Variations in the distribution and migration of centriole duplexes in mitotic ptk2 cells studied by immunofluorescence microscopy

How the formation of the mitotic apparatus is controlled at the molecular level is an intriguing and still open question. One aspect of the formation and the later functioning of the mitotic spindle involves the role of the centrioles. Historically the centrioles have been considered as directly involved in spindle organization and function (Stubblefield & Brinkley, 1967). A variety of information suggests however that the centrioles themselves may not be necessary (Szollosi, Calcarco & Donahue, 1972; Berns & Richardson, 1977; Pickett-Heaps, 1969, 1975; Dietz, 1966), but that intact pericentriolar material may have an active role in spindle formation and function (Berns, Rattner, Brenner & Meredith, 1977; Robbins & Gonatas, 1964; Robbins & Jentzsch, 1968; Robbins, Jentzsch & Micali, 1968). Evidence from drug studies supporting centriole involvement (for discussion, see Berns et al. 1977) and the presence in vitro of microtubules arising both from centriole ends and from pericentriolar material (Gould & Borisy, 1977; Telzer & Rosenbaum, 1979) further complicates the question.

Whether or not the centrioles or the pericentriolar material or both are directly involved, knowledge of the intracellular location of the centrioles with their accompanying pericentriolar material would facilitate the understanding of spindle formation. It is especially important to document the migration of the centrioles and the microtubules associated with them, as well as the behaviour of other cytoplasmic microtubules during this period. The timing of these migrations to form spindle poles versus the timing of other mitotic events, e.g. chromosome condensation, is also of interest.

Detailed studies of centriole location and migration have been limited to light- and electron-microscopic analyses. While the former technique has the advantage of allowing one to follow centriole migration in a living cell and to relate it to other phenomena such as chromosome behaviour, the usefulness of the approach is limited by the difficulty of distinguishing the centriole duplexes from cellular vesicles and granules that appear similar to them in the light microscope. Certain staining procedures have somewhat enhanced centriolar visibility for some studies in the light microscope (Stubblefield & Brinkley, 1967; Brinkley, Stubblefield & Tsu, 1967; Wilson, 1934), however, routinely only a small number of cells has been available for analysis. Furthermore, any information on the accompanying microtubules is below the resolution of the light microscope. Use of the electron microscope, either alone or in combination with light microscopy, has allowed investigation in greater detail of centriolar ultrastructure and simultaneously the state of the nuclear membrane, chromosomes and microtubules. However, the latter technique also has been limited to relatively small numbers of cells and often it has been difficult even from serial sections to get a good detailed overview of the entire cell. Even given the limitations of the 2 techniques, several detailed studies have been attempted and considerable data accumulated (Stubblefield & Brinkley, 1967; Szollosi et al. 1972; Berns & Richardson, 1977; Pickett-Heaps, 1969, 1975; Dietz, 1966; Berns et al. 1977; Robbins & Gonatas, 1964; Robbins & Jentzsch, 1969; Robbins et al. 1968; Brinkley et al. 1967; Wilson, 1934; Roos, 1973; Rattner & Berns, 1976a, Z>; Zeligs & Wollman, 1979). Many unanswered questions as to the generality of some observations: e.g. the variability of location of the 2 centriole duplexes in early prophase (Roos, 1973; Rattner & Berns, 1976a) have not been approached. Further, several important questions have given apparently contradictory answers, possibly due to the small sample sizes under study.

