Where cold sensitive growth has previously been identified as a dominant phenotype for defects in ribosome biogenesis, we set out to first validate such an enrichment strategy with a screen of the E. coli Keio collection (Baba et al., 2006), a comprehensive set of non-essential gene deletion strains (Figure 1—source data 1). We looked for strains that were sensitized to growth at 15°C compared to 37°C (Figure 1—figure supplement 1A,B). A cold sensitivity factor was subsequently generated for each clone, defined as the ratio of growth at 37°C to growth at 15°C, normalized to the mean growth ratio measured for the entire collection (Figure 1A). Strains that displayed a cold sensitivity factor in the top 3.5% (155 clones) were analyzed using clusters of orthologous groups (Tatusov et al., 1997, 2003) to categorize the cellular function of each deleted gene (Figure 1—figure supplement 1C, Supplementary file 1A). To highlight the relative proportion of genes in each functional class, the number of cold sensitive genes in each was divided by the total number of non-essential genes in that same category (Figure 1B). This normalization procedure highlighted ribosome-related genes as exceptionally sensitive to low temperatures, as >20% of genes in this functional class were found to be cold sensitive. Importantly, this screen was also successful in identifying the vast majority of previously reported cold sensitive ribosome biogenesis genes (Supplementary file 1B), providing support that screening compounds for cold sensitivity would enrich for those related to ribosome function and biogenesis.

Figure 1 with 1 supplement see all

Download asset Open asset

Figure 1—source data 1

Screen of the E. coli Keio collection.

Cold sensitivity factors for each strain were normalized to 1, based on the mean of all cold sensitivity factors calculated.

https://doi.org/10.7554/eLife.03574.004

Download elife-03574-fig1-data1-v2.xlsx

Having validated our cold sensitivity enrichment strategy, we proceeded to screen a diverse chemical collection to identify molecules that exhibited a cold sensitive growth inhibition phenotype. This collection, assembled from a variety of vendors, included some 30,000 compounds. These were largely diverse synthetic molecules with a subset of 3500 previously approved drugs and chemicals with known biological activity (Figure 2—source data 1). E. coli was grown in LB media at 15°C and 37°C in the presence of 10 μM of each compound (Figure 2—figure supplement 1). To select compounds for follow-up, we identified those that strongly inhibited growth at 15°C (>3σ below the mean OD₆₀₀ at 15°C), yet displayed little growth inhibition at 37°C (<2σ below the mean OD₆₀₀ at 37°C). These criteria resulted in 49 active compounds (Figure 2A). We removed all antibiotics with known mechanisms of action and filtered the active molecules for diversity in chemical structure. This led to a short-list of 38 active molecules, which were analyzed in dose at 37°C and 15°C. The anticonvulsant drug lamotrigine displayed the largest change in minimum inhibitory concentration (MIC) upon temperature downshift, increasing in potency more than 50-fold from >512 μM at 37°C to 7.8 μM at 15°C (Figure 2B, Figure 2—figure supplement 2).

Figure 2 with 2 supplements see all

Download asset Open asset

Figure 2—source data 1

Small molecule screen for cold senstivity.

OD values were normalized across screening plates to account for plate-to-plate variations.

https://doi.org/10.7554/eLife.03574.007

Download elife-03574-fig2-data1-v2.xlsx

To determine whether lamotrigine resulted in cold sensitivity through perturbation of the ribosome, we harvested ribosomal particles from early-log cultures of E. coli treated with 2× MIC of lamotrigine for 1 hr and 6 hr at 15°C in LB media and resolved them using sucrose density centrifugation. We note that the doubling time of wild-type E. coli at 15°C in LB media was 6 hr. Cultures were pulse labeled with [¹⁴C]-uridine immediately upon drug treatment to visualize the accumulation of newly synthesized particles. Since previous reports have shown that inhibitors of protein translation can cause accumulation of immature ribosomal particles (Siibak et al., 2009, 2011; Sykes et al., 2010), we also tested a panel of antibiotics (Figure 3—figure supplement 1A–D) with known mechanism of action for comparison.

Mock treatment of cells with DMSO and simultaneous pulsing with [¹⁴C]-uridine allowed for the visualization of 30S, 50S, and 70S particle accumulation after 1 hr and 6 hr of growth post-treatment (Figure 3A). After 1 hr of treatment, small quantities of newly synthesized particles were present, and after 6 hr, cells had accumulated labeled ribosomal particles to near steady-state levels. Cultures treated with chloramphenicol (Figure 3B), erythromycin (Figure 3C), and tetracycline (Figure 3D) displayed a substantial accumulation of non-native ribosomal particles after just 1 hr of treatment, illustrating that inhibition of translation can indirectly inhibit ribosomal subunit assembly by limiting the availability of r-proteins. Interestingly, we found that the addition of 2× MIC of vancomycin to E. coli resulted in a detectable perturbation of the ribosome profile (Figure 3E). However, the presence of a ∼40S particle after 6 hr of treatment is likely the result of cell lysis induced by the inhibition of peptidoglycan synthesis (Stokes and Brown, unpublished data). Treatment with lamotrigine resulted in the accumulation of non-native ribosomal particles after 1 hr of incubation and did so in a time-dependent manner (Figure 3F). Further investigations revealed that treatment of E. coli with 2× MIC of lamotrigine for only 5 min (∼1% of the doubling time) caused a significant accumulation of these non-native particles (Figure 3G,H). Consistent with the cold sensitive phenotype induced by lamotrigine, these pre-30S and pre-50S particles that accumulated at 15°C were not present after treatment at 37°C (Figure 3—figure supplement 1E,F).

