Metabolic and evolutionary insights into the closely-related species Streptomyces coelicolor and Streptomyces lividans deduced from high-resolution comparative genomic hybridization
© Lewis et al; licensee BioMed Central Ltd. 2010
Received: 7 June 2010
Accepted: 1 December 2010
Published: 1 December 2010
Whilst being closely related to the model actinomycete Streptomyces coelicolor A3(2), S. lividans 66 differs from it in several significant and phenotypically observable ways, including antibiotic production. Previous comparative gene hybridization studies investigating such differences have used low-density (one probe per gene) PCR-based spotted arrays. Here we use new experimentally optimised 104,000 × 60-mer probe arrays to characterize in detail the genomic differences between wild-type S. lividans 66, a derivative industrial strain, TK24, and S. coelicolor M145.
The high coverage and specificity (detection of three nucleotide differences) of the new microarrays used has highlighted the macroscopic genomic differences between two S. lividans strains and S. coelicolor. In a series of case studies we have validated the microarray and have identified subtle changes in genomic structure which occur in the Asp-activating adenylation domains of CDA non-ribosomal peptide synthetase genes which provides evidence of gene shuffling between these domains. We also identify single nucleotide sequence inter-species differences which exist in the actinorhodin biosynthetic gene cluster. As the glyoxylate bypass is non-functional in both S. lividans strains due to the absence of the gene encoding isocitrate lyase it is likely that the ethylmalonyl-CoA pathway functions as the alternative mechanism for the assimilation of C2 compounds.
This study provides evidence for widespread genetic recombination, rather than it being focussed at 'hotspots', suggesting that the previously proposed 'archipelago model' of genomic differences between S. coelicolor and S. lividans is unduly simplistic. The two S. lividans strains investigated differ considerably in genetic complement, with TK24 lacking 175 more genes than its wild-type parent when compared to S. coelicolor. Additionally, we confirm the presence of bldB in S. lividans and deduce that S. lividans 66 and TK24, both deficient in the glyoxylate bypass, possess an alternative metabolic mechanism for the assimilation of C2 compounds. Given that streptomycetes generally display high genetic instability it is envisaged that these high-density arrays will find application for rapid assessment of genome content (particularly amplifications/deletions) in mutational studies of S. coelicolor and related species.
Bacteria of the Gram-positive genus Streptomyces are of great scientific and economic importance. Streptomycetes are saprophytic, aerobic, soil dwelling bacteria which undergo complex differentiation to form mycelium, aerial hyphae and spores, and produce a diverse array of secondary metabolites, such as antibiotics and bioactive compounds . There are a number of published streptomycete genomes including Streptomyces griseus, Streptomyces avermitilis and Streptomyces coelicolor. The genomes of the sequenced streptomycetes consist of large (6-12 Mb) linear chromosomes with high (72-74%) G+C content . The chromosomes show synteny to each other in the central 'core', a region thought to contain conserved and highly expressed genes, and also contain two flanking 'arm' regions, proposed regions of high horizontal gene transfer (HGT) with high recombination rates [5, 6]. The large sizes of streptomycete genomes are thought to be related to the diversity of environmental niches which Streptomyces encounter.
S. lividans is known to be closely related to S. coelicolor A3(2) and has a virtually identical (99.7%) 16S rDNA sequence [7, 8]. Both species are members of the S. violaceoruber clade in 16S rDNA phylogenetic trees [9–11]. The similarities in basic genomic structure and genetic organisation of the S. coelicolor and S. lividans chromosomes have long been known . However, it is clear that many genomic differences are present, e.g. S. lividans 66 possesses a 93 kb "genomic island" in its chromosome, relative to that of S. coelicolor A3(2) [13, 14], and does not contain the same fertility plasmids as S. coelicolor, A3(2) instead harbouring the plasmids SLP2  and SLP3 . These, and other genetic differences, are reflected in phenotypic differences. For example, S. lividans differs from S. coelicolor in that its DNA is degraded during electrophoresis in buffers containing traces of ferrous iron. This phenomenon is due to a DNA phosphorothioate modification conferred by a cluster of five dnd genes which are only present in the S.lividans genome in the 93 kb island [13, 17–19]. Although, S. lividans exhibits most of the secondary metabolic capability that S. coelicolor possesses it does not produce antibiotics to the same extent, for reasons which are unclear. However, a number of studies have shown that antibiotic gene clusters can be "awakened" in S. lividans through the over-expression of regulatory factors such as AfsS and ActII-ORF4 [20, 21].
S. lividans is one of the most commonly used Streptomyces hosts for DNA cloning [22, 23] and heterologous protein production [24, 25]. It has several features that make it a suitable host for efficient recombinant protein expression, including the absence of a methylation-dependent restriction system  which recognizes and degrades methylated DNA isolated from commonly used Escherichia coli strains [27, 28]. Additionally, S. lividans has very low endogenous extracellular proteolytic activity when compared to other Streptomyces species, leading to higher product recovery . Moreover, in comparison with E. coli, S. lividans is a better host for eukaryotic recombinant protein production because the recombinant proteins produced in S. lividans tend to have higher levels of solubility, therefore avoiding the problem of inclusion body formation.
For streptomycetes for which no genome sequence is available, comparative genomic hybridization (CGH), through the use of microarrays, provides a useful tool for the comparison of genetic content between strains. While microarrays cannot detect chromosomal rearrangements or single nucleotide polymorphisms, oligonucleotide based microarrays can be hybridized using stringent conditions which enable the detection of small numbers of bases changes, and can be a powerful tool in detecting gene duplication, horizontal gene transfer (HGT) and gene loss/divergence. A previous comparative study of S. lividans TK21 against S. coelicolor M145 has been conducted by Jayapal and co-workers . However, because spotted PCR product based arrays were used the findings were limited to identification of large-scale differences, (PCR product probes being more tolerant to nucleotide changes compared to oligonucleotide probes) between gene coding regions. In the present study the genomic differences between the sequenced S. coelicolor M145 strain, (a prototrophic, plasmid-free derivative of the wild-type A3(2) strain) and two S. lividans strains: S. lividans 66 (wild-type), and its plasmid free derivative strain S. lividans TK24, have been determined using our novel high-density S. coelicolor microarrays. Our aim was to determine whether high-resolution CGH could identify small insertions/deletions (indels). It was anticipated that this study would serve not only to experimentally validate the microarray platform, but also shed light on the genotypic differences between the two streptomycete species, and also between the two strains of S. lividans.
