| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Kimmel Cancer Center, Department of Cancer Biology, Thomas Jefferson University, Philadelphia, Pennsylvania 19107
Address all correspondence and requests for reprints to: Matthew LeBaron, Ph.D., Thomas Jefferson University, Kimmel Cancer Center, Department of Cancer Biology, 233 South 10th Street, 330 BLSB, Philadelphia, Pennsylvania 19107-5541. E-mail: mlebaron{at}kimmelcancercenter.org.
 |
Abstract |
|---|
|
Introduction |
|---|
Key insight into the main functions of polypeptide hormones can be gleaned from gene knockout studies of the individual hormones or their receptors in mice. Such genetic models, however, are limited to providing information about gene functions that are critical and therefore cannot be compensated for during development by redundant mechanisms. Furthermore, it is difficult to distinguish between direct and indirect effects of gene loss on tissue and cell function. For instance, infertility of female PRL receptor (PRLR)-null mice is caused by blastocyst implantation failure that is secondary to ovarian progesterone deficiency (1, 2). Likewise, in chronic hormonal overexpression or hyperstimulation models, biological effects also may be direct or indirect. Alternatively, ligand binding assays and mRNA or protein expression studies of receptors have been used to identify candidate hormone target cells at the organ, tissue, or cell level. Complicating these analyses is the existence of multiple isoforms or splice variants of transmembrane receptors that differ in signaling competence or may even be dominant-negative variants. For instance, truncated splice forms of PRLRs have been shown to act as dominant-negative suppressors of at least some functions of the full-length receptor (3) and are expressed at high levels in certain organs such as liver (4). Other factors, including the absence of coreceptors or presence of inhibitors, also may affect cellular responsiveness. Finally, gene transcript levels may not reflect protein expression levels (5). Therefore, the existence of receptors in a given tissue, either at the protein or transcript level, does not prove signaling competence.
To identify direct target cells responsive to PRL, GH, EPO, and GCSF within the physiological setting of the whole animal, we combined in situ detection of rapid receptor-mediated signal transducer and activator of transcription (Stat)-5a/b tyrosine-phosphorylation with high-throughput tissue array analysis using cutting-edge matrix assembly (CEMA) (6, 7). Because these hormones have the capacity to activate both or either of the highly homologous Stat5a and Stat5b gene products (8) and the phospho-Stat5a/b(Y694/699) antibody used does not discriminate between phosphorylated Stat5a and Stat5b (9), the term Stat5 in this text refers to both or either protein. Inducible tyrosine phosphorylation of Stat5 is an immediate, sensitive, and common marker of signaling by receptors for these hormones. CEMA allows arraying of most tissue types, including architecturally complex, thin-walled, and stratified tissues such as gut and skin. Furthermore, analysis of tissue arrays is cost effective, less labor intensive, and limits interassay and interslide variability in sample staining and analysis (10). The experimental protocol involved in situ analysis of nuclear-localized, tyrosine-phosphorylated Stat5 in formalin-fixed rat tissues harvested 30 min after injection of a single, receptor-saturating dose of hormone for maximal receptor activation.
The data provide a high-resolution map of PRL, GH, EPO, and GCSF-responsive cell types in male and female rat tissues based on rapid Stat5 activation as the readout. In general, the map revealed new insight into the broad arrays of target cells for PRL and GH and the more restricted target cell populations of EPO and GCSF. In addition to providing new information on hormone-responsive target cells, the present work also provides proof of concept of in situ biomarker-based mapping of cell types responsive to hormones or drugs. In vivo response-based detection of receptor-mediated signaling at single cell resolution is expected to facilitate understanding of effects and side effects of drugs and hormones.
|
Materials and Methods |
|---|
4 µg/g body weight (bw)], oGH (4 µg/g bw), oPRL + oGH (4 µg/g bw each hormone), human GCSF (1 µg/g bw), or human EPO (10,000 U/rat). Previous publications used similar doses of PRL and GH to analyze activated Stat5 in vivo (6, 9, 11, 12). Highly purified, pituitary oPRL (lot no. AFP9220A) and oGH (lot no. AFP10692C) were provided by Dr. A. F. Parlow (under sponsorship of the National Institute of Diabetes and Digestive and Kidney Diseases, National Hormone and Peptide Program, Torrance, CA), whereas recombinant human EPO (Epogen; Amgen Inc., Thousand Oaks, CA) and recombinant human GCSF (Neupogen; Amgen) were purchased commercially. We did not measure circulating levels of injected hormones in the rats at time the animals were killed. Using the standard formula for calculation of blood volume for the rat (75 ml/kg bw), we estimated the theoretically maximal circulating concentrations after injection as approximately 2500 nM for oPRL and oGH, approximately 100 nM for EPO (444,444 mU/ml), and approximately 800 nM for GCSF. By using these supraphysiological doses (high nanomolar), our intent was to rapidly achieve receptor-saturating doses (low nanomolar) at peripheral cells and tissues to reach all potential target cells for the hormones. Experiments were performed on both male and female rats, with four to six rats included in vehicle injected control groups to establish baseline levels of Stat5 activation, and at least two rats were included for each injection parameter for each gender. This relatively large number of vehicle-injected rats used to establish baseline Stat5 activation levels was especially critical in male rats due to the pulsatile nature of GH secretion (13). In the most GH-responsive tissues (i.e. liver and connective tissues), however, we did not detect basal Stat5 activation in control animals, indicating that Stat5 activation in the GH target tissues under baseline conditions did not reach detectable levels in the six males analyzed. To minimize tissue harvesting time, the injections were staggered so that only one rat was dissected at a time. Thirty minutes after injection, rats were killed, rapidly dissected by three investigators, and the respective tissues and organs collected and immersed in 10% phosphate-buffered formalin at 4 C within 10 min for 18 h and then transferred to 70% ethanol until paraffin infiltration and embedding. Preliminary testing included harvesting of tissues at 20, 30, and 40 min after injection of PRL and GH, and no difference was noted in Stat5 activation status; we therefore concluded that 30 min was representative of maximal activation of Stat5. The same 30-min time point was used for Stat5 activation by EPO and GCSF, which is also consistent with published in vitro studies (14, 15). Furthermore, in a previous study we plotted the dephosphorylation of Stat5 in rat liver over time and noted that phosphorylation was relatively unchanged within the first hour postmortem, indicating the stability of the phosphoepitope for immunohistochemistry (IHC) (6). Samples were sectioned and placed on standard microscope slides; alternate slides were stained with hematoxylin and eosin (H & E) for histological analysis or used for IHC.
