This Article Abstract Full Text (PDF) Purchase Article View Shopping Cart Alert me when this article is cited Alert me if a correction is posted Citation Map Services Email this article to a friend Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Request Copyright Permission Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Dahia, P. L. M. Articles by Grossman, A. B. Search for Related Content PubMed PubMed Citation Articles by Dahia, P. L. M. Articles by Grossman, A. B.

The Molecular Pathogenesis of Corticotroph Tumors

Department of Endocrinology, St. Bartholomew’s Hospital, London EC1A 7BE, United Kingdom

	Abstract

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 

I. Introduction

A. Origin of tumors: hypothalamus vs. pituitary

B. Clonality of pituitary tumors

II. Protooncogenes

III. Tumor Suppressor Genes

A. p53

B. p16/CDKN2/MTS1/INK4 and Rb1

C. p27/KIP1/CDKN4

D. MEN1

E. hZAC

F. NM23

G. Other tumor suppressor genes

IV. Specific Genes

A. Regulatory receptor genes

B. Cytokines and growth factors

C. Developmental genes

V. Miscellaneous

A. Other genes

B. Methylation

VI. Perspectives: Old and New Tools for Understanding Pituitary Pathogenesis

A. Transgenic/knockout animal models

VII. Conclusions

	I. Introduction

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
TO FULLY understand the ontogeny of corticotroph tumors, some understanding is required of normal pituitary cell development. Pituitary ontogeny is directed by a complex myriad of factors, including the homeobox genes, that are expressed at distinct and highly specific phases of pituitary development (1). The anterior pituitary gland arises embryologically from Rathke’s pouch, while the posterior lobe originates from the ventral hypothalamus. Interactions between the two tissues are essential for their differentiation. Once committed to a pituitary ’fate,’ the cells from Rathke’s pouch proliferate and differentiate into specific cell lineages that secrete POMC, GH, PRL, TSH, FSH, and LH (1) (Fig. 1). Control of pituitary cell proliferation and gene expression is provided by hypothalamic peptides and their specific receptors located in the pituitary. The gene encoding the -glycoprotein subunit, -GSU, the common component of the heterodimeric TSH, LH, and FSH, is activated very early in pituitary development. Activation of the transcription factor Pit-1 mRNA occurs as an organ-specific event, initially in all five cell types (2). However, the Pit-1 protein is detected in only three cell types in the mature pituitary gland: the lactotrophs, somatotrophs, and thyrotrophs, but is not found in either gonadotrophs or corticotrophs (2). Interactions between specific activating and restricting factors with Pit-1 in distinct temporal patterns contribute to defining the specificity of the three cell lineages. However, much less is known about the origin and maturation of the corticotroph, although there is evidence from studies in mice that commitment to the corticotroph might occur even earlier than the expression of -GSU (3, 4). The homeobox gene Ptx1 is expressed in most cells of Rathke’s pouch at an early stage of pituitary development, and before the final differentiation of hormone-producing cells. Thus, Ptx1 seems to play a role in the differentiation of pituitary cells and possibly also in the formation of the specific gland. In the adult pituitary, by contrast, Ptx1 appears to be recruited for cell-specific transcription of the POMC gene (4). Pituitary leukemia-inhibitory factor (LIF) is expressed early in the development of the pituitary and enhances POMC expression in synergism with CRH (5). However, LIF exerts an antiproliferative effect on the corticotroph. These interactions between various transcription factors and hypothalamic peptides in coordinating pituitary embryogenesis are outlined in Fig. 1. It is the purpose of this review to explore our current knowledge in understanding how this process may become deranged in the development of corticotroph adenomas.

View larger version (18K): [in this window] [in a new window]   Figure 1. Simplified schematic representation of pituitary ontogeny with putative factors involved in differentiation of specific lineages.

Corticotroph tumors are of special interest in understanding the process of oncogenesis, as many of the biochemical characteristics of such tumors show only relative, rather than absolute, abnormalities compared with the normal corticotroph, which might suggest that the somatic defect that occurs in the corticotroph clone is quite subtle. This renders the corticotroph adenoma a very interesting model of tumor formation and one that should be amenable to systematic analysis. We will attempt to summarize and integrate the current knowledge on the potential involvement of a variety of molecular derangements, and highlight some of the more fruitful areas of present and future research.

B. Clonality of pituitary tumors
As mentioned above, the debate over the origin of pituitary tumors has settled for a local, pituitary origin for at least the majority of tumors, mainly as a result of the clonality studies. Clonal analysis substantiates the fundamental distinction between a polyclonal proliferation in response to a stimulatory factor vs. a monoclonal expansion of a genetically aberrant cell (14, 15, 16, 17). Clonality studies have shown that most of the pituitary adenomas tested are monoclonal in origin, compatible with the hypothesis that somatic defects precede clonal expansion of these cells and are likely to be involved in the tumorigenesis process (11). In the few cases where the samples were found to be polyclonal, admixture of nontumoral cells in the tumor preparation appears to have accounted for the artifactual polyclonality of at least some of these cases (10). It is possible that a minority of such tumors acquire a monoclonal pattern later in their development, after a prolonged phase of polyclonal growth. The mechanisms differentially regulating tissue growth in the monoclonal vs. "polyclonal" pituitary adenomas are still unknown.

	II. Protooncogenes

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
Several protooncogenes have been tested for abnormalities in corticotroph tumors, most of which have mainly been found to be negative. While some genes, such as RAS, c-ERB2/neu, c-MYC, and PKC have been associated with more aggressive pituitary tumors, these findings do not appear to relate specifically to the corticotroph lineage (18, 19, 20, 21). Other genes that have been tested and shown to be unaltered in the majority of pituitary tumors are RET (22), c-MYB, and c-FOS (23). Furthermore, mutations of genes encoding for the stimulatory Gs, inhibitory Gi, and phospholipase C-mediated (Gq) subunits of the G protein complex, which contribute to pituitary signaling, do not seem to be involved to any significant extent in the pathogenesis of corticotropinomas (24, 25). This is in contrast to the situation in somatotropinomas, where approximately 40% of the tumors carry mutations of the stimulatory -subunit of the G protein complex (26). The reasons for this tissue specificity are not clear.

A recently identified rat gene that appears to have strong transforming properties is the pituitary tumor-transforming gene (PTTG), which has been cloned by differential display analysis: a unique sequence was found to be expressed in rat GH- and PRL-secreting tumor cell lines, while it was absent in normal pituitary and in an osteogenic sarcoma cell line (27). In vitro studies have shown that this product, which spans 199 amino acids with no homology to known domains, is able to induce cell transformation and, interestingly, inhibits cell proliferation (27). The human homolog of PTTG has now been cloned and found to possess transforming properties in vivo and in vitro (28). Initial studies have suggested that PTTG functions via SH3-mediated signals and induces fibroblast growth factor (FGF) expression in transfected cells. While an initial analysis has not identified mutations in pituitary tumors, increased expression of PTTG was found in more aggressive pituitary tumors and other human tumors and cell lines (28, 29). Further studies aiming at a functional characterization of the role of the PTTG gene in pituitary cell proliferation and its association with more aggressive tumor phenotypes are warranted.