Results and Discussion
Production of high density 104K × 60-mer S. coelicolor DNA microarray
Due to the high G+C content of S. coelicolor (ca. 72%) it was necessary to experimentally test specificity of a large set of probes in order to select a validated subset for comprehensive coverage of the genome. The resulting 104K microarray comprises almost 104,000 unique 60-mers (see Material & Methods) with an average spacing of 30 nucleotides. The 104K array is designed for CGH, ChIP-Chip and high resolution transcriptome analysis, covering all known genes and all intergenic regions; both strands of the genome are represented for the latter.
High resolution comparative genomics
Prima facie, we were confident that regions of difference identified by multiple contiguous probes indeed represented genuine phenomena and demonstrated the presence of substantial sequence differences between S. coelicolor and the S. lividans strains. However, initially our confidence in the validity of single probe absences was less. The microarray probe design and validation procedure  was such that only probes that exhibited a reduction in signal intensity (and thus represent a non-binding/absent event) following inclusion of a three nucleotide mismatch were selected. Therefore, to experimentally verify in a real world experimental situation whether the apparently missing isolated probes were in fact genuine absences, or false negatives, and whether the microarrays were operating according to specification, we selected several regions targeted by these single, isolated 'absent' probes for further analysis.
Investigation of a single non-binding probe in the CDA biosynthetic gene cluster
Investigation of a single non-binding probe in the actinorhodin gene cluster
From a preliminary, trial array used in our array testing and optimization process (see Methods) we identified a non-binding probe corresponding to positions 5524009-5524068 of the S. coelicolor chromosome, i.e. a region located in between genes SCO5082 and SCO5083 of the actinorhodin biosynthetic cluster, which encode the actII-1 transcriptional regulator and actII-2 actinorhodin transporter proteins respectively. Sequence determination of a PCR product amplified using primers actdelL and actdelR (Additional File 1) spanning the corresponding region of the S. lividans 66 chromosome generated the sequence shown in Additional File 3, which differs from the S. coelicolor probe sequence by seven nucleotides, including three consecutive nucleotide mismatches. It may be that these sequence differences, which are located in the divergent promoter region of SCO5082-3, are at least partly responsible for the differences in actinorhodin production between S. coelicolor and S. lividans, although it is clear that other factors are involved in actinorhodin production/regulation [[20, 21]; see below].
The above results illustrate the sensitivity of the experimentally optimised 60-mer probe set in discriminating relatively small differences in nucleotide sequence.
Establishment of the "region of difference" calling criterion for gene presence/absence
The above studies demonstrate the high specificity of our microarray design, indicating that single non-binding probes can represent genuine genomic differences between strains. In addition to providing detection of small sequence differences and microdissection of individual gene structure our results provide a macroscopic overview of genome content in S. lividans.
Genes classified as absent/divergent from both S. lividans 66 and TK24, and uniquely from TK24 are given, respectively, in Additional Files 7, 8 and 9. The criterion used in this study represents a significant advance in terms of increase in resolution of interspecies genomic differences over the previous amplicon-based CGH study . Furthermore, the present study identifies differences between the intergenic regions of S. coelicolor and S. lividans which have not been previously investigated.
We note that there are 'regions of difference' which do not appear to correlate with absent/divergent protein coding genes. The absence/divergence of these regions may however prove to be significant as these regions may encompass promoter elements and RNA genes, including, for example at present unidentified/uncharacterized cis- or trans-encoded non-coding RNAs.
Comparison of the S. lividans TK24 microarray results with the results of BLAST search
From the annotated 7,824 protein encoding genes in the S. coelicolor genome, the BLAST analysis identifies 657 genes as missing and 7,167 as present in S. lividans TK24. From the array analysis of S. lividans TK24 versus S. coelicolor M145 we identify 619 genes as missing and 7,205 as present. 120 genes were found to be missing by the array analysis but found to be present by BLAST, whilst 159 genes were found to be present by the array analysis but missing by BLAST.
If we classify genes identified as present by the reciprocal BLAST as True Positives (TP) and the number of genes mis-identified (when compared to the reciprocal BLAST) by the array classed as False Negatives (FN) then the sensitivity can be calculated by TP/TP+FN. Thus the sensitivity of the arrays and calling method is 7047/7047+120: 98%. If we class genes identified as absent by the reciprocal BLAST as True Negatives (TN) and the number of genes mis-identified (when compared to the reciprocal BLAST) by the array classed as False Positives (FP) then the Specificity can be calculated by TN/TN+FP. Thus the specificity of the arrays and calling method is 498/498+159: 78%. The overall accuracy (TN+TP/TN+TP+FN+FP) of the arrays and calling method for CGH purposes is 96%.
Macroscopic pattern of gene absence/divergence in S. lividans 66 and TK24 and comparision to S. lividans TK21 and S. coelicolor M145
Previous studies  have categorized regions of difference based on size as either Genomic Islands (GI) (≥25 kb) or as smaller Genomic Islets (Gi), which although smaller than 25 kb contain at least three consecutive genes . The use of this "archipelago" model is understandable given the technical limitations of earlier microarray technologies and the gene absence/presence calling criterion, which tended to bias the analysis towards identifying blocks of consecutive absent genes. Analysis of the S. coelicolor M145 genome sequence identified a series of regions designated as potentially recently laterally acquired  which are broadly consistent with clusters of genes identified by Jayapal and co-workers as absent/divergent. However, the increased resolution of the genomic differences between the species that the 104K microarray of the present study affords has allowed us to identify a more subtle and complicated pattern of differences, so the 'archipelago' model approach was not employed in our study.
