Mesorhizobium huakuii HtpG Interaction with nsLTP AsE246 Is Required for Symbiotic Nitrogen Fixation

The Mesorhizobium huakuii molecular chaperone HtpG interacts with the lipid transfer protein AsE246, which plays an essential role in effective root nodule development and symbiotic nitrogen fixation.

Abstract

Plant nonspecific lipid transfer proteins (nsLTPs) are involved in a number of biological processes including root nodule symbiosis. However, the role of nsLTPs in legume-rhizobium symbiosis remains poorly understood, and no rhizobia proteins that interact with nsLTPs have been reported to date. In this study, we used a bacteria two-hybrid system and identified the high temperature protein G (HtpG) from Mesorhizobium huakuii that interacts with the nsLTP AsE246. The interaction between HtpG and AsE246 was confirmed by far-Western blotting and bimolecular fluorescence complementation. Our results indicated that the heat shock protein 90 (HSP90) domain of HtpG mediates the HtpG-AsE246 interaction. Immunofluorescence assay showed that HtpG was colocalized with AsE246 in infected nodule cells and symbiosome membranes. Expression of the htpG gene was relatively higher in young nodules and was highly expressed in the infection zones. Further investigation showed that htpG expression affects lipid abundance and profiles in root nodules and plays an essential role in nodule development and nitrogen fixation. Our findings provide further insights into the functional mechanisms behind the transport of symbiosome lipids via nsLTPs in root nodules.

Symbiotic nitrogen fixation between legume and rhizobia is not only a major source of fixed nitrogen in natural ecosystems but also the largest natural source of nitrogen in agriculture (Smil, 1999; Vance, 2001; Herridge et al., 2008). Legume-rhizobium interaction results in formation of the root nodule (Oldroyd and Downie, 2008; Ferguson et al., 2010), a specialized plant organ for mutual nutrient exchange. The plant provides reduced carbon derived from photosynthesis and other nutrients to rhizobia, and in turn receives reduced nitrogen and amino acids from the rhizobia (Udvardi and Day, 1997; Udvardi and Poole, 2013). During legume nodulation, rhizobia are released from infection threads into host plant cells by endocytosis and are enclosed by plant-derived symbiosome membranes (Oldroyd and Downie, 2008; Masson-Boivin et al., 2009). The bacterial symbionts subsequently differentiate into functional nitrogen-fixing bacteroids (Kondorosi et al., 2013). Bacteroid development proceeds through several stages, and a number of plant and bacterial genes involved in this process have been identified (Jones et al., 2007; Gibson et al., 2008). With the development of nodules, the number of rhizobia in the infected cell increases dramatically, and is accommodated in up to 50,000 symbiosomes (Maróti and Kondorosi, 2014). Therefore, this developmental process requires a large amount of lipids to synthesize bacterial cell membranes and symbiosome membranes (Verma, 1992; Gaude et al., 2004).

To meet the demand of nodule development as described above, large quantities of lipids need to be synthesized and transported. Several bacterial and plant genes involved in lipid synthesis, modification, and transport have been identified in nodules. For example, the Sinorhizobium meliloti BacA protein, which facilitates the efficient transport of very-long-chain fatty acids out of the cytoplasm and affects the lipid-A fatty-acid content, is essential for S. meliloti to establish chronic intracellular infections within alfalfa-plant cells (Ferguson et al., 2004). The S. meliloti acpXL-lpxXL cluster is critical to the modification of S. meliloti bacteroid lipid A with very-long-chain fatty acids and plays an important role in bacteroid development within alfalfa (Sharypova et al., 2003; Ferguson et al., 2005; Haag et al., 2009, 2011). A recent study reported that a very-long-chain fatty acid (C26:25OH) is important for the successful establishment of symbiosis between Bradyrhizobium strain ORS278 and Aeschynomene species (Busset et al., 2017). The Bradyrhizobium shc is a key gene in the biosynthesis of hopanoids, and its mutation affects the lipid A structure and reduces the ability of Bradyrhizobium to survive intracellularly in Aeschynomene host plants (Silipo et al., 2014). The pmtA and pmtX1 genes in Bradyrhizobium japonicum, both of which encode a phospholipid N-methyltransferase, are required for the efficient symbiotic interaction between B. japonicum and soybean (Minder et al., 2001; Hacker et al., 2008). The methylation pathway and the phosphatidylcholine (PC) synthase pathway are two pathways for PC biosynthesis in S. meliloti, and pmtA and pcs genes are key genes of these two pathways, respectively (de Rudder et al., 1999). PC deficiency in a S. meliloti ΔpmtA Δpcs double mutant showed severe growth defects (de Rudder et al., 2000) and was unable to form nitrogen-fixing nodules on legume host alfalfa (Sohlenkamp et al., 2003; Aktas et al., 2010). In addition, other studies have found that the S. meliloti Δpsd mutant was incapable of decarboxylation phosphatidyl-Ser (PS) to form phosphatidylethanolamine (PE). Hence, the Δpsd mutant lacks PE but accumulates about 18% of PS compared with that of the wild type. The Δpsd mutant exhibited a delayed nodulation phenotype, and the nodules were almost devoid of bacteroids and unable to fix nitrogen (Vences-Guzmán et al., 2008). Orn lipids (OLs) play an important role in the symbiosis of Rhizobium tropici with Phaseolus vulgaris. The R. tropici ΔolsE or ΔolsC and double mutants ΔolsE ΔolsC deficient in OL hydroxylation showed severe nodulation phenotypes (Vences-Guzmán et al., 2011). Some plant proteins have also been shown to be required for lipid synthesis and transport. Medicago truncatula N5 (MtN5), a gene that encodes a lipid transfer protein, is required for the successful symbiotic association between M. truncatula and S. meliloti (Pii et al., 2009). A further study showed that MtN5 regulates the competence of root epidermal cells for rhizobial invasion (Pii et al., 2012). Recent research indicated that MtN5 is involved in modification of the root lipid profile, favoring the early synthesis of galactoglycerolipids (Santi et al., 2017).

Plant nsLTPs (nonspecific lipid transfer proteins) are small, soluble, abundant, and Cys-rich cationic peptides. Almost all the identified nsLTPs contain highly conserved structures characterized by four disulfide bridges formed by eight Cys residues and an N-terminal signal peptide that targets the nsLTPs to the apoplastic space (Jacq et al., 2017; Edqvist et al., 2018). The LTPs were first classified as one of two types, namely LTP1 (molecular mass ∼ 9 kD) or LTP2 (molecular mass ∼ 7 kD), based on their molecular sizes (Kalla et al., 1994). Recently, a novel LTP-classification system has been proposed by Edstam et al. (2011). Based on the presence of glycosylphosphatidylinositol-anchor modification sites, intron positions, sequence similarity and spacing between the Cys residues, LTPs are catergorized into five major types (LTP1, LTP2, LTPc, LTPd, and LTPg) and four minor types (LTPe, LTPf, LTPh, LTPj, and LTPk; Edstam et al., 2011; Salminen et al., 2016). Their exact biological roles in vivo remain unclear, although several potential functions have been proposed in diverse physiological processes, including defense against bacterial and fungal pathogens (Molina and García-Olmedo, 1997; Yeats and Rose, 2008; Ahmed et al., 2017), pollen and seed development (Park et al., 2000; Edstam and Edqvist, 2014; Wang et al., 2015b), plant growth and reproduction (Park and Lord, 2003; Chae et al., 2010), cutin and wax assembly (Kim et al., 2012; Jacq et al., 2017), lipid secretion (Debono et al., 2009; Choi et al., 2012), signaling (Maldonado et al., 2002; Jung et al., 2009), cell wall loosening (Nieuwland et al., 2005), and nodule formation (Pii et al., 2009, 2012; Lei et al., 2014).

In our previous work, we identified the nsLTP AsE246, which contributes to the transport of plant-synthesized lipids to symbiosome membranes and is required for effective legume-rhizobium symbiosis in Astragalus sinicus (Lei et al., 2014). However, the role of nsLTPs in legume-rhizobium symbiosis remains poorly understood, and no proteins in rhizobia that interact with LTPs have been reported.

In this study, by using a bacterial two-hybrid system, high temperature protein G (HtpG) from Mesorhizobium huakuii was identified as an AsE246-interacting protein, and this interaction was confirmed by in vitro and in vivo experiments. Further evidence showed that HtpG affects the content of symbiosome lipids in root nodules and plays an important role in nodule development and nitrogen fixation. Our findings facilitate a better understanding of the role of the nsLTP AsE246 in the transport of symbiosome lipids.