	III. Tumor Suppressor Genes

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
A. p53
The p53 gene has been considered the most frequently mutated gene in human malignancies, being abnormal in approximately 50% of all human malignant tumors (30, 31). Its product has tumor suppressor activity by blocking the progression of the cell cycle from G₁, principally by activating p21, which in turn inhibits certain cyclin-dependent kinases (32, 33, 34, 35, 36, 37, 38). Mutations of the p53 gene have been found to cluster in the DNA-binding domain (DBD) of the protein, represented by exons 5–8 of the gene, and most of the mutation screening studies have concentrated on this ’hot spot’ region (30, 31, 39).

Several studies investigating p53 status in corticotroph tumors are available. While more than 50% of both invasive and noninvasive corticotroph adenomas from one series showed abnormal p53 immunostaining (40), no mutations were found in an independent series examined at the nucleotide level (41, 42). These apparent discordant results of sequence and protein analysis may be explained by the fact that mutations might have been located outside the region of the p53 gene screened. Alternatively, the positive p53 protein staining observed, suggestive of a p53 abnormality, could have been due to increased expression of the wild-type p53 protein as a result of its up-regulation by a regulatory gene such as MDM2 (43, 44), rather than a direct disruption of the p53 gene. On the basis of the existing data, the potential role of p53 as a major contributor to pituitary tumorigenesis, in particular with regard to the corticotroph lineage, cannot be entirely excluded but appears unlikely.

Recently, a gene was cloned on chromosome 1p35, one of the regions found to be a target for loss of heterozygosity (LOH) in endocrine tumors (45, 46, 47, 48). The novel gene, p73, encodes a product that not only resembles p53 structurally, but also appears to interact with it by forming heterodimers (48, 49). Although no structural abnormalities of this gene were found in an initial series of neuroblastoma cell lines and primary tumors with LOH at the 1p35–36 region, it has been noted that most of the samples examined express a single p73 allele, even when the two alleles were present in the germline (48). This p73 monoallelic expression appears to represent a novel mechanism of inactivation of a tumor suppressor gene, and further studies are needed to confirm this initial observation and unravel the potential role of the p73 gene in human tumors.

Another tumor suppressor gene whose product also appears to interact with p53 has recently been identified (50). The novel gene, p33^ING1, interacts with p53 in an interdependent manner in several of known p53 functions, such as negative regulation of cell proliferation, apoptosis, cell anchorage, and control of cellular aging, as well as mediation of p21 transactivation. Because of its role in the p53 signaling pathway, it will be of interest to verify whether p33^ING1 is implicated in the abnormal p53 expression seen in certain tumors, in particular corticotroph adenomas.

Additional p53-related genes have now been reported (51, 52, 53). While these genes are potentially interesting targets for studies in human tumors, their role in pituitary signaling and/or tumorigenesis has not as yet been explored.

B. p16/CDKN2/MTS1/INK4 and Rb1
The Rb1 gene product plays an essential role in regulating the progression of the cell cycle (54, 55, 56, 57). In its dephosphorylated form, the Rb1 product blocks cell cycle progression from G₁ to S phase. Progression of the cycle occurs when the Rb product is phosphorylated by complexes formed between specific subtypes of cyclins and cyclin-dependent kinases (cdk), acting downstream of p53 and p21 (58). Interest in the potential role of the RB1 gene in pituitary tumors arose from knockout mice models: all animals with one inactive RB1 allele developed pituitary tumors of the intermediate lobe (59, 60, 61). However, attempts to identify RB1 gene mutations have failed to show abnormalities of this gene in human tumors of any lineage (62, 63, 64). LOH at 13q14, the RB1 locus, has been identified in all 13 cases of a series of malignant or highly invasive pituitary tumors, in particular of the nonsecreting type, suggesting that another, yet unknown, tumor suppressor gene located in this region may be involved in pituitary tumorigenesis (23, 64, 65, 66), specifically those associated with a more aggressive phenotype.

The p16 gene has arisen as a frequent target for inactivation in several human tumors (67): it maps to chromosome 9p21, a region frequently deleted in human tumors. Through its regulatory role of the cyclin-cdk complex, in particular cdk 4, the p16 product exerts inhibitory effects on cell cycle progression from G₁ to S phase (68, 69, 70, 71). Cyclin D-cdk 4 complexes phosphorylate the retinoblastoma product, rendering it inactive and therefore unable to block cell cycle progression (72). In the absence of the p16 protein, the inhibitory loop ceases to exist, and cell cycle progression becomes unrestrained, giving rise to tumor growth and/or progression.

Because of the interactive pathways of p16 and Rb1 function, it was postulated that disruption of either product might also occur in an interactive manner. In particular, an apparent inverse correlation of their alterations has been observed in certain human tumors, such as lung carcinomas, in which high levels of p16 protein are seen in Rb1-disrupted tumors, while low or undetectable p16 protein occurs in tumors carrying wild-type Rb1 (73). It was therefore postulated that pituitary tumors could give rise to a disrupted Rb1 pathway secondary to a change in p16 function. Indeed, all 25 pituitary tumors from a series in which a nonspecified number of histological subtypes were represented failed to express the p16 protein product (74). More recently, a study has indicated that the mechanism involved in p16 silencing in the majority of pituitary tumors lacking p16 protein expression is hypermethylation of the promoter region of the gene (75). Full characterization of the clinical and biochemical features of pituitary tumors with a hypermethylated p16 gene, as compared with those with full p16 activity, awaits further study. In keeping with the finding of abnormal transcription of the p16 gene without genomic deletions, it has been shown that the p16 locus is intact in the majority of pituitary adenomas. However, about one-third of pituitary tumors from one series, including invasive and noninvasive forms, have been shown to have LOH at the 9p21 region (76). This finding suggests that another tumor suppressor gene located in this region might play a role in pituitary tumorigenesis.

C. p27/KIP1/CDKN4
It has recently been reported that mice with a genomic "knockout" of the p27^KIP1 gene develop multiorgan hyperplasia resulting in increased animal size (77, 78, 79). Similar to the RB1 knockout model, the P27 null mice have shown pituitary tumors that originate from intermediate lobe corticotrophs although, rather surprisingly, no other neoplasms were seen.

