ECF Sigma Factor HxuI Is Critical for In Vivo Fitness of Pseudomonas aeruginosa during Infection

ABSTRACT The opportunistic pathogen Pseudomonas aeruginosa often adapts to its host environment and causes recurrent nosocomial infections. The extracytoplasmic function (ECF) sigma factor enables bacteria to alter their gene expression in response to host environmental stimuli. Here, we report an ECF sigma factor, HxuI, which is rapidly induced once P. aeruginosa encounters the host. Host stresses such as iron limitation, oxidative stress, low oxygen, and nitric oxide induce the expression of hxuI. By combining RNA-seq and promoter-lacZ reporter fusion analysis, we reveal that HxuI can activate the expression of diverse metabolic and virulence pathways which are critical to P. aeruginosa infections, including iron acquisition, denitrification, pyocyanin synthesis, and bacteriocin production. Most importantly, overexpression of the hxuI in the laboratory strain PAO1 promotes its colonization in both murine lung and subcutaneous infections. Together, our findings show that HxuI, a key player in host stress-response, controls the in vivo adaptability and virulence of P. aeruginosa during infection. IMPORTANCE P. aeruginosa has a strong ability to adapt to diverse environments, making it capable of causing recurrent and multisite infections in clinics. Understanding host adaptive mechanisms plays an important guiding role in the development of new anti-infective agents. Here, we demonstrate that an ECFσ factor of P. aeruginosa response to the host-inflicted stresses, which promotes the bacterial in vivo fitness and pathogenicity. Furthermore, our findings may help explain the emergence of highly transmissible strains of P. aeruginosa and the acute exacerbations during chronic infections.

a cytoplasmic membrane-spanning anti-s factor involved in signal transduction from the periplasm to the cytoplasm; and (iii) an extracytoplasmic function (ECF) sigma factor that initiates transcription by directing core RNA polymerase (cRNAP) to the stimulus-responsive target gene(s) (8). The ECFs family is highly diverse, and a comprehensive classification has been reported based on more than 2,700 ECFs from hundreds of bacterial genomes (9). These ECFs often act orthogonally with limited cross talk and allow the partitioning of the transcriptional space. The high stringency of ECFs promoter recognition restricts the number of target genes to mount specific responses (10). In P. aeruginosa, the strains PAO1 and PA14 encode 19 and 21 ECFs factors, respectively. They mediate the functions of cell envelope stress response, production of the exopolysaccharide alginate, iron uptake, and pathogenicity (8,11).
The Hxu CSS pathway, which consists of three adjacent genes hxuIRA encoding ECFs factor, anti-s factor, and TonB-dependent outer membrane receptor, respectively (Fig. 1A), has been recently shown to be involved in heme signaling in P. aeruginosa, and mediates heme acquisition from host hemopexin (12,13). However, the target genes of the ECFs factor HxuI remain unknown. In the present study, we found that HxuI is highly conserved in different P. aeruginosa strains. In addition to heme, the host stresses of iron limitation, oxidative stress, hypoxia, and nitric oxide can all induce the expression of HxuI which, in turn, controls a variety of physiological functions associated with P. aeruginosa infection, including iron acquisition, anaerobic respiration, pyocyanin synthesis, and pyocin production. Most importantly, overexpression of hxuI in PAO1 promoted bacterial colonization and long-term infection in various murine infection models. Together, these studies suggest that HxuI is an important ECFs factor contributing to the in vivo fitness and pathogenicity of P. aeruginosa.

RESULTS
HxuI is highly conserved in P. aeruginosa. To analyze the conservation of the Hxu system, 723 P. aeruginosa clinical isolates with available genome sequences were analyzed by BLASTn (14). All strains possessed the huxIRA genes, and the hxuI gene is highly conserved among various P. aeruginosa strains (Fig. 1B), reflecting its important physiological functions.
