Quorum Quenching of Nitrobacter winogradskyi Suggests that Quorum Sensing Regulates Fluxes of Nitrogen Oxide(s) during Nitrification

ABSTRACT Quorum sensing (QS) is a widespread process in bacteria used to coordinate gene expression with cell density, diffusion dynamics, and spatial distribution through the production of diffusible chemical signals. To date, most studies on QS have focused on model bacteria that are amenable to genetic manipulation and capable of high growth rates, but many environmentally important bacteria have been overlooked. For example, representatives of proteobacteria that participate in nitrification, the aerobic oxidation of ammonia to nitrate via nitrite, produce QS signals called acyl-homoserine lactones (AHLs). Nitrification emits nitrogen oxide gases (NO, NO2, and N2O), which are potentially hazardous compounds that contribute to global warming. Despite considerable interest in nitrification, the purpose of QS in the physiology/ecology of nitrifying bacteria is poorly understood. Through a quorum quenching approach, we investigated the role of QS in a well-studied AHL-producing nitrite oxidizer, Nitrobacter winogradskyi. We added a recombinant AiiA lactonase to N. winogradskyi cultures to degrade AHLs to prevent their accumulation and to induce a QS-negative phenotype and then used mRNA sequencing (mRNA-Seq) to identify putative QS-controlled genes. Our transcriptome analysis showed that expression of nirK and nirK cluster genes (ncgABC) increased up to 19.9-fold under QS-proficient conditions (minus active lactonase). These data led to us to query if QS influenced nitrogen oxide gas fluxes in N. winogradskyi. Production and consumption of NOx increased and production of N2O decreased under QS-proficient conditions. Quorum quenching transcriptome approaches have broad potential to identify QS-controlled genes and phenotypes in organisms that are not genetically tractable.

During nitrification, diverse genera of chemolithotrophic bacteria and/or archaea oxidize ammonia (NH 3 ) to nitrite (NO 2 Ϫ ) and then to nitrate (NO 3 Ϫ ) (11)(12)(13)(14). Generally, NH 3 is oxidized to NO 2 Ϫ by ammonia oxidizers, including bacteria (AOB) and archaea (AOA), while NO 2 Ϫ is oxidized to NO 3 Ϫ by nitriteoxidizing bacteria (NOB). Recently, complete oxidation of NH 3 to NO 3 Ϫ (comammox) was identified in representatives of the genus Nitrospira, a group previously characterized as NOB (15,16). Nitrification is a key part of the nitrogen cycle in natural, agricultural, and industrial systems and is a contributor to gas emissions of nitric oxide (NO), nitrogen oxides (NO x ), and nitrous oxide (N 2 O), which are hazardous gases that contribute to global warming (11,17).
In many proteobacteria, QS is accomplished through the production of acyl-homoserine lactone (AHL) signaling compounds or autoinducers (1). AHLs represent the best-studied class of autoinducer, and they are generally produced by a LuxI homolog autoinducer synthase and detected by a LuxR homolog signal receptor-transcriptional regulator (1). The luxI and luxR genes are commonly located adjacent to each other in the genome and are generally positively autoregulated (1).
One method to study AHL QS is by specifically inactivating all AHL autoinducers through the use of recombinant lactonase to promote a QS-deficient phenotype (18,19). AiiA, an AHL lactonase identified from Bacillus spp., is a well-characterized enzyme that specifically hydrolyzes the homoserine lactone (HSL) ring of AHLs, regardless of the chain length of the acyl group or other moiety (20,21). So-called "quorum quenching" approaches have been implemented through both heterologous expression in a host of interest and addition of purified AHL lactonase (18,20,(22)(23)(24). In this study, we used purified AiiA lactonase to identify QS-controlled gene expression and phenotypes in Nitrobacter winogradskyi, a well-characterized NOB that is currently not genetically tractable.
