Studies on lipid A isolated from Phyllobacterium trifolii PETP02T lipopolysaccharide

The structure of lipid A from lipopolysaccharide of Phyllobacterium trifolii PETP02T, a nitrogen-fixing symbiotic bacterium, was studied. It was found that the lipid A backbone was composed of two 2,3-diamino-2,3-dideoxy-d-glucose (GlcpN3N) residues connected by a β-(1 → 6) glycosidic linkage, substituted by galacturonic acid (GalpA) at position C-1 and partly decorated by a phosphate residue at C-4′ of the non-reducing GlcpN3N. Both diaminosugars were symmetrically substituted by 3-hydroxy fatty acids (14:0(3-OH) and 16:0(3-OH)). Ester-linked secondary acyl residues [i.e. 19:0cyc and 28:0(27-OH) or 28:0(27-4:0(3-OMe))] were located in the distal part of lipid A. A high similarity between the lipid A of P. trifolii and Mesorhizobium was observed and discussed from the perspective of the genetic context of both genomes.

Phyllobacterium trifolii PETP02 T is the type strain of the species. It has been isolated from nodules of Trifolium pratense belonging to natural clover plants of north-west Spain. A comparison of the 16S rRNA gene sequence of P. trifolii indicated that it is closely related to the members of the genus Mesorhizobium. These bacteria can establish an effective symbiosis with Trifolium and Lupinus, plants that differ with regard to the nodule morphology. This property is rare among symbiotic bacteria. The P. trifolii nodD gene sequence shows high similarity (99%) to the homologous genes of Rhizobium etli CFN32 T and Ochrobactrum sp. LUP21, bacteria inducing a determinate type of root nodules on respective legumes, whereas nodules observed on Trifolium, the major macrosymbiont of P. trifolii, represent the indeterminate type (Valverde et al. 2005;Trujillo et al. 2005). The signal molecules (Nod factors) of Rhizobium leguminosarum bv. Trifolii and Bradyrhizobium, typical microsymbionts of clover and lupine, respectively, possess different fine structures. This evidence suggests that Trifolium and Lupinus do not select their microsymbionts strictly and that the host plant receptors could be unspecific (D'Haeze and Holsters 2002;Soulemanov et al. 2002;Schlaman et al. 2006;Renier et al. 2011). Like most Gramnegative bacteria, P. trifolii synthesizes lipopolysaccharide (LPS). The LPS molecules occupy at least 75% of the bacterial cell surface and are composed of three distinct domains: lipid A, core oligosaccharide, and O-specific polysaccharide (O-PS). The domains differ in terms of their structure and biosynthesis pathways (Silipo et al. 2010). Lipid A is a main constituent of the outer leaflet of the outer membrane (OM) and anchors entire LPS in OM through electrostatic and hydrophobic interactions, while the carbohydrate domains of LPS are oriented outwards (Raetz and Whitfield 2002). P. trifolii PETP02 T synthesizes mainly the smooth (S) form of LPS, and its O-PS structure has been described (Zamlynska et al. 2015). The strain synthesized two types of O-polysaccharides, containing hexa-and disaccharide repeats, respectively. The proper structure of the entire LPS determines appropriate integrity and flexibility of the outer membrane, essential for the correct morphology and functionality of bacteroids-the endosymbiotic form of rhizobia in which nitrogen fixation takes place. LPS delays or completely blocks the hypersensitivity reaction (HR) induced by rhizobia and suppresses systemic acquired resistance (SAR) during bacterial infection of root nodules (Dow et al. 2000;Albus et al. 2001;Menezes and Jared 2002;Mathis et al. 2005). For a long time, it was believed that lipid A was not important for development of an effective symbiosis. However, the data published during recent years have shown that mutants in genes encoding lipid A biosynthesis were more sensitive to pH and osmolarity changes, grew slowly, and revealed delayed nodule development. Also, the symbiotic effectiveness of these mutants, measured by nitrogen fixation abilities, was considerably reduced (Vedam et al. 2004;Ferguson et al. 2005;Hagg et al. 2009;Choma et al. 2017).
