Genetic and Molecular Analysis of wings apart-like (wapl), a Gene Controlling Heterochromatin Organization in Drosophila melanogaster

Corresponding author: Maurizio Gatti, Dipartimento di Genetica e Biologia Molecolare, Universitá di Roma “La Sapienza,” Ple. Aldo Moro, 5, 00185 Rome, Italy., gatti{at}axcasp.caspur.it (E-mail)

Communicating editor: R. S. HAWLEY

	ABSTRACT

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mutations in the X-linked gene wings apart-like (wapl) result in late larval lethality associated with an unusual chromosome morphology. In brain cell metaphases of wapl mutants, sister chromatids of all chromosomes are aligned parallel to each other instead of assuming the typical morphology observed in wild type. This effect is due to a loosening of the adhesion between sister chromatids in the heterochromatic regions of the chromosomes. Despite this aberrant chromosome morphology, mutant brains exhibit normal mitotic parameters, suggesting that heterochromatin cohesion is not essential for proper centromere function. On the basis of these observations, we examined the role of wapl in meiotic chromosome segregation in females. wapl exhibits a clear dominant effect on achiasmate segregation, giving further support to the hypothesis that proximal heterochromatin is involved in chromosome pairing during female meiosis. We also examined whether wapl modulates position-effect variegation (PEV). Our analyses showed that wapl is a dominant suppressor of both white and Stubble variegation, while it is a weak enhancer of brown variegation. wapl maps to region 2D of the X chromosome between Pgd and pn. We identified the wapl gene within a previously conducted chromosomal walk in this region. The wapl transcriptional unit gives rise to two alternatively spliced transcripts 6.5- and 5-kb long. The protein encoded by the larger of these transcripts appears to be conserved among higher eukaryotes and contains a tract of acidic amino acids reminiscent of many chromatin-associated proteins, including two [HP1 and SU(VAR)3-7] encoded by other genes that act as suppressors of PEV.

THE term heterochromatin was coined by HEITZ 1928 to denote the chromosomal regions that remain condensed during most of the cell cycle and exhibit positive heteropycnosis at prophase. Since Heitz's original observations, a variety of studies have shown that most chromosomes in higher eukaryotes contain large regions of constitutive heterochromatin; in some cases, entire chromosomes are heterochromatic. Although the cytological differences between heterochromatin and euchromatin are most pronounced during interphase and prophase, these components of the genome can also be distinguished in metaphase: in the heterochromatic segments of metaphase chromosomes, sister chromatids (SCs) are more tightly apposed than they are in the euchromatin (HEITZ 1933 ). That is, the sister chromatids cannot be resolved from each other in heterochromatic regions but are well separated in the euchromatic arms.

In addition to these cytological features, heterochromatic regions have a number of other properties that distinguish them from euchromatin. They are enriched in middle and highly repetitive DNA, are specifically stained by the C-banding technique, replicate later than the bulk of euchromatin during S phase, exhibit reduced meiotic recombination, and contain many fewer genes per length of DNA than euchromatin (reviewed by GATTI and PIMPINELLI 1992 ).

Heterochromatin is mostly transcriptionally inert and, as shown by the phenomenon of position-effect variegation (PEV), it can silence the expression of euchromatic genes that have been relocated next to heterochromatin by a chromosomal rearrangement or a transposition event. This gene inactivation, which is thought to be the consequence of the diffusion of heterochromatic proteins into the neighboring euchromatin, is clonally inherited, leading to a variegated phenotype (for recent reviews see WAKIMOTO 1998 ; WALLRATH 1998 ).

Finally, recent studies have suggested that heterochromatin plays a crucial role in meiotic chromosome pairing and segregation in Drosophila females. In Drosophila oocytes, the heterochromatin of homologous chromosomes is physically associated throughout prophase until metaphase I. This association appears to mediate proper disjunction of achiasmate homologs, in that segregation of nonrecombinant chromosomes is disrupted by heterochromatic deficiencies (HAWLEY et al. 1992 ; DERNBURG et al. 1996B ; KARPEN et al. 1996 ).

The many distinctive features of heterochromatin are likely to reflect the association of these genomic regions with specific chromosomal proteins. Studies in Drosophila have led to the identification of at least ten heterochromatin-associated proteins in this organism: HP1 (reviewed by ELGIN 1996 ), DmORC-2 (PAK et al. 1997 ; HUANG et al. 1998 ), SU(VAR)3-7 (REUTER et al. 1990 ; CLEARD et al. 1997 ), GAGA (FARKAS et al. 1994 ; RAFF et al. 1994 ; HUANG et al. 1998 ), PROD (TOROK et al. 1997 ), MOD (GARZINO et al. 1992 ; PERRIN et al. 1998 ), ARP-4 (FRANKEL et al. 1997 ), and three Drosophila polypeptides that specifically bind the CENP-B box present in human centromeric -satellite DNA (AVIDES and SUNKEL 1994 ). Although all of these proteins are present in both euchromatin and heterochromatin, they are particularly abundant in heterochromatic regions. Moreover, recent studies have shown that HP1, DmORC-2, and SU(VAR)3-7 physically interact with each other and with the actin-related protein ARP-4 (CLEARD et al. 1997 ; FRANKEL et al. 1997 ; PAK et al. 1997 ; HUANG et al. 1998 ). Remarkably, an interaction between ORC and HP1 has also been found in Xenopus (PAK et al. 1997 ). The evolutionary conservation of large complexes of heterochromatin-associated proteins indicates that Drosophila is a highly suitable model system to elucidate the molecular structure of heterochromatin.

Mutations in 6 of the Drosophila genes encoding heterochromatin-associated proteins [Su(var)205, Su(var)3-7, DmORC-2, modulo (mod), Trithorax-like (Trl), and proliferation disrupter (prod)] have been isolated and characterized. Mutations in these genes have a variety of phenotypic effects that suggest roles for their products in gene regulation, chromosome condensation, and chromosome segregation. Except prod, all these genes act as modifiers of PEV and are thus implicated in the direct or indirect control of gene expression. Mutant alleles of Su(var)205 (the gene encoding HP1), Su(var)3-7, DmORC-2, and mod are dominant suppressors of PEV, whereas mutations of Trl (which encodes the GAGA factor) are dominant enhancers of PEV (reviewed by ELGIN 1996 ; WAKIMOTO 1998 ; WALLRATH 1998 ). The molecular, genetic, and cytological data on these genes are consistent with the hypothesis that PEV is modulated by the local concentration of heterochromatin-associated proteins (for recent reviews see WAKIMOTO 1998 ; WALLRATH 1998 ). Over 100 genes that modify PEV have been described and partially characterized to date (reviewed by WAKIMOTO 1998 ; WALLRATH 1998 ), implying that the repertoire of heterochromatin-associated proteins may be very large.

