Functional monopolar spindles caused by mutation in MGR, a cell division gene of Drosophila melanogaster

The poleward movement of chromosomes during ana-phase has attracted the interest of cell biologists since it was beautifully described over one hundred years ago (Flemming, 1878). Though a great amount of work has been devoted to cytological and biochemical aspects of this process, our knowledge of it is still rather unsatis-factory. Genetics provides a powerful tool for identifying the structural or regulatory components involved in chromosome movement and their function during cell division. Genetic analysis has been successful in dis-secting the cell cycle in yeast and other fungi (Pringle & Hartwell, 1981; Nurse, 1985), and it is yielding interesting results in mammalian cell cultures (Simchen, 1978; Marcus et al. 1985) and mouse (Magnuson & Epstein, 1984). Recently, several groups have concentrated on the genetic dissection of cell division in Drosophila melanogaster, where a fairly large collection of mutants is already available (Gatti et al. 1983; Ripoll et al. 1987).

The spindle is the structure responsible for the accurate segregation of chromatids (or chromosomes) to daughter cells during mitosis and meiosis, as well as for the equitable partition of other subcellular organ-elles. Morphologically, the spindle is formed by the centrosomes, which are composed of centrioles and pericentriolar material, and the fibres joining the centrosomes to each other or to the kinetochores. Compared with what we have learned about spindle fibres (Dustin, 1984), centrosomes have managed to escape biochemical and genetical approaches (Fulton, 1971; Mazia, 1984). In this report we present and discuss the phenotype caused in Drosophila melanogaster by mutation in the gene merry-go-round (mgr), which we interpret as resulting in abnormal centrosome behaviour.

Merry-go-round was recovered among a collection of 30 late (1arval and pupal) lethals induced with X-rays in the third chromosome of an isogenic red strain (for description of mutations see Lindsley & Grell, 1968). Zygotic lethality, mitotic phenotypes, and testicular phenotypes, co-mapped at 51·3cM (centiMorgans), based on 28 recombinant chromo-somes between scarlet (44·0cM) and red (53·6cM). The gene was cytologically localized in region 86E3-6; 86E12-20, based on its inclusion in Df(3R)TE61 (86E3-6;87A6-10) but not in Df(3R)TE41 (86El2-20;87Cl-2) (we have redefined the breakpoints of these deficiencies and they do not agree with previous cytological description given by Lindsley & Zimm, 1987). All mutant traits are fully recessive and no alteration has been found in salivary gland chromosomes of either homozygous or heterozygous larvae. Mutant strains were balanced over TM6 B,Hu e Tb ca, and all cultures reared on standard medium at 29, 25 or 17°C.

Homozygous mutant larvae or pharate adults were recognized by the red coloration of their Malpighian tubules, that are yellowish in heterozygous individuals, and the absence of the shortened body shape produced by the dominant mutation Tubby (Tb). Larval brains were stained with orcein following Gatti et al. (1974), and testes following Lifshytz & Hareven (1977). Live cells during spermatogenesis were observed with phase-contrast optics after dissection of adult testes in saline (Hardy et al. 1981).

To quantify mitotic phenotypes we have referred each type of mitotic figure to a ‘field’, which is defined as the area seen under the microscope with the following constants: Zeiss Universal Microscope, plan-neofluar 63×/l·25 oil objective, kpl-W 10×/18 ocular, Optovar in position 1·25X. Under these conditions an average wild-type brain yields some 200 fields. To study a larger population and thus minimize the effect of quantitative variations among individuals, only 15 fields per brain were scored. Determination of the nuclear size of onion stage-spermatids was performed by measuring the diameter of nuclei in micrographs taken under standard conditions with a 6CT2 Nikon Profile Projector.