CEMA tissue array construction
Where possible, tissues were constructed into CEMA tissue arrays, as previously described (6, 7). Tissues and organs needed to be of sufficient size to be used as plates in the construction process. Briefly, paraffin-embedded tissues were trimmed to the desired thickness (typically 250–300 µm) to form primary plates. In the case of thin-walled tissues such as skin or gut, the tissue was directly used as a primary plate, as previously described (6, 7). The primary plates were bonded into a primary stack and sectioned transversely at the appropriate thickness (typically 250–300 µm) to form a secondary plate. These secondary plates were then stacked and bonded into a secondary stack. This bonded tissue array block contained elements of each of the original samples and, when transversely sectioned to micrometer-thin array sections, allowed transfer to support slides for analysis.
IHC
Immunostaining using AX1 antiphospho-Stat5 antibody (Advantex BioReagents, Conroe, TX) detected activated, tyrosine-phosphorylated Stat5, as previously described (9). Briefly, tissues were deparaffinized and rehydrated in ethanol. For antigen retrieval, the samples were microwave treated in a pressure cooker in AR-10 antigen retrieval solution (Advantex BioReagents). Endogenous peroxidase activity was blocked by pretreating the slides with hydrogen peroxide and normal goat serum was added to decrease nonspecific binding. The antiphosphotyrosine Stat5 monoclonal antibody was diluted in BSA and PBS and added to the sample. The secondary antibody was a biotinylated goat-antimouse IgG followed by streptavidin-horseradish-peroxidase complex with the chromogen 3,3'-diaminobenzidine (DAB) used to visualize the antigen-antibody complexes. As a positive control, Stat5 activation status was normalized with lactating mouse mammary gland for each experiment, and all samples were lightly counterstained with Mayer hematoxylin.
IHC scoring
Three separate reviewers (M.J.L., M.T.N., and H.R.) independently scored the tissues using a standard scale: – (negative), + (low staining), 2+ (medium staining), or 3+ (high staining). Where applicable, the cell types or specific subsets of cells within a tissue were identified and scored. All staining stronger than the minimum criteria for 3+ staining was normalized to 3+. Reported scores are representative of the compiled results from the three reviewers on each of the rats used in each treatment. In cases of discrepancy, the samples were reanalyzed by all three reviewers simultaneously and a consensus was agreed on. Throughout this manuscript, the terms activated Stat5 and Stat5 activation refer to nuclear localized Stat5a/b phosphorylated on the conserved residue Stat5a/b-Tyr694/9 as detected by IHC (16). In this work, hormone-inducible Stat5 is used as a readout of hormone-induced receptor activation, and future work will need to determine effects of Stat5 activation in various target cells. There may be instances in which nuclear localized Stat5a/b phosphorylated on the conserved residue Stat5a/b-Tyr694/9 is not functionally active due to presence of inhibitory cofactors such as PIAS-3 (17).
|
Results and Discussion |
|---|
A limitation of this study is that activation of receptors for PRL, GH, EPO, and GCSF is measured solely by Stat5 activation. Although each of these receptors is thought to signal primarily through the receptor-associated Janus kinase (Jak)-2 leading to Stat5 activation, alternative and/or additional signaling pathways may be activated. Therefore, a cell still may be responsive to a given hormone, even if no Stat5 activation is detected, and hence, a negative result does not rule out receptor signaling. Nonetheless, the use of an in situ, immunohistochemical approach provides the best possible readout of receptor signaling in the intact organism because it provides spatial information at the single cell level within tissues. Future approaches with the use of more sensitive, fluorescent-based detection (vs. DAB) may detect additional target cells. It will also be important to expand this work to include other biomarkers of receptor activation. For instance, EPO has been demonstrated to stimulate cardiac and vascular cells through a phospholipase C/protein kinase C pathway (21), but it is unclear whether this is Jak2 independent. Likewise, Jak2-independent signaling through Src tyrosine kinase Fyn (22) and MAPK (23, 24, 25) by PRLR has been proposed, and PRL and GH are capable of activating Stat transcription factors other than Stat5 (26, 27). Although rapid Stat5 activation by hormones in a given cell indicates functional receptors and direct responsiveness, absence of a Stat5 response in a cell type does not rule out hormone responsiveness via alternative pathways. The purity and specificity of these highly purified hormone preparations was supported by distinct target cell responsiveness. For example, a large number of tissues were exclusively responsive to either GH or PRL, indicating that contamination with the other hormone in the ovine pituitary derived GH and PRL preparations was not a critical issue. Furthermore, EPO and GCSF were clinical-grade, recombinant hormone preparations. Whereas it cannot be excluded that some endotoxin and lipopolysaccharide remained in these solutions, endotoxin and lipopolysaccharide are not direct activators of the Stats, (reviewed in Ref. 28).
In addition to performing injections of PRL and GH independently, PRL and GH were also coinjected in an effort to identify additive hormone effects in target cells. In some cases, the combination of the two hormones activated distinct target cells that when combined resulted in an overall increased level Stat5 responsiveness within the tissue (e.g. mammary gland). In other cases both hormones affected overlapping target cells (e.g. liver) and additive effect on staining intensity was assessed. Whereas lack of detectable additive effect in such cases may reflect rate-limiting effects of the joint Jak2-Stat5 pathways when coactivated by PRL and GH, it may also be due to limitations in sensitivity and dynamic range of the DAB chromogen. Lack of additive effects of PRL and GH should therefore be interpreted with caution.