RESULTS

Identification of the Interaction between HtpG and AsE246

In an attempt to identify AsE246-binding proteins, we used AsE246 as a bait protein to screen the genomic pTRG plasmid library of M. huakuii 7653R (bacterial strains and plasmids in Supplemental Table S1). As a result, a few independent clones were isolated. However, after two rounds of confirmative screening and sequence analysis, most of the prey plasmids showed an open reading frame (ORF) shift and were false positive. Thus, only one candidate target protein interacting with AsE246 (HtpG) was obtained. Bioinformatics analysis showed that htpG had an ORF of 1887 bp that encoded 628 amino acids, with a predicted molecular mass of 69.2 kD and an pI of 5.08. Conserved protein domain analysis showed that HtpG belonged to the heat shock protein 90 (HSP90) superfamily, and contained two functional domains, namely an histidine kinase-like ATPases (HATPase-c) domain (Stock, 1999; Dutta and Inouye, 2000) and an HSP90 domain (Prodromou et al., 1997). The amino acid sequences of the HATPase-c domain and the HSP90 domain were conserved in representative bacteria, including Escherichia coli, Salmonella enterica, Bacillus subtilis, and Synechococcus elongates (Supplemental Figure S1), suggesting that these motifs may possess similar functions across species. Phylogenetic analysis showed that HtpG was widely distributed in different genera of rhizobia, with a highly homologous protein in Mesorhizobium (Fig. 1), suggesting that HtpG may play an important role in biological nitrogen fixation.

Figure 1.

Phylogenetic analysis of HtpG. A maximum-likelihood phylogenetic tree was constructed using MEGA6.0 software. Bootstrap values (%) were calculated by 100 resampling repetitions. The homology comparison and phylogenetic tree reveal the relationships between HtpG and homologous proteins from different genera of rhizobia. Organism names are provided after the accession numbers, and the position of M. huakuii 7653R HtpG is indicated by red color. Adjacent genera are shaded with blue and pink, respectively. Genera names are provided at the right of the figure.

HSP90 Domain Is Required for the Interaction of HtpG with AsE246

HtpG contains two functional domains, an N-terminal HATPase-c domain and a C-terminal HSP90 domain. In order to identify which domain(s) of HtpG is responsible for the interaction of HtpG with AsE246, we performed protein structure modeling and molecular docking. With the use of the I-TASSER server, the three-dimensional structures of AsE246 and HtpG were obtained (Supplemental Figure S2A and B). The results showed that AsE246 could dock with the C-terminal HSP90 domain of HtpG (Supplemental Figure S2C).

To confirm the above predictions, two deletions of HtpG were performed in pTRG plasmid for bacterial two-hybrid assay (Fig. 2A). As shown in Figure 2B, a positive cotransformant grew on the selective screening medium, but the corresponding negative cotransformant did not grow. The N-terminal fragment of HtpG (named as HtpG-1) containing the HATPase-c domain did not interact with AsE246. However, bacterial clones expressing the C terminus of HtpG or HtpG-2 were able to grow on the selective medium with the addition of 12.5 µg/mL streptomycin and 5 mM 3-amino-1,2,4-triazole (3-AT; Fig. 2B), indicating that the C-terminal HSP90 domain of HtpG is critical for its interaction with AsE246. This observation is consistent with the predicted complex conformation in protein structure modeling and molecular docking (Supplemental Figure S2C).

Figure 2.

Analysis of protein-protein interactions between HtpG and AsE246 in bacterial two-hybrid system. A, Schematic representation of the two HtpG functional domains, an HATPase-c domain and an HSP90 domain, and two deletion constructs corresponding to truncated versions of HtpG. The names assigned to the mutants are listed on the left. The deleted amino acid residues are also shown. B, Bacterial two-hybrid assays analyzing the interactions between A. sinicus AsE246 and full-length or truncated M. huakuii HtpG. Bacteria cells carrying different combinatory constructs are listed on the left. Left plate: Nonselective screening medium without streptomycin (Str) and 5 mM 3-amino-1,2,4-triazole (3-AT). Right plate: Dual selective screening medium with the addition of str and 5 mM 3-AT. CK⁺, cotransformant containing pBT-LGF2 and pTRG-Gal11^P as the positive control; CK⁻, cotransformant containing pBT and pTRG plasmid as the negative control.

HtpG Interacts with AsE246 in Vitro and in Planta

In order to validate the interaction between HtpG and AsE246, in vitro far-Western blotting and bimolecular fluorescence complementation (BiFC) assays in planta were performed. The far-Western blotting assay was performed with purified glutathione S-transferase (GST)-AsE246, His-HtpG-Strep, His-HtpG-2-Strep, and GST, which were subjected to SDS-PAGE and Western blotting with an antibody anti-Strep (Fig. 3, A and B). The same protein samples were subjected to far-Western blotting, using purified His-HtpG-strep and His-HtpG-2-strep proteins as bait, respectively (Fig. 3, A and B). As shown in Figure 3A, His-HtpG-strep was bound to GST-AsE246. No interaction was detected when GST-tag was used as a control, confirming the specificity of the observed interactions. Similarly, interaction was detected between GST-AsE246 and His-HtpG-2-Strep (Fig. 3B), which is consistent with the result that the HSP90 domain is required for the interaction of HtpG with AsE246 (Fig. 2B; Supplemental Figure S2C).

Figure 3.

In vitro interaction of HtpG with AsE246 in far-Western blotting. The purified GST-AsE246, His-HtpG-Strep (His-HtpG-2-Strep), and GST were separated by SDS-PAGE and transferred to nitrocellulose membranes and subjected to Western blotting analysis with an antibody anti-Strep. The same protein samples were separated by SDS-PAGE and transferred onto nitrocellulose membranes. Proteins were denatured, renatured, and then subjected to far-Western blotting analysis using bait proteins: A, 1 µg/mL of purified His-HtpG-strep. B, 1 µg/mL of purified His-HtpG-2-strep proteins. Binding was visualized using Anti-Strep antibodies and antimouse IgG antibody conjugated to horseradish peroxidase. His-HtpG-Strep (His-HtpG-2-Strep) and GST were used as control.

For BiFC assay, we constructed the plasmids expressing fusion proteins of the C-terminal fragment of SCFP3A protein (SCC; Kremers et al., 2006; Waadt et al., 2008), respectively, with HtpG (SCC::HtpG), HtpG-1 (SCC::HtpG-1), and HtpG-2 (SCC::HtpG-2), as well as a fusion protein of AsE246 with the N-terminal fragment of SCFP3A (AsE246::SCN; primers in Supplemental Table S2). The prepared fusion expression vectors were transiently coexpressed in Nicotiana benthamiana epidermal cells. With a confocal laser scanning microscope, strong cyan fluorescence signals in the plasma membranes of epidermal cells were observed, indicating the interaction between HtpG and AsE246 (Fig. 4). No interactions were detected between HtpG-1 and AsE246. The same signals in the plasma membranes of epidermal cells were detected between HtpG-2 and AsE246 (Fig. 4). These results indicate that HtpG interacts with AsE246 in planta and the C-terminal HSP90 domain of HtpG is essential for the interaction.

Figure 4.

In vivo interaction of HtpG with AsE246 by BiFC. N. benthamiana leaves were cotransformed with the following expression construct pairs: CBL1::SCN/SCC::CIPK24, AsE246::SCN/SCC::HtpG, AsE246::SCN/SCC::HtpG-1, and AsE246::SCN/SCC::HtpG-2. CFP fluorescence was observed in N. benthamiana leaf epidermal cells using a confocal laser microscope. Because the CFP is split into N- and C-terminal halves and fused with different proteins, no cyan fluorescence would be observed if the fusion proteins did not interact. As a positive control, Arabidopsis (Arabidopsis thaliana) CIPK24 (CBL-interacting protein kinase 24) and calcineurin B-like 1 (CBL1) were respectively fused with the C-terminal CFPC155 and N-terminal CFPN173 and coexpressed. No interactions were observed between AsE246 and HtpG-1. Scale bars = 20 μm. SCN, SCFP3A N-terminus; SCC, SCFP3A C-terminus.

Subcellular Colocalization of AsE246 and HtpG in Planta

We constructed plasmids expressing enhanced green fluorescent protein (eGFP)-tagged AsE246, HtpG, HtpG-1, and HtpG-2 fusion proteins under the control of the Cauliflower mosaic virus 35S (CaMV35S) promoter. The plasmid was delivered to N. benthamiana epidermis cells through infiltration of Agrobacterium tumefaciens strains. eGFP alone, which was distributed in the plasma membrane, cytoplasm, and nucleus (Supplemental Figure S3), was used as a control. In cells expressing eGFP-tagged AsE246, green fluorescence signals were observed in the plasma membrane (Supplemental Figure S3). In cells expressing HtpG, HtpG-1, and HtpG-2, strong fluorescence signals were found in plasma membrane as well (Supplemental Figure S3). It could be concluded that the AsE246 and HtpG proteins have similar subcellular localization patterns, further supporting that they interact on the plasma membrane in planta.

