Effects of growth phase and the developmentally significant bldA-specified tRNA on the membrane-associated proteome of Streptomyces coelicolor

Dae-Wi Kim1,2, Keith F. Chater1, Kye-Joon Lee2 and Andy Hesketh1

1 Department of Molecular Microbiology, John Innes Centre, Norwich Research Park, Colney, Norwich NR4 7UH, UK
2 School of Biological Sciences, Seoul National University, Seoul 151-742, Republic of Korea

Correspondence
Andy Hesketh
andrew.hesketh{at}bbsrc.ac.uk


   ABSTRACT
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Previous proteomic analyses of Streptomyces coelicolor by two-dimensional electrophoresis and protein mass fingerprinting focused on extracts from total cellular material. Here, the membrane-associated proteome of cultures grown in a liquid minimal medium was partially characterized. The products of some 120 genes were characterized from the membrane fraction, with 70 predicted to possess at least one transmembrane helix. A notably high proportion of ABC transporter systems was represented; the specific types detected provided a snapshot of the nutritional requirements of the mycelium. The membrane-associated proteins did not change very much in abundance in different phases of growth in liquid minimal medium. Identification of gene products not expected to be present in membrane protein extracts led to a reconsideration of the genome annotation in two cases, and supplemented scarce information on 11 hypothetical/conserved hypothetical proteins of unknown function. The wild-type membrane proteome was compared with that of a bldA mutant lacking the only tRNA capable of efficient translation of the rare UUA (leucine) codon. Such mutants are unaffected in vegetative growth but are defective in many aspects of secondary metabolism and morphological differentiation. There were a few clear changes in the membrane proteome of the mutant. In particular, two hypothetical proteins (SCO4244 and SCO4252) were completely absent from the bldA mutant, and this was associated with the TTA-containing regulatory gene SCO4263. Evidence for the control of a cluster of function-unknown genes by the SCO4263 regulator revealed a new aspect of the pleiotropic bldA phenotype.


Abbreviations: ASB14, tetradecanoylamidopropyldimethylammoniopropanesulphonate (amidosulphobetaine 14); DDM, n-dodecyl {beta}-D-maltoside; IPG, immobilized pH gradient; MALDI-TOF, matrix-assisted laser-desorption ionization time-of-flight; OGP, n-octyl {beta}-D-glucopyranoside; SB3-10, N-decyl-N,N'-dimethyl-3-ammonio-1-propanesulphonate (caprylyl sulphobetaine)


   INTRODUCTION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Uniquely among functional genomics technologies, proteomics offers the opportunity to determine the subcellular or extracellular fate of the products of expression of large numbers of genes. This potentially allows putative functions to be assigned to previously uncharacterized proteins, based on their cellular location.

Proteins targeted to (or across) the cytoplasmic membrane are likely to fulfil roles that require, and reflect, more intimate contact with the external environment of the cell than gene products remaining in the cytoplasm. Such proteins are expected to be more abundant and diverse in organisms that are adapted to cope with very variable environments. Bacteria of the genus Streptomyces, known for their importance as producers of antibiotics and other secondary metabolites, are a particularly good example of this. They are highly adapted to live in soil, which is a very diverse environment where organic matter consists largely of insoluble material accessible only to organisms with suitable extracellular enzymes and appropriate systems for their regulation, in turn often involving extracellular recognition components. Moreover, this environment is subject to great variation in such chemical and physical parameters as oxygen availability, hydration, pH and temperature. It was therefore no surprise that the genome of Streptomyces coelicolor turned out to be very complex (it was annotated to encode 7825 theoretical proteins), with close to a quarter of the genes predicted to encode membrane proteins or extracellular proteins (Bentley et al., 2002; SCO-DB at http://streptomyces.org.uk).

In streptomycetes, some aspects of stationary-phase biology have common control elements, as revealed by the finding that mutations in some genes have drastic effects on both antibiotic production and aerial reproductive growth. Among such bld genes (named because the colonies lack aerial growth and thus are ‘bald’), bldA is unusual: it specifies the only tRNA capable of translating efficiently the leucine codon UUA. Streptomyces DNA typically contains more than 70 mol% G+C, making TTA codons in the genome quite rare; indeed, the S. coelicolor genome annotation predicts that only 145 genes contain TTA codons and are therefore likely to be directly affected by bldA mutations. This number includes 16 encoding proteins likely to be associated with the membrane. The presence of TTA codons in 14 likely regulatory genes further increases the potential for bldA mutations to alter the range of proteins in this location.

In this study, we have partially mapped the membrane proteome of S. coelicolor and explored the influence of bldA on it. We have used two-dimensional (2D) gel electrophoresis coupled with MALDI-TOF peptide mass fingerprint analysis to display, identify and compare proteins extracted from membrane preparations obtained from liquid cultures of wild-type S. coelicolor and an isogenic bldA deletion mutant. Follow-up experiments using molecular genetics revealed a previously unknown bldA-dependent regulon.


   METHODS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Strains and growth conditions.
S. coelicolor M145 and M600 are independently derived, prototrophic, plasmid-free strains of S. coelicolor A3(2) (Kieser et al., 2000). (Note that strain M600 has a duplication of 1005 genes compared with the sequenced strain M145; Weaver et al., 2004.) In the M600 {Delta}bldA strain, kindly provided by M. Tao (National Key Laboratory of Agricultural Microbiology, Huazhong Agricultural University, Wuhan, 430070, P. R. China), bldA is completely replaced by an apramycin-resistance cassette. Strain M600 {Delta}SCO4263 contains a similar replacement of the SCO4263 gene (A. Hesketh and others, unpublished). Strains were cultivated with vigorous agitation at 30 °C in minimal medium supplemented with 0·2 % Casamino acids (SMM) as previously described by Kieser et al. (2000). Briefly, spores (about 1010 c.f.u. ml–1) were pre-germinated in 2xYT medium (Kieser et al., 2000) for 7 h at 30 °C. Germlings were harvested by centrifugation (5 min at 4000 g), resuspended in SMM, and briefly sonicated to disperse any aggregates, before inoculation into 50 ml SMM in 250 ml siliconized flasks containing coiled stainless steel springs. Each flask received the equivalent of 5x107 c.f.u. Growth curves for producing membrane-protein extracts to compare M600 and M600 {Delta}bldA strains were performed in duplicate, as were cultures for producing RNA for S1 nuclease mapping.