The structure of rhizobial lipids A differ considerably from the model one of Escherichia coli. The enterobacterial glucosamine-based hexa-acylated lipid A has endotoxic activity and is the most potent agonist of innate immunity in humans (Zähringer et al. 1994;Beutler and Rietschel 2003). Respective lipids A from Rhizobium and Sinorhizobium genera contain backbones based on glucosamine and 2-aminogluconate (Bhat et al. 1994;Que et al. 2000a, b;Kannenberg and Carlson 2001;Jeyaretnam et al. 2002;Gudlavalleti and Forsberg 2003;Ferguson et al. 2005Ferguson et al. , 2006, whereas the disaccharide backbone in other nodule-forming bacteria is composed exclusively of 2,3-diamino-2,3dideoxy-D-glucose (GlcpN3N) (Choma and Sowiński 2004;Komaniecka et al. 2010;Choma et al. 2012;Brown et al. 2013;Komaniecka et al. 2014;Silipo et al. 2014). Sugar backbones of rhizobial lipids A are substituted by amide-and ester-linked 3-hydroxy fatty residues, whose hydroxyl groups can be acylated by other non-polar and (x-1)-hydroxy/oxo very long chain fatty acid (VLCFA). For example, 27-hydroxyoctacosanoic acid is most often found in lipids A of members of alpha-rhizobia (except for the Azorhizobium genus) and is considered as chemotaxonomic marker of this group of bacteria (Bhat et al. 1991a, b). These very long chain fatty acids could be partially acylated by 3-hydroxybutyric or 3-methoxybutyric acids or by one or two hopanoid residues in some bradyrhizobial lipids A (Komaniecka et al. 2014;Silipo et al. 2014;Choma et al. 2017).
The structures of lipids A isolated from symbiotic bacteria classified outside of the Rhizobiaceae and Bradyrhizobiaceae families, such as that of the Phyllobacterium genus, have been studied sporadically, up to now. The aim of this article was to fill the gap and describe the structure of lipid A from P. trifolii PETP02 T in detail.

Bacterial culture conditions and lipopolysaccharide isolation
The P. trifolii PETP02 T strain was obtained as a kind gift from Dr. E. Velazquez, from the University of Salamanca, Spain. The bacteria were cultivated for 7 days in a liquid Tryptone Yeast medium at 28°C with aeration by vigorous shaking. The bacteria were harvested and the cell pellet was washed twice with saline. Bacterial mass was subjected to delipidation and enzymatic digestion procedures according to the method described in detail by Choma et al. (2012). The lipopolysaccharide was extracted from degraded cells using a hot 45% phenol/water method (Westphal and Jann 1965;Johnson and Perry 1976) and purified by ultracentrifugation, as described by Zamlynska et al. (2015).

Lipid A isolation and O-deacylation
Lipid A was cleaved from LPS by mild acid hydrolysis (aq 1% acetic acid, 100°C, 3 h). After cooling, the hydrolysate was converted to the two-phase Bligh-Dyer system and the chloroform phase containing the lipid A portion was separated by centrifugation (4000 9 g, 15 min, 20°C), collected, and washed with the water phase from a freshly prepared twophase Bligh-Dyer mixture (Que et al. 2000a;Choma et al. 2012). The lipid A preparation obtained was stored at -20°C in chloroform/methanol (2:1, v/v). Lipid A was O-deacylated by incubation of a 1 mg sample in chloroform/methanol/1 M aqueous NaOH, 2:3:1 (v/v/v), for 1 h, at room temperature, according to the method described previously (Que-Gewirth et al. 2004;Komaniecka et al. 2014). Partial lipid A Odeacylation was performed using 25% ammonium hydroxide (Lukasiewicz et al. 2010).

