p27BBP/eIF6 is an evolutionarily conserved protein that was originally identified as p27BBP, an interactor of the cytoplasmic domain of integrin β4 and, independently, as the putative translation initiation factor eIF6. To establish the in vivo function of p27BBP/eIF6, its topographical distribution was investigated in mammalian cells and the effects of disrupting the corresponding gene was studied in the budding yeast, Saccharomyces cerevisiae. In epithelial cells containing β4 integrin, p27BBP/eIF6 is present in the cytoplasm and enriched at hemidesmosomes with a pattern similar to that of β4 integrin. Surprisingly, in the absence and in the presence of the β4 integrin subunit, p27BBP/eIF6 is in the nucleolus and associated with the nuclear matrix. Deletion of the IIH S. cerevisiae gene, encoding the yeast p27BBP/eIF6 homologue, is lethal, and depletion of the corresponding gene product is associated with a dramatic decrease of the level of free ribosomal 60S subunit. Furthermore, human p27BBP/eIF6 can rescue the lethal effect of the iihΔ yeast mutation. The data obtained in vivo suggest an evolutionarily conserved function of p27BBP/eIF6 in ribosome biogenesis or assembly rather than in translation. A further function related to the β4 integrin subunit may have evolved specifically in higher eukaryotic cells.
Integrins are heterodimeric receptors formed by the assembly of one α and one β transmembrane subunit. 16 α and 8 β subunits heterodimerize to produce >20 receptors that are differentially expressed in a wide variety of tissues. Most integrins bind components of the extracellular matrix and can organize the cytoskeleton through their cytoplasmic domain (for review see Hynes, 1992). Moreover, integrins are involved in the control of cell growth, apoptosis, and differentiation through the recruiting of several signal transduction and adaptor molecules (for review see Clark and Brugge, 1995; Giancotti, 1997; Howe et al., 1998). The steadily increasing role of integrins in the organization of cellular metabolic processes has recently encompassed the process of protein synthesis. Indeed, clustering of some integrin subunits at points of focal adhesion results in the relocation of mRNA and ribosomes to focal adhesion contacts (Chicurel et al., 1998). Although the molecular mechanisms of this process are unknown, a picture in which most biochemical processes are spatially regulated, with integrins playing a major role, has emerged.
The integrin subunit β4 associates with α6 to form a multivalent laminin receptor present at high levels in most epithelia. Its ligand engagement correlates with PI3 kinase activation (Shaw et al., 1997) and recruitment of the shc and grb2 adaptors (Mainiero et al., 1995). In addition, in squamous and transitional epithelia, β4 is required for the formation of hemidesmosomes, specialized structures providing firm mechanical links between basal lamina and the intermediate filament cytoskeleton (for review see Giancotti, 1996). Loss of function of β4 both in human and mice results in hemidesmosome disruption, blistering, and perinatal death (Vidal et al., 1995; Dowling et al., 1996; van der Neut et al., 1996).
Mutations in the functional cytodomain of β4 result in an inability to translocate into hemidesmosomes and be targeted to the intermediate filament cytoskeleton (Spinardi et al., 1993; Niessen et al., 1997) accompanied by lethal forms of the blistering disease junctional epidermolysis bullosa associated with pyloric atresia (Vidal et al., 1995). These and other data strongly indicate that the cytodomain of β4 exerts its function through the interaction with cytoplasmic molecules led us to search for protein interactors of the β4 cytodomain. Through an extensive yeast two hybrid analysis, a previously unknown peptide named p27BBP (BBP for beta4 binding protein)1 that binds the β4 cytodomain was discovered. p27BBP directly binds, in vitro and in vivo, a 300–amino acid long stretch of β4 integrin cytodomain, a region required for targeting β4 to the hemidesmosomes and to the intermediate filament cytoskeleton as determined by genetic studies. In addition, p27BBP was found to be present at high levels in the submembrane region of epithelial cells containing β4. Finally, the biochemical association of p27BBP with keratin intermediate filaments, suggested that p27BBP might be the molecular link between β4 and the cytoskeleton (Biffo et al., 1997). The precise ultrastructural localization of p27BBP, in in vivo hemidesmosomes was not yet defined.
The finding that p27BBP homologues exist both in yeast and Drosophila, in which β4 integrin homologues are absent (Biffo et al., 1997) suggested that p27BBP might also have a β4-independent function. Consistently, the cloning of a human cDNA encoding a protein named eIF6 (identical to p27BBP) was almost concomitantly obtained by Si et al. (1997). The biological assay used to clone eIF6 was based on its in vitro ability to inhibit the association between the 40S and the 60S ribosomal subunits, and was not related to integrin function. On the basis of its in vitro determined properties, it was suggested that eIF6 might act as a translation initiation factor. The cloning and sequencing of eIF6 unequivocally indicate that eIF6 and p27BBP are the same protein (Biffo et al., 1997; Si et al., 1997). Much more recently, eIF6/p27BBP has also been identified by another group as a gene induced in mast cells by allergic reaction (Cho et al., 1998). To acknowledge the independent identification of p27BBP and eIF6, the protein will be denoted as p27BBP/eIF6 throughout this work.
Both studies left a set of unresolved questions: (a) Which is the precise cellular localization of p27BBP/eIF6 and is it modulated by the presence of β4 integrin?; (b) Is the association of p27BBP/eIF6 with the intermediate filament cytoskeleton a unique feature of cells that contain β4?; (c) Is p27BBP/eIF6 present in hemidesmosomes?; and (d) Which is the general, evolutionarily conserved, function of p27BBP/eIF6? To address these questions, we used integrated approaches. First, the fine localization of p27BBP/eIF6 was studied in cell lines, either containing β4 integrin or not. Our studies show that p27BBP/eIF6 is a nuclear matrix-associated protein present in the nucleolus of all cells analyzed and enriched at the basal membrane of β4 expressing epithelial cells. Second, the function of p27BBP/eIF6 was addressed in Saccharomyces cerevisiae by constructing and characterizing a null mutant. The yeast p27BBP/eIF6 homologue is essential for cell viability and its depletion results in an abnormal ribosomal profile, with a dramatic reduction of the levels of free 60S ribosomal subunits. Taken together these data indicate that the conserved role of p27BBP/eIF6 is linked to 60S ribosome subunit metabolism, and that this process may be linked to the nuclear matrix. In higher organisms, novel functions of p27BBP/eIF6 may have appeared that link this molecule to epithelial adhesion.