Homozygosity for mgr results in lethality late in development, a majority of the mutant animals dying just before emergence from the pupal case. Many of these pharate adults have slight cuticular abnormalities typical of viable mitotic mutants, such as rough eyes. Since lethality is meiotically inseparable from the mitotic abnormalities seen in larval brains, most probably the failure to emerge from the pupal case is due to general muscular or neurological defects. Larval brains homozygous for mgr show a variety of alterations both in frequency and phenotype of mitotic figures (Figs 1, 2). As shown in Fig. 2 all these mutant traits are stronger when larvae are grown at 17°C than at 29°C; culture at 25 °C results in intermediate values (data not presented). The most conspicuous mitotic abnormalities are: (1) an increase in the mitotic index (Fig. 2A), accompanied by a reduction of the relative number of anaphases (Fig. 2B). (2) 50–60% of the cells have X-shaped over-condensed chromosomes (Fig. 1C). This phenotype is indicative of arrest during prometaphase or metaphase; for simplicity, we will always refer to these figures as metaphases. (3) Around 15% of the mitotic figures seen at 17 °C (3% at 29°C) have numbers of chromosomes exceeding the normal diploid complement of eight (Fig. 2B). Roughly a quarter of these cells seem to be true polyploids, the number of chromosomes being an exact multiple of the haploid complement; the remaining cells are aneuploid, with intermediate numbers of chromosomes (Fig. 1B). Unless specifically stated, we will call both true poly-ploid and aneuploid cells ‘polyploid’. (4) The most conspicuous cellular phenotype is the appearance of circular mitotic figures (CMFs; Fig. 1E); in these figures the major autosomes and sex chromosomes are arranged in a circle with the centromeres pointing towards the centre and the chromatids pointing towards the periphery. The small fourth chromosomes are always located in the centre. This regular configuration is somewhat disturbed when the circular figures are polyploid (Fig. 1G). CMFs represent 25–35% of the mitotic figures in mutant larvae while, even with very permissive criteria, they never exceed 1% in wild type (Fig. 2D). (5) Occasionally, abnormal anaphases are found in which at least one of the poles appears as a circle of chromatids similar to that described for CMFs; the majority of these anaphases are asymmetric, with a circular figure in one pole and a normal anaphasic plate in the other. Fig. 1F shows one of the rare symmetric circular anaphases found in mutant brains. Larval brains hemizygous for mgr (mgr/Df(3R)TE61) show a more dramatic phenotype, indicative of the hypomorphic nature of the mutant allele and of the locus-specific nature of the mutant phenotype. The fraction of normal mitotic figures is decreased 10-fold relative to homozygous mutant brains, the relative numbers of CMFs and polyploid cells increase twice, normal anaphases practically disappear, and CMFs are more frequently polyploid.

Whereas many of the mitotic phenotypes described above are common to other mitotic mutants (Gatti et al. 1983; Ripoil et al. 1985, 1987), CMFs and asymmetric anaphases are typical of mgr. Figures somewhat similar to CMFs have recently been found in individuals mutant for polo (Sunkel & Glover, 1988). When homo-or hemizygous mgr larval brains are treated with colchicine (a drug resulting in depolymerization of microtubules), CMFs as well as asymmetric anaphases are no longer found. Incubating mutant brains in taxol (a microtubule-stabilizing drug) does not have any effect on the frequency of CMFs. Nevertheless, this treatment leads to a change in the typical phenotype of CMFs: although the centromeres are still arranged in a circle with the small fourth chromosomes in its centre, the chromatids are no longer seen pointing towards the periphery. The effects observed after treatment with both antimitotic drugs indicate that the typical CMF structure requires microtubules for its maintenance. Since microtubules are the major component of the mitotic spindle, these results suggest that a functional spindle is needed for CMFs to appear. There is no spindle-specific drug to test this hypothesis, but we can nevertheless substitute drug treatment for the genetic interaction between mgr and mutations known specifically to alter spindles. We have constructed individuals simultaneously mutant for mgr and abnormal spindle (asp, 3-85·2), a mutation thought to alter the spindle but not other microtubule-dependent structures (Ripoll et al. 1985; Wandosell et al., unpublished data). Brains homozygous only for asp show a variety of mitotic phenotypes such as high mitotic index, polyploid cells arrested at metaphase, and absence of anaphases. Since the original mutant allele is cold-sensitive, for our present analysis we have used asp^E3, a strong mutant allele with smaller temperature-dependent phenotypic variations (Fig. 2). While individuals homozygous for asp^E3 die as mature pupae or soon after emergence, and homozygous mgr as pharate adults, double mutant individuals die as third-instar larvae. These larvae lack imaginai discs and their brains are reduced in size (60% of the size of single mutant brains, measured as amount of total protein). Since imaginai discs and larval brains are among the few tissues mitotically active during larval development, this decrease in size suggests a high cell mortality associated with mitotic defects.

