Characterization of CAF4 and CAF16 Reveals a Functional Connection between the CCR4-NOT Complex and a Subset of SRB Proteins of the RNA Polymerase II Holoenzyme*

The CCR4-NOT transcriptional regulatory complex affects transcription both positively and negatively and consists of the following two complexes: a core 1 × 106dalton (1 MDa) complex consisting of CCR4, CAF1, and the five NOT proteins and a larger, less defined 1.9-MDa complex. We report here the identification of two new factors that associate with the CCR4-NOT proteins as follows: CAF4, a WD40-containing protein, and CAF16, a putative ABC ATPase. Whereas neither CAF4 nor CAF16 was part of the core CCR4-NOT complex, both CAF16 and CAF4 appeared to be present in the 1.9-MDa complex. CAF4 also displayed physical interactions with multiple CCR4-NOT components and with DBF2, a likely component of the 1.9-MDa complex. In addition, both CAF4 and CAF16 were found to interact in a CCR4-dependent manner with SRB9, a component of the SRB complex that is part of the yeast RNA polymerase II holoenzyme. The three related SRB proteins, SRB9, SRB10, and SRB11, were found to interact with and to coimmunoprecipitate DBF2, CAF4, CCR4, NOT2, and NOT1. Defects in SRB9 and SRB10 also affected processes at the ADH2 locus known to be controlled by components of the CCR4-NOT complex; an srb9 mutation was shown to reduceADH2 derepression and either an srb9 orsrb10 allele suppressed spt10-enhanced expression of ADH2. In addition, srb9 andsrb10 alleles increasedADR1 c -dependent ADH2expression; not4 and not5 deletions are the only other known defects that elicit this phenotype. These results suggest a close physical and functional association between components of the CCR4-NOT complexes and the SRB9, -10, and -11 components of the holoenzyme.

CCR4-NOT complexes contains CCR4, CAF1 (POP2), the five NOT proteins (NOT1-5), and two other proteins (1,3,4). 1 These proteins can all be coimmunoprecipitated with antibody specific to either CCR4, CAF1, or NOT proteins (1,4). All of the components of the 1-MDa CCR4-NOT complex also comigrate at 1.9 MDa following gel filtration analysis (1,4), 1 and mutations in individual components of the CCR4-NOT complex destroy the ability of these components to migrate at 1.9 MDa (4). 1 Although antibody directed against the core CCR4-NOT proteins does not coimmunoprecipitate any other proteins, a number of other proteins that do not coimmunoprecipitate at their physiological concentrations with CCR4 antibody do interact with the CCR4-NOT complex genetically and can be coimmunoprecipitated when overexpressed. These proteins may possibly be components of the larger 1.9-MDa complex and include such proteins as DBF2, a cell cycle-regulated protein kinase (5), MOB1, a protein that binds DBF2 and is involved in cell cycle regulation (6), and DHH1, a putative RNA helicase (7). Both DBF2 and MOB1 have also been observed to migrate at 1.9 MDa following gel filtration analysis. 1 The internal arrangement of factors in the 1-MDa CCR4-NOT complex has been studied (4). NOT1 appears to be the core component of the complex. CAF1 binds to the central region of NOT1 and links CCR4 to the rest of the NOT proteins. The C terminus of NOT1, in turn, contacts NOT2, NOT5, and NOT4. The physical separation of CCR4 and CAF1 from NOT2, NOT4, and NOT5 agrees with several phenotypic differences between these proteins (1,4). Whereas CAF1 is absolutely required for CCR4 to associate with the 1-MDa complex (1,4), CCR4 can still associate in the 1.9-MDa complex in the absence of CAF1 (1). It is likely, therefore, that within this larger complex CCR4 is making contacts to proteins other than ones found in the core 1-MDa complex. Based on the observation that CCR4 can only immunoprecipitate other components of the core 1-MDa complex, these other interactions in the 1.9-MDa complex may be more susceptible to disruption.
