Oncogene (2000) 19, 6065 ± 6073 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc E2F1 mediates ectopic proliferation and stage-speci®c p53-dependent apoptosis but not aberrant di€erentiation in the ocular lens of Rb de®cient fetuses

Yule Liu1 and Eldad Zacksenhaus*,1,2

1Department of Medicine, Toronto General Hospital Research Institute, University Health Network, University of Toronto, 67 College Street, M5G 2M1, Toronto, Ontario, Canada; 2Department of Medical Biophysics, Toronto General Hospital Research Institute, University Health Network, University of Toronto, 67 College Street, M5G 2M1, Toronto, Ontario, Canada

The retinoblastoma tumor suppressor, Rb, is a transcrip- di€erentiation by modulating the activity of certain tion cofactor that controls cell proliferation, survival and transcription factors. The major partners of Rb include di€erentiation. Mutant mouse embryos lacking Rb exhibit members of the E2F family of transcription factors, ectopic proliferation and apoptosis that are mediated in which regulate the expression of genes required for some tissues by E2F1, a major partner of Rb, and by the progression into S phase and DNA replication (Nevins p53 tumor suppressor. Whether E2F1 and p53 also et al., 1997). Rb binds preferentially to E2F1 ± 3, as well mediate the di€erentiation defects in Rb mutant embryos as E2F4 (Nevins et al., 1997). Of all the E2Fs, E2F1 is, however, not clear. Here we show that partially rescued alone can induce not only S phase entry but also mgRb:Rb7/7 mutant fetuses exhibit ectopic lens apoptosis (DeGregori et al., 1997; Johnson et al., 1993; epithelial cell proliferation, apoptosis and severe cataract. Phillips et al., 1999; Wu and Levine, 1994). Induction The abnormal cell proliferation and apoptosis were of apoptosis by E2F1 is mediated at least in part by signi®cantly suppressed in the lens of compound mutant transcriptional activation of p19ARF, which inactivates fetuses lacking both Rb and E2F1 at embryonic day (E) MDM2, thereby stabilizing p53 (Bates et al., 1998; E15.5. Interestingly however, at E18.5, only ectopic Pomerantz et al., 1998). The combined inactivation of proliferation, not apoptosis, was dramatically reduced in Rb and p53, which often occurs in human cancer, mgRb:Rb7/7:E2F17/7 lenses. In contrast, p53 did results in deregulation of the cell cycle, suppression of not exert such a stage-speci®c e€ect and apoptosis was apoptosis and acceleration of tumorigenicity (Williams invariably suppressed in mgRb:Rb7/7:p537/7 com- et al., 1994). Thus, detailed analysis of the conse- posite mutant lenses throughout embryogenesis. Using quences of disrupting the Rb-E2F1-p53 pathway is RT ± PCR and in situ hybridization analyses, we identi®ed important for understanding the progression of cancer. a subset of lens speci®c genes, most notably the late Mice lacking one copy of Rb, Rb+/7, are viable di€erentiation marker ®lensin, which were not properly but predisposed to pituitary adenocarcinomas and induced during lens development in mgRb:Rb7/7 other malignancies (Clarke et al., 1992; Jacks et al., fetuses. Remarkably, despite the inhibition of cell 1992; Lee et al., 1992). Rb7/7 mutant embryos die at proliferation and apoptosis, the degeneration of lens ®bers mid-gestation and develop speci®c abnormalities in and aberrant expression of ®lensin were only marginally neurogenesis, erythropoiesis and lens development. The corrected in mgRb:Rb7/7:E2F17/7 fetuses at E15.5 spectrum of defects in Rb7/7 mutant embryos is but not at all at E18.5 or in mgRb:Rb7/7:p537/7 consistent with the speci®c expression of Rb during mutant fetuses. Thus, inactivation of E2F1 reduces embryogenesis in the nervous system, lens, liver and ectopic cell proliferation and stage-speci®c p53-dependent skeletal muscles (Jiang et al., 1997). Analysis of the role apoptosis but does not rescue the di€erentiation defects of Rb in older embryos was made possible by the use associated with loss of Rb during lens development. of a mouse Rb minigene, mgRb, which is expressed Oncogene (2000) 19, 6065 ± 6073. exclusively in the nervous system and can rescue the neurogenic defects of Rb7/7 embryos (Zacksenhaus Keywords: Rb; p53; E2F1; lens; crystallins; ®lensin et al., 1996). The mgRb:Rb7/7 mutant mice survive to birth and exhibit severe defects in myogenesis (Jiang et al., 2000a; Zacksenhaus et al., 1996) and lens Introduction development (this study). Developmental abnormalities in Rb mutant mice are The tumor suppressor, Rb, was initially identi®ed as characterized by ectopic DNA synthesis, apoptosis and the retinoblastoma-susceptibility gene (Zacksenhaus et aberrant di€erentiation. Genetic analysis revealed that al., 1993). It was later found that Rb is inactivated in the ectopic cell proliferation and cell death are mediated most human cancers by genetic alterations either in the to a large extent by E2F1 and p53 in the central nervous gene itself or in upstream factors that control the system and lens but not in the peripheral nervous phosphorylation and activity of the protein (Weinberg, system (Macleod et al., 1996; Morgenbesser et al., 1994; 1995). Rb regulates cell cycle exit and terminal Tsai et al., 1998) and skeletal muscles (Jiang et al., 2000a). The di€erentiation defects include morphologi- cal/cellular anomalies and failure to express some *Correspondence: E Zacksenhaus di€erentiation related genes. For example, the neuronal Received 31 July 2000; revised 28 September 2000; accepted 4 growth factor receptors and the muscle creatin kinase October 2000 are not properly induced during peripheral neurogenesis Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6066 and myogenesis, respectively, in Rb de®cient embryos lens ®bers. To determine the temporal e€ects of Rb (Jiang et al., 2000a; Lee et al., 1994). It is unclear loss on lens development, E13.5 ± E18.5 mgRb:Rb7/7 whether expression of these di€erentiation factors is embryos and control littermates were subject