A functional interaction between the histone deacetylase Rpd3 and the corepressor groucho in Drosophila development

Abstract

The Drosophila gene groucho (gro) encodes a transcriptional corepressor that has critical roles in many development processes. In an effort to illuminate the mechanism of Gro-mediated repression, we have employed Gro as an affinity reagent to purify Gro-binding proteins from embryonic nuclear extracts. One of these proteins was found to be the histone deacetylase Rpd3. Protein-protein interaction assays suggest that Gro and Rpd3 form a complex in vivo and that they interact directly via the glycine/proline rich (GP) domain in Gro. Cell culture assays demonstrate that Rpd3 potentiates repression by the GP domain. Furthermore, experiments employing a histone deacetylase inhibitor, as well as a catalytically inactive form of Rpd3, imply that histone deacetylase activity is required for efficient Gro-mediated repression. Finally, mutations in gro and rpd3 have synergistic effects on embryonic lethality and pattern formation. These findings support the view that Gro mediates repression, at least in part, by the direct recruitment of the histone deacetylase Rpd3 to the template, where it can modulate local chromatin structure. They also provide evidence for a specific role of Rpd3 in early development.

Figure 1

Affinity purification of Gro-interacting proteins. (A) Silver-stained SDS–polyacrylamide gel showing results of a small scale affinity purification of Gro-interacting proteins. (Lane 1) Molecular mass markers; (lane 2) unfractionated 0- to 12-hr Drosophila embryonic nuclear extract; (lanes 3–6) SDS elutions of M₂ antibody beads lacking (M₂, lanes 3,5) or containing (M₂Gro, lanes 4,6) epitope-tagged Gro. (Lanes 5,6) The beads were incubated with embryonic nuclear extract and subsequently subjected to extensive washing prior to SDS elution; (lanes 3,4) the nuclear extract was omitted. IgG light (L) and heavy (H) chains are indicated. Polypeptides from the embryonic nuclear extracts that uniquely coprecipitate with the M₂Gro beads are indicated (p140, p110, p68, p38, and p31). (*) Two polypeptides that are retained on the beads in both the presence and absence of Gro. (B) Silver-stained SDS–polyacrylamide gel showing an aliquot of the eluate from a large-scale M₂Gro affinity column used for affinity purification of Gro-interacting proteins. Elution was achieved by washing the beads with excess Flag peptide. Thus, unlike in A, the immunoglobulin chains were not eluted. However, the same two nonspecific polypeptides (*) that eluted with SDS (A), eluted with the peptide, indicating that they are probably binding directly to the antibody. The amino acid sequence denotes a peptide sequence obtained from an internal lysC fragment of p68. This sequence is a perfect match to a predicted lysC fragment in Drosophila histone deacetylase, Rpd3. P38 has been identified as Drosophila histone H1 (H1). (C) Immunblot with anti-Rpd3 antibody. Lanes are the same as in A.

Figure 2

Endogenous Gro and endogenous Rpd3 form a complex in vivo. (A) Immunoprecipitation of embryonic (lanes 1–4) and S2 cell (lanes 5–8) nuclear extracts. (Lanes 1,5) Unfractionated nuclear extracts; (lanes 2,6) immunoprecipitates obtained with nonimmune serum; (lanes 3,7) immunoprecipitates obtained with anti-Gro antibody; (lanes 4,8) immunoprecipitates obtained with anti-Flag antibody. Proteins were analyzed by SDS-PAGE and immunoblotting with anti-Gro (top) or anti-Rpd3 (bottom) antibodies. (B) In vitro histone deacetylase assays of immunoprecipitates from embryonic nuclear extract. Samples assayed are those shown in lanes 1-4 of A. In addition, purified baculovirus-expressed Rpd3 was assayed as a positive control. Each sample was assayed in the presence (+) or absence (−) of the histone deacetylase inhibitor TSA (150 nm).