Chemical analyses of lipid A
The sugar composition was established by hydrolysis of lipid A with 4 M HCl (100°C, 4 h) and conversion of liberated monosaccharides into (amino)alditol acetates (Sawardeker et al. 1965). The absolute configuration of the monosaccharides was established by analysis of trimethylsilylated R-(-)-butyl glycosides according to a modified procedure developed by Gerwig et al. (1978). The fatty acid composition was established after methanolysis of lipid A (2 M HCl/ MeOH, 85°C, 18 h) and conversion of the obtained hydroxy fatty acid methyl esters into trimethylsilyl derivatives. 3-Hydroxy fatty acids were converted to L-phenylethylamides of 3-methoxy derivatives as described previously (Rietschel et al. 1976, Wollenweber et al. 1980) for analysis of the absolute configuration. Sugars and fatty acid derivatives were analysed using a gas chromatograph (Agilent Technologies, instrument 7890A) connected to a mass selective detector (Agilent Technologies MSD 5975C, inert XL EI/CI) (GLC-MS), using helium as a carrier gas. The chromatograph was equipped with a HP-5MS column (30 m 9 0.25 mm). The temperature program was as follows: 150°C for 5 min raised to 310°C (5°C min -1 ), and the final temperature was maintained for 10 min. L-Phenylethylamides of 3-methoxy fatty acid derivatives were also analysed isothermally (270°C) using the same column. Heptadecanoic acid and 3-hydroxy tetradecanoic acids were used as standards for quantitative determination of fatty acids.

NMR spectroscopy
The NMR spectra were recorded at 30°C using a Varian Inova 500 instrument and a standard Varian software. The sample was dissolved in a mixture of CDCl 3 /CD 3 OD (1:1, v/v) with a drop (5 ll) of D 2 O. 1D ( 1 H and 31 P NMR) and 2D NMR spectra ( 1 H, 1 H COSY, DQF-COSY, TOCSY, NOESY, ROESY) were recorded. Proton chemical shifts were measured in relation to TMS as an internal standard (d H 0.00). Phosphorous chemical shifts were measured relative to an external standard of 85% (v/v) phosphoric acid at d P 0.00 p.p.m.

MALDI-TOF MS and MS/MS spectrometry
Matrix-assisted laser desorption/ionization time-offlight mass spectrometry (MALDI-TOF-MS) was performed using a Waters SYNAPT G2-Si HDMS instrument (Waters Corporation, Milford, MA, USA) equipped with a 1 kHz Nd:YAG laser system. Acquisition of the data was performed using MassLynx software version 4.1 SCN916 (Waters Corporation, Wilmslow, United Kingdom). Spectra were recorded in positive and negative ion polarities. For MS/MS experiments, isolated precursor ions were fragmented using collision voltage of 60 V. Data were collected for 120 s for each ion separately. Mass spectra were assigned with a multi-point external calibration using red phosphorous (Sigma).

Bioinformatics tools
Standard BLASTP (with the cut-off E-value of 10 -5 ) was used in searching for putative proteins engaged in the biosynthetic pathway of Phyllobacterium lipid A. Mesorhizobium loti MAFF 303099 protein sequences were used as queries in BLASTP searches against five Phyllobacterium strains registered in the Genomes OnLine Database. Individual protein sequences were then compared across their entire span with an on-line Global Alignment tool (using the Needleman-Wunsch algorithm) provided by the National Center for Biotechnology Information (NCBI).