Our results are broadly consistent with the findings reported by Jayapal et al in that the majority of genes absent in S. lividans 66 and TK24 are present in the GI's and Gi's of TK21 defined previously. In some cases our results correlate exactly with previous results, particularly when considering the smaller genomic islets (Gi's). For example, S. lividans TK24 possesses blocks of absent genes which match Gi-3, (SCO2862-2871), Gi-4, (SCO3250-3270), Gi-8 (SCO4210-4218), Gi-9 (SCO4346-4350), Gi-11 (SCO4615-4631), Gi-13 (SCO5605-5620), Gi-15 (SCO5718-5735) and the Right End terminal deletion (SCO7827-7845) of S. lividans TK21 exactly. Figure 3B illustrates Gi-11 in S. lividans 66. Similarly, S. lividans 66 possesses blocks of absent genes which match Gi-3, Gi-5, Gi-6 (SCO3980-3998), Gi7 (SCO4060-4066) and Gi-8.
However, there are also significant differences between our results and those reported previously , and in many cases these correspond to regions where our results do not provide evidence for the complete absence of large blocks of genes. Instead, we are able to classify at least one gene, and in some cases many more, within most of the GI/Gi's identified previously  as being present, raising doubt as to the GI/Gi's boundaries and in some cases the very existence of a GI/Gi at all. For example, Gi-5 is reported as an absence of SCO3929-SCO3937  whereas we find that in both S. lividans TK24 and 66 that SCO3936 is classed as present. Additionally, Gi-2 was reported to lack SCO2381-2384 whereas we find that SCO2381 is present in both TK24 and 66. Moreover, Gi-1 is defined as an absence of SCO0090-0099  whereas in TK24 we find only SCO0090-0091 and SCO0098 are absent, and in S. lividans 66 only a single gene SCO0098 is absent. Furthermore, out of the 147 missing genes (SCO6806-6953) defined as GI-5, we find that TK24 has 131 missing genes, but significantly S. lividans 66 only has 43 missing. Hence, due to the methods employed in this study we have succeeded in identifying many single genes, and small blocks of genes, located in between genomic islands/islets as absent/divergent. For example, we assign three consecutive secreted protein genes (SCO5013-5015) and the isolated secreted protein gene SCO5995 as absent in both S. lividans 66 and TK24. We also assign SCO3521-3522, encoding an integral membrane protein and a transcriptional regulator respectively, as absent in TK24 only.
Consideration of the absent divergent genes in S. lividans 66 and TK24
The lists of genes which are absent/divergent and are common to S. lividans 66 and TK24 (444) and are specific to TK24 (175) are listed in Additional Files 7, 8 and 9. Hypothetical genes and pseudogenes account for approximately 50% of the genetic differences between S. lividans 66, S. lividans TK24, and S. coelicolor M145. It is clear from the gene annotations and co-localization pattern that a number of genes lost are from integrated plasmids of the pSAM2 family from S. ambofaciens. For example, genes SCO3250-3260 and SCO5336-5349 represent integrated plasmids similar to pSAM2 from which the three S. lividans strains have lost genes.
Other classes of genes well represented in the absent/divergent lists include transposases, and "orphan" membrane proteins, secreted proteins, transcription regulators, unspecified regulator and transporter proteins, lipoproteins, DNA-binding proteins, and ATP-binding proteins. Until such time as the precise functions of these genes are understood we are unable to explain or predict reasons for phenotypic differences between S. coelicolor and the S. lividans strains. Therefore, the main interest of the available datasets lies in the absent/divergent genes of S. lividans whose functional annotations are sufficiently detailed to be useful in hypothesis/prediction generation, i.e. mainly the enzymes.
The differences between the genetic complement of S. lividans TK21 and S. coelicolor M145 have been discussed elsewhere and we do not intend to reiterate the results of the previous analysis  and in view of their current annotations we are unable to usefully comment on the significance of the 28 absent/divergent genes specific to TK24. Of the 55 gene differences found in TK21 but not found in S. lividans 66 and TK24, only three are enzymes and only one (SCO0984) possesses useful functional annotation, being a putative 3 hydroxyacyl-CoA dehydrogenase. However, TK21 also differs significantly from the other S. lividans strains in lacking the φC31 phage resistance gene pglZ (SCO6636), although it and TK24 both lack the φC31 phage resistance gene pglY (SCO6635). Both of these genes are present in S. lividans 66.
The differences in act biosynthesis between S. lividans and S. coelicolor may be due to absence/divergence of genes relating to S-adenosylmethionine
Among the more intriguing of the apparently missing genes in the S. lividans strains are those which are either involved in the biosynthesis of methionine or are S-adenosylmethionine (SAM)-dependent methyltransferases. The indirect involvement of genes possessing these functions in differentiation and antibiotic production has previously been noted . Elevated concentrations of SAM have been shown to lead to expression of actII-ORF4 and subsequent overexpression of actinorhodin biosynthesis in S. lividans TK23 [44, 45], overexpression of undecylprodiginine in S. coelicolor and inhibition of sporulation in S. lividans. SAM has been shown to mediate its effect on sporulation through induction of bldK expression , in addition to other ABC transporters involved in development , and possibly through modulation of bldA expression . We note that the genes identified as missing in S. lividans 66 and TK24 in the present study include SCO0985 (metE, encoding methionine synthase) and the methyltransferase genes SCO0995 (absent in both S. lividans 66 and TK24) and SCO3452 (absent in TK24). It may be significant that both absent methyltransferase genes possess sequence similarity to SAM-dependent methyltransferases. Whilst we do not propose a direct and precise mechanism by which loss of SCO0985, SCO0995 and SCO3452 mediate the differences in antibiotic production and differentiation between S. coelicolor and S. lividans the roles we hypothesize for these genes and their effect on development and secondary metabolite biosynthesis relate to their effect on SAM concentration. This hypothesis is consistent with them playing roles similar to those outlined previously with regard to mutagenesis of the methionine synthase, metH, and a SAM dependent methyltransferase, SCO2525 .
