A Lipopolysaccharide Synthesis Gene rfaD from Mesorhizobium huakuii Is Involved in Nodule Development and Symbiotic Nitrogen Fixation

Rhizobium lipopolysaccharide (LPS) is an important component of the cell wall of gram-negative bacteria and serves as a signal molecule on the surface of rhizobia, participating in the symbiosis during rhizobia–legume interaction. In this study, we constructed a deletion mutant of ADP-L-glycerol-D-mannoheptosyl-6-exoisomerase (rfaD) of Mesorhizobium huakuii 7653R and a functional complementary strain. The results showed that the deletion of rfaD did not affect the free-living growth rate of 7653R, but that it did affect the LPS synthesis and that it increased sensitivity to abiotic stresses. The rfaD promoter-GUS reporter assay showed that the gene was mainly expressed in the infection zone of the mature nodules. The root nodules formation of the rfaD mutant was delayed during symbiosis with the host plant of Astragalus sinicus. The symbiotic phenotype analyses showed that the nodules of A. sinicus lost symbiotic nitrogen fixation ability, when inoculated with the rfaD mutant strain. In conclusion, our results reveal that the 7653R rfaD gene plays a crucial role in the LPS synthesis involved in the symbiotic interaction between rhizobia and A. sinicus. This study also provides new insights into the molecular mechanisms by which the rhizobia regulate their own gene expression and cell wall components enabling nodulation in legumes.


Introduction
The successful establishment of the mutualistic endosymbiosis between rhizobia and legumes results from mutual recognition and exchange of signal molecules [1,2]. Flavonoids secreted by the roots of legumes can activate the nodulation genes of rhizobia to synthesize lipochitooligosaccharides (Nod factors; NFs) [3,4]. Subsequently, NFs are sensed by the cell surface LysM receptor heteromer of leguminous plants, resulting in the root hair curling, plasma membrane invagination, and cell division to form the infection threads and root nodules [5][6][7]. The rhizobia type III secretion system (T3SS) enables rhizobia symbiosis through the NFs-independent signaling pathway [8]. The type III effectors secreted by T3SS, also known as nodulation outer proteins (nops), play an important role in determining symbiosis specificity [9]. Meanwhile, the establishment of a rhizobia symbiosis requires an intimate connection between two symbionts, and cell surface structures will be involved in the early interaction between the rhizobia and legumes. Rhizobia secrete a variety of glycans on their cell surfaces, including lipopolysaccharides (LPS), exopolysaccharides (EPS), capsular polysaccharides (KPS), and cycloglucans (CG) [10,11]. These polysaccharides and glucans are necessary for the successful establishment of rhizobia symbiosis, formation in the rfaD mutant was relatively delayed when compared with that in the wild type. Taken together, our findings revealed that the rfaD gene is required for rhizobia infection and nodulation in A. sinicus, and additionally provides new insights into the molecular mechanism by which the rhizobia orchestrate their own genes and cell wall components in the establishment of endosymbiosis in the roots of legumes.

Symbiotic Phenotype Evaluation
Astragalus sinicus was used for symbiotic phenotype analysis. The seeds were surfacesterilized in 75% (v/v) ethanol for 10 min, washed six times with sterile water for 1 min each, and then treated with 5% sodium hypochlorite for 10 min, followed by 5-7 additional washes in sterile water [31,35]. To synchronize the germination, the sterilized seeds were immersed in distilled water for 18-24 h, and then placed on the sterile plates containing 0.5% sucrose and 0.55% agar. Seedlings were grown under a 16-h/8-h light/dark cycle at 25 • C/18 • C. After 5 days, the plantlets were transferred into sterile sand, and inoculated with 1 mL corresponding rhizobial culture at OD 600 = 0.2 and watered with Fahraeus nitrogen-free plant nutrient solution [35]. At 28 days post-inoculation (dpi), the symbiotic plants were harvested, and the symbiotic phenotypes were examined.

