The Pseudomonas aeruginosa DksA1 protein is involved in H2O2 tolerance and within-macrophages survival and can be replaced by DksA2

In Gram-negative pathogens, the stringent response regulator DksA controls the expression of hundreds of genes, including virulence-related genes. Interestingly, Pseudomonas aeruginosa has two functional DksA paralogs: DksA1 is constitutively expressed and has a zinc-finger motif, while DksA2 is expressed only under zinc starvation conditions and does not contain zinc. DksA1 stimulates the production of virulence factors in vitro and is required for full pathogenicity in vivo. DksA2 can replace these DksA1 functions. Here, the role of dksA paralogs in P. aeruginosa tolerance to H2O2-induced oxidative stress has been investigated. The P. aeruginosa dksA1 dksA2 mutant showed impaired H2O2 tolerance in planktonic and biofilm-growing cultures and increased susceptibility to macrophages-mediated killing compared to the wild type. Complementation with either dksA1 or dksA2 genes restored the wild type phenotypes. The DksA-dependent tolerance to oxidative stress involves, at least in part, the positive transcriptional control of both katA and katE catalase-encoding genes. These data support the hypothesis that DksA1 and DksA2 are eco-paralogs with indistinguishable function but optimal activity under different environmental conditions, and highlight their mutual contribution to P. aeruginosa virulence.


Results
DksA1 contributes to hydrogen peroxide tolerance, and DksA2 can replace DksA1 function. In P. aeruginosa, the dksA1 gene is constitutively expressed while the dksA2 gene is strongly repressed by Zur in the presence of available zinc 24,28 . To achieve comparable expression levels of the two dksA paralogs in a rich laboratory medium, such as LB, we previously generated a set of four P. aeruginosa PAO1 recombinant strains: the wild type strain (PAO1) and the dksA1 dksA2 double mutant (∆dksA1-2), both carrying the pME6032 empty vector, and the ∆dksA1-2 mutant carrying either pDksA1 or pDksA2. The latter are pME6032derivative plasmids for isopropyl β-D-1-thiogalactopyranoside (IPTG)-inducible expression of dksA1 and dksA2, respectively (Supplementary Table S1). As shown previously, the deletion of dksA paralogs or the presence of pME6032 derivatives or IPTG addition did not affect growth in LB. In addition, IPTG induction drove similar levels of dksA paralogs expression in PAO1 ∆dksA1-2 carrying pDksA1 or pDksA2 24 . Here, the same experimental framework has been used to investigate the role of DksA1 and DksA2 in P. aeruginosa response to oxidative stress caused by H 2 O 2 .
The minimum inhibitory concentration (MIC) of H 2 O 2 determined for planktonic P. aeruginosa cells in LB supplemented with 0.1 mM IPTG was 1 mM for all the tested strains. Subsequently, the P. aeruginosa tolerance to H 2 O 2 , i.e., the ability to survive transient exposure to 50X MIC H 2 O 2 concentrations, was determined. Results indicate that the percentage of bacteria survival after 30-min treatment with 50 mM H 2 O 2 was fivefold higher in PAO1(pME6032) than in the ∆dksA1-2(pME6032) mutant (Fig. 1). In addition, complementation with either dksA1 or dksA2 genes in ΔdksA1-2(pDksA1) and ΔdksA1-2(pDksA2) restored H 2 O 2 tolerance, with no significant difference between the two dksA paralogs.
The role of dksA1 and dksA2 in P. aeruginosa biofilm tolerance to H 2 O 2 was also investigated. To this purpose, the MIC of H 2 O 2 was determined in M9-glu-CAA, a chemically-defined medium suited for P. aeruginosa biofilms studies 24 . In M9-glu-CAA, the MIC of H 2 O 2 was 0.1 mM for all the tested strains, thus tenfold lower than that determined in LB. Hence, pre-formed P. aeruginosa biofilms were challenged with supra-MIC concentrations of H 2 O 2 , ranging from 50 to 200X MIC, and compared to the respective untreated controls.
