NfiR, a New Regulatory Noncoding RNA (ncRNA), Is Required in Concert with the NfiS ncRNA for Optimal Expression of Nitrogenase Genes in Pseudomonas stutzeri A1501

Biological nitrogen fixation is an energy-expensive process requiring the hydrolysis of 16 ATPs. Consequently, the expression of nif genes is highly regulated at both transcriptional and posttranscriptional levels through complex regulatory networks. Global regulation involves a number of regulatory proteins, such as the nif-specific activator NifA and the global nitrogen regulator NtrC, as well as various regulatory ncRNAs. We show that the two P. stutzeri ncRNAs, namely NfiS and NfiR (for nitrogen fixation condition-inducible ncRNA), optimize nitrogen fixation and environmental stress responses. NfiS and NfiR respond differently to various environmental signals and differ in their secondary structures. In addition, the two ncRNAs target the mRNAs of nifK and nifD, respectively. Such ncRNA-based posttranscriptional regulation of nitrogenase expression might be an evolved survival strategy, particularly in nitrogen-limiting environments. This study not only highlights the significant roles of regulatory ncRNAs in the coordination and ﬁne tuning of various physiological processes but also provides a new paradigm for posttranscriptional regulation in nitrogen-fixing bacteria.

KEYWORDS NfiR, Pseudomonas stutzeri, nifD mRNA, nitrogen fixation, regulatory ncRNA B acteria belonging to the genus Pseudomonas are found ubiquitously in many biotic and abiotic environments, including soil and water, as well as on the surfaces of plants, where nutrients are probably limited (1)(2)(3)(4)(5). Pseudomonas strains have evolved versatile metabolic capacity and complex regulatory networks for adaptation to nutrient-limited conditions (6)(7)(8). Global regulation involves a number of regulatory proteins, such as the sigma factor RpoN, the sigma factor RpoS, the global nitrogen regulator NtrC, and the noncoding RNA (ncRNA) chaperone Hfq (9)(10)(11). These regulatory proteins play key roles in global responses to environmental stresses.
In addition to regulatory proteins, various regulatory ncRNAs have been reported in pseudomonads, the majority of which appear to adjust bacterial physiology in response to environmental changes or nutrient stress conditions (12)(13)(14). It is particularly noteworthy that nitrogen stress-induced ncRNAs were verified as new players in nitrogen regulatory networks. NrsZ is the first nitrogen-regulated ncRNA reported in Pseudomonas aeruginosa, and it activates the production of rhamnolipid surfactants via posttranscriptional control of the rhamnolipid synthesis gene rhlA (15). Furthermore, several nitrogen stress-induced ncRNAs, the expression of which is activated by HetR, have been identified in the model filamentous nitrogen-fixing cyanobacterium Anabaena sp. strain PCC7120 (16). NsiR1 is the first known nitrogen stress-induced ncRNA present in a cyanobacterium, and it functions as an early marker of cell differentiation (17). Another nitrogen stress-induced ncRNA, NsiR4, is involved in nitrogen assimilation control in cyanobacteria by targeting glutamine synthetase (18). However, until very recently, ncRNAs had not been experimentally identified as regulators directly involved in nitrogen fixation. NfiS in Pseudomonas stutzeri A1501 represents the first ncRNA required for optimal expression of nitrogenase genes, and NfiS directly targets the nifK mRNA (19). In Methanosarcina mazei strain Go1, an ncRNA designated small RNA 154 (sRNA 154 ) was found to play a central regulatory role in nitrogen metabolism via determining the expression of nitrogenase and glutamine synthetase by positively affecting transcript stabilities (20). Most recently, another archaeal ncRNA, sRNA 41 , was identified to regulate nitrogenase expression in an indirect manner by increasing levels of ACDS protein (acetyl coenzyme A-decarbonylase/synthase complexes) under nitrogen limitation (21).
Certain P. stutzeri strains possess a general nitrogen regulatory system (NtrBC and related genes) in the core genome; during evolution, they acquired the capacity to fix nitrogen via horizontal transfer of a 49-kb nitrogen fixation (nif) island (PST1302 to PST1359) that encoded a set of the proteins for synthesis, maturation, and functioning of nitrogenase, as well as two negative/positive regulatory proteins (NifL and NifA) for nif gene expression (1,22,23). Thus, the expression of nif genes in these strains is controlled by two regulatory systems of different evolutionary origins (24,25). This global regulation includes a nitrogen regulator, NtrC, which is encoded in the core genome and is highly conserved among Pseudomonas species (9,26). Under nitrogen fixation conditions, NtrC is phosphorylated, thereby causing transcriptional activation of nifA and carrying a positive regulator of the nif operons within the nif island (27). P. stutzeri strain A1501 is a diazotrophic proteobacterium originally isolated from the rice rhizosphere (28). It normally colonizes root surfaces but can also penetrate the root and grow endophytically. Genes for the synthesis, maturation, and functioning of nitrogenase are clustered in a 49-kb genomic island, suggesting that the nitrogen fixation property is acquired by lateral gene transfer from a diazotrophic ancestor (22). Although the importance of regulatory ncRNAs is generally recognized, their roles in fine-tuning expression of the horizontally acquired nif genes are poorly understood.
Here, we report the identification of a second regulatory ncRNA (termed NfiR for nitrogen fixation condition-inducible ncRNA), which is involved in the posttranscriptional control of the nifD gene. Our study began with transcriptome analysis of P. stutzeri A1501 for cells grown under nitrogen fixation and ammonium shock conditions. This analysis led to the identification of NfiR, whose expression is dependent on NtrC and Hfq and is significantly induced under nitrogen fixation conditions. Subsequent analyses demonstrated that NfiR is capable of binding to nifD transcripts in vitro, and this direct interaction can potentially affect the stability of the transcript. These data allow us to propose a model for the posttranscriptional control of the nitrogenase gene expression that involves two ncRNAs (NfiS and NfiR) in P. stutzeri A1501.