CEMA tissue arraying technology (6, 7) was used to facilitate parallel analysis of hundreds of tissue features, including multilayered tissues such as skin and intestines (Fig. 1
). In parallel, we also used sections from standard tissue blocks for staining and scoring of all samples. Tissues that were highly heterogeneous (e.g. vertebrae or kidneys) or too small for effective arraying (e.g. ovaries or adrenals) were analyzed exclusively in the standard manner. Below follows a presentation and discussion of the data categorized by organ system. For ease of presentation in the text, unless otherwise noted, there was no detectable response to EPO or GCSF in cells of the respective tissue or organ.
|
). Previous work identified very high levels of PRLR mRNA in the rat adrenal gland (29), among the highest of any of the tissues tested. Furthermore, circulating PRL is known to stimulate the production and secretion of adrenal androgens (30), cortisol, and aldosterone (31) and increases adrenal 21-hydroxylase activity (32). Our results support a key role of Stat5 as a mediator of the effects of PRL on adrenal physiology. In contrast, GH did not induce detectable activation of Stat5 in cells of any of the layers of the adrenal cortex. These data are interesting in light of relatively high levels of GH receptor (GHR) mRNA detected in the rat adrenal cortex (33) by RNA in situ hybridization, suggesting lack of a functional receptor or GHR-mediated signaling through an alternate pathway.
|
). The observed unresponsiveness to PRL is consistent with the lack of detectable PRLR mRNA in the cells of the adrenal medulla (34) and the very low levels of GHR mRNA as determined by in situ hybridization (33). Previous reports indicating a role for PRL in inducing catecholamine secretion from the medulla (35, 36) therefore may reflect an indirect effect, which would be consistent with the chronic nature of the hyperprolactinemia over 45 d caused by renal subcapsular transplantation of an anterior pituitary. Numerous studies indicated that PRL is a stress hormone (37, 38, 39, 40), and it is conceivable that a role of PRL in the acute stress response involves regulation of adrenal cortical hormones, whereas chronic hyperprolactinemia also may involve other stress-related effectors such as adrenal catecholamines through indirect mechanisms.
Pancreatic islets of Langerhans.
The pancreas is involved in digestion and nutrient uptake by both endocrine and exocrine functions. Interspersed between exocrine pancreatic acini are distinct endocrine cell masses termed the islets of Langerhans. These islets were another endocrine tissue component in which PRL induced marked Stat5 tyrosine phosphorylation (Fig. 2). Previous work demonstrated direct effects of PRL on many aspects of islet cell biology, including the release of insulin into the circulation (41, 42, 43), and PRLR mRNA expression has been generally localized to the cells of the islets (4). In contrast, in our study GH did not stimulate Stat5 phosphorylation higher than basal levels in any of the subpopulations of cells within the islets, indicating a significant difference between cell responsiveness to PRL and GH. Others also have shown an enhanced effect of PRL over GH in activating Stat5 in rat islet cells; however, the difference was attributed to desensitization by continued exposure to GH, beginning after 1 h of treatment, whereas PRL had a biphasic activation that resulted in prolonged Stat5 activation (44). Several laboratories are investigating PRLR signaling within the islets of Langerhans to gain further insight that could lead to better treatment for type 2 diabetes in humans.
Testis.
The testis produces sperm cells in the seminiferous tubules and androgens in the interstitial cells or Leydig cells. It is well established that PRL promotes the maintenance of testicular Leydig cell morphology (45) and generally promotes androgen production (46, 47, 48, 49, 50) in many mammals. Consistent with a direct effect of PRL on Leydig cells, PRLR mRNA has been identified in the Leydig cells (51, 52) and PRL binding to isolated Leydig cells has been reported (53). PRL–/– (54) and PRLR–/– (55) males, however, are fertile and have no histological abnormalities of the testis, and no alterations of plasma testosterone were detected. Thus, PRL signaling in Leydig cells is not critical for fertility of male mice but may be compensated for by other pathways. Our study indicated that PRLR in a majority of rat Leydig cells are responsive to PRL in vivo (Fig. 2). Although GH or GCSF did not induce Stat5 activation in Leydig cells, marked EPO-induced activation of Stat5 also was detected (see Fig. 5). Previous work on isolated rat Leydig cells has shown that EPO can stimulate testosterone production in these cells (56). The direct effect of EPO on testosterone production was reinforced by work from Foresta et al. (57), which indicated that gonadotropins did not mediate EPO’s androgen-producing effect. It is intriguing that the seemingly diverse hormones, PRL and EPO, acting through separate receptors, share Stat5 as a common target in Leydig cells. It will be of interest to determine whether EPO signaling may compensate for PRLR or PRL loss in the respective gene knockout mice and whether PRL and EPO induce the same subset of genes in Leydig cells. In addition, are other signaling pathways than Stat5 coactivated by these two hormones? Based on our analysis of Stat5 activation, no other cells of the testes were responsive to PRL, GH, EPO, or GCSF.
|
-dihydrotestosterone (65, 66). Normal testosterone levels, however, are observed in patients with mild prolactinemia and 30% of patients with clinically defined prolactinomas (67), and no correlation of testosterone levels was noted between micro- and macroprolactinomas before or after treatment (68). Furthermore, analyses of testes in hyperprolactinemic patients revealed histologically normal Leydig cell morphology but altered Sertoli cell and seminiferous tubule morphology (58). This may be a result of local reduction in androgen levels secondary to Leydig cell dysregulation. Significant differences exist in the reproductive physiology between man and rodents, but our acute injection studies revealed only Leydig cells as PRL target cells based on inducible Stat5 activation.
Ovary.
The ovaries undergo a cyclic maturation process in which egg cells are released and rising levels of the ovarian steroids progesterone and estrogen are produced by luteal cells and interstitial cells, respectively. The surface of the ovary is covered by an epithelial cell layer, germinal epithelium, from which the majority of ovarian malignancies originate. In rodents, PRL has both luteotropic and luteolytic effects (34). Receptors for PRL have been described on progesterone-producing luteal cells at the mRNA (69, 70) and protein binding levels (71). mRNA analysis of rat tissues indicated that PRLR levels are among the highest in the ovary and predominantly express of the long receptor form, but specific localization to cell type within the ovary was not made (29). Furthermore, GH also has been reported to affect ovarian function in rats by influencing ovarian follicle development (72), but a direct target cell for GH has not been identified. Interestingly, highly expressing bovine GH transgenic female mice are infertile as a result of deficiency of luteal function (73), an effect that has been suggested to be associated with a loss of the PRL surges (74), whereas rat PRL transgenic mice (75) and hyperprolactinemic mice (76) have less pronounced fertility problems.