to compromised because Rb is directly required for their histology examination. Control lens from expression or because the excessive cell death and mgRb:Rb+/7 fetuses developed and di€erentiated unscheduled cell proliferation that occur in these tissues normally to form epithelial, transitional and ®ber in the absence of Rb halt the di€erentiation programs. compartments (Figure 1B,a,e). In contrast, the To address these issues, we analysed lens develop- mgRb:Rb7/7 mutant lens failed to polarize and ment in mutant mouse embryos lacking both Rb and elongate properly and contained many nuclei, pyknotic E2F1 or p53. We chose the lens because E2F1 and p53 bodies and vacuoles throughout the lens vesicle, were reported to eciently inhibit cell proliferation and resulting in severe cataract (Figure 1B,b,f). apoptosis in this tissue. In addition, the use of partially rescued Rb mutant embryos allowed us to follow the Impaired lens development in mgRb:Rb7/7:p537/7 consequences of Rb loss during lens development and mgRb:Rb7/7:E2F17/7 compound mutant fetuses throughout fetal development in utero. We found that inactivation of E2F1 leads to stage speci®c suppression Previous analysis has shown that p53 mediates of p53-dependent apoptosis during lens development in apoptosis while E2F1 mediates both apoptosis and Rb mutant mice. Furthermore, we show that Rb ectopic DNA synthesis in the lens of E13.5 Rb mutant mutant lenses fail to induce the expression of certain embryos (Morgenbesser et al., 1994; Tsai et al., 1998). structural proteins and develop severe cataract; these To analyse the temporal e€ects of E2F1 and p53 on defects are only slightly reversed in compound mutant lens development, we generated composite mice lacking both Rb and E2F1, in which ectopic cell mgRb:Rb7/7:E2F17/7 and mgRb:Rb7/7: proliferation and apoptosis are greatly suppressed. p537/7 mutant fetuses (see Materials and methods). These ®ndings indicate that E2F1 is a major partner Sections through lens were analysed by PCNA of Rb but additional factors, required for proper cell immunohistochemistry to detect proliferating cells, cycle exit and terminal di€erentiation, are deregulated TUNEL assay, to label apoptotic nuclei and Hematox- as a consequence of Rb loss. ylin and Eosin (H&E) staining, to reveal histological alterations. The mgRb:Rb7/7 single mutant lenses contain many PCNA- and TUNEL-positive cells (Figures 2b,f,j,n and 3b,f,j,n). The number of PCNA- Results and TUNEL-positive cells was markedly reduced in the lens of mgRb:Rb7/7:E2F17/7 embryos at E15.5, as Lens generation in mgRb:Rb7/7 fetuses previously noted for E13.5 Rb7/7:E2F17/7 em- To study the role of Rb at di€erent stages of lens bryos (Figures 2d,h and 3d,h). However, at E18.5, only development in utero, we used partially rescued Rb cell proliferation but not apoptosis was signi®cantly mutant fetuses, mgRb:Rb7/7, in which the neuro- inhibited in mgRb:Rb7/7:E2F17/7 lenses (Figures genic defect in Rb7/7 embryos is speci®cally 2p and 3p). Loss of p53 had no substantial e€ect on corrected by an Rb mini-gene (Zacksenhaus et al., cell proliferation (Figure 2c,g,k,o) but invariably 1996). The mouse Rb promoter used to create the Rb suppressed apoptosis in lens ®bers throughout embry- mini-gene was fused to the E. Coli b-galactosidase ogenesis in mgRb:Rb7/7:p537/7 compound mutant (lacZ) reporter gene and several independent fetuses (Figure 3c,g,k,o). To quantify the level of RbP(L).lacZ transgenic mice were generated (Jiang et apoptosis in the various mutants, the number of al., 2000b). To determine the expression of the TUNEL-positive nuclei from several sections across transgene in the lens, we analysed sections of the lens of three independent compound-mutant and RbP(L).lacZ transgenic mice through the developing control littermates were counted and normalized for eye by X-gal staining. As shown in Figure 1A, X-gal the lens area. This analysis revealed that at E15.5, loss staining was evident in the brain epithelium, the optic of E2F1 suppressed apoptosis to 12%, slightly better nerve and ganglion layer of the retina but, unlike than loss of p53 (21%), relative to Rb single mutant endogenous Rb (Jiang et al., 1997), was completely lenses (Figure 4). However, by E18.5, the level of absent in the developing lens. The absence of apoptosis in mgRb:Rb7/7:E2F17/7 lenses was only detectable expression of the lacZ reporter gene in the moderately reduced (61% relative to mgRb:Rb7/7 lens was evident in ®ve independent RbP(L).lacZ single mutants). In contrast, the level of apoptosis in transgenic lines. Thus, the mouse Rb promoter used mgRb:Rb7/7:p537/7 double mutant lenses re- to create the mgRb mini-gene, does not direct mained low (15%) relative to single mutants. To detectable levels of linked genes to the developing lens, further establish these results, Hoechst staining was and thus, the mgRb:Rb7/7 fetuses can be viewed as used to identify condensed apoptotic nuclei (Figure null for Rb throughout lens development in utero. 3q ± t). Consistent with the TUNEL analysis, little In mammals, the lens develops from the surface di€erence in apoptosis was observed between ectoderm that invaginates to form the lens vesicle mgRb:Rb7/7:E2F17/7 and mgRb:Rb7/7 lenses within the lens cup (Francis et al., 1999; Oliver and (Figure 3t versus r), whereas mgRb:Rb7/7:p537/7 Gruss, 1997). The epithelial cells lining the posterior fetuses exhibited marked reduction in Hoechst-positive wall of the lens become post-mitotic and elongate to nuclei at E18.5 (Figure 3s). Thus, in Rb de®cient lenses form lens ®bers that ®ll the lumen. In the transitional E2F1 mediates both cell proliferation and apoptosis at zone, at the boundaries between the epithelial and ®ber E15.5 but only proliferation at E18.5. At the latter compartments, cells continue to exit the cell cycle, stage, the loss of Rb induces an E2F1-independent but migrate centrally and di€erentiate to form secondary p53-dependent apoptosis.