The genus Nitrobacter consists of a ubiquitous group of NOB in the family Bradyrhizobiaceae isolated from soil, water, and wastewater treatment systems (13,14). N. winogradskyi is a wellstudied example of NOB due to its superior growth rate and growth yield compared to other NOB, and it was the first NOB shown to produce AHLs (10,13,14). In addition, the genome of N. winogradskyi has been sequenced and this bacterium has been the subject of recent global transcriptome studies (6,25,26). Expression of N. winogradskyi genes nwiI, encoding an autoinducer synthase, and nwiR, encoding a receptor-transcriptional regulator, was shown to be cell density dependent and to correlate with the AHL concentration in culture (10). The structure of the predominant AHL was identified as that of an unsaturated AHL with a 10 carbon acyl chain, C 10:1 -HSL (10). Nuclear magnetic resonance (NMR) spectroscopy analysis of AHL extracts produced via heterologous expression in Escherichia coli identified the isomeric form of C 10:1 -HSL (27) but suggested a location for the double bond that is different from that previously described (10). However, heterologous expression of autoinducer synthases in E. coli often produces AHLs that are different from those in the native strain (27,28).
Previous attempts have been made to identify QS-controlled phenotypes in N. winogradskyi (10,27). Mellbye et al. showed that the growth rate decreased as transcription of nwiI and nwiR in-creased and AHLs began to accumulate (10). Shen et al. observed up to a 2-fold increase or 5-fold decrease in the expression of select genes of the nitrite oxidoreductase (NXR) gene cluster after the addition of purified C 10:1 -HSL to cultures at saturating concentrations but did not observe any statistically significant phenotypic changes (27).
Here, we utilized a quorum quenching approach to identify both primary and secondary regulatory effects of AHL QS in N. winogradskyi. Using purified AiiA lactonase, AHLs were depleted from N. winogradskyi cultures and QS-controlled genes were identified through comprehensive mRNA sequencing (mRNA-Seq) analysis. Our transcriptome analysis showed that depletion of AHLs affected the expression of a significant percentage (52%) of the genetic inventory in N. winogradskyi and also suggested a link between QS and nitrogen oxide fluxes in this bacterium. Our experiments confirm a previous report that N. winogradskyi can produce N 2 O (29) and present new evidence that QS affects NO x fluxes. Our work demonstrates that AiiAmediated quorum quenching coupled with mRNA-Seq is a useful technique to identify QS-controlled genes and phenotypes in difficult-to-study organisms.

AiiA lactonase treatment of N. winogradskyi cultures depletes AHLs.
To determine the effect of QS inhibition in N. winogradskyi, we initiated and monitored three batch culture treatments: (i) AiiA lactonase treatment (QS-deficient), (ii) heat-denatured AiiA lactonase treatment (QS-proficient) (to determine if protein addition had an effect), and (iii) no-added-lactonase treatment (QS-proficient). Depending on cell density, approximately 0.28 or 0.71 g protein ml Ϫ1 was added to both lactonase and heatdenatured lactonase treatments daily (see Text S1 in the supplemental material). Although the treatments showed no significant differences in nitrite oxidation rate, growth rate, or growth yield (Fig. 1A), the addition of AiiA lactonase prevented the accumulation of bioassay-detectable AHL (Fig. 1B). Lactonase-treated and heat-denatured lactonase-treated cultures were harvested on day 3 during peak signal production as observed in our previous work (10) to collect RNA for mRNA-Seq (Fig. 1).
Transcriptome responses to QS inhibition. The transcriptome of N. winogradskyi under QS-deficient (AiiA-treated) conditions was compared to that present under QS-proficient (heatdenatured AiiA-treated) conditions. All changes in gene expression are expressed as the ratio of the number of transcripts seen under the QS-proficient treatment conditions to the number seen under QS-deficient treatment conditions. First, we validated our quorum quenching approach by noting an increase in the transcript abundance of the signal synthase nwiI gene and the signal receptor nwiR gene under QS-proficient conditions (Table 1). As previously noted, many bacterial QS genes, particularly the signal synthase gene, are autoregulated, creating a positivefeedback loop (1,10). In addition, levels of methionine biosynthesis transcripts increased up to 7.7-fold, possibly due to increased use of S-adenosyl methionine for AHL biosynthesis ( Table 1). The transcriptome analysis revealed 1,631 genes showing statistically significant changes in expression in QS-proficient cells, but many changes were Ͻ3-fold (see Table S1 and Dataset S1 in the supplemental material). In total, expression of 1,346 genes changed marginally and expression of 237 genes changed Ͼ3-fold between QSdeficient and QS-proficient conditions (see Dataset S1). Grouping the expression changes into clusters of orthologous groups (COG) functional categories, COGs associated with the process of translation as well as nucleotide, carbohydrate, and amino acid metabolism and transport substantially (Ͼ60% of category COGs) changed in expression (Fig. 2). We observed a similar trend whether an expression cutoff (e.g., 2-fold cutoff) was applied or not; thus, all of the data were included in the COG analysis.