After injection of PRL into rats, luteal cells in the ovaries showed a marked and uniformed Stat5 activation response (Table 1). A consistent but moderate Stat5 activation also was detected in estrogen-producing interstitial cells in response to PRL, whereas granulosa, egg, or ovarian epithelial cells did not respond. These observations support the view that luteal cells and interstitial cells are direct targets for PRL, which is consistent with established data regarding PRL stimulating progesterone and estrogen production in rat ovaries (77, 78, 79), such as the conversion of estrone to estradiol by 17ß-hydroxysteroid dehydrogenase (80) and PRL-induced inhibition of progesterone catabolism (81). Furthermore, PRL-induced Stat5 activation has been shown to be involved in estrogen receptor gene regulation (82) as related to corpus luteum formation. Our study did not find evidence of direct GH target cells in the endocrine-associated cell populations of the ovary in rats; however, although the germinal epithelium of the ovary was unresponsive to PRL, GH induced marked Stat5 activation in the germinal epithelium. Likewise, epithelial cells of the ovarian fimbriae showed moderate response to GH but were unresponsive to PRL.
|
|
). The observed modest GH-mediated response might be strengthened by long-term GH administration, because GHR transcript expression in cultured thyroid cells was up-regulated by GH (84). Although we established that rat thyroid cells are direct targets for PRL and GH, further studies are needed to determine whether follicular epithelial cells and parafollicular medullary cells are differentially responsive to hormone treatment because the two cell populations are difficult to distinguish using a light microscope. Dual immunofluorescence of activated Stat5 and thyroglobulin or calcitonin in the same section, or analysis of Stat5 phosphotyrosine and the respective cellular marker in adjacent serial sections, will be needed to resolve this.
Pituitary.
The pituitary consists of the anterior, intermediate, and posterior lobes and produces at least seven major peptide hormones, including PRL and GH. It is remarkable that the rat anterior pituitary has among the highest levels of expression of PRLR mRNA for any of the organs tested (29), suggesting a sensitive autoregulatory mechanism for synthesis and release of the hormone. Based on transcript analysis, the long form of the PRLR is predominant, with nearly a 100:1 ratio to the short form. Likewise, GHR mRNA was localized to the anterior lobe of the pituitary by in situ hybridization (33). We did not, however, detect PRLR inducible activation of Stat5 in the anterior pituitary by PRL or GH administration above the low constitutive basal levels (Table 1) in either male or female rats. In fact, of the endocrine organs analyzed, the pituitary was the only tissue in which we did not detect PRL-inducible Stat5 activation. The lack of effect may be a result of desensitization by high local hormone levels.
In summary, we documented that a wide array of endocrine cells in the rat are directly responsive to PRL and activate Stat5, whereas the same cells generally are unresponsive to GH. The selective induction of PRLR, but not GHR, signaling in these endocrine-related cell types support the concept that PRL may act as a master coordinator of endocrine regulation (Table 1). In fact, throughout pregnancy, serum PRL levels rise until the onset of labor (85), consistent with the need for a coordinated endocrine response during the increased hormonal and metabolic demands of pregnancy and lactation. Furthermore, it is known that PRL, but not GH, is rapidly released during acute stress in both males and females, and PRL is considered a stress hormone (37, 38, 39, 40). PRL therefore also may help provide a coordinated regulation of endocrine glands during stress, similar to the immunomodulatory effects of PRL documented in burn-stressed animals (86). Finally, in humans, some of the many symptoms of hyperprolactinemia may be related to endocrine dysregulation.
Female reproductive system
Mammary gland.
Development and function of the mammary gland is an intricate process that is regulated by many hormones and growth factors, including estrogen, progesterone, insulin, glucocorticoids, GH, and PRL. The terminally differentiated function of the mammary gland, however, depends on the lobuloalveolar development of secretory alveoli during pregnancy and lactation, and this process is critically dependent on PRLR-Jak2-Stat5 signaling (87, 88, 89, 90). Anatomically, rodent mammary glands consist of branching ducts and distal lobuloalveolar compartments embedded in a mammary stromal compartment (91). The mammary epithelia contain centrally located luminal cells and peripherally located myoepithelial cells. Collaborative work from the Kelly, Kopchick, and Hennighausen laboratories (11) used gene deletion mouse models, mammary transplants, and mammary stromal/parenchymal dissection to investigate the roles of PRL and GH in the mammary gland. Based on these studies of dissected tissue extracts, PRL was reported to mainly activate Stat5 in the epithelial compartment, whereas GH had effects in both epithelial and stromal cells. Furthermore, GHRs have been reported in myoepithelial cells of human and canine mammary glands (92, 93). By using an in situ IHC approach to detect nuclear-localized, tyrosine-phosphorylated Stat5, we determined in the present studies that in nonpregnant rats, PRL markedly activated Stat5 in a large proportion (80%) of the luminal mammary epithelial cells over the low basal constitutive Stat5 activation (Fig. 3). This finding is in agreement with previous work from our laboratory, in which we reported low but consistent basal activation of Stat5 in luminal epithelial cells of nonpregnant mouse epithelium. This activation was lost rapidly after hypophysectomy but could be recovered by injection of PRL but only partially by GH (9). Furthermore, basal Stat5 activation remained active throughout the estrus and luminal mammary epithelial tissue from proestrus, estrus, postestrus, and diestrus showed equivalent levels of basal phosphorylation of Stat5. PRL did not activate Stat5 in myoepithelial cells, stromal fibroblasts, or mammary adipocytes, which have been suggested to resemble brown adipose tissue (94), whereas adipocytes from other tissues composed of white adipose tissue were responsive to PRL.
