Previous Article | Next Article ![]()
Journal of Bacteriology, January 2008, p. 571-580, Vol. 190, No. 2
0021-9193/08/$08.00+0 doi:10.1128/JB.01423-07
Copyright © 2008, American Society for Microbiology. All Rights Reserved.

Christian Sohlenkamp,2,
Meriyem Aktas,1
Otto Geiger,2 and
Franz Narberhaus1*
Lehrstuhl für Biologie der Mikroorganismen, Ruhr-Universität Bochum, Bochum, Germany,1 Centro de Ciencias Genómicas, Universidad Nacional Autónoma de México, Cuernavaca, Morelos, México2
Received 3 September 2007/ Accepted 26 October 2007
|
|
|---|
|
|
|---|
Prokaryotes were originally believed to use only the methylation pathway for PC biosynthesis (43). Indeed, mutants of Rhodobacter sphaeroides and Zymomonas mobilis deficient in Pmt were unable to form PC (1a, 41). However, an alternative pathway in which choline is directly condensed with CDP-diacylglycerol (CDP-DAG) by the PC synthase (Pcs) was later discovered in the legume symbiont Sinorhizobium meliloti (38). Similar activities have been demonstrated in Pseudomonas aeruginosa, Rhizobium leguminosarum, Bradyrhizobium japonicum, Mesorhizobium loti, Legionella pneumophila, Agrobacterium tumefaciens, and Brucella abortus (4, 26, 45, 46). Most intriguingly, evidence for a eukaryotic-like CDP-choline pathway was found in Treponema denticola (23), indicating that prokaryotic PC biosynthesis is very diverse.
The lipid composition of biological membranes defines their stability and integrity (34). It undergoes adaptational changes in response to environmental conditions, such as oxygen tension, temperature, salinity, or medium composition (27, 42). Recently, a fundamental role of PC during infection of eukaryotic hosts has been recognized. A B. japonicum mutant with decreased PC levels was unable to establish an efficient symbiosis with its soybean host, although vegetative growth was largely unaffected (30). An S. meliloti mutant unable to form PC failed to establish a proper symbiosis with its host (8, 39). PC is also required for virulence of the plant pathogen A. tumefaciens (45). Important human pathogens such as P. aeruginosa, L. pneumophila, Borrelia burgdorferi, and Brucella melitensis are also known to possess PC (26, 44, 46). Interestingly, it has been reported recently that PC is necessary for the full virulence of Brucella abortus (4, 5). This clearly supports the earlier hypothesis that PC might play an important role in many host-microbe interactions (14, 39).
Since a B. japonicum pmtA mutant was still able to produce reduced amounts of MMPE, DMPE, and PC, at least one additional Pmt (PmtX) was predicted in this organism (30). In the present study, we set out to determine the complete PC biosynthesis inventory of B. japonicum. We describe a Pcs and multiple Pmt enzymes, that have distinct substrate specificities.
|
|
|---|
|
View this table: [in a new window] |
TABLE 1. Bacterial strains and plasmids used in this study
|
Homology searches were performed by using the National Center for Biotechnology Information's BLAST network server (http://www.ncbi.nlm.nih.gov/BLAST/).
Construction of pmt and pcs mutants. Construction of the pmtA deletion mutant has been described (30). In order to construct a pmtX1 mutant by a gene disruption strategy, an internal fragment comprising nucleotides 38 to 477 was PCR amplified using the oligonucleotides Sig238 and Sig239 (for oligonucleotide sequences, see Table 2). The resulting 440-bp StuI-Ecl136II fragment was directly cloned into the suicide vector pSUP202pol4 linearized with SmaI to yield plasmid pRJ5295 (see Fig. 3B).