An aliquot of lipid A was analysed by negative ion mode MALDI-TOF spectrometry (Fig. 1). Lipid A showed significant heterogeneity due to the 3-OHfatty acid variability ( Table 2). The heaviest group of ions was found between m/z at 2354.6 and 2405.0 showing the presence of a hexaacylated lipid A species composed of two GlcpN3N residues, uronic acid, phosphate residue, 28:0(27-(4:0(3-OCH 3 )) acyloxyacyl residue, 19:0cyc, and 3-hydroxy-acyl chains with different length (D mass = 14 amu). Ions representing lipid A species devoid of phosphate (D mass = 80 amu) and 3-O-methoxybutyryl (D mass = 100 amu) were placed on the other side of the spectrum (signals between m/z at 2159.6 and 2229.7). The middle part of the spectrum is the most crowded region. This group of ions comprises lipid A species modified with phosphate but not with 3-O-methoxybutyryl and vice versa (3-O-methoxybutyryl-containing, but deprived of phosphate). A number of lipid A molecules bearing a 28:0(27-oxo) residue were found in this region. They were always decorated with phosphate. Negative and positive ions derived from the O-deacylated lipid A are depicted in Fig. 2a, b. Negative ions representing   Table 2). The difference in the mass of the intact and O-deacylated lipid A molecules confirmed the previously mentioned findings that both 28:0(27-OH/ oxo) and 19:0cyc constitute ester linked secondary residues. Although MALDI is recognized as a soft ionization technique, it is known that, with the desorption of intact molecules/ions, in-source fragmentation occurs. Very informative B ? type lipid A ions can be found in the positive ion mode spectra of intact as well as Odeacylated lipid A preparations (Fig. 3a, b and Table 2). They were helpful in determination of the fatty acid distribution. The ions at m/z 1340.160 and 1420.136 unequivocally indicated that distal GlcpN3N was partly modified with phosphate (D mass = 80 amu). In addition, asymmetric distribution (4 ? 2) of acyl residues on the disaccharide backbone of intact lipid A was deduced from the analysis of these ions. Thus, the ion at m/z 1340.160 can be interpreted as a tetra-acylated B 1 ? fragment bearing N-linked 16:0(3-OH) and 14:0(3-OH), and esterlinked lactobacillic and 27-oxooctacosanoic acids. The spectrum of O-deacylated lipid A included B 1 ? type ions as well. Ions at m/z 641.509 and 721.475 represented phosphorylated and unphosphorylated lipid A fragments composed of two fatty acids attached to GlcpN3N oxonium ions. Both these ions were accompanied by two mono-and doubly dehydrated product ions (Fig. 3b). Taking together the data from the analyses of intact and O-deacylated lipid A, it can be concluded that lipid A primary fatty acids were distributed symmetrically. MS/MS analyses were performed to assign the exact primary fatty acid positions at the reducing as well as non-reducing GlcpN3N. A set of ions was chosen from the Odeacylated P. trifolii lipid A MALDI TOF MS spectrum. The fragmentation pattern of the ion at m/ z 1498.12 was selected to illustrate our results (Fig. 4a,  b). This sodiated ion represents tetraacylated (exclusively with primary fatty acids) and devoid of phosphate species of P. trifolii lipid A (Table 2).          After loss of GalpA and water residues (176.04 and 18.01 amu, respectively), the precursor ion generated an ion at m/z 1304 containing a double bond between C-1 and C-2 of reducing GlcpN3N. The enamine to imine tautomerization of the compound resulted in elimination of aldehyde from the C-2 amide-bound fatty acid containing a free b-OH group (Choma et al. 2012;Silipo et al. 2014;Di Lorenzo et al. 2017). From two possibilities of GlcpN3N substitution with 3-OH fatty residues, only this one was represented and yielded a final ion at m/z 451.29 (Fig. 4a, b). The presence of this signal in the spectrum (and the absence of an ion peak at m/z 641.51) supported the placement of 16:0(3-OH) at C-2 and 14:0(3-OH) at C-3 of reducing as well as non-reducing GlcpN3N of the P. trifolii lipid A. The presence of other ions in the spectrum shown in Fig. 4b was explained by taking into consideration different fragmentation mechanisms as well (Fig. 4a). According to Silipo et al. (2014), an ion corresponding to the pentaacylated lipid A, containing a VLCFA as the only secondary residue, is the best for determining the VLCFA position in the lipid A molecule. Intact lipid A from P. trifolii contained many hexaacylated species but there were few pentaacylated molecules (Fig. 1). Therefore, we used 25% ammonium hydroxide to remove partly and randomly O-linked fatty acids from the lipid A. Following the methodology proposed by Silipo et al. (2014), an ion at m/z 2000.34 with the composition (GlcN3N) 2 GalA 1 P 1 [C14:0(3-OH)] 2 [C16:0(3-OH)] 2 [28:0(27-OH)] 1 Na (P: represents a phosphate residue) was selected. Two B-type ions (B 2 -type at m/z 1806.4 and B 1 at m/z 1165.9) were found (data not shown). Only the heavier one was accompanied by an ion originating from loss of ketene (D mass = 212.2 amu). These results allowed us to conclude that VLCFA (28:0(27-OH)) was linked to the hydroxyl of the primary fatty acid attached to position C-2 0 of the lipid A backbone. These results are in agreement with the hitherto published VLCFA containing lipid A structures (Choma et al. 2017).