Figure 3 with 1 supplement see all

Download asset Open asset

To determine if lamotrigine directly associated with ribosomal particles, early-log cultures of E. coli were treated with [³H]-lamotrigine in the absence (Figure 3—figure supplement 1G) or presence (Figure 3—figure supplement 1H) of 1× MIC of unlabeled lamotrigine. Ribosomal particles were then separated on a sucrose gradient and individual fractions counted to localize [³H]-lamotrigine. Radiolabeled compound was found exclusively in the soluble fractions eluting early in the gradient, suggesting that lamotrigine does not interact directly with mature or non-native ribosomal particles.

We reasoned that ribosomal particles accumulating during treatment could be immature subunits or degradation products of weakly assembled ribosomes. Thus, we analyzed rRNA processing and r-protein content of all particles that accumulated upon lamotrigine treatment. Because previous investigations have shown that the cleavage of 5′ and 3′ termini of rRNA is among the final events in ribosomal subunit assembly (Lindahl, 1973; Mangiarotti et al., 1974; Srivastava and Schlessinger, 1988), we first performed 5′ primer extension reactions using rRNA purified from sucrose gradients of lamotrigine-treated cells. Early-log cultures of E. coli were grown in the presence of 2× MIC of lamotrigine at 15°C in LB media for 5 min, 1 hr, and 6 hr, at which time the ribosomal particles were resolved on sucrose gradients, and the rRNA corresponding to each discrete particle was purified and reverse transcribed using 5′ carboxyfluorescein-tagged primers. A 16S rRNA-specific primer was used to analyze the 30S subunit rRNA in pre-30S, 30S, and 70S fractions, whereas a 23S rRNA-specific primer was used to analyze the 50S subunit rRNA in pre-30S, pre-50S, 50S, and 70S fractions. Figure 4A displays the 5′ cleavage events during the processing of 16S and 23S rRNAs (Shajani et al., 2011). We note here that our experiments were unable to detect the first 16S cleavage event of 49 nucleotides by Rnase E, and that all immature 16S rRNA species described contain a full-length 5′ terminus of 115 nucleotides.

Figure 4 with 2 supplements see all

Download asset Open asset

Figure 4—source data 1

R-protein occupancy across sucrose gradients, normalized to the maximum value observed.

https://doi.org/10.7554/eLife.03574.013

Download elife-03574-fig4-data1-v2.xlsx

Primer extension analysis of cells treated with DMSO for 6 hr is depicted in Figure 4B (top panel). Analysis of the pre-30S, 30S, and 70S regions of the gradient using the 16S rRNA-specific primer revealed that increasing sedimentation rate correlated with a decreased proportion of immature 16S rRNA relative to total 16S rRNA. The presence of immature 16S rRNA sedimenting in the pre-30S region suggested a heterogeneous composition of 30S particles at various stages of maturation. At the maximum of the 30S peak approximately 40% of 16S rRNA was unprocessed. The 16S rRNA found in fractions corresponding to the 70S subunit was >95% processed, as expected. Overall, a similar trend was seen when analyzing the processing of 23S rRNA in the pre-30S, pre-50S, 50S, and 70S regions of the gradient; increasing sedimentation rate correlated with a decreased proportion of immature 23S rRNA relative to total 23S rRNA. The pre-30S and pre-50S regions of the gradient were devoid of quantifiable 23S rRNA, suggesting very little 50S precursor accumulation in unperturbed cells.

Compared to cells treated with DMSO, those treated with 2× MIC of lamotrigine for 5 min (Figure 4B, bottom panel), 1 hr (Figure 4—figure supplement 1A), and 6 hr (Figure 4—figure supplement 1B) contained similar proportions of immature to total 16S and 23S rRNA in the 30S, 50S, and 70S regions of the gradient. Furthermore, lamotrigine-treated cells displayed almost identical proportions of immature to total 16S rRNA in the pre-30S region, relative to DMSO-treated cells. Unlike DMSO-treatment, however, lamotrigine treatment resulted in the accumulation of immature 23S rRNA in the pre-30S and pre-50S regions. While the presence of unprocessed 23S rRNA in the pre-50S region strongly suggested an immature 50S subunit sedimenting at ∼40S, its presence in the pre-30S region raised questions of whether the dominant species in the pre-30S peak (Figure 3F) was derived from 16S or 23S rRNA.

