Elucidation of the 3-O-Deacylase Gene, pagL, Required for the Removal of Primary β-Hydroxy Fatty Acid from the Lipid A in the Nitrogen-fixing Endosymbiont Rhizobium etli CE3*

Background: R. etli lipid-A de-O-acylation occurs during symbiosis. Results: R. etli produces PagL, which de-O-acylates lipid A, and is active on E. coli lipid A, and a pagL mutant has a symbiotic defect. Conclusion: R. etli CE3 contains PagL; PagL acts on diverse lipid A structures and may be needed for proper R. etli-bean symbiosis. Significance: PagL may be important for proper infection of the bean host plant. Until now, the gene responsible for the 3-O-deacylation of lipid A among nitrogen-fixing endosymbionts has not been characterized. Several Gram-negative animal pathogens such as Salmonella enterica, Pseudomonas aeruginosa, and Bordetella bronchiseptica contain an outer membrane 3-O-deacylase (PagL) that has been implicated in host immune evasion. The role of 3-O-deacylated lipid A among nitrogen-fixing endosymbionts, plant endophytes, and plant pathogens has not been studied. However, D'Haeze et al. (D'Haeze, W., Leoff, C., Freshour, G., Noel, K. D., and Carlson, R. W. (2007) J. Biol. Chem. 282, 17101–17113) reported that the lipopolysaccharide from Rhizobium etli CE3 bacteroids isolated from host bean root nodules contained exclusively tetraacylated lipid A that lacked a lipid A β-hydroxymyristyl residue, an observation that is consistent with the possibility of PagL activity being important in symbiosis. A putative pagL gene was identified in the R. etli genome sequence. With this information, we created a pagL− mutant strain derived from R. etli CE3. Using mass spectrometry, we demonstrated that the mutant lacks 3-O-deacylated lipid A. The parent and mutant LPS were very similar as determined by gel electrophoresis and glycosyl composition analysis using gas chromatography/mass spectrometry. However, fatty acid analysis showed that the mutant lipid A contained larger amounts of β-hydroxypentadecanoic acid than that of the parent. Furthermore, the mutant was adversely affected in establishing symbiosis with its host, Phaseolus vulgaris.

The Gram-negative endosymbiont Rhizobium etli forms nitrogen-fixing symbiosis with the legume host Phaseolus vul-garis (common black bean). Nitrogen-fixing endosymbiosis is initiated through chemical signaling between the symbiont (bacteria) and legume host, which leads to organogenesis of root nodules and the internalization of the bacteria. The invading symbionts travel through infection threads to the root nodule where they are endocytosed. After endocytosis, the bacteria develop into bacteroids within plant-derived peribacteroid membranes and form nitrogen-fixing symbiosomes. For a current review on legume symbiosis, refer to Oldroyd and Downie (1).
Surface and secreted bacterial polysaccharides are necessary for complete symbiosis (2,3). LPS, a surface polysaccharide located in the outer membrane, is the focus of this study. Partial disruption of LPS biosynthesis (e.g., O-chain biosynthesis and lipid A fatty acid acylation) leads to disruption of proper symbiosis in many nitrogen-fixing symbiotic systems (4 -8). For a detailed review on the role of LPS in symbiosis, see Carlson et al. (4). The general structure of R. etli CE3 LPS has been extensively studied (4, 9 -11). D'Haeze et al. (9) determined that subtle structural changes occurred to the O-chain polysaccharide of R. etli CE3 bacteroid LPS when compared with that of free living CE3 whereby a single methyl group was added to the 2-position of a fucosyl residue. In addition, the lipid A of CE3 bacteroid LPS was exclusively tetraacylated as compared with the free living lipid A, which contained a mixture of penta-and tetraacylated lipid A. It was speculated that tetraacylated lipid A resulted from the removal of a primary ␤-hydroxymyristoyl residue located at position 3 of the proximal glucosamine (or 2-aminogluconoate) residue by a 3-O-deacylase. Indeed, Basu et al. (12) reported 3-O-deacylase activity in the membranes of a number of symbionts including R. etli CE3. The outer membrane 3-O-deacylase enzyme PagL has been reported in a number of Gram-negative pathogens including Salmonella typhimurium, Salmonella enterica, Pseudomonas aeruginosa, and Bordetella bronchiseptica (13) and has been implicated in the evasion of the host immune response (14,15). It is possible that plant symbionts, endophytes, and pathogens modify their lipid A in a similar way to promote infection.
