Estrogen Regulates the Synthesis of Epidermal Growth Factor in Mouse Uterine Epithelial Cells
Yvette M. Huet-Hudson*, Chandan Chakrabortyt, Swapan K. De$, Yasunobu Suzuki§, Glen K. Andrews, and Sudhansu K. Dey Departments of Obstetrics-Gynecology and Physiology (Y.M.H.-H., C.C., Su.K.D.), Biochemistry and Molecular Biology (Sw.K.D., G.K.A.), and Pathology (Y.S.) University of Kansas Medical Center Ralph L. Smith Research Center Kansas City, Kansas 66103
Immunocytochemical analyses, using several mouse epidermal growth factor (EGF) polyclonal antibodies, detected immunoreactivity only in uterine luminal and glandular epithelia on late proestrus, estrus, and early on day 1 of pregnancy, but not late on day 1. This immunoreactivity was not detected in the ovariectomized uterus, but after estrogen stimulation it was detected first in the luminal epithelium between 12-24 h and then also in the glandular epithelium by 48 h. After 72 h of estrogen withdrawal, EGF immunoreactivity was no longer detected. This response was specific for estrogen and did not occur after progesterone injection (2 mg/day for 4 days). Using antipeptide antibodies specific for prepro-EGF, no immunoreactivity was detected in the ovariectomized uterus, weak reactivity was detected in the estrogenized uterus and submandibular gland, and strong reactivity was detected in the kidney. Northern blot analysis of uterine RNA failed to detect the expected 4.8-kilobase prepro-EGF mRNA, but, instead, a rare transcript of 2.4 kilobases was detected, which suggests that EGF mRNA is alternately processed in the uterus. The presence of an EGF-coding uterine transcript was further documented by hybridization of an EGF-coding regionspecific oligodeoxyribonucleotide (oligo) to polymerase chain reaction-amplified uterine cDNA. In situ hybridization, using a prepro-EGF cRNA probe as well as an EGF-coding region-specific oligo, showed hybridization that colocalized with the EGF immunostaining (epithelia) and was absent from non-EGFimmunoreactive cells. Pulse labeling experiments coupled with immunoaffinity chromatography showed that estrogen induced an increase in the relative rate of synthesis of an acid-soluble immunoreactive protein which was the same size as au0888-8809/90/0510-0523$02.00/0 Molecular Endocrinology Copyright © 1990 by The Endocrine Society
thentic EGF. Furthermore, analysis of acid-soluble uterine proteins fractionated by DEAE-cellulose chromatography demonstrated a single coincident peak of antigenic activity and receptor-binding activity which coeluted from the column with authentic EGF. Electron microscopy localized EGF immunoreactivity to the Golgi of luminal epithelial cells. Taken together these results suggest that estrogen regulates expression of the EGF gene specifically in uterine epithelial cells. Increased expression of this gene results in an increase in the relative rate of synthesis of this protein and the accumulation of mature EGF. (Molecular Endocrinology 4: 510-523, 1990)
INTRODUCTION
The uterus is comprised of heterogeneous cell populations that respond uniquely to estrogen and progesterone. In the adult mouse and rat, proliferation of luminal and glandular epithelia occurs in response to estrogen stimulation, whereas stromal cell proliferation is dependent upon both progesterone and estrogen (1). The mechanisms by which estrogen and progesterone differentially regulate proliferation of various uterine cell types are unknown. Detection of various growth factors and their receptors in the uterus in response to ovarian steroids provides circumstantial evidence for the concept that estrogen/progesterone actions in the uterus are mediated in an autocrine/paracrine fashion via cell type-specific expression of growth factors and their receptors (2-5). The proliferative response of uterine epithelial cells to epidermal growth factor (EGF) in vitro (6) as well as the presence of EGF receptors in the uterus (4, 5) suggest that this growth factor could be involved in mediating estrogen-induced uterine growth. EGF is a
510
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
small polypeptide (53 amino acids) with well documented mitogenic and dilferentiating effects (7). EGF is synthesized as a large precursor protein (~130 kDa), called prepro-EGF, from which mature EGF and other EGF-like polypeptides are produced by proteolytic cleavage. In the mouse, the organs with the highest EGF levels are submandibular gland and kidney, but low levels have been detected in a variety of other tissues using immunological methods (8, 9). Mature EGF predominates in the submandibular gland, whereas in the kidney prepro-EGF is most abundant (9). An increase in immunoreactive EGF in the rat uterus was noted after estrogen stimulation (10), and a recent report (3) provides some evidence for the presence of EGF mRNA and prepro-EGF in the immature mouse uterus, even in the absence of estrogen stunulation. To investigate in more detail the potential mechanisms of regulation of EGF gene expression in the mouse uterus, the cell type-specific expression of this gene during proestrus, estrus, and on day 1 of pregnancy and after estrogen stimulation of ovariectomzed mice was analyzed. The results indicate that in uterine luminal and glandular epithelia, estrogen regulates expression of a gene that encodes a protein that is similar, if not identical, to mature EGF. The levels of this protein are regulated by the availability of the mRNA rather than by processing of preexisting preproprotein, and the EGF transcript in uterus may be alternately processed.
