Introduction

Rhizobium-legume symbiotic nitrogen (N2) fixation within nodules is a major contributor to soil fertility and the global N2 cycle (Kuypers et al. 2018). Nodules typically occur on roots, but a few legumes of the Papilionoideae and Mimosoideae families also form stem-located and aerial nodules (Boivin et al. 1997). Among these plants, Sesbania rostrata is an important legume green manure that can form both root and stem nodules induced by Azorhizobium caulinodans (Dreyfus and Dommergues 1981). In soil and during symbiosis, A. caulinodans must cope with varying stresses from the environment and from the plant defence against bacterial infections (Den Herder et al. 2006). This bacterium enters S. rostrata via root hairs or cracks (James et al. 2001). Upon first encountering the host plant, A. caulinodans is recognized as an invader, perhaps even as a pathogen, and triggers an oxidative burst in planta (D’Haeze et al. 2003). Notably, hydrogen peroxide (H2O2) is required for optimal symbiosis between A. caulinodans and S. rostrata (Capoen et al. 2010). From the crack entrance, the infection pocket formation induces local cell death due to the massive production of H2O2 (D’Haeze et al. 2003). Abundant H2O2 in infection threads of root hairs is also essential for the invasion of the deep tissue in S. rostrata (D’Haeze et al. 1998; Rubio et al. 2009).

Plant defences produce harmful reactive oxygen species (ROS), including H2O2, superoxide anion (O ·−2 ), and organic peroxides (OHPs), during symbiotic interactions, as reviewed in Apel and Hirt (2004) and Nanda et al. (2010). Beyond these interactions, legumes and rhizobia encounter ROS challenges from normal aerobic metabolic processes such as photosynthesis, respiration, and N2 fixation (Apel and Hirt 2004). ROS are toxic forms of oxygen (O2), often damage proteins, DNA, and cell membranes, and even interfere with symbiosis (Puppo et al. 2013). Indeed, numerous cellular defences against ROS stress are deployed by antioxidant systems, including catalase (Kat), peroxiredoxin (Prx), glutathione (Grx), thioredoxin (Trx) and superoxide dismutase (SOD), in bacteria (Imlay 2008). Among them, bacterioferritin comigratory protein (BCP) is one of the thiol-dependent peroxiredoxins (Mishra and Imlay 2012); BCP reduces a broad range of hydroperoxides with a high redox potential (Reeves et al. 2011) but with low catalytic efficiencies (Jeong et al. 2000).

BCP was first identified in Escherichia coli as a new member of the alkyl hydroperoxide reductase/thiol-specific antioxidant protein (AhpC/TSA) family that displays a general hydroperoxide peroxidase activity (Jeong et al. 2000). BCP reduces fatty acid hydroperoxide prior to H2O2 and t-butyl hydroperoxide (tBOOH) in E. coli (Jeong et al. 2000). BCP-deficient strains of Helicobacter pylori (Wang et al. 2005), Campylobacter jejuni (Atack et al. 2008), and Porphyromonas gingivalis (Johnson et al. 2011) grow slowly in aerobic culture and display an increased sensitivity to the cumene hydroperoxide (CuOOH) rather than to H2O2. Thermococcus kodakaraensis BCP has peroxidase activity towards all three substrates (Pham et al. 2015). Furthermore, the bcp mutant displays a growth-phase-dependent sensitivity to different oxidants in E. coli (Jeong et al. 2000) and Burkholderia cenocepacia (Clarke et al. 2010). However, there are limited reports about the role of BCP in rhizobia.

To investigate whether rhizobial BCP possesses similar functional properties to that of reported bacteria, we first knocked out the bcp gene (AZC_1350) in A. caulinodans ORS 571. Then, we tested the phenotype of the bcp mutant strain under H2O2, tBOOH and CuOOH stresses in vitro, as well as the symbiotic performance of the mutant strain in S. rostrata. We demonstrate that A. caulinodans BCP, which is involved in cellular protection against H2O2, is dependent on the growth phase and is also required for optimal symbiosis with the native host plant of A. caulinodans.

