Mutations in Drosophila heat shock cognate 4 are enhancers of Polycomb

Abstract

The homeotic genes controlling segment identity in Drosophila are repressed by the Polycomb group of genes (PcG) and are activated by genes of the trithorax group (trxG). An F₁ screen for dominant enhancers of Polycomb yielded a point mutation in the heat shock cognate gene, hsc4, along with mutations corresponding to several known PcG loci. The new mutation is a more potent enhancer of Polycomb phenotypes than an apparent null allele of hsc4 is, although even the null allele occasionally displays homeotic phenotypes associated with the PcG. Previous biochemical results had suggested that HSC4 might interact with BRAHMA, a trxG member. Further analyses now show that there is no physical or genetic interaction between HSC4 and the Brahma complex. HSC4 might be needed for the proper folding of a component of the Polycomb repression complex, or it may be a functional member of that complex.

The homeobox-containing clusters of genes are essential for proper body pattern formation in a variety of organisms (1). In Drosophila melanogaster, a correct body plan is laid out by differential expression of the homeotic transcription factors residing within the Antennapedia complex and the bithorax complex. Members of the trithorax group (trxG) genes and the Polycomb group (PcG) genes function to regulate this differential expression. The trxG members can be broadly categorized as activators of the homeotic transcription factors (2). Conversely, the PcG members collaborate to fix the repressed state of the homeotics within particular segments (3). Indirect lines of evidence had previously suggested that the PcG members accomplish at least part of their task through modulation of chromatin structure (4). The products of several genes, including Polycomb (Pc), polyhomeotic (ph), and Posterior sex combs (Psc), reside in a biochemical complex called PRC1 (5). Products of genes in the trxG, including brahma (brm) and moira (mor) are included in the BRM complex (6, 7), the Drosophila counterpart of the SWI/SNF activation complex of Saccharomyces cerevisiae. Recently, PRC1 has been shown to antagonize the chromatin remodeling function of a SWI/SNF complex in vitro (5), supporting the chromatin-effect hypothesis.

To date, at least 15 genes have been classified as PcG members (3, 8, 9). Several different genetic screens have been used to discover PcG member genes. Polycomb itself was recognized for its dominant (haplo-insufficient) phenotype of partial segmental transformations (10). Some PcG mutations were discovered as dominant enhancers of the Pc/+ phenotype, e.g., Polycomblike (Pcl) (11). Some were found as homozygous lethal mutations with homeotic defects, including Additional sex combs (Asx) (12) and super sex combs (sxc) (13). None of these screens was sufficiently large to claim saturation of all possible loci, and probably none of the screening criteria are sufficient to identify all PcG members. Zygotic phenotypes of many PcG mutations are masked by maternal contribution of the wild-type product, as seen most dramatically in the case of extra sex combs (esc) (14). Other PcG mutations show little or no dominant phenotype [such as Su(z)2 (15)]. Virtually all, however, have been shown to enhance the dominant phenotype of Pc.

Jürgens (16) estimated that the PcG group may include as many as 40 members. This number now seems high, but there may be several PcG members yet to be discovered. We performed a new F₁ screen for dominant enhancers of a recessive viable allele of Polycomb, Pc^T3 (17), using 1-ethyl-1-nitrosourea for mutagenesis. Our screen recovered alleles of several known PcG loci but also an allele of the heat shock protein 70 (HSP70) cognate factor HSC4. We present here a genetic and molecular characterization of our hsc4 allele, and of previously isolated alleles, to define the role of HSC4 in regulating the homeobox-containing complexes.

Materials and Methods

Mutagenesis.

Males from a black, cinnabar; claret stock were fed 1-ethyl-1-nitrosourea (18) and mated to Pc^T3 homozygous females. Progeny were examined for transformations of the abdominal tergites, wings, and legs. The mutations recovered were mapped to the second or third chromosome and then tested for complementation with alleles of known PcG genes. These included Asx^XT129, esc², esc¹⁰, Pcl¹, Pcl^D5, Psc^h27, and Su(z)2^1.b8 (on the second chromosome), and Scm^28–16, Scm^D1, Pc³, Sce^b1, and Df(3L)lxd¹⁵ (on the third chromosome).