RESULTS Estrogen Levels after Ovariectomy and 17/?Estradiol (E2) Implant Placement E2 was not detectable in the serum of ovariectomized mice, but increased by 12 h after placement of the E2 implants (20 ± 7.1 ng/ml; n = 5). The level gradually declined to 8.5 ± 4.9 ng/ml at 24 h and 2.2 ± 0.4 ng/ ml at 48 h (n = 5). A large drop in E2 levels was noted by 48 h after withdrawal of the E2 implants (83.6 ± 38.9 pg/ml; n = 4). Cytochemical Localization of EGF Immunoreactivity Uterine sections incubated in preimmune serum, in the absence of primary antibody, or in primary antibody neutralized with excess antigen showed no nonspecific staining (data not shown). As a positive control, cells in the male mouse submandibular gland showed staining for EGF, whereas staining was absent from cells of the parotid gland (Fig. 1a). All cell types in uterine sections from ovariectomized mice were completely devoid of EGF immunostaining (Fig. 1b). The localization of EGF in the uterus late on proestrus, estrus, and day 1 of pregnancy was limited exclusively to the luminal and glandular epithelial cells. The staining was visible at the time of mating (0 h) and continued to be present until 0800 h in the morning; no staining was observed by
511
1000 h (localization for 0800 h is shown in Fig. 1c). After E2 implant placement in ovariectomized mice, EGF immunostaining was not evident at 12 h (data not shown). However, distinct localization of EGF was observed primarily in the luminal epithelium at 24 h (Fig. 1 d), and by 48 h of E2 treatment both the luminal and glandular epithelia were intensely stained (Fig. 1e). The labeling persisted in these cell types 48 h after E2 implant withdrawal. However, by 72 h the staining was no longer detected (data not shown). These results were reproduced using four separate commercially available polyclonal antibodies produced against mature mouse EGF. Therefore, it can be concluded that EGF immunoreactivity in the uterus is induced by estrogen and is restricted to the luminal and glandular epithelium. This effect is specific for estrogen, as injection of progesterone (2 mg/day) for 4 days did not induce uterine EGF immunoreactivity (data not shown). Cytochemical Localization of Prepro-EGF Immunostaining Antibodies specific to either mature EGF or a carboxylterminal peptide of prepro-EGF (amino acids 11831201) were used for immunocytochemistry. Cells in the female mouse submandibular gland stained positively using antibodies directed against mature EGF, albeit not as intensely as those in the male mouse submandibular gland. However, antibodies specific for preproEGF showed weak staining (compare Fig. 2, a vs. b). On the other hand, in sections of the kidney a similar intensity and pattern of immunolocalization were obtained using antibodies to both mature EGF and preproEGF (compare Fig. 2, c vs. d). Uterine sections from E2treated mice showed strong staining for mature EGF and weak staining for prepro-EGF (compare Fig. 2, e vs. f). Neither antibodies to mature EGF (Fig. 1) nor prepro-EGF reacted with the uterine sections from the ovariectomized mouse (data not shown).
Northern Blot Hybridization Analysis of Uterine RNA To determine whether the uterus is the site of synthesis of EGF-immunoreactive protein, uterine RNA was analyzed by Northern blotting. Male mouse submandibular glands contained high levels of the expected 4.8-kilobase (kb) EGF mRNA transcript, which was readily detectable in 60 ng total RNA (Fig. 3A). In contrast, this 4.8-kb transcript was not detected even in 2 ^g poly(A)+ RNA from the estrogenized uterus. This represents a 1000-fold enrichment for mRNA in the uterine samples relative to those from the submandibular gland. The 4.8-kb EGF transcript was detectable using as little as 6 ng total RNA from the male submandibular gland (data not shown). In the uterine poly(A)+ RNA samples, transcripts of 2.4 and 0.5 kb hybridized with the EGF cRNA probe (Fig. 3A). The smaller transcript was also detected in liver poly(A)+ RNA. Treatment of the hybrid-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 512
Vol 4 No. 3
Fig. 1. Cytochemical Localization of EGF Immunoreactivity in Male Salivary Glands, and the Uterus on Day 1 of Pregnancy as well as after Ovariectomy and E2 Implant Placement a, Male submandibular gland (MSG) and male parotid gland (MPG). Magnification, X100; b, Ovariectomized uterus; c, day 1 uterus (0800 h); d, E2-implanted uterus at 24 h; e, E2-implanted uterus at 48 h (magnification, x200). Note that the parotid gland and ovariectomized uterus did not show localization of EGF. LE, Luminal epithelium; GE, glandular epithelium; S, stroma; CM, circular muscle; LM, longitudinal muscle. Cell nuclei were counterstained with hematoxylin.
