Repression of promoter activity by CNOT2, a subunit of the transcription regulatory Ccr4-not complex.

The evolutionary conserved Ccr4-Not complex controls mRNA metabolism at multiple levels in eukaryotic cells. Genetic analysis of not mutants in yeast identifies a negative role in transcription, which is dependent on core promoter structure. To obtain direct support for this we targeted individual core subunits of the human Ccr4-Not complex to promoters in transient transfections of human cells. In this experimental setup we found that the CNOT2 and CNOT9(hRcd1/hCaf40) subunits act as repressors of reporter gene activity. Interestingly, recruitment of other Ccr4-Not subunits did not affect the reporter gene. The major repression function of CNOT2 is localized in a specialized protein motif, the Not-Box. This conserved motif is present in all CNOT2 orthologs and surprisingly also in CNOT3 orthologs. Repression by the Not-Box was sensitive to treatment with the histone deacetylase inhibitor trichostatin A. In addition, mutation of a canonical TATA-box enhanced repression. Our experiments show for the first time direct regulation of promoter activity by components of the Ccr4-Not complex.

Regulation of transcription by RNA polymerase II (pol II) 1 requires the interplay of many transcription factors at different levels (1). This involves proteins recognizing gene-specific DNA elements, which can recruit transcriptional coactivator and corepressor complexes. Several of these cofactors mediate their effect via chromatin, either by ATP-dependent remodeling of nucleosomal structures or by altering the modification status of histone proteins. The latter involves phosphorylation, methylation, or acetylation/deacetylation of histones (2). The end point of transcriptional regulatory pathways is recruitment of the basal pol II transcription machinery. This machinery assembles at the core promoter of the gene, which often contains a combination of the TATA-box and DNA elements like the initiator and the downstream promoter element (3,4). Within the basal machinery two large protein complexes, TFIID and Mediator, are focal points for transcriptional regulation because they can interact with multiple gene-specific activators. Interestingly, these evolutionary conserved complexes can synergize in activation in vitro (5). The TFIID complex, consisting of TATA-binding protein in combination with 13-14 TATAbinding protein-associated proteins (6), nucleates preinitiation complex assembly by binding to the core promoter (7,8). The Mediator complex consisting of more than 20 subunits bridges activators with pol II (9 -11). In crude extracts Mediator complexes are also required for basal levels of transcription (12)(13)(14).
It has become clear that several other regulators control the basal transcription machinery at the level of the core promoter (for reviews, see Refs. 3, 15, and 16). The Ccr4-Not complex is a key representative of this class of regulators. This complex is evolutionary conserved and consists of a core of nine subunits (for reviews, see Refs. 17 and 18). The five NOT (negative on TATA-less) genes were isolated in the yeast Saccharomyces cerevisiae via genetic screens for an increased transcription from the TATA-less promoter of the HIS3 gene. Besides the Not proteins, the core Ccr4-Not complex of ϳ1.0 MDa also harbors Ccr4p, Pop2/Caf1, Caf40, and Caf130. Only Not1p is essential for yeast cell viability, but in most combinations deletion of two nonessential genes results in lethality (19). Two-hybrid interaction studies indicate that the 240-kDa Not1p protein acts as the scaffold for the complex, organizing the Ccr4p and Caf1 at a central region and the Not2p-Not3p-Not4p-Not5p module at its C terminus (17). Recent biochemical analysis revealed two different enzymatic activities associated with the core complex. First, the Ccr4p and Caf1p contain a 3Ј-5Ј exonuclease activity involved in degradation of mRNA via poly(A) tail shortening (20 -23). Second, the human ortholog of Not4p, CNOT4, contains ubiquitin protein ligase activity (24). The relevance of these findings for transcriptional regulation by the Ccr4-Not complex is yet an unresolved question.