We also coexpressed Discosoma sp. red fluorescent protein (DsRED)–tagged AsE246 with eGFP-tagged HtpG/HtpG-1/HtpG-2 in N. benthamiana leaf cells. DsRED and eGFP were used as the control. As shown in Supplemental Figure S4, AsE246-DsRED was colocalized with HtpG-eGFP at the cell plasma membrane, and the HtpG-2-eGFP and AsE246-DsRED fluorescence signals were overlapped in the plasma membrane as well. Notably, we found that the AsE246-DsRED fluorescence signals were also colocalized with those of HtpG-1-eGFP in the cell plasma membrane. Although HtpG-1 protein and AsE246 protein were spatially colocalized, they did not interact with each other (Fig. 4; Supplemental Figures S3 and S4).

HtpG Colocalizes with AsE246 in Infected Nodule Cells and Symbiosomes

In order to determine whether HtpG and AsE246 are colocalized under symbiotic conditions, an immunofluorescence analysis was carried out with A. sinicus nodules formed by HtpG-eGFP–labeled M. huakuii 7653R at 28 d postinoculation (dpi). The anti-AsE246 antibody was previously tested to be specific for AsE246 (Lei et al., 2014). The localization of AsE246 and HtpG was observed by laser confocal microscopy. As shown in Figure 5, green fluorescence was observed in the cytoplasm of infected cells (Fig. 5, A, B, and D) and was evenly distributed in the symbiosomes (Fig. 5, C and E), but was not found in uninfected nodule cells (Fig. 5A). A substantial overlap of CY3-red and green fluorescence was observed in the cytoplasm of infected cells (Fig. 5, A and B), but not in the control (Fig. 5D). After the isolation of symbiosomes from mature nodules, the CY3-red and green fluorescence was overlapped around the surface of symbiosomes (Fig. 5C). However, no red fluorescence was observed in the negative control (without primary antibodies; Fig. 5E). Our previous study has indicated that AsE246 is localized in symbiosome membranes (Lei et al., 2014). Therefore, the results of this work indicate that HtpG and AsE246 are colocalized in infected nodule cells and the symbiosome membranes, further supporting their interaction and functional association in vivo.

Figure 5.

Immunofluorescence colocalization of HtpG and AsE246 in nodules observed by laser confocal microscopy. Rabbit anti-AsE246 and goat-antirabbit CY3 were applied as the primary and secondary antibodies, respectively. M. huakuii 7653R was labeled with HtpG-eGFP fusion protein, and plant nuclei and rhizobial nucleoids were stained by 4′,6-diamidino-2-phenylindole (DAPI). eGFP (green fluorescence), CY3 (red fluorescence), and DAPI (blue fluorescence) were observed in the nodule slides of A and B and the isolated symbiosomes C. D and E, no primary antibodies were added to serve as the controls. Scale bars = 150 µm (A), 20 µm (B and D), and 5 µm (C and E).

Temporal and Spatial Expression Characteristics of htpG during Nodulation

To examine the temporal and spatial expression characteristics of htpG during nodulation, htpG mRNA transcript levels at different nodulation stages and under free-living conditions were measured by reverse transcription quantitative PCR (RT-qPCR). Nodules were harvested at 10, 13, 16, 19, 22, 25, 28, 31, 34, 37, and 40 dpi with M. huakuii 7653R, and free-living bacterial cells were used as a control. In parallel, we also used RT-qPCR to analyze the transcriptional changes of nifD, which encodes the homodimeric iron protein of the nitrogenase enzyme complex and is strongly induced in mature bacteroids (Becker et al., 2004; Capela et al., 2006). The expression of symbiosis marker gene nifD was not detected under free-living conditions, but was significantly induced at 10 dpi and maintained at a constantly high level in nodules thereafter (Fig. 6B). The transcription level of htpG in developing nodules exceeded that in bacterial cells under free-living conditions at 13 dpi and peaked at 19 dpi, which was 2.1-fold higher than that in bacterial cells under free-living conditions; and subsequently the transcription declined significantly at 22 dpi and thereafter remained at a low level, followed by a significant increase at 40 dpi (Fig. 6A). We also examined the mRNA transcription levels of htpG in the roots of early stage infection. The results showed that the transcription of htpG in the infection roots was significantly lower than that in the young nodules at 9 dpi and in bacterial cells under free-living conditions (Fig. 6C), and the expression of marker gene nifD was hardly detected in the infection roots and in bacterial cells under free-living conditions but significantly induced at 9 dpi (Fig. 6D). These results suggest that htpG plays an essential role in nodulation as well as in the free-living growth of rhizobia.

Figure 6.

Changes in the expression level of the htpG gene at different development stages of root nodules. A, Transcript levels of htpG in free-living cells and in nodules at different time points after inoculation, as detected by RT-qPCR. C, Transcript levels of htpG in early stage of infection roots after inoculation, as detected by RT-qPCR. The 2^−ΔΔCt method was used to analyze the data, and rnpB and 16s rRNA were used as the reference genes. B and D, nifD gene expression levels during the same developmental stages described in A and C, serving as the control gene. The experiment was performed in triplicate and error bars represent the SD of three independent experiments. Data with different letters are significantly different as measured by a Duncan's multiple range test (P ≤ 0.05).

In order to analyze the spatial expression of the htpG gene during nodulation, the promoter of htpG, an ∼500-bp genomic DNA fragment upstream of the coding region containing the full promoter region, was cloned and fused to GUS reporter. The recombinant plasmid pRG960-htpG (Supplemental Table S1) was subsequently introduced into M. huakuii 7653R, which was then used to inoculate A. sinicus seedlings. At 5 dpi and 7 dpi, no GUS activity was detected in either infection threads or nodule primordia (Fig. 7, A and B). At 10 dpi, GUS activity was detected in bacteroids in the young nodules (Fig. 7C), but still not in the infection threads, nor in nodule primordia. At 14 dpi and 21 dpi, GUS activity was detected in bacteroids in the mature nodules, and was high in the infection zones of the root nodules (Fig. 7, D and E). Notably, for the free-living bacterial cells, the GUS activity was extremely high, which was consistent with the RT-qPCR results. No GUS activity was detected when using an empty vector pRG960 as the control. The constitutively expressed sigD promoter from M. huakuii was used as the positive control (Fig. 7F). All observations regarding the spatial expression of the htpG gene during nodulation were highly consistent with the RT-qPCR results (Fig. 6, A and C).

Figure 7.

In situ expression pattern of the htpG gene during nodulation. A–E, GUS staining of nodules induced by M. huakuii 7653R carrying P_htpG-GUS fusions. The A. sinicus plants were harvested at 5 (A), 7 (B), 10 (C), 14 (D), and 21 (E) d after inoculation. F, GUS activity in M. huakuii 7653R expressing gusA under the control of different promoters in free-living conditions (from left to right, M. huakuii 7653R carried empty plasmid pRG960, P_htpG-GUS fusions, and P_sigD-GUS fusions). Scale bars = 500 μm.

M. huakuii htpG Mutant Is Defective in Symbiotic Nitrogen Fixation

To determine whether the htpG gene of M. huakuii 7653R plays an essential role in symbiosis, we constructed htpG insertion mutants by using vector pK19mob, and the resulting strain was named as M. huakuii ΔhtpG. The effect of htpG on the growth of M. huakuii was tested in tryptone yeast extract (TY) medium under free-living conditions. It was found that mutation of the htpG gene did not affect the growth of M. huakuii ΔhtpG under free-living conditions (Supplemental Figure S5), but resulted in a defective symbiotic phenotype. At 30 dpi, the A. sinicus plants inoculated with M. huakuii ΔhtpG were dwarf with yellow leaves, and the formed nodules displayed abnormal size and color, whereas the plants inoculated with wild-type 7653R demonstrated normal growth (Fig. 8, A, B, and C). The results of quantitative analysis of the symbiotic phenotypes of the M. huakuii ΔhtpG mutants was summarized in Table 1. Significant decreases were observed in the fresh weight of the above-ground shoot biomass and nodules as well as in the nitrogenase activity of the root nodules after inoculation with the mutant. However, the number of root nodules per plant was similar to that of the plants inoculated with wild-type strain. Meanwhile, the introduction of the complement plasmid pBBR1MCS-5-htpG into the mutant rescued the symbiotic defects of the htpG mutant. The plants inoculated with the complementary strain M. huakuii ΔhtpG-C showed normal growth, and the fresh weight of the above-ground shoot biomass and nodules as well as the nitrogenase activity of the root nodules demonstrated no differences from those of the wild type (Fig. 8, A, B, and C; Table 1).

Figure 8.

Symbiotic phenotypes induced by ΔhtpG strains. The phenotypes were observed 30 d after inoculation. From left to right, the plants were inoculated with 7653R, ΔhtpG, ΔhtpG-C, and without inoculation. A, Aerial plant phenotypes. B, Whole-plant phenotypes. C, Root nodule phenotype.

Table 1. Symbiotic phenotypes induced by different htpG strains tested on Astragalus sinicus.

Nodules were harvested at 30 d postinoculation. The data represent the mean ± SE of three independent experiments. Significance of the data were analyzed using independent-samples T test. Values in each column marked by different letters indicate significant differences (P < 0.05).