ized filter with RNase-A severely reduced the intensity of the 2.4-kb band relative to that of the submandibular gland hybrid and eliminated the 0.5-kb band (Fig. 3B). Although several attempts to obtain sufficient RNA from ovariectomized control uteri failed, the results suggest that the relative intensity of the 2.4-kb band increased after E2 implant placement and declined after withdrawal of the hormone (Fig. 3A). It is unlikely that the 4.8-kb EGF mRNA was degraded during isolation of uterine RNA, because these same RNA samples contained a readily detectable 7.0-kb insulin-like growth factor-l (IGF-I) mRNA transcript (Fig. 3C) which was also E2 dependent, as previously reported (2). The immunocytochemical data suggest that EGF expression is restricted to epithelial cells, which constitute only 3-4% of the total uterine cells (11); thus, Northern blot analysis may not be a sensitive enough method to detect the low abundance prepro-EGF mRNA in total uterine poly(A)+ RNA. However, these results cannot exclude the possibilities that the EGF gene is not transcribed in the uterus or, alternatively, that the uterine prepro-EGF transcript is differently processed. This could account for the RNase-A sensitivity of the 2.4-kb hybrids. However caution must be applied when inter-
preting these data due to the extremely low abundance of these transcripts in the total uterine mRNA preparations. Specific Amplification of Uterine EGF-Coding mRNA Using the Polymerase Chain Reaction (PCR) To determine whether EGF-coding transcripts are present in the uterus, uterine RNA was isolated from ovariectomized mice stimulated with estrogen, and reverse transcribed using an oligo primer complementary to the region of prepro-EGF mRNA coding for amino acids 1045-1038 (12). Mature EGF is encoded in amino acids 977-1029 (12). The resultant cDNAs were amplified via the PCR for 30 cycles using oligo primers complementary to amino acids 957-964 (sense primer) and to amino acids 1045-1038 (antisense primer). The amplified products were analyzed by Southern blot hybridization using an end-labeled oligo (40 bases) probe complementary to the EGF-coding region (amino acids 1015-1028) which is predicted to be amplified in the PCR, since the primers employed flanked this region. As shown in Fig. 4, a single hybridization signal was detected in PCR products generated from estrogenized
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
513
Fig. 2. Cytochemical Localization of Mature EGF and Prepro-EGF Immunoreactivity in the Female Submandibular Gland, Kidney, and E2-lmplanted Uterus a, EGF localization in the submandibular gland; b, prepro-EGF localization in the submandibular gland; c, EGF localization in the kidney; d, prepro-EGF localization in the kidney; e, EGF localization in the E2-implanted uterus at 48 h; f, prepro-EGF localization in the E2-implanted uterus at 48 h. Magnifications were x200. Note that the submandibular gland and E2-treated uterus had minimal staining for prepro-EGF. Cell nuclei were counterstained with hematoxylin.
uterine and submandibular gland RNA. This DNA was of the predicted size [268 basepairs (bp)] for a correctly amplified product, and restriction enzyme digestion also confirmed its authenticity (data not shown). In one experiment this PCR product was also detected in RNA from the ovariectomized uterus (data not shown). No amplification was detected in samples that were not reverse transcribed or in which the RNA was omitted from the PCR reaction, which eliminates the possibility that the PCR product arose from amplification of genomic DNA contamination in these samples. These results establish that the EGF gene is transcribed in the uterus, but do not allow for a quantitative comparison among the uterine RNA samples. Overall, these data are consistent with the concept that prepro-EGF mRNA is alternately spliced in the uterus, but do not provide direct evidence to that effect. In Situ Hybridization Analysis of Uterine Sections To examine the possibility that EGF transcripts were restricted to a subpopulation of the uterine cells, the technique of in situ hybridization was employed. Two different hybridization probes were used in these ex-
periments: the cRNA probe used for the Northern blots, which is complementary to a portion of the prepro-EGFcoding region (amino acids 634-844), and an oligo (40 bases) complementary to the mature EGF-coding region (amino acids 1015-1028). Samples hybridized with the EGF cRNA probe were treated with RNase-A after the hybridization to ensure that only stable hybrids remained, and the oligo hybrids were carried out under high stringency conditions (50% formamide and 0.3 M NaCI; 42 C) (13) to ensure the specificity of the hybridization. Each of these probes hybridized in a cell typespecific manner in the male mouse submandibular gland, and the hybrids colocalized with EGF protein (Fig. 5). In uterine sections, specific autoradiographic signals were not evident in samples from ovariectomized mice (Fig. 6, a and b). However, on day 1 of pregnancy, hybrids were detected specifically in luminal and glandular epithelial cells from 0-0600 h, after which no specific signals were observed. Both the 35S-labeled cRNA (Fig. 6, c, d, and e) and the oligo probe (Fig. 6, f, g, and h) showed similar patterns of localization (hybridization for 0600 h is shown in Fig. 6). In the E2-treated mice, hybrids were detected in the uterine luminal epithelial cells at 24 h (Fig. 6, i and j), and by 48 h both
