Heterochromatin protein 1 is required for correct chromosome ...

6 downloads 0 Views 1MB Size Report
in cycle 9, increasing to a 4 minute lag in cycle 12) and is probably responsible for the asynchrony in the nuclear divisions observed in these embryos. In some ...
1419

Journal of Cell Science 108, 1419-1431 (1995) Printed in Great Britain © The Company of Biologists Limited 1995

Heterochromatin protein 1 is required for correct chromosome segregation in

Drosophila embryos Rebecca Kellum* and Bruce M. Alberts Department of Biochemistry and Biophysics, University of California at San Francisco, San Francisco, CA 94143-0448, USA *Author for correspondence

SUMMARY Heterochromatin protein 1 is associated with centromeric heterochromatin in Drosophila, mice, and humans. Loss of function mutations in the gene encoding heterochromatin protein 1 in Drosophila, Suppressor of variegation2-5, decrease the mosaic repression observed for euchromatic genes that have been juxtaposed to centromeric heterochromatin. These heterochromatin protein 1 mutations not only suppress this position-effect variegation, but also cause recessive embryonic lethality. In this study, we analyze the latter phenotype in the hope of gaining insight into heterochromatin function. In our analyses of four alleles of Suppressor of variegation2-5, the lethality was found to be associated with defects in chromosome morphology and segregation. While some of these defects are seen throughout embryonic development, both the frequency and severity of the defects are greatest between cycles 10 and 14 when zygotic transcription of the Suppressor of variegation2-5 gene apparently begins. By this time in development, heterochromatin

protein 1 levels are diminished by four-fold in a quarter of the embryos produced by parents that are both heterozygous for a null allele (Suppressor of variegation2-505). In a live analysis of the phenotype, we find prophase to be lengthened by more than two-fold in Suppressor of variegation2-505 mutant embryos with subsequent defects in chromosome segregation. The elongated prophase suggests that the segregation phenotype is a consequence of defects in events that occur during prophase, either in chromosome condensation or kinetochore assembly or function. Immunostaining with an antibody against a centromerespecific antigen indicates that the kinetochores of most chromosomes are functional. The immunostaining results are more consistent with defects in chromosome condensation being responsible for the segregation phenotype.

INTRODUCTION

(James and Elgin, 1986). Antibodies to this protein stain the chromocenter of polytene chromosomes from Drosophila salivary glands. Recently, GAGA-factor was shown to be specifically localized to centromeres in early embryos prior to the onset of transcription. Its localization at the centromeres is very similar to the locations of two highly repeated satellite sequences in α-heterochromatin (Raff et al., 1994). Since these satellite DNA sequences provide GA/TC-rich sites to which GAGA factor has been shown to bind in vitro, it is likely that GAGA factor is directly bound to this DNA. Recently, homologues of Drosophila HP1 were also found in mice and humans (Hamvar et al., 1992; Chevillard et al., 1993). Both Drosophila and murine HP1 are located at the centromeres of mitotic chromosomes (Wreggett et al., 1994; Kellum et al., accompanying paper). In addition, HP1 is diffusely distributed around the segregating chromosomes during mitosis in Drosophila embryos. Another heterochromatin-specific protein has been identified in mammals. This protein, centromere protein-B (CENP-B), is found throughout the centromeric heterochromatin of mitotic chromosomes (Cooke et al., 1990) and has been shown to bind to a 17 bp segment of the alphoid DNA repeat in vitro (Masumoto et al., 1989).

The chromosomes of virtually all higher eukaryotes contain regions that retain the cytological properties of condensed metaphase chromosomes throughout the cell cycle (Heitz, 1928). These regions, termed heterochromatin, differ from the bulk of the chromatin (euchromatin) in that they do not condense and decondense in response to cell cycle cues. The condensed appearance of heterochromatin is thought to be due to a specialized nucleoprotein structure, but the nature of this structure is poorly understood. The region of the chromosome that is most generally packaged into heterochromatin is that surrounding the centromere. There is evidence that centromeric heterochromatin is composed of both specific DNA sequences and specific proteins. For example, the DNA in this heterochromatin consists almost entirely of centromere-specific repetitive DNA sequences, both the highly repetitive satellite DNA sequences and middle repetitive elements (Pardue and Gall, 1970; Peacock et al., 1976). Proteins that are enriched in the heterochromatin at the centromeres have also been found. In Drosophila, these include heterochromatin protein 1 (HP1)

Key words: heterochromatin protein 1, chromosome segregation, Drosophila, Suvar2-5, embryo

1420 R. Kellum and B. M. Alberts Many functions have been ascribed to centromeric heterochromatin since its cytological discovery in 1928 (reviewed by John, 1988). Most speculations about heterochromatin function are focussed on its association with DNA sequences that are transcriptionally inert and are replicated late in S phase. Others have found evidence that heterochromatin may play a role in chromosome segregation (Lindsley and Novitski, 1958; Bernat et al., 1991; Wines and Henikoff, 1992; Hawley et al., 1993). Some have even speculated that heterochromatin has no function at all (John, 1988). Drosophila offers a unique opportunity for examining the function of centromeric heterochromatin. Extensive genetic analyses have identified a set of genes, the Suppressor of variegation (Suvar) genes, that are thought to be involved in forming heterochromatin (Sinclair et al., 1983; Locke et al., 1988; Wustmann et al., 1989). Loss of function mutations in these genes decrease the mosaic repression of a euchromatic gene that occurs when it is placed adjacent to centromeric heterochromatin by a genetic rearrangement. Most interestingly, HP1 is encoded by a member of this group of genes, Suvar25 (Eissenberg et al., 1990). Mutations in a subclass of the Suvar genes, including the Suvar2-5 gene, cause recessive embryonic lethality. In this paper we examine the embryonic lethality of mutations in the Suvar2-5 gene (Eissenberg and Hartnett, 1993), in the hope of gaining insight into the function of heterochromatin. To this end, the embryos produced from a cross between Suvar2-5 heterozygous parents were examined for specific developmental defects. Severe defects in nuclear morphology and chromosome segregation were found in roughly one quarter of the progeny. These progeny also had markedly diminished HP1 levels or an abnormal distribution of the protein. These observations suggest that HP1, in addition to its effect on positioneffect variegation, is also required for normal nuclear morphology and mitosis. MATERIALS AND METHODS Analysis of fixed embryos Staged embryos (0-3, 2-6, 6-9, 9-12, and 12-24 hours after egg deposition) were collected at room temperature from wild type, CyO/+, Suvar2-5/+, or Suvar2-5/CyO parents. The Suvar2-5 stocks were provided by the laboratories of T. Grigliatti, G. Reuter, and J. Eissenberg (Sinclair et al., 1983; Wustmann et al., 1989; Eissenberg et al., 1992). In most experiments, embryos were produced from a cross in which both parental genotypes were heterozygous for the same Suvar2-5 allele over a CyO chromosome (data in Table 1). Embryos were also collected from an analogous cross in which the CyO chromosome was replaced by a wild type second chromosome in both parents (data in Table 2). In the tests for a maternal effect of mutations in the Suvar2-5 gene on the pre-cycle 10 chromosome segregation phenotype, only the mothers were heterozygous for the Suvar205 allele, whereas in the test for a paternal effect only the fathers were heterozygous. The wild type parent was Oregon R. The staged embryos were fixed in methanol/EGTA according to Mitchison and Sedat (1983). Embryos were also fixed for up to 15 minutes in 37% formaldehyde before the methanol/EGTA treatment according to Kellogg et al. (1988). Similar results were obtained with the two protocols. The fixed embryos were stained with 0.1 mg/ml of 4,6-diamidino-2-phenylindole (DAPI) for 5 minutes. The chromosome morphology of the fixed embryos was then examined by fluorescence microscopy using a Nikon Microphot-FXA.

