We have established an approach to rapidly access the translatome of yeast cells. In this regard, the translatome refers to the pool of all RNAs that are associated with ribosomes purified via an affinity tag. Using this approach, we compared the relative changes of translatome and global transcript levels in response to five conditions of stress. This analysis provides a catalog of stress-regulated messages comprising approximately 20% of all expressed genes. Most of the messages changed homodirectionally, particularly under severe stress, suggesting a strong coordinate response of global transcript levels and translatome. Our survey suggests that only about 2% of all expressed messages were differentially regulated, preferentially under mild stress conditions and therefore represent candidates for translational regulation. Among these, we showed that the mitochondrial ATPase is prone to translational regulation in response to low doses of the cell-wall–perturbing agent CFW.
We accessed the translatome through the affinity purification of Rpl16, a tagged ribosomal protein of the 60S subunit, and subsequent analysis by microarrays (). This is distinct from previous approaches aimed to study translational regulation, which applied “classical” sucrose density fractionation to separate mRNAs present in polysomes from “free” RNA and those messages present in monosomes [54
]. Our alternative approach captures all messages that are bound by at least one ribosome (monosome) but excludes free RNA and those messages incorporated into other RNP complexes such as P-bodies or pseudo-polysomes, which may be present in classical polysomal gradients [9
]. Possibly, this may also be a reason for the rather fair correlation (r
= 0.3–0.5) seen between the relative enrichments of messages associated with tagged ribosomes and corresponding fractions from classical sucrose density profiles.
We have used affinity-purified ribosomes to assess the recruitment of individual messages to ribosomes on a genome-wide level, covering all steps of translation initiation, which is believed to be a major step for translational regulation [8
]. We wish to note that we may not be able to monitor regulation by alterations of ribosome density or elongation rates. Such an analysis requires monitoring of mRNAs in polysomal profiles as we did for selected candidate messages ( and S6
). Instead, our approach allows several additional applications that cannot be realized with classical sucrose density gradients. For instance, it offers the possibility to study mRNAs associated with paralogous ribosomal protein. Although our data do not indicate preferential association of mRNAs with either tagged paralogous Rpl16a and Rpl16b proteins grown in minimal media (Dataset S1
), the preferential association of selected messages with paralogs may become vital under certain specific growth conditions, especially in the light of recent observations that paralogous ribosomal proteins may be required for proper expression of certain messages [43
]. Tagged ribosomal proteins may also be used to access tissue-specific expression in animals or in complex cell suspensions in vitro. Previously, affinity-tagged poly(A)-binding protein was expressed with tissue-specific promoters to identify muscle- or neuron-specific messages in nematode Caenorhabditis elegans
or in the fruit fly Drosophila melanogaster
and included cross-linking of the proteins to their bound RNAs with formamide [55
]. The use of tagged ribosomal protein may represent a valuable alternative that could involve short treatment with CHX to stall ribosomes on messages. Indeed, while this manuscript was under revision, Heintz and colleagues reported the generation of transgenic mice expressing tagged Rpl10a to identify mRNA expressed in certain cell-types of mouse neurons [57
Here, we used our approach to monitor responses of the translatome upon the application of diverse stress conditions and investigated how they relate to respective changes in the global transcript levels (transcriptome) (). We found that severe stress, leading to growth arrest and ESR, induces highly balanced, correlated, and precise responses of transcript levels and translatome. This strong coordination even within minutes after stress application is intriguing, as it suggests that there is basically no delay of the regulating response of transcription and mRNA decay. In contrast, we found a largely uncorrelated response of global transcript levels and translatome under mild stress conditions: Low doses of CFW and menadione almost exclusively remodeled the translatome with very minor effects on the transcriptome. It appears that under these conditions, the cells adapt primarily at the posttranscriptional level, through the differential recruitment of messages to ribosomes and hence the regulation of protein synthesis. A possible explanation for this reaction could be that cells “buffer” environmental artifacts that do not greatly affect cell function or viability. Cells may favor keeping transcription constant because restructuring requires energy and additional organization.
