| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Endocrinological Oncology |
Groupe d’Etude en Physiopathologie Endocrinienne, INSERM CJF 9208, Institut Cochin de Génétique Moléculaire, Université René Descartes, Paris, France
Address all correspondence and requests for reprints to: Dr. Yves de Keyzer, Groupe d’Etude en Physiopathologie Endocrinienne, INSERM CJF 9208, Institut Cochin de Génétique Moléculaire, 24 rue du Faubourg Saint Jacques, 74014 Paris, France.
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Recently, a new family of POU transcription factors called Brn was cloned from rat and human brain. These genes are expressed throughout development in the brain and some neuroendocrine structures (5). The Brn family has a bipartite DNA-binding domain with a POU-specific N-terminal segment of 74 residues separated by a hinge region from a POU homeodomain of 60 residues at the C-terminus. Six classes of POU proteins have now been identified that share a similar DNA-binding domain structure, but differ by characteristic short regions in the POU-specific domain and POU homeodomain and by the presence of a class-specific domain in the N-terminal region called the POU box (6). Interestingly, one of them, Brn3a, belonging to class IV and originally cloned from rat brain, binds to a DNA motif of the CRH gene that is also present in the central domain of the rat POMC promoter (rCE-2). Moreover, Brn3a is present in AtT-20 cells, binds to the POMC promoter, and stimulates its activity in transfection studies (5).
Alternate promoter and alternative splicing in the 5'-region of Brn3a generate a second messenger ribonucleic acid (mRNA) encoding a protein that contains an additional domain at its amino-terminal end (7), conferring an oncogenic potential to Brn3a when cotransfected with H-ras. The gene coding for its human homolog, RDC1, shows 100% identity with the Brn3a POU domain (8). RDC1, hereafter called human Brn3a (hBrn3a), is expressed in neural lineages and in most neuroepithelioma. In addition, Oct-T1, a cDNA coding for a protein almost identical to Brn3a, except for 45 additional residues at its amino-terminal end, was independently cloned from the human T cell line Jurkat (9). Oct-T1 is most likely the product of differential splicing of the hBrn3a gene reproducing the structure described in the mouse.
We hypothesized that hBrn3a might bind to the Brn3a-like motif conserved in the central domain of the human POMC promoter (hCE-2) and be involved in its activation, particularly in ACTH-secreting tumors. Using a specific and highly sensitive reverse transcription PCR (RT-PCR) approach to quantitate the various mRNA forms of the hBrn3a gene, we examined a series of neuroendocrine tumors. We found that hBrn3a gene expression is not correlated with that of the POMC gene; rather, it is strikingly elevated in highly aggressive tumors, independently of their POMC status and their pituitary or nonpituitary origin. In contrast with the well differentiated and rather indolent ACTH-secreting pituitary adenomas, the mouse tumor cell line AtT-20 had very high levels of Brn3a mRNA.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Pituitary tumors. Nine ACTH-nonsecreting tumors were studied. Four GH- and two PRL-secreting adenomas were diagnosed on the basis of their clinical features, elevated plasma GH or PRL levels, and immunohistochemistry of the tissue. Three gonadotropin-producing tumors were diagnosed by immunohistochemistry. Two GH-secreting adenomas and the two gonadotropin-producing tumors were locally invasive at surgery. Eight ACTH-secreting macroadenomas were studied. Three were obtained in patients after bilateral adrenalectomy, and 5 in patients with Cushing’s disease. Three had features of local invasiveness at surgery, and a fourth patient with a large and aggressive tumor died within 1 yr after diagnosis with putative liver metastases.
Bronchial tumors. Bronchial neuroendocrine tumors consisted of five typical carcinoids, seven atypical carcinoids, and eight small cell carcinomas (SCCLs) (10). The latter group included one original tumor, two lymph node metastases, three xenografts in nude mouse, and the two human tumor cell lines DMS-79 and DMS-53. Nine of the bronchial tumors (seven carcinoids and two SCCLs) secreted ACTH and/or contained POMC mRNAs.
Cell lines. The mouse pituitary tumor cell line AtT-20 and the two human SCCL cell lines (DMS-79 and DMS-53) were grown as described previously (11). DMS-79 expressed the POMC gene, whereas DMS-53 expressed the calcitonin gene.
Tissue collection
Tumor samples were obtained at surgery from patients with pituitary or bronchial neuroendocrine tumors. Xenografts of SCCLs were obtained by immediate transplantation of tumor tissue in nude mice and serially transplanted (a gift from M. F. Poupon) (12). The tissues were immediately frozen in liquid nitrogen and kept at -80 C.
Protein extraction, POMC peptides assays, and RNA extraction were performed as previously described (13, 14, 15, 16).
