- Email: [email protected]
Sequence of centromere separation: Kinetochore formation and DNA replication in dicentric chromosomes showing premature centromere separation in rat cerebral cells
Sequence of Centromere Separation: Kinetochore Formation and DNA Replication in Dicentric Chromosomes Showing Premature Centromere Separation in Rat Cerebral Cells Baldev K. Vig, Neidhard Paweletz, and Dieter Schroeter
ABSTRACT: A subpopu/~tion of rat cerebral endothelial cells, designated B1, exhibits an array of multicentric chromosomes. Because of the jormation of bridges at unuphase, this cell population produced new types of multicentrics at every cell division [71. These chromosomes showed kinetochore proteins ut every centromeric site and all centromeric regions replicated their DNA at the end of the S phase, more or less simultaneously [1]. A new subpopulation of cells, designated B2, obtained from the original sample frozen at Wayne Stnte University displayed several dicentrics. In contrast to B1 these chromosomes exhibit premature centromere separat/on as reported for mouse and human cell lines [sJ. These B2 dicentrics show only one site of
kinetochore protein deposition. The timing of DNA replication around the centric region of prematurely separating centromere is also changed similar to the earlier reported premature DNA synthesis for mouse dicentrics. These observations suggest a universality of relationship between premature centramere separation, a lack of kinetochore formation, and early replica. tion of the centric/per/centric DNA associated with these centromeres. The cause of sudden change from activity to inactivity of these, chromosomes, though interesting, is not clear.
INTRODUCTION With the advent of antikinetochore antibody technique, several reports on the nature and characterization of multicentric chromosomes have appeared. In certain cell lines multicentric chromosomes have been reported to exhibit only one site of kinetochore protein deposition, e.g., in dicentrics in a brain tumor cell line and L cells in mouse [8, 9], human t(X;X) [3], and an octacentric chromosome in L cells of mouse origin [9]. Only in one instance, tdic(9;11) in humans, do both centromeres show deposition of the kinetochore proteins; however, one centromere accumulates much larger quantity than the other [4]. In all these cases, only one centromere is held as
From the Department of Biology, University of Nevada at Reno (B. K. V.) and the Institute of Cell and Tumor Biology, German Cancer Research Center, Heidelberg, F.R.G. (N. P., D. S.).
Address reprint requests to: Baldev K. Vig, Department of Biology, University of Nevada at Reno, Reno, NV 89557. Received November 28, 1988; accepted lanuary 10, 1989.
283 © 1989 Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010
Cancer Genet Cytogenet 38:283-296 (1989) 0165-4608/89/$03.50
284
B.K. Vig et al. one undivided unit until metaphase; others separate at prophase/premetaphase stage of mitosis [8, 9]. Such prematurely separating centromeres have been variously called '~accessory" or "inactive." Apparently these chromosomes segregate equationally like monocentric chromosomes, suggesting that the prematurely separating centromeres do not participate in binding to the spindle microtubules. More recently, it has been shown that the pericentric repetitive DNA and the centric region associated with the prematurely separating centromeres exhibit an altered DNA replication pattern. This DNA, which has the same composition as that flanking the active centromere, replicates much earlier than does the latter [6]. The basic reason for this shift in the timing of replication is not yet clear. It is likely, though, that the DNA associated with the inactive centromere is methylated to lesser extent than that around the active centromere. That undermethylated DNA in mouse does actually replicate earlier in the S phase was shown by studies of Selgi and associates [5]. If so, it can open up interesting avenues of research on replication and function of the centromere. On the other hand, there are reports of multicentric and dicentric chromosomes that exhibit kinetochore protein deposition at every centromeric site. These chromosomes do not show premature centromere separation. An example is seen in a rat cell line, B1, of cerebral endothelial origin [1, 9]. In this case every anaphase and early telophase cell shows chromatid or chromosome type bridges [7]. The breaks in these bridges, followed by rejoining, produce new dicentric and multicentric chromosomes at every cell division. Some centromeres in these multicentric chromosomes do not split until after telophase. In the electron microscope all these centromeres show microtubules binding at the site of the kinetochore structure. Hence, in these chromosomes all centromeres appear to be functional [1]. When pericentric and centric regions associated with the various centromeres on these chromosomes are studied for the timing of their replication, all centromeric regions appear to complete DNA replication toward the end of the S phase. This is in sharp contrast to what is observed in the prematurely separating centromeres in mouse [6], However, it is not known if premature DNA replication of the mouse dicentrics is a unique feature of the species or if dicentrics from any species would utilize the "mouse model" of replication to accommodate premature centromere separation of the accessory centromeres, On the contrary, one may assume that premature separation, a lack of kinetochore formation, and early DNA replication are fortuitous events and may have nothing to do with the centromere activity. It would therefore be of interest to find support for or against the universality of these associations. In view of the past studies on rat, this organism can provide a suitable system. Such material happens to be at hand. A new subsample from the bulk material, from which B1 was derived, exhibited several dicentric chromosomes. These appeared in the descendants without any alteration. This proved to be due to premature separation of one of the centromeres in certain dicentrics in these cells. At the same time similar phenomenon was observed for dicentrics in a clone of cells started from a single B1 cell. The reason for this sudden transformation from activity to inactivity of the accessory centromere is not clear. However, it was thought of interest to find out if, like mouse, these dicentrics show premature centromere separation by forming only one kinetochore (or do these form a kinetochore at every centromeric site), and whether these show differential replication timing of the centric/pericentric region. The present study reports that prematurely separating centromeres in rat behave like those in mouse and that the phenomenon of premature centromere separation is associated with a lack of deposition of kinetochore proteins at the site of the accessory centromeres as well as premature DNA replication in the centric/pericentric re~;ion.