Strains	Fresh Weight of Plant (mg/plant)	No. of Nodules (/plant)	Fresh Weight of Nodules (mg/plant)	Nitrogen Fixation Activity (μmol/g nodule h)
CK	52.33c ± 0.70	0	0	0
M. huakuii 7653R	300.00a ± 14.00	20.99a ± 2.83	13.78a ± 2.49	20.64a ± 4.77
M. huakuii ΔhtpG	65.00c ± 5.77	23.60a ± 3.60	7.85b ± 1.28	0.36b ± 0.27
M. huakuii ΔhtpG-C	286.67a ± 40.02	19.67a ± 2.31	11.88a ± 3.01	12.83a ± 7.17

Although the average number of root nodules induced by the mutant was similar to that of plants inoculated with wild-type strain, the mutant nodules were white in color and small in size (Fig. 8C), with a much lower nitrogenase activity. Hence, a histological analysis was performed on these nodules. The results showed that the ΔhtpG mutant nodules contained a large number of empty and uninfected host plant cells, and only very few developed bacteroids were present in the infection zones (Fig. 9, B and E). By contrast, the nodules of the plants inoculated with wild-type and functional complement strains were fully occupied by symbiotic cells, and the surrounding cortical cell layers were normally formed (Fig. 9A and D, C, and F). To further investigate the influence of htpG mutation on bacteroid development, ultrastructure comparisons of the wild-type and ΔhtpG mutant nodules were carried out by transmission electron microscopes. In the ΔhtpG mutant nodules, infection threads could be normally formed (Fig. 10, B and H), but most of the nodule cells were empty and uninfected, and only a few infected plant cells contained symbiosomes (Fig. 10, B and E). Although the bacteroids could normally differentiate in some infected cells (Fig. 10C), breakdown of bacteroids occurred in most symbiosomes (Fig. 10, B and C). In the magnified views, the symbiosome membrane seemed to be much less intact, possibly due to the degradation and early senescence of symbiosomes (Fig. 10, E, F, and I). By contrast, rhizobia in wild-type nodules differentiated to the characteristic elongated bacteroids enclosed by the symbiosome membrane (Fig. 10, A, D, and G).

Figure 9.

Observation of paraffin sections of different nodules. II: Infection zones; III: Nitrogen-fixing zone. A and D, Inoculation with 7653R. B and E, Inoculation with ΔhtpG. C and F, Inoculation with ΔhtpG-C. Scale bars = 400 μm (A–C) and 150 μm (D–F).

Figure 10.

Ultrastructures of the nodules with ΔhtpG nodules. Transmission electron microscope images of nodules of the wild-type A. sinicus at 28 d postinoculation. A, D, and G, Inoculation with 7653R. B, C, E, F, H and I, Inoculation with ΔhtpG. Scale bars = 2 μm (A–C), 1 μm (D–F, H), and 500 nm (G and I).

Our previous work found that knockdown of AsE246 impairs nodulation, resulting in the formation of fewer and smaller nodules in the AsE246 RNA interference (RNAi) root, and fewer infected cells in the RNAi nodules (Lei et al., 2014). Because the HtpG protein interacts with AsE246, a far more severe phenotypic defect should be expected when the AsE246 RNAi root was inoculated with M. huakuii ΔhtpG. As shown in Figure 11A, when the hairy roots of AsE246 RNAi plant were inoculated with M. huakuii ΔhtpG (n = 40), the plant showed a dwarf phenotype with yellow leaves, which is more severe than that of the empty-vector plant control whose hairy roots were also inoculated with M. huakuii ΔhtpG (n = 41). Moreover, 75% of AsE246 RNAi roots failed to develop any nodules with M. huakuii ΔhtpG (Fig. 11, B, F, J, and O), whereas a normal number of nodules could be formed in the empty-vector control hairy roots inoculated with M. huakuii ΔhtpG, but the nodules were white and small (Fig. 11, D, H, L, and O). No nitrogenase activity was detected in AsE246 RNAi nodules after inoculation with M. huakuii ΔhtpG (Fig. 11N). The nitrogenase activity of the empty-vector control hairy roots inoculated with M. huakuii ΔhtpG was significantly lower than that of the hairy roots inoculated with M. huakuii 7653R (Fig. 11N). After the hairy roots were inoculated with 7653R, the symbiotic phenotypes of the AsE246 RNAi (Fig. 11, C, G, K, N, and O) and empty-vector control plants (Fig. 11, E, I, M, N and O) were consistent with the results reported in previous research (Lei et al., 2014).

Figure 11.

Symbiotic phenotypes induced by ΔhtpG strains on AsE246-RNAi hairy roots. The phenotypes were observed 28 d after inoculation. A, Images of the whole plants. B–M, Root phenotypes. From left to right, AsE246 RNAi hairy roots inoculated with M. huakuii ΔhtpG (B, F, J); AsE246 RNAi hairy roots inoculated with M. huakuii 7653R (C, G, K); the empty-vector control hairy roots inoculated with M. huakuii ΔhtpG (D, H, L); the empty vector-control hairy roots inoculated with M. huakuii 7653R (E, I, M). B–E, appearance of root nodules. F–I, Close-up of boxed area in B–E. J–M, GUS histochemical staining of A. sinicus hairy root transformed with AsE246 RNAi (J, K) and empty vector (L, M). N, The nitrogenase activity of the AsE246-RNAi and the control nodules. O, Nodule numbers on RNAi and control hairy roots at 4 weeks after inoculation with M. huakuii 7653R and M. huakuii ΔhtpG. The error bars represent the standard deviations of three independent experiments. Data with different letters reveal significant differences as measured by a Duncan's multiple range test (P ≤ 0.05). Scale bars = 5 cm (A), 2 cm (B–E), and 1 mm (J–M). CK, the empty vector-control hairy roots.

Taken the above symbiotic phenotypes together, it could be speculated that htpG mutation impairs the development of bacteroids and the functions of nodules, leading to the formation of ineffective nodules.

M. huakuii htpG Mutation Affects Lipid Abundance in Nodules

A previous study has reported that AsE246 is involved in the transport of lipids to nodule symbiosome membranes (Lei et al., 2014). Considering that HtpG was identified to interact with AsE246, we determined whether M. huakuii htpG mutation affects lipid abundance in nodules. The same quantity (100 mg) of the ΔhtpG mutant nodules and wild-type nodules were harvested at 30 dpi for lipid extraction and profiling. The quantitative analyses showed that digalactosyldiacylglycerol (DGDG), phosphatidylinositols (PI), PC, and PE contents were 58.33%, 66.50%, 37.48%, and 57.88% of those in the control nodules, respectively, revealing that the abundance of these lipids is significantly decreased in the nodules inoculated with ΔhtpG (Supplemental Figure S6; Supplemental Tables S3–S6).

For a global overview of the effect of htpG mutation on lipid abundance in nodules, we analyzed the lipidomic profiles of the nodules inoculated with ΔhtpG and the wild-type 7653R. The lipid fractions of seven independent pools of nodules inoculated with ΔhtpG and wild-type 7653R were analyzed by ultra-performance liquid chromatography–mass spectrometry (UPLC-MS). A total of 9957 and 3198 features were detected in positive and negative modes, respectively. The number of features in quality control (QC) samples with RSD ≤ 30% was 8799 in positive mode and 2610 in negative mode, accounting for 88.92% and 84.44% of the total, respectively. Principal component analysis with QC samples was performed to assess the experiment quality. The results showed that the pooled QC samples were well clustered in both ion modes (positive and negative; Supplemental Figure S7, A and B), indicating that the LC-MS analysis process was reliable and resulted in anticipated qualifications (Sangster et al., 2006). For the statistical analysis, we first applied the principal component analysis to evaluate the separation between the htpG mutant nodules and wild-type nodules. The unsupervised multivariate analysis revealed significant differences between the two groups (Supplemental Figure S7, C and D). To further discriminate the two groups, a partial least squares discriminant analysis (PLS-DA), a supervised multivariate data analysis method, was conducted to test the differences between features with P-values < 0.05. As a result, the PLS-DA model clearly distinguished the experimental and control groups based on the lipid dataset (Supplemental Figure S7, E and F).

Based on the PLS-DA analysis and Q-value evaluation, the criteria of variable important for the projection ≥ 1, fold change ≥ 1.2 or ≤ 0.8333, and Q-value < 0.05 were set to identify significant differential features between the experimental and control groups. In total, there were 3904 significant features (2908 in positive mode and 996 in negative mode) satisfying the criteria. We then generated a heat map as a graphical representation of the differential features between the htpG mutant nodules and wild-type nodules (Supplemental Figure S8, A and B). The results show that the lipid compositions of the two types of nodules have remarkable differences.

