Phosphatidylcholine biosynthesis and function in ...

Biochimica et Biophysica Acta 1831 (2013) 503–513

Contents lists available at SciVerse ScienceDirect

Biochimica et Biophysica Acta journal homepage: www.elsevier.com/locate/bbalip

Review

Phosphatidylcholine biosynthesis and function in bacteria☆ Otto Geiger ⁎, Isabel M. López-Lara, Christian Sohlenkamp Centro de Ciencias Genómicas, Universidad Nacional Autónoma de México, Av. Universidad s/n, Apdo. Postal 565-A, Cuernavaca, Morelos, CP62210, Mexico

a r t i c l e

i n f o

Article history: Received 28 June 2012 Received in revised form 10 August 2012 Accepted 13 August 2012 Available online 19 August 2012 Keywords: Membrane lipid biosynthesis Bacterial phosphatidylcholine Phosphatidylcholine synthase Phospholipid N-methyltransferase

a b s t r a c t Phosphatidylcholine (PC) is the major membrane-forming phospholipid in eukaryotes and is estimated to be present in about 15% of the domain Bacteria. Usually, PC can be synthesized in bacteria by either of two pathways, the phospholipid N-methylation (Pmt) pathway or the phosphatidylcholine synthase (Pcs) pathway. The three subsequent enzymatic methylations of phosphatidylethanolamine are performed by a single phospholipid N-methyltransferase in some bacteria whereas other bacteria possess multiple phospholipid N-methyltransferases each one performing one or several distinct methylation steps. Phosphatidylcholine synthase condenses choline directly with CDP-diacylglycerol to form CMP and PC. Like in eukaryotes, bacterial PC also functions as a biosynthetic intermediate during the formation of other biomolecules such as choline, diacylglycerol, or diacylglycerol-based phosphorus-free membrane lipids. Bacterial PC may serve as a specific recognition molecule but it affects the physicochemical properties of bacterial membranes as well. This article is part of a Special Issue entitled Phospholipids and Phospholipid Metabolism. © 2012 Elsevier B.V. All rights reserved.

1. Introduction Phosphatidylcholine (PC) is the major membrane-forming phospholipid in eukaryotes and can be synthesized by either of two pathways, the methylation pathway or the CDP-choline pathway [1]. Bacteria such as the well-studied Escherichia coli or Bacillus subtilis have long served as model organisms for Gram-negative and Grampositive bacteria, respectively, and both have phosphatidylethanolamine (PE), phosphatidylglycerol (PG) and cardiolipin (CL) as major membrane-forming phospholipids [2]. Both “model bacteria”, E. coli and B. subtilis, are devoid of PC and other membrane phospholipids traditionally regarded as typically eukaryotic such as sphingolipids [3], phosphatidylinositol (PI), or methylated derivatives of PE such as monomethyl-PE (MMPE), and dimethyl-PE (DMPE), that occur in some bacteria [4]. For a long time it was believed that PC occurs only in a few specialized bacteria, such as photosynthetic bacteria containing extensive internal membrane structures or bacteria living in association with eukaryotes [5,6]. We previously estimated than more than 10% of

Abbreviations: CL, cardiolipin; CPT, choline phosphotransferase; DAG, diacylglycerol; DMPE, dimethylphosphatidylethanolamine; MMPE, monomethylphosphatidylethanolamine; PAF, platelet-activating factor; PC, phosphatidylcholine; Pcs, phosphatidylcholine synthase; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; PI, phosphatidylinositol; Pmt, phospholipid N-methyltransferase; PS, phosphatidylserine; SAM, S-adenosylmethionine; SAH, S-adenosylhomocysteine ☆ This article is part of a Special Issue entitled Phospholipids and Phospholipid Metabolism. ⁎ Corresponding author. Tel.: +52 777 3290815; fax: +52 777 3175581. E-mail address: [email protected] (O. Geiger). 1388-1981/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.bbalip.2012.08.009

the all bacteria possess PC as a membrane lipid [4] and we support this by our present estimate by which at least 15% of the bacteria have the ability to synthetize PC. The relative amount of PC detected in different bacterial species varies widely and ranges from a few percent of the total membrane lipid (0-4% in Pseudomonas aeruginosa) to up to 73% in Acetobacter aceti [4,7]. Although individual biosynthetic steps of the CDP-choline pathway for PC synthesis exist in some bacteria, catalyzed by the licAC-encoded products that permit the formation of phosphocholine or CDP-choline [4], the only bacterial example for which a complete CDP-choline pathway for PC synthesis has been proposed is the case of the spirochaete Treponema denticola [8]. In more general terms, the two different pathways for PC biosynthesis occurring more frequently in bacteria are the phospholipid N-methylation (Pmt) pathway and the phosphatidylcholine synthase (Pcs) pathway. In the N-methylation pathway, phosphatidylethanolamine is methylated three times to yield PC involving one or more phospholipid N-methyltransferases [4,9], whereas in the PC synthase pathway, choline condenses directly with CDP-diacylglycerol to form PC and CMP [10,11]. Several mutants of different PC-containing bacteria have been generated that are devoid of detectable PC. However, the extent to which the absence of PC affects other observable phenotypes depends on the respective bacterial system and may be severe, as in the case of Sinorhizobium meliloti [12], to hardly detectable, as in the case of P. aeruginosa [13]. Examples are presented where bacterial PC functions as a biosynthetic intermediate during the formation of other biomolecules and where bacterial PC may serve as a specific recognition molecule. We discuss how bacterial PC may affect the physicochemical properties of bacterial membranes.

504

O. Geiger et al. / Biochimica et Biophysica Acta 1831 (2013) 503–513

Fig. 1. Model of glycerophospholipid biosynthesis in Sinorhizobium meliloti (for details see text).


505

2. Biosynthesis of phosphatidylcholine in bacteria

2.2. Phosphatidylcholine synthases

Although glycerophospholipid biosynthesis in bacteria can be more complicated [14] than shown here, we will initially focus on glycerophospholipid biosynthesis in S. meliloti (Fig. 1) which has emerged as a bacterial model organism for the study of phosphatidylcholine biosynthesis and function.

PC synthase condenses choline directly with CDP-diacylglycerol (CDP-DAG) to form PC and CMP [10,11]. PC synthase, phosphatidylinositol synthase, phosphatidylglycerolphosphate synthase, phosphatidylserine synthase type II, cardiolipin synthase type II, aminoalcohol phosphotransferase and choline phosphotransferase all belong to the CDP-alcohol phosphotransferase protein superfamily [4,15]. Most members of this superfamily share the amino acid motif DGX2ARX8GX3DX3D, termed the CDP-alcohol phosphotransferase motif. However, in PC synthases a modified version of this motif is present: DGX2ARX12GX3DX3D [4,11]. Sinorhizobial phosphatidylcholine synthase is an integral membrane protein which works optimally at pH 8.0 and requires the presence of bivalent cations such as Mg 2 + or Mn 2 + for its function, with manganese being much more efficient (more than 20-fold higher activity) than magnesium and with 10 mM Mn 2 + maximal Pcs activity was reached [10]. In cell-free assays, the presence of the detergent Triton X-100 is required in order to detect Pcs activity and 0.2% (w/v) Triton X-100 gave maximal Pcs activity [10]. In an extensive site-directed mutagenesis study, several conserved amino acid residues (H20, T23, D56, D59, G60, D81, D85) of the sinorhizobial Pcs were identified as being important for activity when the enzyme was expressed in an E. coli host [19]. D56 is absolutely essential for Pcs activity and we suggest to include this residue in the future motif for Pcs enzymes: DX2DGX2ARX12GX3DX3D. A topological analysis of the sinorhizobial Pcs shows the presence of eight transmembrane helices, with the N- and C-terminus located in the cytoplasm [19]. Although these are valuable tools for future studies, presently, the substrate binding sites, the enzymatic mechanism, or a detailed structure of Pcs are not known. Members of the CDP-alcohol phosphotransferase superfamily seem to follow a sequential reaction mechanism. For example, phosphatidylinositol synthase follows an ordered Bi-Bi mechanism in which CDP-DAG binds before inositol and phosphatidylinositol (PI) is released prior to CMP in the reaction sequence [20]. Also in the case of phosphatidylserine synthase type II from Bacillus licheniformis, radioisotopic exchange patterns between related substrate and product pairs suggest a sequential Bi-Bi reaction [21] as opposed to the Ping-Pong mechanism exhibited by the well-studied phosphatidylserine synthase of E. coli which belongs to a different protein superfamily [22]. Therefore, we would expect that Pcs, like other CDP-alcohol phosphotransferases, follows a sequential Bi-Bi reaction in which CDP-DAG binds before choline and PC is released prior to CMP. The discovery of Pcs [10,11] also has opened the door for new experimental approaches. For example, expression of the legionellal pcs gene causes PE-independent formation of PC in E. coli [23]. Using this tool in a PE-deficient mutant, these authors studied to which extent PC can replace PE in the proper folding of the membrane protein lactose permease (LacY). E. coli-derived PC and synthetic PC species containing at least one saturated fatty acid also support the native conformation of LacY. Apparently, for proper LacY folding not only the head groups contribute but also acyl residues that affect the physicochemical properties of bacterial membranes [23].

