Antibiotic-Induced Cell Chaining Triggers Pneumococcal Competence by Reshaping Quorum Sensing to Autocrine-Like Signaling

Abstract

Streptococcus pneumoniae can acquire antibiotic resistance by activation of competence and subsequent DNA uptake. Here, we demonstrate that aztreonam (ATM) and clavulanic acid (CLA) promote competence. We show that both compounds induce cell chain formation by targeting the d,d-carboxypeptidase PBP3. In support of the hypothesis that chain formation promotes competence, we demonstrate that an autolysin mutant (ΔlytB) is hypercompetent. Since competence is initiated by the binding of a small extracellular peptide (CSP) to a membrane-anchored receptor (ComD), we wondered whether chain formation alters CSP diffusion kinetics. Indeed, ATM or CLA presence affects competence synchronization by shifting from global to local quorum sensing, as CSP is primarily retained to chained cells, rather than shared in a common pool. Importantly, autocrine-like signaling prolongs the time window in which the population is able to take up DNA. Together, these insights demonstrate the versatility of quorum sensing and highlight the importance of an accurate antibiotic prescription.

Keywords: CSP; Streptococcus pneumoniae; antibiotics; aztreonam; bacterial stress response; clavulanic acid; competence; contact-dependent signaling; quorum sensing; transformation.

Graphical abstract

Figure 1

Competence in S. pneumoniae Is Activated by Several Classes of Antibiotics (A) Schematic overview of competence regulation by the ComD/E two-component system. (B) Growth curves (OD₅₉₅) and bioluminescence activity (RLU/OD₅₉₅) of S. pneumoniae in the presence of several antibiotics. Strain DLA3 (P_ssbB-luc) was grown in C+Y medium at pH 7.3, which is non-permissive for natural competence initiation, with (red lines) or without (black lines) addition of antibiotics. Average of three replicates and SEM are plotted. Concentrations of the antibiotics used: 0.4 μg/mL ciprofloxacin (CIP), 0.15 μg/mL HPUra, 28 μg/mL tobramycin (TOB), 10 μg/mL gentamicin (GEN), 28 μg/mL aztreonam (ATM), 0.12 μg/mL amoxicillin plus 2 μg/mL clavulanic acid (AMC), 0.12 μg/mL amoxicillin (AMX), and 2 μg/mL clavulanic acid (CLA).

Figure 2

AZT and CLA Do Not Shift Gene-Dosage Distribution (A) Effect of antibiotic treatment on origin-terminus ratio. Boxplots represent oriC-ter ratios as determined by real-time qPCR. Whiskers represent the 10th and 90th percentiles of data from Monte Carlo simulations. Strain DLA3 (P_ssbB-luc) was grown in medium without (control) or with the following compounds: 0.15 μg/mL HPUra, 28 μg/mL kanamycin (KAN), 28 μg/mL of ATM, and 2 μg/mL of CLA. Red box (HPura) matches with previous data showing an increase of the oriC-ter ratio (Slager et al., 2014). (B) Transcript copy number changes. Every point is the median fold-change of 51 genes as a function of the central gene’s position. Both ATM and CLA do not affect the oriC-ter ratio. HPUra analysis from Slager et al. (2014) is shown in red, as a positive control of oriC-ter ratio shift. Abbreviations: RE, rapid exposure; AE, adaptive exposure.

