All strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were grown in LB media at 37°C. Azorhizobium caulinodans ORS571 and derivatives were cultured with tryptone-yeast extract (TY) medium or L3 minimal medium at 37°C. Carbon or nitrogen sources in L3 medium can be removed as needed, which are designated as L3-C+N or L3+C-N. To create the environment with different oxygen concentrations, first, the air in 20ml anaerobic tubes was replaced by the oxygen-free gas N2 for hypoxia environment or gas mixture of N₂/O₂ for hyperxia environment. Then 5ml overnight cell liquid culture was injected into the anerobic tubes and was shaken for 30s before putting into incubator with standing or shaking. The final concentrations of antibiotics used in this work are 100μg m⁻¹ of ampicillin, 50μg m⁻¹ of gentamicin and kanamycin, and 25μg m⁻¹ of nalidixic acid.

Two methods were used to construct deleted mutants. The Δ1026 and Δ3153 were constructed by pK18mobsacB, a suicide vector, and the Δ0573 and Δ1546 were constructed by pCM351, an allelic exchange vector. To construct plasmids for azc_1026 and azc_3153 gene deletion, upstream and downstream fragments of them were amplified with primers listed in Table 2. The 3′ end of upstream fragment and the 5′ end of downstream fragment were amplified with same restriction enzyme (EcoRI). After being digested with three restriction enzymes, the two amplicons were integrated into pK18mobsacB to construct a pK18mobsacB::up-down plasmid. The resulting plasmid was introduced into A. caulinodans ORS571 by triparental conjugation with the help of pRK2013, a helper plasmid. The azc_1026 and azc_3153 deletion mutants derived from wild-type strain were designated Δ1026 and Δ3153.

To construct azc_0573 and azc_1546 deletion mutants, the upstream and downstream fragments of them were amplified. These amplicons of upstream fragments were firstly digested and introduced into pCM351 to construct a pCM351::up plasmid. Then the downstream fragments of genes were digested and introduced into the pCM351::up plasmid. The resulting plasmids, pCM351::up-down, were integrated into wild-type strain through triparental conjugation for allelic exchange. Homologous recombinants lacking azc_0573 or azc_1546 were selected and termed as Δ0573 and Δ1546.

Azorhizobium caulinodans ORS571 cells were grown to stationary phase (24h), then the effect of standing time or the role of the hypoxia and electron transport chain inhibitors on localization on the cellular localization of CheZ were observed. To immobile cells, 1% agarose pads were constructed as previously described (Meier and Scharf, 2009). The Olympus BX53 system fluorescence microscope with Olympus DP73 digital camera was used to take photos with a 100×objective. The cellSensDimernsion 1.7 imaging software (Olympus Inc.) was used to capture clear images. The number of cells showed diffuse or polar localization of CheZ was recorded using manual count.

Sequences of intracellular chemoreceptors in A. caulinodans ORS571, Aer in E. coli, and AerC in A. barasilense were collected from Mist 2 database (Ulrich and Zhulin, 2010).¹ Domain analysis of proteins was performed using SMART program (Letunic and Bork, 2018). The protein sequences were aligned by MAFFT program (Madeira et al., 2019) and built a maximum likelihood tree using MEGA (Kumar et al., 2018).

All results in this study were subjected to ANOVA. Tukey’s test was used for multiple comparisons, and student t-test (p<0.01 and p<0.05) was used for significant differences between conditions. All tests were performed using SPSS version 20.0 software (IBM Corp., Armonk, New York).

We previously reported that CheZ in A. caulinodans can locate to cell poles and the localization pattern of CheZ is diverse, including monopolar, bipolar, and diffuse (Liu et al., 2020). Among them, diffuse localization of CheZ is the most common, observed in about 60% of cells (Liu et al., 2020). Unexpectedly, we observed that the ratio of cells with polar CheZ localization becomes very high after standing over 2days (Figure 1A). To get a clear conclusion, we quantified them and sought to determine the factors causing higher polar localization of CheZ.

