| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Studies |
and ß Isoforms in Human Pituitary Tumors1
Division of Endocrinology and Metabolism, Department of Internal Medicine (M.A.S., L.K.P., A.Y.S., A.A.); Department of Neuropathology (M.B.L.); and Department of Neurosurgery (E.R.L.), University of Virginia Health Science Center, Charlottesville, Virginia 22908
Address all correspondence and requests for reprints to: Margaret A. Shupnik, Ph.D., Box 578 HSC, University of Virginia, Charlottesville, Virginia 22908. E-mail: mas3x{at}virginia.edu
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and ERß) exist, with distinct tissue and cell
patterns of expression. Additional ER isoforms, generated by
alternative messenger ribonucleic acid (mRNA) exon splicing, have been
defined in several tissues and are postulated to play a role in
tumorigenesis or in modulating the estrogen response. We examined 71
human pituitary adenomas of varying phenotypes and 6 normal pituitary
specimens for ER mRNA forms by RT-PCR and hybridization blotting
analysis. All prolactinomas (n = 14) contained ER, and several
contained ERß (5 of 14) mRNA. In comparison, 6 tumors that expressed
PRL and GH expressed ERß (4 of 6) more frequently than ER (3 of
6). ERß mRNA was also found more frequently in null cell (8 of 24
ER and 14 of 24 ERß) and gonadotrope (13 of 21 ER and 18 of 21
ERß) tumors. Additionally, ERß was found in 4 of 6 tumors that
contained only GH, although ER was not observed in this tumor type.
Expression of the two ER forms within a tumor type was overlapping, but
some tumors contained only 1 isoform. Expression of ER mRNA splice
variants also varied with cell type. All normal pituitaries contained
ER deletions of exon 4, 5, and 7, whereas only 2 of 6 samples
contained the exon 2 deletion variant. Although the same ER mRNA
variants were observed among the various tumor types, the proportion of
specific splice variants expressed varied. For example, most
ER-positive prolactinomas expressed ER variants with deletions of
exon 2, 4, or 5, whereas gonadotropin tumors preferentially expressed
the ER exon 7 deletion variant. A novel ERß mRNA splice variant,
missing exon 2, was observed in a majority of all ERß-positive
tumors. Immunoblotting analysis of ER and ERß proteins supported
the mRNA results. Because ER and ERß have different biological
responses to selective ER modulators, and the ER deletion variants have
biological effects distinct from those of the full-length ER,
expression of these isoforms may influence the biological properties of
these tumors and affect their ability to respond to estrogen and
antiestrogen therapies. |
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Several groups have used a variety of techniques to detect and localize
the ER messenger ribonucleic acid (mRNA) and protein to specific
cell types in normal pituitary tissue and in distinct classes of
pituitary adenomas (10, 11, 12, 13), which represent clonal expansions of
individual cells (14). In general, the highest levels of ER mRNA and
protein and the highest percentages of ER-containing tumors have
been among PRL-containing tumors, with or without GH (70–100%),
whereas gonadotrope tumors have intermediate levels of protein and a
lower (
50%) percentage of ER-positive tumors. Null tumors with
ER were less frequent, and GH tumors were consistently ER
negative in several studies (10, 11, 12, 13). These results are in agreement
with information obtained from normal human pituitary specimens,
indicating that most PRL and gonadotrope cells have ER, but that
GH-containing cells do not (10, 11). These studies do not, however,
completely explain the uptake of radioactive E2
into GH-containing cells previously described by some investigators or
the effects of E2 on GH secretion (13, 15), and
indirect mechanisms of E2 action have been
proposed. Alternatively, the demonstration of an ER-positive
GH-containing tumor by isoelectric focusing (16) and of positive
ER- and GH-containing cells by in situ hybridization (17)
may indicate the rare occurrence of some GH- and ER-positive
cells.