Centromere Separation
285
MATERIALS AND METHODS A cell line established from a cerebral endothelial tumor of a Wistar-Kyoto rat at Wayne State University [2] has been in use in our laboratory since 1985. This cell line, as originally described, exhibited several multicentric chromosomes in culture studied in Reno. All multicentric chromosomes showed kinetochore antibody binding site at every centromere [9]. In 1986, a sample was sent to the Deutsches Krebsforschungszentrum in Heidelberg where we observed multiple bridges in virtually every cell analyzed at anaphase and as late as the ensuing interphase joining the two daughter cells. It turned out that chromatid and chromosome bridges broke and rejoined in these cells, giving rise to new multicentric chromosomes [7]. This cell line was designated B1. In 1987 we received another sample out of the original cell population kept in the deep freeze at Wayne State. These cells were designated as B2. Upon analysis these cells showed only a few bridges and more dicentrics than the original subpopulation (B1). It was noticed that most dicentrics present in these cells were being carried through mitotic division without causing bridges. The new dicentrics, in contrast to B1, were rare in B2. In order to study certain parameters of cell division in B1 we attempted to raise single cell clones from these cells. Only one cell gave rise to a reasonably growing colony. This was named BLAB. The progeny of these cells turned out to be heterogeneous. Many cells had dicentr~cs and multicentric chromosomes. However, in this heterogeneous population of cells there existed several dicentrics that appeared to perpetuate without any alterations. This is not to say that these two cell lines did not have any mitotic aberrations. We have noticed chromosome instability in that chromatid translocations appear in some cells. These appear to result in bridges at anaphase. However, the present study does not address this question and is confined to the analysis of recognizable transmissible dicentrics. The B2 and BlAB (hereafter called AS) cells were maintained both at Reno and Heidelberg in McCoy's MEM supplemented with 15% fetal calf serum and 100 units each of penicillin and streptomycin. The cells were grown at 37°C with or without CO~. For routine metaphase analysis these were treated with Colcemid for I hour and subjected to 15 minutes 0.075 M KCI as hypotonic treatment. The cells were fixed in acetic acid: methanol (3 : 1). For C banding the slides were treated with 0.2N HCi for 50 minutes at room temperature followed by 14 minutes of treatment in saturated solution of BaOH at 37°C and 2-3 hours incubation in 2 × SSC at 65°C. These were stained with 4% Giemsa. In order to detect the location of kinetochore proteins, the colcemid-treated cells were subjected to a 15-minute hypotonic treatment. These were deposited onto glass slides with the help of a Shadon cytospin. The slides were immediately dipped in 70% ethanol (-40 to -80°C) and then kept at -20°C for at least 20 minutes. The cells, washed in Dulbecco's buffer, were then treated with antinuclear antibody (Davis Antibodies Inc., Davis) for 30 minutes followed by thorough washing and the application of FITC-conjugated goat antihuman lgG (US Biochemical, Cleveland) for another 30 minutes. These steps were carried out at room temperature or in the incubator at 37°C. After another thorough washing the slides were stained with 1:1000 ethidium bromide: water, washed again with Dulbecco's buffer, and mounted in 3% n-propylgallate in glycerine. These were observed under a Zeiss epifluorescence microscope using BP 450/20 excitation filter. For study of DNA replication the exponentially growing cells were given 10 -6 M bromodeoxyuridine (BrdU) for 6 hour. These were treated with Colcemid for I hour
286
B.K. Vig et al. to collect metaphases. The slides were prepared at 15-minute KC1 hypotonic treatment followed by acetic acid: methanol fixation and were flame dried. Denaturation of DNA was achieved by dipping the slides in 0.05 M NaOH for 10-20 seconds. The anti-BrdU antibody (a gift from Dr. L. Stanker, Lawrence Livermore Laboratories) was applied at a dilution of I : 100 for 30 minutes. After washing in Dulbecco's phosphate buffer these were again treated with FITC-conjugated goat antimouse IgG (US Biochemical, Cleveland) for another 30 minutes. Both steps were performed at room temperature. The slides, thoroughly washed in Dulbecco's buffer, were then stained with ethidium bromide to provide counter stain to the FITC. Again, a Zeiss microscope equipped with BP 450/20 excitation filter was used to study fluorescence. Photographs were taken on Kodak HC Technical Pan black and white film (ASA 100) or on Kodak Kodachrome color film (ASA 400). Black and white pictures were made directly from these negatives or diapositives.
RESULTS General Characteristics of the Dicentrics Giemsa-stained cells from the B2 population showed from as few as three to 11 dicentrics in various cells. The number of chromosomes in these cells also varied considerably. The most common dicentrics were those that had one centromere at the terminal and the other somewhere in the arm. The position of the latter was not fixed. Dicentrics of this type also showed variations in their length. At prophase these dicentrics showed the two chromatids held tightly at both centromeres. However, by mid metaphase, one of the centromeres had separated into two daughter units, while the terminal was still held as one entity. Such premature separation was quite evident in a vast majority of our samples {Fig. la). Whereas all cells analyzed had these so-called asymmetric dicentrics, a small proportion also showed one or two symmetric dicentrics. The dicentrics present in the A8 clone, however, were fewer in number and the majority of cells had one to two dicentrics with both centromeres located subterminally near the two ends (Fig, lb), These cells also showed a few dicentrics of the type frequently found in B2 cells, namely, the ones with one centromere located terminally and the other in the arm. Generally, the former type of dicentric separated one of its centromeres later than did those that had a terminal centromere. Obtaining good C bands on rat chromosomes was difficult and all chromosomes did not show C bands. However, we obtained a few slides with reasonable C bands on all types of dicentrics {Fig. 2). These confirmed the presence of C bands at the location of the two centromeres. It is not possible to say whether all dicentrics had C chromatin at all centromeres or only a few, like the ones exemplified in Figure 2, really had them. One interesting observation that could not be clearly made with non-C-banded cells was the presence of the centromeres at each of the two ends of some chromosomes. Premature centromere separation in such chromosomes would be almost impossible to study in non-C-banded Giemsa-stained preparations. For the purpose of this study we would call the two types of dicentrics type I and type 2. The former type is the one that has one terminal centromere; the other being placed subterminally. The study was confined only to those dicentrics that could be identified with certainty. Such unequivocal identification becomes difficult when one considers that the chromosomes stained with ethidium bromide after the application of antikinetochore antibody or BrdU antibody do not lend themselves to morphologic discrimination along their length as do those stained with Giemsa.
0
4b ti
II
Figure 1 Examples of type 1 dicentrics from a B2 cell Ca). This cell has several dicentrics, most of which show premature centromere separation (arrowheads). At a somewhat later stage in mitosis, these dicentrics look like typical metaphase chromosomes with only terminal centromere. Type 2 dicentrics are shown in (b], in which one such dicentric is held at both centromeres (large arrowheads) while some type 1 dicentrics show accessory centromeres already having had separated (small arrowheads). The cell shown in Ca) is from B2 culture, that in (b) is from A8.
.Q
•
itb
0
Q
#
t~
~ 0 ....
Figure 2 C-banded cells showing type 2 dicentric (a) and type I dicentrics (b). The two centromeres and heterochromatin blocks in type 2 are marked by large arrowheads; the small arrow ill (a) points to a dicentric with two terminally located centromeres. Several chromosomes in these photographs show no detectable C bands. Both cells were taken from B2 culture.
':i!,,~
Oo Oo
t~
Centromere Separation
289
o
~ ~i,~.
,.
~!i
~...... W
31 Figure 3 Kinetochore localization on dicentrics in B2. The large chromosome (marked with arrowhead) in (a) is type 2 dicentric, whereas the other two marked by arrowheads appear to be type 1 dicentrics. In (b) a type 1 dicentric is shown with only one pair of kinetochore dots located at the terminal. In both photographs, the arrowheads point to the location of prematurely separated centromere as detected either by rhodamine filter or apparent in these photographs.