Materials and Methods
Antibodies and Cell Lines
The rabbit polyclonal antiserum against the COOH-terminal peptide of p27BBP/eIF6 (NH2-CTIATSMRDSLIDSLT-COOH) was tested for its specificity by Western blotting and immunoprecipitation both on the recombinant protein and on cellular lysates (Biffo et al., 1997). Integrin β4 was detected with the rat mAb3E1 (10 μg/ml; Chemicon International, Inc.), or with the mouse mAb AA3 (Kajiji et al., 1989) at 10 μg/ml (gift of Vito Quaranta, Scripps Research Institute, La Jolla, CA). The human autoantibodies against fibrillarin (Ochs and Smetana, 1991) were a generous gift of Robert Ochs (Scripps Research Institute) and were diluted 1:300. Cytokeratins were detected either with mouse monoclonal anticytokeratin 8:18, IgG2a (Diagnostika) at 1:200, or with mouse monoclonal anticytokeratin 7/17 IgG1, according to the manufacturer's protocol (C46; Euro-Diagnostica). Secondary antibodies were rhodamine- and fluorescein-tagged swine anti–rabbit IgGs (1:50; DAKO Corp.), rhodamine-tagged goat anti–human IgGs (10 μg/ml; Chemicon International, Inc.), rhodamine-tagged goat anti–mouse IgGs (7.5 μg/ml; Molecular Probes Europe) and fluorescein-tagged goat anti–mouse IgGs (1:50; Antibodies Inc.). In control experiments, primary antibodies were replaced by preimmune sera or irrelevant mAbs. In addition, the p27BBP/eIF6 antiserum was preadsorbed with the peptide used for its generation (1 μM, overnight, 4°C), or with the bacterially produced human recombinant full length protein (at 10 μg/ml, 2 h at 4°C) purified by ion exchange chromatography.
The cell lines and primary cells used in this study, as well as the conditions for their propagation, are described in the American Type Culture Collection cell line catalogue or in the references between parentheses. They are as follows: mouse NIH/3T3 fibroblasts, human A431 epidermoid carcinoma, human HeLa epitheloid carcinoma, human pancreatic carcinoma FG2 (Kajiji et al., 1989), human Jurkat T cells, transformed human keratinocytes HaCat (Boukamp et al., 1988), human insulinoma cells Rin2A (Rouiller et al., 1990), and human neuroblastoma SK-N-MC. The 804G rat epithelial cell line clone A was a gift of F. Giancotti (Memorial Sloan-Kettering Cancer Center, New York) and has been described in Spinardi et al. (1993).
Mouse resting splenocytes, human fibroblasts from the umbilical cord, and Xenopus oocytes were gifts of A. Cabibbo, E. Bianchi, and E. Pannese (all at DIBIT, Milano, Italy) and were obtained by standard procedures.
Cells were treated with actinomycin D (Boehringer Mannheim GmbH) at the final concentration of 5 μg/ml for 1, 4, and 12 h, washed, and fixed as described. In some experiments cells were allowed to recover after treatment by switching them to their normal medium.
Electron Microscopy on Human Amnion
Human fresh amniotic membrane (obtained immediately upon delivery from the Department of Obstetrics, San Raffaele Hospital, Milano, Italy) was dissected and pieces of tissue were fixed with 4% paraformaldehyde and 0.25% glutaraldehyde in 125 mM sodium phosphate buffer, pH 7.4, for 45 min at 4°C. The samples were infiltrated with polyvinylpyrrolidone and frozen in a 3:1 (vol/vol) mixture of propane and isopentane cooled with liquid nitrogen. Ultrathin cryosections (50–100 nm thick) were obtained using an Ultracut ultramicrotome equipped with a Reichert FC4 cryosectioning apparatus. The cryosections were processed as described in Villa et al. (1993) using the rabbit anti-p27BBP/eIF6 antiserum and the mouse monoclonal anticytokeratin 7/17 IgG1. Cryosections were examined in an electron microscope (Hitachi H-7000).
Western Blot Analysis
All samples were denatured before loading in Laemmli buffer (Laemmli, 1970) and run on denaturing 12% SDS–acrylamide gel, transferred to Immobilon P membranes (Millipore Corp.), and blotted with the rabbit p27BBP/eIF6 antiserum at 1:1,000 dilution as previously described (Biffo et al., 1997). In the control of the fractionation experiment, a mouse monoclonal anticytokeratin 8/18 IgG2a at 1:200 was used. Detection was always performed by the commercially available chemiluminescence detection system (ECL) technique (Nycomed Amersham).
Extraction of Nuclear Matrix and Ribosomal Proteins
Intermediate filaments/nuclear matrix filaments fractions were prepared exactly according to He et al. (1990). Briefly, all the soluble proteins, the nonintermediate filament cytoskeleton, DNA associated proteins, and proteins loosely associated with the nuclear matrix proteins were removed by sequential washes in buffers (Triton X-100, 250 mM ammonium sulphate, DNase I, and 2 M NaCl). At the end of this procedure, a cytoplasmic and nuclear intermediate filament network containing keratins, lamins, and intermediate filament-associated proteins was left. The efficiency of the extraction was routinely controlled by DNA staining, or by immunostaining for keratins.
Preparation of ribosomes was performed through established procedures and exactly as described in Madjar (1994). 804G clone A cell line monolayer was washed and scraped with cold PBS. The pellet was resuspended in cold buffer A (0.25 M sucrose, 25 mM KCl, 5 mM MgCl2, 50 mM Tris-HCl, pH 7.4), stirred slowly with a vortex, while adding NP-40 to a final concentration of 0.7%, and kept on ice for 10 min. The suspension was centrifuged at 750 g for 10 min at 4°C and the resulting pellet containing nuclei and insoluble proteins was resuspended in Laemmli buffer for biochemical analysis. The supernatant was centrifuged at 12,500 g, 10 min at 4°C and the resulting pellet containing mitochondria was resuspended in Laemmli buffer. The low-speed supernatant was added to 0.32 vol of buffer C (0.25 M sucrose, 2 M KCl, 5 mM MgCl2, 50 mM Tris-HCl, pH 7.4), layered on top of a 1 M sucrose cushion, and ultracentrifuged (TL100; Beckman Instruments, Inc.) at 245,000 g, 2 h at 4°C. The high-speed pellet containing ribosomal proteins was resuspended in Laemmli buffer. The high-speed supernatant was precipitated with cold 10% TCA, for 45 min on ice, centrifuged at 14,000 rpm at 4°C, and resuspended in Laemmli buffer for biochemical analysis.
