Introduction

The CCR4-NOT complex is involved in multiple aspects of mRNA metabolism and is composed of nine subunits: CCR4, CAF1, NOT1-5, CAF40, and CAF130 (CAF1 is also known as POP2, NOT1 as CDC36, and NOT2 as CDC39) (Collart and Struhl 1994; Draper et al. 1995; Bai et al. 1999; Chen et al. 2001; Sakai et al 1992). While the CCR4 and CAF1 components of this complex were originally identified as being activators of mRNA expression (Denis 1984; Sakai et al. 1992), the NOT proteins clearly function as transcriptional repressors, presumably by restricting TFIID access to promoters (Collart and Struhl 1993, 1994; Collart 1996; Badarinarayana et al. 2000; Deluen et al. 2002; Collart and Timmers 2004). More recently, the CCR4-NOT complex has been implicated in transcriptional elongation (Denis et al. 2001) and, importantly, has been shown to comprise the cytoplasmic mRNA deadenylase complex required for mRNA degradation (Daugeron et al. 2001; Tucker et al. 2001; Chen et al. 2002; Tucker et al. 2002). Biochemical and genetic studies have implicated CCR4 and CAF1 as playing the most important roles for mRNA deadenylation, the NOTs as most critical for transcriptional repression, and CAF40 and CAF130 as having little or no effect on either of these processes. These results along with an analysis of the arrangement of the physical positioning of components within the complex (Bai et al. 1999; Chen et al. 2001) has led to the model that the CCR4-NOT complex consists primarily of two modules: CCR4/CAF1 and NOTs.

As expected from a group of proteins that affects different aspects of mRNA metabolism, CCR4-NOT components have been shown to be involved in the control of a number of important pathways and functions within the cell. CCR4-NOT components are required for the mRNA expression of a number of non-fermentative genes (Denis 1984; Denis and Malvar 1990; Sakai et al. 1992; Liu et al. 1998), and thereby control carbon intermediary metabolism, as well as the more general stress response (Lenssen et al. 2002; Collart and Timmers 2004) that includes non-fermentative growth (Gasch et al. 2000). CCR4 and CAF1 exert effects on the cell cycle through interactions with DBF2 (Liu et al. 1998) that is critical for late mitotic processes (Toyn and Johnston 1994), are involved in the control of cell size (Toyn and Johnston 1994; Jorgensen et al. 2002, 2004) and ribosomal gene expression (Grigull et al. 2004), an important determinant of cell size regulation (Jorgensen et al. 2004). The complex has also been found to be important to processes involving DNA repair (Schild 1995; Jelinsky and Samson 1999; Hanway et al. 2002; Traven et al. 2005), histone methylation (Laribee et al 2007; Mulder et al. 2007a) and protein degradation, as NOT4 is an E3 ubiquitin ligase (Albert et al. 2002; Mulder et al. 2007b). These and other pathways under the specific influence of the CCR4-NOT complex suggest that this complex will mediate a number of signaling pathways in the cell.

Because previous analyses have focused on the effects of CCR4-NOT components on a very small set of genes, the correspondences and overlaps between these factors in terms of their actual biochemical roles remains to be determined. In order to clarify the functional overlap of the various CCR4-NOT components and especially to understand the roles of the CAF40 and CAF130 proteins and the relationships between the CCR4/CAF1 and NOT modules, we have conducted whole genome microarray analysis of deletions in seven of the nine genes that encode components of the CCR4-NOT complex. We find that CCR4 and CAF1 proteins under glucose growth conditions alter the patterns of gene expression in very similar ways whereas the NOT3/NOT4/NOT5 proteins act in a similar way and, in particular, control SAGA-responsive genes. CAF40 and CAF130 effects are fairly distinct from either the CCR4/CAF1 or NOTs module. Under non-fermentative growth conditions, a stress condition, all the components share a more common profile on their effects on gene expression. We have also used microarray and immunoprecipitation analysis to confirm that BTT1, a member of the nascent polypeptide complex, is a tenth component of the CCR4-NOT complex. BTT1 along with CAF130 appear to define a third module of the CCR4-NOT complex that specifically affects expression of genes involved in ribosome biogenesis.

Materials and methods

Yeast strains and growth conditions

Yeast strains are listed in Table 1. Yeast were grown on YEP medium (1% yeast extract/2% bactopeptone) supplemented with 4% glucose or 2% ethanol/2% glycerol. Following overnight growth in the indicated media at 30°C, strains were diluted to OD600 of 0.05 and allowed to grow at 30°C until the cultures reached mid-log phase growth (OD600 = 0.5). Duplicates were prepared for each strain at each growth condition.