Embryos were classified according to their developmental age and scored as normal or abnormal for chromosome morphology. The nuclear cycle of each pre-cycle 14 embryo was determined by the number of nuclei and their placement within the embryo (Foe and Alberts, 1983). In Table 1, the cycle 10-14 embryos were scored as normal if at least 80% of the nuclei had normal morphology. This stringent criterion was used to take into account the normal variability within wild type collections of fixed embryos, which can contain up to 10% abnormal nuclei without any apparent effect on viability (Merrill et al., 1988). It also eliminates any embryos that are less severely affected by the Suvar 2-5 mutation and are able to discard the small percentage of defective nuclei. The frequency that a given nuclear defect was observed within the population of abnormal embryos (Table 2) was calculated as the number of times a given defect was observed divided by the number of embryos that were classified as abnormal from a cross in which both parents were heterozygous for a given Suvar2-5 allele over a wild type chromosome. Usually multiple types of defects were observed in a single embryo; therefore the sum of frequencies at which defects were observed for a given allele excedes 100%. In Table 4, the percentage of pre-cycle 10 embryos with any nuclear defect was calculated as the number of embryos between cycles 2 and 9 with any nuclear defect divided by the total number of embryos in these nuclear cycles. These data were obtained from crosses in which both parents were heterozygous for a given allele of Suvar2-5 over a CyO chromosome. The chromosome segregation phenotype was rescued by heat shock-induced expression of the Suvar2-5 gene under the control of an hsp70 promoter (using a white m4; Suvar2-505/CyO; HSHP1(83C) stock graciously supplied by J. Eissenberg (Eissenberg and Hartnett, 1993) (Table 1; Suvar2-505; HSHP1). This stock was placed at 30°C for three hours during which time embryos were collected. Immunostaining To localize HP1 in the abnormal embryos from Suvar2-5/+ parents (Fig. 3), embryos were fixed in formaldehyde by the method of Kellogg et al. 1988 (1:100 dilution of the supernatant from the C1A9 mouse monoclonal line, kindly provided by S. C. R. Elgin (James and Elgin, 1986). Propidium iodide was used to stain the DNA in these embryos, after removing the RNA by a 2 hour treatment with RNAase at 37°C. The staining was detected by laser scanning confocal microscopy using a Nikon optiphot fluorescence microscope equipped with the Bio-Rad MRC-600 laser scanning confocal attachment. The intensity of the HP1 immunofluorescence signal in each embryo (Table 3) was quantitated at low magnification (25×) using the histogram program of the MRC-600 Confocal Microscope Operating Software CoMOS (Bio-Rad Microscience Ltd, 1992-94). To localize the centromeres in abnormal embryos (Fig. 6), formaldehyde-fixed embryos were stained with an affinity purified anti-GAGA factor antibody (Raff et al., 1994). The primary antibody was recognized by a rhodamine-labeled anti-IgG that had been preabsorbed to fixed Drosophila embryos. The DNA in these embryos was visualized with the DNA-specific stain 4,6-diamidino-2phenylindole (DAPI). The embryos were visualized by fluorescence microscopy using a Nikon Microphot-FXA. Analysis of chromosome dynamics in living embryos Pre-cycle 10 embryos from wild type (Oregon R) parents and from a cross in which both parental genotypes were Suvar2-505/+ were injected with histones H2A and H2B prepared by J. Minden (Minden et al., 1989). After injection the embryos were placed in an aerobic microscope slide chamber and incubated at room temperature in the dark for 30 minutes to allow the histones to incorporate into the chromosomes. Chromosome dynamics were observed using confocal microscopy equipped with a laser video disc recorder LVR-5000A for time lapsed recordings (Sullivan et al., 1993). The chromosomes were observed under a 37.5-fold magnification using a #1 or #2 optical

Requirement for HP1 in chromosome segregation 1421 density filter. The lengths of the cell cycle stages were determined from the times that were automatically logged during the recording. Photography Fixed embryos were photographed using a Nikon Mikrophot FXA and Technical Pan ASA 135 film. Still images from the video recording of chromosome dynamics in wild type and Suvar2-5 embryos were photographed using a Shackman Instruments camera and Technical Pan ASA 135 film. The color images in Fig. 3 were printed by a UP5100 Sony color video printer.

RESULTS

Suvar2-5 mutants have defects in nuclear morphology and chromosome segregation The recessive embryonic lethal phenotype associated with Suvar2-5 mutations was analyzed using four alleles of Suvar25, named Suvar205, Suvar2-502, Suvar2-504, and Suvar2-505. Since the mutations are recessive lethal, we were restricted to examining embryos from a cross in which both parents are heterozygous for a given allele. One fourth of the progeny from such a cross are expected to be homozygous for the mutation and display a mutant phenotype. Because the earliest transcription does not begin in the embryo until cycle 10, all of the components required for the nuclei to replicate and divide are supplied by the nurse cells during oogenesis (Edgar and Schubiger, 1986). Therefore, a mutant phenotype that depends on a homozygous loss of HP1 would not be expected to become apparent until the maternal supply of the Suvar2-5

gene product (HP1) is depleted and zygotic transcription of the Suvar2-5 gene begins. The analysis of the embryonic lethality was begun by examining the embryos that did not hatch from a cross between heterozygotes, with each parent carrying the same Suvar2-5 allele. Approximately one fourth of the embryos from these crosses did not hatch. However, no specific developmental defect was observed in the unhatched embryos. When the embryos were fixed in methanol/EGTA and stained with the DNA-specific stain, DAPI, they were found to be mostly devoid of intact nuclei, with any remaining nuclei having defects in chromosome segregation and abnormal morphology (data not shown). Embryos at various stages of the process of nuclear degeneration were seen. In order to observe the nuclear defects at a developmental stage closer in time to when they were generated, staged collections of fixed embryos were stained with DAPI and examined for nuclear morphology defects. Varied defects were found (Fig. 1), including nuclei joined by chromatin bridges (Fig. 1C), chromosomes lagging during anaphase (Fig. 1D), and nuclei that appeared to be under-condensed or overcondensed (Fig. 1E and F). The mitotic synchrony that is characteristic of pre-cycle 14 embryos is lost in a high proportion of the Suvar2-5 embryos. Also, defective nuclei were seen beneath the normal plane of nuclei in many embryos (Fig. 1F). These nuclei appeared to be falling into the interior of the embryo where they presumably will be degraded, leading to the terminal phenotype we observe in unhatched embryos. This mechanism of discarding defective nuclei in Drosophila

Fig. 1. Nuclear defects observed in embryos from a cross in which both parental genotypes are Suvar2-5/CyO; (A) interphase nuclei from a wild type nuclear cycle 10 embryo; (B) anaphase nuclei from wild type nuclear cycle 10 embryo; (C) nuclei joined by chromatin bridges; (D) chromosomes lagging during anaphase; (E) undercondensed nuclei; (F) overcondensed nuclei falling from the embryo surface and asynchronous mitosis. Bar, 10 µm.