Quantification of the mitotic phenotypes shown by larval brains doubly homozygous for mgr and asp is presented in Fig. 2. As happens with whole individuals, most cellular phenotypes suggest synergistic interactions, between both mutations. The reduction in the mitotic index, which is the result of an increased cell mortality, indicates that both mutations lead to cell lethality through independent mechanisms. A similar interpretation is derived from the increase in the relative number of polyploid cells, so that the cause of the failure in chromatid segregation is different in each mutant strain. Only the frequency of anaphases is indicative of an epistatic interaction between both mutations; anaphases are practically absent in the double mutant, as is the case in asp individuals. The same happens with CMFs, even using very loose criteria to define a figure as a circle in doubly mutant brains. Since the effect of asp is thought to be spindle-specific, the absence of CMFs in mgr asp brains shows that a functional spindle is needed for these figures to be formed and maintained.

Mitosis and meiosis are different types of cell division sharing many steps and differing in others. The common steps are likely to be under the control of the same set of genes, and mutations altering both processes have been described (e.g. see Baker et al. 1978; Ripoll et al. 1985). Mutant testes were examined to ascertain if mgr was needed during spermatogenesis.

Observation of homo-and hemizygous testes under phase-contrast optics reveals a series of defects. (1) Absence of recognizable meiotic spindles. In a sample of 30 testes no structure resembling a spindle could be found, while between one and two cysts with clear spindles per testis were observed in wild type. There-fore, meiotic spindles, if they exist, must be highly abnormal. (2) The size of post-meiotic nuclei in early spermatids (onion stage) is noticeably larger than in wild type (Fig. 3A,B). At this stage the large majority of cysts in which all nuclei could be counted had 16 nuclei instead of the normal 64. (3) The mitochondrial derivatives (Nebenkern) are very abnormal: they lack both their typical circular shape and uniform appearance, they are disaggregated, and they are hardly ever found associated with the nuclei (Fig. 3B). Mitochon-drial disorganization is more obvious in testes obtained from pharate adults than from mature larvae. The reason for this age-dependent phenotypic variation is unclear, although it could be related to a decrease in the ability of persisting maternal products to perform the wild-type function partially. (4) Spermatids degenerate during early stages of elongation. During this stage, flagellar basal bodies, which are easily identifiable in wild type, are either absent or unrecognizable in mutant individuals. This phenotypic heterogeneity does not necessarily mean that mgr is needed during different stages of spermiogenesis, since defects in an early stage can result in a cascade of effects producing abnormalities later on (Hardy et al. 1981). All the defects observed in mgr testes can be traced back to abnormal behaviour of components of the meiotic spindle. Failures in spermatid elongation could be due to non-functionality of the mitochondrial aggregates; the abnormalities observed in these aggregates could result either from the absence or functional failure of the spindle (the equitable partition of mitochondria during both meiotic divisions is achieved through their association with this structure) or from defective elongation of the axoneme, a derivative of the centriole (Tokuyasu, 1974).

