sign in sign up
<<
Home , Centromere protein B, Chromosome 1, Chromosome 21, Dicentric chromosome, X chromosome

Journal of Science 109, 2199-2206 (1996) 2199 Printed in Great Britain © The Company of Biologists Limited 1996 JCS3386

Epigenetic control of mammalian binding: does DNA methylation have a role?

Arthur R. Mitchell*, Peter Jeppesen, Linda Nicol†, Harris Morrison and David Kipling MRC Unit, Western General Hospital, Crewe Road, Edinburgh EH4 2XU, UK *Author for correspondence (internet [email protected]) †Present address: MRC Reproductive Biology Unit, Edinburgh, UK

SUMMARY 1 of the inbred mouse strain DBA/2 has a block of minor satellite DNA sequences on . polymorphism associated with the minor satellite DNA at The binding of the CENP-E protein does not appear to be its centromere. The more terminal block of satellite DNA affected by demethylation of the minor satellite sequences. sequences on this chromosome acts as the centromere as We present a model to explain these observations. This shown by the binding of CREST ACA serum, anti-CENP- model may also indicate the mechanism by which the B and anti-CENP-E polyclonal sera. Demethylation of the CENP-B protein recognises specific sites within the arrays minor satellite DNA sequences accomplished by growing of minor satellite DNA on mouse . cells in the presence of the drug 5-aza-2′-deoxycytidine results in a redistribution of the CENP-B protein. This protein now binds to an enlarged area on the more terminal Key words: Centromere satellite DNA, Demethylation, Centromere block and in addition it now binds to the more internal antibody

INTRODUCTION A common feature of many mammalian pericentromeric domains is that they contain families of repetitive DNA The centromere of mammalian chromosomes is recognised at sequences (Singer, 1982). Some of these families of repeated the cytological level as the primary constriction of the chro- appear to be quite specific for any one species. In , mosome. Using the electron microscope Rattner and Bazett- these are the simple-sequence, so-called satellite DNAs, where Jones (1989) and Ris and Wit (1981) identified the particular chromosomes have amplified relatively short as an electron dense component of the centromere with specific oligonucleotides to give the present-day situation where they structural features. These included inner and outer plates and may contain millions of copies of these repetitive DNA a fibrous corona. Phosphorous was one of the elements families (Prosser et al., 1986). A similar situation can be seen detected in this structure indicating that DNA was one of the in the mouse. In Mus musculus both the major and minor structural components. Later, Cooke et al. (1993) came to the satellite DNAs are present on all chromosomes with the conclusion that both the outer plate of the kinetochore and the exception of the (Pardue and Gall, 1969; Jones, matrix between the outer and inner plates either lacked DNA 1970; Wong and Rattner, 1988; Joseph et al., 1989). The same as an integral part of their structure or, if DNA sequences were is true for Mus spretus although the amount of DNA homolo- present, only a small amount (~2 kb) of DNA was involved gous to the Mus musculus major satellite DNA sequences in with these structural components. The initial finding that phos- the M. spretus is greatly reduced (Narayanswami et phorous was present in the outer plate being ascribed more al., 1992). In Mus caroli different families of repetitive DNAs likely to the presence of phosphoproteins rather than DNA have evolved within the centromeric domains of its chromo- sequences (Cooke et al., 1993). somes, and are quite distinct from those in the other two Mus The biological role of the centromere in the cell is of species (Kipling et al., 1995). paramount importance to the chromosome. The loss of cen- It was known that certain families of repetitive DNAs tromere function leads to chromosome instability during cell mapped, by in situ hybridisation, closer to the primary con- division which in turn leads to chromosome loss. striction of chromosomes than others (Mitchell et al., 1985; are recognised as the attachment sites for the spindle micro- Wong and Rattner, 1988; Joseph et al., 1989). Little sequence tubules. The loss of centromere function is associated with the similarity was found between the alphoid DNA in man and the chromosome being unable to capture microtubules. Thus during minor satellite DNA in M. musculus. Masumoto et al. (1989) cell division the alignment onto the metaphase plate and correct were first to report that both of these repetitive DNAs contained segregation of the chromosome during anaphase is prevented a conserved 17 (bp) motif and that this short sequence (Schulman and Bloom 1991; Rattner, 1991). Errors of this type was responsible for binding the CENP-B centromere protein. can lead to complex genetic disorders in higher organisms. CENP-B is one of the detected by autoantibodies in the 2200 A. R. Mitchell