1422 R. Kellum and B. M. Alberts embryos has been previously observed in other mutants with nuclear defects (Sullivan et al., 1990). The nuclear defects were seen in abnormal embryos throughout embryonic development (Fig. 2), but there were differences in the frequency at which they were observed within a given embryo depending on the stage of development. The phenotype is most striking in abnormal embryos between nuclear cycles 10 and 14. Virtually all of the nuclei are defective in abnormal embryos between cycles 10 and 14, and the defects are usually of a single type (Fig. 2E and F). In contrast, only a fraction of the nuclei in abnormal embryos before cycle 10 are defective and these are most often a mixture of types (Fig. 2A-D). After cycle 14, embryos are often seen that are nearly normal but abnormal nuclei are seen in individual mitotic domains (Fig. 2G and H). These may be embryos that are heterozygous for the Suvar2-5 mutation. The percentage of embryos from crosses in which both the mothers and fathers were heterozygous for a given Suvar2-5 allele that were classified as abnormal during cycles 10 through 14 is given in Table 1. The data were derived from crosses in which the Suvar2-5 mutation was placed over a CyO balancer;

however, similar data are obtained when the CyO chromosome is replaced with a wild type second chromosome (e.g. see Table 2). Table 1 also gives the percentage of embryos that failed to hatch from each cross, corrected by subtracting the lethality caused by the CyO balancer where appropriate (Table 1). Within our error, this corrected percentage is roughly equivalent to the percentage of embryos that were classified as abnormal during cycles 10 through 14. We are confident that the chromosome abnormalities that are seen in the progeny are not caused by the CyO balancer chromosome, both because they are observed in experiments without a balancer (e.g. see Table 2) and because they are observed at a much lower frequency in the progeny from CyO/+ heterozygous parents. They are also not likely to be due to a second site mutation on the chromosome containing the Suvar2-5 mutation, because the defects seen during cycles 10 and 14 are also found at a high frequency in the progeny from a cross between parents that are each heterozygous for a different allele (see Suvar205/CyO × Suvar2-504/CyO in Table 1). Table 2 gives the frequency of each type of nuclear defect in the abnormal embryos between cycles 10 and 14. These data

Fig. 2. Normal and abnormal embryos from a cross in which both parental genotypes are Suvar 2-5/CyO at various developmental stages. Embryos were fixed in methanol/EGTA and stained with the DNA-specific stain, DAPI. (A) An abnormal nuclear cycle 3 embryo; (B) a normal nuclear cycle 3 embryo; (C) an abnormal nuclear cycle 6 embryo; (D) a normal nuclear cycle 6 embryo; (E) an abnormal nuclear cycle 13 embryo; (F) a normal nuclear cycle 13 embryo; (G) bottom panel: abnormal stage 7 embryo; top panel: enlarged view of the embryo shown in the bottom panel; (H) bottom panel: normal stage 7 embryo; top panel: enlarged view of the embryo shown in the bottom panel. Bar, 100 µm.

Requirement for HP1 in chromosome segregation 1423 Table 1. Percentage of embryos failing to hatch and percentage abnormal from a cross between Suvar2-5 heterozygotes /×?

Percentage of embryos failing to hatch*

Percentage abnormal embryo cycles 10-14†

Suvar205/CyO × Suvar205/CyO Suvar2-502/CyO × Suvar2-502/CyO Suvar2-504/CyO × Suvar2-504/CyO Suvar2-505/CyO × Suvar2-505/CyO CyO/+ × CyO/+ +/+ × +/+ Suvar205/CyO × Suvar2-504/CyO

33±3.9% (−CyO) 19±2.7% (−CyO) 33±3.4% (−CyO) 43±3.9% (−CyO) 29±3.5% 4±0.5% 32±3.6% (−CyO)

31±2.4% 24±3.8% 28±5.0% 36±5.6% 7±2.2% 6±0.8% 29±5.0%

Heat shock rescue Suvar2-505/CyO; × Suvar2-5 05/CyO; HSHP1 HSHP1

N.D.

6±1.2%

Cross:

*(−CyO) is a corrected value for which the lethality caused by the CyO mutation (29%; see CyO/+ × CyO/+) has been subtracted from the total lethality. Confirmation that these corrected values are representative of the lethality due to the Suvar2-5 mutation alone was obtained from crosses in which the CyO chromosome was replaced by a wild type chromosome. †Calculated as the average from four or more experiments ± standard deviation between experiments. A total of at least 300 embryos were counted from each cross.

also demonstrate that the nuclear defects in the progeny examined for Table 1 are not caused by the CyO chromosome, since the embryos examined for Table 2 were from a cross between Suvar2-5/+ parents. Of the nuclear defects, lagging chromosomes and chromatin bridges were seen at the highest frequency and appeared to be most closely correlated with the percentage of embryos that do not hatch from each allele (Table 1). The abnormal embryos have either decreased levels of HP1 or defects in its distribution We were interested in determining if the abnormal embryos observed during cycles 10 through 14 are the consequence of these embryos being homozygous for the mutant allele of Suvar2-5. In this case, we might expect to find greatly diminished levels of HP1 in the abnormal embryos from Suvar2-505 heterozygous parents. This allele has a frameshift in HP1 after ten amino acids and is almost certainly a null mutation (Eissenberg et al., 1992). The immunolocalization of HP1 during anaphase in a formaldehyde-fixed wild type embryo is shown in Fig. 3A. We have found that HP1 is only conspicuously enriched in the heterochromatin during interphase and telophase. During mitosis the most prominent staining is diffusely distributed around the segregating chromosomes, probably obscuring the fraction of

the protein that is also present on the chromosomes. This chromosome-associated fraction of HP1 is visible in the heterochromatin when unfixed chromosomes from Drosophila tissue culture cells are stained with antibodies against HP1 (Kellum et al., accompanying paper). In the progeny from the Suvar2-505 allele, diminished levels of HP1 were found in the population of embryos that were clearly abnormal during nuclear cycles 10 through 14 (Fig. 3B and C). The intensity of the HP1 immunofluorescence signal in the abnormal embryo in Fig. 3C is roughly one fourth that in the normal embryo of the same developmental age pictured next to it. As expected, this class of embryos with chromosome segregation defects and diminished levels of HP1 represent about 25% (6/23, 26%) of the embryos. The intensity values for the HP1 immunofluorescence signal in normal and abnormal Suvar2-505 embryos during nuclear cycles 5, 8, and 10 are given in Table 3. The intensity of this signal was determined at low magnification for each embryo produced from a cross between parents that are each heterozygous for the Suvar2-505 mutation and was found to vary little between embryos before nuclear cycle 10. However, a clear difference emerges, allowing us to distinguish three classes of embryos during nuclear cycle 10. This observation, and the fact that the class with dramatically diminished levels of HP1 applies to 25% of the embryos, suggest that zygotic transcription of the Suvar2-5 gene begins during nuclear cycle 10, and that the abnormal phenotype in embryos at this time is caused by the disappearance of the maternal supply of HP1 combined with an inability to express a wild type copy of the Suvar2-5 gene in homozygous mutant embryos. The localization of HP1 was perturbed in abnormal (but not normal) embryos in two other Suvar2-5 alleles during nuclear cycles 10 through 14. Again, the abnormal embryos represented roughly one quarter of the embryos at these stages (18/64 or 28.1% for Suvar205 in Fig. 3D; 8/28 or 28.6% for Suvar2-504 in Fig. 3F). In the Suvar205 allele, the total level of HP1 in the abnormal embryos is unaffected (data not shown), but the protein fails to localize in the nucleus. Instead of being dispersed around the chromosomes during anaphase (as it is in the wild type embryo in Fig. 3A), HP1 is excluded from the region containing the aberrantly segregating chromosomes (Fig. 3D). A similar pattern is observed in the mutant embryos from the Suvar2-504 allele (Fig. 3F). The localization patterns for HP1 in the abnormal embryos from the Suvar205 and Suvar2-504 alleles are consistent with the molecular nature of their mutations (see Fig. 4): in the Suvar205 allele, a domain of HP1 is missing which is important for its localization in the nucleus and its association with heterochromatin in salivary glands, while the protein from the Suvar2-504 allele is lacking a domain required for its local-