RESULTS
Transcriptome analysis of ncRNAs under nitrogen fixation conditions. A global transcriptional profiling analysis was conducted with P. stutzeri A1501 to investigate the intracellular nitrogen stress responses, and the results are shown in Table S1. A total of 53 ncRNAs were detected under nitrogen fixation conditions, 17 of which were upregulated under nitrogen fixation conditions (nitrogen-free and microaerobic conditions) but were rapidly downregulated after 10 min of ammonium shock (Table S1). A BLASTN search of the 17 nitrogen-responsive ncRNA sequences against the GenBank database showed that (i) three ncRNAs are specific to strain A1501, without homologs in any other bacterial genomes, (ii) six ncRNAs are restricted to P. stutzeri, and (iii) eight ncRNAs are present in the genomes of other Pseudomonas species or bacteria. Notably, only four ncRNAs (NfiS, CrcZ, CrcY, and signal recognition particle bacterial RNA [SRP bact RNA]) have been functionally characterized, and there were no clues regarding the functions for the remaining 13 ncRNAs. This catalog of candidate regulatory ncRNAs will serve as an important reference point for comprehensive analyses of ncRNA regulation in P. stutzeri and other nitrogen-fixing bacteria.
To gain insights into the potential roles of the 17 nitrogen-regulated ncRNAs in nitrogen stress responses, we monitored the gene transcription levels under both nitrogen fixation and non-nitrogen fixation conditions (Fig. 1). Under nitrogen-free and aerobic (20% oxygen tension) conditions, all ncRNAs were downregulated, with the exception of ncRNA34 (renamed NfiR), the expression of which was significantly upregulated compared to that under nitrogen-sufficient and aerobic conditions (Fig. 1A). In addition, almost all selected ncRNAs were induced under the nitrogen-free and microaerobic conditions appropriate for nitrogen fixation, compared with nitrogen-free, aerobic conditions, implying that they might be involved in the regulation of nitrogen fixation (Fig. 1B). Obviously, expression of NfiS shows the most dramatic increase (Ͼ10-fold), followed by that of ncRNA31 and then that of ncRNA34 (8.8-and 5.5-fold, and 20% oxygen tension, white). (B) Nitrogen-free microaerobic conditions (0.5% oxygen tension, black) versus nitrogensufficient microaerobic conditions (6 mM NH 4 ϩ and 0.5% oxygen tension, white). Inset shows Northern blot detection of the NfiR RNA. Total RNA was extracted from wild-type strain A1501 under nitrogen-free microaerobic conditions (lane 1) and nitrogen excess conditions (lane 2). A1501 is able to fix nitrogen for its growth only under nitrogen-free microaerobic conditions. Data are the means and standard deviations of three independent experiments. respectively), under nitrogen fixation conditions. Since P. stutzeri A1501 was isolated from the rice rhizosphere, it is tempting to speculate that many ncRNAs are induced under nitrogen-free and microaerobic conditions to adapt to the nitrogen-poor and microaerobic rhizosphere environments. Further predictions of the interactions between the selected ncRNAs and nif gene mRNAs revealed that all selected ncRNAs have no binding sites for nif gene mRNAs, except for NfiS, with an experimentally confirmed binding site for nifK mRNA (19), and ncRNA34, with a putative binding site for nifD mRNA (Fig. S1A). Hence, ncRNA34 was redesignated NfiR (nitrogen fixation conditioninducible ncRNA).
Expression of many ncRNAs is associated with environmental stresses and is regulated by various regulatory proteins, such as the sigma factor RpoN, the global nitrogen activator NtrC, the sigma factor RpoS, and the RNA chaperone Hfq (16,18,19). Thus, we compared the expression levels of the 17 ncRNAs between wild-type A1501 and each of four isogenic mutants devoid of RpoN, NtrC, RpoS, or Hfq (Fig. 2). Deletion of rpoS caused decreased expression of 12 ncRNAs and increased expression of ncRNA12 (Ͼ13.8-fold), an ncRNA of unknown function specific to A1501, and CrcZ (Ͼ2.7-fold), a protein-binding ncRNA acting as global regulator of carbon catabolite repression ( Fig.  2A) (29). Most of the ncRNAs were downregulated in the ntrC or rpoN deletion backgrounds ( Fig. 2B and C), except for ncRNA32, whose predicted target is the rnfA gene encoding the subunit A of the electron transport complex and is probably involved in nitrogen fixation. Of particular note is that the transcription of three ncRNAs (NfiS, NfiR, and ncRNA31, whose predicted target is the hesB gene encoding Fe-S cluster assembly protein probably involved in nitrogen fixation) was abolished in the ntrC deletion background (Fig. 2B), suggesting that their expressions are transcriptionally activated by NtrC. Furthermore, Hfq appears to be a pleiotropic regulator of gene expression, as a loss-of-function mutation led to up-or downregulation of the selected ncRNAs, such as nfiR and nfiS, whose expression levels were upregulated by approximately 2-and 3-fold, respectively (Fig. 2D). Taken together, transcriptome sequencing (RNA-seq) analysis of A1501 grown under nitrogen fixation conditions identified 17 nitrogen-regulated ncRNAs that respond differently to nitrogen and oxygen signals and differ in transcriptional regulation, presumably constituting a complex regulatory network to integrate nitrogen fixation with global cellular physiology. Among these ncRNAs, in addition to a previously described NfiS, NfiR is the only ncRNA that has a putative binding site with nitrogenase gene mRNA, implying its potential involvement in the posttranscriptional control of nitrogen fixation.
