Abstract

The DNA-binding protein encoded by the zeste gene of Drosophila activates transcription and mediates interchromosomal interactions such as transvection. The mutant protein encoded by the zeste1 (z1) allele retains the ability to support transvection, but represses white. Similar to transvection, repression requires Zeste-Zeste protein interactions and a second copy of white, either on the homologous chromosome or adjacent on the same chromosome. We characterized two pseudorevertants of z1 (z1-35 and z1-42) and another zeste mutation (z78c) that represses white. The z1 lesion alters a lysine residue located between the N-terminal DNA-binding domain and the C-terminal hydrophobic repeats involved in Zeste self-interactions. The z78c mutation alters a histidine near the site of the z1 lesion. Both z1 pseudorevertants retain the z1 lesion and alter different prolines in a proline-rich region located between the z1 lesion and the self-interaction domain. The pseudorevertants retain the ability to self-interact, but fail to repress white or support transvection at Ultrabithorax. To account for these observations and evidence indicating that Zeste affects gene expression through Polycomb group (Pc-G) protein complexes that epigenetically maintain chromatin states, we suggest that the regions affected by the z1, z78c, and pseudorevertant lesions mediate interactions between Zeste and the maintenance complexes.

THE zeste (z) gene is involved in two different phenomena that require interaction between the two copies of a single gene in the nucleus. zeste1 (z1)-type mutations repress transcription from paired white (w) genes, and za-type mutations block transvection, chromosome pairing-dependent complementation between specific mutant alleles of some genes. The z gene was discovered independently by the two different phenotypes.

Transvection was discovered as partial complementation between mutant Ultrabithorax (Ubx) alleles that depends on allelic pairing (Lewis 1954). Hypomorphic or null za-type mutations, similar to chromosomal rearrangements that disrupt allelic pairing, enhance the heteroallelic mutant phenotype (Lewis 1954; Kaufmanet al. 1973). z also supports transvection-like effects in decapentaplegic (dpp) (Gelbart 1982; Gelbart and Wu 1982), w (Babu and Bhat 1980; Gelbart and Wu 1982), yellow (Geyeret al. 1990), and eyes absent (Leisersonet al. 1994). Interallelic complementation at all of these genes occurs between one allele with a mutant coding region and another allele with a mutant regulatory region. The wild-type regulatory elements have been proposed to activate the wild-type coding region on the other chromosome when the two are brought together by pairing (Pirrottaet al. 1985; Zacharet al. 1985; Geyeret al. 1990). The interchromosomal activity of cis-regulatory sequences is particularly evident in the Cbx1 mutation of Ubx, because Cbx1, a rearrangement in the Ubx regulatory DNA, causes misexpression of both the cis and wild-type trans Ubx genes when the two are paired (Castelli-Gairet al. 1990). Activation of the wild-type homolog paired with Cbx1 absolutely requires Zeste (Goldsborough and Kornberg 1996).

In contrast, the neomorphic mutation z1 represses transcription of w whenever two or more copies of the w regulatory region are brought into proximity by homologous pairing or tandem duplication (Gans 1953; Green 1959; Judd 1961; Jack and Judd 1979; Bingham and Zachar 1985). The part of the w regulatory region required for interaction with z1 is ~1 kb upstream of the promoter (Leviset al. 1985; Pirrottaet al. 1985).

zeste encodes a sequence-specific DNA-binding protein with binding sites distributed throughout the genome (Benson and Pirrotta 1987; Pirrottaet al. 1987; Mansukhaniet al. 1988a). Zeste-binding sites are near the promoters and enhancers of w, Ubx, and dpp (Benson and Pirrotta 1987, 1988), and the protein activates transcription from the Ultrabithorax promoter both in vitro and in Drosophila embryos (Bigginet al. 1988; Laney and Biggin 1992).

Effects of Zeste on gene expression may be mediated through chromatin structure. Enhancers and suppressors of z1 are members of the Polycomb group of genes, the products of which form chromatin complexes that epigenetically maintain the silenced state of gene expression (Jones and Gelbart 1990; Phillips and Shearn 1990; Rastelliet al. 1993; Wu and Howe 1995; Carrington and Jones 1996). Indeed, Zeste colocalizes with the Polycomb group (Pc-G) proteins at many chromosomal sites (Rastelliet al. 1993). Furthermore, zeste null mutations strongly enhance position effect variegation (PEV), which is epigenetic repression of gene activity caused by heterochromatin-induced alterations in chromatin structure (Judd 1995). Moreover, one modifier of zeste also modifies both the phenotype of Pc and PEV (Larssonet al. 1996).