Results from in vivo protein-DNA cross-linking studies in yeast indicate that the Not2p, Not5p, and Ccr4p subunits associate with promoter DNA, and it was suggested that the Gcn4p activator is involved in promoter recruitment of components of the Ccr4-Not complex (25,26). Several genetic and biochemical analyses in yeast provide links between Ccr4-Not proteins and the TFIID and Mediator complexes (for review, see Ref. 18). For example, epitope-tagged Not5p can retain several TATA-binding protein-associated proteins from yeast extracts (26,27). Furthermore, mutant alleles of four core subunits (NOT1, NOT3, NOT5, and CAF1) can suppress a ts allele of SRB4 (28), which encodes an essential subunit of Mediator. This finding is consistent with the proposed repressive function of the Ccr4-Not complex. However, recent analyses obtained in mammalian cells indicate that Ccr4-Not core subunits are involved in transcriptional activation. The murine CAF40 ortholog, CNOT9(mRcd1/mCaf40), is re-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. quired for retinoic acid-induced differentiation of murine teratocarcinoma cells and associates with retinoic acid receptors and the ATF-2 activator (29). In addition, expression of human CNOT6(hCCR4) or CNOT7(hCAF1) potentiates ligand-dependent activation by the estrogen receptor (30). In these cases involvement of the other Ccr4-Not subunits has not been tested.
To investigate transcriptional regulation by the Ccr4-Not subunits in a systematic manner we studied the effects of promoter recruitment of the mammalian subunits. In transient transfection assays we found that CNOT2 and CNOT9(hRcd1/ hCaf40) act as repressors. The major repression function of CNOT2 resided in a novel conserved protein motif, which we designated as the Not-Box. Mechanistic aspects of repression of the Not-Box were investigated further. Our results show for the first time that promoter targeting of Ccr4-Not subunits directly results in repression of pol II transcription.
Plasmids-The expression vector pSG424, which expresses the Gal4-DBD under control of the early promoter of SV40, has been described before (31). pSGCNOT2 was constructed by insertion of the XmaI/XbaI fragment encoding the CNOT2 open reading frame obtained by PCR using B42-hNOT2 (32) as template and the appropriate primers into pSG424. pSGCNOT3 was obtained by insertion of the BamHI/XbaI fragment encoding CNOT3L obtained by PCR using the KIAA0691 cDNA as template into pSG424. pSGCNOT4 was obtained by insertion of the BamHI/XbaI fragment encoding hNOT4N obtained by PCR using B42-hNOT4 (32) as template into pSG424. pSGCNOT8 was constructed by insertion of the BamHI/XbaI fragment encoding CNOT8 open reading frame obtained by PCR using B42-hCALIF (32) as template into pSG424. pSGCNOT1C, containing the 1,290 C-terminal residues of CNOT1, was obtained by digesting LexA-hNOT1C (32) with EcoRI and XbaI, and the insert was cloned into pSG424. pSGCNOT9 was obtained by insertion of the EcoRI/XbaI fragment encoding human CNOT9(hRcd1/hCAF40) obtained by PCR using clone BG479263 as template into pSG424. pCMVMADGal was constructed by insertion of the BglII/EcoRI fragment encoding the Sin3 interaction domain (residues 1-35) of MAD obtained by PCR, into BamHI/EcoRI-digested pcDNA3.
pCGDyNOT2 and pCGDNByNOT2 were constructed by insertion the EcoRI/XbaI fragments obtained by PCR using B42-yNOT2 (33) as template into pCMVBD. pCGDCNOT3 was constructed by insertion of the BamHI/XbaI fragment encoding CNOT3L from pSGCNOT3 into pCM-VDBD. The EcoRI/XbaI-digested fragment of encoding amino acids 661-753 of CNOT3 was cloned into pCMVDBD to obtain pCGDNBC-NOT3. All plasmid constructs were verified by DNA sequencing across the cloning junctions. Details on primer sequences are available upon request.
Immunoblot Detection of Proteins-293T cells were plated on 3-cm dishes. The cells were transfected with 1 g of DNA using FuGENE 6 (Roche Applied Science). The cells were lysed 48 h after transfection in 200 l of SDS loading buffer. 30 l of each sample was analyzed by a 10% or 12% SDS-PAGE. Proteins were detected by immunoblotting using anti-Gal4-DBD mouse monoclonal antibody RK5C1 (Santa Cruz Biotechnology) recognizing epitope 94 -147 and by using enhanced chemiluminescence (ECL, Amersham Biosciences).