Table 2. Frequency of individual nuclear defects in the abnormal embryos from a cross between heterozygotes for each Suvar2-5 allele Genotype of both parents Nuclear defect Lagging chromosome/chromatin bridges Overcondensed chromosomes Undercondensed chromosomes Asynchronous mitosis Nuclei beneath the surface plane

Suvar205/+

Suvar2-502/+

Suvar2-504/+

Suvar2-505/+

72% 31% 3% 8% 36%

45% 23% 35% 36% 36%

67% 13% 23% 27% 17%

73% 41% 14% 34% 21%

1424 R. Kellum and B. M. Alberts

Fig. 3. HP1 localization during anaphase in embryos from wild type parents and from a cross in which both parental genotypes are Suvar25/CyO . Formaldehyde-fixed embryos were immunostained with anti-HP1 that was detected with a fluorescein-labeled secondary antibody (green) and the DNA stain, propidium iodide (red). (A) A wild type nuclear cycle 10 embryo during anaphase; (B) Suvar2-5 05 nuclear cycle 10 embryo with defective chromosome segregation during anaphase; (C) HP1 levels in an abnormal and a normal embryo, siblings from a cross in which both parental genotypes were Suvar2-5 05/CyO; both embryos are in nuclear cycle 10. A nuclear cycle 10 embryo with chromosome segregation defects from a cross in which both parental genotypes were (D) Suvar205/CyO; (E) Suvar2-5 02/CyO; (F) Suvar2504/CyO . Bar, 10 µm.

ization to the nucleus (Powers and Eissenberg, 1993). In contrast, the Suvar2-502 allele has a point mutation in its conserved chromodomain (Shaffer et al., 1993), which would not necessarily be expected to affect its localization, and, accordingly, the localization of HP1 during anaphase (Fig. 3A) is unperturbed in this allele (Fig. 3E). It is of interest that the four-fold decreased level of wild-type HP1 in the mutant Suvar2-505 embryo also appears to affect its localization to the nucleus, because the HP1 that persists at cycles 10 to 14 in these embryos is diffusely distributed throughout the cytoplasm but is absent from the region surrounding the defectively segregating chromosomes (Fig. 3B).

Also, since the earliest zygotic transcription does not begin until nuclear cycle 10, one would not expect to see a difference among the progeny until this time if all the effects depend on the genotype of the embryo (Edgar and Schubiger, 1986). From both western blot analysis and immunofluorescence

The defects seen in pre-cycle 10 embryos are caused by a maternal dosage effect Even though the nuclear defects seen before nuclear cycle 10 were qualitatively different from those between cycles 10 and 14 in being present at a lower frequency and less homogeneous in type within a given embryo, a relatively large percentage of the pre-cycle 10 embryos had detectable nuclear defects (Table 4). These defects were surprising because they are never seen in pre-cycle 10 embryos from wild type or CyO/+ parents.

Nuclear cycle 5 Nuclear cycle 8 Nuclear cycles 10-14

Table 3. Fluorescence intensity of HP1 signal in abnormal and normal embryos Fluorescence intensity Nuclear cycle

Class 1 Class 2

Normal

Abnormal

119.0±2.1* 138.8±8.1* 194.9±18 140.8±13

N.D. N.D. 38.1±7.8 107.9±0.3

Class 3 Class 2

The t-test was applied to determine that the differences between Classes 1, 2 and 3 are significant with 99.999% certainty. Out of 23 embryos, 5 were placed in Class 1 (+/+), 12 in Class 2 (Suvar2-505/+), and 6 in Class 3 (Suvar2-505/Suvar2-505). *Fluorescence intensity of HP1 signal in wild type embryos at these stages is 227.7±27.

Requirement for HP1 in chromosome segregation 1425

Fig. 4. Molecular basis of each Suvar2-5 allele. Cloning and sequence analysis of each allele has been determined and reported by J. C. Eissenberg and co-workers (Eissenberg et al., 1990, 1992; Powers and Eissenberg, 1993; and Shaffer et al., 1993). , N terminal homology domain (chromodomain). , C terminal homology domain. (Powers and Eissenberg, 1993).

staining of pre-cycle 10 embryos (Table 3), it is clear that HP1 is maternally supplied. The Suvar2-5 position-effect variegation phenotype is known to be dose-dependent in adults and to be affected by the maternal genotype (Wustmann et al., 1989; Eissenberg et al., 1992). Therefore, mutations in Suvar2-5 could also have a maternal dosage effect on the embryo. To determine if the early defects in embryos were caused by a decreased dose of the Suvar2-5 gene product (HP1) in their heterozygous mothers, embryos were examined from crosses in which only the maternal parent was heterozygous for the Suvar205 allele. Embryos from this cross were compared to the progeny from the reciprocal cross between Suvar205 heterozygous fathers and wild type mothers. As shown in Table 4 (Test for maternal/paternal effect), nuclear defects were only found at significant levels in the progeny from Suvar2-5 mothers and wild type fathers. The progeny from Suvar2-5 fathers and wild type mothers did not have significantly more defects than those from wild type parents. The maternal genotype also affected the viability of the progeny. Even though all of the progeny from this cross were either heterozygous for the Suvar205 allele or wild type, and, therefore, all of the progeny carried at least one wild type copy of the Suvar2-5 gene, 17% of the embryos did not hatch. This is in comparison to 4% of embryos not hatching from wild type parents or from a cross in which the Suvar205 mutation was present only in the fathers. The percentage of pre-cycle 10 embryos displaying the Table 4. Percentage of embryos with any detectable nuclear defects during cycles 2-9 and the effect of maternal and paternal genotype on these defects /×?

Percentage of embryos with nuclear defects during cycles 2-9*

Suvar2-502/CyO × Suvar2-502/CyO Suvar2-504/CyO × Suvar2-504/CyO Suvar2-505/CyO × Suvar2-505/CyO Suvar205/CyO × Suvar205/CyO

20±5.4% 32±6.8% 15±5.0% 18±1.0%

Test for maternal/paternal effect Suvar205/CyO × +/+ +/+ × Suvar 205/CyO +/+ × +/+

18±7.8% 6±5.1% 0%

Cross:

*Calculated as the average of four or more experiments ± the standard deviation between experiments.

maternal effect varied between the four alleles examined (Table 4). Notably, the effects were most severe in the progeny from Suvar2-504 which produces a protein lacking the domain required for its localization to the nucleus (Powers and Eissenberg, 1993). The greater severity of this allele might be the result of a dominant negative effect of this truncated protein on the full length HP1. The chromosome segregation phenotype can be rescued by heat-shock driven expression of the Suvar2-5 gene The chromosome segregation defect can be rescued by expressing a cDNA for the Suvar2-5 gene from an hsp70 promoter that was previously demonstrated by Eissenberg and Hartnett (1993) to rescue the Suvar2-5 recessive embryonic lethality. As reported by these authors, we found that heat shock-induced expression of the Suvar2-5 cDNA reduces the Suvar2-5 lethality (data not shown). In addition, we found that incubating the parents at 22°C while collecting embryos decreased the lethality from 43% to 28%, presumably as a result of constitutive expression from the hsp70 promoter at this temperature in the mothers. Rescue of the chromosome segregation phenotype by expression of the Suvar2-5 cDNA demonstrates that the embryonic phenotypes we observed are due to a deficiency of HP1, as we have assumed. For this test, a transgene with the Suvar2-5 gene placed under heat shock control was crossed into the Suvar2-505 allele, and adults of this genotype were incubated for 3 hours at 30°C while embryos were being collected. This experimental design allowed expression of the cDNA in both the parents and the progeny and reduced the percentage of abnormal progeny from the Suvar2-505 heterozygotes to a level found in embryos from wild type parents (Table 1; Heat shock rescue). Live analysis of the mutant phenotype We were concerned that the nuclear morphology defects observed in the fixed embryos might be affected by the fixation conditions used. We found that similar results were obtained with a more severe fixation protocol in which the methanol/EGTA fixation is preceded by a 5-15 minute formaldehyde fixation (e.g. Fig. 3). Further evidence that the defects were occurring in vivo came from viewing chromosome morphology and dynamics in living Suvar2-5 embryos. This was accomplished by injecting rhodamine-labeled histones H2A and H2B into pre-cycle 10 Suvar2-505 embryos. The labeled histones were incorporated into the chromosomes,