|
15%), a substantial number of myoepithelial cells (50%) were responsive to GH. Therefore, our analysis specifically identifies a subpopulation of mammary epithelial cells with functional GHR. GH also activated Stat5 in mammary adipocytes and fibroblasts (Fig. 3). Furthermore, GH preferentially activated luminal epithelial cells of the ducts over luminal epithelial cells of the alveoli, with an estimated 2- to 3-fold higher frequency in ductal luminal cells. The observation that PRL and GH activate distinct populations of rat mammary epithelial cells was supported by an additive effect of Stat5-responsive cells when a combination of the two hormones were injected (Fig. 3). The preferential responsiveness of myoepithelial cells to GH also is intriguing and merits further study. We previously have shown that in mammary gland transplants from PRLR-null mice that all of the luminal epithelial cells became responsive to GH (9). This may represent compensatory up-regulation of GHR in all of the luminal epithelial cells or selective expansion of the GHR expressing luminal mammary cells. Consistent with GHR up-regulation, mammary explants from PRLR-null mice showed inducible milk protein expression in response to GH (11). Furthermore, human GH typically is referred to as a lactogen because of its ability to bind and activate the PRLR (95). The present work, however, also indicates that a unique GHR contribution in mammary epithelial signaling through Stat5 activation in myoepithelial cells needs to be considered, in addition to indirect effects of GH on epithelia as a result of its effect on the stromal compartment (11, 96). The well-documented effect of GH on early mammary gland development, side branching, and lactation (11, 97, 98, 99) may therefore in part be related to an effect on myoepithelial cells or other progenitor cells. Likewise, further work is warranted on the involvement of GH in breast cancer (92, 100, 101). In addition, similar experimental analysis performed on pregnant rats may also uncover changes in hormone responsiveness of cell populations during gestation and lactation. For instance, due to the significant up-regulation of circulating PRL levels during pregnancy, positive and negative feedback loops may affect the responsiveness of tissues to PRL.
Uterus.
The uterus has been identified as a site of extrapituitary PRL production involving both the endometrial and myometrial cells (102, 103, 104, 105, 106). In addition, studies have suggested that uterine PRL is directly involved in decidualization (107) and maintenance of full-term pregnancy (2) as well as water-electrolyte balance of the amnion (108) and modulating local immune reactions to prevent rejection of the implanting fetus (107). In the nonpregnant rat uterus, PRLR mRNA levels were about the median of tissues tested for total transcript levels; however, the ratio of full length to short isoform was among the highest of any tissue, with only 2% of the population as the short form (29). Little is known about GH effects on the rat uterus. Injection of PRL or GH did not activate Stat5 in the myocytes of the uterine myometrium or the endometrial cells (Table 1), nor did we detect a response in the endometrium to any of the hormones administered. Our study material included endometrial tissue from proliferative and secretory phases, all with negative results for Stat5 activation. PRL is responsible for the maintenance of full-term pregnancy and circulating PRL levels are dramatically elevated during gestation. It would be of interest to determine whether uterine cell types become hormone responsive with respect to Stat5 activation during pregnancy.
Other female reproductive tissues.
We detected a moderate GH-selective response in the deep layers of the stratified squamous epithelium of the vagina. Additionally, analysis of the vaginal preputial sebaceous glands displayed no basal Stat5 activation, and the epithelial acini were unresponsive to PRL treatment. Administration of GH, however, induced moderate Stat5 activation throughout the gland (Fig. 2). This result is consistent with previous work that has identified an effect of GH on female rat preputial glands (109) and the presence of GHR on rat sebaceous glands (110, 111, 112).
Taken together, these data on the female reproductive system indicate an endocrine cell-centric role of PRL, with the notable exception of PRL’s well-documented function as a growth and differentiation factor in the epithelial cells of the mammary gland in rodents. Interestingly, in contrast to the well-documented role of GH in stromal cell types throughout the body, many of the epithelial cells of the female reproductive tissues were responsive to GH.
Male reproductive system
Prostate.
The rodent prostate is an accessory sex gland that consists of three distinct paired lobes and the coagulating glands, which jointly surround the urethra and contribute enzymes, ions, and other molecules to the seminal fluid. Effects of PRL on rat prostate gland physiology have been well documented, including a proliferative effect in conjunction with androgens, especially in the dorsal and lateral lobes (113, 114, 115), citrate production in the differentiated epithelium of the ventral lobe (116, 117), polyamine biosynthesis (118), and numerous other prostate-specific alterations (116). PRL is produced locally in the rat and human prostate gland and probably also signals in an autocrine or paracrine manner to affect prostate physiology (119, 120). For the rat coagulating gland, long-term overexpression of PRL has indicated an increase in gland size (121), but definitive signaling characteristics of this gland have yet to be described. Furthermore, little work has been done with respect to identifying GH signaling in the rat prostate complex, although in humans a few studies have correlated acromegaly with benign prostate hyperplasia (122, 123).
Analysis of Stat5 activation revealed that the basal Stat5 activation state varied between the lobes of the rat prostate (Table 1). The ventral prostate and the coagulating gland displayed no basal staining, whereas the dorsal and lateral prostates showed low staining (+) and moderate (2+) staining, respectively. After PRL injection, the dorsal (Fig. 3) and lateral lobes of the prostate exhibited strong (3+) staining in essentially all of the luminal epithelial cells, whereas the ventral prostate was induced to a moderate intensity of Stat5 activation, indicating that the dorsal and lateral rat prostate lobes are more sensitive to PRL, compared with the ventral prostate. Moreover, these data are consistent with previous work from our laboratories demonstrating PRL-induced Stat5 activation in ex vivo organ cultures of rat prostate (124). Furthermore, the coagulating glands did not respond to either PRL or GH. In fact, GH did not activate Stat5 in cells of of the prostate lobes.
Seminal vesicle.