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6067

Figure 2 E€ects of E2F1 and p53 on ectopic cell proliferation in Rb-de®cient lens. PCNA immunohistochemical analysis of control, mgRb:Rb7/7, mgRb:Rb7/7:p537/7 and mgRb:Rb7/7:E2F17/7 mutant lenses at E15.5 (a±h) and E18.5 (i±p). Magni®cation: a±d and i±l, 1006; e±h and m±p, 4006. Arrows indicate some PCNA-positive nuclei. Loss of E2F1, but not p53, eciently reduces the number of PCNA- positive cells in mgRb:Rb7/7:E2F17/7 compound mutant lenses (d,h,l,p, versus b,f,j,n)

Figure 1 Lens degeneration in partially rescued Rb mutant fetuses. (A) The Rb promoter used to generate the partially rescued mgRb:Rb7/7 fetuses is not expressed in the lens. Transgenic mRbP(L).lacZ embryos were stained with X-gal, paran embedded and sectioned through the lens. X-gal staining (blue) is detected in the brain, optic nerve and ganglion layer of retina but not in the lens (a,b). (B) Lens development in control, mgRb:Rb7/7, mgRb:Rb7/7:p537/7 and mgRb:Rb7/7:E2F17/7 mutant embryos at E15.5 (a±d) and E18.5 (e±h). Control lenses (a,e) contain three distinct zones: the epithelial zone (E) at the anterior compartment of the lens; the transitional zone (T) where cells undergo di€erentiation and the ®ber zone (B) where cells elongate to form the lens ®bers. In Rb mutant fetuses (b±h), elongation of the lens ®bers is delayed and the lens contain many vacuoles (b±d), the transitional zone is not well de®ned and the ®ber region is occupied by proliferating and apoptotic cells. The micrographs in b±dshow examples of lenses with extreme number and size of vacuoles. Other mutants contained smaller vacuoles. (Magni®cation, 1006)

Despite the signi®cant suppression of lens ®ber proliferation and apoptosis, the mgRb:Rb7/7: E2F17/7 and mgRb:Rb7/7:p537/7 lenses were still grossly deformed and contained many vacuoles similar to Rb single mutant littermates (Figure 1Bc,g and d,h). The distribution and size of vacuoles varied in di€erent mutant fetuses, with the same genotype, but the extensive cellularity and lack of polarized lens Figure 3 E€ects of E2F1 and p53 on apoptosis in Rb-de®cient ®bers were evident in all mutant lenses. Moreover, lens. (a±p) TUNEL analysis on control, mgRb:Rb7/7, examination of multiple sections from several inde- mgRb:Rb7/7:p537/7 and mgRb:Rb7/7:E2F17/7 mutant lenses at E15.5 (a±h) and E18.5 (i±p). Magni®cation: a±d and pendent mutants revealed no signi®cant di€erences i±l, 1006; e±h and m±p, 4006. Arrows indicate some TUNEL- between the single and compound mutant lenses. positive nuclei. Loss of p53 inhibits apoptosis throughout Thus, neither the suppression of apoptosis in the embryonic lens development in mgRb:Rb7/7:p537/7 mutants absence of p53 nor the suppression of both ectopic (c,g,k,o). In contrast, inactivation of E2F1 reduces apoptosis at proliferation and apoptosis in the absence E2F1 are E15.5 (d,h) but not E18.5 (l,p). (q±t) Hoechst staining of lens from E18.5 control, mgRb:Rb7/7, mgRb:Rb7/7:p537/7 and sucient to prevent morphological defects in Rb mgRb:Rb7/7:E2F17/7 mutant embryos (4006). Arrows de®cient lenses. indicate some Hoechst-positive, condensed nuclei

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6068

Figure 4 Quantitative e€ects of E2F1 and p53 on apoptosis in Rb-de®cient lenses. The number of apoptotic nuclei in the indicated lenses was scored in multiple sections from several independent embryos at E15.5 and E18.5 lenses and normalized for the lens area. The relative amount of apoptosis in control and compound mutants was plotted against the mgRb:Rb7/7 mutant lens, which was assigned a value of 1

