Core regulon of the global anaerobic regulator Anr targets central metabolism functions in Pseudomonas species.

A comparative genome analysis of the global anaerobic regulator Anr regulon in five species of Pseudomonas with different life style was performed. Expression of this regulator was detected in all analyzed Pseudomonas. The predicted Anr regulon (pan-regulon) consisted of 253 genes. However, only 11 Anr-boxes located upstream of qor/hemF, hemN, cioA/PA3931, azu, rpsL, gltP, orthologous to PA2867, cspD, tyrZ, slyD and oprG, were common to all species. Whole genome in silico prediction of metabolic pathways identified genes belonging to heme biosynthesis, cytochromes and Entner-Doudoroff pathway as members of Anr regulon in all strains. Extending genome analysis to 28 Pseudomonas spp. spanning all phylogenetic groups showed Anr-boxes conservation in genes related to these functions. When present, genes related to anaerobic metabolism were predicted to hold Anr-boxes. Focused on the genomes of eight P. aeruginosa isolates of diverse origins, we observed a conserved regulon, sharing nearly 80% of the genes, indicating its key role in this opportunistic pathogen. The results suggest that the core Anr regulon comprises genes involved in central metabolism and aerobic electron transport chain, whereas those genes related to anaerobic metabolism and other functions constitute the accessory Anr-regulon, thereby differentially contributing to the ecological fitness of each Pseudomonas species.

Introduction

Pseudomonas species are metabolically versatile bacteria that can thrive in diverse environmental conditions. Aerobic respiration is the privileged energy generation mechanism[1] but several members of this genus are able to develop in a gradient of oxygen conditions, from aerobiosis to anaerobiosis, using different mechanisms for obtaining energy. The Anr regulator (for anaerobic regulation of arginine deiminase and nitrate reduction) is a key factor involved in metabolic plasticity regarding oxygen availability in Pseudomonas species[2,3]. Anr is homologous to Escherichia coli Fnr (fumarate and nitrate reductase regulator)[4,5]. Its ability to sense oxygen levels lies in either a [4Fe–4S]2+or a [2Fe–2S]2+cluster bound to four Cys residues. In E. coli, under low oxygen conditions, the reduced dimeric active form of Fnr binds to conserved DNA binding sites in promoters (Fnr-box) and regulates target genes transcription[6]. Besides the typical metabolic processes related to energy generation such as denitrification, arginine and pyruvate fermentation, Anr also controls a variety of metabolic functions as redox state maintenance, fimbria and cytochrome biosynthesis, secretion type III system, oxidative stress resistance and quorum sensing cascades[3,4,7-10].

Most of the knowledge about Anr modulated functions is based on Pseudomonas aeruginosa, the type species of the genus. In this bacterium, two CRP/FNR transcription factors, Anr and Dnr regulators, are involved in the response to oxygen availability and the presence of N-oxides[11,12]. By contrast, the knowledge of the regulatory role of Anr in other Pseudomonas species is still limited. In P. stutzeri, a facultative anaerobic bacterium, four Fnr-like regulators control the expression of denitrification in response to oxygen levels[13]. In P. putida, an obligate aerobic soil bacterium used as cell factory[14], Anr has an important role on expression level and coordination of the different terminal oxidases belonging to the branched aerobic respiratory chain in response to oxygen availability[2]. The role of Anr in different cellular process including polyhydroxybutyrate metabolism, redox state, oxidative stress resistance and biofilm development has been reported in the Antarctic bacterium P. extremaustralis under low oxygen conditions[15-17]. In the strictly aerobic P. protegens CHA0 (formerly P. fluorescens CHA0), Anr is required for hydrogen cyanide synthesis, a compound that contributes to biocontrol abilities of this strain[18].

Here we confirmed the expression of Anr mRNA and protein in 5 model Pseudomonas with different ecological niches, biotechnological applications and impact in human activities followed by a comparative in silico analysis of the Anr regulon. Bioinformatic analyses together with whole genome metabolic network reconstruction were used to determine the core and accessory regulon in these five bacteria. Inter and intraspecific variability of the functions controlled by this master regulator was examined by extending the analysis to other Pseudomonas spp. Our results predicted a set of core Anr-controlled genes related to major factors and pathways central to energy generation, in both obligate aerobes and facultative anaerobes, whereas a large set of genes showed species-to-species variation with respect to the presence of Anr-boxes, probably as a reflection of their physiological, biochemical and ecological properties.

Results

Characterization of Anr regulator in model Pseudomonas species

In order to examine the interspecific diversity of the Anr regulon, we selected five representative species of the Pseudomonas genus, including environmental and human and plant pathogens strains. The environmental strains comprised two soil isolates (P. putida KT2440 and P. protegens Pf-5) and the extremophile bacteria P. extremaustralis 14-3b isolated from a temporary water pond in Antarctica (Table 1), whereas pathogens included P. aeruginosa PAO1, an opportunistic human pathogen, and P. syringae pv. syringae B728a, a plant pathogen. Genome sizes ranged from 6.09 to 7.07 Mb and G + C content from 59.2 to 66.6% (Table 1). The anaerobic global regulator anr gene was found in P. syringae pv. syringae B728a and P. protegens Pf-5 genomes in addition to P. extremaustralis 14-3b, P. putida KT2440 and P. aeruginosa PAO1, in which it has been reported before[2,5,16]. The genomic region where anr is located showed similarity and high degree of synteny in all Pseudomonas species (Supplementary Fig. S1), with the apt gene, encoding adenine phosphoribosyl transferase, located always upstream of anr. However, the hemN gene, encoding coproporphyrinogen III oxidase, is flanking downstream anr in P. extremaustralis and P. putida KT2440, whereas an orthologous to pemB, encoding a T3SS effector[19], resides between anr and hemN in the remaining analyzed species (Supplementary Fig. S1). Anr sequence was highly conserved among the different species with an average 93.0% of amino acid similarity and 82.7% of identity. All the sequences harboured the four cysteine residues (Cys) that comprise the sensor oxygen domain and the HTH domain necessary for DNA regulation (Supplementary Fig. S2). Phylogenetic tree of Anr deduced protein sequences showed that the sequences clustered according to the species groups reported before[20], suggesting that anr belongs to the core genome of Pseudomonas species (Supplementary Fig. S3).