Investigation on sequence differences in S.coelicolor and S. lividans bldB
Although it is instructive to compare the macrosequence differences between the two S. lividans strains, and S. coelicolor much of the interest of the present dataset lies in the fine detail afforded by the high density coverage of the genomes, as exemplified above in the case of the CDA Asp-activating adenylation domains. A further example of the use of the dataset in microsequence analysis relates to the bldB gene (SCO5723). This gene encodes a small (98 aa) protein required for morphogenesis, antibiotic production and catabolite control in S. coelicolor. S. lividans TK21 apparently lacks this gene  and in the present study we find it is classified as absent/divergent in S. lividans 66 and TK24. A bldB homologue has long been known in S. lividans TK24 , its sequence has been determined (GenBank AF071232) and it is well known that S. lividans sporulates readily and does not exhibit the severe bldB deletant mutant phenotype observed in S. coelicolor mutants . bldB null mutants have a "bald" phenotype when grown on all carbon sources and fail to produce aerial hyphae or antibiotics under any tested conditions [51, 52]. In contrast to  we accept that S. lividans does in fact possess bldB and attribute its apparent absence in microarray studies to technical reasons relating to positioning of probe design and the pattern of sequence similarity between the S. coelicolor and S. lividans genes across their sequences (see Additional Files 10 & 11). It seems that the PCR product used by Jayapal et al spanning the entire TK21 bldB sequence was unable to bind to its target. In our microarrays the probes located 5" of bldB and the probe encompassing the 5" terminus (i.e. the start codon) were not bound in S. lividans 66 or TK24, and of the three remaining probes directly, or partially targeting, bldB on our microarray three bound successfully to the 3' terminus of bldB in S. lividans 66 and two bound successfully to TK24 genomic DNA (as determined by GACK analysis). Thus, in both S. lividans 66 and TK24 the presence of three consecutive absent probes encompassing the translational start means that bldB is classed as absent/divergent when in fact it is present but possesses sequence sufficiently divergent from the S. coelicolor homologue in the 5" region to prevent hybridization. When the sequences of PCR products generated from both S. lividans strains using primers slbldBUp and slbldBDown (Additional File 1) were determined, both were found to be identical to the published S. lividans bldB sequence (GenBank: AF071232) and the TK24 bldB sequence identified by BLAST search of the Broad Institute database, showing that the bldB gene is indeed present in both S. lividans strains as well as in S. coelicolor.
Investigation on metabolic differences between S. coelicolor and S. lividans and alternative mechanisms in S. lividans strains
Flux Balance Analysis (FBA) of Genome Scale Metabolic Reaction Networks (GSMN) has been widely used to predict metabolic capabilities of microbial strains, metabolic engineering of overproducing strains, prediction of essential genes and integration of high throughput data with the literature knowledge on metabolic reactions [53–55]. Here we have used FBA in combination with our CGH data to identify system-level differences in global metabolic flux distribution between S. coelicolor and S. lividans, and have modified the existing S. coelicolor GSMN accordingly and then tested our predictions relating to differences in cell physiology and C2 metabolism between the species.
Reactions requiring products of genes missing from S. lividans but present in the S. coelicolor genome scale metabolic network.
LACTAL + NAD = LLAC + NADH
LACTOSE = GLAC + GLC
KDPG = PYR + G3P
HYDROXYAKG = PYR + GLX
KDG + ATP = KDPG + ADP
KDG + ATP = KDPG + ADP
ICIT = GLX + SUCC
ACCOA + GLX = COA + MAL
HCYS + MTHPTGLU = THPTGLU + MET
OTHIOxt + NADPH = RTHIO + NAD
Perhaps the genes most fundamental to core metabolic activities which the present study identifies as absent in S. lividans are those which encode isocitrate lyase (SCO0982) and malate synthase (SCO0983), which are both involved in the glyoxylate bypass mechanism. This is a key system for the utilisation of compounds, such as fatty acids or acetate, which enter central metabolism at the level of acetyl-CoA. This mechanism involves the condensation of acetyl-CoA and oxaloacetate to form citrate, which is isomerized to isocitrate, which can in turn either be decarboxylated to form α-ketoglutarate in the TCA cycle or which can be cleaved by isocitrate lyase (ICL) to yield succinate and glyoxylate. Malate synthase (MS) is able to condense acetyl-CoA and glyoxylate to generate malate and CoA. Thus ICL and MS, together with enzymes from the Krebs Cycle, catalyse the net formation of succinyl-CoA from two molecules of acetyl-CoA. There is little published evidence for the expression of the glyoxylate bypass enzymes in Streptomyces species and it is well known that the glyoxylate bypass is not the sole mechanism by which C2 units may enter Streptomyces central metabolism [56, 57]. Although S. cinnamonensis possesses ICL and MS genes extracts from cells grown in oil-based media have been shown not to possess ICL activity  and S. collinus has only been shown to possess ICL activity when grown in the presence of Tween, but not in acetate . We note that an isoform of malate synthase (SCO6243) is present in both S. lividans 66 and TK24 and that reaction R388 may be catalyzed by this enzyme. However, as both strains lack ICL the strains must necessarily be deficient in the glyoxylate bypass.