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 514
1 2 i
Vol 4 No. 3
3 4 5 I
I
i
i
EGF
bP
Fig. 3. Northern Blot Detection of EGF and IGF-I mRNAs in the Male Mouse Submandibular Glands, Adult Liver, and E2Treated Uteri RNA samples were separated by formaldehyde agarose gel electrophoresis and Northern blotted to nitrocellulose filters. The filters were hybridized with 32P-labeled cRNA probes for EGF (A and B) and IGF-I mRNAs (C). The hybrids were detected by autoradiography. The filter that was probed for EGF mRNA was subsequently treated with RNase-A (10 ^g/ ml; 10 min at 37 C), and autoradiography was again performed (B). A and B samples were as follows: lane 1, total RNA from male mouse submandibular gland (66 ng); lane 2, poly(A)+ RNA (2 ng) from liver; lanes 3 and 4, poly(A)+ RNA (2 ng) from uteri of mice taken 24 and 48 h after E2 implant placement, respectively; lane 5, poly(A)+ RNA (2 ng) from uteri taken 24 h after E2 implant withdrawal. C samples were as follows: lanes 1 and 2, poly(A)+ RNA (2 ^g) from uteri of mice taken 24 and 48 h after E2 implant placement, respectively; lanes 3 and 4, poly(A)+ RNA (2 /ug) from uteri taken 6 and 24 h after E2 implant withdrawal, respectively. Lane 5, Liver RNA.
the luminal and glandular epithelia were unquestionably labeled (Fig. 6, k and I). Negative controls for this technique included a lack of hybridization of EGF probes in other uterine cells (stroma and muscle) or in the ovariectomized uterine epithelia as well as a lack of detectable cell type-specific hybridization in the day 1 uterus using the G/C-rich metallothionein-l cRNA probe (12). The detection of cell-specific hybrids using the EGF oligo probe under stringent hybridization conditions strongly suggests that mRNA coding for authentic mature EGF was detected. This supports the concept that epithelial cells are the source of synthesis of this growth factor in the uterus. Synthesis of EGF-lmmunoreactive Protein by Uterine Explants To determine more directly whether estrogen can stimulate the synthesis of EGF-immunoreactive protein in the uterus, uterine explants from ovariectomized or E2-
1 2 3 4 I I I I
123Uterus MSG Fig. 4. Amplification and Detection of Uterine EGF-Coding Transcripts Using the PCR Samples of total RNA from the estrogenized uterus and male submandibular gland (MSG) were reverse transcribed using an antisense oligo primer specific for a region of preproEGF mRNA located just 3' to the EGF-coding sequence. The reverse transcription reaction products were amplified for 30 cycles of PCR using oligo primers which flanked the EGFcoding region of prepro-EGF mRNA. The PCR amplification products were separated by 2% agarose gel electrophoresis and Southern blotted to nylon membranes. Membranes were hybridized with a 32P-labeled oligo probe complementary to the EGF-coding region, and hybrids were detected by autoradiography. PCR samples were as follows: lanes 1, 2, and 3, uteri of mice taken 12,24, and 48 h after E2 implant placement, respectively; lane 4, male submandibular gland.
implanted mice were pulse labeled with [35S]cysteine and [35S]methionine. Acid-soluble uterine proteins were then subjected to affinity chromatography, using immobilized EGF antibodies. While about 0.01% of the total labeled acid-soluble protein was retained by the immunoaffinity column in the case of ovariectomized mice, this value increased about 6-fold in E2-treated animals. Equal amounts of radioactivity (7500 cpm) in the bound proteins from the control and E2-treated samples were fractionated by 20% sodium dodecyl sulfate-polyacrylamidegel electrophoresis and detected by fluorography (Fig. 7). The bound labeled proteins from E2-treated uteri migrated as a single radioactive band whose relative mobility, in this sample, was slightly less than that of purified [125I]EGF (Fig. 7, lane 2). A very faint diffuse autoradiographic signal in the same region of the gel was also detected in the bound sample from the ovariectomized mice (Fig. 7, lane 3). The subtle mobility difference between the uterine preparation and the authentic EGF is due to sample preparation, as
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
515
m
. i ' «i .. it.^/ ...
1
!
i*»
?
.
