Ординатура / Офтальмология / Английские материалы / Ocular Oncology_Albert, Polans_2003

induction of cell cycle progression and in the negative regulation of differentiation [180–184]. As in the case of E2Fs, overexpression of Id2 has been associated with both proliferation and apoptosis [184]. pRb binds Id2 [185], suggesting that survival of differentiating neurons and erythrocytes requires direct physiological control of Id2 by pRb [179,183]. The precise nature of pRb’s interaction with Id2 remains to be defined.

3.Transcriptional Activation of Tissue-Specific Genes

While the retinoblastoma protein is best known for its repressor functions, pRb also serves to activate tissue-specific gene expression in terminally differentiating cells in several lineages. In the best studied example, pRb plays an indispensable role in the expression of late skeletal muscle genes through its interaction with MyoD [186], a basic helix-loop-helix (bHLH) family transcription factor which plays a central role in myogenesis [187]. It has been proposed that pRb stimulates the transactivating function of MyoD through direct binding and enhancement of MyoD’s DNA binding ability [188], but data are inconsistent on this point [70]. Recent findings suggest a novel mechanism of transcriptional activation involving the chromatin remodeling enzyme HDAC1. In an early study, HDAC1 was shown to repress the transcriptional activity of MyoD by direct binding to its bHLH domain [189]. Subsequent experiments revealed that during myogenesis, hypophosphorylated pRb competes for HDAC1, displacing it from MyoD. This results in the derepression of MyoD and the activation of late muscle differentiation genes [190]. This model provides an attractive mechanism for precise coordination of terminal cell cycle arrest and induction of differentiation genes by pRb in skeletal muscle tissue.

Some researchers have suggested that pRb also activates bHLH family transcription factors in differentiating neurons [191,192], but evidence for this hypothesis is currently circumstantial. bHLH family proteins do play a critical role in the induction of differentiation genes in nervous tissue, including retina [193–196]. In addition, pRb loss is associated with decreased expression of neuronal differentiation genes in RB1 / mice [142]. However, it remains unclear whether pRb loss plays a proximal role in this outcome [197]. There is also no evidence that pRb directly associates with bHLH proteins in nervous tissue. On the other hand, it is possible the pRb regulates bHLH transcription factors indirectly through interaction with Id2. Like other Id family proteins, Id2 negatively regulates bHLH family proteins by forming inactive Id-bHLH heterodimers. Hence, pRb could positively regulate bHLH-mediated induction of differentiation genes in nervous tissue by negatively regulating Id2 [192]. This proposed mechanism is consistent with recent findings in differentiating cortical progenitor cells [198]. Id2 overexpression in these cells results in the complete suppression of neuron-specific genes and ultimately in apoptosis; both of these effects are prevented by the added introduction of constitutively activated pRb. Conceivably, pRb could also activate bHLH transcription factors in differentiating neurons through an HDAC1-mediated mechanism, as observed in skeletal muscle differentiation (see above).

pRb induces differentiation gene expression in several cell types by enhancing the activity of CCAAT/enhancer-binding protein (C/EBP) family transcription factors, which are important in adipogenesis, hematopoiesis, and hepatogenesis [199–203]. pRb is required for the differentiation of cultured fibroblasts into adipose

tissue, and it promotes this process by stimulating the transcriptional activity of C/EBPa [204] and C/EBPb [50]. pRb binds C/EBPa, C/EBPb, as well as C/EBPd [50,145] suggesting that pRb regulates these C/EBPs through direct interaction [50]. However, abrogation of the pRb binding domain of C/EBPa has no detectable effect on adipogenesis, making this conclusion less certain [205]. pRb also promotes monocyte/macrophage differentiation in lymphoma cells by enhancing the transactivating function of C/EBPb (also called NF-IL6 [199]) [51]. C/EBPs have been implicated in lung development, and pRb has recently been shown to stimulate the transcription of a differentiation-specific gene (surfactant protein D) in embryonic lung epithelial cells through the formation of cooperative, DNA-binding complexes containing C/EBPa, C/EBPb, and C/EBPd [145].

pRb is also required to activate the osteoblast transcription factor CBFA1 [44], which is essential for the expression of late bone differentiation markers [206]. pRb binds CBFA1 and facilitates the ability of CBFA1 to bind osteoblast-specific promoters [44]. In another example, pRb has been proposed to play a role in tissuespecific gene expression in differentiating keratinocytes, where it binds and activates the AP-1 transcription factor c-jun [52]. A general role for pRb in the induction of differentiation-specific genes has been proposed by Khochbin and coworkers [207]. These investigators found that pRb cooperates with HBP1 to induce the expression of histone H10 and possibly other important chromatin remodeling genes which are expressed ubiquitously in terminally differentiated cells. The transactivating function of the pRb-HBP1 complex in this context contrasts with the proposed repressor function of the pRb-HBP1 complex in the induction of cell cycle arrest (see above), but both mechanisms could serve to promote terminal differentiation.

