Activity of N-Acetylcysteine Alone and in Combination with Colistin against Pseudomonas aeruginosa Biofilms and Transcriptomic Response to N-Acetylcysteine Exposure

ABSTRACT Chronic colonization by Pseudomonas aeruginosa is critical in cystic fibrosis (CF) and other chronic lung diseases, contributing to disease progression. Biofilm growth and a propensity to evolve multidrug resistance phenotypes drastically limit the available therapeutic options. In this perspective, there has been growing interest in evaluating combination therapies, especially for drugs that can be administered by nebulization, which allows high drug concentrations to be reached at the site of infections while limiting systemic toxicity. Here, we investigated the potential antibiofilm activity of N-acetylcysteine (NAC) alone and in combination with colistin against a panel of P. aeruginosa strains (most of which are from CF patients) and the transcriptomic response of a P. aeruginosa CF strain to NAC exposure. NAC alone (8,000 mg/L) showed a limited and strain-dependent antibiofilm activity. Nonetheless, a relevant antibiofilm synergism of NAC-colistin combinations (NAC at 8,000 mg/L plus colistin at 2 to 32 mg/L) was observed with all strains. Synergism was also confirmed with the artificial sputum medium model. RNA sequencing of NAC-exposed planktonic cultures revealed that NAC (8,000 mg/L) mainly induced (i) a Zn2+ starvation response (known to induce attenuation of P. aeruginosa virulence), (ii) downregulation of genes of the denitrification apparatus, and (iii) downregulation of flagellar biosynthesis pathway. NAC-mediated inhibition of P. aeruginosa denitrification pathway and flagellum-mediated motility were confirmed experimentally. These findings suggested that NAC-colistin combinations might contribute to the management of biofilm-associated P. aeruginosa lung infections. NAC might also have a role in reducing P. aeruginosa virulence, which could be relevant in the very early stages of lung colonization. IMPORTANCE Pseudomonas aeruginosa biofilm-related chronic lung colonization contributes to cystic fibrosis (CF) disease progression. Colistin is often a last-resort antibiotic for the treatment of such P. aeruginosa infections, and it has been increasingly used in CF, especially by nebulization. N-acetylcysteine (NAC) is a mucolytic agent with antioxidant activity, commonly administered with antibiotics for the treatment of lower respiratory tract infections. Here, we show that NAC potentiated colistin activity against in vitro biofilms models of P. aeruginosa strains, with both drugs tested at the high concentrations achievable after nebulization. In addition, we report the first transcriptomic data on the P. aeruginosa response to NAC exposure.

P. aeruginosa biofilms already established in the lung and that the response to NAC was not related to phenotypic or genotypic features. The few previous studies that have addressed the activity of NAC against preformed P. aeruginosa biofilms have reported similar results (i.e., usually limited and strain-dependent effects), although a direct comparison of data is not straightforward due to different methodological approaches (e.g., different biofilm models and different NAC concentrations tested) and the low number of strains often tested in such studies (i.e., usually reference strains) (8,13,14). This study provided a wider picture on this topic by investigating a panel of characterized P. aeruginosa strains using a standardized in vitro biofilm model and in vivo achievable NAC concentrations. Interestingly, NAC alone (at the concentration used in this study and the same biofilm a CF, cystic fibrosis; RTI ICU , respiratory tract infection in intensive care unit. b According to the MLST Pasteur scheme. c CB r , resistance to carbapenems (imipenem and meropenem); CP r , resistance to cephems (ceftazidime and cefepime); FQ r , resistance to fluoroquinolones (ciprofloxacin); AG r , resistance to aminoglycosides (amikacin and gentamicin); CST r , resistance to colistin. d CST, colistin; NAC, N-acetylcysteine. model) was recently shown to exert relevant activity against preformed biofilms of two relevant CF pathogens, namely, S. maltophilia and Burkholderia cepacia complex (BCC) (7).