Materials and methods

Plant and bacterial culture condition

Escherichia coli strains were cultured in Luria–Bertani (LB) medium at 37 °C (Green and Sambrook 2012). A. caulinodans ORS571 (Lee et al. 2008) and its derivative strains were grown in tryptone-yeast extract (TY) medium at 28 °C (Beringer 1974). All solid media were supplemented with 1.2% agar. Indicated antibiotics were used for the selection at the final concentrations as following: kanamycin (Kan, 100 μg mL−1), ampicillin (Amp, 100 μg mL−1), spectinomycin (Spe, 20 μg mL−1) and gentamicin (Gm, 20 μg mL−1). Seeds of S. rostrata were surface sterilized as the description in (Fernández-López et al. 1998), and the seedlings were inoculated with A. caulinodans and grown in pots in the glasshouse (D’Haeze et al. 1998).

Plasmid and strain construction

Molecular techniques were performed following the standard protocols (Green and Sambrook 2012). Bacterial strains and the primer of the strain constructions from this work are listed in Tables 1 and 2. E. coli DH5α was used for general cloning, while E. coli SM10 λpir was used for cloning in conjugation. A. caulinodans was transformed by electroporation as described previously (Zhao et al. 2016).

Table 1 Strains and plasmids in this work
Full size table
Table 2 Primers in this work
Full size table

The in-frame deletion was used to construct the bcp-defective mutant (Metcalf et al. 1996). First, two flanking fragments of bcp, a 739 bp of the upstream fragment (the primer pair of Ac-bcp-1/2) and a 724 bp of the downstream fragment (the primer pair of Ac-bcp-1/2), were cloned into pEX18Gm which is a sacB (gene encoding the levansucrase) suicide vector, and yielded the pEX18Gm-bcp plasmid. The mutant (∆bcp) was selected for double-cross recombination and confirmed by PCR. For complementation plasmid for (∆bcp), the Ptac-bcp plasmid was constructed through PCR amplification of the coding sequence (beginning with the ATG start codon) with the primer pair of pSRK-bcp-1/2 and cloned into pSRKGm (Khan et al. 2008). The open reading frame of bcp was amplified using indicated primers in Table 2. The amplicon was released by NdeI and HindIII and then cloned into pSRKGm plasmid. The correct orientation allowing the expression of bcp under the control of Ptac promoter was selected by PCR. The corresponding pSRK-bcp plasmid was introduced into the ∆bcp strain by the conjunction, yields the complementation strain bcpc.

Bacterial growth experiment in rich medium

Wild-type A. caulinodans and ∆bcp mutant strains used in this study were grown separately at 28 °C overnight (14–16 h) in 20 mL liquid TY medium. Bacterial cultures were kept in 100 mL conical flasks and shaken at 180 rpm in shaker incubator (SKY-211B, Sukun, China). The overnight cultures were harvested to prepare the bacterial suspension (OD600 = 0.01) in TY liquid medium, and re-incubated for the growth curve experiment at above conditions. Growth was monitored by the determination of OD600 values in an UV-spectrophotometry (Philes Ltd, Nanjing, China) over time. Each experiment was repeated three times.

Peroxide disc diffusion assay

Three peroxides, H2O2, tBOOH and CuOOH were used to examine their inhibitions on the bacterial growth of A. caulinodans, ∆bcp and bcpc via the disc diffusion assay (Wang et al. 2012). Overnight pre-cultures were prepared as described above and adjusted to the OD600 values of 1.0 (109 bacterial cells per mL). Then, those cells were mixed with TY top agar (0.6%) at 42 °C and poured onto the solid TY agar plates (55 mm in diameter). The sterilized Whatman AA disc (6 mm, Whatman International Ltd, Maidsotne, UK) was placed on the Petri plates for carrying 4 μL of 10 M H2O2, 5 M tBOOH, or 4.88 M CuOOH, then incubated at 28 °C for 24 h. The growth inhibition zones were measured and adjusted by subtracting the Whatman AA disc diameter. Each experiment was repeated three times.

Peroxide killing assay

This experiment was used to test the susceptibility of wild-type A. caulinodans and ∆bcp mutant strain to the mM levels of H2O2, tBOOH and CuOOH as reported previously in Sinorhizobium meliloti (Barloy-Hubler et al. 2004; Fontenelle et al. 2011). Overnight cultures of above strains were grown to the early-log phase (OD600 ≈ 0.2) or the stationary phase (OD600 ≈ 1.0) in TY liquid medium in described conditions. For peroxide treatments, H2O2, tBOOH and CuOOH were adjusted to the concentration of 8 mM, 1.5 mM, and 0.6 mM, respectively, for 2 h incubation in the shaker. Then, bacterial cultures were plated onto TY agar plate through a series of ten-time dilutions and incubated at 28 °C for 3 days to determine the colony formation unit (CFU). Survival rate (%) was assessed based on the comparison with the control that was treated with the TY medium. Each trial contained triplicates per strain per condition.

