Ординатура / Офтальмология / Английские материалы / Ocular Oncology_Albert, Polans_2003

ultrastructurally, these tumors display many features of human retinoblastoma, including the presence of Homer-Wright rosettes (Fig. 5), lamelliform nuclear membranes, neurosecretory granules, cytoplasmic microtubules, and cilia with a ‘‘9 þ 0’’ pattern of microtubules, (which are observed both in normal human photoreceptors and differentiated human retinoblastoma) [15]. Immunohistochemically, the tumor cells have been shown to express neuron-specific enolase (NSE) and synaptophysin but not vimentin, glial fibrillary acidic protein (GFAP), or S-100, consistent with a neuronal cell type of origin [20]. However, they do not express opsin or other photoreceptor-specific proteins, as do many human retinoblastomas [21–23]. Based on their location in the inner nuclear layer and their ultrastructural and immunohistochemical characteristics, the tumors arising in LHbeta-Tag mice are thought to be of amacrine cell origin, while human retinoblastomas are generally thought to arise from photoreceptor precursors or retinoblasts having the potential to differentiate into photoreceptors [21,23–26].

Interestingly, approximately one-quarter of the LHbeta-Tag transgenic mice also develop midline intracranial tumors arising in the subependymal midbrain adjacent to the cerebral aqueduct [27]. These are generally poorly differentiated tumors that most closely resemble primitive neuroectodermal tumors (PNET) (Fig. 6). The development of primary midline intracranial tumors is also seen in a small subset of retinoblastoma patients and is referred to as trilateral retinoblastoma [28,29]. In most cases, these tumors appear to arise from the pineal gland, and they display many of the differentiated features of retinoblastoma [28,30,31]. However, a subset of the intracranial tumors in patients with trilateral retinoblastoma are undifferentiated suprasellar-parasellar tumors [32] and more closely resemble the tumors seen in the LHbeta-Tag mice.

Figure 5 High-power view of LHbeta-Tag retinal tumor. Arrows indicate Homer-Wright rosettes. (H&E, x700.)

Figure 6 Midbrain primitive neuroectodermal tumor in LHbeta-Tag mouse. Neoplastic cells are adjacent to and appear to replace a portion of the ventral ependymal lining of the cerebral aqueduct. The tumor has not invaded the aqueductal lumen. (H&E, x315.)

B.Opsin-Tag Transgenic Mice

Although the retinal tumors arising in LHbeta-Tag mice share many histological features with human retinoblastoma, they differ from the majority of retinoblastomas in their lack of photoreceptor differentiatio7n. Thus, several attempts have been made to generate transgenic mouse models of retinoblastoma that more closely

resemble the human tumor by directing T-antigen expression specifically to photoreceptors or photoreceptor precursor cells. In the first attempt, T-antigen expression was placed under the control of an opsin gene promoter, which normally initiates expression as photoreceptor cells terminally differentiate and become postmitotic [33]. However, rather than developing retinal tumors, opsin-Tag transgenic mice undergo rapid photoreceptor degeneration during the early postnatal period [34]. The retinal cell death is accompanied by continued cell proliferation as late as postnatal day 16 [35], a stage when normal photoreceptors are entirely postmitotic. These results indicate that T antigen–induced proliferation in postmitotic cells triggers an apoptotic response, and that in order for T antigen to promote tumorigenesis, its expression must initiate prior to terminal differentiation, while cells are still mitotically active.

C.IRBP-Tag Transgenic Mice

In subsequent efforts to model retinoblastoma in transgenic mice, T antigen was placed under the control of the interphotoreceptor retinoid-binding protein (IRBP) gene promoter [36,37]. In contrast to opsin, IRBP is expressed during early retinal development (as early as day E12), while retinal precursors are still mitotically active [33]. IRBP-Tag transgenic mice develop intraocular tumors with 100% penetrance that are histologically detectable shortly after birth. Importantly, the tumors arising in these mice are not focal but involve the entire developing photoreceptor cell layer (Fig. 7A). Thus, a laminar retina with normal differentiated photoreceptors never forms. By 6–8 weeks of age, the tumors replace the bipolar cell layer (Fig. 7B) and subsequently invade the optic nerve, ciliary body, iris and vitreous. Tumors in these mice resemble undifferentiated retinoblastomas and generally do not form rosettes (Fig. 7C). They express several neuronal markers including NSE and synaptophysin and also express IRBP but not opsin, suggesting that they are derived from a photoreceptor cell precursor that has not yet activated opsin expression [37].

All IRBP-Tag mice also develop early midline intracranial tumors, which in this model are derived from the pineal gland. By 2 weeks of age, these tumors fill the transverse fissure of the cerebrum (Fig. 8A); by 9–12 weeks, they extend diffusely throughout large portions of the brain. In contrast to the retinal tumors arising in these mice, the pineal tumors resemble highly differentiated retinoblastoma, consisting almost entirely of Homer-Wright rosettes (Fig. 8B). The development of pineal tumors in these mice likely reflects the fact that the IRBP gene is expressed in the pineal as well as the retina; thus T-antigen expression is directed to both tissues in IRBP-Tag transgenic mice.