4.Inhibition of Cell Cycle Re-entry

In some cell types, constitutive cell cycle control by pRb is important for maintaining the postmitotic state. For example, pRb is required for the prevention of cell cycle reentry in differentiated muscle cells [155,186]. Mechanisms whereby pRb prevents differentiated cells from re-entering the cell cycle remain unclear, although the involvement of E2Fs and HBP1 has been proposed [42]. RB1 is highly expressed in terminally differentiated nervous tissue [208], including several cell types within the retina [209], suggesting a similar role for pRb in the prevention of cell-cycle re-entry in postmitotic neurons. However, ex vivo studies have demonstrated that pRb and other pocket family proteins are dispensable for the maintenance of terminal cell cycle arrest and dispensable for survival in fully differentiated neurons [156]. Consistent with these findings, pRB-E2F complexes are undetectable in the mature nervous system [210]. While the function of pRb in mature neurons remains to be defined, these observations suggest that its function in such tissues is unrelated to cell cycle control.

III.pRb STRUCTURE

pRb has several important functional domains, the chief of which is an A/B pocket domain, also called the small pocket. The C-terminal region of pRb is also important, containing sequences contiguous with the A/B pocket that are critical for growth suppression. While the role of the N-terminal region of pRb is least defined,

that pRb recruits these complexes consecutively to E2F sites [130]. The chromatin remodeling ATPase BRG1 also contains an LxCxE domain but does do not require this sequence to bind pRb [127,215]. This observation is consistent with the proposal that pRb can simultaneously recruit HDAC and BRG1 to E2F sites [127]. Finally, HBP1, which cooperates with pRb in the induction of terminal cell cycle arrest and differentiation-specific gene expression, also binds to pRb by means of an LxCxE domain [147,207]. The crystal structure of the pRb A/B pocket domain bound to a conserved E7 peptide containing the LxCxE motif has been described [74]. Crystallography demonstrates that the LxCxE motif binds to highly conserved residues in a shallow groove on the B box. While the A box does not bind directly to the LxCxE motif, it is required for LxCxE binding and probably facilitates stable folding of the B box.

The A/B interface, formed by the interaction of the A and B boxes, is also highly conserved. This high degree of conservation suggests that this interface serves as another pocket binding site, possibly for E2Fs, which lack an LxCxE motif [74]. pRb binding of E2F at the A/B interface could provide a means for pRb to simultaneously bind E2Fs and chromatin remodeling complexes. Several missense mutations affecting the A/B interface have been described, providing supporting evidence that this structure is important for tumor suppression by pRb [74]. The precise mechanism whereby viral oncoprotein binding to pRb disrupts E2F binding remains unclear, but it has been suggested that viral oncoproteins compete with E2F for a non-LxCxE binding site [74].

B.The C Terminus

The large pocket domain comprises the A/B pocket plus a portion of the C-terminal domain through residue 869. This domain is required for E2F binding and for pRbmediated growth suppression [216]. The vast majority of naturally occurring RB1 mutations (98% in one large study [217]) affect the large pocket, suggesting the critical significance of this domain for tumor suppression by pRb. The C terminal sequences of the large pocket contain a second binding site for E2Fs as well as a distinct binding site for the c-Abl tyrosine kinase [218,219]. This c-Abl binding site contributes to growth suppression by pRb [219].

pRb binding of several differentiation factors—including MyoD, CBFA1, and c-Jun—requires sequences in both the B box and in the C terminus [44,52,188]. Exon 25 encodes a region of the C terminus that contains a nuclear localization sequence required for full wild-type activity [220,221]. The C-terminal domain also contains a binding site for the oncoprotein mdm2, which negatively regulates p53 by promoting its degradation [222]. While the functional significance of the pRb-mdm2 interaction remains unclear [70,223], pRb has been shown to promote the apoptotic function of p53 by binding to mdm2 [224].

