Bacterial strains, plasmids and planktonic growth conditions

All bacteria and plasmids used in this study are listed in Table 1. Bacteria were grown in either Luria-Bertani broth (LB) (tryptone, 10 g; yeast extract, 5 g; NaCl, 5 g for one liter of broth) or LB supplemented with 50 mM KNO₃ (LBN). When specified, LB or LBN were buffered with 50 mM potassium phosphate to achieve acidic pH 6.5. Aerobic cultures were grown at 37°C with shaking at 200 rpm. Anaerobic cultures were grown in a duel-port, Coy Systems anaerobic chamber at 37°C in an atmosphere of 85% N₂-10% CO₂-5% H₂. Media were solidified with 1.5% Bacto-agar. Frozen bacterial stocks were stored at −80°C in a 1∶1 mixture of 30% glycerol and aerobically grown, stationary phase bacteria.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Bacterial strains, plasmids and oligonucleotides used in this study.

https://doi.org/10.1371/journal.pone.0091813.t001

DNA manipulations

Genomic DNA isolation, PCR, restriction enzyme digestion, ligation, cloning and DNA electrophoresis were done according to standard techniques [43]. All oligonucleotide primers were synthesized by Integrated DNA Technologies (IDT). PCR was performed using either Choice Taq Mastermix (Denville Scientific, Inc. USA) or Pfu DNA polymerase (Strategene, USA). Plasmid isolation was performed using QIAprep Spin miniprep kits (QIAGEN) as recommended by the manufacturer. DNA fragments were purified using either a QIAquick PCR purification kit (QIAGEN) or a QIAquick gel extraction kit (QIAGEN). All cloned inserts were confirmed by automated DNA sequencing performed at the DNA Core Facility of the Cincinnati Children’s Hospital Medical Center. Plasmids were introduced into E. coli by CaCl₂-mediated transformation and into P. aeruginosa by either electroporation or S17-1-mediated conjugation [44].

Construction of P. aeruginosa mutants

The strategy for insertional inactivation of some of the genes listed in Table 1 was facilitated by gene disruption with an 850-bp gentamicin resistance (Gm^R) cassette from pUCGM [45], and the gene replacement vector pEX100T [46], the latter of which allowed for selection of double-crossover events within putative recombinants cultured on agar containing 5% sucrose. All mutants were confirmed by PCR analysis.

DNA cloning and construction of recombinant plasmids and strains

A 300-bp promoter region of the katA gene of P. aeruginosa PAO1 was amplified by PCR with primer pair pkatA/XhSal5′ and pkatA/Bm3′. The amplified promoter pkatA was digested with SalI and BamHI, and cloned within the SalI and BamHI sites of pQF50, creating pkatA-lacZ transcriptional reporter plasmid pQF50pkatA. A 1.7 kb fragment containing katA gene and its upstream promoter region was amplified by PCR with primers pkatA/Bam5′ and katA/Eco3′, restricted with BamHI and EcoR1, and cloned into the vector mini-CTX1[47]. The resultant plasmid mini-CTX1-katA was then used to introduce a wild-type copy of katA at the chromosomal attB1 site of ΔkatA mutant, generating the complementation strain ΔkatA::katA. Open reading frames (ORF) of katA, nirS, and anr were amplified by PCR with primers katA/Eco5′ and katA/Hind3′, nirS/Nco5′ and nirS/Hind3′, and anr/Kpn5′ and anr/Hind3′, and cloned into vector pHERD20T [48], yielding pHERDkatA, pHERDnirS, and pHERDanr, respectively.

Enzymatic assays and Western blot analysis

Catalase activity was measured spectrophotometrically by following the decomposition of 19.5 mM H₂O₂ in potassium phosphate buffer, pH 7.0 at 240 nm [10], [49]. One unit of activity was defined as that which decomposed 1 μmol of H₂O₂ min⁻¹ mg⁻¹ protein. Overnight bacterial cultures (1.5 ml) under either aerobic or anaerobic conditions were harvested by centrifugation. Cell pellet was then suspended in 100 μl of PBS and disrupted by sonication. The suspension was clarified by centrifugation and total soluble proteins were used for catalase activity assay. β-galactosidase assays were performed as described by Miller [50] with slight modifications. Briefly, a 400 μl of bacterial culture at an OD₆₀₀ of 1.0 was transferred to a tube containing 600 μl of Z-buffer, 20 μl of 0.1% SDS and 40 μl of chloroform, vortexed for 10 s, and incubated for 5 min at 28°C. An amount of 200 μl of orthonitrophenyl galactoside (4 mg/ml in 0.1 M phosphate buffer, pH7.0) was added and the reaction was allowed to proceed until yellow color developed. The reaction was stopped by the addition of 500 μl of 1 M Na₂CO₃. Samples were centrifuged to remove cell debris and chloroform, and the supernatant was measured at 420 nm. Aconitase was assayed by the production of cis-aconitate (3.6 mM⁻¹cm⁻¹at 240 nm) as described previously [51]. Protein concentration was estimated by the method of Bradford (4) using bovine serum albumin fraction V (Sigma) as standard. Cell extracts were subjected to 10% SDS-PAGE at 100 V for 1.5 hr. Separated proteins were then examined with coomassie blue staining or electro-blotted to PVDF membranes (Amersham, USA) using the Trans-Blot system (Bio-Rad, USA) at 120 mA for 2 hr. Membranes were washed and blocked in 5% dried skimmed milk in PBS-Tris-HCl buffer, pH 7.5. Then, the membranes were incubated with primary antibody (10⁻⁴ dilution) against KatA of Streptomyces coelicolor, known to cross-react with PA KatA [11] for 1 hr. Excess antibody was removed by three washings with PBS-T. After incubation for 1 hr in PBS-T containing the secondary antibody (10⁻³ dilution of peroxidase-conjugated, goat anti-mouse IgG), the membrane was washed three times with PBS-T. Immunodetection was performed using the ECL select Western blotting detection system according to the manufacturer’s instruction (Amersham, USA). The exposed X-ray film was recorded by scanning, and the relative density of reactive catalase bands was analyzed using the NIH ImageJ 1.47 g software.

Acidified NO₂⁻ sensitivity measurements

(i) Overnight cultures of P. aeruginosa wild-type PAO1, ΔkatA mutant, ΔkatA::katA, FRD1, and FRD1ΔkatA were 1∶100 diluted into either LB broth (pH 6.5) or LBN broth (pH6.5) supplemented with varying concentrations of NaNO₂ (0, 5, 10, 15, 20, 25 and 30 mM) and grown aerobically for 24 hrs, or anaerobically for 48 hrs. 5 μl of cells from each culture were spotted onto LB agar plates and incubated aerobically for 24 hrs at 37°C. The plates were then scanned for figure presentation. (ii) All strains above were also cultured anaerobically for 72 hrs in LBN broth (pH 6.5) supplemented with either 0, 20, and 25 mM NaNO₂ for PAO1, ΔkatA mutant and ΔkatA::katA, or 0, and 5 mM NaNO₂ for FRD1 and FRD1ΔkatA, respectively. Cultures were taken daily, and serial cell dilutions were spotted onto LB broth agar plates. Surviving bacteria were enumerated after a 24 hr incubation at 37°C.