To gain a better understanding of how Zeste affects gene expression, we induced and characterized z78c, a weak z1-like allele, and two mutations of z1 that fail to repress white, but antagonize z1 repression of w+ to a greater extent than does z+. Both of the pseudorevertant proteins self-aggregate, and neither of the mutations affects the DNA-binding domain of the protein. However, neither supports transvection at Ultrabithorax or represses white, suggesting a mechanistic relationship between transvection and the repression of w+ by z1.

MATERIALS AND METHODS

Drosophila culture: Flies were cultured at 25° on a standard cornmeal, yeast, and molasses medium (Wirtz and Semey 1982).

Mutagenesis and isolation of the z78c, z1-35, and z1-42 mutations: For chemical mutagenesis, 3-day-old males were allowed to feed on 0.025 M ethyl methanesulfonate in a 1% sucrose solution for 24 hr. The males were then removed and mated. For X-ray mutagenesis, 3-day-old males were subjected to 3000R at the rate of 732R/min.

Effects of the z1-35 and z1-42 mutations on transvection at Ultrabithorax: To construct flies of the genotypes z1-35/Y; Ubx1 e/bx34e and z1-42/Y; Ubx1/bx34e, stocks were constructed of the genotypes z1-35/z1-35; Ubx1/TM3, Sb Ser and z1-42/z1-42; Ubx1/TM3, Sb Ser. Females from these stocks were mated to bx34e/bx34e males. z+/Y; Ubx1 e/bx34e control males were obtained by mating Ubx1/TM3, Sb Ser females to bx34e/bx34e males. Flies from the crosses were raised at 25°.

Sequencing of mutant zeste genes: Genomic DNA was prepared as described by Levis et al. (1982) from Oregon-R, z1, z1-35, z1-42, and z78c flies and two overlapping fragments of the zeste genes covering the entire coding sequences, which were amplified by polymerase chain reaction (PCR) (Mullis and Faloona 1987). A fragment containing exons 1 and 2 was amplified from each using primers containing nucleotides 5′-534-558-3′ and 5′-2006-1982-3′ (numbers refer to the zeste sequence in the GenBank database, accession number Y00049). At least three independent reactions were pooled for each genotype and cloned into pBluescript vectors (Stratagene, La Jolla, CA). Multiple clones with the same insert orientation were pooled before sequencing single-stranded templates by the dideoxy method (Sangeret al. 1977) with Sequenase 2.0 (United States Biochemical, Cleveland). In addition to the commercially available pBluescript primers and the amplification primers, primers containing nucleotides 5′-847-872-3′, 5′-1297-1322-3′, and 5′-1313-1288-3′ were used in sequencing exons 1 and 2. A fragment containing exon 3 was amplified using primers containing nucleotides 5′-1929-1953-3′ and 5′-2898-2876-3′. Five independent amplification reactions were pooled for each genotype, cloned into pBluescript, and sequenced as described for exons 1 and 2. Primers containing nucleotides 5′-2318-2339-3′ and 5′-2597-2575-3′ were used to sequence the middle of exon 3.

Zeste protein synthesis and aggregation: The ability of the various zeste proteins to aggregate in vitro was determined using an in vitro translation assay (Chenet al. 1992; Chen and Pirrotta 1993). To produce RNA templates for in vitro translation, the NaeI (nucleotide 2057) to NcoI (nucleotide 2888) fragments of the z1, z1-35, and z1-42 pBluescript clones were used to replace the corresponding fragment in the wild-type zeste gene in pET-3CZ (Chenet al. 1992; Chen and Pirrotta 1993, provided by V. Pirrotta). The z78c gene has an additional NaeI site at nucleotide 2434 because of a polymorphism, so the EspI (nucleotide 2125) to NcoI fragment was substituted for the corresponding fragment in pET-3CZ. The clones were sequenced to confirm that they contained the correct mutations. One z1-35 clone was found to contain a PCR-generated stop codon at nucleotide 2652, which produces a truncated protein lacking the C-terminal hydrophobic repeats. This clone was used as a control for a protein that does not aggregate.