Construction of the Deletion Mutants and Functional Complement Strains
In order to explore the biological function of rfaD in rhizobia symbiosis, the cre-lox system was used to construct the in-frame 7653R rfaD gene deletion mutant [36]. The fragments containing the flanking sequences of the 5 -and 3 -terminal coding regions of rfaD were amplified and cloned into the Acc65I/NdeIand ApaI/AgeImultiple cloning sites of pCM351. The recombinant plasmid was verified by PCR and Sanger sequencing, and transformed into E. coli S17-1, then conjugated into M. huakuii 7653R by bi-parental mating. To eliminate the gentamicin resistance gene flanked by the loxP site, the plasmid pCM157 carrying the Cre gene was introduced into the 7653R∆rfaD::Gm by bi-parental conjugation. The transconjugants sensitive to gentamicin were screened, and the strains sensitive to tetracycline were then selected. The candidate strains were further confirmed by PCR and sequencing. To validate the correction of the obtained mutant strain, the genomic DNAs from wild-type and mutant strains were extracted, and amplified by PCR with the rfaD-MAP-F/R primers; the results showed that a large DNA fragment (3.1 Kb) was amplified using the genomic DNA from wild-type M. huakuii 7653R as a template, while a small DNA fragment (2.1 Kb) was specifically amplified using the rfaD mutant genomic DNA as a template ( Figure 1). Meanwhile, these two genomic DNAs were also detected using the rfaD-ORF-F/R primers, and results showed that a 526 bp internal fragment of the rfaD ORF fragment was amplified using the genomic DNA from wild-type M. huakuii 7653R as a template, whereas no fragment was obtained using the rfaD mutant genomic DNA as a template ( Figure S1). All the fragments obtained were subsequently sequenced, and the results showed the absence of the rfaD ORF in the 7653R∆rfaD mutant strain. The strain with the rfaD deletion mutant was named as 7653R∆rfaD. To construct the recombinant plasmid carrying rfaD, the fragment containing the rfaD coding sequence and its promoter region (500 bp) was amplified and cloned into the SmaⅠ/EcoRⅠsites of the broad-host-range vector pBBR1MCS-5 [37]. The resulting construct pBBR1MCS-5-rfaD was transformed into the strain 7653RΔrfaD by bi-parental mating. The strain obtained was named as 7653RΔrfaD-C. The complementary strain was verified by PCR with the rfaD-MAP-F/R and rfaD-ORF-F/R primers ( Figure S1); the DNA fragments represent the absence of the rfaD gene from the genome (2.1 kb) of M. huakuii 7653R and the presence of the rfaD gene in the complement plasmid pBBR1MCS-5-rfaD (526 bp), respectively. The gene-specific primers used for gene amplification in this study are listed in Supplemental Table S2. To construct the recombinant plasmid carrying rfaD, the fragment containing the rfaD coding sequence and its promoter region (500 bp) was amplified and cloned into the SmaI/EcoRIsites of the broad-host-range vector pBBR1MCS-5 [37]. The resulting construct pBBR1MCS-5-rfaD was transformed into the strain 7653R∆rfaD by bi-parental mating. The strain obtained was named as 7653R∆rfaD-C. The complementary strain was verified by PCR with the rfaD-MAP-F/R and rfaD-ORF-F/R primers ( Figure S1); the DNA fragments represent the absence of the rfaD gene from the genome (2.1 kb) of M. huakuii 7653R and the presence of the rfaD gene in the complement plasmid pBBR1MCS-5-rfaD (526 bp), respectively. The gene-specific primers used for gene amplification in this study are listed in Supplemental Table S2.

LPS Extraction and Silver Staining
The LPS from the M. huakuii 7653R and 7653R∆rfaD strains were extracted with an LPS extraction kit (iNtRON 17141) according to the manufactural procedures. The prepared LPS were added with 5 × SDS loading buffer and boiled, then separated by SDS-PAGE. The gel was fixed in 30% ethanol, 10% glacial acetic acid and 7 g/L periodate at 22 • C for 20 min, followed by being washed three times in sterile water. Subsequently, 1 g/L AgNO 3 at 30 • C was added for 30 min for silver staining, with washing in sterile water for 10 s. Subsequently, 30 g/L Na 2 CO 3 (4 • C), 0.02% formaldehyde color development for 20 min (fresh preparation before use) was added. Finally 10% glacial acetic acid was added to terminate the color reaction.

