Conserved repression of Oct-1 and Oct-2 on classical monomeric octamer sequences
The lymphoid expressed proteins Oct-1 and Oct-2 have been described as repressors of transcription (9
). However, the underlying mechanism(s) for this repression has not been determined (11
). A possible clue comes from the observation that a number of other lymphoid transcription factors interact with members of the Groucho/TLE family, e.g. Pax5 (12
) and Pu.1 (13
To investigate whether Groucho/TLE proteins could function as an interaction partner for Oct proteins, we performed transient transfections using the four full-length Grg/TLE1-4 corepressors on a minimal luciferase construct located downstream of multiple copies of the standard octamer sequence. Activation of the reporters was achieved by the addition of Oct-1 () or Oct-2 () and their coactivator OBF-1. As shown in , Grg/TLE1 and 3 could functionally repress Oct-1 transcription to a much greater degree than Grg/TLE2 and Grg/TLE4. This experiment was then repeated using Oct-2 and the same pattern of repression was seen, with the exception that the limited repression seen with Grg/TLE2 on Oct-1 was not observed ().
Figure 1 Grg/TLE proteins can repress Oct-1 and Oct-2. Repression of Oct-1 (A) and Oct-2 (B) on the octamer. 293A cells were transfected with 1 μg reporter containing four copies of the consensus Octamer Sequence 5′-ATGCAAT-3′ linked to (more ...)
To confirm that the Grg/TLE expression plasmids used were functional, a control experiment using an unrelated Grg/TLE-responsive promoter was performed. A construct containing TCF-responsive elements, which are known to be repressed by all Grg/TLE proteins (14
), was analysed. As expected, the activation induced by TCF-1 together with β-catenin could be efficiently repressed with all four of the Grg/TLEs (). Finally, expression levels of the four myc-tagged Grg/TLE proteins were determined by analysing the nuclear extracts used in . As demonstrated in , all four proteins are expressed, with Grg/TLE3 being the most pronounced.
In conclusion, transcription activation by Oct-1 or Oct-2 in conjunction with OBF-1 can be prevented by Grg/TLE1 and 3. In contrast, Grg/TLE2 and 4 had a limited affect on both Oct proteins, despite of the fact that both of these repressors are functionally expressed.
Divergent responses of Oct-1 and Oct-2 to Grg/TLE on dimeric response elements
Having shown that Grg/TLE1 and 3 are functionally capable of repressing Oct-1 and Oct-2 driven transcription on an octamer reporter, we next determined whether the PORE element, which allows for dimeric Oct binding, would respond differently to Grg/TLE members. Therefore, the experiment was repeated using the PORE-D-based reporter. The PORE-D sequence differs from the PORE sequence discovered in the osteopontin enhancer in that it only supports dimeric, and not monomeric Oct binding. The wild-type PORE sequence retains some competency in supporting monomeric binding as well (6
). In contrast to the octamer reporter, only Grg/TLE2 suppressed transcriptional activation (). Importantly, in spite of the greater expression levels of Grg/TLE3 (), compared with the other three members, this was still not sufficient to facilitate repression on Oct-1 on the PORE sequence. In sharp contrast, when the experiment was repeated using Oct-2 (), a different pattern of responses was noted. Although Grg/TLE1 and 3 still repressed, in addition so could Grg/TLE4. The response profile of Oct-2 transcriptional activation on the PORE-D was, therefore, the exact opposite of Oct-1, which was only responsive to Grg/TLE2.
Figure 2 Differential repression patterns on dimeric sequences. Transient transfections on PORE-D sequence with Oct-1 (A) and Oct-2 (B). Increasing repression from monomeric to dimeric reporters (C). In this experiment, fold activation is determined with respect (more ...)
To further confirm that there is a conformational dependency between Oct-2 and Grg/TLE4, a comparison of three different promoters was used: the octamer (monomeric), the PORE (monomeric and dimeric) and the PORE-D (dimeric). If Grg/TLE4 expresses a preference for dimeric binding sites, then one would expect an increase in repression from octamer to PORE, and from PORE to PORE-D. shows that this clearly happens. Although there is some down-regulation on the octamer, as indicated in , this is greatly augmented on the PORE. Furthermore, an increase in repression is seen on the PORE-D when compared with the PORE.
In conclusion, Grg/TLE proteins show a complex pattern of repression on Oct-1 and Oct-2, depending on the binding site examined. These patterns are tabulated in . Of particular interest is that seen on the dimeric PORE-D sequence, where Grg/TLE2 can only repress Oct-1 and not Oct-2, whereas Grg/TLE4 seems only to act on Oct-2 and not Oct-1.
Finally, we wished to determine how Oct-1 would respond to Grg/TLE proteins in the context of a naturally occurring regulatory element, as opposed to the artificial MORE and PORE constructs, which contain multiple binding sites for Oct factors. We tested the four Grg/TLE corepressors on a Vh PORE region derived from the immunoglobulin heavy chain promoter locus, containing one binding site for Oct-1 or Oct-2 (4
). Surprisingly, this reporter responded differently to the multimerized constructs in that Grg/TLE1 and 3, but not Grg/TLE2, were capable of functional repression ().
