A common strategy to generate mouse models of cancer is to express viral oncoproteins in the cell/tissue of interest. These animal models based on viral oncogene overexpression provide limited insight into the mechanisms underlying pituitary-tumor formation. However, they can be considered useful tools to study various aspects of pituitary adenomas such as hormone secretion regulation and pathophysiology of hormone excess.

One of the first mouse models of pituitary tumors was generated by expressing the (PyLT) under the control of the polyoma early region promotor. These transgenic mice developed ACTH-producing microadenomas at 9 months of age, large adenomas at 13–16 months of age, and display features of Cushing’s syndrome (10). Several mouse strains generated using oncogenic viruses were generated specifically targeting the pituitary. Thus, expression of the simian virus 40 (SV40) large T antigen under the control of the proopiomelanocortin (POMC) gene promoter induced the formation of pituitary tumors arising from the intermediate lobe. These tumor cells expressed POMC peptides, but not other pituitary hormones (11). SV40 large T antigen overexpression using another promoter, the arginine vasopressin (AVP) gene promoter, also resulted in pituitary tumors although these tumors were composed of undifferentiated somatotroph cells (12). Transgenic mice carrying a temperature-sensitive mutant of simian virus 40 T antigen driven by regulatory elements of the beta subunit of human follicle-stimulating hormone (FSHβ) gene develop pituitary-tumors reminiscent of human null cell adenomas (13).

As mentioned in the previous section, various hypothalamic hormones and growth factors required for normal pituitary function and formation are overexpressed in pituitary adenomas. Transgenic approaches allow the overexpression of these hormones and factors specifically in the pituitary to evaluate their role in pituitary tumorigenesis. Overexpression of the GH-releasing hormone (GHRH) gene in mice results in hyperplasia of pituitary somatotrophs, lactotrophs, and mammosomatotrophs (cells producing both GH and PRL) by 8 months of age that leads to mixed GH and prolactin secreting pituitary adenomas by 10–24 months of age (14). These results indicate that GHRH might promote proliferation of transformed pituitary cells. However, the extended tumor latency period suggests the requirement for additional molecular alterations in the formation of pituitary tumors in these mice. A transgenic mouse model was generated in which TGFα was specifically overexpressed in the pituitary lactotrophs using the prolactin promoter (15). These transgenic mice developed lactotroph hyperplasia at 6 months of age and prolactin-producing pituitary adenomas with high penetrance by 12 months of age. Pituitary lactotroph-specific overexpression of the N-terminally truncated isoform of FGFR4 (ptd-FGFR4), but not the wild-type FGFR4, in mice leads to the development of tumors reminiscent of human prolactinomas (16). Interestingly, ptd-FGFR4 overexpression causes pituitary adenoma formation without preceding hyperplasia. The mechanism by which ptd-FGFR4 might facilitate pituitary cell transformation appears to involve alterations in neural cell-adhesion molecule/N-cadherin signaling that lead to diminished cell-adhesion (82). The oncogenic role of PTTG1 and HMGA2, two oncogenes previously suggested to play a relevant role in human pituitary adenoma formation, has been analyzed in transgenic mice. Targeted PTTG1 overexpression to gonadotroph cells of the pituitary driven by the alpha-subunit glycoprotein (alphaGSU) promoter results in gonadotroph hyperplasia (17). However, only occasional microadenomas were observed. These observations seem to question the tumorigenic potential of PTTG1 in pituitary. However, when overexpressing-PTTG1 mice were crossed with Rb heterozygous mice, which develop pituitary tumors with high penetrance (see below), increased frequency of pituitary adenomas was observed. On the contrary, PTTG1 deletion diminishes pituitary-tumor development in Rb heterozygous mice (83). Taken together, these results obtained from genetically modified mice indicate that PTTG1 facilitates pituitary adenoma formation. Transgenic mice expressing high levels of the HMGA2 transgene in all tissues (by using the cytomegalovirus promoter) developed mixed growth hormone/prolactin cell pituitary adenomas with high penetrance (85% of female animals) by 6 months of age (19). Of note, transgenic males developed pituitary adenomas with a lower penetrance (40%) and a longer latency period (about 18 months). Interestingly, HMGA1 overexpression also leads to the development of pituitary adenomas secreting prolactin and GH (18). Subsequent studies performed in these transgenic mice have shown that HMGA proteins mediate the formation of these tumors through a mechanism that involves the Rb/E2F1 pathway, thus further supporting a critical role of this pathway in pituitary tumorigenesis (84).