The identification of ER variants, generated by alternative splicing of
the mRNA encoding the protein, in hormone-responsive tissues such as
breast tumors and breast cancer cell lines has provided an additional
layer of complexity. The ER has a defined domain structure, with a
central DNA-binding region, a C-terminal hormone-binding area, and
activating functions conferring the ability to stimulate gene
transcription located in both N- and C-terminal areas (18) (Fig. 1
). Removal of individual exons by
alternative splicing events can thus have significant impact on
receptor function and can affect the ability of the full-length ER to
exert its biological effects. For example, elimination of ER exon 5
(ER
5) results in a truncated protein lacking much of the
hormone-binding domain. In breast cancer cells, this variant protein
has constitutive activity independent of E2 or
antiestrogen binding and may play a role in cellular growth or the
appearance of antiestrogen-resistant cells (19, 20). Other variants,
lacking part of the DNA-binding domain in exon 3 (ER3) or the
activation region in the C-terminus (ER7) have dominant negative
effects on full-length receptor activation of gene transcription in
transfection studies (21, 22). A recent report demonstrated that the
exon 2 deletion variant (ER2) could amplify the ability of the
full-length ER to stimulate gene transcription (23). A previous study
scanned a number of pituitary tumors and suggested that the expression
of some ER splice variants may be pituitary cell type-specific,
allowing potential variation in ER activity in prolactinomas and
gonadotrope tumors (12).
|
)
first identified, but has significant differences in the ability to
bind and affect gene transcription in response to specific
antiestrogens, selective estrogen receptor modulators, and potential
environmental estrogens (27, 28, 29). These data suggest that the
expression of the ERß isoform could have significant impact on the
ability of a specific cell type or tumor to respond to these clinically
relevant compounds. Although ERß mRNA has been detected in rat
pituitary (26), pituitary cell type-specific expression has not been
examined in rodent or human pituitary.
We examined 71 human pituitary adenomas and 6 normal pituitary samples
for the presence of both ER isoforms, ER and ERß. We found
selective tumor-type expression of three ER variants, with most PRL
tumors expressing ER variants with activating (ER2) or
constitutively active (ER5) forms, whereas more gonadotropin tumors
expressed a dominant negative form (ER7) of the ER. Surprisingly, a
majority of GH-only tumors were found to contain mRNA for the ERß
isoform, suggesting that some subset of these tumors may be hormonally
responsive.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Pituitary macroadenomas were obtained from 71 patients who had
transphenoidal surgery. Adenomas were classified according to current
clinical morphological classification (30), using both
immunohistochemistry and transmission electron microscopy analysis.
However, the distinction between male and female type gonadotrope
adenomas was not made. Immunohistochemistry was performed as previously
described (10, 28) and correlated to serum hormone levels. Patients
with prolactinomas (n = 14; 11 women and 4 men) ranged in age from
24–67 yr (median, 33 yr). All had elevated serum PRL levels (from
150->2000 µg/L) and had tumor immunohistochemical staining
consistent with expression of PRL, but not the gonadotropin subunits,
ACTH, or TSH. Six tumors stained for PRL and GH. All patients with
PRL-staining tumors (alone or with GH) had received at least 1 trial of
bromocryptine therapy. Only 1 patient was receiving bromocryptine
therapy at the time of surgery. The tumor from this male patient
stained for PRL only and was positive for ER, but not ERß, similar
to other tumors of this type. Six tumors (2 women and 4 men) were
obtained in which exclusive GH immunostaining was observed; 1 patient
had the occasional cell that also stained for -subunit. These
patients ranged in age from 31–60 yr (median, 39 yr), with elevated
serum GH (10–57 µg/L) and insulin-like growth factor I levels
(529–1179 ng/mL). No patient in this study had been treated with
octreotide. Patients with clinically nonfunctioning tumors were
separated into 2 groups, null and gonadotrope, on the basis of
histology, immunostaining, and serum hormone levels. Patients with
gonadotrope tumors (n = 21, 14 men and 7 women) ranged in age from
26–80 yr (median age, 64 yr). Immunohistochemical staining was
observed in all tumors with one or more specific antibodies against
-subunit, LHß, and FSHß, and one tumor had focal staining for
TSHß. Patients with null cell tumors (n = 24, 20 men and 4
women) ranged in age from 33–81 yr (median, 44 yr). These tumors had
no immunohistochemical staining or focal staining with -subunit
only. Normal pituitary tissue was obtained from 6 individuals (3 women,
aged 26–78 yr, and 3 men, aged 28–70 yr; median age, 46 yr). All
tissue was obtained in accordance with the guidelines of the human
investigation committee at the University of Virginia.
