Proteomic, Microarray, and Signature-Tagged Mutagenesis Analyses of Anaerobic Pseudomonas aeruginosa at pH 6.5, Likely Representing Chronic, Late-Stage Cystic Fibrosis Airway Conditions

Patients suffering from cystic fibrosis (CF) commonly harborthe important pathogen Pseudomonas aeruginosa in their airways.During chronic late-stage CF, P. aeruginosa is known to growunder reduced oxygen tension and is even capable of respiringanaerobically within the thickened airway mucus, at a pH of $~$ 6.5. Therefore, proteins involved in anaerobic metabolism representpotentially important targets for therapeutic intervention.In this study, the clinically relevant "anaerobiome" or "proteogenome"of P. aeruginosa was assessed. First, two different proteomicapproaches were used to identify proteins differentially expressedunder anaerobic versus aerobic conditions. Microarray studieswere also performed, and in general, the anaerobic transcriptomewas in agreement with the proteomic results. However, we foundthat a major portion of the most upregulated genes in the presenceof NO₃^– and NO₂^– are those encoding Pf1 bacteriophage.With anaerobic NO₂^–, the most downregulated genes arethose involved postglycolytically and include many tricarboxylicacid cycle genes and those involved in the electron transportchain, especially those encoding the NADH dehydrogenase I complex.Finally, a signature-tagged mutagenesis library of P. aeruginosawas constructed to further screen genes required for both NO₃^–and NO₂^– respiration. In addition to genes anticipatedto play important roles in the anaerobiome (anr, dnr, nar, nir,and nuo), the cysG and dksA genes were found to be requiredfor both anaerobic NO₃^– and NO₂^– respiration. Thisstudy represents a major step in unraveling the molecular machineryinvolved in anaerobic NO₃^– and NO₂^– respirationand offers clues as to how we might disrupt such pathways inP. aeruginosa to limit the growth of this important CF pathogenwhen it is either limited or completely restricted in its oxygensupply.

INTRODUCTION

Top

INTRODUCTION

Pseudomonas aeruginosa is a gram-negative bacterium of environmentaland clinical importance that is capable of both aerobic andanaerobic respiration, the latter of which requires nitrate(NO₃^–), nitrite (NO₂^–), or nitrous oxide (N₂O) asan alternative electron acceptor (24). The organism can alsoutilize arginine for anaerobic growth via substrate-level phosphorylation,although the final cell yield during this form of growth isabysmally low compared to that observed during anaerobic respiration(55). The most facile means to obtain anaerobic energy, however,is via respiration by NO₃^– reduction. The process of nitratereduction can occur by two routes, the first of which is anassimilatory pathway where the nitrogen from NO₃^– is incorporatedinto macromolecules via formation of NH₃. Assimilation can proceedunder both aerobic and anaerobic conditions. In contrast, respiratoryNO₃^– reduction (denitrification) occurs only under anaerobicconditions and involves the sequential eight-electron reductionof NO₃^– to nitrogen gas (N₂), with intermediates includingNO₂^–, nitric oxide (NO), and N₂O. The anaerobic processgenerates respiratory energy for the cell.

A hallmark of the inherited fatal disease cystic fibrosis (CF)is that patients eventually succumb to lung infection by P.aeruginosa and die at an average age of 36.8 years (www.cff.org).It is now becoming increasingly evident that the oxygen tensionwithin the thickened CF airway mucus found in patients sufferingfrom chronic CF airway disease is either significantly reduced(<2%) (2, 35) or absent (anaerobic) (62). In fact, macrocoloniesof P. aeruginosa growing in vitro as complex communities knownas "biofilms" demonstrate complete oxygen depletion within thetop 30 µm (11). Recently, several laboratories confirmedour findings from 2002 (62) showing that the anaerobic biofilmmode of growth can occur within pockets of the thick airwaymucus of chronically infected CF patients. First, Beckmann etal. (7) showed by using phage display that narG, encoding theanaerobic respiratory nitrate reductase (NAR) ${alpha}$ -chain, was detectedin sera from CF patients within the first year of life. Palmeret al. (42) then showed that NarG was essential for growth ina synthetic CF sputum. Most recently, Son et al. (52) have shownby microarray analysis of CF sputum samples that there is abundantanaerobic gene expression, including the entire anaerobic respiratorypathway (nar, nir, nor, and arc genes), by P. aeruginosa inthe CF airways. Therefore, it appears that the synthesis ofdenitrifying enzymes can take place even in the presence oflow concentrations of oxygen as long as the levels are belowthe biological oxygen demand of the culture (1).

Because humans lack the enzymatic machinery for anaerobic sustenance,such enzymes and the associated processing machinery necessaryfor P. aeruginosa to survive under such conditions representpotentially viable targets for therapeutic intervention. Towardthis end, Yoon et al. (62) revealed the following three majorfindings: (i) P. aeruginosa forms more robust biofilms underanaerobic than under aerobic conditions; (ii) when P. aeruginosalacks the rhl quorum-sensing circuit, the bacterium commitsa metabolic suicide by overproduction of toxic levels of NO;and (iii) organisms lacking the outer membrane protein OprFgrow very poorly during anaerobic respiration. Yoon et al. (61)also found that NO₂^–, at what was found to be the slightlyacidified pH ( $~$ 6.5) of CF airway mucus, effectively kills theantibiotic (23)- and phagocyte (14)-resistant mucoid form ofP. aeruginosa. Two transcriptomic studies, although elegantlydone, were somewhat limited in their breadth in the sense thatthey did not specifically select to rigorously assess the true"anaerobiome" of P. aeruginosa in the context of both NO₃^–-and NO₂^–-grown cells, but only with cells grown in NO₃^–.Specifically, Wagner et al. (57) and Filiatrault et al. (18,19) have investigated certain aspects of anaerobic gene expressionby using GeneChip microarrays. However, many of the genes thatare known to be expressed under anaerobic conditions do notrequire supplementation with NO₃^– or NO₂^– (59).The genes that are known to be induced by strict anaerobiosis,as opposed to the simple addition of NO₃^– or NO₂^–,include narI, narK1 and -2, hemN, arcABC, and napABDFPQ, amongothers. However, Wagner et al. (57) also showed that anaerobicgrowth actually downregulated narK2, narK1, and akk as wellas the napABDF genes, encoding a periplasmic NAR, which havebeen reported to be constitutive and are not required for anaerobicgrowth using NO₃^– (48). Because of the often-ignored differencesbetween wild-type strains of PAO1, the conditions of growthand cell manipulation, the parameters being elucidated experimentally,the necessity for a multifaceted approach, and above all, theimportance of this research, there is justification for a moredetailed and thorough study, which forms the body of this work.Thus, because P. aeruginosa forms more robust biofilms duringanaerobic growth and CF patients succumb to pulmonary insufficiency(60-62), again, the aforementioned anaerobic respiration pathwaymachinery represents plausible gene products for the developmentof novel therapeutic intervention. Thus, the identificationof those gene products required for anaerobic growth is warranted,specifically at pH 6.5.

In this study, a collective (i) proteomic, (ii) transcriptionalprofiling, (iii) signature-tagged mutagenesis (STM), and (iv)targeted mutagenesis approach was initiated in order to identifyP. aeruginosa gene products that are required for anaerobicNO₃^– and NO₂^– respiration at pH 6.5. Ultimately,this study represents a major step in unraveling the molecularmachinery involved in P. aeruginosa anaerobic NO₃^– andNO₂^– respiration and offers clues as to how we might disruptthese pathways, which could ultimately limit the growth of thisimportant pathogen in CF airway mucus.

MATERIALS AND METHODS

Top

MATERIALS AND METHODS

Construction of a P. aeruginosa tryptic library database. First, each of 5,570 open reading frames carried by the P. aeruginosagenome (www.pseudomonas.com) was downloaded from http://www.pseudomonas.com/downloads/sequences/Pseudomonas_aeruginosa_PAO1_2004-Jan-14.fhain Macintosh Stuffit format. The resulting file, designatedP_aeruginosa_Prot.fasta.1, was converted to a Microsoft Worddocument, PA.doc. The protein sequences were downloaded fromPA.doc into ProFound (64) and ProteinProspector (http://prospector.ucsf.edu/),and tryptic fragment libraries were assembled.

