Inhibition of host PARP1 contributes to the anti-inflammatory and antitubercular activity of pyrazinamide

Abstract

The antibiotic pyrazinamide (PZA) is a cornerstone of tuberculosis (TB) therapy that shortens treatment durations by several months despite being only weakly bactericidal. Intriguingly, PZA is also an anti-inflammatory molecule shown to specifically reduce inflammatory cytokine signaling and lesion activity in TB patients. However, the target and clinical importance of PZA's host-directed activity during TB therapy remain unclear. Here, we identify the host enzyme Poly(ADP-ribose) Polymerase 1 (PARP1), a pro-inflammatory master regulator strongly activated in TB, as a functionally relevant host target of PZA. We show that PZA inhibits PARP1 enzymatic activity in macrophages and in mice where it reverses TB-induced PARP1 activity in lungs to uninfected levels. Utilizing a PZA-resistant mutant, we demonstrate that PZA's immune-modulatory effects are PARP1-dependent but independent of its bactericidal activity. Importantly, PZA's bactericidal efficacy is impaired in PARP1-deficient mice, suggesting that immune modulation may be an integral component of PZA's antitubercular activity. In addition, adjunctive PARP1 inhibition dramatically reduces inflammation and lesion size in mice and may be a means to reduce lung damage and shorten TB treatment duration. Together, these findings provide insight into PZA's mechanism of action and the therapeutic potential of PARP1 inhibition in the treatment of TB.

Fig. 1. PZA aligns with the PARP1 active site and shifts the PARP1 thermal stability.

a Structures of pyrazinamide (PZA), nicotinamide (NAM) and benzamide. b Cartoon of the human PARP1 catalytic domain ART fold (green) co-crystalized with the non-hydrolyzable NAD⁺ analog benzamide adenine dinucleotide (BAD; left) or aligned with PZA (right). Ligands and key amino acids are drawn in stick representation, and interacting residues and approximate distances (Å) are indicated. Hydroxyl groups are colored red, nitrogen atoms are blue, and the carbon backbone is yellow (BAD) or teal (PZA). Note that PZA is predicted to form one more bond with PARP1 than BAD (Phe897). The PARP1-BAD crystal structure was resolved by Langelier et al. and accessed from the Protein Data Bank (Accession number: 6BHV). c PZA directly binds to PARP1 and shifts the PARP1 melting temperature (T_m) in a dose-dependent manner. PARP1 T_m (left) and derivative melt curves (right) in the presence of 0.25 mM or 1.0 mM NAM (orange) or PZA (red). Derivative melt curves represent the change in SYPRO orange fluorescence intensity over increasing temperatures of PARP1 alone (green) or with increasing concentrations of NAM (top) or PZA (bottom). The temperature above 40 °C associated with the lowest point in the derivative melt curve (vertical line) was considered the PARP1 T_m (46 °C). Source data are provided as a Source Data file.

Fig. 2. PZA inhibits PARP1 enzymatic activity in macrophages.

a Schematic overview. PMA-differentiated THP-1 cells were incubated with 100 or 250 µM PZA, pyrazinoic acid (POA), NAM or nicotinic acid (NA), or with 10 or 25 nM of the PARP1/2 inhibitor talazoparib (Tp), for 60 min before the addition of PARP1/2 activator MNNG (5 µM). b Representative Western blot showing PAR levels (top) or the loading control β-Actin (bottom). c Densitometric analysis showing the fold change in ß-Actin-normalized PAR levels relative to cells treated with MNNG alone. UNT is untreated (no MNNG). Error bars represent the SEM from seven independent experiments (individual values indicated by scatter plot). ***p = 0.0002 (PZA high dose) or 0.0001 (NAM low dose); ****p < 0.0001; ns not significant; by repeated measures (RM) one-way ANOVA with Dunnett’s multiple comparisons test. PAR levels in cells treated with 250 µM PZA (22.9%) were significantly lower than with 100 µM PZA (43.1%) but not statistically different from cells treated with 100 µM NAM (14.6%). Source data are provided as a Source Data file.

Fig. 3. PZA reverses TB-induced PARP1 activation in mouse lungs.

a Schematic study overview. Female C3HeB/FeJ mice were aerosol infected with M.tb H37Rv (implantation: 111 ± 9 CFU) and administered PZA (150 mg/kg), the PARP inhibitor talazoparib (Tp; 0.5 mg/kg) or vehicle (1.97% DMSO in 0.5% CMC), alone or in combination with the TB antibiotic RIF (10 mg/kg), 5 days/week for 2 months starting 1 month post infection. Bacterial burden and PAR levels were assessed before and after treatment and in age-matched uninfected control mice. b Representative Western blots showing PAR levels (top) or the loading control β-Actin (bottom) in mouse lungs after 2 months of treatment. Each lane represents an individual mouse. c Densitometric analysis showing the change in ß-Actin-normalized lung PAR intensity relative to the mean in vehicle-treated mice. n = 6 (uninfected) or 8 (all other groups). Each symbol represents an individual mouse and bars the group mean ± SEM. ns not significant (RIF vs. vehicle); *p = 0.0159 (PZA vs. vehicle) or 0.0433 (uninfected vs. vehicle); **p = 0.0071 (Tp vs. vehicle) by two-way ANOVA with uncorrected Fisher’s least significant difference (LSD) test. d Corresponding difference in lung bacterial burden following treatments compared to vehicle-treated mice. n = 9 (vehicle) or 8 (all other groups). Each symbol represents an individual mouse and bars the group mean ± SEM. ns not significant (Tp vs. vehicle); *p = 0.0425 (PZA vs. vehicle); **p = 0.0013 (PZA vs. RIF); ****p < 0.0001 (RIF vs. vehicle) by one-way ANOVA with Tukey’s multiple comparisons test. Mice treated with PZA or Tp had significantly reduced lung PAR levels comparable with uninfected mice. PAR levels in RIF-treated mice were not statistically different from vehicle-treated mice, even though the bacterial burden was significantly more reduced by RIF than by PZA or Tp. Source data are provided as a Source Data file.

