Translation is the process by which cellular machines called ribosomes use the information encoded in genes to make proteins . Every organism requires two types of RNA molecules to make new proteins: ribosomal RNAs (rRNAs, which form part of the ribosome) and transfer RNAs (tRNAs, which transport the amino acid molecules that form proteins to the ribosomes). These RNA molecules are coded in the genome, but different organisms have different ‘copy numbers’: some genomes contain just a few copies of each of these genes, while others have thousands.

This apparent redundancy – the presence of several copies of the same gene – is puzzling because it is costly to make and maintain DNA and RNA. This leads to an important question: how does redundancy in these important genes (coding for tRNAs and rRNAs) evolve? The answer is key to understanding how one of the most fundamental cellular processes, the making of proteins from DNA, has evolved.

A possible reason for organisms to have many copies of the genes required to make proteins is to allow rapid translation, which allows cells to divide faster, and populations of cells to grow more quickly. However, this would likely mean that, when nutrients are scarce, carrying and translating many copies of the same gene would become a burden on the cell. Raval et al. set out to test this idea by measuring the costs and benefits of seemingly redundant translation components.

To do this, Raval et al. deleted some of the redundant gene copies in the bacterium Escherichia coli and asked if that changed bacterial growth. The experiments showed that when nutrients were plentiful, cells with more copies of the genes (high redundancy) were better able to use the nutrients and divide rapidly. However, when nutrients were limited, bacteria with extra gene copies divided more slowly, showing that the extra genes are indeed a big burden on the cell.

Raval et al. propose that nutrients available in the environment ultimately determine whether redundancy of the translation machinery is a blessing or a curse. This suggests that the redundancy and underlying growth strategies of different organisms are forged by their experiences of feast and famine during their evolutionary past.

Importantly, by testing the joint effect of many different molecules involved in translation, Raval et al. uncovered several strategies that may maximize bacterial growth and protein production. Their results could thus be useful for optimizing the synthesis of important products that use growing cells as factories – from beer to insulin – where the rate of growth is critical.

In the early 1960s, the degeneracy of the genetic code was revealed in the context of multiple synonymous codons encoding a given amino acid. A large body of work since then has uncovered an astonishing degree of redundancy in the translation apparatus. The redundancy is often qualitative, whereby some components can functionally compensate for others. This includes the pool of tRNA molecules that read mRNA codons and deliver the appropriate amino acid during translation. tRNAs with different anticodons may read distinct synonymous codons but carry the same amino acid (tRNA isoacceptors) and are thus functionally degenerate. In addition, tRNA modifying enzymes (MEs) post-transcriptionally alter specific tRNAs (‘target’ tRNAs, i.e. targets of modification), allowing them to read codons that are otherwise decoded by ‘non-target’ tRNAs that are not modified by MEs (Grosjean, 2009). Hence, some non-target tRNAs could be redundant because their function can be carried out by target tRNAs after they get modified. For instance, in Escherichia coli the codon CCG can be decoded by the non-target tRNA_CGG (encoded by the gene proK), but also by the target tRNA_UGG (encoded by proM) after the U at position 34 is modified to cmo⁵U by the cmo modification pathway (in principle rendering the gene proK redundant). MEs are thus critical for maximizing cellular decoding capacity (Diwan and Agashe, 2018; Grosjean et al., 2010), and maintaining translational capability requires either a very diverse tRNA pool or the presence of MEs that allows for a compact tRNA gene set (Diwan and Agashe, 2018; Marck and Grosjean, 2002; Rocha, 2004; Wald and Margalit, 2014). In addition, quantitative redundancy is conferred by high gene copy number (GCN), such that multiple genes can perform identical functions. For instance, bacterial genomes often carry many copies of genes encoding tRNAs with the same anticodon (tRNA isotypes) (Chan and Lowe, 2009). Similarly, cells typically have many copies of rRNA genes (Roller et al., 2016), which govern ribosome availability (Nomura et al., 1980). Thus, the translation machinery is predicted to be functionally redundant at many levels, with qualitative (multiple codons, tRNA isoacceptors, and MEs) as well as quantitative (GCN) redundancy. However, in many cases, the predicted functional redundancy is not experimentally verified, and it remains unclear whether it influences fitness. We asked: what are the fitness costs and benefits of translational redundancy, and under what conditions do they manifest? Could these costs and benefits explain the evolution of a highly redundant translation apparatus?