Table 1 List of yeast strains
Full size table

RNA isolation, cDNA labeling and microarray hybridization

Total RNA was isolated by the hot acid phenol method (Viswanathan et al. 2004). Further purification was conducted by using RNeasy RNA purification kit (Qiagen). Fluorescently labeled cDNA was generated by a two-step method. Firstly, amino-allyl dUTP (Sigma) was incorporated into cDNA by reverse transcription of 20 mg total RNA using 2 mg oligo dT primers plus 500 mM each dATP, dCTP, dGTP, 300 mM dTTP and 200 mM amino-allyl dUTP. After cDNA synthesis, the RNA was hydrolyzed by bringing the reactions to a final concentration of 100 mM NaOH and 10 mM EDTA and incubating at 65°C for 10 min. The resulting cDNA was then put through a MicroCon-30 column (Millipore). Secondly, purified cDNA was fluorescently labeled by coupling with Cy3 or Cy5 monofunctional dye (Amersham), in the presence of sodium bicarbonate buffer pH 9.0, for 60 min at room temperature. Uncoupled dyes were removed by using the QiaQuick (Qiagen) PCR purification columns. Cy3- and Cy5-labled cDNA probes were combined together applied to yeast cDNA arrays (The Microarray Centre, University Health Network, Toronto) in 1X DIG easy hyb solution (Roche) and 500 mg/ml poly-A RNA at 42°C for 15 h. After washing, hybridized slides were scanned using a GenePix scanner (Axon Instruments) at wavelengths of 635 and 532 nm respectively. Each microarray hybridization experiment was repeated twice.

Data acquiring and processing

Images were acquired by using GenePix Pro 6.0 software (Axon Instruments). The quality of individual spots on each microarray chip was evaluated based on the size of spot, and background level. The spots with obviously small size or with intensity close to background were discarded. Data were then extracted from the image. A global normalization method was employed with each data set based on the assumption that most genes in the cell will not be differentially expressed, and therefore the arithmetic mean of the ratios from every spots on a given microarray chip should be equal to 1. The microarray data and subsequent microarray analysis is available as supplementary data.

Linear regression analysis between each pair of mutants were performed using Microsoft Excel and statistical software S-Plus. Hierarchical clustering analysis was done using Cluster 3.0 (De Hoon et al. 2004). The similarity matrix was centered correlation and the clustering method was complete linkage. The heatmaps were generated using Java TreeView (Saldanha 2004).

Other techniques

Immunoprecipitations and western analyses were conducted as described previously (Chen et al. 2001).

Results

Genome wide expression analysis confirms the functional relationships between components of the CCR4-NOT complex

Whole genome microarray analysis was utilized to determine the functional relationships between components of the CCR4-NOT complex. Deletions in seven of the nine components of the complex were analyzed. not1 was not analyzed as its deletion is lethal, and we were unable to obtain a not2 deletion in our strain background. However, previous results suggest that NOT2 and NOT5 act similarly in the control of gene expression (Bai et al. 1999; Russell et al. 2002; Collart and Timmers 2004). For each experimental determination total RNA was isolated from each of the seven deletion strains and compared to RNA isolated from its isogenic parent. Fluorescently labeled cDNA was synthesized using Cy3 for all experimental samples and Cy5 for all controls (Gasch et al. 2000) and the relative RNA levels for all yeast genes were determined using a GenePix Axon Microarray Scanner and its available software. RNA levels for all yeast genes were determined under glucose conditions (fermentative) and also following growth on glycerol/ethanol-containing medium (non-fermentative) because of the well established role of the CCR4-NOT complex in non-fermentative growth (Denis 1984; Liu et al. 1998). Non-fermentative growth is also known to elicit the stress response (Gasch et al. 2000) in which the CCR4-NOT complex is involved (Lenssen et al. 2002). For each deletion, microarray analysis was conducted in duplicate, and all values presented represent the average of these two determinations.

Our initial focus was to determine the overall level of similarity between the genes affected by any one of the CCR4-NOT factors and those affected by any other component of the complex. This analysis was conducted by correlating the effect of deleting a CCR4-NOT gene on yeast gene expression with that of deleting another component of the complex. In this comparative analysis, only those genes whose gene expression were consistently affected by the CCR4-NOT deletion for each pair of samples (relative standard deviation, RSD, values of 1.6 or less) were considered for further analysis and labeled “good” genes. The number of good genes for each CCR4-NOT deletion ranged from 3,000 to 4,600 of the 6,000 total yeast genes represented on the microarray chips (see Table 2). However, the trends that are described below remained the same when the total 6,200 yeast genes were analyzed (see Fig. 3, below). As shown in Figs. 1 and 2a, ccr4 and caf1 displayed a correlation coefficient of 0.70 for the 1,887 genes that were shared between their two sets of good genes. Similar analyses comparing ccr4 and caf1 from a different set of isogenic strains (EGY188) resulted in a correlation coefficient of 0.74 for 2312 good genes (data not shown). These data indicate that CCR4 and CAF1 are very similar in their global effects on mRNA expression, albeit not identical.