Based on previous work that suggested that QS is autoregulated in N. winogradskyi (10), we used SCOPE to search for luxbox-like inverted repeat motifs upstream of nwiI and nwiR (see Text S1 and Fig. S2 in the supplemental material) (30). Two different motifs (motif A and motif B) were identified that suggested that QS directly activates nirK cluster genes, as well as the stringent response secondary messenger system mediated through GppA phosphatase, and expression of several other genes (see Fig. S1 and Table S2). Nitrite metabolism and signal transduction genes are induced under QS-proficient conditions. An in-depth scan of the QS transcriptome of N. winogradskyi showed that the largest changes in expression involved genes encoding proteins associated with biosynthetic metabolism, nitrogen metabolism, and signal transduction, particularly those associated with nitrite metabolism (Table 1). Under QS-proficient conditions, assimilatory nitrite reductase gene nirBD decreased in expression up to 9.3fold, suggesting imminent growth arrest and induction of the stringent response, as nitrite was the sole nitrogen source in the medium (Table 1). In addition, expression of nitrite reductase gene nirK increased 2.2-fold whereas expression of nirK cluster genes, including ncgABC, increased up to 19.9-fold under QSproficient conditions (Table 1). Furthermore, expression of Nwi0557, a homolog of nnrS, a putative NO-responsive membrane protein gene, increased 8.1-fold (Table 1). These changes suggest a possible link between QS effects on N biosynthesis metabolism and NO x metabolism.
QS-proficient conditions also changed the expression levels of several genes involved in signal transduction and flagellum biosynthesis. Guanine nucleotide secondary messenger (e.g., ci-di-GMP, ppGpp) biosynthesis and response genes, including genes encoding a diguanylate cyclase/phosphodiesterase homolog (Nwi0500), a Crp domain regulator (Nwi2061), a RelA/SpoT homolog (Nwi1922), and Ppx/GppA phosphatase (Nwi2151), increased in expression by 2.3-to 8.9-fold (Table 1). Fourteen genes associated with flagellum biosynthesis and assembly decreased in expression up to 2.6-fold (Table 1). That said, no obvious phenotype differences, such as changes in motility, biofilm, or aggregate formation, were observed under QS-proficient or -deficient conditions.
Quorum sensing in N. winogradskyi influences NO x fluxes. Following the transcriptome analysis prediction that QS affects nitrite metabolism through expression of nirK cluster genes, production and consumption of NO x gases by N. winogradskyi were measured. In order to observe the biggest difference in gas fluxes, cells were incubated at high cell density in sealed vials conducive to AHL accumulation. Preliminary tests detected abiotic NO x accumulation in the headspace above sterile nitrite-containing growth medium, most likely due to the aqueous chemical decomposition of protonated NO 2 Ϫ (nitrous acid [HNO 2 ; also known as HONO]) to NO and NO 2 , collectively referred to as NO x gases (31). Therefore, we included both heat-killed cells and sterile medium controls along with our QS-proficient and -deficient treatments in the experiments. NO x gas measurements were made from such suspensions of concentrated QS-proficient or -deficient N. winogradskyi cells during 24 h of NO 2 Ϫ oxidation. We predicted that an increase in expression of the nirK cluster genes under QSproficient conditions would either increase NO x production as earlier studies have suggested (32)(33)(34) or decrease NO x production by consuming NO as previously reported (34)(35)(36).