Growth of bacteria and preparation of cell extracts. P. aeruginosa PAO1 was grown in Luria (L) broth (10 g/litertryptone, 5 g/liter yeast extract, 5 g/liter NaCl) containingeither 15 mM KNO₃ or 15 mM NaNO₂ at pH 6.5. Bacteria were growneither aerobically with shaking at 300 rpm (volume/flask ratio,1:10) or anaerobically in a Coy anaerobic chamber at 37°Cfor 24 h (NO₃^–-grown cells) or 96 h (NO₂^–-growncells) (see Fig. S1 in the supplemental material). Bacteriawere subjected to three freeze-thaw (–80°C/37°C)cycles, followed by sonication with a Heat Systems Ultrasonicssonic disruptor (Farmingdale, NY) with the microtip at setting5 for 20 seconds on ice. Cell extracts in 10 mM Tris-HCl, pH7.4, were freed of membranes by centrifugation at 100,000 xg for 2 h, and samples were kept frozen at –80°C untiluse. Because NAR and nitric oxide reductase (NOR) are membranebound, these proteins were not expected to be found in the membrane-freeextracts.

2-D gel electrophoresis. Two-dimensional (2-D) gel electrophoresis of P. aeruginosa cellextracts was performed according to the method of O'Farrell(40), as outlined in detail by Sauer and Camper (46). Briefly,crude protein extracts (200 µg) were solubilized in 450µl of a solution containing urea, thiourea, dithiothreitol(DTT), 3-[(3-cholamidopropyl)-dimethylammonio]-1-propanesulfonate(CHAPS), and Pharmalyte 3-10. Samples were applied to ImmobilineDry-Strips (18 cm) (pH 3-10 NL; GE Healthcare) by in-gel rehydration.Isoelectric focusing (IEF) was performed using a Multiphor IIapparatus (GE Healthcare) for a total of 48 kV-h. Upon completionof IEF, the Dry-Strips were equilibrated in a two-step process.In the first step, protein disulfide bonds were reduced by DTTfor 15 min, while in the second step, cysteines were irreversiblymodified by iodoacetamide treatment for 15 min. For the resolutionin the second dimension, a 24- by 20-cm 2-D gel system fromGenomics Solutions, Inc., was used. Crude protein extracts wereseparated in 11% resolving gels at 15°C. 2-D gels were stainedwith silver nitrate (8) and run in triplicate for each growthcondition to confirm the reproducibility of the protein patternsunder planktonic and biofilm growth conditions. A calibratedimage scanner (GE Healthcare) was used for gel scanning to ensureeven spot detection and higher accuracy for the subsequent imageanalysis. Computational image analysis was carried out usingImage Master 2-D Platinum software (GE Healthcare). A fourfolddifference in spot volume was considered significant.

MALDI-TOF MS. Matrix-assisted laser desorption ionization-time-of-flight massspectrometric (MALDI-TOF MS) identification of proteins wasperformed according to previously described strategies (50;http://proteomics.uc.edu). Briefly, protein spots of interestwere excised from the 2-D gels and digested in situ with sequencing-grade,tosylsulfonyl phenylalanyl chloromethyl ketone-modified trypsin(Promega), using a ProGest workstation (Genomics Solutions Inc.,MI). After digestion for 8 h at 37°C, tryptic peptides wereextracted with 50% acetonitrile-0.1% trifluoroacetic acid anddesalted if necessary, using ZipTips (Millipore). An aliquotof the peptide solution was spotted on a MALDI target plate,and mass spectra were recorded on an Ettan MALDI-TOF Pro massspectrometer (GE Healthcare) operated in reflectron mode asdescribed previously (47, 53). As little as 1 pmol of proteinwas sufficient for identification of candidate protein spots.Trypsin peptides were used as internal calibrants for everypeptide sample to ensure high mass accuracy. The peptide massfingerprinting spectra were processed using Ettan evaluationsoftware (GE Healthcare). Briefly, the generated mass lists,composed of monoisotopic [M + H]⁺ masses, were first filteredfor common contaminants (e.g., keratin) and subsequently usedfor database searches using the ProFound search algorithm (64).The database used in this study was composed of current, nonredundantprotein sequences obtained from TIGR (comprehensive microbialresource batch download website [http://www.tigr.org/tigr-scripts/CMR2/batch_download.dbi])and comprised the sequences of Streptococcus pneumoniae R6,S. pneumoniae TIGR4, Streptococcus pyogenes M1, Staphylococcusaureus MRSA252, Streptococcus epidermidis, Enterococcus faecalisV583, Escherichia coli K-12 MG1655, and P. aeruginosa PAO1.All proteins were identified with significant certainty (probabilityscore of <0.03). Proteins were identified with 3 to 15 matchedpeptides and a minimum of 5% sequence coverage.

Protein isolation for subsequent nano-high-performance liquid chromatography-microelectrospray ionization (nHPLC-µESI) MS analyses. P. aeruginosa strain PAO1 was grown under aerobic and anaerobicconditions in LB containing 15 mM KNO₃ or NO₂^–, pH 6.5,at 37°C as described above. The broth-grown samples werepoured over crushed ice and diluted in ice-cold buffer A (0.1M NH₄HCO₃-1 mM DTT-0.05% CHAPS). The bacteria were harvestedby centrifugation at 13,000 x g for 10 min at 4°C. The pelletwas quick-frozen in dry ice-ethanol, thawed on ice, and resuspendedin buffer A. The bacteria were then lysed twice with a Frenchpressure cell at 12,000 lb/in² at 4°C. The samples weretreated with 20 U/ml of both DNase and RNase containing 10 mMMgCl₂ for 15 min on ice. At this point, 1 mM EDTA was added.Debris was removed by centrifugation at 4°C for 15 min at13,000 x g. An aliquot of the supernatant was removed to determinethe protein concentration, and the remainder was frozen at –80°C.To enhance the recovery of membrane proteins, we added 0.3 MNaCl to the 0.1 M NH₄HCO₃-1 mM DTT-0.05% CHAPS.

Sample digestion for subsequent nHPLC-µESI MS. Equal amounts of total protein (70 µg) from P. aeruginosagrown either aerobically or anaerobically as described abovewere diluted to a total volume of 300 µl with 100 mM ammoniumbicarbonate (pH 8.5). The proteins in each solution were reducedwith 200 mM DTT (5 µl) at 51°C for 1 h, carboxyamidomethylatedwith 450 mM iodoacetamide (5 µl) in the dark at room temperaturefor 1 h, and digested with modified trypsin (3.5 µg in7 µl; Promega) at 37°C for 8 h. Proteolysis was terminatedby acidifying the reaction mixture to a pH of 3 with glacialacetic acid (13 µl).

FT-ICR nHPLC-µESI MS analysis. Aliquots of the above digests (5 µl; 1.05 µg protein)were diluted to a total volume of 100 µl with 0.1% aceticacid in water. A small amount of each solution (1 µl;0.0105 µg total protein) was loaded separately, usinga pressure bomb, onto an analytical column with an integratedESI emitter tip (1- to 5-µm diameter) (34). Samples wereanalyzed in duplicate by nHPLC-µESI MS on a home-builtFourier transform ion cyclotron resonance (FT-ICR) mass spectrometerfitted with a custom-designed ESI source. The HPLC gradient(A = 100 mM acetic acid in water, B = 70% acetonitrile-100 mMacetic acid in water) was 0 to 50% B in 50 min, 50 to 100% Bin 5 min, 100 to 0% B in 5 min, and 0% B for 5 min. Full-scanmass spectra (m/z 300 to 5,000) were acquired at a rate of approximately1 scan/s. Mass resolving power ranged from 5,000 to 10,000.