Table 2 Percentage of genes whose expression is affected twofold by CCR4-NOT genes
Full size table
Fig. 1
figure 1

Representative scatter plot analyses of CCR4-NOT effects on gene expression under glucose growth conditions. Top left This scatter plot describes the difference in expression of genes (on a log 2 scale) in a ccr4 background as compared to that in a caf1 background relative to the wild-type strain as determined by microarray analysis. Of the 3,163 good genes affected by ccr4 and the 3,053 good genes affected by caf1, 1,887 were shared between the two sets. The effects on the gene expression of these 1,887 genes are plotted for ccr4 and caf1. Top right Scatter plot comparison of not4 and not5 effects on gene expression. There were 2,512 shared good genes between not4 and not5. Bottom left Comparison of not5 and ccr4 (2,203 genes). Bottom right Comparison of isogenic caf130- and btt1-containing strains from JF13 background (3,288 genes)

Full size image
Fig. 2
figure 2

Correlation coefficients between CCR4-NOT genes for their effects on gene expression. For each pair of CCR4-NOT genes, the correlation coefficient is presented comparing the pair’s effects on gene expression. Below each correlation coefficient in parenthesis is given the number of shared good genes between each CCR4-NOT pair. a Glucose growth conditions. b Non-fermentative growth conditions. 1 caf130-containing strain was isogenic to FY1642. 2 caf130-containing strain was isogenic to JF13

Full size image

In contrast to the shared effects of ccr4 and caf1 on yeast gene expression, as shown in Figs. 1 and 2a, neither of these two deletions’ effects on global gene expression under glucose growth conditions correlated with that of deletions in not3, not4, or not5. On the other hand, not3, not4, and not5 correlated very well with each other (Figs. 1, 2a). These data confirm the previously well described physical and functional separations of the CCR4/CAF1 and NOT modules within the CCR4-NOT complex (Bai et al. 1999; Chen et al. 2001; Maillet and Collart 2002; Tucker et al. 2002) and strongly argue that the CCR4-NOT complex, at least under glucose growth conditions, contains two primarily separate functional groups of proteins.

In addition to clarifying the functional separation of CCR4/CAF1 from the NOTs, we undertook the analysis of the caf40 and caf130 deletions to better understand their roles within the CCR4-NOT complex. The caf40 and caf130 deletions have yet to display phenotypes correlated with defects in other CCR4-NOT components (Chen et al. 2001). A caf130 deletion did not correlate at all with ccr4 or caf1 and correlated slightly with the nots (Fig. 2a). Similarly, caf40 displayed some slight correlation with ccr4 and not3. The known lack of CCR4-NOT-like phenotypes for caf40 and caf130 was supported by the microarray analysis. CAF40 and CAF130 are likely contributing to unique processes controlled by the CCR4-NOT complex and are not principally involved in the functions controlled by either the CCR4/CAF1 or NOT modules.

To obtain a better understanding of the similarity in effects on gene expression of the two modules of the CCR4-NOT complex, we conducted a cluster analysis on all 6200 genes present in the microarray analysis under glucose growth conditions (Fig. 3a, left panel). From the overall cluster analysis the relationships described above for the good genes was confirmed and extended. For example, ccr4/caf1 were most similar in the manner that they affected all yeast genes, as were not3/not4/not5. btt1 will be discussed below. We also further examined by cluster analysis those genes that were affected by more than two-fold by either both ccr4 and caf1 deletions or by both not4 and not5 deletions, as these pairs displayed the greatest correspondences by both cluster analysis and in the correlation data. As displayed in Fig. 3b, under glucose growth conditions, the genes that were affected either positively or negatively by both ccr4 and caf1 were not similarly affected by any of the other deletions. Similar results were obtained for the genes affected by both not4 and not5, except not3 displayed significant similarities with not4/not5 (Fig. 3c).

Fig. 3
figure 3

Heatmap of cluster analysis of genes affected by CCR4-NOT genes. a The effect of deletions in each of the CCR4-NOT on the complete set of yeast genes were clustered and heatmaps were generated. b–e Genes that were affected more than twofold by either both ccr4 and caf1 or by both not4 and not5 were clustered and heatmaps were generated. b Heatmap of cluster analysis of genes commonly affected by both ccr4 and caf1 under glucose growth condition. c Heatmap of cluster analysis of genes commonly affected by both not4 and not5 under glucose growth condition. d Heatmap of cluster analysis of genes commonly affected by both ccr4 and caf1 under non-fermentative growth condition. e Heatmap of cluster analysis of genes commonly affected by both not4 and not5 under non-fermentative growth condition

Full size image

We also determined the genes that were affected twofold by each pair of deletions and the names of the genes affected by each pair are listed in supplementary data. It should be noted that in Fig. 3b there are a few genes that are affected either positively or negatively by both ccr4 and caf1 that also seem to be affected by other members of the CCR4-NOT complex. However, analysis of these genes (15 that were downregulated by CCR4/CAF1 and 6 that were upregulated) did not identify any common functions, and about 50% of them are either unknown genes or genes of dubious function (data not shown). Similarly, we identified the genes (nine that were downregulated and four that were upregulated) most highly affected by both ccr4 and caf1 (greater than 3.5-fold effects on gene expression by both factors) and again found that there were no common functions represented in this pool. These results are expected given the extremely small set of genes that were represented.