Altogether, transgenic mice overexpressing proteins in the pituitary have provided important insights into pituitary tumorigenesis. One of the major limitations of these mouse models is the regulation of the expression of the transgene that ideally should be limited to adult stages, thus mimicking the human situation.

The use of conventional or whole-body knockout mouse models provides a different approach for analyzing the function of important genes in pituitary tumorigenesis. Thus, conventional knockout mice are a crucial tool to analyze the role of tumor suppressor genes in pituitary tumorigenesis. Conventional knockout mice have been generated to assess the direct effect of several genes linked to familial pituitary tumors such as MEN1, CDKN1B, and AIP in pituitary-tumor development. Four different Men1 knockout mouse strains have been generated (20–23). Homozygous Men1 mice are embryonic lethal due to defects in craniofacial development but heterozygous Men1 mice are viable and have been extensively analyzed for tumor formation. In general, all these Men1 heterozygous mouse models correlate closely with the human MEN1 phenotype regarding tumor incidence and spectrum. These mice develop tumors of various endocrine tissues, including pancreatic islet cell tumors, pituitary and parathyroid adenoma, and adrenocortical tumors. Loss of heterozygosity and menin expression was demonstrated in the tumors, consistent with a tumor suppressor role for the Men1 gene. Furthermore, a higher occurrence of anterior pituitary tumors is observed in female Men1 heterozygous mice, similar to what it is observed in MEN1 patients. However, the different Men1 mouse models also show slight phenotypic differences including the type of pituitary adenomas developed. Thus, only prolactin producing adenomas were observed in the Men1 mouse model generated by Crabtree et al. (20) while prolactinomas and non-secreting adenomas were found in the model generated by Loffler et al. (21). Bertolino et al. (22) reported the formation of prolactinomas and GH-secreting tumors. Finally, Harding et al. (23) reported that the majority the pituitary tumors originated from the pars distalis and contained both prolactin and GH while around 10% of the tumors originated from the pars intermedia and contained ACTH. The reason for these differences in pituitary-tumor formation is unclear but it might be related to the types of Men1 mutations (different exons were targeted) and/or differences in genetic background.

Recently, a conventional knockout mouse deficient in the AIP, a gene linked to familial isolated pituitary adenoma (FIPA), and isolated familial somatotropinomas have been generated. Homozygous knockout Aip mice died during embryogenesis, likely due to cardiovascular abnormalities (24). Heterozygous Aip mice were highly prone to pituitary adenomas, developing tumors as early as at 6 months of age (24). Thus, unlike in humans in whom AIP-associated tumors often develop in childhood or early adulthood (85), heterozygous Aip mice develop adenomas in adulthood. Another major difference between the Aip mouse model and the human disease is the complete penetrance of pituitary tumors in the Aip mouse model (at 15 months of age). The majority of adenomas found in the Aip mouse model were somatotropinomas, although mixed GH/prolactin, prolactinomas, and ACTH-corticotropinomas were also found. Biallelic inactivation of AIP was demonstrated in these tumors.

Analysis of conventional knockout mice has also been useful to analyze the effect of disruption of feedback signaling in pituitary-tumor pathogenesis. Thus, analysis of dopamine receptor 2 (Drd2) knockout mice indicates that functional inactivation of this receptor might contribute to the formation of pituitary adenomas, namely prolactinomas. Female Drd2 knockout mice develop massive lactotroph hyperplasia by 12 months of age and invasive prolactinomas at older ages. (25, 26). Male Drd2 knockout mice develop microadenomas in the absence of hyperplasia. The effects of disrupted prolactin feedback on pituitary tumorigenesis have been demonstrated in knockout mice deficient for prolactin and prolactin receptor (Prlr). Prolactin knockout mice develop lactotroph hyperplasia by 6 weeks of age and prolactinomas by 8 months of age (27). Inactivation of Prlr also results in lactotroph hyperplasia and formation of invasive adenomas (28).