MALDI-TOF analysis clearly indicated that P. trifolii lipid A was very similar to its M. huakuii counterpart. Moreover, with respect to their backbones, both lipids A were identical. Thus, we decided to reduce the scope of the NMR analysis to 1 H, 1 H homonuclear correlations sufficient to determine the anomeric configuration and the linkage positions within the lipid A backbone. The NMR data are listed in Table 3. The proton chemical shifts were assigned based on 1 H, 1 H DQF-COSY, TOCSY, and NOESY spectra. Three proton signals were identified in the anomeric region of 1 H NMR indicating that the P. trifolii lipid A backbone was built up of three sugar residues. Three spin systems: A, a-D-GalpA; B, a-D-GlcpN3N; and C, b-D-GlcpN3N were described as well. The sequence of monosaccharides was established in the NOESY experiment by observation of the following interresidue correlations: A-1/B-1 (d 5.19/5.04), C-1/B-6 (d 4.39/3.76), and C-1/B-6 0 (d 4.39/3.84). The small values of 3 J H-1,H-2 coupling constants (\3 Hz) for GalpA (A) and GlcpN3N (B) indicated an a-anomeric configuration of both residues. Thus, they were linked via an a-(1$1)glycosidic bond. Similar proton spectra were published for M. huakuii lipid A (Choma and Sowiński 2004).
The 31 P-NMR spectrum of the intact lipid A revealed a prominent signal with a chemical shift of 0.199 ppm observed in the solvent being a mixture of  Domon and Costello (1988) Monoisotopic masses and m/z values were calculated using a Molecular Weight Calculator, a part of the Waters SYNAPT G2-Si HDMS instrument software Antonie van Leeuwenhoek (2017Leeuwenhoek ( ) 110:1413Leeuwenhoek ( -1433Leeuwenhoek ( 1423 chloroform and methanol (proportion 2:1, v:v, respectively). These properties were indicative of the presence of phosphomonoester. Based on all collected data, the following structure for lipid A being the lipophilic part of P. trifolii lipopolysaccharide was proposed (Fig. 5).

Genetic background of lipid A biosynthesis pathway in Phyllobacterium species
In order to validate the experimental data related to the structure of P. trifolii PETP02 T lipid A obtained in MALDI-TOF mass spectrometry and NMR  Table 2 Antonie van Leeuwenhoek (2017)  CL33Tsu (GOLD Project ID: Gp0115127). All five Phyllobacterium strains were isolated, cultured, and sequenced as part of three independent plant-microbe associated projects. Therefore, we used these genomic resources in searching for candidate genes of the P. trifolii lipid A biosynthetic pathway. A comparative sequence analysis of the 16S rRNA gene indicated that P. trifolii is closely related to the members of the genus Mesorhizobium (Valverde et al. 2005). Our results have confirmed that the structure of P. trifolii lipid A is indeed similar (the backbone structure is even identical) to some mesorhizobial lipids A, i.e. M. huakuii IFO 15243 T and M. loti MAFF 303099 (Choma and Sowiński 2004;Brown et al. 2013). Thus, protein sequences of model rhizobial strain M. loti MAFF 303099 were used as queries in BLAST sequence similarity searching the draft genome sequences of the phyllobacteria, and then respective protein sequences were compared across their entire length with Needleman-Wunsch Global Align. Using this approach, we were able to recognize a set of putative genes coding for common enzymes required for