First, we performed a time-course assay of CheZ localization. Cells with CheZ-GFP fusion were first grown to a late logarithmic phase with shaking and then left standing (without shaking) for 12–60h. The ratio of cells with polar localized CheZ-GFP in standing cultures was recorded at different times. After 24h, the number of cells with a polar localization pattern increased from 20 to 30% (Figures 1B,C). When cells were cultured without shaking for 60h, about 80% cells had polar localized CheZ-GFP (Figure 1C). Azorhizobium caulinodans with a GFP control vector showed a diffuse localization pattern under all conditions tested (Figure 1B).

The concentration of available oxygen becomes limited in cultures when cells are left standing for an extended period (Yuan et al., 2013). Thus, it was hypothesized that the increase in polar localization of CheZ without shaking may be induced by hypoxia. Environments with three kinds of oxygen concentrations were constructed using anerobic tubes to determine the localization pattern of CheZ. Under hypoxia conditions (Oxygen less than 4%) for 12h, around 100% cells showed polar localization (Figure 2). However, under hyperoxia conditions (Oxygen 33%), there were no obvious changes in CheZ localization compared with that under normoxia (Figure 2). These results indicate that oxygen limitation leads to the increase in polar localized CheZ.

Oxygen limitation could affect the energy status of the cells (Wessel et al., 2014). To determine whether the localization of CheZ is energy-related, two electron transport chain inhibitors, sodium azide and CCCP (Danylovych, 2016), were used. Sodium azide (0.2%) or CCCP (25μM) was added to cultures in the late logarithmic phase of growth. After incubating with shaking for 2h, there was no obvious difference between the control and inhibitor treatment in the localization pattern of CheZ (Figure 3). However, after 6h, the polar localization of CheZ increased to about 90% and reached 100% after 12h (Figure 3). These results indicate that the diffuse localization of CheZ relies on the energy status of the cell, and the cellular energy might be necessary to release CheZ from polar location.

Recently, we have mapped the region of CheZ for polar localization and found that the remaining of 70 residues at the N-terminus, CheZN70 (also termed as CheZΔ71-236), is sufficient for polar localization (Liu et al., 2020). To investigate whether the presence of N-terminal region is sufficient for the energy-related localization, the localization pattern of CheZN70 with or without sodium azide was recorded. After incubating with sodium azide for 12h, the CheZ localization pattern of CheZN70 remained unchanged as most cells had diffuse CheZ localization (Figure 4). These results suggest that the energy-related localization of CheZ may need its phosphatase activity.

Then we turned to test whether the energy-related localization of CheZ depends on its phosphatase activity. The conserved phosphatase active sites D165 and Q169 of CheZ in A. caulinodans ORS571 are important for its activity (Zhao et al., 2002; Silversmith et al., 2003; Lertsethtakarn and Ottemann, 2010; Liu et al., 2018b). We substituted these residues by alanine, respectively, and the resulting muteins CheZD165A and CheZQ169A were fused to GFP. When treated with sodium azide for 12h, the polar localization of CheZQ169A-GFP increased to 100%, which was similar as CheZ-GFP, while the localization pattern of CheZD165A-GFP in cells remained diffuse (Figure 4). These results indicate the conserved active site D165 is important for the energy-related localization; however, the conserved active site Q169 does not.

In addition, when some residues at the N-terminal helix of CheZ were deleted, such as CheZΔ2-31 and CheZΔ2-50, which reduced the polar localization of CheZ, almost 100% of cells with these truncated CheZ showed polar localization with the presence of sodium azide (data not shown). These results indicate that the C-terminus of CheZ and a conserved active site D165, but not N-terminal regions, are essential for the energy-associated localization.

Changes in the electron transport system are a source of aerotaxis signals (Taylor et al., 1979; Zhulin et al., 1997a), and cytoplasmic chemoreceptors containing PAS domains can sense the status of redox and oxygen in cells (Zhulin et al., 1996, 1997b). Among A. caulinodans ORS571 genome, six cytoplasmic chemoreceptors were found, and five of them were closely phylogenetically related to E. coli Aer (Supplementary Figure S1). Analyzing sequences of the five chemoreceptors, all of them have PAS domain and the conserved methyl-accepting signal (MA) domain (Figure 5A). Four of them, including AZC_0573, AZC_1026, AZC_1546, and AZC_3153, contain two PAS domains, while AZC_3718 (also named as IcpB) only has one PAS domain (Figure 5A). After aligning PAS domains of these intracellular chemoreceptors, except two PAS domains in AZC_1546, others showed a close relationship with PAS domains in Aer in E. coli, and in AerC in A. barasilense (Figure 5B). Interestingly, two PAS domains from AZC_1026, two PAS domains from AZC_3153, and the PAS2 of AZC_0573 are close to each other on phylogenetic tree (Figure 5B), suggesting they may play redundant functions.