terminally di€erentiated lens ®bers (Masaki and Rb is required for expression of a subset of lens genes Quinlan, 1997; Merdes et al., 1991), was detected in Di€erentiation of lens epithelial cells is associated with wildtype but not in Rb mutant lenses at E15.5 (Figure expression of lens speci®c crystallins and intermediate 5A, arrows). ®laments (e.g. ®lensin), which assemble into highly In situ hybridization analysis was then performed ordered transparent lens (Francis et al., 1999). In with antisense riboprobes speci®c to crystallins gA, gB mammals, there are three classes of crystallins, a, b and and ®lensin (Figure 5B). In accord with the RT ± PCR g, each comprising several antigenically related proteins results, expression of crystallin gB (Figure 5Bc,d) and that are di€erentially regulated during lens develop- ®lensin (Figure 5Be,f) was signi®cantly reduced in ment. Previous analysis of Rb de®cient lenses with pan E15.5 Rb mutant embryos. Expression of gA was also antibody to crystallins showed a decrease in g-crystal- reduced but not to the same extent as gB and ®lensin lins in one study (Morgenbesser et al., 1994), normal (Figure 5Ba,b). Thus, Rb is required for the proper expression in another study (Fromm and Overbeek, expression of a subset of lens speci®c genes during 1996), and abnormal ratio of soluble : insoluble forms embryogenesis. of g-crystallin in a third study (McCa€rey et al., 1999). Since the various g-crystallin genes are independently Impaired expression of filensin in mgRb:Rb7/7:p537/7 regulated during embryogenesis (Murer-Orlando et al., and mgRb:Rb7/7:E2F17/7 compound mutant fetuses 1987), we asked whether some isoforms might be speci®cally a€ected in the absence of Rb. To We next asked whether the suppression of apoptosis and investigate this possibility, we used RT ± PCR analysis ectopic cell proliferation in mgRb:Rb7/7:E2F17/7 to semi-quantify the levels of individual crystallin genes and mgRb:Rb7/7:p537/7 lenses could rescue the in the lens of Rb mutant and wildtype littermates. aberrant expression of ®lensin in Rb mutant fetuses. As Primers were carefully selected so that cross-hybridiza- shown in Figure 6, the level of ®lensin was greatly tion among the di€erent crystallin genes within a reduced in mgRb:Rb7/7:p537/7 lenses to the same family was minimized (see Materials and methods). extent as in mgRb:Rb7/7 fetuses throughout embry- Among the various crystallins, we found that expres- ogenesis (Figure 6c,g versus b,f). By E18.5, expression of sion of crystallin gB was reproducibly reduced in ®lensin in mgRb:Rb7/7:E2F17/7 lenses was also as mgRb:Rb7/7 lenses compared with wildtype litter- low as in mgRb:Rb7/7 single mutant fetuses (Figure mates at E13.5 and E15.5 (Figure 5A, arrows). In 6h). However, at E15.5, a moderate but reproducible contrast, expression of crystallins aA, bA3/A1, gA, gC, increase in expression of ®lensin was observed in gDandgE/F appeared less a€ected. The lens speci®c mgRb:Rb7/7:E2F17/7 lenses relative to ®lament gene, ®lensin, which is only expressed in mgRb:Rb7/7 single mutants (Figure 6d). This increase

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6069

Figure 6 E€ects of E2F1 and p53 on expression of ®lensin in Rb-de®cient lens. Dark-®eld images of 35S-labeled ®lensin antisense ribo-probe, hybridized with lens sections from control, mgRb:Rb7/7, mgRb:Rb7/7:p537/7 and mgRb:Rb7/7: E2F17/7 mutant embryos at E15.5 (a±d) and E18.5 (e±h). Note, the moderate increase in ®lensin signal in mgRb:Rb7/7: E2F17/7 relative to mgRb:Rb7/7 lenses at E15.5 (d versus b) Figure 5 Identi®cation of lens-speci®c genes that are aberrantly expressed during lens development in Rb-de®cient fetuses. (A) RT ± PCR analysis of lens speci®c genes in mgRb:Rb7/7 of transgenic mice at E13.5. However, at E15.5 and in fetuses. At E13.5, expression of gB is clearly reduced in mgRb:Rb7/7 embryos compared with control mice (white post-natal E7:E2F17/7 mice, the absence of E2F1 arrowheads). At E15.5, ®lensin transcripts can be detected in reduces apoptosis to approximately 50% relative to the control but not in mutant lens (white arrowheads). (B) Dark- E7:E2F1+/+ littermates (McCa€rey et al., 1999). ®eld images of 35S-labeled crystallin gA, gB and ®lensin antisense Herein, we found that E15.5 mgRb:Rb7/7: ribo-probes, hybridized with lens sections from E15.5 control and mgRb:Rb7/7 embryos E2F17/7 embryos exhibit very little apoptosis, as previously reported for E13.5 Rb7/7:E2F17/7 embryos, but over 60% apoptosis relative to mgRb:Rb7/7 littermates at E18.5, similar to the in ®lensin gene expression was also evident in E15.5 results found in E7:E2F17/7 neonates. Hence, in the Rb7/7:E2F17/7 mutant embryos (without the mgRb absence of Rb, apoptosis in the lens is only partially mini-gene; data not shown). Densitometery analysis mediated by E2F1 at E18.5 despite the fact that revealed that the ratio of ®lensin gene expression in proliferation in mgRb:Rb7/7:E2F17/7 lenses is wild-type control, mgRb:Rb7/7 and mgRb:Rb7/7: greatly reduced at this stage (Figure 2). In contrast, E2F17/7 lenses was 54 : 1 : 5. Thus, while the level of p53 mediates apoptosis in Rb mutant lenses both at ®lensin increases ®vefold in mgRb:Rb7/7:E2F17/7 E15.5 and E18.5 (Figures 3 and 4). These ®ndings compared with mgRb:Rb7/7 lenses, it is still approxi- suggest that an E2F1-independent pathway may induce mately 11-fold lower than in wildtype embryos. p53-mediated apoptosis in response to Rb loss in the Together, the results indicate that the absence of E2F1 E18.5 lens (Figure 7). This pathway may depend on eciently suppresses abnormal cell proliferation and E2F3, which is also expressed in the lens ®ber apoptosis but only marginally reverses the molecular and compartment (see below). Although, of all E2Fs, only cellular di€erentiation defects in Rb-de®cient lenses. E2F1 can induce apoptosis in ®broblasts (DeGregori et al., 1997), other members of the E2F family may induce apoptosis in other cell types, such as lens epithelium. Alternatively, loss of Rb may signal to p53 Discussion via a novel pathway, yet to be elucidated, which is independent of E2Fs. Stage specific suppression of p53-dependent apoptosis by In the PNS, loss of E2F1 reduces ectopic prolifera- E2F1 in Rb mutant lenses tion to about 50% but only slightly suppresses Previous analysis of Rb7/7:E2F17/7 embryos apoptosis in Rb7/7:E2F17/7 double mutants (Tsai revealed that E2F1 mediates both cell proliferation et al., 1998). Thus, both in the lens at E18.5 and in the and apoptosis in the lens (Tsai et al., 1998). In PNS at E13.5, E2F1 exhibits di€erential e€ects on cell contrast, targeted expression of the papilloma virus proliferation and survival. However, whereas in the E7 oncoprotein, which binds all members of the Rb lens, apoptosis is p53-dependent (Figures 3 and 4), in family, induces E2F1-independent apoptosis in the lens sensory ganglia apoptosis is independent of p53