Figure 3

A direct interaction between Gro and Rpd3. (A) Copurification of baculovirus-expressed Flag-tagged Gro (M₂Gro) and six-histidine-tagged Rpd3 (H₆Rpd3). Nuclear extracts (NE) prepared from insect cells expressing either M₂Gro, H₆Rpd3, or both together were incubated with Ni²⁺–NTA–agarose (Ni) or anti-Flag affinity beads (M₂). After extensive washing of beads, bound proteins were eluted with SDS, resolved by SDS-PAGE, and immunoblotted with anti-Gro (top) or anti-Rpd3 (bottom) antibodies. (B) A Coomassie blue stained SDS–polyacrylamide gel showing baculovirus-expressed and affinity-purified Flag-tagged Gro (M₂Gro) and Flag-tagged Rpd3 (M₂Rpd3). (C) Equal amounts of anti-Flag M₂ beads (M₂) or M₂ beads containing purified Flag-tagged Gro (M₂Gro) were incubated with ³⁵S-labeled Rpd3 produced by in vitro translation. After subjecting the beads to extensive washing, bound ³⁵S-labeled Rpd3 was eluted, and analyzed by SDS-PAGE and autoradiography. (D) In a reciprocal experiment, M₂ beads containing purified M₂Rpd3 were used to examine the interaction with ³⁵S-labeled Gro. (1/10 IP) Ten percent of total ³⁵S-labeled protein input.

Figure 4

Mapping the Rpd3-interaction domain to the GP region of Gro. (A) Schematic diagram of different Gro deletions used in the protein interaction assays described in B–D. The conserved glutamine-rich (Q) and WD repeat (WD) domains of Gro are shaded. GP and SP denote the glycine/proline and serine/proline rich regions of Gro. CcN represents the motif containing putative cdc2 and casein kinase II phosphorylation sites as well as a nuclear localization signal. (B) The WD-repeat domain of Gro is dispensable for the Rpd3 interaction. Equal amounts of nuclear extracts prepared from insect cells coexpressing a Flag-tagged Gro deletion (M₂Gro^N420 or M₂Gro^Δ397) and H₆Rpd3 were affinity purified using Ni²⁺–NTA–agarose (Ni) or anti-FLAG affinity beads (M₂). Purified proteins were immunoblotted with anti-FLAG (top) or anti-Rpd3 (bottom) antibodies. (C) The GP domain of Gro is required for the interaction with Rpd3. Equal amounts of anti-Flag affinity beads alone (M₂) or beads containing purified M₂Rpd3 were used to examine the interac-tions with the indicated ³⁵S-labeledGro deletions. Bound proteins were resolved by SDS-PAGE and visualized by autoradiography. (1/10 IP) Ten percent of each ³⁵S-labeled protein input. (D) GST pull-down assays confirm the requirement of the GP domain for the Rpd3 interaction. Highly purified GST–Gro fusions were immobilized on glutathione beads and incubated with ³⁵S-labeled Rpd3. After extensive washing of the beads, bound proteins were eluted and separated by SDS-PAGE and visualized by autoradiography (top). (Bottom) A Coomassie blue stained SDS–polyacrylamide gel of the purified GST–Gro fusions.

