|Home | About | Journals | Submit | Contact Us | Français|
A growing body of evidence indicates that early mitotic inhibitor 1 (Emi1) is essential for genomic stability, but how this function relates to embryonic development and cancer pathogenesis remains unclear. We have identified a zebrafish mutant line in which deficient emi1 gene expression results in multilineage hematopoietic defects and widespread developmental defects that are p53 independent. Cell cycle analyses of Emi1-depleted zebrafish or human cells showed chromosomal rereplication, and metaphase preparations from mutant zebrafish embryos revealed rereplicated, unsegregated chromosomes and polyploidy. Furthermore, EMI1-depleted mammalian cells relied on topoisomerase IIα-dependent mitotic decatenation to progress through metaphase. Interestingly, the loss of a single emi1 allele in the absence of p53 enhanced the susceptibility of adult fish to neural sheath tumorigenesis. Our results cast Emi1 as a critical regulator of genomic fidelity during embryogenesis and suggest that the factor may act as a tumor suppressor.
Successful cell division requires faithful replication of the genome, and defects in this process can contribute to genomic instability and subsequent malignant transformation (23). A key regulator of the normal cell cycle is the early mitotic inhibitor 1 (EMI1/FBXO5), a zinc finger protein expressed by a variety of adult tissues and especially in proliferating Ki-67-positive cells (39). Studies of the mammalian and Xenopus homologues of EMI1 have shown that it inhibits the anaphase-promoting complex/cyclosome (APC/C), a ubiquitin ligase complex that targets cell cycle-regulated proteins, such as the S- and M-phase cyclins (A and B), securin, and geminin (13, 25, 31). Depletion of EMI1 by small interfering RNA (siRNA) knockdown in cell lines or immunodepletion in cycling Xenopus extracts results in the untimely degradation of APC/C substrates, delaying G1/S- and M-phase progression and inducing rereplication (6, 21, 25, 31). Such rereplication is a consequence of decreased levels of the APC/C substrates cyclin A and geminin, which are regulators of replication licensing (6, 21). The result of EMI1 depletion in some cell lines is senescence (39).
Despite these insights into the molecular underpinnings of EMI1 function, little is known about the role of this protein in development. Knockout of murine Emi1 results in an embryonic-lethal phenotype prior to implantation, while a deficiency of Emi1 in cultured pronuclear zygotes leads to multipolar and tangled spindle structures, orphan chromosomes, large nuclei, and apoptosis by the 16-cell stage (17). Otherwise, the dynamic influence of EMI1 on early vertebrate development remains undefined. We sought to close this gap by taking advantage of the zebrafish model system. Zebrafish embryos harboring homozygous mutations of emi1 (emi1m/m) develop beyond the onset of circulation, providing a unique opportunity to examine the developmental roles of Emi1 in vivo. The zebrafish emi1 mutant (hi2648) line was originally identified by a proviral insertional mutagenesis screen designed to identify genes that are necessary for normal morphological development in embryos (1, 8, 9). Subsequent studies showed that the insertion was located between the first and second exons of the emi1 gene (2). The morphological defects in emi1m/m embryos at 2 days postfertilization (p.f.) are described in the public access zebrafish model organism database (http://zfin.org). Briefly, abnormalities in emi1m/m embryos can be identified as early as 20 h p.f. and include slightly smaller heads and a lack of ventral curving of the posterior presomitic mesoderm. By 25 h p.f., the tail is more ventrally curved than in normal embryos, and increased cell death is observed throughout the central nervous system. Mutant embryos have circulating blood cells, although the onset of circulation is delayed. We became interested in this mutant because it harbors defects in the numbers and morphology of granulocytes, an important myeloid cell type within the innate immune system.
There is evidence that EMI1 may function in cancer pathogenesis, and a variety of human tumors express this factor very highly, although in some cases this may be a consequence of elevated proliferation rates (11, 18). In fact, the human homologue of emi1 resides within chromosome 6q25, a region often deleted in leukemia, which, together with the cell cycle-regulatory role of EMI1, suggested that this factor may also function as a tumor suppressor whose loss of function could promote genetic instability. Thus, in addition to investigating the role of zebrafish emi1 in zebrafish development, with particular emphasis on hematopoiesis, we examined mammalian cells to identify mechanisms that may be important in EMI1-related malignant transformation and explored a putative tumor suppressor role for this cell cycle-regulatory protein.