Catalase assay

Overnight cultures of the wild-type strain and Δbcp were inoculated at 1:100 into TY medium containing appropriate antibiotics and shaken at 28 °C until the early-log or the stationary phases. H2O2 was adjusted to 8 and 15 mM in bacterial cultures incubated at 28 °C for 1 h separately. Rinsed cells were collected and lysed by the sanitation. Equal amounts of lysates from the wild-type strain and Δbcp were subjected to the catalase activity assays using Fluorometric Catalase Activity Assay Kit (Enco Scientific) according to the manufacturer’s instructions.

Nodulation assay and nitrogenase activity

Overnight cultures of wild-type A. caulinodans and Δbcp were diluted OD600 value of 0.1. Bacterial suspensions were stem spread for the stem nodulation or soil drenched onto S. rostrata for the root nodulation. Stem and root nodules were harvested to measure the nitrogen-fixing activity at 5 weeks post-inoculation. To measure the nitrogenase activity, 15 nodules per plant were placed into a 20 mL vial sealed with a butyl rubber septum. Each vial was injected with 2 mL acetylene by syringe and incubated at 28 °C for 2 h. Then, 1 mL gas from the vial was measured the acetylene and ethylene concentrations (Berrabah et al. 2014).

Transmission electron microscopy (TEM)

Five-week-old stem nodules were fixed in 2.5% glutaraldehyde with 0.1 M sodium cacodylate buffer (pH 7.4) at 4 °C for 1 h, then in 2% osmium tetroxide with above buffer for 1 h at the room temperature and embedded with Epon812. Ultrathin sections (80 nm) of stem nodule were cut with a microtome (LKB 8800 Ultrotome III, Swiss) and stained with uranyl acetate for 20 min at 60 °C followed by lead citrate for 10 min at room temperature. The TEM of specimens was analysed in a Hitachi H-7650 transmission electron microscope (D’Haeze et al. 1998).

Results

The putative BCP is essential for resistance to H2O2 but not to tBOOH and CuOOH

In the A. caulinodans ORS571 genome (Lee et al. 2008), AZC_1350 was identified as a putative gene encoding a Prx containing COG1225, i.e., the AhpC/TSA family protein BCP. To study the role of this putative bcp in the response to oxidative stress in A. caulinodans, we constructed an in-frame deletion mutant (Δbcp) based on the genetic background of A. caulinodans ORS571 (WT). As shown in Fig. 1a, this bcp-deficient strain grew similarly to its parental strain ORS571 when it was incubated in liquid TY-rich medium. It is likely that BCP does not play a role in bacterial growth/fitness under normal conditions. Based on the results of disc diffusion assays, the growth of Δbcp on agar plates was strongly inhibited by the addition of H2O2 but not CuOOH and tBOOH (Fig. 1b–d, t test). The inhibition diameter of the 10 M H2O2 treatment drastically increased from 32.8 ± 1.92 cm (WT) to 39.5 ± 1.73 cm (Δbcp, P value = 0.0005). For bcpc, the complemented strain of Δbcp, the inhibition diameter of H2O2 was similar to that of the WT (P value = 0.0646) but significantly lower than that of Δbcp (P value = 0.0125). However, Δbcp grew similarly to the WT on the plate in the presence of 4.88 M CuOOH (Fig. 1c) and 5 M tBOOH (Fig. 2d). The inhibition diameter of CuOOH on Δbcp (33.67 ± 0.73 mm) was not distinguishable from that 34.01 ± 0.01 cm inhibition diameter observed in the WT (P = 0.6611). Meanwhile, the inhibition diameter of Δbcp (54.00 ± 0.58 mm) also showed no significant differences from that of WT (52.75 ± 0.78 mm) on the tBOOH plate (P value = 0.2827). The disc diffusion assays suggested the importance of BCP to the resistance of A. caulinodans to H2O2.

