During each division, the protein content of cells doubles. How does that come about? Is the pattern of overall protein synthesis in the cell cycle constant? For example, is the increase in protein mass proportional to the mass of proteins already present in the cell at any given time? If so, proteins would be produced exponentially, much like money in a bank earning compound interest. The machinery for protein synthesis (i.e., ribosomes) is a significant part of the cell’s protein content 1, and one can see how an exponential pattern of protein synthesis could arise (Figure 1A, bottom panel). Ribosomes make proteins, which make more ribosomes, an ever-increasing amount of proteins, and so on. The above assumes that the newly made ribosomes become immediately available for protein synthesis, making the number of ribosomes per cell the critical determinant of protein synthesis rates 2, 3. Alternatively, if new proteins are made by a fixed absolute amount per unit of time in a cell cycle phase rather than a fixed percentage as in exponential synthesis, then we would have a linear increase in protein synthesis during that phase. Such a linear increase need not be constant in the cell cycle and models of different linear rates of protein synthesis rates at different cell cycle stages are possible 4, 5, 6. Since cells will double their total protein mass in the cell cycle, if the rate of protein synthesis is linear, the rate-to-total protein mass ratio would decline during the cell cycle unless there is an overall doubling of the linear rate of synthesis at one or more points (see Figure 1B, middle panel). In exponential synthesis, however, the rate-to-total protein mass ratio will remain constant throughout the cell cycle (Figure 1B, bottom panel).

–

FIGURE 1: Protein synthesis in the cell cycle. The cell cycle stage is on the x-axis in all cases. Most of the discussion in the text is about continuous synthesis models leading to the doubling of total protein in the cell during one cell cycle (middle and bottom panels). A periodic synthesis model is shown for comparison on the top. While for certain proteins (e.g., histones [3]) this model is valid, no evidence supports such a pattern for ’bulk’ protein synthesis. Adapted from [5, 6]. In (A), the expected amount of protein per cell (y-axis) is shown for three models. In (B), the expected ratios of the rate (of protein synthesis)-to mass (of total protein) graphs are shown. In (C), a schematic of how a given pattern of protein synthesis in the cell cycle might come about. At the top, a simple case of periodic synthesis arises from the periodic transcription of a gene. However, most genes are not transcribed in a cell cycle- dependent manner [7, 8], arguing again that this model cannot account for periodic ’bulk’ protein synthesis. The middle panel visualizes abrupt linear changes in ribosome numbers and/or activity (indicated with the yellow flashes), doubling protein synthesis rates in the later part of the cell cycle. Note that such changes need not be monotonic. A decline may follow a sharp increase, followed by another sharp rise. In all such cases, however, there will be an irregular, jagged pattern of the protein synthesis rate over the cell’s total protein, as shown in (B). The bottom panel in (C) graphically depicts an exponential increase in ribosome numbers in the cell, leading to a continuous, exponential pattern of protein synthesis rates and a constant ratio of the protein synthesis rate over the cell’s total protein. Combinations of the above models are theoretically possible.

We also need to keep in mind that budding yeast cells divide asymmetrically, and different patterns of global protein synthesis may be involved not only in different phases of the cell cycle, but also in cells of different generations. Once a Saccharomyces cerevisiae cell commits to a new round of cell division in late G1, at a point called Start, it will generate a bud. From then on, most of the new cell ‘growth’ in the S, G2, and M phases will be directed to the bud in a polarized manner 9. The continuously increasing bud size serves as a convenient proxy for cell cycle position (i.e., the bigger the bud, the more advanced the cell cycle stage). The bud will eventually become the new daughter cell at the end of cytokinesis.

WHY DOES IT MATTER?

The above considerations bear on the mechanisms involved in cell cycle-dependent translational control. If there are changes in linear rates, the expectation is that the levels of many proteins will be cell cycle-regulated. But if the pattern of protein synthesis is exponential, then the converse will be true, with most proteins not changing in concentration in the cell cycle. If global protein synthesis is exponential, with the concentration of ribosomes in the cell staying constant, translational control would be imparted in a transcript-specific manner (e.g., a particular mRNA interacts with some protein or other factors that change its translatability at different points in the cell cycle 10). If, on the other hand, we are dealing with linear patterns of protein synthesis that change in the cell cycle, then changes in the concentration or activity of the protein synthesis machinery could lead to specific translational control of transcripts with elements that make them more sensitive to global changes in ribosomes 11, 12. For example, the translation of mRNAs with sequences that limit the engagement of the encoded ORF with ribosomes (e.g., uORFs or secondary structure) would be disproportionately inhibited when the ribosome content is low 13.

