In humans and sheep, endotoxin (LPS) administration results in increased growth hormone (GH) concentrations. To determine the role of cytokines in the effect of LPS on GH, sheep were challenged with IL-1β or TNF-α. GH data were compared with results with LH, where the major effects of LPS are known to act via the hypothalamus. Intracerebroventricular (icv) administration of IL-1β or TNF-α did not alter plasma concentrations of GH. Endotoxin was then administered intravenously (iv) in combination with icv injection of IL-1 receptor antagonist (IL-1RA), TNF antagonist (sTNF-R1), or saline. Administration of LPS increased GH (P < 0.0001), although coadministration of IL-1ra or sTNF-R1 icv did not alter GH response to LPS. In contrast, plasma concentrations of LH were profoundly inhibited by icv administration of either cytokine (P < 0.03), but the LH response to LPS was not altered by cytokine antagonists. Intravenous administration of either IL-1β or TNF-α increased plasma concentrations of GH (P < 0.0001). Administration of IL-1RA and sTNF-R1 iv prevented LPS-induced increases in GH. Although LH was suppressed by high iv doses of IL-1β (P = 0.0063), the antagonists did not alter the LH response to LPS. To determine whether LPS might directly activate GH release, confocal microscopy revealed colocalization of CD14, the LPS receptor, with GH and, to a lesser extent, LH and some prolactin (PRL)-containing cells, but not ACTH or TSH. These data are consistent with the effects of LPS on GH secretion originating through peripheral cytokine presentation to the pituitary, as well as a potential to act directly on selective populations of pituitary cells via CD14.
- luteinizing hormone
circulating concentrations of growth hormone (GH) change in response to endotoxin (LPS) in a species-specific manner (reviewed in Ref. 10). In sheep and humans, LPS stimulates an increase in circulating concentrations of GH (7, 12, 24), although the LPS-induced increase in GH is not associated with an increase in insulin-like growth factor I (IGF-I) (5, 23). By contrast, in birds, cattle, and rodents, the effect of LPS on GH secretion is generally considered to be inhibitory (8, 13, 21). In rats, the inhibition of GH release in response to LPS has been documented at the level of the hypothalamus and is the result of the mediation of specific cytokines (31). In species where LPS results in increased circulating concentrations of GH, neither the site of LPS action nor the mechanism is understood.
Concentrations of proinflammatory cytokines are increased in response to LPS and are believed to have a specific role in mediating the myriad of endocrine changes that occur in response to LPS in sheep (5, 7). Specifically, after LPS challenge in sheep, circulating concentrations of tumor necrosis factor-α (TNF-α) have been demonstrated to be increased (2, 7, 18, 28), and there is an increase in expression of interleukin-1β (IL-1β) in the choroid plexus (35). In vitro data suggest that the cytokines IL-1β and TNF-α can alter GH release or production (16, 27), although interleukin-2 and interferon-γ had no effect.
Therefore, a series of studies was designed to determine whether LPS alters GH release at the hypothalamus (as described for rats) or the pituitary and whether IL-1β and TNF-α participate in the observed increase in plasma GH concentrations. Moreover, because LH is inhibited by LPS, and its site of action at the hypothalamus is well established (reviewed in Ref. 9), measurement of LH was utilized for contrast purposes and to assist in determining the specificity of the data obtained for GH.
MATERIALS AND METHODS
All experiments were approved by the Auburn University Institutional Animal Care and Use Committee. For ease of handling and to facilitate measurement of LH without the influence of steroidal feedback, castrated male Hampshire-Suffolk sheep were utilized in all experiments. Sheep were housed under 12:12-h light-dark cycle. Sheep were fed 3.5% body wt of a total mixed ration (12% crude protein) and allowed ad libitum access to water. On the morning of an experiment, all food and water were removed for the duration of the experiment. Sheep were previously trained and acclimated to light halter restraint for blood collection procedures. For all animal experiments, rectal temperatures were monitored, and sheep with elevated temperatures (40°C) after collection of the last blood sample were treated with flunixin meglumine to reduce fever (Banamine; Schering-Plough Animal Health, Kenilworth, NJ).
Effect of intracerebroventricular cytokines.