Wild-type AB stocks of Danio rerio and transgenic and mutant lines were maintained by standard methods and staged as previously established (16, 40). The transgenic zpu.1-EGFP, emi1m (emi1hi2648/fbxo5hi2648), and p53m (tp53zdf1/p53e7) lines have been previously described (2, 4, 14).
Images were obtained using a Nikon SMZ-1500 zoom microscope or a Zeiss Axiovert upright microscope, and color images were captured with Openlab software (Improvision). When necessary, multiple focal layers from separate Openlab images were combined to make a single Photoshop figure; however, we did not enhance, obscure, move, or remove any specific feature within an image. Confocal images were captured with the Zeiss LSM 510 META NLO laser scanning microscope and a Zeiss LD 40× 0.6NA Achroplan objective lens. A multitrack line scan configuration allowed line-by-line pseudosimultaneous capture of red, green, and blue channels. Imaging was performed at room temperature with a Hamamatsu Orca or Nikon DS-Ri1 digital camera. All figures show sibling embryos, or cells are shown at equal magnifications.
Antisense morpholino oligonucleotides (Gene Tools) were designed to target the emi1 translational start site (ATG) and the splice donor site of exon 2. The ATG morpholino sequence is 5′-GTTTGGACACTTCATATTGAGGAGA-3′, the exon 2 splice donor morpholino sequence is 5′-ATTGTCGTTTCACCTCATCATCTGA-3′, and the exon 2 splice donor mismatch control morpholino is 5′-ATTcTCcTTTCAgCTCATgATgTGA-3′ (lowercase letters indicate mismatched base pairs). The efficacy of the emi1 exon 2 splice site-targeting morpholino was tested by single-embryo reverse transcription (RT)-PCR using primers in exon 1 (forward) and exon 3 (reverse). The zebrafish emi1 gene was cloned from wild-type 24-h p.f. zebrafish using a Qiagen OneStep RT-PCR kit, and DNA sequencing confirmed its identity with published mRNA sequences (accession number NM_001003869). Human emi1 was cloned from a cDNA library made from a CD34+ cell fraction of normal human bone marrow, and the DNA sequence was identical to that of a published mRNA (NM_012177). Both EMI1 genes were subcloned into the pCS2+ vector. Zebrafish transcripts were transcribed in vitro for antisense (XhoI and SP6 RNA polymerase) or sense (ClaI and T3 RNA polymerase) mRNA or probes. Human mRNA was prepared in a similar manner (sense, KpnI and SP6; antisense, BamHI and T3). Full-length mRNAs were in vitro transcribed using mMessage Machine kits (Ambion), and Ambion labeling reagents were used to generate all in situ probes. Embryos obtained from crosses of adult fish were injected at the one- or two-cell stage with 200 pl of sense or antisense mRNA (100 ng/μl, with phenol red) or 0.5 mM morpholinos with an injection volume equal to 1 ng of morpholinos per embryo. For the hi2648 rescue studies, the primers used for emi1 hi2648 allele genotyping following mRNA injection were 5′-GCTACCACTGCAATAGCGACAAG-3′ (emi1 sequence) and 5′-GCCAAACCTACAGGTGGGGTC-3′ (viral sequence), and primers to exon 1 and intron 1, corresponding to the region of viral insertion, were used to identify the wild-type allele. The methods for genotyping the wild-type and e7 mutant p53 alleles and use of the p53 morpholino were published previously (4, 34).