Indirect Immunofluorescence and Immunocytochemistry
Immunofluorescence was performed as previously reported (Marchisio et al., 1991). In brief, the following was performed: cell monolayers were fixed in 3% paraformaldehyde in PBS, pH 7.6, containing 2% sucrose for 10 min at room temperature; permeabilized in Hepes–Triton X-100 buffer for 5 min at 4°C (20 mM Hepes, 300 mM sucrose, 50 mM NaCl, 3 mM MgCl2, 0.5% Triton X-100, pH 7.4); and blocked with 5% BSA in PBS for 30 min at room temperature. Next, the cells were incubated in primary antibodies (diluted in 5% BSA in PBS) for 2 h at room temperature, washed in 0.2% BSA in PBS, and treated with secondary antibodies diluted in PBS. Staining for F-actin was performed with 200 nM fluorescein-labeled phalloidin (Sigma Chemical Co.) for 20 min at 37°C in the dark and with 2 μg/ml DNA counterstaining (Hoechst 33342; Sigma Chemical Co.). Once mounted in Mowiol 4-88 (Hoechst AG), coverslips were analyzed with a confocal microscope (MRC-1024; Bio-Rad Laboratories) equipped with a krypton/argon laser. To reduce bleed through, double-label confocal images (XY and XZ sections) were acquired sequentially. Micrographs were taken using either a Focus Imagecorder Plus (Focus Graphics Inc.) on Kodak film or a Professional color Point II dye sublimation printer (Seiko). Stained cells were observed in parallel with a Zeiss Axiophot microscope equipped for epifluorescence and a 63× planapochromatic lens; pictures were taken on Kodak T-MAX 400 films exposed at 1000 ISO and developed at 1600 ISO in T-MAX developer for 10 min at 20°C.
In some experiments, p27BBP/eIF6 was revealed by immunoperoxidase labeling using the avidin–biotin amplification method (ABC Kit Vectastain; Vector Labs Inc.). Briefly, after incubation with the primary antiserum cell monolayers were washed and treated with a goat anti–rabbit biotin-conjugated antibody for 30 min at room temperature, followed by the preformed avidin–biotin complex (ABC). The staining was revealed by horseradish peroxidase and 3,3′-diaminobenzidine as chromogen (BioGenex Labs).
Electron Microscopy on Extracted Cells
FG2 cells were extracted to reveal the nuclear matrix as described above and fixed with fresh 3.7% paraformaldehyde in digestion buffer for 30 min at 4°C. They were washed once in digestion buffer, once in TBS-1 (10 mM Tris-HCl, pH 7.7, 150 mM NaCl, 3 mM KCl, 1.5 mM MgCl2, 0.05% Tween 20, 0.1% BSA, 0.2% glycine), and blocked in 5% BSA in TBS-1 for 30 min at room temperature. The cells were incubated in rabbit p27BBP/eIF6 antiserum, 1:200 in 5% BSA in TBS-1 rocking overnight at 4°C. After this incubation, they were sequentially washed in TBS-1, blocked with 5% BSA in TBS-1, 10 min at room temperature, and incubated with 5-nm gold bead-conjugated goat anti–rabbit antibody (Jackson ImmunoResearch Laboratories, Inc.), 1:40 in TBS-2 (20 mM Tris-HCl, pH 8.2, 140 mM NaCl, 0.1% BSA) rocking for 1 h at room temperature. Cells were washed in TBS-1, postfixed in 2.5% glutaraldehyde in 0.1 M cacodylate buffer, pH 7.4, and centrifuged at low speed (600 g for 5 min). The cell pellets were included in diethylene glycol distearate as described by Nickerson et al. (1994). Resinless sections were examined in a Hitachi H-7000 electron microscope.
Homology searches were performed with the Blast programs available through http://www.ncbi.nlm.nih.gov, or by the alerting system of EMBL (http://www.bork.embl-heidelberg.de/Alerting). The accession number of the sequences retrieved are: Homo sapiens, Y11435; S. cerevisiae, Z49919; Caenorhabditis elegans, Z99709; Arabidopsis thaliana, AC003000; Methanococcus jannaschii, U67463; S. acidocaldarius, P38619; Methanobacterium thermoautotrophicum, AE000920; P. bomkoshii, AB009481; and Archaeoglobus fulgidus, AE000961. The alignment was created using the CLUSTAL W algorithm (Thompson et al., 1994). The phylogram of the aligned proteins were produced with the GROWTREE program in GCG with the Jukes-Cantor distance matrix and neighbor-joining method.
Yeast Strains and Media
All strains used are derivatives of W303 (MATa, ade2-1, trp1-1, leu2-3, 112, his3-11,15, ura3, can1–100). Cells were grown in YEP medium (1% yeast extract, 2% bactopeptone, 50 μg/liter adenine) supplemented with either 2% glucose (YEPD) or 2% raffinose and 1% galactose (YEPRG). Transformants carrying the kanMX4 cassette were selected on YEPD plates containing 400 μg/ml G418 (US Biological).
Plasmid Construction and Genetic Manipulations of Yeast Strains
Standard techniques were used for genetic crosses (Rose et al., 1990) and DNA manipulations (Sambrook et al., 1989). The yeast integrin interacting homologue (IIH1) gene was cloned by PCR using as a template the genomic DNA of strain W303 and oligonucleotides oSP44 (5′CAG-AATAGTCGGAGAAGCGGAC3′) and oSP45 (5′GTAAGGTGCAAGATCAGACAAAG 3′). To construct a IIH1 chromosomal deletion (iih1::kanMX4), the heterologous kanMX4 cassette was amplified by PCR using plasmid pFA6a-kanMX4 (Wach et al., 1994) as a template and oligonucleotides oSP43 (5′CGCATACAACTGTAAACAGACTTGA- GGAAGGAGGGGAATCCCCTCAGGAGATC GATGAAT TC GAGCTC-G3′) and oSP46 (5′GCCTCATCCCTCGTTCTTATAGTATAATTACAAGAAGCAATACGACAGCGTACGCTGCAGGTCGAC3′) as primers. The amplification product contained the kanMX4 cassette flanked by IIH1 sequences (underlined in the oligonucleotide sequences) and was used to transform the diploid strain W303. G418-resistant transformants were shown by PCR analysis to be heterozygous for the replacement of most of the IIH1 chromosomal ORF with the kanMX4 cassette. By sporulation and tetrad analysis of one of these transformants (ySP478), iih1::kanMX4 segregants were shown to be inviable, since all tetrads contained only two viable spores that were always G418 sensitive.