The CCR4-NOT proteins have been found to affect gene expression both positively and negatively (1, 8 -13). Their action as repressors are likely to be the result of the NOT proteins restricting access of TBP 2 to noncanonical TATAAs (8,9). NOT1 has been shown to associate with TBP (14); NOT5 interacts with TFIID (15,47), and NOT2 has been shown to associate with ADA2, a component of the SAGA complex (16). Consistent with these results, deletions of upstream sequences do not apparently affect the ability of CCR4 to affect ADH2 expression (17), and it has been shown that CCR4 acts at a post-chromatin remodeling step in affecting ADH2 derepression (18).
The large sizes of the CCR4-NOT complexes and the possible mode of action of these proteins at or near the TATAA suggest that these complexes would be likely to interact with other proteins acting to control initiation of transcription. In this study we report the identification of two additional factors, CAF4 and CAF16, that interact with the CCR4-NOT proteins and that affect transcription both positively and negatively. Neither CAF16 nor CAF4 was a component of the 1-MDa CCR4-NOT complex, but both proteins were present in a 1.9-MDa complex, and their presence in this complex was dependent on CCR4. CAF16 and CAF4, in turn, were found to interact with SRB9, -10, and -11, components of the RNA polymerase II holoenzyme (19 -23). SRB10 and -11 are a cyclin-dependent protein kinase/cyclin pair that are capable of phosphorylating the CTD of RNA polymerase II (21,24,25). Whereas none of these three SRB genes are essential, the SRB9 -11 proteins have been found to play both positive and negative roles in transcription (20, 26 -28), although they appear to be more predominantly involved in repressing transcription (21)(22)(23). We found that the SRB9, 10, and 11 proteins can coimmunoprecipitate the CCR4 and NOT proteins, and defects in these SRB proteins affect expression at the ADH2 locus in a manner similar to that observed for defects in CCR4 complex components. These results indicate a close physical and functional link between CCR4-NOT components and the SRB9, -10, and -11 proteins.

MATERIALS AND METHODS
Yeast Strains-Yeast strains are listed in Table I. DNA Sequencing and Analysis-CAF16 was sequenced on each strand by double-stranded sequencing using Sequenase (U. S. Biochemical Corp.). Sequence comparison analysis was performed at the National Center for Biotechnology Information, using the BLAST network service MegAlign version 1.05 (BLAST version, 1.8.1). Alignments were performed by using the Clustal method available in the DNASTAR package (DNA Star Inc.).
Gel Filtration Chromatography and Immunoprecipitation-A Superose 6 column HR10/30 was used according to the manufacturer's instructions (Amersham Pharmacia Biotech). Two hundred l of yeast extract at a concentration of about 10 mg/ml was placed over the column following clarification by centrifugation as described previously (1). Running conditions were as described (1), and the standards were eluted as follows: blue dextran (2000 kDa) at 10 ml; thyroglobulin (669 kDa) at 15 ml; bovine serum albumin (66 kDa) at 17.5 ml. Immunoprecipitations were carried out as described previously (3,5). Western analyses were conducted as described (1). Antibody to CAF16 was raised against its C-terminal peptide, CCKRDNQIPDKEIGI, whereas antibody to CAF4 was against CAF4 tagged with three copies of the HA1 epitope at its C terminus and integrated at the CAF4 locus as described (46).
Gene Disruptions and Plasmid Constructions-The caf4::URA3 disruption plasmid was created by replacing the BclI fragment of CAF4 (base pairs 851-2162) with a BamHI fragment of URA3. The CAF16 disruption plasmid was constructed by removing nucleotides 368 -706 base pairs of CAF16 following cutting with NdeI, blunt ending with the large subunit of Escherichia coli DNA polymerase (Klenow), and replacing it with a HindIII fragment of URA3 (blunt-ended with Klenow). The resultant plasmid JP7 was cut with M1uI and SstI prior to transformation to replace the chromosomal copy of CAF16.