Studies performed in mice with targeted disruption of cell-cycle regulators collectively support the notion that cell-cycle dysregulation is central in pituitary tumorigenesis (51). Mice heterozygous for Rb develop ACTH-producing aggressive tumors arising from the intermediate lobe of the pituitary with high penetrance by 12 months of age (29). Subsequent studies have shown that pituitary-tumor incidence and histological type in Rb heterozygous mice is dependent on the mouse strain suggesting an effect of modifier genes in pituitary adenoma formation (86). Homozygous mice deficient for the cyclin-dependent kinase inhibitors p27^Kip1 and p18^Ink4c also develop pituitary adenomas. Homozygous p27^Kip1mice develop tumors from the intermediate lobe by 12 months of age (30, 31) while p18^Ink4c mice develop pituitary adenomas in both the intermediate and anterior lobes (32). Inactivation of other cyclin-dependent kinase inhibitors that are downregulated in human pituitary adenomas, such as p21^Cip1, p15^Ink4b, and p16^Ink4a, does not result in pituitary tumors in mice. The generation of mice with combined inactivation of different cyclin-dependent kinase inhibitors has revealed a cooperation of these cyclin-dependent kinase inhibitors in pituitary-tumor formation. Thus, deletion of p21^Cip1, p27^Kip1, or p16^Ink4a in combination with loss of p18^Ink4c results in increased incidence and decreased latency of pituitary tumors compared to single mutants (32, 87, 88). Of note, double p18^Ink4c p27^Kip1 mutant mice appear to die by 3 months of age due to pituitary adenomas (32). The relevance of INK4 family proteins in pituitary adenoma formation is further supported by the analysis of knock-in mice, which express a CDK4 variant with a mutation that renders the Cdk4 protein insensitive to INK4 inhibitors (Cdk4^R24C/R24C mutant mice) (33, 89). These Cdk4^R24C/R24C knock-in mice develop multiple tumors including pituitary tumors (25% of penetrance) arising in the anterior pituitary by 15 months of age. A dramatic cooperation in pituitary-tumor formation is observed in double Cdk4^R24C/R24C p27^Kip1 mutant mice, which develop poorly differentiated pituitary tumors that kill these mutant mice in 8–10 weeks of age (34). Cooperation between cyclin-dependent kinase inhibitors and Rb in pituitary-tumor formation has also been observed. Inactivation of p27^Kip1 and p21^Cip1in Rb heterozygous mice results in decreased latency of pituitary tumors (90, 91). Similarly, loss of p16^Ink4a strongly accelerates intermediate lobe pituitary tumorigenesis in Rb heterozygous mice (92).

The usefulness of conventional knockout mice as models for evaluating therapeutic agents is highlighted by studies performed with bromocriptine, a dopamine agonist used to treat prolactinomas. Thus, it has been reported that bromocriptine decreases prolactin serum levels in Prlr mutant mice harboring tumors (28) and reduces tumor cell proliferation in p27^Kip1 deficient mice (93). Bromocriptine does not affect prolactin levels in Drd2 knockout mice, as expected from the absence of the dopamine receptor 2 (28) and thus, these mice can be considered a useful model for dopamine-resistant prolactinomas. The effectiveness of antiangiogenic treatments in dopamine-resistant prolactinomas has been analyzed in Drd2 knockout mice. Anti-VEGF strategies resulted in prolactin inhibition and decreased tumor cell proliferation in these mice (94).

The mouse models discussed above illustrate how conventional knockout mice help to elucidate the causal role of genes that were predicted to be important for human pituitary tumorigenesis. These conventional knockout mice can also be useful as preclinical models for testing of therapeutic agents.

The development of conditional gene targeting has allowed to overcome some of the limitations of conventional knockout mice such as potential embryonic lethality or complications due to systemic effects of gene inactivation. The most widely used conditional system is based on the Cre-loxP technology. Combinations of Cre-loxP systems, specific promoters, and inducible systems (i.e., tamoxifen) allow tissue- and time-specific inactivation of the gene of interest. The number of pituitary-specific Cre lines available is rapidly increasing (35, 95–102). The generation of pituitary cell-specific Cre transgenic lines has been instrumental in advancing our understanding of pituitary cell function and cell lineages. However, to date, relatively few studies have exploited these Cre lines to evaluate the role of genes of interest in pituitary tumorigenesis.