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6070 Notably, Rb directly binds PAX6 (Cvekl et al., 1999), a paired domain/homeodomain transcription factor that plays a central role in lens development (Cvekl and Piatigorsky, 1996). In future experiments, it would be of interest to determine whether Rb can form DNA- bound complexes on gB and ®lensin promoter elements and potentiate transcriptional activation of these lens genes by PAX6, Prox1 or other lens-speci®c transcrip- tion factors (Kondoh, 1999; Kumar and Moses, 1997).

Figure 7 E2F1-dependent and -independent pathways down- Marginal reversal of aberrant differentiation by E2F1 in stream of Rb during lens development. Inactivation of Rb results Rb mutant lenses in ectopic DNA proliferation, apoptosis and di€erentiation defects in the ocular lens. At earlier stages, E13.5 ± E15.5, E2F1 Rb7/7:E2F17/7 embryos have been shown to controls both cell proliferation and p53-dependent apoptosis in survive longer than Rb7/7 embryos and exhibit Rb mutant lenses. In older fetuses, E18.5, E2F1 controls cell reduced cell proliferation and apoptosis in the CNS proliferation but not p53-dependent apoptosis. The mechanism by and lens (Tsai et al., 1998). The Rb7/7:E2F17/7 which Rb loss induces p53-dependent apoptosis at E18.5 is not known. At E15.5, when apoptosis and ectopic proliferation are phenotype has suggested that deregulation of E2F1 is greatly reduced in mgRb:Rb7/7:E2F17/7 double mutant the main defect in the CNS and lens in Rb mutant lenses, there are still major di€erentiation defects at both the embryos and that Rb7/7:E2F17/7 fetuses die cellular and molecular levels. Thus, E2F1 reverses some but not several days later because of aberrant development of all the defects associated with Rb loss during lens development. other tissues, such as lung and skeletal muscles, where The abnormal lens di€erentiation in embryos lacking Rb and E2F1 may be attributed to residual cell proliferation and cell E2F1 plays a lesser role (Tsai et al., 1998). Here, we death observed in mgRb:Rb7/7:E2F17/7 mutant lenses, show that although loss of E2F1 signi®cantly sup- perhaps due to deregulation of E2F3, which is also expressed in presses ectopic cell proliferation and apoptosis in the lens ®bers. Alternatively, Rb may be directly required for terminal lens (Figures 2 ± 5), it does not reverse the aberrant lens di€erentiation and expression of certain lens speci®c structural genes through its interaction with PAX6 or other di€erentiation development associated with inactivation of Rb (Figure factors 1B). In addition, at the molecular level, loss of E2F1 only marginally a€ects the impaired expression of ®lensin (Figure 6). Thus, in the mouse, E2F1 is a major (Macleod et al., 1996). Apoptosis during skeletal but not the only partner of Rb even during lens muscle di€erentiation in Rb mutant mice is also development. independent of E2F1 and p53 (Jiang et al., 2000a). In the fruit ¯y, an Rb homolog, RbF, binds two Interestingly, embryos lacking the Cdk inhibitors p27 E2Fs: dE2F1 that acts as an activator and dE2F2 that and p57, and presumably contain hyper-phosphory- functions as a repressor (Du et al., 1996; Dynlacht et lated inactive Rb, develop lens defects similar to Rb al., 1994; Sawado et al., 1998). A dE2F1 allele knockout mice but ectopic proliferation is not harboring a deletion of the c-terminal Rb-binding/ accompanied by increased apoptosis (Zhang et al., transactivation domain can rescue all the develop- 1998). These and other studies (e.g. Agah et al., 1997) mental defects in RbF mutant ¯ies (Du, 2000). Thus, in indicate that multiple pathways operate downstream of Drosophila, dE2F proteins appear to be the sole targets Rb and that a linear Rb-E2F1-p53 pathway operates of RbF. The discrepancy between the mouse and ¯y only in some but not all tissues. Given that Rb is so may be attributed to functional redundancy among often inactivated in human cancer and that induction multiple Rb and E2F factors in mammals. In the of apoptosis has a protective e€ect against cancer, the mouse, there are three Rb related proteins, Rb, p107 identi®cation of the E2F1- and p53-independent and p130, and at least ®ve E2F-like factors, E2F1 ± 5. apoptotic pathways downstream of Rb in these tissues Rb preferentially interacts with E2F1 ± 3 as well as may identify novel targets for cancer therapy in p53- E2F4, whereas E2F4 and E2F5 form complexes with negative tumors. p107 and p130 (Dyson, 1998). In the lens, all E2Fs are expressed in the anterior epithelial cells but only E2F1, E2F3 and E2F5 are expressed in the ®ber cell Effect of Rb on lens specific gene expression compartment (Rampalli et al., 1998), where cells are Using RT ± PCR and in situ hybridization, we unable to exit the cell cycle and die by apoptosis in Rb identi®ed several lens speci®c genes, crystallin gBand mutant lenses (Figure 3). Our results show that the the ®lament protein ®lensin, that are not properly inhibition of cell proliferation and cell death in the induced in Rb mutant embryos during lens develop- mgRb:Rb7/7:E2F17/7 compound mutant mice was ment (Figure 5). The expression of crystallins aA, bA3/ greatly reduced but not completely inhibited (Figures A1, gA, gB, gC and gE/F is less a€ected. This 3 ± 5). It is thus possible that deregulation of E2F3 in abnormal pattern of gene expression in the Rb mutant the Rb mutant lens ®ber induces low level of cell lens is similar to Prox1 knockout embryos (Wigle et al., proliferation and apoptosis. Perhaps it is only when 1999) but not to Sox1 knockout mice, in which all g both E2F1 and E2F3 are inactivated, that Rb7/7 crystallins are downregulated (Wigle et al., 1999). We ®ber cells can completely exit the cell cycle and previously showed that Rb is required for expression of di€erentiate. It will be of interest in this context to MCK and MRF4 but not other muscle speci®c determine the consequences of combined inactivation markers in vivo (Zacksenhaus et al., 1996). Accord- of E2F1 and E2F3 (Humbert et al., 2000), i.e. ingly, Rb was found to potentiate the MCK promoter mgRb:Rb7/7:E2F17/7:E2F37/7 triple knockout in the presence of MyoD in vitro (Novitch et al., 1996). fetuses, on lens development and terminal di€erentia-