A statistically significant (P Ͻ 2 ϫ 10 Ϫ6 ) accumulation of NO x (measured as parts per billion [ppb] by volume) was registered for both QS-proficient and -deficient treatments (approximately 1,699 and 1,240 ppb, respectively) after 2 h of incubation, compared to the accumulations seen with medium alone and with heat-killed cell controls (approximately 87 and 213 ppb, respectively) (Fig. 3A). The peak NO x accumulation in both the QSproficient and -deficient treatments was transient and was followed by the disappearance of NO x as NO 2 Ϫ was consumed by N. winogradskyi (Fig. 3). In contrast, heat-killed cells and medium-alone controls slowly accumulated NO x in the headspace over time and NO 2 Ϫ concentrations did not change significantly (Fig. 3). The pH of N. winogradskyi cultures and controls did not change significantly during the experiments (data not shown). Although these data suggest that some abiotic NO x accumulation occurred in the controls without N. winogradskyi cells, transient NO x production occurring during active NO 2 Ϫ oxidation by N. winogradskyi was significantly greater than that seen with the controls. QS-proficient cells both produced and consumed NO x at significantly greater rates (P Ͻ 0.003) than AiiA-treated, QSdeficient cells (Fig. 3A). Addition of heat-denatured AiiA to either growth medium or heat-killed cell controls did not affect the rate of NO x accumulation or consumption (data not shown). QS-proficient cells produced approximately 756 ppb NO x h Ϫ1 during the initial 2 h of the experiment, while QS-deficient cells produced 514 ppb NO x h Ϫ1 (Fig. 3A). Between h 4 and h 8, the net levels of consumption of NO x by QS-proficient and QS-deficient cells were 262 ppb h Ϫ1 and 191 ppb h Ϫ1 , respectively (Fig. 3A).
QS inhibition increases N 2 O production by N. winogradskyi. Following a previous unsubstantiated report of N 2 O production by Nitrobacter species (29), we measured N 2 O accumulation by both QS-proficient and -deficient concentrated N. winogradskyi cells after 24 h of NO 2 Ϫ oxidation. Despite the absence of a known nitric oxide reductase gene (nor) in the N. winogradskyi genome, N. winogradskyi cells accumulated significantly (P Ͻ 0.002) more N 2 O than either growth medium alone or heat-killed cell controls (Fig. 4). AiiA-treated, QS-deficient cells accumulated approxi-  mately 525 ppb N 2 O, 1.68-fold more N 2 O than QS-proficient cells (313 ppb N 2 O) (P Ͻ 0.0001) (Fig. 4). In addition, the molar ratio of NO x -N at 2 h to normalized N 2 O-N accumulated at 24 h (after all NO x was consumed) was considerably higher (6.90 NO-N/ N 2 O-N ratio) than that seen under QS-deficient conditions (2.78 NO-N/N 2 O-N ratio). These data suggest that QS also affects N 2 O production by N. winogradskyi, likely through changes in NO x fluxes.

DISCUSSION
Quorum quenching transcriptomics is a novel technique to identify QS-controlled genes and phenotypes. We used a quorum quenching transcriptomic technique to investigate the role of QS in N. winogradskyi, a nitrite-oxidizing bacterium that participates in the nitrogen cycle. Our experiments identified a QScontrolled phenotype and revealed a link between QS and NO x metabolism in a nitrifying microorganism. While mutational analysis is the preferred method to study QS, there are many bacteria, including those of environmental importance, with no established genetic system and new techniques are needed to determine the purpose of QS in these organisms. Quorum quenching also represents a different methodology to study animal pathogens and social evolution (20,(22)(23)(24). Mutant construction in most bacteria takes several generations of selection that may introduce unintended changes, particularly during social-evolution experiments. As previously suggested, quorum quenching through addition of a lactonase or chemical inhibitor is a quicker method to induce a QS-deficient phenotype without the need for several generations of selection and the possibility of pleiotropic effects (23,24,37). NO x metabolism in N. winogradskyi. NO production and consumption in N. winogradskyi have generated considerable interest and confusion for almost 30 years. Earlier work by Freitag et al. and Freitag and Bock suggested that Nitrobacter strains consume NO to generate NADH and produce N 2 O under various conditions (29,35). Starkenburg et al. published the sequenced genome of N. winogradskyi and noted that it lacked the gene(s) known to produce N 2 O but possessed a putative NO-producing gene, nirK, and ncgABC, closely related to homologs in N. europaea (6). However, later work by Starkenburg et al. could not confirm production of NO by N. winogradskyi when nirK was expressed but did show that N. winogradskyi was able to consume NO (36). Coculture studies of the AOB Nitrosomonas europaea and N. winogradskyi suggest that N. winogradskyi may consume NO, as less NO accumulated during coculture but expression of nirK in N. winogradskyi increased (25,38). Finally, the studies described above and work on other NOB and AOB were incorporated into a multispecies metabolic model to assess sources and sinks of NO in relation to N 2 O production during wastewater treatment (34). The model predicts that NOB likely oxidize NO to NO 2 Ϫ but do not substantially contribute to N 2 O production (34).