and others serum of some patients with the complex syndrome termed is an important component of some mammalian CREST (Moroi et al., 1980). Comparison of the CENP-B then the prevention of its binding by DNA methylation may be protein from man and mouse (Sullivan and Glass, 1991) has sufficient to cause centromere inactivation. shown 95% conservation at the amino acid level and from an analysis of its structure Pluta et al. (1992) initially showed that MATERIALS AND METHODS a DNA binding domain is present at the amino terminus. The first 158 amino acids were necessary to target the protein to the Chromosome preparations for indirect centromere. Later, Kitagawa et al. (1995) confirmed these immunofluorescence microscopy results and went on to show that the CENP-B protein had a DBA/2 mice were maintained in Edinburgh. Chromosome prepara- protein-protein dimerisation domain at its carboxy terminus. tions from spleen were prepared as described previously (Mitchell et Because it contains these different functional domains it has al., 1993). Cells were accumulated in and metaphase spreads been proposed that its biological role within both the alphoid were prepared as described by Jeppesen et al. (1992). Demethylation arrays in man and the minor satellite arrays in M. musculus is of chromosomal DNA was carried out by culturing cells in the presence of 5-aza-2′-deoxycytidine (Sigma) for 72 hours. The to promote condensation of the containing these optimum condition for growing the lymphocytes in the presence of repetitive DNAs (Kitagawa et al., 1995). this drug was found to be 5×10−6 M. After a of 36 hours the The relationship between CENP-B and the kinetochore cell cultures were replenished with one half volume of fresh medium remains unclear. The CENP-B protein appears localised to the containing 5-aza-2′-deoxycytidine at the above concentration. Control chromatin beneath the kinetochore plate (Pluta et al., 1990) and cultures were set up at the same time and grown under identical con- it does not seem to be directly associated with this structure. ditions. Both control and treated cells were prepared at the same time Where dicentric chromosomes have been studied it now seems and the antibody reactions were run in parallel. clear that CENP-B can be bound at both active and inactive Indirect immunofluorescence with antisera against CENP-B or centromeres (Earnshaw et al., 1989; Sullivan and Schwartz, CENP-E was carried out essentially as described previously (Jeppesen 1995). In contrast the CENP-C protein appears to be associ- et al., 1992). ated only with the active centromere/kinetochore as deter- Anti-CENP-B raised against cloned human CENP-B (a gift from Professor W. C. Earnshaw) and anti-human CENP-E (a gift from Dr mined by the position of the primary constriction on chromo- Tim Yen) were both polyclonal rabbit antisera raised against bacteri- somes (Earnshaw et al., 1989; Page et al., 1995; Sullivan and ally expressed protein. Schwartz, 1995). The CENP-E protein also seems to bind only Anti-CENP-B or anti-CENP-E were diluted 1:300 or 1:500, respec- to the active kinetochore (Sullivan and Schwartz, 1995). In our tively, in KCM (Jeppesen et al., 1992) + 10% normal goat serum previous work on the DBA/2 mouse (Mitchell et al., 1993) we (NGS). The secondary antibody was FITC-conjugated, affinity- showed that chromosome 1 in this animal differed from the purified, anti-rabbit IgG, raised in goat (Sigma Chemical Co.), diluted other chromosomes in the cell by having two separate blocks 1:20 in KCM + 10% NGS. For double antibody labelling, slides were of minor satellite DNA sequences. These blocks are cytologi- first incubated with KCM + 10% NGS containing a 1:100 dilution of cally distinct and separated from each other by intervening a CREST patient anti-centromere serum (CP) together with either major satellite DNA sequences. Only the more terminal of the anti-CENP-B or anti-CENP-E antiserum (both diluted 1:100). After two arrays has the characteristics of a functional centromere, washing, slides were incubated simultaneously with FITC-conju- gated, affinity-purified, anti-human IgG, raised in goat (Sigma i.e. CREST anti-centromere associated labelling (ACA) and Chemical Co.) to detect the CREST signal, and TRITC goat anti- sister attachment. However, in situ hybridisation rabbit IgG to detect rabbit antibody binding, both diluted 1:20 in KCM results revealed the presence of 17 bp CENP-B binding sites + 10% NGS. within both arrays of minor satellite DNA. The presence of CENP-B sites in the more internal array conflicted with the Primed in situ hybridisation (PRINS) apparent absence of CREST ACA to this region. The original technique of Koch et al. (1989) was used in combination In this present paper we extend our previous observations to with immunocytochemistry (Mitchell et al., 1992, 1993) using anti- demonstrate that the CENP-B protein, detected by specific bodies raised against CENP-B or CENP-E. The primers used were polyclonal serum, is normally present at the more terminal of oligos 203 (minor satellite) and C86 (CENP-B-box) (Mitchell et al., the minor satellite arrays on chromosome 1 in DBA/2 animals. 