Drosophila Crosses and Phenotypes.

Flies were raised on standard cornmeal, yeast, and sugar medium at 25°C. Mutations used are described by Lindsley and Zimm (19) or in Flybase. The initial 54.1 mutagenized chromosome was made heterozygous in females with the “ru cu ca” third chromosome (ru h th st cu sr e ca), and these were backcrossed to ru cu ca males. Single F₁ males carrying chromosomes recombined within each interval were crossed to Pc³/TM1 females, to test for the presence of the 54.1 allele. Stocks were prepared for 54.1 mutant recombinant chromosomes with crossover points close to the 54.1 locus; these “cleaned up” chromosomes included ru h th st cu 54.1 and ru 54.1 sr e. Heterozygotes between these chromosomes (ru h th st cu 54.1/ru 54.1 sr e) still showed larval lethality. For the scoring of larval cuticles, the cleaned up chromosomes were balanced with TM6B(Hu), and progeny lacking the Tubby phenotype were chosen for inspection. The tested deficiencies that most closely flank the hsc4 locus were Df(3R)kar⁵¹ (87C7-D1 to 88E2–3) and Df(3R)ea (88E7–13 to 89A1).

Both 54.1 and ru 54.1 sr e chromosomes were tested for the PcG enhancement analysis, with each cross performed in both directions, except for the ph⁴¹⁰ allele. Table 1 presents averaged percentages of progeny with medium or strong transformations. Adult abdomens were mounted by the method of Duncan (11). ABDOMINAL-B protein was detected in embryos by using a mouse mAb as the primary antibody (20) and a horseradish peroxidase-coupled goat anti-mouse antibody (Bio-Rad) as the secondary antibody.

Table 1.

hsc4^54.1 interactions with PcG mutations

Genotype	A4 → A5*	A6 → A7†	Extra sex combs‡
54.1/+	4%	1%	0
54.1/Pc3	100%	100%	34
Pc3/+	ND	0%	0.3
54.1/Asxxt129	26%	7%	0.8
Asxxt129/+	10%	0%	0.1
54.1/Pcl1	70%	35%	1.6
Pcl1/+	4%	0%	0.2
54.1/ScmD1	64%	4%	2.4
ScmD1/+	13%	6%	0

ND, not determined. 

^*Transformation of the fourth abdominal segment to the fifth, expressed as percentage of males with extra black pigmentation on the fourth abdominal tergite. 

^† Transformation of the sixth abdominal segment to the seventh, expressed as the percentage of males with a reduced sixth abdominal tergite. 

^‡ Transformation of the second and third thoracic legs to the character of the forelegs, expressed as the average total number of extra sex comb teeth on the four posterior legs of a male. 

Molecular Analysis.

The coding region was recovered from hsc4^54.1/TM6B heterozygotes by PCR amplification, by using the primer pairs as described (21). Multiple copies of each PCR product were cloned and sequenced. Only one difference with the published sequence (22) was found in two separate clones. This change was verified by direct sequencing of the mixed PCR product. The region of the hsc4^54.1 mutation was also sequenced from two stocks containing the background chromosome on which hsc4^54.1 was induced. These showed only the wild-type sequence.

The cloned PCR products covering the coding region were also used to probe the Southern blots of the various hsc4 alleles; separate blots were done with two different enzymes. The alleles carrying P element insertions near hsc4 (l(3)03550 and E(nd)¹⁹⁵) gave shifted bands on the Southern blots with the enzymes chosen. Thus, changes in these bands in “revertant” lines could be scored in the presence of a wild-type copy of hsc4 from a balancer chromosome.

Analysis of HSC4 Protein.