Answering if and how protein synthesis rates are regulated in the cell cycle in yeast will have broader implications, such as understanding the link between protein synthesis and cancer development in human cells. For example, the metabolic fluxes of the large NCI-60 panel of cancer lines are proportional to their protein synthesis rate and their proliferation rate 14. Furthermore, the levels of ~100 proteins that are the most upregulated in abundance in the fastest proliferating lines in the NCI-60 panel are known outputs of the MYC oncogene 15, which has a well-established role in driving protein synthesis in cancer 16, 17. These observations may have therapeutic consequences. NCI-60 cell lines with high protein synthesis rates were more sensitive to inhibitors of the target of rapamycin (TOR) 14, a kinase complex that promotes anabolic processes in all eukaryotes 18, 19.

HOW IS IT STUDIED? METHODS AND APPORACHES

WHAT ARE THE FINDINGS?

As summarized in Table 1, the approaches and findings vary. Let’s start with the older studies that used metabolic labeling to query cell populations. Focusing on the synthesis of the abundant ribosomal proteins, it was reported that they are made throughout the cell cycle, but the study did not have the resolution to detect linear vs. exponential patterns 34. The widely cited studies introducing the double labeling technique 6, 35, 36 reported exponential protein synthesis (corresponding to the bottom panels in Figure 1). These conclusions were challenged later 4 because while the data supported an exponential pattern of synthesis in the G1 phase, the cell fractionation was not good enough to draw the same conclusion for later stages of the cell cycle. Anecdotally, from experiments in our lab, we can attest that the cell size distribution of age-fractionated elutriated samples becomes irregular and broad as one collects successive fractions, indicative of worsening synchrony. Two more older studies combined metabolic labeling with isopycnic centrifugation for cell fractionation (see Table 1). One found that the synthesis rate of most of the hundreds of proteins they analyzed did not change in the cell cycle 37, corresponding to the bottom panels in Figure 1. However, their cutoffs for significant changes may have been too stringent (>3-4-fold) to draw these conclusions 4. A different group reported that the ratio of the labels in a double labeling scheme was not constant in the cell cycle and reached the highest levels in the G2 phase 38, corresponding to the middle panels in Figure 1. What can we conclude then, if anything, from all these metabolic labeling studies? No study reported changing linear rates of protein synthesis early in the cell cycle in the G1 phase, but the issue is unresolved about later phases. A more effective strategy for these labeling experiments would be to elutriate and collect cells in the early G1 daughter fraction only, which is typically highly homogeneous, and track protein synthesis rates as they progress in the cell cycle.

Other population studies using mass spectrometry measured protein abundances in the cell cycle but not synthesis rates. The first study sampled four points in the cell cycle after release from pheromone arrest 30. Only a small number of proteins (<10%) changed abundance significantly (e.g., 83 proteins were upregulated by at least 2-fold between any two points). A study from our lab used centrifugal elutriation to isolate early G1 cells, which were then sampled at eight different points in the cell cycle. We found 333 proteins whose relative abundance changed by at least 2-fold between any two points 28. From experiment-matched samples, transcripts encoding ribosomal proteins and other gene products involved in ribosome biogenesis were significantly upregulated, peaking in the G1 phase. Still, these changes were not evident at the protein level 28. Furthermore, in the same study, when assembled ribosomes were isolated and quantified by mass spectrometry, there was no change in their relative abundance 28. Hence, the increased levels of transcripts encoding ribosome biogenesis components were somehow buffered, and the ribosome concentration did not change at all during the cell cycle. These studies argue against significant changes in global protein synthesis in the cell cycle. However, another mass spectrometry-based study reported an increase in G1 of the mRNA and protein levels for components of the protein synthesis machinery, including ribosomal proteins 29. Albeit significant, the rise in the abundance of ribosomal proteins in G1 was small (~8%) in that study29 and below the threshold used in the others (2-fold; 28, 30).