1993). Chromosomal DNA was either denatured in 30 mM NaOH, 1 M NaCl for 45 minutes at 4¡C followed by washing in 0.01 M Tris- We also present evidence that supports a role for an epigenetic buffer, pH 7.6, and air-dried or at 70¡C in 2× SSC, 70% formamide modification (Brown and Tyler-Smith, 1995), CpG dinu- for 3 minutes followed by 2 minute washes in 70, 90, and 100% cleotide methylation, in restricting the binding of CENP-B to alcohols (all at 4¡C) and air dried. Digoxigenin-11-dUTP was incor- the more terminal array. Experimental reduction in the amount porated during the PRINS reaction and detected with anti-digoxi- of CpG methylation in the minor satellite DNA was achieved genin-rhodamine FAB fragments (both from Boehringer Mannheim). by growth of DBA/2 cells in the presence of the drug 5-aza- The images were captured using a Zeiss Axioplan fluorescence micro- 2′-deoxycytidine, as judged by the use of Southern hybridisa- scope equipped with a 100 W mercury source and a photometrics tion and methyl sensitive restriction endonucleases. In such CCD camera. Images were acquired and processed using digital cases a change in the binding of the CENP-B protein on chro- software from Digital Scientific (Cambridge, England). The slides mosome 1 was also observed. The result was a redistribution were imaged through a Chroma P1 filter set with excitation filters of this protein which now additionally detected the more mounted in a rotating wheel which gave exact registration between images from the different fluorochromes. internal minor satellite array. Despite the presence of bound CENP-B protein, it does not appear that full centromere Southern hybridisations function has been conferred on the internal minor satellite Lymphocytes from both control and 5-aza-2′deoxycytidine cultures array, because the CENP-E protein remains restricted to the were washed in phosphate buffered saline and resuspended into 10 mM more terminal block. Nevertheless, if CENP-B protein binding Tris, 200 mM EDTA, pH 7.4. Cell lysis was carried out by the addition CenpB binding and methylation 2201 of sodium dodecyl sulphate (SDS) to 0.1%. DNA extraction was chemistry using sera from CREST (Fig. 1A), anti-CENP-B (Fig. carried out using standard methods of phenol/chloroform (1:1) extrac- 1B) and anti-CENP-E (Fig. 1C) each together with PRINS in tions before the DNA solution was dialysed against 10 mM Tris-HCl, situ hybridisation with oligo 203 for minor satellite on pH 7.4, overnight. After raising the salt to 0.1 M Na+ the dialysate was metaphase chromosome spreads from the DBA/2 mouse. In µ digested with RNase A at 50 g/ml for 1 hour at 37¡C before further agreement with our previous results chromosome 1 (arrowed) in phenol/chloroform extractions were carried out. The solution was this animal has two separate blocks of minor satellite DNA extensively dialysed as before. Restriction endonuclease digestions with MspI and HpaII were carried out according to the recommenda- (Mitchell et al., 1993). The immunofluorescence signal from all tions of the manufacturer (Boehringer Mannheim). Approximately 10 three sera gives a characteristic double-dot signal which appears µg of each sample DNA was loaded onto a 1.2% agarose gel and after always to be associated with the more terminal block of minor electrophoresis in 0.5× TAE the DNA was transferred to Hybond nitro- satellite DNA (arrowed in Fig. 1). Again as noted before cellulose (Amersham) according to the method of Southern (1975). (Mitchell et al., 1993) it would appear that in most instances the The M. musculus probe was a cloned trimer (Kipling et al., 1994) of signal is close to the boundary of the terminal array of minor minor satellite DNA labelled by random priming (Feinberg and Vogel- satellite sequences and the major satellite sequences. Figs 2 and stein, 1983) with [32P]dCTP (Amersham). Hybridisation was carried 3 show a double immunofluorescence signal combining CREST out in 5× SSC (1× SSC = 0.15 M NaCl and 0.015 M tri-sodium citrate) × and anti-CENP-B (Fig. 2) and CREST and anti-CENP-E (Fig. 2 Denhardt’s solution (Denhardt, 1966) and 0.1% SDS at 65¡C 3) antibodies. Fig. 2A shows the combined immunofluorescence overnight. The filters were washed in 2× SSC at the same temperature. Filters were exposed using Kodak XAR-5 film and intensifier screens. signal obtained with CREST and anti-CENP-B. Each separate signal is also shown. Fig. 2B is the anti-CENP-B signal and Fig. 2C the CREST signal. Fig. 2D is the DAPI stained chromo- RESULTS somes. Fig. 3A combines the indirect immunofluorescence results using CREST and anti-CENP-E antibodies. Fig. 3B is the Fig. 1 shows the result of combining indirect immunocyto- anti-CENP-E signal alone and Fig. 3C shows the distribution of the CREST antibody. The DAPI image is shown in Fig. 3D.