Immunofluorescence microscopy has been used to study microtubular profiles in both interphase and mitotic cells (see for example Weber et al. 1975 a, b;Fuller, Brinkley & Boughter, 1975). The advantages of immunofluorescence microscopy to study centriole locations in large numbers of cells has been suggested in studies employing both tubulin antibody (Osborn & Weber, 1976) and some nonimmune rabbit sera (Connolly & Kalnins, 1979). Here we report the use of specific tubulin antibodies which we have used to quantitate for the first time the percentages of cells in large populations showing particular locations of centriole duplexes in interphase and during early mitotic events: i.e. from early prophase through metaphase. Centriole duplexes at various stages of separation were analysed. Besides the frequently described migration of duplexes around the nucleus, centriole duplexes having no apparent association with the nucleus migrated to form presumptive spindle poles with the establishment of prometaphase spindles. Cytoplasmic microtubules have been monitored at the same time to ascertain the timing of various stages of centriole movement with the appearance of astral microtubules and the disappearance of cytoplasmic microtubules; the relation of both these events to nuclear membrane breakdown and chromosome condensation has been determined. To increase further the available sample size of mitotic cells, cells were sometimes synchronized in 5-phase by double thymidine block (Rao & Johnson, 1970) and their subsequent progress through mitosis analysed. Several general conclusions are presented demonstrating and quantitating the extreme variability in locations of centrioles in prophase, their numerous paths of migrations, and their different final locations giving rise to prometaphase spindle formation.

The established epithelial cell line, PtK₂, was grown and maintained in culture as described (Aubin, Weber & Osborn, 1979; Webster, Osborn & Weber, 1978). In preparation for immunofluorescence, cells were seeded on to glass coverslips and allowed to grow for at least 24–36 h. When synchrony was used, cells were grown on glass coverslips and treated with a double thymidine block. 24–36 h after seeding at a density of 10⁸ cells per 100-mm diameter Petri dish, medium was removed, and medium prewarmed to 37 °C containing 2 mM thymidine was added. After 24 h, the medium was removed and replaced with fresh medium without thymidine. After a further 24 h, cells were exposed a second time to medium containing thymidine again for a period of 24 h. Then the medium with thymidine was replaced with fresh medium and the population monitored with time for mitotic index (for further details see Aubin, Osborn, Franke & Weber, 1980). The synchronization process itself did not affect the observations on centrioles since untreated cultures gave identical results. In some experiments, cells on coverslips were treated with Colcemid (5 μg/ml) for 4 h and then allowed to recover for 30 min in Colcemid-free medium.

The preparation of rabbit antibodies against porcine brain tubulin has been described as has the isolation of specific tubulin antibodies by passage over Sepharose 4 B coupled with covalently bound tubulin (Weber, Wehland & Herzog, 1976). Fluorescein-labelled goat-anti-rabbit IgGs (Miles-Yeda, Israel) were used as the second antibody. The low level of non-specific binding of these IgGs was eliminated by preabsorption of these antibodies on methanol-fixed PtK₂ cells (see Aubin et al. 1979). Specific tubulin antibodies were used at 0·05 mg/ml; fluorescein-labelled goat-anti-rabbit IgGs were used at approximately 0-5 mg/ml.

PtKa cells on glass coverslips were immersed in − 10 °C methanol for 5 min, followed by a brief wash in phosphate-buffered saline (PBS). Staining with antibodies and washing were performed as described (Aubin et al. 1979; Webster et al. 1978). Coverslips were mounted in Moviol 4-88.

Cells displaying centriole duplexes in various locations were counted by scanning coverslips and focussing through the cells. Centriole duplexes in some locations (e.g. above the nucleus) were difficult to visualize simultaneously with cytoplasmic microtubules. It was necessary to focus up to the duplex, then to focus down to the cytoplasmic array of microtubules to determine their orientation and appearance. Such problems with depth of field sometimes limited the photography as is evident in several of the photographs shown here. Often we have focussed on and exposed for the centriole duplexes resulting in loss of information in the photographic print since: (a) cytoplasmic microtubules may be out of focus, and (ft) cytoplasmic microtubules may be underexposed compared to the bright centriolar duplexes. At the microscope, determination of locations was normally possible and to determine the percentages of cells, 300-500 cells were counted per coverslip. To increase the sampling size of mitotic cells, we frequently used cells synchronized in S-phasc and then allowed to proceed through mitosis (Aubin et al. 1980).