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6071 tion. Unless E2F2, E2F4 or E2F5 are induced when Genetic crosses of knock-out mice and PCR conditions E2F1 and E2F3 are both mutated, one would expect that if deregulation of E2Fs undermines di€erentiation The mgRb:Rb+/7 mice and p537/7 and E2F17/7 in Rb mutant mice, then lens development in knock-out mice were described previously (Field et al., mgRb:Rb7/7:E2F17/7:E2F37/7 triple knockout 1996; Jacks et al., 1992, 1994; Zacksenhaus et al., 1996). Compound mutant embryos were generated by crossing fetuses should be dramatically improved. double heterozygotes, mgRb:Rb+/7:E2F1+/7 or Alternatively, mammalian Rb proteins might have mgRb:Rb+/7:p53+/7. Additional embryos were derived evolved to perform other functions in addition to by crossing mgRb:Rb+/7:E2F17/7 males with regulating E2F-responsive genes. Indeed, substantial mgRb:Rb+/7:E2F1+/7 females or mgRb:Rb+/7: evidence supports a direct role for Rb in terminal p537/7 males with mgRb:Rb+/7:p53+/7 females. For di€erentiation. First, Rb-de®cient embryos fail to timed pregnancy, the morning of vaginal plug observation express a subset of di€erentiation markers such as was considered as E0.5. All mutant embryos were genotyped NGF receptors during neurogenesis (Lee et al., 1994), in duplicates using DNA extracted separately from limb and MCK during myogenesis (Zacksenhaus et al., 1996; tail biopsies. PCR primers for mgRb, Rb and E2F1 were as Jiang et al., 2000a) and crystallin gB and ®lensin during described (Field et al., 1996; Jacks et al., 1992; Zacksenhaus et al., 1996). The primers for p53 were p5371(5'-GTA TCT lens development (Figures 5 and 6). The speci®city of GGA AGA CAG GCA GAC) and p5367(5'-GAT GGT the defects is more consistent with a speci®c regulation GGT ATA CTC AGA GCC) for wildtype allele and neo-2 of a subset of genes by Rb rather than a global, non- (5'-TCC TCG TGC TTT ACG GTA TCG) and p5367 for speci®c e€ect on di€erentiation. Second, Rb was shown mutant allele. Several sections from di€erent areas of the lens to potentiate transcriptional activation by di€erentia- of three independent compound and single mutants were tion factors such as myogenic and adipogenic factors in analysed for each assay. RT ± PCR analysis was performed vitro (reviewed in Sellers and Kaelin (1996)). Rb forms on eyes from 2 ± 3 additional mutant and control littermates. a pentameric complex that includes tal-1, which down- modulates c-kit expression and presumably other Plasmids and probes erythroid genes during erythropoiesis (Vitelli et al., gA was ampli®ed by RT ± PCR from total RNA of eyes using 2000). Rb is also required for interferon-gamma- forward primer (fp-1) CTGCTGGATGCTCTATGAGC and induction of the MHC class II gene (Zhu et al., reverse primer CTGTAACAAGCAAAAGGAGGC. The 1999). Third, ectopic expression of Rb in Rb-de®cient PCR product was puri®ed from agarose gel using Qiagen cells can induce di€erentiation in vitro (Sellers et al., kit and used as a template for PCR ampli®cation with fp-1 1998). Furthermore, some low-penetrant Rb mutant and reverse primer-CGGAATTCGCCCCAGTCGTGG- alleles that are unable to bind E2F1, can still potentiate TAGCGCC. The PCR product was then digested with transcription by myogenic factors and induce terminal HinfI(blunt-ended) and EcoRI, and cloned into the SmaI- di€erentiation in vitro (Sellers et al., 1998). Taken EcoRI site of pBluescript(pBSK). To clone gB into pBKS, the together, these observations and the failure of E2F1 to RT ± PCR was performed using fp-1 and 5'-CGGAATT- rescue the di€erentiation defects in lens development CAACCTTGGCATTTGCAGCC. The PCR product was digested with HinfI(blunt-ended) and EcoRI, and cloned into (Figures 1 and 6) implicate Rb in terminal di€erentia- the SmaI-EcoRI site of pBKS. Filensin was ampli®ed by tion, independently of its control of E2F-responsive RT ± PCR using primers: 5'-AATCTGCAGCGCGCCCGGG genes. and 3'-TCCACCTGCAGTGCCTTT-GG. The PCR product The possible pathways downstream of Rb during was digested by SmaI and BamHI, and cloned into the SmaI- lens development are summarized in Figure 7. E2F1 BamHI site of pBKS. All the plasmid constructs were veri®ed clearly mediates cell proliferation and apoptosis by DNA sequencing. during early stages of lens development. A p53- dependent but E2F1-independent pathway, yet to be RT ± PCR analysis de®ned, mediates apoptosis at E18.5 in the mgRb:Rb7/7 lens. Inactivation of E2F1 only Total RNA was isolated from whole eyes using Trizol reagent (Gibco-BRL). One mg of RNA was reversed transcribed into partially reverses the di€erentiation defects in E15.5 single-stranded cDNA, and subsequently used as a template Rb mutant lenses. It is formally possible that the for PCR ampli®cation of all crystallins and ®lensin sequences. persistent di€erentiation defects in mgRb:Rb7/7 The following primer pairs were used: :E2F17/7 lenses re¯ects the deregulation of E2F3 aA; GACTGTTCGACCAGTTCTTCGG and GAAGGT- or related factors that indirectly inhibit di€erentiation CAGCATGCCATCAGC. bA3/A1; TTATGAACACACCA- by interfering with normal cell cycle exit. Alterna- GCTTCTGTG and TTAGCAAGATGTCATGCTTGAGG. tively, in mammals, Rb may be involved directly in Filensin; AATCTGCAGCGCGCCCGGG and TCCACCT- terminal di€erentiation in conjunction with tissue GCAGTGCCTTT-GG. The 5' primer for the ampli®cation speci®c di€erentiation factors such as PAX6 (Figure of all the g-crystallins was fp-1. The 3' primers were; gA, 7). CTGTAACAAGCAAAAGGAGGC. gB, CAACCTTGGC- ATTTGCAGCC. gC, TTGCAGCGAGCGCACCTCAC. gD, TTCCGTGAACTCTATCACTTGGC. gE/F, CGTGG- AAGGAGTGGAAGTCAC. Materials and methods In situ hybridization Analysis of RbP(L).lacZ transgenic mice In situ hybridization for crystallin and ®lensin genes was The pRbP(L).lacZ transgenic mice were maintained and performed as described (Jiang et al., 1997). The pBSK-gA analysed as previously described (Jiang et al., 2000b). After and pBKS-gB plasmids were linerized by digestion with whole-mount X-gal staining, embryos were dehydrated, BamHI and used as a template for the synthesis of anti-sense embedded, sectioned through the lens at 20 m and counter- riboprobes with T7 and T3 RNA polymerases, respectively. stained with eosin. pBKS-Filensin was linerized by EcoRI and the anti-sense