1426 R. Kellum and B. M. Alberts

Requirement for HP1 in chromosome segregation 1427 allowing their behavior during mitosis to be observed by confocal microscopy under conditions that do not kill normal embryos (Minden et al., 1989; Sullivan et al., 1993; R. Kellum, personal observations). The results confirm that the defects observed in the fixed specimens occur in vivo. Of the fourteen embryos that were observed during nuclear cycles 9 through 13, nine were near normal and five had moderate to severe defects in chromosome segregation. This fraction of embryos with nuclear defects (35.7%) is comparable to that observed in fixed specimens. As controls, seven wild type embryos were also observed during these nuclear cycles. The live analysis of the mutant phenotype allowed us to determine when and how the defects were generated. An example of this analysis is shown in Fig. 5. In Fig. 5A,a-e, a wild type embryo in nuclear cycle 10 is shown. A cycle 10 embryo from a cross between Suvar2-505/+ parents is shown in Fig. 5B,a-e, and the same embryo during nuclear cycle 11 in Fig. 5B,f-j. In the wild type embryo, nuclear morphology was normal throughout the embryo and the nuclear divisions occurred in a synchronous wave across the embryo. In contrast, nuclear morphology in the mutant embryo was heterogeneous and the nuclear divisions were asynchronous. Each of the nuclear defects observed in fixed embryos (see Table 2) was also observed in the living embryos. The most prominent defect observed in the living mutant embryos was chromosomes lagging during anaphase. The chromosomes were not completely separated before they began to decondense during telophase, and consequently, the daughter nuclei were often joined by a bridge of chromatin (Fig. 5B,a-d, short arrow, splitting into a pair of shorter arrows). While some lagging chromosomes were also observed in wild type embryos, they were rare and were usually resolved before telophase. We did not observe any evidence of defects in sister chromatid adhesion or of premature sister chromatid migration. We also did not observe individual chromosomes delayed at the metaphase plate in a large number of nuclei, however this was occasionally observed in the mutant embryos (e.g. left arrow in Fig. 5B,c and top arrow in Fig. 5B,g). Prophase was found to be lengthened more than two-fold in the mutant embryos (Table 5), while the lengths of the other stages of the cell cycle were no greater than in wild type embryos. An elongated prophase was observed throughout each Suvar2-5 embryo with a mutant phenotype, but the effect was greater in the nuclei in the central region of each embryo (on average, prophase was lengthened by 1.7±0.3-fold at the poles and by 2.5±0.2 in the center; see Table 5). Mitosis in the Drosophila embryo normally occurs in a rapid wave that is Fig. 5. Live analysis of nuclear cycle 10 in wild type and Suvar2-505 embryos. (A) (a-e) nuclear cycle 10 in the center of a wild type embryo beginning with: (prophase began at 6:14:01) (a) metaphase 6:16:12; (b) 6:16:41; (c) 6:16:48; (d) 6:17:19; (e) 6:18:14; B) (a-e) nuclear cycle 10 in the center of a Suvar2-505 mutant embryo beginning with: (a) interphase of nuclear cycle 10 4:40:25 (prophase begins at 4:40:52); (b) 4:46:05; (c) 4:48:05; (d) 4:49:11; (e) 4:50:38; (f-j) nuclear cycle 11 of Suvar2-505: (f) 4:52:15 (prophase begins at 4:53:23); (g) 5:01:23; (h) 5:02:17; (i) 5:03:16; (j) 5:06:10. The small arrows denote nuclei with subtle segregation defects during cycle 10 (a-e) which result in a more pronounced phenotype during cycle 11 (f-j); the large arrows denote nuclei that remain hyperchromatic throughout the cell cycle and eventually fall from the surface of the embryo. Bar, 10 µm.

propogated from the poles towards the center of the embryo (Foe and Alberts, 1983), with the nuclei at the center entering mitosis approximately 0.2-0.5 minutes after those at the poles (e.g. Fig. 5A). Lengthening prophase at the center of the mutant embryos generated an increasing lag in the onset of mitosis in this region with each subsequent division (1 minute in cycle 9, increasing to a 4 minute lag in cycle 12) and is probably responsible for the asynchrony in the nuclear divisions observed in these embryos. In some nuclei, the chromosomes failed to condense and congress to the metaphase plate altogether. These nuclei had an intensely fluorescent appearance throughout the cell cycle (Fig. 5B,a-c). The daughter nuclei from an incomplete segregation event (short arrow in Fig. 5B,a-c, which splits into a pair of short arrows in d-j) often took on this brightly staining appearance and failed to condense during prophase of the subsequent nuclear cycle (long arrow in Fig. 5B,f-h). These nuclei and those joined by chromatin bridges ultimately dropped into the interior of the embryo, where they presumably degenerated (pair of arrows in Fig. 5B,i,j). This left holes in the normally regular array of nuclei at the surface of wild type embryos.

Table 5. Lengths of prophase in wild type and abnormal Suvar2-5 embryos Length of prophase in Suvar2-5 Stage

Length of prophase in wild type

At pole

At center

At pole

At center

4.5 4.2 4.3 Avg. 4.3 (1.6×w.t.)

4.1 6.3 5.0 Avg. 5.1 (2.4×w.t.)

2.9 2.3 2.8 Avg. 2.7

2.3 1.9 2.1 Avg. 2.1

4.9 5.3 5.5

6.1 6.0 6.5

Avg. 5.2 (1.9×w.t.)

Avg. 6.2 (2.3×w.t.)

3.0 2.4 3.3 3.0 2.4 Avg. 2.8

2.8 2.4 2.3 3.0 2.8 Avg. 2.7

Cycle 11

6.2 5.0 5.8 Avg. 5.7 (1.4×w.t.)

8.3 7.5 10.7 Avg. 8.8 (2.6×w.t.)

4.0 3.7 4.3 Avg. 4.0

3.9 3.3 3.0 Avg. 3.4

Cycle 12

11.3 (1.9×w.t.)

14.0 (2.5×w.t.)

6.0

5.7

Cycle 9

Cycle 10

The prophase lengths (in minutes) were determined from live recordings of 5 abnormal Suvar2-5 embryos and from 6 wild type embryos. Multiple cell cycles were viewed in each embryo. The beginning of prophase was defined as the time at which chromosome condensation began and the end of prophase was defined as the time at which condensation was complete and the chromosomes were aligned at the metaphase plate. While some ambiguity is inherent in determining the exact timing of each of these events, in the abnormal embryos chromosome condensation was visibly incomplete through much of prophase. The greatest ambiguity was in determining the last 0.5 minutes of prophase, however an error of up to one minute in each case does not significantly alter the factor by which prophase is increased in the abnormal Suvar2-5 embryo relative to wild type or cause a significant lengthening of metaphase in the abnormal Suvar25 embryo. The increase in prophase length in the abnormal embryos (×w.t.) for each cell cycle was calculated as the average length in the abnormal embryos divided by the average length in the wild type embryos.