Superose 6 gel filtration chromatography, SDS/PAGE, and immunoblotting were performed as described by Papoulas et al. (6) with the following changes. For determination of the native molecular mass of HSC4, 4.6 mg of native protein extract from 0–12 h embryos was loaded on a Superose 6 gel filtration column. For immunoprecipitation of size fractionated BRM complexes, 5 mg of extract prepared from wild-type (Oregon R) or w P[w⁺, brm-HA6HIS]92C;brm²/Df(3L)th102, h ri Sb ca² embryos was passed over the column and eluted in IP buffer (10 mM Hepes, pH 8.0/1 mM EDTA/100 mM NaCl/0.05% Tween-20/10% glycerol). In each case, 1.3 ml of pooled fractions 17 to 19 (BRM peak) was incubated with 100 μl of anti-hemagglutinin (HA) immunoaffinity resin for 4.5 h at 4°C. Bound proteins were washed with 40 volumes of IP buffer and eluted with 100 μl of 0.5 mg/ml HA dipeptide (GeneMed Synthesis) in IP buffer (30 min, room temperature). BRM was detected by Western blotting with affinity-purified rabbit anti-BRM polyclonal antibodies (23). For native molecular weight determination, HSC4 was detected by using mouse anti-Hsp70/Hsc70 mAb (clone BB70, StressGen Biotechnologies, Victoria, Canada). HSC4 was detected in immunoprecipitation samples with affinity-purified rabbit anti-Hsc4p polyclonal antibody (24).

Results

Enhancer Screen and Complementation Analyses.

Mutagenized males were crossed to Pc^T3 females, and about 30,000 F₁ progeny were screened for multiple enhancement phenotypes. Five mutants that showed significant enhancement were recovered. The mutations were assigned to chromosomes II or III and then tested for complementation with strong loss-of-function PcG alleles on the relevant chromosome. One mutation was a new Pc allele, and three others failed to complement with mutations in the esc, Pcl, and Scm loci. Unexpectedly, the esc allele appears to be antimorphic; it does not require the removal of the maternal contribution to give lethality against either esc² or esc¹⁰. With apparent null esc alleles, the removal of the maternal contribution is required for lethality (14). The molecular nature of this allele is under investigation. The remaining mutation, on the third chromosome, complemented the mutations in known PcG loci. This mutation, designated as 54.1, is the focus of the studies reported here.

Phenotypic Characterization of the 54.1 Allele.

When 54.1 is tested over Pc³, we observe dramatic transformations along the body axis (Table 1). These include anterior-to-posterior transformations in many abdominal segments (Fig. 1B), a partial wing-to-haltere transformation in the thorax (Fig. 1C), and the occurrence of multiple sex combs on both the second and third pairs of legs in males (Fig. 1 E and F). The ectopic sex combs represent a posterior-to-anterior transformation, caused by misexpression of the homeotic gene Sex combs reduced (Scr), a member of the Antennapedia complex. Nearly 50% of these flies also showed partial antenna-to-leg transformations. The spectrum and severity of the transformations vary from fly to fly among the progeny of a single cross, but the penetrance and expressivity are equivalent whether the 54.1 mutation is transmitted from the father or the mother. However, the progeny of 54.1/Pc³ females often die as first instar larvae, even when the fathers are wild type, but 54.1/Pc³ males give viable offspring when mated to wild-type females.

Figure 1.

Polycomb enhancement phenotypes of hsc4^54.1. (A) The abdominal cuticle of a wild-type female. (B) The abdominal cuticle of a 54.1/Pc³ female. Note that the first abdominal tergite (marked 1t) is transformed to the character of the second abdominal segment. Likewise, the sixth abdominal tergite (6t) and sternite (6s) are transformed to the character of the seventh abdominal segment. The seventh tergite and sternite are almost completely removed, partially transformed to the eighth abdominal segment. (C) The wing from a 54.1/Pc³ adult. Patches of wing tissue are transformed to the character of haltere, especially along the posterior margin (arrowhead). (D–F) Legs from a 54.1/Pc³ male. (D) The first thoracic leg, which shows a row of sex comb teeth (arrowhead) characteristic of the wild-type male foreleg. (E) The second thoracic leg. Note the complete row of ectopic sex comb teeth (arrowhead). (F) The third thoracic leg. There is a partial row of ectopic sex comb teeth (arrowhead).