In the absence of H 2 O 2 treatment, higher levels of biofilm biomass were detected by crystal violet (CV) staining in the ∆dksA1-2(pME6032) mutant compared to PAO1(pME6032) and the ΔdksA1-2(pDksA1) and ΔdksA1-2(pDksA2) complemented strains (Fig. 2a), in accordance with our previous study 24 . However, the fluorescein diacetate (FDA) assay showed that the metabolic activity of PAO1(pME6032) and ∆dksA1-2(pME6032) biofilms  (Fig. 2b), suggesting that in the ∆dksA1-2(pME6032) mutant there could be an increased production of extracellular biofilm matrix components compared to PAO1(pME6032). After challenging with H 2 O 2 , a dose-response biofilm disrupting effect was detected for all the tested strains by using both CV and FDA detection methods, though ∆dksA1-2(pME6032) biofilms were significantly more reduced by the treatment compared to the wild type and complemented strains (Fig. 2). As observed in liquid cultures ( Fig. 1), complementation with either dksA1 or dksA2 genes in the ΔdksA1-2 mutant restored similar levels of H 2 O 2 tolerance. Overall, since dksA2 expression is shut-off in PAO1 grown in the zinc-proficient media, such as LB and M9-glu-CAA 24 , the decreased H 2 O 2 tolerance of the ∆dksA1-2 mutant, relative to the PAO1 parent strain, should in principle be attributable to the lack of dksA1 expression. Hence, the above results suggest that DksA1 is required for H 2 O 2 tolerance of both planktonic and biofilm cultures of P. aeruginosa, and that dksA2 can replace dksA1 function when artificially expressed in zinc-containing media.
DksA1 contributes to macrophages-mediated killing tolerance, and DksA2 can replace DksA1 function. As mentioned in the introduction, the oxidative burst is the first microbicidal mechanism activated by macrophages to kill engulfed bacteria 5,6 . Thus, we analysed the role of DksA1 and DksA2 in counteracting oxidative stress within macrophages. First, RAW macrophages were infected for 15 and 30 min with PAO1 or the ∆dksA1-2 mutant, both carrying the pUCP30T-GFPmut3 plasmid for constitutive expression of the green fluorescent protein (GFP). Flow cytofluorimetry analysis showed that the percentage of fluorescent macrophages was similar in both infections, revealing that the ∆dksA1-2 double mutation has no impact on P. aeruginosa internalization by macrophages ( Supplementary Fig. S1).
Hence, the susceptibility to macrophages-mediated killing was compared in P. aeruginosa PAO1 and its isogenic recombinant strains. After 30 min from the infection, the number of live bacteria recovered from RAW macrophages was higher for PAO1(pME6032) than for the ∆dksA1-2(pME6032) mutant (Fig. 3). Taking into consideration that these strains are similarly phagocytosed, this result indicates that the ∆dksA1-2 mutant is killed more efficiently than the wild type strain. The intra-macrophage survival defect of the ∆dksA1-2 double mutant was complemented by pDksA1-or pDksA2-driven expression of either dksA1 or dksA2, respectively (Fig. 3), providing evidence that DksA2 can replace DksA1 function even inside macrophages. Overall, these results indicate that both DksA paralogs can protect P. aeruginosa against NADPH-oxidase-dependent macrophages oxidative burst.

DksA1 modulates endogenous ROS content by controlling catalases expression, and DksA2
can replace DksA1 function. The results presented above show that both P. aeruginosa dksA paralogs restore wild type levels of tolerance to H 2 O 2 exposure or macrophage oxidative burst. However, the DksA paralogs are also involved in the fine modulation of several central metabolic pathways, and an unbalanced metabolism could also affect endogenous ROS levels 29 . In agreement with this hypothesis, endogenous ROS levels were about 36% lower in PAO1(pME6032) than in ∆dksA1-2(pME6032), both grown in LB (Fig. 4a). Hence, DksA1 contributes to the homeostasis of the intracellular ROS levels naturally produced by cellular metabolism during aerobic growth, even in the absence of exogenous oxidative stress and nutrient starvation. An enzymatic assay revealed that ∆dksA1-2(pME6032) disclosed a 44% reduction of catalase activity compared to PAO1(pME6032) (Fig. 4b), suggesting that the increased ROS levels measured in the ∆dksA1-2 mutant could be related, at least in part, to a decreased expression of the H 2 O 2 detoxifying enzymes. Finally, the expression of either DksA1 or DksA2 in the ΔdksA1-2(pDksA1) and ΔdksA1-2(pDksA2) fully restored wild type catalase activity and ROS levels, providing evidence that also in this case DksA2 can replace DksA1.
To study the impact of the dksA gene products on the expression of different catalase-encoding genes, relative mRNA levels of katA, katB, katE and katN were investigated by means of Real Time PCR analyses in P. aeruginosa As previously reported 10,15 , the mRNA levels of both katA and katB increased in PAO1 exposed to H 2 O 2 ( Fig. 5). Interestingly, the dksA1 dksA2 double deletion had opposite effect on the H 2 O 2 -dependent expression of these genes. Indeed, katA and katB mRNA levels decreased and increased in ∆dksA1-2(pME6032) relative to PAO1(pME6032) in the presence of H 2 O 2 , respectively ( Fig. 5a,b).
