Deletions of Retinoblastoma 1 (Rb1) and Its Repressing Target S Phase Kinase-associated protein 2 (Skp2) Are Synthetic Lethal in Mouse Embryogenesis*

Tumor suppressor pRb represses Skp2, a substrate-recruiting subunit of the SCFSkp2 ubiquitin ligase. Rb1+/− mice incur “two-hit” pituitary tumorigenesis; Skp2−/−;Rb1+/− mice do not. Rb1−/− embryos die on embryonic day (E) 14.5–15.5. Here, we report that Skp2−/−;Rb1−/− embryos died on E11.5, establishing an organismal level synthetic lethal relationship between Rb1 and Skp2. On E10.5, Rb1−/− placentas showed similarly active proliferation and similarly inactive apoptosis as WT placenta, whereas Rb1−/− embryos showed ectopic proliferation without increased apoptosis in the brain. Combining Skp2−/− did not reduce proliferation or increase apoptosis in the placentas but induced extensive apoptosis in the brain. We conditionally deleted Rb1 in neuronal lineage with Nes-Cre and reproduced the brain apoptosis in E13.5 Nes-Cre;Rb1lox/lox;Skp2−/− embryos, demonstrating their synthetic lethal relationship at a cell autonomous level. Nes-Cre-mediated Rb1 deletion increased expression of proliferative E2F target genes in the brains of Skp2+/+ embryos; the increases rose higher with activation of expression of apoptotic E2F target genes in Skp2−/− embryos. The brain apoptosis was independent of p53 but coincident with proliferation. The highly activated expression of proliferative and apoptotic E2F target genes subsided with gradually reduced roles of Skp2 in preventing p27 protein accumulation in the brain in late gestation, allowing the embryos to reach full term with normally sized brains. These findings establish that Rb1 and Skp2 deletions are synthetic lethal and suggest how this lethal relationship might be circumvented, which could help design better therapies for pRb-deficient cancer.

Children who inherit one null allele of the retinoblastoma 1 gene (RB1) are RB1 ϩ/Ϫ in all cells and develop retinoblastoma with full penetrance, and the retinoblastoma cells are invariably RB1 Ϫ/Ϫ . These cases represent the classic "two-hit" RB1-deficient tumorigenesis. Mice that inherited one knock-out allele of Rb1 (the mouse homolog of RB1) are Rb1 ϩ/Ϫ in all cells and develop pituitary melanotroph tumors with full penetrance, and the tumor cells are invariably Rb1 Ϫ/Ϫ . Rb1 ϩ/Ϫ mice therefore model the two-hit RB1-deficient tumorigenesis. Biochemical studies have identified many repressing targets of pRb (encoded by RB1), with E2F1, E2F2, and E2F3 being the best studied (1). It is hypothesized that RB1-deficient tumorigenesis is caused by abnormal activation of pRb repressing targets following the inactivation of pRB. When Rb1 ϩ/Ϫ mice were used to test this hypothesis, the combined deletion of E2f1 (2) or E2f3 (3) inhibited pituitary tumorigenesis in Rb1 ϩ/Ϫ mice to various degrees.
Rb1 is an essential gene in mouse embryogenesis, because Rb1 Ϫ/Ϫ embryogenesis becomes lethal on embryonic day 14.5-15.5 (E14. 5-15.4) 4 (9 -11). Rb1 Ϫ/Ϫ embryogenesis provides another experimental system to test the functional significance of biochemically identified pRb repressing targets. In previous studies, when deletion of a pRb repressing target inhibited pituitary tumorigenesis in Rb1 ϩ/Ϫ mice, the same deletion extended survival of Rb1 Ϫ/Ϫ embryogenesis (12)(13)(14). A straightforward paradigm therefore emerged: if a biochemically identified pRb repressing target is functionally significant, its deletion would lessen the effects of Rb1 inactivation, manifested as inhibition of pituitary tumorigenesis to extend survival of Rb1 ϩ/Ϫ mice and correction of defects in Rb1 Ϫ/Ϫ embryogenesis to extend survival of Rb1 Ϫ/Ϫ embryos. In the current study, we determined how Skp2, whose deletion more effectively inhibited pituitary tumorigenesis in Rb1 ϩ/Ϫ mice than all other pRb repressing targets tested so far, fits into this paradigm.

