Error-prone initiation factor 2 mutations reduce the fitness cost of antibiotic resistance

Mutations in the fmt gene (encoding formyl methionine transferase) that eliminate formylation of initiator tRNA (Met-tRNAi) confer resistance to the novel antibiotic class of peptide deformylase inhibitors (PDFIs) while concomitantly reducing bacterial fitness. Here we show in Salmonella typhimurium that novel mutations in initiation factor 2 (IF2) located outside the initiator tRNA binding domain can partly restore fitness of fmt mutants without loss of antibiotic resistance. Analysis of initiation of protein synthesis in vitro showed that with non-formylated Met-tRNAi IF2 mutants initiated much faster than wild-type IF2, whereas with formylated fMet-tRNAi the initiation rates were similar. Moreover, the increase in initiation rates with Met-tRNAi conferred by IF2 mutations in vitro correlated well with the increase in growth rate conferred by the same mutations in vivo, suggesting that the mutations in IF2 compensate formylation deficiency by increasing the rate of in vivo initiation with Met-tRNAi. IF2 mutants had also a high propensity for erroneous initiation with elongator tRNAs in vitro, which could account for their reduced fitness in vivo in a formylation-proficient strain. More generally, our results suggest that bacterial protein synthesis is mRNA-limited and that compensatory mutations in IF2 could increase the persistence of PDFI-resistant bacteria in clinical settings.

PIC binds (f)Met-tRNA with rate constant k 1 ; the tRNA containing 30S PIC binds then to the 50S subunit with the rate constant k 2 . (B) The same as in (A) but 30S PICs can be either in active 30S A PIC or inactive 30S I PIC conformation and the transition between them occurs with rate constants k 3 and q 3 . (C) The same as in (A) but inactive 30S I PICs do not transform into the active complexes but, instead, can bind 50S subunits slowly with second order rate constant k 3 . Approximate positions of initiation factors IF1, IF2, IF3 as well as the site of SD:anti-SD interactions (SD) on the 30S subunit are shown. Figure S5. Dependence of the total rate of protein synthesis on mRNA concentration in the cell. (A) Total rate of protein synthesis V p (in mM/s) was calculated for different mRNA concentrations according to relations (D12) and (D14) using the following values of parameters: [30S 0 ]=40 µM;, τ clea ,=1.2 s; n c =1200 nts;, v e =20 aa/s (60 nts/s) ; k a =1 µM -1 s -1 ; τ 70S =0.25 s for wild type IF2 and fMet-tRNA i and τ 70S =2.8 s for wild type IF2 and nonformylated Met-tRNA i . (B) Ratio of generation times for bacterial strains having formylation deficient and formylation proficient genetic background harboring wild type IF2 calculated as the ratio of the corresponding rates of protein synthesis in (A). (C) The same as in (A) but k a =10 µM -1 s -1 was used in V p calculations, instead of k a =1 µM -1 s -1 . (D) The same as in (B) but the rates of protein synthesis in (C) were used to calculate the ratio of generation times. Regions where the rate of total protein synthesis in the cell is mRNAlimited according relation (D17) are indicated for the wild type and fmt-mutant strains in the figure.