Luciferase Assays-U2OS or 293T cells were plated on 3-cm dishes and transfected in duplicate or triplicate (as indicated in the figure legends) by the calcium phosphate method (36). To correct for transfection efficiency, 50 ng of pCMV-Renilla-luciferase (U2OS) or 5 ng of pCMV-Renilla-luciferase (293T) was cotransfected. For experiments using trichostatin A (TSA; ICN), transfected cells were treated with carrier (dimethyl sulfoxide) and 100, 400, or 1000 nM TSA for 17 h prior to harvesting. 40 h after transfection the cell lysates were prepared, and luciferase activity was determined using the Dual-Luciferase Reporter Assay System (Promega) and a Lumat LB9507 luminometer (Berthold) according to the manufacturer's protocol. Luciferase values were corrected for the transfection efficiency by determining the firefly luciferase:Renilla luciferase ratio. The -fold repression by Gal4-DBD fusion proteins was calculated by normalization of the firefly luciferase: Renilla luciferase ratios to the ratio obtained by the transfection containing only reporter plasmids and the empty pcDNA3.1 vector.
Sequence Alignments-Amino acid sequences were aligned using the ClustalW algorithm (37), with default settings. The sequences used for the alignment with the GenBank accession number in parentheses are: Homo sapiens Not2 (NP_055330), Anopheles gambiae Not2

RESULTS
The Ccr4-Not Subunits CNOT2 and CNOT9(hRcd1/hCAF40) Repress Reporter Gene Activity-To investigate the ability of the human Ccr4-Not complex to regulate transcription the individual subunits (CNOT2, CNOT3, CNOT4, CNOT8, CNOT9(hRcd1/hCAF40), and the C-terminal 1,290 residues of CNOT1) were targeted to a pol II promoter by fusing each subunit to the Gal4(1-147) DBD. As a positive control a construct expressing the Sin3 interaction domain, amino acids 1-35, of Mad1 fused to the Gal4(1-94) DBD was used (38). The Gal4 expression plasmids were cotransfected into human osteosarcoma U2OS cells with the firefly luciferase reporter plasmid 7xGalTKLuc, which contains seven Gal4 binding sites fused to the herpes simplex virus thymidine kinase (TK) promoter, or with the control reporter plasmid TKLuc lacking any Gal4 binding sites. To correct for differences in transfection efficiency a plasmid expressing the Renilla luciferase gene under control of the strong CMV promoter was also included in each transfection. Fig. 1 shows that cotransfection of the Gal4-CNOT2 and Gal4-CNOT9 plasmids resulted in 8-and 6-fold reduction of luciferase activity, respectively. These effects were dependent on the presence of Gal4 binding sites in the promoter. Luciferase expression from the TKLuc control was only weakly affected by Gal4-CNOT2 (1.5-fold) or Gal4-CNOT9 (2.1-fold). As expected, cotransfection of Gal4-MAD plasmid reduced luciferase expression. Analysis of transfected cell lysates by immunoblotting indicated that all Gal4 fusion proteins are expressed and of the expected size (data not shown). Repression by CNOT2 was also observed using ts20 (Chinese hamster lung), 911 (adenovirus-transformed human embryonic retina), and 293T (adenovirus-transformed human embryonic kidney) cell lines (data not shown).
Our experiments showed that only two of the six Ccr4-Not subunits tested here affect promoter activity. Others found that artificial recruitment of CNOT6 (hCCR4) also does not influence TK promoter activity (39). Although we did not test the CNOT7 (hCAF1) subunit, based on our results with its paralog CNOT8 (hPOP2/CALIF) we expect that CNOT7 will also not affect promoter activity. In conclusion, we observed that promoter targeting of the CNOT2 and CNOT9 subunits of the Ccr4-Not complex resulted in a strong repression of reporter gene activity.