Our quorum quenching transcriptomics approach led us to investigate production and consumption of NO x gases in N. winogradskyi and showed that N. winogradskyi may function as a source and/or a sink of NO x and N 2 O. We hypothesized that if increased expression of nirK and associated genes under QSproficient conditions increased the concentration of NO x gases,  then more NO would be available for abiological or nonspecific mechanism-based reduction to N 2 O. Contrary to our prediction, QS-deficient cells produced significantly more N 2 O than QSproficient cells. A closer inspection of the molar ratio of peak NO x -N and N 2 O-N produced shows that QS-proficient cells not only produced more NO x than QS-deficient cells but also directed considerably less of the NO x to the N 2 O pool than QS-deficient cells. These data possibly suggest consumption of NO x via oxidation back to nitrite and, subsequently, nitrate, as previously suggested (36), or that there are alternate fates for NO such as unspecified signaling roles.
Note that N. winogradskyi produced considerably less N 2 O than NO, as previously predicted (34). N. winogradskyi does not contain a known nitric oxide reductase (nor), and, while our data show that some of the N 2 O formation is dependent on live cells, there may be abiotic reactions of NO with cellular components as has been recently suggested during ammonia oxidation by Thaumarchaeota (39). This hypothesis may partially explain why QSdeficient cells produce more N 2 O, since their consumption of NO x is slower than that by QS-proficient cells.
Considering the close homologies between nirK and ncgABC of N. winogradskyi and N. europaea, the two gene clusters may serve similar purposes. In N. europaea, both nirK and ncgABC were shown to confer tolerance to NO 2 Ϫ , but nirK had a negative fitness effect in ncgABC mutants (40,41). Our observation of an up to 19.9-fold increase in ncgABC transcripts (Table 1; see also Table S1 and Dataset S1 in the supplemental material) under QS-proficient conditions suggests that these genes may play a role in NO x consumption or detoxification in N. winogradskyi, but future work is needed to investigate this prediction.
One interesting observation arising from our work on NO x flux is the abiotic generation of NO x from NO 2 Ϫ , likely via chemical decomposition of aqueous HNO 2 or of its gaseous analog, HONO. Although this phenomenon was studied extensively in the past (31), reviewed in the nitrification and engineering fields (33), and recently appreciated in soils (42,43), most studies in the nitrification and environmental engineering fields have still largely ignored it. Many studies on nitrifying microorganisms, particularly NH 3 oxidizers, routinely measure NO but do not account for NO x generated abiotically from the NO 2 Ϫ end product of NH 3 oxidation. In addition, many metabolic modeling studies, including an important study cited here modeling NO and N 2 O turnover (34), have not included abiotic formation of NO as a significant source. Clearly, future models and studies need to consider the contribution of abiotic reactions to NO x production.
Why does QS regulate NO x metabolism? The immediate rapid generation of NO by concentrated N. winogradskyi cells is an initially puzzling response for a NO 2 Ϫ oxidizer, since reductant would be required to reduce NO 2 Ϫ to NO. However, metabolic modeling of electron flow in Nitrobacter has suggested that generation and consumption of NO would help explain previous experimental data (32). Some support for QS regulation of this metabolic response recently emerged, as Nitrobacter accumulated fewer AHLs under mixotrophic than under autotrophic growth conditions, perhaps also suggesting less generation of NO when organic carbon is available (27). Another possible explanation is the model proposed by Starkenburg et al. that suggests that NO production is a strategy to reversibly block the terminal cytochrome oxidase and redirect electrons toward generation of reductant (36). This hypothesis would make sense if one role for QS is to promote redirection of electrons away from reductive cellular biosynthesis and toward generation of electron-rich storage compounds such as poly-␤-hydroxybutyrate as previously reported (6).