Capped zeste transcripts were synthesized in vitro from the various linearized pET-3CZ templates with T7 RNA polymerase using Megascript reagents and protocols (Ambion, Austin, TX). The transcripts were translated in vitro with either rabbit reticulocyte lysates (Promega, Madison, WI) or wheat germ extracts (Promega) and [35S]methionine, according to the supplier's protocols. Translation reactions were either processed immediately or stored at 4° for the times indicated in the text. The translation reactions were subjected to centrifugation in an Eppendorf centrifuge for 10 min at 4°. The pellet was resuspended in 1× gel buffer (4 m urea, 100 mm Tris·HCl, pH 7.6, 2% sodium dodecyl sulfate, 5% β-mercaptoethanol, 5% Ficoll), and the supernatant was diluted 1:1 with 2× gel buffer. Aliquots containing equivalent amounts of supernatant and pellet were denatured at 95° for 3 min and subjected to SDS-polyacrylamide electrophoresis in a 7.5% gel in Trisglycerine buffer (25 mm Tris base, 250 mm glycine, 0.1% sodium dodecyl sulfate, pH 8.3). The gels were fixed in methanol-acetic acid, dried, and the amount of Zeste protein in each lane was determined by Phosphorimager analysis and visualized by autoradiography. Aliquots of the supernatant fractions were also subjected to gel filtration analysis with a Superose 6 column (Pharmacia, Piscataway, NJ) as described by Chen and Pirrotta (1993), and the amounts of various aggregated forms were quantitated by liquid scintillation counting.

RESULTS

Mutations of z that block the ability of the Z1 protein to repress w: Pseudorevertants of the z1 mutation z1-35 and z1-42 were isolated in a screen for mutations that reverse the repression of white by z1. z1 dominantly represses white when one of the white genes contains a tandem duplication of the regulatory region, such as Dp(1;1)wrdp+. Thus, z1 wrdp+/z+ w+ females exhibit a red-orange eye color intermediate between the dark red color of wild type and the yellow color of hemizygous z1 wrdp+ males (Figure 1). z1 males were mutagenized with ethyl methane sulfonate (EMS) and mated to z1 wrdp+ females. The F1 female progeny were scored for eye colors darker than the red-orange color of z1 wrdp+/z+ w+ females. In addition to revertants of z1, this screen would detect suppressors of z1 and mutations of w+ resistant to repression by z1, even in the presence of two doses of the w regulatory region. z1-35 and z1-42 were isolated in a screen of ~200,000 female progeny. A translocation of w+ to an autosome and an autosomal suppressor of z1 were also identified, but were not characterized further. No w mutations were recovered.

Figure 1.

Eye color phenotypes of z1 pseudorevertants. Clockwise from upper left: z1 wrdp+/Y; z1 wrdp+/+; z1 wrdp+/z1-42; z 1 wrdp+/z1-35.

The z1-35 and z1-42 mutations were shown to be z alleles by recombination experiments. Initially, the mutations were mapped by recombination with a z1 wBwx spl chromosome. This mapping indicated that the mutations are distal to w at a position near z. However, z1-35 and z1-42 cannot be true reversions of z1 because their phenotypes are different from z+. Both counteract the repressive effects of z1 more effectively than z+. The eye colors of both z1 wrdp+/z1-35 w+ and z1 wrdp+/z1-42 w+ females are darker than z1 wrdp+/z+ w+ females (Figure 1). Therefore, an attempt was made to separate the 1-35 mutation from z1. y z1 wBwx/y+ z1-35 w+ females were mated to z1 w65a25 spl sn3/Y males. Separation of the 1-35 mutation from the original z1 allele would produce yellow-eyed females. No recombination between z1 and z1-35 was observed in 153,000 chromosomes tested, indicating that the 1-35 lesion is very close to the original z1 lesion, certainly within the z gene.

Induction of a new z1-type allele: Only a small number of z1-type mutations have been available for analysis. In addition to z1, z58g is a spontaneous mutation (Kaufmanet al. 1973), and zp69a is EMS-induced (Gelbart 1971). We set out to induce new z mutations that would repress w+. z+ males were mutagenized with either X-rays or EMS and mated to za/za females. A z1-type mutation heterozygous with za (z* w+/za w+) would reduce eye color. No mutations were detected among approximately 100,000 female F1 progeny of X-irradiated males. However, a single mutation, z78c, that caused a decrease in eye color was recovered from ~100,000 female progeny of EMS-treated males. The eyes of homozygous z78c flies are red-orange, indicating that z78c represses w+ more weakly than z1. Similarly to z1, z78c does not affect eye color in males, indicating that it also requires multiple copies of w+ to repress. With three copies of w+, z78c homozygotes display the same eye color as z1 homozygotes do with two copies of w+ (data not shown).