Sensitivity of anaerobic biofilm to acidified NO₂⁻

P. aeruginosa wild-type PAO1, DkatA mutant, and DkatA::katA strains were grown aerobically in LB broth to stationary phase followed by a 1∶100 dilution into 3 ml of LBN in confocal “friendly” glass bottomed chambers. Bacterial biofilms were allowed to develop under anaerobic condition as previously described (Yoon et al., [28]. After 24 hrs, biofilms were washed with sterile PBS to remove planktonic cells, and fresh LB broth (pH 6.5) containing 15 mM NO₃ (control), or 15 mM NO₃ plus 15 mM NO₂, or 15 mM NO₃ plus 15 mM NO₂ and 10 mM c-PTIO, was added to the cultures. The biofilms were then incubated under anaerobic condition for an additional 48 hrs, washed 2 times with PBS and biofilm images were viewed by confocal scanning laser microscopy using an LSM 710 confocal microscope (Zeiss, Heidelberg, Germany) and visualized using a live (green cells)/dead (red cells) BacLight stain (Invitrogen, Eugene, OR). The excitation and emission wavelengths for green fluorescence (live cell) were 488 nm and 500 nm, while those for red fluorescence (dead cell) were 490 nm and 635 nm, respectively. All biofilm experiments were repeated at least 3 times independently. The dead/live ratios of the biofilms were calculated using imageJ 1.46r software following the guidelines by Christine Labno of the University of Chicago Integrated Light Microscopy Core. Briefly, the confocal image was imported into ImageJ, and the green (live cells) and red (dead cells) were separated resulting in only one color in each image. The color image was converted into a grey scale image. Then, adjust the threshold to highlight all of structures. At the end, the binary of the interested image was created. Finally, this image was used to analyze particles. Repeat the same cycle for the all other images. The percentage of dead/live ratio between the treated (NO₃⁻ plus NO₂⁻) and control (NO₃⁻ only) was calculated and normalized.

Effect of NO and KatA on anaerobic to aerobic growth transition

Wild-type strain P. aeruginosa PAO1 and its allelic mutants, ΔkatA, ΔnirS, ΔnorCB, ΔkatAΔnirS and ΔkatAΔnorCB, were grown in LBN broth anaerobically for 24 hrs. The cultures were diluted in 10 ml of fresh LBN broth in a 250 ml flask and normalized to an OD₆₀₀∼0.009. The cultures were then incubated aerobically with vigorous shaking and the cell density (as OD_{600 nm}) recorded hourly.

KatA overproduction and protein purification

The KatA-overexpressing bacteria BL21(DE3) (pLysS, pJFM18) were grown aerobically at 20°C to the OD₆₀₀ = 1.0 in LB broth supplemented with 1 mM FeCl₃. The culture was then treated with 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) and 2% ethanol for 10 hrs at 20°C. Bacteria were harvested by centrifugation at 10,000 × g for 10 min at 4°C, washed once with 0.9% saline, and the pellet was resuspended in a buffer containing 20 mM Tris-HCl-250 mM NaCl, pH 8.0. The bacterial cells were lysed by three passages through a French Pressure cell at 1,000 lbs/sq. in. The cell lysates were cleared by centrifugation at 13,800 × g for 30 min at 4°C and by filtration through 0.45 μm filter. Six-His-tagged KatA was first purified using His•Bind purification kit (Novagen, USA) following the manufacturer’s instruction, and then further purified using a GE Superdex 200 26/60 size exclusion column in 20 mM HEPES pH 7.0, 150 mM NaCl. The fractions representing tetrameric KatA were combined and concentrated to 5.5 mg/ml for crystallographic studies. The purity of recombinant KatA was assessed by 10% SDS-PAGE and staining with coomassie blue.

Optical spectroscopy

Optical spectra were recorded on a GBC model 920 spectrophotometer (GBC, Australia). KatA was made anaerobic through repeated evacuation and exchange with argon on a Schlenk line. The concentrated enzyme was resuspended to ∼5 μM in a sealed cuvette with anaerobic buffer containing 50 mM Tris-HCl pH 8.0. Protein-NO bound complexes were formed by flowing NO gas over the sample for 1 min.

Determination of the equilibrium binding constant (K_d) of KatA for NO via spectrophometrically monitored ligand titration

Ferric KatA (as isolated) in 50 mM Tris buffer, pH 8.0, 250 mM NaCl was de-oxygenated under an argon atmosphere. The enzyme was transferred to a screw-cap style cuvette with a silicon septa lid. A saturated nitric oxide solution (1.9 mM, 23°C) was prepared by bubbling 6 M NaOH-scrubbed NO gas through Tris buffer. Each addition of the nitric oxide solution to the enzyme solution in the cuvette was made in an anaerobic glove bag. Optical spectra were recorded immediately after each addition on a Cary 50 Bio. NO was added until no further optical changes were observed. Absorbance increases at 574 nm were used to determine the fraction of NO-bound KatA. This value was plotted versus the amount of free NO (total NO added-KatA NO-bound fraction) and fit to a hyperbolic function to obtain the apparent K_d (Origin 8.6, Origin Lab, MA).

Electron paramagnetic resonance (EPR) spectroscopy

KatA crystallization and structure determination

Purified KatA was screened for crystallization conditions using an Art Robbins Phoenix crystallization robot. A single crystallization condition was obtained from the Hampton Research Index screen, which was optimized to the final conditions: 0.1 M sodium acetate, pH 4.3, 16% PEG 3350. Crystals of the approximate dimensions 50×50×100 μm were harvested and frozen in liquid nitrogen. Synchrotron data were collected at the Advanced Photon Source GMCA beamline 23ID, and a full dataset was obtained to 2.55Å. The structure of Proteus mirabilis catalase (Protein Data Bank accession 1M85) was chosen as a molecular replacement model due its high sequence conservation with P. aeruginosa KatA. Phaser identified four molecules in the asymmetric unit. The model was refined and manually rebuilt using PHENIX and COOT, respectively. Data collection and refinement statistics are shown in Table 2. The final structure was validated using MolProbity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. KatA structure determination statistics.

https://doi.org/10.1371/journal.pone.0091813.t002

Detection of NO in bacteria

NO production by bacterial cells was measured using a fluorescence probe, 5,6 diaminofluorescein diacetate (DAF-2 DA) (Sigma-Aldrich, USA). Overnight aerobic cultures in LB broth were 1∶100 diluted into LBN in the presence of difference concentration of arabinose (0–0.1%), and grown anaerobically at 37°C for 24 hrs. One hundred microliter of bacterial cultures normalized to OD₆₀₀ = 0.1 were incubated with 10 μM of DAF-2 DA in PBS for 30 min at 37°C. The fluorescent reaction product of DAF-2 and NO called DAF-2 T (excitation λ_max  =  495 nm, emission λ_max  =  515 nm) was measured with FL 600 microplate florescence reader, BIO-TEX. The relative NO level as measured by this fluorescent dye was calculated based on a calibration curve, which was generated by using varied concentration of MAHMA-NONOate, the NO donor and fluorescence signal output. The amount of DAF-2 T detected from each sample was then converted to a level of NO.

PA catalase activity is increased under anaerobic conditions, specifically due to KatA expression, and not KatB or KatC

We [15], [19], [52] previously showed that PA generates greater catalase activity during anaerobic versus aerobic respiration yet the molecular basis for this event and the anaerobic function of catalase remained unclear. Increased anaerobic catalase activity was found attributable to KatA since a ΔkatA mutant possessed no catalase activity, while deletion of either KatB or KatC had no effect (Fig. 1A). To elucidate the mechanism underlying the higher anaerobic KatA activity of wild-type strain PAO1, we first assessed the transcription of katA under aerobic and anaerobic conditions. A pkatA-lacZ transcriptional fusion reporter plasmid, pQFpkatA, was constructed and transformed into PAO1 via electroporation. β-galactosidase assays showed that anaerobic PAO1 harboring plasmid pQFpkatA exhibited approximately 2.5-fold higher lacZ reporter activity than aerobic bacteria (Fig. 1B), indicating that transcription of katA is increased anaerobically, similar to catalase activity. Next, we examined the protein level of KatA from total soluble cell lysates of aerobic and anaerobic cultures. Both Coomassie blue staining (Fig. 1C) and Western blotting (Fig. 1D) clearly demonstrated that anaerobic cells produced significant higher KatA enzyme (∼2.5% of total soluble protein).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Analysis of catalase under aerobic and anaerobic conditions.