Fig. 1. (A) Combined image of PRINS in situ using oligonucleotide 203 (minor satellite probe) and indirect immunofluorescence showing the position of the centromere (as defined by CREST ACA serum). The PRINS signal is red and the CREST signal is white (arrowed). (B) As in A above except that the position of the CENP-B Fig. 2. (A) Combined indirect immunofluoresence image of anti- protein is shown in white. (C) As in A above but the indirect CENP-B (red) and CREST ACA (green) showing their immunofluorescence signal now detects the CENP-E protein. The colocalization (arrowed). (B and C) Split images of anti-CENP- arms are stained using DAPI (blue). and CREST ACA, respectively. (D) DAPI stained chromsomes. 2202 A. R. Mitchell and others

Fig. 3. (A) Combined indirect immunofluoresence image of anti-CENP- E (red) and CREST ACA (green) showing their colocalization (arrowed). (B and C) split images of anti-CENP-E and CREST ACA, respectively. (D) DAPI stained chromosomes.

Note that both anti-CENP-B and anti-CENP-E co-localise with presence of 5-azacytidine results in a characteristic deconden- CREST ACA (arrowed in Figs 2 and 3). From this it follows sation of major satellite DNA sequences (Mitchell et al., 1993; that CENP-B and CENP-E co-localise one with the other. Radic et al., 1987). This is not a characteristic when cells are In order to see what effect, if any, demethylation of the CpG cultured in the presence of 5-aza-2′-deoxycytidine. This is the dinucleotides within the minor satellite DNA (especially those CpGs within the CENP-B binding motif) would have on the distribution of the antibody signals, cells were grown in the presence of 5-aza-2′-deoxycytidine as described. For each experiment we prepared both DNA and chromosomes from the same batch of cells grown in the absence and presence of this drug. The DNA was assayed for demethylation by digestion with the endonucleases MspI and HpaII. Both cleave at the recognition sequence CCGG but if either cytosine residue is methylated then only MspI will cleave the DNA. At the same time control cultures were grown and both chromosome spreads and DNA were prepared from these. Fig. 4 is the result of hybri- dising minor satellite DNA sequences to a nitrocellulose filter containing DNA from separate experiments from both control cells and from cells grown in the presence of 5-aza-2′-deoxy- cytidine. DNA from control cells cleaved with MspI and HpaII are shown in the lanes labelled C.Msp and C.Hpa, respectively. Only the C.Msp lane shows a series of digestion products. Lanes 1-4 and 5-8 show the results using DNA prepared from different batches of cells from four separate experiments which were grown in the presence of 5-aza-2′-deoxycytidine. Lanes 1-4 are digested with MspI whilst 5-8 are digested with HpaII. 32 Unlike the controls, where the C.Msp lane only showed Fig. 4. Southern hybridisations using P labelled minor satellite to a digestion products, in these DNA samples digestion products nitrocellulose filter containing DNA from DBA/2 lymphocyte are found with both . Thus growing cells in the cultures. Controls after digestion with MspI are labelled C.Msp or C.Hpa after HpaII digestion. Lanes 1-4 and 5-8 contain DNA presence of this drug has resulted in demethylation of a sub- isolated from cultures grown in the presence of 5-aza-2′- stantial proportion of the CpGs in the minor satellite DNA. deoxycytidine and digested with MspI (1-4) or HpaII (5-8), It is worth noting that the morphological appearance of these respectively. Lanes 1and 5, 2 and 6, 3 and 7, and 4 and 8 are DNA chromosomes is quite distinct from cells cultured in the from the same culture. Each pair represents independent presence of 5-azacytidine. Growing DBA/2 lymphocytes in the experiments. CenpB binding and methylation 2203

sation when the chromosomal DNA is demethylated (described above). Significant differences in the distribution (Fig. 6A) of the CENP-B protein (detected by indirect immunofluores- cence) in comparison to the controls (Fig. 1B and Fig. 2B) can be seen in that both blocks of minor satellite on chromosome 1 are now labelled. This is more easily seen in Fig. 6C. Here the FITC signal (detecting the presence of CENP-B) is shown separately and clearly demonstrates that both blocks of minor satellite are labelled. The PRINS in situ signal is shown in Fig. 6B. The area in the more terminal block of minor satellite sequences on chromosome 1 (this normally binds CENP-B) covered by the anti-CENP-B antibody also appears increased compared to that seen in control cells (compare Figs 1B and 2A,B with 6A and C). Fig. 7 is the result obtained when the anti-CENP-E antibody in combination with oligo C86 labels chromosomal DNA which has been demethylated (as described in Fig. 6). Fig. 7A is the composite image, B and C show the C86 PRINS in situ and the localisation of CENP-E (detected by indirect immuno- fluor-escence), respectively. Fig. 7D is DAPI stained chromo- somes (in grey scale). In Fig. 7 CENP-E remains associated with the more terminal block of minor satellite on chromosome 1 (arrowed). This result is similar to that for untreated chro- mosomes (Fig. 1C) and differs from the result with CENP-B (Fig. 6).

