DDBJ/EMBL/GenBank accession nos AB004649, AB004650
We have used gene targeting in the DT40 cell line to create a cell line which expresses a fusion between CENP-C and a mouse steroid receptor and which behaves as a conditional loss of function mutant of CENP-C. Under restrictive conditions these cells arrest at the metaphase/anaphase junction and after a delay of ~2.5 h die by apoptosis. These results indicate that CENP-C is either necessary for anaphase chromosome movement or for mediating a signal which triggers centromere function during anaphase. Our approach is simple and applicable to a wide range of proteins with general cell autonomous functions in vertebrates.
The centromere is essential for the accurate segregation of eucaryotic chromosomes at cell division. In budding yeast Saccharomyces cerevisiae a combination of biochemical and genetic studies has led to a detailed characterization of the centromeric DNA and associated proteins (1 ,2 ). However, the mechanisms by which the centromere brings about chromosome movement and the principles which relate the structures of vertebrate centromeres to those of yeast are poorly understood. This lack of understanding largely reflects the lack of a genetic system for the systematic dissection of vertebrate centromeres. We have therefore initiated a genetic analysis of the vertebrate centromere by investigating the function of CENP-C (3 ), a conserved centromeric protein.
Earnshaw and colleagues originally used auto-antisera to identify CENP-C as a constitutive human centromere protein of 943 amino acids (3 ). Subsequently others have isolated homologues from mouse (4 ) and sheep (5 ). mif 2 is a gene in yeast which encodes an essential centromere protein (6 ,7 ) which shares two short stretches of sequence similarity with CENP-C suggesting that mif-2p and CENP-C are homologues (8 ). Genetic and biochemical studies demonstrate that mif-2p interacts with the CDEII element of the yeast centromere (6 ,7 ). CDEII also interacts with the S.cerevisiae centromere microtubule binding complex (9 ). It is therefore an attractive hypothesis that CENP-C/mif-2p is a member of a conserved microtubule binding complex. This idea suggests that an understanding of CENP-C/mif-2p would cast light on centromere function and on why centromeric DNA from different organisms appears to be so different.
Anti CENP-C antibody micro-injection into human fibroblasts causes a temporary block in mitosis at the metaphase/anaphase junction but cells ultimately progress to anaphase although they do so abnormally and give rise to daughter cells containing multiple micronuclei (10 ). Genetics is a more versatile way to investigate protein function than the use of antibodies. We have therefore investigated the function of CENP-C using a conditional knockout approach. Our strategy for making a conditional mutation in CENP-C (Fig. 1 ) exploits a combination of the high frequency of sequence targeting characteristic of the chicken lymphoid DT40 cell line (11 ) and the use of a 4-hydroxytamoxifen sensitive steroid binding domain of a mutant mouse oestrogen receptor (conventionally referred to as ERtm; 12 ). In the absence of steroid the receptor is associated with a complex of binding proteins which includes HSP90 (13 ) and which is thought to be present in both the cytoplasm and nucleus. 4-Hydroxytamoxifen elutes the steroid receptor from its binding site and allows the CENP-C fusion to localize at the centromere and function.
We used two short regions of homology between CENP-C and mif-2p to design degenerate primers to isolate a short stretch of chicken CENP-C cDNA by RT-PCR. We used this stretch of cDNA to screen a chicken macrophage cDNA library and isolated four cDNAs encoding the entire CENP-C. These cDNAs corresponded to two alternatively spliced mRNA molecules, the most abundant of which encoded a protein of 865 amino acids. This protein was only 23% identical to human CENP-C over its entire length. However, it contained 18 identities to human CENP-C and 11 identities to mif-2p in the 23 amino acids of the homology block 2 described by M. Brown (8 ); similarly, chicken CENP-C contained 32 identities to human CENP-C and 15 identities to mif-2p in the 52 amino acids of homology block 3 (Fig. 2 A). No identities were detected in the homology block 1 extending doubts (8 ) about the significance of the identities between the human and yeast proteins previously detected in this region. We used the cDNA to isolate the chicken CENP-C gene. Limited sequencing established that the coding region spanned 19 kb of genomic DNA. The DT40 cell line is euploid for all of the chicken macro-chromosomes except for chromosome 2 for which it is triploid. Conventional cytogenetic analysis does not discriminate between most of the mini-chromosomes but their total number is consistent with this compartment of the genome also being nearly euploid. Consistently the DT40 cell line contains two alleles of the CENP-C gene and fluorescent in situ hybridization shows that both are present on an unidentified macro-chromosome (not shown).