CCR4
Interacts with Two Novel Proteins CAF4 and CAF16 -A yeast two-hybrid screen using LexA-CCR4 as a bait was conducted with a library of yeast sequences fused to the B42 activator to identify additional components of the multisubunit CCR4-NOT complexes. In addition to identifying CAF1 and DBF2 (5, 13) as components or associated factors of the CCR4-NOT complex, two additional proteins (designated CAF4 and CAF16) were found to interact with LexA-CCR4 and not with LexA alone (Table II, lines 2 and 3 compared with line 1). Deletion analysis indicated that an intact leucine-rich repeat (residues 345-470) of CCR4 was required for its interaction with both CAF4 and CAF16 (Table II, lines 7 and 8) but not the N-terminal region of CCR4 (lines 5 and 6); the leucine-rich repeat was also required for CCR4 interaction with CAF1, NOT1, and DBF2 (1, 3, 5, 13). Sequencing of CAF4 revealed it to encode a novel protein (yeast protein YKR036c) containing seven WD40 repeats in its C terminus (residues 320 -659). CAF16 when sequenced in its entirety was found to encode a protein (now designated YFL028c) that shares significant homology to the ABC ATPase family of proteins (31,32). ABC ATPases are principally found to play roles in transport across membranes and as membrane receptors (33). CAF16 differs from most of the eucaryotic ABC ATPase proteins in that it lacks the transmembrane domains characteristic of this family. One other eucaryotic ABC ATPase, EF3, involved in translational elongation also lacks these signature transmembrane domains (34). CAF16 also contains only one ABC ATPase domain, whereas most other eucaryotic ABC ATPases contain two domains, suggesting that CAF16 may interact with itself, as was confirmed by two-hybrid analysis (Table II, line 14).
CAF16 and CAF4 Physically Associate with CCR4-NOT Components in Vivo-To determine if the observed two-hybrid interactions were the result of in vivo physical association of CCR4 with CAF16 and CAF4, coimmunoprecipitation analysis was conducted. Immunoprecipitating CCR4 with anti-CCR4 antibody failed, however, to coimmunoprecipitate specifically B42-CAF16 or B42-CAF4 or their cognate unfused proteins (data not shown). These data suggest that CAF16 and CAF4 are not components of the 1-MDa CCR4-NOT complex since all of the NOT proteins and CAF1 can be immunoprecipitated with CCR4 in this core complex (1, 3, 4, 13). 1 Moreover, neither CAF16 nor CAF4 was found to be present in a purified 1-MDa CCR4-NOT complex. 1 To determine whether the CAF16 and CAF4 proteins associated in the 1.9-MDa CCR4-NOT complex, gel filtration analysis was conducted. Following Superose 6 chromatography a subset of the CAF16 protein was found to migrate at 1.9 MDa, coincident with the size of the 1.9-MDa CCR4-NOT complex ( Fig. 1, top panel). A similar analysis with CAF4 protein tagged with the HA1 epitope showed that CAF4-HA migrated at 1.9 MDa (Fig. 1, 2nd from top panel). Deletion of CCR4 was found to remove effectively CAF16 and CAF4 from the 1.9-MDa complex ( Fig. 1, top two panels). A ccr4 deletion did not have this effect on CAF1 or CAF40 (another component of the 1-MDa CCR4-NOT complex) (Fig. 1, bottom two panels), nor did it have this effect on DBF2, which is also a presumed component of the 1.9-MDa CCR4-NOT complex ( Fig. 1, middle panel). The effect of the ccr4 deletion on CAF16 and CAF4 migration at 1.9 MDa supports the physical presence of CAF16 and CAF4 in the larger CCR4-NOT complex and that CCR4 is required for these proteins to associate in this complex. The observation that a majority of the CAF16 in the cell is not apparently in the 1.9-MDa complex suggests that CAF16 may have other functions than those dealing with CCR4 or that the CAF16 association with CCR4 is unstable.
(compare with 54 units/mg ␤-galactosidase in the wild-type strain, Table II, line 2). The fact that CAF4 interacted better with CCR4 in a caf1 background suggests that CAF1 interferes with CAF4 binding to CCR4.
Phenotypic Effects of caf4 and caf16 Alleles-Disruption of the chromosomal loci of CAF4 and CAF16 was conducted, and the resultant phenotypic effects were analyzed. Unlike ccr4 or caf1 alleles neither caf4 nor caf16 affected ADH2 expression, suppressed spt10-enhanced ADH2 expression, resulted in caffeine sensitivity, produced high or low temperature sensitivity, reduced growth in the presence of metal ions, or displayed defects on growth in nonfermentative carbon sources (data not shown). Both caf4 and caf16 alleles, however, did result in increased ADH1 gene expression as did the ccr4-and caf1mutated alleles (Table III).