D.IRBP-E7 and IRBP-E7/p53 / Mice

The fact that retinal tumors in the IRBP-Tag transgenic mice are nonfocal and involve the entire photoreceptor cell layer suggests that expression of T antigen is sufficient to cause oncogenic transformation of photoreceptor precursor cells. In addition to inactivating pRb, T antigen associates with a number of other cellular proteins, most notably the p53 tumor suppressor protein [17,18]. However, the contribution of p53 inactivation to the development of human retinoblastoma is uncertain, since it is not found to be mutated in retinoblastoma tumors (see Chap. 10

inactivates pRb [39,40] but not p53, while the E6 protein binds and promotes the degradation of p53 [41] but does not interact with pRb. Thus, if pRb inactivation were sufficient to promote retinal tumorigenesis, transgenic mice expressing HPV E7 in the retina would be expected to develop retinoblastomas much like the IRBP-Tag mice. However, rather than developing tumors, IRBP-E7 mice undergo rapid and dramatic retinal degeneration resulting from apoptosis of photoreceptor precursors in the first few days after birth, when photoreceptors would normally be undergoing terminal differentiation [38]. This finding suggested that inactivation of pRb promotes a p53-dependent form of apoptosis rather than transformation and that simultaneous inactivation of both pRb and p53 may be required for oncogenic transformation of photoreceptor precursors. To test this hypothesis, IRBP-E7 transgenic mice were bred to p53 knockout mice, in which both alleles of the p53 gene have been inactivated by homologous recombination [42]. As the hypothesis would predict, IRBP-E7/p53 / mice were found to develop retinal tumors similar to those seen in the IRBP-Tag transgenic mice [38]. However, when eyes from IRBPE7/p53 / mice were examined histologically in the early postnatal period, before the development of large ocular tumors, it was found that retinal apoptosis was still occurring at a rate only slightly delayed from that seen in IRBP-E7 mice, but that occasional proliferating lesions could be observed arising from the degenerating retina. Thus, the retinal apoptosis induced by E7 expression is largely p53independent, and p53 loss must be contributing to tumorigenesis in this model by some mechanism other than the abrogation of apoptosis. Additional mouse models that have been created to investigate the sufficiency of Rb inactivation in the genesis of retinoblastoma are discussed in Sec. IV.

E.PNMT-Tag Transgenic Mice

While the IRBP-Tag and IRBP-E7 mice were created with the goal of specifically transforming photoreceptor precursors, T antigen or HPV E6/E7 has been directed to other ocular cell types as well. The phenylethanolamine N-methyltransferase (PNMT) gene is normally expressed in the adrenal medulla and retina [43], and transgenic mice expressing T antigen under the control of the PNMT promoter develop both pheochromocytomas arising from the adrenal medulla as well as retinoblastomas [44]. Within the transgenic retinas, T antigen is expressed both in amacrine cells in the ganglion cell layer and the inner tier of the inner nuclear layer as well as in horizontal cells in the outer tier of the inner nuclear layer. Interestingly, the majority of the horizontal cells disappear in these mice between 2 and 6 weeks of age, with accompanying thinning of the outer plexiform layer. Focal retinal tumors subsequently arise, primarily in the peripheral inner nuclear and ganglion cell layers, between 9 and 16 weeks of age [45]. These tumors express various neuronal markers, such as synaptophysin and neurofilament proteins, and resemble undifferentiated retinoblastomas, lacking rosettes or fleurettes [44].

F.AlphaA-Crystalline-E6/E7 Transgenic Mice

One additional transgenic mouse model for retinoblastoma is a line of mice that express both the HPV E6 and E7 oncogenes under the control of the alphaAcrystalline gene promoter [46]. These mice were originally developed with the goal of

investigating the effects of the E6 and E7 expression in lens epithelial cells, and they display aberrant lens development, with 100% penetrance of bilateral microphthalmia and cataract formation. In addition, a subset of the mice also develop lens tumors at a relatively late age. Unexpectedly, alphaA-crystalline-E6/E7 mice were also found to develop retinoblastomas originating from the bipolar cell layer, resulting from ectopic expression of the transgene in retinal cells. These tumors resemble differentiated human retinoblastomas, with prominent Homer-Wright rosettes. They spread through the optic nerve to the brain, and also metastasize to cervical lymph nodes [46]. Interestingly, the incidence and age of onset of retinoblastomas in this line of transgenic mice is highly dependent on the genetic background of the mice, indicating that other genetic factors in addition to the transgene modulate tumor susceptibility in this model [47].

