Pescadillo is essential for nucleolar assembly, ribosome biogenesis, and mammalian cell proliferation.

Mutation of the zebrafish pescadillo gene blocks expansion of a number of tissues in the developing embryo, suggesting roles for its gene product in controlling cell proliferation. We report that levels of the pescadillo protein increase in rodent hepatocytes as they enter the cell cycle. Pescadillo protein localizes to distinct substructures of the interphase nucleus including nucleoli, the site of ribosome biogenesis. During mitosis pescadillo closely associates with the periphery of metaphase chromosomes and by late anaphase is associated with nucleolus-derived foci and prenucleolar bodies. Blastomeres in mouse embryos lacking pescadillo arrest at morula stages of development, the nucleoli fail to differentiate and accumulation of ribosomes is inhibited. We propose that in mammalian cells pescadillo is essential for ribosome biogenesis and nucleologenesis and that disruption to its function results in cell cycle arrest.

Pescadillo was first identified in a genetic screen for mutations that affect embryonic development of the zebrafish Danio rerio (1,2). Pescadillo Ϫ/Ϫ zebrafish mutants can first be identified on developmental day 3 by reduced brain and eye size and a lack of extension of the jaw, possibly caused by the absence of cartilage in the branchial and jaw arches (2). In addition, the pancreas, gut, and liver fail to expand in pescadillo mutant embryos (2). Interestingly, the most severely affected organs were those that normally expressed high levels of pescadillo mRNA. Moreover, with the exception of the ovaries, pescadillo mRNA was not detected in adult fish tissues. Together these data imply a cell autonomous role for the pescadillo protein during the proliferation of specific cell types in the fish (2).
Pescadillo is well conserved between species with homologues identified in human (PES1), yeast (YPH1, Nop7p), and mouse (Pes1) (2)(3)(4). Consistent with the zebrafish studies, in situ hybridization analyses in mouse embryos reveal a restricted distribution of pescadillo mRNA with the highest levels being found in the developing liver suggesting that its function is also conserved across species (3). Analyses of the predicted amino acid sequence of pescadillo identified the presence of numerous nuclear localization signals, two acidic domains, and a putative SUMO-1 binding site (3,4). In addition, the pescadillo protein was predicted to contain a BRCA1 C-terminal (BRCT)-domain. This domain was first described in the breast cancer susceptibility protein 1 (BRCA1) and has more recently been found in several proteins involved in DNA repair, cell cycle control, and/or recombination (5)(6)(7). The presence of a BRCT domain in pescadillo is provocative and implies that the developmental defects suffered by pescadillo mutant zebrafish may be a consequence of the disruption of some aspect of cell proliferation. In support of this proposal, Kinoshita et al. (4) demonstrated that strains of Saccharomyces cerevisiae harboring mutations in the BRCT domain of the yeast pescadillo gene exhibited delays in cell cycle progression. In addition, the same study demonstrated an increase in expression of pescadillo in malignant astrocytomas and glioblastomas compared with non-malignant tissues, implying that pescadillo may contribute toward tumor progression (4).
Recently, biochemical studies in the yeast, S. cerevisaie, have suggested that pescadillo may have a role in controlling ribosome biogenesis and also the onset of DNA replication (8 -10). A requirement in such ubiquitous processes seems surprising given that several cells and tissues develop normally in zebrafish pescadillo mutants. Moreover, this may imply that multicellular organisms have a different use for pescadillo and that any lessons, regarding pescadillo, learned from yeast may not be generally applicable to vertebrates. Although the zebrafish offers many advantages for the generation and screening of mutant phenotypes, analyses of cell and molecular mechanisms are more advanced in other genetically manipulable systems such as rodents. To address the role of pescadillo in controlling mammalian cell proliferation we, therefore, began an analysis of the murine pescadillo homologue. recovered for immunohistochemical, immunoblotting, and RNA analyses. The integrated density value (IDV) of autoradiographic signals of immunoblots was determined using AlphaImager™ software. Independent immunoblots were normalized by comparing the IDVs of equal amounts of vaccinia-expressed pescadillo protein from each blot. Samples were statistically analyzed using a two-tailed Student's t test with six individual animals (n ϭ 6) representing each time point.
Gene Targeting-The targeting construct (Fig. 5) was created using the pPNT plasmid that contained pgk-neo and HSV-thymidine kinase expression cassettes for positive and negative selection, respectively (12). A 4.5-kb pescadillo genomic ClaI fragment, extending from intron 2 through the ClaI site of exon 5, was cloned into the EcoRI site of pPNT by blunt-end ligation. A 5.5-kb pescadillo genomic SmaI fragment containing 110 bp of exon 12 and extending 3.3-kb downstream of exon 15 was cloned by blunt-end ligation into the XhoI site of pPNT. Linear targeting vector was electroporated into R1 ES cells. 400 G418/gancyclovir-resistant ES cell clones were collected, and their genotype was determined by Southern blot.