Unfortunately, only a few studies have been performed that correlate protein levels with mRNA abundance, and to our knowledge, no systematic analysis of the changes of yeast protein levels upon stress is currently available [59
]. In a recent study investigating mRNA levels and abundance of 450 proteins, it was demonstrated that 73% of the variance in protein abundance can be explained by mRNA abundance [60
]. Conversely, this finding suggests that about one quarter of the variance of protein levels may be additionally controlled by translation or protein turnover. On the basis of our findings, we predict that only a minor fraction of this variance is due to altered ribosome recruitment. It may be possible that under steady-state conditions, the variance is mainly based on altered translation rates, or most likely, altered protein turnover. It will certainly be interesting to simultaneously measure alterations in transcriptome, translatome, and proteome to get a comprehensive view of the regulatory impact on each level of gene expression.
Not surprisingly, we observed that under severe stress conditions, many of the commonly or selectively altered messages in the transcriptome (global transcript levels) and translatome code for functionally related proteins; mRNAs coding for proteins that act in metabolism, transport, or are homed to the generation of energy showed increased relative abundances, whereas those involved in cell growth, protein or DNA/RNA binding and translation were preferentially decreased, many of them being part of the ESR [22
] (). The repression of components for the protein and mRNA synthesis machineries correlates with retarded cell growth, an effect that has been observed in various instances when cells adapted to new conditions [16
]. On the other hand, we speculate that increased production of components acting at the cell periphery (plasma membrane, cell wall) may be dedicated to replace damaged proteins or to intensify sensing of the environment.
Even under mild stress conditions that only affected ribosome association of a minor fraction of messages (up to 1%), we found various examples for concerted themes among messages in the translatome. For instance, the increased ribosome association for most components of the mitochondrial ATPase, which was confirmed by respective shifts of selected messages in polysomal profiles and the CHX-dependent increase of the mitochondrial ATPase activity in treated cells, strongly suggests that this small macromolecular complex undergoes translational regulation. The mitochondrial ATPase provides the major energy source of cells [62
]. Therefore, tight regulation of this complex at all levels of gene expression may be pivotal for proper energy homeostasis. There are no reports on how expression of this complex is achieved in yeast, but its biogenesis and assembly, in particular the uneven stochiometry of components for both F1
which are even encoded in different cellular compartments (B), certainly requires highly coordinate regulation, possibly involving posttranscriptional control. In the green algae Chlamydomonas
, for example, it has indeed been shown that formation of the chloroplast ATPase is dependent on controlled translation of subunits in the cytoplasm and in the chloroplast [63
-acting elements in the mRNAs that may directly affect translation rates of specific messages as seen for GCN4, translational regulation may also occur by trans
-acting factors, such as specific RBPs and noncoding RNAs. Dozens of regulatory RBPs co-sediment with ribosomes and polysomes, and thus could potentially regulate translation of specific sets of mRNAs [34
]. Intriguingly, we and others have shown that regulatory RBPs often bind to messages that encode functionally related proteins [2
]. It is also striking that some RBPs preferentially bind to messages coding for proteins either at the cell periphery (e.g., cell wall) or the nucleus [14
]. Coincidentally or not, we found that messages related to these themes were also preferentially regulated upon application of stress. Furthermore, our studies suggest that regulation of translation, in particular the recruitment of messages to ribosomes, may be a prominent mechanism to escape or balance slight environmental perturbations. Since it appears that such buffering reactions also involve functionally related groups of messages, and such functional coherent sets are often bound by RBPs, we speculate that RBPs may be significantly involved in this process. If so, the plethora of RBPs and their bound mRNAs define a magnitude of overlapping “posttranscriptional operons” and RNA regulons [1
]. This may constitute the primary mediator of environmental responses that may even shape the cell's individual behavior.