RNA isolation
As previously described, pituitary specimens were obtained and partitioned immediately after surgical removal for routine histological examination and RNA isolation. Normal pituitary tissue was obtained 5–14 h after death. Tissue was homogenized in guanidinium isothiocyanate, and total RNA was isolated by centrifugation through a cesium chloride gradient (31, 32). RNA integrity was assessed by observation of intact 28S and 18S bands on an ethidium bromide-containing agarose gel as for Northern blot analysis and by PCR analysis of actin mRNA (31, 33). Specimen contamination with normal tissue was minimized by the use of macroadenomas and careful dissection of the tissue. Lack of contamination was additionally verified by confirming the absence of mRNA (by RT-PCR) for TSHß in gonadotrope, null tumors, and PRL- and GH-containing tumors (which should not express TSH); the absence of PRL mRNA in gonadotrope and null tumors; and the absence of the mRNA for the nuclear orphan receptor steroidogenic factor-1 in tumors not of gonadotrope lineage (34).
RT-PCR analysis of ER and ERß mRNA isoforms
Total isolated RNA was quantitated by spectrophotometry, and 1–2 µg total RNA were used to prepare complementary DNA (cDNA) by RT. Reagents for the RT and PCR reactions were obtained from Perkin Elmer (Norwalk, CT). Each RT reaction also contained random hexamers as primers for first strand cDNA synthesis, 10 U ribonuclease inhibitor, and 40 U Moloney murine leukemia virus reverse transcriptase (Perkin Elmer) in a 20-µL volume containing 50 mmol/L Tris-HCl (pH 8.3), 5 mmol/L KCl, 5 mmol/L MgCl2, 5 mmol/L dithiothreitol, 0.25 mmol/spermidine, and 200 µmol/L each of precursor deoxy-NTPs. Reactions were incubated at 25 C for 10 min (annealing), at 42 C for 15 min (elongation), and at 95 C for 5 min (heat inactivation). PCRs were then performed on each RT reaction, using specific oligonucleotide primers as described below and 25 U Taq polymerase. For each ER mRNA variant tested, at least two independent RT reactions for each tumor RNA sample were independently assessed. Whenever possible, the presence of individual variants was also verified by two independent primer sets. All RNA samples were tested for ER variants in the same PCR amplification experiment. PCR amplifications were performed by the addition of specific primers (15 pmol) to the RT reactions along with Taq polymerase and reagents according to manufacturer’s instructions. Amplifications were performed in a thermocycler for 35 cycles, [95 C for 1 min, optimal annealing and elongation temperature (56–60 C) for 1 min), and one cycle at 72 C for 10 min]. Each experiment also contained a positive control, a PCR reaction with a plasmid containing the cDNA sequence for the entire coding regions of the specific ER. In this case, PCR amplification should result in only one product, corresponding to the full-length ER form. Negative controls included reactions with water and other reagents, but no RNA or DNA, to test for reagent contamination and RNA without reverse transcriptase to test for genomic DNA contamination. No product was seen in the negative controls of the experiments described, and only single bands were observed in plasmid-containing positive control reactions.
PCR primers were synthesized to various sites along both upper
and lower strands of the coding regions of ER and ERß, as shown in
Fig. 1
. Specific primers, their location in the coding region, and
specific sequences are as follows. ER sequences are assigned
nucleotide positions as defined by Chaidrun et al. (12).
ER upper strand primers included H1 (exon 1, bases 226–246;
5'-ACGGACCATGACCATGACCCT-3'), HFC (exon 2, bases 845–862;
5'-TGCAAGGCCCTTCTTCAAG-3'), HU4 (exon 4, bases 1137–1157;
5'-TAAGAAGAACAGCCTGGCCTTG-3'), H4/5 (exon 4, bases 1184–1202;
5'-GCCTTGTTGGATGCTGAG-3'), H5/5 (exon 5, bases 1329–1346;
5'-GCTTTGTGGATTTGACCC-3'), and H6/5 (exon 6, bases 1481–1501;
5'-TGTGTAGAGGGCATGGTGGAG-3'). ER lower strand primers included H3C
(exon 3, bases 885–904; 5'-GGTGGCTGGACACATATAGT-3'), HC5 (exon 5,
bases 1430–1450; 5'-AGGAGCAAACAGTAGCTTCAC-3'), H6/3 (exon 6, bases
1559–1576; 5'-GAGGCACACAAACTCCTC-3'), H 7/3 (exon 7, bases 1768–1785;
5'-CTCATGTGCCTGAATGTGG-3'), and HJ (exon 8, bases 2002–2020;
5'-TCAGACTGTGGCAGGGAAA-3').