Onion-stage spermatid cysts are formed by 16 large nuclei, the same number with which meiosis is initiated in wild type. Variations in the nuclear size of early spermatids have been taken to indicate variations in their chromosome content due to abnormal chromo-some partition during meiosis (Lifschytz & Hareven, 1977; Hardy et al. 1981; Kemphues et al. 1982; Ripoll et al. 1985; Fuller et al. 1987). We have found (unpublished data) that during the onion stage there is a geometric relationship between the amount of DNA contained in a nucleus and its diameter, so that nuclear volume is roughly proportional to chromosome content. A quantification of the amount of DNA contained in mgr postmeiotic nuclei is shown in Fig. 3C. A total of 96% of the nuclei have a chromosome content around four times that of wild-type onion-stage nuclei, while the remaining 4% are similar to wild type.

The number and size of onion-stage nuclei in mutant testes can be explained if spermatocytes can differentiate as spermatids without going through meiosis or, alternatively, if both meiotic divisions take place in the absence of karyo-and cytokinesis. Cytological observation of orcein-stained mutant testes reveals that both meiotic divisions actually take place (Fig. 4), ruling out the first possibility. The first meiotic division starts normally in cysts of 16 primary spermatocytes, with a euploid number of perfectly paired bivalents that end up arranged in nearly normal plates during the first metaphase (Fig. 4B, C). Disarranged metaphase II figures are also found, with their chromosomes showing the typical configuration of this meiotic stage (Fig. 4H); these metaphases differ from wild type (Fig. 4G) in that they remain diploid, indicating that the reductional segregation failed during anaphase I. We have never observed normal anaphases in mutant testes. Instead, cysts where all 16 cells show the phenotype presented in Fig. 4F are found: there is a ring of heavily stained chromatin that, most probably, is formed by all bivalents arranged in a circle. We interpret these circular figures as mutant first meiotic anaphases. Failure of the whole chromosome complement to segregate during the first meiotic division explains the diploid configuration of second division metaphases (Fig. 4H). If a similar failure in segregation also occurs during anaphase II, both the number and the tetrapioid size of postmeiotic nuclei are readily explained. The same result could be obtained if anaphase II is simply absent instead of abnormal. Since we have not found any mutant cyst showing what could be considered as a second division anaphase, whether anaphase II is abnormal or absent remains an open question. In any case, the tetrapioid size of the spermatid nuclei correlates with the cytological observations and indicates that both meiotic divisions have proceeded in the absence of karyo-and cytokinesis. A similar phenotype is observed in mutants lacking β2-tubulin, the testis-specific tubulin isotype, where spermatocytes progress through meiosis in the absence of spindles (Kemphues et al. 1982).

All the mutant traits observed during mitosis in mgr cells can be traced back to structural or functional defects in the mitotic apparatus: aneuploid cells can result from abnormal chromatid segregation, true poly-ploid cells from total failure of chromatid segregation followed by DNA replication, and arrested cells from absence or lack of function of the spindle. Although less obvious, the presence of CMFs could also be attributable to defects in spindle components. The study of mutant brains incubated with anti-mitotic drugs, as well as the analysis of brains from mgr asp doubly mutant larvae, has shown that CMFs need a spindle to be formed and maintained. CMFs could therefore be either metaphases or anaphases, the stages of mitosis or meiosis that require a spindle for their maintenance. While metaphase is a stage of dynamic equilibrium dependent on forces exerted from opposite poles, anaphase is characterized by the synchronous movement of sister chromatids to the poles. Anaphase can therefore be defined either by the separation of sister chromatids or by the initiation of poleward migration, two inseparable phenomena in Drosophila wild-type divisions. CMFs are formed by chromo-somes, not chromatids as in normal anaphases. Similar chromosome behaviour is observed whenever functional monopolar spindles are formed, either spontaneously (Bajer, 1982), due to mutation (Wang et al. 1983), or after experimental manipulation (Mazia, 1960). If the separation of sister chromatids is taken as the peculiarity defining a stage as an anaphase, then the CMFs should be regarded as metaphases. However, if synchronous migration of kinetochores is taken as the characteristic that defines a stage as an anaphase, several observations lead us to consider CMFs not as aborted metaphases but as a peculiar type of anaphase configuration. (1) With the techniques we use, we have never found a figure identical to a typical CMF in wild type, where, even with very permissive criteria, figures resembling a circle are rarely seen (Fig. 2D). (2) Mitotic figures identical to CMFs, formed by chromatids instead of chromosomes, are seen in asymmetric anaphases, where the stage of the cycle is unmistakable. (3) There is an evident correlation between the decrease in the number of normal ana-phases and the increase in the number of CMFs in mgr brains as the culture temperature is decreased (Fig. 2B,D). (4) The orientation of the chromosome arms in these circular figures (Fig. IE) suggests that the chromosomes are being moved towards the centre. In fact, when taxol is added to the culture medium this ‘dynamic’ aspect is lost. (5) The central position attained by the small fourth chromosomes correlates with their position closer to the poles that is usually found in normal anaphases (Kaufmann, 1934).