In bacteria, selection for rapid growth (often facilitated by nutrient availability) is thought to be an important force that shapes the evolution of many translation components. The maximum possible growth rate is determined by translation efficiency, which in turn depends on the concentrations of ribosomes and tRNAs (Ehrenberg and Kurland, 1984; Hu et al., 2020; Kurland and Ehrenberg, 1987; Kurland and Hughes, 1996). Across species, there are striking positive correlations between maximal growth rate and the number of tRNA and rRNA genes (Dethlefsen and Schmidt, 2007; Ikemura, 1985; Rocha, 2004; Roller et al., 2016; Vieira-Silva and Rocha, 2010; Weissman et al., 2021). Thus, we expect that GCN redundancy in key translation components should be especially beneficial during rapid growth in a nutrient-rich niche. In contrast, under nutrient limitation, expressing redundant genes should be costly because translational output remains constrained by nutrients. However, this overarching growth rate-dependent selection may shape the redundancy of translation components differentially. For instance, selection should have the maximum impact on rRNA, whose GCN shows the strongest correlation with growth rate (Rocha, 2004; Roller et al., 2016). Ribosomal RNAs constitute up to 85% of all RNA in rapidly growing E. coli (Bremer and Dennis, 1996) and their concentration is most limiting for translation, likely because rRNA constitutes the core catalyst of ribosome assembly (Bremer and Dennis, 1996). During fast growth, cellular rRNA abundance increases by ~250%, whereas total tRNA increases only by ~80% (Dong et al., 1996). Consequently, the cellular ratio of tRNAs to rRNAs decreases under rapid growth, indicating greater investment in rRNA (Dittmar et al., 2004). Indeed, deleting rRNA operons in E. coli reduces fitness in rich media, but improves fitness in poor media (Condon et al., 1995; Gyorfy et al., 2015; Stevenson and Schmidt, 2004).

In contrast, the growth rate-dependent impacts of redundancy in tRNA GCN remain largely unexplored, with the exception of initiator tRNA genes in E. coli. As predicted, in this case, the loss of some gene copies is deleterious in rich media and advantageous in poor media (Samhita et al., 2014). However, the growth impact of elongator tRNA GCN is not known. The paucity of data for tRNA redundancy is glaring because bacteria show enormous variation in tRNA pools, driven by evolutionary changes in tRNA GCN as well as MEs (Ayan et al., 2020; Bedhomme et al., 2019; Diwan and Agashe, 2018; Saks et al., 1998; Wald and Margalit, 2014). Furthermore, the impacts of redundancy are predicted to vary substantially across different tRNA genes. For instance, mature tRNA pools are biased towards ‘major’ tRNAs that connect frequently used amino acids and codons, and are more strongly correlated with growth rate (Berg and Kurland, 1997; Dong et al., 1996) and rRNA GCN (Mahajan and Agashe, 2018) compared to minor tRNAs. The loss of redundancy in these tRNAs should impose a larger fitness cost, a hypothesis that remains untested. Thus, despite strong comparative evidence for growth rate-driven selection, our understanding of the evolution and impacts of redundancy in translation components is far from complete.

While the fitness costs and benefits of redundancy should ultimately be shaped by nutrient availability, redundancy may itself arise via different mechanisms for different translation components, and hence selection may shape the components in distinct ways. For instance, with rRNAs, the loss of GCN redundancy is buffered by strong compensatory upregulation of ‘backup’ gene copies. An E. coli strain with deletion of 6 out of 7 rRNA operons can produce about half of the normal levels of rRNA (Asai et al., 1999). As a result, the deletion of a few rRNA genes only moderately reduces the growth rate in rich media (Gyorfy et al., 2015), though the direction of the effect is reversed in poor media (see above). Are tRNA pools similarly regulated? While this is possible (e.g. in response to nutrient availability Fessler et al., 2020; Sørensen et al., 2018), we have no data on compensatory regulation of the remaining tRNA copies after gene loss. Furthermore, tRNA redundancy is modulated not only by GCN but also by MEs that allow target tRNAs to perform the function of non-target tRNAs. Thus, tRNA gene loss could be buffered by the regulation of other tRNA gene copies, and/or by the action of MEs. Interestingly, fast-growing bacteria tend to have low tRNA diversity (Rocha, 2004), with their decoding capacity likely maintained by the action of multiple tRNA modification pathways (Diwan and Agashe, 2018). Hence, the fitness consequences of tRNA gene loss should be contingent on the availability of ME backups. Conversely, the joint deletion of non-target tRNAs and MEs is predicted to be more costly than the loss of either component alone (Diwan and Agashe, 2018; Wald and Margalit, 2014). Thus, the mechanisms that mediate redundancy as well as the interactions between translation components are important to fully understand the evolution of translational redundancy.

The patterns noted above suggest a hierarchical organization, whereby redundancy in some components and genes is more important than others (e.g. rRNAs vs. tRNAs, and major vs. minor tRNA isotypes). However, as discussed above we have very limited empirical evidence for the fitness consequences of redundancy in different translation components, particularly in the case of tRNA pools. We addressed these gaps by analyzing the nutrient-dependent impact of changing redundancy in multiple translational components, alone as well as in combination. Specifically, we tested the following predictions: (1) redundancy in tRNA and rRNA GCN, on the whole, should be important to maintain rapid growth, and the benefits of increased redundancy should be proportional to the achievable growth rate (2) broadly, a reduction in rRNA GCN should have stronger fitness impacts than tRNA GCN (3) across tRNAs, the loss of redundancy should be most impactful for major tRNA isotypes, and for non-target tRNAs when combined with the loss of a relevant ME (4) the fitness impact of reduced redundancy should increase with the severity of the loss, e.g., due to the deletion of multiple gene copies or multiple translational components. We worked with E. coli because it has a highly redundant translation machinery (Diwan and Agashe, 2018; Wald and Margalit, 2014) that allowed us to test the impacts of successive losses of redundancy at the level of rRNA genes, tRNA pools, and tRNA modifying enzymes. We first show that, as expected, many components of the translation machinery are indeed redundant with respect to fitness. We then test our predictions by measuring the context-dependent costs and benefits of this redundancy. Our results reveal layered factors that may have shaped the evolution of the translation machinery in bacteria.