In other bacteria, QS generally controls production of a public good that can be used by the entire population (1). According to previous reports that Nitrobacter generates NADH from NO oxidation (35), an increase in NO generation at higher cell densities (QS-proficient conditions) may function as a public good for energy generation. NO generated in a large population of Nitrobacter is more likely to be utilized by nearby Nitrobacter cells and may benefit the population.
Another possible function for QS regulation of the nirK cluster and other genes could be preparation for stationary phase, partially through NO signaling. Signal integration of QS and stress responses has been previously demonstrated in Pseudomonas aeruginosa and other bacterial species (3). Transcriptome data suggest that QS-proficient conditions both prepare for growth arrest via repression of assimilatory nitrite reductase nirBD and induction of the stringent response and promote transition from a motile to sessile state through inhibition of flagellar expression and possible guanine secondary messenger signaling ( Table 1). Many of these and other changes in transcription could be indirect effects of NO signaling. According to transcriptome data, QS activation triggers transcription of a putative nnrS homolog (Table 1). While the role of nnrS in N. winogradskyi is unknown, this gene was previously shown to be transcribed during exposure to NO and to regulate chemotaxis in Rhodobacter sphaeroides (44,45). NnrS has recently been proposed to function as an NO sensor (46), and we speculate that it may serve a similar role in N. winogradskyi. NO signaling would also be a convenient way to detect nearby AOB, since some NO is produced via NH 3 oxidation.
The identification of QS regulation of NO x metabolism in N. winogradskyi raises questions about NO x metabolism in other NOB. A cursory search of genomic databases shows that all NOB and comammox bacteria contain nirK homologs and that all except Nitrococcus and Nitrolancea contain nnrS homologs but that only Nitrobacter species contain both ncgABC and clearly annotated autoinducer synthase and receptor genes associated with QS. Since Nitrobacter species are r-strategists with the ability to exploit higher substrate concentrations and sporadically grow to higher densities, they may make better use of cell-density-dependent QS genetic regulation (10,47). In addition, as r-strategists, Nitrobacter species might use QS-controlled preparation for starvation as an important strategy to recognize transitions to an unfavorable energy-limited situation (48). Future research into nirK function in NOB is needed to confirm the role of nirK and nnrS in NO x fluxes and to determine if NO signaling occurs in these microorganisms.
Bacterial strains, plasmids, and growth medium. Bacterial strains and plasmids used in this study are outlined in Table S3 in the supplemental material. N. winogradskyi was routinely cultivated in 60 mM NaNO 2supplemented mineral salts medium as described previously (26), with minor modifications for NO x and N 2 O measurements (see Text S1 in the supplemental material and the descriptions of AiiA QS inhibition and culturing experiments below). N. winogradskyi cultures were routinely screened for heterotrophic contamination by plating 200-l aliquots of culture on Luria-Bertani (LB) agar plates. Agrobacterium tumefaciens was prepared and cultivated as described elsewhere (49,50). Escherichia coli strains were grown in LB medium on a rotatory shaker at 200 rpm and 37°C.
AiiA lactonase production and activity measurement. Plasmid pDSK519 carrying aiiA was kindly provided by Max Teplitski and Mengsheng Gao of the University of Florida. The aiiA gene was cloned, and AiiA was expressed and purified as outlined in Text S1 in the supplemental material. AiiA-specific activity units were determined by measuring reductions of AHL concentrations after 4 hours as outlined in Text S1 in the supplemental material.
AHL bioassay. AHLs were extracted from N. winogradskyi cultures and quantified in Miller units by broad-range Agrobacterium tumefaciens bioassay as described previously (10,49,50). AHL concentrations (nanomolar) were estimated using standard concentrations of C 10 -HSL (see Fig. S2 in the supplemental material). For determining AiiA activity, 200 l of the assay solution was directly added to A. tumefaciens culture as described previously (49,50).