Plate Inhibition Test
The M. huakuii 7653R, 7653R∆rfaD, and 7653R∆rfaD-C strains were grown in liquid TY medium overnight to OD 600 = 1.0. The cultures were mixed with the heated and melted TY solid medium sufficiently. The mixture was then poured into sterile plates. A sterile filter paper with a diameter of 5 mm was placed in the center of each plate. SDS (detergent), HCl (acid) and NaClO (oxidizing) were then dropped onto the filter paper. The plates were incubated at 28 • C for 3 days.

Construction of Promoter-GUS Reporter System and GUS Activity Assay
A 500-bp of upstream fragment of the rfaD was chosen as the putative promoter region, and then obtained from M. huakuii 7653R genomic DNA by PCR using the primers rfaD-P-F and rfaD-P-R. The amplicon was cloned into the SmaI and PstI sites of pGR960 to create the pGR960-rfaD-GUS construct [38]. The recombinant plasmid was transformed into the wild-type M. huakuii 7653R. The recombinant strain was then inoculated to the roots of A. sinicus.
The roots and nodules of the A. sinicus plants were harvested and washed with distilled water to enable us to detect the GUS expression. The histochemical staining of GUS activity in the roots and nodules was performed according to the previous descriptions [39]. Roots and nodules were stained with the GUS buffer (0.1 M sodium phosphate buffer pH 7.0, 2 mM K 4 Fe (CN) 6 , 2 mM K 4 Fe (CN) 6 , 0.1% Triton X-100, 10 mM EDTA, 0.5 mg mL −1 X-Gluc dissolved in Dimethylformamide). Samples were incubated at 37 • C for 10 h in darkness and then washed with 70% (v/v) ethanol. The root and nodule sections were observed with a light microscope (BX51, Olympus, Japan) to visualize the spatial GUS activity.

Observation and Quantification of Early Infection Events
The wild-type 7653R and mutant 7653R∆rfaD were labeled with a GFP reporter gene from vector pMP2463 [40]. The A. sinicus seedlings were inoculated with the strain 7653R∆rfaD-GFP or 7653R-GFP. The early infection events, including root hair curling, infection threads, and nodular primordia, were observed and quantified at 3, 5, and 7days post-inoculation using a fluorescence microscope (BX51, Olympus, Tokyo, Japan).

Nitrogenase Activity Measurement
The nitrogenase activity of nodules was determined using the acetylene reduction activity method as described previously [33,41]. Briefly, the nodules of A. sinicus plants were harvested at 28 dpi for nitrogenase activity analysis. For each group, nine A. sinicus plants were randomly selected for the measurement. The root systems of each three plants, including roots and nodules, were incubated with 2 mL of acetylene for 2 h at 28 • C. The amount of ethylene was detected by the gas chromatograph (East & West Analytical Instrument GC 4000A, Dongxi, Beijing, China).

Analysis of Histological Nodule Cross-Sections
The microscopic analysis was performed according to methods previously described [31]. For light microscopic analysis, some nodules were fixed in formaldehyde-Microorganisms 2023, 11, 59 6 of 18 acetic acid buffer and dehydrated. The treated nodules were embedded in paraffin and cut longitudinally. The slides were stained with the Toluidine blue O and observed with a light microscope (BX51, Olympus, Tokyo, Japan). For electron microscopic analysis, the other nodules were first fixed in 2.5% glutaraldehyde for 4 h, then immersed in 1% osmium tetroxide for 3 h, dehydrated, and finally embedded in the London resin white [42]. The ultrathin sections were observed with an electron microscope (H-7650; Hitachi, Tokyo, Japan).

Phylogenetic Analysis and Conserved Sequence Alignment
The phylogenetic analysis of the rfaD proteins from bacteria was performed with MEGA11 software [43]. The amino acid sequences of rfaD were searched through the BLASTp program (https://blast.ncbi.nlm.nih.gov/Blast.cgi, accessed on 8 May 2022). The multiple protein sequences were aligned using the ClustalW, and the aligned sequences were loaded into the MEGA11 to generate an unrooted phylogenetic tree. The evolutionary history was inferred using the neighbor-joining method. The conserved sequence alignment was completed with DNAMAN software.