The process of protein synthesis is central to life and is especially important to understand bacterial evolution given the direct link between translation, growth rate, and fitness. Translation rate is affected by several genomic components (e.g. tRNAs, rRNAs, tRNA modifying and charging enzymes) and environmental factors (e.g. nutrient availability). Comparative analyses show that the genomic components have different degrees of redundancy across taxa, and suggest that this redundancy (in the fitness context) should be shaped by the strength of ecological selection for rapid growth (Rocha, 2004; Roller et al., 2016). While lateral gene transfer and gene deletion/duplication make translation components labile even in the absence of selection, the correlation between maximum growth rate and redundancy of tRNA or rRNA GCN suggests a strong role for selection (Rocha, 2004; Roller et al., 2016), likely imposed by nutrients from the environment. Hence, nutrients ultimately limit translation. Cells can meet this environmental limit and maximize fitness by shaping genomic factors to achieve the maximum attainable translational output. Such modulation of the cellular machinery should occur at physiological (short-term) as well as evolutionary timescales. For instance, cells can control translation rates via rapid regulation of translation components (Wilusz, 2015), including via degradation of ribosomes and tRNAs during nutrient starvation (Fessler et al., 2020; Sørensen et al., 2018). Across-species patterns of rRNA and tRNA GCN are consistent with such selection, as discussed in the Introduction. Similar arguments can also be made for redundancy across other translation components. Together, this suggests that the environment sets the limits of translation, according to which natural selection shapes the genomic redundancy of translational components. However, empirical evidence for a common underlying selection pressure shaping redundancy across various translational components has been missing.

Here, we provide such evidence, showing that broadly speaking, several components of the E. coli translation apparatus are indeed functionally redundant and that the costs and benefits of this redundancy vary with nutrient availability. When nutrients permit rapid growth, the loss of redundancy in both tRNAs and rRNAs is detrimental, as these components become limiting for translation. This is especially true for multiple deletions of abundantly used components such as rRNAs, MEs that modify many different tRNAs, and frequently used tRNAs (such as glycine and leucine). These results support prior predictions of larger fitness consequences following the loss of major tRNAs that read abundant codons and respond strongly to fast growth (Dong et al., 1996), or tRNAs that make larger contributions to the tRNA pool (Bloom-Ackermann et al., 2014; Kanaya et al., 1999). The observed variability in fitness impacts across tRNAs is therefore at least partially explained by the relative use of different codons. Our results also support the broad prediction that high tRNA levels should be most critical during rapid growth (Mahajan and Agashe, 2018; Vieira-Silva and Rocha, 2010), when the correlation between tRNA levels and gene copy number is strongest (Dong et al., 1996). Conversely, when nutrients are limiting, both rRNA as well as tRNA gene loss is beneficial. Importantly, we show that the expression cost of high translational redundancy of WT E. coli is not met by increased translation rate in poor media, and imposes a substantial net fitness cost.

While these observations suggest that nutrient availability can guide the evolutionary optimization of the cost-to-benefit ratio of the bacterial translation machinery, interactions (and potential hierarchies) amongst different components suggest multiple routes of optimization (Figure 7). We observe that the loss of rRNA genes is generally more impactful than the loss of tRNAs; and when rRNAs are limiting, the additional loss of tRNAs has a relatively weak effect. We suggest that this is because rRNAs set the first internal limit on translation rate, as predicted by prior work (see the Introduction). However, further experiments are necessary to separate the independent impacts of rRNA and tRNA genes linked within operons. Interestingly, even in complex bacterial communities, the addition of resources enriched for taxa with more rRNA copies (Wu et al., 2017) and the growth rate response to nutrient addition was positively correlated with rRNA copy number (Li et al., 2019). The next internal limiting factor for translation appears to be available tRNA pools, determined by the nested impacts of tRNA gene copy number, transcriptional regulation of gene expression, and MEs. Overall, tRNA gene copy number has a stronger impact on translation and fitness than the regulation of different isotype copies, corroborating prior results with yeast (Percudani et al., 1997). However, in the special case of non-target tRNAs, the relevant MEs have a stronger impact than the tRNA genes. This is not surprising given the predicted functional redundancy between MEs and non-target tRNAs (Diwan and Agashe, 2018; Grosjean et al., 2010). As a result of this layering, a severe loss of tRNA redundancy becomes important for fitness only when many copies of an isotype are lost, or when a non-target tRNA is co-deleted with a relevant modifying enzyme. Conversely, these results predict that an increase in redundancy (e.g. due to occupation of a more nutrient-rich niche and selection for rapid growth) may occur via increasing isotype gene copy numbers, increasing non-target tRNAs, or gaining a relevant modifying enzyme. These predictions from our E. coli data closely match the patterns observed in comparative analyses across bacteria: selection for rapid growth is strongly correlated with gene copy number (Rocha, 2004), and lineages may lose either MEs or non-target tRNAs, but not both (Diwan and Agashe, 2018).