Phenotypic characterization of a mutant lacking the nfiR gene. The nfiR gene, located in the intergenic region between PST2408 (a hypothetical protein) and PST2409 (a putative assimilatory nitrite reductase, NasA) ( Fig. S1B and C), is predicted to encode a 54-nucleotide (nt) transcript (Fig. 1B), representing the smallest ncRNA that has been functionally characterized thus far in Pseudomonas. Furthermore, we detected several nfiR homologs, including an experimentally characterized P. aeruginosa sRNA P11 (30) and a predicted Azotobacter vinelandii sRNA11, but their functions are unknown. In addition, we did not identify homologs in any other bacterial species, suggesting that nfiR is specific to Pseudomonas.
In nitrogen-fixing bacteria, certain nitrogen-regulated ncRNAs were verified as regulatory players in coupling between nitrogen metabolism and stress responses (14,15). To determine whether nfiR is involved in the regulation of nitrogen fixation and stress resistances, we constructed an nfiR-knockout mutant, A1801 (ΔnfiR), and a complementary strain, A1802, corresponding to A1801 harboring a plasmid expressing the wild-type nfiR gene, and we compared their growth under stress conditions by the addition of sorbitol or H 2 O 2 with the growth of wild-type strain A1501. As shown in Fig.  3, both A1801 and A1802 displayed growth rates similar to that of the wild-type strain in Luria-Bertani (LB) medium, indicating that deletion of the nfiR gene had no effect on bacterial survival under normal growth conditions. Furthermore, we found that in the presence of 18 mM H 2 O 2 or 0.3 M sorbitol, the nfiR mutant displayed significantly impaired growth, but the complementary strain recovered the growth capacity to the wild-type level under the same treatment, implying that NfiR might play an important role in the response to oxidative or osmotic stress. To this end, we compared global protein expression changes between wild-type A1501 and its derived nfiR mutant (Fig.  S2). A total of 258 spots with 1.5-fold or greater differences in fluorescence intensity were identified, and the identities of 123 spots were successfully determined using matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry (Table S2). These include three cell mobility-related proteins undetectable in the nfiR mutant and several stress response proteins undetectable in the wild-type strain. In addition, a glutathione S-transferase (GST; PST3481) was found to be 60.0-fold more abundant in the nfiR mutant than in the wild type. GST participates in protecting cells against damage due to oxidative stress in proteobacteria (31); however, its functions in Pseudomonas stutzeri still remain to be elucidated. Unexpectedly, GST seems to be substantially derepressed in the nfiR mutant, although it is not yet known whether NfiR regulates the protein directly or acts indirectly to alter its synthesis. We anticipate that NfiR mutation causes global changes in gene expression that may lead to more complex effects on oxidative stress response than previously thought. Furthermore, we found that deletion of nfiR resulted in a significant reduction of nitrogenase activity by more than half, and the defect was restored by the introduction of a single copy of nfiR (Fig. 4A). The results of quantitative real-time PCR (qRT-PCR) showed that the expression levels of nifA, nifH, nifD, nifK, and their regulatory genes rpoN, ntrC, and glnK were decreased to various extents in the nfiR mutant compared with the wild type, whereas these inductions were fully or partially restored to wild-type levels by the complementation plasmid with a wild-type nfiR gene (Fig. 4B). Furthermore, the Western blot results showed that nitrogenase MoFe protein polypeptides (NifD and NifK) were produced at lower levels in the nfiR mutant background than in the wild type (Fig. 4C). These data consistently indicate that NfiR is a positive regulator required for maximal expression of nitrogenase. These findings are consistent with results from the proteomics analysis (Fig. S2). Proteins associated with nitrogen metabolism and nitrogen fixation, such as the glutamine synthetase GlnA, Fe-S cluster assembly protein NifU, and MoFe protein alpha subunit NifD, were downregulated in the nfiR mutant (Table S2). Together, our data suggest a role for NfiR in regulating the expression of proteins involved in nitrogen fixation, as well as in osmotic and oxidative resistances.
Involvement of Hfq and NtrC in the regulation of NfiR expression. The ncRNA chaperone Hfq is usually required for the function and/or stability of most base-pairing ncRNAs (32). Previous studies have also shown that although expression of the nfiS gene was upregulated in the hfq deletion, the half-life of the NfiS transcript in the wild-type strain was more than twice that of the hfq mutant (19). Similarly, the hfq deletion also resulted in upregulation of nfiR expression (Fig. 2D). When wild-type cells were grown to the mid-exponential phase and treated with rifampin, nfiR mRNA had a half-life of approximately 4 min. In contrast, the hfq mutant had a half-life of 2 min (Fig.  S3). In addition, deletion of hfq resulted in decreased expression of the assayed nif and nif-related genes (Fig. S3B), consistent with the previous results that inactivation of hfq caused a significant decrease in nitrogenase activity (19). Our data indicate that Hfq enhances the stability of both ncRNAs while it functions in a global manner to control their transcriptional activities via unknown mechanisms.
Interrogation of the NfiR promoter DNA indicated the presence of two NtrC-binding sites, which contain the highly conserved TGC and GCA elements with an 11-nucleotide spacing (Fig. 5). Moreover, nfiR expression was abolished in the ΔntrC background (Fig.  2B). Evidence available thus far has strongly suggested that NtrC activates NfiR expression in a direct manner. To this end, we performed a DNase I protection footprint sequencing assay to determine the NtrC-protected region using a 264-bp nfiR promoter DNA fragment as the probe. The results indicate that NtrC was capable of protecting a 43-bp DNA region containing the predicted two putative NtrC-binding sites (Fig. 5). Next, we performed 5= rapid amplification of cDNA ends (5= RACE) to identify the transcription start site of the nfiR gene. Contrary to our expectation, NfiR transcription did not initiate from the 12 position downstream of the predicted 54 -binding site; instead, it started from a cytosine residue located 29 bp downstream of the predicted 54 -binding site (Fig. S1A). Curiously, nfiR expression was reduced (but not abolished) in the genetic background of ΔrpoN (Fig. 2C). Together, these data show that NfiR expression is subject to complex regulation at both the transcriptional and posttranscriptional levels, which involves both Hfq and NtrC.