Location of IQnetochore Protein Deposition on the Dicentrics Unlike rat B1 cells, where dicentric and multicentric chromosomes showed kinetochore proteins deposited at every centromeric site, the dicentrics in B2 as well as A8 showed kinetochore proteins at only one site per chromosome. Type I dicentrics had kinetochore antibody binding site at the position of the terminal centromere both in B2 (Fig.3) and A8 (Fig. 4). In most cases it was not possible to discern the location of the "inactive" centromere. However, because of reduced spreading of the two chroFisure 4 A type 2 dicentric (arrow marks the position of the prematurely separated centromere) in an A8 cell (a) and two type 2 dicentrics in the same cell line (b). Note that various kinetochore dots are of more or less similar size and that no coalescence or compound kinechores are seen, as has been reported for B1 cells [9].
~'L¸~¸~¸ii~
iliiii!~ ~!I¸I~
290
B.K. Vig et al. matids at the site of this centromere or its less dense ethidium bromide staining, the constriction could be recognized in 87 chromosomes in B2 and 20 chromosomes in AS. The type 2 dicentrics also showed kinetochores at only oue site. Because the difference between the lengths of the segments from the distal end to the centromere was definable, it was possible to determin~ which centromere had the kinetochore. The photographs reproduced here both from B2 (Fig. 3) and A8 (Fig. 4) leave little doubt that the active centromere is the one located closer to the terminal of the dicentric. Because this dicentric is present quite frequently in the clone A8 we could study more than 100 such chromosomes for the presence of kinetochore whereas in B2 only 15 such dicentrics could be studied. In all cases the dicentrics had only one kinetochore. It is worth noticing that every analyzable chromosome in these two types of cells had only one kinetochore and most all of them appeared similar in size. There was no instance in which a chromosome showed two adjacently placed kinetochores or had coalescing kinetochores as has been seen in earlier sample, B1 [1, 9].
Replication of Pericentric/Centric DNA After 6 hours of BrdU incorporation, virtually every cell at metaphase showed label. However, the cells of interest were the ones that showed differential incorporation of BrdU in the two centric regions. Such chromosomes were found, though the search for such cells turned out to be time consuming. Nonetheless, a total of 240 dicentric chromosomes met the criterion. Out of these 175 were of type 1, the remaining being of type 2.
Replication oJ type 1 dicentrics. In most cells in which differential replication of the two pericentric regions was clear, the accessory centromere had generally already separated prematurely. Under BP 540/20 filter differentiation between the primary constriction at the site of prematurely separated centromare and the flanking chromatid segments was not clear, However, under rhodamine filter such regions could be distinguished even though some times the constriction was barely discernible. The type I dicentrics showed completion of replication of the accessory centromere and associated heterochromatin when the functional centromere (terminally located) still showed BrdU incorporation. Rather heavy incorporation of BrdU in such dicentrics was also visible even in the euchromatic regions. It appeared that the euchromatic segments at the junction of heterochromatin replicated after the pericentric heterochromatin and the prematurely separating centromere had finished replication (Fig. 5a). In rare cells there also appeared a tricentric in which the only replicating centromere was the one located terminally. The early replication of centromeric/pericentric region of the accessory centromere was somewhat better depicted in other cells (Fig. 5b, arrow). As seen in the corresponding photograph taken by using the rhodamine filter (Fig. 5c) there is a constriction in the region of the chromosome that shows no BrdU incorporation in Figure 5b. Some chromosomes in these cells had completed the replication of their entire DNA before BrdU was added (Fib. 5b, double arrowheads). There were instances in which the main centromere showed only a trace of BrdU incorporation (Fig. 6). Here also the early replication of the accessory centromere was more clearly visible. A chror~osome in the right middle of Figure 6a, with corresponding Giemsa-stained cell in Figure 6b, is another example in which the replication at the junction of heterochromatin/euchromatin is evident. However, three type 1 dicentrics shown in BrdU-incorporated cell in Figure 7a and corresponding Giemsa-stained cell leave little doubt that the accessory centromeres replicate earlier
Centromere Separation
291
Figure 5 (a) Part of a B2 cell showing three dicentrics (single arrowheads) and an apparent tricentric chromosome (two arrowheads). The arrows point to the centric region, which is sandwiched between the fluorescing euchromatin. In all these cases the inactive centromeres have separated and appear to have completed replication before that of the active (terminal) centromeres that show incorporation of BrdU. In (b) the situation is somewhat clearer in that the inactive centromere shows only weak and much reduced incorporation of BrdU compared to the euchromatin located below it (arrow). Location of the accessory centromere can be seen in the corresponding photograph taken under rhodamine filter (c). Certain chromosomes in these cells have t:ompleted their replication and show no sign of BrdU incorporation [double arrowhead in {b}].
than do the functional ones. In these instances there is no indication of any BrdU incorporation in the region of the accessory centromeres.
Replication of type 2 dicentr/cs. The situation regarding early replication of the accessory centromere was quite clear for type 2 dicentrics. In most instances, finding the cells with differential BrdU incorporation in the two centric regions was no problem. In all such dicentrics, the centromere located closer to the terminal of the chromosome showed incorporation of BrdU whereas the other showed completion of DNA replication having had occurred. This late replicating centromere is the same one that deposits the kinetochore proteins (Fig. 4J and apparently is the functional centremere. Figure 7a is an example of such clear diffe:3ntial incorporation. As the accompanying photograph (Fig. 7b) of the same Giemsa-stained cell shows, the early replicating centromere is the one located further away from the terminal and is the one that shows premature separation. Two type 2 dicentrics shown in Figure 8a also present a similar clear picture of premature rep~ication. Figure 8b, taken from a cell
292
B.K. Vig et al.
m
l
P t
mO
iim
b Figure 6 A B2 cell showing a somewhat advanced stage in differential replication of accessory centromere compared to that of the functional one. The fluorescing cell {a) and its Giemsastained counterpart {b) exhibited two type 1 dicentrics showing differential incorporation. In the dicentric marked with an arrowhead in the left of the photograph there is little or no BrdU incorporation at the site of and beyond the arrowhead that marks the position of the accessory centromere. The dicentric to the left shows the last faint incorporation in the active centromere, whereas the inactive centromere is totally devoid of incorporation.
that was Giemsa stained after washing off the BrdU antibody, shows an example in which the least detectable incorporation of BrdU is seen right in the active centromere. The heterochromatin block being minute in these centric regions (Fig. 2), the BrdU bands appear very close to the two centromeres. In summary, the rat accessory centromeres do not appear to deposit kinetochore proteins as do the centromeres on chromosomes that do not express premature separation as seen in B1 cells. The centric and pericentric regions associated with these accessory centromeres also replicate earlier than similar regions associated with the
T'
m
Figure 7 Part of an A8 cell showing both type I and type 2 dicentrics with clearly differential incorporation in the vicinity of active versus inactive centromeres (a). The three type 1 dicentrics exhibit the second centromere in the same cell when stained with Giemsa (b). The region of the chromosome on both sides of and including the inactive centromere is clearly devoid of any BrdU incorporation except for a faint trace in the dicentric shown at the top left of the photograph. These pictures also show a type 2 dicentric that is clearly devoid of any incorporation in the vicinity of and including the inactive centromere, while the region of the active centromere and some large euchromatin blocks show heavy incorporation of the thymidine analog in (a). The corresponding Giemsa-stained dicentric in {b) shows the position of the inactive centromere that is in the process of premature separation but can still be recognized as primary constriction. Note that the distances from the centromeres to terminals in this photograph are similar to what is seen in Fig. lb.
