| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Original Article |
Institut National de la Santé et de la Recherche Médicale Unité 567 (L.P.-L.T., X.B., Y.d.K.), Centre National de la Recherche Scientifique Unité Mixte de Recherche 8104, Université Paris 5, 75014 Paris, France; and Centre chirurgical Marie-Lannelongue (E.D.), 92350 Le Plessis-Robinson, France
Address all correspondence and requests for reprints to: Dr. Yves de Keyzer, Institut National de la Santé et de la Recherche Médicale Unité 567, Centre National de la Recherche Scientifique Unité Mixte de Recherche 8104, Université Paris 5, Department of Endocrinology, 24 rue du Faubourg Saint-Jacques, 75014 Paris, France. E-mail: keyzer{at}cochin.inserm.fr.
Abstract
The proopiomelanocortin (POMC) gene is occasionally expressed in nonpituitary tumors leading to Cushing’s syndrome. Bronchial carcinoid tumors, one of the most frequent source for ectopic ACTH secretion, often display numerous features of the corticotroph phenotype. To identify new markers of corticotroph differentiation in these tumors, we compared the pattern of gene expression in ACTH-secreting (ACTH+) and nonsecreting (ACTH-) bronchial carcinoids by differential display/RT-PCR. Using groups of ACTH+ and ACTH- tumors, we initially selected approximately 300 differentially expressed genes. Fifteen were considered differentially expressed after further characterization by RT-PCR on a larger series of 8 ACTH+ and 12 ACTH- bronchial carcinoids; 11 were restricted to—or overexpressed in—ACTH+ and four in ACTH- tumors. In ACTH+, beside the expected POMC gene, we identified cFos, and KIAA1775, a large expressed sequence tag encoding a putative protocadherin-related protein. On the other hand, the tetraspanin TM4SF5 gene was specifically expressed in ACTH-. Dot blot analysis confirmed the specific expression of KIAA1775 in ACTH+ bronchial carcinoids. However, the expression of most of the differential genes, including KIAA1775, was detected by RT-PCR in pituitary or lung tumors, whether secreting ACTH or not, excepted for TM4SF5, which was only detected in some nonendocrine lung tumors.
Our results show that corticotroph differentiation of bronchial carcinoid tumors is accompanied by induction and repression of specific genes. The nature of some of these genes, identified here, underlines the importance of cell-cell or cell-extracellular matrix interactions in the establishment of neoplastic corticotroph phenotype. These genes should help to better characterize ACTH+ bronchial carcinoids as well as other bronchial carcinoid subtypes.
PITUITARY CORTICOTROPHS ARE characterized by the expression of the proopiomelanocortin (POMC) gene, the precursor to ACTH. POMC gene expression is normally restricted to the pituitary, the arcuate nucleus of the hypothalamus and melanocytes. In pathological situations in man, it may become highly expressed in a subset of nonpituitary endocrine tumors, especially in bronchial carcinoids, leading to ectopic secretion of ACTH with the clinical signs of Cushing’s syndrome (1). By contrast to the molecular mechanisms of POMC gene expression in pituitary corticotrophs, those in nonpituitary tumors remain largely unknown. We previously showed that POMC gene expression in a small cell lung carcinoma (SCLC) cell line did not use the pituitary transcription factors already characterized (2, 3, 4) but rather depended on noncorticotroph-specific factors activated during a particular type of tumoral transformation (5). However, bronchial carcinoid tumors are highly differentiated and differ from SCLCs by their marked endocrine phenotype, with numerous dense core secretory granules, and a slow proliferation rate (6). Bronchial carcinoid transformation is likely to be different from that of SCLCs and may not involve the same factors able to activate POMC promoter. The bronchial carcinoid tumors responsible for ectopic ACTH- secretion (ACTH+) are indistinguishable from silent, non-ACTH-secreting (ACTH-) ones, by morphological/histological examination and analysis of biological properties such as proliferation or presence of metastasis (7, 8). Nevertheless, ACTH+ bronchial carcinoids often display multiple characteristics of the pituitary corticotroph phenotype, making them difficult to differentiate from authentic pituitary corticotroph tumors during clinical investigations (9, 10). We previously showed that POMC mRNA in ACTH+ bronchial carcinoids is qualitatively identical to that of pituitary corticotrophs and is expressed at levels comparable, if not higher, to those of pituitary tumors (11, 12). Similarly, the vasopressin V3 receptor gene, another molecular marker of corticotroph phenotype, is also expressed in these tumors and is functionally coupled to ACTH release (13, 14, 15). Furthermore, ACTH- and ACTH+ bronchial carcinoids are in many aspects very comparable regardless of their secretory status.