The configuration shown by CMFs can be explained if mitosis is started with a single pole. The forces moving the chromosomes towards this single pole could end up placing the centrosome and the centro-meres in the same plane, resulting in a circle with the chromatids pointing outwards. In all respects these figures are identical to the star configuration spontaneously occurring in functional monocentric ana-phases in primary cultures of lung epithelium of the newt Taricha granulosa (Bajer, 1982). Functional monopolar spindles are also found in ts-745, a mutant line of Syrian hamster ovary cells (Wang et al. 1983), where the chromosomes are arranged around the single pole in a spherical configuration like the one found in some round cells in newts (Bajer, 1982). Treatment-induced monopolar spindles have been elegantly shown to be functional by in toto observation after inactivation of centrosome replication with β-mercaptoethanol (Mazia, 1960), and after microtubule breakdown (see Mazia (1961) for a review). After these treatments disorganized chromosomes and chromatids reorganize again following partial reconstruction of the mitotic apparatus. The end result is that chromosomes or chromatids arrange in circles around each pole, a figure that has been called a ‘quasirosette’. These poles are functional since chromosomes or chromatids move towards them. These quasirosettes are similar to the CMFs and asymmetric anaphases found in mgr cells. We have obtained equivalent results after submitting wild-type Drosophila larval brains to low-temperature pulses, and after addition and removal from the culture medium of MTC, a reversible analogue of colchicine (Fitzgerald, 1976). The figures obtained after these treatments are identical to the asymmetric or symmetric circular anaphases found in mgr brains. The high frequency with which CMFs are found in mutant brains contributes to a great extent to the elevated mitotic index they present. This abundance indicates either that many cells arrest their cycle at the stage characteristic of CMFs, or that the duration of this monopolar stage is longer than that of normal ana-phases in wild-type cells. We believe the second interpretation to be correct since the presence of poly-ploid cells, the orientation of chromosome arms pointing outwards, and the absence of chromatin over-condensation, all suggest that the spindle in CMFs is functional. A considerable increase in the time spent during the cell cycle seems to be a characteristic common to all cells with monopolar spindles (Bajer, 1982; Wang et al. 1983).

Monopolar anaphases can also explain the rings of chromatin observed during mutant meiosis (Fig. 3F). These rings would be formed by all bivalents migrating to a single pole during the first meiotic anaphase; as a consequence, the following metaphase would have a diploid chromosome complement (Fig. 3FI). If second-division anaphases during mgr meiosis are also monopolar (or if they never take place) the number of postmeiotic nuclei per mutant cyst corresponds with the one expected, since the establishment of bipolarity is a prerequisite for cytokinesis (discussed by Mazia, 1961).