the biosynthesis of lipid A (lpxA, lpxC, lpxD, lpxH, lpxB, lpxK, and kdtA (waaA)) as well as  5 Proposed structure of the phosphorylated P. trifolii PETP02 T lipid A species decorated with: R -3-methoxy-butyric residue, hydroxy-, or oxo-group, and containing most abundant 14:0(3-OH) and 16:0(3-OH) primary fatty acids encoding specific enzymes involved in the structural modifications of lipid A found in some Gram-negative bacteria (lpxE, rgtF, and acpXL-lpxXL). Moreover, genes coding for enzymes converting GlcpN to GlcpN3N, necessary for biosynthesis of the lipid A disaccharide backbone and characteristic for bacteria from the genera Mesorhizobium, Azorhizobium and Bradyrhizobium (gnnA and gnnB), were identified (Table 4). Putative ORFs of the tested phyllobacterial species shared significant sequence similarity with the respective M. loti MAFF 303099 proteins (Table 4), strongly suggesting their common evolutionary ancestry-homology. We were not able to predict the putative homolog of a-(1$1)-GalpA transferase (rgtF gene) in the genome of Phyllobacterium sp. YR531. We found ORFs coding for putative 4 0 -phosphatase in the genomes of four Phyllobacterium strains (Ga0115491_102664, BR48DRAFT_1751, Ga0115492_1841, and Ga0073264_1478 in Phyllobacterium spp. OV277, UNC302MFCol5.2, YR620, and CL33Tsu, respectively), sharing 50-51% sequence similarity with the LpxF protein of R. leguminosarum bv. viciae 3841 (RL_RS08140), but we cannot identify homologs of RgtD, a putative 4 0 -GalA transferase of R. leguminosarum bv. viciae 3841 (RL_RS03600). Both LpxF and RgtD are involved in transfer of the GalpA residue to lipid A during biosynthesis of LPS in R. leguminosarum bv. viciae 3841 (Brown et al. 2013).
The putative genes described display substantial conservation of organization (synteny and collinearity) among the tested phyllobacteria and in comparison to M. loti MAFF 303099 (Table 5). This was especially apparent with lpxD, lpxA, and lpxB, as well as kdtA and lpxK gene clusters, and the highly conserved acpXL-lpxXL region, typical for bacteria possessing lipid A molecules modified with VLCFAs (Choma et al. 2017). Similar to M. loti, the phyllobacterial lpxC and lpxH were not clustered with other genes related to LPS biosynthesis (Table 5).
There were noticeable differences found in the location of putative genes responsible for the transamination reaction of GlcpN (gnnA and gnnB), and structural modifications of GlcpN3N (lpxE and rgtF). In the case of phyllobacteria, orthologs of gnnA and gnnB are located close to each other but separately from the lpxA, lpxB, lpxD gene cluster, similar to Bradyrhizobium japonicum and Brucella melitensis (Sweet et al. 2004). In contrast, in M. loti MAFF 303099, the putative orthologs of these genes are separated both from each other and from the lpxA, lpxB, lpxD cluster. The rgtF and lpxE genes of M. loti MAFF303099 were clustered together with a putative rgtE gene (encoding putative bactoprenyl-phosphate GalpA transferase) located upstream rgtF. Brown et al. (2013) suggested a functional relationship of these genes in the biosynthesis of lipid A a-(1$1)-GalA. Interestingly, in all the tested phyllobacterial species, the putative rgtE and rgtF orthologs were neighboured, while the putative homolog of lpxE was distantly located from the above-mentioned ORFs.