The cytological analysis of mitotic chromosomes from individuals homozygous for DmORC-2, Su(var)205, Trl, and prod revealed that these genes are required for specific aspects of chromosome organization. Mutations in the DmORC-2 gene affect mitotic condensation of both heterochromatic and euchromatic regions of larval brain chromosomes (GATTI and BAKER 1989 ). Although HP1 accumulates in both centric heterochromatin and telomeres, the cytological effects of Su(var)205 mutations appear to be restricted to telomeres, which in mutant brain cells are often associated, giving rise to monocentric ring and multicentric linear and ring chromosomes (FANTI et al. 1998 ). Su(var)205 mutations also affect chromosome segregation in both embryonic and larval brain cells, probably owing to problems in the resolution of telomeric associations during anaphase (KELLUM and ALBERTS 1995 ; FANTI et al. 1998 ). Abnormalities in embryonic chromosome segregation have also been observed in Trl mutants, suggesting a primary defect in heterochromatin condensation (BHAT et al. 1996A ). Finally, a clear failure in heterochromatin condensation, accompanied by a partial block in the metaphase-anaphase transition, has been observed in larval brain cells of mutants in the prod locus (TOROK et al. 1997 ).

In this article we describe a gene called wings apart-like (wapl) (GVOZDEV et al. 1977 ), which plays a peculiar role in heterochromatin structure: wapl mutations prevent the normal close apposition of sister chromatids in heterochromatic regions, but do not appear to affect either heterochromatin condensation or chromosome segregation. This suggests that wapl is required to hold sister chromatids tightly together in mitotic heterochromatin. In addition, we show that wapl is implicated in heterochromatin pairing during female meiosis and in the modulation of PEV. The wapl gene encodes a novel protein of 1741 amino acids that has some features in common with other polypeptides that modulate PEV.

	MATERIALS AND METHODS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Stocks:
The wapl locus, also called l(1)2Dd (LINDSLEY and ZIMM 1992 ), is defined by lethal and semilethal alleles; in the semilethal escapers the wings are spread apart (GVOZDEV et al. 1977 ). The locus was named wings apart-like after this phenotype (GVOZDEV et al. 1977 ); we have earlier referred to wapl as parallel sister chromatids (pasc) based on the cytological phenotype (see below) caused by wapl mutations. The wapl^A17, wapl^C204, and wapl^HC262 mutations and the wapl region deficiencies Df(1)JC105 (2D4-6; 2D4-6) and Df(1)64c18 (2E1-2; 3C2) were induced by G. Lefevre using X-ray mutagenesis (LEFEVRE 1981 ; PERRIMON et al. 1985 ). Df(1)Pgd-kz (2D3-4; 2F5), also deficient for the wapl region, was synthesized by GERASIMOVA and ANANJEV 1972 using -irradiation, while the allele wapl^11P3 was induced by ethyl methane sulfonate (EMS) by PERRIMON et al. 1985 . All these strains were kindly supplied by N. Perrimon (Harvard University Medical Schoool, Boston, MA). To generate chromosomes carrying both wapl and w, we mated wapl^A17/y w, wapl^C204/y w, wapl^HC262/y w, and y wapl^11P3/w females to FM7, y^31d sc⁸ w^a B males and recovered putative wapl^A17 w, wapl^C204 w, wapl^HC262 w, and y wapl^11P3 w recombinants over FM7; these recombinants were subsequently tested for the presence of wapl. Df(1)JC105 and Df(1)Pgd-kz were marked with w using a similar method. The wapl alleles and deletions of the wapl region were maintained as heterozygotes with either FM7, y^31d sc⁸ w^a B, FM7a, y^31d sc⁸ w^a B1 v^Of, or Binsn, sc^S1 sc⁸ sn^X2 B. The wapl mutations were also kept in stocks of the form wapl/Dp(1;Y)w⁺303 x C(1)DX, y f/Dp(1;Y)w⁺303, where Dp(1;Y)w⁺303 is a duplication of the X chromosome region 2D1-2; 3D3-4 onto an entire Y. l(1)90 carries a lethal mutation induced by EMS that maps to the interval between Phosphogluconate dehydrogenase (Pgd) and prune (pn) but that is not allelic to wapl (ALATORTSEV and TOLCHKOV 1985 ). The l(1)90 mutation defines the complementation group l(1)2Dg (LINDSLEY and ZIMM 1992 ; FROLOV and ALATORTSEV 1994 ). The l(1)90/FM4 stock was obtained from V. A. Gvozdev (Institute of Molecular Genetics, Moscow, Russia).

Further explanations of genetic symbols and stocks can be found in LINDSLEY and ZIMM 1992 or are described below.

Chromosome cytology:
To select wapl mutant larvae, we exploited the colorless Malpighian tubule phenotype caused by white mutations. In wapl/FM7 stocks, w⁺ wapl males can be easily distinguished from their w^a-bearing FM7 brothers. In the w wapl/Binsn stocks both wapl males and wapl/wapl females have colorless Malpighian tubules that distinguish them from their Binsn-bearing siblings. The Malpighian tubule phenotype was also used to identify female larvae hemizygous for wapl that were generated by crossing w Df(1)JC105/Binsn females to either w wapl^A17/Dp(1;Y)w⁺303 or w wapl^HC262/Dp(1;Y)w⁺303 males. Mutant/Df females were selected on the basis of their colorless Malpighian tubules.

Mutant and control brains were dissected, fixed, and squashed in aceto-orcein according to our previously described procedures (GATTI and GOLDBERG 1991 ). To evaluate the frequency of anaphases and the mitotic index, larval brains were squashed without colchicine pretreatment or hypotonic shock. To estimate the mitotic index of wapl mutants relative to controls, we determined the average number of mitotic figures per optic field in brain squashes from FM7 (control), wapl^C204, and wapl^HC262 males. The optic field chosen for this analysis is the circular area defined by a phase-contrast, Neofluar x100 Zeiss objective, using x10 oculars with the Optovar set at x1.25.

Nondisjunction tests and calculations:
For simultaneous measurement of X and 4 nondisjunction in females, we crossed YSX·YL, In(1)EN, v f B/O; C(4)RM, ci ey^R/O males to females carrying suitable markers on their X chromosomes and homozygous for spa^pol. This assay allows recognition of X- and fourth-chromosome nondisjunctional offspring of all the crosses reported in Table 4. For example, for w wapl/cv v f car; spa^pol/spa^pol mothers, regular ova yielded B/+ females(/X; /4) and B⁺males(X/O; /4),whileX-nondisjunctional ova resulted in phenotypically wild-type females (X/X; /4) and B males (/O; (/4), and 4-nondisjunctional ova produced spa^pol/spa^pol (4/4) or ci ey^R (/O) male and female progeny. In a second example, regular ova from w wapl/FM7; spa^pol/spa^pol mothers resulted in either B/+ or B/B females (/X; /4) or y w^a B (X/O; /4) males, while the ex-exceptional-X progeny were w^a B females (X/X; /4) and v f B males (XY/O; /4), and the exceptional-4 progeny were spa^pol/spa^pol (4/4) and ci ey^R (44/O) individuals. For all the crosses reported in Table 4, the haplo-4 Minute progeny, which are poorly viable, were recorded but excluded from all calculations of nondisjunction frequencies.