Experimental Procedures
Mice-Nes-Cre mice were obtained from The Jackson Laboratory. Rb1 ϩ/Ϫ (9), Rb1 lox/lox (15), Skp2 Ϫ/Ϫ (16), and Trp53 lox/lox mice (17) have been described. Nes-Cre was genotyped using primers Cre F (5Ј-ATGCCCAAGAAGAAGAGGAAGGT-3Ј) and Cre R (5Ј-GAAATCAGTGCGTTCGAACG CTAGA-3Ј). Other genotyping details were previously described (18,19). The mice used in this study were on mixed C57BL/ 6Jϫ129SvϫFVB strain background. The mice were maintained under pathogen-free conditions in the Albert Einstein College of Medicine animal facility. All procedures were reviewed and approved by Einstein Animal Care Committee, conforming to accepted standards of humane animal care.
Embryos Gestation Determination and Analyses-Gestation and embryo age were determined by timed mating. Male and female mice were put in the same cage in late evening on day 1 and were separated in early morning on day 2. The embryo age was counted as E0.5 on day 2 if the female mice were found pregnant later. Pregnant females were intraperitoneally injected with BrdU (0.1 mg/g) 2 h before harvest. Embryos at various ages were harvested and were fixed in 10% formalin (Fisher Scientific, SF 100-4), embedded in paraffin wax, and sectioned. For embryos older than E15.5, decalcification was performed before routine processing to paraffin embedding. Embryo histology studies were conducted together with Albert Einstein Cancer Center mouse pathology core. Embryo brains used for RT-PCR and Western blots were dissected, snap frozen, and kept in Ϫ80°C freezer.
Apoptosis in the brain of Rb1 Ϫ/Ϫ embryos was prevented by additional deletion of Trp53 (23, 24). When we additionally

TABLE 2
Nes-Cre;Rb1 lox/lox ;Skp2 ؊/؊ embryos develop to full term Pregnant females were sacrificed to obtain the embryos on three different gestational days (Age column). All embryos were alive; total and expected (in parentheses) numbers for the two key genotypes are shown. These two genotypes also generated similar numbers of live newborns. Live E13. 5  deleted Trp53 in neuronal lineage together with Rb1 in Skp2 Ϫ/Ϫ ; Nes-Cre;Rb1 lox/lox ;Trp53 lox/lox embryos, the brain apoptosis remained, as detected by TUNEL staining (Fig. 4, B, panel b, and D, panel b) and H&E staining (Fig. 4C, panel b). These findings demonstrate that combined deletion of Rb1 and Skp2 can cell autonomously induce apoptosis in neuroepithelium of the developing brain and that the underlying mechanism for this synthetic lethal apoptosis is fundamentally different from that for the non-cell autonomously induced brain apoptosis in Rb1 Ϫ/Ϫ embryos.

Skp2-Rb1 Synthetic Lethal Neuronal Apoptosis Coexists with Further Enhanced
Proliferation-To investigate the mechanisms of Skp2-Rb1 synthetic lethal brain apoptosis, we determined how Skp2 Ϫ/Ϫ affected the effects of Rb1 deletion in the brain. Nes-Cre-mediated deletion of Rb1 in embryos afforded us brain tissues to quantitatively determine E2F target gene expression by RT-qPCR and Western blot. pRb represses the activating E2F subfamily E2F1, E2F2, and E2F3, whose target genes are mostly proliferative, with E2F1 being able to additionally activate expression of apoptotic gene PUMA via ARF-p53 and p73 directly as an E2F1 target gene. Following loss of Rb1, expression of typical proliferative E2F target genes is increased comparing brain tissues of Rb1 lox/lox and Nes-Cre;Rb1 lox/lox embryos (Fig. 5A). Consistent with the lack of brain apoptosis in Nes-Cre;Rb1 lox/lox embryos, expression of E2F1 apoptotic target genes was not increased (Fig. 5B). When we determined expression of proliferative E2F target genes in brains of Skp2 Ϫ/Ϫ ; Nes-Cre;Rb1 lox/lox embryos, we found that Skp2 Ϫ/Ϫ did not correct the increased expression but instead increased the expression further and activated expression of E2F1 apoptotic target genes (Fig. 5, A and B). The further increased expression of proliferative E2F target genes and activation of expression of apoptotic E2F1 target genes in the brains of Skp2 Ϫ/Ϫ ;Nes-Cre; Rb1 lox/lox embryos were more dramatic at protein levels (Fig.  5D). We next show that ectopic proliferation in the intermediate zone, as measured by BrdU labeling (Fig. 5, E and F) and Western blot for pHH3 (Fig. 5D), is increased following Rb1 deletion in brains of Skp2 ϩ/ϩ embryos and increased further in brains of Skp2 Ϫ/Ϫ embryos. Western blot for aCasp 3 more dramatically demonstrated Skp2-Rb1 synthetic lethal brain apoptosis (Fig. 5D). These brain tissue findings indicate the cause of