Discussion
In this study, we have described the structure of P. trifolii PETP02 T lipid A, which contains a trisaccharide carbohydrate backbone. This backbone comprises two D-GlcpN3N connected by a b-(1 ? 6) glycosidic linkage and a D-GalpA residue at position C-1. The substitution of the reducing end of lipid A by a-(1$1)-D-GalpA is unusual among bacteria and has been only described in lipids A from a few representatives of Gram-negative bacteria, including an associative diazotroph-Azospirillum lipoferum (Choma and Komaniecka 2008), and symbiotic bacteria-M. huakuii (Choma and Sowiński 2004), M. loti (Brown et al. 2013), a stalk-forming Caulobacter crescentus (Smit et al. 2008), and a thermophilic bacterium Aquifex pyrophilus (Plötz et al. 2000). The GlcpN3N disaccharide of C. crescentus and A. pyrophilus is bisgalacturonosylated with GalpA located at positions C-1 and C-4 0 (Plötz et al. 2000;Smit et al. 2008), whereas the A. lipoferum backbone does not possess the substituent at the reducing end of GlcpN (Choma and Komaniecka 2008). The P. trifolii lipid A trisaccharide backbone is partially phosphorylated at position C-4 0 and has the same structure as that of M. huakuii IFO15243 T (Choma and Sowiński 2004). One can speculate that this may be related to the lack of an rgtD homolog in the genomes of the tested phyllobacterial species. The presence of phosphate, GlcpN3N, and GalpA has also been reported in lipid A from R. loti, but with a-(1?4 0 )-GalpA, not a-(1$1)-GalpA (Russa et al. 1995). The amino groups of both GlcpN3N of P. trifolii lipid A are symmetrically substituted by 3-hydroxy fatty acids, among which 14:0(3-OH) and 16:0(3-OH) dominate. Moreover, MS/MS data supported the placement of 16:0(3-OH) at C-2 and 14:0(3-OH) at C-3 positions of reducing as well as non-reducing GlcpN3N of the lipid A. Although the backbones of lipids A synthesized by phyllobacteria and mesorhizobia are identical, the substituting primary fatty acids are significantly different. This fact can be explained by the different basic metabolisms of both genera of bacteria. While the Mesorhizobium species synthesize and incorporate branched chain fatty acids into lipids (including lipids A), Phyllobacterium spp. produce predominantly straight chain fatty acids for cellular purposes (Tighe et al. 2000;Choma and Sowiński 2004;Valverde et al. 2005). As can be seen, the biosynthetic pathway of VLCFAs is not subject to this rule. Mesorhizobium as well as Phyllobacterium produces the same VLCFAs.
Because of the distribution of ester-linked fatty acids (19:0cyc and 28:0-(27-OH/oxo)) restricted to the distal GlcpN3N, the entire lipid A is asymmetrically acylated, resembling E. coli and mesorhizobial lipids A (Raetz and Whitfield 2002;Choma and Sowiński 2004;Brown et al. 2013). This pattern of fatty acid distribution can be described by the formula 4 ? 2.
Hydroxyl groups at b positions of the primary fatty acids were additionally acylated by VLCFA and lactobacillic acid. The 27-hydroxyoctacosanoic acid could be partially acylated with 3-methoxybutyric acids. Lactobacillic acid is frequently found as a constituent of bacterial phospholipids. Especially phospholipids extracted from bacteria harvested in the stationary phase of growth are rich in this fatty acid as well as other cyclopropane fatty acids. To the best of our knowledge, cyclopropane fatty acids have not been found yet to be a component of LPS/lipid A. It can be assumed that, similar to phospholipids (Gronan and Cronan 1997), LPS containing unsaturated 18:1x 7 (vaccenic) acyl residue in the lipid A moiety undergoes modification in the bacterial outer membrane. This issue also remains to be resolved in future studies. As mentioned above, we have found structural resemblance of lipid A of P. trifolii and Mesorhizobium and this observation seems to be reflected at the genomic level. Putative ORFs predicted for LPS biosynthesis in the phyllobacteria shared significant sequence similarity and overall similar gene organization with M. loti, with only minor differences. Further genetic analyses are required to answer the question how these differences affect the structure of P. trifolii lipopolysaccharide.  Table   b % identity (% similarity), refers to the entire Table  Table 5 Comparison of the genetic organization of genes engaged in the biosynthesis of lipid A of Mesorhizobium loti MAFF303099 and Phyllobacterium sp. OV277 The Phyllobacterium sp. OV277 was arbitrarily chosen as a representative example of the genetic organization of respective regions, which was highly similar among the tested Phyllobacterium strains