1
t
t~
d
Figure 8 Two type 2 dicentrics showing clear differential incorporation in the two centric regions (a). The accessory centromeres have separated prematurely. These are marked by arrowheads at the point of faintly visible constrictions. Compare the pattern of replication of the two chromosomes with that seen in Fig. 7a. This provides additional support to the fact that the two marked chromosomes are type 2 dicentrics. In (b) is a cell that after BrdU antibody treatment was washed with alcohol and then Giemsa stained. BrdU bands are visible all over. Even though this particular cel~ldid not appear suitable for photography under fluorescence, the type 2 dicentric (the centromeres marked by arrowheads) shows differential BrdU incorporation in the centromeres proper. Note the presence of dark band at the exact site of the active centromere (large arrowhead) and a lack of such incorporation in the accessory centromere. The presence of heavy label as indicated by the BrdU bands all over the chromosome may suggest that the accessory centromere and its associated pericentric heterochromatin replicate much earlier than does most euchromatin.
r' n'f
Centromere Separation
295
active centromeres. Hence, in view of these data and those from mouse cell lines [6], it seems that these mechanisms may constitute universal strategy adopted by the dicentrics to create a condition mimicking monocentricity.
DISCUSSION
Cancer cells show many karyotypic abnormalities. A newly emerging study of such cells is the characterization of dicentric and multicentric chromosomes. For a long time the appearance of such chromosomes had been taken to mean that the cell will not propagate these abnormal chromosomes to the next generation because of formation of bridges at anaphase. It now appears that in spite of the formation of bridges, when broken ends of chromosomes rejoin new dicentric and multicentric chromosomes can emerge [7]. Not only can the cell divide, albeit with constant karyotypic evolution, but sometimes these chromosomes can undergo a meioticlike phenomenon in mitotic cells, as has been reported for some dicentrics in the B1 cell line of rat endothelial origin. Another strategy used by dicentric and multicentric chromosomes in cancer cells is the "silencing" of all centromeres except one. These centromeres separate prematurely during premetaphase before the centromeres that bind to spindle do so. In so doing, somehow, the silent or accessory centromeres fail to bind the kinetochore proteins as detected by antikinetochore antibody tag. A brain tumor cell line in mouse exhibits this characteristic [8], as does the L-929 cell line [9]. It was not known if the formation of kinetochores at the site of silent centromeres is a specialized property of mouse cells and a few human dicentrics, e.g., tdic(X;X) [3] or is a universal phenomenon common to all prematurely separating centromeres. Such studies need to be carried out with cell lines in which the two types of cantromeres could be studied. One type would be the ones in which the dicentrics and multicentrics do not show any inactivity of any centromere, where no centromere in the dicentric shows premature separation, and both centromeres bind to microtubules. The second type would be the one in which one centromere acts as silent or accessory, separates prematurely, and does not bind to the microtubules during cell division. These criteria can be met with the cell line under study in this report. Like the mouse dicentrics, the prematurely separating centromeres in both type of dicentrics studies show only one site of kinetochore proteins, i.e., at the site of the active centromere. In more than 100 dicentrics of both types analyzed in this study, none showed the deposition of kinetochore proteins at more than one site. The inactivity of the accessory centromere can therefore be ascribed to a lack of the kinetochore and not the absence of the centromere. One may thus argue that microtubules bind to the kinetochorema facultative proteinaceous organelle superposed onto the centromeremand not to the primary constriction called the centromere. These studies, along with those from human and mouse dicentrics, lend credence to the idea that premature centromere separation is linked to the lack of formation of the kinetochore. It is not clear why a kinetochore would not form at the site of accessory centremere in these dicentrics. However, it is becoming clear why the accessory centremeres would split prematurely. In mouse, we have shown that the replication of the pericentric region and the centromere proper of the accessory centromere occurs before the replication of the functional centromere takes place [6]. This replication of the hetarochromatin or repetitive DNA in mouse appears to move right into the centromere. Upon replication the centromere splits into two units, if not right away then after some sort of maturation process has taken place. This would explain premature centromere separation. The timing of centromere replication before the
296
B.K. Vig etal. onset of metaphase or availability of spindle appears to be the crucial factor. In rat the present study also appears to point to similar control of premature centromere separation. The evidence presented here shows premature replication of the pericentric/centric DNA as seen in mouse. Whereas this pattern of replication is similar to what is observed for mouse, it is not in conformity with the pattern of replication for the centromeric regions in dicentrics in which premature centromere separation is not seen. In the latter all centromeric regions in a given chromosome replicate almost simultaneously toward the end of S phase [1]. The BrdU incorporation in these centric regions is equivalent at all sites at a time when the entire genome, except the centric regions, has completed replication. Thus our present studies indicate that premature DNA replication is unique to the prematurely separating centromeres and may be the reason for such separation. In summary, it appears that dicentric and multicentric chromosomes in cancer cells need to be studied more carefully. One or a few karyotypes from one cell cycle may not provide information about the nature and characteristics of these multicentric chromosomes. Whether cancer cells have a special potential developed to handle multicentricity is not known. Nonetheless, it is clear that in cancer cells premature centromere separation and early DNA replication of the accessory centromeres may be routine for certain types of multicentrics. The authors thank Drs. M. Tyrkus and C. Diglio for the gift of the cell line. B. K. V. is grateful to the DKFZ for the award of fellowships.
REFERENCES
1. Broccoli D, Paweletz N, Vig BK (1989): Sequence of centromere separation: Characterization of multicentric chromosomes in a rat cell line. Chromosoma (in press). 2. Diglio CA, Wolff DE, Meyers P (1983): Transformation of rat cerebral endothelial cells by Rous sarcoma virus. J Cell Biol 97:15-21. 3. Earnshaw WC, Migeon BR (1985): Three related centromere proteins are absent from the inactive centromere of a stable isodicentric. Chromosoma 92:290-296. 4. Merry DE, Pathak S, Hsu TC, Brinkley BR (1985): Antikinetochore antibodies: Use as a probe for inactive centromeres. Am J Hum Genet 37:425-430. 5. Selgi S, Ariel M, Goitein R, Marcus M, Cedar H (1988): Regulation of mouse satellite DNA replication time. EMBO J 7:419-426. 6. Vig BK, Broccoli D (1988): Sequence of centromere separation: differential replication of pericentric heterochromatin in multicentric chromosomes. Chromosoma 96:311-317. 7. Vig BK, Paweletz N (1988): Sequence of centromere separation: generation of unstable multicentric chromosomes in a rat cell line. Chromosoma 96:275-282. 8. Vig BK, Zinkowski RP (1986): Sequence of centromere separation: A mechanism for orderly separation of dicentrics. Cancer Genet Cytogenet 23:347-359. 9. Zinkowski RP, Vig BK, Broccoli D (1986): Characterization of kinetochores in multicentric chromosomes. Chromosoma 94:243-248.