Family of genes whose expression are downregulated or upregulated by CCR4-NOT factors

Table 2 lists the percentage of good genes whose expression is either increased at least twofold or decreased at least twofold by each of the seven deletions in the CCR4-NOT genes. For each deletion, approximately equal numbers of genes were downregulated as which were upregulated. Among these groups of affected genes we subsequently determined if any of the deletions was preferentially affecting the expression of a particular group of functionally related proteins. Using the Annotated Yeast Gene Data Base (http://mips.gsf.de/) for each functional group of genes, we determined the percentage of genes either increased or decreased two-fold by a CCR4-NOT deletion. Figures 4 and 5 depict all the functional groups that were more highly represented in the group of downregulated and upregulated genes, respectively, than expected as determined by Pearson’s chi-square test with Yate’s continuity correction in which a P value of <0.01 was chosen as the cutoff.

Fig. 4
figure 4

Functional classification of affected genes under glucose growth conditions. The classification of genes is based on the MIPS database available on the Web (http://mips.gsf.de/). Only those families of genes that are overrepresented as being affected (at least twofold effects) by a particular CCR4-NOT deletion are presented (P < 0.01). The percentage of genes affected is given below each gene family name. The data for caf130 2 comes from strain JF13-c130. a Downregulated genes. b Upregulated genes

Full size image
Fig. 5
figure 5

Functional classification of affected genes by specific pairs of CCR4-NOT genes. Analysis was conducted as described in Fig. 4, except all the genes present in the microarray were analyzed. This was done to ensure a large enough pool of genes to be analyzed and compared statistically

Full size image

A number of observations can be made from this analysis. ccr4 and caf1 deletions did not preferentially increase the expression of a large number of families of proteins (Fig. 4a). These observations are consistent with the major effect of CCR4 and CAF1 to be to increase general mRNA stability (Liu et al. 1997; Tucker et al. 2001, 2002; Chen et al. 2002; data not shown). However, both ccr4 and caf1 were found to preferentially increase the expression of genes involved in DNA repair, and previous observations have indicated that ccr4 and caf1 deletions specifically affect the DNA repair pathway (Schild 1995; Jelinsky and Samson 1999; Hanway et al. 2002; Grigull et al. 2004). ccr4 also preferentially increased the expression of mitochondrial genes and those involved in glycolysis and gluconeogenesis. Similarly, CCR4 and CAF1 upregulated few families of proteins (Fig. 4b). Most novel in this regard is that CCR4 tended to be required for expression of nucleotide metabolism genes (Fig. 4b), as possibly was CAF1 (12% of nucleotide metabolism genes were upregulated, P = 0.02) (not shown). This control of nucleotide metabolism genes is significant in that alterations in nucleotide concentrations occurs upon DNA damage and such changes have been correlated with genomic mutation and stability (Bebenek et al. 1992; Chabes et al. 2003; Holmberg et al. 2005). As CCR4 and CAF1 are critical to the response of DNA damage, they may be affecting this response through control of nucleotide metabolism gene expression.

In contrast to ccr4/caf1, the not deletions increased the gene expression of a much larger number of protein families (Fig. 4a). A number of these families of proteins are part of the stress response, a pathway that is known to be controlled by NOTs (Maillet and Collart 2002). These stress response families include genes involved in the stress response per se, C-compound and carbohydrate utilization, proteolytic degradation, TCA cycle, metabolism of energy reserves, mitochondria, and glycolysis and gluconeogenis. The not3 and not5 alleles also disproportionately affected a number of genes involved in transport in the cell and in ionic homeostasis and not4 and not5 affected phosphate metabolism gene expression. These results suggest that under glucose growth conditions the NOTs downregulate mRNA expression as part of larger programmes of control over gene expression. Although the NOTs overall upregulated the gene expression of similar numbers of genes as that they downregulated (Table 2), no families of proteins were preferentially being upregulated by the NOTs (Fig. 4b).

Surprisingly, the caf40 deletion affected several families of proteins that were specifically controlled by the nots. These families clustered in the groups of transporters, membrane proteins and ion homeostasis. These data suggest that CAF40 protein may function with the NOTs in controlling certain subsets of gene families in yeast and support the role for CAF40 in transcriptional control as indicated by studies of the mammalian ortholog of CAF40 (Collart and Timmers 2004).