A. Identification of unknown compensatory mutations using the mini-Tn10 transposon insertion technique.
Transposon pools with random insertions of the mini transposon Tn10Δ16Δ17 [mini-Tn10, (Tet R )] were prepared in two growth compensated mutant strains (DA8740 and DA8799) as described previously (Altman et al., 1996). Each pool consisted of at least 10,000 independent transposon hops. P22 phage was grown on each pool and the phage lysate was used to infect the slow growing parental strain (DA8326 and DA8340, respectively).
Tetracycline resistant colonies were selected on LA plates supplemented with 30mg/L tetracycline and the transductants were screened for acquisition of the growth compensated phenotype by visual examination of colony size. Linkage between the growth compensated phenotype and the tetracycline resistant marker was established by back crossing to the slow-growing mutant. An isogenic pair of tetracycline resistant, slow-growing and tetracyclin resistant, fast-growing strains was saved and frozen at -80°C.
Arbitrary primed PCRs directed outwards from the mini-Tn10 transposon was performed to identify the insertion points of the transposons. The PCR was performed in two steps. First, a PCR reaction was set up with one specific primer for the transposon and a mix of arbitrary primers. The arbitrary primers used consisted of a defined part (20 bp) and a variable part (15 bp) where bases are inserted randomly. Using the PCR products from the first reaction as template, a second nested PCR was performed. The product from the second reaction PCR reaction was then used as a template for sequencing. B. Construction of pBAD30::mut_infB HIS plasmids for complementation studies and IF2 over-expression.
The pBAD30::mut_infB HIS plasmids were constructed as follows: The mutant infB coding region was amplified using PCR with the primer c-infB-forSD-Sac [atatgagctcaaggagatatacatatgcaccaccaccaccaccacacagatgtaaccctaaaagcgc] that contained a SacI restriction site, an optimal Shine-Dalgarno sequence, and a His-tag for purification purposes and the c-infB-rev-xbaI [gcgctctagattaagcgatggtacgttggatct] primer that contained a XbaI restriction site. The resulting PCR product was digested with SacI and XbaI and ligated into the plasmid pBAD30 pre-digested with the same enzymes (Fermentas) Five µl of the ligation mixture was then transformed by heat shock into chemically competent E.
coli following the manufacturer's protocol (NEB 5-alpha competent E. coli). Cells were plated on LA supplemented with 100 mg/L ampicillin and incubated overnight at 37°C.
Colonies were picked and re-streaked the following day. Plasmid (pBAD30::mut_infB HIS ) was prepared from one selected clone of each mutant IF2 (E.Z.N.A.® Plasmid Miniprep Kit, Omega Biotech). The plasmid was then transformed into S. typhimurium LT2 using electroporation. An overnight culture was diluted 1/100, cells were grown to OD 600 ≈0.5 and expression was induced using 0.2% L-arabinose. (Sigma-Aldrich). After 1.5 hours of induction, cells were pelleted and frozen at -80°C.
C. Kinetics of 70S complex formation in experiments in which tRNA and 50S subunits are added to 30S pre-initiation complexes lacking tRNA.

The kinetic model. The intensity of scattered light after rapid mixing of 30S and 50S
subunits supplemented with different factors in a stopped flow instrument reflects the concentration of formed 70S initiation complexes (Antoun et al., 2004). Light scattering data were first fitted to the kinetic scheme in Fig. S4A. Here, the association rate constant k 1 describes tRNA binding to the 30S pre-initiation complex (30S PIC) containing mRNA and all three initiation factors, while the compounded rate constant k 2 describes the subsequent docking of the 50S subunit to the complete tRNA-containing 30S PIC (Antoun et al., 2006). This model did not account for the second, slow phase of 70S formation with small amplitude. We therefore extended the kinetic model in Fig S4A to include two different states, i. e one active and one inactive, of the 30S PIC (Milon et al., 2008). In this model ( Fig. S4B) a small fraction of inactive 30S I PICs must become active 30S A PICs with the rate constant k 3 before they bind tRNA and 50S subunits. It is also assumed that before addition of 50S subunits and tRNA the equilibrium between active and inactive conformations of the 30S PIC has already established. The equilibrium fraction of active 30S A PICs determines the amplitude of the major fast phase in light scattering experiments.
This fraction is in turn determined by the rate constants k 3 and q 3 , so that the model has four parameters k 1 , k 2 , k 3 and q 3 to be fitted.

Effective rate of 70S initiation complex formation.
In the limit of high (f)Met-tRNA i concentration the product k 1 [(f)Met-tRNA i ] in the scheme S4B becomes very large in comparison with k 2 [50S]. If, in addition, the fraction of inactive 30S PICs is small then a much simpler kinetic scheme emerges: The time course of such a mono-phasic formation of 70S ribosomes is given by the formula (Antoun et al., 2006): Here, the apparent rate constant Q depends on the compounded second order rate constant k 2 and the difference between the initial concentrations of the 50S ([50S 0 ]) and 30S ([30S 0 ]) subunits: When the initial concentrations of 30S and 50S subunits are very close the relation (C1) simplifies to: Here, k 2,eff is the product of the compounded rate constant k 2 and the initial concentration of 50S subunits. The time, t 0.5 , when the concentration of 70S complexes reaches 50% of its maximal value is given by: Numerical calculations show that for the scheme S4B, t 0.5 can be approximated by: provided that the fraction of inactive 30S I PICs is low and the concentration of (f)Met-tRNA i is sufficiently high to ensure that ] > k 2,eff . This approximation justifies the use of the effective rate k I , defined as the inverse of t 0.5 , in the main text.
Fitting experimental data to kinetic models. The extended kinetic scheme in Fig. S4B was employed to describe the kinetics of 70S formation when Met-tRNA i was added together with 50S subunits to the 30S PICs ( Fig. 5 and Fig. S2).  Rate constants presented in the table are defined below. Table 1S shows that the fraction, f30S A , of active 30S A PICs for the best fit of curves in Fig.   5 and Fig. S2 generally exceeds 80%. Comparison with Table 2 in the main text shows that k 2,eff corresponds very well with k max values in Table 2 obtained from Linewaver-Burk plots in Fig. 5F. The values of k max /K M in Table 2 were, however, smaller than the association rate constant k 1 for Met-tRNA i binding especially in the case of the A1 IF2 mutant. At the same time, in general k 1 and k 2,eff = k 2 [50S 0 ] correlated well with the corresponding k max /K M and k max values for the different IF2 mutants in Table 2.
We have also tried to fit experimental data to a kinetic model in which inactive 30S I PICs do not transform into active 30S PICs but, instead, bind 50S subunits directly with the second order rate constant k 3 . The fraction, f30S A , of active 30S PICs in the initial population of 30S PICs was also a fitting parameter of the model along with the three rate constants k 1 , k 2 and k 3 . Fitting the 70S formation kinetics to this model (Fig. S4C) gave very similar values for k 1 and k 2 rate constants and a similar initial fraction of active 30S PICs as for the model in Fig. S4B (not shown) indicating that the values of k 1 and k 2 were robust to model choice.