Thus, we hypothesized that such bronchial carcinoid tumors may be a valuable model to identify new genes involved in—or associated with—the corticotroph phenotype. In this context, expression of POMC and the genes associated with the corticotoph phenotype should be one of the main differences between such tumors, thus facilitating their identification by a differential approach. We report here the identification, by differential display RT-PCR (DD/RT-PCR) analysis between groups of ACTH+ and ACTH-, of genes preferentially or specifically expressed in one or the other group. These genes should help characterize and understand corticotroph differentiation in bronchial carcinoids and discriminate subtypes among ACTH-.
Materials and Methods
Tissue collection and RNA isolation
Tumors were obtained at surgery, immediately frozen in liquid nitrogen, and stored at -80 C until processed. The normal human pituitaries were obtained during autopsy, between 12 and 24 h postmortem. Total RNAs were extracted with the guanidium isothiocyanate/phenol method (16) and treated with 1 U deoxyribonuclease I for 30 min at 37 C to remove contaminant genomic DNA traces.
DD/RT-PCR
RNAs (2 µg) were reverse transcribed for 1 h at 42 C with 200 U Maloney murine leukemia virus transcriptase (Invitrogen Life Technologies, Carlsbad, CA) and 25 ng/µl poly-deoxythymidine(12–18), and the cDNA solution was diluted 10-fold. Differential display was performed with the 
RNA Finger Printing kit (CLONTECH Laboratories, Inc., Palo Alto, CA) according to the manufacturer’s recommendations with minor modifications. Briefly, 2.5 U Expand Long Template DNA polymerase (Roche Molecular Biochemicals, Mannheim, Germany) were used in PCRs containing 2 µl diluted cDNA, 350 µM deoxy (d) NTP, 2.5 µCi 
33P-dATP (2000 Ci/mmol, Amersham Pharmacia Biotech, Buckinghamshire, UK), and 0.3 µM each of arbitrary decamer and two base-anchored oligo-deoxythymidine primers. Amplification was carried out for 3 cycles at 94 C for 2 min, 40 C for 5 min, 68 C for 5 min, then 35 cycles at 94 C for 45 sec, 60 C for 1 min, 68 C for 2 min, followed by 72 C for 7 min. Ten percent of the radiolabeled PCR products were separated by electrophoresis onto 4.5% polyacrylamide/7 M urea denaturing sequencing gels (HR-1000, Genomyx, Foster City, CA). The dried gels were exposed to autoradiographic films (Biomax, Kodak, Rochester, NY) for 1–3 d. Differentially expressed bands were identified, excised, eluted, and reamplified with the same primer set. The cDNAs were directly purified with the High Pure PCR kit (Roche) or after gel electrophoresis with the agarose gel DNA extraction kit (Roche) and subcloned in pGemT-easy plasmid (Promega Corp., Madison, WI). Their nucleotide sequence was established by the dideoxy-nucleotide chain termination method and compared with databases with the BLAST and FASTA programs.