Embryos obtained from crosses of adult fish were raised in E3 medium containing 0.2 mM N-phenylthiourea (40). Embryos were fixed in 4% paraformaldehyde (PFA) in phosphate-buffered saline (PBS) overnight at 4°C and dehydrated in 100% methanol at −20°C. To examine cells undergoing DNA replication, live dechorionated embryos were incubated with 10 mM 5-bromo-2-deoxyuridine (BrdU) (Sigma) on ice for 30 min, and then the BrdU was removed and the embryos were transferred to a 28°C incubator for 1 h. The embryos were fixed as described above and permeabilized with an acetone treatment for 7 min at −20°C, followed by a 6-min incubation with 0.25% trypsin in PBS. To allow the detection of BrdU, the embryos were treated with 2 N HCl for 1 h at room temperature, and incorporation was detected using a mouse BrdU antibody (1/100; Roche) and an Alexa-conjugated secondary antibody (1:1,000; Invitrogen). Zebrafish phosphorylated histone H2AX (p-H2AX) was detected using a polyclonal anti-p-H2AX antibody (1:1,000) that was generously provided by James Amatruda, University of Texas Southwestern, and an Alexa-conjugated secondary antibody (1:1,000; Invitrogen). To permeate the embryos for p-H2AX detection, fixed embryos in methanol were treated with acetone. Blocking and antibody incubations were performed in PBS with 0.1% Tween 20, 2% blocking reagent (Roche), 5% fetal calf serum, and 1% dimethyl sulfoxide (DMSO). Using a similar protocol, zebrafish phosphorylated histone H3 (p-H3) was detected using a polyclonal anti-p-H3 antibody (1:1,000; Santa Cruz Biotechnology). The protocols for generating antisense RNA probes to detect mpo, l-plastin, pu.1, and α-globin expression by whole-mount RNA in situ hybridization (WISH) were published previously (32). We examined the difference between the average numbers of myeloid cells in emi1m/m versus wild-type homozygous (emi1wt/wt) or heterozygous (emi1wt/m) siblings using a two-tailed t test and standard error (P < 0.001). Hemoglobin was detected by incubating live embryos in o-dianisidine/H2O2 (Sigma) for 5 to 10 min at room temperature. Using Photoshop, individual o-dianisidine-positive cells were selected, and the area was quantified in pixels. The emi1m/m versus emi1wt/(wt or m) cell size distribution and averages were examined by using a two-tailed t test and standard error (P < 0.001). The protocols for terminal deoxynucleotidyltransferase-mediated dUTP-biotin nick end labeling (TUNEL) staining of fixed embryos and acridine orange staining of live embryos have been previously established (34).
Cells were sorted from zpu.1-EGFP transgenic embryos and stained with May-Grunwald-Giemsa stain as previously described (32). Propidium iodide-based cell cycle analysis was performed as published previously (34) with slight modifications. The modified protocol included the fixation of cell suspensions in 2% glucose and 3% PFA in PBS, which were then dehydrated by dropwise addition of 80% ethanol and resuspended in PBS with 20 mM HEPES, 0.25% NP-40, and 0.1% bovine serum albumin.
Adult zebrafish harboring p53 and emi1 mutations were generated by standard mating practices. Adult progeny from a single cross were genotyped, segregated based on genotype, and examined weekly for the duration of their life spans; thus, all the fish examined were siblings. Fish were kept at a density of no more than 29 fish per 9 liters of water, with some variation in density between the genotypes. Fish with tumors comprising about 5% to 10% of their total body mass were culled with ice, their abdomens were punctured, and the fish were submerged in 4% PFA in PBS. Tissue processing, paraffin embedding, sectioning, and hematoxylin/eosin staining were performed by standard procedures within the Dana-Farber Harvard Cancer Center's Specialized Histopathology Core Laboratory, and the designation of the tumor type was determined by Jeffery Kutok. The tumor incidences were plotted using inverse Kaplan-Meier curves, and the two-sided P values were obtained by a log rank test. A total of 98 adult fish generated from a sibling incross were used in this experiment (see Fig. Fig.9C9C).
Metaphase chromosome spreads were obtained from 28-h p.f. embryos using standard cytogenetic procedures (33). Cells were arrested at metaphase with colcemid, subjected to hypotonic treatments, and fixed onto glass slides with 3:1 methanol-acetic acid, and the chromosomes were visualized with Wright's stain (Sigma). Experiments performed in duplicate had similar results (see Fig. Fig.77 for representative chromosomes and counts from a single experiment).
Fluorescence in situ hybridization of tumor sections was performed as previously published (33). Fluorescently labeled probes included a green ch211-114c12 bacterial artificial chromosome (BAC) DNA probe for zebrafish chromosome 5, encompassing the p53 gene, and a yellow chromosome 5 centromere probe, using BAC zc150k20.
Cell lines were cultured using standard media and practices. Knockdown and Western blot analyses were performed as published previously (21, 30, 34, 39). For knockdown analysis, control siRNA and Emi1 siRNA were purchased from Dharmacon (On-Target plus siControl nontargeting siRNA and Emi1/Fbxo5 On-Target plus Smartpool, respectively). Each siRNA pool was transfected at a final concentration of 20 nM using Hiperfect reagent (Qiagen) according to the manufacturer's instructions. Emi1 protein was detected using a polyclonal antibody (1:1,000; Zymed) and a horseradish peroxidase-conjugated secondary antibody (1:2,000; Cell Signaling Technology). Decatenation assays were performed as previously described with P values determined using Student's t test (41).