Ion-trap nHPLC-µESI MS/MS analysis. Aliquots of each of the above digests (2.5 µl; 0.525 µgtotal protein) were analyzed by nHPLC-µESI MS on an LCQDeca mass spectrometer (Thermo/Finnigan, San Jose, CA). Thepeptides were loaded onto a custom-made C₁₈ microcapillary precolumnby use of a pressure bomb, the precolumn was rinsed to removesalts, and the precolumn was then connected to an analyticalcolumn containing an integrated ESI emitter tip. Peptides wereeluted into the mass spectrometer, using the HPLC gradient detailedabove (spray voltage = 1.7 kV). The instrument was operatedin data-dependent mode and cycled though a single MS (m/z 300to 2,000) and five MS/MS experiments every 12 to 15 s. All MS/MSscans (collision energy = 35%) were performed with an isolationwindow of 3 atomic mass units. The dynamic exclusion optionwas selected, with a repeat count of 1, a repeat duration of0.5 min, and an exclusion duration of 1 min. Peptide sequenceswere assigned using the SEQUEST algorithm (http://fields.scripps.edu/sequest/index.html)and/or de novo sequencing.

RNA isolation and preparation for Affymetrix GeneChip analysis. Growth curves were generated to determine the CFU/ml and growthphase for P. aeruginosa PAO1 under anaerobic and aerobic conditionsin the presence of 15 mM NO₂^– (see Fig. S1 in the supplementalmaterial). Three independent cultures of P. aeruginosa PAO1were grown in buffered (0.1 M potassium phosphate, pH 6.5) LBcontaining 15 mM NaNO₂ for 4 days at 37°C under anaerobicconditions. Three independent cultures of P. aeruginosa PAO1were grown aerobically for 4.5 h in the same medium. These aerobicand anaerobic growth conditions resulted in 2 x 10⁸ CFU/ml attheir respective time points and in mid-log growth phase. Forgrowth in the presence of 15 mM KNO₃, growth curves were generatedto determine the CFU/ml and growth phase for P. aeruginosa PAO1under anaerobic and aerobic conditions. Three independent culturesof P. aeruginosa PAO1 were grown in LB containing 15 mM KNO₃and 0.1 M potassium phosphate, pH 6.5, for 24 h at 37°Cunder anaerobic conditions. Three independent cultures of P.aeruginosa PAO1 were grown aerobically for 5.5 h in the samemedium. These aerobic and anaerobic growth conditions resultedin 3 x 10⁹ CFU/ml at their respective time points and in mid-loggrowth phase. After the cultures were chilled in a dry ice-ethanolbath to stop RNA synthesis, the cells were collected by centrifugationand RNA was isolated as previously described (33).

The quality of the RNA was assessed on an Agilent Bioanalyzer2100 electrophoretic system pre- and post-DNase treatment. TheRNA was treated with 2 U of DNase I (Ambion) for 15 min at 37°Cto remove contaminating DNA. The reaction was stopped by theaddition of 25 µl of DNase stop solution (50 mM EDTA,1.5 M sodium acetate, 1% sodium dodecyl sulfate). The DNaseI was removed by phenol-chloroform extraction followed by ethanolprecipitation. Total RNA (10 µg) was used for cDNA synthesis,fragmentation, and labeling according to the Affymetrix GeneChipP. aeruginosa genome array expression analysis protocol (Affymetrix).Briefly, random hexamers (Invitrogen) were added (25 ng/µl)to the 10 µg of total RNA along with in vitro-transcribedBacillus subtilis control spikes (as described in the AffymetrixGeneChip P. aeruginosa genome array expression analysis protocol).cDNA was synthesized using Superscript II (Invitrogen) accordingto the manufacturer's instructions, under the following conditions:25°C for 10 min, 37°C for 60 min, 42°C for 60 min,and 70°C for 10 min. RNA was removed by alkaline treatmentand subsequent neutralization. The cDNA was purified by a QIAquickPCR purification kit (Qiagen) and eluted in 40 µl of bufferEB (10 mM Tris-HCl, pH 8.5). The cDNA was fragmented by DNaseI (0.6 U per µg cDNA; Amersham) at 37°C for 10 minand then end labeled with biotin-ddUTP, using an Enzo BioArrayterminal labeling kit (Affymetrix), at 37°C for 60 min.Proper cDNA fragmentation and biotin labeling were determinedby gel mobility shift assay using NeutrAvadin (Pierce) on a5% polyacrylamide gel stained with SYBR green I (Roche).

Microarray data analysis. Microarray data were generated using standard protocols generatedby Affymetrix. Absolute transcript expression levels from dataderived from three GeneChip microarrays per condition were normalizedfor each chip by globally scaling all probe sets to a targetsignal intensity of 500. Three statistical algorithms (detection,change call, and signal log ratio) were then used to identifydifferential gene expression in experimental and control samples.The detection metric (present, absent, or marginal) for a particulargene was determined using default parameters in MAS software(version 5.0; Affymetrix). Transcripts that were absent underboth control and experimental conditions were eliminated fromfurther consideration. The data generated in MAS were importedinto Affymetrix Data Mining Tools (version 3.0) to perform batchanalyses in which pairwise comparisons between individual experimentaland control chips were made in order to generate a change calland a signal log ratio value for each transcript. The statisticalsignificance of differences in signals between the control andexperimental conditions (P < 0.05) for individual transcriptswas determined using the t test. We defined a positive changecall as one in which >50% of the transcripts had a call ofincreased or marginally increased for upregulated genes anddecreased or marginally decreased for downregulated genes. Finally,the median value of the signal log ratios from each comparisonfile was calculated. Only those genes that met the above criteriaand had median signal log ratios of $≥$ 1 for upregulated transcriptsand $≤$ 1 for downregulated transcripts were kept in the final listof genes. Signal log ratio values were converted from log₂ andexpressed as x-fold changes.

Construction of STM library and cloning of STM tags into mini-Tn5 transposons. Three mini-Tn5-based transposons, pUTmini-Tn5Km2, pUTmini-Tn5Tet,and pUTmini-Tn5TetGFP, were used for mutagenesis (16, 26, 36).The transposons are located on an R6K-based suicide deliveryplasmid, pUT, where the Pi protein is furnished by the donorcell; the pUT plasmid provides the IS50R transposase tnp genein cis, but external to the mobile element, and its conjugaltransfer to recipients is mediated by RP4 mobilization functionsin the donor (51). Plasmid DNA (0.04 pmol) was ligated with1 pmol of double-stranded DNA tags in a final volume of 10 µlof 1x T4 DNA ligase buffer containing 40 U of T4 DNA ligasein 24 separate reaction mixtures. pUTmini-Tn5Km2 was digestedwith KpnI (New England Biolabs), and recombinant molecules wereconstructed in vitro by blunt-end fill-in with T4 DNA polymerase(Gibco BRL Products). pUTmini-Tn5Tc and -GFP were digested withNotI (New England Biolabs), and recombinant molecules were constructedin vitro by blunt-end fill-in with Klenow fragment (New EnglandBiolabs). Ligated products were purified using Microcon PCR(Millipore) and resuspended in 5 µl of water. The entire5 µl of ligated products was transformed into E. coliS17- ${lambda}$ pir by electroporation using a Bio-Rad apparatus operatedat 2.5 kV, 200 ${Omega}$ , and 25 µF with a 2-mm electroporationgap cuvette. Transformed bacteria containing tagged plasmidswere selected on tryptic soy agar supplemented with 50 µg/mlof ampicillin and 50 µg/ml of kanamycin. Single colonieswere selected, purified, and screened using 10 pmol of one ofthe oligonucleotide tags (29) used to construct the DNA tagsas the 5' primer and 10 pmol of pUTKanaR1 (5'-GCGGCCTCGAGCAAGACGTTT-3')as the 3' primer in the kanamycin resistance gene. Thermal cyclingconditions were set for touchdown PCR in a DNA thermal cycler(Perkin-Elmer Cetus) by using a hot start for 7 min at 95°C;2 cycles at 95°C for 1 min, a temperature ranging from 70°Cto 60°C for 1 min, and 72°C for 1 min; and 10 cyclesat 95°C for 1 min, 60°C for 1 min, and 72°C for1 min. Ten microliters of amplified products was analyzed byelectrophoresis in a 1% agarose gel with 1x Tris-borate-EDTAbuffer and stained for 10 min in 0.5-µg/ml ethidium bromidesolution (45). The amplified product had a size of 500 basepairs.