Preparation of membrane-protein fractions from cultures.
Mycelium for membrane preparation was harvested from cultures by brief centrifugation (30 s at 4000 g) at room temperature and immediately frozen in liquid nitrogen. Typically, mycelium from 50 ml culture samples was collected, with a transfer time from culture flask to frozen sample of 1·5 min. Mycelial pellets were stored at –80 °C until use. For preparation of membranes, frozen cells were thawed on ice in 5 ml washing buffer (40 mM Tris, pH 9·0, 1 mM EGTA, 1 mM EDTA) then pelleted by centrifugation (5 min at 3000 g) at 4 °C. Washed cells were resuspended in lysis buffer [wash buffer containing 5 mM DTT, 4 mM Pefabloc SC protease inhibitor and one-twentieth volume of a protease-free mixture of 0·25 mg RNase I ml–1 (Sigma R-5503) and 1 mg DNase I ml–1 (Fluka 31134)] and disrupted by sonication (Sanyo Soniprep 150; 10x2 s bursts at amplitude 7·5 µm), while cooling in an ethanol–ice bath. Cell debris was removed by centrifugation (15 min, 10 000 g, 4 °C), and membranes in the supernatant were pelleted by ultracentrifugation (30 min at 150 000 g) at 4 °C. Membrane pellets were subjected to a salt wash to remove non-specifically associated proteins by resuspending them in wash buffer containing 250 mM NaCl and 4 mM Pefabloc SC protease inhibitor, then repelleted by ultracentrifugation as above. Salt was subsequently removed by performing an additional wash step using the same buffer but lacking NaCl. Proteins in the washed membranes were extracted by resuspending in the appropriate IEF buffer (see below), and protein extracts stored at –80 °C until use. Mycelium harvested from 25 ml cultures typically yielded 500–1000 µg membrane protein.

Optimization of buffers for extraction and separation of membrane proteins.
Aliquots of a membrane-protein preparation were resuspended in the following IEF buffers, also containing 7 M urea, 2 M thiourea, 40 mM Tris, pH 9·0, 1 mM EDTA, 50 mM DTT and 4 mM Pefabloc SC protease inhibitor unless specifically stated: UTCHAPS [containing 4 % (w/v) CHAPS (Sigma)]; UT Triton X-100 [containing 4 % (w/v) Triton X-100 (BDH)]; UTCHAPS+SB3-10 (5 M urea, 2 % (w/v) CHAPS, 2 % (w/v) SB3-10 (Sigma)]; UTASB14 [1 % (w/v) ASB14 (Calbiochem)]; UTCHAPS+ASB14 [2 % (w/v) CHAPS, 2 % (w/v) ASB14]; UTCHAPS+octyl glucopyranoside (OGP) [2 % (w/v) CHAPS, 2 % (w/v) OGP (Sigma)]; UTCHAPS+dodecyl maltoside (DDM) [2 % (w/v) CHAPS, 2 % (w/v) DDM (Sigma)]. Samples were analysed by 2D gel electrophoresis using pH 4–7 immobilized pH gradient (IPG) strips as detailed below, in each case using the same buffer employed in the IEF step as had been used to dissolve the protein sample applied. Separations were performed at least twice, and their quality assessed visually.

Proteomics techniques.
Protein extracts were subjected to 2D gel electrophoresis as detailed by Hesketh et al. (2002). Samples were not spiked with representative cytoplasmic proteins prior to analysis. Briefly, separation in the first dimension was for 100 000 V h using 18 cm IPG strips, pH 4–7 or 6–11 (Amersham Biosciences) in a Phaser IEF unit (Genomic Solutions). Separation of focussed proteins in the second dimension used in-house fabricated 12·5 % SDS-PAGE gels and the Investigator 5000 vertical format system from Genomic Solutions. For identification of the types of proteins present in a S. coelicolor M145 membrane preparation, gels were stained with colloidal Coomassie G-250 (Neuhoff et al., 1988). For optimization of buffer composition for the separation of membrane proteins (see above) and for quantitative analysis of protein abundance profiles, gels were stained with Sypro-Ruby (Bio-Rad) according to the manufacturer's instructions, and scanned using the Perkin-Elmer ProXPRESS proteomic imaging system with excitation and emission wavelengths of 480 and 630 nm, respectively. To produce a quantitative analysis of protein abundance profiles, gel images were analysed using PHORETIX 2D version 5.1 (NonLinear Dynamics): spot detection was optimized automatically using the ‘spot detection wizard’ and then manually edited; background subtraction was performed automatically using the ‘mode of non-spot’ setting; images were then normalized to the total spot volume for each gel for quantification. Spot filtering was not used, although all spots were manually edited. Histograms of normalized spot volumes displaying changes in spot abundance during growth and between M600 and the {Delta}bldA mutant were generated with this software. Differences between strains were considered significant if normalized volumes for a particular spot were changed twofold or more in both biological duplicate growth curves in at least two time points of four.

Protein spots of interest were excised from colloidal Coomassie- or Sypro-Ruby-stained gels using the Investigator ProPic robot from Genomic Solutions, and identified by tryptic digestion and MALDI-TOF MS as previously described (Hesketh et al., 2002). Identification of proteins from peptide mass fingerprint data was performed using the ‘Mascot’ search engine at http://www.matrixscience.com and was based on their ‘probability based Mowse score’ algorithm. A Mowse score of 60 or higher is significant at the 5 % level or better, and proteins in this work typically gave scores >80 (frequently considerably so). In addition, no identification was accepted unless at least five peptides representing at least 20 % of the protein sequence were detected in the MALDI-TOF peptide mass fingerprint.