Figure 3

ATM and CLA Induce Cell Chaining by Binding to PBP3 (A) CRISPRi-dependent downregulation of pbp1a and pbp3 leads to competence induction. Depletion of pbp1a and pbp3 by induction of dCas9 with IPTG upregulates competence. In contrast, depletion of pbp2b and pbp2x does not have any effect on competence. Detection of competence development was performed in C+Y medium at a non-permissive pH (pH 7.3). 32 μM IPTG was added to the medium at the beginning. Average of three replicates and SEM are plotted. (B) Representation of the PBP profiles of whole cells. D39V and a Δpbp3 were treated with 2 and 20 μg/mL CLA, and 28 and 100 μg/mL of ATM and subsequently labeled with Bocillin-FL. Both ATM and CLA bind PBP3 in the D39V strain. Numbers indicate the different PBPs (i.e., 2b = PBP2B, 3 = PBP3). (C) ATM and CLA induce chain formation. Phase-contrast images. Cells were grown in C+Y, pH 6.8, until OD₅₉₅ 0.4 (stationary phase). Scale bar: 6 μm. (D) Length of the chains. The horizontal red line indicates the average of the number of cells per chain, while the purple line represents the control condition. The addition of ATM or CLA results in the formation of longer chains, as does the deletion and depletion of pbp3 and the depletion of pbp1a. In contrast, pbp2b depletion does not lead to a chain-forming phenotype. The absence of lytB also resulted in an increase of chain length, while its complementation restores normal chain length. ^∗Statistically significantly longer chains than wild-type (mean comparison test, p < 0.05). Cell count for each condition: 2,146, 2,100, 1,827, 2,324, 2,397, 952, 4,033, 5,775, and 3,136, respectively.

Figure 4

Synchronization of Competence Is Affected by Chain Formation Cells were grown in C+Y at competence-permissive pH 7.6, without antibiotics (green lines), in the presence of 2 μg/mL CLA (orange lines) or with 28 μg/mL ATM (red lines). At pH 7.6, cells become naturally competent but with a certain delay compared to pH 7.9. Therefore, the inducing effect of ATM and CLA can be more easily visualized at this pH. Average of three replicates and SEM are plotted for each of five inoculation densities: OD₅₉₅ of 10⁻¹, 10⁻², 10⁻³, 10⁻⁴, and 10⁻⁵. RLU/OD could not be accurately calculated for the two lowest inoculation densities, due to the OD detection limit of the plate reader.

Figure 5

CSP Is Shared in a Common Pool and Synchronizes Initiation of Competence (A) Time-lapse fluorescence microscopy. Two colonies with a fusion of the late competence gene ssbB to gfp are shown; one formed by cells of wild-type D39V (ADP249) and one formed by cells of a ΔcomC mutant D39V (ADP247), which also constitutively expresses a red fluorescent protein. White arrows in the 80-min frame show that both D39V and ΔcomC microcolonies became competent within the same time frame, independent of whether cells touch each other (left microcolony) or not (right microcolony). Scale bar: 4 μm. Note that the overlap, in the ΔcomC strain, of green SsbB-GFP foci with the red background causes the foci to appear yellow. (B) Synchronization of competence at the single-cell level. Cells of strains P_ssbB-ssbB-gfp (ADP249) and P_ssbB-ssbB-gfp, lytB::chl (ADP273) were grown in C+Y at competence-permissive pH 7.9; ADP249 was grown without antibiotics (green lines/areas) or with 28 μg/mL ATM (red lines/areas), and the ΔlytB ADP273 strain (orange lines/areas) without antibiotics. The highly permissive pH 7.9 used in this experiment allowed earlier competence development compared with the pH used in Figure 4, thereby reducing the required number of flow cytometry reads. The average of three replicates and SEM are plotted for each of four inoculation densities: OD₅₉₅ of 10⁻², 10⁻³, 10⁻⁴, and 10⁻⁵. Twelve thousand individual particles (single cells, diplococci, and/or chains) were detected for each replicate every 10 min along the experiment.