Examining the molecular interactions between NfiR and the nifD mRNA. In silico analysis using RNAalifold (33) revealed the presence of an 11-nt sequence in the stem-loop structure of NfiR (nucleotides 12 to 22) pairing with its counterpart in the coding region of nifD mRNA (nucleotides 1194 to 1207) by eight nucleotides (Fig. 6A). To validate the predicted interaction, we first synthesized two 30-nt single-stranded RNA (ssRNA) oligonucleotides containing the wild-type base-pairing sequence of NfiR or nifD mRNA, designated N-NfiR-wt and N-nifDR-wt, plus another two containing an absolute mismatch mutation (N-NfiR-mut) and full complementary mutations (N-NfiRcom). The results of microscale thermophoresis (MST) showed that N-NfiR-wt (82% match level) was capable of binding with N-nifDR-wt, exhibiting a dissociation constant of 36.97 Ϯ 11.59 M (Fig. 6B). Significantly, no binding signal was detected with the N-NfiR-mut molecule carrying five substitutions in the base-pairing sequence of NfiR (Fig. 6C). Next, we measured the binding affinity using N-NfiR-com (100% match level). The complementary mutation resulted in enhanced binding affinity to N-nifDR-wt  (Fig. 6D). Finally, the effect of nifD mRNA mutations in the 11-nt sequence (1194 to 1207 nt) on base pairing with NfiR was investigated using 30-nt ssRNA oligomers containing a mismatch mutation (N-nifDR-mut) and a complementary mutation (N-nifDR-com). The results indicate that N-nifDR-mut (100% mismatch) showed no binding affinity for NfiR (Fig. 6E), whereas N-nifDR-com (perfect match) displayed a stronger interaction with NfiR than N-nifDR-wt (Fig. 6F). Together, these data strongly indicate that NfiR is capable of interacting with the nifD mRNA in vitro.
To further confirm functions of the predicted NfiR sequence in vivo, we measured nitrogenase activities of the ΔnfiR mutant complemented with either wild-type nfiR or each of the four mutant nfiR alleles carrying one or more guanine-to-cytosine substitutions in the base-pairing region (Table 1). To this end, DNA fragments of wild-type and mutant nfiR were individually cloned into the plasmid vector pLAFR3, generating five complementation plasmids, pNfiR-wt, pNfiR-mut1, pNfiR-mut2, pNfiR-mut3 and pNfiR-mut4. The resultant plasmids were then transformed into the nfiR mutant (strain A1801). As shown in Table 1, the nitrogenase activity of the NfiR mutant A1801 was almost fully complemented by the wild-type nfiR allele but was only partially complemented by the mutated nfiR genes. The mutated nfiR gene NfiR-mut1, with one nucleotide substitution (G 13 ¡C), restored 91% of wild-type nitrogenase activity, whereas NfiR-mut4, with five nucleotide substitutions, restored only 19%. Most interestingly, the nitrogenase activity of A1805 carrying NfiR-mut3, with two nucleotide substitutions (G 15 G 16 ¡CC), was restored to 20% of the wild-type level, much lower than that of A1804 carrying NfiR-mut2, with two nucleotide substitutions (G 20 G 21 ¡CC), but almost equal to that of A1806 carrying NfiR-mut4, with five nucleotide substitutions (G 13 ¡C, G 15 G 16 ¡CC, and G 20 G 21 ¡CC), highlighting the importance of the two guanines at positions 15 and 16 in NfiR regulatory function. These results strongly indicate that the 11-nt sequence within its stem-loop structure is functionally required for NfiR to regulate optimal nitrogen fixation.
Both NfiR and NfiS are required for optimal expression of nitrogenase genes. Base-pairing interactions between ncRNAs and their target mRNAs often regulate the stability of the mRNA transcripts (34). We have previously shown that NfiS increases the half-life of nifK mRNA (19). To test whether this function holds for NfiR, we measured the half-life of nifD mRNA in wild-type A1501, nfiR mutant (strain A1801), and the complementation mutant A1802 (Fig. 7A). qRT-PCR was performed for cells grown under nitrogen fixation conditions. Rifampin was added to inhibit RNA synthesis. The results showed that the nifD mRNA half-life was 20 min after rifampin had been added to wild-type cells, and it was reduced to 15 min in the nfiR mutant. The nifD transcript stability of the nfiR mutant was restored with the introduction of wild-type nfiR (Fig. 7A). Consistent with our expectation, the double deletion mutant (A1901) lacking nfiR and nfiS had reduced half-lives for both nifD and nifK mRNAs compared with the wild type (Fig. 7B). Intriguingly, a 70% reduction was observed for the nfiR and nfiS double a The intact wild-type NfiR gene and four mutated nfiR genes with one or more nucleotide substitutions were cloned into plasmid pLAFR3, and the resulting complementation plasmids were introduced into A1801 (ΔnfiR), generating the five complementation strains. b Point mutations introduced into synthesized oligonucleotides are shown in boldface. deletion mutant, whereas deletion of either nfiS or nfiR caused a 40 or 60% reduction of the nitrogenase activity, respectively (Fig. 4A). Together, the data from the nitrogenase activity and mRNA stability assays strongly suggest that both NfiR and NfiS optimize nitrogenase activity in a cooperative manner via determining the stabilities of the nitrogenase gene mRNAs.