»J
Yvette M. Huet-Hudson*, Chandan Chakrabortyt, Swapan K. De$, Yasunobu Suzuki§, Glen K. Andrews, and Sudhansu K. Dey Departments of Obstetrics-Gynecology and Physiology (Y.M.H.-H., C.C., Su.K.D.), Biochemistry and Molecular Biology (Sw.K.D., G.K.A.), and Pathology (Y.S.) University of Kansas Medical Center Ralph L. Smith Research Center Kansas City, Kansas 66103
Immunocytochemical analyses, using several mouse epidermal growth factor (EGF) polyclonal antibodies, detected immunoreactivity only in uterine luminal and glandular epithelia on late proestrus, estrus, and early on day 1 of pregnancy, but not late on day 1. This immunoreactivity was not detected in the ovariectomized uterus, but after estrogen stimulation it was detected first in the luminal epithelium between 12-24 h and then also in the glandular epithelium by 48 h. After 72 h of estrogen withdrawal, EGF immunoreactivity was no longer detected. This response was specific for estrogen and did not occur after progesterone injection (2 mg/day for 4 days). Using antipeptide antibodies specific for prepro-EGF, no immunoreactivity was detected in the ovariectomized uterus, weak reactivity was detected in the estrogenized uterus and submandibular gland, and strong reactivity was detected in the kidney. Northern blot analysis of uterine RNA failed to detect the expected 4.8-kilobase prepro-EGF mRNA, but, instead, a rare transcript of 2.4 kilobases was detected, which suggests that EGF mRNA is alternately processed in the uterus. The presence of an EGF-coding uterine transcript was further documented by hybridization of an EGF-coding regionspecific oligodeoxyribonucleotide (oligo) to polymerase chain reaction-amplified uterine cDNA. In situ hybridization, using a prepro-EGF cRNA probe as well as an EGF-coding region-specific oligo, showed hybridization that colocalized with the EGF immunostaining (epithelia) and was absent from non-EGFimmunoreactive cells. Pulse labeling experiments coupled with immunoaffinity chromatography showed that estrogen induced an increase in the relative rate of synthesis of an acid-soluble immunoreactive protein which was the same size as au0888-8809/90/0510-0523$02.00/0 Molecular Endocrinology Copyright © 1990 by The Endocrine Society
thentic EGF. Furthermore, analysis of acid-soluble uterine proteins fractionated by DEAE-cellulose chromatography demonstrated a single coincident peak of antigenic activity and receptor-binding activity which coeluted from the column with authentic EGF. Electron microscopy localized EGF immunoreactivity to the Golgi of luminal epithelial cells. Taken together these results suggest that estrogen regulates expression of the EGF gene specifically in uterine epithelial cells. Increased expression of this gene results in an increase in the relative rate of synthesis of this protein and the accumulation of mature EGF. (Molecular Endocrinology 4: 510-523, 1990)
INTRODUCTION
The uterus is comprised of heterogeneous cell populations that respond uniquely to estrogen and progesterone. In the adult mouse and rat, proliferation of luminal and glandular epithelia occurs in response to estrogen stimulation, whereas stromal cell proliferation is dependent upon both progesterone and estrogen (1). The mechanisms by which estrogen and progesterone differentially regulate proliferation of various uterine cell types are unknown. Detection of various growth factors and their receptors in the uterus in response to ovarian steroids provides circumstantial evidence for the concept that estrogen/progesterone actions in the uterus are mediated in an autocrine/paracrine fashion via cell type-specific expression of growth factors and their receptors (2-5). The proliferative response of uterine epithelial cells to epidermal growth factor (EGF) in vitro (6) as well as the presence of EGF receptors in the uterus (4, 5) suggest that this growth factor could be involved in mediating estrogen-induced uterine growth. EGF is a
510
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
small polypeptide (53 amino acids) with well documented mitogenic and dilferentiating effects (7). EGF is synthesized as a large precursor protein (~130 kDa), called prepro-EGF, from which mature EGF and other EGF-like polypeptides are produced by proteolytic cleavage. In the mouse, the organs with the highest EGF levels are submandibular gland and kidney, but low levels have been detected in a variety of other tissues using immunological methods (8, 9). Mature EGF predominates in the submandibular gland, whereas in the kidney prepro-EGF is most abundant (9). An increase in immunoreactive EGF in the rat uterus was noted after estrogen stimulation (10), and a recent report (3) provides some evidence for the presence of EGF mRNA and prepro-EGF in the immature mouse uterus, even in the absence of estrogen stunulation. To investigate in more detail the potential mechanisms of regulation of EGF gene expression in the mouse uterus, the cell type-specific expression of this gene during proestrus, estrus, and on day 1 of pregnancy and after estrogen stimulation of ovariectomzed mice was analyzed. The results indicate that in uterine luminal and glandular epithelia, estrogen regulates expression of a gene that encodes a protein that is similar, if not identical, to mature EGF. The levels of this protein are regulated by the availability of the mRNA rather than by processing of preexisting preproprotein, and the EGF transcript in uterus may be alternately processed.