Since DksA2 is expressed at low-basal levels in PAO1(pME6032) grown in LB, the above results show that at least DksA1 is required for the H 2 O 2 -dependent transcriptional upregulation and downregulation of katA and katB, respectively. In addition, DksA1 is also required for full expression of katE, irrespective of H 2 O 2 .
Finally, the IPTG-driven expression of either DksA1 or DksA2 in the ΔdksA1-2(pDksA1) and ΔdksA1-2(pDksA2) strains restored wild type mRNA levels for all the tested genes (Fig. 5), demonstrating that DksA2 can replace the function of DksA1 in promoting katA and katE expression, and in down-regulating katB expression.

Discussion
The stringent response controls bacterial adaption to nutrient starvation and other stressful conditions. In Gramnegative bacteria, (p)ppGpp binds the ß' subunit of RNAP at the interface with the ω subunit and in a second site, sandwiched between the ß′ subunit and the DksA protein, which is strictly required for this interaction 30,31 . Hence, both (p)ppGpp and DksA are required for full stringent response, even if the two factors can also work independently 20,[32][33][34][35] . While previous studies showed that (p)ppGpp positively affects P. aeruginosa tolerance to H 2 O 2 36,37 , the contribution of DksA1 and DksA2 to this process remained unknown so far. In the culture media used in this and our previous study 24 , dksA2 expression is strongly repressed by zinc in the P. aeruginosa PAO1 wild type strain. Hence, the comparison of the wild type and dksA1 dksA2 double mutant phenotypes only reveals the effect of DksA1 depletion. However, complementation of the double mutant with either DksA1 or DksA2 expressed by the corresponding gene under the control of an IPTG-inducible promoter allows comparing the activity of the two DksA paralogs. Overall, the dksA1 and dksA2 expression levels obtained by this approach are adequate to restore wild type levels of all the tested phenotypes.  exposure, both in planktonic and in biofilm growing cultures. In line with these results, the deletion of both dksA1 and dksA2 undermines P. aeruginosa survival within macrophages. The dksA1 dksA2 double deletion leads to an increase in P. aeruginosa biofilm biomass, as already shown in our previous study by using different biofilm models (i.e., CV staining, pellicle formation and Congo-Red binding assays) 24 . In the same study, it was shown that c-di-GMP levels were lower in the ΔdksA1-2 mutant compared to the wild type, indicating that DksA1 negatively controls biofilm formation through a c-di-GMPindependent pathway. Furthermore, the RNA-Seq analysis showed that some of the known P. aeruginosa biofilmrelated genes were up-regulated and others were down-regulated in the ΔdksA1-2 mutant compared to the wild type strain 24 . This study showed that the biofilm cells of the ΔdksA1-2 mutant are less viable after H 2 O 2 exposure www.nature.com/scientificreports/ relative to wild type biofilm cells, despite the increased biofilm biomass. This suggests a defect in the ΔdksA1-2 biofilm matrix composition and development. In the future, it could be interesting to further investigate this issue.
In the absence of exogenous H 2 O 2 , the ΔdksA1-2 mutant showed an increase in the endogenous ROS content and a decrease in the overall catalase activity, compared to wild type PAO1. This could be due to the reduced expression of katE in the ΔdksA1-2 strain, since the expression of the other P. aeruginosa catalase genes was not affected by dksA1 dksA2 double deletion in the absence of oxidative stress. However, since DksA1 controls more than one thousand genes, including primary metabolism pathways 23,24 , the ROS increase measured in the ΔdksA1-2 strain could also be due to a generalized metabolic effect. Moreover, in the presence of exogenous H 2 O 2 , katA and katE genes were downregulated while katB was up-regulated in the double mutant compared to the wild type PAO1.
Overall, it is plausible that the defective expression of katA and katE genes, combined with an altered metabolism leading to increased levels of endogenous ROS, could overcome the effects of the increased katB expression, ultimately reducing the tolerance of the ΔdksA1-2 mutant to exogenous H 2 O 2 and macrophages-mediated killing, relative to wild type PAO1. In this context, it is to note that the role of the KatE catalase in P. aeruginosa H 2 O 2 tolerance has not been clarified yet. Indeed, KatE seems not to be involved in H 2 O 2 detoxification in the PA14 strain 11 , while a protective role of KatE against oxidative stress only at elevated temperatures was described in the PAO6049 strain 38 .