|
View this table: [in a new window] |
TABLE 2. Oligonucleotides used in this study
|
![]() View larger version (15K): [in a new window] |
FIG. 3. Physical and genetic maps of the B. japonicum pmtA (A), pmtX1-4 (B to E), and pcs (F) gene regions. Physical maps are given for SalI, StuI, Ecl136II, and PstI (indicated by S, St, E, and P, respectively). Black arrows indicate the position of chromosomally integrated translational lacZ fusions and point to the corresponding strain designation. Below each map, the insertion sites for antibiotic resistance cassettes (Km in the case of pmtA and pmtX2-4; Sm/Sp in the case of pcs) or the plasmid integration site are depicted. Antibiotic resistance cassettes are not drawn to scale. Km, kanamycin; Sm, streptomycin; Sp, spectinomycin; dnaJ, chaperone protein DnaJ; pmtA, Pmt; blr0682, putative oxidoreductase; pyrF, orotidine-5'-monophosphat-decarboxylase; blr0684, blr6996, bll6993, bll6633, blr8167, bll4801, and bll4584, hypothetical ORFs; trmU, tRNA (5-methylaminomethyl-2-thiouridylate)-methyltransferase; pmtX1, phosphatidylethanolamine N-methyltransferase (EC 2.1.1.17); blr6636, ATP synthase subunit; pmtX2, phosphatidylethanolamine N-methyltransferase (EC 2.1.1.17); bll6635, probable glycosyl transferase; bll8168, Na+/H+ antiporter; pmtX3, Pmt; bll8165, putative alkaline phosphatase; blr4802, two-component response regulator; blr4803, two-component hybrid sensor and regulator; pmtX4, Pmt; pyrG, CTP synthetase; blr4586, putative oxidoreductase protein; pcs, Pcs; qor, quinone oxidoreductase.
|
In order to delete pmtX3, a 1.8-kb DNA fragment carrying the gene was PCR amplified by using primer pair O1-O2. A 1.7-kb EcoRI-NotI fragment from the amplification product was cloned into vector plasmid pBCSK(+). Subsequently, a 249-bp PstI fragment of the pmtX3 coding region was replaced by a 1.2-kb PstI kanamycin cartridge from pBSL86 (see Fig. 3D). The 2.6-kb EcoRI-NotI inserts of the resulting plasmids were ligated into the 6.7-kb EcoRI-NotI vector fragment of pSUP202pol4, yielding plasmids pBO203 and pBO204.
The pmtX4 deletion constructs were cloned as described for pmtX2. The up- and downstream regions (418 and 520 bp, respectively) of pmtX4 were amplified with the primer pairs O63/O64 and O82/O83.
To delete the pcs gene, up- and downstream regions (704 and 662 bp, respectively) were PCR amplified with the primer pairs Sig241-Sig242 and Sig243-Sig244. After PCR products were cloned in pUC18, the 2-kb BamHI fragment of pHP45::
containing a Sm/Sp resistance cassette was inserted in the same orientation as the pcs gene. Finally, the 3.4-kb EcoRI-PstI insert was cloned into pSUP202pol4, yielding pRJ5702.
Finally, the suicide plasmids were mobilized from E. coli S17-1 into B. japonicum 110spc4 for marker replacement mutagenesis as described previously (16). The correct genomic structure of all mutant strains was confirmed by Southern blot analysis of genomic DNA digested with appropriate restriction enzymes.
Construction of chromosomally integrated, translational 'lacZ fusions. For the construction of 'lacZ fusion strains, suitable translational lacZ fusions present on pSUP202pol4 derivates were integrated into the chromosome of wild-type B. japonicum via homologous recombination (see Fig. 3). The correct genomic structure of all mutant strains was confirmed by Southern blot hybridization with appropriate digoxigenin-11-dUTP-labeled DNA probes.
β-Galactosidase assays. The β-galactosidase activity of B. japonicum cells grown aerobically in PSY complex medium or anaerobically in YEM medium was measured according to standard protocols (29).
In vivo labeling of B. japonicum with [14C]acetate and quantitative analysis of lipid extracts. The lipid compositions of B. japonicum wild type and mutant strains were determined after labeling with [1-14C]acetate. Cultures (1 ml) in PSY were inoculated from precultures grown in the same medium. After the addition of 0.5 µCi of [1-14C]acetate (56.0 mCi/mmol) to each culture, the cultures were incubated for 48 h. Cells were harvested by centrifugation, washed with 500 µl of water, and resuspended in 100 µl of water. Lipid extraction was done according to the method of Bligh and Dyer (2). The chloroform phase was used for lipid analysis on HPTLC silica gel 60 plates (Merck, Darmstadt, Germany). After one-dimensional separation with n-propanol-propionic acid-chloroform-water (3:2:2:1) as the running solvent the individual lipids were quantified by using a Bio-Imager (Fujifilm Europe GmbH, Düsseldorf, Germany) and the Aida Image Analyzer v. 4.03 software (Raytest GmbH, Straubenhardt, Germany).