Fluorescence microscopy has been used to continuously record the levels of constitutive reporter proteins in the cell cycle, usually displayed by integrating the fluorescence signal over each cell at each time point and size. However, the deduced pattern of protein production varies among studies, from exponential39 (corresponding to the bottom panel in Figure 1A), to a constant linear rate during each cell cycle 40 to a more dynamic pattern, peaking in late G1 and exceeding the rate of cell size increase 41, and then peaking again later in the cell cycle 42, 43, (corresponding to variations of the middle panel in Figure 1A). Monitoring the levels of two ribosomal proteins tagged with GFP also revealed dynamic changes in their abundance, peaking in G1, then dropping, but rising again in G2/M 42, corresponding again to variations of the middle panel in Figure 1A. These results were reinforced in yet another study that used the abundance of a reporter GFP construct as a proxy for protein production, arguing that protein synthesis is not monotonic 43. An impressive new study interrogated millions of cells from thousands of individual strains, each expressing every ORF from its endogenous promoter as a GFP-tagged protein 22. Then, using fluorescence imaging of unperturbed cells over time, the location of each protein in the cell and its levels were quantified as a function of cell cycle position. This single-cell analysis found ~800 proteins that change in concentration in the cell cycle, about 60% of them peaking in late G1, ribosomal proteins the most prominent group among them 22. The median increase in the concentration of proteins peaking in G1 was about 30% 22. These findings argue for dynamic protein synthesis in the cell cycle (i.e., corresponding again to variations of the middle panel in Figure 1A), but the mechanistic underpinnings still need to be discovered. It should also be pointed out that translational control in the cell cycle is not widespread. Based on two ribosome profiling studies, less than 5% of all mRNAs show differential translational efficiency in the cell cycle 22, 23.

LOOKING AHEAD

How does one reconcile the evidence outlined above? The different conclusions about whether ribosomal protein concentration changes in the cell cycle may reflect the different thresholds used in each study. Still, whether and how these changes also lead to changes in the ribosome content and global protein synthesis rates has yet to be firmly established. Do these extra ribosomal proteins assemble into functional ribosomes? An important consideration here is the stability of ribosomal proteins and ribosomes. Most ribosomal proteins have a half-life of several hours, based on the values listed in the Saccharomyces Genome Database http://yeastgenome.org 44, which exceeds the cell cycle length in typical laboratory conditions. The half-life of ribosomes has also been reported to be very high in proliferating yeast cells, with no decay detected in a 5h period 45. In other organisms, a half-life of ribosomes of several days has been reported for rat liver ribosomes46, 47 and plants 48. The assumption that a short temporal increase in the abundance of ribosomal parts is a direct proxy for ribosome content and protein synthesis rates is far from straightforward.

Another confounding variable among the studies is the different media used (see Table 1). The concentration of ribosomes in the cell is higher in rich nutrients than in poor ones 49. Nutrient-sensitive pathways (e.g., the protein kinase A (PKA) or the target of rapamycin (TOR) signaling networks) respond to the available nutrients (e.g., quality of the carbon and nitrogen sources) to adjust ribosome synthesis. However, it is unknown if these or any other pathways change the rate of protein synthesis in a cell cycle-dependent manner in cells growing and dividing in a steady state in a given nutrient environment. Such causal links would need to be examined in the future, especially since recent work has shown that the activity of the above pathways is dynamic in the cell cycle 42, 43, 50. As to how translation could be affected, while control of protein synthesis rates is often assumed to be exercised primarily at the initiation stage, elongation control due to unusual codon usage and ribosome pausing has gained attention 51. We note, however, that among components of the translation machinery it is mutations in initiation factors that overwhelmingly lead to specific cell cycle phenotypes in budding yeast 52. Furthermore, there is no evidence for altered codon usage in the budding yeast cell cycle 28. Changes in tRNA abundance are also unlikely to impact the cell cycle, because the half-life of mature tRNAs (>10 h to several days; 53) far exceeds the duration of the cell cycle in most conditions. Hence, if rates of protein synthesis change in the cell cycle, pathways that impinge on initiation are likely to be involved.

Future research holds promise in settling the discrepancies and filling the gaps in our knowledge. Including metabolic labeling to query protein synthesis rates, preferably in a single-cell context, could combine the advantages of all the methods. This approach has been recently demonstrated in fission yeast, where it was found that translation rates scale with cell size through most of the cell cycle in this organism, and plateau only in very large cells in mitosis 20. In the near future, we can also expect detailed translatomics of each mRNA in the cell cycle at single-cell resolution, thanks to rapidly evolving technologies 54. Once the patterns of global protein synthesis are firmly established, progress on specific mechanisms of translational control in the cell cycle will follow.