As part of an analysis of insertional zebrafish mutants in the collection identified by Nancy Hopkins, we found that 2 days p.f., emi1m/m embryos have decreased numbers of myeloperoxidase (mpo)-expressing cells, or granulocytes (Fig. 1A and B). Further experiments using WISH analysis showed decreased numbers of mature myeloid cells expressing mpo or l-plastin in emi1m/m embryos at 24 h p.f. (Fig. 1C and D and data not shown). Furthermore, there were fewer pu.1-expressing myeloid progenitor cells in emi1m/m embryos (Fig. 1E and F). By manually quantifying the WISH-positive cells, we confirmed that the average number of cells per embryo was significantly decreased in emi1m/m embryos compared with wild-type emi1wt/(wt or m) siblings (Fig. (Fig.1G)1G) (P < 0.001 for each comparison).
The WISH analysis shown in Fig. Fig.11 also revealed that the residual emi1m/m myeloid cells appeared larger than normal, leading us to examine mutants carrying the zpu.1-EGFP transgene, in which the zebrafish pu.1 promoter drives expression of the gene encoding enhanced green fluorescent protein (EGFP; referred to here as GFP) (14). We observed fewer live GFP+ cells in emi1m/m transgenic embryos (data not shown), consistent with the WISH data. High-magnification microscopy of live GFP-expressing mutant cells showed that these cells were indeed quite large but maintained normal cell shapes (Fig. 2A and B). We then quantified the myeloid progenitor cell size by FACS analysis of GFP+ cells from emi1m/m or emi1wt/(wt or m) siblings at 21 h p.f. (Fig. 2C and D). The forward scatter (x axis) showed a shift to the right, indicating that the mutant cell population contained cells that were on average twice as large as those in the wild-type population. Cytospins of the GFP+ cell populations at 21 h p.f. and 48 h p.f. (data not shown and Fig. 2E and F) confirmed that the emi1m/m myeloid cells were much larger but shared similar morphological features with emi1wt/(wt or m) myeloid cells.
As previously noted, emi1m/m embryos exhibit widespread developmental abnormalities (Fig. 3A and B), suggesting that many tissues may be affected, including additional hematopoietic lineages. A comparison of embryos with circulating erythrocytes showed that emi1m/m mutants had reduced circulating blood (data not shown) and fewer α-globin-expressing cells than their wild-type siblings (Fig. 3C and D). Interestingly, in contrast to normal erythrocytes, the emi1m/m erythroid cells appeared to vary widely in size. Whole-mount o-dianisidine staining of hemoglobin-containing cells further demonstrated the variation in emi1m/m erythroid cell size (Fig. 3E to H). By quantifying the area of o-dianisidine-positive cells from images of stained embryos, we confirmed that the distribution of emi1m/m erythroid cell sizes was significantly different from that of the emi1wt/(wt or m) erythroid population, with emi1m/m erythrocytes having an average size that was twofold larger than that of wild-type cells (Fig. (Fig.3K)3K) (P < 0.001). Furthermore, FACS analysis showed that the total cell population for mutants was larger than that for wild-type cells (Fig. 3I and J), indicating that Emi1 has a functional role affecting many different cell types throughout the developing embryo.
Zebrafish emi1 is expressed maternally and in a ubiquitous pattern throughout the developing embryo (data not shown and Fig. Fig.4B)4B) but was not detected in mature myeloid cells, as indicated by the lack of emi1-positive cells on the yolk of 21-h p.f. embryos, where differentiated granulocytes are normally found. This pattern is consistent with the reported expression of emi1 in proliferating cells (39). At 21 h p.f., emi1m/m embryos showed a severe reduction of emi1 expression (Fig. (Fig.4C).4C). We verified that the hematopoietic and morphological phenotypes of emi1m/m embryos were due to the loss of emi1 expression by phenocopying the mutant with morpholinos targeting either an emi1 splice site (Fig. 4A and D to F) or the emi1 ATG codon (data not shown) and also by rescuing emi1m/m embryos by the forced expression of zebrafish emi1 mRNA (Fig. 4G to I). Importantly, the injection of human EMI1 mRNA into emi1m/m mutants rescued the development of normal-size mpo-positive cells and could return their numbers to wild-type levels (data not shown), indicating that these gene orthologs are functionally conserved in vivo.