Fig. 5. (A) Indirect immunofluorescence with CREST ACA serum (as for Fig. 1A) on chromosomes with demethylated minor satellite DNA (determined as in Fig. 4, lanes 5-8). Note the secondary sites of labelling corresponding to the internal block of minor satellite on this chromosome (arrowed). (B) PRINS in situ signal using oligo 203. (C) Chromosomes stained with DAPI. case even when it is clear from Southern blots that the chro- mosomal DNA has been substantially demethylated. Therefore, there is not necessarily a link in all instances between demethylation of chromosomal DNA, in particular major satellite sequences, and decondensation of chromatin. Indirect immunofluorescence on metaphase spreads of native chromosomes prepared from cells whose DNA showed demethylation when probed with minor satellite DNA (as described in Materials and Methods) is shown in Fig. 5. In Fig. 5A the signal from CREST now consistently detected a Fig. 6. (A) Composite image PRINS in situ using oligo C86 secondary site of label associated with the more internal minor (CENPB-box) and indirect immunofluorescence with the anti-CENP- B serum on chromosomes with demethylated minor satellite DNA satellite array (arrowed in Fig. 5A). This was not observed in sequences. Note the more diffuse signal over the terminal block of chromosomes from control cells. minor satellite (compare this to Fig. 1B) and the fact that the internal Fig. 6A is a composite image of the results obtained when blocks are positive (arrowed). (B) Split image showing PRINS in situ labelling the chromosomes with the anti-CENP-B antibody and only. (C) Split image showing antibody reaction. (C) DAPI stained oligo C86 (against the CENPB-box) by PRINS in situ hybridi- chromosomes (grey image). 2204 A. R. Mitchell and others

Fig. 7. (A) Composite image PRINS in situ using oligo C86 (CENP-B-box) and indirect immunofluorescence with anti- CENP-E serum on chromosomes with demethylated minor satellite DNA sequences. (B) Split image showing PRINS result. (C) Split image showing antibody reaction. (D) DAPI stained chromosomes (grey image). Note that there is no evidence for a redistribution of anti-CENP-E (arrowed) unlike the result obtained with CENP-B (Fig. 7).

DISCUSSION the kinetochore domain providing structural support to this region. Kitagawa et al. (1995) have proposed that one role for CENP-B is the only protein recognised by CREST anti-cen- this protein may be to promote condensation of those minor tromere serum which has been shown to bind directly to a specific satellite arrays which bind this protein. Evidence to support this DNA motif. Neither CENP-A (a -like protein; Palmer et model comes from the finding (Ikeno et al., 1994) that two al., 1987) nor CENP-C (the protein most likely to be involved in separate arrays of alphoid DNA are found on human chromo- the structure of the kinetochore plate; Cooke et al., 1990) have some 21. The array containing the greater number of CENP-B been shown conclusively to bind to a specific sequence motif binding sites (α-21-I) appeared more compact than the array within centromeric DNA sequences. The α-protein (Strauss and with fewer sites (α-21-II) when viewed under the light micro- Varshavsky, 1984) or HMG I (Y) (Bustin et al., 1990), although scope. In this present report we did not observe major changes complexing to alphoid DNA, is somewhat nonspecific in its to the degree of condensation of the centromeric chromatin on target DNAs (Solomon et al., 1986) and would seem to be an chromosome 1 when the minor satellite was demethylated. Our unlikely candidate in centromere activity. Two other proteins, previous observations (Mitchell et al., 1993) on chromosome 1 that described by Gaff et al. (1994) and the MeCP-2 protein in the DBA/2 mouse showed that the minor satellite sequences (Meehan et al., 1992) do bind directly to centromeric DNA associated with CREST ACA staining existed predominantly as sequences of mammalian chromosomes. However, because there a single block i.e. both were fused. The more internal may well be megabases of DNA within each repetitive DNA block of minor satellite DNA was not fused and existed on family in this region (Kipling et al., 1994) their relationship to separated chromatids. Both blocks of minor satellite contained, the position of the centromere as defined by CREST is yet to be as far as we could tell, identical sequences. Our present obser- determined. An understanding of how CENP-B is targeted to its vations imply that even within the terminal block of minor interactive site within chromatin may well provide clues to how satellite only a subset of all the CENP-B binding sites are used. other proteins interact with centromeric DNAs. This is because of the increase in area of binding to CENP-B The role of the CENP-B protein in the formation of an active antibody which occurs when the sequences in this minor satellite centromere is unclear. From the experiments of Cooke et al. block are demethylated (see Fig. 6). Our model (Fig. 8) proposes (1990) it seems that this protein underlies the kinetochore plate that normally within this terminal block of minor satellite a in the area described by Earnshaw (personal communication) as subset of CENP-B binding sites remain unmethylated at their CenpB binding and methylation 2205