To understand the role of PagL concerning symbiosis, pathogenesis, and endophytic association with agriculturally and economically significant host plants, it is important to locate and characterize the gene that encodes PagL in these organisms. Although the membranes from R. etli CE3 demonstrated 3-O-deacylase activity in vitro (12), the isolated enzyme has not been characterized, and consequently, the amino acid sequence and the gene encoding PagL are not known. Furthermore, R. etli PagL could not be identified by amino acid sequence alignment with known PagL proteins because of low sequence identity. The crystal structure of the P. aeruginosa PagL was solved (16), and a conserved domain was elucidated that directly enabled us to identify a putative pagL gene in the R. etli CE3 genome (17). Therefore, we hypothesized that this gene is responsible for 3-O-deacylation of R. etli CE3 lipid A. In addition, we have identified R. etli CE3 PagL homologues in the genomes of a number of endosymbionts including Sinorhizobium meliloti, Sinorhizobium medicae, Sinorhizobium fredii, Rhizobium leguminosarum, R. etli, and Methylobacterium nodulans, as well as the plant pathogens of the genus Agrobacterium and the poplar tree endophyte Methylobacterium populi (18) (supplemental Fig. S1).
In this study, we mutated the pagL gene via insertion mutagenesis in the model organism R. etli CE3 and determined its functional involvement in the 3-O-deacylation of lipid A. We further characterized the effect of the mutation on LPS synthesis and symbiosis with the host plant P. vulgaris.

EXPERIMENTAL PROCEDURES
Bacterial Strain, Plasmids, and Growth Conditions-Bacterial strains and plasmids are described in Table 1. Escherichia coli strains were grown on LB medium at 37°C. Rhizobium strains were grown on tryptone yeast medium containing 10 mM CaCl 2 or Y minimal medium containing 2% sucrose at 30°C (19). Antibiotics were added in the following concentrations where indicated: for E. coli, 15 g/ml gentamicin, 10 g/ml tetracycline, or 50 g/ml ampicillin; and for R. etli, 30 g/ml gentamicin or 20 g/ml tetracycline.
Mutagenesis of the 3-O-Deacylase Gene-The putative 3-Odeacylase (pagL) gene from R. etli bv. phaseoli CE3 was mutated using insertion mutagenesis. PCR cloning was used to create a DNA fragment with the disrupted pagL gene and homologous flanking regions to be cloned into a suicide plasmid. Conjugal transfer of the constructed plasmid into R. etli CE3 was performed, and double recombinants were selected. Plasmid extractions, gel extractions, and PCR/enzyme cleanup was performed using Qiagen mini prep kits. Restriction enzymes were purchased from Promega. PCR reagent (iProof TM high fidelity polymerase) was purchased from Bio-Rad.