RESULTS Estrogen Levels after Ovariectomy and 17/?Estradiol (E2) Implant Placement E2 was not detectable in the serum of ovariectomized mice, but increased by 12 h after placement of the E2 implants (20 ± 7.1 ng/ml; n = 5). The level gradually declined to 8.5 ± 4.9 ng/ml at 24 h and 2.2 ± 0.4 ng/ ml at 48 h (n = 5). A large drop in E2 levels was noted by 48 h after withdrawal of the E2 implants (83.6 ± 38.9 pg/ml; n = 4). Cytochemical Localization of EGF Immunoreactivity Uterine sections incubated in preimmune serum, in the absence of primary antibody, or in primary antibody neutralized with excess antigen showed no nonspecific staining (data not shown). As a positive control, cells in the male mouse submandibular gland showed staining for EGF, whereas staining was absent from cells of the parotid gland (Fig. 1a). All cell types in uterine sections from ovariectomized mice were completely devoid of EGF immunostaining (Fig. 1b). The localization of EGF in the uterus late on proestrus, estrus, and day 1 of pregnancy was limited exclusively to the luminal and glandular epithelial cells. The staining was visible at the time of mating (0 h) and continued to be present until 0800 h in the morning; no staining was observed by
511
1000 h (localization for 0800 h is shown in Fig. 1c). After E2 implant placement in ovariectomized mice, EGF immunostaining was not evident at 12 h (data not shown). However, distinct localization of EGF was observed primarily in the luminal epithelium at 24 h (Fig. 1 d), and by 48 h of E2 treatment both the luminal and glandular epithelia were intensely stained (Fig. 1e). The labeling persisted in these cell types 48 h after E2 implant withdrawal. However, by 72 h the staining was no longer detected (data not shown). These results were reproduced using four separate commercially available polyclonal antibodies produced against mature mouse EGF. Therefore, it can be concluded that EGF immunoreactivity in the uterus is induced by estrogen and is restricted to the luminal and glandular epithelium. This effect is specific for estrogen, as injection of progesterone (2 mg/day) for 4 days did not induce uterine EGF immunoreactivity (data not shown). Cytochemical Localization of Prepro-EGF Immunostaining Antibodies specific to either mature EGF or a carboxylterminal peptide of prepro-EGF (amino acids 11831201) were used for immunocytochemistry. Cells in the female mouse submandibular gland stained positively using antibodies directed against mature EGF, albeit not as intensely as those in the male mouse submandibular gland. However, antibodies specific for preproEGF showed weak staining (compare Fig. 2, a vs. b). On the other hand, in sections of the kidney a similar intensity and pattern of immunolocalization were obtained using antibodies to both mature EGF and preproEGF (compare Fig. 2, c vs. d). Uterine sections from E2treated mice showed strong staining for mature EGF and weak staining for prepro-EGF (compare Fig. 2, e vs. f). Neither antibodies to mature EGF (Fig. 1) nor prepro-EGF reacted with the uterine sections from the ovariectomized mouse (data not shown).
Northern Blot Hybridization Analysis of Uterine RNA To determine whether the uterus is the site of synthesis of EGF-immunoreactive protein, uterine RNA was analyzed by Northern blotting. Male mouse submandibular glands contained high levels of the expected 4.8-kilobase (kb) EGF mRNA transcript, which was readily detectable in 60 ng total RNA (Fig. 3A). In contrast, this 4.8-kb transcript was not detected even in 2 ^g poly(A)+ RNA from the estrogenized uterus. This represents a 1000-fold enrichment for mRNA in the uterine samples relative to those from the submandibular gland. The 4.8-kb EGF transcript was detectable using as little as 6 ng total RNA from the male submandibular gland (data not shown). In the uterine poly(A)+ RNA samples, transcripts of 2.4 and 0.5 kb hybridized with the EGF cRNA probe (Fig. 3A). The smaller transcript was also detected in liver poly(A)+ RNA. Treatment of the hybrid-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 512
Vol 4 No. 3
Fig. 1. Cytochemical Localization of EGF Immunoreactivity in Male Salivary Glands, and the Uterus on Day 1 of Pregnancy as well as after Ovariectomy and E2 Implant Placement a, Male submandibular gland (MSG) and male parotid gland (MPG). Magnification, X100; b, Ovariectomized uterus; c, day 1 uterus (0800 h); d, E2-implanted uterus at 24 h; e, E2-implanted uterus at 48 h (magnification, x200). Note that the parotid gland and ovariectomized uterus did not show localization of EGF. LE, Luminal epithelium; GE, glandular epithelium; S, stroma; CM, circular muscle; LM, longitudinal muscle. Cell nuclei were counterstained with hematoxylin.