Figure 6

Cells in Chains Have Similar CSP Production Levels but Retain More CSP, Leading to an Extended Transformation Period (A) Graphical representation of the HiBiT experiment. Left, ComC (called CSP once outside the cell) and HiBiT are regulated by the comCDE promoter, and both precursors have a leader peptide signal, which is recognized, cleaved, and exported by ComAB. Once outside the cell, HiBiT interacts with the soluble protein LgBiT and yields bioluminescence (Wang et al., 2018). Right, HiBiT lacks the ComC leader peptide and accumulates in the cytoplasm, since it cannot be recognized and exported by ComAB. (B) CSP is exported at a similar rate in wild-type and ΔlytB mutant cells. Bioluminescence (relative luminescence units [RLU]) can be correlated with CSP export. In both the wild-type (ADP308) and the ΔlytB mutant (ADP310), the export rates are similar until the saturation point (1 × 10⁷ RLU). Cells were grown in C+Y at competence-permissive pH 7.6. At pH 7.6, cells become naturally competent but with a delay relative to pH 7.9, facilitating the visualization of the inducing effect of chaining. However, competence development occurs later than the RLU saturation point. Neither the comAB mutant (ADP311) nor the HiBiT version without the leader peptide (ADP312) showed any signal during the first 120 min (values below the threshold line of 100 RLU). After that, potentially due to cell lysis, the signal increased but was negligible compared to the exported version of the peptide. Two replicates are shown for each time point and condition. (C) Graphical representation of the experimental setup. Coincubation (1:1 proportion) of wild-type D39V (black, left panel) or ΔlytB (black, right panel) with ΔcomC mutant (pink) cells. D39V releases more CSP (green dots) into the common pool than the ΔlytB, and more ΔcomC cells become competent (green halo). (D) Competence induction in a comC mutant by wild-type D39V and the lytB mutant. Strains were coincubated (1:1) in C+Y medium (pH 7.9), with an initial density OD₅₉₅ of 10⁻⁴. Three replicates were analyzed by flow cytometry every 10 min to detect GFP signal (12,000 particles each). Green: coincubation of D39V and ADP247 (D39V, ssbB-gfp, comC::ery); orange: coincubation of ADP21 (lytB::chl) and ADP247. Note that on the y axis the percentage of GFP-positive cells reflects the percentage of all cells in the population (including cells not harboring the SsbB-GFP fusion). Average of three replicates and SEM are plotted. (E) Cell chaining widens the transformation time window. D39V (green) and ΔlytB (orange) were grown in C+Y at pH 7.9. Every 20 min, 1 μg of naked linear DNA with homology regions of 1 kb around the ssbB locus (ssbB-luc-kan; see STAR Methods) was added. After 20 min, the cultures incubated with DNA were plated with and without 250 μg/mL kanamycin, to collect the number of transformants and total viable counts, respectively. Three replicates are plotted for each condition. The highlighted area shows the temporal window in which cells were able to take up DNA.

Figure 7

Models of Global Quorum-Sensing Signaling (Left) and Local Quorum-Sensing or Autocrine-Like Signaling (Right) (A) S. pneumoniae secretes CSP to communicate with other cells (purple arrows) and to synchronize competence once a critical CSP (green dots) threshold is reached. In addition, self-sensing CSP (blue arrows) plays a role, as part of the CSP pool is retained by the producer cell (Prudhomme et al., 2016). Diplococci producing more CSP than average (highlighted in red) contribute to the increase of the extracellular CSP pool. (B) Because of neighbor communication, when the CSP threshold is reached, all diplococci synchronize and become competent at nearly the same time (green cells). (C) Microscopy of the ADP249 strain (P_ssbB-ssbB-gfp) after the first competence time point, showing that all diplococci synchronized and expressed GFP at the same time. Cells were collected for microscopy as described in STAR Methods. Scale bar: 4 μm. Two different fields of view are shown. (D) Chain formation shifts quorum-sensing signaling from a global mechanism to a local mechanism. CSP released by cells present in the same chain is retained and sensed by the same chain. (E) As a result of local quorum sensing, the extracellular pool is smaller, thereby reducing the communication with other cells and decreasing competence synchronization. However, as stochastic fluctuations are not buffered through the shared pool of CSP, individual chains of cells will initiate competence earlier than well-mixed populations consisting of diplococci. (F) Microscopy of the ADP273 strain (lytB::chl, P_ssbB-ssbB-gfp) after the first competence time point, showing that not all the chains expressed GFP as a result of the disruption of competence signaling.