30 min. Negative controls excluded the primary or secondary antibody. Immunoreactivities were visualized by incubating sections with Vectastain Elite ABC Solution (Vector) for 30 min and developing with DAB Peroxidase Substrate Solution (Vector) following the manufacturer's instructions. Sections were counterstained with hematoxylin.
For uterine smooth muscle α-actin, the immunohistochemistry protocol followed that for ERα, except for the primary antibody used, which was a monoclonal mouse anti-α smooth muscle actin (A2547 Sigma, St. Louis, MO, USA), at a dilution of 1:2000 and an incubation of 24 h.
We made several attempts to detect ERβ by immunohistochemistry, using different primary antibodies and dilutions, without success. Seeing as Jefferson et al. (2000) did not detect ERβ in the mouse uterus on postnatal day 26, nor transcripts on postnatal days 5, 12, 19 and 26, we did not pursue this further. However, Jefferson et al. (2000) did detect ERβ mRNA (Esr2) in postnatal day 1 mouse uteri, thus we measured Esr2 levels in Siberian hamster uteri at 4 wk of age.
An additional six to 10 females were generated in each photoperiod as above to harvest ovaries and uteri from 4-wk-old individuals to quantify uterine ERα (Esr1) and ERβ (Esr2) and ovarian P450aromatase (Cyp19a1) mRNA levels. Following pentobarbital sodium overdose and exsanguination, ovaries and uteri were removed, dissected free of surrounding adipose and connective tissue, flash-frozen on dry ice, and stored separately at −80 °C until RNA extraction. Frozen tissues were homogenized with a Polytron, and total RNA was extracted with Trizol (Invitrogen, Carlsbad, CA, USA) in accordance to the manufacturer’s instructions. RNA was reverse transcribed using Superscript III First-Strand Synthesis System (Invitrogen) according to the manufacturer’s instructions. RNA samples used for real-time RT-PCR were treated with DNase I (Amp. Grade; Invitrogen) prior to reverse transcription. Quantitative real-time PCR was run under standard conditions using Power SYBR® Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) on an Applied Biosystems 7500 Real-Time PCR System (Applied Biosystems) according to the manufacturer’s instructions. The annealing temperature for all genes was 60 °C.
The specificities of the RT-PCR products were confirmed by melting curve analyses. Esr1, Esr2, and Cyp19a1 mRNA levels are expressed relative to the levels of Ribosomal protein L13a (Rpl13a) mRNA in the same samples. Schroder et al. (2009) noted that Rpl13a was expressed at fairly constant levels in uteri under varying conditions, and our pilot data showed comparable results for Rpl13a in ovaries. For each target gene and the reference gene, samples were run in triplicate. For quantification of each gene, we used relative standard curves (Cikos et al.2007). Briefly, standard cDNAs were prepared from other hamster ovaries and uteri, and were amplified along with sample cDNAs in the same PCR run. Using five to seven serially diluted standard cDNAs, standard curves were prepared for both the target (Esr1, Esr2, and Cyp19a1) and reference (Rpl13a) genes. For each experimental sample, the amount of the target and reference genes was interpolated from the appropriate standard curve. The mean mRNA levels of LD animals were arbitrarily set to 1.
Primers for the genes of interest were designed from known sequences from other rodent species, including the Syrian hamster (Mesocricetus auratus) and Chinese hamster (Cricetulus griseus), using Primer3 software (Rozen and Skaletsky 1998; http://primer3.sourceforge.net/). Primers and PCR product sizes are given in Table 1. PCR products from initial experiments were sequenced to confirm their identities. All primers were designed to span at least one intron to ensure that no genomic DNA-based amplicon would confound cDNA-based amplicons.
Results were analyzed with a commercial statistical program (JMP version 7.0.2, SAS Institute, Cary, NC, USA). For uterine mass and body mass from 1 to 12 wk of age, data were normally distributed but the variances were unequal (Levene's test). Data were log-transformed prior to two-factor ANOVA and posthoc Tukey's HSD tests. Data were back transformed for graphing purposes. Analysis of serum E2 and SHBG concentrations, ovarian Cyp19a and uterine Esr1 and Esr2 mRNA levels in 4-wk-olds were made with t-tests. Differences at p < 0.05 were considered to be significant.
Uterine mass was significantly greater in LD than in SD females at 3 wk of age and beyond (p < 0.05, Fig. 1A), and a similar pattern was apparent when uterine mass was expressed relative to body mass (Fig. 1B). Uterine mass normalized to body mass was also significantly different at 1 wk of age, but the measurement of uterine mass at this age may have been less accurate than at older ages owing to the small size of the pups and their uteri.
Body mass was significantly greater in LD than in SD females at 4 wk of age and beyond (p < 0.01, Fig. 1C). Changes in body mass followed a similar pattern in LD and in SD females, albeit with relatively lower body masses in SD starting at 4 wk. However, age-associated changes in uterine mass followed a different pattern in LD than in SD females. In SD, uterine mass increased significantly until 4 wk of age, and then no significant change occurred through 12 wk of age. Conversely, in LD, uterine mass continued to increase beyond 4 wk of age. Uterine mass normalized to body mass showed similar growth patterns.