Interestingly, a P. aeruginosa mutant lacking the relA and spoT genes, required for (p)ppGpp synthesis, showed several phenotypes similar to those caused by the dksA1 dksA2 double deletion, i.e., increased susceptibility to H 2 O 2 , impaired overall catalase activity and high levels of endogenous ROS with respect to the parental strain 36,37,39 . However, different from what we observed for DksA1, (p)ppGpp positively controls the expression of both katA and katB genes. Interestingly, this (p)ppGpp-dependent regulation is independent of exogenous H 2 O 2 for katA and H 2 O 2 -dependent for katB 36,37 .
By combining our data with previous findings, it can be argued that both (p)ppGpp and DksA1 play a major role in P. aeruginosa tolerance to endogenous and exogenous oxidative stress, even if they differently regulate the expression of H 2 O 2 -scavenger enzymes. Further studies should be carried out to unravel the regulatory mechanisms underlying the different effects of DksA1 and (p)ppGpp on katA, katB and katE genes expression, as well as the impact of KatE-mediated H 2 O 2 detoxification in P. aeruginosa PAO1.
To our knowledge, the importance of DksA in oxidative stress tolerance has so far been documented only for S. enterica and H. ducreyi. Each one of these pathogens expresses only one DksA protein, containing a zinc finger domain, as it is for DksA1 16,18 . Hence the possibility that DksA2 could replace DksA1 deserves special attention, also considering that a previous study showed that a S. enterica ∆dksA mutant expressing the P. aeruginosa dksA2 gene had an impaired tolerance to H 2 O 2 in vitro, reduced survival in macrophages, and reduced virulence in mice relative to the S. enterica ∆dksA mutant expressing the P. aeruginosa dksA1 gene. The different behaviour of DksA1 and DksA2 in the heterologous host Salmonella was supported by experiments carried out with the purified proteins, showing that DksA1 was more resistant to oxidation than DksA2, likely due to the presence of zinc in the globular domain 19 . Different from what observed in S. enterica, we showed by multiple experimental strategies that DksA2 replaces all DksA1 functions related to P. aeruginosa H 2 O 2 tolerance when expressed in its isogenic background, at least under our experimental conditions. Certainly, the possibility that dksA2 overexpression may mask a higher sensitivity of DksA2 to ROS than DksA1 cannot be rule out, and this issue should be further studied under conditions of zinc starvation, mimicking the environmental conditions where dksA2 is naturally expressed. In the wild type genetic background, under zinc-limiting conditions, both dksA1 and dksA2 genes are expressed from their natural promoters 28 and they may have a different impact on P. aeruginosa oxidative stress response. However, while dksA2 expression levels are related to the extent of zinc limitation, DksA1 activity might be impaired under conditions of extreme zinc deficiency, due to incorrect folding of the Zn-finger domain 28 .
Zinc depletion is recognised as an innate immunity mechanism faced by pathogens during the infection. Interestingly, dksA2 is expressed in the sputum of cystic fibrosis patients colonized by P. aeruginosa 40,41 , supporting a role for DksA2 at least in this kind of infection. While further studies should be carried out to better define the relative importance of DksA1 and DksA2 in the protection from H 2 O 2 exposure under zinc starvation conditions, this study provides the evidence that DksA2 has the intrinsic ability to replace the dksA1 gene product under oxidative stress conditions.
Overall, the results of this study are in line with our previous observation that DksA2 can replace DksA1 function in controlling the expression of almost all the DksA1-regulated genes and DksA1-dependent virulence phenotypes 24 strengthening the hypothesis that DksA1 and DksA2 are eco-paralogs as defined by Sanchez-Perez et al. 42 , i.e., paralogs with the same overall function, having optimal activity under different environmental conditions.

Materials and methods
Bacterial strains and growth conditions. Bacterial strains used in this study are listed in Supplementary Table S1. P. aeruginosa strains were routinely grown at 37 °C in shaking conditions in LB 43 , or LB supplemented with 1.5% (w/v) agar. When required, the media were supplemented with 0.1 mM IPTG or 100 µg/mL tetracycline (Tc). M9 medium [0.77% (w/v) Na 2 HPO 4 ·2H 2 O; 0.3% (w/v) KH 2 PO 4 ; 0.05% (w/v) NaCl; 0.025% (w/v) MgSO 4 ·7H 2 O; 0.002% (w/v) CaCl 2 ] 43,44 supplemented with 0.39% (w/v) glucose and 0.5% (w/v) casamino acids (CAA) as carbon sources (M9-glu-CAA) 24   For the phagocytosis assay, macrophages were infected as described above, using GFP expressing bacteria obtained by transformation of PAO1 and ∆dksA1-2 with the pUCP30T-GFPmut3 plasmid (Supplementary  Table S1) 50 . After either 15 or 30 min of infection, infected macrophages were washed twice with PBS, detached from the wells, recovered by centrifugation and finally resuspended in 300 µL PBS for flow cytometry (BD FAC-SCalibur, BD Biosciences, France). Phagocytosis was evaluated by the fraction of GFP positive cells in the bulk populations. Data were analysed using the CellQuest software and images processed using FlowJo.