S1 nuclease mapping.
RNA was isolated from mid-exponential and early stationary phase cultures as described by Strauch et al. (1991). For each S1 nuclease reaction, 20 µg RNA was hybridized in NaTCA buffer (Murray, 1986) to about 0·2 pmol (approx. 105 Cerenkov counts min–1) of each of the following radiolabelled probes. For SCO4248 probe, the oligonucleotide 5'-CAGACGAAGCCGTTGTTGCCGC-3', which anneals within the SCO4248 coding region, was uniquely end-labelled at its 5'-end with [{gamma}-32P]ATP using T4 polynucleotide kinase. This was used in the PCR together with the unlabelled probe 5'-TGGTGGACAACCTGACCCGGCT-3' (which anneals upstream of the SCO4248 gene in the SCO4249 coding sequence) and cosmid SCD49 (Redenbach et al., 1996) as template to generate a 472 bp probe. PCR was performed in the presence of 7 % DMSO using the following conditions: 94 °C for 4 min followed by 26 cycles of 45 s at 94 °C, 45 s at 58 °C and 60 s at 72 °C, then held at 72 °C for 5 min to finish. The probe for the diverging gene SCO4249 was made in exactly the same way, but the primer that was end-labelled was swapped. The probe for SCO4254 was made in a similar way, but using 5'-TTGGAGAGCTGTGCGGTACCGGC-3' as the labelled primer (which anneals within the SCO4254 gene), and 5'-GACCGACGAAGGCCGCCACCGA-3' as the unlabelled primer (which anneals upstream of SCO4254 within the SCO4253 gene). This produced a 512 bp uniquely end-labelled probe product. The location of all the primers above are indicated on the sequences presented in the Results (see Fig. 4). The probe for SCO0762, 365 bp long and producing a protected fragment of 217 bp, was prepared as detailed in Kim et al. (2005). Hybridizations were carried out at 45 °C for 14 h after denaturation at 65 °C for 15 min. S1 nuclease digestions and analyses of RNA-protected fragments were as described by Janssen et al. (1989).



View larger version (46K):
[in this window]
[in a new window]
 
Fig. 4. (a) Gene organization in the chromosomal region of S. coelicolor encoding the bldA-dependent proteins SCO4244 and SCO4252. SCO numbers for ORFs are truncated to the last two digits. Solid arrows are above genes that are separated by <50 bp, and a broken arrow is above those separated by <140 bp. The arrowhead to the left of SCO4241 represents the threonine tRNA for the UGU codon. The circles with arrows mark two sets of divergent promoter regions that potentially direct transcription of four operons; sequences and features of these regions are shown above (SCO4248–49) and below (SCO4253–54) the central map. In these sequences the translational start codons are marked by block arrows; primers used in S1 mapping are underscored with arrows; transcription start points (p1 and p2) are indicated by arrows placed above the sequence [P1 for SCO4253 taken from A. Hesketh and others (unpublished)]; and DNA motifs 1 and 2 (detailed below) are underlined in bold. (b) Alignment of DNA consensus motifs present in the divergent promoter regions, consisting of a pair of highly conserved sequences 10 and 14 bp long (underlined and in bold) separated by an approximately 210 bp (209 and 212 bp) sequence that includes the start codon for the SCO4253 and SCO4248 genes. Note that the orientation of SCO4253 is reversed compared with that shown elsewhere, in order to show the similarity between the conserved sequences.

 

   RESULTS
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Optimization of the analysis of S. coelicolor membrane proteins using 2D gel electrophoresis
Because of their hydrophobicity, membrane proteins are particularly difficult to extract and solubilize using detergents that are compatible with IEF (i.e. detergents with no net charge). Nevertheless, among the range of non-ionic and zwitterionic detergents developed for analysing complex protein mixtures using 2D gel electrophoresis, several have been applied with some success to the separation of membrane proteins (Chevallet et al., 1998; Luche et al., 2003; Molloy et al., 2000; Nouwens et al., 2000). From these studies it is clear that no single extraction buffer recipe is optimal in all organisms, presumably reflecting differences in the lipid composition of their membranes. In order to determine the buffer conditions most suitable for analysing S. coelicolor membrane proteins, buffers with different detergent compositions were evaluated, using 2D gel electrophoresis, for their ability to extract and separate proteins from a membrane preparation (Fig. 1). Fig. 1(a) clearly shows that the IEF buffer containing 4 % CHAPS previously used to analyse the total proteome of S. ceolicolor (Hesketh et al., 2002) was not suitable for producing a good separation of membrane proteins. Replacing the CHAPS detergent with 4 % Triton X-100 improved the separation significantly (Fig. 1b), but a mixture of 2 % CHAPS and 2 % of the sulphobetaine detergent SB3-10 was optimal (Fig. 1c). ASB14, the amidosulphobetaine detergent used successfully in the 2D analysis of membrane proteins from Pseudomonas aeruginosa (Nouwens et al., 2000) and the outer membrane of Escherichia coli (Molloy et al., 2000), did not by itself produce good separation of S. coelicolor membrane proteins in our hands, but when used in a 2 % plus 2 % combination with CHAPS was approximately as effective as the buffer containing 2 % CHAPS and 2 % SB3-10 in Fig. 1(c). Combinations of CHAPS with either DDM or OGP produced similarly good separations. The subsequent experiments described here used UTCHAPS+SB3-10 or UTCHAPS+OGP, two of these equally effective systems.



View larger version (55K):
[in this window]
[in a new window]
 
Fig. 1. Improvement in the separation of S. coelicolor M145 membrane proteins by optimization of the buffer detergent composition. Aliquots of a membrane preparation were extracted with urea/thiourea IEF buffer containing 4 % CHAPS (a), 4 % Triton X-100 (b) or a mixture of 2 % CHAPS plus 2 % SB3-10 (c). 2D gel separations were then performed using the same buffer recipes as in the IEF step. Separations were made at least twice, and representative gels are shown.

 
Identification of the types of proteins present in a S. coelicolor membrane preparation
The strain used in the initial analysis was M145, a prototrophic plasmid-free derivative of the wild-type strain A3(2). This strain was chosen because it was the source of the cosmid library that had been used for the sequencing of the S. coelicolor genome. In some experiments, involving a bldA mutant, another plasmid-free prototrophic derivative of A3(2), named M600, was used, because it had not been possible to disrupt the bldA gene in M145 (M. Tao & K. F. Chater, unpublished).

Membranes were prepared from mycelium harvested from liquid cultures of M145 grown to early stationary phase in SMM. Proteins were extracted using UTCHAPS+SB3-10 buffer, and separated by 2D gel electrophoresis using pH 4–7 and 6–11 IPG strips. The most abundant protein spots were excised from gels and unambiguously assigned to genes using tryptic digestion and MALDI-TOF peptide mass fingerprint analysis. Identifications are listed in Table 1. Fig. 2 shows typical 2D gel separation profiles of S. coelicolor M600 membrane proteins over the pH 4–7 and 6–11 isoelectric point ranges, highlighting selected proteins identified in Table 1.