The p27^KIP1 gene, like p16, is a member of the cyclin-kinase inhibitor family of proteins (80, 81, 82, 83, 84). The p27 protein specifically inhibits complexes formed between cdk2 and cyclin E, required for entry into S phase from late G₁, and also between D-type cyclins and cdk4/cdk6 (85, 86, 87). These latter complexes mediate the phosphorylation of the Rb protein, which renders it inactive, allowing for the cell cycle to progress from mid-G₁. Due to the physiological interaction between p27 and Rb1 pathways, and also from the similarities in the phenotype of either knockout model, p27^KIP1 was clearly a possible candidate gene for pituitary tumorigenesis. We therefore recently examined 23 pituitary tumors, 21 of which were ACTH-secreting, including two with intermediate lobe features, for p27 abnormalities (88). Apart from a previously known polymorphism in the coding region of the gene, no structural alterations were found in the tumors. Also, no specific changes in p27^KIP1 transcription were detected in the tumors as compared with normal pituitaries. At the protein level, most of the tumor samples showed apparently normal p27 staining (although in our more recent ongoing studies we have also noted a relative decrease in p27 protein expression in corticotropinomas as compared with other tumor types). However, two of three malignant corticotropinomas examined, as well as one malignant prolactinoma, were negative for p27 immunocytochemically. This might indicate that p27 inactivation may be related to more aggressive histological subtypes of ACTH-secreting and possibly other pituitary tumors and may be a prognostic factor in such tumors. It has been suggested that p27 immunostaining may be related to more aggressive subtypes of colon and breast tumors, also indicating its potential role as a prognostic marker in such tumors (89, 90, 91, 92). As the distinction between malignant and benign tumors is extremely difficult for endocrine neoplasms, the existence of a marker that helps identify aggressive types in a more premature phase of tumor progression (before the occurrence of metastases) may become a powerful tool in therapeutic planning and prognostic assessment of pituitary tumors. Other recent studies on the potential role of p27^KIP1 in pituitary tumorigenesis have confirmed our findings of a lack of abnormalities at the nucleotide level (93, 94) and lower p27 protein levels in the malignant forms of tumors in comparison with normal pituitaries and benign adenomas (95, 96). One of these studies (95) suggested that the differentiation between benign and malignant tumors might be more quantitative than qualitative, since a graduated diminution of p27 protein levels was observed progressing from normal to benign to malignant samples. These data also suggest that the p27^KIP1 mechanism of inactivation is predominantly a posttranslational event. In support of this hypothesis is the finding of high levels of p27 degradation in proliferative cells (97, 98, 99). However, it still remains unclear as to why both RB1 and p27^KIP1 mice knockouts develop corticotropinomas, but abnormalities of these genes and their products are not seen in human Cushing’s disease. Perhaps these gene products may exert more relevant regulatory functions in other animal species, such as canine and equine models of Cushing’s disease, where the tumors more clearly arise from the pituitary pars intermedia (100, 101, 102), similar to the mouse tumor location. We hope, however, that continuing studies into the relative expression of p27 in corticotroph, as opposed to adenomas of other cell lineages, will help to explain these puzzling findings.

The chromosome 12p13 region, where p27^KIP1 is located, has been defined as a common target for genomic alterations in human malignancies. Using polymorphic markers located at the 12p13 area, we did not find LOH in any pituitary tumor analyzed (88). While seeking LOH in the same area, another study (94) detected, instead, chromosome 12 trisomy in five of eight samples that showed abnormal allelic ratios of polymorphic markers spanning the entirety of chromosome 12. However, as not all abnormal samples were available for fluorescence in situ hybridization analysis, it is possible that some degree of contamination with background normal cells might have accounted for the abnormal allelic ratios seen in some samples. The actual incidence of trisomy of chromosome 12 in pituitary tumors and the potential correlation with histological subtypes awaits further studies.

A recent study reporting a mouse model deficient in the cell-cycle inhibitor p18 revealed the presence of pituitary adenomas of the intermediate lobe (103), in a similar fashion to the p27 knockout mice mentioned above. Mice lacking both p18 and p27 invariably died from pituitary adenomas by 3 months of age. It appears, therefore, that p18 and p27 function in a collaborative manner to suppress pituitary tumorigenesis in mice. The role of p18 in human pituitary tumors still remains to be determined, but it will certainly be relevant to characterize whether, similar to the p27 and Rb knockouts, the mouse model differs from the human in yet another protein involved in the Rb-signaling pathway.

D. MEN1
While most often presenting as sporadic tumors, corticotroph adenomas may be occasionally familial, usually as part of the multiple endocrine neoplasia type 1 syndrome (MEN 1). In this setting, there is usually an association of pituitary tumors with parathyroid and pancreatic tumors as well as less common features, such as carcinoids, adrenal tumors, lipomas, angiofibromas and ependymomas (104, 105).

The susceptibility gene for MEN 1 has been mapped to chromosome 11q13 by linkage analysis (106, 107). The MEN1 gene has been recently identified by positional cloning (108, 109, 110, 111). This gene encodes a 610-amino acid protein product named menin that contains no recognizable structural or functional domains. Mutations of the MEN1 gene have now been reported in a variable number of MEN1 families, with the overall incidence of mutations ranging from 59% to more than 80% of the families studied (108, 109, 112, 113, 114). Although no functional studies involving mutant proteins are available to date, the mutations observed are expected to truncate menin. This finding has important consequences for the screening of members of MEN 1 kindreds. Since 10–30% of sporadic pituitary tumors have also been found to have LOH at the MEN1 locus (23, 115, 116, 117), suggesting that MEN1 acts as a tumor suppressor gene, studies looking for mutations of the MEN1 gene in sporadic pituitary adenomas are relevant. Unlike the findings in MEN 1-related tumors, several reports on the status of MEN1 in sporadic pituitary adenomas have revealed a very low incidence of mutations in such samples, irrespective of cell lineage (113, 118, 119). The expression of MEN1 mRNA also appears to be essentially normal in most sporadic tumors evaluated, including a small number of corticotropinomas (120, 121). The lack of down-regulation of the MEN1 gene was observed despite the presence of LOH at the MEN1 locus in some tumors. Furthermore, while mutations have been detected in all exons of the MEN1 gene, no clear genotype-phenotype correlation has been characterized (122).

Recent studies have demonstrated that menin localizes to the cell nucleus (123). Two nuclear localization signals (NLS) were identified in the C terminus of the protein, a region not found to be targeted by naturally occurring missense mutations and in-frame deletions. However, truncated protein products, resulting from frameshift and nonsense mutations, would lack at least one of these NLS, suggesting that they are important for the full activity of menin. While the protein function is still elusive, its nuclear localization suggests that it might be involved in regulating transcription of responsive elements.

