We have demonstrated that the hSlug protein can function as a transcriptional repressor, and the repression depends on the N-terminal half, which is separable from the DNA-binding zinc fingers. The repression has a dominant effect on neighboring activator-mediated and basal transcription. hSlug appears to be colocalized with SC-35 foci in the nucleus. Such foci have been shown to be sites of active splicing and transcription. Thus, hSlug may repress gene expression by locating itself to the target sites where active transcription occurs.
The analysis of Drosophila
Snail provided much information regarding the molecular function of this protein family. Snail binds directly to at least three target promoters and represses gene expression in the early embryo (6
). The repression domain resides in the N terminus (17
). However, this N terminus of Snail is highly divergent and is approximately twice the length of hSlug (Fig. ). Therefore, it was not clear whether hSlug could function as a repressor, or whether the two proteins use any conserved motif to repress transcription. The results presented in this paper demonstrate that hSlug not only binds to similar target sequences but also represses transcription through the N terminus. The first 32 aa of the N terminus constitute the major repressor activity, and the region contains a partial SNAG domain. It has also been shown that mSnail functions as a repressor in cultured cells and that its SNAG domain is essential (40
). However, we show here that the SNAG domain (20 aa) of hSlug is not sufficient. The most potent repression domain requires the first 32 aa, which are highly conserved among vertebrate Snail and Slug proteins. How this domain mediates repression requires further investigation.
A mechanism of repression is to heterochromatinize the target region to ensure long-range silencing for a long period of time (10
). Our results suggest that hSlug, although it can silence neighboring genes efficiently, does not seem to bring the target genes to heterochromatin domains. The staining of hSlug does not colocalize with the PML pattern, which has been shown to overlap with that of SP100 and heterochromatin protein HP1 (52
). Instead, hSlug colocalizes with the splicing and transcription regions characterized by SC-35 staining (25
). Since hSlug repression is to some extent sensitive to TSA, we postulate that the repressor may recruit HDACs to modify local chromatin as part of the mechanism to inhibit transcription.
Interestingly, structure-function analysis reveals that hSlug may contain an activation domain in the N terminus (Fig. ). Perhaps the Snail family proteins can function as activators or repressors at different target promoters (23
), depending on parameters such as neighboring cofactors and binding sequences. Some other repressor proteins also contain both activation and repression modules (20
). Whether they are artifacts of protein dissection or represent in vivo function remains unclear. This can be verified only after more direct target genes are characterized.
Another issue that awaits investigation is whether the repression and subnuclear localization of mSlug can be linked to the biological functions and to the phenotypes observed in different organisms. mSlug has been demonstrated in prolymphocytes to possess antiapoptotic activity (26
). Interestingly, the Caenorhabditis elegans
protein Ces-1, a Snail family zinc finger protein most related to Scratch, was identified as an antiapoptotic molecule (39
). It has been proposed based on genetic analysis that Ces-1 may be a repressor (39
). The idea that the antiapoptotic activity of hSlug depends on gene repression can now be tested based on the results presented in this report.
Antisense experiments in chick and frog embryos, as well as in rat bladder carcinoma NBT-II cells, showed that the other vertebrate Slug homologs may participate in controlling cell movements during embryogenesis (7
). Gene knockout experiments in the mouse, however, demonstrate that null mutations of mSlug do not lead to any morphological phenotype (31
). One possible explanation is the redundant function provided by other Snail-related proteins, as shown in frog embryos that the antisense-induced phenotype can be rescued by either Slug or Snail (7
). Molecular genetic experiments in Drosophila
also demonstrate possible redundant functions among different members of Snail family (2
). In addition to regulating cell movement, the cSnail can regulate left-right asymmetry (30
), and Drosophila
Snail family proteins have essential functions during nervous system development (2
). Details of how these proteins regulate the various biological processes are not known. However, at least in the case of Drosophila
Snail, repression of the known target genes, though essential for mesoderm specification, is not sufficient to explain the gastrulation phenotype (23
). Whether hSlug and other family members repress the same or different sets of target genes and how such regulation leads to the correct decision in various developmental and physiological processes remain to be determined.