Table 1

Characteristics of selected Pseudomonas species.

Strain	Source	Relevant characteristic	Genome Size (bp)	Genome Accession #	Protein #	%G + C
P. aeruginosa PAO1	Wound isolate	Human opportunistic pathogen	6,264,404	AE004091.2	5,572	66.6
P. extremaustralis 14-3b	Temporary pond in Antarctica	Extremophile	6,586,240	AHIP00000000.1	5,870	60.7
P. protegens Pf-5	Rhizosphere of cotton seedlings	Biocontrol agent	7,074,893	CP000076.1	6,250	63.3
P. putida KT2440	Derived from toluene-degrading soil isolate, mt-2	Soil bacterium, GRAS	6,181,873	AE015451.1	5,564	61.5
P. syringae pv syringae B728a	Snap bean leaflet in Wisconsin	Plant pathogen	6,093,698	CP000075.1	5,089	59.2

Genome information included for comparison, was obtained from National Center for Biotechnology Information (www.ncbi.nlm.nih.gov). G + C, guanine plus adenosine mols percent. GRAS: generally recognized as safe.

Subsequently, microaerobic growth in presence and absence of nitrate was analyzed in the five Pseudomonas spp. All of them were capable to grow under both conditions (Fig. 1A). No significant differences in microaerobic growth with or without nitrate were observed in 14-3b, Pf-5, KT2440 and B728a (t-test, P > 0.05). However, PAO1 showed significant differences that can be attributed to its capability to efficiently use nitrate through a complete denitrification process (t-test, P > 0.05, Fig. 1A). Although P. extremaustralis 14-3b is also able to perform nitrate reduction, it did not show a significant increase in growth at the assayed times. P. syringae pv. syringae B728a presented the lowest growth level while the other analyzed strains reached OD600nm values between 2 and 3 after 30 h of incubation (P < 0.05 one-way Anova-test; Fig. 1A). In addition, expression of anr gene was observed in the five species under microaerobic conditions (Fig. 1B), although significant differences in the expression among them were observed. P. putida KT2440 and P. syringae B728a showed the lowest values for anr expression (P < 0.05 one-way Anova-test; Fig. 1B). However, and for first time, native Anr proteins were detected in both species by western blot assays under microaerobic conditions, using an anti-Anr antibody designed based on amino acid sequence of P. extremaustralis (Fig. 1C). Overall results showed high conservation of anr and its expression in all chosen Pseudomonas species.

Figure 1

Characterization of microaerobic growth and Anr expression in Pseudomonas species. (A) Growth under low oxygen conditions with and without KNO3. (B) Expression of anr measured by Real time qPCR. (C) Western Blot analysis of Anr. Lines: 1. Purified recombinant Anr protein of P. extremaustralis used as positive control 2: Marker PageRuler prestained protein ladder (Thermo scientific); 3: P. extremaustralis; 4: anr mutant of P. extremaustralis[16] used as negative control. 5: P. syringae B728a: 6: P. putida KT2440. Original Western-blot assay with different exposure times used to build C are shown in Fig. S4. In A: microaerobic cultures were carried out in LB medium supplemented or not with KNO3 in sealed bottles with 1:2 medium to flask volume ratio and low agitation (50 rpm), incubated for 30 h. In B and C: cultures were performed with KNO3. Bars represent mean ± SE. *: indicate significant differences P < 0.05. P. aeruginosa PAO1 (PAO1); P. protegens Pf-5 (Pf-5); P. putida KT2440 (KT2440), P. extremaustralis 14-3b (14-3b) and P. syringae pv. syringae B728a (B728a).