E. hZAC
A novel zinc finger protein with tumor suppressor properties has recently been identified and mapped to chromosome 6q24–25 (124, 125). This gene, named hZAC or LOT1 by two independent groups, is the human homolog of a recently isolated transcript from the AtT20 mouse corticotropinoma cell line. By a functional expression-cloning technique, this gene was found to display a unique expression pattern: the pituitary was found to express the highest levels of zac1 transcript among several different tissues examined (126). Interestingly, this functional expression technique also resulted in the isolation of wild-type p53 clones, suggesting that the two products possibly interact. zac1 and p53 were found to induce expression of the gene encoding the type I receptor PACAP-R1, possibly indicating its role in regulating cAMP-mediated pathways. The human homolog has been found to share the same tissue distribution as the mouse gene, with the pituitary, kidney, placenta, and adrenals showing the highest levels of expression (124). Functional studies revealed that the antiproliferative properties of hZAC are related to its ability to induce apoptosis and cell cycle arrest in G₁, similar to p53 (124). This is the first human gene structurally unrelated to p53 that shares its apoptotic and cell cycle-blocking functions. Recently, hZAC expression was noted to be absent in ovarian tumor cell lines (125), and the hZAC1 locus is a commonly deleted area in several human cancers. However, it remains to be established whether this gene plays a role in pituitary tumorigenesis.

F. NM23
A gene encoding for a purine-binding factor with tumor suppressor properties, NM23, has been reported to be down-regulated in multiple invasive human cancers (127, 128, 129, 130, 131). One series of pituitary tumors has been examined for NM23 abnormalities: while no nucleotide alterations have been found, the more aggressive tumors showed reduced expression of one isoform, H2, of the NM23 transcript (132). In particular, an association was found between low H2 levels and cavernous sinus invasion. Whether this phenomenon of down-regulation represents cause or effect of the more aggressive tumor behavior remains to be established.

G. Other tumor suppressor genes
In the course of investigating the potential role of the RB1 gene in pituitary tumors, it had been noted that a subset of such tumors have LOH at the chromosome subregion 13q14 (23, 62, 64, 65). While the RB1 gene has been excluded as the target for deletion in that region, other tumor suppressor genes lying in the 13q14 area have been investigated for their potential role in pituitary tumorigenesis. One such gene is the BRCA2 tumor suppressor, which lies within 25 centiMorgans of the RB1 locus (66). Only two pituitary tumors, neither of which was a corticotropinoma, were found to have LOH at markers spanning both the BRCA2 and RB1 loci (66). These results suggest that yet another tumor suppressor gene in this area might be associated with the development of pituitary tumors.

A study designed to verify potential tumor suppressor loci associated with malignant pituitary tumorigenesis revealed LOH at chromosomes 1p, 3p, 10q26, 11q13, and 22q12 (133). Candidate genes at these regions might potentially be involved in determining or contributing to a more aggressive phenotype in pituitary tumors (23, 133).

	IV. Specific Genes

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
A. Regulatory receptor genes
Certain oncogenes and tumor suppressor genes examined for their role in neoplasia are associated with broad, nonspecific cell growth regulation and may therefore affect any cell type. On the other hand, genes encoding for proteins that participate in tissue-specific mechanisms of cell regulation would appear to be potentially good candidates for tumorigenesis of defined cell types.

Corticotrophs are regulated by several hormones whose actions within the cell take place through specific receptors (134, 135, 136, 137, 138, 139, 140, 141). ACTH secretion and cell growth are determined by a complex balance that regulates stimulatory impulses, such as the CRH receptor (CRH-R), the vasopressin receptor (specifically, the type 1b or type 3 receptor, V₃R), and the LIF receptor (LIF-R), as well as inhibitory regulators, such as the glucocorticoid receptor (GR). While POMC and ACTH secretion might be associated with corticotroph proliferation under the activity of certain factors, occasionally opposing signals result (Table 1). Several other receptors have been found to be expressed by the corticotroph, although their actions do not appear to be specific to this cell type. The potential role of the more corticotroph-specific receptor-encoding genes has recently been assessed, and the results are briefly discussed here. In addition, other receptors with potentially important roles in corticotroph regulation are mentioned.

View larger version (23K): [in this window] [in a new window]   Figure 2. Ethidium bromide-stained 10% polyacrylamide gel of V3R and glyceraldehyde phosphodehydrogenase duplex RT-PCR, the lower band representing V3R. P1 and P2 are two normal pituitaries; T1–T5 are ACTH-secreting tumors; N is negative control and M is a size marker. [Reproduced with permission from P. L. M. Dahia et al.: J Clin Endocrinol Metab 81:1768–1771, 1996 (160 ). © The Endocrine Society.]

View larger version (70K): [in this window] [in a new window]   Figure 3. Ethidium bromide-stained agarose gel of V2R (top band) and V3R (bottom band) PCR products. Lane 1 is a size marker; lane 2, a 10-fold concentrated normal pituitary; lane 3, a normal pituitary; lane 4, the reported bronchial carcinoid ACTH-secreting tumor; lane 5, a corticotroph tumor; and lane 6 is a PCR blank control [Reproduced with permission from W. Arlt et al.: Clin Endocrinol (Oxf) 47:623–627, 1998 (166 ). © Blackwell.]

2. GR. One of the hallmarks of ACTH-secreting tumors is their resistance to corticosteroid feedback, which is partially present in pituitary tumors but is more pronounced in ectopic ACTH secretors (8, 173, 174). This differential regulation suggests that distinct mechanisms are involved in determining the glucocorticoid resistance in these two groups of tumors.

Corticosteroids act via activation of the ubiquitously expressed type II GR (175, 176, 177). The human GR is known to produce two transcripts as a result of alternative splicing of the gene: GR- and GR-ß (178, 179, 180). These highly homologous subforms, with 777 and 742 amino acids, respectively, differ only at their carboxy terminus. This difference renders GR-ß unable to bind glucocorticoids. Only recently has the role of GR-ß been explored in more detail, and some evidence has suggested that it acts as a negative regulator of the classical GR- by forming heterodimers (181, 182, 183, 184). Derangements of GR- have been shown to be associated with glucocorticoid resistance syndromes (185, 186, 187, 188). Furthermore, the wide diversity of phenotypes attributed to the glucocorticoid resistance state has recently been suggested to be related to the type of mutation found in the GR gene structure: different point mutations in the GR gene can affect distinct pathways of gene regulation in a differential fashion (189). Hence, disruption of the GR gene, either at the structural or transcription level, leading to abnormal function of the receptor could potentially contribute to the pituitary-specific glucocorticoid resistance seen in corticotroph adenomas. In fact, a mutation in the DBD of the GR gene has been found in the tumoral tissue of one patient with Nelson’s syndrome (190), an aggressive form of corticotroph adenoma that had its growth accelerated by removal of the adrenal glands in a patient with a pituitary source of dysregulated ACTH secretion (191). A frameshift mutation expected to create a truncation in the GR product was hypothesized to be associated with the hypercortisolemic state in this case (190). A search for similar defects occurring in the DBD, as well as in the steroid-binding domain of the GR- subform and in the splice junction region of the GR, failed to identify mutations in a series of 19 ACTH-secreting tumors, including 2 cases of Nelson’s syndrome, 3 ectopic secretors, and 1 malignant corticotropinoma (192). This suggests that mutations of the GR gene are not commonly involved in the pathogenesis of Cushing’s disease. However, more recently, a study has suggested that deletions of this gene might play a role in the pathogenesis of a number of corticotropinomas: 6 of 22 tumors examined for LOH in 5 known polymorphisms of the gene were found to have monoallelic deletion of GR (193). Because no mutations were found in the remaining GR allele, and no expression studies have been performed in these cases to determine whether GR haploinsufficency was occurring, the functional significance of the deletion of the GR locus in these tumors still remains to be established.

