Endocrine Oncology Site Group Mount Sinai and Princess Margaret Hospitals University of Toronto Toronto, Ontario, M5G 1X5 Canada
Address all correspondence and requests for reprints to: Shereen Ezzat, M.D., University of Toronto, Mt. Sinai Hospital, 600 University Avenue #437, Toronto, Ontario, M5G 1X5 Canada. E-mail: sezzat{at}mtsinai.on.ca.
The pathogenesis of pituitary tumors continues to be an enigma in the field of endocrine oncology. Despite their prevalence and potential for significant morbidity, the etiologies of most pituitary tumors remain unknown. Hypotheses of pituitary oncogenesis have vacillated in the last two decades, moving from one milestone to the next. Like other differentiated neuroendocrine cells, the anterior pituitary displays remarkable plasticity in response to physiological demands, as exemplified by the lactotroph differentiation and proliferation of pregnancy or the thyrotroph hyperplasia of primary hypothyroidism. These reversible changes are mediated by a diverse array of signals that have been interpreted to support a role for hormonal stimulation in the pathogenesis of pituitary adenomas (1). Subsequently, the early years of molecular biology applications permitted clonality assessment based on the X-chromosome inactivation principle. The majority of these studies suggested a clonal pattern, providing the basis for assignment of pituitary adenomas to the list of monoclonal neoplasms. The stage was set for the search for that pivotal genetic event that was necessary and sufficient for pituitary neoplastic transformation.
Historical perspective
One of the first candidate factors to emerge was the G protein -stimulating activity polypeptide (GSP). Activating mutations of this protein that lead to constitutive elevation of adenylyl cyclase activity have been identified in nearly one third of somatotroph adenomas (2). A second major breakthrough was the identification of the putative tumor suppressor gene encoding the nuclear protein menin. Clinical genetic studies linked inactivating mutations of this gene with several families harboring the multiple endocrine neoplasia type 1 (MEN1) phenotype associated with pituitary, parathyroid, and pancreatic endocrine tumors. Disappointingly, however, there has been little evidence that mutations of the MEN1 gene itself play a common role in sporadic pituitary tumors (3). Nevertheless, the possibility that a component of the encoded menin signaling cascade plays a role in pituitary tumor pathogenesis remains open.
Approaches to defining new candidate pituitary oncogenes
Expression profiling.
After seemingly exhausting many of the known candidates implicated in the genesis of nonendocrine solid neoplasms, it became apparent that a more endocrine- or tissue-specific search would be required. Two major strategies have been applied in this regard. The first has been driven by differential display techniques. Identified in a differential mRNA expression analysis of transformed rat pituitary tumor cells, the pituitary tumor-transforming gene PTTG (4) led a series of novel reports. The protein product was subsequently characterized as a member of the securin family that functions in regulating chromatid separation (5). The extent to which it contributes directly or indirectly to pituitary tumor formation remains an important question. Similarly, a cDNA microarray approach singled the folate receptor gene as being significantly overexpressed in nonfunctioning pituitary adenomas compared with other tumor types (6). Subsequently, a cDNA representational differential display study comparing normal human pituitary tissue and clinically nonfunctioning pituitary adenomas identified GADD45 as a candidate gene that was highly underrepresented (7). The biological plausibility of this finding was supported by the well recognized features of GADD45 as a member of a growth arrest and DNA damage-inducible genes.
A maternally expressed gene (MEG3).
In this issue of JCEM, Zhang et al. (8) report a new candidate gene that they noted in their differential display analysis to be underexpressed in pituitary adenomas. The sequence that they identified represents a distinct 141-bp extended form of the maternally imprinted gene MEG3. The failure to splice exon 5 leads to what Zhang et al. (8) termed the MEG3a isoform of this gene. They show the corresponding cDNA to be expressed in human nontumorous anterior pituitary gonadotrophs and other normal adenohypophysial cells. Using primers specific for the MEG3a isoform, the coding sequence was found to be diminished or absent in a number of cancer cell lines as well as in primary pituitary tumors of different types, including gonadotroph adenomas. Introduction of the MEG3a sequence into heterologous cell lines resulted in significant growth arrest, consistent with a suppressive function for this novel gene isoform. It remains to be shown whether this sequence that lacks a classical kozak signal is translated. Deletions of this 14q32.3 locus have been described previously in meningiomas but not in human pituitary tumors (9). The latter findings suggest that loss of MEG3 function in pituitary tumors could be the subject of epigenetic changes. Interestingly, GSP mutations are more frequently detected in the maternal allele (10), providing yet another example of a maternally imprinted gene in pituitary tumorigenesis.
Developmental profiling.
Increasing evidence suggests that highly evolved and conserved tissue-specific signaling events may bear strong overlap with neoplastic events. These biological reflections have provided an increasingly attractive rationale for searching for new candidate oncogenes among components of known crucial events in normal development. In the case of the pituitary, many growth factors and their receptors have been suspected to fit these predictions (1).
Early pituitary development
Early pituitary development is governed by extrinsic patterning signals and endogenous gene expression. The early extrinsic signals that are most important arise from the ventral diencephalon and include members of three protein families: bone morphogenic protein (BMP), Wnt, and the fibroblast growth factor (FGF) family.