Serum E2 concentrations in LD and in SD females were not statistically different at 4wk of age (Fig. 2A). At this age neither LD nor SD females had achieved sexual maturity, i.e., the vagina remained closed and no peri-ovulatory follicles or corpora lutea were seen (see below). Serum SHBG concentrations in LD and in SD females were also not statistically different at 4wk of age (Fig. 2B). Siberian hamster SHBG had a high affinity for DHT and testosterone, whereas the affinity for E2 was relatively low, and only 5% of that for DHT.
Paired ovarian mass (not shown) was significantly greater in LD than in SD females at 4 wk of age (p < 0.01), but this difference disappeared when paired ovarian mass was expressed relative to body mass (p = 0.24). Ovarian histology (not shown) was similar in 4-wk-old LD and SD females. All follicle classes up to early antral follicles were observed in ovaries from LD and SD females, including primordial, primary, and secondary follicles. Late antral (peri-ovulatory) follicles and corpora lutea were not seen in ovaries from 4-wk-old females. The levels of ovarian aromatase (Cyp19a1) mRNA did not differ by photoperiod (Fig. 2C).
The defining histological structures of the uterus (endometrium, endometrial glands, and myometrium) could be readily identified by standard histology (H&E staining) in uteri from 4-wk-old LD females, whereas only the endometrium could be identified with confidence in the uteri from age-matched SD hamsters (Fig. 3A and 3B). The outer layers of the LD and SD uteri were clearly identified as myometrium following immunohistochemistry for smooth muscle α-actin (Fig. 3 C and D). The endometrial glands were readily visible following immunohistochemistry for ERα (Fig. 3 E and F), and this was especially true for SD uteri (Fig. 3F). The relative intensity of the ERα immunostaining among the different uterine compartments was identical to that reported in mice (glandular epithelium > luminal epithelium > myometrial smooth muscle; Tibbetts et al. 1998). Whereas the intensity of ERα immunostaining may appear to be greater in the SD than in the LD uterus, we refrain from any further attempt to quantify these results, as the impression may simply reflect the more compact nature of the SD uterus.
The mRNA levels of uterine ERs (Esr1 and Esr2) were not significantly different in LD and SD females at 4 wk of age (Fig. 4 A and B). Generally, the level of ERα (Esr1) was much higher than ERβ (Esr2) (Fig 4 C).
To our knowledge, the present study is the first to explore the effects of photoperiod on uterine growth and serum E2 concentrations in Siberian hamsters in a single investigation. We have replicated previous observations of a pronounced inhibitory effect of SD on uterine growth (Ebling 1994, Place et al. 2004, Timonin et al. 2006), and extended those findings to show that the LD-SD divergence in uterine size occurs by 3 wk of age. This was 2 wk earlier than had been previously investigated (Ebling 1994). Because we found no LD-SD difference in serum E2 concentration at an age (4 wk) when the photoperiod-induced difference in uterine mass has already been established, we looked to the uterus for possible explanations.
Our assessments of uterine estrogen receptors by immunohistochemistry and RT-PCR showed no significant LD-SD differences at 4 wk of age, and yet, uterine mass was significantly less in SD than in LD. We offer the following as possible explanations: 1. The difference in uterine growth and size could potentially be mediated by differences in MEL secretion, as the pineal MEL rhythm in Siberian hamsters matures before 3 wk of age (Tamarkin et al. 1980, Yellon et al. 1985). MEL interferes with estrogen signaling pathways by inhibiting E2-induced ERα-mediated transcription via induction of conformational changes in the ERα-calmodulin complex (del Rio et al. 2004). These observations were made in breast cancer cell lines (MCF-7), but Kanishi et al. (2000) also showed an inhibitory effect of MEL on an endometrial cancer cell line (Ishikawa) that expresses ERα. However, no effect was seen in a cell line (SNG-II) that is ER-negative. Because MEL secretion is inhibited by light, it circulates at a higher concentration and for a longer duration in SD than in LD hamsters (Darrow and Goldman 1985, Hoffmann et al. 1986). Thus it is conceivable that the SD-MEL signal inhibits E2-induced uterine growth in hamsters held in SD. This is not to say that the uterus in SD females will not grow in response to E2, because Scotti et al. (2007) demonstrated substantial uterine growth in E2-implanted SD hamsters. However, the implants in that study resulted in significantly higher serum E2 concentrations than in their LD controls, and Kanishi et al. (2000) abolished the inhibitory effect of MEL by administering E2 to ER-expressing endometrial cells. 2. Somatic growth was inhibited in SD, as evident by the divergence in body mass after weaning in hamsters, and modulators of somatic growth (e.g., epidermal growth factor [EGF] and insulin-like growth factor [IGF]) may impact uterine growth by estrogen dependent and independent means (Branham and Sheehan 1995, Ignar-Trowbridge et al. 1992, Westley et al. 1994). However, Park et al. (2003) found no LD-SD difference in circulating concentrations of IGF-1 in male P. sungorus, and to our knowledge, the effect of photoperiod on EGF in Siberian hamsters has not been investigated. Whereas the presumed mechanism for SD inhibition of uterine growth in Siberian hamsters may be mostly correct, i.e., the SD-MEL signal suppresses gonadotropin, and in turn, E2 concentrations, the cause for stunted uterine growth in SD by 3 wk of age remains to be elucidated.