The frequencies of X and 4 nondisjunction were calculated by dividing the sum of the nondisjunctional events of each class by the adjusted total of all progeny classes. To determine the numerator for the calculation of X nondisjunction, the exceptional-X progeny were doubled to correct for the inviability of triplo-X and nullo-X individuals. The denominator for X nondisjunction in crosses involving either w wapl⁺/cv v f car or w wapl⁺/FM7 mothers was calculated by doubling the exceptional-X progeny. For all other crosses, due to the inviability of wapl-bearing hemizygous males, we multiplied by ⁴/₃ the number of both regular and 4-nondisjunctional progeny and also doubled the exceptional-X progeny. The numerator for fourth chromosome nondisjunction was calculated by adding the fourth-chromosome exceptions that displayed regular X chromosome segregation to the double of simultaneous X, 4 exceptions. The denominators for fourth chromosome nondisjunction were the same as for X nondisjunction. (It was not necessary to double the fourth chromosome exceptions because, although one-half of these exceptions were inviable, one-half of the regular fourth ova also yielded progeny that were not counted.)

Construction of a wapl-bearing attached-X chromosome:
To construct a wapl-bearing attached-X chromosome we used the T(X;Y)G25 translocation, which has breakpoints in the X heterochromatin and in region h11 of the Y chromosome, and which carries the y w f markers on XL and the y⁺ marker appended to YS (KENNISON 1981 ). By recombination, we generated X^DY^P elements of this translocation carrying y wapl^11P3 or y pn on XL. We then used these marked elements to construct y wapl^11P3·y⁺/y pn·y⁺ females, which were irradiated with X rays (4000 rad) and mated to bb²/B^SY males. This cross yielded two yellow females, both of which generated only matroclinous daughters when backcrossed to bb²/B^SY males, indicating that these yellow females in fact carried compound X chromosomes. One of these compounds was chosen for further experiments and was characterized cytologically by Hoechst 33258 banding. This analysis demonstrated that it is a reverse metacentric in which the two X chromosomes are connected by a small amount of Y chromosome material that includes the Y centromere (data not shown). We named this compound chromosome C(1)RMV (Compound (1) Reverse Metacentric of Vernì).

The C(1)RMV/B^SY stock chosen for nondisjunction analysis is simultaneously heterozygous for wapl and pn (wapl pn⁺/wapl⁺ pn). It was therefore selected at each generation to eliminate wapl⁺ pn/wapl⁺ pn recombinant flies. Although we could not eliminate wapl⁺ pn/wapl⁺ pn⁺ recombinants (because they cannot be distinguished from the desired wapl pn⁺/wapl⁺ pn heterozygotes), this latter category of recombinants should be extremely rare, because wapl and pn are only 20 kb apart (see below, Fig 2). To examine the effects of wapl on Y disjunction, we rapidly expanded a line from a single C(1)RMV/B^SY (wapl pn⁺/wapl⁺ pn) female. To ensure that this founder female was in fact wapl pn⁺/wapl⁺ pn and not wapl⁺ pn/wapl⁺ pn⁺, we examined the brains of several descendant female larvae and found that some of them exhibited the wapl cytological phenotype. These mutant brains must have carried a wapl pn⁺/wapl pn⁺ recombinant compound, indicating that the founder female was indeed heterozygous for wapl. From this original C(1)RMV/B^SY (wapl pn⁺/wapl⁺pn) line we isolated and expanded a subline from a single wapl⁺ pn/wapl⁺pn recombinant derivative that served as a control.

View larger version (136K): In this window In a new window Download PPT slide   Figure 1. The cytological phenotype of wapl larval brain cells. (A–C) Oregon-R controls. (D–F) wapl mutant mitotic figures. (A and D) Colchicine-treated metaphases from Oregon-R (A) and wapl (D) males. Note the wapl-induced parallel sister chromatid arrangement and the separation of SCs in the heterochromatic regions. The latter phenomenon is particularly evident in the Y and fourth chromosomes; for example, in wild-type cells the fourth chromosome is visible as just one dot, but resolves into two distinct dots (arrows) in the wapl mutant. The separation of SCs is not complete as they remain connected by a thread located in the centromeric region (arrowheads). (B and E) Non-colchicine-treated metaphases from Oregon-R (B) and wapl (E) males. Note the complete separation of the fourth chromosome SCs (arrows) and the partial separation of the Y SCs. (C and F) Anaphases from non-colchicine-treated preparations of Oregon-R (C) and wapl males (F). Note that the wild-type and mutant mitotic figures are indistinguishable.

View larger version (16K): In this window In a new window Download PPT slide   Figure 2. The wapl gene region. (A) A genetic and molecular map of the wapl gene region. Left to right on this figure corresponds to the telomere-to-centromere direction along the Drosophila X chromosome. The lines labeled Df(1)JC105 and Df(1)64c18 show the sequences removed by two deletions that delimit the location of the wapl gene to the region -11 to +13 on the molecular map below. The zero coordinate on this map is defined by an EcoRI site just upstream of the CYP4D1 transcript (see GANDHI et al. 1992 ); the distances on the map are measured in kilobases. B, BamHI; G, BglII; R, EcoRI; S, SalI; X, XhoI. Transcriptional units (described in the text) are shown proceeding in the 5' to 3' direction; introns of <500 bp are not shown for simplicity. The black bars at the bottom indicate the extent of genomic sequences included in several transgenes described in the text; the gray shaded area in construct GC does not depict genomic DNA, but instead represents sequences obtained from the C-terminal-encoding portion of a cDNA corresponding to the 6.5-kb wapl transcript. (B) A close-up view of the region from coordinates -11 to -2 on the map in part (A), focusing on the intron/exon structure of the two wapl transcripts. The smaller 5.0-kb transcript (bottom) is truncated at the 3' end relative to the 6.5-kb mRNA (top). Based on available cDNA clones, it appears that the 5' end of the 5.0-kb RNA is transcribed from sequences within the second intron of the 6.5-kb species. Near its 5' end, the smaller transcript has a stop codon in the same frame as that encoding WAPL in the 6.5-kb mRNA.

PEV analysis:
To measure the extent of variegation in In(1)w^m4- and bw^D-bearing flies of various genotypes, we determined the level of red eye pigment according to the method used by REUTER et al. 1982 . Five-day-old flies were frozen in liquid nitrogen and decapitated by rapid shaking; at least five samples of 20 heads were prepared for each genotype. To extract the pigment, each sample was incubated in 1 ml of 30% ethanol acidified to pH 2 for 24 hr at 25° in the dark. These samples were then centrifuged for 15 min at 6000 x g, and the optical density of the supernatants was measured at 480 nm. The level of red pigment present in each genotype is expressed as the mean percentage of wild-type (Oregon-R) pigment ±SE.

The extent of Sb variegation in T(2;3)Sb^V-bearing animals was determined as described by SINCLAIR et al. 1983 . We examined the seven pairs of major dorsal bristles (posterior supraalars, anterior postalars, posterior dorsocentrals, anterior and posterior scutellars, and anterior and posterior sternopleurals) and assigned to each bristle a Sb or Sb⁺ phenotype. The extent of Sb variegation was expressed as the percentage of Sb bristles among the total bristles examined. Since the full expression of Sb produces short bristles, suppression of Sb variegation results in an increase of the frequency of mutant bristles.