Conjugation and insertion of mini-Tn5 into P. aeruginosa and arraying of mutant libraries. E. coli S17- ${lambda}$ pir containing the pUTminiTn5 plasmids with tagswas used as a donor for conjugal transfer into the recipientP. aeruginosa strain PAO1, using a ratio of 1 donor cell ofE. coli to 10 recipient cells of P. aeruginosa. Bacteria weremixed, and 50 µl was spotted on a sterile nylon membranethat was then placed on a nonselective brain heart infusion(BHI) agar plate. Plates were incubated at 30°C for 8 h.Filters were washed with 10 ml of sterile phosphate-bufferedsaline to recover bacteria. Five 100-µl aliquots of thephosphate-buffered saline solution containing the transconjugantswere plated on five BHI agar plates supplemented with the appropriateantibiotics. Kanamycin was used to select transconjugants withmini-Tn5Km2, and tetracycline was used to select those withmini-Tn5Tc and mini-Tn5GFP. Plates were incubated overnightat 37°C. The selected colonies were picked on BHI agar supplementedwith ampicillin to exclude bacterial colonies carrying the suicidedonor pUT plasmid inserted into the chromosome by homologousrecombination. Exconjugants were selected on BHI agar supplementedwith chloramphenicol (5 µg ml^–1) (Sigma) and kanamycin(500 µg ml^–1) for mini-Tn5Km2 or with tetracycline(15 µg ml^–1) for mini-Tn5Tc and mini-Tn5GFP. Kanamycin-resistantand ampicillin-sensitive exconjugants were arrayed as librariesof 96 clones in 96-well microtiter plates, using 1.5 ml of BHIsupplemented with kanamycin and appropriate antibiotics. Toassemble the mutant library, one mutant from each library waspicked to form 96 pools of 72 unique tagged mutants in each2-ml well, labeled, and arrayed. In a defined library, eachmutant had the same tag, but it was theoretically inserted ata different location in the bacterial chromosome.

Cloning and analysis of disrupted STM gene mutants. Chromosomal DNAs from the STM mutants were prepared as describedby the manufacturer (Qiagen). Chromosomal DNA (1 µg) wasdigested with PstI, giving DNA fragments ranging in size from1 to 6 kb. Digested chromosomal DNA was cloned into pTZ18R (AmershamPharmacia Biotech). Ligations were performed using 1 µgof digested chromosomal DNA mixed with 50 ng of digested pTZ18Rin 20 µl of T4 DNA ligase buffer with 40 units of T4 DNAligase. Ligated products were incubated overnight at 16°Cand purified using Microcon PCR (Millipore) as described bythe manufacturer. The recombinant plasmid was electroporatedinto E. coli DH5 ${alpha}$ . Bacterial clones were purified and analyzedfor plasmid content with a Qiagen Mini preparation kit as describedby the manufacturer (Qiagen). Plasmids were sequenced usingthe complementary primer of the corresponding tagged mutantor the 3'-conserved transposon primers encoding antibiotic resistance.Automated sequencing was performed as suggested by the manufacturer.The DNA sequences obtained were assembled and subjected to databasesearches using BLAST, included in the GCG Wisconsin package(version 11.0). Similarity searches with complete genomes wereperformed at the NCBI website, using microbial genome sequences(http://www.ncbi.nlm.nih.gov), or in this specific case, theP. aeruginosa sequence (http://www.pseudomonas.com).

Screening of STM mutants under aerobic versus anaerobic conditions. Approximately 140 microtiter dishes containing the entire P.aeruginosa PAO1 STM library were screened for the ability togrow under aerobic conditions, with each well containing eitherLB-100 mM NO₃^– or LB-15 mM NO₂^– at pH 6.5 for 24h at 37°C. Anaerobic NO₃^–- and NO₂^–-grown cultureswere grown in a Coy anaerobic chamber for 48 h (NO₃^–)and 96 h (NO₂^–). The criterion for a lack of growth indicatedas a negative sign in Table 8 is based upon little or no changein optical density of the control anaerobic mutant, the ${Delta}$ anrmutant, which is known to be incapable of anaerobic growth viadenitrification or arginine substrate-level phosphorylation(65).

View this table:
[in this window]
[in a new window]

TABLE 8. Growth of P. aeruginosa STM mutants and wild-type PAO1 in NO₃- and NO₂-containing medium with or without O₂

RESULTS

Top

RESULTS

Anaerobic versus aerobic expression of P. aeruginosa proteins, determined using 2-D gel and MALDI-TOF identification. We first assessed aerobic versus anaerobic protein expressionin P. aeruginosa via 2-D gel electrophoresis followed by MALDI-TOFMS identification of candidate protein spots. We used mediastrategically buffered at pH 6.5, a value representing the slightlyacidic pH of the mucus lining the CF airways (15, 61). Imagesof 2-D gels derived from cell extracts obtained from all aerobicallyand anaerobically grown organisms in L broth with NO₃^–and NO₂^– were compared and analyzed. On average, >1,200protein spots were detected per 2-D image. When 2-D images ofP. aeruginosa grown in the presence of NO₃^– and oxygen(Fig. 1A and C) were compared to those obtained in the absenceof oxygen (Fig. 1B and D), 68 proteins were significantly moreexpressed with nitrate in the absence of oxygen and 71 wereless expressed. Furthermore, >100 de novo synthesized proteinspots were detected anaerobically with NO₃^– that wereabsent in images obtained from organisms grown in the presenceof NO₃^– and oxygen. When cell extracts from NO₂^–-grownP. aeruginosa with and without oxygen were compared, >140proteins were more abundant with oxygen and >50 proteinswere more abundant without oxygen (Fig. 1B and D). In addition,several de novo synthesized proteins were detected. Comparisonof 2-D images obtained from P. aeruginosa grown anaerobicallyin the presence of NO₃^– or NO₂^– also revealed differentialprotein production. Overall, >170 proteins were induced atleast fourfold in the presence of NO₂^– compared to thosein the presence of NO₃^– under anaerobic conditions. Incontrast, only 30 proteins appeared to be more abundant in thepresence of NO₃^–, primarily in the class of metabolicenzymes (e.g., pyruvate dehydrogenase [PDH], aconitate hydratase2, L-cysteine desulfurase, N-succinylglutamase 5-semialdehydedehydrogenase, trigger factor, glutamate-asparaginase, hypotheticalprotein, and elongation factors G, P, Ts, and Tu) (Table 1).In contrast, a total of nine proteins whose production was lessor undetectable under aerobic conditions were identified. Theseproteins included ATP synthase β-chain, ribonucleosidereductase, flagellin type B, a probable transcriptional regulator(PA4902), a probable peroxidase, an organic solvent toleranceprotein (OstA), two hypothetical proteins, and NarH, a componentrespiratory nitrate reductase that donates electrons to NarG.Interestingly, two isoforms of NarH were detected that differedwith respect to their isoelectric point and production profile.One isoform was detected only in 2-D images of bacteria thatwere grown under anaerobic growth conditions in the presenceof NO₃^–, while the other NarH isoform was expressed at $~$ 16-fold higher levels under anaerobic conditions in the presenceof NO₃^– than in the presence of NO₂^–. Furthermore,five proteins were identified that were absent in the presenceof NO₃^–, including a probable transcriptional regulator(PA4902), leucine dehydrogenase, and three hypothetical proteins(Table 1).

View larger version (72K):
[in this window]
[in a new window]

FIG. 1. 2-D gels of aerobically (A and C) and anaerobically (B and D) expressed P. aeruginosa proteins. Cell extracts (30 µg) derived from the same growth phase (see Fig. S1 in the supplemental material) grown with NO₃^– (A and B)- or NO₂^– (C and D)-containing medium were separated by 2-D gel electrophoresis and stained with silver nitrate. The numbered spots correspond to proteins identified by MALDI-TOF analysis (Table 1).