A. Catalase activity of aerobic (white bars) vs. anaerobic (black bars) cell extracts of P. aeruginosa PAO1 and selected isogenic mutant bacteria. Bacteria were grown for 24 hrs in LBN broth under aerobic (+O₂) or anaerobic (–O₂) conditions. Cell extracts were assayed for catalase activity. The experiment was performed three times in triplicate. Standard deviation (SD) was shown. B. Analysis of katA-lacZ transcriptional activity under aerobic (white bars) vs. anaerobic (black bars) conditions in P. aeruginosa PAO1, ΔnirS and ΔnorCB mutant bacteria. Bacteria harboring pQFpkatA were grown for 24 hrs in LBN broth under aerobic (+O₂) or anaerobic (–O₂) conditions. Cell cultures were then assayed for β-galactosidase activity. The assay was repeated three times. SD was shown. C. SDS-PAGE and Coomassie blue staining of aerobic and anaerobic cell extracts of PAO1. Varied concentration of total soluble proteins (5, 10, 20, 30 and 40 μg) of either aerobic or anaerobic cell extracts of PAO1 were subjected to 10% SDS-PAGE electrophoresis and followed by Coomassie blue staining. Purified recombinant KatA (0.5 μg) was electrophoresed in parallel to indicate the migrating position of KatA on the gel. D. Western blot of aerobic vs. anaerobic cell extracts using anti-KatA antibody. The far right lane contains 1 μg of purified, recombinant KatA (black arrows). The number above each lane is the μg of cell extract loaded from aerobically or anaerobically grown bacteria. See Material and Methods for details.

https://doi.org/10.1371/journal.pone.0091813.g001

Role of NO in KatA activity and katA gene expression

In order to address the possible role of metabolic NO in the observed increase in KatA activity of anaerobic PAO1, we examined catalase and pkatA-lacZ reporter activities in an isogenic NIR mutant (ΔnirS) [42], a strain that is incapable of enzymatically-catalyzed NO production, under aerobic and anaerobic conditions. When compared to PAO1, the ΔnirS mutant yielded wild-type levels of aerobic catalase activity. However, when cultured anaerobically, the mutant produced much lower catalase activity, approximately half of that expressed aerobically and less than that of wild-type PAO1 (Fig. 1A), indicating that NO plays a role in the elevated anaerobic KatA activity of the wild-type PAO1. We also measured catalase and pkatA-lacZ reporter activity in an isogenic ΔnorCB mutant, a strain that is deficient in NO reductase and accumulates higher levels of intracellular NO under anaerobic conditions [42]. As expected, the ΔnorCB mutant produced wild-type levels of aerobic catalase activity. However, when cultured anaerobically, the mutant yielded very little catalase activity (Fig. 1A), which was purportedly inconsistent with the notion that higher KatA activity would be expected because of higher NO production in the ΔnorCB mutant. Because this mutant is devoid of the ability to cope with such high levels of NO, this likely leads to the malfunctioning and/or paralysis of ANR (NO-sensitive)-mediated anaerobic respiratory machinery. This theory was corroborated by very poor anaerobic growth of the ΔnorCB mutant and inactivation of recombinant ANR [42]. The reduced anaerobic catalase activity in both the ΔnirS and especially in the ΔnorCB mutant was also reflected at the transcriptional level where we observed higher katA-lacZ activity in wild-type anaerobic cultures, little, if any, increase in reporter activity in the ΔnirS mutant, and extremely low transcription of the katA gene in ΔnorCB mutant bacteria (Fig. 1A and 1B).

Next, we elected to examine the effect of NO production on catalase activity and cell growth. We used a ΔnirS mutant harboring the plasmid, pHERDnirS, in which expression of nirS is inducible by arabinose. The low levels of NO detected in the ΔnirS mutant are likely a function of the reduction of NO₂⁻ when NO₃⁻ is not being utilized. When compared to the ΔnirS mutant, the ΔnirS(pHERDnirS) strain yielded ∼2.5 fold higher levels of NO (Fig. 2A) and catalase activity (Fig. 2B) and 5-fold greater cell growth (Fig. 2C), even in the absence of arabinose, indicating a leaky expression of nirS via the P_BAD promoter in PA. NO levels were increased after the addition of 0.005% arabinose (Fig. 2A), while no increase was observed in catalase activity and cell growth (Fig. 2B,C). However, the addition of 0.05% arabinose actually impaired cell growth (Fig. 2C) and also reduced catalase activity (Fig. 2B), likely due to harmful NO levels generated compared to the control (ΔnirS mutant bacteria). When arabinose concentrations were raised to 0.1%, NO production, catalase activity and cell growth of ΔnirS(pHERDnirS) dropped to levels similar to that of the ΔnirS mutant (Fig. 2A,B,C). Exposure to an NO donor, sodium nitroprusside (SNP, 10–50 μM), to purified KatA in vitro did not significantly increase or decrease KatA activity (data not shown), in contrast with the results observed by Gusarov and Knudler using purified Bacillus subtilis catalase [53]. However, when SNP was added to live anaerobic ΔnirS mutant bacteria, catalase activity dramatically increased when exposed to 10 μM SNP (Fig. 2D), However, catalase activity decreased dramatically in ΔnirS bacteria when (i) nirS expression was elevated (Fig. 2B) and (ii) when such organisms were exposed to higher SNP concentrations than 10 μM (Fig. 2D), indicating that KatA expression was not sufficient enough to counter the amount of NO generated, and that excess NO was toxic to the bacteria, consistent with our previous results (Yoon et al., [42]).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 2. Effect of NO production on cell growth and catalase activity in ΔnirS mutant, the wild-type PAO1 and its isogenic mutants.

P. aeruginosa ΔnirS mutant and ΔnirS(pHERDnirS) were cultured anaerobically in LBN for 24 hrs. Different concentration of arabinose (0–0.1%) were added to the cultures of strain ΔnirS(pHERDnirS) to induce expression of the nirS gene. Cells were sampled to measure NO production (A), catalase activity (B) and cell growth (OD₆₀₀) (C). NO levels were determined using the fluorescent indicator compound, DAF-2A. Overnight anaerobic cultures were normalized and incubated with 10 μM of DAF-2A for 30 min at 37°C. Catalase activity was measured spectrophotometrically by following the decomposition of 19 mM H₂O₂. See Materials and Methods for detail. All results were presented as a value of fold changed compared to the ΔnirS mutant that was assigned a value of 1. The black, light gray and white bars represented PAO1, ΔkatA mutant and ΔnirS mutant, respectively. The gray, dotted, diagonal striped and vertical striped bars were referred to ΔnirS(pHERDnirS) exposed to 0, 0.005, 0.05 and 0.1% arabinose, respectively. D. Catalase activity of the ΔnirS mutant upon SNP treatment. Exponential phase aerobic culture of ΔnirS in LBN was sub-cultured anaerobically with varying concentration of the NO donor SNP for 24 hrs. Cells were harvested, disrupted by sonication and the cleared cell lysates were used to measure catalase activity. Representative data are shown. E. Aconitase activity assay. Cells of anaerobically-grown PAO1, ΔkatA, ΔnirS and ΔnorCB. The assay was performed three times as described previously [51], and the standard deviation was shown.