Host stresses induce the expression of ECFr factor HxuI. To test whether Hxu responds to the host environment during infection, we infected mice with a laboratory strain PAO1 intranasally and collected bacterial cells from the bronchoalveolar lavage fluid (BALF) 6 h postinfection (pi). Quantitative real-time PCR (qPCR) assays showed that the hxuIRA genes were upregulated 9-, 6.1-, and 2.4-fold, respectively ( Fig. 2A), indicating that Hxu indeed responds to the host environment. To address the in vivo inducing signals, we tested a number of well-known host stress conditions to determine their effects on huxI gene expression. First, we tested hxuI expression under irondeficient conditions. In the PAO1 strain, hxuI expression was increased with the decrease of Fe(III) in ABTG medium (Fig. 2B). During host infection, phagocytic cells generate reactive oxygen species (ROS) such as superoxides, which are involved in antibacterial activity (15). Next, we tested hxuI expression under oxidative stress. In the wild-type (wt) PAO1 strain, exposure to 0.5 mM H 2 O 2 for 30 min induced a 43-fold increase in hxuI expression (Fig. 2C). OxyR is an H 2 O 2 -responsive regulator which activates the expression of defense genes against oxidative stress in P. aeruginosa (16). In an oxyR mutant, the expression of hxuI increased 86-fold even without the H 2 O 2 treatment, indicating a repressive effect of OxyR on hxuI expression, which was restored by complementation with the oxyR gene (Fig. 2C). Possible explanations revolve around the significant regulatory cross-talk in the management of redox-stress and iron homeostasis through ferric uptake regulator (Fur) (17). Nonetheless, these data indicated that hydrogen peroxide induces the expression of HxuI in the presence of OxyR.
P. aeruginosa is able to grow in the absence of oxygen through anaerobic metabolism, which influences infectivity as well as biofilm formation (18). To investigate whether HxuI responds to hypoxia, we determined hxuI expression by qPCR after a short incubation under anaerobic conditions. As shown in Fig. 2D, hxuI expression increased along with the anaerobic culture time in the PAO1 strain. There are two wellknown anaerobic sensors in P. aeruginosa: ANR and DNR (5). The expression of hxuI was increased under anaerobic growth conditions in an anr mutant, but not in a dnr mutant background (Fig. 2D). However, under aerobic conditions, hxuI expression increased by 164-fold in the dnr mutant but did not change in the anr mutant (Fig. 2D), indicating a negative regulation of hxuI by DNR. Complementation with a dnr gene restored hxuI expression levels in the Ddnr mutant (Fig. 2D). Since DNR is known to sense nitric oxide (NO), and NO-dependent DNR activity requires heme (18), we further tested whether NO directly induces the expression of HxuI. When the PAO1 wt strain was treated with 1 to 100 mM NO donor Spermine NONOate (19) for 30 min, the expression of hxuI was increased significantly in a dose-dependent manner (Fig. 2E). The above data suggested that oxygen limitation, likely via NO, induces the expression of HxuI.
Identification of the HxuI regulon genes. To gain insights into the HxuI regulons, a transcriptomic study was performed on P. aeruginosa PAO1 overexpressing the hxuI gene on an inducible expression vector pMMB. Most ECFs are subject to positive auto-regulation and directly induce the expression of corresponding TonB receptor, thereby enhancing their signaling effect for as long as the inducing conditions prevail (11). The expression level of TonB receptor hxuA was monitored by qPCR at various isopropyl b-D-thiogalactopyranoside (IPTG) induction times, and it was found that hxuA expression peaked at 2 h postinduction (Fig. S1). Accordingly, total RNA samples of PAO1/pMMB-hxuI and PAO1/pMMB strains were collected 2 h after induction by 1 mM IPTG, and these were then subjected to RNA-seq analysis. The overexpression of hxuI resulted in the upregulation of 87 genes and the downregulation of 22 genes at rates of more than 2-fold (P value # 0.05). Of the 87 genes significantly upregulated by HxuI, 24 genes are involved in anaerobic respiration and denitrifying redox chain, 18 are involved in metabolism, 16 in iron acquisition, 7 in biofilm formation, 7 in DNA damage response, and 6 in virulence ( Fig. 3A and Table 1).