Fig. 1
figure 1

Bacterial growth and disc diffusion assays. a Bacterial growth of Azorhizobium caulinodans ORS571 (WT) and the Δbcp mutant strains. The cultures were grown in TY medium at 28 °C. OD600 was measured at the time points indicated. In the disc diffusion assay, approximately 107 bacterial cells of WT, Δbcp and its complementary strain (bcpc) were mixed and spread on the top of the TY agar plates. The disc was saturated with 4 µl of 10 M H2O2 (a), 4.88 M CuOOH (b), or 5 M tBOOH (c) and placed in the middle of the plate. Data are the means ± SDs of three independent experiments. **P < 0.01; *P < 0.05; ns no significant difference (t test)

Full size image
Fig. 2
figure 2

Contributions of bcp to peroxide killing at early-log and stationary phases. Wild-type and Δbcp strains were grown in TY liquid medium at 28 °C. The early-log (e.log) and stationary (sta.) phases of growth were defined as 5 h (OD600 of 0.2) and 24 h (OD600 of 1.0), respectively, after inoculation. After incubation, bacterial cultures were treated with 8 mM H2O2 (a), 0.6 mM CuOOH (b), and 1.5 mM tBOOH (c) for 2 h. The cells were then resuspended in sterile water and placed on TY agar plates to determine cell viability. Data are the means ± SDs of three independent experiments. **P < 0.01; ns no significant difference (t test)

Full size image

BCP confers resistance to H2O2 during the early growth phase

To further study the function of BCP in resistance to inorganic and organic hydroperoxides during the growth dynamics of A. caulinodans, we investigated the survival rates of the WT and Δbcp strains in oxidative killing assays using 8 mM H2O2, 1.5 mM tBOOH and 0.6 mM CuOOH (Fig. 2, t test). Our data indicated that Δbcp was more sensitive to H2O2 killing at the early-log stage of bacterial growth in the TY liquid culture (Fig. 2a). The survival rate of Δbcp (0.5027% ± 0.17%) was significantly lower than that of the WT strain (15.17% ± 3.87%) in this stage (P value = 0.0021). In the stationary phase, there were no differences in the survival of the WT and Δbcp strains following treatment with H2O2 (Fig. 2a, P value = 0.5264). As shown in Fig. 2b, c, the mutation of bcp did not alter the survival of A. caulinodans ORS571 relative to the organic peroxide killing at the log phase/stationary phase of growth. Indeed, both the WT and Δbcp strains were equally sensitive to tBOOH and CuOOH. Following treatment with tBOOH there were no clear differences between the survival rates the WT and bcp strains at the log phase (P value = 0.7865) or the stationary stage (P value = 0.7492). For the CuOOH killing assay, the survival rate of Δbcp was analogous to that of the WT strain at the log phase (P value = 0.3464) and the stationary phase (P value = 0.2325).

Δbcp shows a reduction of catalase activity in the presence of H2O2

We found that BCP confers the resistance of A. caulinodans to H2O2 in both the disc diffusion and oxidative killing assays. To further study the role of BCP in H2O2 tolerance, we investigated the catalase activity of the WT and Δbcp strains exposed to 8 mM H2O2 during the growth of A. caulinodans (Fig. 3a, b, t test). No differences were observed between the catalase activities of the WT and Δbcp strains at the log phase/stationary phase of growth (PWT vs Δbcp > 0.05). However, the catalase activity was significantly reduced for WT strain (PWT (+) vs (−) < 0.05) or Δbcp (PΔbcp (+) vs (−) < 0.05) by the addition of H2O2 at both growth stages. Compared with the WT A. caulinodans, the Δbcp strain displayed reduced catalase activity at both the early-log phase (PWT vs Δbcp = 0.0123) and stationary phase (PWT vs Δbcp = 0.0064). In E. coli, the peroxidase activity of BCP is increased with the concentration of H2O2 (Jeong et al. 2000). Thus, we further tested the catalase activity of Δbcp at relatively high levels of H2O2 (15 mM) at the stationary phase because the growth of Δbcp was strongly inhibited at the early-log phase under the stress of 8 mM H2O2 (Fig. 2a) and at both phases under the stress of 20 mM (data not shown). The bcp-defective strain displayed similar catalase activity as its parent strain at both the early-log phase (PWT vs Δbcp = 0.3119) and the stationary phase (PWT vs Δbcp = 0.2786). In this experiment, the catalase activity of Δbcp was similar to that of the WT strain (P value > 0.05) when the bacteria entered the stationary phase. These results indicated that BCP plays an important role in catalase activity at low H2O2 concentrations.