2.1. Phosphatidylcholine biosynthesis in Sinorhizobium meliloti Rhizobia are soil bacteria able to form a symbiosis with legume plants, which leads to the formation of nitrogen-fixing root nodules. S. meliloti can establish such a symbiosis with alfalfa (Medicago sativa). As in most other bacteria, phospholipid biosynthesis in S. meliloti initiates with sn-1 and sn-2 acylation of glycerol-3-phosphate, leading to the formation of phosphatidic acid [2] (Fig. 1). The conversion of phosphatidic acid to CDP-diacylglycerol is catalyzed by CDPdiacylglycerol synthase (CdsA). In S. meliloti two genes (cdsA and cdsA2) are assigned to encode for CDP-diacylglycerol synthase and to date it is not clear whether both are functional. For the biosynthesis of the anionic phospholipids phosphatidylglycerol (PG) and cardiolipin (CL), first phosphatidylglycerolphosphate synthase (PgsA) transfers glycerol-3-phosphate to CDP-diacylglycerol under the release of CMP thereby producing phosphatidylglycerolphosphate (PGP). PGP phosphatase releases inorganic phosphate from PGP to form PG. However, so far no candidate gene for PGP phosphatase activity could be identified in S. meliloti. In S. meliloti and probably most other bacteria, a cardiolipin synthase (ClsB) of the bacterial type condenses two PG molecules to yield CL and free glycerol in a transesterification reaction. This is worth emphasizing as it has been shown recently that some bacteria possess a cardiolipin synthase (ClsE) of the eukaryotic type which condenses CDP-diacylglycerol with PG to form CL and CMP [15]. The first step in the synthesis of phosphatidylethanolamine (PE) is the condensation of CDP-diacylglycerol with serine to form phosphatidylserine (PS) catalyzed by PS synthase (PssA). In a second step, the decarboxylation of PS is catalyzed by PS decarboxylase (Psd) to yield PE (Fig. 1). Both steps have been studied in S. meliloti to some detail [16,17]. A well-known pathway for PC formation occurs by three-fold methylation of PE using S-adenosylmethionine (SAM) as methyl donor and this is catalyzed by phospholipid N-methyltransferase (PmtA). In S. meliloti and many other bacteria, all three subsequent methylations are achieved by a single enzyme [12] (Fig. 1). Many PC-containing bacteria have a second pathway for PC formation, catalyzed by PC synthase (Pcs), in which choline is condensed directly to CDP-diacylglycerol forming PC and CMP [4,10,11,18] (Fig. 1). To date, Pcs has been found only in bacteria. S. meliloti serves as a model organism to study the function of the zwitterionic phospholipids PE and PC in bacteria. Mutants exist in all steps of zwitterionic phospholipid formation (Fig. 1): 1. Mutant CS111 is deficient of PssA, forms no PE but does form PC when grown on complex medium [16]. 2. Mutant MAV01 is deficient of phosphatidylserine decarboxylase, forms no PE but PC and much increased amounts of PS [17]. 3. Individual mutants deficient in PmtA (KDR516) [12] or Pcs (KDR568) [11] as well as a PmtA- and Pcs-deficient double mutant (OG10017) [12] have been generated and whereas the individual mutants produce near normal amounts of PC, the PmtA- and Pcs-deficient double mutant is totally devoid of PC [12]. Transcriptomic and proteomic studies of S. meliloti wild type and the distinct mutants are under way to understand the physiological implications associated with a lack of a distinct zwitterionic membrane phospholipid (PE or PC) or those associated with an accumulation of PS.

2.3. Phospholipid N-methyltransferases Initial studies of PC biosynthesis in bacteria were carried out in the tumor-inducing plant pathogen Agrobacterium tumefaciens by Law et al. [24,25]. They demonstrated a methyltransferase activity in cell-free extracts of A. tumefaciens which could carry out all three methylations to convert PE to PC using the methyl donor SAM (Fig. 1). Later Arondel et al. [26] identified the first bacterial gene for a Pmt (pmtA) in Rhodobacter sphaeroides and rhodobacterial mutants deficient in PmtA

506


did not form PC when grown on a defined minimal medium but showed no other obvious phenotype. S. meliloti also possesses a single gene for a Pmt (pmtA) capable of performing all three methylations to convert PE to PC [12] but sinorhizobial PmtA shows surprisingly little similarity to the corresponding enzyme from R. sphaeroides. Based on these findings it was proposed that at least two bacterial Pmt families exist, one family similar to the PmtA of S. meliloti (Sm-PmtA) and another family similar to the PmtA of R. sphaeroides (Rs-PmtA) [4]. A third family might be represented by the membrane-associated PmtA from Zymomonas mobilis [27]. Although the genomes of several Z. mobilis strains have been sequenced, it is not clear which gene(s) might encode for the Pmt activity. In the genome of Z. mobilis ZM4 the gene ZMO0776 (YP_162511) has been assigned as phospholipid N-methyltransferase but to date experimental evidence is missing that it encodes a Pmt. 2.3.1. Phospholipid N-methyltransferase PmtA from Agrobacterium tumefaciens The Atu300-encoded agrobacterial PmtA (At-PmtA) is a close relative to the PmtA of S. meliloti (Sm-PmtA). A His-tagged version of the At-PmtA has been expressed and purified to near homogeneity [28]. Gel filtration experiments suggest that the AtPmtA exists as a monomer in solution. The purified enzyme catalyzes all three methylation steps converting PE to MMPE, DMPE, and finally to PC. Besides PE, the enzyme can use DMPE, and to a lesser extent MMPE, as externally added substrates. Purified At-PmtA binds to PE, MMPE, DMPE, PC, and surprisingly even more strongly to the negatively charged phospholipids PG and PI both of which are not substrates for PmtA. The presence of PG stimulates the PmtA activity whereas PI does not affect the activity. Binding of the second substrate S-adenosylmethionine (SAM) to the PmtA occurs only in the presence of PE, MMPE, DMPE, or PC, not however, if PG is the only lipid present. From these data one would expect distinct binding sites on the PmtA enzyme for the zwitterionic phospholipid substrates/products and for PG. The first product of the PmtA reaction, S-adenosylhomocysteine (SAH) and the substrate analog sinefungin interfere with SAM binding to PmtA and inhibit the PmtA enzyme activity. The At-PmtA might follow an ordered Bi-Bi mechanism in which the zwitterionic phospholipid substrate (PE, MMPE, or DMPE) binds first causing a structural change of the PmtA in order to allow binding of the second substrate SAM. Subsequently, the first product, SAH might be released followed by the zwitterionic phospholipid product containing one more methyl group than the substrate employed [28]. Using alanine scanning mutagenesis, several conserved amino acid residues (E58, G60, G62, and E84) of the At-PmtA were identified as being important for activity and SAM binding to the enzyme [29]. To date the binding sites for PG, the zwitterionic phospholipid substrates or amino acid residues forming the active site of PmtA are not known. 2.3.2. Multiple phospholipid N-methyltransferases from Bradyrhizobium japonicum In Bradyrhizobium japonicum USDA110 a homologue (Bj-PmtA) of the sinorhizobial PmtA (Sm-PmtA) family was initially discovered [30]. Upon expression of the bradyrhizobial pmtA gene in E. coli, predominantly MMPE was formed from PE which led us to postulate the existence of an additional gene (pmtX) in the B. japonicum genome coding for phospholipid N-methyltransferases able to perform the second and third methylation efficiently. Searches of the B. japonicum genome identified two candidate genes, pmtX1 and pmtX2, that belong to the rhodobacterial PmtA family and two other candidate genes, pmtX3 and pmtX4, that belong, like Bj-PmtA, to the sinorhizobial PmtA family [9]. Expression of the four pmtX candidate genes in E. coli and their coexpression with Bj-pmtA suggest that PmtX1 does not perform the first methylation but can convert MMPE and DMPE to PC, that PmtX3 can perform the first and the second methylation efficiently leading to MMPE and DMPE but not the third one, and that PmtX4 performs efficiently the first methylation (Fig. 2). From the studies performed so far [9], it is not clear to what extent PmtX2 contributes to phospholipid