The reasons for such a diverse response of P. aeruginosa compared to S. maltophilia and BCC should deserve further attention, because they could possibly help identifying critical targets in the complex biofilm environments, to be used for the implementation of new antibiofilm strategies. Activity of NAC-colistin combinations against preformed biofilms. P. aeruginosa Z154 (a mucoid, MDR, colistin-susceptible CF strain) was first used to test the potential antibiofilm synergism of NAC at 8,000 mg/L plus diverse colistin concentrations. As shown in Fig. 1, a relevant synergism was observed already with colistin at 2 mg/L (i.e., the colistin MIC for the tested strain), with a dose-dependent effect at increasing colistin concentrations, and complete biofilm eradication was achieved with the combination of NAC at 8,000 mg/L plus colistin at 8 mg/L (Fig. 1).
The remaining 16 strains were initially tested with the combination of NAC at 8,000 mg/L plus colistin at 8 mg/L. In order to detect a potential synergism, the concentration of colistin was then modified for strains forming biofilms highly susceptible to colistin (n = 7) or particularly resistant (n = 2) ( Fig. 2 to 4). Overall, a relevant synergism of NAC-colistin combinations was observed with all tested strains (including the three colistin-resistant ones), although in two cases (i.e., P. aeruginosa M4 and M32), statistical significance was not achieved (Fig. 2 to 4). These latter strains were also tested with lower colistin concentrations (i.e., 2 and 4 mg/L, respectively), but synergism was not observed (data not shown). Concerning the synergism observed with the three colistin-resistant strains (Fig. 4), it is interesting to note that with strain FC237 (nonmucoid, MDR), an important decrease in viable biofilm cells was observed with a combination including a colistin concentration much lower than the colistin MIC for this strain (i.e., 1/64 MIC) (Fig. 4).
Overall, these data demonstrated that NAC could potentiate colistin activity against preformed biofilms of colistin-susceptible and colistin-resistant P. aeruginosa strains, regardless of the mucoid/nonmucoid phenotype, the resistance pattern, and the ST and O type. Present findings are consistent with the previously observed antibiofilm synergism of NAC-colistin combinations against colistin-susceptible and colistin-resistant strains of A. baumannii and S. maltophilia (9,10). Further studies with a higher number of P. aeruginosa clinical isolates, especially with a colistin-resistant phenotype, are encouraged.
Activity of NAC-colistin combinations in the ASM biofilm model. Two P. aeruginosa CF strains exhibiting different phenotypes were selected for susceptibility assays with the artificial sputum medium (ASM) biofilm model: P. aeruginosa Z34 (nonmucoid, MDR, ST17, O1) and P. aeruginosa Z154 (mucoid, MDR, ST412, O6). Biofilms were grown in ASM, in order to mimic the P. aeruginosa biofilm environmental conditions experienced in the CF mucus. Preformed biofilms were then challenged in the same medium with NAC-colistin combinations.
As shown in Fig. 5, a clear synergism of NAC at 8,000 mg/L in combination with colistin at 64 mg/L was observed with both strains (Fig. 5). Compared to the experiments performed with the Nunc-TSP lid system, the concentration of colistin that allowed observation of a synergism was much higher (i.e., 32Â the MIC), possibly due to colistin strong ionic interactions with ASM components (e.g., extracellular DNA and mucin) (15). Indeed, preliminary experiments carried out with lower colistin concentrations did not show either colistin antibiofilm activity or synergism with NAC (data not shown). In addition, the antibiofilm activity of NAC alone observed against P. aeruginosa Z154 in the Nunc-TSP lid system was not observed in the ASM model (Fig. 5), confirming that Overall, these data demonstrated that the antibiofilm synergism of NAC-colistin combinations against P. aeruginosa strains is preserved also under the environmental conditions mimicking the CF mucus, which is promising for clinical applications. Furthermore, the lower susceptibility to colistin of P. aeruginosa biofilms in the ASM model compared to biofilm susceptibility in standard media observed in this study is consistent with what was previously reported with P. aeruginosa (16).