JOURNAL OF BIOLOGICAL CHEMISTRY 10205
Skp2-Rb1 synthetic lethal neuronal apoptosis to be the further activation of E2F1.
When we additionally deleted Trp53, expression of its target gene PUMA was significantly reduced, whereas expression of E2F1 target gene p73 remained (Fig. 5C). Because the synthetic lethal neuronal apoptosis in embryonic brains is independent of p53 (Fig. 4), these findings further identify E2F1 p53-independent apoptotic activity as the underlying mechanism. Finally, most clusters of TUNEL or aCasp3-positive apoptotic cells in Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox brains were doubly positive for PCNA, product of the proliferative E2F target gene (Fig. 5G), directly linking ectopic proliferation to the apoptosis.
Skp2-Rb1 Synthetic Lethal Neuronal Apoptosis Subsides in Late Gestation-When we determined the impact of Skp2-Rb1 synthetic lethal brain apoptosis on brain development in Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox embryos, it quickly became evident that Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox embryos had normally sized brains (Fig. 6A) without gross structural abnormalities (Fig. 6B) on E18.5. To investigate the molecular basis for this lack of structural impact by the massive apoptosis seen on E13.5, we measured the expression of proliferative and apoptotic E2F target genes in brains on E13.5, E15.5, and E17.5. The sets of PCNA, E2F1, cyclin A, cyclin E (proliferative E2F target genes), p73 (apoptotic E2F1 target gene), pHH3, and aCasp3 (mitosis and apoptosis markers) that were highly expressed in the brains of Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox on E13.5 became less highly expressed on E15.5 and reduced to WT levels on E17.5 (Fig.  6C). These findings provided a biochemical explanation for the development of the brain to normal size and structure when Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox embryogenesis progressed to late gestation.
To investigate how these biochemical changes took place, we determined mRNA expression of proliferative E2F target genes as WT embryogenesis progressed to late gestation. Fig. 6D shows that levels of PCNA, E2F1, cyclin A, and cyclin E mRNA quickly reduced in the WT brains from E13.5. These findings suggest that mitotic exit for neuronal differentiation begins in the last 5 gestational days.
Notably, expression of Skp2 mRNA also reduced significantly in this period (Fig. 6D), consistent with Skp2 being an E2F target gene (5,6). We next found that Skp2 protein levels in WT embryos also gradually reduced from E13.5 to E17.5 (Fig.  6E). To determine the functional impact of these reductions, we determine the protein levels of p27, because mediating p27 degradation is the best known function of Skp2. Unexpectedly, the gradual decrease in Skp2 protein levels did not induce gradual accumulation of p27 protein. Furthermore, Skp2 Ϫ/Ϫ induced accumulation of p27 in the brain, and combined deletion of Rb1 further increased p27 accumulation on E13.5. This two-step p27 accumulation was much reduced on E15.5 and became nonexistent on E17.5. These findings suggest that, with reduced expression, Skp2 roles in the degradation of p27 gradually became dispensable for preventing p27 protein accumulation. The reduced role of Skp2 in preventing p27 protein accumulation could be due to its reduced expression, activation of other ubiquitin ligases for p27, or, more likely, a combination of these. These findings suggest that naturally reduced expression of Skp2 and naturally increased Skp2-independent mechanisms to prevent p27 protein accumulation weakened the Skp2-Rb1 synthetic lethal relationship when embryogenesis progressed to late gestation.