Accession Numbers
The sequence data of this article can be found in the GeneBank according to the following accession numbers: Mesorhizobium huakuii 7653R rfaD (WP_038648977.1). The rfaD homologous sequences for phylogenetic analysis were provided as below:

Statistical Analysis
Statistical significance was determined by an independent-sample t test and singlefactor analysis of the variance method with SPSS16 software (SPSS, Chicago, IL, USA). Three biological replicates were set up for each experiment. The error bars in the figures represent the standard deviation of three biological replicates. The asterisks indicate means that are statistically different at * p < 0.05, ** p < 0.01 and *** p < 0.001.

Identification of rfaD in the Genome of Mesorhizobium huakuii 7653R
In the genome of M. huakuii 7653R, we found a single copy of the rfaD gene encoding an ADP-L-glycero-D-manno-heptose-6-epimerase [34]. Bioinformatics analysis showed that the rfaD gene had a 951 bp length of the open reading frame encoding 316 amino acids, with a predicted molecular weight of 35.01 kD and a theoretical pI of 5.69. Amino acid sequence alignment results showed that the rfaD proteins were relatively conserved across bacteria species, and the homologous proteins were conserved at the N-terminal regions ( Figure 1). Phylogenetic analysis showed that the rfaD protein exists in a variety of Gram-negative bacteria, mainly in Bradyrhizobium, Mesorhizobium and pathogenic bacteria, but not in Sinorhizobium (Figure 2). This result indicated that the function of the rfaD in rhizobia-legume symbiosis may be species-specific. The phylogenetic relationship was inferred by MEGA11 software using the neighbor-joining method [43]. The evolutionary distance was computed using the p-distance method and presented in the number of amino acid differences per site. The branch of M. huakuii 7653R RfaD is marked with a red triangle. Tree scale, 0.1.

The 7653RΔrfaD Mutant Was Sensitive to Abiotic Stresses and Antibiotics
We initially examined the free-living growth of 7653R, 7653RΔrfaD, and 7653RΔrfaD-C strains. The result showed that the deletion of the rfaD gene had no significant effect on the growth of the 7653RΔrfaD strain under free-living conditions ( Figure S2). Subsequently, we examined the LPS synthesis in the wild-type 7653R and 7653RΔrfaD strains. The results showed that the LPS content of the rfaD mutant strain was clearly reduced when compared with that of the wild-type strain ( Figure S2), indicating that the deletion of the rfaD affected the LPS synthesis of the 7653R strain.
In order to further test whether the deletion of the rfaD would affect the stress resistance of the 7653R strain, we selected the reagents SDS (detergent), HCl (acidic) and NaClO (oxidant) to perform the plate inhibition zone experiments. The results showed that the mutation of the rfaD in 7653R led to a significant decrease in the stress resistance of rhizobia, whereas the functional complementary strain in response to abiotic stresses was comparable to that of the wild type ( Figure 3). The previous report has shown that LPS is also resistant to polycationic antibiotics such as polymyxin B [44]. Therefore, we The phylogenetic relationship was inferred by MEGA11 software using the neighbor-joining method [43]. The evolutionary distance was computed using the p-distance method and presented in the number of amino acid differences per site. The branch of M. huakuii 7653R RfaD is marked with a red triangle. Tree scale, 0.1.

