A Comparison of the Costs and Benefits of Bacterial Gene Expression

Abstract

To study how a bacterium allocates its resources, we compared the costs and benefits of most (86%) of the proteins in Escherichia coli K-12 during growth in minimal glucose medium. The cost or investment in each protein was estimated from ribosomal profiling data, and the benefit of each protein was measured by assaying a library of transposon mutants. We found that proteins that are important for fitness are usually highly expressed, and 95% of these proteins are expressed at above 13 parts per million (ppm). Conversely, proteins that do not measurably benefit the host (with a benefit of less than 5% per generation) tend to be weakly expressed, with a median expression of 13 ppm. In aggregate, genes with no detectable benefit account for 31% of protein production, or about 22% if we correct for genetic redundancy. Although some of the apparently unnecessary expression could have subtle benefits in minimal glucose medium, the majority of the burden is due to genes that are important in other conditions. We propose that at least 13% of the cell's protein is "on standby" in case conditions change.

Fig 1. The production of proteins versus their importance for growth.

(A) For each class of gene, we show the distribution of protein expression, in parts per million of amino acids (x axis, log scale). Proteins with little or no expression are shown at 0.1 ppm. (B) The aggregate expression of each class of gene.

Fig 2. The production of “unnecessary” proteins versus their importance in other conditions.

Only genes that were not important for fitness in minimal glucose media are included. The x axis is as in Fig 1A.

Fig 3. Consistency of gene fitness values in minimal glucose medium.

(A) Consistency between replicates at 12 generations. Fitness values less than −3 are shown at −3 so as to focus on the more subtle fitness defects that are more susceptible to noise. 123 genes have fitness under −3 in both replicates. (B) Consistency across time. Genes with significant phenotypes (of either sign) are subdivided into those with weak expression (under 2 ppm of monomers) or above. Fitness values less than −3 are shown at −3. (C) Consistency within each gene. Fitness values at 12 generations were computed separately for the first and second half of each gene that had sufficient coverage. (D) shows the same data as (C), but only for fitness values above −3. In all panels, lines show x = 0, y = 0, and x = y.