RT-PCR analysis
RNAs were reverse transcribed as previously described for DD/RT-PCR. Two microliters of cDNA (undiluted) were amplified for 25–35 cycles in reactions containing 250 µM dNTP, 100 ng sense and antisense specific primers, and 0.5 U Dynazyme II DNA polymerase (Finnzymes). The optimal number of cycles for each RT-PCR was determined in preliminary experiments to prevent inadequate or excessive amplification. Twenty percent of the PCR products were analyzed on agarose gels and blotted onto Nylon membranes (Hybond-N+, Amersham Pharmacia Biotech, Buckinghamshire, UK) after 45 min denaturation in 1.5 M NaCl, 0.5 M NaOH followed by 45-min neutralization in 1.5 M NaCl, 0.5 M Tris-HCl (pH 8.0). Specific primers used for POMC were CCTGCCTGGAAGATGCCGAGA and TGCTGCCGCTGCTGCTGCTGT; TCTCGGGTCAGCAGCATCAAC and ACCCCCACAGCAGGCAAGG for the V3 receptor; and CATCCGGTGCCTGCGAAACA and GGC CCTGGTAGATGGTAGTCGfor the type 1 CRH receptor. RT-PCR blotted membranes were probed under classical conditions with internal primers, end-labeled with (
32P)-ATP (5000 Ci/mmol, NEN Life Science Products, Boston, MA) and T4 polynucleotide kinase (New England Biolabs, Beverly, MA) (17).
Dot and Northern blot analysis
Total RNA (4 µg) was denatured for 10 min at 65 C in 50% formamide, 2.2 M formaldehyde in 10 mM phosphate buffer (pH 7.0), spotted onto Nylon membranes with a dot blot device (Millipore Corp.) as described (11) and hybridized like the Northern blot membranes. RNA blot membranes (CLONTECH Laboratories, Inc.) were hybridized with cDNA probes labeled by random priming (18) with (32P)-dATP (3000 Ci/mmol, NEN Life Science Products) using the Strip-EZ DNA kit (Ambion, Inc., Austin, TX). Membranes were exposed to X–OMAT films (Kodak) at -80 C with intensifying screens (Dupont, Wilmington, DE) for 1–4 d. Dehybridization was performed according to the Strip-EZ DNA kit instructions (Ambion, Inc.).
Results
Tissue selection
We constituted two groups of typical bronchial carcinoids composed of tumors responsible for the ectopic ACTH syndrome (ACTH+) or tumors without characterized endocrine secretion (ACTH-). These tumors were selected to be as close as possible in their morphology and histology. All tumors expressed chromogranins and 7B2, but only those responsible for ectopic ACTH secretion contained detectable (generally very high) levels of ß-endorphin (19). We also analyzed by RT-PCR the expression of POMC gene and of the vasopressin V3 receptor and the CRH receptor 1 (CRH-R1) genes, two markers of pituitary corticotroph cells (Fig. 1
). As expected, the POMC and V3 genes were exclusively expressed in all ACTH-secreting tumors, whereas the CRH-R1 gene was expressed in approximately half of the tumors regardless of ACTH secretion. We selected two ACTH+ and three ACTH- tumors with comparable levels of chromogranins and 7B2 and included in each group CRH-R1 positive and negative tumors, to exclude these known genes from the pool of differentially expressed genes.
|
Genes differentially expressed in the ACTH+ and ACTH- groups were identified by the differential display technique [DD/RT-PCR (20)] combined with an improved electrophoretic separation of DD/RT-PCR products enabling to discriminate fragments as long as 2.5–3 kb and to isolate cDNAs longer than usually reported (21, 22). An extensive analysis was performed, using a total of 90 combinations of arbitrary/anchored-oligo-deoxythymidine primers, allowing to visualize approximately 20,000 cDNAs ranging in size from 0.3 to more than 2 kb. Despite the inherent heterogeneity of the tumor samples, about 300 cDNAs equally distributed in each group were initially isolated as differentially expressed candidates. Differentially expressed genes were defined as genes exclusively detected or overexpressed in all members of one group of tumors compared with the other group. Figure 2 shows the typical profiles of exclusively expressed genes (bands 109.7 in ACTH+, and 67.3 in ACTH-) and others overexpressed in all members of the considered group (bands 56.2 and 22.3 in ACTH+). On the opposite, genes that were expressed in some but not all tumors of a group were excluded from this initial screen (data not shown).