Cloning and expression of the putative pmt genes of B. japonicum in E. coli. Candidate open reading frames (ORFs) pmtX1, pmtX2, and pmtX3 were PCR amplified by using oligonucleotide primers that introduced NdeI/BamHI restriction sites (Table 2) and were cloned into the expression plasmid pET9a (40), resulting in plasmids pCCS20, pCCS36, and pCCS37, respectively. Candidate ORF pmtX4 was PCR amplified by using oligonucleotide primers that introduced NdeI/HindIII restriction sites and cloned into pET24b (40), resulting in plasmid pBO234. Expression plasmids were transformed into the E. coli expression strain BL21(DE3)/pLysS. Cultures were grown in LB medium containing the appropriate antibiotics. At a density of 5 x 108 cells/ml, IPTG (isopropyl-β-D-thiogalactopyranoside) was added to a final concentration of 200 µM. Aliquots of 1 ml were transferred to sterile culture tubes and after 1 h of induction 0.5 µCi of [1-14C]acetate (56.0 mCi/mmol; Amersham) or [14C-methyl]methionine (40 to 60 mCi/mmol; Amersham) were added to each culture. One hour later the cells were harvested and extracted as described above.
To coexpress pmtA from B. japonicum with the candidate ORFs in E. coli, the expression cassette containing the T7 promoter and the pmtA gene was subcloned from pTB2117 (30) as a BamHI/BglII fragment into the broad-host-range plasmid pRK404 (9), resulting in plasmid pCCS27. E. coli BL21(DE3)/pLysS was transformed with pCCS27 and one of the plasmids pCCS20, pCCS36, pCCS37, or pBO234. Growth of the cells, induction, labeling, and extraction of the lipids was done as described above.
To coexpress pmtA from B. japonicum with two or three candidate ORFs in E. coli, new constructs had to be made: the pmtX3 gene was subcloned as a BamHI/BglII fragment from pCCS37 into the BamHI site of pCCS27, resulting in plasmid pCCS67, containing pmtA and pmtX3 both under control of the T7 promoter. The pmtX1 gene was subcloned as a BamHI/SphI fragment from pCCS20 into the BglII/SphI-digested plasmid pBO234, resulting in plasmid pCCS115, containing pmtX1 and pmtX4 under the control of the T7 promoter. Growth of the cells, induction, labeling, and lipid extraction were done as described above.
Plant infection tests. Surface sterilization of soybean seeds (G. max L.Merr cv. Williams 82), inoculation, plant cultivation, and acetylene reduction assays were done as described previously (15, 16). For determination of specific acetylene reduction activity (expressed as µmol of ethylene formed/h/g [dry weight] of nodule) all nodules from each plant were collected, dried overnight at 80°C, and weighed.
|
|
|---|
![]() View larger version (12K): [in a new window] |
FIG. 1. Unrooted phylogenetic tree of selected bacterial Pmt enzymes. The sequences used for the construction of the tree include the B. japonicum Pmt proteins (Bjap_PmtA, Blr0681; Bjap_PmtX1, Bll6994; Bjap_PmtX2, Bll6634; Bjap_PmtX3, Bll8166; Bjap_PmtX4, Blr4804) (19), the PmtA proteins from S. meliloti (Smel_PmtA; accession number AF201699) and R. sphaeroides (Rsph_PmtA, L07247), PmtA from A. tumefaciens (Atum PmtA, AE009001), a Pmt-like ORF from the genome of Rhodobacter capsulatus (Rcaps RRC03911; Ergo-light database; http://www.ergo-light.com/), and one from the genome of M. loti (Mlot Mlr5374, BA000012). The bradyrhizobial Pmt enzymes are highlighted by surrounding black boxes. The dashed line indicates the separation of the different types of Pmt enzymes. The tree was constructed by using the program CLUSTAL X (ftp://ftp-igbmc.u-strasbg.fr/pub/). Distances between sequences are expressed as 0.1 changes per amino acid residue.
|
To summarize, our bioinformatic analysis revealed that in addition to the already-known pmtA gene the B. japonicum genome contains four ORFs coding for putative Pmts. To make PC biosynthesis in this organism even more complex, its genome also contains a pcs gene encoding a Pcs (26).