headgroup methylation. Studies with transcriptional fusions suggest that in wild type B. japonicum besides pcs only pmtA and pmtX1 are significantly expressed. In the pmtA-deficient mutant a much increased expression of pmtX4 is detected [9]. Microarray studies support a slight increase of pmtX3 and a stronger increase of the pmtX4 transcript in the pmtA-deficient mutant with respect to the wild type [31]. 2.4. CDP-choline pathway for phosphatidylcholine biosynthesis in Treponema Bacteria, such as pathogenic and commensal Neisseria, P. aeruginosa, Haemophilus influenzae, or several Streptococcus species, colonizing the respiratory tract or other epithelia frequently display phosphocholine modifications of their surface structures [4]. Usually, these bacteria possess a lic-encoded pathway (Fig. 3) for choline metabolism which is thought to consist of a choline transport system (LicB), a choline kinase (LicA), a CTP:phosphocholine cytidylyltransferase (LicC), and a choline phosphotransferase (LicD). LicD homologues would encode for enzymes catalyzing the transfer of phosphorylcholine from CDP-choline to cell surface structures like lipopolysacharides in the case of H. influenzae and commensal Neisseria, to pili in the case of Neisseria meningitides, and to lipoteichoic acids and teichoic acids in S. pneumoniae [4]. PC is a major phospholipid in Treponema denticola, accounting for 35-40% of total phospholipid [8]. Altough no Pcs homologue can be encountered in the genome of T. denticola, radiolabeled choline was readily incorporated into PC, indicating the presence of a choline-dependent biosynthetic pathway. T. denticola has a licCA gene that encodes a fusion protein with choline kinase and CTP:phosphocholine cytidylyltransferase activity. A partial licCA deletion mutant of T. denticola is viable, does not form any PC but instead has increased amounts of PE [8]. Using the protein

Fig. 2. PC biosynthesis in Bradyrhizobium japonicum. This model is based on phospholipid profiles obtained in E. coli after expression of the individual potential Pmt-encoding genes and after respective coexpression with the bradyrhizobial pmtA gene [9]. Thick arrows and boldface letters indicate the predominant reaction(s) performed by each enzyme. Enzymes marked in brackets (green) are not expressed in B. japonicum wild type.


507

Acidibacillaceae, and the Chromatiales), Pmt might be the only means to synthesize PC. From the δ-proteobacteria, only Desulfohalobium has a Pcs candidate. PC biosynthesis genes are not limited to the proteobacteria but do occur in other bacterial phyla as well (Supplementary Table 1). For example, some members of the Gram-positive Actinomycetales possess a Pcs homologue together with LicCA homologues and Actinomyces viscosus is known to have PC as a membrane lipid [33]. Within the spirochaetes, most members of the genus Borrelia possess a Pcs and the one of Borrelia burgdorferi was shown to be functional [34,35]. Pcs also occurs with some members of the Planctomycetales, in Salinibacter ruber (Bacteroidetes), Rubrobacter xylanophilus (Rubrobacterales), and Truepera radiovictrix (Deinococcales). Some Planctomycetales harbor also genes for Pmt candidates (Supplementary Table 1). Of the roughly 3000 complete and uncomplete bacterial genomes analyzed, about 450 contain PC biosynthesis genes (pcs or pmtA) which confirm our previous estimate [4] that about 15% of the bacteria are able to synthesize PC. A phylogenetic tree including selected Pcs and PssA sequences shows that both groups of enzymes are clearly separated (Fig. 4) and that therefore the predictive value for Pcs enzymes seems to be quite good, especially when taking as additional criterium the motif (DX2DGX2ARX12GX3DX3D) defined for Pcs enzymes [19]. The one exception is the Pcs from Pseudomonas putida, which shows the Pcs motif and for which recently Pcs activity has been shown [36] but which still clearly groups more closely to the PssA enzymes (Fig. 4). 3. Functions of bacterial phosphatidylcholine 3.1. Role of phosphatidylcholine in different bacterial systems Fig. 3. CDP-choline pathway in Treponema. Choline kinase activity is encoded by licA and CTP:phosphocholine cytidylyl transferase activity by licC. In Treponema, a CDPcholine:1,2-diacylglycerol choline phosphotransferase (CPT) activity, probably encoded by the gene TDE0021, might perform the last step of PC biosynthesis via the CDP-choline pathway. The gene licD occurs in many pathogens colonizing epithelia and is predicted to encode a phosphocholine transferase activity transferring phosphocholine from CDP-choline to cell envelope structures like lipopolysaccharides (LPS) in H. influenza or teichoic and lipoteichoic acids in S. pneumoniae.

sequence of CDP-choline:1,2-diacylglycerol choline phosphotransferase from Saccharomyces cerevisiae (CPT1) as query, we identified the gene TDE0021 in the genome of T. denticola ATCC35405 which encodes a CPT homologue that we suggest might perform the last step of PC biosynthesis in this organism. Thus it appears that T. denticola has a complete CDP-choline pathway for PC biosynthesis (Fig. 3) and it remains to be seen whether more bacteria have this pathway. Within the bacteria, CPT1 homologues are restricted to some members of the genus Treponema and are encountered in genomes of T. denticola, T. phagedenis, T. vincentii, P. paraluiscuniculi, and T. pallidum. 2.5. Distribution of phosphatidylcholine biosynthesis genes within bacteria Our present analysis of the published bacterial genome sequences (Supplementary Table 1) reveals that genes encoding for enzymes involved in PC biosynthesis are most abundant within the proteobacteria. In the α-proteobacteria, putative Pcs enzymes are encountered in the classes Rhizobiales (except the Methylobacteriaceae), the Rhodobacterales and the Rhodospirillaceae. The appearance of Pmt candidates in bacteria seems to be much more sporadic and within the α-proteobacteria there maybe none, as in Bartonella, one, as in S. meliloti, or multiple Pmts, as up to five in B. japonicum. Pcs is absent in the Sphingomonadaceae, the Acetobacteraceae and in all β-proteobacteria, however, some bacteria within these groups do have one or several Pmt candidates instead. Within the γ-proteobacteria, Pcs is found in most Pseudomonas species, in the genera Halomonas, Legionella, and Francisella. In Pseudomonas and Francisella, Pmts seem to be absent whereas in Legionella it is known that Pmt contributes less to PC biosynthesis than Pcs [32]. Also within some γ-proteobacteria (Rhodanobacter, Methylococcus, members of the

3.1.1. A phosphatidylcholine-deficient mutant of Sinorhizobium meliloti has pleiotropic phenotypes When PmtA-deficient mutants of S. meliloti are grown on minimal media they cannot form PC and they grow significantly slower than the wild type [12]. Growth of the PmtA-deficient mutant in the presence of choline allows for PC formation via the Pcs pathway and restores wild type-like growth. Double knock-out mutants, deficient in Pmt and in Pcs, are unable to form PC and show reduced growth even in the presence of choline [12]. These results suggest that PC is required for normal growth of S. meliloti [12]. PC-deficient mutants of S. meliloti are also unable to form a nitrogen-fixing symbiosis with their host plant alfalfa [4] maybe due to their sensitivity towards hypo- and hyperosmotic environmental conditions. Finally, PC-deficient mutants of S. meliloti are much more sensitive to freezing in the presence of glycerol than their respective wild type. 3.1.2. Phosphatidylcholine in Bradyrhizobium japonicum is required for efficient nitrogen fixation The rhizobial bacterium Bradyrhizobium japonicum is able to form a nitrogen-fixing symbiosis with soybean. In B. japonicum USDA110 a homologue (Bj-PmtA) of the sinorhizobial PmtA family was initially discovered [30]. A PmtA-deficient B. japonicum mutant still produced low levels of PC (6% of total phospholipids) but they were greatly decreased when compared to the PC content of the wild type strain (52% of total phospholipids). Root nodules of soybean plants infected with B. japonicum pmtA mutants showed a reduced nitrogen fixation activity of only 18% of the wild type level. Nodules generated by the pmtA mutant were beige instead of red, suggesting decreased amounts of leghemoglobin. Within the infected plant cells, the number of bacteroids was greatly reduced in the case of the mutant. From these data it was concluded that wild type amounts of bacterial PC are required for an efficient symbiotic interaction of the bacterium with its soybean host plant and for a proper development of a mature nodule [30]. Besides PmtA which perfoms well the first methylation to yield MMPE, at least 3 more distinct phospholipid N-methyltransferases (PmtX1, PmtX3, and PmtX4) with distinct substrate specificities exist in B. japonicum