DISCUSSION
Biological nitrogen fixation is an ancient trait dating back to more than 3 billion years ago (35,36). It is an energy-expensive process, requiring large amounts of both reducing power and high-energy phosphate (ATP). To adapt to diverse environments and to avoid energy waste, nitrogen-fixing bacteria have evolved various regulatory mechanisms to control the transcription of nif genes and the synthesis and activity of nitrogenases in response to ammonium availability and other environmental factors (37)(38)(39). This global regulation occurs at transcriptional and posttranscriptional levels, including a classical nitrogen regulatory cascade, an ADP-ribosyl-transferase/glycohydrolase (DraT/DraG) system, and a recently identified complex network of ncRNAs. The RpoN-NtrC-NifA regulatory cascade is a major and best-studied regulatory mechanism for nitrogen fixation at the transcriptional level (27,40). In addition, nitrogenase activity can be modulated by ADP ribosylation of NifH, the so-called "switch-off" effect. The best-characterized mechanism of switch-off involves the ADP ribosylation of NifH by the DraT/DraG system in response to the addition of excess ammonium (41). Although A1501 does not carry draTG (22), nitrogenase switching off of this strain has been reported (26), suggesting an inactivation mechanism different from that reported in other nitrogen-fixing bacteria (35). Currently, the ncRNA-based posttranscriptional regulatory mechanisms of nitrogen fixation have not been fully elucidated.
In this study, we conducted a global transcriptional profiling analysis and identified 17 nitrogen-regulated ncRNAs. Further analysis revealed that these ncRNAs respond differently to nitrogen and oxygen signals and differ in transcriptional regulation ( Fig.  1 and 2), presumably constituting a complex regulatory network to integrate nitrogen fixation with global cellular physiology. Furthermore, we employed the computational software sRNATarget to predict their potential targets and found that all selected ncRNAs have no binding sites for nif gene mRNAs, except for NfiS and NfiR. In the case of nitrogen-fixing P. stutzeri, NfiS is identified as the first regulatory ncRNA that regulates optimal nitrogen fixation via direct base pairing with nifK mRNA (19). Here, we identified the second P. stutzeri ncRNA, NfiR, which is potentially involved in base- pairing interaction with the nifD transcript. Both ncRNAs are conserved in P. stutzeri and play global regulatory roles in environmental stress responses in addition to their roles in nitrogenase gene expression. Our findings strongly suggest that in P. stutzeri A1501, NfiR and NfiS act as new riboregulators to integrate the horizontally acquired nif island into host global networks. The precise molecular mechanisms for modulating NfiS and NfiR activities are not known. Although both ncRNAs are expressed at elevated levels under nitrogen fixation conditions, they respond in the opposite way under nitrogenfree conditions, namely, NfiS is downregulated (19), whereas NfiR is upregulated (Fig. 1A). Furthermore, NfiS and NfiR differ in their promoter sequences and transcriptional regulations. We previously identified a conserved 54 -binding site on the nfiS promoter (19). NfiS expression is abolished in the rpoN background, thereby confirming that it is a 54 -dependent ncRNA (19). No NtrC-binding site was identified in the nfiS promoter. In contrast, we characterized the promoter sequence of nfiR and identified a conserved NtrC-binding site (Fig. S1A). The DNase I footprinting assay in vitro confirmed that NtrC binds specifically to the nfiR promoter (Fig. 5).
Nitrogen fixation is significantly affected by environmental stresses, such as nitrogen starvation or oxidative or osmotic stresses encountered by associative nitrogen-fixing bacteria in the rhizosphere. Consequently, certain nitrogen-regulated ncRNAs were verified as regulatory players in coupling between nitrogen fixation and the stress response (14,15). The primary function of NfiS is related to osmotic and oxidative stress responses, but it was later recruited and paired with nifK mRNA as a new riboregulator when ancestral P. stutzeri acquired a nitrogen fixation island through horizontal gene transfer (19). In this study, we found that NfiR is functionally related to osmotic and oxidative stress responses and that it was most likely recruited by nifD mRNA in a manner similar to NfiS. Furthermore, we confirmed that both NfiR and NfiS were upregulated by various transcriptional or posttranscriptional regulators, such as RpoS, NtrC, RpoN, and Hfq (Fig. 2), suggesting that the modulation of many cellular functions by protein regulators could be associated with the two ncRNAs. Such ncRNA-based regulation might be an evolved survival strategy, particularly in nitrogen-limiting environments.
NtrC is a response regulator with a 54 -interacting domain, which requires functional 54 for its activities. Transcription initiation of 54 -dependent promoters usually begins 12 nt downstream of the 54 binding site, which is followed closely by the translation initiation site. It is thus surprising to find that nfiR transcription started from a position 29 bp downstream of the predicted 54 -binding site (Fig. S1A). However, it is still possible that NtrC directly activates nfiR transcription in the conventional 54 -dependent manner; the primary product of NfiR is then subject to RNA processing at the 5= end, producing a 54-nt ncRNA. A similar phenomenon has previously been reported in Pseudomonas putida with CrcY, an ncRNA involved in carbon catabolite repression (29). Alternatively, there may be some yet-unknown transcription factors involved in NtrC/ 54 -mediated gene regulation that cause a shift of the transcriptional start site. It is also possible that NtrC produces additional effects on NfiR expression via a yet-unknown transcription factor(s), the activities of which are independent of 54 . This possibility would explain the observation that nfiR transcription was abolished in the ΔntrC background, while it was reduced only by half in the ΔrpoN mutant. Nevertheless, NfiR expression is subject to complex regulation, and the precise roles of NtrC and 54 require further investigation.