To explore the role of inactivation of the Carney complex gene PRKAR1A on pituitary tumorigenesis, a pituitary-specific knockout of this gene was generated by crossing mice with conditional alleles of Prkar1a with a mouse line expressing the Cre recombinase in pituitary cells using the GHRH receptor promoter (rGhrhr-Cre line) (35). This promoter, drives Cre-mediated recombination in pituitary cells of the Pit1 lineage, including somatotrophs, lactotrophs, and thyrotrophs. Pituitary-specific Prkar1a knockout mice exhibited a moderate increased frequency of pituitary adenomas, although not until advanced age (18 months). These tumors were composed of cells that expressed GH, prolactin, and TSH. Using a different conditional system based on the Flp recombinase, specific inactivation of Rb in pro-opiomelanocortin-expressing cells has been achieved. Mice in which only an allele of Rb was conditionally inactivated in the pituitary displayed atypical lesions in the intermediate lobe of the pituitary by 6–12 weeks of age that resulted in adenoma formation as the mice aged. Accelerated adenoma formation was observed in mice homozygous for the conditional alleles of Rb. These mice developed pituitary adenomas with 100% incidence as early as 12 weeks of age and succumbed to these adenomas by 20 weeks of age (36). The adenomas originated from the intermediate lobe melanotrophs and were indistinguishable from the adenomas observed in conventional Rb heterozygous mice. Pituitary-specific inactivation of Rb has also been achieved using a mouse strain in which Cre expression is under the control of the POMC gene promoter. These mice developed pituitary adenomas with 100% incidence after a relatively short median latency (around 6 months) (37). However, no histological characterization of the adenomas was reported. Interestingly, in this study pituitary adenoma formation in vivo was monitored by crossing conditional Rb knockout mice with transgenic mice that express luciferase under the control of the POMC promoter detecting tumor growth as early as 8 weeks of age.

Pituitary adenomas have been also reported in conditional mouse models using non-pituitary-specific Cre lines. Thus, conditional inactivation of the MEN1 gene using a mouse line in which Cre expression was driven by the insulin promoter resulted in the formation of pituitary adenomas (in addition to islet tumors) by 12 months of age (38). All pituitary adenomas were derived from the pars distalis, as determined by expression of prolactin. The formation of pituitary adenomas in these mice was attributed to aberrant Cre expression in the pituitary. Conditional activation of the polycomb ring finger oncogene using a GFAP-Cre that drives expression in neural and glial lineages developed intermediate and anterior lobe pituitary tumors positive for ACTH and beta-endorphin although with incomplete penetrance and long latency period (1 year) (39). Consistent with these results, Cre expression in the GFAP-Cre line was observed in the pituitary. However, it is unclear whether this Cre activity in the pituitary represents endogenous GFAP expression or ectopic expression of the Cre transgene.

The usefulness of conditional knockout mouse models in the study of pituitary-tumor formation is perhaps best illustrated by two recent studies that analyzed the role of Wnt/β-catenin signaling pathway in pituitary tumorigenesis. Expression of a degradation-resistant mutant form of β-catenin in embryonic pituitary using the Hesx1-Cre line was shown to cause the formation of tumors resembling human adamantinomatous craniopharyngioma (ACP) (40). ACP is a type of pediatric non-hormone-producing tumor derived from pituitary embryonic tissue associated with significant morbidity and mortality (8). Interestingly, expression of mutant β-catenin in differentiated cells in the embryonic or adult pituitary using specific Cre lines (Gh-Cre, Prl-Cre, and Pit1-Cre) did not give rise to tumors (40). Further studies from the same group using specific a tamoxifen-inducible Cre line demonstrated that activation of mutant β-catenin in Sox2-expressing adult pituitary stem cells also result in the formation of ACP-like tumors (41). The molecular characterization of the tumors formed in these mouse models of ACP have uncovered a number of genes and pathways potentially involved in the formation of ACP including the SHH, BMP, and FGF pathways (103). These findings have been validated in human ACP (103) in what is an excellent example of how genetic studies in mice harboring conditional mutations, in addition to providing mechanistic insight into pituitary tumorigenesis, can lead to the generation of new animal models for modeling human disease.