508


Fig. 4. Unrooted maximum-likelihood phylogenetic tree of phosphatidylcholine synthase (Pcs) and Pcs-like ORFs. A multiple sequence alignment of the selected amino acid sequences was made using the program MUSCLE at www.ebi.ac.uk. Based on this alignment, the tree was constructed using the program PhyML 3.0 at www.atgc-montpellier.fr. Distances between sequences are expressed as 0.5 changes per amino acid residue. For the construction of the tree the following Pcs and Pcs-like ORFs were selected: Pcs from Sinorhizobium meliloti 1021 (Pcs_S_meli-NP_385778), Pcs from Rhizobium leguminosarum bv. viciae 3841 (Pcs_R_legu-YP_767960), an ORF from Rhizobium etli CFN42 (ORF_R_etli-YP_469590), Pcs from Agrobacterium tumefaciens C58 (Pcs_A_tume-NP_354778), Pcs from Brucella melitensis bv. abortus 2308 (Pcs_B_abor-YP_418843), Pcs from Mesorhizobium loti MAFF303099 (Pcs_M_loti-NP_102294), an ORF from Bartonella henselae str. Houston-1 (ORF_B_hens-YP_033658), an ORF from Rhodopseudomonas palustris BisB18 (ORF_R_palu-YP_532487), Pcs from Bradyrhizobium japonicum USDA 110 (Pcs_B_japo-NP_771225), an ORF from Rhodobacter sphaeroides 2.4.1 (ORF_R_spha-YP_353639), an ORF from Azospirillum brasilense Sp245 (ORF_A_bras-CCC97530), Pcs from Pseudomonas aeruginosa PAO1 (Pcs_P_aeru-NP_252546), Pcs from Pseudomonas putida KT2440 (Pcs_P-puti-NP_742892), Pcs from Legionella pneumophila subsp. pneumophila str. Philadelphia 1 (Pcs_L-pneu-YP_095613), an ORF from Francisella novicida U112 (ORF_F_novi-YP_899093), an ORF from Actinomyces odontolyticus ATCC 17982 (ORF_A_odon-ZP_02043623), Pcs from Borrelia burgdorferi B31 (Pcs_B_burg-NP_212383), an ORF from Rhodopirellula baltica SH 1 (ORF_R_balt-NP_866142), an ORF from Salinibacter ruber DSM 13855 (ORF_S_rube-YP_444486), an ORF from Rubrobacter xylanophilus DSM 9941 (ORF_R_xyla-YP_643061), an ORF from Truepera radiovictrix DSM 17093 (ORF_T_radi-YP_003704207), phosphatidylserine synthase (PssA) from S. meliloti 1021 (Pss_S_meli-NP_385228), PssA from P. putida KT2440 (Pss_P_puti-NP_746786), and PssA from Bacillus subtilis subsp. subtilis str. 168 (NP_388109).

(Fig. 2) [9]. Single mutants of B. japonicum, generated in pmtX2, pmtX3, or pmtX4 had similar phospholipid composition as the wild type. To date the construction of a bradyrhizobial pmtX1-deficient mutant could not be achieved [9]. A toxin-antitoxin (TA)-like system, encoded by the genes bat/bto (bsl2435/bll2434), was identified in B. japonicum [37]. Deletion of the bat/bto module resulted in a pleiotropic phenotype. Surprisingly, the mutant grew much faster than the wild type in rich media whereas in minimal medium the mutant did not grow at all. Mutant cells were shorter and wider and showed an increased softness when compared with wild type. Wild type B. japonicum forms high molecular weight versions of lipopolysaccharide whereas the mutant can form only low molecular weight versions of lipopolysaccharide. PC synthesis was completely abolished in the bat/bto deletion mutant probably due to the absence of pmtA and pmtX1 transcripts. Vaccenic acid is the predominant fatty acid in membranes of the wild type and it was largely replaced by palmitic acid in the mutant membrane. The bat/bto deletion mutant had a decreased symbiotic capacity. The mutant produced fewer nodules on soybeans than the wild type and plants which had a reduced dry weight [37]. From these data it seems clear, however, that bacterial PC is not an absolute requirement for the formation of a nitrogen-fixing root nodule symbiosis with legumes.

In the peanut-nodulating Bradyrhizobium sp. SEMIA 6144, a pcs homologue and multiple pmt homologues have been identified [38]. A pmtA-deficient mutant of Bradyrhizobium sp. SEMIA 6144 showed a 50 % decrease in the PC content in comparison with the wild type. The mutant was severely affected in motility, showed reduced cell size, and formed smaller colonies on solid complex media. Although the mutant formed wild type-like nodules on its host plant, it was less competitive than the wild type [38] maybe due to its impaired motility. 3.1.3. Phosphatidylcholine in Agrobacterium tumefaciens is required for virulence A. tumefaciens causes tumor formation in plants. Plant signals induce the expression of virulence (Vir) proteins and the formation of a type IV secretion system (T4SS) in the bacteria. On attachment to the plant cells, transfer DNA and Vir proteins are transferred to the host cell through the bacterial T4SS causing alterations in the host that lead to tumor formation [39]. A. tumefaciens C58 has one copy of a Sinorhizobium-like PmtA-encoding gene (Atu300) and a pcs gene (Atu1793), both located on its circular chromosome. Deletion mutants of pmtA or pcs as well as a double mutant with deletions in both genes have been generated and characterized [40]. Growth of wild type and the three mutants was indistinguishable in liquid yeast extract-, beef


extract-, and peptone-containing (YEB) complex medium at 30 °C. On solid YEB medium or AB minimal medium, the double mutant grew slower than the single mutants and wild type at 30 °C whereas at 37 °C, the double mutant hardly grew on complex or minimal medium. Wild type A. tumefaciens C58 caused tumor formation on Kalanchoë leaves whereas the double mutant was unable to do so. In the PCdeficient double mutant, proteins, required for the formation of the T4SS, are absent due to a lack of expression of the virB and virD operons. Expression of the virA and virG genes, encoding a two-component signaling system required for tumor formation, is normal in a PCdeficient mutant of A. tumefaciens [40]. PC is detected in the inner and outer membrane of A. tumefaciens [41]. The PC-deficient double mutant of A. tumefaciens is less motile than the wild type on AB minimal medium [41]. The decreased motility of the mutant might be due to lower levels of the flagellar proteins, i.e. FlaA, FlaB, which is surprising as the transcript levels for many flagellar genes are increased under such conditions [42]. Finally, the PC-deficient double mutant of A. tumefaciens was more inclined to attach to solid surfaces than the wild type [41]. A proteomic and transcriptomic characterization of the PC-deficient double mutant of A. tumefaciens indicated that expression of virulence genes in the PC-deficient double mutant is dramatically (several hundred-fold) reduced whereas the majority of differentially expressed genes unrelated to virulence were altered two- to five-fold, in rare cases up to 16-fold [42]. Under acidic conditions, the plant signal acetosyringone (AS) induces virulence-related genes through the two-component regulatory system VirA/VirG and proteins such as VirB8, VirC1, VirE2, VirH1, Tzs are upregulated by AS in the wild type. Such an upregulation cannot be observed in the PC-deficient double mutant suggesting that the signaling through the VirA/VirG system is not functioning properly in this mutant [42].

3.1.4. Phosphatidylcholine in Brucella is required for full virulence Brucella causes a highly contagious zoonosis usually caused by ingestion of unsterilized milk products from infected animals. In cell-free extracts of Brucella melitensis only Pcs activity and no Pmt activity could be detected [34]. This seemed surprising as genomes of Brucella species contain a homologue of sinorhizobial pmtA and a homologue of pcs. Formation of PC in Brucella abortus depends on a functional Pcs and on choline in the growth medium [43]. Expression of brucellal pmtA in E. coli did not cause the formation of PC in this host [43]. Therefore the methylation pathway of PC synthesis seems not to be functional in the Brucella strains studied by MartínezMorales et al. [34] and Comerci et al. [43]. Notably, potential brucellal Pmts show amino acid changes in the conserved S-adenosyl methionine (SAM)-binding motif VLELGXGXG. The first two G of the motif are critical for activity of the closely related PmtA from A. tumefaciens [29] and their alterations (of the first G in the SAM-binding motif of PmtA of some Brucella suis strains or of the second G in the motif of B. abortus) explain these inactive brucellal PmtA versions. A Pcsand consequently PC-deficient mutant of B. abortus was growing similarly as the wild type and was able to invade and replicate intracellularly in murine macrophages in a wild type-like fashion. However, virulence of the Pcs-deficient mutant of B. abortus was reduced when assayed in mice [43]. A similar study [44] reports slightly reduced growth of pcs-deficient mutants in complex medium when compared to wild type whereas mutants deficient in the pmtA-like gene might be affected during growth on minimal media. Also the intact pmtA-like gene seems to be important for long-term survival in the mouse spleen suggesting that the B. abortus PmtA might not be totally inactive [44] or might have some moonlighting activity. However, upon long-term survival in the mouse spleen, suppressor mutations might have occurred in the bacterium. In this context it is worth noting that a single nucleotide change in the pmtA-like gene of B. abortus would restore a triplet that encodes for a G in the position which corresponds to the second G of the SAM-binding motif of the