We also conducted a similar analysis for all the genes that both CCR4 and CAF1 affected, that NOT4 and NOT5 affected, and that CAF40 and NOT3 affected under glucose growth conditions (Fig. 5Aa, b). Very few families of genes were affected by both CCR4 and CAF1. However, once again NOT4 and NOT5 downregulated the expression of a number of families of genes, including those involved in proteolytic degradation, phosphate metabolism, ion transportation and the stress response. Both CAF40 and NOT3 affected a number of families of genes, as suggested from the data presented in Fig. 4.

The CCR4-NOT factors control similar types of genes under non-fermentative conditions

The genes whose expression was affected by defects in CCR4-NOT components was also determined under non-fermentative growth conditions. As shown in Fig. 2b, ccr4 and caf1 displayed a very high overlap in the genes that they affected under non-fermentative conditions: correlation coefficient of 0.86 for nearly 3,000 genes. Similarly, not4 and not5 acted very similarly with an overall correlation coefficient of 0.77. Most noteworthy, however, was that in almost all other cases all of the seven deletions displayed significant correlations (0.34 or above) in the genes that they affected under non-fermentative conditions. These data indicate that the CCR4-NOT complex as a whole is important to the non-fermentative response. Moreover, as the non-fermentative response is a stress condition, it suggests that the CCR4-NOT complex is critical to the stress response. In addition, we compared by cluster analysis the effect on all 6,200 genes of each of the deletions following non-fermentative growth (Fig. 3a, right panel). In addition to ccr4/caf1 and not4/not5 displaying the greatest similarities, as predicted by the correlation data presented in Fig. 2, we observed that caf40/caf130 were most similar, which also agrees with the correlation data (Fig. 2, bottom panel) on the good genes that we analyzed. However, in contrast to glucose growth conditions, when just the genes affected twofold by either the ccr4/caf1 pair or the not4/not5 pair following non-fermentative growth were clustered (Figs. 3c, d, respectively), many more genes were found to be commonly affected by multiple CCR4-NOT deletions. This result is again in agreement with the correlation data presented for the good genes (Fig. 2).

The types of genes preferentially affected by multiple components of the CCR4-NOT complex under non-fermentative growth conditions, as displayed in Figs. 6 and 7, indicate that a number of processes tend to be downregulated by the CCR4-NOT complex. In addition to genes involved directly in the non-fermentative response (c-compounds and carbohydrate utilization and their transport, amino acid metabolism, TCA cycle, and respiration, mitochondrial transport, and mitochondria), import processes, ion homeostasis, and the stress response were commonly affected. Multiple components of the CCR4-NOT complex appeared to affect stress response genes, for in addition to the not4 and not5 effects shown in Fig. 6b, the caf1 and not3 deletions increased by twofold the expression of 20 and 17% of the stress response genes, respectively, (P = 0.03) (not shown). It is noteworthy that CAF40 and NOT3 appear to share similar patterns of genes that they downregulated with those downregulated by the CCR4/CAF1 module. This observation suggests that CAF40 may also play a particular role under non-fermentative conditions. It should also be noted in Fig. 6b that not5 under non-fermentative conditions also tended to affect genes involved in proteolytic degradation, as possibly did not4 (26% of proteolytic degradation genes were downregulated, P = 0.02) (not shown).

Fig. 6
figure 6

Functional classification of affected genes under non-fermentative growth conditions. Analysis was conducted as described in Fig. 4. a Downregulated genes. b Upregulated genes

Full size image
Fig. 7
figure 7

BTT1 immunoprecipitates with the CCR4-NOT complex. Immunoprecipitations were conducted with antibody against CAF130, CCR4, and BTT1 as indicated. Western analyses were used to detect CAF130, CCR4, NOT1, NOT5, CAF1, CAF40, BTT1, EGD1 and EGD2 as indicated. CE-crude extract. a BTT1 immunoprecipitates with CAF130 and CCR4. Lanes 1, 4, and 7-strain JF13; lanes 2, 5, and 8-strain JF13-c130; lanes 3 and 6-strain JF14. b BTT1 immunoprecipitates the CCR4-NOT complex. Strains are the same as used in (a). c EGD1 and EGD2 are not part of the CCR4-NOT complex. Strains are the same as used in (a)

Full size image

Once again, however, under non-fermentative growth conditions, few families of genes were preferentially upregulated by CCR4-NOT genes. Given that similar numbers of genes were upregulated by the CCR4-NOT factors as downregulated, these data suggest that the increase in gene expression by the CCR4-NOT complex, as in the control of ADH2 gene expression (Denis 1984; Draper et al. 1995; Liu et al. 1998), may result from very general effects on mRNA levels or from indirect effects. CAF40, however, was responsible for the preferential upregulation of mRNA transcription and synthesis genes under both glucose and non-fermentative conditions, suggesting that it, in contrast to other CCR4-NOT components, may play a direct role in gene activation.