RT-PCR amplification
Random primed cDNAs were synthesized from 500 ng total RNA with
100 U Moloney murine leukemia virus reverse transcriptase following the
manufacturer’s instructions (Life Technologies, Grand Island, NY), and
40% of the reaction was used for PCR. The following oligonucleotides
were used as PCR primers for hBrn3a cDNAs (see Fig. 1
):
POU1 (5'-GGCCCACCTCAAGATCCCGG-3') and POU2 (5'-AGTTTCTCGGCGATGGCGGC-3')
for competitive PCR, and BRN2 (5'-GCCTGCCTGCCCACGCCG-3'), BRN3
(5'-TCCACTGCCCCCAAACCCG-3'), BRN5 (5'-CACGGCACGCTGTTCATCG-3'), and BRN6
(5'-CGAGCGACGGCGAGGAGATG-3') to differentiate among the various hBrn3a
mRNAs forms. PCR was performed in 50 mmol/L KCl, 25 mmol/L Tris (pH
8.3), 2 mmol/L MgCl2, and 2.5 U Taq DNA
polymerase (BRL, Bethesda, MD). After 30 cycles of amplification
consisting of 45 s at 94 C, 45 s at 57–62 C, and 45 s
at 72 C, followed by 10 min at 72 C, 20% of the PCR products were
separated by electrophoresis, blotted onto a Nylon membrane (Hybond,
Amersham, Arlington Heights, IL), and probed with internal end-labeled
oligonucleotides. High sequence identity between mouse and human Brn3a
POU domains enabled the amplification of mouse Brn3a RNA with POU1 and
POU2 primers. The PCR products specific for the large and short Brn3a
mRNAs were probed with internal 32P end-labeled specific
oligonucleotides.
|
Mutant construction. Random primed cDNAs from a pituitary corticotropic adenoma were used to amplify a fragment of hBrn3a with POU1 and POU2 primers. This 294-nt product was subcloned in pCRII (Invitrogen Corp., San Diego, CA) and identified by sequencing. A mutant was obtained by insertion at a unique StuI site of a head to tail dimer of an unrelated 52-bp RsaI fragment. The resulting plasmid should, therefore, yield a 398-nt PCR product with POU1 and POU2 primers. Human Brn3a mutant RNA was synthesized with SP6 RNA polymerase from this plasmid linearized 960 nt downstream of the 3' cloning and serially diluted in siliconized tubes to give arbitrary concentrations ranging from 1- to 10-7-fold that of the original solution. These reference solutions were aliquoted and stored at -80 C.
Semiquantitative RT-PCR. Increasing concentrations of the synthetic mutant RNA were added to 500 ng total RNA and simultaneously reverse transcribed with random hexamers and Moloney murine leukemia virus. PCR was then performed as described with POU1 and POU2 primers. All experiments included control reactions in which RT was omitted to visualize a putative genomic DNA contamination. PCR products were subjected to nondenaturing gel electrophoresis on a 4.5% polyacrylamide gel, electroblotted, and hybridized as described above. Filters were first autoradiographed, and the precise amount of radioactivity in each band was determined with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA). The signal ratio (mutant/wild type) was plotted against the mutant RNA dilution. The value of hBrn3a mRNA in the sample is equivalent to the mutant dilution corresponding to a PCR products ratio of 1. Arbitrary units were defined as original mutant solution dilutions.
Whole cell protein microextraction and gel mobility shift assays
Microextraction of proteins from whole DMS-79 cells was carried
out on 2 x 107 log phase growing cells as described
previously (17). The protein concentration was determined by UV
absorbance and with the protein bioassay kit (Bio-Rad Laboratories,
Richmond, CA). Gel mobility shift assays were performed according to
classical methods (18). The double stranded probe was end labeled with
[
-32P]ATP. DNA (0.1 ng) were incubated with 3–5 µg
cellular proteins for 10 min at 4 C in 20 µL 10 mmol/L HEPES (pH
8.0), 50 mmol/L NaCl, 50 mmol/L KCl, 5 mmol/L MgCl2, 4
mmol/L spermidine, 2 mmol/L dithiothreitol, 0.1 mmol/L ethylenediamine
tetraacetate, 100 µg/mL albumin, 15% (vol/vol) glycerol, and 1 µg
poly(dI-dC), which was used as nonspecific DNA competitor. Protein-DNA
complexes were resolved on 6% nondenaturing polyacrylamide gels and
exposed to XAR-Omat films (Eastman Kodak, Rochester, NY) at -80 C with
an intensifying screen (DuPont, Wilmington, DE). DNA-binding sites were
POMC hCE-2 oligonucleotide (-263/-228 of the human promoter), CRH
promoter element (-139/-108 of the rat promoter) (5), and herpes
simplex virus (HSV)-Oct-binding site (5) flanked by the 5'- and
3'-sequences of the hCE-2 oligonucleotide.