Metabolite identification was performed using Progenesis QI (version 2.2). The LipidMaps (Sud et al., 2007) and the Kyoto Encyclopedia of Genes and Genomes (KEGG) database (Kanehisa et al., 2017) were used for MS1 identification, and theoretical fragments were used for MS/MS identification. The mass tolerance in MS1 and MS2 was 10 ppm. Based on the composite score of mass error, fragmentation score, and isotope similarity, the lipids with the highest scores were defined as the identified lipids. In total, 11580 lipids were identified in positive mode and 2398 lipids in negative mode, among which 3410 lipids and 1150 lipids were significantly differential, respectively. To analyze these lipid data, we summarized all the lipid species and calculated the percentages of significantly changed lipids (up-regulated and down-regulated) in the total amount of each type of identified lipid species. As shown in Figure 12, a large amount of lipids were down-regulated in the ΔhtpG mutant nodules compared with that in the wild-type nodules, especially the glycerolipids and gycerophospholipids, which accounted for 21.83% and 39.54% in positive mode (Fig. 12A) and 32.49% and 42.91% in negative mode (Fig. 12B), respectively. Further statistical analysis showed that phosphatidic acids, PC, PE, phosphatidylglycerol, PI, and PS were all significantly decreased in the ΔhtpG mutant nodules (Fig. 12, C and D). More interestingly, almost all detected glycoglycerolipids were significantly decreased, including monogalactosyldiacylglycerol, DGDG, and sulfoquinovosyldiacylglycerol (Fig. 12, C and D; Table 2). In earlier work, DGDG was found to accumulate in the symbiosome membranes of nodules (Gaude et al., 2004; Lei et al., 2014). Taken together, all these results support the speculation that HtpG acts as a functional partner of the nsLTP AsE246 and therefore affects the abundance of lipids in nodules.

Figure 12.

Differentially expressed lipids between ΔhtpG mutant and wild-type nodules. A and B, Percentage of differentially expressed lipids (up-regulated [Up] and down-regulated [Down]) in different kinds of lipid species in positive mode (A) and negative mode (B). C and D, Percentage of differentially expressed lipids (up- and down-regulated) in different kinds of glycerolipids in positive mode (C) and negative mode (D). The number of differentially expressed lipids is indicated in parenthesis. FA, fatty acyls; GL, glycerolipids; GP, glycerophospholipids; PK, polyketides; PR, prenol lipids; SL, saccharolipids; SP, sphingolipids; ST, sterol lipids; DGDG, digalactosyldiacylglycerol; MGDG, monogalactosyldiacylglycerol; SQDG, sulfoquinovosyldiacylglycerol; PA, phosphatidic acids; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositols; PS, phosphatidyl-Sers.

Table 2. Differential glycoglycerolipids between htpGmut nodules and wild-type nodules.

RT, retention Time; VIP, variable important for the projection; Score, composite score of mass error, fragmentation score, and isotope similarity; FS, fragmentation score, calculated by progenesis QI with theoretical fragmentation matches; ME, mass error; IS, isotope similarity, calculated by progenesis QI with theoretical isotope distribution.

m.z	Compound ID	RT (min)	VIP	Formula	Adducts	Score	FS	ME (ppm)	IS	Description	Fold Change HtpGmut /Wild Type
789.5345	4.58_789.5345m/z	4.58	1.68	C43H78O10	M+Cl	34.5	0	7.45	80.99	MGDG(16:0/18:2(9Z,12Z))	0.05
811.5186	5.36_811.5186m/z	5.36	1.51	C45H76O10	M+Cl	41.8	37.1	6.88	79.80	MGDG(18:2(9Z,12Z)/18:3(9Z,12Z,15Z))	0.07
777.5513	5.69_777.5513m/z	5.69	1.61	C45H78O10	M-H	45.1	48.8	−1.15	78.04	MGDG(18:2(9Z,12Z)/18:2(9Z,12Z))	0.06
813.5333	6.38_813.5333m/z	6.38	1.68	C45H78O10	M+Cl	34.4	0	5.61	78.41	MGDG(18:2(9Z,12Z)/18:2(9Z,12Z))	0.05
739.5312	7.91_739.5312m/z	7.91	1.05	C43H78O10	M-CH3	51.4	90.1	−7.03	74.69	MGDG(16:0/18:2(9Z,12Z))	0.29
841.5142	4.70_840.5070n	4.70	1.30	C45H76O12S	M+H	42.8	23.2	1.44	92.73	SQDG(22:5(5Z,8Z,11Z,14Z,17Z)/16:1(13Z))	0.30
950.5868	5.21_950.5868m/z	5.21	1.07	C51H80O15	M+NH4	37.7	3.05	3.52	89.35	DGDG(18:4(6Z,9Z,12Z,15Z)/18:4(6Z,9Z,12Z,15Z))	0.42
739.5685	10.00_756.5718n	10.00	1.29	C43H80O10	M+H-H20	44.3	34.1	−4.43	92.68	MGDG(16:0/18:1(9Z))	0.32
764.5308	5.16_764.5308m/z	5.16	1.04	C43H70O10	M+NH4	39.4	8.25	0.10	88.83	MGDG(18:3(9Z,12Z,15Z)/16:3(7Z,10Z,13Z))	0.47
794.5794	5.26_794.5794m/z	5.26	1.47	C45H76O10	M+NH4	57.2	94.3	2.22	94.40	MGDG(18:2(9Z,12Z)/18:3(9Z,12Z,15Z))	0.23
799.5353	5.27_799.5353m/z	5.27	1.12	C45H76O10	M+Na	49.3	53	2.84	96.85	MGDG(18:2(9Z,12Z)/18:3(9Z,12Z,15Z))	0.39
800.6168	8.03_800.6168m/z	8.03	1.16	C45H82O10	M+NH4	40.8	18.7	−9.98	96.01	MGDG(18:1(9Z)/18:1(9Z))	0.38
805.5782	8.83_805.5782m/z	8.83	1.18	C45H82O10	M+Na	38.4	0.385	−2.37	94.27	MGDG(18:1(9Z)/18:1(9Z))	0.41
805.5782	8.83_805.5782m/z	8.83	1.18	C45H82O10	M+Na	38.4	0.423	−2.37	94.27	MGDG(18:0(9Z)/18:2(9Z,12Z))	0.41

DISCUSSION

Symbiotic nitrogen fixation in legumes takes place in specialized organs called root nodules during legume nodulation, and large amounts of lipids need to be transported from the plant to the symbiosome membranes (Gaude et al., 2004; Lei et al., 2014). Although several LTPs have been identified as necessary for the successful symbiotic association between rhizobia and legume plants (Pii et al., 2009, 2012; Lei et al., 2014; Bogdanov et al., 2016), no proteins have been found to interact with these LTPs. The current study is the first report of a rhizobial protein that functionally interacts with a LTP of the host plant.

Using a bacterial two-hybrid system, we identified that a rhizobial protein HtpG interacts with AsE246, a nsLTP in legumes. HtpG is a heat shock protein previously predicted to contain two functional domains: an HATPase-c domain (Stock, 1999; Dutta and Inouye, 2000) and an HSP90 domain (Prodromou et al., 1997). We demonstrated that the C-terminal HSP90 domain of HtpG is critical for its interaction with AsE246 by in vitro and in vivo experiments. Similar results were reported in E. coli (Motojima-Miyazaki et al., 2010) and cyanobacterium (Sato et al., 2010). In E. coli, the fragment from the middle to C-terminal domain of HtpG is important for the interaction with protein L2, which corresponds to the C-terminal HSP90 domain of HtpG in this study (Motojima-Miyazaki et al., 2010). Therefore, the C-terminal HSP90 domain of HtpG might be extremely critical for its biological function.

Hsp90 is a ubiquitous and highly conserved molecular chaperone that constitutes up to 1% to 2% of cellular proteins in most cells under normal physiological conditions (Csermely et al., 1998; Chen et al., 2006). In eukaryotes, Hsp90 is essential for cell viability. It is not only required for the activation and stabilization of a wide variety of client proteins but also involved in important cellular pathways such as signal transduction, intracellular transport, protein degradation, protein folding, protein trafficking, receptor maturation, and innate/adaptive immunity (Taipale et al., 2010; Li and Buchner, 2013; Schopf et al., 2017). Although eukaryotic Hsp90 has been studied extensively, the function of its bacterial homolog HtpG remains elusive. Researchers found that the deletion of htpG in E. coli did not affect bacterial growth except for a slight growth disadvantage at high temperatures (Bardwell and Craig, 1988; Thomas and Baneyx, 1998). Similar results were reported in B. subtilis (Versteeg et al., 1999) and Actinobacillus actinomycetemcomitans (Winston et al., 1996). However, in cyanobacteria, HtpG is essential for thermal stress management (Tanaka and Nakamoto, 1999) and plays a role in the acclimation to cold and oxidative stress (Hossain and Nakamoto, 2002, 2003). Other research indicated that HtpG interacts with the linker polypeptides of phycobilisome in cyanobacterium (Sato et al., 2010). Another study showed that HtpG functions as a virulence factor contributing to Salmonella persistence in pigs (Verbrugghe et al., 2015). Thus, HtpG is involved in the protection of cells against a variety of environmental stresses.