BMP signaling by BMP4 is required for normal pituitary development (11). Targeted expression of a BMP2/4 antagonist results in the arrest of Rathkes pouch invagination, whereas deletion of BMP4 causes embryonic death with complete failure of pouch invagination. Interestingly, mRNA differential display has recently singled the BMP inhibitor noggin as being down-regulated and BMP4 as up-regulated in lactotroph adenomas from dopamine D2 receptor-deficient mice and in human pituitary tumors. BMP4 was further shown to selectively stimulate lactotroph cell proliferation through a Smad4-dependent pathway (12).
In contrast, the role of FGF signaling seems to be more critical in pituitary development after initial invagination of Rathkes pouch. The deletion of FGF10 or its receptor, the FGFR2 IIIb isoform, leads to failure of primordial pituitary gland development (13). Mid-gestational expression of a soluble dominant negative FGFR results in complete agenesis or severe dysgenesis of the pituitary along with craniofacial and limb abnormalities (14). The normal human pituitary expresses mRNAs for FGFR 1, 2, and 3 (15). An interesting finding was that truncated mRNAs for the first and second Ig-like loops of FGFR4 are expressed by nontumourous pituitary (15) and a kinase-containing variant of FGFR4 with an alternative initiation site was expressed by pituitary tumors (16). The latter is a pituitary tumor-derived kinase-containing FGFR4 isoform that causes transformation in vitro and in vivo and when selectively expressed results in pituitary tumor formation in transgenic mice (16). In marked contrast, transgenic mice that express wild-type FGFR4 do not develop hyperplasia or adenomas.
Validation of new candidates
Do the current studies indicate that MEG3a is responsible for pituitary tumorigenesis? The answer depends on the extent to which a candidate factor is able to satisfy basic criteria. Establishing a causal link between a putative factor and human tumorigenesis requires multiple lines of evidence. First, the factor must be shown to be expressed in the cell type of interest. Even this fundamental step is sometimes difficult in the pituitary. Limited amounts of human tumorous tissue often preclude conclusive studies such as Northern blotting. Similarly, the lack of reliable antibodies often precludes Western immunoblotting. Second, the expression of the candidate factor must be significantly deregulated (underexpressed or overexpressed) compared with other unaffected cells of the same lineage. Normal human pituitary cells are even more scarce in quantity and quality, further limiting the number of studies that can be performed. Third, experiments of nature have dictated that the deregulated expression of candidate genes is typically the subject of genetic alteration such as a somatic mutation or alternatively an epigenetic change in methylation or acetylation of the control elements of the gene. Fourth, gain or loss of function of the putative factor in a controlled cell model should influence some measure of growth, including transformation and/or proliferation. This is often performed, as in the study by Zhang et al. (8), by means of transfection of the candidate factor in a cell type with recognizable growth properties in an assay such as colony formation in soft agar. Fifth, gain or loss of function of the putative factor in a controlled mouse model should display similar properties. This is often performed by means of stable introduction of the gene of interest into a cell that is inoculated into an immune-compromised host such as with the nude mouse model. This provides evidence of transforming and invasive growth potential but does not necessarily establish tumorigenic causality. Sixth, gain or loss of function of the putative factor in a genetically altered mouse model should yield morphological features of a pituitary adenoma. There are many examples of factors that, intentionally or unintentionally, resulted in aberrant pituitary growth. However, detailed histological examination of such genetically altered mice have frequently displayed features of hyperplasia alone or with adenoma transition (17). Neither phenotype, however, accurately mimics what is seen in the human situation. This raises questions about the biological relevance of the candidate factor or the model in which it is examined. In fact, one of the more characterized somatic mutations in endocrine oncology has been the Gs oncogene. Introduction of the activating mutant GSP allele has been better displayed in transgenic mice, leading to hyperplasia and adenoma formation in thyroid follicles but not in the pituitary (18). Another mouse model of the MEN1 inactivation has been developed by gene targeting. Heterozygous mice develop multiple endocrine tumors, including prolactin-producing pituitary adenomas characterized by somatic loss of the wild-type allele reminiscent of patients with MEN1 (19). Seventh, ideally, the mechanisms of action of the candidate factor should bring functional integration with known signaling cascades recognized to be important in pituitary development and/or tumorigenesis. Finally, specific interruption of the signaling cascade should lead to reversal of the underlying pituitary pathology.
Future perspectives
The end of the human genome project has heralded a new era in postgenomic biology. We have come to know fundamental molecular structures and have the tools to predict an ever-growing list of protein sequences, their conformations, and their interactions. The individual contributions of many of these components in the complex cascade of events leading to human neoplasia is finally unraveling. It is, thus, becoming increasingly evident in many systems that not one single factor can effectively explain the many facets of the tumorigenic process. Nevertheless, cellular and molecular biologists are now poised to make functional links in an expanding biological dictionary. The challenge for translational researchers, including endocrinologists, will be to systematically apply this knowledge in a manner that validates biological predictions. Identifying and validating the functions of novel factors such as MEG3a in an endocrine model such as the pituitary is well in line with this mission.
Footnotes
Abbreviations: BMP, Bone morphogenic protein; FGF, fibroblast growth factor; FGFR, FGF receptor; GSP, G protein -stimulating activity polypeptide; MEG, maternally expressed gene; MEN1, multiple endocrine neoplasia type 1.
Received September 18, 2003.
Accepted September 18, 2003.
References
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 |