of CENP-B does not necessarily indicate an active centromere. Instead, CENP-B binding may have an accessory role, such as altering the centromeric chromatin structure to facilitate efficient and stable association of other centromere proteins with this region. Our findings are in agreement with such a model in that we find that acquisition of CENP-B binding by the more internal domain does not appear to lead to the formation of a functional centromere, as judged by the binding of the centromeric protein CENP-E and sister chromatid association, features which are seen for the more terminal array. Sullivan and Schwartz (1995) have also come to the conclusion that CENP-E binding reflects the position of the active centromere in dicentric chromosomes. In these dicentrics CENP-B is found at both active and nonactive centromeres whereas CENP-E remains at the active centromere. The mechanism of this proposed model (Earnshaw et al., 1989) is unclear. It is possible that other DNA elements absent from the more internal array are essential for centromere function. Another possibility is that centromere function is inherited in an epigenetic fashion, in a similar manner to the stability of large transcription factor complexes through DNA replication. In this model additional centromere proteins would be targetted to those Fig. 8. Kinetochore imprinting model showing the arrays of minor regions of the genome where such proteins already exist, and satellite (small arrows) with a small unmethylated region (area that following DNA replication some of these proteins are dis- without round symbols) which is the primary signal for CENP-B binding. tributed equally to the daughter strands of the replication fork. A new kinetochore would be formed following division at arrays of minor satellite which bind CENP-B (to facilitate a stable inter- action) and which already have attached some molecules of CpG dinucleotides. It is this subset of unmethylated sites which other additional and essential kinetochore components. This act as the primary signal for the CENP-B protein to bind to the model is attractive as it explains how centromere inactivation chromosomal DNA. Because this domain as a whole is more could occur by DNA methylation and the removal of CENP-B compact than the internal block of minor satellite it follows that binding. The model also predicts that centromere reactivation full occupancy of all the CENP-B binding sites by the protein is would not occur simply by the re-acquisition of CENP-B not necessary for chromosome condensation as envisaged by binding sites. Ikeno et al. (1994). Support in favour of our model comes from One new observation reported here is the effect of regaining the observations of Haaf and Ward (1994). They showed using additional CENP-B binding to regions of minor satellite arrays extended chromatin fibres that although CENP-B binding sites which already associate with an active kinetochore. We show appeared uniformly spread throughout alphoid arrays CREST that demethylation causes CENP-B not only to bind to the ACA labelling covered only a proportion of the alphoid domain. more internal array, but also that it now decorates a much This result is what one would expect to see if only a subset of larger area of the terminal (active) array. As this array is the potential binding sites for the CENP-B protein are used at already active and ‘tagged’ in our model for binding of addi- one time. This subset would be those which remained tional centromere proteins, it would predict a redistribution of hypomethylated at the CpG dinucleotides. other centromere proteins to accompany the change in CENP- Initially we noted that a secondary site for CREST ACA was B binding. However, as CENP-E binding remains similar created in the internal block of minor satellite sequences on unaltered, no such change would appear to be taking place. chromosome 1 after cells were cultured with 5-aza-2′deoxy- This is at odds with the simple model presented above unless cytidine in the medium. This can now be explained by an the kinetochore is a self-assembling organelle with internal size increase in the number of demethylated CpGs in the CENP-B constraints; that is, it is always formed with a defined size and binding site (see above). It remains possible that additional its interaction with the chromosome is to provide attachment binding sites for other centromere proteins such as CENP-A points rather than to define its