We have used the S. coelicolor and S. lividans models to make predictions about the growth of both species on medium comprising palmitate as the sole carbon source. Simulations of maximal biomass synthesis rate predicted that S. coelicolor is able, and S. lividans is not able, to use palmitate as a sole carbon source (maximal biomass synthesis rate of the S. lividans model was 0). This contradicted previously published experimental data indicating that S. lividans TK24 is capable of growth in media comprising triacylglycerides or oleic acid as sole carbon source [58, 59] in addition to our own observations which demonstrate that S. lividans 66 and TK24 are capable of growth in minimal media where the sole carbon sources are Tween 80 or palmitic acid (data not shown). The apparent non-essentiality of the glyoxylate bypass enzymes may be explained by the presence of an alternative metabolic pathway. A number of possible alternatives for C2 metabolism to the glyoxylate bypass mechanisms have been suggested, although they have not been fully characterized/confirmed [56, 60]. However, the recently identified ethylmalonyl-CoA pathway [61, 62] is the most likely candidate mechanism which operates as an alternative to the glyoxylate shunt in S. lividans; indeed, this pathway has recently been shown to operate in S. coelicolor (D. A. Hodgson, personal communication).
The results presented here provide a powerful demonstration of the application of high-density microarrays to CGH studies. As shown here high-density IJISS arrays are capable of distinguishing between extremely similar sequences allowing specific discrimination on the basis of as few as three nucleotide mismatches.
The results presented here regarding single probe absences provide evidence for recombination between the aspartate-specific adenylation domain modules (4, 5 & 7) of the CdaPS genes of the CDA biosynthetic gene cluster revealing them to be "mosaic" genes, relative to the corresponding S. coelicolor sequences. We also identify an intergenic sequence divergence from S. coelicolor in the actinorhodin biosynthetic gene cluster and suggest that this, and/or the differences in the S. lividans complement of genes involved in SAM biosynthesis, or SAM-dependent methytransferases, may be involved in mediating the phenotypic differences in actinorhodin production between S. coelicolor and S. lividans.
Taking a broader view, the results indicate that the pattern of genetic differences between S. lividans 66 and TK24 are different to those of S. coelicolor M145 and S. lividans TK21. We propose that more widespread genetic drift and recombination has occurred in S. lividans than the "archipelago model" developed previously, suggests (which focuses on "hotspots" of genetic difference).
In a series of case studies we confirm that the pattern of probe binding to S. lividans genes correlates with their sequence differences relative to S. coelicolor and use this to explain the published contradictory results previously reported for S. lividans bldB.
We have developed a GSMN for S. lividans taking into account the genetic differences relating to the differences in central metabolism between S. lividans 66 and S. coelicolor, including the absence of genes encoding the enzymes of the glyoxylate bypass-isocitrate lyase and malate synthase. The fact that the GSMN indicated that both S. lividans strains should not be able to grow on fatty acids as sole carbon source, so contradicting the literature, suggests an alternative pathway to the glyoxylate shunt exists in S. lividans and it is likely that the hypothesized ethylmalonyl-CoA pathway fulfils this role.
It is clear from the case studies presented here regarding bldB and the genes which encode the glyoxylate bypass enzymes that the microarray data presented here has enormous potential to explain previously published observations and inform new hypotheses. We note that observations published in the previous study  regarding the putative identification of genes involved in DNA methylation systems of S. lividans have stimulated and informed further studies into this phenomenon . We expect that the present, more extensive and detailed study will do likewise.
Finally, we envisage that these high-density arrays will find widespread application for rapid assessment of genome content in mutational studies of S. coelicolor and related species. It is well known that streptomycetes generally display high genetic instability and second-site mutations can arise frequently in the S. coelicolor genome when conducting targeted mutagenesis studies (our unpublished observations). The 104K array provides an efficient tool for comparing a new mutant with its immediate parent, to allow identification of mutants that have acquired additional, unwanted, deletions/duplications (which may be missed by 'next-generation' sequencing) and hence exclude them from further study.
Strains and culture conditions
Strains used were S. lividans 66 SLP2+ SLP3+ corresponding to NRRL Code B-16637 and ICSSB Number 1023 (corresponding to stock number 1326 from the John Innes Centre collection) , S. lividans TK24 (str-6, SLP2-, SLP3-)  and S. coelicolor A3(2) strain M145 . S. coelicolor and S. lividans cultures for genomic DNA preparation were grown in YEME plus 34% sucrose medium supplemented with 0.5% glycine and grown at 30°C for 72 h until early stationary phase. DNA was extracted using the Kirby mix procedure .
A S. coelicolor high-density IJISS microarray comprising almost 104,000 60-mer experimentally assessed probes was produced and validated for CGH in this study. The initial rounds of probe design have been described previously  where the authors reported the use of a 44,000 60-mer probe array; briefly, a large database of all possible 60-mer probes based on the S. coelicolor A3(2) M145 genome (EMBL accession AL645882 version 2) was developed and the best performing probes, validated in terms of hybridisation quality and reduction of signal with the introduction of mismatches, were selected to optimise the array sensitivity and specificity. For the present study 103,695 experimentally validated probes, were selected to cover both coding and non-coding genomic regions, with an average spacing of 30 nucleotides; the probes were distributed randomly on the microarray. The S. coelicolor 104K microarrays are available from Oxford Gene Technology Ltd (UK).