View this table:
[in this window]
[in a new window]

TABLE 1. MALDI-TOF MS analysis of aerobic versus anaerobic proteins identified from 2-D gels, in order of PA number from www.pseudomonas.com

Under anaerobic conditions, several proteins were not detectedin the presence of NO₃^– (a probable transcriptional regulator,leucine dehydrogenase, and three hypothetical proteins), whilesome others were not produced in the presence of NO₂^–.The latter included the organic solvent tolerance protein OstAand the nitrate reductase subunit NarH. However, the isoformof NarH was detectable under both growth conditions, but proteinexpression was greater in the presence of NO₃^– than inthe presence of NO₂^–. A similar protein production profilewas observed for the rod shape-determining protein MreB, ribonucleosidereductase, and a hypothetical protein. Proteins that were producedin greater abundance under anaerobic growth conditions in thepresence of NO₂^– included flagellin type B, elongationfactors Ts and Tu, N-succinylglutamate 5-semialdehyde dehydrogenase,a probable DNA-binding stress protein, trigger factor, a probableperoxidase, glutamate-asparaginase, and two hypothetical proteins.In contrast, those proteins that were demonstrably lower inabundance in the presence of NO₂^– were MreB (rod shape-determiningprotein), ribonucleotide reductase (NrdA), the outer membraneprotein OprF, succinate dehydrogenase (SdhA), and another putativeribonucleotide reductase (NrdJb), encoded by PA5496. Finally,proteins that were detected in NO₃^–-grown bacteria butnot in NO₂^–-grown organisms were NarH (respiratory NAR),OstA (organic solvent tolerance protein), and isocitrate dehydrogenase(Icd).

Identification of differentially expressed peptides/proteins using nHPLC-µESI MS and MS/MS. A second, more sensitive MS approach coupling nHPLC-µESIMS and MS/MS was employed to improve on our initial 2-D gel/MALDI-TOFanalyses. Because of the time and effort required for theseexperiments, we focused on protein expression that was mostdramatically upregulated anaerobically only in the presenceof NO₃^–. Cell extracts from aerobically and anaerobicallygrown P. aeruginosa cultures were reduced, alkylated, and treatedwith trypsin, in solution, to generate peptide fragments. Thesesamples were then subjected to nHPLC-ESI MS analysis on an FT-ICRmass spectrometer to determine the peptide m/z ratios. Figure2A and B show the chromatograms from the two replicate analysesof the anaerobic samples and demonstrate the complexity of thesamples as well as the reproducibility of the FT-ICR analysis.A comparison of the peak areas for a tryptic peptide (AQAAEIVEQAK;m/z 579.31) from the constitutively expressed ATP synthase β-subunitrevealed that the level of this peptide in every sample didnot differ by more than a factor of 4 (Fig. 3). Using this information,a "differentially expressed peptide" was defined as a peptidewhose peak area increased by at least a factor of 10 in theanaerobic sample relative to each of the aerobic samples. Onepeptide peak (m/z 488.77) meeting the criteria for being considereddifferentially expressed is shown in Fig. 3. Using in-housesoftware, the peptide masses from all of the FT-ICR analyseswere deconvoluted to their +1 monoisotopic masses and comparedto one another. As expected, the comparisons between similarsamples (both anaerobic or both aerobic samples) revealed only370 differences, on average, while the comparisons between dissimilarsamples revealed over 1,360 differentially expressed peptides,on average. Further analysis of these comparisons yielded alist of 489 differentially expressed peptide masses that appearedconsistently in each aerobic-anaerobic analysis. Utilizing 120of these highly accurate mass measurements as target valuesin an nHPLC-µESI MS/MS analysis on an ion-trap instrumentallowed the sequence for each of these peptide m/z ratios tobe determined. The MS/MS spectrum of the differentially expressedpeptide peak shown in Fig. 3 (m/z 488.77) is shown in Fig. 4.As depicted, these data were used to assign the sequence VVQPEYNKto this m/z ratio, revealing that this differentially expressedpeptide peak was derived from NirS (also known as cytochromecd₁ or nitrite reductase [NIR]). A list of all differentiallyexpressed proteins obtained by nHPLC-µESI MS/MS is shownin Table 2. We found that the most commonly identified peptideswere those derived from NirS. In fact, there were 13 "hits"of NirS alone, with scores ranging from 2 (++) to 4 (++++) (from10- to 1,000-fold induction). The next most abundant peptideswere those derived from ArcA, ArcB, NosZ, NrdB, and cobalamin.Recall that ArcA and ArcB are required for anaerobic argininesubstrate-level phosphorylation (20), NosZ is a regulator involvedin the disposal of anaerobically produced N₂O (5), NrdB is partof the class II ribonucleotide reductase complex that is requiredfor anaerobic growth (27, 56), and cobalamin, too, has beenpurported to be required for this process (44). Some paradoxicalresults included enhanced anaerobic production of the majorcatalase KatA (PA4236) (32) and an alkyl hydroperoxide reductase,AhpCF (PA0139-140) (39), under anaerobic conditions. Finally,the magnitude of upregulation of the anaerobically induced proteinswas consistent, in most cases, with the transcriptional profilingexperiments that are discussed below. Finally, using both MALDI-TOFand nHPLC-µESI MS/MS techniques, a synopsis of proteinswas assembled in Table 3 for comparative purposes. Note thatfar more peptides were identified using the sensitive nHPLC-µESIMS/MS system. However, some were identified only using MALDI-TOFMS. It should be noted here that MALDI-TOF data are more quantitative,while LC-MS data are far more sensitive. Therefore, for comparativepurposes, it was important to compare both methodologies toexpose the most consistent findings using both techniques.

View larger version (10K):
[in this window]
[in a new window]

FIG. 2. Chromatograms from FT-ICR analysis of anaerobic P. aeruginosa digests. Total ion chromatograms 1 (A) and 2 (B) from anaerobic FT-ICR analysis are shown. Note the complexity as well as the reproducibility of the chromatography samples between replicate analyses.

View larger version (25K):
[in this window]
[in a new window]

FIG. 3. Determination of differentially expressed peptide peak areas. The peak at m/z 579.31, corresponding to the constitutively expressed tryptic peptide AQAAEIVEQAK from the P. aeruginosa ATP synthase β-subunit, was observed at approximately the same level in each of the four analyses (anaerobic [A and B] and aerobic [C and D]). Since the peak areas varied maximally by a factor of 4, a 10-fold change in area was used in the subsequent determination of differentially expressed m/z values. The peak shown at m/z 488.77 represents one such differentially expressed peptide m/z ratio. Note the consistency of the chromatographic elution profiles that easily permits alignment of the proper peaks.

View larger version (12K):
[in this window]
[in a new window]

FIG. 4. MS/MS spectrum of the differentially expressed peptide VVQPEYNK at m/z 488.77. Predicted masses for the ions of type b and y are shown above and below the sequence, respectively. Ions observed in the spectrum are underlined.

View this table:
[in this window]
[in a new window]

TABLE 2. Identification of peptides by nHPLC-µESI MS, using both ion-trap and FT-ICR MS, in anaerobically grown P. aeruginosa in the presence of NO₃^–

View this table:
[in this window]
[in a new window]

TABLE 3. Summary of MALDI-TOF and LC-MS analyses of differentially expressed proteins

Microarray analysis of P. aeruginosa anaerobic gene expression: effect of NO₃^– versus NO₂^–. Transcriptional profiling experiments were initiated using P.aeruginosa PAO1 grown to the same phase in LB-NO₃^– orLB-NO₂^– under aerobic versus anaerobic conditions. Organismswere harvested at the same optical density for isolation ofRNA as described in Materials and Methods. Because of the shearmass of data collected in these experiments, all genes, separatedby (i) an arbitrary level of induction or repression, (ii) theirputative gene products, and (iii) whether they were induced,as assessed by nHPLC-µESI MS/MS (given as "LC-MS" is tables),are provided in Tables S1 to S4 in the supplemental material.For this paper, we elected to simplify the data by presentingit in tabular form and by curated PseudoCyc metabolic pathwaysfor ease of interpretation.