Consistent with the above results, we further observed that (i) pyoverdine production in the hxuI-overexpressing strain was significantly higher than that of the wt strain during late exponential phases (Fig. 3D); (ii) under anaerobic condition, the growth rate of the hxuI deletion mutant was slower than that of the parent strain PAO1, while no growth defect was observed under aerobic conditions (Fig. 3E); and (iii) booming pyocyanin production was observed in PAO1 which overexpressed hxuI (Fig. 3F). In P. aeruginosa, a pair of tandem small RNAs, PrrF1 and PrrF2, promote the production of Pseudomonas quinolone signal (PQS), which activates pyocyanin production (20). In a prrF1,2 double mutant strain background, the activation of pyocyanin production by HxuI disappeared (Fig. 3F), indicating that HxuI-mediated activation of pyocyanin production requires the PrrF small RNAs.
PA2384 (Fur2) plays a major role in the regulation of hxuI regulon. HxuI was classified as an iron-responsive ECFs in previous studies, as its promoter region carries a Fur box (21). The ferric uptake regulator (Fur) plays a central role in iron response and is an essential gene in P. aeruginosa (22). The Fur protein employs Fe(II) as a cofactor and binds to a so-called "Fur box" in the promoters of iron-regulated genes, resulting in repression of the target genes; under low-iron conditions, the Fur protein is released from the operator sites and transcription takes place (21). Interestingly, RNA-  seq data analysis showed that PA2384 encoding a Fur homologue (designated Fur2) was upregulated 2.89-fold in the PAO1 overexpressing HxuI (Table 1). A HxuI-mediated transcriptional activation was observed in P fur2 -lacZ reporter with a 12-fold increase in b-galactosidase activity (Fig. 3C). Fur2 shares 35% amino acid identity with the N-terminal DNA-binding domain of Fur (PA4764), but does not bear the C-terminal domain of Fur which is responsible for iron binding and dimerization (23). To determine whether Fur2 is involved in the regulation of hxuI regulon, we examined the transcriptional activation effects of HxuI on fpv, nir, nos, and phz2 promoters in a Dfur2 mutant.
Overexpression of HxuI in the PAO1 strain led to significant increases in b-galactosidase activity in P fpv -lacZ, P nir -lacZ, P nos -lacZ, and P phz2 -lacZ fusions in the wild-type strain (Fig. 3C); however, these HxuI-mediated activations were diminished in the Dfur2 mutant background (Fig. S2), suggesting that HxuI-mediated activation of the fpv, nir, nos, and phz2 genes requires the presence of Fur2. Similarly, overexpression of hxuI resulted in 5-fold increases in b-galactosidase activity in PAO1 harboring P hxuI -lacZ fusion reporter, but not in the fur2 mutant background (Fig. 3G), indicating that Fur2 is also required for HxuI self-regulation. HxuI activates pyocin and bacterial cell lysis-related genes. To establish infection, bacteria must establish a strong foothold for colony development and also outcompete resident microbes. One strategy that potentially addresses both needs is the use of phage tail-like bacteriocins, which are broadly called pyocins in P. aeruginosa (24). Pyocins are released into the environment through explosive cell lysis which kills the producer and nearby competitor bacteria (25). This event also releases extracellular DNA which structurally supports biofilm formation (26). Looking at the RNA-seq data, we noticed that the whole gene sets encoding all three types of pyocins in P. aeruginosa were upregulated in the HxuI-overexpressing strain (Table S1), including the soluble S-type pyocin S2, S4, S5 (PA0985 in Table 1), the contractile R-type pyocin, and the noncontractile F-type pyocin (PA0646 in Table 1). To test whether HxuI is involved in pyocin production, neat supernatants from wt PAO1, DhxuI, and the complemented strain DhxuI/pAK1900-hxuI were spotted onto an L agar overlay containing the indicator P. aeruginosa strain PAK. As shown in Fig. 4A, the growth inhibition zone of the DhxuI/pAK1900-hxuI strain was larger than that of the wt and the DhxuI mutant, indicating higher intraspecies competitiveness that might be mediated by pyocin production. In addition, two sets of cell lysis genes, PA0807 (ampDh3)-PA0808 (immunity of AmpDh3) and alpDE (27), were upregulated at average rates of ;2.7-fold and ;2.1fold, respectively, in the HxuI overexpressor (Table 1). AmpDh3, a cell wall amidase, is thought to be delivered