z1-35 and z1-42 mutations block transvection at Ultrabithorax: z gene activity is required for transvection at a number of loci, including Ubx. The z1 mutation, however, supports transvection at Ubx. We tested the z1-35 and z1-42 pseudorevertants to determine whether they retain the ability to support transvection between the bx34e and Ubx1 alleles of Ultrabithorax. Ubx1/bx34e flies exhibit a weak transformation of metathorax to mesothorax. Indicative of a transformation of halteres to wings, the halteres are enlarged and flattened and have a row of bristles along the anterior edge. The Ubx1/bx34e flies lack any dorsal mesothoracic tissue on the dorsal metathorax. The z1-35 and z1-42 mutations enhance the Ubx1/bx34e phenotype substantially. The halteres of both z1-35/Y; Ubx1/bx34e males and z1-42/Y; Ubx1/bx34e males are larger than those of males z+/Y; Ubx1/bx34e. More strikingly, unlike z+/Y; Ubx1/bx34e, both z1-35/Y; Ubx1/bx34e males and z1-42/Y; Ubx1/bx34e males have dorsal mesothoracic tissue on their metathoraces (Figure 2). z1-42, which is a weaker antagonist of z1, causes the weaker transformation of the dorsal metathorax. z1-42/Y; Ubx1/bx34e males have a small scutellum and dorsal notum on the metathorax (Figure 2C). z1-35/Y; Ubx1/bx34e males have a larger area of mesothoracic notum on the metathorax, and the transformed scutellum is nearly twice as large as that of z1-42/Y; Ubx1/bx34e males (Figure 2B). Therefore, both the z1-35 and z1-42 mutations fail to support transvection between Ubx1 and bx34e, and the relative expressivities of the mutations for antagonizing z1 are similar to that for blocking transvection. We conclude that the activity of Zeste1 required for repression of w is also required for transvection.

Sequence of the z1-35, z1-42, and z78c mutant genes: Wild-type zeste from Drosophila melanogaster (Pirrottaet al. 1987; Mansukhaniet al. 1988b) and D. virilis (Chenet al. 1992), as well as several mutant D. melanogaster alleles (Pirrottaet al. 1987) have been sequenced. In combination with directed mutagenesis, this has allowed the functions of various domains of the Zeste protein to be examined (Figure 3). The DNA-binding domain is at the N terminus (Mansukhaniet al. 1988a), and the hydrophobic repeats at the C terminus are required for aggregation of the Zeste protein, transvection, and repression of white (Bickel and Pirrotta 1990; Chenet al. 1992). As described above, the z1-35 and z1-42 pseudorevertants of z1 counteract white repression by z1 more effectively than z+ and do not support transvection at Ultrabithorax. Based on these observations, one might predict that the mutations alter the hydrophobic repeats involved in aggregation. Because the effects of the z78c mutation on white expression are similar to z1, it might be expected to affect the same domain as z1.

Figure 2.

z1-35 and z1-42 mutations do not support transvection at Ultrabithorax. The two zeste mutations enhance the phenotypes of the Ubx1/bx34e mutants, causing the formation of obvious mesothoracic tissue with bristles and a scutellum in the dorsal mesothorax. Mesothorax and metathorax of (A) z+/Y; Ubx1 e/bx34e (B) z1-35/Y; Ubx1 e/bx34e, and (C) z1-42/Y; Ubx1/bx34e males are shown. Arrows in B and C indicate dorsal mesothoracic tissue formed in the metathorax.

We tested these predictions by sequencing the three mutant genes isolated by PCR amplification from genomic DNA. All three mutations were isolated in screens using EMS as a mutagen and are, therefore, likely to be point mutations. To avoid mistakes in identification of the lesions because of introduction of point mutations by PCR, DNA from multiple independent amplification reactions was pooled and cloned into a bacterial plasmid vector, and multiple clones for each mutant gene were pooled before sequencing. We also determined the sequences of both the z1 and the Oregon-R wild-type gene as controls. In all cases, both strands were sequenced multiple times with primers that gave overlaps in the regions sequenced.