A primer pair was engineered with XbaI (underlined) and BamHI (underlined) restriction sites (AGTCCTCAGACCGT-TATTGCCGCTGT and AGTCGGATCCAGCTGTTGCTG-CCAGC) to construct an ϳ1.1-kb PCR product containing 225 bp of the pagL gene (5Ј region) and 873 bp DNA upstream of the pagL gene. The PCR product was cloned into pUC18 to create plasmid pPagL-Up. A second PCR product was constructed using a primer pair engineered with BamHI (underlined) and XmaI (underlined) cloning sites to create an ϳ1.5-kb product containing 239 bp of the pagL gene (3Ј region) and 916 bp of downstream DNA. The product was cloned into pUC18 to create the plasmid pPagL-Dwn. The two inserts were tethered together by subcloning the insert from pPagL-Up into pPagL-Dwn (double digest with XbaI and BamHI) to create pPagL-UpDwn, which contained an insert region with upstream DNA, the pagL coding region (ϳ100-bp deletion near the center of the pagL gene), and downstream DNA. The center of the pagL region was engineered with a BamHI cloning site. The gentamicin-resistant cassette aacC1 from plasmid pMS255 (20) was cloned into the pagL region at the BamHI cloning site to create plasmid pPagL-UpGmDwn. The insert from pPagL-UpGmDwn was subcloned into the suicide vector pEX18-Tc (21), which contains the sacB lethal gene (double digest with XbaI and XmaI) to create pPagL-KO. The plasmid pPagL-KO was moved into R. etli CE3 via tri-parental mating as described previously (6). A pagL Ϫ mutant was obtained and was given the strain number R. etli ODB31. A construct containing the pagL gene was created to complement the mutant strain. The pagL gene was PCR-cloned into plasmid pRK404E1 (22) to create plasmid pRePagL using a forward primer engineered with a BamHI cloning site (underlined) and homologous DNA starting 14 bp upstream of the pagL gene (ATCGATGGATCCGC-TTGGTGCGAATCAAAG). The reverse primer was engineered with an EcoRI cloning site (underlined) that recognized homologous DNA 30 bp downstream of the pagL coding region (ATCGATGAATTCGCAGGCTTTTCTCGTAAAG). The insert was oriented in parallel with the lac promoter of pRK-404E1 to promote transcription. Plasmid pRePagL was moved into strain ODB31 via tri-parental mating as described above.
All of the strains were verified by PCR with the above primers that were used to create pRePagL (see Fig. 1). In addition, the pagL gene containing the native promoter was PCR-cloned into the EcoRI and HindIII cloning sites of pRK404E1 using the primer set AAGTCAGAGAATTCCAGGAATTGACCGC-AGGC and CATTAGGCAAGCTTGTGATCGGGATGGTG-TTTGAA. The pagL gene was oriented anti-parallel to the pRK404E1 lac operon. Lipopolysaccharide Isolation and Analysis-Briefly, LPS was extracted by the hot water phenol extraction method (23) and dialyzed against deionized water using 12,000 -14,000 molecular weight cutoff dialyzes bags and subsequently lyophilized. Composition of LPS was determined by GC/MS of trimethylsilyl-derived methyl glycosides and fatty acid methyl esters as described previously (10, 24) using a Hewlett Packard 5890 Series II GC/MS analyzer fitted with a 30-m DB-1 column (J&W Scientific).

Matrix-assisted Desorption Ionization of Lipid A-Lipid
A was isolated by mild acid hydrolysis as described previously (25). Briefly, ϳ5 mg of extracted LPS was treated with 1% acetic acid (100°C, 1 h) to partition the lipid A. Partitioned Lipid A was separated by 3ϫ chloroform/methanol/water (2:2:1.8, v/v/v) extraction. The lipid A containing chloroform layers were pooled and dried under a stream of air. Isolated lipid A was dissolved in a 3:1 (v/v) chloroform/methanol mixture, mixed 1:1 (v/v) with 2,4,6-trihydroxyacetophenone matrix and spotted on a MALDI plate. Spectra were acquired by a MALDI-TOF analyzer (Applied Biosystems, 5800) in the positive reflectron mode operating at a 20 kV extraction voltage.
Deoxycholic Acid-PAGE and Silver Stain of Lipopolysaccharide-Polyacrylamide gel electrophoresis in the presence of deoxycholic acid (DOC) 3 was adapted from Reuhs et al. (26). A loop full of bacteria grown on TY agar medium was dissolved in 100 l of lysis buffer (1% SDS, 50 mM Tris, and 10 mM EDTA) and incubated at 100°C for 10 min to release LPS from the cell envelope. The sample was microcentrifuged (14,000 rpm) to pellet cell debris. The supernatant (containing LPS) was mixed 1:1 with sample loading buffer, and 2 l was loaded for DOC-PAGE. The gel was silver-stained as described previously (27).