size. This latter hypothesis is are also created. However, we see no noticable increase in supported (Earnshaw et al., 1989) by the observation that CENP-E binding (Fig. 7) and although additional sites for although the amount of CENP-B labelling varies between chro- other CENPs may be created we have no evidence in support mosomes in the human , CENP-C labelling of this at the present time. Others (Merry et al., 1985; Peretti (probably a better criterion for the size of the kinetochore) et al., 1986; Wandall, 1989) have also noticed secondary appears to be uniform. It may also be possible that CENP-B labelling for CREST ACA serum in human chromosomes. binding has little to do with facilitating centromere function These sites are associated with what is presumed to be the and that the position and size of the mammalian centromere is inactive centromere in dicentric chromosomes. We would defined by interactions of other proteins with other DNA suggest that the labelling of these secondary sites with CREST elements. The apparent absence of minor satellite DNA ACA serum also reflects demethylation of the CpGs within the sequences in the M. musculus Y chromosome (Broccoli et al., CENP-B binding sites of this chromosomal DNA. 1990) and the absence of CENP-B binding sites on the M. It has been argued (Earnshaw et al., 1989) that the presence caroli (Kipling et al., 1995) may be examples 2206 A. R. Mitchell and others of this. The precise role of the highly conserved centromeric Merry, D. E., Pathak, S., Hsu, T. C. and Brinkley, B. R. (1985). Anti- protein CENP-B in centromere biology thus remains to be kinetochore antibodies: use as probes for inactive centromeres. Am. J. Hum. determined in more detail. Genet. 37, 425-430. Mitchell, A. R., Gosden, J. R. and Miller, D. A. (1985). A cloned sequence, p82H, of the alphoid repeated DNA family found at the centromeres of all We thank William C. Earnshaw and Tim Yen for providing us with human chromosomes. Chromosoma 92, 369-377. the antibodies used in this study. Nick Hastie and Howard Cooke are Mitchell, A., Jeppesen, P., Hanratty, D. and Gosden, J. (1992). The thanked for their critical reading of the manuscript. The Photographic organisation of repetitive DNA sequences on human chromosomes with Department of the Unit, comprising Norman respect to the kinetochore analysed using a combination of oligonucleotide Davidson, Sandy Bruce and Douglas Stuart, are also thanked for their primers and CREST anticentromere serum. Chromosoma 101, 333-341. expert advice and help. Mitchell, A. R., Nicol, L., Malloy, P. and Kipling, D. (1993). Novel structural organisation of a Mus musculus DBA/2 chromosome shows a fixed position for the centromere. J. Cell Sci. 106, 79-85. Moroi, Y., Peebles, C., Fritzler, M. J., Steigerwald, J. and Tan, E. M. REFERENCES (1980). Autoantibody to centromere (kinetochore) in scleroderma sera. Proc. Nat. Acad. Sci. USA 77, 1627-1631. Broccoli, D., Miller, O. J. and Miller, D. A. (1990). Relationship of mouse Narayanswami, S., Doggett, N. A., Clark, L. M., Hildebrand, C. E., Weier, minor satellite DNA to centromere activity. Cytogenet. Cell Genet. 54, 182- H.-U. and Hamkalo, B. A. (1992). Cytological and molecular 186. characterisation of centromeres in Mus domesticus and Mus spretus. Mam. Brown, W. and Tyler-Smith, C. (1995). Centromere activation. Trends Genet. Gen. 2, 186-194. 11, 337-339. Palmer, D. K., O’Day, K., Wener, M. H., Andrews, B. S. and Margolis, R. Bustin, M., Lehn, D. A. and Landsman, D. (1990). Structural features of the L. (1987). A 17-kD centromeric protein (CENP-A) copurifies with HMG chromosomal proteins and their . 1049, 231-243. core particles and with . J. Cell Biol. 104, 805-815. Cooke, C. A., Bernat, R. L. and Earnshaw, W. C. (1990). CENP-B: a major Page, S. L., Earnshaw, W. C., Choo, K. H. A. and Shaffer, L. G. (1995). human centromere protein located beneath the kinetochore. J. Cell Biol. 110, Further evidence that CENP-C is a necessary component of active 1475-1488. centromeres: studies of a dic (X;15) with simultaneous immunofluorescence Cooke, C. A., Bazett-Jones, D. P., Earnshaw, W. C. and Rattner, J. B. and FISH. Human Mol. Gen. 4, 289-294. (1993). Mapping DNA within the mammalian kinetochore. J. Cell Biol. 120, Pardue, M. L. and Gall, J. G. (1969). Chromosomal localisation of mouse 1083-1091. satellite DNA. Science 168, 1356-1358. Denhardt, D. T. (1996). A membrane filter technique for the detection of Peretti, D., Maraschio, P., Lambiase, S., Curto, F. Lo and Zuffardi, O. complementary DNA. Biochem Biophys. Res. Commun. 23, 641-646. (1986). Indirect immunoflourescence of inactive centromeres as indicators of Earnshaw, W. C., Ratrie, H. and Stetten, G. (1989). Visualisation of centromeric function. Hum. Genet. 