Figure 7

Download asset Open asset

Together with prior work, our results also suggest that fast-growing organisms such as E. coli have evolved to rely strongly on gene copy numbers to maximize translation rates. Presumably, such species are either able to bear the costs of maintenance of surplus tRNA genes during periods of slow growth, or across longer evolutionary timescales the extra tRNA copies provide a net benefit. We therefore suggest a model whereby bacterial growth rate is primarily limited by external nutrient availability, then by rRNA molecules, and finally by tRNA pools (determined by tRNA GCN and MEs, and secondarily via tRNA gene regulation). Thus, the layered costs and benefits of high translational redundancy – both within and across distinct components – are ultimately determined by the environmental context. Importantly, this model predicts that prolonged selection for rapid growth should cause successive evolutionary changes in the redundancy of different translation components, with the general order of events determined (and parallel routes offered) by the hierarchical layering of components (Figure 7).

We hope that future work will test this model and enrich it by considering additional translational components and factors that may drive their evolution. For instance, genome GC content is strongly associated with tRNA GCN and diversity across the bacterial phylogeny (Diwan and Agashe, 2018; Wald and Margalit, 2014), as well as with codon bias (Hershberg and Petrov, 2010). An understanding of the ecological and evolutionary pressures that drive shifts in GC content would thus be useful to understand the impact of selection for rapid growth on genome GC. Parallel to our results, prior work shows that the two copies of tuf genes – encoding the E. coli translation elongation factor EF-Tu – also have distinct, medium-dependent impacts on growth rate (Zuurmond et al., 1999). It would thus be interesting to include EF-Tu redundancy in our model. It is also worth considering other evolutionary processes that can alter redundancy, such as genetic drift that may facilitate the loss of MEs (Diwan and Agashe, 2018). In addition, here we have focused on selection acting on translation rate, primarily tested using growth rate. However, our results hint at interesting effects of the loss of redundancy on other growth parameters such as yield and lag phase that may be orthogonal to growth rate. These measurements had a low resolution in our study, but explicit and better analysis of such impacts may reveal the effects of selection in niches where yield or survival (rather than growth rate) determine fitness. Finally, we note that selection may act on other cellular functions performed by some translational components (Shepherd and Ibba, 2015), or on translational accuracy (Gingold and Pilpel, 2011). While global mistranslation can provide fitness benefits in stressful contexts (Jones et al., 2011; Samhita et al., 2020), the nature and strength of selection acting on translational accuracy and the relationship between translation rate and accuracy remain poorly understood (Drummond and Wilke, 2009). Nonetheless, prior work suggests that both tRNA pools and MEs influence translation accuracy (Manickam et al., 2016) and protein aggregation (Fedyunin et al., 2012). Thus, selection for accuracy could also shape the evolution of tRNAs and MEs. We therefore suggest that expanding the layers of organization of translational components in our model will prove fruitful in gaining a deeper understanding of the evolution of translational redundancy.

In summary, our experiments demonstrate that several components of the translation machinery are redundant in E. coli, the costs and benefits of which vary based on nutrient availability – an environmental variable that likely shaped the redundancy in the first place. Our results support the broad idea that translational limits imposed by different components and their interactions generate multiple translational optima and make many paths feasible (Grosjean et al., 2014; Higgs and Ran, 2008), depending on the selective context (Rocha, 2004). We propose a model with hierarchies of translation components that may allow for the maximization of translational output and fitness. It remains to be seen to what extent these hierarchies operate across diverse taxa, and which of the possible parallel routes are taken during the course of bacterial evolution in nature. Further studies on the layers and hierarchies connecting translational components will thus shed light on the molecular toolkits underlying evolutionary transitions between slow versus rapid translation.

All experimental data are provided as supplementary files. Raw YAMAT-seq reads and analysis files are available at the NCBI Gene Expression Omnibus (GEO accession number GSE198606).

1. 1. Gallie J
   2. Ngan WY
   3. Raval PK
   4. Agashe D

   (2022) NCBI Gene Expression Omnibus

   ID GSE198606. YAMAT-seq of mature tRNA pools in the bacterium Escherichia coli str. K-12 substr. MG1655 and derivatives.

   https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE198606

1. 1. Anders S
   2. Huber W

   (2010) Differential expression analysis for sequence count data

   Genome Biology 11:R106.

   https://doi.org/10.1186/gb-2010-11-10-r106

   • PubMed
   • Google Scholar
2. 1. Ardell DH
   2. Kirsebom LA

   (2005) The genomic pattern of tDNA operon expression in E. coli

   PLOS Computational Biology 1:e12.

   https://doi.org/10.1371/journal.pcbi.0010012

   • PubMed
   • Google Scholar
3. 1. Asai T
   2. Condon C
   3. Voulgaris J
   4. Zaporojets D
   5. Shen B
   6. Al-Omar M
   7. Squires C
   8. Squires CL

   (1999) Construction and initial characterization of Escherichia coli strains with few or no intact chromosomal rRNA operons

   Journal of Bacteriology 181:3803–3809.

   https://doi.org/10.1128/JB.181.12.3803-3809.1999

   • PubMed
   • Google Scholar
4. 1. Ayan GB
   2. Park HJ
   3. Gallie J

   (2020) The birth of a bacterial tRNA gene by large-scale, tandem duplication events

   eLife 9:e57947.

