Toxin-mediated depletion of NAD and NADP drives persister formation in a human pathogen

Abstract

Toxin-antitoxin (TA) systems are widespread in bacteria and implicated in genome stability, virulence, phage defense, and persistence. TA systems have diverse activities and cellular targets, but their physiological roles and regulatory mechanisms are often unclear. Here, we show that the NatR-NatT TA system, which is part of the core genome of the human pathogen Pseudomonas aeruginosa, generates drug-tolerant persisters by specifically depleting nicotinamide dinucleotides. While actively growing P. aeruginosa cells compensate for NatT-mediated NAD⁺ deficiency by inducing the NAD⁺ salvage pathway, NAD depletion generates drug-tolerant persisters under nutrient-limited conditions. Our structural and biochemical analyses propose a model for NatT toxin activation and autoregulation and indicate that NatT activity is subject to powerful metabolic feedback control by the NAD⁺ precursor nicotinamide. Based on the identification of natT gain-of-function alleles in patient isolates and on the observation that NatT increases P. aeruginosa virulence, we postulate that NatT modulates pathogen fitness during infections. These findings pave the way for detailed investigations into how a toxin-antitoxin system can promote pathogen persistence by disrupting essential metabolic pathways.

Keywords: Pseudomonas aeruginosa; NADase; Persisters; RES Domain; Toxin–antitoxin System.

Figure 1. The NatT toxin confers drug tolerance to P. aeruginosa.

(A) Schematic of the natR–natT locus and domain structure of the NatT toxin and NatR anti-toxin. (B, C) Expression of natT limits P. aeruginosa growth and increases tolerance. A ΔnatR-ΔnatT mutant carrying plasmids with an IPTG-inducible copy of natT and a cumate-inducible copy of natR was grown in LB supplemented with different concentrations of IPTG and cumate, as indicated. Growth (B) and tolerance to ciprofloxacin (2.5 μg/ml) (C) were recorded for each combination. Heatmaps in (B) show ODs reached in the presence of the inducers normalized to the highest OD value of the analyzed group. Heatmaps in (C) show the fraction of surviving cells after treatment with ciprofloxacin (pVC control plasmid). (D) Growth of P. aeruginosa wild-type and natT_E29D mutant in LB complex media. Stationary phase cultures were diluted into fresh medium at 15 h post inoculation. (E) Survival of P. aeruginosa wild-type and natT_E29D mutant upon exposure to tobramycin (16 μg/ml, solid lines) and ciprofloxacin (2.5 μg/ml, stippled lines) (average ± SEM, n = 3). (F) Tolerance is growth phase-dependent. Cultures of P. aeruginosa wild-type (triangles) and natT_E29D mutant (circles) were harvested at different stages of growth (indicated by OD) and exposed to tobramycin (16 μg/ml) for 3 h. Fractions of surviving cells from overnight stationary phase cultures are indicated in red (average ± SEM, n = 3). Source data are available online for this figure.

Figure 2. NatT is a NAD-dependent phosphorylase.

(A) Kinetics of NAD⁺ phosphorolysis using purified NatR–NatT_E29D complex (50 nM). The kinetic of the reaction was determined by quantifying the NAD⁺ concentration over time in a series of 1D ¹H NMR spectra. Km and Vmax values were determined by nonlinear regression analysis with the Michaelis–Menten equation (average ± SEM, n = 3). (B) 2D ¹H³¹P HMBC NMR NMR spectrum identifying ADPR-1P as the reaction product of NAD⁺ (5 mM) with 40 nM purified NatRT_E29D complex. Phosphate atoms from the ADP-ribose moiety are marked in blue and the phosphate derived from an added orthophosphate group is in red. (C) NatT depletes cellular NAD⁺ pools of P. aeruginosa. NAD⁺ and NADH concentrations were determined in P. aeruginosa wild-type (wt) and in strains harboring plasmid-encoded copies of natT_E29D and natT_{E29D R82A} growing exponentially (filled bars) or in stationary phase (open bars) (technical replicates, average ± SEM, n ≥ 2). (D) Expression of natT_E29D induces global changes of P. aeruginosa metabolites. Comparison of metabolites in strains harboring a plasmid with an IPTG-inducible copy of natT_E29D or a control plasmid. Metabolites significantly enriched or depleted in the natT_E29D strain are shown in red and green, respectively (n = 3). The P value was obtained from a t test of the metabolite levels. (E) NatT toxin activity is required for P. aeruginosa drug tolerance. Survival of different P. aeruginosa strains is shown during exposure to tobramycin (16 μg/ml) (technical replicates, average ± SEM, n > 3). Source data are available online for this figure.