Microarray hybridization and processing
Labelling reactions were performed using the BioPrime kit (Invitrogen). DNA (0.1-1 μg) was denatured at 94°C for 3 min in 40 μl including 20 μl 2.5 × random primer mix and kept on ice. Nucleotide mix, 5 μl (2 mM dATP, 2 mM dGTP, 2 mM dTTP, 0.5 mM dCTP), 3.75 μl Cy3/Cy5-dCTP (Perkin Elmer) and 1.5 μl of Klenow fragment (1.5 units) were added and the reaction was incubated at 37°C overnight. The labelled DNA was purified using the Minielute PCR purification kit (Qiagen) and the incorporated Cy3/Cy5-dCTP was quantified with the NanoDrop ND-1000 spectrophotometer. Two different genomic DNA preparations for both S. coelicolor M145 and S. lividans 66 and one preparation of S. lividans TK24 were each analysed in a 'dye-balanced' experimental design (to remove any dye bias). Labelled gDNA (50 pmol) of each pair of strains was hybridised to the 104K arrays in a buffer containing 1 M NaCl/50 mM MES, pH 7/20% formamide/1% Triton X-100 and rotated at 55°C over 60 h. Each glass slide contains 2 × 104K arrays and thus both Cy3/Cy5 dye orientations for each genomic DNA sample of each strain pair were hybridized on the same slide. The arrays were then washed with Wash 1 [6 × standard saline phosphate/EDTA (0.18 M NaCl/10 mM phosphate, pH 7.4/1 mM EDTA) (SSPE)/0.005% N-lauryl sarcosine) and Wash 2 (0.06% SSPE/0.18% polyethylene glycol 200), both for 5 min at room temperature. Hybridised arrays were scanned using an Agilent Technologies microarray scanner (5 μm resolution) and the resultant images analysed using Agilent Technologies Feature Extraction software (Version 18.104.22.168) with local background correction.
Microarray data processing
All microarray data were imported into R (version 2.5, R, R Development Core Team) and processed using the Bioconductor package Limma [63–65] and checked for spatial effects, of which none were found. By assuming the null hypothesis that S. lividans strains 66 and TK24 are similar to S. coelicolor M145 the data was transformed to log2S. lividans/S. coelicolor ratios and normalised using within-array loess normalisation followed by the between-array scale function; by normalising all arrays together the data (and thus S. lividans strains) become directly comparable due to the M145 genome being present on each array. Probes on each array were flagged as poor quality if the signals of both channels were classified as outliers in at least one of the binary variables (1 for bad, 0 for good) of the Feature Extraction software (Agilent Technologies, variables: gIsFeatNonUnifOL/rIsFeatNonUnifOL, gIsBGNonUnifOL/rIsBGNonUnifOL, gIsFeatPopnOL/rIsFeatPopnOL and gIsBGPopnOL/rIsBGPopnOL). Probes that had at least two dye balanced values for each S. lividans strain experiment (i.e. good signals in two of the four arrays for the 66 vs M145 experiment and both arrays for TK24 vs M145) were selected for further processing. Note, where probes had three (non dye-balanced) good quality signals for the 66 vs M145 experiment one of the two values for the over-represented dye was selected randomly. For large intergenic regions the 104K arrays include complementary probes targeting both DNA strands to detect the presence of non-coding RNAs, as all samples used in this study are DNA, able to hybridise to both complementary probes, these signals were averaged. In total 97,611 probes targeting unique DNA regions were used for genomotyping analysis. The microarray design and data are available from ArrayExpress (Accession numbers, respectively, A-MAXD-28 and E-MAXD-58).
For each of the 97,611 probes that passed filtering, median log2 ratios (taking the dye swapping into consideration) across biological replicates were calculated, resulting in one value for S. lividans 66/S. coelicolor M145 and one value for S. lividans TK24/S. coelicolor M145. The resultant data was converted into suitable data inputs (pcl format) for GACK (Charles Kim, Stanford University) and processed using the default settings. Binary outputs of 0 or 1 were obtained to denote probe absence, or presence, respectively. "Regions of difference" between S. lividans 66 or TK24 and S. coelicolor M145 were classified by sets of at least three probes. Thus, one or more probes with a GACK value of 1 (presence) separate regions of difference. Once all regions of difference were found absent genes could subsequently be identified by their annotated translational start codon; if a gene's translational start coordinate resides within an identified region of absence then it too was deemed as missing/divergent.
Reciprocal BLAST between Streptomyces genomes
S. coelicolor protein-encoding gene nucleotide sequences were obtained from the EMBL genome file (Accession No. EMBL:AL645882.2]) in FASTA format. Protein-encoding gene nucleotide sequences for S. lividans TK24 were obtained from the genes.fasta file downloaded from the BROAD institute  on 12/12/2009. Local BLAST nucleotide databases were created from the obtained FASTA files and each gene sequence was blasted against the other database (i.e. S. coelicolor genes against the S. lividans database and vice versa). A gene was classed as present if in each of the blast results it had a sequence match of greater than 60 nt in length (at least the length of the probes on the array) with greater than 60% sequence identity and an expected value less than 0.01. If these criteria were not met then the gene was classified as absent.
Codon Adaptation Index
The Codon Adaptation Index (CAI) value, a measure of synonymous codon usage bias, for each S. coelicolor protein encoding gene was calculated using the CAI tool of EMBOSS  The nucleotide sequences from the EMBL genome file (accession EMBL: AL645882.2) in FASTA format and the 'background' (genome frequency) codon usage table from the codon usage database [68, 69] were used as input to EMBOSS:cai for calculating bias using default settings.
Primer pairs, each specific for a particular region of S. coelicolor and/or S. lividans DNA sequence (Additional File 1), were obtained from Eurofins-MWG-Operon and used in colony PCR. One independently obtained PCR product for each S. coelicolor MT1110 CdaPS Asp-specific adenylation domain (Modules 4, 5 & 7) and two independently obtained PCR products for the corresponding sequences of S. lividans 66, were separated on a 1% agarose/Tris-Acetate-EDTA gel, the relevant bands excised and the DNA extracted using a Promega Wizard™ SV Gel & PCR Clean-up system. The purified DNA was treated with PNK (NEB) and blunt end-cloned into Sma I cut dephosphorylated pUC18 (Fermentas) using T4 DNA ligase (NEB). The ligation mix was transformed into competent E. coli JM109 and insert containing clones were identified by blue/white selection. Plasmid DNA was prepared from such clones using Promega Wizard Plus™ minipreps DNA purification system and supplied to Eurofins-MWG-Operon for BigDye™ (ABI) sequencing using M13 forward (-43) and reverse (-49) universal primers.