509

PmtA-like protein of B. abortus. Such a change in this protein might restore PmtA activity. 3.1.5. Phosphatidylcholine is required for full virulence of Legionella pneumophila Legionella pneumophila is the causative agent of Legionnaires’ disease. Although L. pneumophila can synthesize PC via the PmtA or the Pcs pathway [4,34], the Pcs pathway is the main route for PC formation [32]. Double mutants of L. pneumophila, deficient in both pathways cannot form any PC and are affected in multiple ways [32]. First, there is reduced binding of PC-deficient L. pneumophila to macrophages due to the loss of bacterial PC or a derivative formed from PC that is recognized by the host cell. Macrophages detect the platelet-activating factor (PAF) in body fluids via their PAF receptor and this interaction modulates the macrophage immune response. Initial binding of L. pneumophila to macrophages can occur via several different receptors, however, the major binding is via the PAF receptor. PAF has the same glycerophosphocholine head group as PC [4], but in contrast to PC, PAF is soluble in an aqueous environment. Legionellal PC might therefore mimic the PAF on a molecular level and cause bacterial adhesion to the PAF receptor. Second, a PC-deficient mutant has only a poorly functioning type IVB secretion system (Dot/Icm apparatus) and is severely impaired in delivering virulence protein substrates which are required for bacterial intracellular growth into the cytosol of infected cells. Third, strains lacking PC show lowered cytotoxicity, are non-motile, and have low levels of flagellin protein [32]. 3.1.6. Phosphatidylcholine in Pseudomonas aeruginosa P. aeruginosa is an opportunistic human pathogen of clinical relevance. In P. aeruginosa PAO1, homologues for rhodobacterial PmtA (PA0798) and Pcs (PA3857) were encountered [45]. PAO1 is able to synthesize PC when grown on complex medium or on defined minimal media containing choline. PAO1 grown on choline-free media does not contain PC. A PA0798-deficient mutant of PAO1 still forms PC at wild type levels whereas a PA3857-deficient mutant forms no detectable amounts of PC. Expression of PA3857 in E. coli causes the formation of PC whereas the expression of PA0798 does not. All these data seem to indicate that the PmtA homologue PA0798 is not functional and that Pcs constitutes the only functional PC biosynthesis pathway in PAO1. However, PA0798 is a distant homologue of rhodobacterial PmtA and with our present search (Supplementary Table 1) it would not have been assigned as a good Pmt candidate. It is surprising, however, that the viability of PAO1 is not only affected in a Pcs-deficient mutant, but also in PA0798-deficient one, and most strongly in a mutant deficient in both, Pcs and PA0798 [45]. These latter data leave doubts that PA0798 might still be responsible for the formation of very minor amounts of PC which might not have been detected by the methods employed so far or that PA0798 might be responsible for some still unknown moonlighting activity. A recent study confirmed that PC biosynthesis via the Pcs pathway is widespread in P. aeruginosa laboratory and clinical strains [13]. PC constitutes about 4% of the total phospholipids in P. aeruginosa when the bacterium is grown in choline-containing media. Pcs-deficient mutants of PAO1 or PA14 did not produce detectable amounts of PC but behaved like their respective wild types when assayed for many phenotypes, among them motility, biofilm formation, colonization, or virulence [13]. 3.1.7. Atypic phosphatidylcholine synthase in Pseudomonas putida Phylogenetically, the gene pp0731 from P. putida KT2440 encoding a potential CDP-alcohol phosphatidyltransferase is one of the most distant members of possible Pcs enzymes (Fig. 4). Whereas wild type P. putida forms PC upon growth on complex medium, a mutant deficient in pp0731 cannot. A pp0731-deficient mutant complemented with an intact pp0731 in trans can again synthesize PC [36]. These data strongly suggest that pp0731 encodes a functional Pcs enzyme. A P. putida wild type strain can absorb much more Al3+ ions than a PC-deficient mutant

510


[36] and it is therefore thought that binding of Al3+ ions by means of bacterial PC might provide some resistance against Al 3+ toxicity. 3.2. Bacterial phosphatidylcholine as biosynthetic intermediate S. meliloti usually synthesizes PG, CL, PE, MMPE, and PC as its major membrane lipids when grown in culture media rich in phosphate [4,46]. However, under phosphorus-limiting conditions, S. meliloti replaces most of its phospholipids by membrane-forming lipids that do not contain phosphorus, such as the sulfolipid sulfoquinovosyl diacylgycerol (SL), an ornithine-containing lipid, and diacylglyceryl-N, N,N-trimethylhomoserine (DGTS) [46]. These phosphorus-free membrane lipids are not important for the symbiotic life style of S. meliloti, but they are required for optimal growth under phosphorus-limiting conditions [47]. Recently we could show that under such phosphorus-limiting conditions of growth, PC and PE are degraded in S. meliloti [48]. This degradation of zwitterionic lipids is achieved by a phospholipase C (PlcP) which is induced by an active PhoB response regulator. The phospholipase C PlcP converts PC to phosphocholine and diacylglycerol (DAG) (Fig. 5). DAG, in turn, is the lipid anchor from which biosynthesis is initiated during the formation of the phosphorus-free membrane lipids SL and DGTS. Two structural genes (btaAB) are required for DGTS biosynthesis in S. meliloti [47]. BtaA converts DAG into diacylglyceryl-homoserine

(DGHS), and BtaB catalyzes the threefold methylation of DGHS to yield DGTS. Expression of BtaA and BtaB requires induction by the PhoB regulator. Another minor DAG-containing, phosphorus-free membrane lipid in S. meliloti is SL and sqdB is required for its formation [49]. In some related bacteria, such as Mesorhizobium or Agrobacterium, distinct glycolipids are among the phosphorus-free membrane lipids formed under phosphorus limitation [50]. Also in these cases, DAG formed by the PlcP reaction is the substrate used by glycosyltransferases. PlcP (SMc00171) is the first example of a bacterial phospholipase C that degrades endogenous phospholipids and it belongs to the PfamPF00149 protein family of phosphoesterases [48]. The four motifs (DxH, GD, GNHD, GHxH) of the calcineurin-like phosphoesterase superfamily are present in PlcP (SMc00171) [48]. PlcP-like phospholipases are also encountered in most α-proteobacteria, but are notably absent in Bartonella, Brucella, and the Rickettsiales which are pathogens usually not confronted with low phosphorus conditions. Also PlcP-like phospholipases exist in many β- and γ-proteobacteria, such as P. aeruginosa or Ralstonia solanacearum [48]. PlcP is PhoB-regulated [51], and upon phosphorus limitation, the plcP (smc00171) transcript is increased about 20-fold. Remarkably, also pmtA (smc00414) of S. meliloti is induced about fourfold [51] under phosphorus-limiting conditions. Similarly, genes encoding for enzymes of the initial steps of SL and DGTS biosynthesis (Fig. 5), SqdB (SMc03961) and BtaA (SMc01848), respectively, are preceded by a

Fig. 5. Membrane lipid formation, turnover, and recycling in Sinorhizobium meliloti. The major membrane phospholipids of S. meliloti, phosphatidylethanolamine, phosphatidylcholine, phosphatidylglycerol, and cardiolipin, are formed by well-known pathways (see Fig. 1). Under phosphorus-limiting conditions, zwitterionic phospholipids are degraded to diacylglycerol whereas during low osmolarity conditions phosphoglycerol head groups of PG are transferred to form anionic cyclic glucans and diacylglycerol. Diacylglycerol can be recycled to phosphatidic acid or serve as lipid anchor during the formation of diacylglycerol-based phosphorus-free membrane lipids such as sulfolipid or diacylglyceryl trimethylhomoserine. CdsA: CDP-diacylglycerol synthase; Pcs: phosphatidylcholine synthase; PssA: phosphatidylserine synthase; Psd: phosphatidylserine decarboxylase; PmtA: phospholipid N-methyltransferase; PgsA: phosphatidylglycerolphosphate synthase; ClsB: cardiolipin synthase, bacterial type; DgkA: diacylglycerol kinase (SMc04213); SqdB: UDP-sulfoquinovose synthase (SMc03961); BtaA: S-adenosylmethionine:diacylglycerol 3-amino-3-carboxypropyl transferase (SMc01848); BtaB: diacylglyceryl homoserine N-methyltransferase (SMc01849); CgmB: cyclic glucan-modifying phosphoglycerol transferase (SMc04438); PlcP: phospholipase C (SMc00171); BetC: choline sulfatase (SMc00127); BetA: choline dehydrogenase (SMc00093); BetB: betaine aldehyde dehydrogenase (SMc00094). Steps increased under phosphorus limitation (red) or at low osmolarity (blue) are highlighted. Also, the degradation of phosphocholine to choline and further conversion to glycine betaine is indicated (green). In symbiosis, dgkA is induced suggesting an increased recycling of diacylglycerol (DAG) to phosphatidic acid and thereby a reintroduction of the DAG lipid anchor into phospholipid biosynthesis (purple).