ABSTRACT: A subpopu/~tion of rat cerebral endothelial cells, designated B1, exhibits an array of multicentric chromosomes. Because of the jormation of bridges at unuphase, this cell population produced new types of multicentrics at every cell division [71. These chromosomes showed kinetochore proteins ut every centromeric site and all centromeric regions replicated their DNA at the end of the S phase, more or less simultaneously [1]. A new subpopulation of cells, designated B2, obtained from the original sample frozen at Wayne Stnte University displayed several dicentrics. In contrast to B1 these chromosomes exhibit premature centromere separat/on as reported for mouse and human cell lines [sJ. These B2 dicentrics show only one site of
kinetochore protein deposition. The timing of DNA replication around the centric region of prematurely separating centromere is also changed similar to the earlier reported premature DNA synthesis for mouse dicentrics. These observations suggest a universality of relationship between premature centramere separation, a lack of kinetochore formation, and early replica. tion of the centric/per/centric DNA associated with these centromeres. The cause of sudden change from activity to inactivity of these, chromosomes, though interesting, is not clear.
INTRODUCTION With the advent of antikinetochore antibody technique, several reports on the nature and characterization of multicentric chromosomes have appeared. In certain cell lines multicentric chromosomes have been reported to exhibit only one site of kinetochore protein deposition, e.g., in dicentrics in a brain tumor cell line and L cells in mouse [8, 9], human t(X;X) [3], and an octacentric chromosome in L cells of mouse origin [9]. Only in one instance, tdic(9;11) in humans, do both centromeres show deposition of the kinetochore proteins; however, one centromere accumulates much larger quantity than the other [4]. In all these cases, only one centromere is held as
From the Department of Biology, University of Nevada at Reno (B. K. V.) and the Institute of Cell and Tumor Biology, German Cancer Research Center, Heidelberg, F.R.G. (N. P., D. S.).
Address reprint requests to: Baldev K. Vig, Department of Biology, University of Nevada at Reno, Reno, NV 89557. Received November 28, 1988; accepted lanuary 10, 1989.
283 © 1989 Elsevier Science Publishing Co., Inc. 655 Avenue of the Americas, New York, NY 10010
Cancer Genet Cytogenet 38:283-296 (1989) 0165-4608/89/$03.50
284
B.K. Vig et al. one undivided unit until metaphase; others separate at prophase/premetaphase stage of mitosis [8, 9]. Such prematurely separating centromeres have been variously called '~accessory" or "inactive." Apparently these chromosomes segregate equationally like monocentric chromosomes, suggesting that the prematurely separating centromeres do not participate in binding to the spindle microtubules. More recently, it has been shown that the pericentric repetitive DNA and the centric region associated with the prematurely separating centromeres exhibit an altered DNA replication pattern. This DNA, which has the same composition as that flanking the active centromere, replicates much earlier than does the latter [6]. The basic reason for this shift in the timing of replication is not yet clear. It is likely, though, that the DNA associated with the inactive centromere is methylated to lesser extent than that around the active centromere. That undermethylated DNA in mouse does actually replicate earlier in the S phase was shown by studies of Selgi and associates [5]. If so, it can open up interesting avenues of research on replication and function of the centromere. On the other hand, there are reports of multicentric and dicentric chromosomes that exhibit kinetochore protein deposition at every centromeric site. These chromosomes do not show premature centromere separation. An example is seen in a rat cell line, B1, of cerebral endothelial origin [1, 9]. In this case every anaphase and early telophase cell shows chromatid or chromosome type bridges [7]. The breaks in these bridges, followed by rejoining, produce new dicentric and multicentric chromosomes at every cell division. Some centromeres in these multicentric chromosomes do not split until after telophase. In the electron microscope all these centromeres show microtubules binding at the site of the kinetochore structure. Hence, in these chromosomes all centromeres appear to be functional [1]. When pericentric and centric regions associated with the various centromeres on these chromosomes are studied for the timing of their replication, all centromeric regions appear to complete DNA replication toward the end of the S phase. This is in sharp contrast to what is observed in the prematurely separating centromeres in mouse [6], However, it is not known if premature DNA replication of the mouse dicentrics is a unique feature of the species or if dicentrics from any species would utilize the "mouse model" of replication to accommodate premature centromere separation of the accessory centromeres, On the contrary, one may assume that premature separation, a lack of kinetochore formation, and early DNA replication are fortuitous events and may have nothing to do with the centromere activity. It would therefore be of interest to find support for or against the universality of these associations. In view of the past studies on rat, this organism can provide a suitable system. Such material happens to be at hand. A new subsample from the bulk material, from which B1 was derived, exhibited several dicentric chromosomes. These appeared in the descendants without any alteration. This proved to be due to premature separation of one of the centromeres in certain dicentrics in these cells. At the same time similar phenomenon was observed for dicentrics in a clone of cells started from a single B1 cell. The reason for this sudden transformation from activity to inactivity of the accessory centromere is not clear. However, it was thought of interest to find out if, like mouse, these dicentrics show premature centromere separation by forming only one kinetochore (or do these form a kinetochore at every centromeric site), and whether these show differential replication timing of the centric/pericentric region. The present study reports that prematurely separating centromeres in rat behave like those in mouse and that the phenomenon of premature centromere separation is associated with a lack of deposition of kinetochore proteins at the site of the accessory centromeres as well as premature DNA replication in the centric/pericentric re~;ion.