Analysis of DEGs revealed that NAC mainly acted as Zn 21 chelator, inducing a strong Zn 21 starvation response. DEGs associated with such response were consistent with data reported in previous studies addressing zinc homeostasis in P. aeruginosa and other bacteria ( Table 2) (17)(18)(19)(20)(21)(22). In particular, 31 of the 46 upregulated DEGs belonged to the zur regulon and are known to be activated in response to Zn 21 starvation (Table 2) (17)(18)(19)(20)(21)(22). Such genes mainly included operons involved in zinc uptake (e.g., the PA4063-PA4064-PA4065-PA4066 operon, cntOLMI operon, and znuABC operon) and genes encoding zinc-independent paralogs of cellular proteins (i.e., type B 50S ribosomal proteins L31 and L36, RNA polymerasebinding protein DksA2, adn GTP-cyclohydrolase FolE2) ( Table 2) (17)(18)(19)(20)(21)(22)(23). Upregulated DEGs belonging to the zur regulon also included genes encoding an N-acetylmuramoyl-L-alanine amidase (AmiA, involved in splitting of septal peptidoglycan during cell division), a g-carbonic anhydrase (Cam, involved in reversible hydration of carbon dioxide and important for growth under low-CO 2 conditions), and three modulators of the membrane FtsH protease (i.e., HflC and HflK family modulators) ( Table 2). The membrane FtsH zinc-dependent protease is required for the expression of diverse unrelated phenotypes (e.g., swimming and twitching motility, biofilm formation, autolysis, production of secondary metabolites, maintenance of plasma membrane integrity by degrading misfolded proteins), and it has been recently demonstrated to represent an important virulence factor in P. aeruginosa clone C (23). HflC and HflK family modulators interact with FtsH at the level of the plasma membrane, usually with an inhibitory effect (23). The NAC-mediated effects on the phenotypes related to FtsH would deserve further attention.
The remaining 15 upregulated DEGs included genes encoding a recently described transcriptional regulator, PA2100 (also named MdrR2) (24), an AhpC-like alkyl hydroperoxide reductase (involved in protection from oxidative stress) (25), and proteins possibly involved in copper and iron uptake (Table 2).
MdrR2, together with MdrR1, has been demonstrated to repress the mexAB-oprM operon (independently from the MexR repressor), activate the EmrAB efflux pump, and indirectly inhibit biofilm formation (Table 2) (24). The effect of NAC on the MdrR1-MdrR2 dual-regulation system should be further investigated. Nonetheless, a previous study aimed at investigating the potential antagonism of high NAC concentrations (i.e., as those tested in this study) on the activity of the major classes of antibiotics used in the clinical practice, did not show major effects (with the exception of carbapenems, due to a chemical instability of carbapenems in the presence of NAC) (26), suggesting that the activation of the EmrAB efflux could not be relevant or circumvented by compensatory mechanisms.
Analysis of downregulated DEGs identified genes involved in denitrification, in particular norB (encoding the nitric oxide reductase subunit NorB), nosR (encoding the NAC of colistin antibiofilm activity was observed with all tested strains, although in two cases, statistical significance was not achieved (i.e., strains M4 and M32). CST 2, colistin at 2 mg/L; CST 4, colistin at 4 mg/L; CST 8, colistin at 8 mg/L. Biofilms not exposed to NAC or CST represent the control. Black lines indicate median values. The x axis is set at the limit of detection (20 CFU/peg). regulatory protein NosR), and nosZ (encoding the nitrous oxide reductase NosZ) ( Table 2). These data suggested that NAC might affect P. aeruginosa anaerobic respiration (which is crucial in the deeper biofilm layers and in the CF mucus) (27), because the nitric oxide reductase NorBC and the regulatory protein NosR have been recently demonstrated to constitute the nucleus of the denitrification protein network (28). NAC-mediated inhibition of the P. aeruginosa denitrification pathway might be implicated in the observed antibiofilm synergism of the NAC-colistin combination. Indeed, colistin has been demonstrated to exert increased antibiofilm activity against P. aeruginosa under anaerobic conditions, possibly due to a lower ability to implement the tolerance mechanism (e.g., lipopolysaccharide [LPS] modification) because of the low metabolism accompanying anaerobic growth (29). In this perspective, the inhibition of anaerobic respiration by NAC would further inhibit a P. aeruginosa adaptive response to colistin toxicity. This could be particularly relevant in P. aeruginosa biofilm in the CF mucus, where the anoxic conditions of biofilm cells are related not only to the position of the bacteria within the biofilm (i.e., anoxic conditions in the deeper layers), but also to the intense O 2 depletion caused by polymorphonuclear leukocytes (PMNs), determining entire biofilm growth without aerobic respiration (29).