View this table:
[in this window]
[in a new window]
 
Table 1. Proteins identified in 2D gel separations of membrane protein extracts of S. coelicolor

 


View larger version (98K):
[in this window]
[in a new window]
 
Fig. 2. Representative 2D gel separations of membrane-protein extracts of S. coelcolor. Numbers correspond to protein spot identifications in Table 1.

 
Reassuringly, the majority of proteins identified in the membrane preparations possessed primary amino acid sequences predictive of their membrane localization (Table 1). Thus, of 115 proteins detected from M145, and an additional five from the M600/M600 {Delta}bldA experiments reported below, 26 were predicted to have lipid attachment sites, 29 possessed motifs characteristic of ATP-binding components of ABC transporters, and an additional 29 were predicted to contain one or two transmembrane helices (Table 1c). In all, 50 of the 120 proteins are believed to be involved in transport across the membrane. A further two proteins had predicted N-terminal secretion signals. The remaining 34 examples had no sequence-based predictions to account for their observed location. The most abundant spots on the gels (numbers 5, 8, 9, 14, 15, 17 and 20) all corresponded to predicted substrate-binding lipoproteins (represented in Fig. 3c, d). ATP-binding proteins from the corresponding transport systems of some of these lipoproteins (represented in Fig. 3b) were notably less abundant on the gels (i.e. spot numbers 4, 7, 10, 13 and 16).



View larger version (61K):
[in this window]
[in a new window]
 
Fig. 3. Time-course abundance profiles of selected proteins in membrane preparations of S. coelicolor M600 and M600 {Delta}bldA during growth in liquid minimal medium supplemented with 0·2 % Casamino acids (SMM). (a) Proteins significantly different between M600 and M600 {Delta}bldA. (b–d) Growth-stage-associated abundance of some ATP-binding proteins from ABC transport systems during growth (b); some substrate-binding lipoproteins from the ABC transport systems represented in (b), see (c); some substrate-binding lipoproteins from other ABC transport systems (d). The numbers 1 and 2 indicate duplicate experiments in which cultures were sampled at four time points, as illustrated in the stylized growth curve shown at the top right. Histogram bars are normalized spot intensities following staining with Sypro Ruby, arranged from left to right in the same order as the arrows in the growth curve. For each protein, the bar extending to the top of the display represents the greatest abundance observed.

 
Although extraction and display were successful for some membrane-associated proteins, proteins with predicted multiple transmembrane domains were not detected, consistent with previous reports that highly hydrophobic proteins cannot be separated using IPG strips (Nouwens et al., 2000; Santoni et al., 2000). An apparent exception in this study was SCO1806, which was represented on the 2D gels even though it is predicted to contain six transmembrane helices. However, this spot appeared at an observed molecular mass corresponding to approximately 50 % of its predicted value, and the MALDI-TOF data showed peptides only from the N-terminal region (aa 24–399; data not shown), suggesting that the SCO1806 spot was truncated from the C-terminal region containing all the predicted transmembrane helices.

Observed localization provides clues for annotation of gene function
There are no clues to the functions of ~30 % of the genes in the S. coelicolor genome that are annotated as encoding hypothetical or conserved hypothetical proteins (Bentley et al., 2002; SCO-DB at http://streptomyces.org.uk). Eleven such proteins were detected in this study, not only indicating that these putative genes of unknown function do indeed encode protein products but also assigning a subcellular localization to the proteins, which may assist in future assignment of function. In addition, we established that two of these hypothetical proteins, proteins with no close database homologues, depend on bldA (see below).

Four proteins annotated as being secreted were identified in the membrane preparations (listed in Table 1c), stimulating closer examination of their annotation. In the case of SCO0681, a protein that acts as a reductive partner to P450 cytochromes, the peptide mass fingerprint showed that the N terminus from the seventh amino acid residue of this protein was present (data not shown), ruling out cleavage of the putative signal sequence. Use of the SignalP tool (http://www.cbs.dtu.dk) showed that the prediction for a signal sequence was in fact not strong. The prediction for a signal peptide in SCO2837 was similarly poor. Probably both of these proteins have N-terminal transmembrane domains rather than being secreted. The SignalP prediction for SCO3184, a putative secreted penicillin acylase, was much more convincing, but the presence of an additional transmembrane helix at the C terminus suggested that this secreted protein would remain anchored to the membrane. Interestingly, only the C-terminal half of this protein was represented in the peptide mass fingerprint of the protein spot, and its apparent molecular mass was approximately 60 kDa compared with the predicted value of 102 kDa, suggesting that it corresponds to only the C-terminal half of the annotated protein sequence. SCO1517, a protein of unknown function, also possessed a convincing signal sequence, with predicted processing between amino acid residues 23 and 24. Consistent with this prediction, a peptide assigned to amino acid residues 27–34 was closest to the N terminus detected in the MALDI-TOF data for membrane-associated SCO1517. Detection of this protein in the membrane fraction therefore suggests that it remains associated with the membrane after export, and the location of the SCO1517 gene adjacent to the integral membrane protein secretion genes secD and secF could offer an explanation for this. None of the four proteins was detected in an analysis of extracellular proteins harvested from the culture supernatants of the mycelium used in this study (Kim et al., 2005), providing additional support for the arguments above.

Changes in the membrane proteome during the growth cycle, and the effects of mutation in bldA
S. coelicolor M600 and a bldA deletion mutant derived from it were grown in SMM, and mycelium was harvested by centrifugation at four comparable points during growth as illustrated in Fig. 3. Membrane fractions were prepared, and the proteins were extracted and separated using the UTCHAPS+OGP buffer system. [Proteins in the culture supernatants were also harvested by precipitation in TCA-acetone and used in an analysis of the extracellular proteome described elsewhere (Kim et al., 2005).] The protein composition of the membrane extracts was analysed by 2D gel electrophoresis, and differences in the spot abundance profiles between the two strains were identified by image analysis of scanned Sypro Ruby-stained gels. All the protein spots that showed at least twofold changes in abundance between the strains, together with a limited number of other spots whose abundance profiles were of interest, were identified using peptide mass fingerprint analysis. The results are presented in Fig. 3.