The caf16 and caf4 alleles also affected the expression from several lacZ reporters (1). Deleting caf16 reduced the function of HO-lacZ and FKS1-lacZ reporters by 3-4-fold while having no effect on GAL1-lacZ (data not shown). A caf4 deletion, in turn, increased GAL1-lacZ expression by 3-fold and reduced the expression of the HO-lacZ and FKS1-lacZ reporters by 2-fold (data not shown). These data confirm that CAF16 and CAF4 are required for full or proper gene expression in certain promoter contexts.
We had previously shown that caf1 and ccr4 reduce the transcriptional function of several different LexA-activators by 2-4-fold (3,13). We similarly found that the ability of a similar set of LexA activators to activate transcription was diminished 2-4-fold by caf4 and caf16 alleles (data not shown). These results confirm a role for CAF4 and CAF16 in activated transcription and are similar in degree and nature to the effects of ccr4 or caf1 disruptions on LexA activator function (3,13). Moreover, the caf4 and caf16 effects on LexA activator functions appeared to be independent of the type of activator used, suggesting they are affecting a function common to the core transcriptional process.
CAF16 and CAF4 Interact with SRBs and CCR4 Is Required for This Interaction-To further our understanding of CAF16 function, LexA-CAF16 was used as a bait in a two-hybrid screen. Only two B42 fusion proteins were identified that interacted specifically with LexA-CAF16 as follows: B42-CAF16 (data not shown; see also Table II, line 14) and B42-SRB9 (Table IV, line 1). SRB9 is a component of the RNA polymerase II holoenzyme, and various genetic and biochemical data have suggested that SRB9, -10, and -11 function closely together in a subcomplex within the holoenzyme (20,22,23). B42-SRB9 failed to interact with LexA alone (Table IV, (Table IV, lines [13][14][15][16]. The above two-hybrid results indicated a cluster of interactions in which, for example, CCR4 interacted with CAF16, CAF16 with SRB9, SRB9 with SRB10, and SRB10 with NOT1 and NOT2. Such clusters have been shown to have biological relevance and to be common in or indicative of multisubunit complexes and signaling and development pathways (35,36).
Because we had shown that CCR4 is required for CAF16 association in the 1.9-MDa complex, we tested whether CCR4 was required for CAF16 and CAF4 association with SRB9. A ccr4 disruption completely abrogated the two-hybrid interaction between B42-SRB9 and either LexA-CAF4 or LexA-CAF16 (Table V). Deleting ccr4 did not in general have this effect on two-hybrid interactions (3). For instance, LexA-CAF1 interactions with B42-DBF2 were only affected 2-fold by a ccr4 disruption (Table V), an extent expected for ccr4 effects on general transcriptional activator function (2-3-fold, Ref. 3). In contrast, a caf1 deletion did not have a similar effect on LexA-CAF4 or FIG. 2. CAF4 immunoprecipitates with DBF2 and NOT1. Yeast extracts from diploid strain EGY188/EGY191 containing LexA and B42 fusion proteins as indicated were immunoprecipitated with anti-LexA antibody, and LexA and B42 fusion proteins were detected by Western analysis following SDS-polyacrylamide gel electrophoresis using LexA and HA1 antibodies, respectively. LexA-CAF4 contained residues 61-659, LexA-SRB10, and LexA-NOT1 contained full-length SRB10 and NOT1, respectively, and B42-DBF2 and B42-SIP1 contained full-length DBF2 and SIP1, respectively. Crude extracts are represented in lanes 1-6, whereas the immunoprecipitates are depicted in lanes 7-12. Similar volumes were loaded in the SDS-polyacrylamide gel electrophoresis, but 5-fold more crude extract (Ex.) was immunoprecipitated than was analyzed in the crude extract lanes. LexA-CAF16 interactions with B42-SRB9 (Table V). Since a caf1 deletion does not completely remove CCR4 from the 1.9-MDa complex (1) and is not required for CCR4 association with either CAF4 or CAF16, these results suggest that the CAF16 and CAF4 interactions with SRB9 are occurring through the larger CCR4 complex and are mediated by or require CCR4. CCR4 