To gain direct access to the neuroventricular compartment and its unique route of hypothalamic communication, indwelling intracerebroventricular (icv) cannulas were surgically implanted in sheep as described previously (29). Sheep were treated with flunixin meglumine as an analgesic (1 ml) after the surgery and on the following day, and with florfenicol to prevent possible infections (Nuflor; Schering-Plough Animal Health) after the surgery and again 2 days after the surgery. The surgical wound was cleaned and treated with 1% silver sulfadiazine cream (BASF, Mt. Olive, NJ) for 5 days after the surgery. Placement of the cannula was confirmed by injection of a radio-opaque dye (Omnipaque; Amersham Health, Oslo, Norway), followed by contrast radiography. Sheep were allowed ≥2 wk to recover before initiation of an experiment. Due to the critical and delicate nature of the surgical procedures, animals were used in more than one challenge procedure, where adequate time was permitted between uses for animals to recover.
The day before infusion, sheep (n = 4 per treatment) were fitted with an indwelling jugular cannula. Blood samples were collected every 15 min for 1 h before icv challenge with TNF-α (0.05 or 0.5 μg/kg recombinant human TNF-α; PeproTech, Rocky Hill, NJ,), IL-1β (0.005 or 0.05 μg/kg recombinant ovine IL-1β, G. Barcham; University of Melbourne, Parkville, Victoria, Australia,), or saline and every 15 min for 8 h after the challenge. Immediately before the challenge, the delivery icv cannula was placed inside the icv guide. Challenges were delivered in a volume of <500 μl over 30 s. Core body temperatures were measured every hour. A minimum of 1 wk was allowed between challenges.
Effect of icv cytokine antagonists.
Intracerebroventricular cannulas were placed as described in the previous section. The day before infusion, the sheep (n = 4 per treatment) were fitted with an indwelling jugular cannula. Blood samples were collected every 15 min for 1 h before challenge with LPS (0.6 μg/kg iv, Escherichia coli 055:B5 lot 20k4083; Sigma, St. Louis, MO,) or saline and every 15 min for 8 h after the challenge. Endotoxin challenge was administered via the jugular cannula. Sheep were treated via the indwelling icv cannula with IL-1 receptor antagonist (IL-1RA); (20 μg/kg, Amgen, Thousand Oaks, CA), soluble TNF-receptor 1 (sTNF-R1, 2 μg/kg Amgen), or saline icv immediately before LPS challenge and every hour after LPS challenge for the remainder of the sample collection period. A minimum of 2 wk recovery was allowed between LPS challenges.
Effect of iv cytokines.
Sheep (n = 5 per treatment) were fitted with an indwelling jugular cannula the day before the experiment. Blood samples were collected every 15 min for 1 h before challenge with IL-1β (0.5 and 5 μg/kg recombinant human IL-1β, PeproTech), TNF-α (0.5 and 5 μg/kg recombinant human TNF-α, PeproTech), or saline, and every 15 min for 8 h after the challenge. Cytokines were administered via the jugular cannula and followed with a 3-ml flush of sterile saline. Core body temperatures were determined every hour. A minimum of 1 wk recovery was allowed between challenges.
Effect of iv cytokine antagonists.
Sheep (n = 5 per treatment) were fitted with an indwelling jugular cannula the day before the experiment. Blood samples were collected every 15 min for 1 h before injection of LPS (0.6 μg/kg) or saline and every 15 min for 8 h after the challenge. Endotoxin was administered via the jugular cannula. Immediately before LPS challenge and every hour after the challenge for the remainder of the sample collection period, sheep were treated with IL-1RA (200 μg/kg), sTNF-R1 (20 μg/kg), or saline iv. A lower dose of IL-1RA (20 μg·kg−1·h−1) was also administered but did not alter GH or LH response to LPS (P ≥ 0.2072); therefore, results of only the higher dose of IL-1RA (200 μg·kg−1·h−1) are presented. Core body temperatures were measured every hour. A minimum of 2 wk recovery was allowed between LPS challenges.
All blood samples were immediately placed into EDTA-containing tubes and centrifuged at 4°C. Plasma samples were stored on ice until transported to the lab and stored at −20°C until assayed. Plasma concentrations of GH were determined in duplicate by RIA as described previously (29, 30). Intra- and interassay coefficient of variations were <17% for all assays. Plasma concentrations of LH were determined in duplicate by RIA as described previously (7). Intra- and interassay coefficient of variations were <12.3% for all assays.
Endotoxin receptor, CD14, and localization in sheep pituitary cells.