While calculating proportions of immature to total rRNA of the same species ([immature 16S/total 16S] and [immature 23S/total 23S]) provides detail of rRNA processing efficiency in each region of the gradient, it does not inform on the absolute quantity of one species (16S rRNA) relative to the other (23S rRNA). To answer this question, we quantified absolute cDNA fluorescence from 5′ primer extension reactions (Figure 4C, Figure 4—figure supplement 1C). Samples of rRNA purified from single fractions of sucrose gradients were reverse transcribed in parallel reactions using either the 16S- or 23S-specific primer. In this study, cDNA production is proportional to the amount of rRNA transcript in the sample and therefore reflects absolute quantities of each rRNA species (16S and 23S) present. The quantity of immature 16S rRNA from DMSO-treated cells was minor relative to cells treated with lamotrigine. This is consistent with previous reports, which have shown that immature ribosomal particles in unperturbed cells account for only a small proportion of total ribosomal material (Mulder et al., 2010; Chen et al., 2012). Furthermore, while immature 23S rRNA slowly accumulated in this region as a function of lamotrigine treatment length, immature 16S rRNA did so at a significantly greater rate. These results indicate that, by far, the major species of rRNA residing within the pre-30S region in lamotrigine-treated cells was unprocessed 16S rRNA. Thus, 5′ primer extension results strongly suggested that lamotrigine treatment results in the accumulation of an immature 30S subunit that sediments at ∼25S and an immature 50S subunit that sediments at ∼40S.

To further test this hypothesis, we used quantitative mass spectrometry to determine the relative stoichiometry of r-proteins across sucrose gradients of DMSO- and lamotrigine-treated cells. Early-log cultures of E. coli grown at 15°C in ¹⁴N-labeled LB media were treated with DMSO or 2× MIC of lamotrigine for 6 hr, lysed, and the ribosomal particles were separated through sucrose gradients. Fractions spanning the pre-30S to the 70S regions were spiked with a fixed concentration of 70S ribosomes purified from cells grown in ¹⁵N-labeled media. These spiked samples were then digested with trypsin and prepared for mass spectrometry. This approach resulted in multiple independent peptide measurements for each r-protein in every fraction (Figure 4D, Figure 4—figure supplement 2A,B). Protein occupancy was calculated as ¹⁴N/[¹⁴N + ¹⁵N]. Direct inspection of the protein occupancy profiles revealed distinct patterns for early- (e.g., S15, L24) and late- (e.g., S3, L28) binding proteins. In the DMSO samples, all r-proteins within a given subunit displayed highly correlated occupancy patterns with maximal occupancy corresponding to ‘peak’ fractions as determined by rRNA absorbance (Figure 4D). In contrast, treatment with lamotrigine resulted in significant occupancy of the early-binding proteins in pre-30S and pre-50S fractions, whereas the late-binding r-proteins exhibited relatively unperturbed profiles. Indeed, protein S15 is found at significantly greater occupancy in the pre-30S fractions upon lamotrigine treatment (dark blue). This effect on early binding proteins is particularly pronounced in the L24 profile with peak occupancy shifted six fractions earlier in the gradient (from orange to green). To facilitate further analysis, this large data set (∼20,000 measurements) was compressed to a 53-protein × 28-fraction heat map using the median protein occupancy value for each protein in each fraction (Figure 4E, Figure 4—source data 1). As expected, the 70S peak from both DMSO- and lamotrigine-treated cells exhibited stoichiometric occupancy of each r-protein.

In both DMSO- and lamotrigine-treated samples, we observed sub-stoichiometric occupancy of the late-binding r-proteins S21, S2, and to a lesser extent S3, within the 30S peak. Notably, this effect was enhanced in the lamotrigine-treated samples. The depletion of these proteins is consistent with prior in vivo analysis of small subunit biogenesis at 37°C, which found S2, S3, and S21 to be the latest-binding small subunit proteins (Figure 4—figure supplement 2C; Chen and Williamson, 2013). Consistent with sucrose gradient traces monitoring UV absorbance and [¹⁴C]-uridine incorporation, we observed a subtle broadening of the 30S protein occupancy peak upon treatment with lamotrigine. Analysis of the leading edge of this peak revealed an enrichment of relatively early-binding proteins S13, S7, S16, S8, S15, S6, and S18, consistent with the presence of 30S subunit assembly intermediates in lamotrigine-treated cells. Inspection of large subunit protein occupancy revealed drastic changes as a result of lamotrigine treatment, resulting in the accumulation of an immature particle depleted of the late-binding r-proteins L35, L36, L16, L30, and L28 as well as earlier-binding proteins, L34, L6, and L21 (Figure 4—figure supplement 2D). Particles with this heterogeneous protein composition could not be found in the DMSO-treated samples, indicating that if they do form they rapidly convert to mature particles that migrate later in the gradient.

Formally, these r-protein occupancy patterns could have resulted either from immature assembly intermediates or from the degradation of mature particles initiated by the removal of late-binding proteins. To distinguish between these possibilities, we used mass spectrometry to determine the occupancy pattern for the ribosome biogenesis factors RbfA and DeaD (Figure 4F). These proteins were completely absent from mature 70S particles, and thus we reasoned that their presence could be used as markers of immature particles. The 30S-specific maturation factor RbfA co-sedimented with the 30S particles in both DMSO- and lamotrigine-treated samples. Further, the 50S-biogenesis factor DeaD co-migrated with the pre-50S peak in the lamotrigine-treated samples and the 50S peak in the DMSO-treated samples. These data further suggest that the pre-30S and pre-50S particles are immature subunits and not the result of degradation of mature ribosomal particles.