Analysis of Genotype by PCR-PCR genotyping of embryos was performed using a multiplex PCR scheme using the following primers: Pes62 (5Ј-CCTCTCCACCCTGCAGGTACCCCACATT-3Ј), Pes3 (5Ј-CCTACGGCACAACTGAATGGTC-3Ј), and NeoG (5Ј-TTAAGGGC-CAGCTCATTCCTCCACTCAT-3Ј). Cycling conditions were 94°C for 5 min and then 37 cycles of 94°C for 30 s, 60°C for 45 s, and 72°C for 45 s, followed by an extension at 72°C for 7 min. This PCR yields products of 138 bp for the wild-type pescadillo allele and 281 bp for the mutated allele.
Electron Microscopy-Preimplantation embryos were collected and fixed in 2% glutaraldehyde in caocodylate buffer. Embryos were transferred to agarose, dehydrated, and embedded in epoxy resin. 80 -100-nm sections were contrasted with uranyl acetate and lead citrate and examined using a Hitachi 600 electron microscope. Polyribosomes were identified following the method of Piko and Clegg (13).

Pescadillo Protein Levels Are Increased in Proliferating
Cells-If pescadillo has a role in controlling cell proliferation we believed it likely that it would be most highly expressed in proliferating tissues. To allow us to test this prediction we generated affinity-purified antibodies to peptides derived from the predicted amino acid sequence from the amino-(PesN), middle-(PesM) and carboxyl-regions (PesC) of the murine pescadillo protein. To confirm that these antibodies could specifically identify pescadillo protein we performed immunoblot analyses using extracts from cells infected with a recombinant vaccinia virus encoding mouse pescadillo. Fig. 1A shows that the anti-PesC antibody identified a single band in infected cell extracts but not in uninfected cells or in cells infected with control virus. Based on its mobility relative to known M r standards, the protein identified by anti-PesC antibodies was estimated to be around 68 kDa. This is close to the 67.8 kDa that had been calculated through the analysis of the pescadillo predicted primary amino acid sequence. Similar results were obtained with anti-PesN antibodies, however, anti-PesM antibodies were unable to detect the vaccinia-expressed pescadillo protein (data not shown). From these results we were able to conclude that anti-PesC and -PesN antibodies can specifically identify the pescadillo protein. During this analysis we noted that the anti-PesC antibodies were unable to detect any pesca-dillo protein in uninfected BSC-40 primate cell extracts. In contrast to these data, Kinoshita et al. (4) had detected the human homologue of pescadillo in several human cancer cell lines. When we compared the amino acid sequence of the peptide used to generate the anti-PesC antibodies to the corresponding human sequence we observed that 4 of 11 amino acids were different. This suggested that the anti-PesC antibodies interacted with a pescadillo epitope that was absent from the human homologue. To test this hypothesis we probed immunoblots containing extracts from a variety of primate and mouse cell lines. Fig. 1B shows that, while anti-PesC antibodies detected pescadillo in all mouse cell lines (Hepa1, NIH3T3, STO, L-cells) tested, no protein was identified in any of the primate cell lines (HeLa, BSC-40, HepG2). These data show that the anti-PesC antibodies are mouse-specific.
Having generated antibodies that specifically recognized pescadillo we were able to use them to compare the relative abundance of pescadillo in extracts of tissues that were considered relatively quiescent versus tissues containing a significant number of proliferating cells. Fig. 1C shows that immunoblot experiments using anti-PesC antibodies detected a 68-kDa band, which co-migrated with the vaccinia virus-expressed pescadillo protein in fetal liver, ovary, testis, and intestine, all of which contain proliferating cells. In contrast, only minimal amounts of the 68-kDa pescadillo protein were identified in adult liver, heart, lung, and kidney. A larger band that appeared to be Ͼ110 kDa was identified in some tissues. Whether this represents a modified form of pescadillo is unknown; however, it should be noted that this band was not detected by anti-PesN antibodies. From these results we conclude that pescadillo expression shows a restricted distribution in adult mouse tissues and that all tissues in which the pescadillo protein was detected contain a population of proliferating cells.