PCR products for S. coelicolor M145 and S. lividans 66 and TK24 bldB were similarly obtained, and treated, to determine the sequences of the central portions of bldB in these organisms/strains.
PCR products for a region spanning SCO5082 and SCO5083 incorporating the intergenic region for S. coelicolor M145 and S. lividans 66 were also similarly obtained, and treated, to determine their sequences.
Flux Balance Analysis
The genome scale metabolic reaction network of Borodina et al has been used as an initial model in the investigation of the system-wide metabolic differences of S. coelicolor and S. lividans. The reactions which require products of the genes, which according to our CGH are missing in S. lividans were removed from the network. Subsequently, the model has been refined by incorporation of additional reactions, which have been experimentally demonstrated to operate in Streptomyces, but were omitted from the initial model. The feasibility of growth of different in silico strains on different media has been simulated by the calculation of the maximal biomass synthesis rate with Flux Balance Analysis. Details of FBA methodology are described in detail elsewhere . Linear Programming calculations have been done with GLPK library run from our software, which has been previously used to model metabolic reaction networks of M. tuberculosis and S. coelicolo r. Models used in this work are available in SBML format from the ScoFBA website . The ScoFBA server also allows interactive simulations of these models via a web-based interface to our software.
We thank David Hodgson and Georgios Efthimiou for helpful discussions on metabolism.
This work was funded by BBSRC grants G18886 to CPS and BB/D0115821 to AK and CPS and the European Commission (FP6 Programme ActinoGEN IP005224 to CPS).
- Hopwood DA: Streptomyces in nature and medicine: the antibiotic makers. 2007, New York, Oxford University PressGoogle Scholar
- Ohnishi Y, Ishikawa J, Hara H, Suzuki H, Ikenoya M, Ikeda H, Yamashita A, Hattori M, Horinouchi S: Genome sequence of the streptomycin-producing microorganism Streptomyces griseus IFO 13350. J Bacteriol. 2008, 190: 4050-4060. 10.1128/JB.00204-08.PubMed CentralPubMedView ArticleGoogle Scholar
- Ikeda H, Ishikawa J, Hanamoto A, Shinose M, Kikuchi H, Shiba T, Sakaki Y, Hattori M, Omura S: Complete genome sequence and comparative analysis of the industrial microorganism, Streptomyces avermitilis. Nature Biotechnol. 2003, 21: 526-531. 10.1038/nbt820.View ArticleGoogle Scholar
- Bentley SD, Chater KF, Cerdeno-Tarraga AM, Challis GL, Thomson NR, James KD, Harris DE, Quail MA, Kieser H, Harper D, Bateman A, Brown S, Chandra G, Chen CW, Collins M, Cronin A, Fraser A, Goble A, Hidalgo J, Hornsby T, Howarth S, Huang CH, Kieser T, Larke L, Murphy L, Oliver K, O'Neil S, Rabbinowitsch E, Rajandream MA, Rutherford K, Rutter S, Seeger K, Saunders D, Sharp S, Squares R, Squares S, Taylor K, Warren T, Wietzorrek A, Woodward J, Barrell BG, Parkhill J, Hopwood DA: Complete genome sequence of the model actinomycete Streptomyces coelicolor A3(2). Nature. 2002, 417: 141-147. 10.1038/417141a.PubMedView ArticleGoogle Scholar
- Karoonuthaisiri N, Weaver D, Huang J, Cohen SN, Kao CM: Regional organization of gene expression in Streptomyces coelicolor. Gene. 2005, 353: 53-66. 10.1016/j.gene.2005.03.042.PubMedView ArticleGoogle Scholar
- Choulet F, Aigle B, Gallois A, Mangenot S, Gerbaud C, Truong C, Francou FX, Fourrier C, Guerineau M, Decaris B, Barbe V, Pernodet JL, LeBlond P: Evolution of the terminal regions of the Streptomyces linear chromosome. Mol Biol Evol. 2006, 23: 2361-2369. 10.1093/molbev/msl108.PubMedView ArticleGoogle Scholar
- Kawamoto S, Ochi K: Comparative ribosomal protein L11 and L30 sequence analyses of several Streptomyces spp. commonly used in genetic studies. Int J Syst Bacteriol. 1998, 48 (Pt 2): 597-600.PubMedView ArticleGoogle Scholar
- Krasil'nikov NA, Sorokina EJ, Alferova VA, Bezzubenkova AP: Biology of Selected Groups of Actinomycetes. Edited by: Krasil'nikov NA. 1965, Institute of Microbiology, MoscowGoogle Scholar
- Taguchi S, Kojima S, Miura K, Momose H: Taxonomic characterization of closely related Streptomyces spp. based on the amino acid sequence analysis of protease inhibitor proteins. F.E.M.S. Microbiol Lett. 1996, 135: 169-173. 10.1111/j.1574-6968.1996.tb07984.x.View ArticleGoogle Scholar
- Monson AM, Bradley SG, Enquist LW, Cruces G: Genetic homologies among Streptomyces violaceoruber strains. J Bacteriol. 1969, 99: 702-706.PubMed CentralPubMedGoogle Scholar
- Kutzner HJ, Waksman SA: Streptomyces coelicolor Mueller and Streptomyces violaceoruber Waksman and Curtis, two distinctly different organisms. J Bacteriol. 1959, 78: 528-538.PubMed CentralPubMedGoogle Scholar