The Conserved C-terminal Part of CNOT2 Contains the Major Repression Function-To investigate further the strong repressive effect on transcription by CNOT2 we decided to map the region responsible for this. The human CNOT2 gene displays a strong sequence homology with the yeast ortholog Not2p in the last 190 residues (32). In addition, two short amino acid stretches (258 -285 and 333-365) of CNOT2 share homology with orthologs from other metazoans but are absent in yeast orthologs. We constructed CMV promoter-based expression plasmids for CNOT2 deletions in the context of Gal4 fusions as indicated ( Fig. 2A). The various Gal4-CNOT2 constructs were transiently transfected into U2OS cells together with the 7xGalTKLuc reporter plasmid. As expected full-length CNOT2 strongly repressed luciferase activity (9.1-fold). Constructs only expressing the nonconserved N-terminal part of CNOT2 did not repress the reporter. On the other hand the C-terminal constructs, spanning amino acid regions 256 -540 or 334 -540, repressed transcription between 8-and 10-fold. Progressive truncations indicated that the CNOT2(437-540) region is minimally required to observe the repressive effect, although it is reduced to 5-fold. Truncating 46 residues from the C-terminal end of CNOT2 completely abolished repression of the reporter as evidenced by the CNOT2(334 -494) construct. Immunoblot analysis of transfected 293T cell lysates using a monoclonal antibody specific for Gal4 indicated that all proteins were expressed and of the expected size (Fig. 2B). Although some Gal4 fusions were expressed at higher levels than others, this variation did not correlate to repression properties. For example, the CNOT2(386 -540) fusion is expressed to lower levels than CNOT2(1-225) or CNOT2(349 -540) but showed a strong reduction in reporter activity. In conclusion, we found that the major repression function of CNOT2 is contained in the last 104 residues of the C-terminal conserved region of the protein.
Sequences Similar to the CNOT2 Region Responsible for Repression Are Also Present in CNOT3 Orthologs-Analysis of the genomic data bases for proteins containing sequences homologous to the minimal repression domain of CNOT2 revealed that this region is not unique to CNOT2 and its orthologs. Surpris-FIG. 1. Promoter-targeted CNOT2 and CNOT9(hRcd1/hCAF40) repress expression of a reporter gene. U2OS cells were transiently cotransfected with a luciferase reporter construct (7xGalTKLuc or TKLuc, containing or lacking Gal4 binding sites, respectively) and an expression plasmid of the human Ccr4-Not subunits as indicated. pC-MVMADGal was included as a positive control for repression. The TK promoter contains region Ϫ108 to ϩ19, which encompasses the core promoter and Sp1 and CCAAT binding transcription factor binding sites (53). Luciferase values were corrected for the transfection efficiency by determining the firefly luciferase:Renilla luciferase ratio. The graph presents the -fold repression by Gal4-DBD fusion proteins and is relative to the ratio obtained for the pcDNA3.1 empty vector control transfection. Each transfection was performed in triplicate, and the standard error is indicated. The experiment shown is representative of at least three independent ones. ingly, we found that CNOT3 orthologs (including yeast Not3p and Not5p, which could be considered paralogs; see Ref. 17) also contain sequences at their C termini, which are significantly similar to region 437-540 of CNOT2. Because these sequences are only present in the core subunits of the Ccr4-Not complex, we tentatively named this protein motif the Not-Box, which encompasses residues 438 -524 of CNOT2. Fig. 3A shows the alignments of this region with human CNOT3 and with orthologs of CNOT2 and CNOT3 from the mosquito A. gambiae, the fruit fly D. melanogaster, the fission yeast S. pombe, the nematode C. elegans, and the budding yeast S. cerevisiae. Orthologs from other mammals such as mouse and monkey were not included to prevent a bias toward higher eukaryotes. Fig.  3B depicts the percentages of identity between the Not-Boxes in a pairwise comparison. The identity between Not-Boxes from CNOT2 orthologs ranges between of 32 and 74% (depicted in the upper left quadrant). The identity between Not-Boxes of CNOT3 orthologs varies from 8 to 91% (depicted in the lower right quadrant). Clearly, yeast Not3p and Not5p show the lowest percentage identity, and when they are excluded the lower limit is 62%. Comparison of the Not-Box of the CNOT2 and CNOT3 proteins within one species indicates that the identity ranges between 19 and 38% (depicted in bold). Clearly, Not-Boxes of CNOT2 orthologs are more similar among each other than when compared with Not-Boxes from CNOT3 orthologs and vice versa.
Taken together, we found that the region of CNOT2 capable of transcription repression is not only conserved between CNOT2 orthologs, but is also present in the C-terminal regions of CNOT3 orthologs. In addition, the sequence comparisons indicated that these Not-Boxes were separated into CNOT2 and CNOT3 types of Not-Boxes.
The Not-Box Is an Autonomous Transcription Repression Domain in Mammalian Cells-The finding that the C-terminal part of human CNOT2 can repress reporter gene activity was surprising in light of the findings of Struhl and co-workers (40), who reported that promoter targeting of yeast Not2p resulted in strong transcriptional activation of a yeast promoter. We also noted that fusion of human CNOT2 to the LexA DBD could strongly activate a promoter bearing LexA binding sites in yeast (32). To investigate this issue further we decided to test the transcriptional effect of yeast Not2p and its isolated Not-Box in human cells. In addition, we decided to analyze transcriptional properties of the isolated Not-Box of CNOT3 as we found that promoter targeting of full-length CNOT3 did not alter reporter gene activity (Fig. 1).