Discussion
A key strategy to study pRb is to identify its repressing targets, defined as pRb binding proteins whose functions are repressed when bound by pRb. The best established pRb repressing targets are E2F1, E2F2, and E2F3, the activator members of the E2F family. pRb binds these activator E2Fs to inhibit their transactivating functions and to recruit repressive chromatin modifiers onto the promoters of their target genes to achieve long term repression. When a pRb repressing target is functionally significant, the effects of Rb1 deletion should be reduced when this pRb repressing target is inactivated. Indeed, pituitary tumorigenesis in Rb1 ϩ/Ϫ mice is inhibited in E2f1 Ϫ/Ϫ ; Rb1 ϩ/Ϫ mice (2) or E2f3 Ϫ/Ϫ ;Rb1 ϩ/Ϫ mice (3) and combining deletion of E2f1 (12), E2f2 (25), or E2f3 (11,13) corrected certain defects in Rb1 Ϫ/Ϫ embryos and placentas to improve Rb1 Ϫ/Ϫ embryogenesis. A weakness shared by these three pRb repressing targets is that their individual deletion only partially inhibited pituitary tumorigenesis in Rb1 ϩ/Ϫ mice. Deleting E2f1, E2f2, and E2F3 together is embryonic lethal (26); its  cule to allow biochemical studies to understand the underlying mechanisms.
We next used cultured MEFs to study the mechanisms of apoptosis by co-deletion of Skp2 and Rb1 (20). As expected, acute deletion of Rb1 in WT MEFs induced increased expression of E2F target genes. Unexpectedly, E2F target gene expression increased further when Rb1 was deleted in Skp2 Ϫ/Ϫ MEFs. Cyclin A binds E2F1 to repress its transactivating activity, and cyclin A is a E2F target gene; activated E2F1 following Rb1 deletion is therefore restrained by cyclin A. Skp2 Ϫ/Ϫ MEFs accumulate a significant amount of p27 protein, which competed against E2F1 for binding to cyclin A. The freed E2F1 is therefore "superactivated" to further increase expression of proliferation and apoptotic target genes. Remarkably, E2f1 deletion reduced oncogenic expansion of pituitary intermediate lobe in Skp2 ϩ/ϩ ; Pomc-Cre;Rb1 lox/lox mice but increased the thickness of the pituitary intermediate lobe in Skp2 Ϫ/Ϫ ;Pomc-Cre;Rb1 lox/lox mice. However, important questions remained. Skp2 Ϫ/Ϫ MEFs in culture accumulated p27 to higher degrees than most tissues in Skp2 Ϫ/Ϫ mice, and the role of p53 in apoptosis is unclear.
In the present study, we used embryogenesis to study the functional relationship between Skp2 Ϫ/Ϫ and Rb1 Ϫ/Ϫ . By mating Skp2 ϩ/Ϫ ;Rb1 ϩ/Ϫ mice, we have now established that Skp2 Ϫ/Ϫ and Rb1 Ϫ/Ϫ are synthetic lethal at the organismal level. We demonstrate that Skp2 Ϫ/Ϫ did not inhibit ectopic proliferation in Rb1 Ϫ/Ϫ neuroepithelium intermediate zone but induced apoptosis specifically in the neuroepithelium of Rb1 Ϫ/Ϫ embryos. These findings provide in vivo evidence for a synthetic lethal relationship that is currently unique among the growing list of pRb repressing targets. By generating Skp2 Ϫ/Ϫ ; Nes-Cre;Rb1 lox/lox embryos, we have further established the Skp2-Rb1 synthetic lethal relationship in vivo at a cell autonomous level with a biochemical understanding. In this cell autonomous setting, Rb1 deletion activated E2Fs to increase expression of proliferative genes but not apoptotic genes; codeletion of Skp2 and Rb1 not only further activated E2F proliferative target genes but activated expression of E2F1 apoptotic target genes. The role of a further activated E2F1 in this Skp2-Rb1 synthetic lethal apoptosis is dramatically demonstrated by the findings that (i) ectopic proliferation in the neuroepithelium is further increased, (ii) most apoptotic cells expressed the proliferative marker PCNA, and (iii) the apoptosis is independent of p53.
Finally, it is evident that most other tissues in E10.5 Skp2 Ϫ/Ϫ ; Rb1 Ϫ/Ϫ embryos do not contain significantly increased apoptosis, thereby documenting that the Skp2-Rb1 synthetic lethal relationship is context-dependent. Our study of brains of Skp2 Ϫ/Ϫ ;Nes-Cre;Rb1 lox/lox embryos in late gestation further demonstrated a number of mechanisms that could nullify Skp2-Rb1 synthetic lethal neuronal apoptosis, including lower expression of Skp2 and developing additional mechanisms of  Fig. 5D, for the indicated proteins in brains of the indicated genotypes were labeled as in Fig. 5D. Three gestational ages were studied. D, RT-qPCR gene expression analysis of WT brains of the indicated gestational ages. E, Western blot analyses of Skp2 and p27 levels in brains of the indicated embryos at the indicated gestational ages. Loading control is the same as in C. Protein molecular mass markers in kDa are on the right of the blots. p27 degradation to render Skp2 dispensable in preventing p27 accumulation.
Inactivation of pRb is a highly recurrent feature of cancer cells, and inhibiting pRb repressing targets is a logical and intensively pursued therapeutic strategy for pRb-deficient cancer. The identification of Skp2-Rb1 synthetic lethal relationship illuminated a highly attractive new strategy for pRb-deficient cancer. It is increasingly clear that cancer as an heterogeneous disease cannot be cured by a single treatment, no matter how effective it might be for certain cases. It is further evident that no matter how effective a treatment is at the beginning, cancer always evolves to resist it. The Skp2-Rb1 synthetic lethal apoptosis in embryogenesis show these two features, and our findings suggest the cancer characteristics that are best suited for the Skp2-Rb1 synthetic lethal treatment strategy (genetic inactivation of pRb with high expression of Skp2), how these cancers might develop resistance to this treatment (select for additional abilities to degrade p27), and how the resistant cancer might be treated (combining targeting the other p27 ubiquitin ligases).