To construct the GAL–IIH1 fusion (pSP43), the IIH1 XbaI/SspI fragment was cloned in XbaI (SalI) of a Yiplac211-derived plasmid (Gietz and Sugino, 1988) that carried the BamHI–EcoRI GAL1-10 promoter fragment (c2139). To construct the GAL-Hsp27BBP/eIF6 fusion (pSP40), a PCR fragment containing a hemagglutinin (HA)-tagged version of Hsp27BBP/eIF6 was amplified using as a template the pSG5-p27 expression plasmid in which the entire open reading frame of human p27BBP/eIF6 gene was cloned in frame with a 10–amino acid HA tag in the NH2 terminus (the original vector is described in Green et al., 1988, the HA-modified vector was a gift of V. Zappavigna, DIBIT). The oligonucleotides YSTAG (5′CGG-AATTCAACAATAATGTACCCATACGAGCTTCCA3′) and YAS-TAG2 (5′CGGAATTCCTAGGTGAGGCTGTCAATGAGGGA3′), where the sequences of the human gene and of the HA tag are underlined were used as primers. The obtained PCR product was cut with EcoRI and cloned in the EcoRI site of c2139. Both GAL-IIH1 and GAL-Hsp27BBP/eIF6 fusions were integrated at the ura3 locus of ySP478 by cutting pSP43 and pSP40 with ApaI before transformation. As a result, strains ySP650 (GAL-IIH1 single copy), ySP653 (GAL-Hsp27BBP/eIF6 single copy), and ySP652 (GAL-Hsp27BBP/eIF6 multiple copies), respectively were generated. The copy number of the integrated plasmids was checked by Southern analysis. Strain ySP661 (MATa, iih1::kanMX4, and ura3::URA3:: GAL-IIH1) was obtained after sporulation and tetrad dissection of ySP650, whereas ySP664 (MATa, iih1::kanMX4, and ura3::URA3::GAL-Hsp27BBP/eIF6) was obtained after sporulation and tetrad dissection of ySP653.
Polysome and Western Blot Analysis of Ribosomal Fractions
Polyribosome preparation and polysome analysis were done exactly according to Foiani et al. (1991). Briefly, cell cultures of W303 (wt), ySP661 (iih1Δ, GAL-IIH1), and ySP664 (iih1Δ, GAL-HSp27BBP/eIF6) were grown in YEPRG medium and shifted to YEPD at time 0 to repress the GAL promoter. Yeast extracts were prepared from 300 ml of cell culture at OD = 0.5–1 (Foiani et al., 1991), layered on a 7–47% sucrose gradient in 50 mM Tris-acetate, pH 7.0, 50 mM NH4Cl, 12 mM MgCl2, and 1 mM dithiothreitol and centrifuged at 4°C in a SW41 Beckman rotor for 2 h at 39,000 rpm. Gradient analysis was performed with a gradient collector with continuous monitoring at A254.
For protein analysis, the collected fractions were precipitated with TCA to a final concentration of 10% and left on ice for 30 min. Fractions were centrifuged at 15,000 g, for 15 min at 4°C, and resuspended in Laemmli buffer. Equal amounts of extracts were run on denaturing 12% acrylamide gels and blotted as described above.
p27BBP/eIF6 Is Present in all Cell Lines, at Early Developmental Stages and Is Associated with the Cytoskeleton
It was previously observed that p27BBP/eIF6 mRNA was highly expressed during mouse embryonic development and that highly conserved homologues were present in the unicellular organism S. cerevisiae (Biffo et al., 1997). These data suggested that p27BBP/eIF6 might have a general role in cellular processes that is not limited to epithelial cells expressing β4 integrin only. To test this hypothesis, the expression of p27BBP/eIF6 was first measured by Western blot analysis with a polyclonal antiserum directed against the COOH terminus of p27BBP/eIF6 on total protein lysates from immortalized cell lines of various origin (see Materials and Methods for original references). So far, p27BBP/eIF6 has been detected in all cell lines analyzed. Fig. 1 (left) shows the levels of p27BBP/eIF6 in the immortalized cell lines NIH/3T3 (nontransformed mouse fibroblasts), Jurkat (human T cells), SK-N-MC (human neuroblastoma), A431, HeLa, HaCaT, and FG2 (transformed human epithelial cell lines), and Rin2A (human insulinoma). Constitutive expression of p27BBP/eIF6 was also detected in two out of two primary cultures tested, respectively, human primary fibroblasts, and mouse resting splenocytes.
It was previously observed that p27BBP/eIF6 mRNA was abundant during embryonic development and declined in the adult, where it was mainly retained in epithelial tissues (Biffo et al., 1997). To test the hypothesis that p27BBP/eIF6 protein may be present already at early phases of development, its onset was studied in embryos. The protein was found to be expressed from the earliest developmental stage and later on. Fig. 1 (right) shows p27BBP/eIF6 in the Xenopus egg, between fertilization and the beginning of segmentation. Comparable results were obtained in mice. As previously observed for p27BBP/eIF6 mRNA, in the adult, high levels of p27BBP/eIF6 protein were retained mostly in epithelial tissues, and testis (not shown).
In epithelial cells containing β4 integrin, ∼50% of p27BBP/eIF6 was associated with the intermediate filament cytoskeleton (Biffo et al., 1997). The association of p27BBP/eIF6 with the cytoskeleton was analyzed in cell lines not containing β4 integrin. For this purpose, cells were first extracted with detergent containing buffers (Materials and Methods) and the various fractions were analyzed by Western blot. Part of p27BBP/eIF6 was always found in the cytoskeletal fraction. However, the extent of the association varied according to the cell line (not shown). Fig. 1 (right) shows the results of the fractionation experiments in Xenopus eggs, where at least half of p27BBP/eIF6 was found to be resistant to detergent extraction and associated with the cytoskeleton.
Nucleolar Localization of p27BBP/eIF6
The topographical distribution of p27BBP/eIF6 was studied in detail by immunofluorescence, immunocytochemistry, and electron microscopy. To summarize our findings, as shown in Figs. 2–4, p27BBP/eIF6 was present in the nucleus, with a clear nucleolar pattern. The nucleolar staining of p27BBP/eIF6 was present in all the organisms analyzed so far (from worms to humans) and in all cell lines. In addition, in some cell lines containing β4 integrin, p27BBP/eIF6 was clearly evident in the cytoplasm (see Fig. 5). All the immunoreactivity described is specific, since both the nuclear and the cytoplasmic staining could be routinely abolished by preincubating the antiserum with either the peptide used for immunization or with the recombinant protein (Fig. 2, A and B). In addition, a similar nucleolus-enriched staining pattern could be seen on NIH/3T3 fibroblasts transfected with a HA-tagged version of p27BBP/eIF6, followed by immunofluorescence with a mouse anti-HA mAb (not shown).