Centromere Separation
285
MATERIALS AND METHODS A cell line established from a cerebral endothelial tumor of a Wistar-Kyoto rat at Wayne State University [2] has been in use in our laboratory since 1985. This cell line, as originally described, exhibited several multicentric chromosomes in culture studied in Reno. All multicentric chromosomes showed kinetochore antibody binding site at every centromere [9]. In 1986, a sample was sent to the Deutsches Krebsforschungszentrum in Heidelberg where we observed multiple bridges in virtually every cell analyzed at anaphase and as late as the ensuing interphase joining the two daughter cells. It turned out that chromatid and chromosome bridges broke and rejoined in these cells, giving rise to new multicentric chromosomes [7]. This cell line was designated B1. In 1987 we received another sample out of the original cell population kept in the deep freeze at Wayne State. These cells were designated as B2. Upon analysis these cells showed only a few bridges and more dicentrics than the original subpopulation (B1). It was noticed that most dicentrics present in these cells were being carried through mitotic division without causing bridges. The new dicentrics, in contrast to B1, were rare in B2. In order to study certain parameters of cell division in B1 we attempted to raise single cell clones from these cells. Only one cell gave rise to a reasonably growing colony. This was named BLAB. The progeny of these cells turned out to be heterogeneous. Many cells had dicentr~cs and multicentric chromosomes. However, in this heterogeneous population of cells there existed several dicentrics that appeared to perpetuate without any alterations. This is not to say that these two cell lines did not have any mitotic aberrations. We have noticed chromosome instability in that chromatid translocations appear in some cells. These appear to result in bridges at anaphase. However, the present study does not address this question and is confined to the analysis of recognizable transmissible dicentrics. The B2 and BlAB (hereafter called AS) cells were maintained both at Reno and Heidelberg in McCoy's MEM supplemented with 15% fetal calf serum and 100 units each of penicillin and streptomycin. The cells were grown at 37°C with or without CO~. For routine metaphase analysis these were treated with Colcemid for I hour and subjected to 15 minutes 0.075 M KCI as hypotonic treatment. The cells were fixed in acetic acid: methanol (3 : 1). For C banding the slides were treated with 0.2N HCi for 50 minutes at room temperature followed by 14 minutes of treatment in saturated solution of BaOH at 37°C and 2-3 hours incubation in 2 × SSC at 65°C. These were stained with 4% Giemsa. In order to detect the location of kinetochore proteins, the colcemid-treated cells were subjected to a 15-minute hypotonic treatment. These were deposited onto glass slides with the help of a Shadon cytospin. The slides were immediately dipped in 70% ethanol (-40 to -80°C) and then kept at -20°C for at least 20 minutes. The cells, washed in Dulbecco's buffer, were then treated with antinuclear antibody (Davis Antibodies Inc., Davis) for 30 minutes followed by thorough washing and the application of FITC-conjugated goat antihuman lgG (US Biochemical, Cleveland) for another 30 minutes. These steps were carried out at room temperature or in the incubator at 37°C. After another thorough washing the slides were stained with 1:1000 ethidium bromide: water, washed again with Dulbecco's buffer, and mounted in 3% n-propylgallate in glycerine. These were observed under a Zeiss epifluorescence microscope using BP 450/20 excitation filter. For study of DNA replication the exponentially growing cells were given 10 -6 M bromodeoxyuridine (BrdU) for 6 hour. These were treated with Colcemid for I hour
286
B.K. Vig et al. to collect metaphases. The slides were prepared at 15-minute KC1 hypotonic treatment followed by acetic acid: methanol fixation and were flame dried. Denaturation of DNA was achieved by dipping the slides in 0.05 M NaOH for 10-20 seconds. The anti-BrdU antibody (a gift from Dr. L. Stanker, Lawrence Livermore Laboratories) was applied at a dilution of I : 100 for 30 minutes. After washing in Dulbecco's phosphate buffer these were again treated with FITC-conjugated goat antimouse IgG (US Biochemical, Cleveland) for another 30 minutes. Both steps were performed at room temperature. The slides, thoroughly washed in Dulbecco's buffer, were then stained with ethidium bromide to provide counter stain to the FITC. Again, a Zeiss microscope equipped with BP 450/20 excitation filter was used to study fluorescence. Photographs were taken on Kodak HC Technical Pan black and white film (ASA 100) or on Kodak Kodachrome color film (ASA 400). Black and white pictures were made directly from these negatives or diapositives.
RESULTS General Characteristics of the Dicentrics Giemsa-stained cells from the B2 population showed from as few as three to 11 dicentrics in various cells. The number of chromosomes in these cells also varied considerably. The most common dicentrics were those that had one centromere at the terminal and the other somewhere in the arm. The position of the latter was not fixed. Dicentrics of this type also showed variations in their length. At prophase these dicentrics showed the two chromatids held tightly at both centromeres. However, by mid metaphase, one of the centromeres had separated into two daughter units, while the terminal was still held as one entity. Such premature separation was quite evident in a vast majority of our samples {Fig. la). Whereas all cells analyzed had these so-called asymmetric dicentrics, a small proportion also showed one or two symmetric dicentrics. The dicentrics present in the A8 clone, however, were fewer in number and the majority of cells had one to two dicentrics with both centromeres located subterminally near the two ends (Fig, lb), These cells also showed a few dicentrics of the type frequently found in B2 cells, namely, the ones with one centromere located terminally and the other in the arm. Generally, the former type of dicentric separated one of its centromeres later than did those that had a terminal centromere. Obtaining good C bands on rat chromosomes was difficult and all chromosomes did not show C bands. However, we obtained a few slides with reasonable C bands on all types of dicentrics {Fig. 2). These confirmed the presence of C bands at the location of the two centromeres. It is not possible to say whether all dicentrics had C chromatin at all centromeres or only a few, like the ones exemplified in Figure 2, really had them. One interesting observation that could not be clearly made with non-C-banded cells was the presence of the centromeres at each of the two ends of some chromosomes. Premature centromere separation in such chromosomes would be almost impossible to study in non-C-banded Giemsa-stained preparations. For the purpose of this study we would call the two types of dicentrics type I and type 2. The former type is the one that has one terminal centromere; the other being placed subterminally. The study was confined only to those dicentrics that could be identified with certainty. Such unequivocal identification becomes difficult when one considers that the chromosomes stained with ethidium bromide after the application of antikinetochore antibody or BrdU antibody do not lend themselves to morphologic discrimination along their length as do those stained with Giemsa.
0
4b ti
II
Figure 1 Examples of type 1 dicentrics from a B2 cell Ca). This cell has several dicentrics, most of which show premature centromere separation (arrowheads). At a somewhat later stage in mitosis, these dicentrics look like typical metaphase chromosomes with only terminal centromere. Type 2 dicentrics are shown in (b], in which one such dicentric is held at both centromeres (large arrowheads) while some type 1 dicentrics show accessory centromeres already having had separated (small arrowheads). The cell shown in Ca) is from B2 culture, that in (b) is from A8.
.Q
•
itb
0
Q
#
t~
~ 0 ....
Figure 2 C-banded cells showing type 2 dicentric (a) and type I dicentrics (b). The two centromeres and heterochromatin blocks in type 2 are marked by large arrowheads; the small arrow ill (a) points to a dicentric with two terminally located centromeres. Several chromosomes in these photographs show no detectable C bands. Both cells were taken from B2 culture.
':i!,,~
Oo Oo
t~
Centromere Separation
289
o
~ ~i,~.
,.
~!i
~...... W
31 Figure 3 Kinetochore localization on dicentrics in B2. The large chromosome (marked with arrowhead) in (a) is type 2 dicentric, whereas the other two marked by arrowheads appear to be type 1 dicentrics. In (b) a type 1 dicentric is shown with only one pair of kinetochore dots located at the terminal. In both photographs, the arrowheads point to the location of prematurely separated centromere as detected either by rhodamine filter or apparent in these photographs.
Location of IQnetochore Protein Deposition on the Dicentrics Unlike rat B1 cells, where dicentric and multicentric chromosomes showed kinetochore proteins deposited at every centromeric site, the dicentrics in B2 as well as A8 showed kinetochore proteins at only one site per chromosome. Type I dicentrics had kinetochore antibody binding site at the position of the terminal centromere both in B2 (Fig.3) and A8 (Fig. 4). In most cases it was not possible to discern the location of the "inactive" centromere. However, because of reduced spreading of the two chroFisure 4 A type 2 dicentric (arrow marks the position of the prematurely separated centromere) in an A8 cell (a) and two type 2 dicentrics in the same cell line (b). Note that various kinetochore dots are of more or less similar size and that no coalescence or compound kinechores are seen, as has been reported for B1 cells [9].
