Interaction of Short-Range Repressors with Drosophila CtBP in the Embryo

See allHide authors and affiliations

Science  03 Apr 1998:
Vol. 280, Issue 5360, pp. 101-104
DOI: 10.1126/science.280.5360.101


Human CtBP attenuates transcriptional activation and tumorigenesis mediated by the adenovirus E1A protein. The E1A sequence motif that interacts with CtBP, Pro-X-Asp-Leu-Ser-X-Lys (P-DLS-K), is present in the repression domains of two unrelated short-range repressors in Drosophila, Knirps and Snail, and is essential for the interaction of these proteins with Drosophila CtBP (dCtBP). A P-element–induced mutation in dCtBP exhibits gene-dosage interactions with a null mutation in knirps, which is consistent with the occurrence of Knirps-dCtBP interactions in vivo. These observations suggest that CtBP and dCtBP are engaged in an evolutionarily conserved mechanism of transcriptional repression, which is used in both Drosophila and mammals.

Transcriptional repression is essential for establishing localized stripes (1), bands (2), and tissue-specific patterns (3) of gene expression in the early Drosophila embryo (4). Various modes of repression have been proposed, including competitive binding of repressors to activator elements and the local quenching (inhibition) of transcriptional activators (5). Repressors can be classified according to range of action (6). Short-range repressors work over distances of less than 100 base pairs (bp) to quench upstream activators or the core transcription complex. This form of repression allows enhancers to work independently of one another to direct complex, additive patterns of gene expression, including seven-stripe patterns of eve and hairyexpression (1, 7, 8). Long-range repressors can function over distances of >1 kb to silence the transcription complex and inhibit multiple enhancers (6), thereby resulting in simple on/off patterns of gene expression.

Sequence-specific repressors can recruit co-repressor proteins to the DNA template (9). For example, two unrelated long-range repressors, Dorsal (6) and Hairy (10), recruit a common co-repressor protein, Groucho (11), which is related to a general repressor protein in yeast, Tup1p (9). Here, we show that two unrelated short-range repressors, Knirps (12) and Snail (13), recruit dCtBP, which is theDrosophila homolog of the human CtBP protein (14).

Knirps is a nuclear receptor protein that controls the segmentation of the abdomen (12), whereas Snail is a zinc finger protein (13) that establishes a boundary between the presumptive mesoderm and neurogenic ectoderm (15). Both protein sequences were used as bait in yeast two-hybrid assays (16). Knirps identified three different recombinant cDNA clones that contain overlapping versions of a common protein coding sequence; Snail identified the same coding sequence. The cDNAs that were selected by both Knirps and Snail, dCtBP, specify a putative protein of 383 amino acid residues that shares ∼60% overall identity with the corresponding human CtBP protein (Fig.1).

Figure 1

Conservation of CtBP in humans andDrosophila, as shown by alignment of the dCtBP and human CtBP protein sequences (17). Identical residues are denoted by asterisks. Most of the Drosophila sequence was derived from one of the four dCtBP cDNA clones isolated in the yeast two-hybrid assays. This particular cDNA begins at codon 8 (the proline residue); the sequence of the first seven codons was obtained from a different cDNA clone. The analysis of this latter clone suggests that the dCtBP mRNA contains a ∼300-bp 5′ UTR and an optimal translation initiation sequence. The putative dCtBP protein appears to contain 383 amino acid residues, although sequence analysis of another dCtBP cDNA suggests that there may be three additional amino acid residues, VFQ, located between codons 298 and 299. Moreover, it is conceivable that the dCtBP gene encodes differentially spliced mRNAs, which encode divergent COOH-terminal sequences beginning at codon 373 (30). The histidine residue at position 314 of CtBP corresponds to the catalytic center of d-isomer 2-hydroxy acid dehydrogenases (14). The complete dCtBP sequence can be obtained from the DDBJ/EMBL/GenBank databases (accession number AB011840).

CtBP interacts with a conserved sequence in the adenovirus E1A protein, P-DLS-K (14, 17). Mutations in this sequence eliminate E1A-CtBP interactions so that CtBP no longer inhibits E1A-mediated transcriptional activation and tumorigenesis in mammalian cell cultures (14). The P-DLS-K sequence is present in the repression domains of Knirps (12) and Snail (13); the latter protein also contains the related sequence P-DLS-R (13).