While the presence of pescadillo in tissues containing proliferating cells was provocative it could easily be coincidental. If pescadillo does indeed contribute to some aspect of cell proliferation we predicted that its expression should be increased as cells transition from quiescent to replicative. 60% of the liver cell mass consists of hepatocytes, which under normal conditions are quiescent with mitosis observed in ϳ1 in every 20,000 hepatocytes (14). However, when a portion of the liver tissue is surgically removed the remaining hepatocytes proliferate to restore normal liver mass (15,16). We, therefore, performed partial hepatectomies on mice to test whether pescadillo expression was up-regulated as hepatocytes left G 0 and entered the cell cycle. Livers were collected from sham-operated mice or liver remnants from hepatectomized mice at 0, 16,36,40,44,48, and 72 h after surgery. Liver tissue obtained from six mice at each time point was processed for immunoblot analysis using anti-PesC antibodies. A representative immunoblot shown in Fig. 2A, and statistical analyses shown in Fig. 2B, demonstrate that pescadillo protein levels increase beginning at 16-h posthepatectomy and continue to climb until they plateau at around 40-h posthepatectomy. Expression of pescadillo was found to precede that of proliferating cell nuclear antigen (PCNA), which is a marker of S-phase. In agreement with BrdUrd labeling studies performed by others, PCNA expression was first identified in liver remnants at 36 h and reached maximal levels at 44-h posthepatectomy (17,18). The finding that pescadillo expression precedes that of PCNA implies that pescadillo expression is induced during the G 1 phase of the first cell cycle after surgery. We confirmed the immunoblot data by immunohistochemical analyses of liver sections taken from the same experiment. Fig. 2C shows that anti-Pes antibodies gave diffuse background staining throughout the liver in shamoperated mice and in liver remnants collected immediately after hepatectomy. By 16-h posthepatectomy faint nuclear staining could be detected in the liver remnants, which increased in intensity at 40-and 44-h posthepatectomy. Surprisingly, at 48-h postsurgery staining was reproducibly detectable but diffuse, possibly due to cells entering mitosis, before returning to a nuclear staining pattern at 72 h. Cumulatively, these data demonstrate that expression of pescadillo is upregulated as quiescent cells enter the cell cycle and support the hypothesis that pescadillo plays a role in cell proliferation.
Pescadillo Is a Nucleolar Protein That Associates with Chromosomes and Prenucleolar Bodies during Mitosis-Careful analysis of the subcellular location of pescadillo should help decipher its function. Previously, we had identified the presence of several nuclear localization motifs within the predicted amino acid sequence of pescadillo and had demonstrated that epitope-tagged pescadillo protein localized to the nucleus when expressed in cultured hepatoma cells (3). To determine whether the endogenous protein was also nuclear and to address whether its subcellular location changed during the cell cycle we performed immunostaining on NIH3T3 cells (Figs. 3 and 4).
When NIH3T3 cells were analyzed using anti-PesC antibodies, pescadillo was found to co-localize with the DAPI-stained cell nuclei (Fig. 3, A and B). As expected, BSC-40 primate cells did not react with the anti-PesC antibodies demonstrating that the nuclear labeling seen in NIH3T3 cells accurately identified the location of pescadillo protein and was not due to any cross reactivity of the antibody (Fig. 3, C and D). At higher magnification it was clear that pescadillo was distributed in distinct domains throughout the nucleus (Fig. 3E). The largest of these domains resembled nucleoli and so we compared the location of pescadillo to nucleolin, a characteristic nucleolar protein. Fig.  3, E-H show that both pescadillo and nucleolin co-localize in the largest spots, indicating that pescadillo is present within the nucleoli. This is consistent with previous work that found pescadillo in the nucleoli of HeLa cells (4). Of note, in addition to the prominent nucleolar staining, we also identified several smaller dots within the nucleoplasm that reacted solely with the anti-PesC antibody indicating they were distinct from the nucleolus (Fig. 3

, E-H).
We initially used an asynchronous cell population in our analysis of pescadillo distribution. However, we noted that pescadillo was still present, although nucleoli are absent, in mitotic cells. We therefore analyzed the location of pescadillo protein at various stages of mitosis in a population of NIH3T3 cells made synchronous by release from serum starvation (Fig.  4). As the chromosomes condensed during mitosis, pescadillo protein appeared as nebulous accumulations within the dissi- pating cell nucleus (Fig. 4, A and E). These regions of intense pescadillo staining did not coincide with the centrosomes or spindle apparatus, as was demonstrated in Fig. 4E by costaining prophase cells with anti-␤-tubulin antibodies. Fig. 4, B and F represent a polar view of the metaphase plate showing condensed chromosomes stained with DAPI, anti-PesC, and anti-␤-tubulin antibodies. In metaphase cells pescadillo appeared to closely associate with the condensed chromosomes, localizing to distinct stripes that appeared to coat the outside of the chromosomes. This staining of the chromosome periphery persisted during anaphase, although additional distinct spots of pescadillo staining could also be seen in the vicinity of the spindle poles (Fig. 4, C and G). As the nuclear envelope reformed during telophase these spots were localized to the cytoplasm (Fig. 4, D and H). Punctate staining could also be seen within the nucleus during cytokinesis, although this staining was relatively less intense compared with that of the cytoplasmic dots (Fig. 4, D and H).