The sequence of the z1 mutant gene was identical to that determined previously, including the lesion that causes the critical K425M substitution in the protein (Pirrottaet al. 1987; Table 1). As expected, both the z1-35 and z1-42 mutant genes contained both the critical K425M mutation and most of the polymorphisms found in z1 (Table 1; Figure 3). Each of the pseudorevertant genes differed from the z1 sequence by a single amino acid change (Table 1; Figure 3). In the case of z1-35, a C to T change at position 2407 in the nucleotide sequence leads to a P441L change in the predicted protein sequence. The z1-42 mutant gene has a similar mutation that predicts a P433S change in the protein sequence. Thus, both pseudorevertants alter proline residues just C-terminal to the K425M z1 lesion. Besides the proline residues at 441 and 433 altered by the z1-35 and z1-42 mutations, residues 450, 454, and 456 are also prolines. Thus, the pseudorevertants affect the two N-terminal prolines in a cluster of prolines.

The z78c mutant gene contains polymorphisms that more closely match the wild Oregon-R sequence than the previously determined wild-type sequence in GenBank (Table 1). The only nucleotide change that introduces an amino acid not found in any of the wild-type zeste genes is a C to T change at position 2283 in the nucleotide sequence that predicts a H400Y change in Zeste (Table 1; Figure 3). This is between two glutaminealanine-rich stretches (opa repeats) found in several Drosophila proteins (Whartonet al. 1985) and thought to form α-helices. The histidine residue affected by the z78c lesion is at the N terminus of the last opa repeat, and the lysine residue altered by the z1 mutation is at the C terminus of the same repeat (Figure 3).

Figure 3.

The Zeste protein. The DNA-binding domain (DNA) near the N terminus (N) is shown as a black box. Opa repeats are indicated by light gray boxes, and hydrophobic repeats (H) involved in aggregation are indicated by dark gray boxes. The proline residues in the cluster (P) are shown as vertical lines. The sites of the critical mutations in various zeste alleles (z78c, z1, z1-42, z1-35, zop6, and z11G3) are indicated below. All except z78c also contain the lesion in z1.

View this table:
TABLE 1

Nucleotide and amino acid differences between zeste alleles

Aggregation of the Z1-35, Z1-42, and Z78c mutant proteins: Z1 protein has been proposed to repress white, because its ability to aggregate is greater than wild-type Zeste protein (Chen and Pirrotta 1993). The protein encoded by the z11G3 pseudorevertant of z1 has a deletion of the Y510 residue in one of the C-terminal hydrophobic repeats, and its ability to aggregate is less than wild-type Zeste, supporting the idea that enhanced aggregation contributes to z1 repression (Bickel and Pirrotta 1990).

The z1-35 and z1-42 mutations also block the repression phenotype of z1 but more effectively than wild-type zeste (Figure 1). If increased aggregation of Zeste is the primary factor responsible for repression of white, then the Z1-35 and Z1-42 proteins would be expected to have lost the ability to aggregate. However, the z1-35 and z1-42 lesions do not affect the hydrophobic repeats involved in aggregation (Table 1; Figure 3). Rather, the z1-35 and z1-42 lesions are similar to those in the zop6 allele, a z1 derivative that more strongly represses white, blocking the expression of a single unpaired w+ (Lifschytz and Green 1984). Besides the original z1 K425M lesion, zop6 contains a P454L mutation (Pirrottaet al. 1987) similar to the z1-35 P441L lesion.

To understand the effects of the z1-35 and z1-42 mutant proteins on white expression, their ability to aggregate in vitro was determined using assays developed previously (Chenet al. 1992; Chen and Pirrotta 1993). The mutant proteins were synthesized by in vitro translation in rabbit reticulocyte lysate and labeled with [35S]methionine. After translation, aggregates were sedimented in a microfuge, and aliquots of the supernatants and pellets were subjected to electrophoresis to determine the extent of aggregation. Subtle differences in the abilities of the wild-type and z1 proteins to aggregate cannot be detected with all rabbit reticulocyte preparations (Chen and Pirrotta 1993). Therefore, the in vitro translations were performed with multiple batches of commercial rabbit reticulocyte lysate. With most batches, more than 90% of the wild-type, z1, z1-42, z1-35, and z78c mutant proteins were found in the pellet, and less than 10% of the full-length transcription products remained in the supernatant (data not shown). However, with a mutant protein lacking the hydrophobic repeats, more than 90% of the protein was in the supernatant, indicating that aggregation requires the hydrophobic repeats. Based on these results, we concluded that the z1-42, z1-35, and z78c mutant proteins have not lost the ability to aggregate through the hydrophobic repeats.

Figure 4.