In order to construct a cell line expressing a conditionally active CENP-C protein we first deleted the entire coding region of one of the two alleles with a replacement targeting construct containing a neomycin resistance gene (neo) (Fig. 2 B and C). We screened this targeting reaction by probing Asp718I digests with a probe beyond the 3' end of the targeting construct. Successful targeting was expected to lead to the replacement of a 26 kb cognate fragment with a 15 kb cognate fragment which also hybridized to a neo specific probe. In this way we analysed 22 stably transfected clones and isolated two in which one of the two alleles was disrupted. Next we built a dicistronic minigene targeting construct which included a 5' targeting sequence containing 7.4 kb of genomic DNA extending 5' from the ATG start codon and which was therefore likely to include the entire promoter. This sequence was followed by a cDNA encoding chicken CENP-C fused to a ERtm tag. (The fusion protein used in these experiments encodes all of the amino acids of CENP-C and of the ERtm linked by the peptide arginine-aspartic acid-proline.) This cDNA was followed by an internal ribosome entry site (IRES) of encephalomyocarditis virus, the coding region of Salmonella typhimurum hisD gene (his), a polyadenylation signal from the herpes simplex virus thymidine kinase gene and 4 kb of genomic DNA from beyond the 3' end of the coding region. Homologous recombination between the genomic DNA and the sequences at the ends of this targeting construct would replace the entire CENP-C gene with DNA encoding the dicistronic CENP-C ERtm hisD mRNA. We transfected cells deleted for one CENP-C allele with this second round targeting construct, selected for stable transfectants in the presence of neomycin, histidinol and the steroid agonist 4-hydroxytamoxifen and analysed 30 clones by restriction digestion with Asp718I and filter hybridization with the CENP-C 3' probe. This demonstrated that in 24 of the clones the construct had integrated into the CENP-C gene by homologous recombination (Fig. 2 C). However the mapping revealed that while the 3' targeting event was as intended, the 5' targeting reaction occurred between the CENP-C cDNA sequence and the coding region of the gene itself. In 22 of the 24 targeted clones the targeting reaction resolved within an exon which includes the EcoRI site marked with an asterisk in Figure 2 B. Thus probing of an EcoRI digest of genomic DNA from one of these clones with the complete cDNA revealed the replacement of the 3' 6.2 kb and 4.2 kb cognate fragments with the 2.1 kb 3' fragment derived from the construct while the 16 kb 5' EcoRI fragment remained intact (Fig. 2 B and C). In two of the other clones the resolution of the targeting reaction was 1 kb 5' from the EcoRI site and the 5' EcoRI fragment was reduced to ~15 kb in size (not shown). We do not know why the targeting frequency was 10-fold higher in the second round of the procedure nor why the 5' end of the targeting construct resolved within such a short stretch of homology. One explanation for both these observations is that the central region of the CENP-C gene is very active recombinationally. Another explanation for the observation that the recombination event was resolving within the coding region of the gene is that there were undiscovered isoforms encoded within the 5' of the gene and that the replacement of the entire gene by our cDNA was being selected against. Although the targeting in this step did not occur as originally planned the consequences of the targeting reaction were as we had desired in that the single remaining CENP-C gene was replaced with a gene encoding a CENP-C ERtm fusion (Fig. 2 C). In order to check that the targeted cell line expressed the ERtm tagged CENP-C fusion protein we stained metaphase chromosomes with an antiserum recognizing the ERtm domain and showed that it specifically detected a protein with a localization expected of a centromere component (Fig. 2 D). Thus we had isolated a cell line in which expression of CENP-C had been replaced by expression of a CENP-C ERtm fusion protein.