https://doi.org/10.1371/journal.pone.0091813.g002

One enzyme that is critical for normal cellular metabolism and very sensitive to NO is the [4Fe-4S]²⁺ cluster TCA cycle enzyme, aconitase [54]. We postulated that the reduction in KatA activity and cell growth upon exposure to elevated NO, as well as the significantly reduced (43-fold) anaerobic KatA activity in the ΔnorCB mutant relative to wild-type cells (Fig. 1A) were attributed to the harmful level of NO reached by the cells and failure to detoxify NO. If this is the case, a reduction or loss of aconitase activity would be demonstrated in those cells. As expected, the ΔnorCB mutant, that produces nearly 13.6 μM NO anaerobically, [42]) possessed no detectable aconitase activity (Fig. 2E). The ΔnirS mutant, that produced no enzymatically-driven NO, produced 16-fold higher aconitase activity than wild-type bacteria (Fig. 2E). Based on our previous studies [28], where ΔrhlR mutant bacteria were found to overproduce NO, they also possessed reduced aconitase activity. We hypothesize that NO-mediated inactivation of proteins such as ANR and aconitase, especially in the case of the ΔnorCB mutant, is likely to be a result of NO-mediated destruction or modification (e.g., [2Fe-S] of the [4Fe-4S]²⁺ cluster of these proteins [42]). Such an event likely represents a last ditch, “circuit breaker” effort for the anaerobic ΔnorCB mutant that is under considerable endogenous NO stress (13.6 μM) to retain cell viability, and indicative of abysmally slow growth patterns.

Role of ANR in katA transcription and KatA activity

Trunk et al., [16] reported that the putative promoter region of katA has a weakly conserved ANR “box”, representing bases TTGTCTTCCGCCAA located 73-bp upstream of the translational start codon of katA. A slight Anr dependence of katA was detected in their transcriptome analysis and an Anr-dependent production of KatA was confirmed in their proteomic analysis under aerobic to anaerobic growth conditions [16]. To test that ANR participates in the regulation of the katA gene during anaerobic conditions, we first introduced plasmid pHERDanr into an Δanr mutant that is incapable of anaerobic growth, in which the expression of ANR is under the control of the arabinose-inducible P_BAD promoter. The resultant strain, Δanr(pHERDanr), was cultured anaerobically for 24 hrs in the presence of increasing arabinose levels, and both cellular growth (OD_{600 nm}) and catalase activity were determined. As shown in Fig. 3, as little as 0.005% arabinose was sufficient and optimal to induce expression of catalase in the complemented Δanr mutants via the induction of ANR expression. Increasing cellular levels of ANR beyond that point, however, decreased catalase activity and cell growth (Fig. 3). This is likely due to the high iron and sulfur demands necessary for optimal ANR core Fe-S cluster formation and function, and also limiting these critical elements to serve other Fe-S cluster (estimated at >200 in E. coli, [32]) and other cellular proteins requiring these elements for normal function. Similar to the pHERDnirS plasmid (Fig. 2A-C), the expression plasmid pHERDanr was also leaky, manifested by cultures reaching their highest cell densities in the absence of arabinose. This is in marked contrast to the exceedingly tight arabinose control of the P_BAD system in E. coli [55].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 3. ANR upregulates the expression of the katA gene.

The Δanr mutant harboring plasmid pHERDanr, in which the expression of ANR is under the arabinose-inducible P_BAD promoter, was cultured anaerobically for 24 hrs in LBN supplemented with increasing concentration of arabinose up to 1%. Cultures were sampled for optical cell density (OD₆₀₀) measurements (dotted line) and catalase activity (white bar). The experiment was performed three times independently and the standard deviation error bars are shown.

https://doi.org/10.1371/journal.pone.0091813.g003

KatA helps protect planktonic PA against the potential therapeutic agent and NO donor acidified nitrite (A-NO₂⁻)

Our underlying hypothesis was that KatA plays a role in protection of anaerobic bacteria against metabolically generated NO, but especially against NO generated from outside the cell. An example of this is that during human infection, inducible NO synthases (iNOS) that generate NO contribute to the antimicrobial activity of macrophages and neutrophils [56]. Thus, we elected to use an NO donor, A-NO₂⁻, that we have previously shown kills the antibiotic-resistant mucoid mucA22 form of PA [40]. Both aerobically- and anaerobically-grown PAO1 ΔkatA mutant bacteria exhibited greater sensitivity to increasing concentrations of A-NO₂⁻ than wild-type bacteria or an in cis complemented ΔkatA mutant (Fig. 4A). Under both aerobic and anaerobic conditions, all A-NO₂⁻-treated wild-type bacteria showed cell growth on plates, though decreased growth of anaerobic cells was observed when exposed to higher concentrations of A-NO₂⁻ (Fig. 4A, right panel). In contrast, the ΔkatA mutant bacteria exhibited varied sensitivities to A-NO₂⁻, as manifested by poor or no growth on LB plates at specified exposure times and test conditions (i) 20 mM A-NO₂⁻ or above in LB + O₂; (ii) 25 mM A-NO₂⁻ or above in LBN + O₂; (iii) 15 mM A-NO₂ or above in LB –O₂; and (iv) 25 mM A-NO₂⁻ or above in LBN –O₂. As expected, the complemented ΔkatA mutant, ΔkatA::katA, with wild-type katA in cis at the chromosomal attB site were not susceptible to A-NO₂⁻, similar to that of wild-type PA. As a control, wild-type PAO1, its ΔkatA mutant, and the complemented mutant, ΔkatA::katA, exhibited similar anaerobic cell growth in the absence of A-NO₂⁻ where bacteria could grow only via NO₃⁻ respiration (Fig. 4B). Since a single administration of A-NO₂⁻ may not kill the entire ΔkatA mutant population due to the presence of NIR activity, we performed viability (CFU) assays with bacteria exposed to 20 or 25 mM A-NO₂⁻. In both cases that the ΔkatA mutant was more sensitive to A-NO₂⁻ than PAO1 and complemented ΔkatA mutant organisms (Fig. 4C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 4. Sensitivity of the ΔkatA mutant to A-NO₂⁻.