Several reports have shown that HtpG is present in root nodules. By using the two-dimensional gel electrophoresis coupled with HPLC tandem mass spectrometry (LC-MS/MS), 51 proteins from M. truncatula symbiosome membranes were identified, among which 23 were bacterial proteins, including the Hsp90 heat shock chaperone protein (Catalano et al., 2004). HtpG was also detected in nodule bacteria (Djordjevic, 2004). Two chaperones, HtpG and DnaK, were sulfenylated in 4-week-old nodules (Oger et al., 2012).

However, the role of the HtpG protein in legume-rhizobium symbiosis has not been characterized yet. In this study, we identified that HtpG interacts with AsE246 and plays an essential role in nodule development and nitrogen fixation. The experimental evidence can be summarized into three major aspects. First, the temporal and spatial expression patterns of the htpG gene during nodulation indicate its role in legume-rhizobium symbiosis. We found that the htpG gene started to be expressed in nodules at 9 dpi, and was mainly expressed in the infection zones, because the bacteria start to be released from the growing tip of the infection thread into the host cell cytoplasm through endocytosis at this time point, and continuously infect the new host plant cells produced by apical meristem in the infection zones (Jones et al., 2007; Ferguson et al., 2010). The released bacteria are surrounded by a plant-derived membrane called peribacteroid membrane, and thus a large amount of lipids are required for the biosynthesis of bacterial cell membranes and symbiosome membranes (Verma, 1992; Gaude et al., 2004). These findings explain the high expression of the htpG gene in the infection zones.

It is interesting that the htpG gene was highly expressed in free-living cells (Fig. 7F), which is consistent with the RT-qPCR results (Fig. 6), suggesting that the htpG gene plays some other roles in free-living cell growth and development of M. huakuii 7653R besides nodulation. The ability of HtpG to form complexes with other bacterial proteins, especially those involved in fundamental functions, is indicative of its cellular role (Grudniak et al., 2015).

Second, the mutation of htpG impaired the development of bacteroids. After being released from the infection threads, only a few rhizobia could differentiate into bacteroids, and the bacteroids quickly began to be degraded (Fig. 10), leading to the formation of ineffective nodules. These results are consistent with the symbiotic phenotypes of AsE246 RNAi plant (Lei et al., 2014). A far more severe phenotypic defect was observed when the AsE246 RNAi roots were inoculated with M. huakuii ΔhtpG (Fig. 11).

Third, HtpG may act as a functional partner of the lipid transfer protein AsE246 and therefore affect the abundance of lipids in nodules. The quantitative analyses of lipid abundance showed that DGDG, PI, PC, and PE contents significantly decreased in htpG mutant nodules, which is consistent with the results of a previous report on AsE246. Lipidomic data further confirmed that htpG mutation affects the lipid abundance in nodules. The defect in the transport of symbiosome lipids impacts the release of bacteria from infection threads into host cells and impairs the symbiosome development as well (Lei et al., 2014). As a result, the abnormal nodules induced by ΔhtpG displayed significantly decreased nitrogenase activity. Interestingly, the number of root nodules per plant was similar to that of plants inoculated with the wild-type strain, implying that htpG mutation does not affect the bacterial infection but impacts the release of bacteria from infection threads into host cells in the infection zones of nodules. These findings are consistent with the symbiotic expression pattern of htpG in nodules. In addition, considering the result that HtpG and AsE246 were colocalized in the infected nodule cells and symbiosome membranes, it can be speculated that HtpG participates in the transport of the lipids synthesized by the host plants such as DGDG to the symbiosome membranes.

Recent research indicated that the Hsp90 protein could directly interact with lipids of different compositions, and preferentially binds with more unsaturated phospholipid species (Zhang et al., 2018). Hsp90 could also directly interact with and deform membranes via an evolutionarily conserved amphipathic helix, promoting the release of exosomes (Lauwers et al., 2018). It was reported that the expression of heat shock proteins (hsp27, hsc70, hsp70, and hsp90) in exosomes is increased in response to heat stress, and these stress proteins are encapsulated within the exosome lumen and not present on the exosome surface (Clayton et al., 2005). A recent study has indicated that Rhizobium etli CE3 grown in succinate-ammonium minimal medium could excrete outer membrane vesicles (OMVs), which contain proteins including Rhizobium heat shock proteins (Taboada et al., 2018).

Chlamydia species are obligate intracellular Gram-negative bacterial pathogens that infect human genital, ocular, and pulmonary epithelial surfaces. After internalization, Chlamydia is surrounded by a membrane-bound compartment, which is termed as inclusion and requires host-derived lipids for intracellular growth and development (Escalante-Ochoa et al., 1998; Hammerschlag, 2002; Elwell and Engel, 2012). The formation of the inclusion and its membrane is similar to that of a symbiosome and its membrane. It was found that the ceramide transfer protein CERT of the host and the Chlamydia trachomatis inclusion protein are colocalized in the inclusion membrane and interact with each other. The C. trachomatis inclusion protein-CERT interaction mediates the nonvesicular transfer of ceramide from the endoplasmic reticulum to the inclusion (Derré et al., 2011). Although human lipid transfer protein CERT mediates the nonvesicular transfer of ceramide, it is completely unrelated to plant LTPs.

Based on the results of this work and the latest research reports, we proposed a preliminary model for the mechanism through which HtpG participates in the transport of lipids in nodules (Supplemental Figure S9). The plant nsLTP AsE246 is synthesized and processed by endoplasmic reticulum and Golgi, and then transported to the plasma membrane. The heat shock protein HtpG is synthesized by the rhizobium and possibly transferred to symbiosome membranes by OMVs and then released to the cytoplasm of plant cells. The HtpG protein is recruited to the plasma membrane and interacts with AsE246, facilitating/directing the transport of lipids from plant cells to symbiosome membranes. Although this working model is speculative, it is fundamentally consistent with the current knowledge of the bacterial chaperones and plant LTPs. We found that AsE246 interacts with HtpG and both of them are located in plasma membranes and symbiosome membranes (Lei et al., 2014). Notably, recent studies have indicated that Hsp90 could directly interact with lipids (Lei et al., 2014; Zhang et al., 2018). In addition, although the HtpG protein is not an effector protein or secretory protein, it might be transported outside the cell by exosomes or OMVs (Clayton et al., 2005; Lauwers et al., 2018; Taboada et al., 2018). Moreover, it has been reported that bacterial proteins could interact with host LTPs and mediate the lipid transfer from host cells to the symbionts (Derré et al., 2011).

In conclusion, we identified a rhizobial protein HtpG that interacts with the nsLTP AsE246 in legumes, and the C-terminal HSP90 domain of HtpG is critical for the interaction. The interaction of HtpG with AsE246 may facilitate the transport of lipids from plant cells to symbiosome membranes and is therefore required for legume-rhizobium symbiosis. Our findings provide a better understanding of the functional mechanism for the transport of symbiosome lipids via nsLTP AsE246, as well as some new insights into legume nodule organogenesis and symbiosome development.

MATERIAL AND METHODS

Bacterial Strains and Growth Conditions

The bacterial strains and plasmids used in this study are listed in the Supplemental Table S1. Mesorhizobium huakuii 7653R was cultured for 3 d at 28°C in TY medium (Beringer, 1974). Escherichia coli was grown in Luria-Bertani (LB) medium (Sambrook et al., 1989) at 37°C. E. coli XL1-Blue MRF', the host strain for BacterioMatch II two-hybrid analysis, and its derivatives were grown according to the manufacturer's specifications (Stratagene). Agrobacterium tumefaciens GV3101(pMP90) strains and p19 strains (Voinnet et al., 2003) were cultured in LB liquid medium at 28°C. A. tumefaciens EHA105 strains were cultured in LB liquid medium at 28°C. Antibiotics used were as follows: ampicillin, 100 µg/mL; gentamycin, 20 µg/mL; kanamycin, 50 µg/mL; tetracyclin 12.5 µg/mL; streptomycin, 50 µg/mL.

Plant Materials and Growth Conditions

Astragalus sinicus seeds were sterilized, germinated, planted, and inoculated with M. huakuii 7653R strains as described previously (Chou et al., 2006). Nicotiana benthamiana plants were grown in a growth chamber at 22°C and 70% relative humidity under a 16-h-light/8-h-dark photoperiod for ∼6 weeks before infiltration with A. tumefaciens. After infiltration, the plants were maintained under the same growth conditions.

Bacterial Two-Hybrid Assays

The BacterioMatch II Two-Hybrid System Library Construction Kit (Stratagene) was used to detect the protein-protein interactions between AsE246, an A. sinicus nodule-specific lipid transfer protein, and M. huakuii 7653R proteins. A M. huakuii 7653R genomic library was prepared, and the experiment was carried out according to the manufacturer’s instructions. Recombinant pTRG vectors containing M. huakuii 7653R genes and their different domains were constructed (Supplemental Table S1). The pBT-AsE246 plasmid carrying the A. sinicus AsE246 gene (without signal sequences) was used as the bait to screen the M. huakuii 7653R genomic library. Protein-protein interactions were screened based on the expression of his3 and aadA, which confer His prototrophy (His⁺) and streptomycin resistance (Str⁺), respectively. Positive growth cotransformants were selected on the selective screening medium plate containing 5 mM 3-AT (Stratagene), 12.5 µg/mL streptomycin, 12.5 µg/mL tetracycline, 50 µg/mL kanamycin, and 25 µg/mL chloramphenicol. A cotransformant containing pBT-LGF2 and pTRG-Gal11^P (Stratagene) was used as a positive control for expected growth on the selective screening medium. A cotransformant containing empty vectors pBT and pTRG was also used as the negative control.

