We herein revisit the contribution of translation to global mechanisms allowing bacteria to modulate abundance of single proteins with respect to the growth rate. Current theories suggest that translation efficiency (i.e. number of proteins produced per mRNA) remains invariant with increasing growth rate (Klumpp et al., Cell, 2009), which is inconsistent with the scanty correlation between mRNAs and cognate proteins abundances at the genome-scale level (Buescher et al., Science, 2012; Vogel and Marcotte, Nat Rev Genet., 2012). We tackled this apparent paradox using a systems biology approach.

We developed a knowledge-based, nonlinear mathematical model of translation. The in-depth analysis of the model led us to reassess experimentally, using high-throughput and genome-wide technologies, each measurable RNA entity at different growth rates. In contrast to the current knowledge, the total mRNA abundance was not constant but linearly increased with respect to the growth rate. A model-driven integration of genome-wide and molecular experimental datasets demonstrated that the drop in abundance of a constitutively expressed protein with increasing growth rate is not only due to the dilution but also to an unexpected up to 4-fold decrease of translation efficiency. Our model revealed that this drop relies on a drastic decrease in free (untranslating) ribosomes, a non-measurable entity. Using a set of 18 Bacillus subtilis strains combining 9 synthetic translation initiation regions (TIRs) and 2 constitutive promoters, we show that TIRs together with free ribosome abundance strongly contribute to a nonlinear modulation of single proteins as a function of the growth rate. The nonlinearity accounted for the loss of correlation between mRNAs and cognate proteins abundances. Altogether, our results evidenced a unique, hard-coded and global growth-rate-dependent regulation of single bacterial proteins without dedicated regulators.