A. Wild-type strain P. aeruginosa PAO1, ΔkatA mutant, and the complemented mutant ΔkatA::katA were cultured aerobically in either LB broth (pH 6.5) or LBN broth (pH 6.5) for 24 hrs (left panel), or anaerobically in either LB broth (pH 6.5) or LBN broth (pH 6.5) for 48 hrs (right panel) supplemented with 0, 15, 20, 25 and 30 mM NaNO₂, respectively. Cells were plated out directly on LB agar medium to determine the sensitivity to A-NO₂⁻. B. Anaerobic cultures in LBN (pH 6.5) without A-NO₂⁻ were serially diluted and spotted on LB agar plates to demonstrate the similar growth occurred in all test strains. C. Wild-type PAO1, ΔkatA mutant, and the complemented mutant ΔkatA::katA were cultured anaerobically for 72 hrs in LBN broth (pH 6.5) supplemented with 0, 20 mM, and 25 mM NaNO₂, respectively. Culture aliquots were removed daily, and serial cell dilutions were plated out on LB agar medium to enumerate CFU. The experiments were repeated three times and the representative data were shown.

https://doi.org/10.1371/journal.pone.0091813.g004

KatA is important for protection of A-NO₂⁻ sensitive mucoid, mucA22 bacteria

One hallmark of airway diseases such as COPD and CF is the frequent isolation of mucoid strains that have mutations within the mucA gene (notably mucA22) [57], [58] encoding a membrane-bound anti-sigma factor and the mutant was demonstrated to be sensitive to A-NO₂⁻ [40]. Strain FRD1, a mucA22 mutant coupled with a secondary katA mutation, renders such bacteria more sensitive to A-NO₂⁻ than strain FRD1 alone (Fig. 5A). Serially diluted control cells demonstrated identical CFU under anaerobic conditions without A-NO₂⁻ (Fig. 5B). Notably, FRD1 ΔkatA was sensitive to as little as 5 mM A-NO₂⁻ relative to strain FRD1 alone (Fig. 5C). This is in marked contrast to the higher levels (20-25 mM) of A-NO₂⁻ required to kill nonmucoid PAO1-based strains (Fig. 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 5. Sensitivity of the ΔkatA mutant of mucoid mucA22 strain FRD1 to A-NO₂⁻.

A. Undiluted samples of strains FRD1 and FRD1ΔkatA that were subjected to various concentrations of anaerobic NO₂⁻ treatment for 48 hrs were spotted on to LB agar plates. Top: samples anaerobically grown in LB, pH 6.5. Bottom: samples anaerobically grown in LBN, pH 6.5. B Anaerobic cultures in LBN (pH 6.5) without A-NO₂⁻ were serially diluted and spotted on LB agar plates to demonstrate the similar growth occurred in test strains. C. Quantification of the bacterial load for each strain as measured by CFU/ml. Aliquots of anaerobic cell suspensions were removed from strains grown in LBN, pH 6.5 with the indicated concentration of A-NO₂⁻. Serial dilutions were performed and the number of colonies were recorded after plating the dilutions on LB agar.

https://doi.org/10.1371/journal.pone.0091813.g005

KatA also helps protect anaerobic PA in biofilms upon exposure to A-NO₂⁻

To test our hypothesis that KatA also plays a role in protection of anaerobic biofilm bacteria against A-NO₂⁻, we treated mature anaerobic biofilm cells of PAO1 with A-NO₂⁻. Strain PAO1 forms robust anaerobic biofilms compared to strain FRD1 [59] that has an LPS-rough phenotype [60] and lacks flagella [61], an appendage critical for biofilm formation [62]. The ΔkatA mutant and the complemented mutant, ΔkatA::katA was exposed to 15 mM A-NO₂⁻ for 2 days, and viability assessed in such biofilms by live/dead staining of biofilm cells using confocal laser scanning microscopy. As shown in Fig. 6A, all three strains formed similar biofilms in LBN, pH 6.5, under anaerobic conditions (top panel). In contrast, upon exposure to A-NO₂⁻, ΔkatA mutant bacteria exhibited greater levels of dead biofilm bacteria (in red, bottom panel) than wild-type PA. Genetic complementation by a chromosomal copy of wild-type katA at the non-essential attB site restored wild-type biofilm viability in the ΔkatA mutant (Fig. 6A, bottom panel). Upon chemical treatment with 10 mM c-PTIO, a well known NO scavenger [28], [42], biofilm cells of the ΔkatA mutant exhibited the wild-type viability (Fig. 6A, bottom panel). Based on the viability of untreated cells (Fig. 6A), we further calculated and normalized the ratio of dead/live cells in A-NO₂⁻ -treated biofilm. The treated ΔkatA mutant biofilms harbored 14-fold more dead biofilm bacteria relative to wild-type bacteria (Fig. 6B).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 6. Susceptibility of ΔkatA mutant biofilms to A-NO₂⁻.

Biofilms of wild-type P. aeruginosa PAO1, its ΔkatA mutant, and the complemented mutant ΔkatA::katA were grown anaerobically in LBN broth for 24 hrs. After gently washing off the planktonic cells, the biofilm cells were then cultured anaerobically in fresh LB, pH 6.5 with 15 mM KNO₃, or 15 mM KNO₃ + 15 mM NaNO₂, or 15 mM KNO₃ + 15 mM NaNO₂ + 10 mM c-PTIO for additional 48 hrs. The biofilms were then stained with a viability stain containing the DNA binding agent Syto 9 (live, green cells) and dead (propidium iodide, red) according to the Materials and Methods, and such bacteria were observed by confocal laser scanning microscopy. A. Biofilm formation of P. aeruginosa PAO1, ΔkatA mutant and ΔkatA::katA in the absence of NaNO₂ (top panel, top and sagittal views, serving as control biofilms) and in the presence of NaNO₂ (lower panel, top and sagittal views, treated biofilms). Biofilms of the ΔkatA mutant treated with 10 mM c-PTIO under both conditions were shown in far right panels. B. The ratio of dead/live cells in treated biofilms in Fig. 6A was calculated and normalized using ImageJ. The experiment was performed three times independently. The average values were plotted with standard error.

https://doi.org/10.1371/journal.pone.0091813.g006

KatA and NO: Roles in anaerobic to aerobic growth transition

We next assessed the role of KatA in an anaerobic to an aerobic growth transition during which the H₂O₂ levels can suddenly elevate [63]. As shown in Fig. 7, the ΔkatA mutant grew slower during the transition and took an extra hour to reach a similar rate of cell growth relative to wild-type PA, indicating that KatA activity was necessary for such a transition and may be one reason why anaerobic catalase is elevated, especially if organisms are suddenly exposed to molecular oxygen. Unexpectedly, the ΔnirS mutant grew more rapidly than even wild-type bacteria upon the transition from anaerobic to aerobic conditions, suggesting that an absence of endogenous NO (and possibly influenced by increased aconitase activity, Fig. 2E) enhanced the optimal transition from anaerobic growth to a more rapid aerobic growth rate. In striking contrast, the ΔnorCB mutant, that is incapable of disposing anaerobic NO, grew much slower than even the ΔkatA mutant during the transition. The mutant required two extra hours to reach wild-type growth rates and exhibited a much longer exponential growth phase. It was noteworthy that introduction of a ΔkatA mutation in either ΔnirS mutant or ΔnorCB mutant slowed cell growth, suggesting the importance of KatA during an anaerobic to aerobic growth transition, in which the growth rate of all strains in a simplified synopsis are as follows: ΔnirS > ΔnirSΔkatA > PAO1 > ΔkatA >ΔnorCB >ΔnorCBΔkatA (Fig. 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 7. The importance of KatA in the transition from anaerobic to aerobic growth.

Bacteria were grown anaerobically in LBN for 17 hrs. Cultures were diluted 1∶100 into fresh LB broth and incubated aerobically with vigorous shaking at 250 rpm at 37°C. The optical density was measured at hourly intervals. Open circles, PAO1; filled circles, ΔkatA; open triangles, ΔnirS; open squares, ΔnorCB; strikethrough circles, ΔnorCBΔkatA; strikethrough asterisk; ΔnirSΔkatA. The experiments were repeated three times and the representative data were shown.

https://doi.org/10.1371/journal.pone.0091813.g007

Ferric KatA binds NO

With the knowledge in hand that KatA is critical for resistance to A-NO₂⁻ in both planktonic and biofilm culture and important for an optimal growth transition from anaerobiosis to aerobiosis, recombinant KatA was purified to homogeneity to assess its potentially anaerobic NO-binding properties, a known property of catalases [64]. NO typically binds with much higher affinity to heme proteins in the ferrous rather than ferric state. However, attempts to reduce KatA to study the nitrosyl complexes were largely unsuccessful (data not shown). For example, we found that KatA was not reducible with typical reductants such as NADPH and dithionite, even in the presence of mediators such as methyl viologen. Photoreduction of KatA was attempted using deazaflavin and EDTA irradiated with visible light under anaerobic conditions as previously described for other catalases [65]. Even these conditions produced only partial reduction to the ferrous state (data not shown). Based on these observations, in vivo reduction as a means to form a strong NO complex seems unlikely, although, at present, this possibility cannot be ruled out.