The designation ‘close beside’ the nucleus refers to duplexes lying less than 0·5 mm from the nuclear membrane (on 3 5-mm film with 160 × magnification) or less than ∼ 3 μm from the membrane. Duplexes located further away could be seen to be removed from the nuclear membrane and located in the cytoplasm. Centriolar duplexes could be found in the cytoplasm at any distance from the nucleus; some were measured up to approximately 25 μm from the nuclear membrane.

Prophase was taken to be that period of the mitotic cycle from earliest detectable chromosome condensation up to breakdown of the nuclear envelope. Prometaphase was taken as beginning after loss of the nuclear envelope.

In interphase cells stained with tubulin antibody, the centriole duplexes can often not be visualized against the extensive background of microtubules. However, in a normal population of PtK2 cells we were able to distinguish in about 20–30% of the cells the centriole duplex as a bright dot of fluorescence, sometimes with a hollow core, which served as a focal point for many cytoplasmic microtubules. A corresponding phase-dense black dot could be seen only occasionally. Intracellular locations of the duplexes in several hundred cells were determined relative to the substrate (coverslips) to which the cells were attached. In normally growing cells, the centriole duplexes were found at variable locations: (a) in direct juxtaposition to the nuclear membrane, i.e. close beside the nucleus. This was the most common location (∼ 56%); (b) ‘on top of’ or below the nucleus (22%); and (c) noticeably removed from the nucleus and often far out in the cytoplasm (22%) (Fig. 1 A). In category (ft), when it was possible to distinguish the orientation, the majority of the centriole duplexes were on top of the nucleus relative to the substrate; for convenience we shall continue to refer to them here as ‘on top’. To try to determine how representative of the whole population these percentages were, a similar quantitation was attempted after treatment of the PtK₂ cells with Colcemid and recovery in Colcemid-free medium for 30 min. Under these conditions the cytoplasmic microtubules which are destroyed by the drug treatment have only just started to regrow and the centriolar position is clearly visible as a bright dot or ring of fluorescence (cf. Osborn & Weber, 1976). After Colcemid treatment fields of cells were chosen in which the positions of the centriole duplexes could be distinguished in > 95% of the cells (e.g. Fig. 1 B). The relative percentages of cells with centrioles in various locations were: (a) 60% with centriole duplexes close beside the nucleus; (6) centrioles on top (or below) the nucleus (19%); and (c) centrioles noticeably removed from the nucleus (21%).

The earliest indication of prophase detectable in immunofluorescence in cells stained with tubulin antibodies was a striking increase in the size and intensity of the bright fluorescent spot marking the centriolar region (Figs. 1, 2). At this very early time, the general display of cytoplasmic microtubules appeared unchanged, with no breakdown or decrease in abundance of the tubules yet detectable. Occasionally 2 very close black dots, presumably the 2 centriole duplexes, could be distinguished in phase contrast, although while they remained very close together, their fluorescent images appeared fused probably because of the wealth of microtubules radiating from them (e.g. Fig. 5A, B, p. 188). Chromosome condensation had always begun already by this stage. The bright fluorescent dots could be seen in about 95% of the early prophase cells identified by observing chromosome condensation either in phase-contrast microscopy or using Hoechst stain (Hilwig & Gropp, 1972). In the remaining prophase cells, chromosome condensation was seen to have begun without coincidence of the increase in fluorescence staining of the centriolar regions (see below).

Although it has been suggested or implied that the centriole duplexes migrate to the nuclear envelope prior to their separation in prophase (see e.g. Rattner & Berns, 1976 a, b), we have noted that in PtKz cells this does not always occur. When prophase cells with centriole duplexes at various locations were counted, we found: 72% with centriole duplexes closely juxtaposed to the nuclear membrane, 22% with centriole duplexes on top of the nucleus, and 6% with centriole duplexes still far removed from the nucleus in the cytoplasm (e.g. Figs. 1, 2). Comparing these numbers with those from interphase cells would suggest that some centriole duplexes migrated closer to the nucleus prior to prophase, but that prophase development of centrioles could occur in the cytoplasm as judged by the high percentage of cells that displayed prophase centrioles located there.