Figure 3. Interaction of NatR and NatT mediates toxin activation.

(A) Crystal structure of a NatR–NatT hexamer with a central NatT dimer and two flanking dimers of NatR (NatR’ and NatR”). (B) NatR–NatT interaction is mediated by the Flap region. Left: Surface representation of NatT [protomer 1] with its Flap region (yellow) sandwiched between the interacting NatR’ and the neighboring NatT [protomer 2]. Right: Surface representation of NatT with NatR in cartoon style and individual helices marked. NatR’ interacts with the Flap region of NatT and α7 of NatR’ extends into the active site groove of NatT. (C) Detailed view of NatT-NatR’ interface with negatively charged residues of the Flap (yellow) and positively charged residues of NatR’ indicated in stick representation. Phosphate moieties in the active site and in the NatR–NatT interface are indicated as space-filled molecules. (D) Conservation of Flap region (NatT residue 26–66). Positions of conserved residues interacting with NatR’ are highlighted. K54 and Y66 coordinate the phosphate molecule in the NatR–NatT interface (see: E). (E) Zoom-in views comparing the interaction of NatR’ with NatT wild-type and NatT_E29D. Residues with altered conformation are highlighted in stick representation with coloring scheme as in (C). The movement of residue E26 away from its interaction with the backbone of α7 of NatR’ (blue) is marked by a red circle. (F) Mass photometry analysis of NatR–NatT (blue) and NatR–NatT_E29D (yellow) at 10 nM. The mass of the NatR–NatT_E29D complex corresponds to a NatR’-NatR”-NatT complex. Source data are available online for this figure.

Figure 4. NatR is an anti- and a co-toxin of NatT.

(A) Schematic model of NatT activation and autoregulation. A catalytically inactive hexameric complex (4R:2T) acts as repressor of natR–natT transcription. Transition to a 2R:1T complex leads to derepression of natR–natT transcription and under specific conditions to NatT NADase activation. The postulated regulatory role of the NAD precursor nicotinamide (NAM) is indicated. (B) NatR (turquoise) and NatT (orange) levels are increased in a natT_E29D mutant. Proteins were quantified by mass spectrometry and plotted as a function of growth (optical density) of P. aeruginosa wild-type (wt) and natT_E29D mutant. (C) Transcription of natT is controlled by the NatR–NatT toxin complex. Transcription was determined in reporter strains containing a chromosomal copy of gfp downstream of natT. Fluorescence of a control strain lacking the gfp reporter is indicated (dotted lines). (D) NatR protein levels are increased in natT_E29D and ΔnatT mutants. Levels of NatR were determined by quantitative mass spectrometry analysis of P. aeruginosa wild-type and mutants indicated. (E) Transcription of natR–natT is induced in the natT_E29D background. Cultures of P. aeruginosa wild-type, natT_E29D and natT_E29D ΔrelA ΔspoT mutants harboring a chromosomal natT::gfp reporter were assayed by flow cytometry during different phases of growth (see: Fig. EV4C). Black lines show baseline auto-fluorescence of a strain lacking a gfp reporter gene. (F) NatT protein levels are reduced in a strain lacking NatR. Immunoblots of P. aeruginosa wild-type and ΔnatR mutants with chromosomal natT-FLAG or natT_E29D-FLAG alleles. Cell extracts were harvested at different ODs and stained with anti-FLAG and anti-RpoB antibodies and NatT levels normalized to RpoB. (G) NatR–NatT binds to its own promoter. Mass photometry of NatR–NatT (blue) and NatR–NatT_E29D (yellow) without and with DNA (red) containing the natR promoter region. The mass of dsDNA is shown in black. (H) Transcription of natR–natT activity is coupled to NatT activity. Expression of a chromosomal natR-ΔnatT::gfp reporter was assayed by flow cytometry in a P. aeruginosa ΔnatT mutant expressing different natT alleles from a plasmid, as indicated. The black line represents the fluorescent signal in a control strain lacking a gfp reporter. One representative experiment is shown (pVC = plasmid control). Source data are available online for this figure.