~'L¸~¸~¸ii~
iliiii!~ ~!I¸I~
290
B.K. Vig et al. matids at the site of this centromere or its less dense ethidium bromide staining, the constriction could be recognized in 87 chromosomes in B2 and 20 chromosomes in AS. The type 2 dicentrics also showed kinetochores at only oue site. Because the difference between the lengths of the segments from the distal end to the centromere was definable, it was possible to determin~ which centromere had the kinetochore. The photographs reproduced here both from B2 (Fig. 3) and A8 (Fig. 4) leave little doubt that the active centromere is the one located closer to the terminal of the dicentric. Because this dicentric is present quite frequently in the clone A8 we could study more than 100 such chromosomes for the presence of kinetochore whereas in B2 only 15 such dicentrics could be studied. In all cases the dicentrics had only one kinetochore. It is worth noticing that every analyzable chromosome in these two types of cells had only one kinetochore and most all of them appeared similar in size. There was no instance in which a chromosome showed two adjacently placed kinetochores or had coalescing kinetochores as has been seen in earlier sample, B1 [1, 9].
Replication of Pericentric/Centric DNA After 6 hours of BrdU incorporation, virtually every cell at metaphase showed label. However, the cells of interest were the ones that showed differential incorporation of BrdU in the two centric regions. Such chromosomes were found, though the search for such cells turned out to be time consuming. Nonetheless, a total of 240 dicentric chromosomes met the criterion. Out of these 175 were of type 1, the remaining being of type 2.
Replication oJ type 1 dicentrics. In most cells in which differential replication of the two pericentric regions was clear, the accessory centromere had generally already separated prematurely. Under BP 540/20 filter differentiation between the primary constriction at the site of prematurely separated centromare and the flanking chromatid segments was not clear, However, under rhodamine filter such regions could be distinguished even though some times the constriction was barely discernible. The type I dicentrics showed completion of replication of the accessory centromere and associated heterochromatin when the functional centromere (terminally located) still showed BrdU incorporation. Rather heavy incorporation of BrdU in such dicentrics was also visible even in the euchromatic regions. It appeared that the euchromatic segments at the junction of heterochromatin replicated after the pericentric heterochromatin and the prematurely separating centromere had finished replication (Fig. 5a). In rare cells there also appeared a tricentric in which the only replicating centromere was the one located terminally. The early replication of centromeric/pericentric region of the accessory centromere was somewhat better depicted in other cells (Fig. 5b, arrow). As seen in the corresponding photograph taken by using the rhodamine filter (Fig. 5c) there is a constriction in the region of the chromosome that shows no BrdU incorporation in Figure 5b. Some chromosomes in these cells had completed the replication of their entire DNA before BrdU was added (Fib. 5b, double arrowheads). There were instances in which the main centromere showed only a trace of BrdU incorporation (Fig. 6). Here also the early replication of the accessory centromere was more clearly visible. A chror~osome in the right middle of Figure 6a, with corresponding Giemsa-stained cell in Figure 6b, is another example in which the replication at the junction of heterochromatin/euchromatin is evident. However, three type 1 dicentrics shown in BrdU-incorporated cell in Figure 7a and corresponding Giemsa-stained cell leave little doubt that the accessory centromeres replicate earlier
Centromere Separation
291
Figure 5 (a) Part of a B2 cell showing three dicentrics (single arrowheads) and an apparent tricentric chromosome (two arrowheads). The arrows point to the centric region, which is sandwiched between the fluorescing euchromatin. In all these cases the inactive centromeres have separated and appear to have completed replication before that of the active (terminal) centromeres that show incorporation of BrdU. In (b) the situation is somewhat clearer in that the inactive centromere shows only weak and much reduced incorporation of BrdU compared to the euchromatin located below it (arrow). Location of the accessory centromere can be seen in the corresponding photograph taken under rhodamine filter (c). Certain chromosomes in these cells have t:ompleted their replication and show no sign of BrdU incorporation [double arrowhead in {b}].
than do the functional ones. In these instances there is no indication of any BrdU incorporation in the region of the accessory centromeres.
Replication of type 2 dicentr/cs. The situation regarding early replication of the accessory centromere was quite clear for type 2 dicentrics. In most instances, finding the cells with differential BrdU incorporation in the two centric regions was no problem. In all such dicentrics, the centromere located closer to the terminal of the chromosome showed incorporation of BrdU whereas the other showed completion of DNA replication having had occurred. This late replicating centromere is the same one that deposits the kinetochore proteins (Fig. 4J and apparently is the functional centremere. Figure 7a is an example of such clear diffe:3ntial incorporation. As the accompanying photograph (Fig. 7b) of the same Giemsa-stained cell shows, the early replicating centromere is the one located further away from the terminal and is the one that shows premature separation. Two type 2 dicentrics shown in Figure 8a also present a similar clear picture of premature rep~ication. Figure 8b, taken from a cell
292
B.K. Vig et al.
m
l
P t
mO
iim
b Figure 6 A B2 cell showing a somewhat advanced stage in differential replication of accessory centromere compared to that of the functional one. The fluorescing cell {a) and its Giemsastained counterpart {b) exhibited two type 1 dicentrics showing differential incorporation. In the dicentric marked with an arrowhead in the left of the photograph there is little or no BrdU incorporation at the site of and beyond the arrowhead that marks the position of the accessory centromere. The dicentric to the left shows the last faint incorporation in the active centromere, whereas the inactive centromere is totally devoid of incorporation.
that was Giemsa stained after washing off the BrdU antibody, shows an example in which the least detectable incorporation of BrdU is seen right in the active centromere. The heterochromatin block being minute in these centric regions (Fig. 2), the BrdU bands appear very close to the two centromeres. In summary, the rat accessory centromeres do not appear to deposit kinetochore proteins as do the centromeres on chromosomes that do not express premature separation as seen in B1 cells. The centric and pericentric regions associated with these accessory centromeres also replicate earlier than similar regions associated with the
T'
m
Figure 7 Part of an A8 cell showing both type I and type 2 dicentrics with clearly differential incorporation in the vicinity of active versus inactive centromeres (a). The three type 1 dicentrics exhibit the second centromere in the same cell when stained with Giemsa (b). The region of the chromosome on both sides of and including the inactive centromere is clearly devoid of any BrdU incorporation except for a faint trace in the dicentric shown at the top left of the photograph. These pictures also show a type 2 dicentric that is clearly devoid of any incorporation in the vicinity of and including the inactive centromere, while the region of the active centromere and some large euchromatin blocks show heavy incorporation of the thymidine analog in (a). The corresponding Giemsa-stained dicentric in {b) shows the position of the inactive centromere that is in the process of premature separation but can still be recognized as primary constriction. Note that the distances from the centromeres to terminals in this photograph are similar to what is seen in Fig. lb.