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6072 riboprobe was made by T7 RNA polymerase. All probes single mutant fetuses was considered as one, and the levels of were used at 26105 c.p.m./ml. The slides were coated with apoptosis in other lenses were plotted against this value. For NBT-2 emulsion (Kodak) and exposed for 1 week. Hoechst staining, sections were deparanated, hydrated and immersed in Hoechst 33258 (Sigma, 0.5 mg/ml in water) for 30 min, washed, mounted in dH2O and photographed under Histology, immunohistochemistry, TUNELS and a ¯uorescent microscope. Hoechst staining Embryos were embedded in 4% paraformaldehyde in PBS, dehydrated, paran embedded and sectioned through the lens at 8 m. Sections were deparaned in xylenes and Acknowledgments rehydrated in decreasing concentrations of ethanol. Immu- We are grateful to Drs Tyler Jacks and Michael Greenberg nohistochemistry was performed with PCNA monoclonal for permission and Rod Bremner and Lina Dagnino for antibody (Sigma) at 1 : 200 dilution followed by biotin-labeled provision of knockout mice, Rod Bremner for critical anti-mouse and ABC kit from DAKO. A modi®ed TUNEL comments on the manuscript and Zhe Jiang for analysis of procedure (Gavrieli et al., 1992) was performed as described the mRbP(L).lacZ mice. E Zacksenhaus is a scholar of the in (Jiang et al., 2000a). To quantify the levels of apoptosis, Cancer Research Society Inc./Medical Research Council of the number of TUNEL-positive cells was counted and Canada. This work was funded by a grant to E normalized for the lens area from several sections from Zacksenhaus from the Medical Research Council of independent mutant embryos. Apoptosis in mgRb:Rb7/7 Canada (MT-14314).