Expression of the individual pmt genes in E. coli. Analysis of B. japonicum mutants deficient in the pmtA gene and heterologous expression of the B. japonicum pmtA in E. coli indicated that PmtA appears to be responsible mainly for the first methylation step in PC biosynthesis (30). Additional Pmts should therefore carry out the subsequent methylation steps converting MMPE into DMPE and PC. To detect this activity, we expressed PmtA, PmtX1, PmtX2, PmtX3, and PmtX4 in E. coli BL21(DE3)/pLysS. Lipids were radiolabeled by [14C]acetate or [methyl-14C]methionine, and the lipid composition of the strains was analyzed by using thin-layer chromatography (TLC) (Fig. 2). The amount of [14C]methionine-labeled lipids was quantified (Table 3). E. coli does not have Pmts and therefore only produces the membrane lipids PE, phosphatidylglycerol (PG), and cardiolipin. Expression of pmtA from B. japonicum led to the formation of mainly MMPE and minor amounts of DMPE (Fig. 2A and Table 3) as described earlier (30). If at all, only minute amounts of methylated PE derivatives were formed when PmtX1 or PmtX2 were expressed in E. coli. When PmtX3 was expressed, MMPE and DMPE were produced in almost equal amounts. Expression of PmtX4 led to the formation of large quantities of MMPE and only small amounts of DMPE.
![]() View larger version (26K): [in a new window] |
FIG. 2. Lipid formation after heterologous expression of pmtA from B. japonicum and four genes encoding putative Pmt enzymes from B. japonicum in E. coli. E. coli BL21(DE3)/pLysS cells expressing the indicated enzymes were labeled with [14C]acetate or [methyl-14C]methionine during growth on LB medium. Panel A shows the resulting phospholipid profile after single expression of (putative) Pmts in E. coli: panels B and C show the resulting phospholipid profile after coexpression of the phospholipid methyltransferase PmtA from B. japonicum with ORFs encoding putative Pmts in E. coli. After extraction, lipids were separated by one-dimensional TLC. Lane M, marker (S. meliloti lipids labeled with [14C]acetate); lane C, control with BL21(DE3)/pLysS/pET9a (+pRK404 for panels B and C); lane A, pmtA; lane X1, pmtX1; lane X2, pmtX2; lane X3, pmtX3; and lane X4, pmtX4 (genes encoded in the indicated vector backbone [see also Materials and Methods and Table 1]); lane Bj, B. japonicum lipids labeled with [14C]acetate. The lipids PG, cardiolipin (CL), PE, MMPE, DMPE, and PC are indicated.
|
|
View this table: [in a new window] |
TABLE 3. Formation of methylated derivatives of PE upon expression of genes encoding putative Pmts from B. japonicum in E. colia
|
Wild-type B. japonicum grown in complex PSY medium produces mainly PC and only small amounts of MMPE and DMPE (30). In an attempt to simulate this lipid composition heterologously, we coexpressed two, three, or four different Pmts from B. japonicum in E. coli. However, the expression of PmtA and PmtX1 caused the maximal formation of DMPE and PC methylation products in E. coli hosts. The relative amounts of DMPE and PC could not be increased by expressing additionally PmtX3 or PmtX3 and PmtX4 (Fig. 2C), indicating that PmtX3 and PmtX4 have no important role for overall PC formation.
Expression of pmt genes and the pcs gene in B. japonicum. In order to understand the complexity of PC biosynthesis in B. japonicum, we studied the expression of all potential PC biosynthesis enzymes using the corresponding reporter gene fusions to these and some flanking genes (Fig. 3). We first measured β-galactosidase activity during aerobic growth in complex medium. The reporter strains harboring chromosomally integrated, translational lacZ fusions were assayed 24, 48, and 72 h after inoculation to an optical density at 600 nm (OD600) of 0.02, which corresponds to early-exponential (OD600 = 0.3), late-exponential (OD600 = 1.0), and stationary-phase (OD600 = 1.2) cultures, respectively. B. japonicum wild type, which does not encode an endogenous β-galactosidase, served as a negative control.