Depletion of EMI1 in human cell lines results in rereplication (6, 21, 39), suggesting that an essential role of the protein is to preserve genomic integrity by blocking rereplication. To address whether the EMI1 function is conserved in zebrafish, we performed propidium iodide-based FACS analysis of cells from 21 h p.f. emi1m/m or emi1wt/(wt or m) zpu.1-EGFP transgenic siblings. At this age, hematopoietic progenitors and more differentiated cells are present in the developing embryos, and this is the earliest age at which we could morphologically distinguish emi1m/m mutants from emi1wt/(wt or m) siblings. The total cell population and the GFP+ subset from wild-type embryos showed typical distributions of cycling normal cells (Fig. (Fig.5A)5A) in G1 (64% and 61%, total and GFP+, respectively), S (31% and 33%), and G2/M (5% and 6%). In contrast, the distributions of emi1m/m total and GFP+ cycling cells were G1, 36% and 31%; S, 33% and 30%; and G2/M, 31% and 38%, with a large increase in both a sub-G1 population and a cell fraction greater than the 4 N complement of DNA (Fig. (Fig.5B).5B). Previous studies of mammalian cell lines showed that severe siRNA depletion of EMI1 caused rereplication, which is consistent with our findings, and prevented cells from entering into mitosis, as indicated by a very low percentage of cells with nuclear envelope degradation or p-H3 (6, 21). Interestingly, using confocal microscopy to assess p-H3 staining, we found that emi1m/m mutants and emi1wt/(wt or m) siblings had identical mitotic indexes, in that a wild-type sibling contained 10 p-H3+ nuclei per 198 DAPI+ nuclei that were analyzed (~5% mitotic cells) whereas an emi1m/m mutant embryo showed 5 p-H3+ nuclei out of a total of 98 DAPI+ nuclei enumerated (~5% mitotic cells). Representative images are shown in Fig. 5C to H, illustrating that in the trunks of the embryos, especially in the p-H3+ and DAPI merged panels (E and H), the DAPI-stained emi1m/m mutant nuclei are larger and fewer in number but the proportion of p-H3+ nuclei is the same as in emi1wt/(wt or m) siblings. In previous studies (6, 21), severe siRNA depletion of EMI1 prevented cells from entering into mitosis, suggesting that the cells entering mitosis in emi1m/m mutant zebrafish have residual maternal Emi1 or low levels of zygotic Emi1 that allow mitotic entry. However, when viewed in the context of the FACS analysis of nuclear DNA content, most of the increased cells with a 4 N or greater complement of DNA in emi1m/m mutant zebrafish represent G2-phase cells or cells continuing in cycle with rereplicated DNA.
Published studies using human cell lines show that loss of EMI1 leads to an increase in double-strand DNA breaks, as indicated by analysis of p-H2AX (6, 21, 39). We therefore evaluated the levels of p-H2AX in zebrafish embryos and detected increased levels of this phosphorylated protein in emi1m/m embryos by Western analysis (Fig. (Fig.6Q)6Q) and whole-mount immunohistochemistry (Fig. 6R and S). Using confocal microscopy, we found that the levels of p-H2AX in zpu.1-GFP myeloid cells were strikingly increased in emi1m/m embryos (Fig. 6I to P) relative to emi1wt/(wt or m) siblings (Fig. 6A to H).
We consistently observed increased BrdU incorporation, reflecting increased DNA replication, in cells throughout emi1m/m embryos at 24 h p.f. (Fig. 7C and D). We next examined whether emi1m/m embryos manifested changes in chromosome numbers by analyzing metaphase spreads of cells from emi1wt/(wt or m), and emi1m/m siblings. In multiple independent experiments, metaphase spreads from wild-type embryos had a modal chromosome count of 50 chromosomes per cell, which is normal for zebrafish cells (Fig. 7E, H, and K). In contrast, almost half of the metaphases from emi1m/m mutants showed rereplicated, unsegregated chromosomes (Fig. 7F, I, and K) or polyploidy (Fig. (Fig.7G,7G, J, and K). These data indicate that zebrafish cells lacking emi1 exhibit aberrant DNA rereplication during embryogenesis.