The 7653R∆rfaD Mutant Was Sensitive to Abiotic Stresses and Antibiotics
We initially examined the free-living growth of 7653R, 7653R∆rfaD, and 7653R∆rfaD-C strains. The result showed that the deletion of the rfaD gene had no significant effect on the growth of the 7653R∆rfaD strain under free-living conditions ( Figure S2). Subsequently, we examined the LPS synthesis in the wild-type 7653R and 7653R∆rfaD strains. The results showed that the LPS content of the rfaD mutant strain was clearly reduced when compared with that of the wild-type strain ( Figure S2), indicating that the deletion of the rfaD affected the LPS synthesis of the 7653R strain.
In order to further test whether the deletion of the rfaD would affect the stress resistance of the 7653R strain, we selected the reagents SDS (detergent), HCl (acidic) and NaClO (oxidant) to perform the plate inhibition zone experiments. The results showed that the mutation of the rfaD in 7653R led to a significant decrease in the stress resistance of rhizobia, whereas the functional complementary strain in response to abiotic stresses was comparable to that of the wild type ( Figure 3). The previous report has shown that LPS is also resistant to polycationic antibiotics such as polymyxin B [44]. Therefore, we further detected the growth of M. huakuii 7653R, 7653R∆rfaD and 7653R∆rfaD-C strains on the TY plates in response to the polymyxin B addition. It was found that the 7653R∆rfaD strain was more sensitive to the polymyxin B (50 µg/mL) than the wild-type strain ( Figure S4). Collectively, these results suggest that the rhizobia LPS producing the gene rfaD plays a crucial role in the bacterial cell resistance to external stresses and antibiotics. further detected the growth of M. huakuii 7653R, 7653RΔrfaD and 7653RΔrfaD-C strains on the TY plates in response to the polymyxin B addition. It was found that the 7653RΔrfaD strain was more sensitive to the polymyxin B (50 μg/mL) than the wild-type strain ( Figure S4). Collectively, these results suggest that the rhizobia LPS producing the gene rfaD plays a crucial role in the bacterial cell resistance to external stresses and antibiotics.

The In-Situ Expression of the rfaD Gene in Nodules during Symbiosis
To investigate the in-situ spatial expression of the rfaD gene during nodulation in A. sinicus, we cloned the rfaD promoter, a 500 bp genomic DNA upstream fragment. This promoter fragment was fused with the GUS reporter in the pRG960 vector. The recombinant plasmid pRG960-rfaD was subsequently introduced into the wild-type M. huakuii 7653R, which was then inoculated into A. sinicus seedlings. The samples including roots and nodules were harvested separately at 3, 5, 9, 14 and 28 dpi for GUS staining analysis (Figure 4). At 3 dpi and 5 dpi, no GUS activity signals were detected at the curled root hairs and nodule primordia ( Figure 4A,B). At 9 dpi, strong GUS activity was detected at the base of young nodules ( Figure 4C). At 14 dpi and 28 dpi, GUS activity signals were found in the infection zone of mature nodules ( Figure 4D,E). In addition, we also detected high GUS activity in the free-living bacterial cells of the 7653R strain carrying pRG960-rfaD; however, no GUS activity was detected in the M. huakuii 7653R strain with the empty vector pRG960 ( Figure 4F). The spatiotemporal expression results showed that the rfaD gene was highly expressed in the infection zone of the mature nodules, suggesting that it may play a role in the infection process during symbiosis.

The In-Situ Expression of the rfaD Gene in Nodules during Symbiosis
To investigate the in-situ spatial expression of the rfaD gene during nodulation in A. sinicus, we cloned the rfaD promoter, a 500 bp genomic DNA upstream fragment. This promoter fragment was fused with the GUS reporter in the pRG960 vector. The recombinant plasmid pRG960-rfaD was subsequently introduced into the wild-type M. huakuii 7653R, which was then inoculated into A. sinicus seedlings. The samples including roots and nodules were harvested separately at 3, 5, 9, 14 and 28 dpi for GUS staining analysis (Figure 4). At 3 dpi and 5 dpi, no GUS activity signals were detected at the curled root hairs and nodule primordia ( Figure 4A,B). At 9 dpi, strong GUS activity was detected at the base of young nodules ( Figure 4C). At 14 dpi and 28 dpi, GUS activity signals were found in the infection zone of mature nodules ( Figure 4D,E). In addition, we also detected high GUS activity in the free-living bacterial cells of the 7653R strain carrying pRG960-rfaD; however, no GUS activity was detected in the M. huakuii 7653R strain with the empty vector pRG960 ( Figure 4F). The spatiotemporal expression results showed that the rfaD gene was highly expressed in the infection zone of the mature nodules, suggesting that it may play a role in the infection process during symbiosis.