   https://doi.org/10.7554/eLife.57947

   • PubMed
   • Google Scholar
5. 1. Bedhomme S
   2. Amorós-Moya D
   3. Valero LM
   4. Bonifaci N
   5. Pujana MÀ
   6. Bravo IG
   7. Gonzalez J

   (2019) Evolutionary changes after translational challenges imposed by horizontal gene transfer

   Genome Biology and Evolution 11:814–831.

   https://doi.org/10.1093/gbe/evz031

   • PubMed
   • Google Scholar
6. 1. Berg OG
   2. Kurland CG

   (1997) Growth rate-optimised tRNA abundance and codon usage

   Journal of Molecular Biology 270:544–550.

   https://doi.org/10.1006/jmbi.1997.1142

   • PubMed
   • Google Scholar
7. 1. Bloom-Ackermann Z
   2. Navon S
   3. Gingold H
   4. Towers R
   5. Pilpel Y
   6. Dahan O

   (2014) A comprehensive tRNA deletion library unravels the genetic architecture of the trna pool

   PLOS Genetics 10:e1004084.

   https://doi.org/10.1371/journal.pgen.1004084

   • PubMed
   • Google Scholar
8. 1. Bremer H
   2. Dennis PP

   (1996) Modulation of chemical composition and other parameters of the cell at different exponential growth rates

   EcoSal Plus 3:3.

   https://doi.org/10.1128/ecosal.5.2.3

   • PubMed
   • Google Scholar
9. 1. Chan PP
   2. Lowe TM

   (2009) GtRNAdb: a database of transfer RNA genes detected in genomic sequence

   Nucleic Acids Research 37:D93–D97.

   https://doi.org/10.1093/nar/gkn787

   • PubMed
   • Google Scholar
10. 1. Chan PP
    2. Lowe TM

    (2016) GtRNAdb 2.0: an expanded database of transfer RNA genes identified in complete and draft genomes

    Nucleic Acids Research 44:D184–D189.

    https://doi.org/10.1093/nar/gkv1309

    • PubMed
    • Google Scholar
11. 1. Condon C
    2. Liveris D
    3. Squires C
    4. Schwartz I
    5. Squires CL

    (1995) rRNA operon multiplicity in Escherichia coli and the physiological implications of rrn inactivation

    Journal of Bacteriology 177:4152–4156.

    https://doi.org/10.1128/jb.177.14.4152-4156.1995

    • PubMed
    • Google Scholar
12. 1. Datsenko KA
    2. Wanner BL

    (2000) One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products

    PNAS 97:6640–6645.

    https://doi.org/10.1073/pnas.120163297

    • PubMed
    • Google Scholar
13. 1. Delaney NF
    2. Kaczmarek ME
    3. Ward LM
    4. Swanson PK
    5. Lee MC
    6. Marx CJ

    (2013) Development of an optimized medium, strain and high-throughput culturing methods for Methylobacterium extorquens

    PLOS ONE 8:e62957.

    https://doi.org/10.1371/journal.pone.0062957

    • PubMed
    • Google Scholar
14. 1. Dethlefsen L
    2. Schmidt TM

    (2007) Performance of the translational apparatus varies with the ecological strategies of bacteria

    Journal of Bacteriology 189:3237–3245.

    https://doi.org/10.1128/JB.01686-06

    • PubMed
    • Google Scholar
15. 1. Dittmar KA
    2. Mobley EM
    3. Radek AJ
    4. Pan T

    (2004) Exploring the regulation of tRNA distribution on the genomic scale

    Journal of Molecular Biology 337:31–47.

    https://doi.org/10.1016/j.jmb.2004.01.024

    • PubMed
    • Google Scholar
16. 1. Dittmar KA
    2. Sørensen MA
    3. Elf J
    4. Ehrenberg M
    5. Pan T

    (2005) Selective charging of tRNA isoacceptors induced by amino-acid starvation

    EMBO Reports 6:151–157.

    https://doi.org/10.1038/sj.embor.7400341

    • PubMed
    • Google Scholar
17. 1. Diwan GD
    2. Agashe D

    (2018) Wobbling forth and drifting back: the evolutionary history and impact of bacterial tRNA modifications

    Molecular Biology and Evolution 35:2046–2059.

    https://doi.org/10.1093/molbev/msy110

    • PubMed
    • Google Scholar
18. 1. Dong H
    2. Nilsson L
    3. Kurland CG

    (1996) Co-Variation of tRNA abundance and codon usage in Escherichia coli at different growth rates

    Journal of Molecular Biology 260:649–663.

    https://doi.org/10.1006/jmbi.1996.0428

    • PubMed
    • Google Scholar
19. 1. Drummond DA
    2. Wilke CO

    (2009) The evolutionary consequences of erroneous protein synthesis

    Nature Reviews. Genetics 10:715–724.

    https://doi.org/10.1038/nrg2662

    • PubMed
    • Google Scholar
20. 1. Edgar R
    2. Domrachev M
    3. Lash AE

    (2002) Gene expression omnibus: NCBI gene expression and hybridization array data Repository

    Nucleic Acids Research 30:207–210.

    https://doi.org/10.1093/nar/30.1.207

    • PubMed
    • Google Scholar
21. 1. Ehrenberg M
    2. Kurland CG

    (1984) Costs of accuracy determined by a maximal growth rate constraint

    Quarterly Reviews of Biophysics 17:45–82.