The separation of the 2 duplexes to form spindle poles proceeded by a number of routes as would have been expected from their variable cytoplasmic locations observed in early prophase (see above). It was evident that even amongst those duplexes found close to the nuclear envelope, there was variation in their paths of migration. For example, centrioles could move progressively further apart while remaining in close proximity to the nuclear envelope, until finally reaching opposite sides of the nucleus (Fig. 3). The microtubules associated with this class of duplexes seemed to radiate from both duplexes and often curved around the nucleus. Alternatively, the duplexes could be removed a short distance from the nucleus (arbitrarily defined here as ⩽ 3 μm), and could apparently move apart on a straight line path at an oblique angle to the nuclear envelope. Often one duplex was found close to the nucleus while the other was found in the cytoplasm at various angles to the nucleus. It seemed that probably nearly all angles of the centrioles relative to the nucleus were possible. The second type of movement involved those duplexes found, relative to the substrate, above the nucleus. Duplexes on top of the nucleus appeared to migrate across the top to final positions at opposite sides of the nuclear membrane (Fig. 4A-C). Occasionally, one duplex was found close beside the nuclear envelope and the other apparently migrated across the top (Fig. 4D-G). The third, and perhaps most intriguing path, involved those centriole duplexes located far away (> 3 μm) from the nucleus in the cytoplasm. Numerous examples of such duplexes were found in which the duplexes moved apart to apparently normal distances, but without any detectable association with the nucleus (e.g. Fig. 5).

In all these cases, astral microtubules could be seen radiating from the migrating centriolar duplexes, with microtubules from each overlapping frequently. Often the centrioles and the astral microtubules were at angles to each other, i.e. there was not a straight line between the centrioles (e.g. Fig. 3 E). Although occasional microtubules could be found there, normally microtubules could not be detected which ran directly between the separating duplexes, that is beginning on one and ending on the other. Astral microtubules grew noticeably in length as the migration process progressed. A prometaphase spindle began to form when centriole duplexes reached some distance apart, whether they were at opposite sides of the nucleus (Fig. 3), with one beside the nucleus and one in the cytoplasm (Fig. 4), or with both in the cytoplasm (Fig. 5). It has not been possible to determine this distance exactly, since varying degrees of flatness exist in the mitotic PtK₂ cells used and the precise orientation of the substrate to the axis joining the centriole duplexes cannot be determined in the fluorescence microscope. Fluorescent dots such as are seen in Figs. 4B and c are occasionally observed; their significance is not known, although they have been observed in a second cell line (Marchisio, Osborn & Weber, 1978; Fig. 11).

The onset of breakdown of cytoplasmic microtubules occurred sometime after chromosome condensation began. It could be detected always in cells where centriole duplex migration had begun. Thus, in cells where 2 spots of fluorescence could be distinguished indicating that the centrioles had begun to separate, some disappearance of microtubules could be seen. Breakdown of microtubules began apparently from the periphery of the cell in one particular area and then moved in the direction of the nucleus. Although microtubular breakdown could be followed in many cells, no preferential location of this early breakdown could be determined. Although often the breakdown appeared to progress more rapidly on the side of the nucleus opposite from where the centrioles were migrating, this was not always so. Occasionally, growth of astral microtubules somewhat obscured the appearance of cytoplasmic microtubules and the determination of the breakdown process. Focussing up and down through the cell normally helped to determine what kinds of microtubules were present, and in particular, to identify those originating from the migrating centrioles.