To identify lamotrigine's target in vivo, we generated suppressor mutants and sequenced the resulting genomes to identify the mutation(s) that were responsible for resistance. Briefly, E. coli BW25113 was grown at 15°C in the presence of 5× MIC (39 μM) of lamotrigine in LB media to saturation. Putative suppressors were then serially passaged in the presence and absence of lamotrigine to purify and to ensure mutation stability. After 20 independent strains had been isolated, three were selected at random, sequenced using an Illumina MiSeq platform, and analyzed against the E. coli MG1655 genome using BreSeq. At this time, the chromosome of BW25113 had yet to be sequenced, thus this strain was sequenced in parallel to be used as a reference genome. The sequencing data revealed mutations solely in domain II of initiation factor IF2 (Figure 5A, Figure 5—figure supplement 1A). Subsequent Sanger sequencing of the infB genes from each of the remaining 17 suppressor strains revealed that all mutations mapped to domain II of IF2 and fell into only four categories. Three classes of mutant contained in-frame chromosomal deletions in this region and one mutant class contained a short duplication.

Figure 5 with 1 supplement see all

Download asset Open asset

To understand the phenotypic characteristics of these four unique lamotrigine suppressor strains, cells from each class were first analyzed for growth rate and resistance to lamotrigine. All displayed wild-type growth at 15°C in the absence of lamotrigine and little sensitivity to lamotrigine treatment up to 512 μM in LB media (Figure 5—figure supplement 1B,C). We subsequently analyzed these strains to determine the composition of ribosomal particles upon lamotrigine treatment. Suppressor strains were grown to early-log phase, treated with 2× MIC of lamotrigine, and grown for 6 hr at 15°C. Immediately after the addition of lamotrigine to the cultures, cells were pulse labeled with [¹⁴C]-uridine to monitor accumulation of non-native ribosomal particles. Treatment of suppressor strains with lamotrigine did not result in the accumulation of non-native ribosomal particles (Figure 5—figure supplement 1D).

To test the hypothesis that IF2 was the target of lamotrigine, we conducted in vitro binding studies using recombinant E. coli IF2 and [³H]-lamotrigine (Figure 5B). Wild type and mutant forms of E. coli IF2 were purified and incubated with [³H]-lamotrigine in the presence of GDP or GTP. After incubation for 3 hr at 15°C, the reaction mixtures were passed through a pre-cooled Sephadex G-25 column, and the flow-through was collected and scintillation counted to detect the presence of lamotrigine-IF2 complexes. Lamotrigine was found to associate with wild-type E. coli IF2 in a G-nucleotide-dependent manner, with lamotrigine-IF2 complex formation favored in the presence of GDP over GTP (Figure 5C). We note here that previous studies have not reported measurable GTP turnover by IF2 in the absence of ribosomal subunits (Severini et al., 1991). Analyses of lamotrigine binding with mutant IF2 in the presence of GTP and GDP failed to reveal association (Figure 5C). Consistent with these results, lamotrigine was found to have activity solely against members of the Enterobacteriaceae (Supplementary file 2), which is the only bacterial family that contains the domain II sequence outlined in Figure 5A. Interestingly, the potency of lamotrigine at 15°C against the Enterobacteriaceae was directly correlated with IF2-α sequence homology, further suggesting that the sequence of domain II defines essential structural features for lamotrigine association (Figure 5—figure supplement 1E,F).

Given the known role of IF2 in protein translation, we tested whether lamotrigine was indirectly perturbing ribosome biogenesis by inhibiting IF2-dependent translation. We first monitored [³⁵S]-methionine incorporation into bulk cellular protein. Early-log cultures of E. coli grown in M9 minimal media were treated with lamotrigine and a collection of known antibiotics for 2.6 hr at 15°C (doubling time = 16 hr). Immediately after the addition of drug, cells were pulsed with [³⁵S]-methionine to monitor the production of newly synthesized proteins. Cells were then pelleted, washed, lysed, and treated with TCA. The precipitated proteins were captured on glass filters and counted. These investigations revealed that lamotrigine had no impact on [³⁵S]-methionine incorporation, even when cells were treated with 8× MIC of lamotrigine (Figure 6A, C, black dots). Similarly, cells treated with these same concentrations of lamotrigine at 37°C for three doublings did not display any inhibition of translation (Figure 6—figure supplement 1A). We found that when cells were treated with 8× MIC of tetracycline, chloramphenicol, and erythromycin, there was a marked decrease in protein labeling. As expected given its known mechanism of action, cells treated with 8× MIC of vancomycin did not display inhibition of protein biosynthesis after 2.6 hr of treatment.

Figure 6 with 1 supplement see all

Download asset Open asset

Figure 6—source data 1

In vivo r-protein synthesis rates measured using qMS.