by the type VI secretion system locus II (H2-T6SS) to bacterial competitors and degrade the cell wall peptidoglycan of prey, thereby providing a growth advantage for P. aeruginosa (28). AlpDE belongs to the AlpBCDE self-lysis cassette which responds to DNA damage inflicted by the host immune system and enhances the virulence of P. aeruginosa (29). Under scanning electron microscopy (SEM), Promoter-lacZ fusions assay. P. aeruginosa PAO1 cells containing the lacZ reporter fusions in pDN19 and either plasmid pMMB (empty plasmid) or pMMB-hxuI were grown in LB with 1 mM IPTG until late exponential growth phase and analyzed for b-galactosidase activity. Error bars represent SD. *, P , 0.05; **, P , 0.01; ***, P , 0.001. more bacterial cell lysis was observable in the DhxuI/pAK1900-hxuI culture than in the wt strain culture (Fig. 4B); cells that overexpressed hxuI were inclined to gather together on the coverslips and form colony-like architectures, while DhxuI cells were scattered evenly (Fig. 4B). To investigate the transcriptional activation effects of HxuI on the above genes, the promoter regions upstream of PA0614 (R-pyocin), PA0646 (Fpyocin), pyoS5, ampDh3, and alpD were fused to the lacZ reporter and introduced into a PAO1 strain harboring the plasmid pMMB-hxuI. Significant increases in b-galactosidase activity were observed in all five fusions when HxuI expression was induced (Fig. 4C).
HxuI promotes P. aeruginosa infection in mice. A mouse lung infection model was used to determine the role of HxuI in acute infection. Mice were intranasally infected with the same amount of wt PAO1, DhxuI mutant, and hxuI complementary strain, respectively. At 12 h postinfection, the hxuI deletion mutant exhibited a significantly lower bacterial load in lungs compared to that of wt PAO1, and complementation with hxuI restored bacterial colonization capacity to wt levels (Fig. 6A). These data indicated that HxuI is critical for colonization in P. aeruginosa. A murine cutaneous abscess model was further employed as a chronic infection model (31) to determine the role of HxuI in long-term infection. To avoid the loss of HxuI expression vector, hxuI driven by tac promoter was inserted into the PAO1 chromosome via a mini-Tn7 vector (PAO1::P tac -hxuI), resulting in a constitutive expression of the hxuI gene (32,33). Mice were subcutaneously inoculated with 5 Â 10 6 CFU of wt PAO1, DhxuI, or PAO1::P tac -hxuI. On day 3 postinfection, the DhxuI mutant-infected group exhibited a lower bacterial burden in lesions than those infected by wt PAO1 or PAO1::P tac -hxuI (Fig. 6B). Histological examinations of skin abscesses indicated intense inflammatory infiltration, local tissue necrosis, and thickening of the epidermis in both PAO1 and PAO1::P tac -hxuI infection groups, while infection by DhxuI resulted in very mild inflammations (Fig. 6C). On day 7, a large abscess with overlying crust/scab was formed on the dorsum skin of 75% (6/8) mice infected by PAO1::P tac -hxuI, but on only 25% (2/8) and 12.5% (1/8) of those infected by PAO1 and DhxuI, respectively ( Fig. 6D and E). Histological sections of the PAO1::P tac -hxuI-infected group showed thickened epidermis, collagen fiber necrosis, lysis of subcutaneous muscle fibers, and inflammation (Fig. 6C). In comparison, the PAO1 and DhxuI infection groups exhibited much lower bacterial loads inside abscesses and fewer scattered inflammatory cells (Fig. 6B and C). These results indicated that forced expression of the HxuI enables P. aeruginosa to better adapt to the host environment, promoting the establishment of long-term infection.

DISCUSSION
In this study, we found that ECFs factor HxuI is highly conserved in different P. aeruginosa strains and can be induced by several host-inflicted stresses, including iron deprivation, oxidative stress, and hypoxia, as well as NO. Physiological adaptation to varied environmental stresses, such as changes in oxygen levels encountered within diverse niches, is an important capability for pathogenic bacterial species (34). The viability of P. aeruginosa within robust anaerobic biofilms requires NO reductase to modulate or prevent the accumulation of toxic NO, a byproduct of anaerobic respiration (35). Our data indicate that the NO sensor DNR negatively regulates HxuI, which further activates denitrification to reduce NO into nitrogen gas (36), revealing a novel ECFsmediated nitrosative stress-response pathway in P. aeruginosa.