The transition from prophase to prometaphase is generally regarded as the time when the nuclear membrane disintegrates (Roos, 1973). In those cases where centrioles moved around the nuclear membrane, they appeared to reach opposite sides before nuclear membrane breakdown (Fig. 3F). Similarly, when centriole duplexes migrated through the cytoplasm, they appeared to reach distances of separation equivalent to those reached by centrioles proximal to the nucleus before nuclear membrane disintegration occurred (Fig. 5). It is possible that some localized areas of membrane breakdown might have been present and yet been undetected in the phase microscope, but in almost all cases looked at in the phase microscope, a phase-dense line could be observed at these times. A nucleolus and recognizable state of chromosome condensation and packing were also evident. Immediately after nuclear membrane breakdown, when the nucleolus was also gone and chromosomes were further condensed, pole-to-pole microtubules were clearly visible for the first time in the corresponding fluorescent images (Fig. 3 G). At this point, a few cytoplasmic microtubules still persisted. The last cytoplasmic microtubules disappeared shortly after this as the prometaphase spindle developed. Well before metaphase, no microtubules outside the arrays of astral, pole-to-pole, and chromosome-to-pole microtubules were visible (Fig. 3 H).

Apparently the centriole duplexes which migrated through the cytoplasm away from the nucleus could form spindle poles. In Fig. 5, a prometaphase spindle far to the side of the chromosome mass is shown. Such a configuration of prometaphase occurred with approximately the same frequency (8%) as the early prophase centriole duplexes in the cytoplasm removed from the nucleus (6%).

A further interesting prometaphase type was observed. Fig. 6 summarizes this behaviour. In a small percentage of cells in prometaphase (3%), the chromosomes were seen after nuclear membrane breakdown and at a late stage of chromosome condensation to be grouped around a single fluorescent spot. Cytoplasmic microtubules were absent as expected for this stage of nuclear events. The observation of a single spot did not appear to be due merely to viewing the spindle end on, looking down along the pole axis, since focussing through the cell did not reveal the other pole. Rather it seemed that in those few cells, the centriole duplexes had not yet migrated apart. That they were however capable of doing so and forming spindle poles a short time later is shown by Fig. 6. In these micrographs, it is evident that centriole duplexes could be found among the condensed chromosomes at various stages of separation.

The migration of centriole duplexes and its temporal relationship to other micro-tubular events has been investigated by immunofluorescence microscopy in interphase and mitotic PtK.2 cells with antibodies to tubulin. By allowing visualization of a large number of cells both in normal cultures and in synchronized populations, an unambiguous determination of the location of centrioles and their spatial orientation to the nucleus has been obtained. Thus the technique has been useful in defining and in quantitating a number of possible paths of centriolar migration and early spindle development. It has been possible also to determine the overall state of the cytoplasmic microtubules relative to centriole movement and the progression of prophase. This study therefore extends earlier studies of mitotic cells by immunofluorescence using tubulin antibodies in which only isolated photographs of cells in different mitotic stages have been presented (Fuller et al. 1975; Weber et al. 1975a-, Brinkley, Fuller & Highfield, 1976).

Centrioles could be seen in about 25% of interphase cells without any drug treatment as dots of fluorescence with a sometimes extensive array of microtubules radiating from the dots. In essentially all prophase cells (95%), the centrioles were clearly visible as very bright fluorescent spots, such that prophase cells could easily be distinguished in fluorescence. The staining intensity of the centriolar duplexes in prophase cells was much increased over that observed in interphase cells and in the prophase cells the centriolar regions were obvious foci for increased numbers of microtubules. These observations support previous conclusions obtained from electron-microscopic observations which suggest an increase during mitosis in the abundance of pericentriolar material and microtubules radiating from it (Robbins & Gonatas, 1964; Robbins & Jentzsch, 1969; Robbins et al. 1968;Roos, 1973;Telzer&Rosenbaum, 1979; Snyder & McIntosh, 1975; Rattner & Phillips, 1973).