ized filter with RNase-A severely reduced the intensity of the 2.4-kb band relative to that of the submandibular gland hybrid and eliminated the 0.5-kb band (Fig. 3B). Although several attempts to obtain sufficient RNA from ovariectomized control uteri failed, the results suggest that the relative intensity of the 2.4-kb band increased after E2 implant placement and declined after withdrawal of the hormone (Fig. 3A). It is unlikely that the 4.8-kb EGF mRNA was degraded during isolation of uterine RNA, because these same RNA samples contained a readily detectable 7.0-kb insulin-like growth factor-l (IGF-I) mRNA transcript (Fig. 3C) which was also E2 dependent, as previously reported (2). The immunocytochemical data suggest that EGF expression is restricted to epithelial cells, which constitute only 3-4% of the total uterine cells (11); thus, Northern blot analysis may not be a sensitive enough method to detect the low abundance prepro-EGF mRNA in total uterine poly(A)+ RNA. However, these results cannot exclude the possibilities that the EGF gene is not transcribed in the uterus or, alternatively, that the uterine prepro-EGF transcript is differently processed. This could account for the RNase-A sensitivity of the 2.4-kb hybrids. However caution must be applied when inter-
preting these data due to the extremely low abundance of these transcripts in the total uterine mRNA preparations. Specific Amplification of Uterine EGF-Coding mRNA Using the Polymerase Chain Reaction (PCR) To determine whether EGF-coding transcripts are present in the uterus, uterine RNA was isolated from ovariectomized mice stimulated with estrogen, and reverse transcribed using an oligo primer complementary to the region of prepro-EGF mRNA coding for amino acids 1045-1038 (12). Mature EGF is encoded in amino acids 977-1029 (12). The resultant cDNAs were amplified via the PCR for 30 cycles using oligo primers complementary to amino acids 957-964 (sense primer) and to amino acids 1045-1038 (antisense primer). The amplified products were analyzed by Southern blot hybridization using an end-labeled oligo (40 bases) probe complementary to the EGF-coding region (amino acids 1015-1028) which is predicted to be amplified in the PCR, since the primers employed flanked this region. As shown in Fig. 4, a single hybridization signal was detected in PCR products generated from estrogenized
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
513
Fig. 2. Cytochemical Localization of Mature EGF and Prepro-EGF Immunoreactivity in the Female Submandibular Gland, Kidney, and E2-lmplanted Uterus a, EGF localization in the submandibular gland; b, prepro-EGF localization in the submandibular gland; c, EGF localization in the kidney; d, prepro-EGF localization in the kidney; e, EGF localization in the E2-implanted uterus at 48 h; f, prepro-EGF localization in the E2-implanted uterus at 48 h. Magnifications were x200. Note that the submandibular gland and E2-treated uterus had minimal staining for prepro-EGF. Cell nuclei were counterstained with hematoxylin.
uterine and submandibular gland RNA. This DNA was of the predicted size [268 basepairs (bp)] for a correctly amplified product, and restriction enzyme digestion also confirmed its authenticity (data not shown). In one experiment this PCR product was also detected in RNA from the ovariectomized uterus (data not shown). No amplification was detected in samples that were not reverse transcribed or in which the RNA was omitted from the PCR reaction, which eliminates the possibility that the PCR product arose from amplification of genomic DNA contamination in these samples. These results establish that the EGF gene is transcribed in the uterus, but do not allow for a quantitative comparison among the uterine RNA samples. Overall, these data are consistent with the concept that prepro-EGF mRNA is alternately spliced in the uterus, but do not provide direct evidence to that effect. In Situ Hybridization Analysis of Uterine Sections To examine the possibility that EGF transcripts were restricted to a subpopulation of the uterine cells, the technique of in situ hybridization was employed. Two different hybridization probes were used in these ex-
periments: the cRNA probe used for the Northern blots, which is complementary to a portion of the prepro-EGFcoding region (amino acids 634-844), and an oligo (40 bases) complementary to the mature EGF-coding region (amino acids 1015-1028). Samples hybridized with the EGF cRNA probe were treated with RNase-A after the hybridization to ensure that only stable hybrids remained, and the oligo hybrids were carried out under high stringency conditions (50% formamide and 0.3 M NaCI; 42 C) (13) to ensure the specificity of the hybridization. Each of these probes hybridized in a cell typespecific manner in the male mouse submandibular gland, and the hybrids colocalized with EGF protein (Fig. 5). In uterine sections, specific autoradiographic signals were not evident in samples from ovariectomized mice (Fig. 6, a and b). However, on day 1 of pregnancy, hybrids were detected specifically in luminal and glandular epithelial cells from 0-0600 h, after which no specific signals were observed. Both the 35S-labeled cRNA (Fig. 6, c, d, and e) and the oligo probe (Fig. 6, f, g, and h) showed similar patterns of localization (hybridization for 0600 h is shown in Fig. 6). In the E2-treated mice, hybrids were detected in the uterine luminal epithelial cells at 24 h (Fig. 6, i and j), and by 48 h both