Sheep (n = 3 per treatment) were injected with LPS (0.6 μg/kg iv) or nonpyrogenic saline. Four hours after LPS challenge, sheep were euthanized with pentobarbital sodium. Pituitaries were collected and preserved in Bouin's solution for 24 h and transferred to 70% ethanol for subsequent paraffin embedding and confocal microscopy as previously described (26). Briefly, tissue sections were cut (5 μm thick), deparaffinized with xyline, and rehydrated to water through decreasing ethyl alcohol concentrations. Nonspecific immunoreactivity was neutralized by blocking with a mixture of 3% donkey and 3% goat normal sera (Jackson Immunoresearch Labs, West Grove, PA) in Tris-saline. Slides were then incubated overnight at 4°C in a mixture of mouse monoclonal anti-ovine CD14 (1:500 dilution, Serotec, Oxford, UK,) and a rabbit polyclonal antibody against a pituitary hormone [1 hormone per mixture, 1:1,000 dilution: adrenocorticotropin (ACTH; Serotec), GH (15, R-1–1-3), prolactin (PRL; a generous gift from Dr. Douglas Bolt, USDA), LH (a generous gift from the National Institute of Arthritis, Diabetes, and Kidney Diseases, National Institues of Health), and TSH (procured from Dr. A. Parlow)]. The next day, the sections were incubated for 1 h with a mixture of bodipy-goat anti-rabbit (Molecular Probes, Eugene, OR) and Texas red-donkey anti-mouse (Jackson Immunoresearch) both at a final concentration of 1:200. After thorough washes, the slides were mounted in GelMount (Biomeda, Foster City, CA). Confocal visualization of antigen localization and colocalization was accomplished using a Zeiss Laser Scanning Microscope 510 equipped with four lasers. Digital pictures were taken of 6–10 fields/specimen for different pituitary regions with high densities of the respective hormones. CD14 was represented by red fluorescence, individual hormones represented by green fluorescence, and the colocalization within the same cell represented by convergence to a yellow signal. Quantification signals as single or colocalized antigens were accomplished by digital image analysis as previously described and validated by Elsasser et al. (14). Each digital photograph was captured using identical settings for illumination through the microscope and analyzed using the Image-Pro Image Analysis Software (version 4.5.1; Media-Cybernetics, Silver Spring, MD) through a standardized protocol. Each field captured averaged 30 cells, as determined by counting cell nuclei; specific hormone-containing cells within a given field ranged between 20 and 90% of the presented cells, a reflection of the relative abundance of hormone-secreting cells in a pituitary region. After importation of the image into the analysis field, the image was automatically optimized for brightness, contrast, and gamma by using a software-driven internal best-fit algorithm. The pixel density of red-green fluorescence, color-specific staining was obtained by defining through the color cube-based segmentation option a spectrum-specific range of wavelengths, hues, and intensities that corresponded to the specific emissions spectrum for a given flurophore. Yellow colocalization was manually established using the pixel selector to acquire the needed spectrum information. Specificity of staining was further defined in the software application cutoff criteria, where object counts lower than a 3 × 3 pixel unit were eliminated. As defined in this manner, this analytic solution was found to be most commonly associated with false negative staining, thus minimizing errors of false positive inclusion through this conservative rubric. This spectral information was filed into three separate retrievable *.rge formats (one for each color, files available on request) that were subsequently used for analyzing all images. In this fashion, unbiased continuity of analysis across tissue sections was achieved. Quantification of specific hormone cell-associated CD14 was achieved by first isolating and defining the border of a given hormone-secreting cell using the yellow fluorescence and then resampling the available pixels through the red fluorescence channel. For each sheep, a CD14 quantitative statistic was generated as the average cell-specific pixel density as derived mean of 30–100 counted cells, depending on the population density of a cell type in a given pituitary region.
Data for GH were analyzed using a general linear model (GLM) procedure for ANOVA for repeated measures in the Statistical Analysis System software (version 8; SAS Institute, Cary, NC). Differences between treatments were established as significant with P ≤ 0.05. Although the sample collection interval was not ideal for description of pulse parameters, plasma concentrations of LH after cytokine or LPS challenge were subjected to Cluster analysis (34) for determination of mean concentration of LH, area under the curve, peak frequency, peak height, and nadir to facilitate comparison with previous studies (1, 7, 11, 18, 36). Pulse parameters of LH were then tested for effects of treatment utilizing GLM procedures of SAS. Peak frequency was compared between treatments by ANOVA of the ranks of the peaks. Means separation procedures were preformed using LSMEANS statements if the main effect was determined to be significant. Images were statistically analyzed as a function of the mean color-specific pixel density/field for individual antigens, as well as the ratio of colocalized pixels to the total hormone pixels with and without LPS treatment.