Downregulated DEGs also included the following: (i) two genes involved in flagellar biosynthesis (i.e., fliF, encoding the flagellar M-ring protein FliF, and flhF, encoding the flagellar biosynthesis protein FlhF); (ii) a NAD(P)H-quinone oxidoreductase protecting against ROS-induced oxidative stress, which was recently demonstrated to be part of the core biofilm transcriptome (PA1137) (30); and (iii) nalD, encoding a second repressor of the mexAB-oprM operon (31). Finally, consistent with previous studies on Pseudomonas response to zinc starvation, downregulation of copA and copZ, involved in copper efflux, was observed, suggesting interplay between zinc and copper homeostasis (Table 2) (32).
NAC-mediated inhibition of P. aeruginosa denitrification pathway. The role of NAC in the inhibition of the denitrification pathway was confirmed by measuring NO 3 2 and NO 2 2 concentrations during anaerobic growth of the P. aeruginosa Z154 strain (i.e., the strain used for transcriptomic analysis) in culture media supplemented with 10 mM NaNO 3 or KNO 2 , in the presence or absence of NAC at 8,000 mg/L.
As expected from previous studies (33), in NaNO 3 -containing medium, the levels of NO 3 2 and its reduction product, NO 2 2 , fell below the detection limit after 24 h, in the absence of NAC (Fig. 6A). However, in the presence of NAC at 8,000 mg/L, the depletion of NO 3 2 was followed by an accumulation of NO 2 2 (evident at both 24 and 48 h), indicating that further reduction of NO 2 2 was inhibited in the presence of NAC (Fig. 6A). In order to consolidate these data, the experiments were repeated using a medium supplemented with KNO 2 . In the absence on NAC, complete reduction of NO 2 was observed after 48 h (Fig. 6B), as expected (33). On the contrary, in the presence of NAC at 8,000 mg/L, NO 2 levels did not decrease (Fig. 6B).
These results were consistent with the transcriptomic data and showed that NAC was able to inhibit the denitrification pathway in anaerobic environments, such as those encountered in endobronchial CF mucus. This feature might contribute to the observed antibiofilm synergism of NAC-colistin combinations, as previously discussed.
Time-kill assays of the NAC-colistin combination against planktonic cultures grown under anaerobic and aerobic conditions. Transcriptomic and biological data from this study suggested a role of NAC in inhibiting the P. aeruginosa denitrification apparatus, which could contribute to the observed antibiofilm synergy of NAC-colistin combinations. In order to further investigate this issue, time-kill assays of the NAC-colistin combination were performed with P. aeruginosa Z154 (i.e., the strain used for transcriptomic analysis) planktonic cultures, under both anaerobic and aerobic conditions.     . 7A and B). Interestingly, a clear bactericidal effect of colistin at 0.25 mg/L (i.e., 1/8 MIC) in combination with NAC at 8,000 mg/ L was observed in planktonic cultures grown under anaerobic conditions, with eradication achieved after 24 h of exposure (Fig. 7A). The wide error bars were due to the fact that in 2 out of 8 replicates (related to two independent experiments), no synergism was observed (Fig. 7A). This discrepancy was probably related to the low colistin concentration tested and the possible presence of heteroresistant subpopulations. On the contrary, cultures grown in the presence of oxygen were not affected by the NAC-colistin combination, demonstrating the influence of the growth conditions on the susceptibility of P. aeruginosa to such combination (Fig. 7B). These results supported the hypothesis that, under anoxic conditions like those present in the deeper biofilm layers and in CF mucus, NAC-mediated inhibition of anaerobic respiration would prevent an adaptive response of P. aeruginosa to protect from colistin toxicity.
