We demonstrate that most human retinoblastomas in which
RB1 is inactivated (
RB1
-/-) show low ARF protein despite high
ARF mRNA. In contrast,
RB1
+/+ retinoblastoma tumors show low
ARF mRNA. This data confirms previous findings [
3] and our results in the Rb knockout mouse (data not shown) that loss of pRB leads to an increase in
ARF mRNA due to transcriptional activation by E2F1. High
ARF mRNA but low ARF protein also suggests attenuation of synthesis or instability of ARF in most
RB1
-/- retinoblastoma cell lines. For the cell line Y79, low ARF was at least in part due to low
ARF mRNA. Although Y79 has an
RB1
-/- genome, a truncated pRb protein is produced; this may affect its own transcription and interplay with E2F family members [
22,
23].
Through viral overexpression of ARF, we showed in WERI-Rb1 cells that: i) exogenous ARF protein expression induces the p53 response even when MDM2 and MDM4 are expressed; and ii) the ARF-induced p53 response results in a functional decrease in cell viability. These data demonstrate that the integrity of the p53 pathway is maintained in retinoblastoma cells, and that ARF protein expression may play an important role in controlling p53 responses. We have shown that proteasomal degradation does not account for low ARF protein in the WERI-Rb1 cell line, thus other mechanisms of ARF protein regulation may exist. Indeed, miR-24 has documented expression in normal retinas and retinoblastomas, and demonstrated translational repression of
p16
INK4a
mRNA, which shares 100% homology with the 3' untranslated region of
ARF [
20,
24,
25]. In our expression analyses, miR-24 is less abundant in human retinoblastomas than in normal fetal and adult retinas, with highest expression in adult retinas, consistent with reports of elevated miR-24 expression in terminally differentiated cells [
21]. We showed that fetal retinas in general demonstrate higher miR-24 expression than retinoblastoma tumors. This data suggests regulation of miR-24 expression during retinal development, perhaps due to its role in regulation of p53 via ARF to suppress p53 hyperactivation and unwanted cell death. Indeed miR-24 was shown to be involved in developmental apoptosis in Xenopus retina [
26]. Paradoxically, when pRB is inactivated during retinal tumorigenesis, ARF protein regulation mediated by high miR-24 intrinsic levels may impede the tumor suppressor functions of p53. Moreover, the pleiotropic miR-24, in addition to its regulation of ARF, has been shown to repress expression of MYC and E2F2 [
21], which are important in driving proliferation and imperative to tumor growth. The reduced miR-24 level in retinoblastomas relative to normal retinas may be explained by the requirement of sustained expression of proliferation genes in tumor cells while maintaining sufficient amounts of miR-24 to repress the ARF tumor suppressor. Intriguingly, the low miR-24:protein ratios in
RB1
+/+ retinoblastomas are consistent with the notion that, in the absence of selection pressure from
ARF activation induced by
RB1 loss, tumor cells may maximize proliferation through the derepression of MYC and E2F2 [
21] without the compromise of tumor suppressor activation. Thus the observed miR-24 in retinoblastoma tumors might be at an optimal level that maximizes tumor cell growth and survival. Some
RB1
-/-
primary tumors showed low miR-24 expression, comparable to
RB1
+/+
cell lines. These "low miR-24"
RB1
-/-
tumors raise the possibility that the level of miR-24 expression may be insufficient to compromise the p53 response in some retinoblastoma tumors. However other factors besides expression, such as components of the RISC complex, may influence the effect of miR-24 activity in retinoblastoma cells [
27].
ARF could also be regulated by other microRNAs in retinoblastomas. For example, expression of miR-125b and miR-24 is similar in retinoblastomas [
28], and miR-125b has been shown to regulate the 3' untranslated region of
p16
INK4a
and
ARF mRNA [
20]. To gain insight on the role of miR-24 on the ARF-p53 axis, a thorough examination of the biological and functional implications of miR-24 expression in human retinal and retinoblastoma tumor development is thus warranted.
Regulation of ARF protein expression by miR-24 in retinoblastoma cell lines points to a possible mechanism through which ARF protein is decreased in retinoblastoma cells. We uncovered an intact p53 response in WERI-Rb1 cells through overexpression of ARF protein; however, whether targeting miR-24 in retinoblastoma cells would increase ARF protein expression to optimal levels to elicit a p53 response remains to be investigated.
Our miR-24 and ARF protein expression data are in agreement with ARF protein regulation by miR-24 that is unique to
RB1
-/- retinoblastomas. However, SaOS2 and OVCAR3 cell lines show similar miR-24:protein ratios as
RB1
-/- retinoblastoma cell lines, suggesting that miR-24 regulation of ARF may apply to other cell types. Given that the other cell lines tested display different genetic or protein changes to those seen in retinoblastoma cell lines, such as wild-type
RB1 (SKOV3 and OVCAR3), inactivated pRb (HeLa and HEK-T) or mutant
TP53 (SKOV3, OVCAR3 and SaOS2) [
29,
30], further investigation into the functional mechanisms of ARF regulation via miR-24 are required. Differential miR-24:protein ratios between
RB1
-/- retinoblastomas, fetal retina and adult retina, and
RB1
+/+ retinoblastomas, combined with ARF-mediated apoptosis and miR-24-regulated ARF expression in WERI-Rb1 cells nonetheless demonstrate a unique mechanism in
RB1
-/- retinoblastoma tumors through which proliferation could be maintained by miR-24 suppression of ARF.