It has also been shown that in vivo models of glucocorticoid resistance, such as a rare type of glucocorticoid-resistant asthma and also the New World primates, appear to be associated with an increase in the expression of GR-ß subform (187, 188, 194). A potential mechanism for the glucocorticoid resistance at the pituitary level observed in corticotroph tumors could be related, therefore, to an altered ratio of the two isoforms of GR, with an excess of GR-ß being potentially able to counteract the activity of the steroid-binding form, GR-, or alternatively, actively transrepress glucocorticoid-responsive genes. Relative quantitation of the GR- and GR-ß transcription in corticotroph tumors revealed the two isoforms to be expressed at levels similar to those seen in normal pituitary, with a clear predominance of the steroid-binding form over the nonbinding form in all tumors (192). Additionally, while variable levels of expression of the GR-ß subform were observed among the tumors, no correlation with the degree of glucocorticoid resistance, as measured by the response to the high-dose dexamethasone suppression test, was found. Hence, although corticotropinomas show glucocorticoid resistance, this feature cannot be attributed to mutations of critical GR functional domains or abnormal splicing of the GR gene. However, since there has been a report of an apparent lack of correlation between the transcription and posttranslational relative abundance of the GR- and GR-ß forms (184), abnormal expression of these two subforms at the posttranslational level, contributing to the glucocorticoid-resistant phenotype in Cushing’s disease, has not been excluded.

3. CRH-R. CRH, also known as CRF, is the main neuroregulator of the HPA axis and plays an important role in coordinating the endocrine, autonomic, and behavioral responses to stress and immune challenge. There are at least two known subforms of CRH-R, types I and II (195, 196, 197, 198, 199, 200, 201, 202). CRH-R type I is predominantly expressed at the pituitary level (200), while type II is more widely expressed throughout the central nervous system and other tissues such as the heart (195). The gene coding for the CRH receptor type I was cloned from a human corticotropic tumor library (200). The cDNA encodes a 415-amino acid protein comprising seven membrane-spanning domains (type Ia). An alternatively spliced form of the receptor, which includes an insert of 29 amino acids in the first intracellular loop, has also been reported (type Ib) (200). Another variant of the human CRH-R has been characterized, which has a 40-amino acid deletion in the amino-terminal domain of the receptor and is the only form of CRH-R detectable in the individual from which it was isolated (203). CRH binds to this variant receptor with lower affinity as compared with the wild-type receptor, and high concentrations of human CRH are required to elevate intracellular cAMP levels in cells transfected with the variant receptor. While the existence of two alternatively spliced transcripts of the CRH-R type II has been reported in the rat, no such variation has yet been characterized in humans (195, 204). The functional significance of the variant transcripts of the CRH type I receptor gene is not fully understood, and their potential role in the regulation of corticotrophs is yet to be established. However, studies in mice have suggested that posttranslational modifications of CRH-R product might occur without corresponding changes in the CRH-R mRNA transcription rate (205).

Recently, increased expression of the CRH-R, without mutations of the coding sequence of the gene, has been reported in corticotroph tumors (206, 207). It is currently unknown whether the up-regulation of the CRH-R1 transcript in corticotroph tumors corresponds to an increased receptor number at the cell surface. If so, this increased transcriptional level of CRH-R1 might reflect an abnormal response of the tumor cells to the CRH stimulus or might solely denote an attempt to counterregulate abnormal processing of the receptor at a posttranslational level (208). A recent study revealed that CRH administration to cultured corticotropinoma cells increased the mRNA levels of CRH-R type I, whereas down-regulation of CRH-R type I was observed with dexamethasone and vasopressin infusion (209). This may be contrasted to the situation in rat anterior pituitary cells, in which CRH administration results in down-regulation of its receptor mRNA (210). Taken together, these findings might explain, in part, the exaggerated ACTH response to CRH injection observed in patients with Cushing’s disease, but not in cured patients or normal individuals, although such responses must also relate to the relative corticosteroid resistance of such tumors.

An important model for Cushing’s disease study is the CRH transgenic mouse, which is discussed in detail later in this review.

4. Nur77. Nur77, also known as NGFI-B (nerve growth factor I-B), a member of the steroid hormone receptor superfamily with no known ligand, is encoded by a growth factor-inducible immediate early gene (211, 212, 213). Nur77 has been proposed to be a mediator of ACTH’s ability to activate the expression of genes that encode steroidogenic enzymes. Since Nur77 is capable of activating such genes as CYP21 (214), it is possible that ACTH induces the synthesis and activity of Nur77, which, in turn, might activate the transcription of steroidogenic enzyme genes. In addition, in vitro studies have suggested that phosphorylation may play an important role in regulating Nur77 function. However, it is still unclear whether this phosphorylation affects Nur77 activity in vivo and how such phosphorylation is regulated by ACTH.

The recent identification of Nur77 as a mediator of the CRH induction of POMC transcription led to the study of the mechanism of glucocorticoid antagonism of receptor activation (215, 216, 217). It has been demonstrated that positive signals mediated by Nur77 (and also probably by related family members), and negative signals exerted by GR, appear to be mechanisms for the control of transcription of both corticotroph and lymphoid cells (218). In agreement with these findings, it has also been demonstrated recently that AtT20 cells transfected with the human Nur77 homolog showed lower glucocorticoid-induced inhibition of POMC mRNA transcription and ACTH secretion (219). Thus, the Nur77-signaling pathway appears to combine stimulatory signals and glucocorticoid repression in both endocrine and lymphoid systems. The role of Nur77 in the corticotroph tumor phenotype has yet to be explored in detail, but the ability to control ACTH and CRH activity renders it a natural candidate for a role in corticotroph cell regulation. For example, overexpression of Nur77 could lead to apparent corticosteroid resistance and a phenotype similar to Cushing’s disease. However, while a transgenic model of Nur77 overexpression is still not available, a Nur77-deficient mouse was not reported to develop pituitary abnormalities (220).