(i) Genes regulated by NO₃^– (anaerobically induced). As shown in Table 4 and Table S1 in the supplemental material,the genes most activated by anaerobic relative to aerobic growthare those that would be predicted to be involved in classicalanaerobic NO₃^– reduction. According to an analysis ofcurated PseudoCyc metabolic pathways, disproportionately moreof the activated genes are involved in the processes of denitrification(P = 0.0013) and heme d₁ biosynthesis (P = 0.014) than wouldbe expected. Of the top 100 upregulated genes, the most inducedclass of anaerobic respiratory genes are norCBD, encoding subunitsof the protective NOR. NOR functions to detoxify potentiallyharmful NO during anaerobic respiration in P. aeruginosa (63).Not surprisingly, these genes are conveniently localized ina predicted operon on the P. aeruginosa genome (Fig. 5). Thenext most extensively represented group of genes include anoperon involved in biosynthesis of the respiratory NO₃^–reductase complex, narK1-narK2-narGHJI (PA3871). The narK1 andnarK2 genes encode extrusion pumps to rid the cell of potentiallytoxic levels of NO₂^– (49). PA3871 encodes a putative probablepeptidyl-prolyl cis-trans isomerase, and moaA1 encodes a molybdenumcofactor biosynthesis protein. Molybdenum cofactor biosynthesisproteins are required for NO₃^– reduction in P. aeruginosa(38). Another predicted operon that was highly transcribed andinvolved in the reduction of NO₂^– was the nirSMCFDLGHJENoperon. Activation of this operon required, among others, thetranscriptional activator NirQ (PA0520), whose gene expressionwas also upregulated but was below the 30-fold induction cutoffused for Table S1 in the supplemental material. The final geneclass involved in the denitrification pathway that was in thetop 100 most activated genes was nosRZDFYL, encoding membersof the nitrous oxide reductase enzyme and regulators. Interestingly,the class of genes that we did not expect to be upregulateddramatically were bacteriophage-related genes. In fact, withinthe top 100 most activated genes, 44 were related to bacteriophageproduction. Most of these genes, encompassing PA0613 throughPA0648, are localized between trpG and trpE on the chromosome.However, transcription of another set of phage-related genes,PA0717 to PA0729, was also dramatically induced.

View this table:
[in this window]
[in a new window]

TABLE 4. Genes activated in NO₃^–-containing PseudoCyc metabolic pathways^a

View larger version (36K):
[in this window]
[in a new window]

FIG. 5. Schematic depiction of genes whose transcription is either activated anaerobically (x-fold activation [in green]) or repressed/not activated (in red) in the presence of NO₃^– or NO₂^–. The boxes that are empty represent those whose cutoff values were below the designated detection limit of Tables S1 to S4 in the supplemental material. The figure includes genes involved in the transcription of nar genes (A), nir/nor genes (B), NADH dehydrogenase genes (nuo) (C), and two different predicted operons encoding PF1 bacteriophages (D and E) and TCA cycle genes that were repressed when organisms were grown in the presence of NO₂^– (in red) (F).

(ii) Genes regulated by NO₂^– (anaerobically induced). We next examined gene expression by P. aeruginosa grown aerobicallyversus anaerobically in 15 mM NO₂^– at pH 6.5. By the sameanalysis using curated PseudoCyc metabolic pathways, disproportionatelymore of these activated genes are involved in denitrification(P = 0.000060) and biosynthesis of heme d₁' (P = 0.00070) (Table5; see Table S2 in the supplemental material), similar to thecase for NO₃^–-grown cells. Of particular note, all ninegenes in the genome classified as being involved in biosynthesisof heme d₁' were identified in this study. Interestingly, similarto the case for anaerobic NO₃^–-grown organisms, 32 ofthe top 100 most activated genes were genes involved in bacteriophageproduction. However, unlike the two bacteriophage classes observedwith the NO₃^–-grown bacteria discussed above, only thegenes from the PA0612-to-PA0648 operon were activated. Similarto the case for NO₃^–-grown bacteria, the nirSMCFDLGHJEN,norCBD, and nosRZDFYL operons were found to be activated. ThearcDABC operon was significantly more activated in NO₂^–-grownbacteria than in NO₃^–-grown organisms. One noticeabledifference between NO₃^–- and NO₂^–-induced gene expressionduring anaerobic growth was overexpression of the adhA genein NO₂^–-grown cells but not in NO₃^–-grown organisms.

View this table:
[in this window]
[in a new window]

TABLE 5. Genes activated in NO₂^–-containing medium^a

(iii) Genes regulated by NO₃^– (anaerobically repressed). We next examined genes that were repressed by anaerobic growthin NO₃^–-containing medium. According to a COG-based analysis,significantly more of these repressed genes are classified asbeing involved in amino acid transport and metabolism than wouldbe expected (P = 0.0069) (Table 6; see Table S3 in the supplementalmaterial). Of note, the identified genes were classified inmany PseudoCyc amino acid metabolic pathways, supporting theseCOG findings. For example, the most downregulated genes werethose involved in branched-chain amino acid transport, i.e.,PA1070 to PA1074. However, none of these specific PseudoCycpathway categories reached statistical significance. Among themost logically repressed genes were those encoding two dioxygenases,homogentisate-1,2-dioxygenase (PA2009) and 4-hydroxyphenylpyruvatedioxygenase (PA0865), both of which are members of the tyrosinedegradation pathway and dependent upon the presence of oxygen(63). Two other gene members of the tyrosine degradation pathway,maleylacetoacetate isomerase (PA2007) and fumarylacetoacetase(PA2008), were also downregulated. The oxidative stress genekatB, which is only responsive to aerobic H₂O₂ in an OxyR-dependentfashion (39), was also repressed 7.5-fold.

View this table:
[in this window]
[in a new window]

TABLE 6. Genes repressed in NO₃^–-containing medium^a

(iv) Genes regulated by NO₂^– (anaerobically repressed). Table 7 and Table S4 in the supplemental material indicate thata disproportionately high level of repressed genes are classifiedby COG as being involved in energy production and conversion(P = 3.18 x 10^–8) and in translation, ribosomal structure,and biogenesis (P = 0.025). According to PseudoCyc analysis,a disproportionately higher level of repressed genes are involvedin the trichloroacetic acid (TCA) cycle (P = 0.0019) than wouldbe expected, supporting the above COG data that significantlymore of these repressed genes are involved in energy productionand conversion. Interestingly, the entire postglycolytic metabolicmachinery, including members of the PDH complex, the TCA cycle,and the electron transport (oxidative phosphorylation) cascade,appears to be repressed significantly. For example, the aceEFgenes (PA5015 and PA5016) encode the PDH and dihydrolipoamidedehydrogenase components of the PDH complex. Genes encodingTCA cycle enzymes, including sucCD (PA1588 and PA1589; succinylcoenzyme A [CoA] synthase), icd (PA2623; isocitrate dehydrogenase),gltA (PA1580; citrate synthase), acnB (PA1787; aconitase), andsdhAC (PA1583; succinate dehydrogenase), were all downregulated.Finally, the majority of the nuo class of genes, encoding subunitsof the NADH dehydrogenase (complex I) of the electron transportchain, were also substantially repressed. Two putative cytochomeoxidase components (PA1856 and PA1553) were downregulated 14-and 11-fold, respectively. Interestingly, the anr gene (PA1544),encoding the FNR/CRP-like transactivator ANR that is absolutelyrequired for anaerobic growth of P. aeruginosa (65), was alsorepressed in NO₂^–-containing medium. This essentiallyindicates that the entire TCA cycle and electron transport chainare repressed. Potential reasons for decreased expression ofthese genes are offered in Discussion. Also, note that manygenes involved in translation, ribosomal structure, and biogenesiswere disproportionately repressed but, similar to the case inTable 5, were distributed relatively evenly among multiple categories,and no specific PseudoCyc annotated pathway reached statisticalsignificance.