1428 R. Kellum and B. M. Alberts Which portion of the chromosomes are lagging during anaphase? In both the in vitro and in vivo analysis of the Suvar2-5 phenotype, chromosomes were seen lagging during anaphase. The lengthening of prophase we observed in the mutant embryos in live analysis suggests that defects in events occurring during this stage of the cell cycle contribute to the phenotype seen during anaphase. In order to better understand the cause of the failure of these chromosomes to separate completely, we compared the position of the centromeres during anaphase in wild type embryos to that in abnormal Suvar2-5 embryos. If the principle cause of the defects during anaphase is defective kinetochore function, we would expect to find centromeres delayed at the metaphase plate. In contrast, if the defect is the consequence of insufficient chromosome condensation we would expect to find that the centromeres move normally towards the poles, with the lagging chromatin consisting instead of the chromosome arms.

Fig. 6. Localization of centromere-specific staining in chromosomes displaying defects in chromosome segregation. Normally segregating anaphase chromosomes in a wild type embryo stained with (A) DAPI and (B) anti-GAGA factor; defectively segregating anaphase chromosomes in an abnormal embryo stained with (C) DAPI and (D) anti-GAGA factor in which GAGA factor localization can be traced along stretched chromosomal arms from spindle pole (arrowheads) towards the metaphase plate (arrow); defectively segregating telophase chromosomes in an abnormal embryo stained with (E) DAPI and (F) anti-GAGA factor in which some GAGA factor staining remains at the metaphase plate (arrow). The cell cycle stage was determined by anti-tubulin staining, with anaphase nuclei having a microtubule spindle and telophase nuclei having a microtubule mid-body structure. Bar, 10 µm.

To discriminate between these two possibilities, we used an antibody against GAGA factor that stains centromeres throughout the cell cycle in early Drosophila embryos (Raff et al., 1994). We have used this antibody to mark the positions of the centromeres in the Suvar2-5 embryos with abnormally segregating chromosomes. Ideally, we would have liked to use an antibody which marks only the positions of each kinetochore (rather than flanking heterochromatin) in the abnormal embryos. Further complicating the analysis of kinetochore position, GAGA factor localization has an abnormally stretched appearance in the mutant embryos. Nonetheless, in most defective nuclei (in 10 fields, 196/216, 91% of the nuclei) GAGA factor localization can be traced from the spindle pole along stretched chromatin towards the metaphase plate (e.g. Fig. 6C and D, short arrow at spindle pole and long arrow at metaphase plate). In a few nuclei in telophase (in 10 fields, 20/216, 9% of the nuclei) clusters of GAGA factor staining are seen at the metaphase plate that do not appear to be associated with the chromatin that is stretching

Requirement for HP1 in chromosome segregation 1429 towards the poles (e.g. Fig. 6E and F, arrowheads at spindle pole and arrow at metaphase plate). These observations led us to conclude that, in spite of the decondensed appearance of the GAGA factor-staining chromatin, in the majority of nuclei each kinetochore has migrated to the spindle pole and therefore appears to be functional. However, close examination of chromosome morphology at high magnification revealed inadequate condensation. DISCUSSION We undertook this analysis of the Suvar2-5 embryonic lethal phenotype for the purpose of gaining insight into the function(s) of heterochromatin in the cell. Most of the speculations about heterochromatin function in Drosophila are focused on its effects on transcription and replication. While euchromatic genes are mosaically repressed when juxtaposed to heterochromatin, a number of genetic loci are located in centromeric heterochromatin and their correct expression is dependent upon their proximity to this specialized chromatin and on the dose of the Suvar2-5 gene product (Hilliker, 1976; Hearn et al., 1991; Wakimoto and Hearn, 1990). Consequently, one function of heterochromatin is in regulating correct expression of these genes. Other studies have demonstrated a role for heterochromatin at the centromeres in chromosome segregation (Lindsley and Novitski, 1958; Bernat et al., 1991; Wines and Henikoff, 1992; Hawley et al., 1993). In these analyses of the Suvar2-5 phenotype, we wished to determine whether the embryonic lethality is the result of the known role of heterochromatin in regulating gene expression or if it is, instead, a consequence of more basic functions for heterochromatin in forming a normal nucleus and in mitosis.

Suvar2-5 mutants have defects in chromosome morphology and segregation In these analyses of the Suvar2-5 embryonic phenotype, we found embryos between cycles 10 and 14 with severe chromosome defects at a frequency roughly equivalent to the lethality for each allele. The defects which appeared to be most closely correlated with the lethality were lagging chromosomes and nuclei joined by chromatin bridges (Table 2). Our analysis of living embryos attribute these defects to inadequate chromosome condensation during prophase (Fig. 5). Beginning with nuclear cycle 10, the level of HP1 is diminished by one fourth in abnormal Suvar2-505 embryos relative to its level in normal sibling embryos (Fig. 3C and Table 3). Abnormal embryos with diminished levels of HP1 during nuclear cycle 10 are present at a frequency that is consistent with these embryos being homozygous for the Suvar2-505 mutation. Thus, we believe that the severe chromosome segregation phenotype observed between nuclear cycles 10 and 14 reflects the failure to express a wild type copy of the gene in the homozygous Suvar2-5 embryos. Similar nuclear defects were also seen in pre-cycle 10 embryos. These defects, however, are qualitatively different from those in nuclear cycle 10 through 14 embryos in that they are present at a lower frequency within individual embryos and they are usually of mixed type. The early defects were shown to be dependent upon the genotype of the mother but independent of paternal genotype. The maternal genotype also

affected the viability of the progeny in a test cross for maternal effect, where all of the embryos were either heterozygous for the Suvar2-5 mutation or wild type. The maternal dosage effect is consistent with the dose-dependence of the Suvar2-5 position-effect variegation phenotype in adults, which is thought to reflect an extreme sensitivity of heterochromatin assembly to the level of its components (Locke et al., 1988). Our observation that the diminished level of wild type HP1 in the Suvar2-505 abnormal embryos affects its localization in the nucleus supports this hypothesis (Fig. 3B). The occurrence of the chromosome morphology and segregation defects throughout embryonic development suggests a cell vital function for HP1. This is consistent with the results of Eissenberg and Hartnett (1993) in rescuing the Suvar2-5 embryonic lethality with a heat shock-inducible Suvar2-5 cDNA. Complete rescue of Suvar2-5 homozygous embryos required a 30 minute heat shock every day until eclosion. Interestingly, one half of the expected homozygous progeny could be rescued even after delaying heat shock for 5 days, but all of the resultant flies had severe reductions in eye size and the number of ocelli. These abnormalities might be explained by mitotic defects in these cell types. How might a deficiency of HP1 cause chromosome segregation defects? Perturbations in heterochromatin during interphase could potentially affect chromosome segregation indirectly by perturbing the expression of a heterochromatic gene that is required for chromosome segregation. For example, the rolled gene, which requires proximity to heterochromatin for normal expression (Eberl et al., 1993), was recently found to encode a MAP kinase (Brunner et al., 1994). However, defects in this gene (and all other genes located within heterochromatin for which the lethal period has been determined) cause lethality in late larvae and early pupae, much later than Suvar2-5 mutants die (Hilliker, 1976). The defects that are seen before the onset of zygotic transcription also argue against this interpretation, although a gene, such as rolled, could be misexpressed during oogenesis and be responsible for the early defects. Heterochromatin organization could also directly affect a number of processes in chromosome segregation. Although we recently found that a large pool of HP1 is dispersed around the segregating chromosomes during mitosis, there is the possibility that a fraction of the protein remains associated with the heterochromatin throughout the cell cycle. When Drosophila tissue culture cells are prepared in a way which removes soluble protein and avoids fixation of the mitotic chromosomes, HP1 is visible in the heterochromatin at the centromeres and telomeres. HP1 is not localized around the segregating chromosomes in the defective anaphase nuclei in two alleles (see Fig. 3D and F), probably reflecting its exclusion from the heterochromatin throughout the cell cycle. Recent studies have attributed a role to centromeric heterochromatin in sister chromatid adhesion (Wines and Henikoff, 1992) and in the assembly of a functional kinetochore (Bernat et al., 1991). Prophase is lengthened two-fold in the Suvar2-5 embryos (Table 5), suggesting that the segregation phenotype seen during anaphase is a consequence of defects that occurred earlier during prophase, either in kinetochore assembly or chromosome condensation. In both the in vitro and in vivo analyses we see evidence of failed kinetochore function in only

