In this mini-review, we summarize our current knowledge about the cross-talk between the different levels of gene expression. We introduce the Ccr4 (carbon catabolite repressed 4)–Not (negative on TATA-less) complex as a candidate to be a master regulator that orchestrates between the different levels of gene expression. An integrated view of the findings about the Ccr4–Not complex suggests that it is involved in gene expression co-ordination. Since the discovery of the Not proteins in a selection for transcription regulators in yeast [Collart and Struhl (1994) Genes Dev. 8, 525–537], the Ccr4–Not complex has been connected to every step of the mRNA lifecycle. Moreover, it has been found to be relevant for appropriate protein folding and quaternary protein structure by being involved in co-translational protein complex assembly.
Gene expression buffering
In eukaryotic cells, genes are transcribed in the nucleus and the produced mRNAs are exported for translation and subsequent degradation to the cytoplasm. Despite this separation in time and space for the different stages of gene expression, compelling evidence has accumulated these last years to indicate that the processes are intimately connected. Components of the different cellular machines that participate to production and degradation of mRNAs circulate between the different compartments. For several of these factors, it has now been shown that they exert their major function in one compartment, but can moonlight in the other compartment to oppositely affect gene expression. This serves the purpose of buffering gene expression, namely by inducing compensatory changes in production and degradation of mRNAs to maintain homoeostasis.
Instrumental in revealing the circuitry of gene expression have been studies by Tirosh and colleagues  in which they found that mRNAs of a given gene were expressed at a relatively similar level despite very different decay rates in different yeast species. Similar observations could be made by comparing different metazoans . These findings were supported and extended by evidence that mRNA synthesis and decay rates were oppositely affected by mutations in the synthesis or decay machines in Saccharomyces cerevisiae .
Coupled changes in synthesis and decay rates of different yeast have been linked in cis to transcription factor-binding sites in gene promoters, suggesting that binding of transcription factors to promoters might co-ordinate transcription and decay rates. Indeed, replacement of an upstream activating sequence in a yeast reporter gene without changing the sequence of the produced transcript alters the kinetics of mRNA degradation . Moreover, single-molecule FISH (fluorescent in situ hybridization) experiments demonstrated that the specificity and timing of decay of two cell cycle-regulated genes in yeast was completely dependent upon their promoter . In fact, at a genomic scale, co-ordination of mRNA synthesis and decay rates is defined by promoter sequences, in yeast and also in human . Interestingly a promoter's function may not only be mediated through transcription factors, but through other promoter-associated factors such as a kinase that can remain associated and be exported with an mRNA .
Gene expression buffering requires that communication occurs forward, from mRNA synthesis to degradation, but also in reverse direction, namely from degradation to synthesis. The first evidence for forward buffering from a component of the mRNA synthesis machinery itself was provided in 2005 . A subunit of RNA polymerase II (RNA Pol II), Rpb4 (RNA polymerase B subunit 4), was shown to be required for the decay of a specific class of mRNAs in yeast. Rpb4 shuttles between the nucleus and the cytoplasm and in the cytoplasm associates with components of the major mRNA decay pathway and even localizes to P-bodies, where it is thought that mRNAs are decapped and degraded [7–9]. Consistently with this role for Rpb4, the most prominent trans mutations that correlate with coupled changes in mRNA synthesis and decay rates in diverse yeast are linked to RPB4 .
Rpb4 forms a heterodimeric complex with another subunit of RNA Pol II, Rpb7 and the dimer associates with the largest subunit of RNA Pol II, Rpb1, at a region that is situated near to the transcript exit groove and the C-terminal repeat region of Rpb1 (CTD) known to serve as a platform for assembly of components of post-transcriptional regulatory machines. This dimer tends to dissociate readily from the rest of the polymerase and is dispensable in vitro for stable recruitment of polymerase to active pre-initiation complexes, though it is necessary for promoter-dependent transcription initiation [10–17]. Whereas Rpb4 is dispensable for viability in S. cerevisiae, Rpb7 is essential. Rpb7 has also been reported to be important to connect mRNA synthesis and decay rates . The Rpb4–Rpb7 dimer was proposed to associate with transcripts at the completion of transcription in a co-transcriptional manner (mRNA imprinting), be exported with the mRNAs and govern mRNA decay in the cytoplasm . Subsequent work has argued against a cytoplasmic function of Rpb4 because a yeast strain in which Rpb4 was fused to Rpb2 compensated altered synthesis and mRNA decay rates due to the Rpb4 deletion . However, our study published last year has indicated that the presence of Rpb7 in the cytoplasm may be connected to its co-translational assembly into RNA Pol II . We showed that Rpb2 was also present in polysomes for this purpose, such that a fusion of Rpb4 and Rpb2 would similarly be able to localize to polysomes where Rpb4 might be able to exert its cytoplasmic function. It is unlikely that co-transcriptional Rpb4 mRNA imprinting can take place if Rpb4 is fused to Rpb2. However, if the mRNA is imprinted instead by a factor that usually tethers Rpb4, it would still be possible for Rpb4 to have a cytoplasmic effect via it's recruitment to the mRNA by the interacting factor in the cytoplasm. Ccr4 (carbon catabolite repressed 4)–Not (negative on TATA-less) complex is certainly a candidate for such a factor. Indeed, cytoplasmic presence of Rpb4 was shown to be dependent on Not5 . The co-translational assembly of the transcription machinery in the cytoplasm that we reported is likely to participate in the global scheme of gene expression buffering (see below) . In this scheme, cytoplasmic Rpb4 could play a role.