Four proteins are significantly different in abundance in the {Delta}bldA mutant.
The 2D gel spot abundance patterns obtained from the parent and mutant strains were very similar. Of 352 spots present on a composite master reference gel image created by the software, only four reproducible differences were observed between the two strains: two hypothetical proteins, SCO4244 and SCO4252, were completely absent from the membranes of the {Delta}bldA mutant; the abundance of a putative nucleotide-binding protein, SCO5249, was greatly reduced; and the accumulation profile of the putative lipoprotein SCO7399 was significantly increased (Fig. 3a).

Proteins from transport systems change in abundance during growth.
Protein spots that displayed interesting changes in abundance during the growth curves, but no significant differences as a result of {Delta}bldA mutation, included many lipoproteins and ATP-binding proteins from gene clusters predicted to be involved in the transport of small molecules across the membrane (Fig. 3b–d). For example, PstB and PstS, which are the ATP-binding protein and phosphate-binding lipoprotein components, respectively, of the phosphate transport system, both increased sharply in abundance in transition and stationary phase cultures. This may suggest that phosphate (present in SMM at 1 mM) was growth-limiting. Curiously, these proteins were not previously detected in cultures of M145, a different ‘wild-type’ derivative of S. coelicolor A3(2) (Hesketh et al., 2002), grown in the same medium. The observed accumulation of PstB and PstS during growth contrasted with the substrate-binding lipoprotein AtrA and its ATP-binding protein partner AtrC, which both tended to decrease in abundance, and the equivalent proteins GluB and GluA from the glutamate-uptake system, which remained approximately constant at all time points. The ATP-binding proteins SCO5115 and SCO5479 from putative oligopeptide-transport systems were most abundant in the mid-exponential phase sample, and decreased in abundance at later time points. However, their partner oligopeptide-binding lipoproteins remained approximately equally abundant throughout the growth curve, which could reflect differences in the stability of the respective proteins to proteolysis, or alternatively reflect the transcriptional regulation of these transport systems.

Three of the lipoproteins predicted to be involved in iron uptake displayed potentially interesting differences in their abundance profiles (Fig. 3a, d). SCO0494, believed to be responsible for the uptake of iron scavenged by the siderophore coelichelin, tended to increase in abundance throughout growth, while SCO2780, located adjacent to the desferrioxamine siderophore biosynthetic cluster, displayed an obvious decrease. The abundance profiles for SCO7399, a possible ferrichrome-binding lipoprotein, were less well-defined, although the protein was present at all time points. SCO6009 and SCO2008, putative substrate-binding lipoproteins predicted to be involved in sugar transport and amino acid transport, respectively, were also detected in all three phases of growth.

Further analysis of the suggested bldA-dependence of a cluster of genes of unknown function (SCO4242–58)
Two of the three membrane proteins downregulated in the bldA mutant, SCO4244 and SCO4252, are the products of genes located close to one another on the S. coelicolor chromosome (Fig. 4a). Neither of these genes contains TTA codons that could account for the bldA-dependence of their encoded proteins, but we have shown elsewhere that production of the three proteins SCO4251–3 depends on the nearby TTA-containing regulatory gene SCO4263, since deletion of SCO4263 abolished transcription from the putative promoter upstream of SCO4253, and the SCO4251–3 proteins could not be detected in the mutant strain (A. Hesketh and others, unpublished). Sequence analysis of the chromosomal region SCO4242 to SCO4258 using SCO-DB (http://streptomyces.org.uk) suggests that it is arranged in four operons controlled by two divergent promoter regions (Fig. 4a). The absence of both the SCO4244 and SCO4252 proteins in membranes of the bldA mutant strain raises the possibility that they are co-regulated, and this is supported by their similar abundance profiles in the parent strain (Fig. 3a). DNA sequence analysis using MEME (http://meme.sdsc.edu) revealed consensus motifs in each of the divergent promoter regions that could possibly serve as recognition sites for the regulatory protein SCO4263 (Fig. 4b). These motifs each consist of a pair of highly conserved sequences 10 and 14 bp long (designated motifs 1 and 2, respectively) separated by a segment of approximately 210 bp (209 and 212 bp) that includes the start codons of SCO4253 or SCO4248 in each divergent promoter region (Fig. 4b). This arrangement of motifs is not found anywhere else in the chromosome. In order to investigate the role of SCO4263 in regulating the transcription of other promoters in this region in addition to that already verified for SCO4253, S1 nuclease protection experiments were performed using probes for SCO4248, SCO4249 and SCO4254 (Fig. 5). In the M600 parent strain, transcription of SCO4254 from two promoters was readily detected in exponential phase cultures, with transcript abundance decreasing in the stationary phase (Fig. 5a). Transcription from promoter P2 (protected fragment approx. 270 bp) was weaker than from P1 (protected fragment approx. 300 bp), and was undetectable at the later time. The approximate locations of the transcription start sites are indicated in Fig. 4(a). [Note that the start point recently determined for SCO4253 transcription (A. Hesketh and others, unpublished) does not give rise to a transcript that would overlap with the SCO4254 transcripts.] Interestingly, the P1 site is likely to be located only 5–10 bp downstream of one of the DNA motif sequences (motif 1 in Fig. 4b). In the {Delta}SCO4263 mutant, SCO4254 transcription from both promoters was completely absent (Fig. 5a), while transcription of SCO0762, used as an internal control for RNA quality, was similar to that observed in M600 (Fig. 5c). Similar results were obtained for SCO4248, where two promoters were again observed, and both found to be dependent on SCO4263 (Fig. 5b). In this case P1 and P2 produced protected fragments approximately 340 and 280 bp long, respectively, and the P1 transcription start site mapped to the middle of DNA sequence motif 2 present in this divergent promoter region (Fig. 4). Transcription from SCO4249 could not be detected in either strain (data not shown).



View larger version (86K):
[in this window]
[in a new window]
 
Fig. 5. Transcription analysis of the divergent promoter regions SCO4248–9 and SCO4253–4. (a) Transcription of SCO4254 during growth of M600 (SCO4263+) and M600 {Delta}SCO4263 in SMM. RNA was isolated during the mid-exponential (E) and early stationary (S) phases of growth and subjected to S1 nuclease protection analysis using a uniquely end-labelled PCR-generated probe. (b) Analysis of SCO4248 transcription using the same RNA samples. (c) Analysis of SCO0762 transcription as an internal control for RNA quality, using the same RNA samples as above. S1 analyses were performed on duplicate biological samples, and representative results are shown.