1430 R. Kellum and B. M. Alberts a small percentage of nuclei in the Suvar2-5 mutant embryos; therefore these defects probably cannot account for the entire phenotype. GAGA factor immunolocalization was used as a marker for the centromeric regions of the chromosomes in these defective anaphase and telophase nuclei, and the results of these analyses are more consistent with an defective condensation of the chromosomes causing the phenotype. Incomplete chromosome condensation in the Suvar2-5 mutants is likely to affect chromosome segregation by causing the chromosome arms to become tangled and lag at the metaphase plate during anaphase. Supporting this view, the phenotype we observe for Suvar2-5 embryos is similar to the effects of injecting the drug VM26, an inhibitor of topoisomerase II activity, into living Drosophila embryos (Buchenau et al., 1993). It is also most similar to the phenotype of one of three classes of cut mutants in S. pombe, in which a portion of the nuclear chromatin is stretched by the elongated spindle, but the entire nucleus is not separated (Samejima et al., 1993). This class of cut mutants includes mutations in the gene encoding topoisomerase II. How might HP1 be involved in condensing the mitotic chromosomes? How the organization in the heterochromatic regions of the chromosomes might have a global effect on the condensation of the mitotic chromosomes is not straight-forward. It is possible that the chromosomes fail to segregate completely because of inadequate condensation in the heterochromatin only, causing these regions to stretch and fail to translate the pulling force of the spindle to the remainder of the chromosome. However, on close morphological examination, the chromosomes appear to be decondensed globally. We present two models for the role of HP1 in this process. By the simplest model, HP1 provides a catalytic activity that helps maintain the compact state of heterochromatin during interphase and spreads throughout the chromosomes during prophase. This model is supported by the presence of HP1 at a low level throughout unfixed mitotic chromosomes and the similarity of the Suvar2-5 phenotype to that of mutants for the lodestar gene, which encodes a DNA helicase that enters the nuclei during prophase and has an anaphase distribution similar to that of HP1 in embryos (Girdham and Glover, 1991). A second model emphasizes the observation that heterochromatin is associated with the nuclear envelope in virtually every cell type (Bouteille et al., 1983; Hancock and Boulikas, 1982; Franke et al., 1981; Brown, 1966). Evidence that this association may be important for chromosome condensation during prophase comes from the identification of specific chromosomal focal points for initiating chromosome condensation along the euchromatic arms that are associated with the nuclear envelope in Drosophila (Hiraoka et al., 1989). Many of these appear to be sites of intercalary heterochromatin (Hochstrasser et al., 1986). By this model HP1 is required to assemble heterochromatin or to organize the heterochromatin in the interphase nucleus, and a correctly organized interphase nucleus is a prerequisite for its reorganization during mitosis. Furthermore, the dispersed pool of HP1 during mitosis may facilitate the process of reassembling the interphase nucleus after mitosis. By either of these models for the role of HP1 in chromosome condensation, sister chromatids fail to completely

separate as a consequence of the poor condensation in Suvar25 mutant embryos. The defective nuclei are then eliminated by a process of falling into the interior of the embryo where their DNA is degraded, leaving embryos that are largely devoid of intact nuclei. Thus, the lethality associated with Suvar2-5 mutations is ultimately the result of extensive nuclear loss. We thank T. A. Grigliatti, G. Reuter and J. C. Eissenberg for kindly providing Suvar 2-5 mutant stocks and the Suvar2-505; HSHP1(83) transgenic stock; and S. C. R. Elgin for providing the C1A9 mouse hybridoma cell line that produces monoclonal antibodies against HP1. We also thank B. Sullivan, J. Sedat, M. Moritz, Y. Zheng, R. Aroian, W. Marshall and A. Dernburg for critical reading of the manuscript. This work was supported by grants from the National Institutes of Health (GM24020) to B.A. and by a Damon Runyon-Walter Winchell Cancer Research Fund Fellowship, DRG-1118 to R.K.

REFERENCES Bernat, R. L., Delannoy, M. R., Rothfield, N. and Earnshaw, W. C. (1991). Disruption of centromere assembly during interphase inhibits kinetochore morphogenesis and function in mitosis. Cell 66, 1229-1238. Bouteille, M., Bouvier, D. and Seve, A. P. (1983). Heterogeneity and territorial organization of the nuclear matrix and related structures. Int. Rev. Cytol. 83, 135-182. Brown, S. W. (1966). Heterochromatin. Science 151, 417-425. Brunner, D., Oellers, N., Szabad, J., Biggs, W. H. III, Zipursky and Hafen, S. L. S. (1994). A gain-of-function mutation in Drosophila MAP kinase activates multiple receptor tyrosine kinase signaling pathways. Cell 76, 875888. Buchenau, P., Saumweber, H. and Arndt-Jovin, D. J. (1993). Consequences of topoisomerase II inhibition in early embryogenesis of Drosophila revealed by in vivo confocal laser scanning microscopy. J. Cell Sci. 104, 1175-1186. Chevillard, C., Reik, W., McDermott, M., Fontes, M., Mattei, M. and Singh, G. P. B. (1993). Chromosomal localization of human homologs of the Drosophila heterochromatin protein 1 (HP1) gene. Mammalian Genome 4, 124-126. Cooke, C. A., Bernat, R. L. and Earnshaw, W. C. (1990). CENP-B: A major human centromere protein located beneath the kinetochore. J. Cell Biol. 110, 1475-1488. Dorer, D. R. and Henikoff, S. (1994). Expansions of transgene repeats cause heterochromatin formation and gene silencing in Drosophila. Cell 77, 9931002. Eberl, D. F., Duyf, B. J. and Hilliker, A. J. (1993). The role of heterochromatin in the expression of a heterochromatic gene, the rolled locus of Drosophila melanogaster. Genetics 134, 277-292. Edgar, B. A. and Schubiger, G. (1986). Parameters controlling transcriptional activation during early Drosophila development. Cell 44, 871-877. Eissenberg, J. C., James, T. C., Foster-Hartnett, D. M., Hartnett, T., Ngan, V. and Elgin, S. C. R. (1990). Mutation in a heterochromatin-specific protein is associated with suppression of position-effect variegation in Drosophila melanogaster. Proc. Nat. Acad. Sci. USA 87, 9923-9927. Eissenberg, J. C., Morris, G. D., Reuter, G. and Hartnett, T. (1992). The heterochromatin-associated protein HP1 is an essential protein in Drosophila with dosage-dependent effects on position-effect variegation. Genetics 131, 345-352. Eissenberg, J. C. and Hartnett, T. (1993). A heat shock-activated cDNA rescues the recessive lethality of mutations in the heterochromatin-associated protein HP1 of Drosophila melanogaster. Mol. Gen. Genet. 240, 333-338. Foe, V. E. and Alberts, B. M. (1983). Studies of nuclear and cytoplasmic behaviour during the five mitotic cycles that precede gastrulation in Drosophila embryogenesis. J. Cell Sci. 61, 31-70. Franke, W. W., Scheer, U., Krohne, G. and Jarash, E. (1981). The nuclear envelope and the architecture of the nuclear periphery. J. Cell Biol. 91, 39-50. Gilson, E., Laroche, T. and Gasser, S. M. (1993). Telomeres and the functional architecture of the nucleus. Trends Cell Biol. 3, 128-134. Girdham, C. H. and Glover, D. M. (1991). Chromosome tangling and breakage at anaphase result from mutations in lodestar, a Drosophila gene encoding a putative nucleoside triphosphate binding protein. Genes Dev. 5, 1786-1799.