Administration of roIL-1β and recombinant human TNF-α (rhTNF-α) icv did induce a significant increase in body temperature [41.0°C (SD 0.7) high IL-1β, 40.6°C (SD 0.7) low IL-1β, 40.6°C (SD 0.8) high TNF-α, 40.1°C (SD 0.5) low TNF-α, and 39.2°C (SD 0.3) saline; P = 0.0007], but did not affect plasma concentrations of GH (P = 0.0976; Fig. 1). However, central administration of low- and high-dose IL-1β and high-dose TNF-α decreased the mean concentration of LH, as well as area under the concentration-by-time curve and number of detected peaks (P < 0.05; Table 1). Additionally, low-dose TNF-α icv administration also decreased LH area under the curve and number of peaks (P < 0.04).
In a preliminary experiment to confirm that the antagonists would block responses to cytokines, the cytokines roIL-1β or rhTNF-α were administered icv in combination with icv IL-1RA (20 μg·kg−1·h−1) or sTNF-R1 (2 μg·kg−1·h−1). Temperature was recorded every hour. Administration of IL-1RA prevented IL-1β-induced fever (body temperature actually decreased, P = 0.0279; Fig. 2A) and administration of sTNF-R1 prevented TNF-α-induced fever (P = 0.0069; Fig. 2B). However, administration of IL-1RA did not prevent TNF-α-induced fever (Fig. 2B), indicating that the antagonist was specifically blocking the effects of IL-1β and not a general inhibitor of fever. In addition, the ability of recombinant human IL-1-receptor antagonist to inhibit actions of roIL-1β suggests that the human antagonist functions at the oIL-1 receptor. Endotoxin significantly increased GH concentrations (P < 0.0001; Fig. 3), but the administration of IL-1RA or sTNF-R1 icv did not prevent LPS-induced increases in circulating concentration of GH (P = 0.2572; Fig. 3, A and B). However, LPS treatment did significantly affect mean concentration of LH (P = 0.0009), area under the curve (P = 0.0105), peak number (P = 0.0009), peak height (P < .0001), and nadir (P = 0.0003; Table 2). Neither cytokine antagonist had an effect on the LPS inhibition of LH.
In contrast to icv administration of IL-1β and TNF-α, administration of both rhIL-1β and rhTNF-α iv stimulated a significant increase in plasma concentrations of GH relative to the saline-treated animals (P < 0.0001; Fig. 4). Administration of a low dose of TNF-α iv stimulated a smaller but significant increase in plasma concentrations of GH, but there was no effect of the low dose of IL-1β on plasma concentrations of GH. However, only administration of high-dose IL-1β iv decreased mean concentration of LH (P = 0.0063), although high- and low-dose IL-1β and high-dose TNF-α iv did result in decreased number of peaks (P < 0.03; Table 1).
Administration of LPS stimulated two increases in plasma concentrations of GH, the first significant increase beginning at 105 min posttreatment and a second significant increase 225 min posttreatment. Administration of IL-1RA iv prevented LPS-induced increases in GH at most time points (P < 0.05, Fig. 5), and sTNF-R1 prevented the LPS-induced increase in plasma concentrations of GH at all time points (P < 0.05, Fig. 5). LH area under the curve was affected (P = 0.0422) by treatment with LPS (Table 3). Mean concentration of LH (P = 0.07091) and nadir (P = 0.083) had a tendency to be affected by treatment with LPS. Peak height and peak number were not significantly affected by treatment with LPS. Administration of antagonist and the interaction of LPS and antagonist administration did not significantly affect any LH pulse parameters examined.