In order to analyse the consequences of loss of CENP-C from the centromere we cultured the CENP-C ERtm mutant cells in the absence of 4-hydroxytamoxifen. Firstly we examined the cells for the distribution of the CENP-C ERtm. In the cells grown in the presence of 4-hydroxytamoxifen the CENP-C ERtm had a punctate distribution in the interphase nucleus and could be localized at the ends of the chromosomes nearest the spindle poles of dividing cells (marked with an arrowhead in Fig. 3 A) but after culture in the absence of 4-hydroxytamoxifen the exclusively nuclear localization of the CENP-C ERtm was gradually lost and by 72 h CENP-C ERtm was localized throughout the cell (Fig. 3 B). Cells started to die 1 day after removal of 4-hydroxytamoxifen and by 4 days almost all of the cells were dead (Fig. 4 A). The results illustrated in Figure 4 A were reproduced with similar kinetics in three experiments. The cell cycle time of the DT40 cell in the presence and absence of the 4-hydroxytamoxifen line and of the CENP-C ERtm line in the presence of 4-hydroxytamoxifen (see below) is ~10 h indicating firstly that neither the presence of the tag nor of 4-hydroxytamoxifen had a detectable effect upon the passage of the cells through the cycle and secondly that the CENP-C ERtm tagged cells had to go through several divisions before they died. This delay before the cells started to die was consistent with the progressive loss of the CENP-C ERtm from the nucleus and the idea that centromeric CENP-C had to fall below a critical level before it had an effect.
It seemed likely that the cells were dying as a result of a failure at mitosis because CENP-C is a centromere protein and so we used DNA staining and immunocytochemical staining of microtubules to measure the proportion of cells in mitosis (Fig. 4 B). Culture in the absence of 4-hydroxytamoxifen indeed led to the accumulation of mitotic cells (Fig. 4 B) and many of these were aberrant in that they contained decondensed chromosomes which extruded from the spindle. Figure 3 D shows a field of cells from a culture grown for 48 h in the absence of 4-hydroxytamoxifen, immunocytochemically stained for tubulin which appears green and then counterstained for the DNA which appears red. The four cells marked with arrowheads are mitotic cells seen either down (top left hand cell) or across the spindle axis. The cells appear to be blocked at metaphase because the DNA is located within the spindle but not segregated to the poles. The chromatin in these metaphase cells is, however, not distributed on a discrete metaphase plate and appears puffy and decondensed. This is emphasized by comparison with the metaphase cells from a culture grown in the presence of 4-hydroxytamoxifen. Figure 3 C shows three cells from such a culture. The top left hand cell is a metaphase cell viewed down the spindle axis (the other two cells in Figure 3 C are interphase cells and show staining for interphase microtubule bundles). By comparison with the cells in Figure 3 D the chromosomes in the metaphase cell in Figure 3 C are well condensed and appear sharp as they all lie at the plane of focus. The simplest interpretation of these data was that removal of CENP-C from the centromere caused the cells to fail to enter anaphase and that after an interval they died. We repeated the experiment and obtained quantitatively similar results; results are presented in Figure 4 B.
It was, however, possible that the mutant cells were dying as a result of a failure to execute some point in interphase. In order to determine whether removing the cells from 4-hydroxytamoxifen had any consequences for their passage through interphase we used flow cytometric analysis to measure the DNA content of the cells before and after 48 h in the absence of 4-hydroxytamoxifen. These experiments demonstrated that loss of CENP-C from the centromere has no detectable effect upon progress through interphase. Thus 43 and 32% of the cells grown in the presence of 4-hydroxytamoxifen contained two and four haploid equivalents, respectively, of DNA while the 33 and 39% of the cells grown in the absence of the steroid contained two and four haploid equivalents of DNA. The increase in the proportion of cells with a 4N DNA content was consistent with the accumulation of mitotic cells that we had detected by single cell analysis (Fig. 4 B). A colcemid block experiment showed that the rates at which the two populations moved through interphase into mitosis was also similar. Thus during a 3 h incubation with colcemid of the cells grown in the absence of 4-hydroxytamoxifen the proportion of the population with a 4N DNA content increased by 17.5% while the proportion of a population of cells grown in the presence of 4-hydroxytamoxifen with a 4N content increased by 16%. Furthermore, analysis of the same experiments showed that there was no significant difference between the rates at which the two types of cultures moved out of G1. Thus the cytological and cell cycle results show that removing CENP-C from the kinetochore causes cells to fail to exit metaphase and that these cells go on to die. We estimate that cells spend ~2.5 h in metaphase before they die (see methods for details of the calculation). These results thus demonstrate that CENP-C is either necessary for anaphase chromosome movement or for mediating a signal which initiates anaphase centromere function and that CENP-C is unnecessary for transition through any other part of the cell cycle.