    https://doi.org/10.1017/s0033583500005254

    • PubMed
    • Google Scholar
22. 1. Elf J
    2. Nilsson D
    3. Tenson T
    4. Ehrenberg M

    (2003) Selective charging of tRNA isoacceptors explains patterns of codon usage

    Science 300:1718–1722.

    https://doi.org/10.1126/science.1083811

    • PubMed
    • Google Scholar
23. 1. Fedyunin I
    2. Lehnhardt L
    3. Böhmer N
    4. Kaufmann P
    5. Zhang G
    6. Ignatova Z

    (2012) tRNA concentration fine tunes protein solubility

    FEBS Letters 586:3336–3340.

    https://doi.org/10.1016/j.febslet.2012.07.012

    • PubMed
    • Google Scholar
24. 1. Fessler M
    2. Gummesson B
    3. Charbon G
    4. Svenningsen SL
    5. Sørensen MA

    (2020) Short-term kinetics of rRNA degradation in Escherichia coli upon starvation for carbon, amino acid or phosphate

    Molecular Microbiology 113:951–963.

    https://doi.org/10.1111/mmi.14462

    • PubMed
    • Google Scholar
25. 1. Gingold H
    2. Pilpel Y

    (2011) Determinants of translation efficiency and accuracy

    Molecular Systems Biology 7:481.

    https://doi.org/10.1038/msb.2011.14

    • PubMed
    • Google Scholar
26. Book

    1. Grosjean H

    (2009) DNA and RNA Modification Enzymes

    CRC Press.

    https://doi.org/10.1201/9781498713153

    • Google Scholar
27. 1. Grosjean H
    2. de Crécy-Lagard V
    3. Marck C

    (2010) Deciphering synonymous codons in the three domains of life: co-evolution with specific tRNA modification enzymes

    FEBS Letters 584:252–264.

    https://doi.org/10.1016/j.febslet.2009.11.052

    • PubMed
    • Google Scholar
28. 1. Grosjean H
    2. Breton M
    3. Sirand-Pugnet P
    4. Tardy F
    5. Thiaucourt F
    6. Citti C
    7. Barré A
    8. Yoshizawa S
    9. Fourmy D
    10. de Crécy-Lagard V
    11. Blanchard A

    (2014) Predicting the minimal translation apparatus: lessons from the reductive evolution of mollicutes

    PLOS Genetics 10:e1004363.

    https://doi.org/10.1371/journal.pgen.1004363

    • PubMed
    • Google Scholar
29. 1. Gyorfy Z
    2. Draskovits G
    3. Vernyik V
    4. Blattner FF
    5. Gaal T
    6. Posfai G

    (2015) Engineered ribosomal RNA operon copy-number variants of E. coli reveal the evolutionary trade-offs shaping rRNA operon number

    Nucleic Acids Research 43:1783–1794.

    https://doi.org/10.1093/nar/gkv040

    • PubMed
    • Google Scholar
30. 1. Hershberg R
    2. Petrov DA

    (2010) Evidence that mutation is universally biased towards AT in bacteria

    PLOS Genetics 6:e1001115.

    https://doi.org/10.1371/journal.pgen.1001115

    • PubMed
    • Google Scholar
31. 1. Higgs PG
    2. Ran W

    (2008) Coevolution of codon usage and tRNA genes leads to alternative stable states of biased codon usage

    Molecular Biology and Evolution 25:2279–2291.

    https://doi.org/10.1093/molbev/msn173

    • PubMed
    • Google Scholar
32. 1. Hu XP
    2. Dourado H
    3. Schubert P
    4. Lercher MJ

    (2020) The protein translation machinery is expressed for maximal efficiency in Escherichia coli

    Nature Communications 11:5260.

    https://doi.org/10.1038/s41467-020-18948-x

    • PubMed
    • Google Scholar
33. 1. Hu XP
    2. Lercher MJ

    (2021) An optimal growth law for RNA composition and its partial implementation through ribosomal and tRNA gene locations in bacterial genomes

    PLOS Genetics 17:e1009939.

    https://doi.org/10.1371/journal.pgen.1009939

    • PubMed
    • Google Scholar
34. 1. Ikemura T

    (1985) Codon usage and tRNA content in unicellular and multicellular organisms

    Molecular Biology and Evolution 2:13–34.

    https://doi.org/10.1093/oxfordjournals.molbev.a040335

    • PubMed
    • Google Scholar
35. 1. Jensen KF

    (1993) The Escherichia coli K-12 “ wild types” W3110 and MG1655 have an rph frameshift mutation that leads to pyrimidine starvation due to low pyrE expression levels

    Journal of Bacteriology 175:3401–3407.

    https://doi.org/10.1128/jb.175.11.3401-3407.1993

    • PubMed
    • Google Scholar
36. 1. Jones TE
    2. Alexander RW
    3. Pan T

    (2011) Misacylation of specific nonmethionyl tRNAs by a bacterial methionyl-tRNA synthetase

    PNAS 108:6933–6938.

    https://doi.org/10.1073/pnas.1019033108

    • PubMed
    • Google Scholar
37. 1. Kanaya S
    2. Yamada Y
    3. Kudo Y
    4. Ikemura T