73, 12-16. centromere proteins CENP-B and CENP-C on a stable Pluta, A. F., Cooke, C. A. and Earnshaw, W. C. (1990). Structure of the in cytological spreads. Chromosoma 98, 1-12. human centromere at metaphase. Trends Biochem. Sci. 15, 181-185. Feinberg, A. P. and Vogelstein, B. (1983). A technique for radiolabelling Pluta, A. F., Saitoh, N., Goldberg, I. and Earnshaw, W. C. (1992). DNA restriction endonuclease fragments to high specific activity. Anal. Identification of a subdomain of CENP-B that is necessary and sufficient for Biochem. 132, 6-13. targetting to the human centromere. J. Cell Biol. 116, 1081-1093. Gaff, C., Dusart, D., Kalitsis, P., Lannello, R., Nagy, A. and Choo, K. H. A. Prosser, J., Frommer, M., Paul, C. and Vincent, P. C. (1986). Sequence (1994). A novel nuclear-protein binds centromeric alpha-satellite DNA. J. relationships of three human satellite DNAs. J. Mol. Biol. 187, 145-155. Hum. Genet. 3, 711-716. Radic, M. Z., Lundgren, K. and Hamkalo, B. A. (1987). Curvature of mouse Haaf, T. and Ward, D. C. (1994). Structural analysis of α-satellite DNA and satellite DNA and condensation of . Cell 50, 1101-1108. centromere proteins using extended chromatin and chromosomes. Hum. Mol. Rattner, J. B. (1991). The structure of the mammalian centromere. BioEssays Gen. 3, 697-709. 13, 51-56. Ikeno, M., Masumoto, H. and Okazaki, T. (1994). Distribution of CENP-B Rattner, J. B. and Bazett-Jones, D. P. (1989). Kinetochore structure: electron boxes reflected in CREST centromere antigenic sites on long-range α- spectroscopic imaging of the kinetochore. J. Cell Biol. 108, 1209-1219. satellite DNA arrays of human . Hum. Mol. Genet. 3, 1245- Ris, H. and Witt, P. L. (1981). Structure of the mammalian kinetochore. 1257. Chromosoma (Berl. ) 82, 153-167. Jeppesen, P., Mitchell, A., Turner, B. and Perry, P. (1992). Antibodies to Schulman, I. and Bloom, K. S. (1991). Centromeres: an integrated defined histone epitopes reveal variations in chromatin configuration and protein/DNA complex required for chromosomes movement. Annu. Rev. underacetylation of centric heterochromatin in human metaphase Cell Biol. 7, 311-336. chromosomes. Chromosoma 101, 322-332. Singer, M. F. (1982). Highly repetitive sequences in mammalian . Int. Jones, K. (1970). Chromosomal and nuclear location of mouse satellite DNA in Rev. Cytol. 76, 67-112. individual cells. 225, 912-915. Solomon, M. J., Strauss, F. and Varshavsky, A. (1986). A mammalian high Joseph, A., Mitchell, A. R. and Miller, O. J. (1989). The organisation of the mobility group protein recognises any stretch of six AT base pairs in duplex mouse satellite DNA at centromeres. Exp. Cell Res. 183, 494-500. DNA. Proc. Nat. Acad. Sci. USA 83, 1276-1280. Kipling, D., Wilson, H. E., Mitchell, A. R., Taylor, B. A. and Cooke, H. J. Southern, E. M. (1975). Detection of specific sequences among DNA (1994). Mouse centromere mapping using oligonucleotide probes that detect fragments separated by gel electrophoresis. J. Mol. Biol. 98, 503-517. variants of the minor satellite. Chromosoma 103, 46-55. Strauss, F. and Varshavsky, A. (1984). A protein binds to a satellite DNA Kipling, D., Mitchell, A. R., Masumoto, H., Wilson, E., Nicol, L. and Cooke, repeat at three specific sites that would be brought into mutual proximity by H. J. (1995). CENP-B binds a novel centromeric sequence in the Asian DNA folding in the nucleosome. Cell 37, 889-901. mouse Mus caroli. Mol. Cell Biol. 15, 4009-4020. Sullivan, K. F. and Glass, C. A. (1991). CENP-B is a highly conserved Kitagawa, K., Masumoto, H., Ikeda, M. and Okazaki, T. (1995). Analysis of mammalian centromere protein with homology to the helix-loop-helix protein-DNA and protein-protein interactions of family of proteins. Chromosoma 100, 360-370. (CENP-B) and properties of the DNA-CENP-B complex in the cell cycle. Sullivan, B. A. and Schwartz, S. (1995). Identification of centromeric antigens Mol. Cell Biol. 15, 1602-1612. in dicentric Robertsonian translocatons: CENP-C and CENP-E are necessary Koch, J. E., Kolvraa, S., Petersen, K. B., Gregersen, N. and Bolund, L. components of functional centromeres. Hum. Mol. Gen. 4, 2189-2197. (1989). Oligonucleotide-priming methods for the chromsome-specific Wandall, A. (1989). Kinetochore development in two dicentric chromosomes labelling of alpha satellite DNA in situ. Chromosoma 98, 259-265. in man. A light and electron microscopic study. Hum. Genet. 82, 137-141. Masumoto, H., Masukata, H., Muro, Y., Noaki, N. and Okasaki, T. (1989). Wong, A. K. C. and Rattner, J. B. (1988). Sequence organisation and A human centromere antigen (CENP-B) interacts with a short sequence in cytological organisation of the minor satellite of mouse. Nucl. Acids Res. 16, alphoid DNA, a human centromeric satellite. J. Cell Biol. 109, 1963-1973. 11645-11661. Meehan, R. R., Lewis, J. D. and Bird, A. P. (1992). Characterisation of MeCP2, a vertebrate DNA binding protein with affinity for methylated DNA. Nucl. Acids Res. 20, 5085-5092. (Received 15 January 1996 Ð Accepted 29 May 1996)