|
Nucleotide sequence was established for the remaining DD/RT-PCR fragments, allowing to design the PCR primers necessary for further characterization and to identify some of the differentially expressed genes. The expression of each of these candidate gene was analyzed by RT-PCR in a larger series of bronchial carcinoids including 8 with proven POMC expression and 12 without clinical endocrine manifestations (Table 1 and Fig. 3C). Two cDNAs, 109.7 and 71.4 (Fig. 3A), showed a pattern of expression restricted to ACTH+ tumors, and two others, 67.3 and 66.4 (Fig. 3B) to the ACTH- group. The cDNAs (24.3, 43.2, and 60.1) were overexpressed in ACTH+ group and the cDNA 10.2 in ACTH- group, whereas the other cDNAs had a more heterogeneous pattern, being expressed in most carcinoids of one group with occasional detection in some samples of the other group (for example, see cDNAs 22.3, 50.1, 51.1, or 102.5 in Fig. 3A). The equivalent RT-PCR signals detected for the ubiquitously expressed glyceraldehyde 3-phosphate dehydrogenase gene (Fig. 3C), demonstrated that the variations of RT-PCR signals observed for the differential candidates reflected authentic variations of expression. Table 2 lists the cDNAs selected as differentially expressed after this stringent RT-PCR analysis. DD/RT-PCR fragments corresponding to several genes were repeatedly isolated with different combinations of primers, among which POMC, cFos, and a large expressed sequence tag (EST) named KIAA1775. The 108.2 and 109.1 cDNAs were identical and represented the 5' part of the longer 109.7 cDNA, which is likely to correspond to a 3' variant generated by the use of alternative polyadenylation sites. They were identical to several long EST (DKFZp434A132, KIAA1775) encoding a putative protein with cadherin-like domains (23). The DD/RT-PCR fragment 67.3 corresponded to TM4SF5, a member of the tetraspanin gene family. The 22.3 cDNA corresponded to an EST (DKFZp586N2022) encoding a putative protein with so-called sushi-repeats, presenting homologies with other known proteins (24, 25). The other cDNAs were similar to genomic sequences or short ESTs present in GenBank or EMBL databases, and two were only identified in our study. Interestingly, this analysis also suggested that the ACTH- carcinoid group may be divided in subgroups according to the profiles of expression of several genes studied here. For instance, both 67.3 and 66.4 cDNAs were detected at much lower levels in ACTH- tumors of rows 2, 5, 7, 9, and 12 in Fig. 3 than in the other tumors of the same group.
|
|
|
). RT-PCR analysis showed very variable distributions among the genes. As shown in Fig. 4A, low levels of the 22.3, 43.2, and 51.1 cDNAs were detected in normal pituitary and higher levels in ACTH-producing tumors. However, these genes were also expressed in PRL and GH-secreting tumors indicating that they are not specifically associated with the corticotroph phenotype. This observation was a general feature of cDNAs characterized as ACTH+ in bronchial carcinoids, and none of them had a pattern of expression following that of POMC outside the bronchial carcinoid tumors group. The 109.7 cDNA, specific of the ACTH+ carcinoid group, was also detected at high levels in all corticotroph pituitary adenomas and in the majority of the other types of pituitary tumors examined, whereas it was barely detectable in normal pituitary. In contrast, it was not present in normal lung and only occasionally in nonendocrine bronchial tumors (Fig. 4B). The expression of the 67.3 cDNA (TM4SF5), initially identified in ACTH- tumors, was strikingly restricted to some nonendocrine bronchial tumors; and was only detected in three pituitary tumors including two corticotropes (Fig. 4B).
|
|
). In addition, 109.7 mRNA was detected by Northern blot in poly-A+ RNA from normal brain as a 5.5- to 6-kb band and in the colon as a 7.5-kb and a 6-kb mRNA species (Fig. 5B). It was not detectable in endocrine tissues such as pancreas, thyroid, adrenals, thymus, or gonads (data not shown). However, some of these tissues contain only a small population of neuroendocrine cells, and the apparent absence of 109.7 expression may be due to a dilutional effect. Further studies with samples enriched for endocrine cells such as purified pancreatic islets or islet-cell tumors should help to determine 109.7 expression profile in endocrine tissues.