When we analyzed the families of genes downregulated by both CCR4 and CAF1 under non-fermentative conditions, it became clear that a number of families were represented, implying a critical role for these two factors in the non-fermentative response (Fig. 5b). Almost all of these families involve carbohydrate metabolism and the response to non-fermentative growth. Under the same conditions, NOT4 and NOT5 once again downregulated genes involved in the stress response, and we observed that NOT3 displayed considerable overlap with that of CCR4 and CAF1 in the downregulation of genes involving carbohydrate metabolism and ion transport. These results extend the observations that we made when analyzing the effects of single components of the CCR4-NOT complex on gene expression (Fig. 4) and indicate that certain pairs or other combinations of factors will play particular roles in downregulation of gene expression.

The CCR4-NOT complex preferentially represses SAGA-controlled genes

Previously, several NOT proteins have been shown to interact physically with various components of either TFIID or SAGA (Benson et al. 1998; Badarinarayana et al. 2000; Lemaire and Collart 2000; Deluen et al. 2002; Russell et al. 2002; Sanders et al. 2002). The NOT genes are also particularly important for repressing gene expression from non-canonical TATA sequences (Collart and Struhl 1994; Badarinarayana et al. 2000). In that nearly all of gene expression in yeast is either controlled by TFIID or SAGA (Huisinga and Pugh 2004), we examined whether any of the CCR4-NOT factors tended to specifically affect TFIID- or SAGA-controlled genes. As displayed in Table 3, SAGA-controlled genes were significantly over represented in the genes downregulated by the NOT factors under glucose growth conditions. For example, while NOT5 downregulated about 16% of all genes in yeast, 35% of all SAGA-controlled genes were downregulated by NOT5. The CCR4/CAF1/CAF40/CAF130 proteins did not preferentially downregulate either TFIID- or SAGA-responsive genes. Importantly, under non-fermentative growth conditions, all of the CCR4-NOT factors, except for CAF130, dwornregulated two- to four times as many SAGA genes as expected (Table 3). In contrast, neither TFIID- nor SAGA-controlled genes were over represented in the genes upregulated by any of the CCR4-NOT factors under either glucose or non-fermentative growth conditions (data not shown). These results imply that a substantial part of the specific effects of the NOTs on stress related genes, either on glucose or non-fermentative growth conditions, is mediated by or in conjunction with the SAGA complex.

Table 3 CCR4-NOT complex preferentially represses SAGA-controlled genes
Full size table

BTT1 is a component of the CCR4-NOT complex

The above results indicate that CAF130 is a distinct member of the CCR4-NOT complex that does not appear to overlap in its gene expression pattern with either the CCR4/CAF1 or NOT module of proteins. Previously, a comprehensive genomic two-hybrid search in yeast identified BTT1 as a possible partner for CAF130 (Ito et al. 2001). BTT1 is a component of the nascent polypetide associated complex (NAC) that is known to associate with the ribosome (Franke et al. 2001), suggesting that CAF130 may function in a unique role for the CCR4-NOT complex. BTT1 is also known to affect mRNA transcription (Hu and Ronne 1994). We consequently investigated whether BTT1 did associate with the CCR4-NOT complex through its interaction with CAF130.

When CAF130 was immunoprecipitated in a wild-type strain, BTT1 was co-immunoprecipitated (Fig. 5a, lane 4) along with CCR4. This antibody precipitation of BTT1 required the presence of CAF130, for in a caf130 strain no BTT1 was immunoprecipitated (lane 5). The band identified with the BTT1 antibody was shown to be BTT1 as it was absent in a btt1 strain background (lanes 3 and 6). Immunoprecipitating CCR4 pulled down both CAF130 and BTT1 (lane 7) and in a caf130 background CCR4 failed to co-immunoprecipitate BTT1. As CCR4 and other members of the CCR4-NOT complex when expressed at their normal physiological concentrations only co-immunoprecipitate other components of the CCR4-NOT complex (Bai et al. 1999; Chen et al. 2001; Denis and Chen 2003), these results indicate that BTT1 is a bona fide member of the CCR4-NOT complex. Moreover, they indicate that BTT1 is linked to the CCR4-NOT complex through its interaction with CAF130.