C.The N Terminus

The N-terminal region of pRb is the least studied region of this protein, and its functional significance is not well defined. This region does contain binding sites for several proteins [225–228], including p202, a protein that negatively regulates the cell cycle through the pRB/E2F pathway and is targeted by the adenovirus E1A

oncoprotein [229–231]. In addition, in vitro analyses of N-terminally mutated pRb indicate that this domain is important for normal protein function [43,232].

Deletion analyses in mice suggest that this region does play a significant role in tumor suppression by pRb. Introduction of N-terminal-deleted RB1 fails to rescue pituitary tumors which arise in RB1þ/ mice (discussed below), suggesting that the N-terminal region contributes to tumor suppression by pRb [233]. N-terminal deletions have also been reported in low-penetrance retinoblastoma kindreds, providing further evidence for this region’s importance in tumor suppression [43,234,235]. Surprisingly, however, N-terminally truncated pRb has also been associated with enhanced tumor suppressive function in cell culture systems [236,237], and it has been suggested that N-terminal truncation of pRb may represent a cellular mechanism for modulating pRb function [237].

D.Insights from Low-Penetrance pRb Mutants

Although RB1 has no known hot spots for mutation, truncating mutations are very common. Ninety percent of clinically significant mutations are characterized by frameshift, nonsense, or splice site mutations, which produce a premature stop codon and a truncated transcript [217]. These mutations are distributed fairly evenly throughout the N-terminal and large-pocket domains, invariably resulting in partial or complete loss of large pocket coding sequences. The high frequency of truncating mutations in retinoblastoma suggests that tumorigenesis is most favored by mutations that globally inactivate pRb. This provides further support for the view that multiple functional domains contribute to tumor suppression by pRb.

This view is also consistent with genetic findings in low-penetrance retinoblastoma kindreds, where truncating RB1 mutations are notably absent and disease is less severe. RB1 mutations reported in low-penetrance kindreds include point mutations in the promoter region [239,240], splice site mutations [241–243], amino acid substitutions (missense mutations) [242,244–249], and in-frame deletions [221,234,245]. These mutations have consistently been predicted to only mildly alter pRb expression or structure, permitting residual protein function [250]. Consistent with these predictions, retinoblastoma in these kindreds often skips generations [23,221,234,235,239,240,244,245,247,251,252], and is associated with unilateral involvement [23,221,235,239,245,247], benign retinoma [23,221,235,244,246], delayed onset [243], few intraocular recurrences [244], and early and complete response to therapy [244].

A model has been proposed to explain how minimally deranging germline RB1 mutations results in low-penetrance, low-expressivity retinoblastoma (Fig. 4) [234,239,253]. This model relies on the observation that in approximately 70% of cases, the second hit leading to RB tumor formation occurs following loss of heterozygosity (LOH) for the normal RB1 allele. In heritable RB, the usual consequence of LOH is two copies of the germline mutant allele. In cases where the germline mutant allele encodes a protein with residual activity, biallelic expression of the mutant protein as a consequence of LOH is proposed to result in pRb activity which is suboptimal yet sufficient to prevent tumorigenesis (Fig. 4).

Functional studies of low-penetrance pRb mutants have provided further insight into both the mechanisms of low-penetrance retinoblastoma and the tumor

muscle, so the precise role of the LxCxE binding site in retinogenesis and retinoblastoma tumorigenesis remains unclear.

IV. pRb AND THE CELL CYCLE

pRb functions at the nexus of a signaling network that controls the cell’s replicative and developmental fate. The upstream effectors of this network include mitogenic and antiproliferative factors that act in an antagonistic fashion. On the one hand, mitogens produce their effects through activation of cyclin D-cdk4/6 complexes, which phosphorylate and inactivate pRb. This critical action derepresses E2F target genes, enabling the cell to begin its progression through the cell cycle. In contrast, antiproliferative signals induce the expression of CDKIs, which positively regulate pRb by inactivating its negative regulators, the cyclin-cdk complexes. When activated by CDKIs, hypophosphorylated pRb induces cell cycle arrest through the repression of E2F target genes. In this manner, pRb functions as a sensitive indicator and effector of the balance of proliferative and antiproliferative signals emanating both from within the cell and from its external environment [39,40].

The cell cycle consists of a presynthetic gap phase (G1), followed by a DNA synthesis phase (S), a second gap phase (G2), and finally mitosis (M). Expression of cdks is relatively invariant, but the kinase activity of these enzymes requires association with cyclins, whose expression is strictly regulated throughout the cell cycle [40,255,256] (Fig. 5). Tightly controlled fluctuations in cyclin expression result in assembly and activation of distinct cyclin-cdk complexes at sequential stages throughout the cell cycle, beginning in early G1 with the assembly of cyclinD-cdk4/6. Cyclin E-cdk2 complexes are assembled in mid to late G1, and cyclin A- and B-dependent kinases are activated later in the cell cycle.