To this end gene fusions were constructed of the Gal4(1-147)-DBD with full-length yeast Not2p, CNOT3, the Not-Boxes of yNot2p (residues 96 -191) or of CNOT3 (residues 661-753) under control of the CMV promoter. The effects of these expression plasmids on reporter gene activity were tested by transfection into U2OS cells. Interestingly, a clear repression of reporter activity, which was dependent on the presence of Gal4 binding sites, was observed for both full-length yNot2p and by its isolated Not-Box (Fig. 4A, compare 7xGalTKLuc with TK-Luc). In contrast to full-length CNOT3, the isolated Not-Box from CNOT3 reduces expression from the Gal4 binding site reporter gene (7xGalTKLuc). We repeated this experiment in 293T human embryonic kidney cells and also observed Gal4 binding site-dependent repression by yeast Not2 (data not shown). However, in this cell line expression of the fusion of Gal4 with yeast Not2 also represses the TKLuc reporter construct lacking Gal4 binding sites. The reason for this is unclear to us. Immunoblot analysis of the transfected cell lysates indicated that proteins of the expected size were expressed (Fig.  4B). This also showed that the inability of the full-length CNOT3 fusion to repress transcription is not caused by low levels of expression as it is expressed to higher levels than the NOT2 fusion proteins (see also Supplemental Fig. 1).
In conclusion, our experiments showed that the isolated Not-Boxes from yNot2 and CNOT3 also repressed gene activity upon promoter targeting.
Not-Box-mediated Repression Is Sensitive to the Histone Deacetylase Inhibitor TSA-Our experiments suggest that the Not-Box motif as present in CNOT2 and CNOT3 orthologs can actively repress transcription. At present it is not evident how the Ccr4-Not complex is recruited to promoters, but its effect is reminiscent of the action of other repressors (e.g. MAD), which in certain cases can recruit histone deacetylases (41). We decided to test whether repression by CNOT2 is sensitive to the histone deacetylase inhibitor TSA. Gal4 fusion constructs for full-length CNOT2, its isolated Not-Box, or the unrelated Wilms' tumor 1 (WT1) repressors were cotransfected with the luciferase reporter into 293T cells. Gal4-WT1 served as a neg- ative control, as we observed previously that transcription repression by WT1 is insensitive to TSA (34). Different concentrations of TSA were added to cell cultures prior to harvesting transfected cells. Interestingly, TSA treatment reduced the repression by Gal4-CNOT2 from 11-fold to 6.5-fold (Fig. 5). The repression by the Not-Box of CNOT2 is reduced from 5.8-fold to 1.8-fold. This effect required 400 nM TSA. Repression by WT1 was not affected by the addition of TSA, which indicates that TSA specifically affected CNOT2-mediated repression. Immunoblotting of transfected cell lysates showed that Gal4 proteins are expressed at slightly higher levels in TSA-treated cells (Fig.  5B). This indicated that the TSA-mediated reduction in repression is not caused by a reduced expression of Gal4 proteins. Taken together, these results showed that transcriptional repression by the Not-Box of CNOT2 is TSA-sensitive, which suggests involvement of histone deacetylases.
Repression by CNOT2 Is Affected by the Sequence of the Core Promoter-Genetic experiments in yeast indicate that the NOT genes specifically repress transcription from weak TATA elements. To investigate TATA dependence in our system we analyzed CNOT2-mediated repression on different core promoters. We found that a synthetic promoter, containing five Gal4 binding sites fused to the consensus TATA-box from the adenovirus major late promoter and the retinoic acid receptor-␤ initiator element, was also repressed by CNOT2 (Fig. 6). Mutation of the second residue of the TATA-box (5xGal.TGTA-AAA.Luc) enhanced repression about 2-fold by full-length CNOT2, MAD, or the isolated Gal4-DBD. However, repression by the Not-Box of CNOT2 increased 8-fold. Interestingly, repression by the isolated Not-Box became as strong as repression by full-length CNOT2. We also analyzed repression of other core promoters by targeting of CNOT2 and found that the adenovirus E1B core promoter was also repressed, but that, for example, the T-cell receptor V␤ 8.1 core promoter was not affected (data not shown). Whereas E1B contains a canonical TATA-box, the V␤ 8.1 core promoter has a nonconsensus GATAAAA sequence around position Ϫ30 (42), which indicates that elements outside the TATA-box may also contribute to the effect of CNOT2.