To confirm these results, we conducted these immunoprecipitations in the reverse direction utilizing BTT1 antibody. In this case, BTT1 co-immunoprecipitated multiple components of the CCR4-NOT complex (Figs. 5b, lane 4). Deleting CAF130 blocked the ability of BTT1 to co-immunoprecipitate any of the CCR4-NOT proteins, although BTT1 was efficiently immunoprecipitated (lane 5). In a btt1 strain, the BTT1 antibody also did not immunoprecipitate the CCR4-NOT factors (lane 6), although a very low amount of NOT1, CCR4, and CAF130 may be observed, suggesting some non-specific immunoprecipitation with this preparation of antibody. These results confirm that BTT1 is a component of the CCR4-NOT complex and is linked through the complex through its interaction with CAF130. Superose 6 gel chromatography of yeast crude extracts further confirmed that BTT1 migrated in 1.9 and 1.0 MDa complexes coincident with that of CCR4 and CAF40 (data not shown).

The linkage of CAF130 and the CCR4-NOT complex to BTT1 suggests that the CCR4- NOT complex may be linked functionally to polypeptide folding and the ribosome via the NAC complex. To explore this possibility, we examined if the other two components of the NAC complex, EGD2 (the BTT1 homolog) and EGD1, associate with the CCR4-NOT complex. As shown in Fig. 5C, however, immunoprecipitation of either CAF130 or CCR4 did not co-immunoprecipitate either EGD1 or EGD2 (lanes 3 and 5, respectively). We conclude that the interaction between CCR4-NOTs and BTT1 is not functionally linked to BTT1’s role in the NAC complex. Our results are in contradiction to those of Panasenko et al. 2006, which demonstrated that EGD1 and EGD2 of the NAC complex interacts with NOT4/NOT5/NOT1, a difference that is probably the result of alternate means to purify the CCR4-NOT complex. Since immunoprecipitation with CCR4 antibodies reliably co-precipitates the core and integral components of the CCR4-NOT complex, we believe that BTT1, but neither EGD1 or EGD2, is a component of the 1 MDa CCR4-NOT complex (Chen et al. 2001; Liu et al. 1998).

Microarray analysis indicates that BTT1 and CAF130 function similarly in control of gene expression

Because of the close physical contact between BTT1 and CAF130, we used microarray analysis to determine if they functioned similarly in terms of controlling gene expression. Whole genome microarray analysis was conducted for isogenic btt1 and caf130 strains under glucose conditions. caf130 and btt1 displayed a very high degree of overlap in the degree to which they affected the same genes: correlation coefficient of 0.73 for over 3,200 genes (Fig. 2). Although the strain used for this analysis was different than that which was used for the analysis of the other CCR4-NOT components, when we compared the sets of good genes from the two caf130 strains, we found that they displayed a correlation coefficient of 0.81 (Fig. 2), suggesting that the use of a different strain did not bias our results. BTT1 and CAF130, therefore, appear to act very similarly in the control of gene expression and to comprise a third module of the CCR4-NOT complex. These results were also confirmed by cluster analysis of the effect of btt1/caf130 on all the genes present in the yeast genome (Fig. 3a). In this case, BTT1 acted most similarly to CAF130.

While the function of the CAF130 and BTT1 proteins of the CCR4-NOT complex remains unknown, both btt1 and caf130 in the JF13 strain background preferentially increased expression of genes involved in ribosome biogenesis (Fig. 4a) and caf130 in the FY1642 strain background also affected this gene family under non-fermentative conditions. Other CCR4-NOT factors did not share this propensity (although not5 did repress preferentially repress ribosome biogenesis genes under non-fermentative conditions), suggesting that CAF130 and BTT1 function to control processes in a manner different than the other two modules of the CCR4-NOT complex. In addition, some type of auto-regulation of the CCR4-NOT complex might be one role employed by CAF130/BTT1 in that deletions in either caf130 or btt1 reduced CCR4 mRNA expression by two-fold, and caf130 has previously been shown to reduce CCR4 protein expression (Chen et al. 2001).

Discussion

Using whole genome microarray analysis we have compared the effects on gene expression of seven of the nine known components of the CCR4-NOT complex. Our results indicate that based on similarities of gene expression under glucose growth conditions, the CCR4-NOT complex consists of three modules: CCR4/CAF1, NOT3/NOT4/NOT5, and CAF130/BTT1. The separation of CCR4/CAF1 from NOT3/NOT4/NOT5 agrees with previous data (Chen and Denis 2003; Collart and Timmers 2004). Therefore, while there do exist overlap in phenotypes between these two modules in regards to transcriptional activation, repression, elongation, and mRNA degradation (Liu et al. 1998; Denis et al. 2001; Tucker et al. 2002), the two modules function quite differently in affecting overall mRNA levels. Although we did not analyze the effects of NOT2 on global gene expression, previous data suggest that NOT2 belongs in the NOT3/NOT4/NOT5 module (Russell et al. 2002). NOT1’s role is less obvious. In contacting both modules with clearly separate protein regions, NOT1 might be expected to functionally bridge both modules (Denis and Chen 2003; Collart and Timmers 2004).