As discussed previously, pRb induces reversible cell cycle arrest (quiescence) through formation of corepressor complexes at E2F-responsive promoters (Fig. 1). Continuous mitogenic signaling is required to induce the cell out of quiescence (also termed G0) and back into the cell cycle. The cell remains dependent on this stimulation until it passes through a critical checkpoint in late G1 called the restriction point, after which it becomes irreversibly committed to progressing through the cell cycle (mitogen-independent). A current model for the cell’s progression out of quiescence is depicted in a simplified fashion in Figure 6 (for critical discussion, see Refs. 79 and 257).

The cell’s emergence from quiescence requires the activity of Ras family proteins. These proteins relay exogenous growth signals from the inner cell membrane to the nucleus via a cascade that activates transcription of cyclin D [258–260]. Induced cyclin D binds to its partners cdk4 or cdk6 to form complexes whose assembly requires the cooperation of the CDKIs p21Cip1 and p27Kip1 [82]. Hence, these cdk ‘‘inhibitors’’ actually serve as positive regulators of cell cycle progression in early G1. Once assembled, cyclin D-cdk4/6 must also be activated by cdk activating kinase (CAK), which adds a phosphate group.

Activated cyclin D-cdk4/6 partially inactivates pRb by phosphorylating the protein at specific sites [72,79,257]. This phosphorylation results in disassociation of HDAC complexes from pRb and derepression of HDAC-regulated E2F target genes, notably cyclin E (Fig. 6). Upregulation of cyclin E expression results in the

Figure 5 Cell cycle-dependent expression of cyclins. While cdk expression is relatively constant throughout the cell cycle, cyclin expression fluctuates in a cell cycle-dependent manner. D-type cyclins are induced in early G1 by mitogenic signaling, and their activity results in the expression of cyclin E later in G1. Cyclin E is required for S-phase entry and for expression of cyclin A, which is in turn required for the completion of S phase. Cyclin B is induced during S phase and regulates the cell’s progression through mitosis.

formation of activated cyclin E-cdk2 complexes, which further inactivate pRb through additional phosphorylation events. This action disrupts pRb’s binding to E2F and BRG1 and results in the derepression of BRG1-regulated E2F target genes. These genes include additional cell cycle genes (notably cyclin A) and DNA synthesis genes with activities that promote S phase progression.

While it is widely held that upregulation of cyclin E and functional inactivation of pRb is critical for the cell’s passage through the restriction point (Fig. 6), recent evidence suggests that the converse may be true. Detailed analysis of individual cycling cells indicates that passage through the restriction point occurs prior to the accumulation of cyclin E [261]. These findings suggest that upregulation of cyclin E is more closely associated with the initiation of DNA synthesis than with the establishment of mitogen independence.

Levels of cyclin E-cdk2 decline precipitously upon the cell’s entry into S phase. According to an early model, cyclin A- and B-dependent kinases maintain pRb in a hyperphosphorylated, functionally inactivated state as the cell progresses through S and G2/M [40]. However, more recent data suggests that pRb is only partially inactivated in G1 and that pRb performs significant functions during later phases of the cell cycle, including induction of cell cycle arrest in G2/M [262–269].

pRb is dephosphorylated in late M phase by a type 1 protein phosphatase (PP1a) [270–272]. According to the current consensus, dephosphorylation of pRb at

Figure 6 Model for the cell’s emergence from quiescence. pRb blocks cell cycle progression by binding and inhibiting E2Fs and by recruiting repressor complexes to E2F sites. For simplicity, only one of the pRb corepressor complexes depicted in Figure 1 is depicted in this model. Chronic mitogenic signaling induces the expression of D-type cyclins, resulting in the assembly and activation of cyclin D-cdk4/6 complexes. Cyclin D–dependent kinase phosphorylates pRb at specific sites, abrogating the association between pRb and RBP1HDAC complexes. This results in the derepression of HDAC-regulated E2F target genes, including cyclin E. Induced cyclin E complexes with cdk2, and activated cyclin E-cdk2 phosphorylates pRb at additional sites, disrupting its association with E2F and BRG1 complexes. This results in the derepression of BRG1-regulated E2F target genes, stimulating the cell’s entry into S phase. The restriction point is depicted in its customary position subsequent to the induction of cyclin E. However, recent evidence suggests that cyclin E is actually induced only after the cell attains mitogen independence (see text for details). Ink4 family CDKIs positively regulate pRb by inhibiting the action of cyclin D–dependent kinase. Cip/Kip family proteins promote the assembly of cyclin D-cdk4/6 complexes in early G1 but also serve as potent inhibitors of cyclin E-cdk2 and cyclin A-cdk2.