The POU domain of Oct-2 interacts with the SP domain of Grg/TLE4
To further investigate the interaction between Oct-2 and Grg/TLE, we specifically assessed the conformationally recruited Grg/TLE4. The ability to distinguish between different cis
motifs suggested to us that the recruitment of Grg/TLE could occur through the POU domain. However, the N-terminal region of Oct has previously been shown as a repressor of a number of different promoters (15
). To exclude the N-terminal repression region as being the Grg/TLE interacting domain, we assessed different deletion mutants of Oct-2. Removal of the N-terminal (Oct-2ΔN) still permitted repression, as did deletion of the C-terminal (Oct-2ΔC), although to a lesser extent, while Oct-1 was still refractory the effects of Grg/TLE4 (). Therefore, the previously reported repression domain in the N-terminus of Oct-2 is unlikely to recruit corepressors of the Grg/TLE family.
Figure 3 The POU domain of Oct-2 interacts with the SP domain of Grg/TLE4. The N-terminal deletion of Oct-2 (Oct-2ΔN) lacking the first 187 amino acids of Oct-2 is efficiently repressed by Grg4 as is the C-terminal mutant Oct-2 (Oct-2ΔC) on the (more ...)
We next wished to determine whether Oct-2 and Grg/TLE complex formation occurs in vivo
. Grg/TLE proteins were immunoprecipitated from 293A cells, a non-lymphoid cell line that does not express Oct-2 but does produce Grg/TLE proteins (14
). Immunoprecipitation with a pan-Grg/TLE antibody, but not protein A alone, resulted in Oct-2 only being co-precipitated when Oct-2 had been transfected into the cell line (), indicating that over expressed Oct-2 interacts with endogenous Grg/TLE proteins. To further confirm that Oct-2ΔN could interact with Grg/TLE members, immunoprecipitation was performed again and Oct-2ΔN could indeed be precipitated with endogenously expressed Grg/TLE proteins.
The functional and immunoprecipitation data indicated an interaction between Oct-2 and Grg/TLE4; we therefore wished to determine the binding domain(s) responsible. First, Grg/TLE4 lacking various C-terminal domains were assessed for the ability to repress Oct-2/OBF-1 potentiated transcription on the PORE-D driven promoter (). The removal of the most C-terminal region, the WD40 repeats, had no effect on the repression process. However, subsequent removal of the next C-terminal region, the SP domain, alleviated the Grg/TLE4-mediated transcriptional repression. Western analysis also confirmed that these deleted mutants were stably expressed ().
indicates that the SP domain appears to be crucial for the formation of a functional repression complex with Oct-2. To determine whether the interaction between Oct-2 and Grg/TLE4 is direct, GST pull-down reactions were performed. GST, or GST fused to the POU domain of Oct-2 was used as bait for in vitro radiolabelled Grg/TLE4 and a mutant lacking the WD40 and SP domain. As shown in , GST-POU, but not GST alone could interact with full-length Grg/TLE4. In contrast, only very low levels of Grg/TLE4-ΔSP could be detected.
Grg/TLE prevents Oct-driven transcription and the interaction is evolutionarily conserved
The removal of the inhibitory N-terminal region generates an Oct-2 mutant, which is both competent for Grg/TLE-mediated repression and has greatly elevated transcriptional activation properties, as previously demonstrated (15
). This allowed us to determine whether repression could occur in the absence of OBF-1, on both the PORE-D and the OBF-1 excluding MORE motif. Using transient transfections, we demonstrated that Grg/TLE4 could efficiently extinguish OBF-1 independent, Oct-2 driven transcription on both a PORE-D and a MORE driven reporter ().
Figure 4 The evolutionary conserved interaction prevents Oct transcription. (A) Oct-2ΔN can be repressed by Grg/TLE1 on both the PORE-D and MORE sequences. (B) Transfections preformed in the presence of TSA (50 ng/ml) fails to relieve repression. (C) EMSA (more ...)
Groucho and Grg/TLE in certain situations mediate repression through interaction with Histone deacetylases (HDAC) (14
). Transfections performed in the presence of the HDAC inhibitor Trichostatin A () failed to relieve repression. Therefore, at least in the case of Oct-2 and Grg/TLE4, an HDAC-independent mechanism is in operation.
As the underlying repression mechanism seems to a level distinct from HDAC recruitment, it is possible that Grg/TLE prevents DNA binding of Oct proteins. EMSA analysis using endogenously expressed Oct-1 and transfected OBF-1 indicated that Oct-1 is fully competent in both binding DNA and forming a stable DNA/Oct-1/OBF-1 ternary complex in the presence of Grg/TLE1 (). This strongly suggests that Grg/TLE-mediated repression is not simply sequestering of Oct from its DNA target. Therefore, the Grg/TLE-mediated transcriptional repression prevents a fully active Oct-1/OBF-1 complex from successfully transcribing a reporter gene.
Oct-2 is part of large conserved family of POU domain proteins (18
), and an interaction with Drosophila
Groucho may be indicative of a greater evolutionary relationship between POU domains and Grg/TLE/Groucho corepressors. GST pull-down experiments indicate that Drosophila
Groucho can directly interact with the POU domain of Oct-2 but not with GST alone (). In concordance with this data, Groucho was able to repress Oct-2/OBF-1-mediated transcriptional activation (). Given that the interaction between Oct-2 and Grg/TLE/Groucho is evolutionary conserved, it will be interesting to determine whether other POU domain containing factors can also interact with the Grg/TLE family of corepressors.