54.1 also enhances the phenotypes of several loss-of-function PcG alleles to varying degrees (Table 1). The table presents observations of progeny from 54.1/+ mothers. In 54.1/+ progeny, there is no correlation between the strength of enhancement and maternal versus paternal transmission of the 54.1 allele. However, transformations in Asx, Pcl, or Scm/54.1 progeny were more penetrant when the 54.1/+ parent was the mother. Of particular note is that 54.1 gives synthetic lethality against the polyhomeotic mutation ph⁴¹⁰. The ph locus encodes two related products; ph⁴¹⁰ is an inversion that breaks the proximal polyhomeotic transcript (25) and it is null for the ph-proximal product (26). The ph⁴¹⁰; 54.1/+ males die as pharate adults that carry complete ectopic sex combs on the second and third thoracic legs. These animals show malformed or missing abdominal tergites and sternites, a reduced number of thoracic bristles, and missing aristae.

Heterozygous 54.1 males do not display the extra sex combs trait seen in many Polycomb group mutations (Table 1). However, they occasionally show a partial posterior transformation of the fourth abdominal tergite, seen as spots of black pigmented cuticle (Table 1). These anterior-to-posterior conversions are characteristic of loss-of-function PcG alleles, and they are presumably caused by anterior misexpression of the Abdominal-B (Abd-B) gene. A small fraction of embryos carrying 54.1 (<5%) show spotty anterior expression of the Abd-B gene product (Fig. 2A).

Figure 2.

Embryonic and larval phenotypes of hsc4^54.1. (A) ABD-B staining of a stage 14 embryo obtained from crossing the ru hsc4^54.1 sr e and ru h th st cu hsc4^54.1 stocks. The embryo was cut open along the dorsal midline and flattened. Note the ectopic expression of ABD-B, especially in the central nervous system, up through the second thoracic segment, or parasegment 4 (arrowhead). A bracket marks parasegment 10, the anterior limit of ABD-B expression in wild-type embryos. (B) The ventral cuticle of a ru h th st cu hsc4^54.1/+ second instar larva. The denticle belts of the third, fourth, and fifth abdominal segments are marked. Most of the denticles of the fourth abdominal segment are missing (arrow).

54.1 homozygous individuals survive embryogenesis but die as larvae before the third instar. (The mutagenized chromosome was first “cleaned” by recombination; see Materials and Methods.) We examined both 54.1 heterozygous and homozygous larvae for signs of homeotic transformations. Neither class displays any obvious homeotic phenotypes. However, we noticed that 7% of the larvae carrying one copy of 54.1 from their mothers show either partial or complete loss of one or more ventral denticle belts, predominantly in the A4 segment (Fig. 2B). This percentage is higher for 54.1 homozygous larvae (13%).

Recombination and Deletion Mapping Analyses.

Salivary polytene chromosomes from 54.1/+ heterozygotes were examined for potential rearrangements on the third chromosome; none were found. We proceeded to map the 54.1 mutation by meiotic recombination with a multiply marked third chromosome (ru h th st cu sr e ca). 54.1 was assigned to a region between the curled (cu) and stripe (sr) loci. The 54.1 allele was then mapped relative to the dominant Stubble marker (at 89B9–10 on the cytological map); it lies approximately 1 map unit proximal to Stubble. We also tested a series of overlapping deletions spanning the interval 86C1–88E2–3 and another series spanning 88E13–89E. None of them uncovered our 54.1 allele. This analysis suggested that 54.1 resides within the 88E2–3 to 88E13 region, for which we had no deficiencies, or that the 54.1 allele might be viable over a deficiency for its site.

Allelism with HSC4.

We first checked candidate genes in the 88D-E region. 54.1 fully complemented a mutation in E(Pc)88D (27). It also failed to modify the position-effect variegation of the w^m4h inversion, and so it seemed unlikely to be allelic to E(var)88D (28) or Su(var)3–9 (29). We then tested a collection of lethal P-element insertions within the 88E segment. The original 54.1 stock and the derivative recombinants ru 54.1 sr e and ru h th st cu 54.1 ca all gave lethality against the insertion line l(3)03550 in the 88E region. This P-insertion behaves as a strong hypomorphic allele of the locus that encodes for the HSC4 product (21). The P element is inserted 431 bp upstream of the likely RNA start site for the hsc4 mRNA (Berkeley Fly Database; ref. 22). Schmucker et al. (30) named the locus scattered before its identity to hsc4 was discovered (21). However, the scattered name had already been given to a locus on the second chromosome; we prefer to call the third chromosome locus hsc4 to avoid confusion.