Measurement of intracellular ROS levels.
The intracellular ROS levels were measured as previously described 51 , with minor modifications. P. aeruginosa strains were grown in LB with 0.1 mM IPTG to the late exponential phase (OD 600 of ≈ 2.5). Then, cells were collected by centrifugation, washed with PBS, and resuspended in PBS at an OD 600 of about 1.25. Bacteria were incubated with 10 µM 2′,7′-dichlorodihydrofluorescein diacetate (H 2 DCFDA; stock 2 mM, dissolved in dimethyl sulfoxide [DMSO]; Sigma Aldrich) for 20 min at 37 °C in the dark. Cells treated with an equal volume of DMSO were used as a negative control. Aliquots (200 µL) of each bacterial suspension were dispensed in a 96-well black microtiter plate. Fluorescence was measured at 485 nm excitation and 535 nm emission wavelengths with an automated luminometer-spectrometer plate reader (Tecan Spark 10 M) and normalized to the OD 600 of each sample.
Catalase activity assay. Catalase activity was measured as described previously 52 , with minor modifications. Bacteria were grown in LB with 0.1 mM IPTG to the late exponential phase (OD 600 of ≈ 2.5). Hence, cells were collected by centrifugation, resuspended in 50 mM potassium-phosphate buffer (PPB, pH 6.8), lysed by sonication and subsequently centrifugated to collect the supernatants. Total proteins from the soluble fraction were quantified by the Bradford assay 53 , with bovine serum albumin as the standard. Aliquots containing 10 µg of proteins were added to 50 mM PPB, and catalase activity was monitored by following the decomposition of 20 mM H 2 O 2 in 50 mM PPB at 240 nm (OD 240 ) for 1 min (readings every 10 s) by using UV-transparent dispos- www.nature.com/scientificreports/ able cuvettes. One unit of catalase activity corresponded to 1 μmol of H 2 O 2 hydrolysed per min by 1 mg of total proteins at 25 °C.
RNA extraction and expression profiling experiments. P. aeruginosa strains were grown at 37 °C with shaking at 200 rpm in 10 mL of LB supplemented with 0.1 mM IPTG until reaching an OD 600 of ≈ 2.5. Exponential phase cells were incubated with or without 1 mM H 2 O 2 for 20 min in shaking conditions and 1 mL of each culture was mixed with 2 mL of RNA Protect Bacteria Reagent (Qiagen) for RNA extraction. Total RNA was extracted as previously described 24,54 . Briefly, RNA isolation was performed using RNeasy Mini Kit (Qiagen), including the on-column DNase I digestion step, followed by 1 h treatment at 37 °C with TURBO DNase (0.2 U per μg of RNA; Ambion) and SUPERase-In (0.4 U per µg of RNA; Ambion), and subsequent purification with the RNeasy Column Purification Kit (Qiagen). After confirmation of the lack of contaminating chromosomal DNA by PCR with the primer pair FWPpqsL and RVPpqsL (Supplementary Table S2), 1 μg of total RNA was used to synthesize cDNA using the iScript Reverse Transcription Supermix for RT-qPCR kit (BioRad). Real Time PCR analyses were performed using the iTaq Universal SYBR Green Supermix (BioRad), the AriaMx Real-Time PCR system (Agilent Technologies; software version 1.0) and the target-specific primers obtained by means of the Primer-Blast designing tool (www. ncbi. nlm. nih. gov/ tools/ primer-blast; Supplementary Table S2). The thermal cycling protocol was denaturation for 2 min at 95 °C, followed by 40 cycles of amplification at 95 °C for 15 s and 60 °C for 45 s. Fluorescence was registered in the last 15 s of the 60 °C step. The relative fold change in gene expression was calculated by the 2 −ΔΔCt method using rpoD as the housekeeping gene. The average data and standard deviations (SD) were calculated from two independent experiments each performed on three technical replicates.

Statistical analysis.
Statistical analysis was performed with the software GraphPad Prism 6.01, using oneway analysis of variance (ANOVA) followed by Tukey-Kramer multiple comparison test. Statistical analysis of macrophages infections used two-way ANOVA. A p value of < 0.05 was considered statistically significant.

Data availability
All data generated or analysed during this study are included in this published article (and relative Supplementary Information file).