Fig. 4. Adjunctive PARP inhibition reduces TB lung inflammation.

Lung histopathology of M.tb-infected female C3HeB/FeJ mice (implantation: 111 ± 9 CFU) 3 months post infection following 2 months of treatment with PZA (150 mg/kg), talazoparib (Tp; 0.5 mg/kg) or vehicle (1.97% DMSO in 0.5% CMC), alone or in combination with RIF (10 mg/kg). a Schematic overview. b Lung bacterial burden at the end of treatment. n = 8 (Tp), 10 (vehicle, RIF, RIF + PZA) or 11 (PZA, RIF + Tp). Statistical differences were determined by one-way ANOVA with Šidák’s multiple comparisons test. **p = 0.0067; ***p = 0.0006; ****p < 0.0001. c Representative H&E-stained lung sections. d–f Quantified areas of lung inflammation (d), % lung involvement (e), or quantified vimentin-positive area indicative of fibroblasts expressed as a percent of total lung area (f). Each symbol represents an individual mouse, with mean ± SEM indicated. n = 2 (Tp) or 3 (all other groups). Statistical differences were determined by one-way ANOVA with uncorrected Fisher’s LSD test (d, e) or Dunnett’s multiple comparisons test (f) with a single pooled variance. *p < 0.05; **p < 0.01; ***p < 0.001; ****p < 0.0001; exact p values are provided in the Source Data file. Histopathology and IHC analyses were performed on randomized and coded slides by a veterinary pathologist blinded to experimental design. Adjunctive PARP inhibition reduced lung inflammation independently of bacterial burden. Source data are provided as a Source Data file.

Fig. 5. PZA’s anti-inflammatory effects are PARP1-dependent.

a Schematic overview. Male and female PARP1-null (PARP1^−/−) or 129S1 (WT) mice were aerosols infected with the PZA-resistant M.tb H37Rv ∆pncA (A146V) mutant (implantation: 54 ± 5 CFU). Starting one month after infection, half of the mice were administered PZA (150 mg/kg) 5 days/week for 2 months before lung bacterial burden and cyto-/chemokine levels (Luminex multiplex assay) were determined. n = 9 (PZA-treated WT) or 10 (all other groups). b, c Bacterial burden (b) and cytokine/chemokine concentrations (c) in untreated (−) or PZA-treated (+) mice at the end of treatment. d Change in cyto-/chemokine levels in PZA-treated mice relative to the levels in untreated mice of the same strain. Values below 1 indicate levels that are lower in the lungs of PZA-treated than untreated mice. Each symbol represents an individual mouse, with mean ± SEM indicated. Statistical differences between groups were determined by two-way ANOVA with Šidák’s multiple comparisons test. p Values for all relevant comparisons are indicated in the figure. PZA reduced lung cyto- and chemokine levels in WT but not in PARP1^−/− mice and independently of bacterial burden. Source data are provided as a Source Data file.

Fig. 6. PZA has reduced efficacy in PARP1^−/− mice.

a Study outline to compare the efficacy of PZA in WT and PARP1^−/− mice. Male and female PARP1-null (PARP1^−/−) or 129S1 (WT) mice were aerosols infected with M.tb H37Rv (implantation: 80 ± 7.5 CFU). Starting one month after infection, half of the mice were administered PZA (150 mg/kg) 5 days/week for 2 months. Lung bacterial burden was assessed at the beginning and end of treatment. b–d Lung bacterial burden on the day after infection (b; n = 8), at the start of treatment (c; n = 6), and in untreated (−) or PZA-treated (+) mice at the end of treatment (d; n = 5 (untreated WT), 6 (untreated PARP1^−/−) or 8 (PZA-treated)). Each symbol represents an individual mouse, with mean ± SEM indicated. Statistical differences between groups were determined by two-tailed unpaired t-test (b, c) or two-way ANOVA with Šidák’s multiple comparisons test (d). *p = 0.0135 (PARP1^−/− untreated vs. PZA) or 0.0361 (PZA: WT vs. PARP1^−/−); ***p = 0.0005; ns not significant. After 2 months of PZA treatment, significantly more bacilli remained in the lungs of PARP1^−/− than of WT mice (5.234 vs. 4.437 log₁₀ CFU, respectively). e Proposed model: PARP1 inhibition is a key component of PZA’s mechanism of action in TB therapy. Without treatment (left), M.tb infection activates PARP1 which promotes the production of inflammatory mediators, immune cell activation and M.tb containment at the cost of lung damage. Inhibiting PARP1 during TB therapy (right), by PZA or adjunctive PARP inhibition, can accelerate bacterial clearance and the resolution of lung damage while dampening inflammation. However, PARP1 inhibition without adequate antibiotics may impair bacterial containment. Source data are provided as a Source Data file.