|
We compared the genes expressed in two groups of bronchial carcinoid tumors by the differential display technique to isolate those overexpressed, or repressed, in ACTH-secreting tumors and thus identify new markers of their corticotroph phenotype. This technique was preferred over other differential methods such as microarrays, representational difference analysis or SAGE for its ability to identify and clone new genes or ESTs (20). Despite the careful selection of the tumors, the DD/RT-PCR analysis clearly showed an important variability between the tumors within each group. This variability, inescapable with tumor samples, was indeed an advantage because DD/RT-PCR was performed on a group of tumors sharing the principal traits of the ACTH+ phenotype. Thus, within a given group the genes that are variably expressed cannot be specifically associated with the principal traits that define the group.
The differential expression of the initial approximate 300 candidate cDNAs was analyzed by several techniques including a sensitive and stringent RT-PCR analysis on larger populations of ACTH+ and ACTH- carcinoid tumors. Most of the candidates were eliminated during this step because they were only detected in a limited number of tumors (essentially those used for DD/RT-PCR screening) of one or both groups or because they were true false-positive, widely expressed.
We finally selected the cDNAs corresponding to 15 differentially expressed genes, including 11 specific or overexpressed in ACTH+ tumors, of which two had not been isolated before, and 4 in ACTH- tumors. The relatively small proportion of ACTH- specific genes may be related to the potentially larger heterogeneity of tumors within this group essentially characterized as silent, yet two genes appeared very specifically excluded from the ACTH+ tumors. Furthermore, these cDNAs were detected in an identical subset of ACTH- tumors, and their characterization may help define different subpopulations within the silent carcinoids. Most cDNAs were identified through nucleotide sequence comparison with databases. However, it has not been possible to determine the nature of the corresponding genes and proteins because most of these database entries arise from the human genome sequencing project and remain functionally uncharacterized. Besides the expected differentially expressed POMC gene, we also repeatedly isolated cDNAs coding for cFos. They were isolated from ACTH+ tumors, where cFos mRNA was generally detected at low levels by RT-PCR, yet superior to those in the majority of ACTH- tumors. Although it has been reported to play a role in POMC gene expression (26, 27), it is also well known for its wide expression and its actions on many genes. This increased expression may, however, participate in the high levels of POMC expression detected in such tumors, although no clear relation between the levels of the POMC and cFos mRNA could be established (28).
Three other cDNAs (109.7, 67.3, and 22.3) presented sequence identities with known genes. The 109.7 cDNA, together with the shorter 108.2 and 109.1 cDNAs, were only detected in ACTH+ carcinoids and corresponded to several large ESTs and to a single gene located on chromosome 10 and called KIAA1775. The putative coding region of this gene encodes a large protein (845 residues at least) with one transmembrane domain at its C-terminal end and six cadherin-like motifs in its putative extracellular domain (23), defining this protein as a nonclassical member of the cadherin family. Protocadherins have been mainly involved in development processes, particularly in the brain and also in various neuronal functions (for review, see Ref. 29). In the case of KIAA1175, it was originally reported to be specifically expressed in the olfactory bulbs of the brain, but partial ESTs have also been identified in Retin-A, pineal gland, as well as tumors from the colon, lung, and uterus. The identity between 109.7 and KIAA1775 was confirmed by RT-PCR using primers designed from both sequences, which showed the same tissue distribution among carcinoids and other tumors, whether endocrine or not. The expression of this gene appeared restricted to the ACTH+ carcinoids but was not specifically associated with the corticotroph phenotype outside of the bronchial carcinoid tumors. Its wide expression in pituitary tumors, together with its specific expression in ACTH+ carcinoids and its absence or weak expression in nonendocrine lung tumors, suggests that it may be involved in neuroendocrine functions of pituitary cells, which are also present in ACTH+ carcinoid tumors. It is also possible that multiple mRNAs with different coding sequences may be generated from the KIAA1775 gene. Indeed, preliminary 5' RACE PCR results with 109.7 primers showed the presence of several splicing variants within the coding region (de Keyzer, Y., unpublished observation) and Northern blot analysis detected mRNA with different size in normal brain and colon. Further studies are needed to determine the structure and function of the KIAA1775 gene products in ACTH+ bronchial carcinoids and other tissues.