Figure 4

Mapping the Rpd3-interaction domain to the GP region of Gro. (A) Schematic diagram of different Gro deletions used in the protein interaction assays described in B–D. The conserved glutamine-rich (Q) and WD repeat (WD) domains of Gro are shaded. GP and SP denote the glycine/proline and serine/proline rich regions of Gro. CcN represents the motif containing putative cdc2 and casein kinase II phosphorylation sites as well as a nuclear localization signal. (B) The WD-repeat domain of Gro is dispensable for the Rpd3 interaction. Equal amounts of nuclear extracts prepared from insect cells coexpressing a Flag-tagged Gro deletion (M₂Gro^N420 or M₂Gro^Δ397) and H₆Rpd3 were affinity purified using Ni²⁺–NTA–agarose (Ni) or anti-FLAG affinity beads (M₂). Purified proteins were immunoblotted with anti-FLAG (top) or anti-Rpd3 (bottom) antibodies. (C) The GP domain of Gro is required for the interaction with Rpd3. Equal amounts of anti-Flag affinity beads alone (M₂) or beads containing purified M₂Rpd3 were used to examine the interac-tions with the indicated ³⁵S-labeledGro deletions. Bound proteins were resolved by SDS-PAGE and visualized by autoradiography. (1/10 IP) Ten percent of each ³⁵S-labeled protein input. (D) GST pull-down assays confirm the requirement of the GP domain for the Rpd3 interaction. Highly purified GST–Gro fusions were immobilized on glutathione beads and incubated with ³⁵S-labeled Rpd3. After extensive washing of the beads, bound proteins were eluted and separated by SDS-PAGE and visualized by autoradiography (top). (Bottom) A Coomassie blue stained SDS–polyacrylamide gel of the purified GST–Gro fusions.

Figure 4

Mapping the Rpd3-interaction domain to the GP region of Gro. (A) Schematic diagram of different Gro deletions used in the protein interaction assays described in B–D. The conserved glutamine-rich (Q) and WD repeat (WD) domains of Gro are shaded. GP and SP denote the glycine/proline and serine/proline rich regions of Gro. CcN represents the motif containing putative cdc2 and casein kinase II phosphorylation sites as well as a nuclear localization signal. (B) The WD-repeat domain of Gro is dispensable for the Rpd3 interaction. Equal amounts of nuclear extracts prepared from insect cells coexpressing a Flag-tagged Gro deletion (M₂Gro^N420 or M₂Gro^Δ397) and H₆Rpd3 were affinity purified using Ni²⁺–NTA–agarose (Ni) or anti-FLAG affinity beads (M₂). Purified proteins were immunoblotted with anti-FLAG (top) or anti-Rpd3 (bottom) antibodies. (C) The GP domain of Gro is required for the interaction with Rpd3. Equal amounts of anti-Flag affinity beads alone (M₂) or beads containing purified M₂Rpd3 were used to examine the interac-tions with the indicated ³⁵S-labeledGro deletions. Bound proteins were resolved by SDS-PAGE and visualized by autoradiography. (1/10 IP) Ten percent of each ³⁵S-labeled protein input. (D) GST pull-down assays confirm the requirement of the GP domain for the Rpd3 interaction. Highly purified GST–Gro fusions were immobilized on glutathione beads and incubated with ³⁵S-labeled Rpd3. After extensive washing of the beads, bound proteins were eluted and separated by SDS-PAGE and visualized by autoradiography (top). (Bottom) A Coomassie blue stained SDS–polyacrylamide gel of the purified GST–Gro fusions.

Figure 4

Mapping the Rpd3-interaction domain to the GP region of Gro. (A) Schematic diagram of different Gro deletions used in the protein interaction assays described in B–D. The conserved glutamine-rich (Q) and WD repeat (WD) domains of Gro are shaded. GP and SP denote the glycine/proline and serine/proline rich regions of Gro. CcN represents the motif containing putative cdc2 and casein kinase II phosphorylation sites as well as a nuclear localization signal. (B) The WD-repeat domain of Gro is dispensable for the Rpd3 interaction. Equal amounts of nuclear extracts prepared from insect cells coexpressing a Flag-tagged Gro deletion (M₂Gro^N420 or M₂Gro^Δ397) and H₆Rpd3 were affinity purified using Ni²⁺–NTA–agarose (Ni) or anti-FLAG affinity beads (M₂). Purified proteins were immunoblotted with anti-FLAG (top) or anti-Rpd3 (bottom) antibodies. (C) The GP domain of Gro is required for the interaction with Rpd3. Equal amounts of anti-Flag affinity beads alone (M₂) or beads containing purified M₂Rpd3 were used to examine the interac-tions with the indicated ³⁵S-labeledGro deletions. Bound proteins were resolved by SDS-PAGE and visualized by autoradiography. (1/10 IP) Ten percent of each ³⁵S-labeled protein input. (D) GST pull-down assays confirm the requirement of the GP domain for the Rpd3 interaction. Highly purified GST–Gro fusions were immobilized on glutathione beads and incubated with ³⁵S-labeled Rpd3. After extensive washing of the beads, bound proteins were eluted and separated by SDS-PAGE and visualized by autoradiography (top). (Bottom) A Coomassie blue stained SDS–polyacrylamide gel of the purified GST–Gro fusions.