Another male sex organ of the rat with a documented response to PRL is the seminal vesicle, which secretes fructose and prostaglandins into the semen. Similar to the prostate, PRL and androgens coordinately stimulate the secretory function and morphology of the seminal vesicles (113, 114, 115) and specifically alter the chemical composition of the secretions in both rodents and primates (125, 126, 127). Furthermore, human GH transgenic mice had dramatically enlarged seminal vesicles; however, this effect may have been mediated through the PRLR and not GHR because mice with transgenic expression of the nonlactogenic bovine GH did not show similar hypertrophy (128). Injection of PRL activated Stat5 in luminal epithelial cells of the seminal vesicles to a maximal level (3+), from a low (+) basal activation state. Importantly, acute high-dose GH did not induce Stat5 activation in the seminal vesicle, which supports the concept that hyperplasia in human GH overexpressing transgenic mice is mediated by PRL receptors.
Epididymis.
The epididymis represents a key segment of the excurrent duct system for spermatozoa and serves the dual functions of absorbing fluid from the seminiferous tubules, defective sperm, and other bodies and secreting factors that promote maturation of sperm. PRL has been implicated as a regulator of rat epididymal function, including stimulating energy metabolism (129) and affecting the chemical composition of the secretions (130, 131, 132). A role for GH in epididymal function has not been well defined; however, evidence from rats suggested improved sperm maturation and increased motility after GH treatment (133). In the present study, rat epididymis displayed a low (+) baseline of Stat5 activity, and after PRL administration the luminal epithelial cells became markedly activated (3+). In contrast, GH did not affect Stat5 activation levels in cells of the epididymis. The reported effect of GH on the viability of spermatozoa in the absence of inducible Stat5 activation in GH-injected rats may reflect an indirect biological effect of GH mediated by IGF-I or other secondary factors.
Preputial gland.
The male preputial gland is a sebaceous gland that is well developed in rats, and its secretion is mixed with urine to mark territory and identify members of a group. Previous studies have identified effects of PRL and GH on the male preputial gland in rats. Transplantation of a pituitary under the renal capsule elevated circulating PRL and increased the weight of the male preputial gland (121). Furthermore, GH-mediated signaling effects were differentiation related and not proliferative and resulted in increased secretion when assayed in a rat preputial cell culture model (134). Our analyses revealed no basal Stat5 activation; however, treatment with either PRL or GH induced a strong (3+) and uniform activation of Stat5 in the epithelium throughout the sebaceous gland acini (Fig. 3). These observations suggest that previously reported biological effects of PRL and GH on the male preputial gland are direct.
Other male reproductive tissues.
Although not uniquely related to the reproductive functions of the male rat, the urethral transitional epithelium is involved critically in transporting semen from the reproductive organs during ejaculation. There was no basal or inducible Stat5 activation in the urethral epithelium in response to either PRL or GH injection.
In general, sex steroid-producing cells of gonads in both male and female rats were responsive to PRL but not GH, consistent with the noted broad responsiveness of endocrine cells to PRL. In contrast, nonsteroid producing reproductive tissues in male and female rats showed a divergent responsiveness to PRL and GH. With the exception of the skin-derived preputial glands and mammary glands, male reproductive tissues were preferentially responsive to PRL and not GH, whereas female reproductive tissues were preferentially responsive to GH and not PRL. This differential responsiveness of male and female reproductive tissues to PRL and GH perhaps is related to differences in GH secretory patterns of male and female rats. In male rats, GH secretion is pulsatile and, acting through Stat5, is responsible for sexual dimorphic growth patterns (13, 135). On the other hand, intrinsic differences in male and female reproductive tissues (e.g. gene expression levels) may be a more likely explanation of the differences in responsiveness to PRL and GH.
Digestive system
Liver.
The liver is the largest internal organ and also is the largest glandular tissue mass in mammals. This complex organ has both endocrine and exocrine functions and hence is involved in many physiological processes, including metabolism, digestion, serum protein production, and IGF-I secretion. Absolute PRLR mRNA levels in rat liver are among the highest of any tissue tested (29), and numerous pathways and growth-related genes have been identified that are activated by PRL (34). It should be noted, however, that the majority of PRLR transcripts in rat liver encode the short form of the receptor (29). Additional reports have shown that PRL has a somatogenic effect in the liver, including stimulation of IGF-I production (136, 137). Others have reported that PRL treatment is not able to activate Stat5 in the liver (138), possibly because of the dominant-negative effect of the short PRLR on Stat5 activation by the long PRLR (3, 139). The differences in results between the studies also may be a function of the sensitivity of the assays used. The report that did not detect PRL-induced Stat5 activation in rat liver relied on immunoblotting and EMSAs (138). By way of comparison, previous studies identified basal Stat5 activation in the nonpregnant mammary gland by IHC (9), whereas immunoblotting or EMSA was not sensitive enough to detect the activated Stat5 (140, 141, 142, 143). Although it is generally accepted that Stat5 tyrosine phosphorylation is required for DNA binding, it may not be sufficient, depending on the specific nucleotide sequence of the probe used in an EMSA assay and the presence of additional cofactors that may modulate DNA binding (17). Furthermore in the liver, a very recent study (144) identified low but specific expression of the long form of the PRLR on the cholangiocytes of the intrahepatic bile ducts at the mRNA and protein levels and that regulation of the PRL isoforms was opposite of that detected within hepatocytes. Lastly, a large body of work identified GH-induced activation of Stat5 in the liver and stimulation of IGF-I, which has been established as a mediator of cell and organism growth, and GH-induced liver secretion of IGF-I forms the basis of the long-standing somatomedin hypothesis (145).