ERß primers were assigned base positions according to the numbering system of Mosselman et al. (25). Exon/intron boundaries were assigned as demonstrated by Enmark et al. (35). Upper strand primers were HB10 (bases 10- 29; 5'-CCTGCTGTGATGAATTACAG-3'), HB862 (bases 862–882; 5'- TGGATGGAGGTGTTAATGATG-3'), and HB1173 (bases 1173–1193; 5'-TGCTTTGGTTTGGGTGATTGC-3'). Lower strand primers were HB559 (bases 541–559; 5'-TTCTCTGTCTCCGCACAAG-3'), HB936 (bases 918–936; 5'-GATCTGGAGCAAAGATGAG-3'), and HB1422 (bases 1402–1422; 5'-TTTGCTTTTACTGTCCTCTGC-3').
Southern blot and hybridization analysis
Products from PCR reactions were subjected to electrophoresis on
a 1.5% agarose TBE gel (88 mmol/L Tris-HCl, 50 mmol/L boric acid, and
2 mmol/L ethylenediamine tetraacetate) containing ethidium bromide, as
previously described (33). PCR product bands were visualized under UV
light, transferred to nitrocellulose membranes, and hybridized with
32P-labeled DNA probes specific for ER-coding sequence.
Probes included cDNAs for the entire coding regions of the human ER
cDNA (10) or ERß cDNA (25) labeled by random priming and
incorporation of labeled deoxy-CTP (33) or oligonucleotides
corresponding to specific exon sequences, which were end labeled by T4
polynucleotide kinase (36). After hybridization, blots were washed and
subjected to autoradiography at -70 C for 2–16 h.
Cloning and DNA sequence analysis of ER variant PCR products
To confirm the identity of the specific ER mRNA splice variants, specific PCR products of the appropriate size were isolated from an agarose gel and cloned into the PCR2.1 cloning vector (Invitrogen, San Diego, CA) according to the manufacturer’s instructions. Plasmids containing the PCR inserts were subjected to DNA sequencing by the dideoxy method of Sanger (37). At least two independent clones were used to confirm sequence identity.
Immunoblot analysis
To determine whether variant ER proteins could be detected in
either normal pituitary or pituitary tumors, immunoblot analysis was
performed on total cellular protein or nuclear protein as previously
described (33, 38). In all cases tested, the use of nuclear proteins
gave the same immunopositive protein pattern as whole cell proteins,
but with a stronger signal (not shown). Protein samples (35–100 µg
as indicated) were boiled for 5 min in buffer containing 50 mmol/L
Tris(hydroxymethyl)aminomethane (pH 7.6), 2% (wt/vol) SDS, and 2%
ß-mercaptoethanol. Proteins were separated by electrophoresis on 12%
polyacrylamide gels containing 1% (wt/vol) SDS and transferred to
nitrocellulose membranes. ER was detected with an antibody (1D5,
Dako Corp., Carpinteria, CA) that recognizes amino acids
65–78 in the N-terminal region of human ER (39). Primary antibody
was used at a dilution of 1:100 and was incubated for 3 h at room
temperature. The secondary antibody was peroxidase-conjugated goat
antimouse IgG (Jackson ImmunoResearch Laboratories, Inc.,
West Grove, PA) and was used at a dilution of 1:40,000 for 1 h at
room temperature. The ERß antibody was provided by Dr. Nira
Ben-Jonathan, University of Cincinnati, Cincinnati, OH) and is a
polyclonal antibody to the 18-amino acid N-terminal peptide of ERß,
homologous in rat, mouse, and human. Primary antibody was used at a
dilution of 1:1000 at room temperature for 1 h, followed by
incubation with secondary antibody (peroxidase-conjugated donkey
antirabbit IgG, Amersham, Arlington Heights, IL) at a
dilution of 1:500 dilution for 1 h. Enhanced chemiluminescence (ECL,
Amersham) was used to detect immunopositive protein
bands.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
mRNA variants
Both normal pituitary and pituitary adenoma RNA samples were
analyzed for the presence of full-length and variant ER mRNA by
RT-PCR techniques. Full-length ER mRNA was detected in all normal
pituitaries, all lactotrope tumors expressing only PRL, and a majority
(62%) of gonadotrope tumors (Table 1). A
lower percentage of null (33%) tumors and tumors expressing both PRL
and GH (50%) contained ER mRNA. No tumors that expressed only GH
contained any ER mRNA. Analysis with defined primer pairs also
demonstrated the existence of mRNA splice variants in all tumor types.