Figure 5

Histone deacetylase activity contributes to Gro-mediated repression in cultured cells. (A) Schematic diagram of various Gal4–Gro fusion constructs used in the cotransfection assays described in B–D. (TD) The tetramerization domain of p53 (residues 309–371). (B) TSA treatment dramatically reduces Gal4–Gro mediated repression. The structure of the firefly luciferase reporter (G5DE5tkLuc) is depicted at top. This reporter and an internal control reporter (p-37tkRLuc) encoding Renilla luciferase were cotransfected into S2 cells with vectors expressing Dorsal, Twist, and the Gal4 DNA-binding domain (G4) or the Gal4–Gro fusion protein (G4Gro). Twenty-four hr post-transfection, cells were treated with the specified amounts of TSA and luciferase activities were measured 10 hr later. All firefly luciferase activities (normalized first to the control Renilla luciferase activities) are normalized to the activity in the presence of Dorsal, Twist, and Gal4 DBD alone, which is set at 100%. Each bar represents the average plus standard deviation of three independent duplicate assays (left). The fold repression (right) reflects the ratio of the activity observed with Gal4 DNA-binding domain alone to that observed with Gal4–Gro at each TSA concentration. (C) The Gro GP domain is able to repress transcription when fused to the Gal4 DBD and the p53 TD (G4TDGP). However, the GP domain alone (G4GP) and the p53 TD alone (G4TD) fail to repress transcription when fused to the Gal4 DNA-binding domain. Cotransfection assays were conducted as described in B. (D) The Gal4–GP fusion protein synergizes with the enzymatically active form of Rpd3 to repress transcription. S2 cells were transfected with the reporters and expression constructs encoding Dorsal, Twist, and the indicated Gal4 fusion protein in the absence (−, open bars) or presence of a vector expressing either wild-type (WT, shaded bars) or single-point mutant (H196 F, solid bars) forms of Rpd3.

Figure 5

Histone deacetylase activity contributes to Gro-mediated repression in cultured cells. (A) Schematic diagram of various Gal4–Gro fusion constructs used in the cotransfection assays described in B–D. (TD) The tetramerization domain of p53 (residues 309–371). (B) TSA treatment dramatically reduces Gal4–Gro mediated repression. The structure of the firefly luciferase reporter (G5DE5tkLuc) is depicted at top. This reporter and an internal control reporter (p-37tkRLuc) encoding Renilla luciferase were cotransfected into S2 cells with vectors expressing Dorsal, Twist, and the Gal4 DNA-binding domain (G4) or the Gal4–Gro fusion protein (G4Gro). Twenty-four hr post-transfection, cells were treated with the specified amounts of TSA and luciferase activities were measured 10 hr later. All firefly luciferase activities (normalized first to the control Renilla luciferase activities) are normalized to the activity in the presence of Dorsal, Twist, and Gal4 DBD alone, which is set at 100%. Each bar represents the average plus standard deviation of three independent duplicate assays (left). The fold repression (right) reflects the ratio of the activity observed with Gal4 DNA-binding domain alone to that observed with Gal4–Gro at each TSA concentration. (C) The Gro GP domain is able to repress transcription when fused to the Gal4 DBD and the p53 TD (G4TDGP). However, the GP domain alone (G4GP) and the p53 TD alone (G4TD) fail to repress transcription when fused to the Gal4 DNA-binding domain. Cotransfection assays were conducted as described in B. (D) The Gal4–GP fusion protein synergizes with the enzymatically active form of Rpd3 to repress transcription. S2 cells were transfected with the reporters and expression constructs encoding Dorsal, Twist, and the indicated Gal4 fusion protein in the absence (−, open bars) or presence of a vector expressing either wild-type (WT, shaded bars) or single-point mutant (H196 F, solid bars) forms of Rpd3.