Synthesis rates are log2 transformed after normalization to the measured synthesis rate of E. coli in the presence of DMSO. Growth rates are reported as 1/doubling time (hr).

https://doi.org/10.7554/eLife.03574.019

Download elife-03574-fig6-data1-v2.xlsx

To determine if lamotrigine had a direct effect on protein biosynthesis in vitro, we employed a commercially available E. coli K-12 cell-free transcription/translation system producing luciferase. Reactions in the presence of 8× MIC of lamotrigine and the aforementioned antibiotics were incubated at 15°C for 4 hr (see Figure 6—figure supplement 1C for in vitro translation kinetics at 15°C), at which time luciferin was added to quantify the luciferase produced. As expected, all translation inhibitors blocked the production of luciferase while vancomycin did not (Figure 6B). Lamotrigine failed to block the production of luciferase at either 15°C or 37°C (Figure 6B, C, white dots, Figure 6—figure supplement 1B). To ensure that the in vitro translation assay required IF2 activity, we tested the effect of evernimicin, an oligosaccharide antibiotic known to inhibit IF2-dependent 70S initiation complex formation (McNicholas et al., 2000; Belova et al., 2001). Evernimicin prevented luciferase synthesis in vitro at both 15°C and 37°C (Figure 6D).

Having ruled out a direct effect on bulk protein biosynthesis, we wondered if lamotrigine might have a specific effect on r-protein synthesis that could lead to the accumulation of immature ribosome subunits. To test this, we measured the synthesis rate of each r-protein in vivo using a mass spectrometry-based pulse labeling technique. At 15°C, cells were grown in ¹⁴N-labeled M9 minimal media to mid-log phase at which point they were diluted twofold into ¹⁵N-labeled M9 media and concurrently treated with DMSO, lamotrigine, or chloramphenicol. Cells were harvested after 1 hr, 2.6 hr, 4 hr, 8 hr, and 16 hr and spiked with equal quantities of ¹⁵N-labeled 70S ribosomes as an internal reference standard. After cell lysis, these spiked samples were digested with trypsin for analysis by mass spectrometry. Using a Fourier transform deconvolution algorithm (Sperling et al., 2008; Chen et al., 2012), we independently quantified the r-proteins produced before the pulse (¹⁴N) and those synthesized post-pulse (50% ¹⁵N) from the cellular lysate (Figure 6—figure supplement 1D).

Inspection of ¹⁴N abundance as a function of time revealed that most ribosomal proteins were stable over this time course (Figure 6—figure supplement 1E,G,I), consistent with our prior work (Chen et al., 2012). We then carefully inspected the rate of 50% ¹⁵N incorporation into each ribosomal protein in each treatment condition. Using a linear approximation of the synthesis rate based on the 4-, 8-, and 16-hr time points, we found that each protein was synthesized at a similar rate in the DMSO- and lamotrigine-treated cells up to 6× MIC. However, we found significant inhibition of r-protein synthesis with increasing concentrations of chloramphenicol, our positive control compound (Figure 6E, Figure 6—figure supplement 1F,H,J, Figure 6—source data 1).

To establish if the pre-30S and pre-50S particles that accumulate upon lamotrigine treatment represented immature subunits on pathway to maturity, we endeavored to monitor the impact of relieving inhibition by lamotrigine. We hypothesized that cells relieved of lamotrigine stress would assemble immature 30S and 50S particles into mature 30S and 50S subunits. E. coli was grown to early-log phase in LB media at 15°C and treated with either 2× MIC of lamotrigine or DMSO as a mock treatment. After 5 min, [¹⁴C]-uridine was added and the cells were grown an additional 3 hr, at which point cells were pelleted, washed, and resuspended in fresh LB media supplemented with a 1000-fold excess of non-labeled uridine. Cells were harvested immediately preceding the chase, and after 30 min, 1 hr, 2 hr, and 3 hr of this chase period (Figure 7A).

Figure 7

Download asset Open asset

In each DMSO-treated sample (Figure 7B), we found significant quantities of [¹⁴C]-uridine-labeled 30S and 50S subunits. Because the quantity of labeled subunits did not change as a function of the length of the chase, these particles likely represent fully mature subunits that have simply dissociated. Interestingly, DMSO-treated cells harvested immediately after the 3-hr pulse and before the addition of the chase show a slight decrease in the levels of complete 70S ribosomes relative to any of the samples harvested post-chase (Figure 7B). This small but significant change likely results from the presence of an intracellular pool of [¹⁴C]-uridine, which is incorporated into 70S particles during the initial 30-min chase. This pool may consist of free nucleotides that are not washed away during the chase or, as described previously, may exist as transcribed rRNA that has not completed the assembly process (Chen et al., 2012; Chen and Williamson, 2013). Cells harvested at subsequent times during the chase period showed no change in the quantities of 30S, 50S, and 70S particles, indicating that all newly synthesized rRNA is incorporating exclusively non-labeled uridine. This result allowed us to analyze the maturation of the lamotrigine-induced pre-30S and pre-50S particles, confident that they were generated during the initial pulse and not synthesized de novo between 30 min and 3 hr post-pulse.

We next analyzed the ability of cells treated with lamotrigine to process pre-30S and pre-50S particles (Figure 7C). Cells harvested immediately after the 3-hr pulse period displayed a significant accumulation of pre-30S and pre-50S material. As shown earlier (Figure 3F), we also noticed a large decrease in the relative accumulation of 70S ribosomes during drug treatment. Some 30 min after removal of lamotrigine and non-labeled uridine chase, the relative proportions of ribosomal particles began to adjust. Specifically, the levels of pre-30S and pre-50S particles decreased with a corresponding increase in 70S ribosomes. This trend continued throughout the 3-hr chase period, after which there were no apparent differences between DMSO-treated and lamotrigine-treated cells. Interestingly, after 1 hr of non-labeled uridine chase, a cluster of three particles that sedimented at approximately 40S (discussed above as the pre-50S), 45S, and 50S appeared. With each successive time point, the levels of 50S increased at the expense of the other particles, suggesting our time course had captured cells actively assembling the pre-50S particles into 50S subunits.