Overexpression of HxuI remarkably activated the transcription of genes associated with pyoverdine-dependent iron acquisition, denitrification, pyocyanin biosynthesis, and the production of pyocins involved in intraspecies competition. Fur2 is positively regulated by HxuI and plays a critical role in HxuI-mediated transcriptional regulation, and even in the auto-activation of HxuI. Most notably, forced expression of the hxuI gene promotes the establishment of long-term P. aeruginosa infection in vivo; therefore, HxuI functions as an important regulator that senses host stresses and enables P. aeruginosa to tune metabolic strategies for adaptation to the host environment and express virulence factors which promote persistent infection.

MATERIALS AND METHODS
Bacterial strains, plasmids, and growth conditions. The bacterial strains, plasmids, and primers used in this study are listed in Table S2. Gene deletion and complementation were performed as previously described (37,38). Bacterial cells were grown at 37°C in LB (Luria-Bertani) broth or in M9 medium with 0.1% (wt/vol) glucose (39). The following concentrations of antibiotics were used: for P. aeruginosa, gentamicin at 30 mg/mL in LB, tetracycline at 50 mg/mL in LB, and carbenicillin at 150 mg/mL in LB; for Escherichia coli, tetracycline at 10 mg/mL, gentamicin at 10 mg/mL, kanamycin at 50 mg/mL, and ampicillin at 100 mg/mL. For iron-limitation condition, strains were grown in ABTG medium [15.1 mM (NH4) 2 SO 4 , 33.7 mM Na 2 HPO 4 , 22.0 mM KH 2 PO 4 , 0.05 mM NaCl, 1 mM MgCl 2 , 100 mM CaCl 2 , 0.5% (wt/vol) glucose, and 1 to 10 mM FeCl 3 ] (40). Anaerobic conditions were established by an anaerobic workstation (Don Whitley Scientific) with an oxygen content of 0.07%, and bacteria were statically cultured in various media supplemented with 50 mM NaNO 3 . All experiments were done in the biosafety level 2 laboratory at Nankai University.
Gene conservation analysis. The population structure of P. aeruginosa can be divided into five groups (41). The complete genomes of 723 P. aeruginosa strains that covered all five groups were analyzed in this study. The nucleotide sequences of the huxIRA of PAO1 were used as reference. We aligned each genome sequence of the 723 strains against the reference using BLASTn (14) with the criteria set as E value , 1e-5 and length coverage of the gene . 85% to find the homologous sequences. Finally, the identities between each strain and reference were illustrated using the R package (http://www.r -project.org/). Ethics statement. All animal studies complied with National and Nankai University guidelines regarding the use of animals in research. All animal experiment protocols were approved by the Institutional Animal Care and Use Committee of the College of Life Sciences of Nankai University with the permit number NK-04-2012.
Murine lung infection. The infection of mice was performed as previously described (42). Briefly, overnight bacterial culture was diluted 1:100 in fresh LB and grown at 37°C until the OD 600 reached 1.0. Bacterial cells were collected by centrifugation and washed once with phosphate-buffered saline (PBS). The bacterial cell concentration was adjusted to 5 Â 10 8 CFU/mL in PBS. Each female BALB/c mouse (Vital River, Beijing, China), at the age of 6 to 8 weeks, was anesthetized with an intraperitoneal injection of 7.5% chloral hydrate and inoculated with 20 mL of the bacterial suspension, resulting in 1 Â 10 7 CFU per mouse. Bronchi alveolar lavage fluid (BALF) was collected as previously described (43). At 6 h postinfection, mice were euthanized via CO 2 inhalation. One mL PBS containing 0.05 mM EDTA was injected into the lungs via the trachea by a vein detained needle (BD, Angiocath). After 1 min of detaining, BALF was collected.