Characterization of genomic DNA:
Preparation of DNA from recombinant cosmids, phage, and plasmids, as well as techniques for colony screening, Southern blotting, and nucleic acid hybridization have been previously described (GUNARATNE et al. 1986 ; MANSUKHANI et al. 1988 ) or were based on standard procedures (SAMBROOK et al. 1989 ). Phage containing overlapping segments of DNA from the 2D region of the Drosophila genome were previously isolated in a chromosomal walk through this interval (DURA et al. 1987 ) and were kindly provided by Dr. Hugh Brock (University of British Columbia, Vancouver, BC, Canada). Cosmids containing longer contiguous segments from the 2D region were isolated by screening a Drosophila cosmid library (J. Tamkun, University of California, Santa Cruz, CA) with probes from the genomic phage clones.

Analysis of whole genome Southern blots suggested that the wapl^C204 allele was associated with a DNA insertion in the wapl locus. To isolate this polymorphic DNA segment, genomic DNA isolated from wapl^C204/FM7 females was treated with BamHI and cloned into the BamHI site in the Lambda ZapII vector (Stratagene, La Jolla, CA) to generate a library of genomic clones. Sequences containing the wapl locus were isolated from this library using plaque hybridization. Plasmids containing Drosophila DNA inserts were obtained from recombinant phage using an in vivo excision protocol detailed in the Lambda ZapII instruction manual. Plasmids of interest containing the wapl^C204 insertion were selected on the basis of a restriction enzyme digestion pattern showing the presence of 1.6 kb of exogenous sequences in the wapl gene region.

Characterization of transcripts and corresponding cDNAs:
Preparation of poly(A)⁺ RNA from tissues of Oregon-R flies at various developmental stages has been previously described (GUNARATNE et al. 1986 ). A total of 10 µg of each poly(A)⁺ RNA sample was glyoxalated in a total volume of 25 µl containing 4 µl of deionized glyoxal and 12 µl of dimethylsulfoxide in 10 mM sodium phosphate buffer, pH 6.5. The RNA was then fractionated on 1% agarose gels run in the same sodium phosphate buffer. The RNA was subsequently transferred to Genescreen hybridization membrane (NEN Life Science Products, Boston, MA) as previously described (GUNARATNE et al. 1986 ; MANSUKHANI et al. 1988 ).

To obtain cDNA clones corresponding to transcriptional units in the wapl region, labeled fragments of genomic DNA were used to screen embryonic and imaginal disk cDNA libraries in the plasmid vector pNB40 (BROWN and KAFATOS 1988 ), kindly provided by Nicholas Brown (Cambridge University, UK). cDNAs with wapl inserts were sequenced using a T7 polymerase-based sequencing system (Amersham Pharmacia Biotech, Piscataway, NJ) according to the manufacturer's recommendations.

P-element-mediated germline transformations:
Three different plasmids were prepared for P-element-mediated germline transformation. The first of these consists of the BamHI genomic restriction fragment between coordinates +6 and -6 in Fig 2 (fragment A) inserted into the BamHI site of the pW8 vector (KLEMENZ et al. 1987 ). The second plasmid contains the BglII genomic restriction fragment between coordinates +0.5 and -8 in Fig 2 (fragment B) inserted into the BamHI site of pW8. The third plasmid, named construct GC in Fig 2, is a hybrid containing both genomic and cDNA sequences from the wapl transcriptional unit. This was prepared by inserting both fragment B described above and a BglII-NotI fragment obtained from a cDNA clone corresponding to the larger wapl 6.5-kb transcript into the pW8 vector that had been digested with NotI and BamHI in a one-step ligation.

Germline transformants were obtained by microinjection of the three constructs described above into w; Sb e 2-3/TM6 embryos as described by SIMON et al. 1985 . For all of the above constructs, several independent transformants were obtained. Autosomal insertions were selected on the basis that males bearing the transposon yielded male and female progeny with pigmented eyes when mated with z¹w^11e4 females. Insertions were placed into stock by subsequent matings with w; CyO; TM2/T(2:3)ap^Xa as previously described (WILLIAMS et al. 1992 ).

To check whether DNA within any of the three constructs above would rescue the lethality associated with mutations in wapl or in the nearby essential gene l(1)2Dg defined by the mutation l(1)90 (see above), z¹w^11e4 males carrying autosomal transpositions of the three constructs were mated with wapl^C204/FM7 females, with wapl^HC262/FM7 females, or with l(1)90/FM7 females. The presence of non-Bar male progeny indicates rescue of lethality by the transposon; in all successful experiments using construct GC, the numbers of Bar and non-Bar male progeny were approximately the same.

	RESULTS

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

Mutations in wapl affect mitotic chromosome structure:
In this study we have analyzed four lethal wapl alleles: wapl^C204, wapl^HC262, wapl^11P3, and wapl^A17. Animals homozygous or hemizygous for the first three of these alleles die at the larval-pupal boundary and exhibit small imaginal discs; wapl^A17 causes lethality during the early pupal stage but has apparently normal discs (PERRIMON et al. 1985 ). When examined in germline clones, the three small-disc wapl alleles exhibit a maternal effect lethal phenotype that is not seen in wapl^A17. Mutant embryos usually arrest prior to blastoderm formation but manage to form small patches of cuticle (PERRIMON et al. 1985 ; R. GANDHI and M. L. GOLDBERG, unpublished observations).

A preliminary examination of wapl mutant larval brain cells suggested that mitotic chromosomes are morphologically abnormal (PERRIMON et al. 1985 ). We reexamined the mitotic chromosomes of these mutants in squashed preparations obtained from larval brains incubated with the microtubule poison colchicine to induce mitotic arrest. The cytological analysis of these preparations revealed that all wapl mutants with the exception of wapl^A17 exhibit a characteristic defect in heterochromatin organization. In Drosophila melanogaster the entire Y chromosome, the proximal 40% of the X chromosome, the centric 25% of chromosomes 2 and 3, and most of the fourth chromosome are composed of constitutive heterochromatin (reviewed by GATTI and PIMPINELLI 1992 ). Due to the tight adhesion of sister chromatids in heterochromatic regions, chromosomes in wild-type metaphase figures have distinctive morphologies. The major autosomes and the X chromosome appear X and V shaped, respectively; the Y chromosome has a rod-like structure with the sister chromatids closely apposed along the entire chromosome; and the fourth chromosomes are visible as dots in which the two sister chromatids cannot be discerned (Fig 1). In contrast, the vast majority of metaphases in wapl^C204, wapl^HC262, and wapl^11P3 mutant brains display a very different chromosome morphology. The sister chromatids of the major autosomes and of the X chromosome are aligned parallel to each other, instead of assuming the X- and V-shaped arrangements seen in wild-type cells. Moreover, contrary to the wild-type situation, the Y and the fourth chromosomes in wapl mutant metaphase figures are resolved into two distinct sister chromatids. The sister chromatids of all wapl mutant chromosomes are not completely separated but are instead connected by a thread located in the centromeric region (Fig 1 and Table 1). As discussed below, we believe that the morphological abnormalities in wapl chromosomes reflect the same primary defect: a loosening of sister chromatid apposition in the heterochromatic regions.