NAC-mediated inhibition of P. aeruginosa swimming and swarming motility. Transcriptomic results indicated that NAC downregulated two genes belonging to P. aeruginosa flagellar apparatus (i.e., fliF and flhF), which are necessary for the first step of flagellum assembly (36). In order to confirm the potential NAC-induced inhibition of flagellum-mediated motility, we performed classical swimming and swarming tests with the reference strain P. aeruginosa PAO1 and the CF strain P. aeruginosa Z154 (i.e., the strain used for transcriptomic analysis). P. aeruginosa Z154 was not capable of swarming motility under our laboratory conditions, so only the effect of NAC on swimming motility could be tested with this strain. Overall, the results showed a clear inhibition of both swimming and swarming motility in the presence of NAC at 8,000 mg/L ( Fig. 8 and 9). Such inhibition could be related to the downregulation of crucial genes of the flagellar apparatus and/or the induction of a zinc starvation response. Indeed, zinc starvation has been demonstrated to affect the ability of P. aeruginosa to express several virulence phenotypes, crucial for the ability of this pathogen to colonize CF lung, including motility, biofilm formation and siderophore synthesis (37).
Conclusions. In conclusion, the results of this study demonstrated a relevant antibiofilm synergism of NAC-colistin combinations (at the high concentrations achievable by inhalation) against P. aeruginosa, which would deserve further investigation for potential clinical applications of inhaled formulations. Transcriptomic and biological experiments suggested that NAC inhibited P. aeruginosa anaerobic respiration, which could be relevant for the observed antibiofilm synergism with colistin. In addition, although NAC alone was not demonstrated to be effective against preformed P. aeruginosa biofilms, transcriptomic analysis of NAC-exposed planktonic cultures revealed that NAC could attenuate P. aeruginosa virulence, mainly by inducing a zinc starvation response, affecting anaerobic respiration and inhibiting flagellum-mediated motility (with the last two features confirmed experimentally). In this perspective, NAC, at the high concentrations achievable by inhalation, might have beneficial effects in the very first steps of lung infection, possibly preventing biofilm formation and the establishment of a chronic colonization, which should be further investigated.

MATERIALS AND METHODS
Bacterial strains. Seventeen strains were investigated, including 15 clinical isolates from CF patients, an MDR clinical isolate from a respiratory tract infection (RTI) from an intensive care unit (ICU), and the reference strain, P. aeruginosa PAO1 (Table 1). Identification was performed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS) (Bruker, Shimadzu). Antimicrobial susceptibility was determined using the reference broth microdilution method (38). Whole-genome sequencing of clinical isolates was performed with the Illumina (San Diego, CA, USA) MiSeq platform, using a 2Â 150-bp paired-end approach. Raw reads were assembled using SPAdes (39), and draft genomes were used to determine multilocus sequence types (MLSTs) and O types at the Oxford PubMLST site (https://pubmlst.org/) (40) and at the  Center for Genomic Epidemiology site (https://cge.food.dtu.dk/services/PAst/) (41), respectively. The complete genome of P. aeruginosa Z154 was obtained by combining results from Illumina with those obtained using the Oxford Nanopore Technologies (Oxford, United Kingdom) MinION platform, and de novo assembly was generated using Unicycler v0.4.4 as previously described (42).
Preparation of culture media. NAC stock solutions (100 g/L) were prepared immediately before use. NAC powder (Zambon, Bresso, Italy) was dissolved in sterile distilled water, the pH was adjusted to 6.5 to 6.8 with NaOH at 10 M, and the solution was filtered through a 0.22-mm-pore membrane filter. All experiments were performed in cation-adjusted Mueller-Hinton broth (CAMHB) (Becton Dickinson, Milan, Italy), unless otherwise specified, starting from an appropriately concentrated medium to avoid broth dilution when NAC solution was used. The artificial sputum medium (ASM) was also used in selected experiments and was prepared as previously described by Kirchner et al. (43).
