Retinoblastoma beyond RB1: recent advances in genetic biomarkers

Keywords::

• biomarkers
• nutlin-3
• RB pathway
• retinoblastoma
• therapeutic targets

Retinoblastoma is an eye cancer that accounts for approximately 4% of all pediatric malignancies and is the most common malignant ocular tumor in children [1]. The disease can be unilateral (when it affects only one eye, as seen in two-thirds of patients) or bilateral (when it affects both eyes). Although most cases of retinoblastoma do not show family history, the importance of hereditary factors for this cancer is well established [2]. The familial disease is transmitted as a typical Mendelian autosomal dominant trait, with high penetrance. The heritable form affects all patients with bilateral retinoblastoma and approximately 15% of patients with the unilateral form, is generally multifocal and is associated with increased risk of second primary cancers. Approximately 40% of retinoblastoma cases are heritable, meaning that patients are at risk of transmitting the disease to their offspring, although most heritable cases arise as sporadic disease. Nonheritable retinoblastoma, which occurs in approximately 60% of cases, is always unilateral, unifocal and is not associated with increased risk of other cancers.

Clinical studies on retinoblastoma led to the identification of the first tumor-suppressor gene – retinoblastoma (RB1), located at 13q14 – and the development of the ‘two-hit’ model [3]. According to Knudson, in fact, two mutational events (M1 and M2) or ‘hits’ are required for tumor onset. Thus, an individual with a germline mutation – M1 – in a tumor-suppressor gene (inherited or occurring de novo) is predisposed to cancer because only another somatic mutation in the same gene – M2 – will be enough to knock out gene function and trigger tumor development [3].

Indeed, biallelic mutations of RB1 have long been recognized as the causative genetic alteration underlying retinoblastoma [4–8]. RB1 functions as the gatekeeper of the G1–S transition of the cell cycle by binding to members of the E2F family of transcription factors and repressing genes that are required for cell proliferation. Consistent with its role as tumor suppressor, mutations in the RB1 pathway are found in most human cancers. However, in the past decades, a myriad of studies focusing on the role of RB1 in cancer development implicated RB1 in many other cellular processes, which could all contribute to its tumor-suppressor function, suggesting that the role of RB1 in cancer is much more complex than previously thought. Moreover, the other members of the RB family, retinoblastoma-like 1 (RBL1 or p107) and retinoblastoma-like 2 (RBL2 or p130), both have overlapping and distinct functions compared with RB1, making it more difficult to dissect specific gene functions.

Thus, despite over two decades having passed since RB1 loss was found to be a crucial event for retinoblastoma development, many questions remain open. For example, why is the high susceptibility of individuals bearing RB1 mutations to specifically develop retinoblastoma not recapitulated in mouse models? In mice, retinal tumors only develop when RB1 is lost together with at least one other member of the RB family [9,10], suggesting that, at least in mice, other family members can compensate for RB1 loss. What makes the human retina so extraordinarily susceptible to RB1 loss? Retinal development is a complex process that leads to the generation of at least seven different specialized cell types. Understanding what is – among all these cell types – the origin of retinoblastoma, could explain how the cellular context affects the requirement of specific mutations for cancer initiation and progression [11–13].

Although RB1 mutations are crucial for retinoblastoma development, some authors argued that the two-hit model oversimplifies the disease process, and suggest that other molecular mechanisms of tumorigenesis must be involved [14,15]. Beyond disputes, the importance of mutational events other than RB1 mutations is almost universally recognized. Many cytogenetic and comparative genomic hybridization (CGH) studies, which have been comprehensively reviewed by Gallie et al., have shown that mutational events affecting RB1 are not the only genomic changes in retinoblastoma [16]. The same authors recently showed that, in humans, loss of both copies of RB1 does not lead directly to retinoblastoma, but to the benign precursor lesion retinoma, which commonly progresses to retinoblastoma after further mutational events – defined M3 to Mn in keeping with Knudson nomenclature [17]. Further studies seem to support this model [18,19].