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
To determine the level of hBrn3a transcripts, random primed cDNAs
were amplified with POU1 and POU2 primers. These primers are
complementary to sequences specific for class IV POU proteins in the
POU domain (Fig. 1
). The RT-PCR reaction thus amplifies a portion of
the hBrn3a-coding POU domain that is present in both the long and the
short hBrn3a transcripts. Using the POU1 and POU2 primers, we amplified
a similar 294-nt RT-PCR product with RNA from a human SCCL line
(DMS-79) and a human pituitary corticotroph tumor. After subcloning,
all clones analyzed showed the expected hBrn3a sequence, demonstrating
the specificity of the PCR reaction. This primer set, however, did not
allow discrimination among the various forms of hBrn3a mRNAs (Fig. 1).
When synthetic mutant RNA was simultaneously reverse transcribed with
RNA samples, the subsequent PCR amplified the expected 294-nt wild-type
fragment and the 398-nt mutant PCR products. Increasing the
concentration of the mutant RNA progressively decreased the amount of
the 294-nt fragment, whereas that of the 398-nt fragment increased
conversely. Areas representing the respective signals, as measured with
a PhosphorImager, were plotted against mutant RNA concentrations (Fig. 2) for three samples with wide differences in hBrn3a
mRNA contents. Competitive amplification between wild-type and mutant
RNAs gave opposite linear curves over a wide range of mutant dilution
and was useful even for tissues with very high or very low hBrn3a RNA
content.
|
All GH-, PRL-, and gonadotropin-secreting adenomas had low levels
of hBrn3a transcripts (<5 x 10-6 arbitrary units;
Fig. 3, left panel). Seven of eight
ACTH-secreting tumors had similarly low hBrn3a transcript levels. Only
one had a higher level (3 x 10-5), corresponding to
the highly invasive tumor of the patient who eventually died with liver
metastases. The highest hBrn3a levels were found in the corticotrope
mouse cell line AtT-20 (4 x 10-4 arbitrary units;
Fig. 3, left panel).
|
Three groups of bronchial neuroendocrine tumors were distinguished
according to their histological status: each group contained several
tumors responsible for the ectopic ACTH syndrome (Fig. 3, black
circles, right panel). All five typical bronchial
carcinoids had barely detectable levels, similar to those in pituitary
adenomas, whereas SCCLs had extremely high levels (up to 1000 times
more), and atypical carcinoids had intermediate values.
There was no obvious association between hBrn3a levels and POMC gene expression; the two tumors with the highest POMC mRNA contents were two typical bronchial carcinoids with low Brn3a mRNA levels (data not shown).
The different molecular forms of hBrn3a transcripts
Two sets of oligonucleotides were designed in the 5'-region of the
hBrn3a gene to frame sequences specific for the mRNA encoding the large
form of hBrn3a (BRN2 and BRN5) or the short one (BRN3 and BRN6; Fig. 1). Both forms were detected in the subset of tumors examined (data not
shown), and no gross variation in their respective abundance was
observed among the various types of tumors.
Gel mobility shift assay for hBrn3a protein
We next examined the content in hBrn3a proteins by their
DNA-binding ability in cells in which the Brn3a mRNA had been detected.
The rat CRH promoter contains a high affinity binding site for Brn3a,
which was used in the original characterization of Brn3a (5). We used
this sequence as a probe to detect the presence of Brn3a by mobility
shift experiments. Figure 4A shows the pattern observed
with extracts of AtT-20, where the presence of the Brn3a protein has
previously been demonstrated (5). Three retarded complexes (a, b, and
c, lane 1) are specifically competed by an excess of unlabeled probe
(lane 2), but not by an excess of unrelated DNA (lane 5). Among them,
complex c is competed by an excess of HSV-Oct oligonucleotide (lane 3).
Moreover, the HSV-Oct probe forms a similar complex in the presence of
AtT-20 extracts (lanes 6–8). This complex is totally competed by an
excess of hCE-2 oligonucleotide (lane 4), whereas complex b is
partially competed, and complex a is unaffected. The binding
characteristics of the Brn3a protein (5) are in part summarized in Fig. 4B; Brn3a binds with high affinity to the CRH probe, has a weaker
affinity for a CRH-binding site mutated in the three central nt, a
sequence close to that of hCE-2, and has no affinity for the HSV-Oct
sequence. Only complex b presents these characteristics. We used these
same tools to look for the presence of Brn3a in the extracts of DMS-79.