Figure 5. The NAD salvage pathway neutralizes NatT toxin activity and abolishes drug tolerance.

(A) Schematic of NAD⁺ de novo and salvage pathways in P. aeruginosa (top) with intermediates and catalyzing enzymes indicated. Salvage pathway regulation by NrtR is shown below with enzymes and genes being highlighted in matching colors. The position of the chromosomal gfp reporter used for these studies is in green. Impact of natT_E29D expression on the growth of P. aeruginosa wild-type and different salvage pathway mutants is shown at the bottom. Strains carrying an IPTG-inducible copy of natT_E29D on a plasmid were grown in LB medium with or without IPTG. The heatmap shows the OD after 8 h of growth normalized to the OD of wild-type. (B) Transcription of NAD salvage pathway genes in individual cells of P. aeruginosa wild-type, natT_E29D, and ΔnrtR mutants harboring a pncB1::gfp reporter was analyzed by flow cytometry. GFP signals are shown for individual cells and as histograms. A strain lacking the gfp reporter is shown as control (gray dots and stippled line). The area with average GFP signals higher than wild-type (GFP⁺) is marked in yellow. Please note the small subfraction of natT_E29D mutant cells (orange dots) with increased expression of the pncB1::gfp reporter (yellow stippled box). (C) NatT-mediated activation of salvage pathway genes declines in a growth phase-dependent manner. Fractions of cells expressing salvage pathway genes are shown as a function of the optical densities of cultures analyzed. (D) Nicotinamide (NAM) abolishes NatT-mediated drug tolerance. P. aeruginosa natT_E29D grown in LB supplemented with different concentrations of NAM was challenged with tobramycin (32 μg/ml) for 3 h and fractions of surviving cells were determined. (E) NAM abolishes NatT-mediated NAD⁺ salvage pathway induction. Fractions of cells with induced NAD⁺ salvage pathway were determined as in (C) in the presence or absence of NAM (20 mM) (technical replicates, average ± SEM, n = 2). (F) NAM blocks NatT-mediated derepression of the natR–natT operon. Growing cultures of the natRT_E29D::gfp reporter strain were supplemented with NAM (25 mM) and analyzed by at times as indicated. A control strain lacking a gfp reporter is shown (dotted line). (G) Salvage pathway derepression partially abolishes NatT-mediated drug tolerance (technical replicates, average ± SEM, n = 3). Experiments were carried out as indicated in (D). Source data are available online for this figure.

Figure 6. NatT activation generates persister subpopulations with extended lag phase.

(A, B) The P. aeruginosa natT_E29D mutant generates growth-arrested cells. Cultures of P. aeruginosa wild-type and natT_E29D mutant expressing TIMER were grown in LB and samples were removed at indicated the time points (t1–t6), diluted into fresh medium (OD 0.1) and analyzed by flow cytometry after three hours of growth. Fractions of arrested cells are plotted in (A) (solid lines). (B) Cells harvested at time point t5 (A) were spotted on LB agar patches, and their growth was analyzed by time-lapse microscopy (wild-type, n = 450; natT_E29D, n = 150). Bar plots indicate fractions of cells with delayed re-growth and different lag times (black line=median). (C) The natT_E29D mutant generates growth-arrested cells with increased drug tolerance. Stationary phase cultures of wild-type and natT_E29D mutant expressing TIMER were harvested at time point t5 (A), diluted into fresh LB medium for three hours, and sorted by FACS using the red and green fluorescence channels (left panels). Subpopulations were harvested and survival was determined upon exposure to tobramycin (right panel) (technical replicates, average ± SEM, n = 3). (D) Dual-input mother machine with P. aeruginosa cells expressing TIMER growing in parallel side channels supplied by media influx from a central media channel. Fluorescence images are assembled from one single channel representing the growth of cells over time as indicated by the arrows. Cells were loaded into the microfluidic device, supplied with LB for 3 h, then gradually switched to spent LB (SLB) for 12.5 h before they were again supplied with fresh LB for 3 h (outgrowth), followed by exposure to LB containing tobramycin (16 µg/ml) for 3 h (treatment) and finally growth resumption in LB. TIMER fluorescence indicates growing (green) and resting (red) cells. (E) Persister frequency drops with increasing length of outgrowth. The frequency of cells surviving antibiotic exposure was measured in a microfluidic device as indicated in (D) with variable length of outgrowth periods. The analysis included a total of 122,704 cells with 241 persisters lineages. (F) Representative examples of persister lineages (blue) grown in microfluidic devices as indicated in (D). The growth and division of individual cells is indicated by length measurements over time. Representative susceptible lineages (orange) are shown as control. Stages during which specific media were supplied are indicated in the same color code as in (D). Source data are available online for this figure.