A detailed understanding of the sequence of molecular events that underlie retinoblastoma pathogenesis is crucial to identify new possible therapeutic targets and markers for early diagnosis and prevention. In fact, although in developed countries the survival rate among children with retinoblastoma is as high as 95% or more, children who survive this disease suffer serious morbidity. Most children with unilateral disease are treated by enucleation, while those with bilateral disease are treated with globe salvage therapies, which are associated with serious toxicities [20]. Furthermore, in less economically developed countries the survival rates are dramatically worse, and retinoblastoma still poses a serious life-threatening problem for children, mostly owing to late diagnosis. A tumor spreading beyond the confines of the eye is still impossible to cure worldwide [21,22]. Therefore, there is currently a great need for the identification of new molecular markers that could serve to open novel therapeutic avenues for more advanced disease, or that could minimize the adverse effects of the current strategies.

Alterations of many other genes beyond RB1 have been found in retinoblastoma. However, for most of these it remains to be established whether they are drivers or passenger mutations. This will be necessary to understand whether such genes have a causal role in tumor development or progression and thus if they can be exploited as cancer-predictive/prognostic biomarkers or, ideally, as therapeutic targets. Moreover, RB1 deficiency itself induces genomic instability, not only by causing defects in the DNA replication process but also by causing chromosome gains, losses and breakage during mitosis [23–27]. Therefore, it needs to be determined whether mutations in other genes are necessary for tumor progression or, rather, are a by-product of cancer cell development.

Several studies implicate RBL2 as one of the genes involved in retinoblastoma development. These are consistent with data from mouse models, as mentioned before, in which at least another RB family member has to be ablated along with RB1 to develop retinoblastoma [28]. We reported that loss of RBL2 expression in sporadic retinoblastomas correlates with a minor degree of cellular differentiation and a lower apoptotic index [29]. Moreover, we identified sequence variants in exon 1 of the RBL2 gene that seem to predispose to promoter hypermethylation and a decreased protein expression in retinoblastomas [30]. We also showed that the treatment of the retinoblastoma cell line Weri-RB1 with the demethylating agent 5-aza-2´-deoxycytidine induced an increase in the expression levels of RBL2 and a significant growth rate reduction. Recently, another study showed that RBL2 is also inactivated by loss of heterozygosity in a high percentage of retinoblastomas [31]. As for the role of RBL2 in retinal tumorigenesis, Dimaras and colleagues suggest that it might be involved in the progression from retinoma to retinoblastoma [17].

Other genes affecting the RB pathway seem to be relevant for retinoblastoma development. For example, p16INK4A – the cell cycle inhibitor that indirectly regulates RB phosphorylation – was proposed to block the transformation of RB1^-/- retina cells at the stage of the nonproliferative precursor lesion retinoma, whereas its inactivation seemed to promote retinoblastoma development [17]. We consistently found a significant downregulation of p16INK4A in approximately half of the tumor and blood samples from the analyzed retinoblastoma patients, and in most parents of the patients showing p16INK4A deregulation. Therefore, p16INK4A alterations might be novel inheritable susceptibility markers to retinoblastoma [32].

Among other cell cycle-related genes, the mitotic kinesin KIF14, which is essential for proper spindle formation and cytokinesis, is often found to be overexpressed in retinoblastomas compared with normal retina, owing in part to gene copy number alteration [33–35]. An increase in KIF14 copy number and mRNA expression was also found in retinomas, although its expression was much higher in the adjacent retinoblastoma tissue, suggesting that it might play a part in tumor progression [17]. Despite no significant correlation being observed between KIF14 expression and laterality, invasion, differentiation status and duration of the disease, an older median age at diagnosis was significantly correlated with a higher KIF14 expression, suggesting that different mutational pathways probably underlie early- or late-onset tumors [34]. Interestingly, KIF14 is also overexpressed in breast tumors and non-small-cell lung cancer, which correlates with a poor prognosis, and in medulloblastoma cell lines [16,33,36].