Pho box that mediates PhoB-controlled expression under phosphorus limitation [51]. PlcP (SMc00171) from S. meliloti degrades zwitterionic phospholipids of the bacterial membrane to DAG and phosphoalcohol (Fig. 5). The betCBA operon is required for the formation of the compatible solute glycine betaine [52] and in S. meliloti, it is induced under phosphoruslimiting conditions of growth [51]. Although BetC is annotated as a choline sulfatase, it is clear that it can act as phosphocholine phosphatase and form inorganic phosphate (Pi) and choline [52]. The liberated choline might then further be converted to glycine betaine by BetBA (Fig. 5), or it might be recycled by Pcs into the PC pool. The liberated Pi can be used for the synthesis of essential phosphorus-containing biomolecules such as nucleic acids. Therefore, membrane phospholipids might provide an important internal phosphorus pool for Gram-negative environmental bacteria under phosphorus-limiting conditions of growth. During adaptation to hypo-osmotic conditions, S. meliloti synthesizes periplasmic cyclic β-1,2-glucans, which can be substituted with sn-1-phosphoglycerol moieties derived from PG [53]. The phosphoglycerol transferase CgmB (SMc04438) of S. meliloti [54] is thought to catalyze a reaction similar to the sn-1-phosphoglycerol transferase MdoB from E. coli, and produce DAG as a byproduct. Given the multiple sources of DAG in bacteria, it will be of great interest to determine whether DAG or DAG-derived molecules function as stress signals in bacteria. It is remarkable that one of the genes induced in early symbiosis is dgkA [55]. This suggests that in symbiotic conditions, DAG is phosphorylated to phosphatidic acid thereby reentering the biosynthesis pathway for phospholipids (Fig. 5). Bacteria, such as S. meliloti, have at least two totally different life styles, one as a soil bacterium exposed to phosphorus limitation stress and another as a symbiotic bacterium in association with the legume host plant where, at least for the bacterium, phosphorus supplies are abundant. Between the two different life styles of S. meliloti, membrane lipid composition is distinctly different. In symbiotic conditions, the membrane is formed by phospholipids whereas upon phosphorus limitation phospholipids are largely replaced by membrane lipids that do not contain any phosphorus in their structure. 3.3. Can we learn phosphatidylcholine metabolism and about metabolic fluxes from operon organization in genomes? Genes that are organized in an operon or in a gene cluster often encode for proteins that physically interact or that belong to the same metabolic pathway. In Rhodobacter sphaeroides, the gene pmtA forms an operon with the genes pssA and psd (Fig. 6), encoding PssA and Psd, respectively. This operon therefore encodes individual steps of the biosynthetic pathway from CDP-DAG to PC. A second example is the pmtX2 gene from R. palustris BisB18 which is clustered with genes coding for a PlcP homologue and a glycosyltransferase. In this case, the predicted pathway would start with PE. The Pmt would then be responsible for PC formation. The phospholipase C would hydrolyze PC to DAG and choline phosphate and finally DAG might be the substrate of the glycosyltransferase and be converted into a phosphorus-free glycolipid. Remarkably, in B. japonicum the expression of the gene downstream of pmtX3 (bll8165), encoding a potential alkaline phosphatase, is similarly upregulated in the pmtA-deficient mutant. Upstream of pmtX4 are two genes (blr4802 and blr4803) encoding for a twocomponent response regulator and a sensor histidine kinase, respectively and although both are upregulated in the pmtA-deficient mutant [31] they seem not to form an operon with pmtX4. It is tempting to speculate about specific functions for the 5 Pmt activities reported for B. japonicum considering their genomic context. In A. tumefaciens, pmtA (atu300) seems to form an operon with the downstream genes atu0299, pyrF (atu298), and atu0297 and a similar organization is observed in S. meliloti where pmtA (smc00414) is followed by smc00413 (putatively encoding a phosphoglycerate mutase), pyrF

511

(smc00414, encoding orotidine 5′-phosphate decarboxylase), smc00411, and smc00410 (probably encoding NADH-ubiquinone oxidoreductase). It is not immediately obvious why there should be a correlation between PC biosynthesis, pyrimidine biosynthesis, and the conversion of reductive power into energy via respiration. In A. tumefaciens also pcs (Atu1793) seems the first gene of an operon that includes four genes (abc1-4) probably encoding for an ABC transporter [41]. A mutant lacking abc1-4 seems not to be affected in PC formation by the Pcs pathway [41]. Sinorhizobial pcs (smc00247) might be the first gene of an operon followed a gene qor (smc00246) encoding for a probable NADPH:quinone reductase, and three genes (smc00245–smc00243) encoding for a potential ABC transporter. These examples show that in many cases we might learn about function of genes from operon organization in genomes. In a few cases the functional relation of the genes is trivial, but in many other cases our knowledge is still too limited to make meaningful conclusions. The other question that arises is whether we can learn something about distinct metabolic fluxes in specific organisms if functionally identical genes are encountered in different genomic contexts. 3.4. Modulation of physicochemical properties of bacterial membranes by phosphatidylcholine Mutants of different PC-containing bacteria that are devoid of detectable PC display a wide range of phenotypes distinct from those observed in their respective wild type strains. In general terms, PC-deficient mutants are impaired in their symbiotic (S. meliloti and B. japonicum) or pathogenic interactions (A. tumefaciens, B. abortus and L. pneumophila) with eukaryotic hosts. They produce less flagellin and are less motile (Bradyrhizobium SEMIA, A. tumefaciens, and L. pneumophila) and they are impaired in the formation of a type IV secretion system (A. tumefaciens and L. pneumophila) required for infection of their host. PC-deficient mutants are also more sensitive to high or low osmotic stress (S. meliloti), don't grow at elevated temperatures (A. tumefaciens), and are sensitive to freezing (S. meliloti and P. aeruginosa). Already Goldfine noted that bacteria with high proportions of unsaturated fatty acids often have PC in their membranes and he suggested that larger effective polar head groups are required to compensate for the increased volume provoked by cis-unsaturated fatty acyl residues in order to maintain stable bilayers [56]. In this context it is worth noting that the toxin/antitoxin-deficient mutant of B. japonicum which was devoid of PC also had much reduced amounts of the unsaturated fatty acid cis-vaccenic acid [37]. In general, the lack of PC in mutants is compensated by a respective increase in the relative amount of PE which should result in a greater tendency to form a non-lamellar phase if the relative amount of cis-unsaturated fatty acyl residues remains constant. One might expect that such PC-deficient membranes are more fluid, are more permeable for small molecules and ions, and might provoke a decreased membrane potential. However, if a reduction of PC is accompanied by a reduction of cis-unsaturated fatty acids (favoring the formation of the lamellar phase) a certain compensation of the original membrane properties would be expected. The fact that PC-deficient bacterial mutants are severely affected in their growth at elevated temperatures might also be explained as a result of the increase of PE which undergoes lamellar to non-lamellar transition at lower temperatures than PC. Determinations of membrane permeability and fluidity as well as of membrane potentials [57] in PC-deficient mutants are needed in order to understand better how the presence of PC affects the physicochemical properties of bacterial membranes. 4. Conclusions Phosphatidylcholine (PC) is unequally distributed in the three domains of life. In Eukarya PC is nearly ubiquitous, in Archaea PC is essentially unknown, and in the Bacteria the ability to synthesize PC is found in about 15% of its members. Usually bacteria synthesize their PC via the phospholipid N-methylation pathway or via the

512


Rhodobacter sphaeroides 2.4.1 pssA

pmtA

psd

Rhodopseudomonas palustris BisB18

pmtX2

phosphomethyl pyrimidine kinase

plcP

glycosyltransferase

Bradyrhizobium japonicum USDA110 pmtA

dnaJ

dnaK

trmU

pmtX1

glycosyltransferase

pmtX2

pmtX3

phoA

bll6993

blr0682

pyrF

tRNAMet

bll6633

plcP

glycosyltransferase

bll8167

blr4802 (RR)

blr4803 (HHK)

pmtX4

300 nt Fig. 6. Genomic context of genes encoding phospholipid N-methyltransferases. Genes coding for phospholipid N-methyltransferases are often found within operons or gene clusters. Genomic contexts for pmtA from R. sphaeroides 2.4.1 (YP_353639), pmtX2 from Rhodopseudomonas palustris BisB18, and pmtA (NP_771225), pmtX1 (NP_773634), pmtX2 (NP_773274), pmtX3 (NP_774806), and pmtX4 (NP_771444) from B. japonicum USDA110 are shown. If a functional prediction was possible, it is indicated below each arrow. Otherwise only the gene tag ID is mentioned. pssA: phosphatidylserine synthase; psd: phosphatidylserine decarboxylase; pmtA: phospholipid N- methyltransferase; plcP: phospholipase C; dnaK: chaperone, also known as HSP-40; dnaJ: chaperone, also known as HSP-70; pyrF: orotidine 5’-phosphate decarboxylase; trmU: tRNA methyl transferase; phoA: alkaline phosphatase; RR: response regulator; HHK: hybrid histidine kinase. Genes coding for proteins with phospholipid N-methyltransferase activity are shaded in grey.