The 67.3 cDNA corresponded to the TM4SF5 gene, which belongs to the tetraspanin gene family, and was found exclusively expressed in ACTH- carcinoids with the exception of some nonendocrine lung tumors. This large gene family encodes small proteins characterized by a common structure with four transmembrane domains. The TM4SF5 protein is a member of the tetraspanin, but it has original characteristics that are also found in the L6 antigen, TM4SF4 and il-TMP proteins and defined the L6 subfamily (30, 31, 32). Some tetraspanins are found in almost all tissues, whereas others are highly restricted, and they might act as molecular facilitators in cell adhesion, migration and proliferation processes by bringing together with various proteins at the cell surface (33, 34). For instance, membrane proteins such as integrins have been shown to modulate cell signaling through interaction with numerous proteins, among which several of the tetraspanin family (35). Interestingly, the TM4SF5 gene was also reported to be overexpressed in some pancreatic adenocarcinomas (36).
The last differentially expressed gene identified during this study was the 22.3 cDNA. It is preferentially expressed in the ACTH+ group and was identified as a long EST isolated from uterus and coding for a putative protein containing short consensus repeats also called sushi-repeats. Partial identities have also been observed with other proteins containing such repeats, which have been already described but whose functions remain obscure (24, 25, 37). The sushi-repeats were first identified in plasma ß2 glycoprotein and factor XIII (38, 39) as well as in members of the selectin family (40), where it may play a role in cell adhesion. The putative sushi-repeat containing protein encoded by the 22.3 cDNA in ACTH+ carcinoids may also be a membrane protein but its role remains to be elucidated.
Therefore, three genes identified in our study potentially encode transmembrane proteins, pointing to the importance of membrane-associated mechanisms in the differentiation of carcinoid tumors. Interestingly, recent studies suggested a role for proteins of the extracellular matrix in the regulation of ACTH production and proliferation of tumoral pituitary corticotrophs (41, 42). In addition, these surface proteins could provide possible antigen-based diagnostic approaches for carcinoids, whether ACTH+ or ACTH-. Only one transcription factor has been identified so far, but others may be present among the unknown genes, which may play a role in POMC or other corticotroph specific genes.
In conclusion, our study shows that differential display is a useful approach to isolate genes overexpressed or repressed by comparing different phenotypes within a single type of tumor. The genes isolated here and already identified, by their putative cell membrane location, might constitute targets for pharmacological agents and help to locate or characterize the, often occult, carcinoid tumors responsible for the ectopic-ACTH syndrome. The proteins encoded by this set of cDNAs should help understand the mechanisms underlying the corticotroph phenotype of the carcinoids and allow comparison with those active in normal and tumoral pituitary corticotrophs.
Acknowledgments
We thank Sophie Boucher for setting up the DD/RT-PCR system, Nicolas Lebrun and Franck Letourneur from the sequencing facility of our institute, and Dr. Marc Lombès for critical reading of the manuscript.
Footnotes
This study was supported by grants from the Ligue Nationale Française Contre le Cancer and from the Legs Poix (to Y.d.K.). L.P-L.T. is the recipient of a fellowship from the Ministère de l’Education Nationale de la Recherche et de la Technologie.
Abbreviations: DD/RT-PCR, Differential display RT-PCR; d, deoxy; EST, expressed sequence tag; POMC, proopiomelanocortin; SCLC, small cell lung carcinoma.
Received April 16, 2002.
Accepted July 30, 2002.
References
This article has been cited by other articles:
 |
|
 |
|
  M Messager, C Carriere, X Bertagna, and Y de Keyzer RT-PCR analysis of corticotroph-associated genes expression in carcinoid tumours in the ectopic-ACTH syndrome Eur. J. Endocrinol., January 1, 2006; 154(1): 159 - 166. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| 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 |