Figure 7. The NatRT module is widespread in proteobacteria and is under selection during infections.

(A) Frequency distribution of natT alleles in 8286 P. aeruginosa strains analyzed. Single SNPs and SNP combinations are indicated in different colors. Arrows indicate the natT alleles that were tested in this study. (B) Different natT alleles confer different levels of drug tolerance. Cultures of P. aeruginosa carrying plasmids with different natT alleles from (A) were scored for survival after 3 h of treatment with tobramycin (technical replicates, average ± SEM, n = 3). (C) Salvage pathway induction by different natT alleles. A P. aeruginosa pncB1::gfp reporter strain carrying plasmids with different natT alleles under IPTG control was grown with (filled bars) or without (open bars) IPTG. Fractions of cells with a derepressed salvage pathway (GFP⁺) are indicated. (D) NatT activity controls P. aeruginosa virulence in a simple insect larvae model. Survival rates of G. mellonella larvae are shown for different P. aeruginosa strains as indicated. Experiments were carried out with five larvae per strain tested. (E) Bacterial phylogeny with the distribution of NatR–NatT TA system orthologs (red bars). Source data are available online for this figure.

Figure EV1. The NatT toxin confers drug tolerance to P. aeruginosa.

(A) Reduced fitness of a P. aeruginosa natT_E29D mutant. P. aeruginosa strains expressing different fluorophores were mixed 1:1 and subjected to consecutive cycles of growth and re-dilution. Subpopulations were analyzed by flow cytometry at indicated time intervals. (B, C) Survival of ΔnatT and ΔnatR-ΔnatT mutants during treatment with tobramycin (8 μg/ml) (B) or ciprofloxacin (2.5 μg/ml) (C) (technical replicates, average ± SEM, n = 3). (D) Expression of natT impairs P. aeruginosa growth. Cultures of a ΔnatT mutant containing plasmids with IPTG-inducible natT alleles were grown in LB medium with or without IPTG (average ± SEM, n > 3) (pVC = control plasmid). (E) Expression of natT increases tolerance to tobramycin. Cultures of P. aeruginosa containing plasmids with IPTG-inducible natT or natR alleles were grown in LB medium without or with IPTG. Fractions of surviving cells were determined after three hours of treatment with tobramycin (16 μg/ml). Median values are indicted (n ≥ 3) (pVC = control plasmid). (F) Expression of natT increases tolerance to ciprofloxacin. Cultures of P. aeruginosa containing plasmids with IPTG-inducible natT or natR alleles were grown in LB medium without (empty boxes) or with 250 μM IPTG (filled boxes). Fractions of surviving cells were determined after three hours of treatment with ciprofloxacin (2.5 μg/ml) (n ≥ 5; lines mark medians) (pVC = control plasmid). (G) Expression of natT_E29D gradually limits P. aeruginosa growth. Cultures of a ΔnatT mutant containing a control plasmid (stippled line) or a plasmid expressing natT_E29D from an inducible promoter, were grown in LB with increasing concentrations of IPTG as indicated (average ± SEM, n > 3). (H) Expression of natT_E29D increases P. aeruginosa tolerance. P. aeruginosa cultures containing a plasmid with an IPTG-inducible natT_E29D were grown as in (G) and survival was determined after three hours of treatment with tobramycin (16 μg/ml) (average ± SEM).

Figure EV2. NatT is a NAD-dependent phosphorylase.