phosphatidylcholine synthase pathway (Pcs) or via both. Pcs is only found in the domain Bacteria and in some important pathogens, Pcs is the only (Brucella, Bartonella, Pseudomonas, Francisella, and Borrelia) or the predominant (Legionella) pathway for PC biosynthesis. A search for specific inhibitors of Pcs might lead to the development of antibiotics that act against these pathogens. It had been speculated that PC plays an important role in the interactions between symbiotic and pathogenic bacteria and their eukaryotic hosts. At least in the case of the pathogens B. abortus, L. pneumophila, and A. tumefaciens, mutants deficient in PC formation were less virulent than the respective wild types in interactions with their hosts. This is consistent with the observation that many of the bacteria presenting a Pcs homolog are known to show an either symbiotic (S. meliloti, R. leguminosarum, and M. loti) or pathogenic lifestyle (A. tumefaciens, B. burgdorferi, L. pneumophila, and B. abortus). Besides its structural function as a membrane-forming phospholipid, bacterial PC can function as intermediate during the biosynthesis of other biomolecules or form part of lipid metabolic cycles. Physicochemical studies are needed to advance in the understanding how the presence or absence of PC affects the properties of bacterial membranes. Supplementary data to this article can be found online at http:// dx.doi.org/10.1016/j.bbalip.2012.08.009.

Acknowledgements This research was supported by grants from CONACyT-Mexico (158359 and 178359) and DGAPA/UNAM (IN203612).

References [1] D.L. Nelson, M.M. Cox, Lehninger: Principles of Biochemistry, Fifth ed. W.H. Freeman and Company, New York, 2008. [2] C.O. Rock, Fatty acid and phospholipid metabolism in prokaryotes, in: D.E. Vance, J.E. Vance (Eds.), Biochemistry of Lipids, Lipoproteins and Membranes, Fifth ed., Elsevier, Amsterdam, 2008, pp. 59–96. [3] O. Geiger, N. González-Silva, I.M. López-Lara, C. Sohlenkamp, Amino acid-containing membrane lipids in bacteria, Prog. Lipid Res. 49 (2010) 46–60. [4] C. Sohlenkamp, I.M. López-Lara, O. Geiger, Biosynthesis of phosphatidylcholine in bacteria, Prog. Lipid Res. 42 (2003) 115–162. [5] H. Goldfine, Lipids of procaryotes — structure and distribution, in: S. Razin, S. Rottem (Eds.), Current Topics in Membranes and Transport, vol. 17, Academic Press, New York, 1982, pp. 1–43. [6] I.M. López-Lara, O. Geiger, Novel pathway for phosphatidylcholine biosynthesis in bacteria associated with eukaryotes, J. Biotechnol. 91 (2001) 211–221. [7] M. Aktas, M. Wessel, S. Hacker, S. Klüsener, J. Gleichenhagen, F. Narberhaus, Phosphatidylcholine biosynthesis and its significance in bacteria interacting with eukaryotic cells, Eur. J. Cell Biol. 89 (2010) 888–894. [8] C. Kent, P. Gee, S.Y. Lee, X. Bian, J.C. Fenno, A CDP-choline pathway for phosphatidylcholine biosynthesis in Treponema denticola, Mol. Microbiol. 51 (2004) 471–481.

O. Geiger et al. / Biochimica et Biophysica Acta 1831 (2013) 503–513 [9] S. Hacker, C. Sohlenkamp, M. Aktas, O. Geiger, F. Narberhaus, Multiple phospholipid N-methyltransferases with distinct substrate specificities are encoded in Bradyrhizobium japonicum, J. Bacteriol. 190 (2008) 571–580. [10] K.E. de Rudder, C. Sohlenkamp, O. Geiger, Plant-exuded choline is used for rhizobial membrane lipid biosynthesis by phosphatidylcholine synthase, J. Biol. Chem. 274 (1999) 20011–20016. [11] C. Sohlenkamp, K.E. de Rudder, V. Röhrs, I.M. López-Lara, O. Geiger, Cloning and characterization of the gene for phosphatidylcholine synthase, J. Biol. Chem. 275 (2000) 18919–18925. [12] K.E. de Rudder, I.M. López-Lara, O. Geiger, Inactivation of the gene for phospholipid N-methyltransferase in Sinorhizobium meliloti: phosphatidylcholine is required for normal growth, Mol. Microbiol. 37 (2000) 763–772. [13] A.A. Malek, M.J. Wargo, D.A. Hogan, Absence of membrane phosphatidylcholine does not affect virulence and stress tolerance phenotypes in the opportunistic pathogen Pseudomonas aeruginosa, PLoS One 7 (2012) e30829. [14] O. Geiger, C. Sohlenkamp, I.M. López-Lara, Formation of bacterial membrane lipids: pathways, enzymes, reactions. , (chapter 27) in: K.N. Timmis (Ed.), Handbook of Hydrocarbon and Lipid Microbiology; Hydrocarbons, Oils, and Lipids: Diversity, Properties and Formation, Springer-Verlag, Berlin Heidelberg, Germany, 2010, pp. 395–408. [15] M. Sandoval-Calderón, O. Geiger, Z. Guan, F. Barona-Gómez, C. Sohlenkamp, A eukaryote-like cardiolipin synthase is present in Streptomyces coelicolor and in most actinobacteria, J. Biol. Chem. 284 (2009) 17383–17390. [16] C. Sohlenkamp, K.E. de Rudder, O. Geiger, Phosphatidylethanolamine is not essential for growth of Sinorhizobium meliloti on complex culture media, J. Bacteriol. 186 (2004) 1667–1677. [17] M.A. Vences-Guzmán, O. Geiger, C. Sohlenkamp, Sinorhizobium meliloti mutants deficient in phosphatitidylserine decarboxylase accumulate phosphatidylserine and are strongly affected during symbiosis with alfalfa, J. Bacteriol. 190 (2008) 6846–6856. [18] K.E. de Rudder, J.E. Thomas-Oates, O. Geiger, Rhizobium meliloti mutants deficient in phospholipid N-methyltransferase still contain phosphatidylcholine, J. Bacteriol. 179 (1997) 6921–6928. [19] R.L. Solís-Oviedo, F. Martínez-Morales, O. Geiger, C. Sohlenkamp, Functional and topological analysis of phosphatidylcholine synthase from Sinorhizobium meliloti, Biochim. Biophys. Acta 1821 (2012) 573–581. [20] A.S. Fischl, M.J. Homann, M.A. Poole, G.M. Carman, Phosphatidylinositol synthase from Saccharomyces cerevisiae: reconstitution, characterization, and regulation of activity, J. Biol. Chem. 261 (1986) 3178–3183. [21] A. Dutt, W. Dowhan, Purification and characterization of a membrane-associated phosphatidylserine synthase from Bacillus licheniformis, Biochemistry 26 (1985) 1073–1079. [22] T.J. Larson, W. Dowhan, Ribosomal-associated phosphatidylserine synthase from Escherichia coli: purification by substrate-specific elution from phosphocellulose using cytidine 5’-diphospho-1,2-diacyl-sn-glycerol, Biochemistry 15 (1976) 5212–5218. [23] M. Bogdanov, P. Heacock, Z. Guan, W. Dowhan, Plasticity of lipid–protein interactions in the function and topogenesis of the membrane protein lactose permease from Escherichia coli, Proc. Natl. Acad. Sci. U. S. A. 107 (2011) 15057–15062. [24] T. Kaneshiro, J.H. Law, Phosphatidylcholine synthesis in Agrobacterium tumefaciens. I. Purification and properties of a phosphatidylethanolamine N-methyltransferase, J. Biol. Chem. 239 (1964) 1705–1713. [25] S.I. Sherr, J.H. Law, Phosphatidylcholine synthesis in Agrobacterium tumefaciens. II. Uptake and utilization of choline, J. Biol. Chem. 240 (1965) 3760–3765. [26] V. Arondel, C. Benning, C.R. Somerville, Isolation and functional expression in Escherichia coli of a gene encoding phosphatidylethanolamine methyltransferase (EC 2.1.1.17) from Rhodobacter sphaeroides, J. Biol. Chem. 268 (1993) 16002–16008. [27] Y. Tahara, T. Yamashita, A. Sogabe, Y. Ogawa, Isolation and characterization of Zymomonas mobilis mutant defective in phosphatidylethanolamine N-methyltransferase, J. Gen. Appl. Microbiol. 40 (1994) 389–396. [28] M. Aktas, F. Narberhaus, In vitro characterization of the enzyme properties of the phospholipid N-methyltransferase PmtA from Agrobacterium tumefaciens, J. Bacteriol. 191 (2009) 2033–2041. [29] M. Aktas, J. Gleichenhagen, R. Stoll, F. Narberhaus, S-Adenosylmethionine-binding properties of a bacterial phospholipid N-methyltransferase, J. Bacteriol. 193 (2011) 3473–3481. [30] A.C. Minder, K.E. de Rudder, F. Narberhaus, H.M. Fischer, H. Hennecke, O. Geiger, Phosphatidylcholine levels in Bradyrhizobium japonicum membranes are critical for an efficient symbiosis with the soybean host plant, Mol. Microbiol. 39 (2001) 1186–1198. [31] S. Hacker, J. Gödeke, A. Lindemann, S. Mesa, G. Pessi, F. Narberhaus, Global consequences of phosphatidylcholine reduction in Bradyrhizobium japonicum, Mol. Genet. Genomics 280 (2008) 59–72. [32] G.M. Conover, F. Martínez-Morales, M.I. Heidtman, Z.Q. Luo, M. Tang, C. Chen, O. Geiger, R.R. Isberg, Phosphatidylcholine synthesis is required for optimal function of Legionella pneumophila virulence determinants, Cell. Microbiol. 10 (2008) 514–528. [33] P.N. Pandhi, B.F. Hammond, The polar lipids of Actinomyces viscosus, Arch. Oral Biol. 23 (1978) 17–21. [34] F. Martínez-Morales, M. Schobert, I.M. López-Lara, O. Geiger, Pathways for phosphatidylcholine biosynthesis in bacteria, Microbiology 149 (2003) 3461–3471.