The nuclear staining of p27BBP/eIF6 and its dynamic features will be described using the FG2 cell line as a model. In the interphase nucleus, p27BBP/eIF6 was clearly concentrated in the nucleolus (Fig. 2, A and C). This pattern was similar to the one obtained with an antiserum recognizing the nucleolar protein, fibrillarin (Fig. 2 E). The nucleolar colocalization was supported by double immunofluorescence studies with fibrillarin and p27BBP/eIF6 (not shown).
To establish whether p27BBP/eIF6 was dynamically associated with the nucleolus, epithelial cells were treated with low doses of actinomycin D and p27BBP/eIF6 localization was analyzed after 1, 4, and 12 h. This treatment caused the collapse of the nucleolus and the redistribution of nucleolar-associated proteins (Schofer et al., 1996). In actinomycin D–treated cells, both p27BBP/eIF6 (Fig. 2 D) and the nucleolar antigen, fibrillarin (Fig. 2 F), reversibly weakened their association with the nucleolus and became mostly diffuse in the cell's nucleus. Importantly, no effect on p27BBP/eIF6 localization was seen when cells were treated with the protein synthesis inhibitors, cycloheximide and puromycin (not shown). Nucleolar localization of p27BBP/eIF6 was confirmed by immunoelectron microscopy, using the anti-p27BBP/eIF6 antiserum, followed by 5-nm gold-labeled secondary antibodies. 5-nm gold beads were strongly concentrated within the nucleolus (Fig. 2 G).
Next, we tested whether p27BBP/eIF6 was stably associated with ribosomal proteins in the cytoplasm of the 804G-clone A epithelial cell line. For this purpose, ribosomes and ribosomal proteins were separated from all of the following: mitochondria, nuclear matrix/intermediate filaments, and soluble proteins. Afterwards, the different fractions were tested for the presence of p27BBP/eIF6 by Western blot analysis. As shown in Fig. 2 H, most of the protein was present in the nuclear matrix/intermediate filament cytoskeleton fraction. A faint band was associated with the ribosomal fraction.
p27BBP/eIF6 Redistributes during Mitosis
The strong nucleolar-associated pattern of p27BBP/eIF6 was visible in all cell lines during interphase, as well as in various normal and neoplastic tissues (Sanvito, F., manuscript in preparation). Therefore, it was expected that during mitosis, when the nucleolus disappears, the protein would be redistributed. Indeed, during cell division p27BBP/eIF6 dramatically changed its topographical pattern. At prophase, the immunoreactivity tended to become more dispersed at first. Later, it became associated with the periphery of condensed chromosomes (Fig. 3, A and B). At metaphase, p27BBP/eIF6 was enriched in the central mass of chromatin formed by the condensed chromosomes of the metaphasic plate (Fig. 3, C and D) and this pattern was even more noted at anaphase (Fig. 3, E and F). With the onset of telophase and the reappearance of the nucleolar organization, p27BBP/eIF6 first scattered and then regained its association with the nucleolus (Fig. 3, G and H). The redistribution of p27BBP/eIF6 during the mitotic phases was not associated with its proteolytic degradation. Furthermore, no obvious physical association of p27BBP/eIF6 with tubulin was observed (not shown). A similar redistribution was observed for some nucleolar antigens, chromosome passengers, which redistribute around chromosomes during mitosis (Earnshaw and Bernat, 1991), as well as for some nuclear matrix-associated antigens, whose immunoreactivity become more dispersed during mitosis (Nickerson et al., 1992).
p27BBP/eIF6 Is Associated with the Nuclear Matrix
To investigate whether the nucleolar p27BBP/eIF6 was associated with the nuclear matrix, FG2 cells were extracted with a sequential treatment by means of detergents, DNase, RNase, and high salts (He et al., 1990), and then analyzed by immunofluorescence and electron microscopy. This treatment removed >90% of the proteins, and 95% of the DNA. In addition, the treatment uncovered a nuclear matrix consisting of a nuclear lamina connected to the cytoplasmic intermediate filaments and of an internal meshwork of polymorphic fibers connecting the lamina to masses within the nucleus. In conditions that lead to the complete loss of DNA (Fig. 4 D), the p27BBP/eIF6 staining, associated with the nucleolus was clearly retained (Fig. 4, C–E). Also, the nuclear staining of p27BBP/eIF6 was unaffected after digestion of residual RNA with RNase (not shown).
To establish whether the residual staining of p27BBP/eIF6 was present in specific structures, extracted cells were examined by immunoelectron microscopy. By this analysis, immunoreactivity of p27BBP/eIF6 was always found to be associated with the residual thick filaments of the nuclear matrix (Fig. 4 F). Taken together these data show that in the nucleolus and in the nucleus a relevant part of p27BBP/eIF6 is tightly associated with the nuclear matrix.
Topographical Relationships of p27BBP/eIF6 and β4 at Hemidesmosomes in Epithelial Cells
In epithelial cells containing the β4 integrin, the pattern of immunoreactivity of p27BBP/eIF6 was slightly different and is briefly described using the epithelial cell line 804G clone A. This cell line contains human β4 integrin, clustered in rosettes of hemidesmosomes (Spinardi et al., 1993). As a result, when stained with antibodies against β4, these cells exhibit a typical Swiss cheeselike pattern in which intense β4 staining surrounds cytoplasmic areas devoid of integrin (Fig. 5, A and D). Confocal laser scanning microscopy analysis in the horizontal section (x, y) of p27BBP/eIF6 immunolocalization in these cells clearly showed a cytoplasmic staining partially superimposable to the one for β4 integrin (Fig. 5, A and D, β4; B and C, p27BBP/eIF6; E, β4– p27BBP/eIF6). Most importantly, in the vertical (x, z) and horizontal (x, y) sections, both β4 and p27BBP/eIF6 stainings were excluded from the small circular areas forming the holes of the Swiss cheeselike pattern (Fig. 5, A′ and D; B′ and C). However, staining with the labeled actin-binding drug, phalloidin, showed that these holes contained other cytoskeletal components such as actin and actin-binding proteins (data not shown; Spinardi, L., manuscript in preparation). These data suggest that in epithelial cells that require β4 to form hemidesmosomes, p27BBP/eIF6 can be specifically recruited in the intermediate filament's cytoskeleton converging on these adhesion structures.