Neither pmtX2, pmtX3, and pmtX4 nor the genes potentially forming an operon with pmtX2 and pmtX3 (bll6635 and bll8165, respectively) were expressed in B. japonicum (Fig. 4). This finding was confirmed by primer extension analysis (data not shown) and by recently deposited microarray data in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/; accession number GSE8478). Consistent with our previous primer extension analysis (30), significant expression between 200 and 300 MU was detected for the pmtA'-'lacZ fusion. The only other expressed Pmt gene was pmtX1, with approximately 10 MU throughout the growth curve. The trmU gene, a potential tRNA methyltransferase that is located upstream of pmtX1 (Fig. 3B) shows about 50 MU (Fig. 4). The pcs gene is only weakly (3 to 4 MU) expressed.
![]() View larger version (10K): [in a new window] |
FIG. 4. Expression of B. japonicum PC biosynthesis genes. The expression of chromosomally integrated, translational lacZ reporter gene fusions (see Fig. 3) was measured as described previously (see Materials and Methods). Samples were collected after 24, 48, and 72 h. The results of a typical experiment, in which two parallel cultures of each strain were grown and assayed in duplicate, are shown. Four independent experiments were performed with individual strains.
|
Construction and phospholipid profiling of B. japonicum PC biosynthesis mutants. To examine the phenotypic importance of each putative PC biosynthesis gene, we constructed a series of individual deletion mutants (Fig. 3). Despite numerous attempts, the pmtX1 gene could not be inactivated by a simple plasmid insertion strategy.
The lipid composition of B. japonicum wild type and all obtained PC biosynthesis mutants was examined after in vivo labeling with [1-14C]acetate. The wild-type strain possesses PE (52.2% of total phospholipids) and PC (42.4%) as major phospholipids and contains only small amounts of MMPE (2.3%) and DMPE (1.8%) in the membrane (Fig. 5). Except for the pmtA mutant, all other strains showed a wild-type-like membrane composition, as one might expect from the expression data described above. The pmtA mutant, on the other hand, showed a marked decrease of PC and a concomitant increase of PE in the membrane (30.4 and 63.7% of total phospholipids, respectively), The MMPE and DMPE levels remained unchanged (2.5 and 1.8%).
![]() View larger version (37K): [in a new window] |
FIG. 5. Phospholipid profile of PC biosynthesis mutants after in vivo labeling with [14C]acetic acid. B. japonicum strains were labeled with [14C]acetate during growth in PSY medium. After extraction, lipids were separated by one-dimensional TLC. The lipids PC, DMPE, MMPE, and PE are indicated. The relative PC content (as the percentage of total phospholipids) is given below each PC spot.
|
![]() View larger version (28K): [in a new window] |
FIG. 6. Expression of the PC biosynthesis genes in the pmtA mutant compared to the wild type. The expression of chromosomally integrated, translational lacZ reporter gene fusions (see Fig. 3) was measured as described previously (see Materials and Methods). Samples were collected after 24, 48, and 72 h. The results of a typical experiment, in which two parallel cultures of each strain were grown and assayed in duplicate, are shown. Four independent experiments were performed with individual strains.
|
![]() View larger version (13K): [in a new window] |
FIG. 7. Symbiotic phenotype of B. japonicum pmt and pcs mutants. The nitrogen fixation activity was measured as the amount of C2H2 reduced per hour per gram (dry weight) of nodule. A 100% wild-type (WT) activity corresponds to 93.7 µmol of C2H2/h/g. Mean values and standard deviations derive from at least eight individual plants, except for the pcs mutant with six individual plants.
|
|
|
|---|
Why are there so many pmt genes in some organisms but not in others? We studied the complexity of prokaryotic PC biosynthesis using B. japonicum as a model organism. Apart from PmtA, at least one additional enzyme is required since PmtA alone produces predominantly MMPE. PmtX1 was shown to be responsible for converting MMPE via the DMPE intermediate to PC. Only pmtA and pmtX1 genes were significantly expressed in B. japonicum under all conditions tested. However, expression experiments with pmtX2, pmtX3, or pmtX4 in E. coli show that at least PmtX3 and PmtX4 function as Pmt enzymes. The fact that B. japonicum retains these Pmts intact suggests that there is some physiological demand for them. Otherwise, cryptic genes tend to accumulate missense mutations, as was shown for a member of the small heat shock gene family in Bradyrhizobium sp. (Parasponia) (33). B. japonicum is notorious for encoding multigene families such as the sigma factors RpoH1 to RpoH3 (32), the groESL-like chaperonin family (13), small heat shock proteins (31), and the transcription activators FixK1 and FixK2 (28). Members of these families often have overlapping functions but differ in the way they are regulated.