5. Other receptors. GH can be released by stimulation of GH-releasing peptides and their nonpeptide analogs (GHSs) (221). The receptor mediating such signaling has recently been identified (GHS-R) and found to encode a G protein-coupled product distinct from the GH-releasing hormone receptor (222). GHSs induce GH and ACTH secretion in patients with somatotroph and corticotroph tumors, respectively (223, 224). We have recently analyzed a series of 40 pituitary tumors of several lineages for the expression of GHS-R (225): all 8 GH-secreting tumors and 5 of 18 corticotroph tumors (and one ectopic ACTH secretor) were found to express higher transcription levels of GHS-R in comparison with normal pituitaries (Fig. 4). In contrast, the majority of the corticotroph tumors, as well as pituitary adenomas arising from other cell types and other neuroendocrine tumors, including ectopic ACTH-secretors, showed lower or equivalent GHS-R transcription levels as compared with normal pituitaries. In addition, some tumor samples, including two corticotroph tumors, had no detectable GHS-R transcripts. While further studies are required to evaluate the biological meaning of GHS-R overexpression in these tumor types, it is possible that this finding helps explain the increased responses to GHS seen in patients with GH- and ACTH-secreting tumors (223, 224).

View larger version (9K): [in this window] [in a new window]   Figure 4. Relative expression of the GHS-R gene in tissue from 7 normal pituitaries, 8 somatotroph tumors, 4 prolactinomas, 7 nonfunctioning adenomas, 18 corticotroph adenomas, an FSH-oma, and nonpituitary tumors, including 3 ectopic ACTH-secreting tumors (•), 3 insulinomas (*), 1 gastrinoma (), and 1 nonsecreting thymic carcinoid tumor (). Open symbols represent samples with no detectable expression under any conditions; solid symbols at the zero GHS-R/GAPDH ratio represent samples with very low levels of GHS-R expression. [Reproduced with permission from M. Korbonits et al.: J Clin Endocrinol Metab 83:3624–3630, 1998 (225 ). © The Endocrine Society.]

1. LIF and LIF receptor. LIF is a member of a family of structurally and functionally related cytokines that include interleukin (IL)-6, IL-1, oncostatin M (OSM), and ciliary neurotrophic factor (226). Acting via its own receptor (LIF-R), it induces ACTH secretion and POMC mRNA transcription (5). Although LIF potentiates CRH effects on ACTH, this response occurs via cAMP-independent pathways. Intracellular mediators of LIF include Jak/Trk kinases, p91, and mitogen-activated protein kinases (227). LIF has been found to be expressed in the developing human fetal pituitary and in both normal and tumorous adult pituitary tissue (5). More recent studies performed in murine AtT20 cells have indicated that LIF reduces proliferation, induces ACTH secretion, and potentiates the effects of CRH on ACTH secretion while suppressing the mitotic effects of CRH, suggesting its role as a differentiation factor for the corticotroph (228). Furthermore, there is evidence to suggest that LIF inhibits pituitary angiogenesis (229, 230) which, in addition to its expression in pituitary tumors and its potential antiproliferative effect in corticotrophs, might implicate this factor in pituitary tumorigenesis. In keeping with the presumed functions of LIF from previous in vivo studies, low levels of ACTH were found to be a prominent feature of LIF knockout mice (231). However, transgenic mice with overexpressed LIF, under regulation of the rat GH promoter, show arrested pituitary maturation, invagination of Rathke’s cleft predisposing to cyst formation, marked GH deficiency, and an increased mass of ACTH-secreting cells without a corresponding change in POMC mRNA transcription (232). The expected antiproliferative effects of LIF on corticotrophs anticipated by the studies in AtT20 cells (228) were not seen clearly in the transgenic model. The reasons for this apparent discordance between the antimitotic effects of LIF in corticotroph cells in previous studies and the corticotroph hyperplasia observed in the transgenic animals are not clear. It is possible that the effects of LIF on specific pituitary cell types may vary cyclically according to different phases of embryonic development, a phenomenon that would not be reproduced by this transgenic model. In fact, a recently reported transgenic mouse in which LIF expression was driven by the -GSU promoter was shown to exhibit corticotroph hyperplasia and cushingoid features, as well as reduced gonadotroph, somatotroph, and lactotroph function (233). Also, LIF has been seen to influence the development of ciliated epithelium. Certain differences seen between the phenotypes of the two transgenic mice are possibly attributed to cell-specific expression, time of LIF expression relative to embryonic development, and level of LIF expression in each one of the models. Further studies designed to determine the specific role of LIF in corticotroph tumors are needed to help clarify its contribution to pituitary tumorigenesis.

2. Epidermal growth factor (EGF) and EGF receptor (EGF-R). Some studies have suggested a potential role for EGF in stimulating corticotroph proliferation and ACTH secretion in vitro (234). However, these effects may have been secondary to the ability of EGF to stimulate CRH synthesis (235). A series of pituitary tumors was investigated in relation to the immunohistochemical expression of EGF and EGF-R (236): expression of both products was found in all normal pituitaries tested and in 60% of corticotroph adenomas, whereas it was less frequent in other tumor cell lineages. Another study detected EGF and EGF-R expression in all types of human pituitary adenomas, albeit at variable levels (237). Conversely, lack of both EGF and EGF-R expression in corticotroph adenomas has been noted in another study (238). In the light of these controversial results, it is not clear whether these products play any role in pituitary oncogenesis.

3. Other cytokines and growth factors. Although many cytokines and growth factors mediate cell division, there has been no clear evidence so far for a prominent role of any cytokine in the development of pituitary tumors. More important effects of these substances are indicated by their critical role in regulating specific pituitary hormone regulation (239). The best studied conditions of cytokine hyperactivation in animal models are those in which the effects of transforming growth factor- (TGF-) and nerve growth factor (NGF) have been analyzed. However, overexpression of both genes has been associated with lactotroph cell proliferation (240, 241).

IL-6 is a known potent ACTH secretagogue; however, its effect on cell division is less clear. Opposing results have been observed in studies using different cell models with regard to the ability of IL-6 to induce cell proliferation (242).

FGF-4 has been shown to be expressed only in neoplastic tissue in the adult pituitary. However, the potent mitogenic effects of FGF-4 have been characterized only in PRL-secreting cells (243).

Galanin, a peptide retaining little homology with other neuropeptides, has been detected by immunostaining in normal pituitary and more prominently in corticotroph tumors, whereas it was less reactive in other tumor types (244). In vivo studies, on the other hand, have implicated galanin predominantly in GH and PRL regulation (245). Galanin appears to mediate estrogen-stimulated lactotroph hyperplasia, which may implicate this peptide in prolactinoma development (245), but less is known about its role in corticotroph regulation.

TGF-ß, unlike TGF-, has been suggested to act as an inhibitor of tumor cell proliferation, due to its effects on p21, p27, and p15 induction (246). Interestingly, human colon cancer cell lines with high rates of microsatellite instability were found to harbor mutations in the type II TGF-ß receptor II gene (TGF-ßR-II) (247). By inducing the escape of cells from TGF-ß-mediated growth control, TGF-ß-R-II mutations link DNA repair defects with a specific pathway of tumor progression. Such an association has not yet been explored in pituitary tumors.