Figure 5

Histone deacetylase activity contributes to Gro-mediated repression in cultured cells. (A) Schematic diagram of various Gal4–Gro fusion constructs used in the cotransfection assays described in B–D. (TD) The tetramerization domain of p53 (residues 309–371). (B) TSA treatment dramatically reduces Gal4–Gro mediated repression. The structure of the firefly luciferase reporter (G5DE5tkLuc) is depicted at top. This reporter and an internal control reporter (p-37tkRLuc) encoding Renilla luciferase were cotransfected into S2 cells with vectors expressing Dorsal, Twist, and the Gal4 DNA-binding domain (G4) or the Gal4–Gro fusion protein (G4Gro). Twenty-four hr post-transfection, cells were treated with the specified amounts of TSA and luciferase activities were measured 10 hr later. All firefly luciferase activities (normalized first to the control Renilla luciferase activities) are normalized to the activity in the presence of Dorsal, Twist, and Gal4 DBD alone, which is set at 100%. Each bar represents the average plus standard deviation of three independent duplicate assays (left). The fold repression (right) reflects the ratio of the activity observed with Gal4 DNA-binding domain alone to that observed with Gal4–Gro at each TSA concentration. (C) The Gro GP domain is able to repress transcription when fused to the Gal4 DBD and the p53 TD (G4TDGP). However, the GP domain alone (G4GP) and the p53 TD alone (G4TD) fail to repress transcription when fused to the Gal4 DNA-binding domain. Cotransfection assays were conducted as described in B. (D) The Gal4–GP fusion protein synergizes with the enzymatically active form of Rpd3 to repress transcription. S2 cells were transfected with the reporters and expression constructs encoding Dorsal, Twist, and the indicated Gal4 fusion protein in the absence (−, open bars) or presence of a vector expressing either wild-type (WT, shaded bars) or single-point mutant (H196 F, solid bars) forms of Rpd3.

Figure 5

Histone deacetylase activity contributes to Gro-mediated repression in cultured cells. (A) Schematic diagram of various Gal4–Gro fusion constructs used in the cotransfection assays described in B–D. (TD) The tetramerization domain of p53 (residues 309–371). (B) TSA treatment dramatically reduces Gal4–Gro mediated repression. The structure of the firefly luciferase reporter (G5DE5tkLuc) is depicted at top. This reporter and an internal control reporter (p-37tkRLuc) encoding Renilla luciferase were cotransfected into S2 cells with vectors expressing Dorsal, Twist, and the Gal4 DNA-binding domain (G4) or the Gal4–Gro fusion protein (G4Gro). Twenty-four hr post-transfection, cells were treated with the specified amounts of TSA and luciferase activities were measured 10 hr later. All firefly luciferase activities (normalized first to the control Renilla luciferase activities) are normalized to the activity in the presence of Dorsal, Twist, and Gal4 DBD alone, which is set at 100%. Each bar represents the average plus standard deviation of three independent duplicate assays (left). The fold repression (right) reflects the ratio of the activity observed with Gal4 DNA-binding domain alone to that observed with Gal4–Gro at each TSA concentration. (C) The Gro GP domain is able to repress transcription when fused to the Gal4 DBD and the p53 TD (G4TDGP). However, the GP domain alone (G4GP) and the p53 TD alone (G4TD) fail to repress transcription when fused to the Gal4 DNA-binding domain. Cotransfection assays were conducted as described in B. (D) The Gal4–GP fusion protein synergizes with the enzymatically active form of Rpd3 to repress transcription. S2 cells were transfected with the reporters and expression constructs encoding Dorsal, Twist, and the indicated Gal4 fusion protein in the absence (−, open bars) or presence of a vector expressing either wild-type (WT, shaded bars) or single-point mutant (H196 F, solid bars) forms of Rpd3.