In vitro aggregation of wild-type and mutant Zeste proteins. The panels are autoradiograms of an SDS-PAGE gels. Aggregation of Zeste protein was determined by centrifugation of in vitro translation reactions to obtain pellet (P) and supernatant (S) fractions. In the top panel, sedimentation was conducted immediately after termination of the translation reactions, and in the bottom panel, after incubation overnight at 4°. The proteins were translated from z+ (+), z1 (1), z1-42 (1-42), z1-35 (1-35), and z78c (78) RNA templates using a particular lot of reticulocyte lysate that gives lower levels of aggregation than most lysates, allowing subtle aggregation differences to be detected. Equivalent amounts of supernatant and pellet fractions were loaded in all lanes, and the amount of radioactively labeled full-length Zeste protein (arrows) in each was quantitated with a phosphorimager. The percent of in vitro translated protein that forms sedimentable aggregates after incubation at 4° for various numbers of days is shown in the bar graphs to the right. Each bar represents averages of independent experiments, and the error bars indicate the range of experimental values.

With one particular batch of reticulocyte lysate, more subtle differences in the aggregation properties between different Zeste proteins were observed. With this lysate, ~10% of the wild-type Zeste protein formed sedimentable aggregates compared to 40–50% of the z1 mutant protein (Figure 4). With this extract, ~30% of the z78c protein formed sedimentable aggregates. Because z78c has effects on white similar to z1, this observation is consistent with the idea that increased aggregation contributes to white repression. The differences in aggregation were reproduced in multiple experiments.

The aggregation properties of the z1-35 and z1-42 mutants are similar to wild-type Zeste protein if the sedimentation is performed immediately after the in vitro translation reactions (Figure 4), but aggregation increases to z1 and z78c levels upon incubation at 4° overnight. Incubation at 4° does not alter the aggregation of the wild-type, z1 or z78c proteins, and incubation of the z1-35 and z1-42 proteins up to 13 days does not significantly increase aggregation over the levels obtained with an overnight incubation (Figure 4). Thus, the z1-35 and z1-42 proteins do not display lower aggregation levels than wild type and can aggregate as much as the z1 protein, although at a slower rate.

Chen and Pirrotta (1993) observed that after removing large Zeste aggregates from in vitro translation reactions by sedimentation, the supernatant contains smaller aggregates that are separable by gel filtration chromatography. They also observed that a fraction of the Zop6 mutant protein forms larger nonsedimentable aggregates than either wild-type Zeste or the Z1 mutant protein. Because z1-35 and z1-42 lesions are similar to zop6, we examined the possibility that the pseudorevertant proteins also form larger nonsedimentable aggregates than wild type. The aggregates left in the supernatant after sedimentation of in vitro translation reactions were separated by gel filtration chromatography (Figure 5). Z+ and Z1 controls produce lower molecular weight forms (Figure 5, complexes 1 and 2) and complexes ~1200 kD in size (Figure 5, complex 3). These are virtually identical to the patterns previously observed with Z+ and Z1 (Chen and Pirrotta 1993). The complexes formed by the Z78c mutant protein are very similar to those formed by Z1 (Figure 5). Z1-35 and Z1-42, however, form complexes similar to those formed by Zop6. Like Zop6 (Chen and Pirrotta 1993), Z1-35 and Z1-42 both form larger complexes eluting in the void volume (Figure 5, complex 4). The presence of complex 4 correlates with a reduction in the level of complex 3. Therefore, the aggregation properties of the Z1-35 and Z1-42 pseudorevertant proteins are more like Zop6 than they are wild type, although their effects on white expression are opposite to that of Zop6. These results suggest that increased Zeste aggregation may contribute to, but is not sufficient for, repression of white. Furthermore, because both pseudorevertants still aggregate, we conclude that the ability of the protein to aggregate is also not sufficient to support transvection at Ultrabithorax.

DISCUSSION

zeste has intrigued geneticists because different mutant forms of the gene were found to have novel genetic interactions with other genes (Gans 1953; Lewis 1954). Zeste protein has positive effects on gene expression. For example, Zeste activates transcription from the Ubx promoter in vitro and in vivo (Bigginet al. 1988; Laney and Biggin 1992) and is required for transvection (Lewis 1954), which occurs when regulatory elements of a gene on one chromosome activate expression of the homologous gene on the other chromosome (Wu and Goldberg 1989; Geyeret al. 1990; Pirrotta 1991; Goldsborough and Kornberg 1996). Furthermore, loss-of-function zeste mutations enhance repression of gene activity caused by heterochromatin-induced PEV, indicating that Zeste promotes expression of the genes subject to PEV (Judd 1995).