To extend these observations, the presence of p27BBP/eIF6 was analyzed by immunoelectron microscopy on cryosections of human amnion, a tissue that contains hemidesmosomes clustered at the basal cell surface. Consistent with the pattern observed in the 804G clone A cells, p27BBP/eIF6 was detected at the level of inner plaque of the hemidesmosome, where it seemed associated with a thin filament network (Behzad, 1995) running between the intermediate filaments and the hemidesmosomal dense plaque (5-nm gold beads; Fig. 5, F and H, arrowheads). A specific immunolabeling was also noticed in the cytoplasm associated with filamentous structures (e.g., the area indicated by the arrow in Fig. 5, G and H), and also at the inner face of desmosomes (Fig. 5 I). In agreement with the association with the intermediate filament cytoskeleton, p27BBP/eIF6 immunolocalization was resistant to high salt extraction (not shown). However, the p27BBP/eIF6 positive structures (5-nm gold beads) were within intermediate filament bundles, as shown by a double staining with antikeratin antibodies (15-nm gold beads; Fig. 5, G and H, arrows).
p27BBP/eIF6 Is Essential for Yeast Cell Viability
To gain more insights into p27BBP/eIF6 function, several approaches were taken, but our efforts to manipulate the levels of p27BBP/eIF6 in mammalian cell lines were not successful. Briefly, the expression of p27BBP/eIF6 antisense mRNA in NIH/3T3 cells led only to a small decrease of protein levels and established clones could not be derived (Sanvito, F., unpublished observations). Furthermore, transient expression of several mutated constructs in COS cells led in some cases to accumulation of p27BBP/eIF6 either in the nucleus or in the cytoplasm, and was toxic to the cells (Sanvito, F., unpublished observations). These observations, together with the nucleolar localization and the fact that the protein is conserved from yeast to humans (Biffo et al., 1997; Si et al., 1997), might suggest a conserved function for this protein, which should be independent of β4 integrin (S. cerevisiae does not have β4 homologues). The possibility that p27BBP/eIF6 has an ancestral function is further supported by the finding that putative genes encoding peptides homologous to human p27BBP/eIF6 are present in the genome of different Archibacteria and are also found in plants (Fig. 6).
The analysis of the conserved amino acid sequences does not provide any insight into p27BBP/eIF6 function. However, the fact that S. cerevisiae contains a p27BBP/eIF6 homologue, 80% identical to the human protein, allowed us to analyze the functional role of the protein in the yeast model. The yeast protein is encoded by a single copy gene, which we called IIH1. We disrupted one chromosomal copy of the IIH1 gene in a diploid strain (see Materials and Methods), followed by sporulation of the obtained IIH1/iih1Δ heterozygous strain. Tetrad dissection and analysis showed that all tetrads contained only two viable spores (Fig. 7), none of which carried the disruption marker KanMX4, indicating that deletion of IIH1 was lethal. Spores carrying the iih1Δ allele were able to germinate, but arrested cell division either in the first or the second cell cycle.
We asked whether the human protein could rescue the lethality caused by deletion of the IIH1 gene. For this purpose, we constructed fusion genes where the yeast or the human p27BBP/eIF6 coding sequences were expressed under control of the yeast galactose inducible GAL1-10 promoter. These fusions were integrated in either single or multiple copies at the yeast URA3 locus of IIH1/iih1Δ heterozygous diploid strains. Subsequently, these integrated fusions underwent induced sporulation to analyze viability of their meiotic segregants under galactose-induced conditions. As shown in Fig. 7, most tetrads derived from any of these diploid strains contained either three or four viable spores, as expected if expression of human p27BBP/eIF6 (Hsp27BBP/eIF6) was able to rescue the lethality caused by the iih1Δ allele. These data indicate that human and yeast p27BBP/eIF6 share a common function. However, expression of human p27BBP/eIF6 seems to complement the defect less efficiently than its yeast counterpart; as indicated by the slower growth of the clones derived from spores expressing a single copy of the human gene and the iih1Δ allele (Fig. 7). This might be due to inefficient translation of the human mRNA gene in yeast (CAI-S.c. = 0.076); consistently with this hypothesis, the slow growth phenotype was substantially abolished when multiple copies of the GAL- Hsp27BBP/eIF6 fusion were integrated at the ura3 locus (Fig. 7).
Depletion of p27BBP/eIF6 Causes Accumulation of G1 Cells
To study the function of p27BBP/eIF6 in yeast cells, we characterized the phenotype caused by its depletion. For this purpose, wild-type and iih1Δ cells, carrying either the GAL-IIH1 or the GAL-Hsp27BBP/eIF6 fusion and logarithmically growing in galactose, were transferred to glucose-containing medium, to switch off the GAL promoter. The switch to a glucose-containing medium resulted in the progressive loss of the p27BBP/eIF6 protein (not shown). Since the shut-off of the GAL-Hsp27BBP/eIF6 fusion caused a much quicker arrest of cell division than that of the GAL-IIH1 fusion, we used the GAL-Hsp27BBP/eIF6 fusion-expressing strain for all the described depletion experiments. As shown by the FACS® profiles in Fig. 8, yeast cells depleted of p27BBP/eIF6, progressively stopped growing and accumulated as G1 cells with 1C DNA content. This phenotype is consistent with a role of p27BBP/eIF6 in protein synthesis since yeast cells need to grow in cell mass and reach a critical size before they can enter the S phase.
p27BBP/eIF6 Depletion Correlates with the Loss of Free 60S Ribosomal Subunit
The arrest of p27BBP/eIF6 depleted cells in G1, the fact that p27BBP/eIF6 has been independently identified as a putative translation initiation factor (Si et al., 1997) and our observation that p27BBP/eIF6 is detected in the nucleolus of all cell lines, suggested that this protein might be involved in protein synthesis and/or ribosome assembly. To understand the relevance of p27BBP/eIF6 in one of these processes in yeast, the polysome profiles of wild-type– and p27BBP/eIF6-depleted cells were analyzed. For this purpose, wild-type and iihΔ strains carrying the GAL-Hsp27BBP/eIF6 were grown in galactose-containing medium, and then shifted to glucose-containing medium to switch off the GAL promoter. As a control, a strain where the iih1Δ allele lethality was rescued by the GAL-IIH1 fusion was also used.
