Complex transcriptional regulatory networks around the retinoblastoma tumor suppressor gene

The retinoblastoma tumor suppressor RB belongs to a cellular pathway that is altered on some epistatic level in virtually all human cancers. When RB itself is not mutated, amplifications or activating mutations in Cyclin D or CDK4, or loss of upstream inhibitors of these proteins, result in RB hyperphosphorylation, causing the loss of RB inhibition on E2F transcription factors and leading to deregulation of the cell cycle and other tumorigenic processes.¹ Even though RB is a potent inhibitor of cell cycle progression and is expressed in most, if not all cell types, children born with a mutant RB allele are only at a high risk to develop retinoblastoma and osteosarcoma. Furthermore, RB mutant mice specifically develop pituitary, thyroid and adrenal gland tumors. One explanation for this tissue-specific sensitivity is that the two other RB family members 107 and p130 compensate for loss of RB in specific settings, an idea supported by the development of multiple tumor types in mice with combinations of mutations in RB family genes.²

Multi-member gene families, like the RB family, are thought to exist due to the evolutionary retention of individual family members that have gained either unique functions or unique expression patterns; more recently, it has been suggested that direct transcriptional control of one family member over another may create genetic backup circuits, as an additional reason for retaining similar genes.^3,4

Because RB family members function as transcriptional co-factors, it is possible that the molecular mechanisms underlying the functional overlap between RB, p107 and p130 are based on the ability of one or more family members to affect the transcription of itself or others. Indeed, we and others have shown that loss of RB function results in the transcriptional upregulation of p107, probably through the activation of E2F transcription factors (reviewed in ref. 5); strikingly, this negative feedback between RB and a second RB-related protein has independently evolved in plants⁶ and in flies.⁷

In order to further delve into the potential compensatory feedback mechanisms that may exist between RB, p107 and p130, we generated transgenic mice in which the gene coding for a GFP reporter was inserted at the first codon of each gene within the context of a large BAC. This system allowed us to examine the transcription of RB, p107 and p130, without affecting the expression levels of the endogenous proteins (reviewed in refs. ⁵ and ⁸; and unpublished data). Using RB-GFP transgenic mice,⁸ we found that RB is differentially transcribed in a tissue-specific pattern, as well as between different lineages within the same organ. Furthermore, although the promoter of RB, like that of p107, contains a highly conserved E2F binding site, we found no correlation between RB expression and the proliferative index of cells in vivo, while a clear correlation was observed between p107 expression and proliferation. These observations suggest that, unlike p107, tissue-specific regulation of RB transcription overrides its regulation by E2F (Fig. 1A). Accordingly, RB transcription is only slightly altered in the absence of RB alone. While we found that some myeloid cells demonstrate altered RB promoter activity in the absence of p107 or p130, we did not observe evidence of a more general transcriptional compensation. Thus, the absence of gross phenotypes in p107 or p130 knockout mice² is probably not due to a compensatory upregulation of RB transcription. Only upon deletion of the whole RB family did we observe a significant increase in RB promoter activity in several cell types in vivo. This triple mutation recapitulates the functional inactivation of the three family members by viral oncoproteins or by mutations in upstream regulators of the RB pathway. Upregulation of RB transcription in this context may act as a negative feedback mechanism to increase RB levels and suppress cancer development by overcoming or saturating the functional inactivation. Alternatively, RB has been shown to have pro-survival functions,¹ such that this increase in RB expression may counterintuitively be supporting tumor growth.

Figure 1.

A model for transcriptional regulatory networks around RB. (A) RB family members control RB transcription in mammalian cells, potentially through their interactions with SP1, ATF and E2F transcription factors in the proximal RB promoter. In addition, variations in RB transcription levels in different cell types in mouse organs and tissues suggest that RB levels are controlled by still unidentified tissue-specific transcription factors, which may interact with RB family members. (B) RB family members also control the transcription of p107, probably by interacting with E2F family members at the p107 promoter. The control of RB and p107 by RB family members creates a complex transcriptional regulatory network whose balance may be important to control cell cycle progression in wild-type cells and in cells with mutations in the RB pathway.

These studies illustrate the complexity of the transcriptional networks around RB (Fig. 1B). They also underscore the fact that the regulation of RB transcription may be an important aspect of cell cycle progression under physiological and pathological conditions. Finally, our studies with RB-GFP BAC transgenic mice suggest that RB mRNA levels may also be regulated at the post-transcriptional level, for example in the embryonic liver.⁸ Future studies will continue to examine the possibility that this other level of regulation may be under the control of p107 and p130. Understanding the interconnectivity between RB family members may enable us not only to understand why some tumors lose RB and others inactivate the whole family, but also how to take advantage of these connections to artificially increase compensatory expression of RB family members in the cells of cancer patients.

Abbreviations:

GFP

green fluorescent protein

RB

mouse or human retinoblastoma gene

RB

retinoblastoma protein

CDK

cyclin-dependent kinase

E2F

adenovirus E2 promoter binding factor

BAC

bacterial artificial chromosome