Taken together, these results indicate that core promoter sequences play an important role in repression by the Not-Box of CNOT2 in mammalian cells.

DISCUSSION
Components of the Ccr4-Not complex regulate cellular mRNA metabolism at the levels of synthesis and degradation (for reviews, see Refs. 17 and 18). The initial genetic identification of the Not components indicated that they act as repressors of RNA pol II-mediated transcription. At present it is not known how the Ccr4-Not complex is recruited to pol II promoters. To study transcriptional regulation by Ccr4-Not we circumvented this issue by fusing individual subunits of the human complex to the DBD of the Gal4 transcription factor and tested them for transcription regulation in human cells. We showed that components of the Ccr4-Not complex could actively repress gene expression. In transient transfection assays using various human cell lines a strong repression by the CNOT2 and CNOT9(hRcd1/hCaf40) subunits was observed (Fig. 1). We found that its major repression function was localized in the conserved C-terminal 104 residues of CNOT2 (Fig. 2). We named this protein motif the Not-Box and found that its transcriptional repression function was sensitive to the histone deacetylase inhibitor TSA and was dependent on core promoter sequences (Figs. 5 and 6).
Repression and Activation by Not2 Proteins-Although we observed strong transcriptional repression by Not2 proteins in human cells, the situation is clearly different in yeast cells. Promoter targeting of human CNOT2 or yeast Not2p by the LexA DBD leads to strong transcriptional activation in yeast (32,40,43). In this respect it is important to note that the relatively weak TK promoter used in most of our experiments would allow detection of both promoter activation and repression, whereas only activation can be observed in the yeast experiments. Possibly, the activation function is not revealed in human cells. Another indication that transactivation and repression are separate functions is the observation that activation in yeast by yNot2p maps to its first 102 residues (40), whereas the repressive Not-Box encompasses residues 107-191 (Fig. 4). Initial analysis of Not2p-mediated activation in yeast revealed a correlation with binding to the evolutionary conserved Ada2p (40), which is a subunit of different Gcn5p-containing histone acetyltransferase complexes (44). However, analysis of different not2 alleles indicated that transactivation did not correlate well with Ada2p binding (43). Although his-tone acetyltransferase activity has mostly been implicated in activation of transcription, SAGA components can be involved in transcriptional repression as well (45). Therefore, it would be interesting to test whether the interaction between Not2 and Ada2 proteins is evolutionary conserved and whether histone acetyltransferase complexes are involved in CNOT2-mediated repression in human cells. However, the results with TSA (Fig.  5) indicate involvement of histone deacetylases. It is important to note that inhibition of repression by the Not-Box of CNOT2 requires 400 nM TSA. In our experiments we noted that TSA treatment induced morphological changes of the cells. It is also known that prolonged TSA treatment can have adverse effects on cell cycle progression (46). Interestingly, the observation that subcellular localization of the CNOT7(hCAF1) protein is cell cycle-regulated (30) suggests that Ccr4-Not function may also be regulated in the cell cycle. In this way the observed TSA sensitivity could represent an indirect effect on CNOT2-mediated repression rather than direct involvement of histone deacetylases. Clearly, further experiments are required to establish a direct connection between CNOT2-dependent repression and histone deacetylase function.
Does the Not-Box Function as an Independent Module?-Experiments with yeast Not2p indicated that integrity of its Not-Box is not essential for either interaction with Not5p or for stable association with the complex (43). This suggests that the Not-Box may act as an independent module with other regulatory proteins to mediate transcriptional repression. Several observations support this. First, TSA treatment of cells relieves repression by full-length CNOT2 protein to a lesser extent than repression by the isolated Not-Box (Fig. 5). Second, whereas the full-length CNOT3 protein affects reporter activity very weakly (if at all), extraction of its Not-Box creates a repressive module. However, although the major repression function is localized in the Not-Box, it is less potent compared with repression by the full-length Not2 proteins. These observations indicate that an additional repression function may reside outside the Not-Box, and this could involve (other subunits of) the Ccr4-Not complex.