In vitro biofilm susceptibility testing. Biofilm susceptibility testing was first performed using the Nunc-TSP lid system (Thermo Fisher Scientific, Waltham, MA, USA), as described previously (44). Briefly, biofilms were grown for 24 h in CAMHB at 35°C under static conditions. Preformed biofilms were then exposed to NAC at 8,000 mg/L and colistin (colistin sulfate; Applichem, Darmstadt, Germany) at 2 to 32 mg/L, alone and in combination. The colistin concentration was selected according to preliminary results of antibiofilm susceptibility testing and the colistin MIC for each strain. After 24 h of exposure (i.e., 35°C, static conditions), biofilms were washed twice with 200 mL of phosphate-buffered saline (PBS) (Sigma-Aldrich, Milan, Italy) to remove loosely adherent bacteria, and sessile cells were removed from pegs by sonication for 30 min (Elma Transsonic T 460; Elma, Singen, Germany) in 200 mL of tryptic soy broth (TSB) (Oxoid, Milan, Italy) supplemented with 1% Tween 20 (Sigma-Aldrich) (i.e., the recovery medium). The median number of CFU per peg was then determined by plating 10 mL of appropriate dilutions of the recovery medium onto tryptic soy agar (TSA) (Oxoid) and incubating for 24 h at 35°C (detection limit, 20 CFU/peg). The colony count was also double-checked after 48 h of incubation.
The potential antibiofilm synergism of NAC-colistin combinations was further investigated using an in vitro ASM biofilm model (43) in order to mimic P. aeruginosa biofilm conditions within the CF mucus.
The study was carried out with two selected CF strains (P. aeruginosa Z154 and Z34), exhibiting different features (i.e., mucoid/nonmucoid phenotype, antimicrobial susceptibility pattern, MLST, and O type) ( Table 1). In brief, biofilms were grown in 2 mL ASM in 24-well plates (Sarstedt, Nümbrecht, Germany), for 72 h at 35°C under static conditions. Preformed biofilms were then exposed to NAC at 8,000 mg/L and colistin at 64 mg/L, alone and in combination. Preliminary experiments carried out with lower colistin concentrations (i.e., 2 to 32 mg/L) did not show evident synergistic antibiofilm activity, while higher colistin concentrations (i.e., .64 mg/L) led to eradication of the biofilm cultures even in the absence of NAC (data not shown). After 24 h of exposure (i.e., 35°C, static conditions), bacterial biofilms were disrupted by 30 min of sonication followed by manual pipetting, and the median number of CFU per milliliter was determined following the same protocol described for the Nunc-TSP lid assay.
Data from both biofilm models were obtained in at least three independent experiments, with at least 12 replicates per condition per experiment.
RNA-seq and transcriptomic analysis. P. aeruginosa Z154 (i.e., colistin-susceptible CF strain, mucoid, MDR, ST412, O6) ( Table 1) was selected for studies aimed at investigating the transcriptomic response of P. aeruginosa to NAC exposure. A CF strain, rather than a reference strain (such as P. aeruginosa PAO1), was selected for this analysis because of the known adaptive diversification of P. aeruginosa into "specialized" types during chronic/recurrent infections in CF patients (3).
Because these represented the first data on the transcriptomic response of P. aeruginosa to NAC exposure, and considering the complex and still largely unknown effects of NAC on microbial physiology, we decided to perform the experiments with planktonic cultures, which represent a more homogenous and better standardized model for transcriptomic studies.
Overnight cultures in CAMHB were diluted at 1:50 in the same medium and incubated at 35°C with agitation to achieve an optical density at 600 nm (OD 600 ) of 1.0. The cells were then exposed to NAC at 8,000 mg/L for 30 min at 35°C under static conditions. Cultures treated in the same way but not exposed to NAC represented the control. Total RNA extraction was performed using the SV total RNA isolation system (Promega, Madison, WI, USA) following the manufacturer's instructions. rRNA depletion, cDNA library construction, and Illumina HiSeq 4000 platform-based transcriptome sequencing (RNA-seq) were performed by Eurofins Genomics Europe Sequencing (Constance, Germany). The transcriptome libraries were single-end sequenced with 50-bp reads for a total of 10 million reads per sample. Bioinformatic analysis was performed using the SeqMan NGen v17.3 software tool (DNASTAR Lasergene, Madison, WI, USA), with default parameters. Reads were aligned using P. aeruginosa Z154 complete genome (n = 6,344 coding DNA sequences [CDSs]) as a reference. Differentially expressed genes (DEGs) of the NAC-exposed cultures compared to the control were analyzed considering false-discovery rate (FDR) adjusted P values of ,0.05 from DeSeq2. DEGs with a 99% confidence interval (CI) were discussed. Results were obtained from two independent experiments. In order to favor comparison with data present in the literature, genes without a univocal name have been indicated as P. aeruginosa PAO1 locus tags throughout the text and reported in Table 2 also as P. aeruginosa UCBPP-PA14 locus tags.