M. huakuii 7653R rfaD Gene Engages in the Early Infection Events during Rhizobia-A. sinicus Interaction
To further explore the effect of the rfaD gene on the infection process during M. huakuii 7653R-A. sinicus interaction, we introduced the pMP2463-GFP plasmid into the rfaD mutant and 7653R strains. Therefore, the 7653R∆rfaD mutant strain was marked by the GFP reporter gene, and could constitutively express the GFP protein. This transgenic strain was named 7653R∆rfaD-GFP. Subsequently, we inoculated 7653R∆rfaD-GFP into the roots of A. sinicus seedlings, while the wild-type 7653R-GFP was used as a control line. By visualizing the GPF signals during infection processes, we observed and quantified the rhizobia infection events during M. huakuii 7653R-A. sinicus interaction. In the A. sinicus roots with 7653R-GFP, the root hair curling was formed at the tips of root hairs, and infection threads were present in the curled root hairs, while the infection threads extended along the root hairs to the cortex, thereby forming a nodule primordium to develop the root nodules ( Figure 5A,C). The early infection events in A. sinicus roots carrying the 7653R∆rfaD-GFP were also observed and quantified at 7 dpi. The results showed that the number of curled root hairs and infection threads was significantly increased in the roots of A. sinicus inoculated with the 7653R∆rfaD-GFP, while the number of nodule primordia was obviously decreased in this rfaD mutant when compared with the control strain 7653R-GFP ( Figure 5B,D). Collectively, these findings suggest that the M. huakuii 7653R rfaD gene is involved in the early infection events during rhizobia-A. sinicus interaction and nodule formation.

M. huakuii 7653R rfaD Mutant Results in the Defective Nodules in A. sinicus
To further investigate the role of the rfaD gene from M. huakuii 7653R in the processes of the rhizobia-legume symbiosis, we inoculated the strains of 7653R, 7653R∆rfaD, or 7653R∆rfaD-C into the roots of the host legume A. sinicus, while the plants without inoculation were used as the control lines. At 28 dpi, we harvested all the inoculated and non-inoculated plants. The A. sinicus plants inoculated with 7653R∆rfaD exhibited a defective symbiotic phenotype: the biomass of the shoots of the plants decreased, the leaves turned yellow, and the morphology and color of the formed nodules were abnormal, being small white round nodules. However, the A. sinicus plants inoculated with the functional complementary strain exhibited normal growth and were almost similar to the 7653R wild-type plants; the shoot biomass was larger, the leaves were green, and the morphology and color of root nodules were normal with a pink color ( Figure 6A-F).
The quantitative analysis showed that, compared with the 7653R wild-type, the symbiotic phenotype of the M. huakuii 7653R∆rfaD strain, the fresh weight of the shoot biomass and nodules were significantly reduced, whereas the number of root small nodules was significantly increased, and the nitrogenase activity of root nodules after being inoculated with 7653R∆rfaD was not detectable (Figure 6G-J). Additionally, combined with the results of early infection events, it is suggested that the lack of rfaD results there is a delay in the plant-rfaD mutant symbiosis, and that they finally form ineffective root nodules. On the other hand, there was no significant difference in the shoot biomass, root nodules, nodules number, or the nitrogenase activity of nodules between the 7653R∆rfaD-C and the wild-type 7653R strains. These results indicated that the 7653R∆rfaD-C strain restored the defective nitrogen fixation ability of the 7653R∆rfaD mutant.
To next dissect the structures of root nodules of the A. sinicus, we performed a histological analysis of the nodules from the 7653R, 7653R∆rfaD, and 7653R∆rfaD-C strains. The results showed that the nodules from the 7653R∆rfaD mutant contained a considerable number of uninfected root cells, while only a small number of rhizobia cells were found in the infective zone of the nodules. In contrast, the nodules of plants inoculated with the 7653R wild-type or complement strain 7653R∆rfaD-C contained large numbers of symbiotic cells with bacteroid ( Figure 7). In order to further examine the effect of the rfaD deletion on the development of bacteroid, we observed the ultrastructure of wild-type and ∆rfaD mutant nodules by transmission electron microscopy. The results showed that the rhizobia in the wild-type nodules differentiated into normal elongated bacteroid surrounded by symbiosome membranes (Figure 8A,B). On the contrary, in the ∆rfaD mutant nodules, the differentiation in the bacteroid was abnormal; they were small spheres ( Figure 8C,D).
Taken together, these findings reveal that the rfaD deletion impairs the normal differentiation in the bacteroid, resulting in the formation of defective nodules.