Glutathione S-transferase (GST) pull-down assays were conducted to determine whether dCtBP interacts with Knirps or Snail (18). These experiments involved the use of in vitro–translated, 35S-labeled repressor proteins and a GST-dCtBP fusion protein produced in bacteria (19). Both Knirps and Snail strongly interacted with the GST-dCtBP fusion protein (Fig. 2) but exhibited little or no binding to control GST. Mutations in the Knirps P-DLS-K motif (Gal4-knirps 75-340M) abolished interactions with GST-dCtBP (Fig. 2A). Similarly, mutations in the Snail P-DLS-R motif caused a severe reduction in Snail-dCtBP interactions (Fig. 2B). Thus, dCtBP might interact directly with Knirps and Snail, although we cannot exclude the possibility that binding is mediated by one or more proteins in the reticulocyte lysate.

Figure 2

In vitro binding assays using a GST-dCtBP fusion protein. Full-length Knirps and Snail proteins were translated with a rabbit reticulocyte lysate and labeled with [35S]methionine. Each protein was incubated with either a GST nonfusion protein or a GST-dCtBP protein containing dCtBP amino acid residues 8 to 383. (A) Knirps. Aliquots containing 10 μl of each of the indicated 35S-labeled proteins were incubated with 6 μg of either GST or the GST-dCtBP fusion protein bound to glutathione-agarose beads. Bound proteins were fractionated by SDS-PAGE and visualized by autoradiography. The input lanes contain 20% (2 μl) of each 35S-labeled protein used in the binding assays. The full-length Knirps protein exhibited efficient binding to the dCtBP-GST fusion protein (arrow). A Gal4-Knirps fusion protein containing amino acid residues 75 to 340 also exhibited efficient binding. A mutagenized derivative of this protein, 75-340M, containing alanine residues in place of the DLS sequence within the P-DLS-K motif did not exhibit specific binding to GST-dCtBP. (B) Snail. The full-length Snail protein exhibited efficient binding to the GST-dCtBP fusion protein (arrow). However, there was a substantial reduction in binding when P-DLS-R was mutagenized to A-AAA-R (to produce the Snail M2 protein). Presumably, the residual binding that is observed (arrow) is mediated by the P-DLS-K motif, which was left intact. Removal of both motifs abolished dCtBP interactions (30).

The dCtBP cDNA was used as a hybridization probe to screen an arrayedDrosophila genomic DNA library (Genome Systems Inc.). The smallest P1 recombinant phage that was identified, DS06433, maps to the 87D7-9 region of chromosome 3. This region contains a single P-element–induced lethal mutation, l(3)03463 (20). Plasmid rescue assays and polymerase chain reaction (PCR)–mediated sequence analysis suggest that this P-element maps within the 5′ untranslated region (UTR) of the dCtBP transcription unit (21). l(3)03463 homozygotes are lethal and embryos derived from germline clones exhibit severe patterning defects, including segment fusions and the loss of ventral tissues (22).

Gene dosage assays suggest that knirps and dCtBP interact in vivo. Embryos that are heterozygous for theknirps 9 (23) null mutation (knirps 9/+) exhibited occasional defects in theeve expression pattern, including reduced staining of stripe 5 (Fig. 3A). Combining the dCtBP andknirps 9 mutations [mating l(3)03463/+ females with knirps 9/+ males] resulted in more severe disruptions in the eve pattern, including the fusion (Fig.3B) or loss (Fig. 3C) of stripes 4 to 6. The latterknirps 9/dCtBP transheterozygous phenotype (Fig.3C) is virtually indistinguishable from that observed forknirps embryos (Fig. 3D).

Figure 3

Genetic interactions between dCtBP andknirps. Heterozygous females carrying the P-element–induced mutation in the dCtBP gene [l(3)03463/TM3, Sb] were mated with heterozygous males carrying a null mutation inknirps (knirps 9/TM3, Ser). Embryos were hybridized with a digoxigenin-labeled eveantisense RNA probe and visualized by histochemical staining (3). The embryos are undergoing cellularization (anterior is to the left; dorsal is up). (A) Control embryo obtained from the mating of knirps 9/TM3, Ser males with normal females. There are occasional reductions ineve stripe 5. This particular embryo lacks stripe 5 expression (arrow), which is the most extreme phenotype that is observed for such heterozygotes. (B) F1 embryo obtained by mating dCtBP heterozygous females withknirps 9 heterozygous males. This mutant phenotype [fusions of eve stripes 3, 4, and 5 (arrow)] represents the most common class of defects that were observed. (C) A different knirps 9/dCtBP transheterozygous embryo. There is almost a complete loss ofeve stripes 4, 5, and 6 (arrow). About 5 to 10% of the mutant embryos exhibited this severe phenotype. (D) Aknirps 9/knirps 9homozygous mutant embryo showing an eve staining pattern. Stripes 4, 5, and 6 are fused and reduced (arrow), similar to the pattern observed in knirps 9/dCtBP transheterozygotes [compare with (C)].