the mitotic exit enables pRb to resume its function as a negative regulator of the G1/S phase transition during the cell’s next passage through G1 [39,40]. However, exclusively hyperphosphorylated Rb has been detected in several exponentially growing cell lines, suggesting that cell-cycle dependent phosphorylation and dephosphorylation of pRb is not necessary for passage through the cell cycle in actively proliferating cells [273,274]. This view is consistent with the observation that embryonic cells in RB1 / mice appear cycle normally until relatively late in development [45–47,141].

Cell cycle arrest is effected by antiproliferative signals that convey their effects through upregulation of Ink4 and Cip/Kip family CDKIs (Fig. 6). Ink4 family proteins (p15Ink4b, p16Ink4a, 18Ink4c, and p19Ink4d) specifically inactivate cyclin D-cdk4/6 complexes. Cip/Kip family proteins (p21Cip1, p27Kip1, and p57 Kip2)

function as potent inhibitors of cyclin E-cdk2 and cyclin A-cdk2 complexes. Antiproliferative signals that regulate CDKIs include transforming growth factor b, which induces p15Ink4b [85], and cell-to-cell contact, which is associated with the induction of Cip/Kip family proteins [85]. Senescence signals, which are often associated with shortening of telomeres at the distal ends of chromosomes, induce both p16Ink4a [82] and p27Kip1 [275]. p53 induces pRb-mediated cell cycle arrest in response to DNA damage by upregulating p21Cip1 [286,287]. Finally, Cip/Kip family proteins also respond to differentiation signals. Interestingly, like pRb, Cip/Kip family proteins have been shown to activate the expression of differentiation genes by mechanisms distinct from their cell cycle regulatory functions [83,84,282].

V.pRb AND CANCER

Malignant transformation is generally regarded as a multistep, Darwinian process in which the cell sustains a series of mutations that endow it with growth and survival advantages [279]. The earliest traits required for malignant transformation include replicative self-sufficiency, insensitivity to antiproliferative signals, and resistance to apoptosis. Later requirements for tumor growth and spread include avoidance of replicative senescence, new blood vessel formation (angiogenesis), and the acquisition of tissue invasiveness (metastatic potential). Consistent with a proposed universal requirement for deregulated antiproliferative signaling in malignancy, functional inactivation of pRb is observed in most if not all tumors. It should be emphasized, however, that inactivation of pRb occurs most frequently through dysfunction of its upstream regulators. For example, mutational inactivation of p16Ink4a is commonly observed in many malignancies [280]. Amplification of cyclin D is another common mechanism of pRb inactivation in tumorigenesis [281]. These findings support the concept of a core ‘‘Rb pathway,’’ p16Ink4a\cyclin D-cdk4/6\ pRb, which must be disabled for malignant transformation to occur [39–41].

The high frequency of Rb pathway defects upstream of pRb nevertheless suggests that this pathway is not linear, and that other proteins in the pocket family, i.e., p107 and p130, are also important downstream effectors of this pathway. This conclusion is supported by recent findings indicating that p16-mediated cell cycle arrest depends upon the activity of both pRb and either p107 or p130 [282]. Since p16 is a critical positive regulator of all three pocket proteins, mutational inactivation of p16 may be functionally equivalent to loss of both alleles of all three pocket proteins, a set of events that is unlikely to occur in a single cell [60]. Similarly, the induction of cell cycle progression by cyclin D-dependent kinase activity appears to involve phosphorylation and inactivation of not only of pRb but also p107 and p130 [283–286]. Constitutive phosphorylation of all pocket family proteins by deregulated cyclin D–dependent kinase represents an oncogenic mechanism similar to that which has evolved in viral oncoproteins, which bind and inactivate all pocket family proteins [61]. Employing a mechanism that is even more akin to the action of viral oncoproteins, the helix-loop-helix (HLH) transcription factor Id2 also binds and inactivates all pocket family proteins when it is overexpressed in cell culture [185,287]. This mechanism has recently been