Fig. 3
figure 3

Catalase activity of Azorhizobium caulinodans strains. Wild-type and ∆bcp were grown in TY medium at 28 °C. The early-log (e.log) and stationary (sta.) growth phases were defined as 5 h (OD600 of 0.2) and 24 h (OD600 of 1.0) after inoculation, respectively. When indicated, 8 mM H2O2 (e.log, a), 8 mM H2O2 (sta., b), and 15 mM H2O2 (sta., c) were added to the cell cultures and incubated for 1 h. Proteins were extracted from these two strains and determined as described in materials and methods. Data are the means ± SDs of three independent experiments. **P < 0.01; *P < 0.05; ns no significant difference (t test)

Full size image

BCP is required for nodulation and nitrogenase activity

The sensitivity of the A. caulinodans mutant to H2O2 is potentially relevant to symbiosis since S. rostrata cells respond to rhizobial infection with an enhanced production of ROS (D’Haeze et al. 2003; Zhao et al. 2016). To test the effect of bcp mutation on nodulation and nitrogen fixation, we inoculated ~ 108 WT and Δbcp cells onto a S. rostrata stem. At 5 weeks post inoculation, fewer nodules were clearly observed in the plants infected with Δbcp (Fig. 4a). The nodulation efficiency of Δbcp was significantly reduced in comparison to that of plants inoculated with the parental strain ORS571 (P value = 0.0068, Fig. 4b). Moreover, stem nodules induced by the bcp mutant displayed substantially lower acetylene reduction activity than the WT-induced nodules (P value = 0.0294, Fig. 4c). However, there were differences between the nitrogenase activity of the 5-week-old root nodules formed by the Δbcp and WT strains, even though the root nodule number was significantly reduced in the former (Fig. S1). To further investigate whether the lower nitrogenase activity of the nodules was impaired by the development of Δbcp bacteroids, we performed a TEM test on the stem nodules induced by both the Δbcp and WT strains. As shown in Fig. 4d, Δbcp formed enlarged, rod-shaped, and round bacteroids within the cells of stem nodules. Our data offer that BCP is important for nodulation and fixation but has no impact on the development of bacteroids within stem nodules.

Fig. 4
figure 4

Effect of BCP on stem nodule formation and nitrogenase activity. Stem nodule formation (a). Stationary cultures of wild-type (WT) and Δbcp strains were inoculated on the stem surface of S. rostrata. Nodules were photographed 5 weeks post inoculation. The number of stem nodules on S. rostrata formed by WT and Δbcp were assessed (b). Nitrogen fixation activities of stem nodules (c). The stem nodules formed by each strain were harvested at 5 weeks post inoculation for nitrogenase activity measurements. Data are the means ± SDs of three independent experiments. **P < 0.01; ns no significant difference (t test). d TEM of the stem nodule structure. Transmission electron micrographs of stem nodules. The micrographs were taken from infected plant stem cells of 5-week-old nodules induced by WT and Δbcp bacterial cells. Scale bars = 1 µm

Full size image

Discussion

In this work, we performed a functional analysis of the TSA/AhpC family protein BCP in A. caulinodans. Like other bacteria, A. caulinodans is often confronted with broad oxidative stresses generated in the environment or during host plant infection dynamics. Here, our data offer a preliminary report that BCP is essential for A. caulinodans defence against H2O2 rather than against two OHPs (i.e., CuOOH and tBOOH) under free-living conditions, and BCP is also required for nodulation and N2 fixation during symbiotic associations with S. rostrata.