Results of microscale thermophoresis and genetic complementation strongly suggest that nifD mRNA is a direct target of NfiR. However, given the global effects associated with nfiR deletion, it is reasonable to speculate that NfiR targets multiple mRNAs, particularly those in the 54 regulon. The expression levels of rpoN (the gene encoding 54 in Pseudomonas) and other genes under the control of 54 , e.g., nifA, nifD, nifK, ntrC, and glnK, were decreased in the nfiR strain (Fig. 4B), indicating that their transcriptional activities were controlled in an ncRNA-dependent manner. Furthermore, proteomic analysis of the ΔnfiR mutant also identified a subset of differentially expressed proteins whose expression profiles are 54 dependent (Table S2). Hence, it would make sense that 54 is one of the primary targets for NfiR. However, no putative NfiR-binding sites were identified in the rpoN locus. Thus, the observed phenotypic effects related to 54 function are likely mediated by NfiR directly targeting a regulator in the 54 regulon.
Efficient nitrogenase activity greatly relies on the sufficient accumulation of nif mRNAs. Some diazotrophs can accumulate nitrogenase at levels of as much as 10% of the total soluble proteins within the cells (42), which would require relatively higher levels of nifHDK mRNAs than of mRNA specifying other proteins. Thus, nitrogen-fixing cells must have a robust capacity to produce nifHDK mRNAs at a level sufficient to sustain maximal nitrogenase activity. Such a cellular physiological requirement is consistent with the proposed functional roles of the P. stutzeri NfiS/NfiR and the M. mazei sRNA 154 ncRNAs in enhancing the stability of their target mRNAs. It is interesting to note that sRNA 154 was shown to be directly involved in regulating nitrogen fixation in Methanosarcina mazei (20). This ncRNA specifically targets the nifH mRNA. Despite originating from phylogenetically distinct organisms, sRNA 154 possesses similar modes of action as NfiS and NfiR. Considering the high levels of conservation of the nifHDK genes among nitrogen-fixing bacteria, it is reasonable to speculate that P. stutzeri A1501 may carry a third as-yet-unidentified ncRNA that specifically targets the nifH transcript, as previously described for sRNA 154 (20). Involvement of three ncRNAs targeting each of the three nitrogenase components (NifH, NifD, and NifK) would certainly strengthen the regulation of nitrogen fixation and may also help prevent unbalanced synthesis of the three proteins. More importantly, it would allow integration of additional environmental signals into the precise control of nitrogenase synthesis. In conclusion, our work provides evidence that two Hfq-dependent P. stutzeri ncRNAs, namely NfiS and NfiR, optimize nitrogen fixation and abiotic stress responses. In addition, the two ncRNAs target the mRNAs of nifK and nifD, respectively, increasing the half-life of the transcripts and, consequently, nitrogenase activity. This study not only highlights the significant roles of regulatory ncRNAs in the coordination and fine tuning of various physiological processes but also provides a new paradigm for posttranscriptional regulation in nitrogen-fixing bacteria.

MATERIALS AND METHODS
Bacterial strains, media, and growth conditions. A summary of bacterial strains used in this study is provided in Table 2. P. stutzeri strains were grown at 30°C in Luria-Bertani (LB) medium or in modified minimal lactate-containing medium (medium K) (26). When necessary, spectinomycin (Spc), kanamycin (Km), and chloramphenicol (Cm) were added at final concentrations of 34 g/ml, 50 g/ml, and 40 g/ ml, respectively.
Total RNA-seq and ncRNA prediction. Strain A1501 was cultured for 5 h under nitrogen fixation and ammonium shock conditions as described previously (24,25). RNA was extracted using TRIzol LS reagent (Invitrogen, USA) following the manufacturer's instructions. Host-cell RNA was depleted using a MICROBEnrich kit (Ambion, USA), and bacterial 23S and 16S rRNAs were subsequently depleted with a MICROBExpress bacterial mRNA enrichment kit (Ambion, USA). Total RNA-seq libraries were then constructed and sequenced using an Illumina HiSeq 2500 instrument with the paired-end method, and ncRNAs were predicted based on the mapping of read pairs by Tianjin Biochip Corporation (Tianjin, China).
Growth conditions for transcriptional expression assay. To test the expression levels of the selected 17 ncRNAs, cells of wild-type and mutant strains were grown in medium K and harvested at the early exponential phase. Subsequently, the cells were washed once with fresh medium K and used to start experiments at a final optical density at 600 nm (OD 600 ) of 0.5 under the following four growth conditions: nitrogen starvation conditions (medium K, nitrogen-free, and aerobic conditions), nitrogen fixation conditions (medium K, nitrogen-free, and 0.5% oxygen tension), oxygen limitation conditions (medium K, nitrogen sufficient, and 0.5% oxygen tension), and normal conditions (medium K, nitrogen sufficient, and aerobic conditions). After incubation for 0.5 h at 30°C, cells were harvested and centrifuged for 10 min at 4°C, and pellets were quick-frozen in liquid nitrogen and stored at Ϫ80°C until ready for quantitative real-time PCR (qRT-PCR) assays.
RNA isolation for qRT-PCR. Total RNA was isolated from bacteria cultured under the described conditions using the SV total RNA isolation system (Promega, Madison, WI) according to the manufacturer's instructions. Total RNA was quantified using microspectrophotometry (NanoDrop Technologies, Inc.). RNA integrity was measured using an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). RNA samples with RNA integrity numbers (RINs) above 7.0 and threshold cycle (C T ) values above 32 were used for qRT-PCR.