M. huakuii 7653R rfaD Mutant Results in the Defective Nodules in A. sinicus
To further investigate the role of the rfaD gene from M. huakuii 7653R in the processes of the rhizobia-legume symbiosis, we inoculated the strains of 7653R, 7653RΔrfaD, or 7653RΔrfaD-C into the roots of the host legume A. sinicus, while the plants without inoculation were used as the control lines. At 28 dpi, we harvested all the inoculated and noninoculated plants. The A. sinicus plants inoculated with 7653RΔrfaD exhibited a defective symbiotic phenotype: the biomass of the shoots of the plants decreased, the leaves turned yellow, and the morphology and color of the formed nodules were abnormal, being small white round nodules. However, the A. sinicus plants inoculated with the functional com- type plants; the shoot biomass was larger, the leaves were green, and the morphology and color of root nodules were normal with a pink color ( Figure 6A-F).  type and ΔrfaD mutant nodules by transmission electron microscopy. The results showed that the rhizobia in the wild-type nodules differentiated into normal elongated bacteroid surrounded by symbiosome membranes (Figure 8A,B). On the contrary, in the ΔrfaD mutant nodules, the differentiation in the bacteroid was abnormal; they were small spheres ( Figure 8C,D). Taken together, these findings reveal that the rfaD deletion impairs the normal differentiation in the bacteroid, resulting in the formation of defective nodules.

Discussion
The symbiotic nitrogen fixation between rhizobia and legumes is a delicate and complex process, which is involved in a large number of genes and chemical components from both symbionts [2,45]. Among them, the lipopolysaccharide (LPS) is an important component of the cell wall of gram-negative bacteria and plays an important role in the symbiosis between rhizobia and leguminous plants [10,18]. The LPS in E. coli was extensively studied [27,[46][47][48], and some rhizobia also have the gene information of the LPS synthetic