The parallel sister chromatid phenotype appears to characterize the null state of the wapl gene, as similar effects are seen in hemizygotes and homozygotes for the strong wapl^HC262 allele and in heterozygotes for the same allele and a wapl^- deficiency (Table 1). However, this cytological phenotype is not common to all wapl mutants. Although the wapl^A17 allele is lethal in combination with all other wapl mutations, the mitotic chromosomes of wapl^A17 mutant males and wapl^A17/Df females appear completely normal.

Additional observations provide further insight into the cytological phenotype elicited by wapl mutations in larval brain cells. First, the phenotype is not an artifact of colchicine treatment, because it is also clearly seen in untreated cells that received only a short hypotonic shock (Fig 1). Second, an examination of hundreds of aberrant wapl metaphases showed no evidence of abnormal chromosome condensation, chromosome breakage, aneuploidy, or polyploidy. Third, the mitotic index, a measure of the number of cells undergoing mitosis, and the frequency of anaphases are similar in wild-type and wapl brains (Fig 1 and Table 2). These last two points suggest that sister chromatid adhesion in heterochromatin is not essential for chromosome stability or segregation, or for normal mitotic progression.

The observation that wapl affects sister chromatid adhesion in larval brain cell heterochromatin prompted us to examine the chromocenter of polytene chromosomes from wapl hemizygous males and homozygous females. The chromocenter results from the fusion of the heterochromatic regions during the process of polytenization and contains two types of heterochromatin, designated as - and ß-heterochromatin. -Heterochromatin appears as a compact mass located in the middle of the chromocenter and includes the bulk of the mitotic heterochromatin, which is severely underpolytenized relative to euchromatin. The rest of the chromocenter is composed of ß-heterochromatin, a diffusely banded mesh-like material that connects the euchromatic chromosome arms to -heterochromatin (reviewed by MIKLOS and COTSELL 1990 ). When we examined lacto-acetic orcein squashes of wapl^HC262 polytene chromosomes, the chromocenters appeared normal (data not shown). Thus, wild-type wapl activity is not required for chromocenter assembly during salivary gland chromosome polytenization.

The role of wapl in female meiosis:
It has been suggested that the segregation of achiasmate homologs during the first meiotic division of Drosophila females depends on pairing between homologous heterochromatic regions (HAWLEY et al. 1992 ; DERNBURG et al. 1996B ; KARPEN et al. 1996 ). The observation that wapl is required for maintaining sister chromatid adhesion in heterochromatic regions suggested the possibility that this locus may also be involved in heterochromatin pairing during female meiosis. Since the available wapl alleles all cause lethality in homozygotes (all the semilethal wapl alleles have been lost), we tested this hypothesis by examining X chromosome nondisjunction in wapl/wapl⁺ heterozygous females. Thus, w wapl/FM7, y w^a B and w wapl/w females were crossed to Oregon-R males and their progeny were examined for the presence of diplo-X and nullo-X exceptions. As shown in Table 3, females heterozygous for wapl^HC262, wapl^11P3, wapl^A17, or wapl^C204 and the FM7 balancer all exhibit a significant increase in X chromosome nondisjunction with respect to w/FM7 control females. Moreover, females carrying the FM7 balancer and either Df(1)JC105 or Df(1)Pgd-kz [two deficiencies that delete wapl⁺; hereafter together abbreviated as Df(1)wapl] also show small but significant increases in X nondisjunction. In the offspring of wapl/FM7 or Df(1)wapl/FM7 females we never observed FM7/FM7 females, indicating that wapl dominantly increases nondisjunction only during the first meiotic division. However, when either wapl^C204 or wapl^HC262 is made heterozygous with an X chromosome of normal sequence, X nondisjunction remains at the control level. Since the FM7 multiply inverted chromosome virtually abolishes exchange on the X chromosome (HAWLEY et al. 1992 ), these results strongly suggest that wapl affects only the segregation of nonexchange homologs.

To assess whether wapl enhances nondisjunction of the fourth chromosomes, which are always achiasmate, we simultaneously measured the frequencies of meiotic nondisjunction of the X and fourth chromosomes (Table 4). wapl heterozygous females, homozygous for the fourth chromosome marker spa^pol, were crossed to v f B/O; , ci ey^R/O males, and the progeny were examined for both X and fourth meiotic nondisjunction (see MATERIALS AND METHODS). As shown in Table 4, females heterozygous for wapl, but carrying two structurally normal X chromosomes, do not display a significant increase in either X or fourth chromosome nondisjunction with respect to controls (although the overall frequency of nondisjunctional ova appears to be higher in wapl/+ females than in +/+ females). However, wapl/FM7 females produce significantly more X chromosome exceptions than control wapl⁺/FM7 females, but exhibit only a slight increase in fourth chromosome nondisjunction. Moreover, most of the fourth chromosome exceptions produced by wapl/FM7 females are X and fourth double nondisjunction events. Previous studies have demonstrated an excess of X-4 double exceptions in the progeny of recombination-defective meiotic mutants (BAKER and CARPENTER 1972 ), indicating that fourth chromosome missegregation can be a secondary effect of X nondisjunction. Thus, we conclude that wapl has no direct effect on fourth chromosome meiotic behavior and, instead, we believe that the low level of fourth chromosome nondisjunction is caused by the presence of nondisjoining X chromosomes.

Because the females utilized to assay the effect of wapl on nondisjunction in the presence of a normal sequence X chromosome were heterozygous for the w, cv, v, f, and car recessive markers (see Table 4 and MATERIALS AND METHODS), we could test whether wapl affects recombination in a dominant manner in the intervals defined by these genes. As shown in Table 5, examination of X/O males obtained from these crosses established that wapl has no effect on meiotic exchange. This result further supports the notion that wapl⁺ is specifically involved in meiotic disjunction of nonexchange chromosomes.

Finally, we asked whether wapl mutations disrupt the segregation of the Y chromosome from a compound X chromosome. In a wild-type genetic background, these heterologous chromosomes are faithfully segregated by the achiasmate system (GRELL 1976 ). To examine the effects of wapl mutations on Y disjunction, we constructed a reverse metacentric attached-X chromosome heterozygous for both wapl^11P3 and pn (wapl pn⁺/wapl⁺ pn; see MATERIALS AND METHODS). From this compound chromosome we isolated a wapl⁺ pn/wapl⁺ pn recombinant derivative that served as a control. We found that heterozygosity for wapl^11P3 increases missegregation of the Y from the attached-X chromosome to only a small (less than three times the control value) but significant extent (Table 6). Thus, taken together, our data indicate that wapl is primarily involved in homologous segregation of nonexchange X chromosomes. We do not know at present whether wapl also affects nondisjunction of nonexchange large autosomes.

wapl modulates position-effect variegation:
Because most mutations that affect heterochromatin structure modulate PEV (see Introduction), we asked whether wapl is also a PEV modifier. We therefore examined the dominant effects of wapl on three different variegating systems: In(1)white^mottled4 (w^m4), T(2;3) Stubble^Variegated (Sb^V), and brown^Dominant (bw^D).