Plant Growth and Inoculation-Black turtle beans (lot HP-BTB-04) from Wheatgrasskits.com (Springville, UT) were surface-sterilized in 95% ethanol for 1 min followed by three rinses in sterile deionized water. The beans were further sterilized in 5% hypochlorite (2 min) followed by 10 rinses in sterile deionized water. Then the beans were transferred to 0.8% tap water agar plates and germinated in the dark for 3 days at room tem-perature. The sprouts were transferred to sterile 500-ml Erlenmeyer flasks containing a foam top, hydroponic beads immersed in defined, nitrogen-free liquid plant medium (Fahreus medium, (28)) and surrounded with a brown paper bag. The plants were kept in a Conviron growth chamber and were maintained at 24°C with 14 h of light and 20°C with 10 h of darkness. Bacteria used as inoculants were grown to mid/late log phase in liquid TY medium containing appropriate antibiotics. The bacteria were pelleted, washed in sterile phosphatebuffered saline (pH 7.4), and resuspended to an approximate A of 0.15 (600 nm wave length). Then 1 ml of inoculate was added to the plants 11 days after germination. Plant phenotypes were observed and scored. The experiment was repeated three times. 30 plants per inoculate strain was used in one experiment, and 15 plants per inoculate strain was used in subsequent experiments.
Acetylene Reduction Assay-The plants were inoculated with appropriate strains as described above and assessed for their ability to fix nitrogen by the acetylene reduction assay as previously described (6). Statistical analysis was performed using the Student's t test.

RESULTS
Creation of the pagL Ϫ Mutant in R. etli CE3-To identify the gene responsible for 3-O-deacylase activity in R. etli and study its function, the putative pagL gene was mutated. The putative pagL gene (accession number ABC91065.1) in R. etli CE3 (isogenic to R. etli CFN 42) is located on the chromosome and contains the conserved PagL catalytic domain (16,17) and predicted outer membrane protein structure. The pagL Ϫ mutant was created via insertion mutagenesis as described under "Experimental Procedures" and was given the strain name R. etli ODB31. To validate phenotypic differences observed between the mutant and parent strain, the mutant ODB31 was complemented with the native pagL gene from parent strain CE3 (Fig. 1).
Lipopolysaccharide Composition of the pagL Ϫ Mutant-To study differences between parent and mutant strain LPS, the composition of each LPS was determined. The LPS of R. etli CE3 has been studied extensively, and its structure is known (4,9,29). When compared with parent strain CE3, the mutant ODB31 contained the known glycosyl residues found in the O-chain polysaccharide (GlcA, N-acetylquinovosamine, 3-Omethyl-6-deoxytalose, Man, and fucose), the core (Man, Gal, galacturonic acid, and 3-deoxy-d-manno-2 octulosonic acid), and lipid A (GlcN and galacturonic acid). In addition, the DOC-PAGE profiles of parent and mutant LPS displayed similar banding patterns (Fig. 2). These results suggest that the mutant maintained the general carbohydrate structure as the parent strain LPS. The removal of fatty acids by PagL produces LPS with a lower molecular weight. Because the mutant does not have a functional PagL, the LPS molecules migrate at a slightly higher molecular weight when compared with parent and complemented strain (Fig. 2).