These studies indicate that, with the possible exception of LIF, which has yet to be further characterized in terms of its regulation of corticotroph growth and differentiation, no other cytokine or growth factor has been demonstrated to play a major part in corticotroph tumorigenesis. It is possible that these substances may exert a regulatory influence on an already altered pituitary cell or preestablished clonal structure, contributing to tumor development, rather than being the primary initiator(s) of tumoral formation.

C. Developmental genes
The essential roles of homeodomain proteins in cell fate determination during development have been demonstrated in organisms as divergent as Drosophila and higher mammals (248). The genes involved in the differentiation of the embryonic pituitary are important in determining the degree of specialization of each cell lineage, but little is known about the ontogeny of the human corticotroph. Immunoreactive ACTH is already detected in Rathke’s pouch at the fifth gestational week, while a more mature hypophyseal-portal system only occurs between weeks 8 and 14 (1). Because the fetal corticotroph is sensitive to CRH only after week 14, other factors are believed to be involved in the specificity of the corticotroph in earlier weeks (249, 250, 251). Commitment to developing a distinct pituitary gland occurs when the rudimentary Rathke’s pouch, which represents the primordium of the anterior and intermediate lobes, comes into contact with ventral hypothalamus, of neuroectodermal origin. From this point on, the Rathke’s epithelium starts to differentiate into pituitary-specific cell lineages (1, 2). A knockout model has been developed in mice where the homeobox gene Lhx3 (also known as mLim-3 and P-Lim) was targeted (3). Homozygous null mice formed Rathke’s pouch, but no anterior or intermediate pituitary lobes developed. The growth arrest observed in the Rathke’s pouch was shown to be caused by a failure in proliferation rather than increased apoptosis. All cell lineages were affected, except for the corticotrophs, indicating that these cells are regulated by a factor or factors that are distinct from the remaining pituitary cells very early in the development, possibly before the activation of -glycoprotein subunit primitive cells (3). Although POMC expression was detected in some corticotroph cells, these failed to proliferate, indicating a potential role for Lhx3 product in the growth of both differentiated and undifferentiated pituitary cells (3). Alternatively, it could indicate its role in providing pituitary cells with trophic factors from adjacent structures such as the hypothalamus or mesenchyme. Recent studies in AtT20 cells have demonstrated that Lhx3 may play a key role in inducing PRL gene expression in lactotrophs independently of the Pit-1/GHF-1 pathway (252). The Lhx3 knockout is therefore an invaluable tool for studying early development of the pituitary and might also provide clues on the role of proliferation-inducing factors in the progress and development of pituitary tumors. However, as the human homologue of Lhx3 has not as yet been characterized, the specific role of this gene in humans remains speculative.

1. PTX gene family. Ptx1 (also known as Otx-1) is a member of the small bicoid family of homeobox-containing genes (4, 253). It was isolated as a tissue-restricted transcription factor of the POMC gene. The pattern of Ptx1 expression supports the hypothesis that Ptx1 defines a primordial structure that may be involved in the commitment to the corticotroph phenotype (4, 254). Two highly homologous subforms of Ptx1, Ptx1a and Ptx1b, have now been identified, and they both appear to specifically regulate genes expressed exclusively at the pituitary level, including POMC (255). Interestingly, Ptx1a appears to be critical to the expression of the -subunit of glycoprotein hormones (256).

Another homeobox gene involved in the pathogenesis of the Rieger syndrome, RIEG1, and its mouse homologue, Rieg1 (also known as Ptx2), have been recently identified by positional cloning and found to be highly homologous to the Ptx1 gene product (257). Ptx2 gene produces two alternatively spliced mRNA transcripts, which encode proteins of 271 (Ptx2a) and 317 amino acids (Ptx2b), respectively. Ptx2 is expressed in both developing and adult pituitary gland, eye, and brain tissues, suggesting an important role in development and maintenance of anterior structures. Ptx2 was mapped close to EGF on mouse chromosome 3, in a region having extensive syntenic homology with human chromosome 4q. Mutations of the human RIEG1 gene have been identified in the Rieger syndrome, an autosomal-dominant disorder with variable craniofacial, dental, eye, and pituitary anomalies (258, 259).

Another new murine gene, Ptx3, was isolated that encodes a homeoprotein with strong homology to the other Ptx proteins, which may suggest that this is a third member of the family (260). However, the embryological expression pattern of Ptx3 was much more restricted than the remaining isoforms: only the developing lens appears to express the Ptx3 transcript (260).

The potential role of the Ptx genes in mouse pituitary tumorigenesis has not yet been explored, but their important role in embryological pituitary development (in particular, Ptx1 and Ptx2) makes these genes natural candidates for an involvement in pituitary growth abnormalities. The full characterization of the human homologs of the Ptx family of genes will shed some light into their actions involving pituitary regulation. A potential role for homeobox genes in human malignancies has already been suggested (261, 262), as well as their interaction with cytokines such as TGF (259). Studies investigating the potential role of Ptx genes in pituitary tumor formation might provide clues as to the involvement of homeobox genes in tumor pathogenesis.

2. POU factors. Specific families of transcription factors mediate events in the sequential maturation of distinct cell phenotypes (263, 264). POU is a family of transcription factors that specifically recognizes two classes of cis-acting regulatory elements that bear little sequence similarity, the octamer motif ATGCAAAT and the TAATGARAT motif (265, 266). The most well known of these factors is possibly the Pit1 product, which has been shown to be related to the embryological development of the somatotroph, mammotroph, and thyrotroph lineages (1). Members of the class IV POU domain transcription factor Brn-3a (also known as Brn3.0) and two highly related factors, Brn-3b and Brn-3c, are differentially expressed in the developing and mature mammalian nervous system (266, 267). While Brn-3a and Brn-3c products activate their target genes, Brn-3b has repressor activities that can occasionally interfere with activation by Brn-3a or c (268). Analysis of mice null for the Brn-3a locus shows that Brn-3a is required for the survival of subpopulations of proprioceptive, mechanoceptive, and nociceptive sensory neurons (269). Targeted deletion of Brn-3a also alters differentiation, migration, or survival of specific neuronal populations, resulting in defective suckling and in uncoordinated limb and trunk movements, ultimately leading to early postnatal death. No abnormalities have been reported at the pituitary level. Thus, Brn-3a exerts its major developmental effects in somatosensory neurons and in brainstem nuclei involved in motor control (263, 270, 271).