As shown in Fig. 9, the ribosomal profiles of wt and iih1Δ GAL-IIH1 strains were very similar at time 0, whereas iih1Δ GAL-Hsp27BBP/eIF6 cells, consistently with their slow growth phenotype, already showed a marked decrease in the amount of both the 60S subunit and the polysome fraction at the same time point. In contrast, the levels of the free 40S subunit seemed unaffected or slightly increased. This phenotype was even more dramatic 6 h after shifting to the glucose-containing medium of iih1Δ GAL-Hsp27BBP/eIF6 cells (Fig. 9). Furthermore, an accumulation of half-mer polysomes (i.e., 80S + 60S) was detectable under these conditions. These data suggest that p27BBP/eIF6 might have a primary function in the correct assembly of the 60S ribosomal subunit in yeast.
Cofractionation of human p27BBP/eIF6 in yeast cells was analyzed in parallel. For this end, fractions from the ribosomal gradients were precipitated with TCA and analyzed by Western blot using antibodies against the human protein. As shown in Fig. 10, p27BBP/eIF6 was detected in the 80S and in the free 60S fractions, but absent from polysomes.
p27BBP/eIF6 was simultaneously identified by two laboratories using two different approaches. It was isolated in our laboratory as a cytoplasmic interactor of the β4 integrin subunit, and we have shown that it can specifically bind the cytodomain of β4 in vitro (Biffo et al., 1997). However, the discovery of p27BBP/eIF6 homologues in organisms that do not contain β4 indicated that this protein might have a function independent of β4. Along this line, p27BBP/eIF6 was independently identified by Si et al. (1997) as a putative translation initiation factor, able to inhibit the association between the 60S and the 40S ribosomal subunits.
In this study, we have shown that although in epithelial cells p27BBP/eIF6 is coherent with β4 at hemidesmosomes, its association with the cytoskeleton is not a unique feature of epithelial cells. Indeed the protein is in the nuclear matrix of all growing cells. Consistently with its conserved nucleolar expression pattern and sequence, p27BBP/eIF6 is necessary for growth in yeast cells where its loss correlates with a reduced level of the free 60S ribosomal subunit. The in vivo findings were unexpected because they were consistent with a role of p27BBP/eIF6 in ribosomal biogenesis rather than in mRNA translation. In addition, the association of p27BBP/eIF6 with the nuclear matrix suggested that this process was linked to the nuclear cytoskeleton. The ability of the human protein to complement yeast mutation further suggested a conserved function for p27BBP/eIF6.
An Evolutionarily Conserved Function for p27BBP/eIF6 in 60S Metabolism
Database analysis indicates that p27BBP/eIF6 is a very ancient, evolutionarily conserved protein. It is striking to note that the homology is not restricted to a particular domain of the protein, and that even the length of the protein is constant among different species (246 amino acids in C. elegans; 245 in humans, fly, yeast, and A. thaliana; and 215–222 in different Archibacteria). These data suggest that p27BBP/eIF6 may have a critical and conserved function. Indeed, we have shown that the deletion of the S. cerevisiae IIH1 gene, encoding the p27BBP/eIF6 homologue, is lethal to yeast cells, and that the human protein can complement the yeast-null mutation. Some lines of evidence suggest that also in mammalian cells, p27BBP/eIF6 may be required for growth because of the following: (a) the inability to produce stable p27BBP/eIF6 mRNA antisense expressing mammalian cells (not shown); (b) the ubiquitous p27BBP/eIF6 expression in all immortalized cell lines so far analyzed; (c) and the presence of a single p27BBP/eIF6 gene in the human genome (Sanvito et al., 1998). The generation of p27BBP/eIF6-null mice will help to understand whether p27BBP/eIF6 is also necessary for growth in higher vertebrates. Unfortunately, extensive sequence analysis did not yield significant clues to understand p27BBP/eIF6 function.
To gain some insights into this problem, we used two complementary approaches: the depletion of p27BBP/eIF6 in the genetically manipulable yeast model, and the study of its topographical localization and biochemical properties in mammalian cell lines and tissues. Yeast cells depleted of p27BBP/eIF6 are progressively arrested in G1, a phenotype consistent with a defect in either protein synthesis or ribosomal biogenesis. This fact, and the localization of p27BBP/eIF6 in nucleoli prompted us to analyze the effect of its depletion on the polysome profile. These experiments provide useful information about how p27BBP/eIF6, based on its in vitro ribosomal anti-association activity, could be a translation initiation factor (Si et al., 1997). Polysome profiles of p27BBP/eIF6-depleted yeast cells showed a dramatic reduction in the peak of free 60S subunits and the appearance of half-mer polysomes. Similar polysome profiles have been observed for mutants defective in ribosomal proteins of the 60S ribosomal subunit (Moritz et al., 1991; Deshmukh et al., 1993; Vilardell and Warner, 1997), or for components involved in pre-rRNA processing and 60S ribosomal subunit assembly (Ripmaster et al., 1992; Sun and Woolford, 1994; Hong et al., 1997; Weaver et al., 1997; Zanchin et al., 1997; Kressler et al., 1998). Thus, the primary function of p27BBP/eIF6 in yeast is likely related to the 60S ribosomal subunit metabolism.
Polysome profiles of yeast cells, defective in translation initiation factor proteins, are generally characterized by the reduction of the rate of polysomes accompanied by the gradual accumulation of both the free 40S and 60S subunits. Therefore, the polysome profile of p27BBP/eIF6-depleted yeast cells does not support its primary function as a translation initiation factor. However, on the basis of the in vitro data of Si et al. (1997), and in view of the presence of p27BBP/eIF6 also in the cytoplasm of some human cells, the possibility that this protein might have a function also as a cytosolic initiation factor cannot be ruled out, as such activity could be masked by the predominant defect in 60S metabolism.
The polysome profile does not enlighten the precise role played by p27BBP/eIF6 in 60S metabolism. The protein may be necessary for ribosome assembly/transport, or may act as a structural ribosomal protein. On the basis of the available data, this last possibility is less likely to be true. In fact, the amount of p27BBP/eIF6 sedimenting with the ribosomal fraction in several cell lines represents only a minor fraction of the total p27BBP/eIF6 content. Furthermore, no p27BBP/eIF6 was detected in the polysome fraction.
p27BBP/eIF6 accumulates in the nucleolus of all the analyzed cell lines, where its pattern follows nucleolar evolution (redistribution at mitosis, when the nucleolar organizing region disappears, and redistribution after actinomycin D treatment). Since the nucleolus is the site where ribosomal subunits are assembled, it seems plausible to speculate that p27BBP/eIF6 might be involved in 60S ribosomal biogenesis. The process of ribosome biogenesis is complex and involves several factors (for review see Woolford and Warner, 1991; Eichler and Craig, 1994) including proteins with diverse functions such as RNA helicases, transcription factors, and nucleases. Further studies will address the precise role that p27BBP/eIF6 might play in this process.