We hypothesized that rereplication in the absence of chromatid disjunction might result in highly intertwined, or catenated, DNA. Decatenation, or untangling of the chromosomes, requires topoisomerase (Topo) IIα, and this process can be blocked by ICRF-193, a chemical inhibitor of Topo IIα (3, 15). Hence, we used HCT116 human colon carcinoma cells to test whether EMI1 depletion affects the proportion of cells that are dependent on Topo IIα to progress through mitosis, thus indirectly assessing the percentage of mitotic cells harboring catenated DNA. EMI1 was partially depleted with an siRNA specific to human EMI1 mRNA (Fig. (Fig.8T).8T). Catenation can inhibit cell cycle progression at two points, preventing progression from G2 to M (7) and progression within M phase from metaphase to anaphase, ultimately blocking cell division if the chromosomal junctions are unresolved (5, 24, 35, 38). We studied the percentage of cells in metaphase by immunofluorescence microscopy of asynchronous cells stained with DAPI and a tubulin antibody, a control for spindle assembly (data not shown). Equal percentages of metaphase cells were identified in control and EMI1-depleted cells following exposure to DMSO, which is consistent with the normal mitotic index in emi1m/m embryos and probably reflects the residual EMI1 protein evident by Western blot analysis in the EMI1-siRNA-treated HCT116 human colon carcinoma cells (Fig. (Fig.8T).8T). Following ICRF-193 exposure, EMI1-depleted cells showed a statistically significant increase in the percentage of cells in metaphase in comparison with control cells (Fig. (Fig.7L)7L) (P < 0.01). Thus, the mitotic obstruction that occurs when Topo IIα is challenged with ICRF-193 is exacerbated by EMI1 depletion, indicating that decreased EMI1 results in an increased proportion of cells with mitotic catenated DNA, which may contribute to genomic instability and the production of cells with greater than 4 N DNA content.
When analyzed by FACS, emi1m/m embryos showed an increased population of sub-G1 DNA content cells indicative of apoptotic cells, which we confirmed by TUNEL (Fig. 8A and B) and acridine orange staining (Fig. 8C and D). Cell death was most prevalent in the central nervous system within the head and spinal cord of mutant embryos but was also observed in cells on the yolk and in the tail. We could block cell death in emi1wt/(wt or m) and emi1m/m embryos by p53-morpholino knockdown or in a p53 mutant line (Fig. 8E and F and data not shown) (4). However, the aberrant granulocytic cell phenotype of emi1m/m embryos was p53 independent by morphology and mpo WISH analysis (Fig. 8G to J). Furthermore, EMI1 knockdown in HCT116 cells in the presence or absence of p53 resulted in identical cell cycle distributions, indicating that the cell cycle defects were not caused by p53 activation (Fig. 8O to T).
Zebrafish embryos heterozygous for emi1 mutations appear to develop normally, and the adult fish do not develop tumors at a higher rate than wild-type strains. To determine whether emi1 haploinsufficiency could synergize with p53 mutations in tumorigenesis, we bred this line into a p53 mutant zebrafish line in which the adult fish develop peripheral neural sheath tumors (4). We analyzed tumors formed in adults harboring mutations in p53 (p53wt/m or p53m/m) with wild-type alleles of emi1 (emi1wt/wt) or emi1 haploinsufficiency (emi1wt/m). The tumors identified in p53 null fish were consistent with previous studies and were predominantly malignant peripheral neural sheath tumors of the spindle cell type (Fig. 9A and B), with a single case showing a more epithelioid morphology (data not shown). Fish heterozygous for the p53 mutation also developed malignant peripheral neural sheath tumors, although other tumor types, such as rosette-bearing neural tumors with morphology consistent with peripheral primitive neuroectodermal tumors, were occasionally observed. Fish heterozygous for both the p53 mutation and the emi1 mutation displayed decreased tumor incidence compared to p53 heterozygotes with wild-type emi1wt/wt (P = 0.007), indicating that emi1 haploinsufficiency in this context did not promote tumorigenesis and suggesting that heterozygous emi1 mutation does not promote the loss of the wild-type allele of p53. To verify this finding, we examined the regions on zebrafish chromosome 5 encompassing the zebrafish p53 gene and the centromeric region, using fluorescently labeled BACs in tumors arising from double-heterozygous fish (Fig. 9D and E). The fish analyzed were not siblings of those in Fig. Fig.9C.9C. Paraffin sections of the double-heterozygous tumors from two fish were probed, and in every cell examined from both tumors (cell count, >100 for each tumor), there were two signals for p53, suggesting that the p53 wild-type allele is retained, although we cannot rule out the possibility that the wild-type allele had not been inactivated either by gene conversion from the mutant allele or by de novo mutation.