Discussion
The symbiotic nitrogen fixation between rhizobia and legumes is a delicate and complex process, which is involved in a large number of genes and chemical components from both symbionts [2,45]. Among them, the lipopolysaccharide (LPS) is an important component of the cell wall of gram-negative bacteria and plays an important role in the symbiosis between rhizobia and leguminous plants [10,18]. The LPS in E. coli was extensively studied [27,[46][47][48], and some rhizobia also have the gene information of the LPS synthetic pathway, but the precise functions of LPS synthesis-related genes in the process of symbiotic nitrogen fixation remain elusive. It has been reported that in some Gram-negative bacteria, the deletion of the rfaD gene (encoding an ADP-L-glycero-D-manno-heptose-6-epimerase) will affect the biosynthesis of the inner core of the LPS [16,27]. Here, we focused on the rfaD gene in M. huakuii 7653R, by investigating its roles in LPS synthesis and stress resistance, its expression patterns during symbiosis and its roles in the early infection and nodulation.
In this study, we constructed a rfaD gene deletion mutant using the cre-loxp system and explored its functions in free-living growth and during symbiosis. Quantitative detection of the LPS extracted from 7653R∆rfaD and wild-type 7653R strains showed that the deletion of the rfaD gene did directly lead to a lower level of LPS synthesis. Furthermore, the ability of the 7653R∆rfaD strain to resist external stresses decreased significantly under free-living conditions. It is reasonably proposed that this strain sensitivity to stresses may have resulted from the lower LPS production of this ∆rfaD mutant. Because the LPS is an important component of the cell wall of Gram-negative bacteria, the very low amounts of LPS may change the permeability of the cell wall, thereby affecting the ability of bacteria to resist external stress. These results further indicate that the epimerase encoded by the rfaD converts the ADP-D-glycerol-β-D-mannoheptose into the ADP-Lglycerol-β-D-mannoheptose; this reaction is necessary for LPS synthesis [27]. Therefore, the ADP-L-glycerol-β-D-mannoheptose can participate in the synthesis of the inner core of the LPS. The core polysaccharide is connected to the other two important components of the LPS: lipid A and O antigen; thus, the defective synthesis of the core polysaccharide has an impact on the structure of the LPS.
LPS is also considered as one of the essential components in the symbiosis between rhizobia and legumes [19,49]. The mutant of the rfaD gene of Azorhizobium caulinodans ORS571, which forms nodules on the stems and roots of Sesbania rostrata, was more advanced than the wild-type strain, but the mutant nodules were ineffective [50]. In this work, the inoculated 7653R∆rfaD into the roots of A. sinicus, the symbiotic phenotype analysis showed that the shoots of the symbiotic plants grew poorly, and the leaves were yellow in color (an indicator of lacking N). Furthermore, the development of root nodules was abnormal and defective: in the ∆rfaD mutant roots, nodules were small and white, and the nitrogenase activity of nodules was not detectable. This led us to think about and analyze the possible causes of these symbiotic phenotypes. Because of the deletion mutation of the rfaD, the structure or synthetic quantity of the LPS changed in the mutant strain, which further interfered with the balance of symbiosis and immunity during the interactions between the rhizobia and the host, thus affecting the normal release and survival of the rhizobia from the infection threads (ITs), and significantly reducing the number of infected plant cells in the root nodules; the mutant forms were therefore ineffective and white nodules. Furthermore, the enhanced sensitivity of the outer membrane of the rfaD mutant may affect the normal differentiation of the bacteroid, which leads to a loss of nitrogen fixation ability of the mutant nodules. The results indicated that the rfaD is an important gene in the process of symbiotic nitrogen fixation, suggestive of the activation of symbiotic nitrogen fixation by regulating the LPS synthesis. The previous studies reported that the LPS could regulate symbiotic nitrogen fixation at the multiple infection stages, such as the inhibition of plant defense responses during infection, and participation in the formation of infection threads [18][19][20]. Core polysaccharide and O-antigen of the LPS may change during development and symbiotic interaction [20]. The rhizobia LPS would become more hydrophobic during bacterial development, and would also change due to plant exudate [51][52][53]. Therefore, the LPS synthesis-related genes may play different roles in the autogenesis growth and symbiotic interaction. The results from the in-situ expression experiment showed that the rfaD gene was highly expressed in the infection zone of root nodules, but not in the early infection process. This result may be due to the low expression of the gene in the early stage of infection. Alternatively, it is predicted that the rfaD gene also functions in the infected zone at the late stage of nodulation, such as protecting bacterial cells in the nitrogen fixation zone as a barrier, or participating in the molecular information exchange at the symbiotic interface. The researches on rfaD in some pathogenic bacteria reported that the deletion of this gene would lead to a short-of-LPS-sugar chain, thereby reducing the adhesion of bacteria to the host cells and reducing the pathogenicity of bacteria [26,27]. Moreover, the observation on the 7653R∆rfaD inoculated into A. sinicus roots found that the number of infection threads increased at the early stage of infection; this result may be caused by the decrease in the adhesion ability of the rhizobia 7653R∆rfaD (due to the lower LPS synthesis) during the infection processes, resulting in a defective nodule development. In the process of symbiosis with the rhizobia, plants increase their nitrogen fixation ability by regulating the number of nodules [54,55]. However, the low number of nodule primordia in the 7653R∆rfaD strain may be due to the lagging process of nodule formation. The above findings indicate that the rfaD gene plays a role in the infected zone after nodule formation, regulating nodule development and nitrogen fixation, although it is not highly expressed at the early stage of nodule formation.
In conclusion, the present study showed an ADP-L-glycerol-D-mannoheptosyl-6exoisomerase coding gene, rfaD from M. huakuii 7653R. The expression, localization and function experiments revealed the rfaD engaged in the LPS synthesis required for nodulation during the M. huakuii 7653R-A. sinicus interaction. These new findings advance our understanding of the function of LPS synthesis-related genes from rhizobia in the establishment of endosymbiosis in legumes.