Complex Components Can Be Immunoprecipitated by SRB9, -10, and 11-We subsequently examined the physical interaction between proteins associated with the CCR4 complex and the SRB9, -10, and -11 proteins using coimmunoprecipitation. Initially, to confirm the two-hybrid interactions between LexA-CAF4 and B42-SRB9, yeast extracts expressing LexA-CAF4 and either B42-SRB9 or B42-SIP1 were treated with LexA antibody, and the resulting immunoprecipitates were analyzed by Western analysis using HA1 antibody. B42-SRB9 was found to immunoprecipitate with LexA-CAF4 (Fig.  3a, lane 4), whereas B42-SIP1 failed to coimmunoprecipitate (Fig. 2, lane 7). When the immunoprecipitation was conducted from extracts expressing just LexA alone and B42-SRB9, no B42-SRB9 was coimmunoprecipitated with LexA (Fig. 3a, lane  6). Similarly, LexA-CCR4-(496 -837) and B42-SRB9 coimmunoprecipitated after extracts containing these fusions were treated with LexA antibody (Fig. 3a, lane 5).
Overproduction of CAF4 Specifically Impairs LexA-SRB11 Activation-The above results indicate that the SRB9, -10, and -11 proteins can be physically and functionally associated with the CCR4-NOT group of proteins. No synthetic phenotypes were observed, however, when these SRB genes were deleted in combination with either CCR4, CAF1, CAF4, or CAF16 (data not shown). We did observe, however, that overexpression of a C-terminal portion of CAF4 (residues 544 -659) specifically impaired the ability of LexA-SRB11 to activate the LexA-LEU2 reporter (Table VII). This CAF4 fragment had no effect on the function of LexA-CAF1, LexA-SRB9, or other LexA activators (Table VII; data not shown). Overexpression of larger fragments of CAF4 did not result in this phenotype, suggesting that residues 545-659 of CAF4 were specifically blocking an interaction of LexA-SRB11 important to its recruitment of the transcriptional machinery to the LexA-LEU2 promoter (37,38).

Identification of Two New Factors That
Associate with CCR4 -We have identified two novel factors that can physically interact with the CCR4-NOT proteins, CAF4 and CAF16. CAF4 interacted in the two-hybrid assay with CCR4 and NOT1, whereas CAF16 interacted with CCR4. These interactions were shown to be the result of in vivo physical interactions by two pieces of evidence. First, CAF4 immunoprecipitated with NOT1. Second, the ability of CAF16 and CAF4 to associate in a 1.9-MDa complex following gel filtration analysis was dependent on the presence of CCR4. These experiments suggest that CAF4 and CAF16 are associated with or components of the CCR4-NOT complex. However, neither CAF4 nor CAF16 was a component of the 1-MDa CCR4-NOT complex. In contrast, the gel chromatography results suggest that CAF16 and CAF4 are components of the 1.9-MDa CCR4-NOT complex. The ability of CAF16 and CAF4 to interact with CCR4 in the two-hybrid assay in a cafl deletion background confirms that CAF16 and CAF4 lie outside of the core 1-MDa CCR4-NOT complex. CAF4, in turn, physically interacted with DBF2 which is also known to associate with CCR4-NOT proteins but not to be present in the 1-MDa CCR4-NOT complex. The identification of DBF2, CAF4, and CAF16 in the 1.9-MDa CCR4-NOT complex will require its purification and characterization of its constituents. Our previous analysis has shown that the 1.9-MDa CCR4-NOT complex is not identical to the SRB holoenzyme (1), and others (13,40) have not identified CCR4-NOT components in the SRB polymerase II holoenzyme. A number of CCR4-NOT phenotypes were analyzed using caf4 and caf16 deletions, but in general caf4 and caf16 did not affect processes controlled by the CCR4-NOT proteins. We did observe, however, that deletions in both of these genes resulted in enhanced ADH1 gene expression, a phenotype shared by ccr4 and caf1 defects. HO-lacZ and FKS1-lacZ reporter gene expression were also reduced 2-3-fold by caf4 and caf16 deletions similar to the 2-fold differences observed for ccr4 and caf1 defects (1). These concurrences suggest that both CAF16 and CAF4 may share a subset of phenotypes with CCR4-NOT proteins.