D. Dependence of total protein synthesis in the cell on initiation time and mRNA concentration
The loading rate, V i , of ribosomes onto mRNA according to the scheme is given by Here, [30S] is the concentration of fee 30S subunits and k a is the rate constant for 30S subunit association to the ribosomal binding site (RBS) on mRNA. We define regeneration time, τ reg , of the RBS of mRNA as:  (Bremer et al., 2003;Dennis et al., 2009;Ehrenberg et al., 2009): The maximal initiation rate is given by The K M value is given by The total rate, V p , of protein synthesis per cell volume, proportional to the growth rate of the cell population (Ehrenberg and Kurland, 1984), is given by: Here, [mRNA 0 ] is the total mRNA concentration in the cell, N r is the average number of translating ribosomes per mRNA and v e is the average rate of protein elongation per ribosome. N r can be obtained from the average time, τ transl , to translate all codons in the ORF and the initiation time τ init (or the initiation rate ) through : The total average number of ribosome, N tot , per mRNA (the translating ones plus those in the initiating state C1 in the scheme D1) is given by The maximal total average number, of ribosome per mRNA is obtained when the initiation rate V i is maximal. Using relation (D5) one obtains: The average protein translation time is given by The total rate of protein synthesis per cell volume can now be explicitly written as The concentration of translating ribosomes, To determine the concentration of the free 30S subunit, we recall that the free 30S concentration plus the concentration of 30S subunit bound to mRNA either as subunit or in 70S ribosomes equals the total concentration of the 30S subunits, [30S 0 ] in the cell: Taking into account relations (D9) for N tot and (D2) for V i one gets:  Fig. S5 for realistic values for these parameters as discussed in the main text (see the legend of Figure S5 for more details). [ ] >> 1, relation (D12) shows that the total rate of protein synthesis per cell volume is given by This defines the condition of mRNA limitation, where V p is proportional [mRNA 0 ] and inversely proportional to τ reg . In addition, in this limit relation (D14) is approximated by: The ratio [ ] >> 1 for mRNA-limited protein synthesis requires that Our model calculations ( Figure S5) show that expression (D17) approximates the region for mRNA-limited protein synthesis even for k a =1 µM -1 s -1 , a value considerably below experimental estimates of the rate constant for association of 30S subunits to the RBS of mRNAs (Studer and Joseph, 2006).

Ribosome-limited protein synthesis.
When the mRNA concentration increases above: Substituting (D20) into (D12) gives the total rate of protein of protein synthesis as: This is the condition of ribosome limitation, where protein synthesis does not depend on the mRNA concentration, but only on the total ribosome concentration. This approximation holds well for realistic parameter values already when the total mRNA concentration is just above the boundary set by inequality (D17) for mRNA-limited protein synthesis (Fig. S5). Furthermore, the transition between mRNA-limited and ribosome-limited protein synthesis becomes sharper with increasing k a values, also illustrated in Fig. S5. Figure S5 also shows that the maximal difference in generation times (maximal ratio of generation times) between the wild type and fmt-strains is observed for mRNA concentrations where protein synthesis is mRNA-limited in both wild type and mutant strain. From (D15) follows that in this case the ratio between generation times is given by: The ratio between generation times decreases gradually as the total mRNA concentration increases above the boundary for the mRNA-limited protein synthesis in the wild type strain and is minimized when protein synthesis is ribosome-limited for both wild type and fmt-strain. From (D21) follows that the ratio between generation times under ribosome limitation is given by: € n c /v e + τ 70S ( fmt − mutant) n c /v e + τ 70S (wild − type) = 20s + 2.8s 20s + 0.25s ≈ 1.13 (D23)