The dynamic distribution of pescadillo during mitosis was reminiscent of several proteins and rRNAs with roles in nucleolar assembly and ribosome biogenesis (19,20). The nucleolus is disassembled early in mitosis and then reformed during late mitotic stages in a process that is dependent upon repression and the subsequent re-activation of RNA polymerase I-mediated transcription (21,22). During this procedure molecules involved in ribosome biogenesis and nucleolar assembly distribute to defined locations within the mitotic cell. Some localize to the periphery of chromosomes (perichromosomal sites) early in mitosis, then become associated with prenucleolar bodies (PNBs) that are present within nascent nuclei during telophase (20,21,(23)(24)(25)(26). Others are present within discrete cytoplasmic particles called nucleolus-derived foci (NDF) that arise transiently during anaphase and disappear by G 1 (20). To determine whether the distinct spots of pescadillo staining in the telophase nucleus and cytoplasm were PNBs and NDFs, respectively, we co-stained mitotic cells using anti-nucleophos- During metaphase and anaphase pescadillo was seen to associate with the periphery of condensed chromosomes (white arrow) and was found in cytoplasmic nucleolus-derived foci (cyan arrow) and in prenucleolar bodies (bar) in the nucleus during telophase. p, pescadillo; d, DAPI; t, ␤-tubulin; b, nucleophosmin. min (also called B23) antibodies, which recognize these nucleolar-related structures (27). Fig. 4, I-L shows that anti-nucleophosmin and anti-PesC antibodies recognize the same structures within cells during telophase. Overall, we conclude that pescadillo is associated with all structures, nucleoli, decondensing chromosomes, NDFs, and PNBs, which have roles in reassembly of the nucleolus following mitosis.
Pes Ϫ/Ϫ Embryos Arrest during Preimplantation Stages of Development-The data presented above suggested a role for pescadillo in controlling cell proliferation, possibly by being required for ribosome biogenesis or nucleolus formation. To directly test this proposal we attempted to generate pescadillo (pes) Ϫ/Ϫ mice using the strategy shown in Fig. 5. A successful gene-targeting event in ES cells was predicted to replace the final 158 bp of exon 5 through 110 bp of exon 12 with a cassette containing a neomycin phosphotransferase gene expressed from the phosphoglycerate kinase promoter (pgk-neo). Southern blot analysis identified 2 ES cell clones, one of which is shown in Fig. 5B that contained a correctly targeted pescadillo allele. Chimeric mice were produced from both cell lines but only one gave rise to chimeras that transmitted the mutant pescadillo allele through their germline. pes ϩ/Ϫ mice were viable, having lifespans in excess of 19 months, and exhibited no obvious abnormal phenotype.
If pescadillo were essential for cell proliferation, we predicted that pes Ϫ/Ϫ embryos would be unlikely to survive past the earliest stages of development. We, therefore, interbred pes ϩ/Ϫ mice and determined the genotype of offspring and postgastrulation stage embryos by Southern blot analysis. Table I shows that of 126 offspring 41 were pes ϩ/ϩ and 85 were pes ϩ/Ϫ . No pes Ϫ/Ϫ mice were recovered, demonstrating that pescadillo is essential for embryonic development. Analysis of embryos collected from 7.5-10.5 d.p.c. (not shown) found no pes Ϫ/Ϫ embryos nor sites of re-adsorption. We next developed a PCR-based assay to genotype preimplantation stage embryos. Fig. 6A shows that oligonucleotide primers designed to hybridize to pgk-neo or flanking genomic pescadillo nucleotide sequences (Fig. 5A) could be used in a PCR to differentiate between pes ϩ/ϩ , pes ϩ/Ϫ , and pes Ϫ/Ϫ embryos. Of 182 blastocyst stage embryos collected 49 were pes ϩ/ϩ , 109 were pes ϩ/Ϫ , and 24 were pes Ϫ/Ϫ . Similarly, of 104 embryos at the morula stage of development 31 were pes ϩ/ϩ , 57 were pes ϩ/Ϫ , and 16 were pes Ϫ/Ϫ . These results demonstrate that, while pes Ϫ/Ϫ morulae could be identified, the frequency at which they were recovered was non-Mendelian indicating that approximately half of the pes Ϫ/Ϫ embryos arrested at an even earlier stage of development.