Turning to the p53m/m fish, we found a significant increase in the rapidity of onset and the overall cumulative incidence of tumors in emi1wt/m p53m/m fish compared with emi1wt/wt p53m/m or emi1wt/wt p53wt/m fish (P = 0.047) (Fig. (Fig.9C).9C). Thus, emi1 functions as a haploinsufficient tumor suppressor that synergizes with loss of p53 function, which is found in at least 50% of human cancers overall.
We have shown that inactivation of emi1 in the zebrafish germ line has marked developmental defects, including multilineage hematopoietic abnormalities. Human EMI1 mRNA rescued the phenotype of zebrafish emi1m/m embryos, and depletion of EMI1 in zebrafish or human cells led to similar cell cycle defects, indicating that the function of this mitotic regulator is conserved from zebrafish to humans. EMI1 depletion in human cells and zebrafish leads to increased cell size, while FACS analysis of human cell lines, as well as zebrafish cells, in the present study indicated that this factor is essential for inhibiting rereplication (6, 21, 39, 42). It is likely that our observation of rereplicated and unsegregated chromosomes or polyploid karyotypes has not been reported before because knockdown of EMI1 in human cell lines induces senescence and the murine Emi1 knockout has a very early lethal phenotype (17, 39). Nonetheless, we have shown that emi1 depletion is a mechanism for the genesis of near-tetraploid cells, suggesting that emi1m/m zebrafish may be a useful model for examining the molecular mechanism regulating this process and the ensuing cellular consequences, such as the derivation of aneuploid cells, as well as identifying additional mechanisms through which emi1 may contribute to leukemogenesis. Of note, increased ploidy has been associated with an increase in cell size (10, 33, 36), suggesting that these two phenotypes may be inherently linked in our mutant.
To account for the presence of metaphase spreads with polyploidy or rereplicated, unsegregated chromosomes in emi1m/m cells, we asked whether EMI1 depletion affected the percentage of cells that are dependent on Topo IIα in order to progress beyond metaphase, thus indirectly assessing the percentage of mitotic cells harboring catenated DNA. Our results showed that loss of EMI1 in human cells enhanced the percentage of cells in metaphase when Topo IIα was inhibited, suggesting that these cells retained mitotic catenated DNA and were unable to undergo chromosome segregation. Thus, loss of EMI1 resulted in an increased population of cells with catenated DNA, a defect that may be due to rereplication. Interestingly, Emi1-deficient mouse pronuclei showed an abundance of abnormal mitoses, which may be due in part to catenation-related defects in chromosome disjunction (17). Our working hypothesis is that unresolved DNA catenation in emi1-deficient cells may prevent cells from completing cell division or facilitate unbalanced chromosomal segregation, resulting in aneuploid cells, a process that may be facilitated by a cellular environment that precedes apoptosis.
Zhang et al. recently described an alternate allele of emi1 (emi1tiy121) in which a premature stop codon near the amino terminus results in a truncated protein, and the resultant zebrafish mutant appears to have more severe morphological defects than does our mutant (42). The emi1tiy121 studies focused on the morphologically abnormal somites, while we observed normal-appearing chevron-shaped somites. The emi1tiy121 mutant cells did have larger nuclei than wild-type cells, consistent with our studies. In that study, as well as in our own, homozygous emi1 mutations could be rescued by forced expression of emi1 mRNA, indicating that both mutant phenotypes are due to the loss of emi1 expression. A small amount of emi1 mRNA can be detected by WISH in our insertional mutant, which may account for the less severe phenotype of our mutant in comparison to the emi1tiy121 mutant and the ability of emi1hi2648 mutant cells to enter into mitosis. Furthermore, heterozygous mutation of zebrafish or murine emi1 did not give rise to obvious defects during embryogenesis, and the adults remained healthy, indicating that heterozygous mutation of emi1 alone is not sufficient to disturb essential developmental pathways (17).