Mass Spectroscopic Analysis of the pagL Ϫ Mutant Lipid A-Mass spectroscopic analysis of mutant ODB31 lipid A was performed to characterize the loss of 3-O-deacylase activity on its structure. The spectra are shown in Fig. 3, and proposed com-positions for the various ions are listed in Table 2. The spectrum obtained from parent strain lipid A (Fig. 3, panel I) contains two major ion groups, each displaying a mixture of microheterogenous ions. Microheterogeneity occurs because of variation in acyl chain length (Ϯ28 mass units [C 2 H 4 ] or Ϯ14 mass units [CH 2 ]), the presence of ␤-hydroxybutyryl group (Ϯ 86) at the C27 position of 27-hydroxyoctacosanoic acid, and the oxidation of the proximal glucosamine to 2-aminogluconoate (Ϯ 16). The higher molecular weight ion group, m/z 1914 -2022, represents penta-acylated lipid A, whereas the lower molecular weight ion group, m/z 1624 -1774, represents tetraacylated lipid A. Two forms of tetraacylated lipid A are present. One form is an artifact of the isolation process caused by the acid-catalyzed ␤-elimination of the 3-O linked primary fatty acid during the mild acid hydrolysis procedure as reported by Que et al. (30). This occurs on lipid A species in which the proximal 2-aminogluconate residue forms a lactone during mild acid hydrolysis of the LPS. The resulting structure(s) on MS analysis, structure(s) C, show ions of m/z 1738.7, 1710.6, 1652.5, and 1624.4 (Fig. 3). The second tetraacylated lipid A is the product of PagL activity that results from the hydrolysis of the 3-O-linked hydroxyl fatty acid by 3-O-deacylase (PagL) leaving a hydroxyl group at the C3 position of the proximal glucosamine/2-aminogluconate (Fig. 3, structures D and E) The parent strain CE3 produced a mixture of ions consistent with A, B, C, and D structures, whereas ions caused by E structures were low in intensity (Fig. 3, panel I, and Table 2). The lipid A from the mutant ODB31 did not contain detectable ions consistent with structures D or E (Fig. 3, panel II, and Table 2). However, structures A, B, and C were present. The lipid A from the complemented mutant recovered the presence of D and E structures (Fig. 3, panel III, and Table 2). Because the expected product resulting from PagL activity is structure D, these results strongly suggest that the R. etli CE3 pagL gene product is responsible for the removal of a primary ␤-hydroxy fatty acid from the lipid A.
It was noticed that the relative intensities of the D and E to the A, B, and C ions for the complemented mutant lipid A were significantly greater than for the parent lipid A, suggesting that the PagL activity in the complemented mutant may be greater than in the parent. In fact the D:B ion intensity ratio for the parent lipid A was 1.4, whereas this ratio for the complement lipid A was 3.1, indicating that the PagL activity in the complement was two to three times greater than in the parent. It was thought that this may be due to the fact that the complement PagL was under lac promoter control rather than the R. etli native promoter. Therefore, a complement was prepared using a plasmid that contained pagL with its native promoter. This complement also gave results indicating that it had more PagL activity than the parent, i.e., repeated isolation of the parent and mutant lipids A, and comparison with the native promoter pagL complement showed a D:B ratio of 1.1, 0.0, and 3.8, respectively, indicating that this complement had 3-fold greater PagL activity than the parent (data not shown). It seems possible that for both complements the increase in PagL activity may be due to the presence of multiple copies of the pagL-containing plasmid rather than due to a difference in the lac versus native promoter.
E. coli LPS Is Deacylated by the R. etli PagL Enzyme-The Rhizobium 3-O-deacylase, PagL, has less than 20% amino acid sequence similarity when compared with the PagL of enteric bacteria Salmonella spp. and P. aeruginosa. Enteric bacteria share a similar general lipid A structure defined by a bis-phosphorylated diglucosamine backbone containing four ␤-hydroxymyristic primary fatty acids at the 2Ј, 3Ј, 2, and 3 positions of the distal and proximal GlcN residues, respectively. The 2Ј and 3Ј primary fatty acids contain ester-linked myristoyl and lauroyl subsitutents, respectively, at their ␤-OH groups creating the acyloxyacyl lipid A moieties (Fig. 4, structure A). A 3-Odeacylase has not been detected in E. coli, and this is consistent with the fact that the E. coli genome does not contain a PagL homologue. Therefore, the R. etli PagL was introduced into the E. coli Top 10 cloning strain (E. coli Top 10/pRePagL) to determine whether R. etli PagL can de-O-acylate E. coli lipid A.
Isolated E. coli LPS was subjected to the mild acid hydrolysis procedure as described under "Experimental Procedures" and subjected to MALDI-TOF MS analysis. The mass spectrum of the lipid A from the parent E. coli strain (Fig. 4, panel I) contains   (Fig. 5), and although pink in color, the root nodules were considerably smaller compared with those found on plants inoculated with the parent strain (Fig. 5). The acetylene reduction assay was employed to determine how efficiently strain ODB31 fixed nitrogen during symbiosis (Fig. 6). At 15 dpi, the ODB31-inoculated plants were significantly impaired in nitrogen fixation compared with parent strain CE3-inoculated plants. By 22 dpi, the ODB31-inoculated roots fixed nitrogen at the same level as the CE3-inoculated roots and maintained comparable levels at 31 dpi. Unexpectedly, plants inoculated with the ODB31 pagL complemented strain maintained a similar phenotype as the plants inoculated with the mutant strain ODB31 (data not shown).