In silico comparison of Anr regulon in Pseudomonas species

Once anr expression was verified in the five Pseudomonas model species, an exhaustive genome analysis was performed in order to detect putative Anr-consensus binding sequences (Anr-box) using the PRODORIC software. The complete list of the putative Anr target genes was summarized in Supplementary Table S1. The comparison of the Anr regulons in all species showed the presence of a relatively similar number of putative Anr-boxes in P. extremaustralis (136) and P. aeruginosa PAO1 (122), followed by P. putida KT2440 (113), P. protegens Pf-5 (96) and P. syringae B728a (92) (Supplementary Table S1), being boxes present in: orthologous genes present in all species, genes not shared among all species, and strain-specific genes. The predicted Pseudomonas Anr regulon (pan-regulon) consisted of 253 genes that contains Anr-box in at least one species and covered a wide range of functional categories (Supplementary Fig. S5). Among them, we focused on a subset of 105 orthologous genes that were present in all species although not in all of them contained Anr-boxes. The comparison of these 105 genes revealed a core Anr regulon shared in all analysed species, which putatively could regulate 13 genes, as the Anr-boxes were located at adequate distance in the intergenic region of rpsL, hemN, azu, gltP, orthologous to PA2867, encoding a protein related to chemotaxis, cspD, sylD, tyrZ and oprG, but also upstream of genes transcribed in opposite direction: qor/hemF and cioA/PA3931 (Fig. 2A, Supplementary Table S1). On the other hand, the analysis showed that P. extremaustralis had 3 unique Anr-boxes regulating groEL, ftsB and hisG (Fig. 2A). In P. aeruginosa PAO1, 5 unique Anr-boxes were found upstream of PA0864, PA1558, yhbH, PA4902 and yieO, whereas in P. putida KT2440, 4 unique Anr-boxes were identified upstream of glxR, capB, dcd and PP1135 and only 1 for P. syringae B728a located upstream of Psyr_3920 (rpmE2) (Fig. 2A). No unique Anr-boxes were found for P. protegens Pf-5. When non-pathogenic model species (P. extremaustralis 14-3b, P. protegens Pf-5 and P. putida KT2440) were compared, Anr-boxes were identified in the promoter zone of algZ, hemC, dnaK, PP2161, ppx, PP0175 and pepN, that can be grouped into chaperone, hypothetical protein and regulator protein functions (Fig. 2A, Supplementary Fig. S5). Surprisingly, we could find only one Anr-box shared between the pathogenic species P. aeruginosa PAO1 and P. syringae B728a upstream of the radA gene (Fig. 2A), encoding a DNA repair protein previously related to antibiotic resistance[21].

Figure 2

Venn diagram showing the number of genes containing Anr-boxes. (A) Comparison of Anr-boxes located upstream of conserved genes present in all Pseudomonas species. (B) Venn diagram showing the number of genes with predicted Anr box shared between P. extremaustralis 14-3b and the other strains. Areas within the pairwise Venn diagrams are drawn to scale and the total number of genes in each species is indicated in parentheses. The numbers in brackets represent unique genes (absent from the other genome). P. extremaustralis 14-3b (14-3b); P. aeruginosa PA01 (PA01); P. protegens Pf-5 (Pf-5); P. putida KT2440 (KT2440); P. syringae pv. syringae B728a (B728a).

As an approach to validate the prediction of Anr controlled genes, experimental data available in GEO database[22] or from literature were analyzed. Interestingly, in P. aeruginosa the core Anr-controlled genes, hemN, azu, PA2867 and gltP were upregulated in anaerobiosis, whereas rpsL and tyrZ were found downregulated. Furthermore, in PAO1 anr deletion mutants the azu, hemN, hemF and oprG genes presented lower expression respect to the wild type strain[3,10] whereas cioA showed higher expression in the mutant[10]. In P. extremaustralis, 5 of the predicted core Anr-controlled genes showed differential expression under microaerobic conditions, with azu, hemN, orthologous to PA2867 and gltP being upregulated and cioA downregulated; accordingly this gene showed increased expression in an anr mutant in comparison with the wild type strain, suggesting that Anr acts as repressor of cio genes[15,23]. Additionally, in P. putida KT2440 the cioA was reported to be repressed by Anr[2] and evidences regarding Anr dependent upregulation of azu and oprG were also found[24].

Anr regulon pairwise comparison’s between P. extremaustralis and the other Pseudomonas species

To analyze variations in Anr regulated genes from species to species, we performed comparisons between P. extremaustralis 14-3b regulon and the regulons of the other Pseudomonas. As mentioned, this extremophile species showed the largest set of putatively regulated genes (136) followed by P. aeruginosa PAO1 with 122, with which it shared 69 putatively Anr-regulated genes (Fig. 2B). Similar overlap of regulated genes was observed when Anr regulon of P. extremaustralis 14-3b was compared with P. protegens Pf-5 and P. putida KT2440 regulons sharing 73 and 69 genes, respectively (Fig. 2B). P. syringae B728a presented the smallest Anr-controlled regulon with 92 genes (Fig. 2B) and shared only 53 genes with P. extremaustralis 14-3b.

The higher relatedness between regulons belonging to P. aeruginosa PAO1 and P. extremaustralis seem to be associated to the presence of several anaerobic metabolic functions (Supplementary Table S1, Fig. S6), as both species can use nitrate as electron acceptor and can perform arginine and pyruvate fermentation under low oxygen conditions[4,8,25]. In fact, a shared set of genes was related to anaerobic metabolism like ack, dnr, nosR, narK1, narX, norC, arcD, and oxygen high affinity cytochromes (Supplementary Table S1, Fig. S6). Moreover, Anr-boxes were also detected in several genes important for environmental adaptability and to cope with stressful conditions like pilG, related to biofilm formation, the catalase katA, aer, related to aerotaxis, and alkB, a gene that was found to be involved in microaerobic degradation of alkanes in P. extremaustralis[23]. Remarkable, hcnA, one of the first targets of Anr regulation in P. aeruginosa[26] was not shared with P. extremaustralis as hydrogen cyanide synthesis genes are absent in this last bacterium (Supplementary Table S1, Fig. S6). However, boxes on hcnA were also predicted for the plant associated bacteria P. protegens Pf-5 and P. syringae B728a (Supplementary Table S1). Interestingly, as a strain-specific function, P. extremaustralis 14-3b showed the presence of Anr-boxes in phbR/phbC, related to polyhydroxybutyrate (PHB) metabolism that were previously analyzed in an anr mutant[16]. This bacterium presents two different clusters for polyhydroxyalkanoates biosynthesis one for polyhydroxybutyrate (PHB), a short length PHA, and another for medium chain length polyhydroxyalkanoate (mclPHA)[27,28]. Interestingly, in P. extremaustralis Anr-box was found only in the promotor zone of the PHB related genes, but not upstream the mclPHA, encoding the operon typical of Pseudomonas species. In concordance, boxes in genes related to mclPHA were not found in P. aeruginosa.