Whereas forward buffering (inducing changes in mRNA decay to compensate for changes in transcription) is very easy to grasp as mRNA can be marked during transcription, reverse buffering (inducing changes in transcription to compensate for changes in mRNA decay) from the cytoplasm to the nucleus is much less intuitive. The Cramer laboratory investigated which components of the mRNA degradation machines were most important for buffering of mRNA levels in yeast . They tested a large number of components of the RNA degradation machines and identified the 5′- to 3′-exonuclease Xrn1 (exoribonuclease 1) as absolutely key for buffering. Cells lacking Xrn1 are unable on a global scale to compensate for reduced mRNA degradation by reduced synthesis. They found that the catalytic function of Xrn1 and the presence of Xrn1 in the nucleus were required for buffering. They argued against a direct repression of transcription by Xrn1 since they found no cross-linking to genes and no effect on in vitro transcription by extracts. Instead, they observed that the synthesis of the mRNA encoding a global repressor, Nrg1 (negative regulator of glucose-repressed genes 1), was increased in xrn1Δ and also in other mRNA degradation mutants (such as caf1Δ or ccr4Δ) indicating that this repressor could be part of the buffering system. Others have argued instead that Xrn1 does directly impact on transcription and is present at promoters and during transcription elongation [21,22].
Ccr4–Not complex in gene expression buffering
After synthesis, mRNAs are translated before they are degraded and hence missing in our knowledge of the buffering of gene expression in eukaryotes is how to integrate translation in the circuitry. It is more than certain that it will play an essential role. For instance comparison of S. cerevisiae and Saccharomyces paradoxus mRNA abundance and ribosome occupancy revealed that translation regulatory divergence often buffers species differences in mRNA abundance . Our recent finding that a subunit of the Ccr4–Not complex, Not5, connects transcription to translation and backwards translation to transcription, playing essential roles in the nucleus and in the cytoplasm, seems to indicate that an important master regulator integrating translation in the buffering circuitry could be the Ccr4–Not complex . Subunits of this complex are present at the sites of transcription [24–37] and impact on transcription elongation [36,38,39], the major eukaryotic deadenylases are subunits of this complex  and finally Ccr4–Not subunits are present at polysomes during translation and have been implicated in co-translational quality control [41–44].
The core Ccr4–Not complex is composed of between nine and 11 identified subunits in different organisms [28,45–50]. In yeast, these subunits are the five Not proteins, Caf1 (Ccr4 associated factor 1), Ccr4, Caf40 and Caf130. The large Not1 protein serves as a scaffold for assembly of the different subunits into the complex, but it also serves as a scaffold to bring mRNAs in contact with the major eukaryotic deadenylases, Ccr4 and Caf1 [51,52]. The other Ccr4–Not subunits dock on to the Not1 scaffold and can in certain cases contribute to bring the deadenylases on the Not1 scaffold together with mRNAs, but their primary function does not seem to be deadenylation [46,53–56]. The Not4 subunit has an E3 ligase activity and ubiquitylates a variety of substrates [57–63]; it is also involved in co-translational quality control [42–44,64]. Not4 binds to the C-terminal region of Not1 [65,66].
The Not2 and Not5 subunits function together and form a heterodimer that also binds the C-terminus of Not1 [65,67]. The binding of Not4 and Not2/5 to the Not1 scaffold occur on largely separate surfaces of Not1 . Not2 and Not5 were first described to function in transcription [24,26,45,54,65,68,69]. This idea will need to be revisited with our current knowledge that cells lacking Not5 contain improperly assembled RNA Pol II . Indeed, Not5 is essential during translation for proper interactions of nascent Rpb1, the largest subunit of RNA Pol II, with its chaperone and subsequent functional co-translational assembly of Rpb1 into polymerase complexes . Efficient Pol II assembly hence is connected to translation and this suggests that the status of the translation apparatus, the availability of actively translating ribosomes may feedback to transcription by modulating the quantity of newly assembled Pol II. Hence, Not5 connects translation with transcription. Another interesting finding that connects the Ccr4–Not complex with integrity of other protein complexes and assembly of newly produced proteins into complexes as previously proposed , is that the deletion of Not4 has an effect on the integrity of the proteasome .