Total RNA isolation and quantitative real-time PCR. Total bacterial RNA was isolated using an RNAprep Pure Cell/Bacteria Kit (Tiangen Biotec, Beijing, China). cDNAs were synthesized with reverse transcriptase and random primers (Takara Bio, Dalian, China). Real-time (RT) PCR was performed using SYBR II Green Supermix (Bio-Rad, Beijing, China). Specific Primers (Table S3) were used for quantitative RT-PCR. The peptidyl-prolyl cis-trans isomerase D gene ppiD was used as an internal control.
Transcriptome sequencing and data analysis. Both PAO1/pMMB and PAO1/pMMB-hxuI cultures (OD 600 =0.6) were grown in LB with 1 mM IPTG for 2 h. Total RNA was isolated using an RNAprep Pure Cell/Bacteria Kit (Tiangen Biotec, Beijing, China). Three replicates were prepared for each strain. Sequencing and analysis were performed as previously described (44).
Promoter-lacZ reporter assay. The promoter region (500 bp upstream from the start codon) of each gene was cloned into pDN19lacX to construct the promoter-lacZ reporter construct. The reporter constructs, as well as the pMMB-hxuI or the empty plasmid pMMB, were introduced into PAO1 by electroporation, and the transformants were selected on an L agar plate containing Tc and Cb. After inducing the expression of HxuI with 1 mM IPTG for 2 h, bacterial cells were collected by centrifugation and resuspended in 500 mL of Zbuffer (16 g/L Na 2 HPO 4 Á7 H 2 O, 4.8 g/L NaH 2 PO 4 , 0.746 g/L KCl, 0.246 g/L MgSO 4 Á7 H 2 O, 3.5 mL/L b-mercaptoethanol [pH = 7]). To permeabilize the cells, 10 mL of 0.1% SDS and 10 mL of chloroform were added and vortex for 10 s. After this, 100 mL of 4 mg/mL ONPG (o-nitrophenyl-b-D-galactopyranoside) was added to the cells. The samples were incubated at 37°C until the yellow color became apparent, and 500 mL of Na 2 CO 3 (0.5 M) was added to stop the reaction. Sample absorbance was read at 420 nm, and b-galactosidase activity was calculated as Miller units = 2,000 Â OD 420 /OD 600 /incubation time (min). Each assay was repeated three times.
Pyocin toxicity assays. Zones of clearance were observed for the P. aeruginosa PAK strain using the supernatants of wt PAO1, hxuI mutant and hxuI-overexpressing strains. A 0.05 mg/mL volume of ciprofloxacin was used to induce the production of pyocins in the PAO1-derived strains. PAK was used as an indicator strain, diluted to OD 600 = 0.6, and plated on LB agar. Finally, 200 mL of supernatants of the test strains were added to sterile Oxford cups placed on the PAK plate and cultured overnight at 37°C.
Scanning electron microscopy (SEM). Bacterial cultures (OD 600 = 1.0) were co-incubated with 0.1% gelatin-coated glass slides at 37°C for 4 h. The unattached bacterial cells were discarded. The glass slides with sessile bacteria were washed once with PBS and fixed with 4% paraformaldehyde. The bacterial cells were dehydrated with a gradient (30%, 50%, 70%, 90%, 100%) of alcohol, air dried, and imaged under an electron microscope.
Mouse cutaneous abscess model. The infection of mice was performed as previously described (31). Briefly, mice were clipped in the dorsal area by a shaver and depilatory cream. Fifty mL of either 5 Â 10 6 CFU bacterial suspension or saline were subcutaneously injected into the dorsum of each mouse. At 3 and 7 days postinfection, mice were euthanized with carbon dioxide, and then the skin abscesses were excised, homogenized in saline, and subjected to plating for CFU counting.
Statistical analysis. Statistical evaluations were performed using GraphPad Prism 7.0 (GraphPad Software Inc., La Jolla, CA). P values were calculated using one-way analysis of variance (ANOVA), a twotailed unpaired Student's t test. Data were considered significant when P values were below 0.05, as indicated.
Data availability. The transcriptome (RNA-Seq) data have been deposited in NCBI BioProject with the accession code PRJNA717102.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.3 MB.