Prediction of Anr regulation on the Pseudomonas metabolic network

According to our results, the core Anr-controlled gene set comprises only around 10% of the Anr regulon from each of the five analyzed species. However, Anr could control the expression of different genes belonging to the same metabolic pathway leading to the whole regulation of a particular metabolic route.

To test this hypothesis, we performed an in silico prediction of metabolic networks (MN) of P. extremaustralis using genome information as was described in the methods section and the metabolic reactions containing boxes were identified for all the species. The comparative analysis revealed that, the superpathway of heme biosynthesis, the electron transport chain and the Entner-Doudoroff pathway (EDP) involved different genes in each species that presented Anr-boxes, suggesting that these metabolic pathways are Anr dependent (Fig. 3).

Figure 3

Metabolic network of Pseudomonas species. Each node represents a predicted reaction in the Pseudomonas metabolism. Heme biosynthesis, Entner Doudoroff pathway and cytochromes are shown in detail.

Heme biosynthesis is a key process that provides this component for different cellular proteins[29]. In addition to qor/hemF and hemN that were identified as belonging to the Anr core regulon, hemE showed boxes in four of the five model species being absent in P. aeruginosa PAO1, hemC presented Anr boxes in 3 of the 5 analyzed species (P. extremaustralis 14-3b, P. putida KT2440, P. protegens Pf-5) while hemB showed Anr box in 2 of them (P. extremaustralis 14-3b and P. protegens Pf-5). P. extremaustralis showed putative Anr binding sites in all the mentioned genes, suggesting a redundant and a high regulation of this pathway. Reactions predicted as Anr controlled in heme biosynthesis pathway showed different normalized betweenness centrality with values ranging from 0.004 for hemF to 0.074 for hemC (Fig. 3).

EDP converts glucose into pyruvate with the production of NADH[30]. Although absent in the Anr-core, the presence of Anr-boxes in different genes belonging to EDP, such as fbp (fructose 1,6 biphosphate), gap (glyceraldehyde 3-phosphate dehydrogenase) or edd (phosphogluconate dehydratase) (Fig. 3) in the five Pseudomonas species suggests that regulation of these functions by Anr is spread among this genus. EDP and glycolysis represent fundamental metabolic pathways and include one of the most central reactions in the whole metabolism catalyzed by pyruvate kinase (0.73 normalized betweenness centrality), a key node representing a very interconnected reaction (Fig. 3). Reactions predicted as Anr controlled, showed intermediate centrality with the higher value for the reaction catalyzed by edd product (0.053 normalized betweenness centrality) in comparison with those catalyze by the enzymes encoded by fbp and gap (0.028 and 0.024 normalized betweenness centrality, respectively).

Regarding the electron transport chain, besides cioA that belongs to the core Anr regulon, nuoA presented boxes in 3 species (Fig. 3). The cbb3-type cytochrome c oxidases, cbb3-1 and cbb3-2, encoded by the gene clusters ccoN1O1Q1P1 and ccoN2O2Q2P2 were studied in P. aeruginosa, as well as in P. putida and P. extremaustralis. However, P. putida cbb3-1 corresponds to cbb3-2 of P. aeruginosa[2]. Then, to avoid confusion, in Fig. 3 we decided to use ccoN (without number) to include both designations. Boxes were detected in ccoN for all species, with the exception of P. syringae (Fig. 3). For P. extremaustralis, Anr boxes were observed in the promoter zones of the three genes (cioA, nuoA and ccoN). All the predicted Anr controlled reactions involving cytochromes presented similar normalized betweenness centrality showing values around 0.1 (0.129, 0.100, and 0.089 for ccoN, cioA and nuoA, respectively).

Interspecific Anr regulon among Pseudomonas species

Our integrative Anr regulon analysis allowed the identification of both core Anr-controlled genes and core Anr-controlled pathways in the five representative Pseudomonas species. To better characterize core targets of Anr-regulation in silico analyses were extended to 28 Pseudomonas species belonging to different taxonomical groups or subgroups described before[20]. We searched for Anr-boxes in the promoter zone of genes belonging to the core regulon (orthologous to PA2867, hemN, azu and cioA) and, based on the metabolic network analysis, the presence of Anr-boxes upstream of genes identified in EDP (gap, edd, fbp) was also analyzed. All of analyzed species presented Anr-box like sequences upstream of hemN reinforcing the importance of heme biosynthesis as key target of this regulator and recognizing this gene as a core target of Anr regulation (Fig. 4). Regarding core Anr regulon genes involved in electron transport chain (cioA and azu) Anr-boxes were predicted in the promoter zone of both genes in most of the analyzed species (19/28), but were absent in species belonging to P. stutzeri subgroup (Fig. 4). Importantly, although cioA was not predicted as Anr-regulated in this subgroup, it was reported that the Anr-like regulator, fnrA, controls the cytochrome ccoN in P. stutzeri[13], reinforcing the idea that core regulation may be focused in the pathway and not in individual genes. It has been also reported that, unlike P. aeruginosa PAO1, in P. putida KT2440 the cytochrome cyoA is controlled by Anr[2]. Since only those Anr boxes located up to 300 bp upstream of the ATG translation initiation codon (see Material and Methods section) were considered, and the Anr box for cyoA in P. putida KT2440 was located outside (upstream) of this window was not considered in our analysis. However, this report also gives support to the idea that electron transport chain is an important target for Anr regulation. Anr-boxes were detected upstream the orthologous to PA2867 related to chemotaxis in most of the species except in P. stutzeri subgroup (Fig. 4). In addition, 26 of the 28 analyzed species presented at least one Anr-box like sequence in gap or edd involved in EDP (Fig. 4). Only 2 P. stutzeri strains did not show Anr-box in gap or edd. Remarkable, an Anr box upstream fbp promoter zone was only presented in 5 of the 28 analyzed Pseudomonas (Fig. 4), suggesting that regulation in this target gene could not be generalized, in concordance with the metabolic network analysis, in which only 3 of the five model species had Anr-boxes in this gene (Fig. 3).