The Knirps repression domain (amino acid residues 255 to 429) contains a copy of the P-DLS-K motif. A Gal4-Kni255-429 fusion protein was expressed in ventral regions of transgenic embryos by means of a mesoderm-specific enhancer derived from the twist gene (Fig.4A) (3, 24). An evestripe 2–lacZ reporter gene was introduced into embryos expressing the Gal4-Kni255-429 fusion protein (see diagram above Fig.4A). The reporter gene contained two tandem copies of the yeast Gal4 operator site (UAS) and is normally expressed equally in dorsal and ventral regions. However, the Gal4-Kni255-429 fusion protein bound to the UAS sites and repressed stripe 2–lacZ staining in ventral regions (arrowhead, Fig. 4A). A mutant form of the Gal4-Kni255-429 protein, which contained the sequence AAAA in place of P-DL in the P-DLS-K motif, failed to mediate repression of the stripe 2 pattern in ventral regions (Fig. 4B). This result suggests that P-DLS-K is essential for Knirps-mediated repression in theDrosophila embryo.

Figure 4

The Knirps P-DLS-K motif is essential for repression in vivo. (A to C) Transgenic embryos express an eve stripe 2–lacZ reporter gene that contains two Gal4 (UAS) binding sites [see diagram above (A)]. The reporter gene was introduced into strains that express different Gal4-Knirps fusion proteins [(A) and (B)] or a Gal4-dCtBP fusion protein (C) under the control of a mesoderm-specific enhancer from thetwist promoter region (see diagrams below the panels). Reporter gene expression was visualized by hybridizing the embryos with a digoxigenin-labeled lacZ antisense RNA probe. Embryos are oriented as in Fig. 3. In (A), a lateral view of a cellularizing embryo that expresses a wild-type Gal4-Knirps 255-429 fusion protein, stripe 2 staining is selectively repressed in ventral regions (arrowhead). In (B), the conditions are as in (A) except that the Knirps fusion protein was mutagenized to convert the P-DLS-K (PMDLSMK) motif into the sequence AAAASMK. The mutant protein fails to repress eve stripe 2, so that there is uniform staining in ventral and dorsal regions [compare with (A)]. In (C), a stripe 2–lacZ staining pattern is seen in a transgenic embryo that expresses a Gal4-dCtBP fusion protein (containing amino acid residues 8 to 383 of dCtBP). Staining is reduced in ventral regions (arrowhead), although the repression is less complete than that observed for the Gal-Knirps 255-429 fusion protein [compare with (A)]. The staining in anterior, dorsal regions is attributable to vector sequences in the P-element transformation vector (7, 8). (D) Normal embryo stained with a rat dCtBP antiserum (31). Staining is primarily detected in nuclei.

To determine whether dCtBP is sufficient for transcriptional repression, we expressed a Gal4-dCtBP fusion protein in ventral regions of transgenic embryos (Fig. 4C). There was a slight reduction in the ventral expression of the stripe 2–lacZ reporter gene, although repression was not as complete as that observed with the Gal4-Kni255-429 fusion protein (Fig. 4A). These results raise the possibility that dCtBP is part of a larger co-repressor complex that assembles on the Knirps/DNA template.

Taken together, in vitro binding assays, gene dosage studies, and transgenic repression assays suggest that dCtBP is essential for Knirps-mediated repression in the early Drosophila embryo. Snail might also require dCtBP (see Fig. 2), because embryos derived from dCtBP oocytes exhibit dorsoventral patterning defects (22). The COOH-terminal repression domain of a third short-range repressor, Kruppel, contains a sequence that is related to the CtBP and dCtBP interaction sequence: P-DLS-H (25). Mutations in this sequence nearly abolish Kruppel-mediated repression in human osteocarcinoma cells (25). Thus, short-range and long-range repressors may recruit distinct co-repressor complexes; dCtBP is essential for short-range repression, whereas Groucho mediates long-range repression.

Recent studies suggest that repression involves changes in chromatin structure (26). For example, certain mammalian regulatory factors, including the thyroid hormone receptor, interact with a protein complex that includes histone deacetylases. CtBP and dCtBP may mediate repression through the enzymatic modification of chromatin because both proteins are related to d-isomer 2-hydroxy acid dehydrogenases (14). Despite this rather unexpected homology, immunolocalization assays indicate that the dCtBP protein accumulates in nuclei (Fig. 4D). Perhaps CtBP and dCtBP cause local changes in chromatin structure by introducing subtle changes in core histones. Alternatively, it is possible that CtBP and dCtBP are components of an enzymatic cascade that modulates the activities of histone deacetylases or other co-repressor proteins.

Note added in proof: dCtBP has been shown to interact with Hairy (32).


View Abstract

Stay Connected to Science

Navigate This Article