1
t
t~
d
Figure 8 Two type 2 dicentrics showing clear differential incorporation in the two centric regions (a). The accessory centromeres have separated prematurely. These are marked by arrowheads at the point of faintly visible constrictions. Compare the pattern of replication of the two chromosomes with that seen in Fig. 7a. This provides additional support to the fact that the two marked chromosomes are type 2 dicentrics. In (b) is a cell that after BrdU antibody treatment was washed with alcohol and then Giemsa stained. BrdU bands are visible all over. Even though this particular cel~ldid not appear suitable for photography under fluorescence, the type 2 dicentric (the centromeres marked by arrowheads) shows differential BrdU incorporation in the centromeres proper. Note the presence of dark band at the exact site of the active centromere (large arrowhead) and a lack of such incorporation in the accessory centromere. The presence of heavy label as indicated by the BrdU bands all over the chromosome may suggest that the accessory centromere and its associated pericentric heterochromatin replicate much earlier than does most euchromatin.
r' n'f
Centromere Separation
295
active centromeres. Hence, in view of these data and those from mouse cell lines [6], it seems that these mechanisms may constitute universal strategy adopted by the dicentrics to create a condition mimicking monocentricity.
DISCUSSION
Cancer cells show many karyotypic abnormalities. A newly emerging study of such cells is the characterization of dicentric and multicentric chromosomes. For a long time the appearance of such chromosomes had been taken to mean that the cell will not propagate these abnormal chromosomes to the next generation because of formation of bridges at anaphase. It now appears that in spite of the formation of bridges, when broken ends of chromosomes rejoin new dicentric and multicentric chromosomes can emerge [7]. Not only can the cell divide, albeit with constant karyotypic evolution, but sometimes these chromosomes can undergo a meioticlike phenomenon in mitotic cells, as has been reported for some dicentrics in the B1 cell line of rat endothelial origin. Another strategy used by dicentric and multicentric chromosomes in cancer cells is the "silencing" of all centromeres except one. These centromeres separate prematurely during premetaphase before the centromeres that bind to spindle do so. In so doing, somehow, the silent or accessory centromeres fail to bind the kinetochore proteins as detected by antikinetochore antibody tag. A brain tumor cell line in mouse exhibits this characteristic [8], as does the L-929 cell line [9]. It was not known if the formation of kinetochores at the site of silent centromeres is a specialized property of mouse cells and a few human dicentrics, e.g., tdic(X;X) [3] or is a universal phenomenon common to all prematurely separating centromeres. Such studies need to be carried out with cell lines in which the two types of cantromeres could be studied. One type would be the ones in which the dicentrics and multicentrics do not show any inactivity of any centromere, where no centromere in the dicentric shows premature separation, and both centromeres bind to microtubules. The second type would be the one in which one centromere acts as silent or accessory, separates prematurely, and does not bind to the microtubules during cell division. These criteria can be met with the cell line under study in this report. Like the mouse dicentrics, the prematurely separating centromeres in both type of dicentrics studies show only one site of kinetochore proteins, i.e., at the site of the active centromere. In more than 100 dicentrics of both types analyzed in this study, none showed the deposition of kinetochore proteins at more than one site. The inactivity of the accessory centromere can therefore be ascribed to a lack of the kinetochore and not the absence of the centromere. One may thus argue that microtubules bind to the kinetochorema facultative proteinaceous organelle superposed onto the centromeremand not to the primary constriction called the centromere. These studies, along with those from human and mouse dicentrics, lend credence to the idea that premature centromere separation is linked to the lack of formation of the kinetochore. It is not clear why a kinetochore would not form at the site of accessory centremere in these dicentrics. However, it is becoming clear why the accessory centremeres would split prematurely. In mouse, we have shown that the replication of the pericentric region and the centromere proper of the accessory centromere occurs before the replication of the functional centromere takes place [6]. This replication of the hetarochromatin or repetitive DNA in mouse appears to move right into the centromere. Upon replication the centromere splits into two units, if not right away then after some sort of maturation process has taken place. This would explain premature centromere separation. The timing of centromere replication before the
296
B.K. Vig etal. onset of metaphase or availability of spindle appears to be the crucial factor. In rat the present study also appears to point to similar control of premature centromere separation. The evidence presented here shows premature replication of the pericentric/centric DNA as seen in mouse. Whereas this pattern of replication is similar to what is observed for mouse, it is not in conformity with the pattern of replication for the centromeric regions in dicentrics in which premature centromere separation is not seen. In the latter all centromeric regions in a given chromosome replicate almost simultaneously toward the end of S phase [1]. The BrdU incorporation in these centric regions is equivalent at all sites at a time when the entire genome, except the centric regions, has completed replication. Thus our present studies indicate that premature DNA replication is unique to the prematurely separating centromeres and may be the reason for such separation. In summary, it appears that dicentric and multicentric chromosomes in cancer cells need to be studied more carefully. One or a few karyotypes from one cell cycle may not provide information about the nature and characteristics of these multicentric chromosomes. Whether cancer cells have a special potential developed to handle multicentricity is not known. Nonetheless, it is clear that in cancer cells premature centromere separation and early DNA replication of the accessory centromeres may be routine for certain types of multicentrics. The authors thank Drs. M. Tyrkus and C. Diglio for the gift of the cell line. B. K. V. is grateful to the DKFZ for the award of fellowships.
REFERENCES
1. Broccoli D, Paweletz N, Vig BK (1989): Sequence of centromere separation: Characterization of multicentric chromosomes in a rat cell line. Chromosoma (in press). 2. Diglio CA, Wolff DE, Meyers P (1983): Transformation of rat cerebral endothelial cells by Rous sarcoma virus. J Cell Biol 97:15-21. 3. Earnshaw WC, Migeon BR (1985): Three related centromere proteins are absent from the inactive centromere of a stable isodicentric. Chromosoma 92:290-296. 4. Merry DE, Pathak S, Hsu TC, Brinkley BR (1985): Antikinetochore antibodies: Use as a probe for inactive centromeres. Am J Hum Genet 37:425-430. 5. Selgi S, Ariel M, Goitein R, Marcus M, Cedar H (1988): Regulation of mouse satellite DNA replication time. EMBO J 7:419-426. 6. Vig BK, Broccoli D (1988): Sequence of centromere separation: differential replication of pericentric heterochromatin in multicentric chromosomes. Chromosoma 96:311-317. 7. Vig BK, Paweletz N (1988): Sequence of centromere separation: generation of unstable multicentric chromosomes in a rat cell line. Chromosoma 96:275-282. 8. Vig BK, Zinkowski RP (1986): Sequence of centromere separation: A mechanism for orderly separation of dicentrics. Cancer Genet Cytogenet 23:347-359. 9. Zinkowski RP, Vig BK, Broccoli D (1986): Characterization of kinetochores in multicentric chromosomes. Chromosoma 94:243-248.