No amplified product for a splice variant mRNA appeared without a
corresponding product for full-length mRNA, and not all possible
variants were detected. For example, no normal pituitary or pituitary
adenoma RNA appeared to express mRNA variants lacking either exon 3 (as
tested with primer pairs H1 and HC5, or HF5 and HC5) or exon 6 (tested
with primer pairs HU4 and H7/3, H4/5 and HJ, and H5/5 and H7/3).
|
depicts RT-PCR analysis for ER mRNA variants in several tumors.
Overall, the ER2 variant mRNA, was contained in all prolactinomas
(n = 14), but to a lesser extent in gonadotrope (6 of 13) or null
cell (4 of 8) tumors. Splice variant products representing the deletion
of exon 4 (ER4) were observed in the majority of all tumors that
expressed ER regardless of tumor type. Similarly, expression of the
ER5 variant was expressed in all prolactinomas and was present in
nearly half of the tumors representing other cell types. In contrast,
the ER7 splice variant was preferentially expressed in
gonadotrope (n = 13) and null (n = 7) tumors and appeared in
less than half of prolactinomas (6 of 14) or in tumors expressing both
PRL and GH (1 of 6). No splice variant appeared without a corresponding
full-length ER mRNA, and no ER mRNA variants appeared in any tumor
that expressed only GH. Incorporation of radioactive nucleotide
precursors into amplification products gave identical results as
experiments performed with blotting of DNA products (not shown).
|
mRNA splice variants were also expressed in normal pituitary
tissue. Deletion variants for exons 4 and 7 appeared in all normal
pituitary RNA samples, and exon 5 deletions appeared in four of six
samples. In some cases, the intensity of the labeled variant band was
low compared to that of the full-length PCR product (not shown). In
contrast, only two normal pituitaries expressed the exon 2 deletion
splice variant even though it was contained in many pituitary tumors
(Table 1). No pituitary or tumor RNA samples contained double exon
deletions.
ER protein
The biological significance of any ER mRNA variant depends on the
ability of the mRNA to be translated into a functional protein.
Immunoblotting analysis of proteins from one normal pituitary and six
pituitary adenomas is depicted in Fig. 3.
Tumor samples were divided for both protein and RNA analysis, and the
ER mRNA variants contained in each sample are shown below each protein
lane. ER mRNA exon deletion variants are indicated by the number of
the exon deleted. For example, in this study the prolactinoma contained
mRNA for ER (full-length and 2, 4, and 5 variants), but not
for ERß. The normal pituitary contained all ER and ERß mRNA and
splice variants shown in Table 1, except for exon 5. The GH and PRL
tumor and both gonadotropin (GO) tumors contained mRNA for ER and
ERß. Both null tumors contained low levels of ER, and one of two
tumors expressed ERß mRNA.
|
(1D5) was used to
detect immunopositive protein in the samples. This antibody would be
expected to detect full-length ER (65–67 Da) as well as the
in-phase exon 4 deletion protein (54–56 kDa), and the prematurely
truncated proteins corresponding to the exon 5 (40–42 kDa), exon 7
(50–52 kDa), and exon 2 (18–19 kDa) deletion variants. In the
ER-positive tumors and in normal pituitary, the major immunopositive
protein band, in general, corresponded to the full-length ER. In
four tumors and in normal pituitary, an immunopositive band at
approximately 41 kDa was observed, and this may correspond to the
ER5 variant protein. This protein migrated with the same relative
mobility as the exon 5 deletion protein in BT-20 breast cancer cells,
which have been demonstrated to contain this deletion variant ER
protein (37). In the normal pituitary and prolactinoma, in which the
ER2 variant mRNA was detected, an immunopositive band at
approximately 18–20 kDa was observed. No such protein band was
detected in other tumor samples. No immunopositive protein band
corresponding to the expected molecular masses for ER4 or
ER7 was detected in any sample, even though the mRNAs for these
variants were detected by RT-PCR. Although the absolute identity of
each immunopositive band cannot be made from this study, these data
suggest that at least some of the ER mRNA variants in both normal
pituitary and pituitary tumors can be translated into functional
proteins, which could then influence the biological responses of the
tissue to estrogen. ERß mRNA variants
Primers specific for 2 regions of ERß mRNA, spanning both
N-terminal- and C-terminal-coding regions of the cDNA, were used to
analyze normal pituitary and pituitary adenoma samples in RT-PCR
reactions. As the data were essentially the same for both sets of
primers, and no splice variants were observed or characterized from the
C-terminal set of primers, only the N-terminal results are shown (Fig. 4). ERß mRNA was expressed in all
normal pituitary RNA samples, but was preferentially expressed in
gonadotrope tumors (19 of 21), compared to null tumors (15 of 24) and
prolactinomas (6 of 14). Quantitative data cannot be obtained by these
methods. However, the intensity of the amplified ERß bands appeared
to be greater in many of the gonadotrope tumor samples, suggesting that
more ERß mRNA might be contained in those tumors. Of interest, 4 of 6
somatotrope tumors (expressing only GH) contained no ER mRNA, but
expressed easily detectable levels of ERß mRNA. No striking
differences in patient levels of serum PRL, GH, or insulin-like growth
factor I were apparent based on receptor status (Table 2). In addition, 4 of 6 tumors expressing
both GH and PRL expressed ERß. Two of these tumors expressed only
ERß, 2 expressed only ER, and 2 expressed both ER mRNA isoforms.