The SRB9, -10, and -11 Proteins Interact with Components of the CCR4-NOT Complex and Associated Factors-To extend our understanding of CAF16, a two-hybrid search demonstrated that it could interact with SRB9. CAF4, in turn, was found to associate with the SRB9 and SRB10 proteins, components of the yeast RNA polymerase II holoenzyme. CCR4, CAF1, the NOT proteins, and DBF2 were subsequently shown to interact with the SRB9, -10, and -11 proteins. These interactions were indicated by two-hybrid analysis and by coimmunoprecipitation studies. For example, LexA-CAF4 and LexA-CCR4 coimmunoprecipitated B42-SRB9, whereas immunoprecipitating LexA-SRB10 brought down CCR4 and B42-DBF2, and immunoprecipitating either LexA-SRB11 or LexA-SRB9 coimmunoprecipitated NOT1, B42-NOT2, and CCR4. The two-hybrid interactions between CAF4 and CAF16 and that of SRB9 was further shown to be absolutely dependent on the presence of CCR4 in the cell (Table VI). This CCR4 dependence could be explained by either a CCR4 requirement for the integrity of the SRB9 interaction with CAF4 and CAF16 or CCR4 being an intermediary of this interaction. These observations indicate that components of CCR4-NOT complex, CAF4, and CAF16 can contact the SRB9, -10, and -11 proteins. Relatedly, the ccr4 and srb8 to -11 mutations were recently identified in causing increased expression from a defective promoter, drawing an additional link between CCR4 and these SRB proteins (39).
It is clear from these immunoprecipitation studies, however, that the strength of the interactions between SRB9, -10, and -11 and that of CCR4-NOT components and/or CAF16 and CAF4 is not strong. In most cases the interactions were only observed when the proteins were overexpressed. Interactions between large complexes such as the 1.9-MDa CCR4-NOT complex and the yeast holoenzyme might be expected to be weak and not as stable as interactions that are within the complex.
The SRB9, -10, and -11 Proteins Are Phenotypically Related to the CCR4-NOT Complex and Associated Factors-The physical interactions described above suggest that the SRB9, -10, and -11 proteins and the CCR4-NOT complex should be involved in regulating some similar processes in the cell. Defects in the SRB9, -10, and -11 factors have generally indicated that they act as repressors, although they have been shown to function as activators in some cases (20,25,37). Similarly, CCR4, CAF1, and the NOT proteins have both positive and negative effects on transcription (1), although in most cases the NOT proteins are considered as repressors (9). We found that at the ADH2 locus, SRB9 is required for full derepression of ADH2 as are CCR4, CAF1, and the NOTs (1, 10, 13). Components of the core CCR4-NOT complex and DBF2 are also the only known suppressors of spt10-enhanced ADH2 expression (1, 5, 3, 10, 13) and defects in either SRB9 or SRB10 elicited this same phenotype. In these above cases the SRB9 or SRB10 proteins function as activators, but their repressive function was also observed at the ADH2 promoter. srb9, srb10, not4, and not5 deletions all shared the phenotype of enhancing ADR1-5 c activity under glucose growth conditions. It should be noted, however, that many of the effects of srb9, ccr4, caf16, and caf4 or various processes were weak (2-4-fold effects). Although it is possible that these changes could arise from indirect effects, it should be stressed that many of the processes involved in activation/repression appear to occur through redundant mechanisms. There are several activator and repressor complexes (e.g. SAGA, TAF II s, SRBs, SNF/SWI, and SSN6-TUP1), inactivating any one of which might only impair or augment transcription to a limited extent (24,(41)(42)(43).