    (1999) Studies of codon usage and tRNA genes of 18 unicellular organisms and quantification of Bacillus subtilis tRNAs: gene expression level and species-specific diversity of codon usage based on multivariate analysis

    Gene 238:143–155.

    https://doi.org/10.1016/s0378-1119(99)00225-5

    • PubMed
    • Google Scholar
38. 1. Kurland CG
    2. Ehrenberg M

    (1987) Growth-optimizing accuracy of gene expression

    Annual Review of Biophysics and Biophysical Chemistry 16:291–317.

    https://doi.org/10.1146/annurev.bb.16.060187.001451

    • PubMed
    • Google Scholar
39. Book

    1. Kurland CG
    2. Hughes D

    (1996)

    Limitations of Translational Accuracy Escherichia coli and Salmonella: Cellular and Molecular Biology

    American Society for Microbiology Press.

    • Google Scholar
40. 1. Li J
    2. Mau RL
    3. Dijkstra P
    4. Koch BJ
    5. Schwartz E
    6. Liu XJA
    7. Morrissey EM
    8. Blazewicz SJ
    9. Pett-Ridge J
    10. Stone BW
    11. Hayer M
    12. Hungate BA

    (2019) Predictive genomic traits for bacterial growth in culture versus actual growth in soil

    The ISME Journal 13:2162–2172.

    https://doi.org/10.1038/s41396-019-0422-z

    • PubMed
    • Google Scholar
41. 1. Love MI
    2. Huber W
    3. Anders S

    (2014) Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2

    Genome Biology 15:550.

    https://doi.org/10.1186/s13059-014-0550-8

    • PubMed
    • Google Scholar
42. 1. Mahajan S
    2. Agashe D

    (2018) Translational selection for speed is not sufficient to explain variation in bacterial codon usage bias

    Genome Biology and Evolution 10:562–576.

    https://doi.org/10.1093/gbe/evy018

    • PubMed
    • Google Scholar
43. 1. Manickam N
    2. Joshi K
    3. Bhatt MJ
    4. Farabaugh PJ

    (2016) Effects of tRNA modification on translational accuracy depend on intrinsic codon-anticodon strength

    Nucleic Acids Research 44:1871–1881.

    https://doi.org/10.1093/nar/gkv1506

    • PubMed
    • Google Scholar
44. 1. Marck C
    2. Grosjean H

    (2002) TRNomics: analysis of trna genes from 50 genomes of Eukarya, Archaea, and Bacteria reveals anticodon-sparing strategies and domain-specific features

    RNA 8:1189–1232.

    https://doi.org/10.1017/s1355838202022021

    • PubMed
    • Google Scholar
45. Book

    1. Miller JH

    (1972)

    Experiments in Molecular Genetics

    Cold Spring Harbor Laboratory, Cold Spring Harbor.

    • Google Scholar
46. 1. Nomura M
    2. Yates JL
    3. Dean D
    4. Post LE

    (1980) Feedback regulation of ribosomal protein gene expression in Escherichia coli: structural homology of ribosomal RNA and ribosomal protein mRNA

    PNAS 77:7084–7088.

    https://doi.org/10.1073/pnas.77.12.7084

    • Google Scholar
47. 1. Percudani R
    2. Pavesi A
    3. Ottonello S

    (1997) Transfer RNA gene redundancy and translational selection in Saccharomyces cerevisiae

    Journal of Molecular Biology 268:322–330.

    https://doi.org/10.1006/jmbi.1997.0942

    • PubMed
    • Google Scholar
48. 1. Quan S
    2. Skovgaard O
    3. McLaughlin RE
    4. Buurman ET
    5. Squires CL

    (2015) Markerless Escherichia coli rrn deletion strains for genetic determination of ribosomal binding sites

    G3: Genes, Genomes, Genetics 5:2555–2557.

    https://doi.org/10.1534/g3.115.022301

    • PubMed
    • Google Scholar
49. Software

    1. R Development Core Team

    (2021) R: A language and environment for statistical computing

    R Foundation for Statistical Computing, Vienna, Austria.

    http://www.r-project.org
50. 1. Rocha EPC

    (2004) Codon usage bias from tRNA’s point of view: redundancy, specialization, and efficient decoding for translation optimization

    Genome Research 14:2279–2286.

    https://doi.org/10.1101/gr.2896904

    • PubMed
    • Google Scholar
51. 1. Roller BRK
    2. Stoddard SF
    3. Schmidt TM

    (2016) Exploiting rRNA operon copy number to investigate bacterial reproductive strategies

    Nature Microbiology 1:16160.

    https://doi.org/10.1038/nmicrobiol.2016.160

    • PubMed
    • Google Scholar
52. 1. Saks ME
    2. Sampson JR
    3. Abelson J

    (1998) Evolution of a transfer RNA gene through a point mutation in the anticodon

    Science 279:1665–1670.

    https://doi.org/10.1126/science.279.5357.1665

    • PubMed
    • Google Scholar
53. 1. Samhita L
    2. Nanjundiah V
    3. Varshney U

    (2014) How many initiator tRNA genes does Escherichia coli need?