Figure 4

Extended analysis of Anr putative common conserved target genes among 28 Pseudomonas species belonging to different phylogenetic groups or subgroups. Orange color indicates presence of Anr box in the promoter region of the gene and white color indicates absence of Anr box. NP: gene not present.

As mentioned, Anr was traditionally related to denitrification, pyruvate and L-arginine fermentation and some cytochromes genes[26]. According to our prediction genes related to these functions that presented Anr-boxes in P. extremaustralis and P. aeruginosa PAO1 represent species-variable sets of Anr-controlled genes (accessory genes). Then, three genes belonging to anaerobic metabolic pathways were chosen for extending the analysis to the selected 28 species: ack (pyruvate fermentation), arcD (arginine–ornithine transporter for arginine fermentation) and narX (nitrate sensor). Pseudomonas spp. that presented genes related with nitrate metabolism (only 11/28, Supplementary Fig. S7) always showed an Anr-box upstream the ATG of narX gene (Fig. 4). Similarly, a search for arginine fermentation genes showed that they were present in 24 of the 28 analyzed species (Supplementary Fig. S7). In the case of arcD, an Anr-box like upstream ATG was present in 20 species. Only in P. putida subgroup and P. mediterranea CFBP 5447 the Anr box was absent (Fig. 4). Anr boxes upstream of ack were only observed in 4 species including P. fulva and those belonging to P. aeruginosa and P. oleovorans subgroups (Fig. 4).

Anr regulon in environmental and clinical strains of P. aeruginosa

The opportunistic pathogen P. aeruginosa is able to get energy from versatile respiratory mechanisms as well as from fermentative systems, which contribute to its ubiquitous distribution and persistence in diverse environments either under aerobic or anaerobic conditions. Therefore, we next wondered about the variability of functions controlled by Anr among different strains of P. aeruginosa. In addition to PAO1 strain, we analyzed the intraspecific diversity of the Anr regulon by exploring the set of genes controlled by Anr in seven P. aeruginosa strains from diverse origins: three isolated from environmental habitats (MTB-1, YL84, SJTD-1) and four clinical isolates from chronic cystic fibrosis (CF) lung infections (LESB58, DK2, RP73, SCV202065) (Supplementary Table S2).

Interestingly, the in silico analysis on the seven additional strains of P. aeruginosa did not render a significant increase in the size of Anr regulon observed in PAO1. Only 9 additional genes were incorporated to the whole P. aeruginosa pan regulon of 131 genes, 25 of which (20%) were absent or did not show the upstream Anr box in at least one of the analyzed genomes, indicating that the biological functions controlled by Anr are highly conserved among P. aeruginosa strains (Supplementary Table S2). Furthermore, no differential patterns were observed in relation to the environmental origin of the strains. As expected, the previously identified interspecific set of core genes, including fbp, was highly conserved in all P. aeruginosa strains and entirely belonged to the core genome of this species (Supplementary Table S2). In addition, the core Anr regulon was also composed by other genes related to the metabolic response to low-oxygen conditions (ccoN2) and the cytochrome c peroxidase (ccpR), denitrification (dnr, nirS, nirQ, norC, nosS, narK1, narX, narL), arginine fermentation (arcD) and other fermentation pathways (adhA, PA2119 and ackA). Furthermore, genes involved in key adaptive strategies, such as biofilm formation (pilG), fimbria production (cgrA), hydrogen cyanide production (hcnA) and biodegradation (alkB) were also included. Interestingly, the set of shared genes under the control of Anr include genes that were described to be regulated by quorum sensing[10], such as hsiA2 (type IV secretion locus), PA3913 (putative collagenase) and PA5232 (part of a putative ABC transporter), in addition to hcnA, nosR, narK1 and ccpR.

Most of the genes that belong to the core Anr regulon are also part of the P. aeruginosa core genome (Supplementary Table S2). On the other hand, the accessory genome of P. aeruginosa includes genes with conserved putative regulatory Anr-domains as well as genes of unknown function, which showed high strain-to-strain variability with respect to the presence of Anr boxes (Supplementary Table S2). Among them, there was a small set of genes (9) which contained Anr-boxes and was not present in the genome of PAO1 (extraPAO1).

The survey of the 13 interspecific Anr-core genes in the 95 assembled genomes available in Pseudomonas DB showed that they were conserved in almost 100% among the P. aeruginosa strains (Supplementary Table S3). In contrast, the set of strain-variable extraPAO1 genes was present in less than 50% of the genomes (Supplementary Table S3).