In the present study, no Stat5 activation was detectable in hepatocytes of either male or female rats under basal conditions. After PRL or GH injection, however, the majority of hepatocytes exhibited strong Stat5 activation (Fig. 4), and there was no difference in responsiveness between hepatocytes of male and female rats. No additive effect was detected by Stat5-pY staining intensity with the coinjection of PRL and GH in the hepatocytes. The marked activation of Stat5 by PRL was unexpected in light of the predominant expression in rat liver of the short PRLR isoform as determined by RT-PCR. We did not detect responsiveness to PRL or GH in any of the other cell types of the liver, including cholangiocytes in which low levels of the PRLR were recently reported (144). In that report, the authors indicated that the long form of the PRLR was up-regulated by cholestasis, suggesting that under normal conditions, the receptor levels are below threshold levels for detectable response.
|
), without any response in the ductal epithelium. In contrast, GH did not affect Stat5 activation status in any cell type in the parotid gland. The submaxillary gland (also called the submandibular gland) is a mixed salivary gland containing a serous compartment and a mucinous-secreting compartment. Previous studies have reported PRLR mRNA by in situ hybridization in the acini of the submaxillary gland of the fetal rat (146), and PRL has been shown to stimulate secretion of immunoregulatory factors from the submaxillary gland of adult rats (147). With respect to GH, an early study identified significant specific binding of rat GH to isolated membranes from rat submaxillary glands (148), indicating specific presence of the GHR protein. In addition, another report indicated a role for GH in hypertrophy and hyperplasia of the submaxillary gland in dwarf rats and also identified the expression of GHR in the tubular component of the gland (149). Furthermore, GH induced epithelial growth factor production in the submaxillary gland in hypophysectomized rats (150). Interestingly, hormone responsiveness for the submaxillary gland differed somewhat from that of the parotid gland. We detected a moderate induction in response to PRL within the serous compartment (Fig. 4), whereas a low, but consistent, GH-induced activation of Stat5 was detected within the serous acini of the submaxillary gland. Regardless of the hormone administered, we did not detect any hormone responsiveness within the ductal epithelium of the serous compartment of either the parotid or submaxillary gland. Notably, in the mucinous-producing salivary glands, there was no hormone responsiveness for any of the cytokines tested. This included the mucinous component of the submaxillary gland as well as the entire sublingual gland (Table 1). Our studies of rat salivary glands suggest that GH and PRL may directly modulate the function of protein producing serous acini but may not affect the mucus-producing salivary glands.
Exocrine pancreas.
Like the liver, the pancreas is both an exocrine and endocrine organ. The exocrine pancreas consists of serous acini and a ductal tree that secrete digestive enzymes into the gastrointestinal (GI) tract. The responsiveness of endocrine islets of Langerhans to PRL and GH as measured by in situ Stat5 activation was presented under the endocrine system above (Fig. 2). mRNA for PRLR is detectable in homogenates of pancreas tissue (29). Although the effects of PRL and GH on the endocrine pancreas have been studied in great detail, relatively little work has been reported on the exocrine pancreas. In one study, chronic hyperprolactinemia induced by a pituitary graft under the renal capsule was associated with increased proliferation of pancreatic acinar cells and an increase in pancreatic secretions; however, this increase was transient and reverted back to control levels over time (151). In the present study, injection of PRL or GH revealed that exocrine pancreas acinar cells showed a low but direct response to PRL and GH in both male and female rats (Table 1). Furthermore, a combination of the two hormones did not have an additive effect on Stat5-pY staining intensity in the exocrine pancreas.
Alimentary canal.
The alimentary canal is made up of the esophagus plus the GI tract and is responsible for digestion, absorption, movement, and excretion of nutrients and waste. PRL has been reported to induce hyperplasia, increased villus height, and mass of the intestinal mucosa (152, 153). Furthermore, a protective effect against gastric ulcers was observed in chronic hyperprolactinemic rats (154). Effects on growth and changes in metabolism within the intestine by PRL and GH have been reported (155). Also, acromegalic patients have relatively longer overall colon length, and these patients have increased incidence of colon polyp formation, colon cancer incidence, and higher colon cancer mortality (156). Lastly, studies involving EPO also have identified some GI tract-specific effects in the gastric mucosa (157) and small intestine (158); however, in the present study, we did not detect EPO-induced activation of Stat5 in the alimentary canal (Table 1).
Throughout the entire length of the alimentary canal, we detected a low to moderate activation of Stat5 by PRL and/or GH in the mucosal epithelial cells (Table 1). We did not detect basal levels of Stat5 activation in the GI tract, with the exception of low levels in the epithelial cells in the base of the crypts of the colon. Specifically, moderate levels of inducible Stat5 activation were observed in the basal layers of the stratified squamous epithelium of the esophagus in response to GH or PRL treatment. No additive effect on Stat5-pY staining intensity of PRL and GH was detected in these target cells. We also noted a low, but detectable, activation of Stat5 in response to PRL, but not to GH, in the gastric epithelium. The different regions of the small intestine appeared to be somewhat differentially responsive to PRL treatment but uniform in the response to GH. The CEMA tissue arraying technology was particularly effective for systematic analysis of the entire GI tract at high resolution. In particular, Stat5 activation was scored as low in the duodenum in response to both PRL and GH, whereas in the ileum and jejunum PRL induced moderate activation and GH induced low levels of Stat5 activation (Table 1).
Integumentary system
Skin.
As the largest organ in the body, the skin provides a barrier and is important in sensory perception. PRL is known to play a role in various skin functions in many vertebrates. In mammals, PRL has been shown to be involved in the control of proliferation of melanocytes (159) and keratinocytes (160), hair loss in nesting animals (161), and hair growth (162, 163). Absolute PRLR mRNA expression levels in the skin ranked in the middle of the organs tested but notably had among the highest ratio of the full-length variant of the receptor to the short form (29), presumably resulting in more efficient intracellular signaling. GH also is known to have a growth-promoting effect on the skin, and a significant pharmacological benefit is seen in elderly patients receiving supplemental GH (164). Consistent with this, GHR has been identified in the epidermis of rat skin at the transcript level (33) and the protein level (110).
We detected low baseline levels of Stat5 activation in a subset of the basal epidermis cells and marked Stat5 activation in response to PRL throughout this basal layer of the epidermis. GH had a moderate effect in activating Stat5 in the basal epithelial cells (Table 1). We noted a strong (3+) basal activation of Stat5 in the dermal papilla of the hair follicle, but these cells did not show further activation of Stat5 in response to either PRL or GH (Table 1). Local production of PRL and PRLR has been reported in murine and human hair follicle, specifically in the inner and outer root sheath and matrix keratinocytes (165, 166, 167) but not the dermal papilla, suggesting the mechanism of Stat5 activation in the dermal papilla is independent of the autocrine/paracrine PRL-PRLR signaling proposed to exist within the hair follicle. Reduced hair growth, however, was noted in Stat5b-null mice (135), indicating PRL-Stat5b signaling may be involved in growth regulation of hair, presumably by an alternate mechanism than PRL-induced apoptosis of the follicle (167). Clearly, additional work is needed to understand the role of Stat5 and its regulation in hair follicle biology.