 

   DISCUSSION
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
The membrane proteome of S. coelicolor
Bacterial membranes are selectively permeable and determine what molecules can enter and leave the organism. Water, dissolved gases and lipid-soluble small molecules simply diffuse across, but other molecules require specific protein-based transport systems. Important transactions that take place between the bacterial cell and its environment to ensure its survival include the uptake of peptides, sugars, amino acids and inorganic ions, and the export of proteins (often including exoenzymes) and toxic molecules (waste products and exogenously or endogenously produced bioactive molecules).

In a recent in silico analysis, Bertram et al. (2004) identified 81 ATP-binding cassette (ABC) permeases in the S. coelicolor genome. Although the permease proteins themselves, possessing multiple membrane-spanning domains, could not be detected by 2D gel electrophoresis, the substrate-binding lipoproteins and ATP-binding proteins identified in our work (see Table 1) represented components of 33 of the 81 different ABC transporter systems. This indicated that even in SMM about 40 % of the unusually large number of these systems were available for use. However, only five of them corresponded to the 45 carbohydrate transporters identified by Bertram et al. (2004), perhaps because the utilizable carbohydrate composition of the medium was very simple: it contained only glucose (though we cannot rule out the possibility that traces of carbohydrates might have been introduced along with the Casamino acids supplement). Presumably, many carbohydrate transport systems are inducible by their substrates and/or repressed by glucose and probably the five observed in this study are not subject to these forms of control. On the other hand, the presence of proteins from six of the nine ABC systems predicted to be involved in amino acid transport was consistent with the Casamino acids supplementation. In addition, evidence for two of seven systems predicted to be responsible for the import of oligopeptides from the environment was obtained (Table 1), two of the most abundant spots present in the 2D gels being the oligopeptide-binding lipoproteins SCO5113 and SCO5477 (spot numbers 8 and 17, respectively, in Fig. 2).

Five of the eight systems annotated as being involved in the import of iron were detected. The multiplicity of these systems presumably reflects the importance to S. coelicolor of obtaining this metal from its natural soil environment, and the high proportion of the available iron-uptake proteins detected may indicate that iron is limiting under the growth conditions used (FeSO4 was added as part of the trace elements solution, but was probably largely oxidized to the relatively inaccessible Fe3+ form). The abundance patterns of different iron-uptake lipoproteins showed interesting diversity (see SCO0494 and SCO2780 in Fig. 3d), suggesting that desferrioxamine-mediated iron uptake is more prevalent in conditions of rapid growth and/or low cell-density, while coelichelin-mediated uptake may be more prevalent in stationary phase/high cell-density. Challis & Hopwood (2003) have recently proposed that different siderophores may differ in their vulnerability to piracy by other bacteria. The differential regulation that we observed adds a further dimension to this argument.

Proteins presumptively associated with other important processes occurring at the cell membrane were also detected in this study. They included: several ATP-synthase subunits, responsible for energy production; the cell-division-associated protease FtsH2 (SCO3404); and several proteins annotated as being involved in biosynthesis of lipids [SCO1814, SCO2390 (FabF) and SCO6717] and peptidoglycan [FemX (SCO3904) and the penicillin-binding protein homologue SCO4439]. The four lipoproteins not assigned to any transport systems included a putative secreted peptidyl-prolyl cis–trans isomerase (SCO1639) believed to play a role in ensuring correct folding of proteins after they have been exported. No data for the abundance of these proteins during growth were obtained.

Interestingly, around 25 % of the proteins detected in the salt-washed membrane preparations did not have sequence motifs that predicted their observed localization (Table 1d). The presence of ribosome and ribosome-associated proteins was probably the result of co-sedimentation during the isolation process, which employed ultracentrifugation of cell-free extracts. However, the ATP synthase {alpha}-, {beta}- and {delta}-subunits identified are known to be intimately associated with the integral membrane protein subunits A, B and C, and this raises the possibility that other members listed in Table 1(d) similarly interact with integral membrane protein partners. Scrutiny of the chromosomal context of the genes provided circumstantial support for this in two cases. Thus, SCO1814 is immediately downstream of, and possibly translationally coupled to, fabG, a 3-oxoacyl-(acyl carrier protein) reductase involved in fatty acid biosynthesis, which encodes a protein with a strongly predicted transmembrane helix; and SCO4252, encoding a protein of unknown function, is also next to a gene predicted to code for an integral membrane protein (SCO4253).

The TTA-containing regulatory gene SCO4263 is associated with the absence of SCO4244 and SCO4252 from the {Delta}bldA membrane proteome, and appears to regulate a cluster of operons of function-unknown genes
Only four membrane-associated proteins showed obvious bldA-dependence (Fig. 3a). SCO4244 and SCO4252, both annotated as being hypothetical proteins of unknown function, were completely absent from the mutant membranes. SCO4252 had previously been identified as a bldA-dependent protein in an analysis of the whole-cell proteome of S. ceolicolor, and this dependence was shown to be mediated via the nearby TTA-containing regulatory gene SCO4263 (A. Hesketh and others, unpublished). The possibility that the SCO4244 gene is also dependent on this regulator was raised by the similarity of the abundance profiles for SCO4244 and SCO4252 in the membranes of the parent strain (Fig. 3a), and by the occurrence of the same DNA sequence motif upstream of each gene, located at the start of the SCO4248 and SCO4253 genes that are believed to be at the beginning of two operons containing the genes in question (Fig. 4). An M600 {Delta}SCO4263 deletion mutant was completely defective in transcription of SCO4248 (Fig. 5b), linking the observed absence of SCO4244 in M600 {Delta}bldA directly to the TTA-containing regulatory gene SCO4263. This is analogous to the dependence of the SCO4251–3 operon on SCO4263 observed by A. Hesketh and others (unpublished). Furthermore, transcription from SCO4254, divergent from SCO4253, was also absent in the M600 {Delta}SCO4263 mutant, indicating that three of the four ‘arms' of the two sets of divergent promoters illustrated in Fig. 4 depend on the SCO4263 regulator. SCO4263 encodes a protein homologous to members of the MalT-related LAL subfamily (large ATP-binding regulators of the LuxR family; De Schrijver & De Mot, 1999) of transcriptional regulators, and is most similar to members reported from antibiotic biosynthetic clusters in Streptomyces species [e.g. nystatin (Sekurova et al., 2004), amphotericin (Carmody et al., 2004), candicidin (Campelo & Gil, 2002) and geldanamycin (Rascher et al., 2003)]. An M600 {Delta}SCO4263 mutant, however, has no obvious phenotype with respect to antibiotic production or morphological differentiation (A. Hesketh and others, unpublished). MalT activates transcription of the maltose operon in E. coli by forming a nucleoprotein complex with a second activator protein CRP (reviewed by Boos & Shuman, 1998), and the two DNA motifs present in each of the divergent promoter regions SCO4248–9 and SCO4253–4 may indicate that the SCO4263 regulator similarly requires a partner activator protein. In this context it is interesting to note that SCO4261 encodes a putative GerE family transcriptional regulator. In the M600 parent strain, transcription from the SCO4263-dependent promoters was clearly higher in the exponential phase sample than in the stationary phase sample, while the corresponding SCO4263-dependent proteins were most abundant at the end of exponential growth. This suggests that bldA, in addition to its well-documented affects on stationary-phase processes, is important for the correct functioning of at least some processes that do not persist into the stationary phase.