AiiA QS inhibition and culturing experiments. Batch cultures of N. winogradskyi were prepared in 100 ml 60 mM NO 2 Ϫ -supplemented medium at pH 7.5 as outlined above, inoculated to an optical density at 600 nm (OD 600 ) of 0.001 from mid-exponential-phase cultures, and grown in Erlenmeyer flasks on a rotatory shaker at 100 rpm and 30°C. All experiments, including mRNA-Seq experiments, included 4 biological replicates. For NO x and N 2 O measurements, batch cultures were either treated with AiiA lactonase as outlined in Text S1 in the supplemental material or left untreated and were harvested by centrifugation. Harvested cells were suspended to an OD 600 of 0.2 in 25 mM (NO x measurement) or 60 mM (N 2 O measurement) NO 2 Ϫ -supplemented medium and treated with AiiA lactonase or left untreated or subjected to heat killing by incubation at 110°C for 20 min. Cultures were placed in 41-ml serum vials. The serum vial cultures were then capped with gray-butyl stoppers, crimp sealed, and incubated for 24 h as outlined. NO x and N 2 O levels were routinely measured in the headspace as outlined below. Experimental cultures were routinely monitored every 24 h to check cell density (OD 600 ), NO 2 Ϫ concentration by the Griess assay (51), and AHL concentration as described above.
For QS inhibition transcriptome experiments and other experiments, a specific number of activity units of AiiA lactonase (filtered using a 0.2m-pore-size filter) was added into N. winogradskyi batch cultures every 24 h (QS-deficient conditions) as outlined in Text S1 in the supplemental material.
RNA preparation and sequencing. RNA was extracted using an RNeasy minikit (Qiagen, Germantown, MD) (10), and mRNA was enriched and prepared for Illumina mRNA-Seq as described previously (52). The libraries were sequenced (150mer paired-end sequencing) on a HiSeq 3000 Sequencer (Illumina, San Diego, CA) at the Center for Genome Research and Biocomputing Core Laboratories at Oregon State University.
Transcriptome data analysis. The mRNA-Seq data were analyzed using the CLC Genomics Workbench (CLC bio, Prismet, Denmark) as previously described (52). Briefly, mRNA-Seq reads were normalized to reads per kilobase of transcript per million mapped reads (RPKM) and the module for empirical analysis of differential gene expression (DGE) was used as described previously (53,54). Quantitative reverse transcription-PCR (qRT-PCR) was used to corroborate gene expression of selected genes with total RNA from biological replicates and primers as described in Text S1 in the supplemental material and outlined in Table S4 and  Table S5.
Analytical methods. NO 2 Ϫ levels were determined by chemical assay as described previously (51). NO x levels were measured using a portable NO x detector (LMA-3D and LNC-3D; Unisearch Associates Ltd., Con-cord, Ontario, Canada) that passes air through a CrO 3 filter to convert NO to NO 2 and then measures parts per billion by volume of NO 2 by chemiluminescence as described previously (55), with modifications. Briefly, 1 ml or 5 ml of samples was injected into the intake line of the instrument and the peak NO x level recorded. NO x peaks were quantified by comparison to both NO and acidified NO 2 Ϫ standards. N 2 O was measured by gas chromatography as described previously (52). Cell density was measured as OD 600 , and the protein concentration was measured with a Pierce bicinchoninic acid (BCA) protein assay kit (Thermo Scientific, Rockford, IL).
Accession number(s). Raw datasets and processed datasets are available at Gene Expression Omnibus (GEO) at the National Center for Biotechnology Information (NCBI) under accession no. GSE84969.

ACKNOWLEDGMENTS
Special thanks to Ajai Dandekar for helpful advice on the use of AiiA lactonase to inhibit AHL QS, Max Teplitski and Mengsheng Gao for providing the pDSKaiiA and pDSK constructs, and Jun Zhu for providing the bioassay strain A. tumefaciens KYC55(pJZ372) (pJZ384) (pJZ410) for this and previous projects. We also thank David Myrold for gas chromatography instrument use, Steve Perakis for use of a NO x detector, and Chris Catricala, Rebecca Ferrell, Anne Taylor, Michael Dobie, Neeraja Vajrala, and Aiden Maxwell for helpful discussions and laboratory assistance. The funding agencies had no role in study design, data collection and interpretation, or the decision to submit the work for publication.