It seemed possible that the cells cultured in the absence of 4-hydroxytamoxifen were dying apoptotically and so we incubated cells with terminal nucleotidyl transferase and fluorescein-dUTP in order to detect the double strand breaks characteristic of early stage apoptotic cells. We observed that 2.5% of the cells derived from cultures grown in the absence of 4-hydroxytamoxifen specifically contained chromatin fragments with double strand DNA breaks (Fig. 3 E). This compared with a figure of 0.1% of cells grown in the presence of 4-hydroxytamoxifen (Fig. 4 B) and <0.05% of DT40 cells grown in either the presence and absence of 4-hydroxytamoxifen.
Our results thus demonstrate that CENP-C is either necessary for anaphase chromosome movement or for mediating a signal which initiates anaphase centromere function and that CENP-C is unnecessary for transition through any other part of the cell cycle. The phenotype that we detect has several features in common with the effect of the injection of anti CENP-C antibodies. However, the anti CENP-C antibodies only caused a delay in the metaphase-anaphase transition and thus it would appear that they were less effective at blocking CENP-C function than the removal of 4-hydroxytamoxifen in our system. Antibody microinjection has often been used to investigate the functions of vertebrate cell autonomous proteins in eucaryotic cells but our results and those of others (15 ) emphasize that this method may sometimes be only partially effective in blocking protein function. This difference has allowed us to make a stronger statement about the possible function of the CENP-C protein than was possible on the basis of the antibody micro-injection results. However, while we are able to demonstrate that CENP-C is essential for centromere function, the mechanism by which CENP-C functions remains to be established. It is possible that CENP-C mediates one or more of three possible functions: it may transduce a signal from the cell cycle control machinery to the microtubule binding complex; it may, as originally suggested by Tomkiel and colleagues (10 ), be involved in the assembly of the kinetochore; or it may be directly involved in microtubule binding. The DT40 system might allow us to make progress in resolving this issue. Thus it will be interesting to locate CENP-C to non-centromeric positions within chromosomes and the cell and explore the consequences. Ultimately it will be necessary to identify a biochemical activity associated with the CENP-C protein complex and to characterize it. The ease of genetic manipulation in DT40 may enable us to engineer a complex with properties which will facilitate this. One limitation of the DT40 system, however, is that few chicken homologues of genes involved in chromosome segregation have been identified. Thus, we would like to know whether CENP-E is associated with kinetochores lacking CENP-C and to know what role the Mad/Bub pathway plays in the apoptotic response to the loss of CENP-C. The availability of a cell line in which the CENP-C is conditionally active, however, means that we will be able to address these questions once the appropriate reagents become available.
The success of our strategy emphasizes the potential utility of DT40 cells for investigating the functions of general cell autonomous functions in vertebrates. Targeting is so efficient in DT40 that it is possible to carry out complex modifications of endogenous genes very quickly. Our strategy for making a conditional mutant is, in principle, preferable to the use of site specific recombination or translational control in that it is quicker in its action and requires fewer manipulations. One of the advantages of the yeasts as systems for dissecting biological processes is that they exist as haploids and thus it is often simple to select for loss of function or conditional mutations. DT40 is nearly diploid and it is not clear how new genes might be identified by mutational analysis. In yeast and in other systems, however, biochemical approaches incorporating the use of epitope tags are now proving powerful alternatives to the use of genetic methods for the identification of the components of protein complexes. This approach is directly applicable to the DT40 system. Together this combination of genetics based upon engineered conditional mutants and biochemistry may permit the dissection of the vertebrate centromere with a rigour necessary to compare it to that of yeast and thereby lead to an understanding of how the kinetochore functions and evolves.
DT40 cells were cultured and transfected by electroporation as described by Buerstedde and Takeda (11 ). 4-hydroxytamoxifen was used at 100 nM final concentration. G418 was used at a final concentration of 2 mg/ml and histidinol at a final concentration of 1 mg/ml to select for stable transfectants.
Chicken CENP-C was isolated by RT-PCR on mRNA extracted from DT40 cells. The sequences of the degenerate primers were CC(A/G)(C/T)TGGA(A/G)TA(C/T)TGG(A/C)G and TT(A/C)C- (T/C)(G/C/T)GA(A/G/C)GG(A/G/C)AC(A/G)TAGAA. The RT-PCR product was identified as being derived from CENP-C by sequencing and used to screen a chicken macrophage cDNA library, four cDNAs encoding the entire coding region were isolated and sequenced and used to screen a library of DT40 DNA made in [lambda]FIX. About 50 kb of genomic DNA covering the entire CENP-C gene was isolated and mapped. The extent of the coding region was identified by filter hybridization and limited sequencing. Plasmids were constructed by standard methods.