DISCUSSION
The lipid A 3-O-deacylase, PagL, has been characterized in a number of Gram-negative animal pathogens (13) and has been implicated in the evasion of the host immune response and resistance to antimicrobial peptides (14,15,32). It was determined in vitro that the membranes of several nitrogen-fixing endosymbionts contained 3-O-deacylase activity able to remove the 3-O linked ␤-hydroxymyristoyl moiety from the lipid IV A precursor molecule (12). In addition, mass spectrometric evidence suggests that the endosymbionts R. etli CE3 (9) and S. fredii NGR 234 (33), when grown under laboratory conditions, produce 3-O-deacylated lipid A molecules. Furthermore, R. etli CE3 bacteroids exclusively produce tetraacylated lipid A, presumably because of 3-O-deacylation (9). Therefore, we hypothesized that R. etli CE3 produces a PagL enzyme that may play a role in symbiotic infection.   Fig. 3. Structure types D and E are those that result from PagL activity. Structure type C results from acid catalyzed ␤-elimination of the 3-O-acyl residue from the proximal GlcNonolactone during mild acid hydrolysis of the LPS, which converts that residue into 2,3-unsaturated GlcN(acyl)onolacton.

TABLE 2 Lipid A ions ([M-H] ؊ ) observed in the MALDI-TOF-MS spectra and their predicted compositions
In this study we describe the preparation of a mutant that lacks a functional pagL gene in R. etli bv. phaseoli CE3. The pagL Ϫ mutant does not produce tetraacylated lipid A caused by the action of an acylase, indicating that enzymatic de-O-acylation of lipid A has been disrupted. In addition, the mutant lipid A contained a larger amount of ␤-⌷⌯C15:0 than that of the parent strain. When the native pagL gene is introduced into the mutant, the enzymatically tetraacylated lipid A structure was recovered, and the amount of ␤-OHC15:0 decreased to parent strain levels. Based on these results, we conclude that the putative pagL gene described in this study is indeed responsible for the deacylation of lipid A and that when ␤-OHC15:0 is present it is located almost exclusively at position 3 of the lipid A, i.e., the target of PagL activity.
The P. aeruginosa PagL crystal structure has been solved (16), and a catalytic domain was proposed to contain a HXSX (nϩ18) E motif by which the histidine, serine, and glutamate residues form a classical serine esterase catalytic triad. The R. etli PagL contains the HXS motif, but instead of glutamic acid, a threonine was observed (supplemental Fig. S1). The threonine residue is conserved in PagL homologues from organisms found in the Rhizobiales order (supplemental Fig.  S1) and can theoretically perform a similar function as glutamate (i.e., stabilization of the deprotonated catalytic histidine by hydrogen bonding). In addition, the R. etli PagL contained the conserved phenylalanine hydrophobic residue (supplemental Fig. S1) previously implicated in positioning the lipid A molecule for hydrolysis by the P. aeroginosa PagL (13). However, a lipid A-binding motif has not been fully characterized for PagL enzymes. The substrate specificity for the R. etli PagL enzyme extends to E. coli lipid A (Fig. 4). E. coli and R. etli lipid A structures are quite different, and our results suggest that the R. etli PagL has a broad substrate specificity that may extend to a variety of Gram-negative lipid A structures.