The optical Soret band of anaerobic ferric KatA is shown in Fig. 8A (black line). When NO was introduced into the system (Fig. 8A, red line), the spectrum of KatA immediately changed, exhibiting a Soret absorbance at 427 nm and Q bands at 540 and 574 nm, which are characteristic of NO binding to catalase [66]. At the same time, the charge transfer band at 630 nm typical of high spin ferric heme disappeared. Anaerobic titration of ferric KatA with NO showed that NO bound rather tightly to the heme iron, with an apparent K_d of 6 ± 1 μM (Fig. 8A, inset). These results are consistent with those found previously using bovine liver catalase [67]. After early fluctuations from 90 nM to nearly undetectable levels, the internal NO concentration for PA growing under anaerobic conditions is on the order of 0.41 μM ([42]); thus, significant complex formation in vivo seems likely.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 8. Effect of NO bound to oxidized KatA.

A. Optical Spectrum of KatA in as-isolated state (black) and bound to NO (red). Inset to part A: Titration curve of NO binding to KatA. The titration was performed as described in Materials and Methods. Data from two independent titrations are shown superimposed (•,▪). The hyperbolic fit shown utilizes both data sets. B. EPR spectrum of ferric KatA (black line) and NO-bound KatA (red line). 100 μM KatA in the as-purified state in 50 mM Tris, 20% glycerol was frozen in liquid nitrogen. The spectral region below g  =  2.03 was obscured by the signal from dissolved NO. X-band EPR spectra were recorded using a Bruker Elexsys E-500 spectrometer equipped with an Oxford Instruments ESR-10 liquid helium cryostat, under the following conditions: Temp  =  2K, modulation amplitude  =  1 mT, and a microwave power of 100 μW. All experiments were performed three times and the representative spectrum was shown.

https://doi.org/10.1371/journal.pone.0091813.g008

The EPR spectrum at 2.2 K for KatA (Fig. 8B, black spectrum) shows that the metal center is homogeneous exhibiting g–values at 6.3, 5.54 and 1.99, characteristic of a high spin (S  =  5/2) ferric heme with a small distortion in axial symmetry. Generally, this is indicative of a 5-coordinate heme, which is consistent with the KatA crystal structure described below. This indicates that the heme iron in KatA is very likely to have an available binding site for small molecules such as NO. Paralleling the NO-mediated optical changes (Fig. 8A), the EPR spectrum of KatA was eliminated on NO binding (Fig. 8B), presumably due to spin coupling between the S  =  5/2 metal center and S  =  ½ NO. There are no new EPR-active NO-mediated KatA species generated or DNICs formed as evidenced by the absence of any new EPR signals down to g  =  2.03. Evaluation of the region at g  =  2 to determine if there is any ferrous KatA-NO formed was obscured by dissolved NO. However, the optical spectrum of the sample in the EPR tube shows no evidence for any species other than ferric KatA-NO (data not shown).

KatA X-ray structure

The rapid and relative tight binding of NO to KatA as opposed to proteins without iron or heme suggests that a binding cavity large enough to accommodate NO and perhaps containing stabilizing moieties must be present. This was investigated by solving its crystal structure at a 2.55 Å resolution; the structural determination statistics are listed in Table 2. KatA is a class 3 catalase that forms a homotetramer (Fig. S1A). Each monomer of KatA binds a deeply buried heme group and an NADPH molecule at a peripheral solvent exposed site (Fig. S1A and Fig. S1C). The NADPH molecules co-purified with KatA during the protein purification steps, suggesting that KatA has a high affinity for NADPH. The proximal axial ligand of KatA is Tyr338 (Fig. S1B), the anionic charge of which may account in part for the difficulties encountered in reducing the enzyme. This charge would also be expected to weaken the affinity for NO, but this effect may be modulated by the interaction with nearby Arg334 as noted by Bandara et al., [66], for the NO complex of bovine catalase. The small molecule-binding pocket on the distal side of the KatA heme contains the conserved His (H55), which is essential for catalytic activity and its side chain forms a hydrogen bond with a nearby water molecule. Nearby, we modeled Met54 as a methionine sulfone, similar to what has been observed in the catalase of Proteus mirabilis [68]; the function of this sulfone modification has never been determined (Fig. S1D). The structure in the vicinity of the heme is very similar to that reported for bovine liver catalase. The NO complex of the latter enzyme has been structurally characterized [67]. Superposition of the two structures suggests that there would be no spatial constraints to NO binding in the KatA active site (Fig. S1D). Based on this alignment, KatA His55 and Phe141 and the active site solvent could potentially stabilize the binding of NO to the heme iron in the active site. One significant difference is that bovine liver catalase has no methione sulfone, but instead has a valine in its position ∼4 Å from the NO (Fig. S2). If NO binds in the same orientation in KatA as it does in bovine catalase, then the Met-sulfone would be ∼5 Å from the NO, thus, far from hydrogen bonding distance. However, the sulfone side chain has considerable rotational freedom and could potentially assume a new conformation when NO is bound to KatA).

KatA may modulate NO levels in anaerobic PA that lead to formation of dinitrosyl iron complexes (DNIC)

DNIC formation in PA is a result of exposure of iron-containing proteins (e.g., Fe-S or Fe) to endogenously synthesized NO [69]. DNICs inherently have two NO molecules bound to an iron center. Fe-S metal centers are most often the sites of DNIC formation in E. coli [32]. A wide variety of such complexes may be produced, but they are all characterized by the appearance of an EPR resonance of g  =  2.04 [35]. DNICs have been proposed to have a variety of functions (i) act as signaling agents indicating NO stress [70], (ii) exert an “iron starvation” state in cells [71], (iii) compromise the function of proteins to which they are formed [32], or, more recently, (iv) proposed to facilitate repair of Fe-S clusters [72]. We hypothesize that PA KatA may help to decrease DNIC formation that could compromise the function of other iron-containing proteins, including ANR, the master regulator of anaerobic metabolism in PA. To test this hypothesis, we first measured catalase activity in a ΔkatA mutant harboring plasmid pHERDkatA, in which the expression of KatA was inducible by arabinose. As shown in Fig. 9A, KatA activity was induced gradually with increasing concentrations of arabinose, and reached the highest activity (17,000 U/mg) when exposed to 0.5% arabinose. Thus, we elected to use 0.5% arabinose for optimal expression of KatA and examined the effect of KatA on DNIC formation in vivo. Whole cell EPR experiments were performed with wild-type PA, ΔkatA and other strategically selected mutants. The bacteria were grown anaerobically in LBN at pH 6.5, the approximate pH of the COPD and CF airways [40]. The frozen mutant suspensions normalized to total colony forming units were examined using liquid nitrogen EPR. Fig. 9B illustrated that the isogenic DkatA mutant (red line) generates more DNIC (4.4-fold higher signal intensity at g  =  2.04) than wild-type bacteria (black line). Introduction of the plasmid, pHERDkatA, into the DkatA mutant reduced DNIC formation to the wild-type level (blue line). In contrast, Fig. 9C clearly showed the maximal formation of DNIC (63-fold higher signal intensity at g  =  2.04) at pH 6.5 in a DnorCB mutant (red line) which we had shown accumulates micromolar levels (∼13.6 mM) of endogenous NO under anaerobic conditions [42]; RAW 264.7 (mouse macrophage cell line) cells grown in the presence of NO were used as a positive control for DNIC production Fig. 9C, purple line) [35]. We next postulated that deletion of the katA gene would be expected to increase the DNIC peak in a ΔnorCBΔkatA double mutant (Fig. 9C, blue line). However, this was not observed in this strain due to the inherent high intensity of the signal coupled with the already extremely low KatA expression in DnorCB mutant bacteria (recall Fig. 1A). Introduction of the katA gene in trans via the pHERD plasmid where the katA gene was overexpressed to levels shown in Fig. 9A, significantly reduced the DNIC signal by about 60% (Fig. 9C, black line).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 9. Whole cell EPR spectra of anaerobic bacteria.