We tested a variety of hsc4 alleles, including a likely null mutation, for complementation with 54.1. Schmucker et al. (30) had treated the l(3)03550 P insertion with x-rays to induce deletions for the locus. Among the recovered chromosomes lacking the P element selectable marker were several derivatives that gave more severe phenotypes. We obtained three of these revertant lines, examined them cytologically, and analyzed their DNAs by Southern blots using a probe for the hsc4 coding region. The II-R35-R2 line showed a small deficiency in the 88E3–5 region and an altered restriction map according to the Southern blots. The II-R35-R7 line looked normal (like the parental P insertion) in both assays. The third line, I-R35-R7, lacked an hsc4 band on Southern blots, although it was cytologically normal. This line is our best candidate for a complete null. Hing et al. (21) isolated the hsc4¹⁹⁵ allele as an enhancer of notchoid. In the same study, they reverted the enhanced phenotype with additional mutations in hsc4. We also obtained hsc4¹⁹⁵ and a revertant called hsc4^PB2, which had been induced by P element-mediated dysgenesis. All five of the hsc4 alleles mentioned above were lethal when heterozygous to 54.1.

54.1 showed no abnormality on Southern blots in the hsc4 coding region. The coding region was recovered from 54.1/TM6B heterozygotes by PCR amplification and sequenced. A single sequence change was found relative to the published sequence (22), a T to C transition changing Leu₃₈₀ to Pro. This change was absent in the background chromosome on which 54.1 was induced. Thus, the 54.1 mutation is designated below as hsc4^54.1.

The Leu₃₈₀ residue is conserved among HSP70 homologs from bacteria to mammals (31). A crystal structure has been determined for the 44-kDa amino-terminal ATPase domain of bovine HSP70 (32). The equivalent leucine residue lies within an extended α-helix, although it is very near the end of the peptide studied, and so it is not possible to describe its environment in the intact protein.

PcG Interactions with Additional hsc4 Alleles.

The hsc4^54.1 allele was compared with six other hsc4 mutations described above for enhancement of the Pc³ mutation in trans-heterozygotes (Table 2). Although three other alleles showed some enhancement of the segmental transformations, none were as pronounced as those of hsc4^54.1, and the likely null allele, I-R35-R7, showed no clear interaction. Because hsc4^54.1 gives synthetic lethality against the ph⁴¹⁰ allele, we checked the additional hsc4 alleles for interaction with ph⁴¹⁰. All three revertants of the l(3)03550 P insertion, including the likely null, also caused lethality of late pupae, although with less severe segmental transformations than in the hsc4^54.1 case (Table 2). Other hsc4 alleles, not likely to be null, were viable in combination with ph⁴¹⁰ (Table 2). ph⁴¹⁰; hsc4^54.1/+ males carrying a duplication for hsc4 (P[hsc4]) (21) occasionally survive, and they show less severe phenotypes than pharate adults without the duplication. Likewise, hsc4^54.1/Pc³ trans-heterozygotes with a duplication for the hsc4 gene still show some enhancement of the Pc transformations, although less pronounced than the enhancement without the duplication. These observations suggest that hsc4^54.1 is an antimorphic allele.

Table 2.

Interaction phenotypes of hsc4 alleles

hsc4 allele (from father)	Pc3			ph410		trxB11
	T2,3 → T1*	A4 → A5†	A6 → A7‡	Lethal?§	T2,3 → T1	T1 → T2¶	A5 → A4‖
54.1	+++	+	++	l.	++	0	0/−
l(3)03550	++	0	0	v.	+	0	0
II-35-R2	+	+/0	0	l.	+	0	+
II-35-R7	+	+/0	0	l.	+	0	0/+
I-35-R7	0	+/0	0	l.	+	++	++
E(nd)195	−	+	0	v.	−	0	−
195rvPB2	−	+	0	v.	−/0	0	−
Dp-P[hsc4]	0	0	0	v.	0	0	−

^*Transformation of the second and third thoracic legs toward that of the first, seen as extra sex combs on the second and third legs of the males. Enhancements are categorized as complete (+++), strong (++), weak (+), or not more than seen with wild-type fathers (0). −, Denotes suppression of the indicated transformation. 