In contrast to wild-type Zeste, the Z1 mutant protein represses white gene transcription in the eye (Jack and Judd 1979; Bingham and Zachar 1985). This repression is unusual in that it increases with the number of white gene targets. Repression requires at least two w+ alleles on paired homologous chromosomes or two w+ alleles in proximity to each other on the same chromosome (Jack and Judd 1979). Furthermore, although z1 is recessive toz+ with two paired w+ alleles, z1 is semidominant in the presence of three doses of w+ (tandem duplication of w+ paired with a wild-type w+), and is completely dominant with four doses (paired tandem duplications) (Jack and Judd 1979). The requirement for multiple local copies of the white gene indicates that repression by Z1 involves interallelic interactions.

A number of possible explanations could account for the alteration of Zeste activity so that Z1 observably represses transcription of w+. The z1 lesion could simply ablate the ability of Zeste to activate transcription. In this case, the Z1 protein would repress passively by occupying binding sites and failing to activate. A second possibility is that the z1 lesion creates a repression domain in Zeste de novo. Alternatively, the z1 lesion could deregulate an inherent repression activity of the Zeste protein. Although wild-type Zeste has not been observed to repress, this activity may not be apparent with the Zeste target genes that have been examined.

The genetic and molecular characteristics of the z78c, z1-35, and z1-42 mutants presented here are most consistent with the idea that wild-type Zeste has an inherent repression activity that is made apparent at white by the z1 and z 78c lesions. The z1 and z78c lesions both affect the same part of the Zeste protein, but alter different amino acids. The creation of a new activity in a protein is likely to be a rare event and unlikely to be effected by either one of two different amino acid substitutions. The mutations are more likely to have inactivated a domain of the protein that allows an existing repression activity to become apparent.

Furthermore, repression of white by the Z1 and Z78c mutant proteins is not likely to be the result of a simple loss of the ability of the proteins to activate transcription, because both proteins retain the ability to support transvection and are, therefore, able to activate. Moreover, the hypothesis that Z1 and Z78c have simply lost the ability to activate does not explain the z1 pseudorevertants, z1-35 and z1-42, which not only fail to repress, but also antagonize Z1 repression more effectively than wild-type Zeste. The enhanced antagonism of Z1 by the pseudorevertants can be accounted for by the hypothesis that Z1 reveals a repressor activity present in wild-type Zeste. A prediction of this hypothesis is that a protein lacking the repressor activity would antagonize Z1 better than wild-type Zeste, because wild-type Zeste has repressor activity itself.

The z1-35 and z1-42 mutations both substitute for prolines in a proline-rich region outside the region altered in z1 and z78c. Because the pseudorevertant mutations abolish the ability of Z1 protein to repress and support transvection, the proline-rich region, which is still wild type in Z1, is required for both the repression and activation functions of Z1. Although we do not know how the z1 and z78c lesions lead to white repression, one possibility is that they alter presentation of the proline-rich region of Zeste to other transcription factors, thereby allowing inappropriate interactions at the white locus.

Figure 5.

Gel filtration of nonsedimentable wild-type and mutant Zeste protein aggregates formed in vitro. The graphs show the elution profiles for each protein, with the complexes labeled 1–4 as in Chen and Pirrotta (1993). The Z+ and Z1 protein profiles are virtually identical to the profiles observed previously (Chen and Pirrotta 1993), and the Z78c profile is very similar to Z1. The Z1-42 and Z1-35 profiles are very similar to each other, with the gain of complex 4 and loss of complex 3 as previously observed for Zop6 (Chen and Pirrotta 1993).

The simultaneous loss of repression and activation in the Z1-35 and Z1-42 proteins cannot be explained by effects on Zeste-Zeste interactions. Although both white repression and transvection require the Zeste domain involved in self-interaction (Bickel and Pirrotta 1990; Chenet al. 1992), that domain is not affected in the Z1-35 and Z1-42 mutants. Both pseudorevertant proteins retain the ability to aggregate in vitro. Because the DNA-binding domain of Zeste is also unaffected by the z1-35 and z1-42 mutations, our data indicate that both repression of white and transvection require interactions other than Zeste-DNA and Zeste-Zeste interactions, and that these other interactions involve the proline-rich region. We postulate that the proline-rich region interacts with other transcription factors to mediate repression and activation and that the region affected by the z1 and z78c lesions regulates these interactions.