Comparisons of pes ϩ/ϩ , pes ϩ/Ϫ , and pes Ϫ/Ϫ morulae, shown in Fig. 6B, revealed no striking morphological differences, although often the blastomeres of the mutant morulae appeared to be less organized than in control embryos. Morulae of all genotypes appeared to contain the same number of blastomeres and had successfully undergone compaction (the process by which the outermost blastomeres of the morula become polarized). However, striking differences were observed between control and pes Ϫ/Ϫ embryos as they formed blastocysts. At 3.5 d.p.c. control embryos had progressed from a compacted morula to a blastocyst containing an identifiable inner cell mass, trophectoderm, and blastocoel cavity. In contrast, all pes Ϫ/Ϫ embryos examined (n ϭ Ͼ50) contained only 8 -16 cells, which is the number of cells seen in wild-type morulae at 2.5 d.p.c., and none of the pes Ϫ/Ϫ embryos had formed blastocysts. While the 3.5 d.p.c. pes Ϫ/Ϫ embryos strongly resembled morulae the individual blastomeres appeared disorganized, often being misshapen and having a granular appearance (Fig.  6B). We saw no significant increase in TUNEL staining in the mutants compared with controls (data not shown) suggesting that apoptotic cell death of the blastomeres was not a major cause of the phenotype. We, therefore, conclude that the primary defect associated with loss of pescadillo is a failure of the blastomeres to proliferate past morula stages of development.
To begin to understand why proliferation was blocked in pes Ϫ/Ϫ blastomeres, we characterized the expression and localization of the pescadillo protein during the development of wild-type preimplantation embryos using confocal microscopy (Fig. 7). Prior to the 8-cell stage, expression of pescadillo could be faintly identified in ϳ50% of embryos examined using anti-PesC antibodies. In interphase blastomeres of 2-and 4-cell stage embryos pescadillo expression was restricted to a subnuclear band that encircled one or more large vacuoles present within the nucleus (Fig. 7, A-D). These dense vacuoles, which are believed to give rise to mature functional nucleoli during embryonic development, have been described as nucleolus precursor bodies or fibrillar primary nucleoli (28 -30). Similar patterns of localization around the periphery of nucleolus precursor bodies have been described for other nucleolar proteins, such as fibrillarin, nucleolin, and nucleophosmin in preimplantation mouse embryos as well as in similar staged embryos of other species (29,31,32). In contrast to our findings in NIH3T3 cells, in embryos prior to the 8-cell stage, pescadillo protein was  seen to co-localize with the spindle apparatus in dividing blastomeres and was not associated with the condensed chromosomes (Fig. 7, A-C). At the 8-cell stage pescadillo was detected in all embryos examined. Moreover, the intensity of the antipescadillo staining was seen to increase and its boundaries had expanded to occupy a greater volume within the nucleus (Fig.   FIG. 6. Phenotype of pes ؊/؊ mouse embryos. A, PCR genotyping of 3.5 d.p.c. embryos from pes ϩ/Ϫ intercrosses. Following microscopy, DNA was extracted from embryos and pescadillo or neo-specific DNA fragments were amplified using primers Pes62, Pes3, and NeoG (shown in Fig. 5, described under "Experimental Procedures"). Pescadillo primers amplified a 138-bp DNA fragment from pes ϩ/ϩ and pes ϩ/Ϫ embryos but not from pes Ϫ/Ϫ embryos. Neo primers amplified a 281-bp fragment from both pes ϩ/Ϫ and pes Ϫ/Ϫ embryos but not from wild-type embryos. B, embryos were recovered from pes ϩ/Ϫ crosses at 2.5 d.p.c. (morulae) or 3.5 d.p.c. (blastocyst), and the genotype determined to be wild type (ϩ/ϩ), heterozygous (ϩ/Ϫ) or null (Ϫ/Ϫ) for pescadillo by PCR of genomic DNA following photography. The zona pellucida was removed for clarity and embryos fixed prior to microscopy. pes Ϫ/Ϫ morulae appear similar to control embryos with evenly sized blastomeres (arrow) and well-defined cell-cell borders. At 3.5 d.p.c. pes Ϫ/Ϫ embryos are easily distinguished from control embryos which have undergone cavitation to form an inner cell mass (ϩ), trophectoderm cell layer (solid line), and blastocoel cavity (*). In contrast, pescadillo-null embryos have not formed blastocysts and appear disorganized with unevenly sized blastomeres (arrows). conjugated secondary antibody. Embryos were counterstained with propidium iodide (red) to visualize DNA and examined by confocal microscopy. Both wild type (ϩ/ϩ) and pes (Ϫ/Ϫ) morula were examined. Panels show staining for pescadillo (P), propidium iodide (PI), or both (p ϩ PI). Pescadillo protein is associated with the mitotic spindle in mitotic 2-cell embryos (white arrow) but not the chromosome periphery (bar) (A-C). In interphase blastomeres pescadillo localizes to the periphery of nucleoli in all stages examined (D, E, and G-I). Nucleolar staining was not detected in pes Ϫ/Ϫ embryos (F). By blastocyst stage of development (G-I), pescadillo could be detected around the chromosomes in mitotic cells (cyan arrow) and in nucleoli as was observed in NIH3T3 cells. 7E). Although the nucleolus precursor bodies were still present at this stage they were diminished in size and were, in fact, undetectable by the blastocyst stage of development. We believe that this expansion of pescadillo localization reflects the maturation of the nucleolus and coincides with an increase in ribosome biogenesis that is observed in 8-cell stage embryos (28,30,33). We also noted the presence of additional intense fluorescence surrounding the periphery of the cell. However, Fig. 7F shows that this staining remained in pes Ϫ/Ϫ embryos in which all nucleolar staining was lost. This implies that the peripheral staining is likely due to artifactual antibody interactions with denatured zona pellucida glycoproteins. At the blastocyst stage of development the localization of pescadillo resembles that found in NIH3T3 cells. Fig. 7, G-I, shows that strong anti-pescadillo staining was identified in nucleoli of interphase cells and was associated with chromosomes in dividing cells. In sum, pescadillo is expressed at the onset of nucleologenesis in structures that are believed to give rise to the nucleolus.