All gonadotrope tumors and null tumors that expressed ER expressed
ERß mRNA, and the remaining ER-positive tumors expressed ERß mRNA
alone. In contrast, all prolactinomas that expressed ERß (6 of 14)
also expressed ER mRNA, with the remaining 9 tumors containing only
ER mRNA.
|
|
and 5). In the 5'-region of the cDNA, a
variant corresponding to the precise deletion of exon 2, encoding the
first zinc finger in the DNA-binding region, was cloned and
characterized. The exon 2 deletion variant was observed in all
ERß-positive tumors expressing GH and GH plus PRL, in the majority
(14 of 19) of gonadotrope tumors, and in half of the ERß-positive
null (8 of 15) tumors and prolactinomas (3 of 6). The splice variant
results in a mRNA that would encode a truncated protein of 76 amino
acids; the coding frame is out of phase and results in an inappropriate
threonine (instead of histidine) residue at the end of exon 1 followed
by 2 stop codons (Fig. 5).
|
Immunoblotting analysis of pituitary tumors was performed for
ERß using the identical samples analyzed for ER in Fig. 3, but
with a separate gel and blot. Figure 6
demonstrates that immunopositive ERß protein (50–55 kDa) can be
detected in several of the samples, including normal pituitary and
several gonadotropin tumors, but not in the prolactinoma. This tumor
contained detectable ER protein (Fig. 3) and was positive for ER,
but not ERß, mRNA. Exposure of the film for longer times did not
reveal additional immunopositive proteins (not shown). BT-20 breast
cancer cells also were positive for ERß mRNA and protein, as has been
recently reported for some breast cancer cell lines (35). For each
sample shown, the presence of ER or ERß mRNA correlated with the
presence or absence of the appropriate receptor protein (Figs. 3 and 6). A smaller immunopositive band was noted in two of the tumor lanes.
The identity of the protein is not known, but it cannot correspond to
an exon 2 deletion protein, which would migrate below 10,000 Da; no
such protein was observed in our samples.
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
mRNA expression occurs in all
prolactinomas, a majority of gonadotrope tumors, and some null tumors,
but not in tumors expressing only GH. This agrees with previous data
from several laboratories (10, 11, 12, 13) using RT-PCR, nuclease protection,
and immunohistochemistry of pituitary tumors that demonstrated
preferential expression of ER in lactotrope and gonadotrope cells.
One group reported a single GH adenoma containing ER protein
detected by isoelectric focusing (16), although other investigators
using similar techniques failed to observe this protein in
GH-containing tumors (11). Using in situ hybridization
techniques for ER, Stefaneaunu et al. (17) found that
nearly all types of cells contained ER mRNA. Given some homology
between the ER and the recently cloned ERß (24, 25), other nuclear
receptors or both ER isoforms may have been detected by this technique
and could explain these results. For example, we have found that
GH-containing tumors express ERß mRNA; we did not test ACTH- or
TSH-containing tumors.