In general, hormones could be detected throughout the anterior pituitary, but high-density clusters of immunoreactive cell regions were photographed and used for the image analysis. The pictures illustrate that some low level of CD14 was present in most pituitary cells before LPS challenge. Although the percentage of colocalization of CD14 with a particular type of hormone-containing cell was somewhat variable, virtually 100% of the GH somatotropes contained CD14 (Fig. 6). Cells containing ACTH, LH, PRL, and TSH presented decreasing colocalization ratios, with CD14 ranging from 32 to 9% (P < 0.001 vs. GH cells). In GH cells, treatment of sheep with LPS resulted in a 70% decrease (P < 0.004) in CD14 pixel intensity (Fig. 7). CD14 pixel intensity was ≥28% higher (P < 0.05) in GH cells than in any other cell type. Although the decrease in pixel intensity after LPS in LH cells approached 50%, the numerical change was not statistically significant because of large variability in the positive staining populations.
Under normal circumstances of sound health, secretion of GH by the pituitary is under the influence of releasing and inhibiting hormones of hypothalamic origin, including GH-releasing hormone and somatostatin (4, 17, 32, 33). Likewise, secretion of LH by the pituitary is primarily under the influence of gonadotropin-releasing hormone (GnRH) from the hypothalamus. The regulatory scenario becomes more complicated in states of toxemia-associated decompensation due to additional influences from induced differential regional blood flow, localized ischemia, acid-base balance, the dose or level of toxin incurred, and the temporal relationship in the cascade of postchallenge effectors.
Although TNF-α and IL-1β increased fever via central administration in our present study, the data appear to indicate limited effects of these cytokines on hypothalamic regulation of circulating concentrations of GH. In contrast, central administration of TNF-α and IL-1β clearly inhibited circulating concentrations of LH, indicating that these peptides or downstream effectors were biologically active and presumably reached hypothalamic targets with known effects on the regulation of LH. Central administration of antagonists to these cytokines had little effect on LPS action on either GH or LH, suggesting that the two cytokines have little involvement in mediating the LPS effects at the hypothalamus. These data suggest that the effect of LPS on circulating concentrations of GH is either via other cytokines acting at the hypothalamus or LPS and/or cytokines acting at the level of the pituitary. In further support of the peripheral site of action for LPS stimulation of GH, research conducted by Briard et al. (6) indicated that LPS administration to sheep did not alter pituitary portal vein concentrations of GH-releasing hormone and actually increased pituitary portal vein concentrations of somatostatin. Taken alone, the alterations in hypothalamic-releasing hormone concentrations observed by Briard et al. would suggest that LPS should have suppressed circulating concentrations of GH, but Briard et al. observed increased circulating concentrations of GH in animals used in that study (6). Thus LPS stimulation of increased circulating concentrations of GH does not appear to involve hypothalamic mechanisms.
Increased GH release does not appear to be influenced at a hypothalamic level by cytokines; however, this is not the case for LH. Although LH is inhibited at a central level, the effect of LPS is not mediated exclusively by TNF-α or IL-1β. Indeed, multiple other factors, including cortisol, prostaglandins, and opioids have been suggested to possibly be involved in LPS suppression of LH (11, 18, 23, 37). The ability of these cytokines to inhibit LH centrally without an effect on GH further strengthens the concept that the hypothalamus does not mediate LPS actions on GH.
In the present study, iv administration of TNF-α and IL-1β dramatically increased circulating concentrations of GH. Because icv administration of these cytokines had little effect on GH release, the actions of these iv cytokines appear to be expressed at the pituitary. In contrast, only the highest dose of IL-1β altered mean concentrations of LH. Because IL-Iβ results in increased LH release from cultured pituitary cells (3), the inhibitory effects of iv injection of IL-1β likely require uptake across the blood-brain barrier. Furthermore, the antagonists to the two cytokines inhibit LPS effects on GH, suggesting a pituitary site of action of LPS via TNF-α and IL-1β. Interestingly, inhibition of TNF-α mediation of LPS actions had the greatest effect on GH release. Since TNF induces the initiation of the cytokine cascade, inhibition of TNF-α may serve as a key molecule regulating GH or, by inhibition of the cytokine cascade, prevent the production and release of other cytokines (such as IL-1β) that might subsequently release GH. Culture of sheep pituitary cells with TNF-α increased GH mRNA expression (27). However, culture of sheep pituitary cells with TNF-α did not affect the medium concentration of GH (16). Both IL-1α and IL-1β stimulated an increase in GH mRNA and GH concentration in the media from cultured pituitary cells (16), although TNF-α, IL-2, and interferon-γ had no effect on the pituitary cells to release GH. These in vitro data, combined with the effects of the cytokine antagonists in the present study, argue for an effect of IL-1β in mediating the effects of LPS on GH release from the pituitary. Collectively, these data clearly support a role of the cytokines TNF-α and IL-1β in LPS-induced increases in circulating concentrations of GH. Moreover, because the GH response to LPS, cytokines, and cytokine antagonists differs from the response of LH, it suggests that the effects observed with GH are specific actions.