BCP contributed to the bacterial defence against high concentrations of H2O2 when A. caulinodans was grown on agar plates and against low H2O2 levels in cell culture (Fig. 3a). The roles of the bcp-associated phenotype seem to be heterogeneous among bacterial species. While bcp-deleted A. caulinodans, E. coli (Jeong et al. 2000) and B. cenocepacia (Clarke et al. 2010) are highly sensitive to H2O2, bcp-deleted H. pylori (Wang et al. 2005), C. jejuni (Atack et al. 2008), and P. gingivalis (Johnson et al. 2011) are not. This phenotypic heterogeneity of BCP is probably attributable to variations in the composition and the abundance of other antioxidant enzymes within the cells of different species. In H. pylori, BCP is considerably less abundant than another thiol-based Prx AhpCD. Compared to the wild-type strain of H. pylori, the AhpCD-defective mutant displays a strong oxidative stress resistance phenotype (Wang et al. 2005). Indeed, we also found functional overlap between BCP and AhpCD due to the antioxidant activities of AhpCD in A. caulinodans (submitted data). Furthermore, a high concentration of H2O2 can induce a more efficient Kat system in E. coli (Seaver and Imlay 2001), S. meliloti (Barloy-Hubler et al. 2004) and A. caulinodans (Zhao et al. 2016). This may explain why BCP-deficient A. caulinodans grows similar to its parent strain during bacterial growth under high H2O2 stress.

BCP appears to be unimportant for bacterial defence against organic ROS in A. caulinodans. The bcp-deleted mutant showed comparable growth to the wild-type A. caulinodans strain both on the agar plate and in liquid medium in the presence of CuOOH and tBOOH. This phenotype is inconsistent with previous reports that E. coli BCP can reduce a broad range of hydroperoxides, including CuOOH and tBOOH (Jeong et al. 2000; Reeves et al. 2011). However, in addition to BCP, OHPs are reduced by other proteins, such as AhpCD in E. coli (Seaver and Imlay 2001), Xanthomonas campestris (Mongkolsuk et al. 2000) and Bacillus subtilis (Bsat et al. 1996) and Ohr (organic hydroperoxide resistance protein) in B. subtilis (Bsat et al. 1996), X. campestris (Mongkolsuk et al. 1998), Agrobacterium tumefaciens (Chuchue et al. 2006), and Pseudomonas aeruginosa (Ochsner UA 2001), as well as the N2-fixing bacterium S. meliloti (Fontenelle et al. 2011). In our recent work, we also showed that Ohr, rather than AphCD (alkyl hydroperoxide reductase), is essential to the tolerance of stress from CuOOH and tBOOH in A. caulinodans (unpublished data). We argued that it would be ideal for A. caulinodans to produce different enzymes to cope with many types of ROS through an optimized metabolism and ROS signalling network, such as other bacteria or rhizobia (Mishra and Imlay 2012; Puppo et al. 2013).

In A. caulinodans, BCP is required for nodulation on stems and roots as well as N2 fixation in stem nodules. In the early stage of infection, this bacterium induces the high production of H2O2 by defence reactions in S. rostrate cells (D’Haeze et al. 2003). It is possible that BCP reduces H2O2 and facilitates rhizobial infection in planta. The function of H2O2 in host plants during symbiosis is not only restricted to plant defence against bacteria but is also important for symbiotic processes (Nanda et al. 2010). In S. rostrate, the formation of infection pockets is induced by the A. caulinodans nod factors (D’Haeze et al. 1998) and involves the induction of local cell death associated with a massive production of H2O2 (D’Haeze et al. 2003). Thus, the balance between production and reduction of H2O2 is critical for deeper invasion of S. rostrate tissues. This characteristic explains why the H2O2-associated antioxidant proteins, such as BCP in this work, AphCD (submitted data) and KatG (Zhao et al. 2016) of A. caulinodans are required for the nodulation of S. rostrate. Furthermore, BCP significantly affects the nitrogenase activity of stem nodules (Fig. 4c) rather than that of root nodules (Fig. S1c) in S. rostrate. BCP is more effective at detoxifying ROS in stem nodules, which may have a lower oxygen microenvironment than root nodules (James et al. 1998).

Azorhizobium caulinodans encounters many ROS generated from environmental stresses under free-living conditions and during symbiosis. In this work, we preliminarily documented that BCP plays an important role in A. caulinodans in the defence reaction to oxidative stresses and in nodulation and N2 fixation in S. rostrate. A. caulinodans has different antioxidant pathways to deal with ROS and eventually establishes normative symbiosis with the host plant. Our work shed light on deciphering the molecular bases of BCP to ROS resistance during symbiotic interactions between rhizobia and plants.