In w^m4, the w⁺ gene, juxtaposed to the X heterochromatin, is inactivated in a mosaic manner (LINDSLEY and ZIMM 1992 ). We generated w^m4/wapl females by crossing either w wapl/Dp(1;Y)w⁺303 males to w^m4/w^m4 females or w^m4/Y males to w wapl/FM7a females. We performed these crosses with all four mutant wapl alleles described above, as well as with the two Df(1)wapl deficiencies that delete the gene. As shown in Table 7, regardless of the method of generation, all w wapl/w^m4 and w Df(1)wapl/w^m4 females displayed higher amounts of red pigment than that measured in control w/w^m4 females. Thus, wapl mutations act as dominant suppressors of PEV.

To confirm and extend these results, we tested whether wapl influences the Sb variegation associated with T(2;3)Sb^V (SINCLAIR et al. 1983 ). This translocation juxtaposes the Sb mutation and the centric heterochromatin of the second chromosome, resulting in mosaic flies with both Sb and normal bristles. The normal bristles result from transcriptional inactivation of the dominant neomorphic mutation Sb. We crossed T(2;3)Sb^V males to either w wapl/FM7a, w Df(1)wapl/FM7a, or w/FM7a females and counted the Sb and Sb⁺ bristles in the w wapl/+; T(2;3)Sb^V/+, w Df(1)wapl/+; T(2;3)Sb^V/+, and w/+; T(2;3)Sb^V/+ (control) female progeny (see MATERIALS AND METHODS). wapl^11P3, wapl^A17, Df(1)JC105, and Df(1)Pgd-kz all cause significant increases in the frequency of Sb bristles compared with the control. However, the wapl^C204 and wapl^HC262 mutant alleles, which suppress the white⁺ variegation of In(1)w^m4, do not significantly modify Sb variegation (Table 8). Despite this discrepancy, the fact that two wapl deficiencies and two of the four wapl mutant alleles tested suppress Sb variegation strongly suggests that a reduction of wapl⁺ activity relieves the silencing of Sb in T(2;3)Sb^V.

To obtain further insight into the role of wapl in PEV, we tested the effects of wapl mutations or deficiencies on bw^D. In the bw^D chromosome, about a megabase of 2L heterochromatin is inserted next to the brown (bw) gene in region 59E. This rearrangement, as well as other rearrangements that juxtapose bw to heterochromatin, result in a peculiar dominant effect leading to bw variegation in the eye (reviewed by HENIKOFF et al. 1993 ). To determine the effects of wapl on bw^D, we crossed w wapl/FM7a and w Df(1)wapl/FM7a females to bw^D/bw^D males and measured the amount of pigment in the resulting w wapl/+; bw^D/+ or w Df(1)wapl/+; bw^D/+ female progeny. As a control we used w/+; bw^D/+ females generated by crossing w/FM7a females to bw^D/bw^D males. The presence of wapl^C204, wapl^11P3, or either of the two wapl deficiencies significantly reduced the amount of eye pigment, while wapl^HC262 and wapl^A17 heterozygous females did not differ from the control (Table 9). Thus, these data taken as a whole surprisingly indicate that wapl mutations generally act as dominant enhancers of bw^D variegation, but as dominant suppressors of w^m4 and Sb^V variegation.

Molecular analysis of the wapl locus region:
Previous genetic studies placed the wapl gene within the 2D region of the Drosophila X chromosome, 0.03–0.08 map units from the Pgd locus (GVOZDEV et al. 1977 ; see Fig 2A). The wapl gene is uncovered by the small deletion Df(1)JC105 but not by the overlapping deletion Df(1)64c18 (PERRIMON et al. 1985 ; see Fig 2A). The corresponding part of the genome was first cloned by DURA et al. 1987 in a molecular walk conducted to isolate the polyhomeotic (ph) gene. These clones were kindly provided to us by Dr. Hugh Brock, and our preliminary analysis of the region surrounding wapl was previously reported in a different context (GANDHI et al. 1992 ). We mapped the breakpoints for the deletions Df(1)JC105 and Df(1)64c18 with the aid of whole genomic Southerns and in situ hybridization to polytene chromosomes (Fig 2A; data not shown). Assuming the wapl gene does not overlap the Pgd locus, these breakpoints localize the wapl gene to a subregion ~25 kb in length.

The interval between the Pgd and pn genes that includes wapl has recently been entirely sequenced by the European Drosophila Genome Project (EDGP). Fig 2A presents a map of the 10 genes either known from previous work (GUTIERREZ et al. 1989 ; TENG et al. 1991 ; GANDHI et al. 1992 ; FROLOV and ALATORTSEV 1994 ; DUNKOV et al. 1996 ) or inferred by EDGP to exist in this interval. Within the 25-kb subregion defined by the two deletions mentioned above, 5 candidate genes could potentially correspond to wapl. These include two members of the cytochrome P450 family [CYP4D1 (GANDHI et al. 1992 ) and CYP4D14 (previously EDGP open reading frame 152A3.2)], an open reading frame that could potentially encode a NADH-ubiquinone oxidoreductase subunit (152A3.7), and an open reading frame partially conserved in Caenorhabditis elegans that could encode a protein of unknown function (152A3.3). We present evidence below that wapl corresponds to the remaining candidate, a gene located in the region between coordinates -11 and -2.5 on Fig 2A that is transcribed into two poly(A)⁺ RNAs of 6.5 and 5 kb in length.

Identification of the wapl gene:
To better localize the wapl gene within the 25-kb region of interest, we searched for polymorphisms associated with mutant alleles of wapl, using whole genome Southern blot analysis (data not shown). The X-ray-induced, apparent null allele wapl^C204 displayed polymorphisms consistent with the insertion of ~1.6 kb of DNA into the 1.1-kb SalI restriction fragment between coordinates -6.9 and -8 in Fig 2A. This conclusion was verified by cloning and sequencing the mutant DNA (see below). We also observed polymorphisms in the same region associated with a different X-ray-induced, apparent null mutation (wapl^HC262), although we have not characterized these polymorphisms further. These data taken together suggest, but do not prove, that the wapl gene extends over the 1.1-kb SalI fragment, which is contained within the -11 to -2.5 transcriptional unit described above.