View this table:
[in this window]
[in a new window]

TABLE 7. Genes repressed in NO₂^–-containing medium

Simplified synopsis of transcriptional profiling results from an operonic perspective. Given the shear mass of data collected in our transcriptionalprofiling experiments, we next elected to select predicted operonsin the P. aeruginosa genome that collectively were either up-or downregulated for the purpose of consistency. These operonswere downloaded from the P. aeruginosa genome website (www.pseudomonas.com),and genes that were activated (green numbers) or repressed (rednumbers) are boxed in Fig. 5. Figure 5A indicates that the narK1-narK2-narGHJI(PA3871)-moaA1 operon was upregulated 37- to 72-fold in anaerobicNO₃^– but not NO₂^– culture. Surprisingly, the narX-narLgenes, encoding a two-component regulatory system that was recentlyshown to be required for activation of narK1-narK2-narGHJI (PA3871)-moaA1in cooperation with the second-tier regulator DNR (48), werenot upregulated during anaerobic NO₃^– growth. Unlike thegenes encoding the NAR machinery, several, but not all, genesinvolved in production of NIR and NOR activity were significantly(between 32- and 244-fold) induced by both anaerobic NO₃^–and NO₂^– growth (Fig. 5B). NO₂^–, but not NO₃^–,was found to significantly repress the nuo genes encoding theNADH dehydrogenase complex (complex I) of the respiratory chain(Fig. 5C). One of two major gene regions on the P. aeruginosachromosome encoding bacteriophage, from PA0713 to PA0729, wasupregulated only by NO₂^– (Fig. 5D). This was in contrastto the very large chromosomal region encompassing PA0613 toPA0648, which was activated by both NO₃^– and NO₂^–(Fig. 5E). Finally, anaerobic NO₂^– growth significantlydownregulated two PDH genes, aceE and aceF, and eight genesencoding components of five enzymes within the TCA cycle (Fig.5F).

Transposon library screen for genes required for anaerobic growth using NO₃^– or NO₂^– as a terminal electron acceptor. An STM library comprised of 7,968 mutants (43) was next screenedfor mutants that could not grow anaerobically using NO₃^–or NO₂^– as a terminal electron acceptor. As shown in Table8, the [4Fe-4S]²⁺ transactivator ANR and its downstream regulator,DNR, were required for anaerobic growth using both NO₃^–and NO₂^–. Similar mutants harboring transposons withinthe cysG (encoding siroheme synthase), nuoD, nuoG, nuoM (encodingcomponents of the NADH dehydrogenase complex), and dskA (encodinga C₄ TraR zinc finger protein) genes were isolated. The narI(encoding the membrane-binding ${gamma}$ component of NAR) and narK2(encoding NO₂^– extrusion pump 2) genes were required foranaerobic growth using NO₃^– but not NO₂^–. No mutantswere identified that could grow anaerobically with NO₃^–but not with NO₂^–.

DISCUSSION

Top

DISCUSSION

Although many processes in humans can occur under anaerobicconditions (e.g., glycolysis during vigorous aerobic exercise),humans require oxygen for life, and oxygen limitation or excesscan damage human and even bacterial cells (25). In contrast,P. aeruginosa can easily survive and even thrive without oxygenif it is provided with NO₃^–, NO₂^–, and to a lesserextent, arginine. Arguably, proteins and, most importantly,enzymes that are essential for survival or growth under suchconditions could represent targets for novel therapeutic interventionstrategies. Importantly, the treatment of highly refractoryP. aeruginosa biofilms in CF airway disease has been a sourceof immense frustration for both basic and clinical scientistssince one initial feature of the disease is manifested, in part,as pancreatic insufficiency (17). In 2002, Worlitszch et al.(60) and our research group (62) discovered that P. aeruginosagrowing in the thickened CF airway mucus is capable of growthunder reduced oxygen tension and even anaerobic conditions,especially during chronic, end-stage CF airway disease. Giventhat humans cannot undergo anaerobic respiration, this studywas initiated with the goal of identifying P. aeruginosa genesand gene products that are modulated during anaerobic growthby NO₃^– or NO₂^– respiration specifically at theslightly acidic pH ( $~$ 6.5) of the CF airway mucus. Given thathighly advanced techniques and diverse expertise were availableto our collaborative research groups, we elected to use a combinedand exhaustive proteomic, microarray, and STM mutagenesis approachto identify potential therapeutic targets for intervention ofthe anaerobic biofilm mode of P. aeruginosa.

Differences in anaerobic growth patterns between NO₃^– and NO₂^– growth. (i) NO₃^– or NO₂^– reduction machinery and arginine substrate-level phosphorylation. For anaerobic respiration using NO₃^– or NO₂^–, P.aeruginosa has dissimilatory NAR, NIR, NOR, and nitric oxidesynthase activities for the purpose of ATP generation. Duringboth our proteomic and microarray studies, we found clear evidenceof gene and protein expression patterns that are consistentwith some but not all members of the denitrification pathwaybeing activated during anaerobic but not aerobic growth withNO₃^– as well as with NO₂^–. Predictably, the norCBDgenes, encoding the protective enzyme NOR, were induced underboth conditions. We have recently shown that a P. aeruginosamutant lacking NOR (e.g., norCB) grows abysmally slowly duringanaerobic growth but does not die (63). This is due to an eleganttwo-pronged mechanism. It involves (i) an NO-mediated inactivationof the master anaerobic regulator ANR coupled with (ii) overexpressionof two paradoxically oxygen-dependent gene products that aretypically the most repressed proteins, namely, homogentisate-1,2-dioxygenase(HmgA) and 4-hydroxyphenylpyruvate dioxygenase (Hpd), both ofwhich possess inherent NO-scavenging activity (63). Despitethe fact that only NO₃^– and NO₂^– were used as alternativeelectron acceptors in these studies, genes/proteins involvedin the arginine deiminase pathway, including arcA/arcB, werealso activated, especially transcriptionally, by NO₂^–(see Table S2 in the supplemental material), and proteomicallyusing LC-MS (Table 2), confirming previous results indicatingthat some of the master anaerobic regulators, including ANR,DNR, and ArgR, are fully operative as regulators of the argininedeiminase pathway (20, 31). Furthermore, our LC-MS proteomicdata indicated that NirS and Arc(ABC) represent some of themost frequently encountered proteins during anaerobic NO₃^–respiration (Table 2). The anaerobic transcriptional profilingresults of Filiatrault et al. (19) also confirm the activationof arcA and arcB (and arcD) by NO₃^– alone. The transcriptionof the various nar genes was elegantly shown to be under thecontrol of DNR and NarX/L (4, 48), both of which require activationby the global transcriptional activator ANR by oxygen limitation(48). In contrast, DNR, NarX/L, and NirQ each participate inthe regulation of nir, nor, and nos genes (3, 48).

(ii) Somewhat unexpected findings. Why are genes and proteins involved in postglycolytic energymetabolism repressed during anaerobic NO₂^– respiration?The mucus lining the CF airways was recently found to be slightlyacidic (pH $~$ 6.5) (61), and such conditions allow a product ofanaerobic NO₃^– respiratory metabolism, namely, NO₂^–,to be reduced nonenzymatically to NO. The elegance of this findingis that the formidable mucA mutant mucoid strains of P. aeruginosa,which are resistant to antibiotics and phagocytes, are exquisitelysusceptible to anaerobic NO (61). This was recently shown byYoon et al. (61) and offers some hope for eradication of themucoid form of P. aeruginosa from the CF airways. One of themajor frontline antibiotics used to treat P. aeruginosa biofilminfections is the aminoglycoside tobramycin. The efficacy oftobramycin is either markedly reduced or absent in the absenceof oxygen (11, 13, 37, 54), yet a recent study indicates thatbiofilm P. aeruginosa is more susceptible to tobramycin whencultures are amended with either nitrate or arginine (10). Ourstudy reveals that the entire postglycolytic metabolism of P.aeruginosa is shut down under these conditions. For example,in the presence of NO₂^–, PDH, five of eight genes encodingTCA cycle enzymes, and various members of the NADH dehydrogenasecomplex are downregulated (Fig. 5F). The potential rationalebehind these metabolic events is as follows. Two enzymes, inparticular (aconitase and fumarase), are strategically positionedimmediately prior to the production of TCA cycle enzymes (isocitratedehydrogenase and ${alpha}$ -ketoglutarate dehydrogenase) that generatereducing power in the form of NADH. Both aconitase and fumarasehave been shown to be susceptible to inactivation not only byaerobic free radicals, such as the superoxide anion (O₂^–)(22, 30), but also by NO (21). The inherent cellular logic forthese events is that the cell uses a "circuit-breaker" mechanismto compromise the activity of enzymes immediately precedingthose that generate NADH. The "circuit-breaker" term was coinedin the metabolic sense by Irwin Fridovich and colleagues atDuke University 16 years ago, after they revealed the O₂^–and NO sensitivity of these enzymes (21, 22). This allows cellsto survive while growing slowly in the presence of 15 mM NO₂^–(see Fig. S1 in the supplemental material), an amount that wasshown nearly 30 years ago to be the limit of toxicity for P.aeruginosa (59). This feature of NO-mediated inactivation ofcellular proteins is not unique to aconitase and fumarase. Infact, the ribonucleotide reductase NrdA is an iron-sulfur proteinthat is known to be poisoned by NO (D. J. Hassett, unpublishedresults).