Phylogenetic Analysis and Prediction of Protein-Protein Interactions

Homology search was performed through the BLAST programs (http://www.ncbi.nlm.nih.gov/). DNAMAN software was used to identify the ORF and translate the corresponding protein. The alignment of HtpG protein sequences was performed by ClustalX 2.0 (Larkin et al., 2007), and the results were used to generate the phylogenetic tree with the maximum-likelihood method of MEGA6.0 (Tamura et al., 2013). The RPS-BLAST program was used to identify the conserved domain (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi; Marchler-Bauer et al., 2017). SignalP 4.0 Server (http://www.cbs.dtu.dk/services/SignalP/) was used to identify the signal peptide. ProtParam (https://web.expasy.org/protparam/)was used to predict the molecular weight, amino acid composition, and theoretical pI. The Berkeley Drosophila Genome Project program (http://www.fruitfly.org/seq_tools/promoter.html; Reese, 2001) was used to predict the potential promoters.

Protein structural models for AsE246 and HtpG were generated by the I-TASSER (Iterative Threading ASSEmbly Refinement) server (Zhang, 2008; Roy et al., 2010; Yang et al., 2015). The model quality was assessed by using VARIFY_3D, Ramachandran Plots and ERRAT generated with PROCHECK (http://services.mbi.ucla.edu/SAVES). Structural models were refined by energy minimization using the web-based application Chiron (Ramachandran et al., 2011). Proteins were prepared for docking with Autodock Tools v. 1.5.6 (Morris et al., 2009). Protein-protein docking was performed by GRAMM-X Protein-Protein Docking Web Server v.1.2.0 (http://vakser.compbio.ku.edu/resources/gramm/grammx; Tovchigrechko and Vakser, 2005, 2006). The images were produced by PyMol.

BiFC Assay

The BiFC assay was carried out using the method described by Li et al. (2015) and Waadt and Kudla (2008) with some modifications. The full-length complementary DNA (cDNA) of AsE246 without stop codon was amplified by PCR (primers in Supplemental Table S2) and inserted between Bam HI and Kpn I sites of pSCYNE (Waadt and Kudla, 2008) to obtain AsE246::SCN (SCFP3A N-terminus). A full-length HtpG (amino acids 1–629) and truncated HtpG-1 (amino acids 1–166) and HtpG-2 (amino acids 187–629) were amplified by PCR (primers in Supplemental Table S2) and inserted between Bam HI and Xho I sites of pSCYCE(R) (Waadt and Kudla, 2008) to obtain SCC::HtpG, SCC::HtpG-1, and SCC::HtpG-2 (SCFP3A C-terminus). The constructs were respectively transferred into A. tumefaciens strain GV3101/pMP90 by electroporation for future N. benthamiana transformation. GV3101/pMP90 strains were cultured in LB liquid medium, collected by centrifugation, and resuspended in infiltration buffer (Witte et al., 2004). Agrobacterium strains containing the BiFC constructs and the p19 silencing plasmid were mixed and adjusted to a final OD₆₀₀ of 0.5:0.5:0.3. After incubation for 3 h at room temperature, the GV3101 mixture was injected into the leaves of 5- to 6-week-old N. benthamiana plants. Three to six days after injection, fluorescence was observed using a Zeiss LSM510 laser-scanning microscope with CFP (excitation wavelength, 405 nm; emission wavelength, 477 nm) for SCFP3A_C/SCFP3A_N complexes.

Subcellular Localization of Proteins in Epidermal Cells of N. benthamiana

The subcellular localization assay was carried out using the method described by Wang et al. (2013). In brief, the full-length cDNA of AsE246 without a stop codon was amplified by PCR (primers in Supplemental Table S2) and cloned into Spe I /Kpn I site of pCAMBIA1302-eGFP to obtain AsE246-eGFP. A full-length cDNA of AsE246 without stop codon was amplified by PCR (primers in Supplemental Table S2) and cloned into Nco I /Spe I site of pCAMBIA1302GUS-DsRED to obtain AsE246-DsRED. A full-length HtpG (amino acids 1–628) without stop codon and truncated HtpG-1 (amino acids 1–166) and HtpG-2 (amino acids 187–628) without stop codon were amplified by PCR (primers in Supplemental Table S2) and inserted between Spe I and Kpn I sites of pCAMBIA1302-eGFP to obtain HtpG-eGFP, HtpG-1-eGFP, and HtpG-2-eGFP, respectively. These constructs were transferred into A. tumefaciens strain EHA105 for N. benthamiana transformation. Fluorescence signals were observed in leaf cells using a Zeiss LSM510 laser scanning microscope at 2 to 4 d after infiltration with Agrobacterium cells.

Expression and Purification of Fusion Proteins

AsE246 coding sequence was amplified by PCR and inserted in frame at the Bam HI and Xho I site of pGEX-6P-1 vector (GE Healthcare) to generate pGEX-6p-1- AsE246, which was then checked by sequencing. A full-length htpG gene without stop codon (1–1884 bp) and truncated htpG-2 (541–1875 bp) were amplified by PCR (primers in Supplemental Table S2) and inserted between Nde I and Not I sites of pHSTag (Liu and Chen, 2009) to obtain pHSTag-htpG and pHSTag-htpG-2, which were subsequently verified by sequencing. For protein expression, E. coli Rosetta DE3 (Invitrogen) harboring plasmids were induced with 1 mM isopropyl β-D-1-thiogalactopyranoside at 37°C for 4 h. His-HtpG-strep and His-HtpG-2-strep proteins were purified from the lysate using a nickel-agarose column (Bio-Rad) according to the manufacturer's instructions. AsE246 protein was purified from E. coli lysate using a GST cartridge (Bio-Rad).

Far-Western Blotting Assay

The far-Western blotting assay was carried out using the method described by Wu et al. (2007) in phosphate-buffered saline buffer supplemented with 0.05% (w/v) Tween-20 and 5% (w/v) skim milk powder. The purified GST-AsE246 protein was subjected to SDS-PAGE and transferred to membranes. After the denaturation/renaturation procedure using guanidine-HCl, the blot was incubated overnight at 4°C with 1µg/mL His-HtpG-strep or His-HtpG-2-strep proteins in the protein-binding buffer. Bound protein was detected with commercial antibody against strep (Sungene Biotech).

Immunofluorescence Experiments

A plasmid containing the HtpG-eGFP fusion protein was constructed to determine the colocalization of AsE246 and HtpG in nodules. The full-length ORF of htpG without the termination codon was amplified by PCR (primers in Supplemental Table S2). After digestion with Kpn I and Bam HI, the htpG fragment was inserted between Kpn I and Bam HI sites of pMP2463 (Stuurman et al., 2000) and fused to the N-terminal region of eGFP. The recombinant plasmid was named as pMP2463-htpG. The HtpG-eGFP fusion protein was driven by the lac promoter. The immunofluorescence experiments were carried out according to previous studies (Lei et al., 2014; Wang et al., 2015a).

Construction of Promoter-GUS Fusion and GUS Activity Assay

DNA fragments of ∼500 bp containing the htpG promoter regions were amplified with the htpG-P-F and htpG-P-R primer pairs (Supplemental Table S2). The amplicons were digested with Bam HI and Pst I and cloned into pRG960 (Van den Eede et al., 1992), resulting in the recombinant plasmid pRG960-htpG (Supplemental Table S1). Histochemical analysis of GUS activity in nodules was performed according to the protocol of Wilson et al. (1995). The plants were harvested at different time points after inoculation, and the whole root segment was stained. Root and nodule tissues were immersed in phosphate buffer (Jefferson et al., 1987) containing 5-bromo-4-chloro-3-indolyl-β-glucuronic acid at 10 mg/mL and 10 mM potassium ferricyanide. Samples were incubated at 37°C overnight, and then visualized to analyze the β-glucuronidase activity of root nodules under a light microscopy.

Construction of htpG Mutants and Functional Complement Strains

Mutation of the M. huakuii 7653R htpG gene was conducted by using the pK19mob suicide vector system (Schäfer et al., 1994). A DNA fragment (1032 bp) internal to htpG was PCR amplified with primers htpG-mut-F and htpG-mut-R (Supplemental Table S2). The PCR product was ligated into the pMD18-T vector to generate plasmid pMD18T-htpG, which was verified by DNA sequencing. The fragment digested from pMD18T-htpG by Bam HI and Eco RI was subcloned into pK19mob vector to generate the plasmid pK19mob-htpG. The resulting plasmid was then transformed into E. coli S17-1 (Simon et al., 1983) and conjugated into M. huakuii 7653R by biparental mating. The transconjugants were selected on TY plates containing Str (50 μg/mL; to kill E. coli S17-1 donor strains) and neomycin (400 μg/mL; resistance conferred by pK19mob-htpG) and a single-crossover integration into htpG was verified by PCR followed by sequencing. Strains with verified mutation in the gene htpG was named as M. huakuii ΔhtpG.