Consistent with the hypothesis that the proline-rich region interacts with other transcription factors, the zop6 mutation, which has an effect on white expression opposite to z1-35 and z1-42, produces an amino acid substitution strikingly similar to z1-35 and z1-42. zop6 alters another proline in the same cluster affected by z1-35 and z1-42. The Zop6 protein is actually a stronger repressor than Z1, repressing a single copy of white. Although the characteristics of Zeste-Zeste complex formation by Zop6 in vitro are subtly different from complexes formed by Z1, as shown here, the Z1-35 and Z1-42 proteins display the same subtle changes in aggregation, indicating that differences in ability to aggregate do not account for the fact that Zop6 is a stronger repressor than Z1. Instead, Zop6 may be a better repressor than Z1, because its proline-rich region interacts differently with other transcription factors.

The hypothesis that Zeste mediates repression and activation by interactions with other transcription factors is further supported by observations that suggest that members of the Pc-G of chromatin proteins mediate repression by Z1. Genes whose mutations dominantly enhance or suppress the z1 repression of white are either Pc-G genes or enhancers of Pc-G genes (Wuet al. 1989; Jones and Gelbart 1990; Phillips and Shearn 1990; Pelegri and Lehmann 1994; Larssonet al. 1996). Complexes of Pc-G proteins associate with Polycomb response elements (PRE) in genes to epigenetically maintain a silenced chromatin state after the state is determined by transcriptional repressors (for reviews, see Pirrotta and Rastelli 1994; Bienz and Müller 1995; Orlando and Paro 1995; Paro 1995; Pirrotta 1995; Simon 1995; Pirrotta 1997). The genetic data suggest that the Pc-G chromatin proteins are involved in repression of white by Z1. Like Z1 repression, PRE repression of white in artificial constructs variegates, and is both Pc-G and pairing-dependent. Based on the data presented here, Z1 could recruit Pc-G proteins to the white gene or induce repression by the Pc-G complex, possibly through direct interactions between the Z1 proline-rich region and one or more members of the Pc-G. Indeed, the normal function of wild-type Zeste may involve interactions with the Pc-G proteins. Zeste is present at many of the ~100 sites on the polytene chromosomes at which the Pc-G proteins colocalize (Frankeet al. 1992; Martin and Alder 1993; Rastelliet al. 1993; Lonieet al. 1994; Carrington and Jones 1996).

The phenotypes of the z1-35 and z1-42 pseudorevertants of z1 suggest a relationship between the mechanism for Z1 repression of white and Z+ activation of gene expression. One manifestation of the ability of Zeste to activate gene expression is transvection. The z1-35 and z1-42 mutations each alters only one proline in a proline-rich region, and each blocks both the ability of the protein to repress w+ and to support transvection at Ultrabithorax. As discussed above, the available evidence suggests that the Pc-G proteins likely mediate Z1 repression of w+ by control of chromatin conformation. If so, alteration of the proline-rich region of Zeste alters the activity of the Pc-G complex at white. Our observation that the same amino acid changes in the proline region render the Z1 protein unable to support transvection raises the possibility that, like Z1 repression of white, transvection is mediated by effects on chromatin conformation through interactions with Pc-G complexes or other chromatin proteins.

Similarly, Judd (1995) has proposed that Zeste acts with other proteins to establish and maintain chromatin domains. Enhancement of heterochromatin-induced PEV by zeste null and hypomorphic mutations simultaneously includes several genes repressed by the spreading heterochromatin (Judd 1995). Thus, Zeste is more likely to counter the inactive chromatin conformation caused by heterochromatin than to activate transcription of each individual gene. Taken together, the available data suggest that Zeste is involved in the establishment or maintenance of active and inactive chromatin states through interactions with the Pc-G and possibly other chromatin proteins. The data presented here suggest that the proline-rich region of Zeste, which is required for both positive and negative effects of Zeste on gene expression, may mediate interactions with chromatin proteins.

Acknowledgments

The authors thank Guy Myette for technical assistance, Burke Judd for a critical reading of the manuscript and Vince Pirrotta for stimulating discussions, sharing data prior to publication and providing plasmid DNAs. This work was supported by research grant NP-715 from the American Cancer Society to D.D. and by a Faculty Research Grant from the University of Connecticut Health Center to J.J.

Footnotes

  • Communicating editor: L. L. Searles

  • Received July 31, 1997.
  • Accepted December 22, 1997.

LITERATURE CITED

View Abstract