The existence of monopolar spindles points to defects in the centrosome cycle as the primary cause of the appearance of CMFs. These defects could consist of the absence of centrosome replication (as happens after treatment with β-mercaptoethanol) or of failure in the segregation of centrosomes (as is the case of the mutation in the SHO cell line or in the spontaneous production of monocentric anaphases in the newt). In theory, the complete lack of function of a gene essential for either function (replication or segregation) should result in the formation of a single giant cell with as many chromosomes as the continuous cycle ‘chromo-some duplication-monopolar mitosis’ could permit. The difference between.this theoretical phenotype and what is found in mutant individuals could be due, at least in part, to the action of wild-type products left by the maternal heterozygous genome in the oocyte: mutant cells would divide normally until these maternal products are diluted out and/or degraded. If the mutant mgr phenotype is only due to persisting wild-type products, all the cells found in mitosis should show similar phenotypes, i.e. true polyploid meta-phases and anaphases (CMFs). This is what happens during meiosis in testes homo-or hemizygous for mgr, where practically all the nuclei show the expected phenotype. However, this is not the case in mutant brains, where a variety of mitotic figures are found.

Although the phenotypes shown by larval brains homo-and hemizygous for mgr deviate from the phenotype expected for the complete lack of function of the gene, the latter is closer to the amorphic phenotype. The same happens with the homozygous brains grown at low temperature. These observations clearly show that the mutation is leaky (hypomorph), so that the mutant allele is still capable of performing in part its role during cell division. As a consequence, cells that in a given cycle are unable to divide normally can behave as either mutant or normal in subsequent cycles, and vice versa. Owing to the combined action of persistence of maternal products and hypomorphism, what we see as the final mgr phenotype is only the addition of defects progressively accumulating during brain development.

As discussed above, several phenotypes can be explained by the complete absence of either centrosome replication or separation: true polyploid cells, CMFs, and number, size and chromosome complement of spermatid nuclei. The abnormal shape of the mito-chondrial derivatives in early spermatids (Fig. 3B) as well as their lack of association with the adjacent nuclei could be explained either as secondary effects of the abnormal segregation of mitochondria during ana-phase, or as due to failure in the development of the axoneme, a derivative of the centriole. Other mutant phenotypes can be explained as due to a delay in the formation of a functional bipolar spindle, a phenomenon spontaneously occurring with relatively high frequency in primary cultures of newt cells (Bajer, 1982). This late establishment of bipolarity could be due to delayed replication, segregation or acquisition of functionality of centrosomes, and it can explain the existence of both asymmetric anaphases and aneuploid cells. Starting from a typical CMF, the subsequent appearance of bipolarity would result in the attachment of a set of chromatids to the new functional pole. The original pole would frequently keep its circular configuration, while the new pole would in most instances form a normal anaphasic plate. Failure of some chro-matids to attach to the second pole would explain the presence of aneuploid cells. Finally, cells with over-condensed chromosomes, typical of arrest at metaphase, could result from absence of functional centrosomes. A possible general explanation for this abnormal centrosome behaviour could consist of a diminution of its microtubule-nucleating ability, which would affect both its segregation and the structure of the resulting spindle (Brinkley, 1985), as well as the elongation of the axoneme (Tokuyasu, 1974).

Centrosomes have until now been resistant to biochemical and genetic analyses, mgr provides a promising starting point for the genetic analysis of centrosomes in an organism where the power of genetics is known to everybody: the mutational analysis of the locus will yield both new mutant alleles and, it may be hoped, mutations in other loci related to the centrosome that will interact in trans with mgr, similar to the second site non-complementing mutations found for B2t (Fuller, 1986).

We thank T. Fitzgerald and M. Suffness for their generous gifts of MTC and taxol, respectively, and D. Mathog for comments. We also thank C. Sunkel and D. Glover for their interest and for sharing their unpublished results. Many of the mutant strains used for this work were kindly provided by the Drosophila Stock Centers in Oak Ridge and Bowling Green. The work was supported by grants from Comisión Asesora para la Investigación Cientifica y Técnica, and an institutional grant from Fondo de Investigaciones Sanitarias. J.C. was supported by a fellowship from Plan de Formacion del Personal Investigador.