Discussion

Pseudomonas genus shows remarkable metabolic and physiological versatility, enabling colonization of diverse ecological niches. From human opportunistic pathogens, as P. aeruginosa, to P. extremaustralis an Antarctic bacterium with biotechnological interest, a wide range of cellular functions can be found in these species in response to oxygen variations. These functions include the presence of high oxygen affinity cytochromes, the utilization of nitrate as alternative electron acceptor and arginine and pyruvate fermentation[2,4,8,31] that have been described as controlled by Anr. However, anaerobic metabolism is not conserved among Pseudomonas species, while P. aeruginosa PAO1 can carry out a complete denitrification process[32], P. extremaustralis is only able to reduce nitrate to nitrite, and both species can perform arginine and pyruvate fermentation[16,25]. On the contrary, P. putida KT2440 is not able to use nitrate as electron acceptor and neither carry out pyruvate fermentation[33]. However, less experimental information is available for other Pseudomonas species. In this work, we found that less than 50% of the 33 analyzed Pseudomonas genomes presented genes encoding nitrate reduction enzymes while arginine fermentation appeared as a more extended metabolism among them. Cytochromes with different oxygen affinities and characteristics like azurin and cioA, among others, could be found in the majority of the species suggesting a wide battery of genes for oxygen utilization.

Current knowledge about the Anr role in Pseudomonas is mainly based on P. aeruginosa PAO1 physiology and in lesser extent in P. extremaustralis and P. putida KT2440. P. aeruginosa PAO1 possesses not only the Anr regulator but also Dnr, which is sensitive to NO and is mostly related to denitrification control[34]. Notably, the Anr-box and Dnr-box are indistinguishable[3]. Due to this observation our in-silico analysis did not differentiate these regulators. It is important to point out that in the non-denitrifier analysed species, this regulator seems to be absent when a BlastP analysis is carried out.

Results of this work show Anr expression at mRNA and protein level in five Pseudomonas species including P. protegens Pf-5 and P. syringae pv. syringae B728a in which was not previously studied and suggest that Anr not only control a set of genes of the core genome, but also elements of the accessory genome, encoding functions probably associated with adaptation of each species. Whole genome in silico prediction of metabolic pathways results indicate that primitive or more conserved functions that can be attributed to Anr regulon are those related to carbon central metabolism and the optimization of oxygen utilization (heme metabolism and different oxygen affinity cytochromes). Although Anr was first described controlling anaerobic related genes, these functions seem to belong to the accessory Anr regulon, since the genes encoding key enzymes of anaerobic functions were absent in the majority of the species.

This is particularly interesting for biotechnological applications. For example, P. putida KT2440 has emerged as a new engineerable bacterium that displays as advantages a great metabolic adaptability, high tolerance to stress and is generally recognized as safe (GRAS) microorganism[14,35]. The manipulation of central metabolism is a key step for the production of different metabolites of interest[35], particularly in the context of oxygen supply. In this bacterium glycolysis is reported to proceed through ED pathway, particularly through a cycle including not only ED but also pentose phosphate pathway and gluconeogenic reactions from the upper glycolysis[36]. In this context, the presence of Anr-boxes detected in this work in edd and gap genes could be relevant for metabolic engineering. Moreover, hexR and gntR regulators for glucose usage[37,38] also presented Anr-boxes. In addition, reactions involved in these central metabolic routes endow to P. putida KT2440 with high NADPH regeneration rates[36], allowing this bacterium to cope with oxidative stress and making it a remarkable host for redox-intensive reactions[14]. The coordination of carbon usage with oxygen availability could improve the efficiency in metabolism and can be used for biotechnological purposes. It has been shown that this bacterium possesses, in addition to the originally described Anr, two proteins belong to the Fnr/Crp family which are less sensitive to oxygen[39]. The presence of at least three oxygen sensitive regulators in this aerobic bacterium suggests their importance for environmental adaptability, and for fast and efficient response to changes in oxygen availability.

Soil presents oxygen gradients and microenvironments, which are variable according to distinct environmental conditions and/or bioactivities and interactions that are highly heterogeneous in the rhizosphere, whereas oxygen tension is low in soils with high humidity[40]. Thus, the control of several functions in plant-associated bacteria such as P. protegens Pf-5, P. extremaustralis or P. syringae B728a could be crucial for environmental survival. Indeed, biocontrol functions are regulated by Anr in different Pseudomonas species[18,41] and in this work we detected Anr-boxes not only in genes related to central metabolism but also to other genes related with stress resistance and colonization (Fig. 5). These functions included some genes belonging to the core Anr regulon such as cspD encoding a cold shock protein, and the orthologous to PA2867 related to chemotaxis, besides others stress resistance associated genes like katA, dnaJ, dnaK and groEL. Genes related to motility that are also relevant for colonization such as pilG, encoding a component of Type IV pili implicated in cellular adhesion and morA, involved in signal transduction and flagella development, were detected. Decreased expression of both genes was observed in an anr mutant of P. extremaustralis[17].

Figure 5

Proposed model showing conservation of functions controlled by Anr.

P. aeruginosa is a versatile opportunistic pathogen widely distributed, capable to respire oxygen as well as nitrogen oxides. Currently different reports indicate that its mode of respiration can be related to its distinct infection strategies, linking aerobic respiration to acute infections and microaerobic and anaerobic respiration to chronic infections mediated by biofilm mode of growth[42,43]. Particularly in the cystic fibrosis context, P. aeruginosa growths in oxygen restricted environments imposed by the thickened CF mucus, with local exacerbated hypoxia, and its anaerobic physiology is believed to play a key role in persistence functions[44-46]. Recently, it was reported that oxygen metabolism is a target in evolution of P. aeruginosa clinical clones[47]. In this sense, overexpression of denitrification and arginine fermentation, as well as certain high-affinity cytochrome oxidases genes[3], indicate that Anr activity is high in vivo[48,49]. Moreover, several genes expressed under the control of Anr, like azu, cytochromes, ccpR and icd were reported as genetic markers for the metabolic adaptation to the CF lung environment, emphasizing the importance of oxygen metabolism during infection[50]. Our results are in line with these reports, since the Anr regulon is highly conserved among P. aeruginosa strains. The core Anr regulon was extensive and no major differences among the reference PAO1 strain, environmental and CF strains were observed, suggesting a pivotal role for Anr not only in the CF host but also in natural environments. Recently, examination of the quorum-sensing regulon from different free-living and host-associated P. aeruginosa strains revealed that quorum control of gene expression has a strain variable component, exhibiting the habitats from which strains were isolated[51]. On the contrary, we observed that genes included in Anr core regulon mainly belong to the core genome, supporting the view that Anr regulatory interactions appear to be mostly conserved. Thus, it seems that only a small strain-variable set of genes, related to accessory functions, would have been coupled to the Anr regulon during niche diversification among P. aeruginosa strains.