Brn-3a has been found to stimulate POMC transcription and also to exert mitogenic effects on targeted cells when cotransfected with H-RAS (268). Furthermore, Brn3a has been suggested to act as a potential oncogene in neuroectodermal tumors (268). A recent study examining the levels of Brn3a in a series of pituitary and ectopic tumors secreting ACTH reported a high expression of this factor to be associated with more aggressive forms of ACTH-secreting tumors, both of pituitary and nonpituitary origin, but no correlation was seen with POMC expression in such tumors (272). It appears that while Brn3a induces POMC transcription, its potential association with corticotroph tumor aggressiveness is not dependent on this activation. It is believed that activation of the POMC promoter requires that various factors interact synergistically (273). It appears from the study on ACTH-secreting tumors that Brn3a is not necessary for POMC transcription, as the majority of pituitary corticotroph tumors had levels of Brn3a similar to GH- and PRL-secreting tumors. However, a clearer inverse correlation was seen between the degree of tumor differentiation and the levels of Brn3a among the ectopic ACTH secretors (272). Analysis of Brn3a in other tumor types will determine whether its correlation with more aggressive tumor forms is also observed and will shed light on the mechanisms accounting for such an association.

	V. Miscellaneous

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
A. Other genes
Other genes such as the retinoid X receptor, involved in the regulation of growth and differentiation, the apoptosis inhibitors such as the BCL2 gene, and others have been examined for their role in pituitary tumorigenesis (274, 275). While some abnormalities have been found in a number of tumors, it is not clear whether the defects observed are specific to the pituitary tumorous phenotype or whether they represent a general phenomenon related to neoplastic transformation. If specific, it remains unclear whether the finding represents the original cause, or whether it is simply an effect of another, still unknown, primary abnormality.

B. Methylation
It has been generally accepted that an important mechanism of gene inactivation in mammals is methylation (276, 277, 278). In general, it is found that the promoters of expressed genes are unmethylated, whereas those of genes that are not expressed are methylated (279). Unmethylated CpG islands located in gene promoter regions are a principal target of hypermethylation. This hypermethylation correlates with transcriptional repression in a similar manner to the effect of inactivating mutations of tumor suppressor genes (277, 278, 280, 281, 282, 283). A tumorigenesis model attempting to reconcile the role of the nonspecific phenomenon of methylation and more gene- and/or tissue-specific alterations has been proposed (280). According to this model, tumor progression results from an initial clonal expansion of heterogeneous cell populations directed by continuous interaction between methylation abnormalities and classic genetic events.

Pituitary tumor cells have been found to be targets of abnormal methylation: as noted above, the p16 gene has been found to be transcriptionally silenced due to hypermethylation of its promoter in the majority of pituitary adenomas examined (75). Recently, it has been found that the specific "pituitary" POMC promoter (the promoter that drives the transcription of the translated POMC peptide, as opposed to an alternative promoter that leads to the transcription of a nonsecreted peptide) is unmethylated in all POMC-expressing tissues and cells and is methylated in nonexpressing tissues (284). In addition, the activity of the promoter was lost by in vitro methylation in the POMC-secreting small-cell lung cancer cell line DMS-79. The alternative promoter was found to be methylated in all tissues. In ectopic ACTH-secreting tumors, the "pituitary" promoter was found to be specifically unmethylated, while the alternative promoter was methylated, suggesting that this regulation is not a random process and might be related to tumorigenesis in these cells. It is possible that the expression of the POMC promoter confers some growth or survival advantage to certain cell types, such as DMS-79 cells. Forthcoming studies might reveal the role of POMC regulation in pituitary tumors and shed some light into the mechanisms regulating methylation in a tissue-specific manner.

	VI. Perspectives: Old and New Tools for Understanding Pituitary Pathogenesis

Top Abstract I. Introduction II. Protooncogenes III. Tumor Suppressor Genes IV. Specific Genes V. Miscellaneous VI. Perspectives: Old and... VII. Conclusions References

 
A. Transgenic/knockout mice models
The epidemiology of cancer in humans and in animals suggests that multiple genetic events are responsible for the genesis of malignant tumors. During the development of many tumors, distinctive changes can be recognized: normal unaffected tissue, hyperplasia with a high incidence of proliferating cells, induction of tumor angiogenesis with the new growth of capillaries, solid tumor formation (neoplasia), and finally, metastasis. The molecular analysis of tumorigenesis is often restricted by the lack of availability of tissue specimens from the multiple stages. For this reason, the genetic reproducibility and the accessibility of tissue specimens have made transgenic mice an important tool in the study of the molecular events that are involved in the multistep progression to the tumor phenotype.

Many transgenic mouse models of human cancer exhibit similar patterns of stepwise tumor development (285). However, there are many other models in which an obvious parallel with the corresponding human disease system is not observed. We have referred to some of these examples earlier when the RB1 and p27^KIP1 null mice were mentioned (48, 59, 77, 78, 79, 286). The reasons for the lack of correspondence between the murine and human models is not entirely understood, but differences in the regulatory mechanisms involved in certain redundant systems, as well as the involvement of modulating factors contributing to the final phenotype, might play a role in this distinction. In addition, variations of the genetic background of the mice used to create specific disease models might contribute to differences in the phenotype resulting from such systems. It is possible that specific strains of animals produce distinct phenotypes when the same genomic manipulation is performed. Nevertheless, despite the obvious limitations and drawbacks represented by these models of study, as stated above, transgenic/knockout models are still considered invaluable tools for the understanding of in vivo effects of specific gene alterations. Important advances into the mechanisms involved in HPA axis regulation have been provided by animal models of excess or null CRH expression.

A transgenic mouse model overexpressing the CRH gene driven by a metallothionein promoter showed many features characteristic of Cushing’s disease, which were mostly reversed by adrenalectomy (287). Compared with endogenous CRH tissue distribution, the transgenic animals showed a broader CRH expression in the brain, possibly due to a less efficient glucocorticoid-mediated CRH down-regulation in these tissues. Transgenic mice were also noted to exhibit an anxiogenic-like state, similar to the behavior assumed by animals under chronic stress (288). Furthermore, an impaired immune response, related to both quantitative and qualitative B cell activity, was observed in the transgenic mice. Both endocrine and behavioral features of these animals were counteracted by injection of the CRH antagonist, -helical CRH(9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41), indicating that the effects observed were specifically caused by the excess of CRH. However, despite the fact that high levels of CRH produced a hypercortisolemic state with features suggestive of Cushing’s disease, no tumors were seen in the mice pituitary corticotrophs. It remains to be established whether chronic corticotroph hyperplasia in humans facilitate the development of a functioning corticotroph adenoma.

Targeted disruption of CRH, CRH-R1, and GR genes in mice have also provided important insights into the mechanisms involved in HPA axis regulation at all stages of animal development. The gene-deficiency model helps delineate not only the specific role of a certain product from the phenotype of its absence, but also provides clues as to the compensatory mechanisms that are activated or arise when it is not present. These compensatory systems, in turn, may shed light on novel pathways that might play a critical role in the maintenance of homeostasis and might therefore contribute to the understanding of tumorigene