We have previously shown by quantitative ribonuclease protection assays
that the ER mRNA is more abundant in prolactinomas than in
gonadotrope or null tumors, and that pituitary tumor expression of the
ER mRNA is reflected in the proportion of normal pituitary cells
expressing the ER protein (10). Expression of the mRNA splice
variants similarly exhibited some cell type selectivity, in that all or
most prolactinomas contained the exon 2 and 5 deletion variants,
whereas a higher proportion of gonadotrope tumors contained the exon 7
deletion variant. These results vary somewhat from the report by
Chaindrun et al., who found some expression of ER3
variants, and a higher expression of the ER7 splice variant mRNA
in prolactinomas (12). We have verified expression of these isoforms
using two separate sets of primers; thus, the basis for this difference
in expression is not apparent, but may reflect different patient
populations.
Normal pituitary tissue contains the same ER mRNA splice variant
forms as pituitary tumors, although the exon 2 deletion is much more
prevalent in tumor tissue, particularly in prolactinomas. Because
lactotropes represent a large proportion of cells in normal
pituitaries, it is more likely that this difference represents enhanced
expression in tumor tissue, rather than expression in a small
population of cells. No double exon mRNA deletions were observed in
normal or tumor tissue, although double exon 3/4 and 5/6 deletions have
been observed in rat pituitaries (33).
Variant immunopositive forms of ER protein, corresponding to the
predicted sizes of proteins translated from the splice variant mRNAs,
occur in normal and tumor pituitary tissues, suggesting that variant
proteins can occur and modulate the physiological response to estrogen
in those cells. Although little information is available on the
potential role of such ER protein variants in human pituitary, a large
body of data is available from human breast cancer cells and
transfected cell lines. Because of the domain structure of the
receptors, mRNA deletions can result in ER proteins with dramatically
altered functions. Deletion of exon 4 removes a portion of the
DNA-binding, hinge, and hormone-binding regions and results in a
protein with no discernible biological activity or influence on
full-length ER activity (40). This variant, expressed in many tissues,
would be neutral for estrogen actions. The ER7 protein can bind
to EREs on DNA but cannot bind hormone and acts in a dominant negative
fashion to inhibit the effects of the wild-type ER (22). In contrast,
ER5 mRNA is translated into a truncated protein that binds DNA
but cannot bind estrogen and has constitutive, hormone-independent
effects on ERE-regulated gene transcription (19, 41). The mRNA and
protein have been characterized in human breast cancer tissues and cell
lines and are postulated to contribute to estrogen-independent
ER-mediated effects on growth and gene expression (20). The ER5
variant is expressed at high levels in some ER-negative, progesterone
receptor-positive tumors. Expression of the variant may explain the
phenotype, in that the variant protein may bind to EREs and activate
the progesterone receptor without estrogen, thus
representing a step in escape from estrogen and antiestrogen
responsiveness (19). A recent report (23) describes studies in which
the ER2 variant cDNA, transfected into U-2 human osteosarcoma
cells, had little activity alone, but amplified the ability of
full-length ER to stimulate c-fos reporter gene
expression. The mechanism for this effect is unknown, because the
expected protein would not bind to DNA or hormone. It is difficult to
simply extrapolate data from transfected cell lines to pituitary
tissue, particularly given the cell-specific effects of ER proteins and
isoforms on different promoters (27, 28, 29, 42). However, it is intriguing
that consititutively active (ER5) or ER-stimulating (ER2)
forms of the receptor are preferentially expressed in prolactinomas,
whereas a suppressive form of ER (ER7) is preferentially expressed
in gonadotrope tumors.
In contrast to the results with ER, ERß mRNA is expressed
preferentially in gonadotrope tumors, much less frequently in
prolactinomas, and in the majority of tumors expressing only GH or GH
plus PRL. One exon deletion splice variant was characterized in these
studies. ERß2 mRNA was expressed in normal pituitary and many
tumor types, and would result in a severely truncated protein
consisting only of exon 1. The biological activity of such a protein,
if any, is difficult to predict, but might correspond to the
stimulatory activity described for the ER2 variant (23). Several
ERß mRNA splice variants have been characterized in rat tissues,
including the pituitary (43, 44, 45). These include deletions of exon 3,
which eliminates the second zinc finger and the ability to bind DNA,
and a variant (ERß2), in which insertion of 18 amino acids between
exons 5 and 6 results in a protein with altered estrogen binding
affinity (43, 44). We have no evidence for either of these variant mRNA
forms in any of our tumor samples. Recently, an ERß5 variant was
reported in breast cancer cells (45), but this variant was not observed
in our experiments, using primers that would amplify most of the coding
region (not shown).