Overnight cultures of E. coli BW25113 (including Keio strains) grown in LB media at 37°C were diluted 1/1000 in fresh LB, and incubated at 15°C (48 hr) and 37°C (24 hr) in duplicate without shaking in a final volume of 100 μl. Cells were grown in Corning (Corning, NY) Costar 96-well clear-bottom plates. For the small molecule screen, compounds were added to E. coli BW25113 to a final concentration of 10 μM. All screens were performed in duplicate. Molecules, dissolved in DMSO, were sourced from ChemBridge (San Diego, CA), Maybridge (Waltham, MA), MicroSource Discovery Systems (Gaylordsville, CT), Prestwick Chemicals (Washington, DC), and Biomol-Enzo Life Sciences (Farmingdale, NY). Liquid handling was performed using a Beckman Coulter (Brea, CA) FX^P Laboratory Automated Workstation. After incubation, plates were read using a Perkin Elmer (Waltham, MA) EnVision plate reader at 600 nm.

25 ml cultures of early-log E. coli BW25113 (OD = 0.2) grown in LB media at 15°C were treated with the appropriate concentration of each antibiotic (purchased from Sigma, St. Louis, MO) and, when applicable, pulse labeled with [¹⁴C]-uridine (purchased from American Radiolabeled Chemicals, St. Louis, MO) to a final concentration of 0.2 μCi/ml (specific activity 55 mCi/mmol). Cells were incubated as necessary, harvested by centrifugation, and lysed using a Constant Systems (Daventry, England) cell disruptor at 13 kpsi in 3 ml ice-cold ribosome buffer (20 mM Tris–HCl, pH 7.0, 10.5 mM MgOAc, 100 mM NH₄Cl, 3 mM β-mercaptoethanol). Cell lysates were clarified using a Beckman Coulter MLA-80 rotor at 24,000 rpm for 45 min, at which time they were loaded onto 35 ml 10–40% sucrose gradients and centrifuged for 18 hr at 18,700 rpm in a Thermo (Waltham, MA) SureSpin rotor. The volume of lysate added to each gradient was adjusted based on OD₆₀₀ of the DMSO-treated control culture to ensure reproducibility across experiments. Gradients were either fractionated using an AKTA Prime FPLC (GE Healthcare, Little Chalfont, England) outfitted with a continuous flow UV cell at 260 nm or analyzed via continuous flow UV and scintillation counting using an AKTA Prime FPLC in series with a Perkin Elmer 150TR flow scintillation analyzer.

Sucrose density gradients, loaded with clarified cell lysates normalized for OD₆₀₀, were ran as described above. Total rRNA from 500 μl sucrose gradient fractions was purified using phenol chloroform extraction followed by sodium acetate precipitation and dissolved in 5 μl of water. 1 μl of rRNA from each sample was added to 9 μl of water and 1 μl (2.4 pmol) of the necessary primer was added. Each 11 μl reaction was incubated at 80°C for 10 min and allowed to cool to room temperature in order to denature the rRNA. rRNA was subsequently reverse transcribed at 45°C for 24 hr using RevertAid H Minus Reverse Transcriptase from Thermo Scientific in a reaction volume of 20 μl according to the manufacturers instructions. cDNA products from each reaction were precipitated using sodium acetate and 90% ethanol and washed once in 70% ethanol. Purified cDNA samples were analyzed via capillary electrophoresis using a GeneScan 350 TAMRA size standard (Thermo Scientific). 16S rRNA and 23S rRNA primers containing a 5′ 6-oxyfluorescein marker were purchased from Sigma. 16S rRNA primer sequence: 5′-CTGTTACCGTTCGACTTG-3′. 23S rRNA primer sequence: 5′-CTTATCGCAGATTAGCACG-3′.