(A) Purification of NatR–NatT complex from P. aeruginosa. NatR and NatT proteins were copurified from a P. aeruginosa ΔnatT mutants harboring plasmids containing FLAG-tagged natT alleles by affinity chromatography using anti-FLAG beads. (B) Degradation of NAD⁺ by NatT_E29D is phosphate-dependent. Purified NatR–NatT_E29D complex was incubated with NAD⁺ with and without phosphate, and the reaction was analyzed by 1D ¹H and ³¹P NMR spectra. (C) Purified NatR–NatT wild-type does not degrade NAD⁺. (D) Purified NatR–NatT_E29D degrades NADP⁺ in a phosphate-dependent reaction. (E) NatT depletes cellular NADP pools of P. aeruginosa. NADP⁺ and NADPH concentrations were determined in cultures of P. aeruginosa wild-type (wt) and in strains harboring plasmids expressing natT_E29D or natT_{E29D R82A} growing exponentially (filled bars) or from stationary phase (open bars) (average ± SEM, n = 2). (F) The active site residue R82 is required for NatT-mediated toxicity. P. aeruginosa with plasmids containing IPTG-inducible copies of natT_E29D (black line) or natT_{E29D R82A} (green line) were grown in LB with (dotted lines) or without IPTG (solid lines) (average ± SEM, n = 3).

Figure EV3. Interaction of NatR and NatT mediates toxin activation.

(A) SEC-MALS analysis of NatR–NatT and NatR–NatT_E29D complexes reveal different oligomeric structures. Experiments were performed with 7 µM (red) and 2 µM (orange) NatR–NatT and with 7 µM (green) and 2 µM (blue) NatR–NatT_E29D. (B) Structural homology between NatT and diphtheria toxin indicates analogous NAD⁺ binding modes. NatT is shown in light brown and yellow (Flap), diphtheria toxin (1tox) is in gray and its active site NAD⁺ is in pink. (C) Superposition of P. aeruginosa NatT (light brown) with Flap (yellow) and RES domain proteins RESpp (Skjerning et al, 2019), MbcT (Freire et al, 2019) and ParT (Piscotta et al, 2019) (gray). (D) Structure-guided sequence alignment of NatT and RES domain proteins shown in (B). Predicted active site residues are marked with purple stars. Charged residues of the Flap involved in interaction with NatR’ are marked with red asterisks.

Figure EV4. NatR is an anti- and a co-toxin of NatT.

(A) Introduction of a chromosomal natT::gfp reporter does not affect P. aeruginosa drug tolerance. Survival of P. aeruginosa was scored during exposure to tobramycin (16 μg/ml) (technical replicates, average ± SEM, n > 3). (B) NatR is a repressor of natT transcription. Transcription of natT was determined by flow cytometry in ΔnatR mutants with a gfp reporter downstream of natT. Strains contained a control plasmid (orange) or a plasmid expressing natR (green). Control strains lacking the gfp reporter are indicated by a dotted black line. (C) Growth of P. aeruginosa wild-type, natT_E29D and natT_E29D ΔrelA ΔspoT in LB. Samples from different time points (t1-t10) were used for flow citometry analysis in Fig. 4E (n = 1). (D) Drug tolerance is not affected by an engineered chromosomal natT-FLAG allele (technical replicates, average ± SEM, n = 3). (E) NatT protein levels increase in P. aeruginosa natT_E29D upon entry into stationary phase. Levels of NatT and NatT_E29D were determined during growth (see: C) by immunoblot analysis using anti-FLAG antibodies and anti-RpoB antibodies as control. (F) Ectopic expression of natT leads to derepression of natR–natT transcription. Transcription of natR–natT was determined in P. aeruginosa harboring a chromosomal natT::gfp reporter and plasmids with IPTG-inducible natT alleles as indicated (pVC = control plasmid). (G) NatT is degraded in strains lacking NatR. Concentrations of NatT were determined by immunoblot analysis after treating P. aeruginosa cultures with chloramphenicol and plotted as relative values of the initial concentration (0 h) (technical replicates, average ± SEM, n ≥ 3). (H) NatT is insoluble in strains lacking NatR. NatT protein was quantified by immunoblot analysis of fractions harvested from different P. aeruginosa strains and in different growth phases as indicated. F: full lysate; S: soluble fraction; I: insoluble fraction. (I) NatT-mediated toxicity is abolished in a ΔnatR mutant. Growth of P. aeruginosa wild-type and ΔnatR mutant carrying a plasmid with an IPTG-inducible natT_E29D allele was recorded in LB medium with or without IPTG as indicated (average ± SEM, n = 3). (J) NatR is required for NatT-mediated drug tolerance. Survival of P. aeruginosa wild-type and mutants indicated during exposure to tobramycin (average ± SEM, n > 3). (K) NatR is required for NatT-mediated drug tolerance. Survival of P. aeruginosa wild-type and ΔnatR mutant carrying plasmids with IPTG-inducible copies of natT or natT_E29D was determined after treatment with tobramycin (16 μg/ml). Cultures were grown with (filled circles) or without IPTG (empty circles) (average ± SEM, n = 3). (L) Ectopic expression of natR restores drug tolerance of a ΔnatR–natT_E29D mutant. Survival to tobramycin is shown for strains indicated. Plasmid pnatR harbors an IPTG-inducible copy of natR. The box extends from the lower (25th percentile) to upper quartile (75th percentile) values, with a line at the median (50th percentile). Whiskers indicate the minimum and maximum values within 1.5 times the interquartile range (n = 3).