Previous data from other investigators suggest that ER is expressed
at much higher levels than ERß in pituitaries of rats (26, 43) and in
wild-type or ER gene-disrupted mice (46). However, these studies do
not distinguish between low levels of expression in many cell types
vs. intermediate or high levels of expression in a selected
cell population. Given that gonadotropes comprise a small proportion of
pituitary cells, low overall ERß expression would still be consistent
with expression in a high proportion of gonadotropes. We have shown
that a high percentage of lactotrope and gonadotrope cells in normal
pituitary expresses the ER protein (10). Thus, many of these cells
would express both ER isoforms. Because ERß and ER can form
heterodimers (47), both proteins could influence estrogen action in
these cells. The observation that several GH-only and GH plus PRL
tumors contain ERß is particularly interesting. GH-only tumors and
GH-expressing cells in normal pituitary do not express ER mRNA or
protein (10, 11, 12, 13), and only half of PRL- and GH-containing tumors or
normal cells do. Thus, ERß would be the only or the major mediator of
estrogen action in these tumors, and ERß may account for some of the
actions of estrogen on GH previously thought to be indirect (48).
Because the two ER isoforms have different ligand binding affinities
for natural, environmental, and synthetic estrogens (26), differential
or exclusive expression of the proteins in different cell types might
have physiological or therapeutic significance.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
Received May 12, 1998.
Revised July 20, 1998.
Accepted July 28, 1998.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and ß. Endocrinology. 138:863–870. and ER ß
at AP1 sites. Science. 277:1508–1510. and ß. Endocrinology. 139:111–118. (ER) and estrogen receptor-ß (ERß) messenger
ribonucleic acid in the wild-type and ER-knockout mouse. Endocrinology. 138:4613–4621.. Mol
Endocrinol. 11:1486–1496.This article has been cited by other articles:
 |
|
 |
|
  C. Montani, M. Penza, M. Jeremic, G. Biasiotto, G. La Sala, M. De Felici, P. Ciana, A. Maggi, and D. Di Lorenzo Genistein is an Efficient Estrogen in the Whole-Body throughout Mouse Development Toxicol. Sci., May 1, 2008; 103(1): 57 - 67. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Cosma, J. Bailey, J. M. Miles, C. Y. Bowers, and J. D. Veldhuis Pituitary and/or Peripheral Estrogen-Receptor {alpha} Regulates Follicle-Stimulating Hormone Secretion, Whereas Central Estrogenic Pathways Direct Growth Hormone and Prolactin Secretion in Postmenopausal Women J. Clin. Endocrinol. Metab., March 1, 2008; 93(3): 951 - 958. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 N. Ben-Jonathan, C. R. LaPensee, and E. W. LaPensee What Can We Learn from Rodents about Prolactin in Humans? Endocr. Rev., February 1, 2008; 29(1): 1 - 41. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. D. Veldhuis, D. M. Keenan, and C. Y. Bowers Peripheral estrogen receptor-{alpha} selectively modulates the waveform of GH secretory bursts in healthy women Am J Physiol Regulatory Integrative Comp Physiol, October 1, 2007; 293(4): R1514 - R1521. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. A. Arreguin-Arevalo, T. L. Davis, and T. M. Nett Differential Modulation of Gonadotropin Secretion by Selective Estrogen Receptor 1 and Estrogen Receptor 2 Agonists in Ovariectomized Ewes Biol Reprod, August 1, 2007; 77(2): 320 - 328. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
U. Ohnemus, M. Uenalan, J. Inzunza, J.-A. Gustafsson, and R. Paus The Hair Follicle as an Estrogen Target and Source Endocr. Rev., October 1, 2006; 27(6): 677 - 706. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. M.A. Dreijerink, K. W. Mulder, G. S. Winkler, J. W.M. Hoppener, C. J.M. Lips, and H.Th. M. Timmers Menin Links Estrogen Receptor Activation to Histone H3K4 Trimethylation. Cancer Res., May 1, 2006; 66(9): 4929 - 4935. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. Veldhuis, J. N. Roemmich, E. J. Richmond, and C. Y. Bowers Somatotropic and Gonadotropic Axes Linkages in Infancy, Childhood, and the Puberty-Adult Transition Endocr. Rev., April 1, 2006; 27(2): 101 - 140. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. J. Lee and J L. Jameson Gene therapy of pituitary diseases J. Endocrinol., June 1, 2005; 185(3): 353 - 362. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