Anaerobic bacteriophage gene activation. One of the most intriguing results was the discovery of a dramaticincrease in bacteriophage gene activation during anaerobic growthusing both NO₃^– and NO₂^– as terminal electron acceptors(see Tables S4 and S5 in the supplemental material). Specifically,of the top 62 most upregulated genes during anaerobic growthin the presence of NO₃^–, 32 (NO₃^–) and 33 (NO₂^–)were the phage genes PA613 to PA648, while 11 genes were ofthe PA713-to-PA729 cluster when cells were grown in the presenceof NO₃^–. Interestingly, PA0713, the first gene of thePA713-to-PA729 cluster, was the most upregulated gene in thepresence of NO₂^– (121-fold). These phage genes encodePf1-type (pilus/flagellum) proteins that are known to be upregulatedin biofilms (58). Because slightly acidified NO₂^– generatesNO, the latter gas could be a cellular signal indicating thatthe prophage incorporated into the bacterial genome is underattack and that it would behoove the phage to find a suitablehost that is not under such stress. Thus, it is not surprisingthat bacteriophages have been shown to cause death in P. aeruginosabiofilms (58). Relatedly, NO has also been shown to cause dispersalof P. aeruginosa from biofilms (6). We also have shown thatan absence of the rhl quorum-sensing circuit causes metabolicsuicide by overproduction of NO via anaerobic respiration inbiofilms and planktonic culture (62). In addition, we also showedthat slightly acidified NO₂^– actually kills the formidablemucoid, alginate-overproducing form that is highly resistantto conventional antibiotics and phagocytic killing (61). Thus,both the quorum-sensing mutants and the mucoid strains may perishby an as yet unappreciated mechanism involving activation ofbacteriophage(s), a hypothesis that certainly merits experimentaltesting.

Clearly, organisms exposed to NO₂^– during anaerobic growthexhibited the most revealing metabolic pattern with respectto the therapeutic potential of acidified NO₂^–. Remarkably,the organism appeared to be shutting down the production ofreducing power in the form of NADH and FADH from a dramaticallyreduced PDH complex (aceEF genes), a reduced TCA cycle (gltA,acnB, icd, sucABCD, and sdhA), and also reduced ATP generationby a reduction in NADH dehydrogenase activity (nuoBDEFHIJKMgenes). In fact, even two genes encoding putative cytochromeoxidases (PA1553 and PA1856) were the most repressed genes duringanaerobic growth in the presence of NO₂^– at pH 6.5. Inreducing the proton motive force by reduction of NADH dehydrogenaseI activity, there would be reduced transport of NO₂^– intoP. aeruginosa. With reduced NO₂^– transport coupled witha reduction in cellular NADH because of the likely poisoningof P. aeruginosa aconitase by NO (21), the organism grows poorly.As such, we have shown that acidified NO₂^– at pH 6.5 canact almost in a bacteriostatic fashion (61). These results areconsistent with those of Wagner et al. (57) showing that membersof the NADH dehydrogenase complex (nuo genes) were absolutelyrequired for anaerobic growth in the presence of NO₃^–.However, two new genes that were found to be required for anaerobicgrowth in the presence of both NO₃^– and NO₂^– werecompletely unexpected. One of these genes, dksA, controls theintercellular signaling process of quorum sensing in P. aeruginosa(12, 28). It is for this reason that a dksA mutant is not ableto perform quorum sensing, and perhaps its inability to growin NO₃^– or NO₂^– medium may be a function of a defectsimilar to the metabolic overproduction of NO akin to anaerobicrhlR mutant bacteria (62). Finally, the multifunction productof the cysG gene, siroheme synthase, was shown by Zumft andcolleagues (56) to be essential for incorporation of protoporphyrinIX into heme (9). Since heme is required for many enzymes involvedin the anaerobic respiratory pathway, these results are consistentwith their cellular function.

Limitations of this study. (i) Differences in growth rate and cell yield between NO₃^–- and NO₂^–-grown bacteria. Recall that the growth characteristics of P. aeruginosa in NO₃^–versus NO₂^– medium are very different. The anaerobic growthrate of P. aeruginosa in NO₂^– medium is significantlylower than that in NO₃^– medium, and the organisms do notachieve the same maximal cell density (see Fig. S1 in the supplementalmaterial). NO₂^– at concentrations approaching 20 mM arenot well tolerated by P. aeruginosa, and the cells cannot generatethe ATP of NO₃^–-respiring cells because they have no substrate(NO₃^–) for the NAR (proton-pumping) coupling step. Thesedifferences in growth rate could provide an alternative explanationfor some of the differences in gene expression. Bacteria thatare growing more rapidly may increase the expression of genesin the central pathways (e.g., PDH complex and TCA cycle), whichis essential for synthesis of many cellular metabolites andstructures relating to anaplerotic functions. This may alsobe the reason for differences in expression of genes for energyproduction and conversion as well as for translation, ribosomalstructure, and biogenesis.

(ii) Results derived from rich medium versus those derived from the host during infection. Despite the exhaustive and comprehensive nature of this study,we must stress caution in attempting to relate the significanceand potential correlates of our results to the behavior of P.aeruginosa in the CF airway and other disease settings. First,the proteome and transcriptome were determined in vitro andin complex growth media. Recently, a synthetic medium containingmany of the factors in CF sputum was used to monitor the behaviorof P. aeruginosa (41), and this medium may be useful for furtherstudies involving a similar research theme. The argument thatthe "proteogenome" of P. aeruginosa in situ is identical oreven similar to the results of these work conditions would beconsidered tenuous at best. However, we would also argue thatthe basic challenge of a slightly decreased pH ( $~$ 6.5) (61) andanaerobic or oxygen-limiting conditions (2, 60) would in allprobability require similar regulation of cellular pathwaysirrespective of additional environmental stress that would ensuein an actual infection. Finally, we must also note that theP. aeruginosa research community almost always uses strain PAO1.This strain can behave very differently from laboratory to laboratory,and while the vast majority of results using this strain arehighly reproducible, some differences may occur.

In conclusion, we have accomplished comprehensive (i) MALDI-TOF/LC-MSproteomic analysis, (ii) transcriptional profiling using GeneChipmicroarrays, and finally, (iii) STM analysis of genes that arerequired for anaerobic growth of P. aeruginosa at pH 6.5. Wehope that these studies provide a partial framework for futureinvestigations to unravel new clues as to mechanisms by whichwe might help to eradicate this notorious pathogen during thecourse of various disease states.

ACKNOWLEDGMENTS

This work was supported in part by Public Health Service grantsGM-69845 (D.J.H.) and GM-37537 (D.F.H.), Cystic Fibrosis Foundationgrants HASSETT07G0 and R457-CR01 (D.J.H.), a Cure Finders grant(D.J.H.), the National Science Foundation 0311307 (K.S.), NationalInstitutes of Health grant HL-073835 (K.S.), The Canadian Institutefor Health Research (R.C.L.), and Cystic Fibrosis Foundationpilot funds.

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular Genetics, Biochemistry and Microbiology, University of Cincinnati College of Medicine, 231 Albert Sabin Way, Cincinnati, OH 45267-0524. Phone: (513) 558-1154. Fax: (513) 558-8474. E-mail: Daniel.Hassett{at}UC.Edu

Published ahead of print on 18 January 2008.

Supplemental material for this article may be found at http://jb.asm.org/.

REFERENCES

Top