Nucleologenesis and Ribosome Biogenesis Is Disrupted in Embryos Lacking Pescadillo-The nucleolus is the major site of ribosome biogenesis in the cell and is formed during specific stages of preimplantation development. Indeed, the formation of a fully differentiated nucleolus is defined by the act of generating ribosomes, and this is reflected by changes in nucleolar architecture during nucleologenesis (21,34). The architecture of an active nucleolus can be divided into three major structures: the fibrillar component (FC) that contains proteins involved in rDNA transcription, the dense fibrillar component (DFC) that contains newly synthesized rRNAs, and the granular component (GC) that consists of preribosomal particles (30). It has been proposed that initiation of rDNA transcription begins between the 2-4-cell stage of mouse development (28,33). Autoradiographic labeling studies have shown that during these early stages of development rDNA transcription is confined to the outermost reticulated layer of the differentiating nucleolus whereas the dense inner layer of the prenucleolar body contains nucleolar matrix proteins (33). Between 8-cell and blastocyst stages the extent of active fibrillo-granular nucleolar material increases, progressing toward the center of the nucleolus precursor body. The observation that pescadillo is restricted to the periphery of the nucleolus precursor body of 2-8 cell stage embryos suggests that it is associated with regions of the nucleolus in which rDNA is being actively transcribed. Consistent with this model we observe that the localization of pescadillo tracks the expansion of the fibrillo-granular regions of the differentiating nucleolus (Fig. 7). We, therefore, addressed whether pescadillo was essential for formation of mature functional nucleoli by comparing nucleolar structure in mutant and control embryos using electron microscopy.
Embryos were collected at the blastocyst stage of development (3.5 d.p.c.) and processed for electron microscopy. We were unable to genotype embryos that had been prepared for EM, so we classified embryos as control or mutant based on the phenotypic differences shown in Fig. 6. Fig. 8 shows that control embryos had formed nucleoli containing reticulated structures characteristic of an active nucleolus. FC, DFC, and GC regions were all clearly recognizable (Fig. 8A). In some cases an electron dense fibrillar sphere, believed to be the remnant of the nucleolus precursor body, was identifiable within the nucleolus. In contrast to controls, differentiation of the nucleolus in mutant embryos was dramatically abnormal (Fig. 8B). Although spherical masses of presumptive nucleolus precursor bodies were discernable, formation of fibrillogranular structures indicative of ribosome biogenesis remained rudimentary.