Figure EV5. The NAD salvage pathway neutralizes NatT toxin activity and abolishes drug tolerance.

(A) Ectopic expression of natT mediates NAD⁺ salvage pathway induction. Fractions of cells inducing salvage pathway genes were determined in P. aeruginosa pncB1::gfp reporter strains harboring plasmids expressing different natT alleles from an IPTG-inducible promoter. Cultures were grown with or without IPTG (average ± SEM, n = 3) (pVC control plasmid). (B) Ectopic expression natT_E29D induces salvage pathway genes. Representative microscopy images of a pncB1::gfp reporter strain expressing natT_E29D from a plasmid with or without IPTG. (C) Limiting natR expression induces salvage pathway genes. P. aeruginosa ΔnrtR pncB1::gfp carrying a plasmid with an IPTG-inducible natR was analyzed at different IPTG concentrations as indicated. Fractions of cells with derepressed salvage pathway were scored as a function of natR expression. (D, E) NAM neutralizes NatT_E29D-mediated growth defect in P. aeruginosa wild-type (D) or ΔnadD2 salvage pathway mutant (E). Strains harboring a plasmid with an IPTG-inducible natT_E29D allele were grown with (orange) or without IPTG (black) and with (solid lines) or without NAM (20 mM) (dotted lines). (F) NAM overrides NatT-mediated growth arrest. Cultures of P. aeruginosa wild-type and natT_E29D mutant constitutively expressing TIMER were grown in LB with or without NAM for 3 h before analyzing populations by flow cytometry. Fractions of slow-growing or arrested cells were calculated using GFP/DsRed ratios of individual cells. (G) NAM limits NatR and NatT protein levels. A P. aeruginosa natR_E29D mutant was grown in LB with (open circles) or without NAM (closed circles), followed by the analysis of relative levels of NatR and NatT by mass spectrometry (technical replicates, average ± SEM, n = 3). (H, I) (p)ppGpp is required for NatT-mediated drug tolerance. (H) Cultures of P. aeruginosa strains indicated were treated with tobramycin and survival was determined over time (technical replicates, average ± SEM, n = 3). (I) Survival of P. aeruginosa wild-type and ΔrelA ΔspoT mutant carrying plasmids with IPTG-inducible natT (wt) or natT_E29D (E29D) alleles was determined after three hours of treatment with tobramycin. Cultures were grown with (filled circles) or without IPTG (open circles). Solid lines mark median values.

Figure EV6. NatT activation generates subpopulations with extended lag phase during outgrowth.

(A) Expression of TIMER from the chromosomal attb locus does not affect P. aeruginosa drug tolerance. Cell survival during tobramycin treatment was determined for isogenic strains with (dotted line) or without (solid line) TIMER (technical replicates, average ± SEM, n = 3). (B) The natT_E29D allele induces growth arrest in a subpopulation of P. aeruginosa cells. Samples of P. aeruginosa wild-type and natT_E29D mutant cultures expressing TIMER were harvested at different time points (see: Fig. 6A), diluted into fresh medium for three hours, and analyzed by flow cytometry. Fractions of slow-growing cells (black gate) are indicated. (C) Deletion of nrtR abolishes the population of slow-growing natT_E29D mutant cells. Experiments were carried out as in (B) with cells being harvested at time point t5. Strains are indicated above the panels.