The identification of the third module of CAF130/BTT1 arises from our determination that BTT1, a member of the NAC complex, is a bona fide member of the core CCR4-NOT complex. This conclusion is based on our immunoprecipitation and deletion analysis and similarities in effects on gene expression based upon our microarray analysis. We found no evidence of other members of the NAC complex as being core CCR4-NOT components, although other studies do indicate that the NAC interacts with the NOT factors (Panasenko et al. 2006). The functional role for CAF40 remains, in contrast, more elusive as it shares similarities in effects on gene expression to multiple CCR4-NOT components.

Under non-fermentative growth conditions the relationships between these modules still inheres, but the most striking observation was that under this growth condition all of the CCR4-NOT factors behaved much more similarly in their effects on global gene expression. This observation probably results from the fact that non-fermentative growth is a stress growth condition and the CCR4-NOT complex, as shown by us and others, plays a particular role in the general stress response (Lenssen et al. 2002).

These observations about the role of the CCR4-NOT complex in the stress response relates to another major observation obtained in these studies: the preferential effects on SAGA-controlled gene expression by the CCR4-NOT complex. This connection is important for several reasons. First, as the SAGA-controlled genes tend to be involved in the stress response, it is likely that a substantial part of the specific effects of the NOTs on stress related genes under glucose growth conditions is mediated by or in conjunction with the SAGA complex. Second, under non-fermentative growth conditions, it appears that the whole CCR4-NOT complex (except for CAF130) plays a specific role in governing SAGA functions and therefore the stress response. Third, these results suggest that the NOT-SAGA interactions that have previously been identified (Collart and Timmers 2004) will turn out to be critical to affecting SAGA-controlled gene expression. Fourth, the NOT control of the stress response also correlates with the SAGA complex preferentially affecting gene expression from TATA-containing promoters in contrast to TFIID which affects TATA-less promoters (Basehoar et al. 2004). The TATA- containing promoters controlled by SAGA include the non-canonical TATAs that have been shown to be affected by not mutations (Collart and Struhl 1993, 1994; Collart 1996; Badarinarayana et al. 2000). These observations indicate that the NOT factors act to repress gene expression from non-canonical TATA promoters, thereby affecting the stress response mediated by SAGA.

Two further observations should be mentioned in regards to these analyses. First, certain CCR4-NOT factors act very similarly in their effects on gene expression (CCR4/CAF1, for example). The concordance of a number of cellular phenotypes between a ccr4 deletion and that of caf1 (Draper et al. 1995; Liu et al. 1997, 1998; Tucker et al. 2001, 2002) is consistent with the high correlation observed in this microarray analysis. However, other observations indicate that CAF1 may affect processes not controlled by CCR4 (Bai et al. 1999; Chen et al. 2002; Deluen et al. 2002; Viswanathan et al. 2004; Ohn et al. 2007). Therefore, it is unlikely that any one of the CCR4-NOT proteins is truly redundant with any of the others. Second, our results do not imply, for instance, that the NOTs have no role in mRNA deadenylation, a role primarily assigned to CCR4/CAF1. Instead, the NOTs may still affect the degradation rates of mRNA in a manner similar to CCR4/CAF1, although to a lesser extent (Tucker et al. 2002), but may also affect other aspects of the mRNA expression profile of yeast genes such that the absolute levels of mRNA levels in strains carrying not mutations are not similar to those observed in ccr4/caf1 backgrounds.

In analyzing the families of proteins preferentially affected by either individual CCR4-NOT components or particular modules, many of the previously identified functional roles of the CCR4-NOT complex in processes such as the stress response, DNA repair, carbohydrate metabolism and protein degradation were substantiated. The microarray results also suggest a number of new functional roles for CCR4-NOT components. Further studies will be necessary to validate these newly identified effects on specific cellular biological processes. It was also observed that in general the CCR4-NOT factors played very little roles in upregulating gene expression of particular families of genes. Instead, their major effects on particular classes of genes were on the downregulation of their gene expression. This difference implies the activation of gene expression by CCR4-NOT factors may result from either very global effects on gene expression or from indirect effects on the downregulation of other factors.

In conclusion, microarray analysis is a powerful tool to view patterns of whole genome expression and to investigate the interactions between different proteins. Our microarray results and previous data indicate that the CCR4-NOT complex consists of three functional modules under glucose growth conditions: CCR4/CAF1, the NOTs, and CAF130/BTT1. Part of this separation of function may result from their distinct biochemical roles. CCR4/CAF1 are primarily responsible for mRNA deadenylation and the NOTs function in repression of mRNA transcription. These microarray and previous results also suggest that the NOTs act with or through the SAGA complex in the control of the stress response. While the CAF130/BTT1 module is specifically responsible for control of genes involved in ribosome biogenesis, the molecular means by which this is accomplished is not yet understood. Under non-fermentative conditions, in contrast, the whole CCR4-NOT complex, except for CAF130, tends to function more similarly in the control of families of genes involved in the stress response.