Reference standard ribosomal particles were prepared by growing E. coli strain NCM3722 in supplemented M9 (48 mM Na₂HPO₄, 22 mM KH₂PO₄, 8.5 mM NaCl, 10 mM MgCl₂, 10 mM MgSO₄, 5.6 mM glucose, 50 µM Na₃·EDTA, 25 mM CaCl₂, 50 µM FeCl₃, 0.5 µM ZnSO₄, 0.5 µM CuSO₄, 0.5 µM MnSO₄, 0.5 µM CoCl₂, 0.04 µM d-biotin, 0.02 µM folic acid, 0.08 µM vitamin B1, 0.11 µM calcium pantothenate, 0.4 nM vitamin B12, 0.2 µM nicotinamide, and 0.07 µM riboflavin) bearing 7.6 mM of either ¹⁴N or ¹⁵N-labeled (NH₄)₂SO₄. Cells were harvested at OD = 0.5 and lysed in buffer A (20 mM Tris–HCl, 100 mM NH4Cl, 10 mM MgCl2, 0.5 mM EDTA, 6 mM β-mercaptoethanol; pH 7.5) using a mini bead beater. Clarified lysates (5 ml) were layered above a 5 ml sucrose cushion (20 mM Tris–HCl, 500 mM NH4Cl, 10 mM MgCl2, 0.5 mM EDTA, 6 mM β-mercaptoethanol, 37% sucrose; pH 7.5) and were spun for 22 hr at 37.2k rpm in a Ti 70.1 rotor. Pellets bearing 70S ribosomes were solubilized in buffer A at 4°C and saved at −80°C. Lamotrigine- and DMSO-treated ribosomal particles were separated on a sucrose density gradient and fractions were collected as described above. A mixed reference standard bearing 10 pmol of ¹⁴N-labeled and 30 pmol of ¹⁵N-labeled 70S ribosomal particles was added to 20 pmol of each experimental fraction. The use of this mixed reference ensured that every ¹⁵N-labeled peptide bore a ¹⁴N-labeled peptide pair irrespective of the abundance of that peptide in the experimental sample. Additionally, one sample bearing only the reference standard was mixed with an equal volume of buffer A. These samples were then prepared for LC/MS via precipitation, reduction, alkylation, and tryptic digestion as described previously (Jomaa et al., 2014). Peptides were eluted from a C18 column using a concave acetonitrile gradient and detected using first an Agilent (Santa Clara, CA) G1969A ESI-TOF and second, to improve proteomic coverage and to identify non-ribosomal proteins, using an AB/Sciex (Framingham, MA) 5600 Triple-TOF run in MS² mode. In each case, the entire isotope distribution of each extracted MS¹ spectrum was fit using a Least Squares Fourier Transform Convolution algorithm (Sperling et al., 2008) providing accurate quantitation of the ¹⁴N and ¹⁵N species' abundance. To account for the reference standard's contribution to the measured ¹⁴N peptide abundance, each spectrum was normalized using the paired ¹⁵N abundance. Having measured the reference standard alone in triplicate, we then subtracted these normalized spectra, resulting in the corrected peptide abundance for each peptide in each experimental sample. Data sets from the ESI-TOF and Triple-TOF were merged and filtered for interference from co-eluting peptides. As a proof of principle, a series of standards bearing various quantities of ¹⁴N-labeled 70S particles were also analyzed to assess the linearity of our detection technique (Gulati et al., 2014). Non-ribosomal proteins were quantified across the gradient using the aforementioned MS² Triple-TOF data sets. These data sets were acquired as IDA experiments with 200 ms MS¹ scans followed by 50 MS² scans, each with 50 ms of ion accumulation. Precursor ions were excluded from MS² analysis 12 s after one occurrence. In each fraction, spectral counts for each non-ribosomal protein were normalized to the total number of spectral counts in that fraction. These values were then normalized to the maximal spectral counts in any gradient fraction, and the occupancy profile was smoothed using a 3-fraction sliding Gaussian window.

30 ml mid-log cultures of E. coli BW25113 (OD = 0.4) grown in M9 media were pulsed with 50% ¹⁵N by adding 30 ml of M9 containing ¹⁵N ammonium chloride as the sole nitrogen source. Cells were introduced to the necessary concentrations of chloramphenicol or lamotrigine during the pulse by supplementing the ¹⁵N M9 with antibiotic. Cells were pulsed for 0 hr, 1 hr, 2.6 hr, 4 hr, 8 hr, and 16 hr, at which times 10 ml of each culture was removed and the cells harvested via centrifugation. Cell pellets were frozen at −80°C prior to processing for mass spectrometry.

Pulse-labeled cells were spiked with 20 pmol of ¹⁵N-labeled 70S ribosomal particles and prepared for analysis on the ESI-TOF as described above. Extracted MS¹ spectra were fit using Least Squares Fourier Transform Convolution algorithm with three species: 0% ¹⁵N (pre-pulse), 50% ¹⁵N (post-pulse), and 100% ¹⁵N (Sperling et al., 2008). Each species was normalized to the reference resulting in the following: pre-pulse material [0%/100%], post-pulse synthesis [50%/100%], and total material ([0% + 50%]/100%). Synthesis rates were calculated for each r-protein independently by fitting the median post-pulse synthesis measurement for the 0, 4, 8, 16-hr time points to a line. The synthesis rate of each of r-protein in the lamotrigine or chloramphenicol treatment was normalized to that of the DMSO treatment and log-transformed. The resultant values were presented as notched box and whisker plots, centered at the growth rate for each treatment (Figure 6E). Whiskers extend to the most extreme data point within 1.5 times the inner quartile range (IQR) whereas notches extend from the median 1.57 × IQR/(number of points)^1/2. All aforementioned data analysis was performed using a series of Python scripts available at https://github.com/joeydavis/StokesDavis_eLife_2014.