Quantitative real-time PCR. Expression levels of selected genes were determined by qRT-PCR with a Power SYBR green PCR master mix using an ABI Prism 7500 sequence detection system (Applied Biosystems, USA) according to the manufacturer's instructions. Primers were designed based on sequences of selected genes, which were imported into OligoPerfect (Invitrogen), a primer design software program designed to generate primer pairs suitable for real-time PCR. Primers used for qRT-PCR are listed in Table 3. All qRT-PCRs were performed in triplicate using a 25-ml mixture containing cDNA (5 ml of a one-fifth dilution), 1ϫ brilliant SYBR green quantitative PCR master mixture (Stratagene), and approximately 5 pmol of each primer. Amplification and detection of specific products were performed using the following procedure: 95°C for 10 min, followed by 40 cycles of 95°C for 30 s, 55°C for 1 min, and 72°C for 30 s and then a dissociation curve analysis. The 16S rRNA gene was used as the endogenous reference control, and relative gene expression was determined using the 2 ϪΔΔCT relative quantification method. To obtain a standard curve for the real-time PCR (RT-PCR), PCR was performed with each primer set by using calibrated amounts of chromosomal DNA; these reactions were performed at the same time as the qRT-PCR. The amplification efficiencies (E) of primer pairs used in the study were calculated from the standard curve obtained from a five-point 10-fold serial dilution series of cDNA template according to the following equation: E (%) ϭ [10 (Ϫ1/slope) Ϫ 1] ϫ 100. All experiments were performed with three biological replicates.
Nitrogenase activity assays. Nitrogenase activity was determined with bacterial suspensions incubated to an OD 600 of 0.1 in N-free minimal medium (0.5% oxygen and 10% acetylene) at 30°C (26). Protein concentrations were determined using a standard protein assay (Bio-Rad, Hercules, CA) with bovine serum albumin as a standard. The specific activity of nitrogenase was expressed as nmol ethylene per hour per milligram of protein. Each experiment was repeated at least three times.
Abiotic stress-resistance assays. Strain A1501 and mutant derivatives were grown in LB medium at 30°C to an OD 600 of 0.6 and were then transferred into fresh LB medium in the presence or absence of 0.3 M sorbitol and 18 mM H 2 O 2 . At the time indicated (sorbitol stress, 1 h; H 2 O 2 stress, 10 min), 10-fold serial dilutions were made, and 8 l of each dilution was spotted onto solid LB plates. These plates were incubated at 30°C for 24 h before colony growth was observed and enumerated.
Construction of the nfiR deletion mutant and complementation plasmids. To generate the nfiR mutant strain A1801, a nonpolar deletion (covering residues from Ϫ10 to ϩ54) was introduced into the Pseudomonas stutzeri chromosome by homologous recombination (19), as shown in Fig. S4. Amplifica-  tion of a 674-bp DNA fragment located upstream of nfiR was performed using the primer set upF/upR, and amplification of a 637-bp DNA fragment located downstream of nfiR was performed using the set downF/downR (Table 3). Restriction enzyme sites (BamHI and HindIII) incorporated into the oligonucleotide primers to facilitate vector construction are underlined in the oligonucleotide sequences shown in Table 3. An 851-bp DNA fragment containing the Cm resistance cassette was amplified from the plasmid pKatCAT5 by PCR using the primers CmF and CmR. The three amplicons were fused into a 2.162-kb fragment in which the Cm gene is located between the other two amplicons by overlap extension PCR according to the strategy of PCR-based fusions (43). The fusion PCR product was then cloned into the multiple cloning site of the pMD18-T vector (TaKaRa, Japan). The resulting plasmid DNA was double digested with BamHI/HindIII and then cloned into the BamHI/HindIII site of pK18mobsacB (44). The resulting plasmid, pK18/delR, was mobilized from Escherichia coli into P. stutzeri strain A1501 by conjugation using pRK2013 (19) as the helper plasmid. After mating, cells were spread on LB plates containing 50 g/ml Km and 40 g/ml Cm to screen for clones in which pK18/delR had integrated into the A1501 genome via a single recombination event. Another recombination event was then induced to replace nfiR with cat and for removal of the Km r and sacB genes from the genome. A colony of a single recombinant was then grown in a nonselective LB medium at 30°C. Cultures were diluted and spread onto LB agar supplemented with 10% (wt/vol) sucrose and 40 g/ml Cm. The nfiR mutant strain was selected for kanamycin-sensitive and sacB-negative colonies. Correct recombination was checked using the primers testF and testR, followed by nucleotide sequencing of the amplicon obtained. The resulting nfiR deletion mutant, named A1801, was used for further study. The expression levels of flanking genes (PST2408 and PST2409) were compared in the wild-type strain and the nfiR mutant by qRT-PCR. The experimental data indicated that nfiR deletion did not significantly alter the expression levels of the two flanking genes (data not shown). To explore the function of the NfiR stem-loop containing the 11-nt site that pairs with nifD mRNA, four 501-bp complementation DNA fragments carrying nfiR genes with different mutated sites, including sites within their promoter and terminator regions, were synthesized by BGI Co. Ltd. and cloned into the BamHI/HindIII site of pLAFR3 as described previously (19). The resulting four different complemented plasmids, named pNfiR-mut1, pNfiR-mut2, pNfiR-mut3, and pNfiR-mut4, were used for further studies. Construction of the double mutant strain lacking nfiR and nfiS. A 2.2-kb fragment was double digested with BamHI/HindIII, in which the Spc gene is located between the upstream and downstream DNA fragments of nfiS by overlap extension PCR, and was then cloned into the BamHI/HindIII site of pK18mobsacB, generating pK18/delS (19). To engineer a ΔnfiR ΔnfiS double mutant, the homologous recombination plasmid pK18/delS was transformed into the genome of the nfiR mutant (A1801) by triparental mating, as described previously (19), and was chromosomally integrated upon first crossover selection for Km and Spc resistances. The second crossover cells were selected by culture on LB plates containing 10% (wt/vol) sucrose, spectinomycin (34 g/ml), and chloramphenicol (40 g/ml). Correct recombination was checked using the primers testSF and testSR (Table 3), followed by nucleotide sequencing of the amplicon obtained. The resulting ΔnfiR ΔnfiS double deletion mutant, named A1901, was used for further study.