References

Agah R, Kirshenbaum LA, Abdellatif M, Truong LD, Lee EYHP, Chang CY, Hu N, Wang YCJ, Lai CC, Herrup Chakraborty S, Michael LH and Schneider MD. (1997). K, Lee WH and Bradley A. (1992). Nature, 359, 288 ± 294. J. Clin. Invest., 100, 2722 ± 2728. Macleod KF, Hu Y and Jacks T. (1996). EMBO J., 15, Bates S, Phillips AC, Clark PA, Stott F, Peters G, Ludwig 6178 ± 6188. RL and Vousden KH. (1998). Nature, 395, 124 ± 125. Masaki S and Quinlan RA. (1997). Gene, 201, 11 ± 20. Clarke AR, Maandag ER, Van Roon M, Van der Lugt McCa€rey J, Yamasaki L, Dyson NJ, Harlow E and Griep NMT, Van der Valk M, Hooper ML, Berns A and Te Riele AE. (1999). Mol. Cell. Biol., 19, 6458 ± 6468. H. (1992). Nature, 359, 328 ± 330. Merdes A, Brunkener M, Horstmann H and Georgatos SD. Cvekl A, Kashanchi F, Brady JN and Piatigorsky J. (1999). (1991). J. Cell. Biol., 115, 397 ± 410. Invest. Ophthalmol. Vis. Sci., 40, 1343 ± 1350. Morgenbesser SD, Williams BO, Jacks T and DePinho RA. Cvekl A and Piatigorsky J. (1996). Bioessays, 18, 621 ± 630. (1994). Nature, 371, 72 ± 74. DeGregori J, Leone G, Miron A, Jakoi L and Nevins JR. Murer-Orlando M, Paterson RC, Lok S, Tsui LC and (1997). Proc. Natl. Acad. Sci. USA, 94, 7245 ± 7250. Breitman ML. (1987). Dev. Biol., 119, 260 ± 267. Du W. (2000). Development, 127, 367 ± 379. Nevins JR, Leone G, DeGregori J and Jakoi L. (1997). J. Du W, Vidal M, Xie JE and Dyson N. (1996). Genes Dev., 10, Cell. Physiol., 173, 233 ± 236. 1206 ± 1218. Novitch BG, Mulligan GJ, Jacks T and Lassar AB. (1996). J. Dynlacht BD, Brook A, Dembski M, Yenush L and Dyson Cell. Biol., 135, 441 ± 456. N. (1994). Proc. Natl. Acad. Sci. USA, 91, 6359 ± 6363. Oliver G and Gruss P. (1997). Trends Neurosci., 20, 415 ± Dyson N. (1998). Genes Dev., 12, 2245 ± 2262. 421. Field SJ, Tsai FY, Kuo F, Zubiage AM, Kaelin WG, Phillips AC, Ernst MK, Bates S, Rice NR and Vousden KH. Livingston DM, Orkin SH and Greenberg ME. (1996). (1999). Mol. Cell., 4, 771 ± 781. Cell, 85, 549 ± 561. Pomerantz J, Schreiber-Agus N, Liegeois NJ, Silverman A, Francis PJ, Berry V, Moore AT and Bhattacharya S. (1999). Alland L, Chin L, Potes J, Chen K, Orlow I, Lee HW, Trends Genet., 15, 191 ± 196. Cordon-Cardo C and DePinho RA. (1998). Cell, 29, 713 ± Fromm L and Overbeek PA. (1996). Oncogene, 12, 69 ± 75. 723. Gavrieli Y, Sherman Y and Ben-Sasson SA. (1992). J. Cell. Rampalli AM, Gao CY, Chauthaiwale VM and Zelenka PS. Biol., 119, 493 ± 501. (1998). Oncogene, 16, 399 ± 408. Humbert PO, Verona R, Trimarchi JM, Rogers C, Sawado T, Yamaguchi M, Nishimoto Y, Ohno K, Sakaguchi Dandapani S and Lees JA. (2000). Gene Dev., 14, 690 ± K and Matsukage A. (1998). Biochem. Biophys. Res. 703. Commun., 251, 409 ± 415. JacksT,FazeliA,SchmittEM,BronsonRT,GoodellMA Sellers WR and Kaelin WG. (1996). Biochim. Biophys. Acta., and Weinberg RA. (1992). Nature, 359, 295 ± 300. 1288, M1 ± M5. Jacks T, Remington L, Williams BO, Schmitt EM, Halachmi Sellers WR, Novitch BG, Miyake S, Heith A, Otterson GA, S, Bronson RT and Weinberg RA. (1994). Curr. Biol., 4, Kaye FJ, Lassar AB and Kaelin WGJ. (1998). Genes. Dev., 1±7. 12, 95 ± 106. Jiang Z, Gou Z, Saad F, Ellis J and Zacksenhaus E. (2000b), Tsai KY, Hu Y, Macleod KF, Crowley D, Yamasaki L and J. Biol. Chem., in press. Jacks T. (1998). Mol. Cell., 2, 293 ± 304. Jiang Z, Liang P, Leng R, Guo Z, Liu Y, Liu X, Bubnic S, Vitelli L, Condorelli G, Lulli V, Hoang T, Luchetti L, Croce Keating A, Murray D, Goss PE and Zacksenhaus E. CM and Peschle C. (2000). Mol. Cell. Biol., 20, 5330 ± (2000a), Dev. Biol., in press. 5342. Jiang Z, Zacksenhaus E, Gallie BL and Phillips RA. (1997). Weinberg RA. (1995). Cell, 81, 323 ± 330. Oncogene, 14, 1789 ± 1797. Wigle JT, Chowdhury K, Gruss P and Oliver G. (1999). Nat. Johnson DG, Schwarz JK, Cress WD and Nevins JR. (1993). Genet., 21, 318 ± 322. Nature, 365, 349 ± 352. Williams BO, Remington L, Albert DM, Mukai S, Bronson Kondoh H. (1999). Curr.Opin.Genet.Dev.,9, 301 ± 308. RT and Jacks T. (1994). Nat. Genet., 7, 480 ± 484. Kumar J and Moses K. (1997). Genes Dev., 11, 2023 ± 2028. Wu X and Levine AJ. (1994). Proc. Natl. Acad. Sci. USA, 91, Lee EY-HP, Hu N, Yuan S-SF, Cox LA, Bradley A, Lee W- 3602 ± 3606. H and Herrup K. (1994). Genes Dev., 8, 2008 ± 2021.

Oncogene Effects of E2F1 inactivation on lens development in Rb mutant fetuses Y Liu and E Zacksenhaus 6073 ZacksenhausE,BremnerR,JiangZ,GillRM,MuncasterM, Zhang P, Wong C, DePinho RA, Harper JW and Elledge SJ. Sopta M, Phillips RA and Gallie BL. (1993). Adv. Cancer (1998). Genes Dev., 12, 3162 ± 3167. Res., 61, 115 ± 141. ZhuX,PattendenSandBremnerR.(1999).Oncogene, 18, Zacksenhaus E, Jiang Z, Chung D, Marth J, Phillips RA and 4940 ± 4947. Gallie BL. (1996). Genes Dev., 10, 3051 ± 3064.

Oncogene