513

[35] X.G. Wang, J.P. Scagliotti, L.T. Hu, Phospholipid synthesis in Borrelia burgdorferi: BB0249 and BB0721 encode functional phosphatidylcholine synthase and phosphatidylglycerolphosphate synthase proteins, Microbiology 150 (2004) 391–397. [36] P.S. Boeris, G.I. Lucchesi, The phosphatidylcholine synthase of Pseudomonas putida A ATCC12633 is responsible for the synthesis of phosphatidylcholine which acts as a temporary reservoir for Al3+, Microbiology 158 (2012) 1249–1257. [37] P.S. Miclea, M. Péter, G. Végh, G. Cinege, E. Kiss, G. Váró, I. Horváth, I. Dusha, Atypical transcriptional regulation and role of a new toxin-antitoxin-like module and its effect on the lipid composition of Bradyrhizobium japonicum, Mol. Plant Microbe Interact. 23 (2010) 638–650. [38] D.B. Medeot, C. Sohlenkamp, M.S. Dardanelli, O. Geiger, M. García de Lema, I.M. López-Lara, Phosphatidylcholine levels of peanut-nodulating Bradyrhizobium sp. SEMIA 6144 affect cell size and motility, FEMS Microbiol. Lett. 303 (2010) 123–131. [39] A. Pitzschke, H. Hirt, New insights into an old story: Agrobacterium-induced tumour formation in plants by plant transformation, EMBO J. 29 (2010) 1021–1032. [40] M. Wessel, S. Klüsener, J. Gödeke, C. Fritz, S. Hacker, F. Narberhaus, Virulence of Agrobacterium tumefaciens requires phosphatidylcholine in the bacterial membrane, Mol. Microbiol. 62 (2006) 906–915. [41] S. Klüsener, M. Aktas, K.M. Thormann, M. Wessel, F. Narberhaus, Expression and physiological relevance of Agrobacterium tumefaciens phosphatidylcholine biosynthesis genes, J. Bacteriol. 191 (2009) 365–374. [42] S. Klüsener, S. Hacker, Y.L. Tsai, J.E. Bandow, R. Gust, E.M. Lai, F. Narberhaus, Proteomic and transcriptomic characterization of a virulence-deficient phosphatidylcholinenegative Agrobacterium tumefaciens mutant, Mol. Genet. Genomics 283 (2010) 575–589. [43] D.J. Comerci, S. Altabe, D. de Mendoza, R.A. Ugalde, Brucella abortus synthesizes phosphatidylcholine from choline provided by the host, J. Bacteriol. 188 (2006) 1929–1934. [44] R. Conde-Alvarez, M.J. Grilló, S.P. Salcedo, M.J. de Miguel, E. Fugier, J.P. Gorvel, I. Moriyón, M. Iriarte, Synthesis of phosphatidylcholine, a typical eukaryotic phospholipid, is necessary for full virulence of the intracellular bacterial parasite Brucella abortus, Cell. Microbiol. 8 (2006) 1322–1335. [45] P.J. Wilderman, A.I. Vasil, W.E. Martin, R.C. Murphy, M.L. Vasil, Pseudomonas aeruginosa synthesizes phosphatidylcholine by use of the phosphatidylcholine synthase pathway, J. Bacteriol. 184 (2002) 4792–4799. [46] O. Geiger, V. Röhrs, B. Weissenmayer, T.M. Finan, J.E. Thomas-Oates, The regulator gene phoB mediates phosphate stress-controlled synthesis of the membrane lipid diacylglyceryl-N, N, N-trimethylhomoserine in Rhizobium (Sinorhizobium) meliloti, Mol. Microbiol. 32 (1999) 63–73. [47] I.M. López-Lara, J.L. Gao, M.J. Soto, A. Solarez-Pérez, B. Weissenmayer, C. Sohlenkamp, G.P. Verroios, J. Thomas-Oates, O. Geiger, Phosphorus-free lipids of Sinorhizobium meliloti are not required for the symbiosis with alfalfa but contribute to increased yields under phosphorus-limiting conditions of growth, Mol. Plant Microbe Interact. 18 (2005) 973–982. [48] M. Zavaleta-Pastor, C. Sohlenkamp, J.L. Gao, Z. Guan, R. Zaheer, T.M. Finan, C.R.H. Raetz, I.M. López-Lara, O. Geiger, Sinorhizobium meliloti phospholipase C required for lipid remodeling during phosphorus limitation, Proc. Natl. Acad. Sci. U. S. A. 107 (2010) 302–307. [49] B. Weissenmayer, O. Geiger, C. Benning, Disruption of a gene essential for sulfoquinosyldiacylglycerol biosynthesis in Sinorhizobium meliloti has no detectable effect on root nodule symbiosis, Mol. Plant Microbe Interact. 13 (2000) 666–672. [50] E.A. Devers, V. Wewer, I. Dombrink, P. Dörmann, G. Hölzl, A processive glycosyltransferase involved in glycolipid synthesis during phosphate deprivation in Mesorhizobium loti, J. Bacteriol. 193 (2011) 1377–1384. [51] E. Krol, A. Becker, Global transcriptional analysis of the phosphate starvation response in Sinorhizobium meliloti strains 1021 and 2011, Mol. Genet. Genomics 272 (2004) 1–17. [52] M. Osterås, E. Boncompagni, L.T. Smith, M.C. Poggi, D. Le Rudulier, Presence of a gene encoding choline sulfatase in Sinorhizobium meliloti bet operon: choline-O-sulfate is metabolized into glycine betaine, Proc. Natl. Acad. Sci. U. S. A. 95 (1998) 11394–11399. [53] K.J. Miller, R.S. Gore, A.J. Benesi, Phosphoglycerol substituents present on the cyclic β-1,2-glucans of Rhizobium meliloti 1021 are derived from phosphatidylglycerol, J. Bacteriol. 170 (1988) 4569–4575. [54] P. Wang, C. Ingram-Smith, J.A. Hadley, K.J. Miller, Cloning, sequencing, and characterization of the cgmB gene of Sinorhizobium meliloti involved in cyclic β-glucan biosynthesis, J. Bacteriol. 181 (1999) 4576–4583. [55] X.S. Zhang, H.P. Cheng, Identification of Sinorhizobium meliloti early symbiotic genes by use of a positive functional screen, Appl. Environ. Microbiol. 72 (2006) 2738–2748. [56] H. Goldfine, Bacterial membranes and lipid packing theory, J. Lipid Res. 25 (1984) 1501–1507. [57] L. Vigh, P.V. Escriba, A. Sonnleitner, M. Sonnleitner, S. Piotto, M. Mareca, I. Horváth, J.L. Harwood, The significance of lipid composition for membrane activity: new concepts and ways of assessing function, Prog. Lipid Res. 44 (2005) 303–344.