NO 3 2 and NO 2 2 quantification. NAC-mediated inhibition of the denitrification pathway was investigated by measuring the concentration of NO 3 2 and NO 2 2 in anaerobic cultures of P. aeruginosa Z154 (i.e., the strain used for transcriptomic analysis). For this purpose, the Griess nitrite/nitrate colorimetric assay (Cayman Chemicals, Ann Arbor, MI, USA) was used according to the manufacturer's recommendations and as previously described, with some modification (33). CAMHB was supplemented with 10 mM NaNO 3 or KNO 2 and allowed to equilibrate for 3 days at 35°C in an anaerobic atmosphere by using the AnaeroGen kit (Oxoid). Overnight cultures were then diluted in 20 mL of each anoxic culture medium to reach a concentration of 10 6 CFU/mL and challenged with NAC at 8,000 mg/L. At times 0, 24, and 48 h of incubation under anoxic conditions at 35°C, supernatants were harvested and subjected to Griess colorimetric reaction in order to detect NO 3 2 and NO 2 2 levels. NAC-free cultures represented the control. Experiments were carried out in triplicate with one replicate per time point per condition.
Time-kill assays. Time-kill assays were performed according to CLSI guidelines (45) with the colistinsusceptible strain P. aeruginosa Z154 (i.e., the strain used for transcriptomic analysis). Colistin at 0.25 mg/L was tested alone and in combination with NAC at 8,000 mg/L under both aerobic and anaerobic conditions. We decided to use this colistin concentration since a higher concentration led to eradication of the planktonic cultures (data not shown). The medium (CAMHB) used to obtain anoxic cultures was placed under an anaerobic atmosphere by using the AnaeroGen kit (Oxoid) for 3 days prior to use and during the whole experiment. The killing curves were carried out in borosilicate glass bottles with a final volume of 20 mL of CAMHB. At 0, 2, 4, 8, and 24 h of exposure, CFU per milliliter were determined by plating 60 mL of appropriate dilutions of each condition onto TSA and incubating for 24 h at 35°C (detection limit, 17 CFU/mL). Data were obtained from at least four independent experiments with two replicates per condition per experiment.
Motility tests. NAC-induced inhibition of flagellum-mediated motility (i.e., both swimming and swarming motility) was investigated with the reference strain P. aeruginosa PAO1, which has been used for similar motility experiments in several previous studies (46), and P. aeruginosa Z154 (i.e., the strain used for transcriptomic analysis). P. aeruginosa Z154 was not capable of swarming motility under our laboratory conditions (perhaps due to the known reduction of flagellar expression in mucoid CF-adapted strains) (47), so only the effect of NAC on swimming motility could be tested with this strain. Swim plates consisted of Luria-Bertani (LB) broth (Oxoid) containing 0.3% agar (46). Swarm plates consisted of nutrient broth (Oxoid) with 0.5% glucose and 0.5% agar (46). Overnight cultures in CAMHB were diluted in the same medium to a final OD 600 of 3.0, and 5 mL was spotted onto swim and swarm plates, with or without NAC at 8,000 mg/L. Results were observed after incubation at 35°C for 48 h. Assays were performed in at least three independent experiments with three replicates per condition per experiment.
Statistical analysis. Statistical analysis of biofilm susceptibility assays was performed using GraphPad Prism version 8.0 (San Diego, CA, USA). Multiple-comparison tests were performed by the Kruskal-Wallis test with Dunn's correction. A P value of #0.05 was considered significant. RNA-seq statistical analysis was performed using the SeqMan NGen v17.3 software tool.
Data availability. The complete genome sequence of P. aeruginosa Z154 was deposited in GenBank under accession no. CP069177. RNA-seq data were also deposited in the NCBI Gene Expression Omnibus (GEO) database under accession no. GSE190946.