Little is known about regulation of PC biosynthesis in prokaryotes. It is unclear under which conditions, if at all, pmtX2, pmtX3, and pmtX4 are expressed in B. japonicum. In contrast to the pmtA mutant, deletion of these genes did not confer a symbiotic defect, indicating that they do not play an essential role in plant-microbe interaction under the conditions tested. Accordingly, anaerobic conditions did not induce expression of the pmtX genes. Expression of pmtA was affected by the medium composition rather than the oxygen supply. This seems to contradict results from a previous study showing a decrease in PC levels under anaerobic conditions and suggesting a possible oxygen-dependent regulation of PC biosynthesis gene expression (42). However, both the medium ingredients and the cultivation conditions were different in both studies. There is more evidence that external conditions and the nutrient supply influence PC biosynthesis in B. japonicum. The amount of pmtA transcripts was reduced three- to fourfold under heat shock conditions (30). Recent experiments with Bradyrhizobium SEMIA 6144 showed that PC amounts were decreased by heat stress and increased with salinity (27). Finally, a microarray analysis of an irr mutant lacking the iron-responsive regulator Irr showed that pmtX1 (and its upstream trmU gene) was downregulated approximately threefold under iron-limited conditions compared to the parental strain (47). This result was reproduced with our pmtX1'-'lacZ fusion (data not shown). Interestingly, transcription of the gene for the phosphatidylglycerophosphate synthase (pgsA), catalyzing the first reaction from CDP-DAG to PG, was decreased 2.6-fold under the same conditions (47). These findings suggest a possible regulatory connection between iron availability and phospholipid biosynthesis in B. japonicum.
Although it has been reported that B. japonicum is unable to take up choline (3), it expresses a functional Pcs, which requires choline as precursor. A previous study already showed Pcs activity in cell extracts of B. japonicum (26). Expression of the pcs gene was low but significant under all conditions tested. As speculated for A. tumefaciens and B. abortus (5, 20), B. japonicum might be able to synthesize choline.
On the basis of our analysis we propose an updated model for PC biosynthesis in B. japonicum (Fig. 8). The previously postulated Pmt activity catalyzing the second and third methylation step was shown to derive from PmtX1. PC biosynthesis is achieved by the concerted action of PmtA and PmtX1. Although PmtX4 and PmtX3 are able to perform the initial one or two methylation reactions when expressed in E. coli, their role in B. japonicum is not clear. The conditions under which they come into play remain to be elucidated. In an alternative route, PC can be synthesized in a one-step reaction by the Pcs pathway. One interesting avenue for future studies will be to study the mechanistic differences between the various Pmts biochemically, especially in cases in which closely related sequences produce different product spectra.
![]() View larger version (16K): [in a new window] |
FIG. 8. Model of PC biosynthesis in B. japonicum. The model is based on the phospholipid profiles after expression of each enzyme in E. coli and integrates the corresponding gene expression data from B. japonicum grown aerobically in complex culture medium. Thick arrows and boldface letters indicate the predominant reaction(s) performed by each enzyme. Enzymes in brackets are not expressed in B. japonicum wild type but are functional when expressed in E. coli. SAH, S-adenosylhomocysteine.
|
A fellowship from the Allgemeines Promotionskolleg der Ruhr-Universität Bochum to M.A. and funding by the German Research Foundation (DFG SFB 480) to F.N. are gratefully acknowledged. This research was supported by grants from CONACyT-Mexico (46020-N and 42578-Q), DGAPA/UNAM (IN217907) and the Howard Hughes Medical Institute (HHMI 55003675).
Published ahead of print on 9 November 2007. ![]()
S.H. and C.S. contributed equally to this study. ![]()
|
|
|---|
3)-β-D-glucan (curdlan). J. Bacteriol. 184:4114-4123.This article has been cited by other articles:
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Copyright © 2009 by the American Society for Microbiology. For an alternate route to Journals.ASM.org, visit: http://intl-journals.asm.org | More Info»