5= rapid amplification of cDNA ends. The transcriptional start site of the nfiR gene was determined using 5= rapid amplification of cDNA ends (5= RACE) assay (Invitrogen, USA), following the manufacturer's instructions. The sequences of the primers GSP1 and GSP2, which are specific for the nfiR gene tested here, are listed in Table 3. Products were cloned into a pMD18-T vector (TaKaRa, Japan) and sequenced to map the 5= RACE end of the transcript.
Measuring the stability of nfiR, nifD, and nifK transcripts. To assay nfiR, nifD, and nifK mRNA stability, 2-ml bacterial samples incubated for 5 h under nitrogen fixation conditions were collected at different times (0, 1, 3, 5, and 7 min, or 0, 5, 10, 15, 20, 25, and 30 min) right after the addition of rifampin (400 g/ml). Two volumes of RNAprotect (Sigma, USA) were added to each sample to prevent RNA degradation, and the samples were immediately frozen in liquid nitrogen. Total RNA was isolated from each sample for use in estimating mRNA levels by qRT-PCR. The primers used for measurement of the half-life of the selected gene mRNAs are listed in Table S2 in the supplemental material, and qRT-PCR was performed as described above. Data are presented as percent mRNA levels relative to time point zero. DNase I footprinting assay. The DNA probe was prepared by PCR amplification of a 264-bp nfiR promoter region using the primers FP-F and FP-R ( Table 3). The assay was performed in a total volume of 40 l containing 500 ng of probe and various concentrations of purified NtrC protein. After incubation for 30 min at 25°C, 10 l of solution, containing approximately 0.015 units of DNase I (Promega, USA) and 100 nmol of freshly prepared CaCl 2 , was added, and the sample was further incubated for 1 min at 25°C. The reaction was stopped by the addition of 140 l of DNase I stop solution (200 mM unbuffered sodium acetate, 30 mM EDTA, and 0.15% sodium dodecyl sulfate [SDS]). Samples were first extracted with phenol-chloroform and then precipitated with ethanol, and the pellets were dissolved in 30 l of Milli-Q water. The preparation of the DNA ladder, electrophoresis, and DNA sequencing were performed as described previously (45,46).
Microscale thermophoresis measurements. MST experiments were performed according to Zhan et al. (19). A set of 30-nt ssRNA oligonucleotides containing wild-type or mutated base-pairing regions of NfiR (N-NfiR probes) or regions complementary to nifD mRNA (N-nifDR competitors) was synthesized by the GenePharma Company, and their sequences are listed in Table 4. The wild-type and mutated N-NfiR probe molecules were labeled with 6-carboxyfluorescein (FAM). Four microliters of sample, containing 500 nM labeled probe and increasing concentrations (from 10 nM to 300 M) of a nonlabeled competitor, was loaded on standard treated silicon capillaries. Thermophoresis was monitored at 280 nm using a Monolith NT.115 instrument (NanoTemper Technologies GmbH), and the dissociation constants (K d ) were calculated as previously described (19,47). Data analyses were performed using NanoTemper Analysis software v.1.2.101 (NanoTemper Technologies, Germany).
Northern blotting. Total RNAs used for Northern blot analysis were prepared from P. stutzeri cells grown under nitrogen fixation conditions as described above using the SV total RNA isolation system (Promega, Madison, WI, USA). [␣-32 P]-dCTP-labeled probes were obtained as follows: PCR fragments of the nfiR gene were synthesized with P. stutzeri A1501 genomic DNA as the template and pairs of primers listed in Table 3. Ten micrograms of RNA was mixed with formaldehyde load dye, loaded into separate wells of a 1.5% Tris-acetate-EDTA (TAE)-agarose gel, and separated by electrophoresis. RNA was transferred to nylon membrane via capillary blotting, UV cross-linked, and probed overnight with an ncRNA-specific probe. Following hybridization and washing, Northern blots were developed and exposed to film.
Western blot assays for NifDK expression. Western blotting was performed using protein extracts of bacterial cells incubated for 5 h under nitrogen fixation conditions. A total protein loading control was used to quantify the expression levels of the target proteins in various samples. Protein concentrations were determined using the Bio-Rad protein assay reagent kit. Equal amounts of protein from each sample were separated by SDS polyacrylamide gel electrophoresis (SDS-PAGE) with an acrylamide/bisacrylamide ratio of 172:1 and were transferred to a polyvinylidene difluoride (PVDF) membrane (Amersham, USA) by electroblotting. The membrane was incubated with antisera raised against the MoFe protein (kindly provided by Ying Li, China Agricultural University, Beijing) for 2 h at 37°C and then washed three times in Tris-buffered saline (TBS)/Tween before being incubated with anti-rabbit secondary antibodies (Sangon, China) for 2 h. Detection was performed using an HRP-DAB chemistry kit (Tiangen, China).
Proteomic analysis. Bacterial cells were harvested and centrifuged at 8,000 ϫ g for 5 min at 4°C. Protein samples were then purified by phenol saturated with Tris-HCl (pH 8.6) and precipitated with five volumes of 0.1 M ammonium acetate in methanol at Ϫ20°C overnight. Supernatant containing the soluble protein fraction was immediately subjected to two-dimensional polyacrylamide gel electrophoresis (2-DE) for protein separation, and each sample was analyzed in triplicate. For MALDI-TOF mass spectrometry (MS) analysis, protein spots were excised from gels and digested with trypsin as previously described (48), and the various peptides generated were analyzed using a 4700 proteomics analyzer (Applied Biosystems, USA). Proteins were identified by automated peptide mass fingerprinting using GPS Explorer v3.5 software (Applied Biosystems, USA). The identified differentially expressed proteins were functionally categorized using the Gene Ontology Tool. Gene ontology (GO) enrichment analysis was performed using GOEAST.
Accession number(s). The RNA-seq data described here have been deposited in the NCBI Sequence Read Archive (SRA) database under the accession numbers SRR8369128 and SRR8369127.