On the basis of this information, we synthesized three different plasmids that included DNA from this transcriptional unit inserted into a vector for P-element-mediated germline transformation of Drosophila (see MATERIALS AND METHODS). Two of these plasmids contain overlapping fragments of genomic DNA (fragments A and B in Fig 2A) comprising sequences from the N-terminal half of the putative wapl transcriptional unit as well as the entire adjacent CYP4D1 gene (GANDHI et al. 1992 ). The large size of the putative wapl transcription unit and a paucity of convenient restriction sites in the region hindered the cloning of genomic DNA containing the complete candidate wapl gene. To create a plasmid including all anticipated wapl coding sequences, we thus made construct GC, which consists of fragment B fused at a common BglII site to C-terminal-coding sequences obtained from a cDNA clone corresponding to the larger 6.5-kb putative wapl transcript (Fig 2A). Construct GC should in theory contain all instructions for synthesis of the protein product of the 6.5-kb RNA as well as the normal promoter and upstream regulatory sequences for this transcriptional unit.

Plasmids A, B, and GC were injected into Drosophila embryos to obtain germline transformants. We then tested the ability of the three transduced DNA fragments to rescue the lethality associated with mutations in wapl or in the nearby l(1)2Dg gene (which maps to the same genetic subregion; see FROLOV and ALATORTSEV 1994 ) by complementation (see MATERIALS AND METHODS). The constructs containing either genomic fragments A or B alone failed to rescue the lethality caused by wapl or l(1)2Dg mutations. These fragments of DNA contain the entire CYP4D1 gene, so CYP4D1 is unlikely to correspond to either lethal complementation group. Construct GC was able to rescue the lethality associated with mutations in wapl but not in l(1)2Dg. Construct GC also rescues the cytological phenotype described above seen in animals carrying mutant wapl alleles (data not shown). These findings allow us to conclude that the wapl gene does in fact correspond to the transcriptional unit located between coordinates -11 and -2.5 on the map in Fig 2A.

Characterization of wapl transcripts and the wapl protein product:
Probing Northern blots containing developmentally staged Drosophila poly(A)⁺ RNA with wapl gene sequences reveals two species, 6.5 and 5 kb long, that we propose to be alternatively spliced wapl mRNAs (Fig 3). These transcripts are detectable as discrete bands only in embryos. Smears of homologous RNA are also detectable at other times in development, particularly in adult females. Smearing is not the result of general RNA degradation in these samples, as control hybridizations are unaffected. All signals are relatively weak and are seen only after prolonged exposure of the autoradiograms, suggesting that wapl RNAs are low in abundance.

View larger version (83K): In this window In a new window Download PPT slide   Figure 3. Northern blot analysis of wapl transcripts. Poly(A)+ (10 µg) from the indicated Drosophila developmental stage was fractionated on each lane of a 1% agarose gel, transferred to Genescreen membranes, and hybridized with the labeled EcoRI fragment of cloned Drosophila DNA extending from coordinates -11 to -7 on the map in Fig 2. Identical results were obtained with the adjacent EcoRI fragment (coordinates -7 to -3) as a probe (not shown). E, 24-hr embryo collection; L1, L2, and L3, first, second, and third instar larvae, respectively; P1 and P2, early and late pupal stages, respectively; M, adult males; F, adult females. Discrete bands of 6.5 and 5.0 kb are observed only in the lane with embryonic mRNA, while smears above background are observed in all other lanes, particularly that from adult females. Controls showing that these samples contained both equivalent amounts of poly(A)+ RNA and also other undegraded RNA species up to 4.3 kb in length have been previously published (GANDHI et al. 1992 ; WILLIAMS et al. 1992 ). Similar results were obtained with several independent preparations of poly(A)+ RNA (data not shown).

We isolated clones containing cDNA inserts corresponding to both transcripts from an embryonic cDNA library, as well as a clone derived from the shorter, 5-kb wapl mRNA from a library made from imaginal disc tissue RNA (see MATERIALS AND METHODS). The Berkeley Drosophila Genome Project (BDGP) has also identified expressed sequence tags (ESTs) from the 5' ends of several embryonic cDNA clones corresponding to both wapl mRNAs; these are cataloged as clot 1327 in the BDGP database (URL: http://www.fruitfly.org/EST/). By comparing the DNA sequences of these cDNA clones with the DNA sequence of this region of the X chromosome, we have been able to reconstruct the structure of the wapl transcriptional unit (Fig 2B). The larger 6.5-kb mRNA is assembled from eight exons. Although the bulk of the 5.0-kb transcript is similar, sharing five exons with the 6.5-kb RNA, the 5' and 3' ends of these two RNA species are different (Fig 2B).

We determined the complete sequence of the longest of our embryonic cDNA inserts corresponding to the 6.5-kb wapl mRNA. The sequence contains an open reading frame predicting a protein 1741 amino acids long. The nucleotide sequence of this wapl cDNA and the amino acid sequence of the predicted protein product have been deposited in GenBank under the accession no. U40214. Three types of supportive evidence verify that the 6.5-kb cDNA contains an open reading frame of this length. First, we made constructs that in bacterial cells express wapl/glutathione S-transferase fusion proteins of sizes predicted by the sequence (data not shown). Second, coupled in vitro transcription and translation of this 6.5-kb cDNA synthesizes a protein of ~200 kD, the size predicted by this open reading frame (data not shown). Finally, the DNA insertion associated with the null allele wapl^C204 interrupts the predicted open reading frame and would be expected to produce a truncated product (see Fig 2A).

Even though the predicted wapl protein is quite large, computerized searches of databases with the deduced amino acid sequence have failed to reveal significant homologies to any protein sequence of known function. However, the protein is unquestionably conserved through evolution, as C. elegans, rats, and humans contain wapl homologs. The predicted protein does contain a long stretch of acidic amino acids (residues 439–457), a feature that has been observed in many different chromatin binding proteins (see DISCUSSION). The wapl protein also includes a putative ATP binding site near its C terminus.

The function of the shorter 5.0-kb transcript is currently unclear. Sequence analysis shows that the exon at the 5' end of the cDNAs for this RNA species contains a stop codon in the same frame as that encoding the 200-kD protein described above (Fig 2B). In addition, this 5' exon does not contain any ATG initiating codons. One possibility is that this shorter RNA encodes a protein starting from an initiation codon further downstream, producing an N-terminal truncated form of the 200-kD protein. However, we cannot exclude the possibility that the 5.0-kb RNA does not in fact encode a protein at all.

	DISCUSSION

TOP ABSTRACT MATERIALS AND METHODS RESULTS DISCUSSION LITERATURE CITED

The mitotic phenotype of wapl mutants:
The metaphase chromosomes in larval brain cells of animals homozygous or hemizygous for wapl^C204, wapl^HC262, and wapl^11P3 mutations exhibit a common cytological defect. In the heterochromatic regions of these chromosomes, sister chromatids are separated, in contrast to their tight apposition in wild type. These effects are most obvious in the unusual ability to resolve SCs of the mostly heterochromatic fourth and completely heterochromatic Y chromosomes (Fig 1). We believe that other aspects of the cytological phenotype are the indirect result of this primary defect. As an example, in colchicine-treated wapl autosomes, SCs lie parallel to each other in a rail-track fashion, so that in distal euchromatic regions SCs are usually closer to each other than in wild type (Fig 1). This observation is at first glance at odds with the well-known ability of colchicine to disrupt SC cohesion in euchromatic arms but not in heterochromatic regions (GONZALEZ et al. 1991 ; reviewed by MIYAZAKI and ORR-WEAVER 1994 ). However, this aberrant behavior of the distal euchromatin can be simply rationalized as the consequence of the wapl mutant-induced release of the forces that hold SCs together in the heterochromatin. To visualize this concept, one could imagine the two sister chromatids as two pieces of rubber tubing. In wild type, these tubes would be squeezed together in the middle (i.e., at the centromere), forcing their ends to diverge. However, when the squeezing forces are released, as would occur in wapl mutants, the tubes would assume a parallel configuration. We thus believe that wapl mutations specifically affect heterochromatin and have little or no direct effect on euchromatin.