An explanation for the relatively high E2 values in young SD females as reported by van den Hurk (2002) warrants discussion. van den Hurk et al. (2002) made no mention of a pre-RIA ether extraction of serum samples, thus the method used to measure serum E2 in the present study was significantly different from theirs. Ether extraction of samples prior to RIA for E2 serves three potential benefits: 1. Displacement of E2 from its binding protein (SHBG), 2. Conjugates that may cross-react with the antibody are left in the aqueous phase of the extraction, and 3. Proteins that may interfere with the RIA are removed (Wheeler 2001). The manufacturer of the RIA states the antibody for E2 is highly specific, with cross-reactivities to other estrogens being less than 2%, save for estrone (10%). This modest level of cross-reactivity with estrone is unlikely to explain the discrepancy between extracted and unextracted samples, as estrone (E1) would also be recovered from the ether fraction. In the present study, we also ruled out a LD-SD difference in SHBG concentrations as a possible explanation for the spuriously high E2 concentrations in SD that were reported by van den Hurk et al. (2002). The relatively low binding affinity that Siberian hamster SHBG has for E2 (present study, Gustafson et al. 1989) and for E1 (Gustafson et al. 1989) also argues against SHBG causing high E2 in unextracted serum samples. Alternatively, other substances within the serum could bind non-specifically to the E2 antibody of the RIA, and these yet-to-be-determined substances could be found in higher concentrations in SD than in LD hamsters. Remembering that RIAs work on an inverse relationship basis (less radioactivity means more endogenous hormone), substances that compete with the RIA tracer for antibody binding sites could falsely elevate E2 values. Whatever the offending agent(s) may be, the problem does not seem to be limited to the RIA format, as Scotti et al. (2007) also found higher serum E2 concentrations in SD than in LD hamsters when unextracted samples were analyzed by enzyme immunoassay.
When van den Hurk et al. (2002) sought to explain the high concentration of E2 in SD hamster serum, they logically looked to the ovaries as a potential source. They described the SD ovary as being “highly steroidogenic”, especially the so-called “luteinized atretic follicles”, which showed substantial 3β-hydroxysteroid dehydrogenase (3β-HSD) activity. The so-called “luteinized atretic follicles” are numerous in the SD ovary and are composed of hypertrophied granulosa cells that surround atretic oocytes (Kabithe and Place 2008). van den Hurk et al. (2002) speculated that these structures were the source of the elevated E2 concentrations in SD. Whereas increased 3β-HSD activity could conceivably provide a mechanism for increased E2 concentrations in SD, by increasing the substrates for P450aromatase, in the present study we did not find evidence for higher levels of mRNA (Cyp19a1) for this enzyme in the SD ovary.
The day length regimes used in the present study and by van den Hurk et al. (2002) were not identical, but variations in the photoperiods used are unlikely to explain the study differences in serum E2 outcomes. In our investigations of development in SD females, we routinely gestate and maintain hamsters in 10L (Place et al. 2004, Timonin et al. 2006, Kabithe and Place 2008, Place and Cruickshank 2009) and this was the protocol used in the present study. In contrast, van den Hurk et al. (2002) held females in 16L (LD) during gestation and transferred them to 4L at birth. However, several studies have shown a static SD photoperiod (e.g., 8L or 10 L) during gestation and postnatal development was equally effective at inhibiting sexual development as a dynamic transition from prenatal LD to postnatal SD (Hoffmann 1982, Stetson et al. 1989, Shaw et al. 1995, Goldman and Goldman 2003). Even though van den Hurk et al. (2002) presented no uterine size data, there is every reason to believe that uterine growth in their SD females was profoundly photo-inhibited, and that the reproductive states of SD females in the two studies were comparable.
In conclusion, we have found that photoperiod-induced differences in uterine growth are evident in Siberian hamsters at an early age when no differences in serum estradiol or uterine estrogen receptors could be detected. The possibility that other mediators of uterine growth (e.g., EGF) are modulated by photoperiod warrants investigation.
The authors would like to thank the staff of Laboratory Animal Services at Cornell University, and Jackie Belliveau in particular, for the exceptional care of our animals. We thank Esther Kabithe and Stella Vincent for technical assistance, Dr. Geoffrey Hammond for the analysis of SHBG, and Dr. Matt Paul and two anonymous referees for their helpful comments on the manuscript. The authors declare that there is no conflict of interest that would prejudice the impartiality of this scientific work. This study was supported by NIH grant HD-050358.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.