A. Catalase activity in ΔkatA(pHERDkatA) was induced maximally by 0.5% arabinose. Bacteria were grown anaerobically for 18 hrs in LBN broth supplemented with increased concentration of arabinose. Cells were harvested and lysed for catalase activity assay. EPR spectra were recorded on an X-band (9.33 GHz) Bruker Elexsys E-500 spectrometer at 10 G modulation amplitude, 1 mW microwave power at 150 K. B. Representative DNIC EPR spectra of anaerobic ΔkatA mutant (red line) compared to overnight wild-type PAO1 cultures (black line) and ΔkatA mutant (pHERDkatA, blue line). C. EPR spectra of anaerobic cultures of ΔnorCB (red line); ΔnorCB ΔkatA (blue line); and ΔnorCB (pHERDkatA, black line). D. DNIC EPR spectra of anaerobic cultures of ΔnirS (light blue line); ΔnirS ΔkatA (brown line); PAO1 (black line); ΔrhlR (light green line); ΔrhlR ΔkatA (red line), ΔkatA (purple line); and ΔkatA ΔbfrA (blue line).

https://doi.org/10.1371/journal.pone.0091813.g009

Elimination of NO production by creation of ΔnirS or ΔnirS ΔkatA mutants would be expected to result in less DNIC production as observed in Fig. 9D (light blue and brown lines). In contrast, anaerobic NO overproduction was induced using another mutant (ΔrhlR) devoid of the RhlR quorum sensing regulator (that we have shown overproduces NO during anaerobic growth [28]). Fig. 9D demonstrated that the ΔrhlR mutant bacteria (green line) produces substantial levels of DNIC, but the ΔrhlR ΔkatA mutant strain (red line) increases this level by 50%. This shows, again, that the presence of active KatA attenuates DNIC formation by reduction of NO. We next tested a ΔkatA ΔbfrA strain because we discovered that the bfrA gene, encoding an iron storage protein, BfrA, [10]), and potential NO binding protein, is immediately downstream of katA on the PA genome, but controlled by different promoters [10]. As expected, the DNIC spectrum in the ΔkatA ΔbfrA mutant (blue line) is slightly greater than that of the ΔkatA mutant (purple line). PAO1 (black line) also displayed little, if any, DNIC signal, presumably due to background KatA activity and normal NOR activity.

NO is required for optimal activation of anaerobic katA transcription and the master anaerobic regulator ANR controls this process

The surprising drop in anaerobic catalase activity in the ΔnirS mutant relative to wild-type PA suggested that NO is required for optimal transcription of the katA gene and/or enhances KatA protein stability (Fig. 1A). Clearly, the expression of nirS from a plasmid pHERDnirS and the exogenous addition of SNP resulted in increased NO levels and increased KatA activity in the ΔnirS mutant (Fig. 2A,D). Thus, at least basal levels of NO are required for high levels of KatA expression. This is likely due to the possible requirement of basal NO for optimal ANR transcriptional activity. Supportively, the [4Fe-4S] cluster of the PA ANR homolog in E. coli, FNR, reacts with eight NO molecules, resulting in oxidation of the cluster sulfide ions (S²⁻) to sulfane (S⁰). In fact, some NO molecules can remain associated with the protein as Cys persulfide [73]. This allows for FNR to act optimally as an anaerobic transcription factor, a mechanism that may also occur with PA ANR. Another alternative explanation is that reduced oxygen tension activates ANR, which, in turn, transcriptionally activates the katA gene as well as the NO-responsive second-tier regulator, DNR and other regulators for the denitrification (nir, nor and nos genes). Without NO, however, there is no transcriptional activation of DNR-dependent genes [74], leading to nonfunctional anaerobic respiration, resulting in poor growth of cells and less expression of KatA.

Based upon the suggestion of Trunk et al., [16] that katA is under the control of ANR, we found that this event required production of minimal or basal NO levels. However, production of too much cellular NO or very high levels of ANR in the cell decreased catalase activity as well as cell density (Figs. 2C,D and Fig. 3). Similarly, very high NO levels dramatically reduced KatA activity in a ΔnorCB mutant (Fig. 1A) and also eliminated NO-sensitive aconitase activity in this strain (Fig. 2E). Thus, when wild-type bacteria have low basal levels of NO, they produce more anaerobic than aerobic catalase activity (refer to our model in Fig. 10) and this process requires both the transcriptional activator ANR and at least a basal level of NO mediated by normal NIR activity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 10. Model of anaerobic respiration in PA and the effects of KatA and NO.

Depiction of the PA anaerobic anaerobic regulatory hierarchy and NO levels produced wild-type (A) (estimated baseline of 0.41 μM NO from Yoon et al., [42]), ΔkatA mutant (B), ΔnorCB mutant (C), and ΔnorCB (pHERDkatA) (D) strains. The central regulatory theme is based upon that described by Schreiber et al., [88]. The red “lightening bolts” indicate the nitrogen oxide species that control activation of specific anaerobic respiratory regulators (ANR (red spheres), DNR (blue spheres), NarXL (green spheres), NirQ (black spheres) with the exception of the katA gene. The colored lines emanating from ANR, DNR, NarXL and NirQ are those directed at downstream genes controlling the anaerobic respiratory cascade. The level of NO produced (in this case measured using an NO electrode or via extrapolation from DNIC levels in Fig. 9) and from Yoon et al., [42], in each organism is given on the top caption of parts A-D and also by a red color and font size, the size of which is directly correlative to levels of NO produced in each cell. The black circular “strike through” signs are assigned where genes are not activated. The arrow widths indicate the amount of DNIC produced in each mutant. The width of the black upraised arrows indicates the relative levels of DNIC (dinitrosyliron complexes) produced in each cell. CIP = chelatable iron pool.

https://doi.org/10.1371/journal.pone.0091813.g010

KatA is required for optimal anaerobic resistance to the NO generating agent, A-NO₂⁻

In nature and during human airway diseases such as CF and COPD, bacteria often form antibiotic/biocide refractory, highly organized, enmeshed communities known as biofilms [28], [75], [76]. We have previously shown that KatA is essential for optimal viability of bacteria grown in aerobic flow-through biofilm systems when exposed to H₂O₂ [21]. In this study, we demonstrated that KatA was essential for optimal anaerobic viability of planktonic and biofilm bacteria and enhanced survival when cells were challenged with the potential therapeutic agent A-NO₂⁻ in both the wild-type PAO1 and its formidable antibiotic-resistant, mucoid mucA22 mutant (Fig. 4–6). The aforementioned mutant strains are frequently isolated from the lungs of both CF and COPD patients and severely complicate the clinical course for such patients. Like mucoid mucA22 mutant of PA, B. cepacia has also recently been shown to be susceptible to killing by anaerobic A-NO₂⁻ [77]. When oxygen is present, NO has been found to potentiate killing of B. cepacia, by formation of reactive oxygen intermediates (ROI) [78].