^† Transformation of the fourth abdominal segment toward the fifth, seen as black pigmentation appearing on the fourth abdominal tergite of the male. 

^‡ Transformation of the sixth abdominal segment to the seventh, seen as reduction of the sixth abdominal tergite in males and females. 

^§ Pharate lethality of males with the ph⁴¹⁰ mutation, indicated as lethal (l.) or viable (v.). 

^¶ Transformation of the first leg toward the second, seen as loss of sex comb teeth in the male. 

^‖ Transformation of the fifth abdominal segment toward the fourth, seen as loss of black pigment on the male fifth abdominal tergite. 

HSC4 and the trxG.

A simple interpretation of the genetic data is that hsc4 is required for the function of one or more PcG proteins. Alternatively, hsc4 could be a negative regulator of a trxG protein. Consistent with this second interpretation, HSC4 was identified as a component of the purified BRM complex (6). The brm gene was first identified as a suppressor of Pc mutations (33), and it encodes the ATPase subunit of a protein complex related to the SWI/SNF chromatin-remodeling complex in yeast. However, the presence of HSC4 in purified BRM complexes could represent contamination by this very abundant protein. To resolve this issue, we examined the physical association of HSC4 with BRM by comparing their native molecular weights in whole embryo extracts. Western blotting of fractions from a gel-filtration column showed some overlap between the elution peaks of these two proteins (Fig. 3A), consistent with the prior copurification from similar extracts. However, it was not clear whether the small fraction of HSC4 in the BRM-containing fractions was physically associated with BRM. When BRM complexes were immunoprecipitated from these partially purified fractions, no specific association with HSC4 was detected (Fig. 3B). Because these immunoprecipitations were performed by using near-physiological salt and buffer conditions, even a relatively weak interaction should have been detected. We note also that the BRM complex purified by Kal et al. (7) did not include HSC4. It is therefore unlikely that HSC4 is a subunit of the BRM complex.

Figure 3.

HSC4 separation from the BRM complex in embryos. (A) Superose 6 gel filtration chromatography of embryo extracts, with fractions separated on Western blots and probed for BRM or HSC4. Lanes are designated by column fraction number or L, material loaded. Elution volumes of native molecular mass standards are indicated above. There is only a small amount of overlap between the BRM and HSC4 peaks. (B) BRM immunoprecipitates. An anti-HA mAb was used to immunoprecipitate extracts from wild-type control embryos (C) or embryos containing HA-epitope-tagged BRM protein (HA). Immunoprecipitation was from the pooled BRM peak fractions of a Superose 6 column. Western blotting of the starting material (total), supernatant (sup), and pellet samples revealed only trace amounts of HSC4 in both pellets, indicating no specific association with BRM.

To determine whether we could detect a functional interaction between hsc4 and brm, we used two genetic assays. First, we asked whether hsc4 mutations could either enhance or suppress phenotypes created by a dominant-negative brm allele, brm^K804R. brm^K804R induces a rough eye phenotype when expressed from a UAS-brm^K804R transgene under control of an eyeless-Gal4 driver (O.P., J.A., and J.T., unpublished observations). This eye phenotype is enhanced by alleles of brm or mor. The three tested hsc4 alleles (hsc4^54.1, l(3)03550, and II-R35-R7) neither enhanced nor suppressed the eye phenotype. We then tested a second phenotype of the antimorphic brm^K804R transgene. An insertion of this transgene on the X chromosome (P[w⁺, brm^K804R]22D) reduces the amount of functional brm in males to levels barely sufficient for survival to adulthood. Mutations in brm or mor cause dominant lethality of males in this sensitized background (6). hsc4^54.1, l(3)03550, and II-R35-R7 did not produce male lethality in this test.