Madhavan and colleagues also described, in association with KIF14 overexpression, the upregulation of the RB-regulated transcription factor E2F3[35]. E2F3 maps on the 6p22 locus, and 6p gains are among the most common changes identified by CGH studies in retinoblastoma [16]. On the same locus lies the DNA-binding protein DEK, which exerts a pleiotropic role in cancer, affecting both chromatin remodeling and RNA processing [37]. Thus, E2F3 and DEK are both candidate oncogenes for retinoblastoma. Moreover, E2F3 has recently been shown to regulate DEK transcription, and Gallie and colleagues suggest that they might cooperate when both are upregulated in retinoblastoma [16].

MYCN amplification is another feature of retinoblastomas, both in humans and in mouse models. MYCN is a member of a small family of transcription factors (MYC, MYCL and MYCN), which regulate many cellular processes, including cell growth and proliferation, differentiation and apoptosis. MYCN expression is normally restricted to the central and peripheral nervous systems, kidney, lung and spleen during embryonic development [38], whereas MYCN amplification occurs in a variety of tumors, most notably in neuroblastoma [39]. In retinoblastoma, MYCN amplification is associated with a high proliferation index [40] and seems to correlate with distant metastases [41,42], suggesting a role in tumor progression. Interestingly, in neuroblastoma, MYCN has shown a paradoxical function because it is able to induce cell proliferation and also to sensitize cells to apoptosis [43]. This apparent inconsistency for a tumor as aggressive and therapy resistant as neuroblastoma might be explained by the high levels of the inhibitor of apoptosis, survivin. High levels of survivin have in fact been associated with MYCN amplification and a poor prognosis [43]. Remarkably, retinoblastomas also show high levels of survivin, which, on this basis, warrants further investigation and might represent a useful therapeutic target [44].

Considering the invasive nature of retinoblastoma, genes involved in extracellular matrix degradation and cell adhesion must also be relevant to tumor development and progression. Two studies have addressed this issue thus far [45,46]. Most notably, the gene encoding CDH11, a type II classical cadherin from the cadherin superfamily that mediates cell–cell adhesion, is frequently lost in both human retinoblastomas and in retinoblastomas arising in mouse models [47,48]. Recent data show that CDH11 exerts tumor-suppressor functions in retinoblastoma through the promotion of cell death [49].

Similarly to the RB pathway, the p53 pathway is deregulated in most human tumors. In retinoblastoma, despite an intact p53 molecule, the p53 pathway is altered by amplification of p53-negative regulators such as MDM4 (also known as MDMX; amplified in 65% of retinoblastomas) and MDM2 (amplified in 10% of retinoblastomas) [50]. Interestingly, targeting the p53–MDM2/4 interaction with the small molecule nutlin-3 has proven effective against retinoblastoma both in vitro and in vivo[50,51]. Other studies are investigating the potential role of other p53 family members in retinoblastoma [52].

Recently, epigenetic modifications have also been increasingly recognized as one of the most common molecular alterations in cancer, associated both with the inactivation of tumor-suppressor genes and chromosomal instability [53]. Several aberrantly methylated genes have also been identified in retinoblastoma tumors, including RB1 itself [54], RBL2[32], CASP8 and RASSF1A[55–57], MGMT and MLH1[57,58], and NEUROG1[59]. Because these alterations might have an important role in retinoblastoma progression, the potential anticancer properties of epigenetic drugs should be further investigated [60].

The list of other candidate genes, beyond RB1, which are crucial to retinoblastoma development and progression, is rapidly growing owing to powerful new techniques and high-throughput screenings. Although for many of these genes the functional role in retinal tumorigenesis still needs to be clarified, it is likely that some will serve to identify new possible therapeutic strategies. Furthermore, the recently developed animal models of retinoblastoma will be an invaluable tool to analyze novel therapeutic strategies at the preclinical level, and to help move the clinical translation of these findings forward.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.