In addition, the mutant blastomeres contained multiple smaller electron dense spheres that were closely associated with the periphery of the nucleus. Although we have not positively identified these structures, we propose that they exist as a consequence of aberrant nucleolar differentiation. In addition to the differences within the nucleus, we observed that the cytoplasm of mutants appeared sparse compared with controls. Under higher magnification we could discern clusters of ribosomes throughout the cytoplasm of control embryos that were undetectable in the mutants (Fig. 8, C and D). In order to quantify the relative number of ribosomes in mutant and control embryos we followed the procedure of Piko and Clegg (13). Ribosomes were counted in eleven mutant and twelve normal cells using two micrographs/cell, each of which covered 15 m 2 of cytoplasm. Control embryos contained an average of 253 ribosomes/15 m 2 whereas the average number of ribosomes in the same area of cells from pescadillo mutant embryos was only 11. This represented a striking 23-fold reduction in the number of ribosomes found in the mutant embryos compared with wild type. These cumulative data demonstrate conclusively that pescadillo is required for ribosome biogenesis and, as a consequence, formation of the nucleolus during mammalian embryonic development. DISCUSSION A Role for Pescadillo in Nucleologenesis and Ribosome Biogenesis-In the current study we have shown that the mammalian pescadillo protein associates with several structures that are associated with nucleologenesis. Furthermore, we FIG. 8. Nucleologenesis and formation of ribosomes is disrupted in embryos lacking pescadillo. 3.5 d.p.c. embryos from crosses between pes ϩ/Ϫ mice were collected, separated based on normal or mutant phenotype, and processed for electron microscopy. Blastomeres of normal embryos (A) have mature nucleoli with differentiated nucleolar regions including fibrillar components (FC), dense fibrillar components (DFC), and granular components (GC). Occasional fibrillar spheres (FS), which are presumed to be remnants of the nucleolus precursor body (NPB) were seen in the center of the fibrillo-granular nucleolar material. Blastomeres of mutant embryos (B and D) failed to form well-differentiated nucleoli although electron dense material that was presumed to be the NPB was readily identifiable as well as the presence of electron dense spheres (DS) that lay close to the nuclear periphery. At high magnification (C and D; ϫ55,000) ribosomes could be identified in control embryos (arrows) but were sparse in mutant embryos. demonstrated that loss of pescadillo function in mice results in embryonic lethality due to a disruption in ribosome biogenesis. This is consistent with recent findings in yeast, discussed below, which have also implicated pescadillo in the generation of the 60 S ribosomal subunit (8 -10). Although the nucleolus was identified over 150 years ago, the molecular mechanisms through which it is formed and how it controls formation of ribosomes is only now beginning to be understood. Part of the difficulty in unraveling nucleolar mechanisms is the surprising complexity of this subnuclear compartment. Recently, large scale proteomic analyses using MALDI-TOF spectroscopy of nucleoli purified from HeLa cells identified 271 proteins (35). 80 of these proteins were novel or previously uncharacterized, confirming that the make-up of the nucleolus is extremely complex. Because the major role of the nucleolus, indeed its raison d'etre is to synthesize preribosomal particles it seems likely that many of these proteins will be involved in ribosome biogenesis (36). However, a number of nucleolar functions that appear to be unrelated to ribosome biogenesis have recently been discussed (21,37). These include the processing and export of a subset of mRNAs and processing of non-rRNAs such as signal recognition particle RNA, telomerase RNA, and U6 snRNA, all of which are components of ribonucleoprotein catalytic complexes (38 -41). While we cannot exclude a possible role for pescadillo in such non-ribosomal associated functions we believe that several pieces of data imply a direct and active role for pescadillo in ribosome formation.
First, the immunostaining experiments using anti-pescadillo antibodies described in this report show that the pescadillo protein not only associates with the nucleolus but is present in several structures that are believed to be involved in the assembly of the nucleolus during both mitosis and embryogenesis. During mitosis the nucleolus disassembles and, while the rRNA transcriptional machinery remains closely associated with rDNA in areas called nucleolar organizing regions, the proteins involved in rRNA processing and formation of ribosomes are redistributed to distinct cellular domains. Some proteins and pre-rRNA sequences are located at the periphery of chromosomes where they "piggy-back" into the daughter cells. During anaphase these proteins migrate from the chromosome periphery to form the PNBs, which reside within newly formed nuclei at the end of mitosis. Others are found in the NDFs that transiently appear within the cytoplasm late in mitosis; proteins from these NDFs are believed to subsequently enter the nucleus where they join with PNBs and nucleolar organizing regions. Although the functional significance of mitotic sites of nucleolar proteins is still to be established, it has been proposed that they participate in the reassembly of the nucleolus at the end of mitosis by contributing preribosomal material toward nascent nucleoli in the daughter cells (20). In this regard they can be thought to "seed" the daughter cells with the essential elements needed to generate new ribosomes in an epigenetic manner. The fact that pescadillo is found associated with PNBs and NDFs, structures that are crucial for the onset of ribosome formation, therefore implies that pescadillo is an integral component of the ribosome-generating machinery. In addition to pescadillo, several other factors with known roles in ribosome biogenesis are found in PNBs, the chromosome periphery and NDFs. These include nucleophosmin (B23), which has ribonuclease and molecular chaperone activities, fibrillarin, which is implicated in pre-rRNA processing and ribosome assembly, and U3 and U8 snoRNAs (19,27,(42)(43)(44)(45)(46). It will be of interest, and instructive, to determine whether and how pescadillo interacts with these other components to facilitate ribosome synthesis.