Requirement for HP1 in chromosome segregation 1431 Hamvar, R. M., Reik, W., Gaunt, S. J., Brown, S. D. and Singh, P.B. (1992). Mapping of a mouse homolog of a heterochromatin protein gene on the X chromosome. Mammalian Genome 2, 72-75. Hancock, R. and Boulikas, T. (1982). Functional organization in the nucleus. Int. Rev. Cytol. 79, 165-214. Hawley, R. S., Irick, H., Zitron, A. E., Haddox, D. A., Lohe, A., New, C. Whitley, M. D., Arbel, T., Jang, J., McKim, K. et al. (1993). There are two mechanisms of achiasmate segregation in Drosophila, one of which requires heterochromatic homology. Dev. Genet. 13, 440-467. Hearn, M. G., Hedrick, A., Grigliatti, T. A. and Wakimoto, B. T. (1991). The effect of modifiers of position-effect variegation on the variegation of heterochromatic genes of Drosophila melanogaster. Genetics 128, 785-797. Heitz, E. (1928). Das heterochromatin der moose. I Jahrb. Wissensch. Bot. 69, 762-818. Hilliker, A. J. (1976). Genetic analysis of the centromeric heterochromatin of chromosome 2 of Drosophila melanogaster: Deficiency mapping of EMSinduced lethal complementation groups. Genetics 83, 765-782. Hiroaka, Y., Minden, J. S., Swedlow, J. R., Sedat, J. W. and Agard, D. A. (1989). Focal points for chromosome condensation and decondensation revealed by three-dimensional in vivo time-lapse microscopy. Nature 342, 293-296. Hochstrasser, M., Mathog, D., Gruenbaum, Y., Saumweber, H. and Sedat, J. W. (1986). Spatial organization of chromosomes in the salivary gland nuclei of Drosophila melanogaster. J. Cell Biol. 102, 112-123. James, T. C. and Elgin, S. C. R. (1986). Identification of a nonhistone chromosomal protein associated with heterochromatin in Drosophila melanogaster and its gene. Mol. Cell. Biol. 6, 3862-3872. John, B. (1988). The biology of heterochromatin. In Heterochromatin: Molecular and Structural Aspects (ed. R. S. Verma), pp. 1-128. Cambridge University Press, Cambridge. Kellogg, D. R., Mitchison, T. J. and Alberts, B. (1988). Behaviour of microtubules and actin filaments in living Drosophila embryos. Development 103, 675-686. Kellum, R., Raff, J. W. and Alberts, B. M. (1995). Heterochromatin protein 1 distribution during development and during the cell cycle in Drosophila embryos. J. Cell Sci. 108, 1407-1418. Lindsley, D. and Novitski, E. (1958). Localization of the genetic factors responsible for the kinetic activity of the X chromosome of Drosophila melanogaster. Genetics 43, 790-798. Locke, J., Kotarski, M. A. and Tartof, K. D. (1988). Dosage-dependent modifiers of position effect variegation in Drosophila and a mass action model that explains their effect. Genetics 120, 181-198. Masumoto, H., Masukata, H., Muro, Y., Nozaki, N. and Okazaki, T. (1989). A human centromere antigen CENP-B interacts with a short specific sequence in alphoid DNA, a human centromeric satellite. J. Cell Biol. 109, 1963-1973. Merrill, P. T., Sweeton, D. and Wieschaus, E. (1988). Requirements for autosomal gene activity during precellular stages of Drosophila melanogaster. Genetics 10, 345-350. Minden, J. S., Agard, D. A., Sedat, J. W. and Alberts, B. M. (1989). Direct

cell lineage analysis in Drosophila melanogaster by time lapse, threedimensional optical microscopy of living embryos. J. Cell Biol. 109, 505516. Mitchison, T. J. and Sedat, J. (1983). Localization of antigenic determinants in whole Drosophila embryos. Dev. Biol. 99, 261- 264. Pardue, M. L. and Gall, J. G. (1970). Chromosomal localization of mouse satellite DNA. Science 168, 1356-1358. Peacock, W. J., Appels, R. Dunsmuir, P., Lohe, A. R. and Gerlach, W. L. (1976). Highly repeated DNA sequences: Chromosomal location and evolutionary conservatism. In Int. Cell Biol. (ed. B. K. Brinkley and K. R.), pp. 494-506. Rockefeller University Press: New York. Powers, J. A. and Eissenberg, J. C. (1993). Overlapping domains of the heterochromatin-associated protein HP1 mediate nuclear localization and heterochromatin binding. J. Cell Biol. 120, 291-299. Raff, J. W., Kellum, R. and Alberts, B. (1994). The Drosophila GAGA transcription factor is associated with specific regions of heterochromatin throughout the cell cycle. EMBO J. 13, 5977-5983. Samejima, I., Matsumoto, T., Nakaseko, Y., Beach, D. and Yanagida, M. (1993). Identification of seven new cut genes involved in Schizosaccharomyces pombe mitosis. J. Cell Sci. 105, 135-143. Shaffer, C. D., Wallrath, L. L. and Elgin, S. C. R. (1993). Regulating genes by packaging domains: Bits of heterochromatin in euchromatin? Trends Genet. 9, 35-37. Sinclair, D. A. R., Mottus, R. C. and Grigliatti, T. A. (1983). Genes which suppress position-effect variegation in Drosophila melanogaster are clustered. Mol. Gen. Genet. 191, 326-333. Sullivan, W., Minden, J. S. and Alberts, B. M. (1990). daughterless-abo-like, a Drosophila maternal-effect mutation that exhibits abnormal centrosome separation during the late blastoderm divisions. Development 110, 311-32 Sullivan, W., Daily, D. R., Fogarty, P., Yook, K. J. and Pimpinelli, S. (1993). Delays in anaphase initiation occur in individual nuclei of the syncytial Drosophila embryo. Mol. Biol. Cell. 4, 885-896. Talbert, P. B., LeCiel, D. S. and Henikoff, S. (1994). Modification of the Drosophila heterochromatic mutation brownDominant by linkage alterations. Genetics 136, 559-571. Wakimoto, B. T. and Hearn, M. G. (1990). The effects of chromosome rearrangements on the expression of heterochromatic genes in chromosome 2L of Drosophila melanogaster. Genetics 125, 141-154. Wines, D. R. and Henikoff, S. (1992). Somatic instability of a Drosophila chromosome. Genetics 131, 683-691. Wreggett, K. A., Hill, F., James, P. S., Hutchings, A., Butcher, G. W. and Singh, P. B. (1994). A mammalian homologue of Drosophila heterochromatin protein 1 (HP1) is a component of constitutive heterochromatin. Cytogenet. Cell Genet. 66, 99-103. Wustmann, G., Szidonya, J., Taubert, H. and Reuter, G. (1989). The genetics of position-effect variegation modifying loci in Drosophila melanogaster. Mol. Gen. Genet. 217, 520-527. (Received 13 September 1994 - Accepted 16 December 1994)