The region of the chromosome shown in Fig. 4 is rich in function-unknown genes. Of 22 putative ORFs, 16 encode proteins of unknown function, three produce possible hydrolytic enzymes, two are regulatory genes and one encodes an AAA family ATPase. The likely bldA-dependence of many of these hypothetical proteins reveals a new aspect of the regulation of this region, and of the pleiotropic bldA phenotype. A comparable genetic region is not present in the Streptomyces avermitilis genome (http://avermitilis.ls.kitasato-u.ac.jp/), although the flanking regions SCO4230–40 and SCO4264–69 show synteny with the SAV3972 to SAV3953 genes of S. avermitilis. In S. coelicolor, the SCO4241 to SCO4263 genes therefore interrupt this region of apparent synteny and it may be that they have been laterally acquired. In this context it is interesting to note that two of the proteins encoded by the cluster possess domains found in phage structural proteins (SCO4245 and SCO4246), and that the only tRNA for the threonine-ACA codon is located between SCO4240 and SCO4241 (Fig. 4). This tRNA determinant is present in S. avermitilis at exactly the same position at which synteny between the genomes is interrupted (i.e. between SAV3961 and SAV3962). Often, tRNAs are used as attachment sites by integrating genetic elements such as prophages and plasmids. However, the region is not one of the 14 regions designated by Bentley et al. (2002) as being potentially recently laterally acquired insertions in the S. coelicolor genome, and, unlike those regions, has an overall G+C content similar to that of the S. coelicolor chromosome. Because there is no trace of a partial tRNA determinant at the right-hand end of the insertion, it seems possible that a deletion of the right-hand end followed the initial acquisition of the proposed inserted element.

Other bldA-dependent proteins in the membrane proteome
The abundance of SCO5249, a putative cyclic nucleotide-binding protein, was dramatically reduced in the bldA mutant. In M600 this protein was undetectable in membranes of exponentially growing cells, but increased markedly in samples prepared from transition and stationary phase cultures. Since the SCO5249 gene lacks a TTA codon, the absence of SCO5249 in the bldA mutant is presumably an indirect effect. The protein does not contain a predicted transmembrane helix and in the parent strain it may interact with an integral membrane protein absent from the bldA mutant. SCO5249 was also lacking in total protein extracts of a bldA mutant, suggesting that it has not merely been released from a membrane-bound partner (A. Hesketh and others, unpublished). SCO5249 is a close homologue of EshA, a 52 kDa cyclic nucleotide-binding protein of Streptomyces griseus that forms large multimers of approximately 20 monomers, allowing it to co-sediment with ribosomes during ultracentrifugation (Saito et al., 2003). Co-enrichment for ribosomes and the membrane fraction, hinted at by the finding of many ribosomal proteins in the membrane fraction (Table 1), may similarly account for the appearance of SCO5249 in the membrane preparations in this study.

In contrast to the three proteins above, SCO7399, a putative lipoprotein predicted to form part of an iron-transport system, was readily detectable in the {Delta}bldA strain but its abundance was reproducibly increased twofold or more in the early time points (Fig. 3a). Two other lipoproteins believed to be involved in siderophore-mediated uptake of iron from the environment were, however, unchanged in the mutant (SCO0494 and SCO2780 in Fig. 3d).

It is curious that we found no products of TTA-containing genes in the list of differences between the M600 and M600 {Delta}bldA strains reported here, or in a previous analysis of the extracellular proteome (Kim et al., 2005). Among several possible explanations for this, we think a major contributing factor may be that these genes are likely to be of adaptive value only in particular environmental circumstances, and therefore may not be expressed in the culture conditions used.


   ACKNOWLEDGEMENTS
 
We are grateful to Mike Naldrett and staff of the JIC Proteomics Facility for their help, and to Govind Chandra for help with bioinformatics analysis of the S. coelicolor genome. We thank Meifing Tao for generously providing the M600 {Delta}bldA strain, and David Hopwood for thoughtful comments on the manuscript. This work was funded by BBSRC grant EGH16080 by a grant-in-aid to the John Innes Centre from the BBSRC, and by research grant (KISTEP)-M603010 00019-04A0200-01510 from the Korea Institute of Science and Technology Evaluation and Planning. Dae-Wi Kim is also supported by a BK21 Research Fellowship from the Korea Ministry of Education and Human Resource Development.


   REFERENCES
TOP
ABSTRACT
INTRODUCTION
METHODS
RESULTS
DISCUSSION
REFERENCES
 
Bentley, S. D., Chater, K. F., Cerdeno-Tarraga, A.-M. & 40 other authors (2002). Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature 417, 141–147.[CrossRef][Medline]

Bertram, R., Schlicht, M., Mahr, K., Nothaft, H., Saier, M. H. & Titgemeyer, F. (2004). In silico and transcriptional analysis of carbohydrate uptake systems of Streptomyces coelicolor A3(2). J Bacteriol 186, 1362–1373.[Abstract/Free Full Text]

Boos, W. & Shuman, H. (1998). Maltose/maltodextrin system of Escherichia coli: transport, metabolism and regulation. Microbiol Mol Biol Rev 62, 204–229.[Abstract/Free Full Text]

Campelo, A. B. & Gil, J. A. (2002). The candicidin gene cluster from Streptomyces griseus IMRU 3570. Microbiology 148, 51–59.[Medline]