Further evidence supporting a direct role for pescadillo in ribosome biogenesis comes from recently published studies of the yeast pescadillo homologue (YPH1, Nop7p) in S. cerevisiae (8 -10). The ability to combine biochemical and genetic approaches has led to a greater understanding of ribosome biogenesis in yeast compared with mammalian systems. It also appears that many of the proteins involved in yeast ribosome biogenesis are evolutionarily conserved, and so it is likely that information garnered from the study of yeast is directly applicable to higher species. Recently, Adams et al. (9) identified the yeast homologue of pescadillo (YPH1, Nop7p) in a synthetic lethal screen for proteins that could affect ribosome biogenesis. Using a temperature sensitive pescadillo allele the investigators found that accumulation of 60 S ribosomal subunits declined compared with 40 S ribosomal subunits when the yeast were grown at the restrictive temperature. Analyses of pre-rRNA accumulation found that processing of the 27 S rRNA was delayed in yeast lacking pescadillo (9). The defect in 27 S rRNA processing was characterized further by the Tollervey laboratory who found that pescadillo functions at multiple steps in the processing of the 60 S ribosomal subunits (10). However, they demonstrated that loss of pescadillo function specifically resulted in inhibition of 5Ј-3Ј exonuclease digestion of 27 S rRNA, which is necessary to form the 5Ј-end of mature 5.8 S rRNA (10). Consequently, yeast lacking pescadillo function failed to accumulate mature 5.8 S rRNA. In addition to a defect in rRNA processing the authors found that yeast pescadillo mutants may have defects in export of 60 S ribosomal subunits from the nucleus into the cytoplasm. Fusion of the ribosomal protein Rpl25 with eGFP allowed the authors to track export of 60 S ribosomal subunits from the nucleus into the cytoplasm. In control cells Rpl25-eGFP accumulated predominantly in the cytoplasm. In contrast, when pescadillo expression was repressed Rpl25-eGFP was retained in the nucleus (10). In line with a requirement for nuclear export, pescadillo was found to associate with a 60 S pre-ribosomal particle that has been proposed to act as a nucleoplasmic transport complex containing several non-ribosomal proteins with roles in exporting ribosomes from the nucleus (47). The possibility that pescadillo is associated with such an export particle is also consistent with our identification of pescadillo protein in the nucleoplasm of NIH3T3 cells (Fig. 3). A question that was raised by the study of the yeast pescadillo homologue was whether the findings would be applicable to higher eukaryotes. This was pertinent because loss of pescadillo appears to insignificantly affect proliferation of many cells and tissues in zebrafish (2). Our findings show that pescadillo function appears to be conserved between mice and S. cerevisaie suggesting that findings in yeast will be applicable to mammals.
A Link to Cell Proliferation-We have shown that pescadillo is highly expressed in proliferating cell types and that its expression is induced as cells enter the cell cycle. We have also demonstrated that loss of pescadillo function results in a disruption of nucleologenesis, which raises the question of whether there exists a link between nucleolar function (i.e. ribosome biogenesis) and cell proliferation. Recent studies have shown that ribosome biogenesis may be linked to a cell cycle checkpoint that controls passage through S-phase. In one case conditional ablation of the gene encoding the 40 S ribosomal protein S6 in livers of adult mice resulted in a block to biogenesis of 40 S ribosomal subunits (48). When partial hepatectomies were performed on these mice the cells of the liver remnant failed to proliferate in the absence of S6 protein. Analyses of cyclin gene expression showed that while cyclin D1 activity increased in response to hepatectomy, there was a failure to induce cyclin E and cyclin A expression (48). These data indi-cate that mammalian cells maintain a cell cycle checkpoint that senses the status of ribosome levels during the cell cycle. Recently, studies of a non-ribosomal nucleolar protein called Bop1 have supported and extended this model. Bop1 was identified in a genetic screen for cDNAs that caused a reversible perturbation of cell cycle progression (49). Inhibition of Bop1 activity was found to prevent processing of 5 S and 28 S rRNAs and the consequent formation of 60 S ribosomal subunits. Proliferation of these cells arrests in G 1 phase of the cell cycle (50,51). Interestingly, when p53 is inactivated, cell cycle progression is restored in the absence of Bop1 although ribosome biogenesis is still deficient (52). These results imply that the G 1 /S-phase checkpoint arrest, which is induced through a deficiency in the assembly of ribosomes, is mediated by p53. Biochemical studies of pre-ribosomal particles have shown that the yeast homologues of both pescadillo and Bop1 can be readily co-purified suggesting that they may be intimately associated. This being the case we propose that, like Bop1, the failure of cells to proliferate in the absence of pescadillo is caused by a deficit in the formation of ribosomes, which results in nucleolar stress and consequently induces a cell cycle checkpoint arrest.
The mechanisms linking ribosome biogenesis to control of the cell cycle remain obscure. However, recent biochemical studies from Du and Stillman (8) have shown that the yeast pescadillo homologue associates with the origin of recognition complex (ORC). ORC acts as a scaffold upon which a variety of proteins assemble during different cell cycle stages to influence cell cycle progression and DNA replication (8). The authors, therefore, suggest that in yeast pescadillo may directly link cell growth to cell division through the action of pescadillo. How pescadillo could coordinate these processes will clearly be an important line of future study that will be aided by the availability of pes Ϫ/Ϫ embryos.