Could then the Ccr4–Not complex be the, or a component of the, master regulator that orchestrates the different levels of gene expression? Would the site at which this buffering is regulated be the translating ribosome? The co-ordination of the buffering might occur through the docking of the different subunits on the Not1 scaffold. But at the same time the various subunits might affect first on association of the mRNA with Not1. The latter would occur in the nucleus as the Ccr4–Not complex can bind transcription elongation complexes [36,38,39] whereas the former would occur in the cytoplasm. The idea that the different Ccr4–Not subunits may impact on Not1 mRNA association stems from the finding that several subunits are able to bind mRNA. Indeed, UV cross-linking of poly(U) stretches to Not1, Not2 and to Not5 showed that these subunits can bind RNA in complex but individually also . Moreover Not4 has an RNA recognition motif . Hence, mRNAs might be imprinted in the nucleus and then marked by Not1 in polysomes where Not1 can sense and integrate environmental cues and modify the imprinted state. A differentially assembled complex on the mRNA and Not1 scaffold could serve as a protein signature for other proteins to identify what to do with the mRNA. Globally, the Ccr4–Not complex would define the number of mRNAs to be regulated immediately upon reception of environmental signals. At the same time define the responsive pool of mRNAs inside the cells. The pool of imprinted mRNAs that the cell can deal with is not infinite as cellular resources are limited (there are much more mRNAs in cells than ribosomes for example) and definition of the functional mRNA pool is therefore necessary. In the signalling cascades to define the responsive pool of mRNAs in given conditions, RNA-binding proteins and miRNAs that have the capacity to tether the Ccr4–Not complex to specific mRNAs are obviously also relevant . The different roles of the Ccr4–Not complex in gene expression buffering discussed above are summarized on Figure 1.
Schematic view of the Ccr4–Not complex's role in gene expression buffering
There is ample evidence that the Ccr4–Not complex exists in several forms. For instance the level of several Ccr4–Not subunits is variable between different types of tissues when the amount of the other subunits remains stable . However, there is also evidence that depletion of certain Ccr4–Not subunits leads to depletion of all subunits [75,76] and this has been argued to suggest that the complex works as a unique entity. These latter experiments however, do not exclude dynamic assembly and disassembly of the complex for instance during translation where the Not1 scaffold might be protected from degradation through its interaction with the mRNA and translation machinery.
The existence of a unique machine that buffers the different levels of gene expression provides the cell with means to give a fast internal response to environmental signals and optimally use limited resources. Energy metabolism is very essentially connected to responses to the environment. Under certain conditions, typically under stress, cells have to shut down the expression of certain genes almost immediately; otherwise they run out of energy to maintain housekeeping functions. Hence, if a complex or machinery is responsible for orchestrating the different levels of gene expression, then this complex or machinery should also be directly linked to mitochondrial functions and/or biogenesis. In this context, it is interesting to note that several lines of evidence connect the Ccr4–Not complex to energy metabolism. A transcriptome analysis of the homozygous ccr4 null mutant in Candida albicans showed that a large proportion of the differentially-expressed genes are related to the function of mitochondria . Specific effects on gene expression of mitochondrial functions have also been observed in S. cerevisiae ccr4–not mutants [37,50] In mice, CNOT3 is important for stress-induced cardiac functions and control of lipid storage [78,79], both linked to energy metabolism.
Dynamic responses to environmental changes require that the active pool of mRNAs can be modified. Some mRNAs must be stored, others should be degraded and their transcription should be turned off rapidly to maintain cellular homoeostasis. It seems that the Ccr4–Not complex has the capacity to provide this dynamic regulation and define the active pool of mRNAs; it can control imprinting of mRNAs in the nucleus and modify this nuclear imprinting in the cytoplasm upon receiving a signal in this compartment. It is able to initiate the destruction of the mRNA it is engaged with by its deadenylase subunits and our recent results suggest that, via Not5, it even can modify the pool of translated mRNAs . Finally, it can determine how much new polymerase is produced. All these evidences determine that the Ccr4–Not complex fits the criteria required for the ability to respond fast to environmental changes by means of co-ordination of the different levels of gene expression. Exploring the signalling cascades of the Ccr4–Not regulation and the dynamics of Ccr4–Not assembly with mRNAs will be one of the most interesting future challenges and will greatly add to our understanding of how the different levels of gene expression are orchestrated in eukaryotes.
This work was supported by grant 31003a_135794 from the Swiss National Science foundation.
Translation UK 2015: Held at the University of Aberdeen, U.K., 7–9 July 2015.