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
MOL ENDO-1990 514
1 2 i
Vol 4 No. 3
3 4 5 I
I
i
i
EGF
bP
Fig. 3. Northern Blot Detection of EGF and IGF-I mRNAs in the Male Mouse Submandibular Glands, Adult Liver, and E2Treated Uteri RNA samples were separated by formaldehyde agarose gel electrophoresis and Northern blotted to nitrocellulose filters. The filters were hybridized with 32P-labeled cRNA probes for EGF (A and B) and IGF-I mRNAs (C). The hybrids were detected by autoradiography. The filter that was probed for EGF mRNA was subsequently treated with RNase-A (10 ^g/ ml; 10 min at 37 C), and autoradiography was again performed (B). A and B samples were as follows: lane 1, total RNA from male mouse submandibular gland (66 ng); lane 2, poly(A)+ RNA (2 ng) from liver; lanes 3 and 4, poly(A)+ RNA (2 ng) from uteri of mice taken 24 and 48 h after E2 implant placement, respectively; lane 5, poly(A)+ RNA (2 ng) from uteri taken 24 h after E2 implant withdrawal. C samples were as follows: lanes 1 and 2, poly(A)+ RNA (2 ^g) from uteri of mice taken 24 and 48 h after E2 implant placement, respectively; lanes 3 and 4, poly(A)+ RNA (2 /ug) from uteri taken 6 and 24 h after E2 implant withdrawal, respectively. Lane 5, Liver RNA.
the luminal and glandular epithelia were unquestionably labeled (Fig. 6, k and I). Negative controls for this technique included a lack of hybridization of EGF probes in other uterine cells (stroma and muscle) or in the ovariectomized uterine epithelia as well as a lack of detectable cell type-specific hybridization in the day 1 uterus using the G/C-rich metallothionein-l cRNA probe (12). The detection of cell-specific hybrids using the EGF oligo probe under stringent hybridization conditions strongly suggests that mRNA coding for authentic mature EGF was detected. This supports the concept that epithelial cells are the source of synthesis of this growth factor in the uterus. Synthesis of EGF-lmmunoreactive Protein by Uterine Explants To determine more directly whether estrogen can stimulate the synthesis of EGF-immunoreactive protein in the uterus, uterine explants from ovariectomized or E2-
1 2 3 4 I I I I
123Uterus MSG Fig. 4. Amplification and Detection of Uterine EGF-Coding Transcripts Using the PCR Samples of total RNA from the estrogenized uterus and male submandibular gland (MSG) were reverse transcribed using an antisense oligo primer specific for a region of preproEGF mRNA located just 3' to the EGF-coding sequence. The reverse transcription reaction products were amplified for 30 cycles of PCR using oligo primers which flanked the EGFcoding region of prepro-EGF mRNA. The PCR amplification products were separated by 2% agarose gel electrophoresis and Southern blotted to nylon membranes. Membranes were hybridized with a 32P-labeled oligo probe complementary to the EGF-coding region, and hybrids were detected by autoradiography. PCR samples were as follows: lanes 1, 2, and 3, uteri of mice taken 12,24, and 48 h after E2 implant placement, respectively; lane 4, male submandibular gland.
implanted mice were pulse labeled with [35S]cysteine and [35S]methionine. Acid-soluble uterine proteins were then subjected to affinity chromatography, using immobilized EGF antibodies. While about 0.01% of the total labeled acid-soluble protein was retained by the immunoaffinity column in the case of ovariectomized mice, this value increased about 6-fold in E2-treated animals. Equal amounts of radioactivity (7500 cpm) in the bound proteins from the control and E2-treated samples were fractionated by 20% sodium dodecyl sulfate-polyacrylamidegel electrophoresis and detected by fluorography (Fig. 7). The bound labeled proteins from E2-treated uteri migrated as a single radioactive band whose relative mobility, in this sample, was slightly less than that of purified [125I]EGF (Fig. 7, lane 2). A very faint diffuse autoradiographic signal in the same region of the gel was also detected in the bound sample from the ovariectomized mice (Fig. 7, lane 3). The subtle mobility difference between the uterine preparation and the authentic EGF is due to sample preparation, as
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 15 October 2015. at 03:52 For personal use only. No other uses without permission. . All rights reserved.
EGF in the Uterus
515
m
. i ' «i .. it.^/ ...
1
!
i*»
?
.
»J