We also tested the hsc4 alleles for interactions with a strong allele of trithorax (trx). trx is the prototypic activator gene of the trxG, but its biochemical function is not understood. For some alleles of hsc4 (especially I-R35-R7), the hsc4, trx^B11 double heterozygous adult males showed enhancement of the mild anterior transformations of the fifth, sixth, and seventh abdominal segments seen in trx^B11/+, and/or reduced numbers of sex comb teeth on the first thoracic legs (Table 2). Conversely, an extra copy of hsc4 suppressed the transformations of trx^B11/+ adults. The simultaneous effects on repression (Polycomb) and activation (trithorax) functions might explain why hsc4 mutations alone have only subtle and variable effects on segmental specification.

Discussion

The hsc4 gene encodes the major heat shock cognate protein, which is expressed at all times of development, but is particularly abundant in ovaries and embryos (34). A variety of phenotypes have been associated with hsc4 mutations. Removal of both the zygotic and maternal contributions of hsc4 causes embryonic lethality accompanied by variable segmental deletions (35). Loss of zygotic activity alone leads to larval lethality (21). These dying larvae develop melanotic tumors (36) and are defective in the development and projection of the larval optic nerve, referred to as Bolwig's nerve (30). We have observed that heterozygotes for all alleles show malformations of the fourth abdominal tergite in a small fraction of adults. A dominant negative allele of hsc4, hsc4¹⁹⁵, was found to modulate Notch signaling during development (21). The same allele enhances the dominant adult phenotypes of mutations in the ecdysone receptor (37). We described above the interaction with mutations in the PcG, which constitutes another distinctive class of phenotypes. We also noted a seemingly contradictory enhancement of trithorax mutation, itself an antagonist of Polycomb. One mechanism that might connect all of these phenotypes is the one suggested by the protein sequence of HSC4, namely, the protein folding process. Protein targets of the HSC4 chaperone function might be involved in many diverse developmental pathways.

The interaction of the hsc4^54.1 allele with Polycomb is distinctive. The HSC4 protein is not particularly limiting, because heterozygotes for a null allele appear nearly wild type. Heterozygotes for hsc4^54.1 do show a phenotype, suggesting that the hsc4^54.1 protein product might act as a poison. Perhaps the hsc4^54.1 protein product sequesters a target protein or misfolds it into a nonfunctional product. If POLYCOMB were the target, this could inactivate 50% of the POLYCOMB protein, assuming half of the POLYCOMB nascent chains interact with the mutant form of HSC4. A 50% reduction in functional POLYCOMB would explain the weak phenotypes of hsc4^54.1/+, and the lethality of hsc4 alleles with ph⁴¹⁰ (Table 2), because another ph-proximal null, ph⁴⁰⁹, is lethal in a Pc⁻/+ background (38).

An alternative explanation for the hsc4^54.1 phenotype is that HSC4 is more directly involved in the process of Polycomb group repression. Recently, HSC4 has been identified among the components of the Polycomb complex purified from Drosophila embryos (A. Saurin, Z. Shao, H. Erdjument-Bromage, P. Tempst, and R. Kingston, personal communication). Because HSC4 is so abundant, it could be merely a contaminant, but the purification included immunoprecipitation of an epitope-tagged complex, and it was demonstrated that some HSC4 protein coelutes with the complex on a sizing column. It is possible that HSC4 remains bound to one of its targets and becomes incorporated into the complex with that protein. It is also possible that HSC4 has a catalytic role. HSC4 is related to a variety of ATPases, including actin (39). A bacterial homolog of HSC4, DnaK, is thought to rearrange the protein complex involved in phage lambda DNA replication (40). Other actin-related proteins have been found in the SWI/SNF and RSC chromatin remodeling complexes (41), and it has been suggested that they catalyze conformational changes in those complexes or their substrates (42). HSC4 could have a similar function in PRC1 repression.

Abbreviations

trxG

trithorax group

PcG

Polycomb group

HSC4

heat shock cognate 4

HSP

heat shock protein