The majority of previous data has suggested that each duplex migrates as a single unit through interphase cells and ends as a duplicated unit close to the nucleus prior to separation to form the spindle poles (e.g. Roos, 1973; Rattner & Berns, 1976a). We have verified the observation that the centriolar duplex may be located in a wide variety of places in the interphase cell and quantitated these locations. That at least some duplexes must migrate from their interphase positions closer to the nucleus is indicated by the higher percentage of cells with centrioles removed from the nucleus in interphase versus prophase cells. However, a considerable fraction (6%) of the PtK₂ cells observed in this study revealed centriole duplexes which remained in the cytoplasm far removed from the nucleus and which separated normally as prophase proceeded. Thus, it seems that a fixed spatial orientation of centriole duplexes to the nucleus is not required for duplex migration and ultimately for spindle formation. Further, amongst those centriole duplexes located close to the nucleus, either beside or on top, much variability was observed in their paths of migration. The straight line joining pole-to-pole was observed at all possible angles to the substrate and nucleus. Variation in possible paths of migration has been seen in light-microscope examination of prophase cells (Rattner & Berns, 1976a). Interpolating from the variety of positions noted here in hundreds of cells, we have also found that centrioles migrate around or over the top (relative to the substrate) of the nucleus. One centriole duplex was often found in the cytoplasm, whereas the other was proximal to the nucleus. In all of these cases, however, the centrioles seemed to move apart equally well reaching opposite sides of the nucleus or some distance apart in the cytoplasm before nuclear envelope breakdown. Frequently astral microtubules from the migrating duplexes appeared to be at angles to each other rather than along the straight line joining the poles. This could be seen in cells even after the breakdown of the nuclear membrane. This has also been observed in electron-microscopic studies (Roos, 1973; Rattner & Berns, 1976 ft) and may indicate that some final realignment of centriole duplexes can occur before formation of the actual spindle axis.

The timing of centriole migration relative to other events, for example the degree of chromosome condensation, was variable. The majority of cells (∼ 95%), showing early prophase chromosome condensation, also displayed the increase in intensity of centriolar duplexes stained with tubulin antibody. A few prophase cells (∼ 5%) in which chromosome condensation had already begun, did not display this increase in staining in the duplex region. A comparable number of cells displayed duplexes separated by only short distances when chromosome condensation was well progressed. The most extreme case occurred in a small percentage of cells in which the duplexes appeared not to migrate until after nuclear membrane breakdown. For this class of cells, duplexes separated normally through the region of the prometaphase chromosomes. This occurred in the absence of cytoplasmic microtubules which seemed to have disappeared with the usual timing and were completely gone after nuclear membrane breakdown (see below).

Only after nuclear membrane breakdown did clearly identifiable pole-to-pole microtubules appear. The loss of the nuclear membrane with accompanying loss of the nucleolus and further chromosome condensation coincided with the completion of depolymerization of cytoplasmic microtubules. Duplex separation is thus not always coupled to the trigger for a cell’s entry into mitosis as evidenced by chromosome condensation and ultimately in loss of the nuclear envelope. Breakdown of cytoplasmic microtubules may be tied to some extent to these events. However, maturation of pericentriolar material as evidenced by the increase in staining intensity, and centriole separation are more variably timed events. Thus, what the trigger is for prophase as defined by the onset of chromosome condensation and the trigger for and mechanism of duplex separation remain intriguing questions.

This report verifies a number of light- and electron-microscopic observations. More importantly, the study has extended the understanding of the variability in both the locations of centriole duplexes in prophase and their paths of migration. A preliminary study of centriole movement in mouse 3T3 cells has shown that the percentages of cells displaying certain types of centriole behaviour is cell-line dependent. For example, a higher percentage of 3T3 cells displayed centrioles separating after nuclear membrane breakdown than in PtK₂ cells; no 3T3 cells were observed with duplexes separating in the cytoplasm. Such a rapid technique for visualization both of the centrioles and of their milieu against the depolymerizing cytoplasmic microtubules is excellent for quantitation and for the general definition of centriole migration. If these processes can be studied at the light- and electron-microscopic levels using the same mitotic cells, understanding of the molecular events which accompany mitosis, as well as their sequence, should be advanced.

J.E.A. was a Postdoctoral Fellow of the Medical Research Council of Canada.