Thus, we proposed that the landscape of Anr regulon functions includes a set of more conserved functions such as those related to the central metabolism (core functions), others with an intermediate degree of conservation such as those related to stress responses and adaptability and finally, those that would be accessory, including anaerobic metabolism, which was surprisingly the first target described for Anr (Fig. 5).

Evolutionary expansion of the different species of Pseudomonas led to a great variety of the enzymes for energy metabolism. However, the regulatory machinery for energy metabolism based on Anr seems to be less variable, targeting key metabolic pathways that are common within the genus, inter and intraspecifically. The information provided in this study might be worth considering for exploring biotechnological innovations based on microbial metabolism.

Methods

Bacterial strains

The following Pseudomonas species were selected for analysis: P. aeruginosa PAO1[52], P. putida KT2440[53], P. protegens Pf-5[54], P. syringae pv syringae B728a[55] and, P. extremaustralis 14-3b[56]. Characteristics and genome accession number of the selected species are shown in Table 1.

In silico analysis of Anr sequence

Anr nucleotide and protein sequences were obtained from public data bases (anr locus_ tag for P. aeruginosa PAO1: PA1544, P. putida KT2440: PP_4265, P. protegens Pf-5: PFL_1910, P. syringae pv.syringae B278a: Psyr_3425 and P. extremaustralis 14-3b: PE143B_0108070). Sequence were compared using Clustal Omega[57].

Phylogenetic analysis

Sequences were analyzed using the following programs available on line: BLAST (http://blast.ncbi.nlm.nih.gov/Blast.cgi), and bioinformatic tools included in the RAST server (http://rast.nmpdr.org/). The Pseudomonas genome database[58] was also used to obtain comparative information. Phylogenetic analyses of proteins were performed using MEGA 5[59]. Phylogenetic trees were constructed using the neighbor-joining (NJ) method bootstrap analysis of 500 replicates and root on midpoint. Nucleotide analysis was carried out for the 16S rRNA subunit gene sequence. Accession number: PA5369.5 (P.aeruginosa PAO1), PFL_0119 (P.protegens Pf-5), Psy_RNA52 (P.syringae pv. syringae B728a), PP_16SA (P. putida KT2440) and PE143B_0130605 (P. extremaustralis).

In silico determination of the Anr regulon

Potential Anr regulated genes in the Pseudomonas species genomes were evaluated with PRODORIC software using the P. aeruginosa Anr-Dnr box and the Fnr-box of Escherichia coli K12 matrix[60]. This search criterion was chosen because Anr and Dnr regulatory proteins of P. aeruginosa share an indistinguishable DNA consensus binding site[3] and Anr is able to recognize Fnr binding sites[4]. For this reason, Anr, Dnr and Fnr consensus box sequences were included in this analysis. This approach, that involved the combination of several matrices included in PRODORIC, assigned a different score to each base position. All matrices used are very conservative in the bases located at both extremes of the box (4 bases each side) that are coincident in Anr and Dnr, and also for the homolog regulator Fnr, but allowed certain degree of flexibilization in the middle region of the box that has less conservation. The genome sequences of P. aeruginosa PAO1, P. putida KT2440, P. protegens Pf-5 and P. syringae pv. syringae B728a, available in the PRODORIC server were used, while for P. extremaustralis a manual genome search was performed. Only putative Anr-boxes located in intergenic zones up to 300 pb upstream of the ATG of the gene target were used for further analysis. Venn diagrams were constructed using InteractiVenn[61]. Manual analysis was also performed with selected genes in other representative Pseudomonas species. In case that the identity of the gene was not annotated, a BLAST analysis was carried out to compare with public data bases. Sequence of genes of other Pseudomonas species were obtained from Pseudomonas database[58].

For P. aeruginosa Anr pan regulon determination, the presence of Anr-boxes was predicted by using the “Search DNA motif” tool from the Pseudomonas Genome DB (http://www.pseudomonas.com) and the Anr 5′-TTGATNNNNATCAA-3′-motif previously established[3]. Whole genomes sequences from seven P. aeruginosa strains were used: PACS2, AES-1R, FRD1, RP73, c7447m, DK2, LESB58, SCV20265 all of them available in Pseudomonas Genome DB.

Intergenic regions containing Anr-boxes in orthologous genes that were not present in PAO1 regulon (extraPAO1 regions) were determined by using the same procedure mentioned before and the presence of the Anr box was confirmed by PRODORIC using the Anr-Dnr-40 matrix[3]. In order to assess the distribution of these extraPAO1 regions in other P. aeruginosa strains, Blastn alignments of upstream regions were performed against the 95 complete genome sequences that are deposited at the Pseudomonas Genome DB. In order to compare the distribution of the Anr core/accesory regulon, this extended analysis was also performed with the 13 genes included in the Anr core regulon obtained from the interspecies analysis.