The effects of LPS on pituitary function represent a complex interplay between various cytokines (such as IL-I and TNF) and other downstream modifiers. Indeed, the pattern of GH release in response to individual cytokines does not directly replicate the response to LPS. In addition to LPS inducing cytokine release (which in turn stimulates the somatotrope to release GH), LPS could also act directly on pituitary cells to regulate GH release. The ability of pituitary cells to respond directly to LPS would require the presence of a receptor capable of binding LPS and initiating a signal transduction response. The membrane-associated form of CD14 satisfies this capacity. Indeed, colocalization of GH and, to a lesser extent, LH with CD14 indicate that somatotropes and gonadotropes have one of the biochemical requirements necessary to respond directly to LPS. The importance of CD14 was suggested in experiments in mice that were CD14 deficient. When a dose of LPS was administered to CD14−/− mice, 10-fold the LD100 for normal mice only in mild illness was evidenced (19, 20). Furthermore, CD14-deficient mice also produced less TNF-α than normal mice in response to LPS (19, 20). The presence of CD14 in somatotropes and gonadotropes provides a mechanism consistent with the data of Coleman et al. (7), indicating that exposure of cultured dispersed pituitary cells to LPS resulted in increased concentrations of both GH and LH in the media. Therefore, the colocalization data in the present study would indicate that the GH secretion in response to LPS is due in part to direct actions of LPS on the somatotrope. Indeed, the delay in the increase in circulating concentrations of GH after LPS administration observed in the present study and by previous researchers (6, 7) suggests that additional factors are necessary at the level of the pituicyte for LPS to induce increases in GH. Additionally, it is unlikely that LPS is acting directly on the gonadotrope to decrease circulating concentrations of LH in vivo, as the whole animal response is the opposite of the in vitro response of increased concentrations of LH. However, investigations have indicated a decreased effectiveness of GnRH to induce LH pulses in the presence of LPS (1, 36). The coexpression of the CD14 molecule with LH suggests that LPS may directly cause a disconnection between GnRH and LH pulses. Perhaps LPS and cytokines act in concert at the pituitary to alter concentrations of GH and LH.
The generalized decrease in cell presentation of CD14 in the pituitary is a new observation, albeit one consistent with the observations made previously on membrane CD14 in alveolar macrophages (25). The generalized decrease in membrane presence at the 4-h sampling point is consistent with the concept that the development of systemic tolerance to LPS may be mediated by the loss of membrane receptor through internalization or shedding of the CD14 receptor with the consequence of overall lower responsiveness. It is interesting that the largest measure of CD14 was present in cells coexpressing GH and that, similarly, the downregulation was significant mainly for the somatotropes. The next-highest level of CD14 expression was evident in PRL-associated lactotropes. The rather large level of variability in this cell population may be related to the presence of so-called “mammosomatotropes,” a subset of pituitary cells coproducing both PRL and GH and detected as such by using an in vitro reverse hemolytic plaque assay and the R-1–1-3 antibody used in the present study (22).
In conclusion, the results of these experiments indicate that one component of the molecular mechanisms for responding to LPS, CD14, is present on somatotropes and gonadotropes. The cytokines IL-1β and TNF-α induce increases in circulating concentration of GH through a peripheral mechanism and are involved in mediating LPS-induced increases in circulating concentration of GH. Furthermore, these results indicate that LPS-induced increases in circulating concentrations of GH are mediated through a peripheral mechanism likely at the somatotrope. Additionally, the inflammatory cytokines IL-1β and TNF-α suppress circulating concentrations of LH, although neither appears to be primary mediators of LPS suppression of LH concentrations.
Preliminary reports have appeared in the Program of the 83rd Annual Meeting of the Endocrine Society, Denver, Colorado, 2001 (Abstract P2-597). This research was supported by United States Department of Agriculture Grant 99-03361.
We thank J. Baker for assistance with this project.
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 by American Physiological Society