Finally, it is possible that p27BBP/eIF6 may be involved in the transport of the 60S subunit from the nucleus to the cytoplasm. To date, very little is known about this process (for review see Shaw and Jordan, 1995), and only a few nucleolar proteins have been found to shuttle between the nucleolus and the cytoplasm. In this context, three observations are particularly intriguing: (a) the presence of p27BBP/eIF6 both in a soluble pool and in a cytoskeletal bound compartment; (b) the existence of trace amounts of soluble cytoplasmic p27BBP/eIF6 in all cells; and (c) the ability of p27BBP/eIF6 to bind also the mature 60S subunit (Si et al., 1997).
It is also worth noting that the nucleolar localization of p27BBP/eIF6 is observed in the absence of a consensus nuclear localization signal. Therefore, either p27BBP/eIF6 carries an unknown sequence for nuclear targeting or it is targeted into the nucleus by binding an additional factor in the cytoplasm. The second hypothesis is supported by the fact that even in its most soluble form, p27BBP/eIF6 partitions in gel filtration as a high molecular weight complex (unpublished observation). The molecular dissection of this high molecular weight complex may shed light on the mechanism by which p27BBP/eIF6 is transported to the nucleus.
p27BBP/eIF6 in the Nuclear Matrix/Intermediate Filaments Fraction
Our study shows that a relevant fraction of p27BBP/eIF6 is highly insoluble in vivo and is associated both with the nuclear matrix and with the intermediate filament pool. In the cytoplasm, electron microscopy studies have detected p27BBP/eIF6 on thin cytoplasmic filaments of unknown composition that are spatially separated from the classical keratin intermediate filaments, and converge both upon hemidesmosomes and desmosomes. To our knowledge, beside keratins, only another intermediate filament associated protein, IFAP300, has been described both in hemidesmosomes and desmosomes (Skalli et al., 1994). In this context, it is interesting to note that a recent thorough electron microscopy analysis of human hemidesmosomes has shown the presence of a novel filamentous structure in the proximity of the inner plaque of the hemidesmosome (Behzad et al., 1995).
Nuclear matrix consists of both thick polymorphous filaments and of thin filaments known as core filaments (He et al., 1990). In the nucleus, p27BBP/eIF6 is associated with polymorphous thick filaments, and is absent from the core filaments. This observation is fully consistent with the notion that core filaments may be formed by nuclear RNA, and that p27BBP/eIF6 distribution is resistant to RNase digestion (He et al., 1990). The localization of p27BBP/eIF6 in the nuclear matrix is of extreme interest in the context of ribosome biogenesis. Our data provide an intriguing link between the nuclear cytoskeleton and the process of ribosome assembly.
In recent years growing evidence has indicated that most nuclear and cytoplasmic processes including transcription, DNA replication, and protein synthesis are spatially organized in association with the cytoskeleton. The combined roles of p27BBP/eIF6 protein in 60S assembly, its association with the cytoskeleton, and its ability to bind β4 integrin (Biffo et al., 1997) and the mature 60S ribosome subunit (Si et al., 1997) belong to an integrated view of cell regulation that encompasses structure as well as biochemical processes (Chicurel et al., 1998).
p27BBP/eIF6 and β4 Integrin
We have previously shown that p27BBP/eIF6 binds specifically to the cytodomain of β4 integrin in vitro and in yeast (Biffo et al., 1997). Our previous data, and specifically the association of p27BBP/eIF6 with keratin intermediate filaments, strongly suggested that this interaction could occur also in vivo and be necessary for targeting β4 to hemidesmosomes and intermediate filaments. Since intermediate filament-associated proteins can be solubilized only upon SDS treatment, rendering the maintenance of biochemical interactions impossible, an association between β4 and p27BBP/eIF6 in tissues could not be proved. We now provide two further elements suggesting that p27BBP/eIF6 is functionally associated to the β4 integrin in vivo: (a) its peculiar Swiss cheese distribution is superimposable to that of β4 in cells that form hemidesmosomes; and (b) the presence of the protein, in vivo, in hemidesmosomes of the human amnion. Further experiments are needed to clarify the functional significance of β4–p27BBP/eIF6 interaction, and specifically whether p27BBP/eIF6 may direct β4 to hemidesmosomes. Alternatively, as it has been recently suggested, on the basis of in vitro evidence and yeast two-hybrid assays, the crucial step in targeting β4 to hemidesmosomes is the interaction with the large intermediate filament-associated protein, HD-1 (Niessen et al., 1997; Rezniczeck et al., 1998). If this is the case also in vivo, then the role of p27BBP/eIF6 binding to β4 may be related to a nonstructural function of β4 integrin, similar to that shown in the case of the recruitment of shc and grb2 (Mainiero et al., 1995) or of PI3 kinase (Shaw et al., 1997).
In the absence of further evidence, we may reasonably suggest that p27BBP/eIF6 has an evolutionarily conserved function linked to 60S ribosome biogenesis, and one acquired during evolution in epithelial cells containing β4 integrin. At least one precedent of a protein with a dual function acquired during evolution, i.e., β-catenin/armadillo, has already been reported. This remarkable protein can be found both at sites of cell–cell adhesion in connection to cadherins and in the nucleus where it can signal in conjunction with LEF-1 (for review see Willert and Nusse, 1998).
We thank R. Ochs for the kind gift of human antibodies directed against fibrillarin, F. Giancotti for permission to use clone A of the 804G cell line, and J. Nickerson for advice on nuclear matrix preparation. We are indebted to N. Offenhaeuser for useful suggestion and criticism throughout this work, to A. Hinnebusch and M. Foiani for useful suggestions, G. Serini for having pushed us to perform some experiments, E. Bianchi for her criticism, and E. Rizzo for the preparation and purification of the recombinant p27BBP protein used in the preadsorption experiments and the experiments with actinomycin D.
Abbreviations used in this paper
F. Sanvito and S. Piatti contributed equally to this work.
Address correspondence to Stefano Biffo, Lab. Istologia Molecolare DIBIT, San Raffaele Scientific Institute, V. Olgettina 58, 20132 Milano, Italy. Tel.: 39-22-643-4857. Fax: 39-22-643-4855. E-mail: email@example.com
The financial support of Telethon-Italy (grant 762 and E.712) is gratefully acknowledged. The work was supported also by Associazione Italiana per la Ricerca sul Cancro, Giovanni Armenise-Harvard Foundation, and MURST to P.C. Marchisio, and by CNR Target Project on Biotechnology Grant CT.97.01180.PF49(F).