Information on whether Anr-controlled genes belong to the core or accessory P. aeruginosa genome was assessed through the BACTOME database, which includes 101 genomes (99 clinical isolates, and reference strains PAO1 and UCBPP PA14)[62]. Genes with >90% homology which are present in all P. aeruginosa isolates are considered to be part of the core genome.

Growth conditions

Pseudomonas cultures were performed under microaerobic conditions in LB medium and LB supplemented with KNO3. Microaerobic cultures were performed using 50 ml of medium in 100-ml hermetically sealed bottles incubated at 30 °C and 50 rpm and harvested after 30 h of growth to ensure low oxygen conditions[4].

Quantitative real-time reverse-transcriptase PCR (qRT-PCR) experiments

Bacterial pellets of 5 ml of microaerobic cultures in LB medium supplemented with KNO3 were used for total RNA extraction using Total RNA Extraction Kit (RBC Biosciences). After treatment with DNaseI, cDNA was obtained using random hexamers (Promega) and Revert Aid Reverse Transcriptase (ThermoFisher Scientific, Waltham, USA) following the manufacturer’s instructions. qRT PCR was performed using a MyiQ2 Real-Time PCR Detection System (Bio-Rad Laboratories, Hercules, USA) and Real Time PCR mix (EvaGreen qPCR Mix Plus, no ROX, Solis Biodyne). The cycling conditions were as follows: denaturation at 95 °C for 5 min, 40 cycles at 95 °C for 25 s, 58 °C for 15 s, and 72 °C for 15 s, with fluorescence acquisition at 80 °C in single mode. For normalization 16S rRNA gene was used and relative changes in the expression of anr gene for microaerobic conditions was obtained through the relative standard curve method. Oligonucleotides used are detailed in Table S4.

Anr antibody development and Western Blot assays

Antibody against native Anr was developed by a rational design. Anr sequences were aligned using ClustalO. Exposed amino acids corresponding to a functional domain were identified by comparing with Fnr from Escherichia coli which structure is available in public database. An immunogenic peptide was chosen by taking into account the conservation in Pseudomonas species and its immunogenic characteristics. The immunogenic peptide was specifically designed to do not recognize Dnr protein from Pseudomonas species. The designed peptide was synthesized and the polyclonal antibody was produced in rabbit by Abmart (Berkeley, USA). Then, the antibody was tested using P. extremaustralis total proteins to ensure its specificity and sensibility. Western Blot analysis was performed to detect Anr in the Pseudomonas spp. Aliquots of 1.5 ml of cultures grown under microaerobic conditions in LB medium supplemented with KNO3 were centrifuged at 12,000 rpm for 2 min. Pellets were resuspended in 1 ml PBS buffer with 1 mM PMSF and sonicated (Cole Parmer, ultrasonic homogenizer 4710) on ice. The lysates were centrifuged for 2 min at 12,000 rpm and the supernatants were transferred to a new tube. Total proteins in the supernatant were quantified by Bradford method. Aliquots containing 50 µg of proteins were electrophoresed using standard protocols in 4–12% polyacrylamide gel and then transferred to a PDVF membrane using a Semi-dry transfer system (TE 70 Amersham). Membranes were blocked overnight at 8 °C in a solution containing 0.02 M Tris-HCl, 0.5 M NaCl (pH 7.5), 0.05% v/v Tween-20 (TBST) and 5% w/v skimmed milk. The membrane was incubated with the rabbit polyclonal anti-Anr for 1.5 h and afterwards with anti-rabbit horseradish peroxidase-conjugated secondary antibody (1:5000, Santa Cruz Biotechnology) for 1.5 h. The resulting blots were incubated with enhanced chemiluminescence (ECL) reagent (GE Healthcare) and detected using Amersham imager 600 equipment. Negative controls were performed using an anr mutant of P. extremaustralis[16] and positive controls using purified Anr protein of P. extremaustralis.

Construction of the whole genome in silico metabolic network of Pseudomonas strains

To build the in silico metabolic networks (MN) of P. extremaustralis genome, we used the PathoLogic module of Pathway Tools v. 18.0[63]. PathoLogic creates a Pathway/Genome Database (PGDB) containing the predicted metabolic pathways of a given organism using as input an annotated genome in GenBank format. Metabolic reconstruction includes determining gene-protein-reaction-pathway associations, which are primarily based on the corresponding enzyme commission (EC) number. After automatic reconstruction, manual curation of pathways of interest was performed (tetrapyrrole biosynthesis I (from glutamate), tetrapyrrole biosynthesis II (from glycine), superpathway of heme biosynthesis from uroporphyrinogen-III, heme biosynthesis II (anaerobic), basically by filling of enzymatic “holes”, reactions in metabolic pathways for which no enzyme is identified, by looking for functional orthologs using BLAST. Afterwards, the metabolic network was exported in systems biology mark-up language (SBML) format for downstream analyses. Reactions involving only small-molecules were considered. After exporting MN construction, we filtered potential currency compound[64,65]. Cytoscape v. 3.1.0[66] was then used for network visualization and calculation of topological metrics. In this representation, nodes represent reactions, and there exists an edge between two nodes if a product of a reaction is used as substrate on the reaction that follows. Betweenness centrality (BC) was also calculated for each node in MN. The size of the nodes on the graph representation is proportional to their centrality value. High values of Betweenness centrality for a node indicate its participation as an important communication path, bridging different metabolic parts[67].

Statistical analysis

Differences between means were determined through one-way analysis of variance (ANOVA) and t-test, with a P < 0.05 significance level.

Supplementary Figures

Table S1

Table S2

Table S3

Table S4