Elevated anaerobic KatA is important for an optimal anaerobic to aerobic physiological transition

During biofilm culture and in some biofilm-related diseases (e.g., CF, COPD), there are macrocolonies harboring bacteria that are likely significantly deprived of oxygen [25] or even completely anaerobic [24], [28]. However, when bacteria become dislodged from the biofilm (often termed biofilm dispersion, [79]), they may be immediately exposed to ambient oxygen, leaving them vulnerable to a sudden and potentially lethal oxidative stress. Our data in Fig. 7 indicate that KatA may facilitate the difficult transition from anaerobic to aerobic growth after suddenly being subjected H₂O₂ via aerobic respiration. In E. coli, fumarate reductase and the flavin-dependent desaturating dehydrogenase, NadB, are capable of generating H₂O₂ after a sudden anaerobic to aerobic transition [63], [59]. Intriguingly and unexpectedly, the ΔnirS mutant grew more rapidly when exposed to molecular oxygen than even wild-type bacteria (Fig. 7), indicating that the ΔnirS mutant cells have lesser cellular problems to deal with during the transition than other strains tested. In contrast, the ΔnorCB mutant faced significant difficulties to grow up as manifested by the slowest growth (Fig. 7). Such differential bacterial growth during the transition was most likely attributable to the varied degree of NO-mediated cell damage or nitrosative stress occurred in anaerobic cells. Introduction of a ΔkatA mutation in either ΔnirS mutant or ΔnorCB mutant slowed cell growth, suggesting the importance of KatA during an anaerobic to aerobic growth transition.

Anaerobic KatA likely acts as an internal buffer or a sensor to help initiate the detoxification process

Because NO is produced during anaerobic respiration by PA and many other denitrifying bacteria [80], and it has a high potential for interaction ­with iron, its levels must be stringently controlled by NO-detoxifying enzymes such as NorCB [31] and flavohemoglobin [81]. Our first hypothesis for elevated anaerobic KatA activity and its very high cellular protein levels is that it acts as a scavenger or “sponge” for NO at levels approaching its K_d (e.g., ∼1 μM). The NO scavenging hypothesis for KatA, which crystallographically resembles the shape of a rectangular brick, with the long axis ∼100 Å and the short axis ∼50 Å, was initially derived from in vitro NO-KatA binding data, showing that the K_d for KatA with NO was relatively high, 6 ± 1 μM (Fig. 8A, inset). These results are consistent with those of bovine liver catalase [67]. The hypothesis was strongly supported by our in vivo EPR data obtained from several isogenic mutant strains. First, the ΔkatA mutant generated 4.4-fold more DNIC than wild-type bacteria (Fig. 9B). Second, the overexpression of KatA from a plasmid not only restored the wild-type level of DNIC in the ΔkatA mutant, but also reduced the culminating level of DNIC in ΔnorCB mutant bacteria by ∼60% (Fig. 9B,C). Third, the level of DNIC in ΔrhlR mutant was increased by ∼50% with the introduction of katA mutation (Fig. 9D). The molecular basis for the protective effects of KatA on DNIC formation in PA may involve either buffering of free NO or a signaling event that may recruit other cellular defenses as toxic NO concentrations are approached. Hence, KatA catalase would serve to help protect the bacteria by buffering the NO, binding it when NO concentrations are elevated, but releasing it as concentrations fall so that NOR can degrade it. Such systems are likely overwhelmed in a ΔkatA mutant, which demonstrated clear sensitivity to the NO generator, A-NO₂⁻ in planktonic and biofilm bacteria (Fig. 4, 5 and 6).

KatA and similarities to anaerobic homogentisate-1,2-dioxygenase and hydroxyphenylpyruvate dioxygenase, non-obvious enzymes that bind to NO

Anaerobic KatA is similar in many respects to two oxygen-dependent proteins that are expressed in the anaerobic ΔnorCB mutant, the iron-containing dioxygenases, homogentisate-1,2-dioxygenase (HmgA) and 4-hydroxyphenylpyruvate dioxygenase (Hpd) [42], respectively, that are typically only expressed aerobically. Initially, this finding seemed metabolically counterintuitive, since these latter enzymes require oxygen for enzymatic activity, yet their ANR-dependent, anaerobic de-repression (HmgA and Hpd overproduction) helped provide protection against impending cell death in biofilms because they help scavenge NO (supported by the EPR experiments with purified enzymes [42]), similar to experiments performed with KatA in this study. The NO produced in norCB mutant bacteria was high enough (∼13.6 μM) to impair the transcriptional activity of ANR via disruption of its [4Fe-4S]²⁺ cluster as shown by Mössbauer spectroscopy [42]. This, in turn, makes sense since ANR poisoning by NO in the ΔnorCB mutant, enabling ANR-DNIC formation, would impair anaerobic katA transcription (Fig. 1B), resulting in significantly reduced KatA activity in this mutant (Fig. 1A) and resulted in a dramatic delay in the anaerobic to aerobic growth transition (Fig. 7). Similar to ANR and its homolog in E. coli, FNR, [82], NO also impairs other critical Fe-S proteins such as aconitase in E. coli [54] and the activity of this enzyme is essentially undetectable in a PA ΔnorCB mutant (Fig. 2E). Fridovich and colleagues postulated that NO-mediated aconitase inactivation represented a “circuit breaker” like mechanism to prevent formation of additional reducing power in the form of NADH formed by the downstream TCA cycle enzyme, isocitrate dehydrogenase. In both ANR and aconitase, and likely many other Fe-S cellular proteins (>500 have been identified in different organisms, [83]), this compromised protein function is likely via formation of DNIC complexes. For example, Ren et al., [32] have observed, using anaerobic E. coli, more than 27 DNIC protein fractions with characteristic EPR g  =  2.04 signals ranging in molecular weight from >679 kDa to less than 29 kDa. Thus, while the identities of the proteins contributing to the very intense DNIC signal in anaerobic ΔnorCB PA (Fig. 9C) remain unknown, we offer a model of how such events are influenced by KatA and NO in Fig. 10A-D. Again, this model is supported by the substantial decrease in the DNIC EPR signal when the katA gene is overexpressed in the norCB mutant in Fig. 9C. In wild-type anaerobic PA, NO levels rise to ∼90 nM within 15 minutes and then are reduced to nearly undetectable levels and yet eventually rise to ∼0.46 μM (based upon our DNIC conversions). This is due to the fact that ANR and its downstream regulator, DNR, which is activated by NO [84], are fully functional and activate the necessary NO detoxifying enzymes (e.g., NorCB) or proteins to which NO binds, Fe-S proteins (ANR, aconitase, KatA, HmgA, Hpd, etc.) that help the cell keep intracellular NO levels at very low levels (0.41 μM, Fig. 10A). Fig. 10C shows a schematic diagram of very high NO levels in the ΔnorCB mutant with the chelatable iron pool (CIP) to generate very high DNIC levels, impair ANR, and thereby reduce anaerobic KatA activity by 43-fold (Fig. 1A). However, again, overexpression of KatA in the ΔnorCB mutant bacteria, reduced DNIC production by ∼60% (Fig. 9C).