The ability of the srb9, -10, and -11 alleles to result in similar phenotypes as observed for defects in the CCR4-NOT complex suggests that the physical interactions between these protein groups represent shared regulatory interactions. The fact that overexpression of a C-terminal portion of CAF4 can specifically block LexA-SRB11 ability to activate a LexA-LEU2 reporter further supports the connection of these complexes and their associated factors. It has been suggested that LexA-SRB11 can activate transcription by its recruitment of the RNA polymerase II holoenzyme, of which it is a part (37,38). Overproduction of this fragment of CAF4 may interfere with this process. However, it is unlikely that CAF4 is a direct intermediary in this interaction between SRB11 and the holoenzyme. Deletion of CAF4 did not have a specific effect on LexA-SRB11 activation of transcription, for it also reduced activation by most LexA activators tested (data not shown). Conversely, deletion of srb9 or srb10 did not impair LexA-CAF4, LexA-CCR4, or LexA-CAF1 in their ability to activate (data not shown). These results suggest that in contrast to the SNF1 protein that appears to act through SRB9, -10, and -11 in affecting transcription (44), the functional interactions between the CCR4-NOT complex, CAF4, and CAF16 with the SRB 9, -10, and -11 proteins remain less clear.
Three possible models may explain the interactions between the CCR4-NOT complex and the SRB9 -11 proteins. In the first model, the SRB proteins stabilize, recruit, or otherwise influence CCR4-NOT function. For example, the CCR4-NOT proteins play a role in restricting TBP access to noncanonical TATAAs (8,15) and are involved in activated transcription under some circumstances (1). The CCR4-NOT complex may be controlled by the SRB9 -11 proteins for both these activated and repressor functions, suggesting a means by which TBP stabilization at certain promoters is connected to holoenzyme function. However, deletions of the SRB9 -11 genes did not elicit phenotypes characteristic of the not alleles at the HIS3 locus or of ccr4 or caf1 defects in terms of caffeine or glycerol B42-CAF4 ϩ temperature sensitivity. It seems unlikely, therefore, that the SRB9 -11 proteins are required for CCR4-NOT function unless it were restricted to only a few promoter contexts such as at the ADH2 locus. In the second model the CCR4-NOT proteins regulate SRB9 -11 function. This can be imagined in two ways. The activator or repressor function of the CCR4-NOT complex may exert some of its effects by affecting the SRB9 -11 subcomplex activity. The SRB10 protein may become stimulated or inhibited by the CCR4-NOT complex in its phosphorylation of its target proteins (45,48). For instance, SRB10 phosphorylation of the CTD of RNA polymerase II may be enhanced and thereby prevent efficient preinitiation complex formation as has been recently proposed (45). Alternatively, the CCR4-NOT complex could aid in recruiting or stabilizing the SRB9 -11 proteins and thereby aid in bringing the holoenzyme to the promoter. This could occur by virtue of CCR4-NOT contacts to TFIID (15,47). In this model the observation that overexpression of a segment of CAF4 impairs SRB11 function would suggest that CAF4 regulates or otherwise affects SRB11 contacts within the holoenzyme. A more detailed analysis of the effects of CCR4-NOT defects on SRB10 protein kinase activity or on the association of the SRB9 -11 proteins within the holoenzyme might illuminate these possible interactions.
A third model would suggest that the physical connections between the CCR4-NOT complex and the SRB9 -11 proteins derive from their proximity at the promoter in shared regulatory events but may not be restricted to one complex specifically controlling the function of the other. Identification of the genes controlled by the CCR4-NOT proteins using whole genome microarray analysis would be a step toward identifying the genes controlled both by the CCR4-NOT complex and by the SRB9 -11 proteins (49). If this set of genes is very small, then the two groups of proteins may be in physical contact without functional implication. In contrast, direct functional interactions would be supported by shared control of a number of genes.
Although we favor the above described second and third models, it is apparent that the ability of the CCR4-NOT complex to have multiple contacts and functions is not unique to it. The SAGA complex, the RNA polymerase II holoenzyme, TAFs, and the SNF/SWI complex all display varied roles in the cell. Identifying the connections between these large complexes of proteins should be one means of clarifying how transcriptional processes are controlled.