To complement the htpG mutant, full-length htpG was PCR amplified with primers htpG-mut-C-F and htpG-mut-C-R (Supplemental Table S2) and was cloned into broad-host-range vector pBBR1MCS-5 (Kovach et al., 1995). The resulting plasmid, pBBR1MCS-5-htpG, was introduced into the M. huakuii ΔhtpG by biparental mating. The resulting complemented strain was named as M. huakuii ΔhtpG-C.

Plant Transformation and Nodulation of Hairy Roots

Agrobacterium rhizogenes–mediated plant transformation was carried out using the method described in a previous study (Lei et al., 2014). Plants harboring transgenic hairy roots were transferred to pots filled with sterilized sand and grown in a chamber in a 16-h/8-h day/night cycle at 22°C. The plantlets were watered with Fahraeus nitrogen-free nutrient solution. After 3 to 5 d, plants were inoculated with M. huakuii 7653R and M. huakuii ΔhtpG, respectively. Roots and nodules were collected at 28 dpi for further analysis.

RT-qPCR

The nodules were harvested at different days postinoculation, and free-living M. huakuii 7653R cells grown in 100 mL TY liquid medium were harvested by centrifugation at 12000 g at 4°C for 1 min when the OD₆₀₀ reached up to 1 to 2. These materials were frozen in liquid nitrogen and stored at −80°C until the isolation of RNA. Total RNA was extracted from each sample using the method described by Peng et al. (2014). An aliquot of 1 μg total RNA was used for reverse transcription. The cDNA was synthesized by RevertAid Reverse Transcriptase (Fermentas) using random primers. Real-time quantitative PCR was performed with the SYBR Green Master Kits (Roche). The data were analyzed by the 2^-ΔΔCt method (Livak and Schmittgen, 2001) with rnpB (Xie et al., 2011) and 16s rRNA as the reference gene, and the experiment was performed in triplicate.

Nitrogenase Activity Measurement and Microscopic Analysis

Nitrogenase activity was assessed by acetylene reduction activity method (Hardy et al., 1973). At 30 dpi, the plants were harvested for nitrogenase activity measurement. For each sample, nine randomly chosen plants were analyzed. Every three hypogeal parts of the plant roots (including nodules and roots) were incubated in 2 mL acetylene for 2 h at 28°C in 20-mL glass bottles with rubber seals. The amount of ethylene was measured using an East & West Analytical Instrument GC 4000A gas chromatograph (Dongxi). For light microscopy, ΔhtpG mutant nodules and wild-type nodules were fixed in formaldehyde-acetic acid buffer and dehydrated by graded ethanol. After being embedded in paraffin, the nodules were cut longitudinally. Then, the slides were stained with toluidine blue and observed with an Olympus light microscope. To investigate the changes in the ultrastructure of ΔhtpG mutant nodules, transmission electron microscopy analyses were performed as described by Wang et al. (2016). Ultrathin sections were observed with an electron microscopy (Tecnai G² 20 TWIN, FEI).

Lipid Extraction and Quantification

The lipid extraction was carried out using the method described by Lei et al. (2014). To determine the differences in lipid composition between the ΔhtpG mutant nodules and wild-type nodules, nodules at 30 dpi were harvested (100 mg) for lipid extraction and profiling. The lipid extracts were introduced into the Quadrupole Time-of-Flight (Q-TOF) mass spectrometer (Agilent 6540 Accurate Mass Q-TOF LC-MS unit) operated in the positive mode.

LC-MS, Metabolite Annotation, and Data Analysis

All samples were acquired by the LC-MS system followed machine orders. First, all chromatographic separations were performed using an UPLC system (Waters). An ACQUITY UPLC CSH C18 column (100 mm × 2.1 mm, 1.7 μm; Waters) was used for the separation. The column oven was maintained at 55°C. The flow rate was 0.4 mL/min and the mobile phase consisted of solvent A (acetonitrile : H₂O = 60:40, 0.1% [w/v] formate acid and 10 mM ammonium formate) and solvent B (isopropyl alcohol: acetonitrile = 90:10, 0.1% [w/v] formate acid and 10 mM ammonium formate). Gradient elution conditions were set as follows: 0–2 min, 40–43% phase B; 2.1–7 min, 50–54% phase B; 7.1–13 min, 70–99% phase B; 13.1–15 min, 40% phase B. The injection volume for each sample was 10 μL. A high-resolution tandem mass spectrometer Xevo G2 XS QTOF (Waters) was used to detect metabolites eluted from the column. The Q-TOF was operated in both positive and negative ion modes. For positive ion mode, the capillary and sampling cone voltages were set at 3.0 kV and 40.0 V, respectively. For negative ion mode, the capillary and sampling cone voltages were set at 2.0 kV and 40 V, respectively. The mass spectrometry data were acquired in MS^E mode (Plumb et al., 2006). The TOF mass range was from 100 to 2000 D in positive mode and 50 to 2000 D in negative mode. The survey scan time was 0.2 s. For the MS/MS detection, all precursors were fragmented using 19–45 eV, and the scan time was 0.2 s. During the acquisition, the LE signal was acquired every 3 s to calibrate the mass accuracy. Furthermore, in order to evaluate the stability of the LC-MS during the whole acquisition, a quality control sample (pool of all samples) was acquired after every 10 samples, and MS data were collected and processed with Progenesis QI (version 2.2; Waters) and data analysis using software metaX (Wen et al., 2017).

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under the following accession numbers: AsE246 (ABB13623), HtpG (WP_038648151.1). GenBank accession numbers for the genes used to build the phylogenetic tree also are presented there. HtpG homologs for alignment: Mesorhizobium loti (WP_019859476.1), Mesorhizobium japonicum (WP_010910602.1), Sinorhizobium meliloti (ASP69478.1), Sinorhizobium fredii (WP_097587048.1), Bradyrhizobium diazoefficiens (AND92683.1), Escherichia coli (NP_415006.1), Salmonella enterica (NP_455083.1), Bacillus subtilis (AIX09623.1), Synechococcus elongates (BAA85851.1). HtpG homologs for phylogeny tree: Mesorhizobium.erdmanii (WP_027052330.1), Mesorhizobium opportunistum (WP_013894354.1), Mesorhizobium australicum (WP_015316884.1), Mesorhizobium plurifarium (WP_073987335.1), Sinorhizobium americanum (WP_064251992.1), Ensifer adhaerens (WP_065374778.1), Ensifer shofinae (WP_065996970.1), Aminobacter sp. J41 (WP_024848909.1), Microvirga vignae (WP_047187987.1), Microvirga guangxiensis (WP_091138055.1), Rhizobium etli CIAT 652 (ACE94682.1), Rhizobium phaseoli Brasil 5 (ARM16423.1), Rhizobium leguminosarum WSM1689 (AHF87606.1), Azorhizobium caulinodans ORS 571 (BAF90350.1), Bradyrhizobium japonicum (AJA60319.1), Cupriavidus basilensis (AJG20328.1), Methylobacterium nodulans ORS 2060 (ACL55095.1), Burkholderia gladioli BSR3 (AEA61641.1), Burkholderia thailandensis (ATF34482.1).

SUPPLEMENTAL DATA

The following supplemental materials are available.

• Supplemental Figure S1. Sequence alignment of HtpG proteins in some representative organisms.

• Supplemental Figure S2. Protein structure modeling and molecular docking for AsE246 and HtpG.

• Supplemental Figure S3. Subcellular localization of AsE246 and HtpG in Nicotiana benthamiana cells.

• Supplemental Figure S4. Subcellular colocalization of AsE246 and HtpG in Nicotiana benthamiana cells.

• Supplemental Figure S5. Effects of htpG mutation on bacterial growth under free living conditions.

• Supplemental Figure S6. Differences in abundance of DGDG, PI, PE and PC between wide-type and ΔhtpG root nodules.

• Supplemental Figure S7. PCA and OPLS-DA score plots.

• Supplemental Figure S8. Heatmap of differential features between the htpG mutant and wild-type nodules.

• Supplemental Figure S9. model for the role of AsE246 and HtpG in nodule development.

• Supplemental Table S1. Bacterial strains and plasmids used in this study.

• Supplemental Table S2. Primers used in this study.

• Supplemental Table S3. DGDG contents in nodules inoculated with ΔhtpG and 7653R.

• Supplemental Table S4.. PI contents in nodules inoculated with ΔhtpG and 7653R.

• Supplemental Table S5. PC contents in nodules inoculated with ΔhtpG and 7653R.

• Supplemental Table S6. PE contents in nodules inoculated with ΔhtpG and 7653R.