IV. Rb GENE INACTIVATION IN GENETICALLY ENGINEERED MICE

A.Rb Heterozygous and Homozygous Knockout Mice

Each of the transgenic mouse models described above involves the expression of a viral oncoprotein that binds and inactivates pRb, and thus these models at least in part molecularly mimic human retinoblastoma, in which no functional pRb protein is produced. However, a caveat to this approach is that each of these viral proteins potentially interacts with a number of other cellular proteins, and these interactions may contribute to the resulting phenotype [48]. A more genetically accurate model for hereditary retinoblastoma would be one in which one copy of the Rb gene is inactivated in the mouse germline using a gene targeting approach, and several groups have generated such mice [49–51]. Surprisingly, retinoblastomas are never observed in the heterozygous Rb knockout (Rb þ / ) mice; instead, 100% of these mice develop intermediate lobe pituitary tumors at a mean age of approximately 6–9 months [52–55]. As with human retinoblastomas, the pituitary tumors in these mice invariably show loss of expression of the wild-type Rb allele. Thus, although these mice model the human disease in that loss of one Rb allele creates a marked tumor susceptibility, the tumor spectrum is distinctly different.

When Rb heterozygotes are interbred to generate homozygous knockouts, the Rb / embryos die in midgestation (days E12–15), displaying defects in liver hematopoiesis, neurogenesis, and lens development [49–51,56]. The hematopoietic defect appears to result from an impairment in end-stage differentiation of red blood cells, as Rb / embryos have severely reduced numbers of mature (enucleated) erythrocytes, both in circulation and in the liver [49–51], the major site of erythropoiesis at this stage of mouse development. In addition, Rb / embryos show several areas of massive apoptosis in both the central and peripheral nervous system, associated with inappropriate cell cycle entry in cells that would normally be exiting the cell cycle at this stage of development [57,58]. Similarly, abnormalities are seen in the lens of Rb / embryos, resulting from inappropriate proliferation and accompanying apoptosis [56]. However, there are no obvious abnormalities in retinal development in Rb / embryos.

Because pRb is a key regulator of the cell cycle, directly controlling transcription of genes required for entry of cells into S-phase, one might have

expected that homozygous inactivation of the Rb gene in mice would result in an early embryonic lethal phenotype. The fact that embryos survive as long as they do, approximately two-thirds of the way through gestation, suggests that pRb is not central to the control of every cell cycle. Rather, its major role during development appears to be in the initiation of terminal differentiation, and Rb-deficient cells that fail to exit the cell cycle in response to differentiation signals are rapidly eliminated by apoptosis.

B.Rb-Deficient Chimeric Mice

The early embryonic lethality caused by Rb deficiency precludes the analysis of its role in later stages of development or the role of Rb mutation in tumorigenesis. One strategy that has been employed to circumvent the embryonic lethality is to create chimeric mice by introducing Rb / ES cells into wild-type blastocysts [53,54,59]. Interestingly, such chimeras are readily obtained and display relatively few pathological abnormalities, despite sometimes extensive contribution of Rb / cells to most tissue types, including mature erythrocytes, the liver, and the central nervous system (CNS). The fact that Rb / cells can contribute to tissues in the chimeric mice that show profound defects in fully Rb-deficient embryos suggests that wild-type cells can ‘‘rescue’’ the Rb / cells, either by direct cell-cell contact or by paracrine mechanisms, thereby enabling them to undergo normal terminal differentiation. Thus, the central defect in hematopoiesis, for example, may not lie entirely in the erythroid cells themselves but also in cells contributing to the liver microenvironment necessary to support erythrocyte maturation. This is not the case for all cell types, however, as the lens of chimeric mice exhibits the same defects as in the Rb knockout mice, with abnormal proliferation and apoptosis of Rb / lens fiber cells. The inner retinas of E16.5- to E18.5-day chimeric mice also display regions of ectopic mitosis and apoptosis, and there is a reduced contribution of Rb / cells to the adult retina as compared to other tissues [54]. Nevertheless, the retinas of adult chimeric mice are grossly normal in appearance [53] and retinoblastomas are never observed. Instead, nearly all chimeras develop intermediate lobe pituitary tumors, at significantly earlier ages than the Rb heterozygous knockout mice, consistent with the lack of requirement for somatic loss of the second Rb allele [53,54].

C.The Role of p107 Inactivation in Murine Retinoblastoma

These results indicate that inactivation of Rb is insufficient to induce retinoblastoma, at least in the mouse, and that additional genetic alterations are required. One possibility is that other members of the Rb gene family, p107 or p130, might share functional overlap with Rb and would therefore also need to be inactivated to generate retinoblastoma. This possibility has been explored by creating p107 knockout mice and mice with mutations in both the Rb and p107 genes [60]. p107 / mice are viable and display no obvious abnormalities. However, p107 deficiency exacerbates the defects seen in Rb heterozygous and homozygous knockouts. Rb / :p107 / embryos die approximately 2 days earlier in gestation than Rb / embryos, with accelerated apoptosis in both the liver and the CNS. Rbþ/ :p107 / mice show pronounced growth retardation and increased mortality