Carmody, M., Byrne, B., Murphy, B., Breen, C., Lynch, S., Flood, E., Finnan, S. & Caffrey, P. (2004). Analysis and manipulation of amphotericin biosynthetic genes by means of modified phage KC515 transduction techniques. Gene 343, 107–115.[CrossRef][Medline]

Challis, G. L. & Hopwood, D. A. (2003). Synergy and contingency as driving forces for the evolution of multiple secondary metabolite production by Streptomyces species. Proc Natl Acad Sci U S A 100, S14555–S14561.[CrossRef]

Chevallet, M., Santoni, V., Poinas, A. & 7 other authors (1998). New zwitterionic detergents improve the analysis of membrane proteins by two-dimensional electrophoresis. Electrophoresis 19, 1901–1909.[CrossRef][Medline]

De Schrijver, A. & De Mot, R. (1999). A subfamily of MalT-related ATP-dependent regulators in the LuxR family. Microbiology 145, 1287–1288.[Medline]

Hesketh, A. R., Chandra, G., Shaw, A. D., Rowland, J. J., Kell, D. B., Bibb, M. J. & Chater, K. F. (2002). Primary and secondary metabolism, and post-translational protein modifications, as portrayed by proteomic analysis of Streptomyces coelicolor. Mol Microbiol 46, 917–932.[CrossRef][Medline]

Janssen, G. R., Ward, J. M. & Bibb, M. J. (1989). Unusual transcriptional and translational features of the aminoglycoside phosphotransferase gene (aph) from Streptomyces fradiae. Genes Dev 3, 415–429.[Abstract]

Kieser, T., Bibb, M. J., Buttner, M. J., Chater, K. F. & Hopwood, D. A. (2000). Practical Streptomyces Genetics. Norwich: John Innes Foundation.

Kim, D. W., Chater, K. F., Lee, K. J. & Hesketh, A. R. (2005). Changes in the extracellular proteome caused by the absence of the bldA gene product, a developmentally significant tRNA, reveal a new target for the pleiotropic regulator AdpA in Streptomyces coelicolor. J Bacteriol 187, 2957–2966.[Abstract/Free Full Text]

Luche, S., Santoni, V. & Rabilloud, T. (2003). Evaluation of nonionic and zwitterionic detergents as membrane protein solubilizers in two-dimensional electrophoresis. Proteomics 3, 249–253.[CrossRef][Medline]

Molloy, M. P., Herbert, B. R., Slade, M. B., Rabilloud, T., Nouwens, A. S., Williams, K. L. & Gooley, A. A. (2000). Proteomic analysis of the Escherichia coli outer membrane. Eur J Biochem 267, 2871–2881.[Abstract/Free Full Text]

Murray, M. G. (1986). Use of sodium trichloroacetate and mung bean nuclease to increase sensitivity and precision during transcript mapping. Anal Biochem 158, 165–170.[CrossRef][Medline]

Neuhoff, V., Arold, N., Taube, D. & Ehrhardt, W. (1988). Improved staining of proteins in polyacrylamide gels including isoelectric focusing gels with clear background at nanogram sensitivity using Coomassie Brilliant Blue G-250 and R-250. Electrophoresis 9, 255–262.[Medline]

Nouwens, A. S., Cordwell, S. J., Larsen, M. R., Molloy, M. P., Gillings, M., Willcox, M. D. & Walsh, B. J. (2000). Complementing genomics with proteomics: the membrane subproteome of Pseudomonas aeruginosa PAO1. Electrophoresis 21, 3797–3809.[CrossRef][Medline]

Rascher, A., Hu, Z., Viswanathan, N., Schirmer, A., Reid, R., Nierman, W. C., Lewis, M. & Hutchinson, C. R. (2003). Cloning and characterization of a gene cluster for geldanamycin production in Streptomyces hygroscopicus NRRL 3602. FEMS Microbiol Lett 218, 223–230.[CrossRef][Medline]

Redenbach, M., Kieser, H. M., Danapiete, D., Eichner, A., Cullum, J., Kinashi, H. & Hopwood, D. A. (1996). A set of ordered cosmids and a detailed genetic and physical map for the 8 Mb Streptomyces coelicolor A3(2) chromosome. Mol Microbiol 21, 77–96.[CrossRef][Medline]

Saito, N., Matsubara, K., Watanabe, M., Kato, F. & Ochi, K. (2003). Genetic and biochemical characterization of EshA, a protein that forms large multimers and affects developmental processes in Streptomyces griseus. J Biol Chem 278, 5902–5911.[Abstract/Free Full Text]

Santoni, V., Molloy, M. & Rabilloud, T. (2000). Membrane proteins and proteomics: un amour impossible? Electrophoresis 21, 1054–1070.[CrossRef][Medline]

Sekurova, O. N., Brautaset, T., Sletta, H., Borgos, S. E., Jakobsen, O. M., Ellingsen, T. E., Strom, A. R., Valla, S. & Zotchev, S. B. (2004). In vivo analysis of the regulatory genes in the nystatin biosynthetic gene cluster of Streptomyces noursei ATCC 11455 reveals their differential control over antibiotic biosynthesis. J Bacteriol 186, 1345–1354.[Abstract/Free Full Text]

Strauch, E., Takano, E., Bayliss, H. A. & Bibb, M. J. (1991). The stringent response in Streptomyces coelicolor A3(2). Mol Microbiol 5, 289–298.[Medline]

Weaver, D., Karoonuthaisiri, N., Tsai, H. H. & 9 other authors (2004). Genome plasticity in Streptomyces: identification of 1 Mb TIRs in the S. coelicolor A3(2) chromosome. Mol Microbiol 51, 1535–1550.[CrossRef][Medline]

Received 1 March 2005; revised 26 April 2005; accepted 28 April 2005.



This Article
Abstract
Full Text (PDF)
Alert me when this article is cited
Alert me if a correction is posted
Citation Map
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Download to citation manager
Google Scholar
Articles by Kim, D.-W.
Articles by Hesketh, A.
Articles citing this Article
PubMed
PubMed Citation
Articles by Kim, D.-W.
Articles by Hesketh, A.
Agricola
Articles by Kim, D.-W.
Articles by Hesketh, A.


HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
INT J SYST EVOL MICROBIOL MICROBIOLOGY J GEN VIROL
J MED MICROBIOL ALL SGM JOURNALS
Copyright © 2005 Society for General Microbiology.