With the CRH probe, the pattern is the same as that observed with
AtT-20 extracts (lane 9). A complex migrates at the same location as
complex b, and Fig. 4C shows that this complex is specific (lanes
1–3). In addition, an hCE-2 probe of the same size as the CRH probe
gives two retarded complexes similar to those obtained with the CRH
probe (lanes 4–6).
|
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
As hBrn3a has been shown to stimulate the rat POMC gene promoter, how can we explain why some tumors with high contents of Brn3a transcripts did not express POMC? We examined whether one tissue expressing the Brn3a mRNA contained proteins able to bind to a Brn3a target. To demonstrate the presence of Brn3a protein in ACTH-producing tumor cells, we performed gel shift analysis. Brn3a binding features have been defined with in vitro translated protein using a rat CRH promoter element that constitutes a high affinity binding site (GCATAAATAAT). Mutations within the CAT or TAAT sequences (such as the consensus binding site of the widely distributed, POU-related protein Oct-1, HSV-Oct) abolish binding, whereas mutations in the three central nt have a minor effect. We first used AtT-20 extracts, known to contain Brn3a, and the CRH probe to visualize Brn3a complexes. As expected AtT-20 extracts showed several specific complexes that were further characterized with several mutated sequences. Of the three specific complexes detected, complex b presents the expected properties; it binds to the CRH element, is not competed by HSV-Oct, and is moderately competed by the human POMC motif hCE-2, which represents a natural mutant of the central region. In addition, this complex is not found with rat liver extracts where the Brn3a gene is not expressed (data not shown). The same Brn3a-specific complex was detected in DMS-79 extracts with the CRH probe, demonstrating that in this cell line functional Brn3a proteins are produced. DMS-79, which has the highest level of Brn3a mRNA among the ACTH-producing samples, gives an abundant Brn3a complex, suggesting that the protein levels are related to the mRNA level. Interestingly, when hCE-2 was used as a probe, the same complex b was seen, demonstrating for the first time that Brn3a is able to bind to the human POMC gene.
An important feature of the POMC promoter is that its function requires the synergistic effects of various proteins acting on different domains. It is conceivable that Brn3a is only a minor contributor to POMC gene transcription, and its full effect requires high levels and the coexpression of other trans-activating factors, such as corticotropin upstream transcription element-binding protein, for example (4). That would offer an explanation for why not all SCCLs with high Brn3a contents express POMC.
Whereas no link could be found between Brn3a transcripts and POMC gene expression, it was clear that high levels of Brn3a transcripts were associated with highly aggressive tumors. Within the bronchial tumors of neuroendocrine origin, the carcinoids and the SCCLs are well recognized tumors with contrasted features; the former are a homogeneous group of tumors with an indolent course and a high degree of neuroendocrine differentiation, as exemplified by their high content of specific biochemical markers of secretory granules (19) and prohormone convertases (20, 21); the latter are highly aggressive tumors that have lost some of their neuroendocrine characteristics (22). The so-called atypical carcinoids have been identified by the pathologist on specific morphological features and have an intermediary prognosis between the typical carcinoids and the SCCL (10). Interestingly, this group of tumors showed a tendency toward elevated Brn3a transcript levels, although much less than those observed in SCCLs. It appears, therefore, that Brn3a levels vary in parallel with tumor aggression.
Brn3a is highly expressed in neuroepithelioma and Ewing’s sarcoma, which are malignant neoplasias derived from neuroectodermal tissues. It has been proposed that Brn3a plays a role in their pathogenesis. More precise studies established a preferential oncogenic action of the long form of Brn3a compared to the short form. As reported here, both forms of Brn3a mRNA were present in all of the bronchial tumors examined. These data further suggest that Brn3a may exert an oncogenic action in neuroendocrine tumors of the lung, and that the level of Brn3a transcripts bears a poor prognosis.
|
Footnotes |
|---|
Received July 19, 1996.
Revised September 25, 1996.
Accepted October 4, 1996.
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-lipotropin, and ß-endorphin in plasma of normal human subjects,
patients with ACTH/LPH hypersecretory syndromes, and patients
undergoing chronic hemodialysis. J Clin Invest. 67:124–133.
This article has been cited by other articles:
 |
|
 |
|
  V. Budhram-Mahadeo, P. J. Morris, M. D. Smith, C. A. Midgley, L. M. Boxer, and D. S. Latchman p53 Suppresses the Activation of the Bcl-2 Promoter by the Brn-3a POU Family Transcription Factor J. Biol. Chem., May 21, 1999; 274(21): 15237 - 15244. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P. L. M. Dahia and A. B. Grossman The Molecular Pathogenesis of Corticotroph Tumors Endocr. Rev., April 1, 1999; 20(2): 136 - 155. [Abstract] [Full Text]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Endocrinology | Endocrine Reviews | J. Clin. End. & Metab. |
| Molecular Endocrinology | Recent Prog. Horm. Res. | All Endocrine Journals |
, SCCL; •, SCCL cell line DMS-53; 
,
GH-secreting adenoma.