CENP-C ERtm was immunolocalized on metaphase chromosomes as described by Earnshaw and colleagues (16 ) using the rabbit anti mouse ER antibody (A. Sewing, unpublished) at a final concentration of 1/1000 and FITC conjugated goat anti rabbit antibody (Sigma) at 1/100. Tubulin and CENP-C ERtm were localized in cytospun mitotic cells by indirect immunofluorescence. Cells were fixed in 3% paraformaldehyde in 250 mM HEPES for 15 min at room temperature, permeabilized in 0.5% NP-40 in phosphate buffered saline (PBS) for 15 min at room temperature, rinsed three times in 0.5% bovine serum albumin (BSA) in PBS and stained for 1 h at 37°C in either neat supernatant of the anti-tubulin antibody YL1/2 (17 ) or in anti ER antibody diluted 1/1000 in PBS/0.5% BSA; these antibodies were then detected using either FITC conjugated rabbit anti rat antibody at 1/66 (Vector) or FITC conjugated goat anti rabbit IgG at 1/100 (Sigma) in PBS/0.5% BSA. Antibody stained cells were counterstained with DAPI at 0.5 µg/ml or propidium iodide at 0.5 µg/ml in Vectorshield antifade. Images were collected using a Digital Pixel (KAF1400) 12bit slow scan, cooled CCD camera mounted on a Zeiss Axioscop microscope with a Plan neofluar 100*/NA1.3 objective and then manipulated in IPlab before composition using Adobe photoshop.
Cell cycle analysis was as described in ref. 18 . Propidium iodide at a final concentration of 10 µg/ml was used to stain nuclei and the fluorescent signal was measured using a Becton-Dickinson FACS SCAN.
In order to estimate the length of time that the cells spend in mitosis before dying we assumed that the culture consisted of two subpopulations; one which was dying (termed D) and one which was not (termed A); and that the A population was cycling with the same kinetics as the cells grown in the presence of 4-hydroxytamoxifen. From these assumptions it follows that the proportion of the cells at metaphase Fm = A.f pile {size 8 italic m above size 8 italic a} + D.f pile {size 8 italic m above size 8 italic d} where A is the fraction of the population that is cycling and does not die, D is the fraction of the population that is cycling to death, f pile {size 8 italic m above size 8 italic a} is the fraction of the cell cycle that the alive cells spend in mitosis and f pile {size 8 italic m above size 8 italic d} is the fraction of the cell cycle that the D cells spend in mitosis before dying. CENP-C ERtm mutant cells double every 10 h in the presence of 4-hydroxytamoxifen and spend 1.7% of the cycle in metaphase. We have shown that the rate of passage through interphase is the same in the presence and absence of 4-hydroxytamoxifen and thus in the cell cycle lasting between 48 and 58 h the data of Figure 4 A show that A = 0.7 and D = 0.3. From this it follows that f pile {size 8 italic m above size 8 italic d} = 0.2066 which, given an interphase lasting 10 h, corresponds to a metaphase arrest of ~2.5 h. This figure will be an overestimate if the cells which do not die (referred to as A) delay in mitosis.
We thank Jean-Marie Buerstedde of The Basel Institute for Immunology for cells, plasmids and advice, Trevor Littlewood of the ICRF for the ERtm domain and Andreas Sewing for the antiserum recognizing it, John Young of The Institute of Animal Health, Compton for chicken cDNA libraries, Austin Smith for the IRES and Austin Smith and Andy Porter for comments upon the manuscript. The work was supported by a JSPS fellowship to TF and by the Cancer Research Campaign.
Human Molecular Genetics
Pages
Introduction
Results
Isolation of the chicken CENP-C gene
Mutation of the chicken CENP-C gene
Removing CENP-C from the centromere causes cell death
Removing CENP-C from the centromere causes a metaphase/anaphase block
Removing CENP-C from the centromere has no detectable effect upon passage through interphase
Removing CENP-C from the centromere leads to cell death by apoptosis
Discussion
Materials And Methods
Cell culture
Molecular biology
Cytology
Cell cycle analysis
Acknowledgements
References
REFERENCES
This page is maintained by OUP admin. Last updated Sat Oct 18 13:29:50 BST 1997. Part of the OUP Journals World Wide Web service.
Copyright
Oxford University Press, 1997