The effect of these chemoreceptors on the localization of CheZ was further investigated. Deletion mutants of these chemoreceptors were constructed, and the CheZ-GFP fusion was introduced into each of them. The pattern of localization of CheZ in these mutants was compared with that in wild-type strain. When cells were cultured to the stationary phase, the ratio of cells showed polar localization of CheZs in Δazc_0573, Δazc_1026, and Δazc_1546 increased from 6% to around 20, 20, and 15%, respectively, while the localization pattern of CheZs in Δazc_3153 and Δazc_3718 remained unchanged (Figure 6). These results suggest that some intracellular chemoreceptors, AZC_0573, AZC_1026, and AZC_1546, participated in the regulation of CheZ localization.

Nitrogenase activity can only be detected at low concentration of oxygen (less than 4%) and nitrogen limiting conditions (Desnoues et al., 2003), suggesting that the hypoxia induced polar localization of CheZ may be related to nitrogen fixation. Before testing the CheZ localization under nitrogen fixation condition (hypoxia and nitrogen-limitation), we first tested the effect of nitrogen limiting on the CheZ localization. Under nitrogen limiting conditions without shaking, nitrogen-fixing rhizobacteria can move toward suitable regions with microaerobic for nitrogen fixation (Greer-Phillips et al., 2004), which was called aerotaxis. To avoid the effect of aerotaxis, we compared the localization pattern of CheZ in the presence or absence of ammonium with shaking. Compared to the group with nitrogen source (L3+C+N), the nitrogen limitation (L3+C-N) decreased the ratio of CheZ polar localization from 10 to 5% of cells showed (Figures 7A,B). For some bacteria, under nitrogen limiting conditions and excessive carbon source, they could accumulate glycogen, polyhydroxyalkanoates, or other substances to response to harsh environments (Blaby et al., 2013; Mozejko-Ciesielska et al., 2018; Lai et al., 2019). Interestingly, under nitrogen limitation, CheZ-GFP proteins in many cells are squeezed from both ends and converge in the middle of cells (Figure 7A), which is different from common diffuse localization pattern, suggesting A. caulinodans may also produce glycogen-like molecules affecting the distribution of CheZ-GFP. The cells cultured with carbon starvation (L3-C+N) were used as a control group, and CheZ-GFP in about 90% cells locates to cell poles, which is similar as the localization pattern in the presence of sodium azide (Figures 7A,B).

Then, we tested the role of free-living nitrogen fixation on CheZ localization. Under nitrogen fixation condition (hypoxia and absence of ammonium), about 90% cells showed polar localization of CheZ (Figures 7C,D). However, 100% cells showed polar localization of CheZ under hypoxia with nitrogen source (Figures 7C,D). Because no CheZ-GFP is squeezed in the middle of cells under nitrogen fixation conditions (Figure 7C), the role of nitrogen limitation can be excluded. Interestingly, when we transferred cells from hypoxia conditions of nitrogen fixation (hypoxia without nitrogen source) or non-nitrogen fixation (hypoxia with nitrogen source) to normoxia conditions, the ratio of cells with CheZ polar localization of them both decreased significantly (Figures 7C,D). With the decrease in CheZ polar localization, the squeezed localization pattern of CheZ appeared again when cells transferred from nitrogen fixation condition to nomoxia condition (Figure 7C), suggesting without nitrogen fixation, cells begin to suffer nitrogen limitation again. These results suggest the free-living nitrogen fixation of bacteria might partially release the degree of CheZ polar localization. When sodium azide was added into cell cultures under nitrogen fixation conditions, 100% cells showed a polar localization of CheZ (Figure 7E), supporting the hypothesis that the decreased localization pattern caused by nitrogen fixation is also energy dependent.