Although wapl mutations disrupt SC associations in heterochromatic regions, including centromeric heterochromatin, it appears that the mutations do not affect the function of the centromere per se. In both colchicine-treated and untreated wapl chromosomes, SCs appear to be held together at the centromere. This is indicated by three observations. First, in most wapl chromosomes, SCs are not completely separated in the centromeric regions but remain physically connected by a chromatin thread (Fig 1). Second, we never observed wapl metaphase chromosomes with unaligned or staggered SCs. Third, the fact that wapl mutant brain cells display normal mitotic chromosome segregation argues strongly against the possibility of precocious centromere splitting in the absence of the wapl gene product (see below).

The effects of wapl mutations contrast with those associated with three different classes of mutations that have previously been described to generate aberrant or precocious sister chromatid separation (PSSC). (1) Mutations in Saccharomyces cerevisiae genes encoding SMC family proteins that are components of the cohesin complex (Smc1p, Smc3p, and Mcd1/Scc1) cause the complete precocious separation of chromatids, including both the arms and the centromeres (GUACCI et al. 1997 ; MICHAELIS et al. 1997 ; reviewed by BIGGINS and MURRAY 1998 ). These yeast mutations also disrupt chromosome condensation, in contrast to the absence of such effects associated with wapl. (2) Mutations in several organisms inappropriately produce PSSC during the first meiotic division, resulting in high levels of meiotic chromosome nondisjunction. These mutations, which include the mei-S332 and ord mutants of Drosophila, pc mutants of Lycopersicon esculentum, and desynaptic mutants of Zea mays, cause precocious and complete SC separation involving the splitting of both the centromere and the chromosome arms and result in massive meiotic chromosome missegregation (reviwed by MIYAZAKI and ORR-WEAVER 1994 ). It appears that these latter genes are mainly required for holding together sister centromeres during the first meiotic division. Recent data indicate that the protein product of mei-S332 localizes to mitotic as well as meiotic chromosomes, though null mei-S332 mutations have no effect on mitotic SC separation or segregation (MOORE et al. 1998 ). (3) Yet other mutations appear specifically to allow separation of sister centromeres but not of sister chromatids. In barren mutants of Drosophila, as well as in Schizosaccharomyces pombe Mis6 mutants, sister centromeres separate and migrate toward the spindle poles, but the sister chromatid arms remain associated, causing chromatin bridges (BHAT et al. 1996B ; SAITOH et al. 1997 ). The effects of wapl mutations, which primarily affect sister chromatid apposition in heterochromatic regions, but which produce neither precocious splitting of sister centromeres or chromatids nor detectable chromosome missegregation, are very different than those associated with any of the three classes of previously described mutations.

Interestingly, a cytological phenotype very similar to that elicited by wapl mutations is observed in Roberts syndrome (RS), a rare human genetic disorder with an autosomal recessive mode of inheritance. Roberts syndrome is characterized by pre- and postnatal growth retardation, craniofacial malformations, and tetraphocomelia. Mitotic cells from affected individuals exhibit a characteristic cytological phenotype consisting of the "repulsion" of constitutive heterochromatin so that the chromosomes display a "railroad-track appearance" (reviewed by VAN DEN BERG and FRANCKE 1993 ). Published photographs of RS colchicine-treated metaphases reveal a striking similarity with the wapl metaphase chromosomes shown in Fig 1. Despite these morphological similarities, RS and wapl mutants differ in one important aspect of their phenotype. Whereas wapl chromosomes with defective heterochromatin segregate normally, mitosis is abnormal in RS mutants with affected chromosomes. Although these RS chromosomes have morphologically regular kinetochores, they often exhibit defective anaphase segregation, eventually leading to chromosome loss and the formation of micronuclei (JABS et al. 1991 ; VAN DEN BERG and FRANCKE 1993 ). We have recently cloned the human homolog of wapl (P. SOMMA, F. VERNÌ and M. GATTI, unpublished results). Given the similarities as well as the differences in the RS and wapl phenotypes, it will be of interest to establish whether this human wapl gene is implicated in Roberts syndrome.

The absence of mitotic abnormalities in wapl mutants is somewhat at odds with observations made on other Drosophila mutants that affect mitotic heterochromatin structure (see Introduction). For example, mutations in the genes that encode the GAGA and PROD proteins needed for proper heterochromatin condensation cause mitotic chromosome missegregation (BHAT et al. 1996A ; TOROK et al. 1997 ). In addition, altered chromosome segregation leading to the formation of hyperploid and polyploid cells has been observed in several mutants that exhibit abnormal mitotic chromosome condensation (GATTI and BAKER 1989 ). It is therefore conceivable that alterations in pericentric heterochromatin condensation inevitably impair centromere function, leading to abnormal chromosome segregation. This generalization is not demolished in the case of wapl mutants, where heterochromatin is normally condensed despite the lack of SC cohesion in heterochromatic regions. The proper condensation of heterochromatin in wapl mutants could thus allow normal centromere function and regular chromosome segregation. These considerations imply that SC cohesion in pericentromeric regions is required neither for heterochromatin condensation nor for normal centromere behavior during mitotic cell divisions of Drosophila brain cells.

Although wapl mutant brain cells have normal mitotic parameters, the three mutants with abnormal metaphase chromosomes (wapl^C204, wapl^HC262, and wapl^11P3) also show small imaginal discs, whereas wapl^A17 affects neither chromosome structure nor disc morphology and dies later than the other mutants alleles (PERRIMON et al. 1985 ; F. VERNÌ, R. GANDHI, M. L. GOLDBERG and M. GATTI, unpublished data). However, wapl^A17 is as strong as the other mutants for the phenotypes of meiotic chromosome missegregation and PEV suppression. Combined with the finding that wapl^A17 hemizygotes do not exhibit SC separation in the heterochromatin, these observations suggest that the wapl^A17 mutation impairs a WAPL domain required for some of the protein's functions, but not for SC cohesion.

It is currently unclear why wapl^C204, wapl^HC262, and wapl^11P3 exhibit small imaginal discs, while the analysis of brain cells in these mutants did not reveal alterations in either the rate of cell division (the mitotic index) or the accuracy of mitotic chromosome segregation. We envisage two possibilities. Because of differences in the physiology of cell division between disc and brain cells, wapl mutations that alter heterochromatin organization might affect mitosis only in disc cells. Alternatively, the