Dense overnight cultures of E. coli BW25113 were diluted 1/1000 in 10 ml of LB media supplemented with 39 μM lamotrigine and grown at 15°C until cultures became dense. This occurred after 7 days of incubation. Potential suppressor clones from these cultures were subsequently passaged three times on LB agar. Single colonies from LB agar plates were then re-streaked onto LB agar supplemented with 39 μM lamotrigine to assess mutation stability and purify individual suppressor clones. Individual colonies were isolated based on colony diameter and analyzed via UV absorbance at 600 nm in liquid LB media to determine growth kinetics and lamotrigine MICs at 15°C. Kinetic growth assays were conducted in a temperature-controlled Tecan (Mannedorf, Switzerland) Sunrise plate reader. The ribosomal particles of these clones were analyzed using sucrose density gradient centrifugation as described above. Genomic DNA from wild-type E. coli BW25113 and lamotrigine suppressor mutants were purified using a Qiagen (Venlo, Netherlands) Gentra Puregene kit and sequenced using an Illumina (San Diego, CA) MiSeq platform. Paired-end 250 bp read data for wild type and mutant samples were aligned to the E. coli MG1655 chromosome (NC_000913) using BowTie2, and mutations were visualized and annotated using BreSeq and Tablet.

Wild type and mutant #3 E. coli infB genes were cloned into the pDEST17 plasmid containing an N-terminal His tag using the Invitrogen (Carlsbad, CA) Gateway cloning system. Protein expression was conducted in E. coli BL21-AI cells grown in LB at 15°C. Expression was induced at OD ∼0.6 using 0.2% arabinose, and cells were harvested after 16 hr of induction. Cells were lysed using a Constant Systems cell disruptor at 20 kpsi in IF2 lysis buffer (50 mM HEPES–KOH, pH 7.4, 1 M NH₄Cl, 10 mM MgCl₂, 0.1% Triton X-100, 7 mM β-mercaptoethanol) containing EDTA-free protease inhibitor tablets from Roche (Basel, Switzerland). Cell lysates were clarified via centrifugation at 20,000 rpm for 45 min in a Beckman Coulter JA-25.50 rotor. Clarified lysates were loaded onto a 1 ml GE Healthcare HisTrap FF column and eluted with IF2 elution buffer (50 mM HEPES–KOH, pH 7.4, 1 M NH₄Cl, 10 mM MgCl₂, 7 mM β-mercaptoethanol, 400 mM imidazole). Purified wild type and mutant IF2 was buffer exchanged into ice-cold ribosome buffer using an Amicon (Millipore, Billerica, MA) Ultracel filtration unit with a 50 kDa cutoff filter. 50 µl reactions containing 2 mg/ml BSA, 20 µM IF2, 200 nM [³H]-lamotrigine (specific activity 5 Ci/mmol; purchased from American Radiolabeled Chemicals), and 30 mM G-nucleotide (purchased from Sigma) were incubated at 15°C in ribosome buffer for 3 hr. Reactions were loaded onto 200 µl pre-wet Sephadex G-25 resin beds (resin purchased from GE Healthcare) and centrifuged at 400×g for 3 min. Flow-through samples were scintillation counted using Perkin Elmer Ultima Gold scintillation fluid.

1 ml early-log cultures of E. coli BW25113 (OD = 0.2) grown in M9 media were treated with 8× MIC of various antibiotics and were concurrently pulse labeled with [³⁵S]-methionine (purchased from Perkin Elmer) to a final concentration of 5 μCi/ml (specific activity 1175 Ci/mmol). Cells were incubated for 2.6 hr at 15°C, at which time they were harvested via centrifugation and washed twice in 1 ml 0.85% saline. Cells were lysed using 100 μl Millipore BugBuster Master Mix reagent and proteins were precipitated using 25 μl ice-cold 25% TCA. Protein pellets were then washed twice in 25 μl ice-cold 10% TCA and passed through Whatman (GE Healthcare) GF/C filters using a Millipore vacuum manifold. Filters were washed three times in ice-cold 10% TCA, dried overnight at room temperature, and scintillation counted using Perkin Elmer Ultima Gold scintillation fluid.

Cell-free translation was conducted using the E. coli S30 transcription–translation system for circular DNA from Promega (Finchburg, WI) according the manufacturer's instructions. 10 μl reactions containing the necessary antibiotics and plasmid DNA encoding the firefly luciferase gene were incubated either at 15°C for 4 hr or 37°C for 1 hr. Reactions were halted on ice for 5 min prior to addition of 25 μl of room-temperature luciferin (purchased from Promega). Immediately after the addition of luciferin, samples were analyzed for luminescence output in a Nunc (Roskilde, Denmark) 384-well clear bottom plate using a Tecan Ultra Evolution luminometer.

The GenBank accession numbers for the IF2 variants described in this paper are KJ752767 (wild-type E. coli BW25113 IF2); KJ52768 (mutant #1 IF2); KJ752769 (mutant #2 IF2); KJ752770 (mutant #3 IF2); and KJ752771 (mutant #4 IF2).

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

We thank Dr Mike Surette and Nicholas Waglechner from McMaster University for assistance with genome analysis; members of the Center for Microbial Chemical Biology at McMaster University for technical assistance with screening robotics; Dr Stephan Nord, formerly in the lab of EDB, for assistance in 5′ primer extension and lamotrigine suppressor generation; and Kella Kapnisi, formerly in the lab of EDB, for assistance with small molecule screening.

1. Received: June 4, 2014
2. Accepted: September 17, 2014
3. Accepted Manuscript published: September 18, 2014 (version 1)
4. Version of Record published: October 7, 2014 (version 2)

© 2014, Stokes et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.