    Journal of Bacteriology 196:2607–2615.

    https://doi.org/10.1128/JB.01620-14

    • PubMed
    • Google Scholar
54. 1. Samhita L
    2. Raval PK
    3. Agashe D

    (2020) Global mistranslation increases cell survival under stress in Escherichia coli

    PLOS Genet 16:e1008654.

    https://doi.org/10.1371/journal.pgen.1008654

    • Google Scholar
55. 1. Shepherd J
    2. Ibba M

    (2015) Bacterial transfer RNAs

    FEMS Microbiology Reviews 39:280–300.

    https://doi.org/10.1093/femsre/fuv004

    • PubMed
    • Google Scholar
56. 1. Shigematsu M
    2. Honda S
    3. Loher P
    4. Telonis AG
    5. Rigoutsos I
    6. Kirino Y

    (2017) YAMAT-seq: an efficient method for high-throughput sequencing of mature transfer RNAs

    Nucleic Acids Research 45:e70.

    https://doi.org/10.1093/nar/gkx005

    • PubMed
    • Google Scholar
57. 1. Sørensen MA
    2. Fehler AO
    3. Lo Svenningsen S

    (2018) Transfer RNA instability as a stress response in Escherichia coli: rapid dynamics of the tRNA pool as a function of demand

    RNA Biology 15:586–593.

    https://doi.org/10.1080/15476286.2017.1391440

    • PubMed
    • Google Scholar
58. 1. Stevenson BS
    2. Schmidt TM

    (2004) Life history implications of rRNA gene copy number in Escherichia coli

    Applied and Environmental Microbiology 70:6670–6677.

    https://doi.org/10.1128/AEM.70.11.6670-6677.2004

    • PubMed
    • Google Scholar
59. 1. Thomason LC
    2. Costantino N
    3. Court DL

    (2007) E. coli genome manipulation by P1 transduction

    Current Protocols in Molecular Biology 1:mb0117s79.

    https://doi.org/10.1002/0471142727.mb0117s79

    • PubMed
    • Google Scholar
60. 1. Vieira-Silva S
    2. Rocha EPC

    (2010) The systemic imprint of growth and its uses in ecological (meta) genomics

    PLOS Genetics 6:e1000808.

    https://doi.org/10.1371/journal.pgen.1000808

    • PubMed
    • Google Scholar
61. 1. Wald N
    2. Margalit H

    (2014) Auxiliary tRNAs: large-scale analysis of tRNA genes reveals patterns of tRNA repertoire dynamics

    Nucleic Acids Research 42:6552–6566.

    https://doi.org/10.1093/nar/gku245

    • PubMed
    • Google Scholar
62. 1. Weissman JL
    2. Hou S
    3. Fuhrman JA

    (2021) Estimating maximal microbial growth rates from cultures, metagenomes, and single cells via codon usage patterns

    PNAS 118:e2016810118.

    https://doi.org/10.1073/pnas.2016810118

    • PubMed
    • Google Scholar
63. 1. Wilusz JE

    (2015) Controlling translation via modulation of tRNA levels

    Wiley Interdisciplinary Reviews. RNA 6:453–470.

    https://doi.org/10.1002/wrna.1287

    • PubMed
    • Google Scholar
64. 1. Wu L
    2. Yang Y
    3. Chen S
    4. Jason Shi Z
    5. Zhao M
    6. Zhu Z
    7. Yang S
    8. Qu Y
    9. Ma Q
    10. He Z
    11. Zhou J
    12. He Q

    (2017) Microbial functional trait of rRNA operon copy numbers increases with organic levels in anaerobic digesters

    The ISME Journal 11:2874–2878.

    https://doi.org/10.1038/ismej.2017.135

    • PubMed
    • Google Scholar
65. 1. Zuurmond AM
    2. Rundlöf AK
    3. Kraal B

    (1999) Either of the chromosomal tuf genes of E. coli K-12 can be deleted without loss of cell viability

    Molecular & General Genetics 260:603–607.

    https://doi.org/10.1007/s004380050934

    • PubMed
    • Google Scholar

Author details

1. Parth K Raval

   National Centre for Biological Sciences (NCBS-TIFR), Bengaluru, India

   Contribution

   Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing – review and editing

   For correspondence

   parth_science@yahoo.com

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0001-6151-437X

2. Wing Yui Ngan

   Max Plank Institute for Evolutionary Biology, Plön, Germany

   Contribution

   Data curation, Formal analysis, Investigation, Visualization, Writing – review and editing

   Competing interests

   No competing interests declared

3. Jenna Gallie

   Max Plank Institute for Evolutionary Biology, Plön, Germany

   Contribution

   Resources, Data curation, Formal analysis, Supervision, Funding acquisition, Visualization, Methodology, Writing – review and editing

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0003-2918-0925

4. Deepa Agashe

   National Centre for Biological Sciences (NCBS-TIFR), Bengaluru, India

   Contribution

   Conceptualization, Resources, Supervision, Funding acquisition, Visualization, Writing - original draft, Project administration, Writing – review and editing

   For correspondence

   dagashe@ncbs.res.in

   Competing interests

   No competing interests declared

   "This ORCID iD identifies the author of this article:" 0000-0002-0374-8159

• 2,521

  views

• 284

  downloads

• 6

  citations

Views, downloads and citations are aggregated across all versions of this paper published by eLife.