In case the Not-Box functions independently of the rest of the Ccr4-Not complex, it becomes important to identify interacting proteins. Interestingly, Tamura and co-workers (47) identified an interaction partner of the Not-Box of CNOT3, the TIP120B protein. TIP120B is a muscle-specific paralog of TIP120A, which was identified as a TATA-binding protein-interacting protein (48,49). The genetic specificity of the yeast Not proteins for the TATA-less promoter of the HIS3 gene and our observation that mutation of the TATA-box enhances transcription repression by the Not-Box (Fig. 6) make the TIP120B protein a potential regulator or mediator of CNOT3 Not-Box function.
Involvement of the Other Ccr4-Not Subunits in Repression by CNOT2 and CNOT9 -We speculated previously that ubiquitination of TFIID by the E3 ligase activity of CNOT4 may be involved in transcription regulation by the Ccr4-Not complex (24). Several observations make it unlikely that ubiquitination plays a role in the observed Not-Box-mediated repression as reported here. First, coexpression of CNOT4 or with a mutant unable to interact with E2 enzymes did not affect CNOT2mediated repression (data not shown). Second, coexpression of CNOT3 or CNOT4 with Gal4-tagged CNOT1C did not result in transcriptional repression (data not shown). And third, we examined repression by CNOT2 in the ts20 cell line, which harbors a ts version of the E1 enzyme. At the restrictive temperature ubiquitination is blocked, but CNOT2 function did not seem to be affected (data not shown).
Although genetic characterization of Ccr4-Not subunits in- FIG. 6. Mutation of the TATA-box affects the transcriptional repression by CNOT2 protein. 293T cells were transiently transfected with luciferase reporter construct, containing either wild type or mutated adenovirus major late TATA-box (indicated here as 5xGal. TATAAAA.Luc and 5xGal.TGTAAAA.Luc) and expression constructs for the Gal4 fusion proteins as indicated. Luciferase values were corrected as described in the legend of Fig. 1. The basal activity of the mutant TATA luciferase construct is comparable with the wild type. The graph presents the -fold repression by Gal4-DBD fusion proteins and is relative to the pcDNA3.1 empty vector control transfection. Each transfection was performed in triplicate. The experiment shown is representative of at least three independent ones. dicated that they have overlapping but not identical phenotypes (17), it is surprising that repression is not observed with all core subunits. CNOT7 and CNOT6 are involved in mRNA degradation in the cytoplasm (20,50). However, this does not exclude a nuclear role for these proteins. A significant portion of CNOT6 and CNOT7 proteins is associated with the nuclear form of the human Ccr4-Not complex (30,51). Also, overexpression of human CNOT6 and CNOT7 has been shown to support estrogen-dependent transcription (30). Several explanations are possible for the observations that individual Ccr4-Not components do not show a similar behavior in different assays. The individual proteins may have distinct functional properties outside of the Ccr4-Not complex. In fact, gel filtration analyses of both yeast and human cell extracts indicate that significant amounts of CNOT6/Ccr4p and CNOT7/Caf1p proteins do not coelute with Not proteins (30,52). Second, it is also possible that the Ccr4-Not complex can exert control both at the mRNA synthesis level by promoter interactions and at the mRNA degradation level by direct associations with mRNA itself (18). Alternatively, the core Ccr4-Not complex may represent a regulatory platform, which associates with other proteins to carry out specific functions, as with the Dhh1p subunit of the decapping complex for mRNA degradation. 2 How does this help an understanding of the selective repression by the CNOT2 and CNOT9 subunits in our assays? Either these proteins carry out specific functions independently of the other core subunits or CNOT2 and CNOT9 are the limiting components in our transient transfections. In this latter scenario targeting of these subunits would result in recruitment of the complete Ccr4-Not complex(es) to the promoter. This predicts that cotransfection of other subunits may either enhance or abolish repression. This was not observed (data not shown), which suggests that CNOT2 and CNOT9 mediate transcriptional repression without involvement of other core subunits. Rigorous testing of this hypothesis requires development of in vitro assays. The results presented in this study provide the framework for these in vitro studies, which should lead to elucidation of the mechanism of repression by CNOT2 and CNOT9 and a more detailed understanding of transcription regulation by the Ccr4-Not complex.