Hematopoietic system
Effects of PRL on many aspects of the hematopoietic system have been well documented and reviewed extensively and include effects on lymphoid cells, monocytes, macrophages, and thymic epithelial cells (34, 168, 169). The effects of PRL on these cells have been categorized as redundant and overlapping with other cytokines because PRL and PRLR knockout mice generally have normal immunity and hematopoietic status (1, 170, 171, 172). Similarly, GHR transcript and protein expression also has been identified in both B and T lymphocytes and monocytes (173), and polymorphonuclear leukocytes have been shown to express EPO receptor (EPOR) (174) and GCSF receptor (175) mRNA and specifically bind EPO. Our study did not examine responsiveness of circulating leukocytes to PRL, GH, EPO, or GCSF but included examination of hematopoietic cells in lymphatic organs and bone marrow.
Spleen.
The spleen is the largest lymphatic organ and acts to filter the blood for antigens or defective blood cells. The spleen contains aggregations of B and T lymphocytes, erythrocytes, plasma cells, macrophages, and granulocytes and in rodents also megakaryocytes for platelet production. The spleen has been shown to express PRLR mRNA (29) and GHR mRNA (33), and studies have implicated an important role for PRL signaling in the organ (176, 177). We did not, however, detect definitive activation of Stat5 after PRL treatment, but GH stimulated Stat5 to a low, but detectable, level in an estimated 10% of cells in the white pulp of the spleen (Table 1). EPO strongly activated Stat5 in an estimated 30% of cells in the red pulp of the spleen, presumably erythroblasts (Fig. 5), whereas GCSF had a moderate induction of Stat5 activation in neutrophils in the red pulp and lymphocytes associated with the periarterial lymphatic sheath. Further work is needed to identify which subpopulations of cells within the spleen respond to the individual hormones.
Thymus.
In the thymus, stem lymphocytes are transformed into T lymphocytes, which are important for cell-mediated immunity as well as development of B lymphocytes. Several lines of evidence have suggested a significant role for PRL and GH in the thymus. For example, as reviewed by De Mello-Coelho et al. (178), expression of PRLR and GHR has been identified on both thymocytes and thymic epithelial cells, and both cell types can produce PRL and GH. Furthermore, differentiating T cells are responsive to PRL and GH, and GH administration increases total thymocyte numbers (178). The present study revealed a subset of thymic cells that were responsive to PRL and GH (Table 1). The data suggested that the same cells were responsive to the two hormones because roughly the same percentage (1–3%) and localization (cortex and medulla) of lymphocytes were responsive to either PRL or GH, and no further Stat5-pY staining intensity was detected after combined injection of both hormones.
Bone marrow.
The spaces within long or spongy bones are filled with red bone marrow if there is active hematopoiesis and yellow marrow if there is no active hematopoiesis. The hematopoietic cells mature within the marrow and must actively migrate into the circulation. The role of PRL and GH on bone marrow cells is not clearly defined, although a recent study in human tissue demonstrated that GH can directly and indirectly increase the hematopoietic activity of bone marrow by several mechanisms (179). We analyzed the bone marrow from a long bone (femur) and an irregular bone (vertebra). Although a subpopulation of cells (5%) in the red marrow of femur displayed a moderate response to PRL, no PRL-responsive cells were detected in red bone marrow of the vertebra (Table 1). In contrast, subpopulations of cells from either femur or vertebra bone marrow were equally responsive to GH, displaying a moderate response in approximately 5% of cells. We identified an additive effect of PRL and GH on Stat5-pY signal intensity of responsive cells in the femur but not on cell numbers, suggesting that the same cells were responsive to both PRL and GH. No additional Stat5-pY staining intensity above that induced by GH stimulation alone was detected in the vertebral marrow when PRL and GH were coinjected. Furthermore, vertebral marrow cells showed a marked activation of Stat5 in response to EPO, presumably in erythroblast cells (Fig. 5). A small population of cells in the vertebral marrow exhibited low Stat5 activation in response to GCSF, consistent with granulocyte/monocyte colony-forming cells (data not shown).
Gut-associated lymphatic tissue.
One component of the gut-associated lymphatic tissue is aggregates of lymphatic nodules located in the distal small intestine (ileum) called Peyer’s patches. These nonencapsulated aggregations of lymphocytes are highly reactive when presented with antigens and form germinal centers. Peyer’s patches had a moderate (2+) basal activation status that was enhanced to 3+ with PRL administration, demonstrating the presence of PRL-responsive cells within Peyer’s patches (Table 1). GH treatment did not induce a discernable increase in Stat5 phosphorylation status above basal levels in any population of lymphocytes in Peyer’s patches. Likewise, EPO and GCSF treatment also did not stimulate Stat5 activation above basal levels in the lymphatic aggregations of the ileum and jejunum. Another type of gut-associated lymphatic tissue (GALT) is classified as diffuse lymphatic tissue and is present in the lamina propria. Interestingly, lymphocytes present within in the small intestinal villi generally had consistent moderate levels of basal Stat5 activation but were unresponsive to PRL or GH treatment (Table 1). Because Stat5 activation is associated with general activation of lymphocytes (180), the observations suggest that lymphocytes in the lamina propria are activated.
|
Circulatory system |
|---|
). The endothelial cells of the aorta, however, displayed a moderate response to GH but not PRL. No inducible activation of Stat5 was detected in cardiac cells or cardiac vascular cells in response to EPO or GCSF. Nevertheless, further work is needed to localize EPO-responsive cardiac cells through Stat5-independent pathways. |
Musculoskeletal system |
|---|
 TopAbstractIntroductionMaterials and MethodsResults and DiscussionCirculatory system Musculoskeletal system Nervous systemUri |
|---|