A striking phenotype of our emi1m/m embryos is robust apoptosis, which is evident from acridine orange- and TUNEL-positive cells in the central nervous system, on the yolk, and in posterior trunk regions consistent with apoptosis in hematopoietic cells. Apoptosis in myeloid cells was confirmed by FACS analysis, showing an increase in the sub-G1 (apoptotic) population of the emi1m/m myeloid cell population (Fig. (Fig.5).5). The acridine orange-positive apoptotic cells in emi1m/m embryos could be blocked by p53 depletion, indicating that this apoptosis is p53 dependent in all tissues. However, the emi1m/m myeloid cell phenotype of reduced cell numbers and the emi1 siRNA-induced cell cycle defects observed in human cell lines are p53 independent. We interpret this to mean that the cell cycle defects in the emi1m/m myeloid cells are primarily responsible for the myeloid cell phenotype, which is p53 independent and likely reflects the known effects of loss of Emi1-mediated inhibition of the activity of the APC/C in rapidly dividing myeloid cell progenitors (21, 25, 31).
Our results led us to hypothesize that cooperation between the heterozygous loss of emi1, or haploinsufficiency, and p53 mutation might play a role in cancer pathogenesis. Genome-wide approaches have recently revealed cryptic chromosomal changes and unexpected gene mutations in leukemia, suggesting that there are many as yet unidentified factors that can contribute to this disease (19, 27, 29). EMI1 has been shown to be oncogenic, as evidenced by proviral insertion into the Emi1 locus that accelerated lymphomagenesis in mice transgenic for the c-myc gene under the control of the immunoglobulin heavy-chain enhancer (E mu-myc) (12). In humans, high levels of EMI1 protein and RNA have been found in a variety of malignant tumors (11, 18). In contrast, deletion of the long arm of chromosome 6 encompassing the EMI1 gene has been observed in primary human leukemias and as an abnormality associated with progression of myelodysplastic syndrome to acute myeloid leukemia (20, 22, 26, 37). In addition, haploinsufficiency or complete deficiency of E2F2, a transcriptional activator of the EMI1 gene, also accelerated the onset and progression of lymphomas in E mu-myc transgenic mice (28), indirectly suggesting that EMI1 may function as a tumor suppressor. We found that compound homozygous p53 mutant and heterozygous emi1 mutant adults developed tumors at an accelerated rate and with higher penetrance than did fish with wild-type emi1, establishing that emi1 functions as a haploinsufficient tumor suppressor when combined with loss of p53 in zebrafish. This finding suggests that haploinsufficiency for the EMI1 gene is likely to be associated with inactivation of p53 in leukemia and other human cancers.
In summary, our data support an essential and evolutionarily conserved function for emi1 in regulating the integrity of the genome. Our studies are in agreement with previous reports showing that EMI1 depletion leads to rereplication and the activation of p53 (6, 21, 39). In addition, we showed that the EMI1 depletion is associated with mitotic catenated DNA and that the cell cycle defects are independent of p53. Similarly, the decrease in myeloid cell numbers and increase in myeloid cell size in zebrafish embryos are also p53 independent, even though apoptosis is markedly decreased in p53-depleted embryos. Importantly, we have provided the first evidence that emi1 acts as a tumor suppressor in p53 mutant zebrafish, and we hypothesize that emi1 may have a similar role in human malignancy.
We thank Nancy Hopkins for generously providing the emi1hi2648 mutant zebrafish line. Thanks go to John Gilbert for manuscript editing and helpful comments. We thank members of the Look laboratory for discussions and reagents, especially Samuel Sidi, Raymond Hoffmans, and Hui Feng. We acknowledge Tingxi Liu and Min Deng for the zebrafish emi1 clone and Adolfo Ferrando for use of his panel of T-ALL patient sample DNAs. We are grateful to Thomas Diefenbach and Lihong Bu for their assistance in the Mental Retardation and Developmental Disabilities Research Center's imaging facility at Children's Hospital, Boston, and Tianyu Li in the FCCC Biostatistics and Bioinformatics Facility for her help analyzing the zebrafish tumor data.
This work was supported by NIH grants R01 CA93152 (A.T.L.), K01 DK69672 (J.R.), P01 CA66996-11A1 (R.M.S. and J. D. Griffin), R01 CA111560 (C.L.), and R01 RR12589 (A.A. and Nancy Hopkins).
Published ahead of print on 24 August 2009.