The pagL Ϫ mutant formed abnormal symbiosis with the host plant Phaseolus (Figs. 5 and 6), indicating that disruption of pagL adversely influenced symbiotic infection. However, when complemented with the constitutively expressed native pagL gene via a plasmid vector, the complemented mutant displayed the same symbiotic phenotype as the mutant when inoculated on Phaseolus. We had thought perhaps this may be due to overexpression of PagL activity in the complement, possibly caused by the fact that pagL expression is under control of a lac promoter. However, a second complement prepared using a plasmid containing pagL under the control of its native promoter gave the sample result; namely two to three times greater PagL activity and a failure to complement the symbiotic phenotype. The common feature of both complements is this apparent overexpression of PagL activity, likely because of multiple copies of the complementing plasmid in both cases. Although the role of PagL in symbiosis remains to be determined, we suggest from these results that both over-and underexpression of pagL is deleterious to symbiosis; that is, we hypothesize that the regulation of PagL activity is important for this R. etli CE3-bean symbiosis. This hypothesis is consistent with the work of D'Haeze et al. (9), who reported the lipid A structure of R. etli CE3 bacteroids and demonstrated that these bacteroids produced exclusively tetraacylated lipid A (i.e., the PagL lipid A structure D), whereas free living bacteria produced a mixture of penta and tetraacylated lipid A. However, the mechanism of PagL regulation is unknown, and until this report, the pagL gene in R. etli was unknown. It is possible that the pagL gene is subject to transcriptional regulation via an as yet unidentified upstream promoter region. Another possibility is that the PagL protein is subject to post-translational regulation or substrate specific latency as is the case with the Salmonella PagL enzyme (15,35). In any case, the results by D'Haeze et al. strongly suggest that regulation of PagL is important for proper symbiotic infection, and therefore, overexpression (i.e., lack of regulation) in the complemented mutants (as indicated by the mass spectrometry results) may explain its inability to recover normal symbiotic infection. Although elucidating the regulatory mechanisms underlying PagL expression and/or enzyme activity is beyond the scope of this manuscript, it is important to devote attention to this area in future studies to fully understand the role PagL plays in nitrogen fixing endosymbiosis.
In mammalian systems, LPS is recognized by the Toll-like receptor 4 (TLR-4), which in turn induces a proinflammatory response (36,37). The TLR-4 ectodomain contains leucinerich repeats (LRRs), a common feature of the TLR receptor family responsible for the perception of a wide variety of conserved microbial associated molecular patterns. The modification or synthesis of unique lipid A structures has been implicated in the evasion of the animal LPS recognition system (13, 38 -40). Although it is unknown how LPS is recognized in plants, there are numerous transmembrane pattern recognition proteins containing an LRR ectodomain and an intracellular receptor-like kinase domain (LRR-RLK proteins) that are  . The ability of the pagL ؊ mutant to fix nitrogen during symbiotic infection. Dinitrogenase activity was assessed by the acetylene reduction assay at 15, 22, and 31 dpi. Displayed are the relative amounts of ethylene produced by roots inoculated with mutant strains ODB31 compared with roots inoculated with parent strain CE3. ***, extremely significant difference (p Ͻ 0.0001). ns, no significant difference. n ϭ 3 for each experiment.
It is important to note that the pagL gene is not present or expressed in all symbionts and is not absolutely required for complete symbiosis in these systems. Recent work by Brown et al. (6) showed that the bean nodulating strain R. leguminosarum bv. phaseoli 8002 does not produce deacylated lipid A under laboratory or symbiotic conditions and likely does not contain the pagL gene (the genome of strain 8002 is not published). The genome of the closely related organism R. leguminosarum bv. viciae 3841 has been published and does not contain the pagL gene (supplemental Fig. S2). However, strain 3841 contains remnants of the pagL gene, suggesting that these organisms may have lost the pagL gene possibly through the loss of selective pressure in their natural environment or in the laboratory. Another closely related clover-nodulating strain, R. leguminosarum bv. trifolii WSM 1325 does contain the pagL gene. It is not known whether or not strain WSM 1325 produces 3-O-deacylated lipid A. As stated above, S. meliloti contains both deacylated lipid A and the pagL gene. The absence of the pagL gene from some endosymbionts suggests that pagL is not absolutely required for complete symbiosis. However, the results presented in this study supports the conclusion that R. etli CE3 requires pagL expression for normal symbiosis (Fig. 5).