Figure 6

Functional analysis of a single point mutation of Rpd3. (A) A Coomassie blue-stained SDS–polyacrylamide gel showing equal amounts of purified wild-type (WT) and mutant forms of Rpd3. In mutant Rpd3, a conserved histidine residue (H196) has been changed to phenylalanine (H196F). (B) In vitro histone deacetylase assays of purified wild-type (WT) and mutant (H196F) forms of Rpd3. (C) GST pull-down assays show that wild-type (WT) and mutant (H196F) forms of Rpd3 bind to Gro with comparable affinities. GST-pull down assays are conducted as described in Fig. 4D. (D) Gal4 fused to wild-type Rpd3 (WT) is able to repress transcription in S2 cells, whereas Gal4 fused to the single-point mutant form of Rpd3 (H196F) is not able to repress transcription in S2 cells. Cotransfection assays were conducted as described in Fig. 5C.

Figure 7

Spatial expression pattern of rpd3 during Drosophila oogenesis and embryogenesis. (A) Genomic organization of the rpd3 gene. The map of the rpd3 transcription unit (at cytological map location 64C1-2) is based on previous reports (DeRubertis et al. 1996; Maixner et al. 1998). BamHI (B), EcoRI (E), HindIII (H), and SalI (S) sites are indicated. The P1633 and P15/1 P-element insertion sites are depicted by inverted triangles. (B–E) Staining of wild-type ovaries and embryos with anti-Rpd3 antibodies. (B) Ubiquitous germ-line nuclear expression of Rpd3 is observed in the wild-type ovary during early oogenesis (before stage 8). (C) Uniform expression of Rpd3 is observed in follicle cell nuclei by stage 10 of oogenesis. (D) Rpd3 is uniformly distributed throughout the nuclei of precellular embryos. (E) Patches of zygotic expression (arrows) are observed in the anterior region of stage 9–10 embryos. (F,G) Characterization of Rpd3 expression in embryos carrying the P1633 P-element insertion. (F) Expression of Rpd3 in the an-terior of stage 9–10 embryos (arrows) heterozygous for the P-element insertion is reduced relative to wild-type embryos (see E). The seven-stripes of lacZ expression due to the ftz–lacZ marker on the TM3 balancer chromosome are indicated with asterisks. (G) Expression of Rpd3 in stage 9–10 embryos homozygous for the P-element insertion is undetectable. Homozygous embryos were recognized by the absence of the ftz–lacZ stripes.

Figure 8

Genetic interaction between gro and rpd3. (A) Embryos were collected from females of the indicated genotypes that had been mated with wild-type males. After 48 hr at 28°C, unhatched embryos were scored to calculate embryonic lethality. The number of embryos scored and percent lethality (%) are indicated to the right and the inside of each bar, respectively. (B) Cuticle of an embryo derived from a wild-type mother (WT). (C,D) Cuticles of unhatched embryos derived from mothers trans-heterozygous for either of two gro alleles (E48 or BX22) and the P-insertional allele of rpd3 (P1633) showing the bicaudal phenotype including a duplicated posterior spiracle (arrowheads) and a mirror-image duplication of the posterior abdominal segments in place of normal anterior segments. (E) Cuticle of unhatched embryo derived from mother trans-heterozygous for gro^E48 and a deficiency that removes rpd3 (Df10H). Anterior duplication of the posterior spiracle (arrowheads) is accompanied by disorganized denticle belts. (F) Cuticle of an embryo derived from a female containing germ-line clones homozygous for P1633, the P-insertional allele of rpd3. The embryo exhibits a pair-rule phenotype.