- LeBlond P, Redenbach M, Cullum J: Physical map of the Streptomyces lividans 66 genome and comparison with that of the related strain Streptomyces coelicolor A3(2). J Bacteriol. 1993, 175: 3422-3429.PubMed CentralPubMedGoogle Scholar
- He X, Ou HY, Yu Q, Zhou X, Wu J, Liang J, Zhang W, Rajakumar K, Deng Z: Analysis of a genomic island housing genes for DNA S-modification system in Streptomyces lividans 66 and its counterparts in other distantly related bacteria. Mol Microbiol. 2007, 65: 1034-1048. 10.1111/j.1365-2958.2007.05846.x.PubMedView ArticleGoogle Scholar
- Zhou X, He X, Li A, Lei F, Kieser T, Deng Z: Streptomyces coelicolor A3(2) lacks a genomic island present in the chromosome of Streptomyces lividans 66. Appl Environ Microbiol. 2004, 70: 7110-7118. 10.1128/AEM.70.12.7110-7118.2004.PubMed CentralPubMedView ArticleGoogle Scholar
- Chen CW, Yu TW, Lin YS, Kieser HM, Hopwood DA: The conjugative plasmid SLP2 of Streptomyces lividans is a 50 kb linear molecule. Mol Microbiol. 1993, 7: 925-932. 10.1111/j.1365-2958.1993.tb01183.x.PubMedView ArticleGoogle Scholar
- Hopwood DA, Kieser T, Wright HM, Bibb MJ: Plasmids, recombination and chromosome mapping in Streptomyces lividans 66. J Gen Microbiol. 1983, 129: 2257-2269.PubMedGoogle Scholar
- You D, Wang L, Yao F, Zhou X, Deng Z: A novel modification by sulfur: DndA is a NifS-like cysteine desulfurase capable of assembling DndC as an iron-sulfur cluster protein in Streptomyces lividans. Biochemistry. 2007, 46: 6126-6133. 10.1021/bi602615k.PubMedView ArticleGoogle Scholar
- Xu T, Liang J, Chen S, Wang L, He X, You D, Wang Z, Li A, Xu Z, Zhou X, Deng Z: DNA phosphorothiolation in Streptomyces lividans : mutational analysis of the dnd locus. BMC Microbiology. 2009, 9: 41-10.1186/1471-2180-9-41.PubMed CentralPubMedView ArticleGoogle Scholar
- Zhou X, Deng Z, Firmin JL, Hopwood DA, Kieser T: Site-specific degradation of Streptomyces lividans DNA during electrophoresis in buffers contaminated with ferrous iron. Nucleic Acids Research. 1988, 16: 4341-4352. 10.1093/nar/16.10.4341.PubMed CentralPubMedView ArticleGoogle Scholar
- Bruheim P, Sletta H, Bibb MJ, White J, Levine DW: High-yield actinorhodin production in fed-batch culture by a Streptomyces lividans strain overexpressing the pathway-specific activator gene actII-ORF4. J Ind Microbiology & Biotechnology. 2002, 28: 103-111.View ArticleGoogle Scholar
- Floriano B, Bibb M: afsR is a pleiotropic but conditionally required regulatory gene for antibiotic production in Streptomyces coelicolor A3(2). Mol Microbiol. 1996, 21: 385-396. 10.1046/j.1365-2958.1996.6491364.x.PubMedView ArticleGoogle Scholar
- Hu Z, Hopwood DA, Khosla C: Directed transfer of large DNA fragments between Streptomyces species. App Environ Microbiol. 2000, 66: 2274-2277. 10.1128/AEM.66.5.2274-2277.2000.View ArticleGoogle Scholar
- Penn J, Li X, Whiting A, Latif M, Gibson T, Silva CJ, Brian P, Davis J, Miao V, Wrigley SK, Baltz RH: Heterologous production of daptomycin in Streptomyces lividans. J Ind Microbiol Biotechnol. 2006, 33: 121-128. 10.1007/s10295-005-0033-8.PubMedView ArticleGoogle Scholar
- Ayadi DZ, Chouayekh H, Mhiri S, Zerria K, Fathallah DM, Bejar S: Expression by Streptomyces lividans of the Rat alpha Integrin CD11b A-Domain as a Secreted and Soluble Recombinant Protein. J Biomed Biotechnol. 2007, 1: 54327-Google Scholar
- Fornwald JA, Donovan MJ, Gerber R, Keller J, Taylor DP, Arcuri EJ, Brawner ME: Soluble forms of the human T cell receptor CD4 are efficiently expressed by Streptomyces lividans. Biotechnology N Y. 1993, 11: 1031-1036. 10.1038/nbt0993-1031.PubMedView ArticleGoogle Scholar
- Gonzalez-Ceron G, Miranda-Olivares OJ, Servin-Gonzalez L: Characterization of the methyl-specific restriction system of Streptomyces coelicolor A3(2) and the role played by laterally acquired nucleases. F.E.M.S. Microbiol Lett. 2009, 301: 35-43. 10.1111/j.1574-6968.2009.01790.x.View ArticleGoogle Scholar
- Binnie C, Cossar JD, Stewart DI: Heterologous biopharmaceutical protein expression in Streptomyces. Trends Biotechnol. 1997, 15: 315-320. 10.1016/S0167-7799(97)01062-7.PubMedView ArticleGoogle Scholar
- Kieser TB, Buttner MJ, Chater KF, Hopwood DA: Practical Streptomyces Genetics. 2000, The John Innes Foundation, NorwichGoogle Scholar
- Nakashima N, Mitani Y, Tamura T: Actinomycetes as host cells for production of recombinant proteins. Microbial Cell Factories. 2005, 4: 7-10.1186/1475-2859-4-7.PubMed CentralPubMedView ArticleGoogle Scholar
- Jayapal KP, Lian W, Glod F, Sherman DH, Hu WS: Comparative genomic hybridizations reveal absence of large Streptomyces coelicolor genomic islands in Streptomyces lividans. BMC Genomics. 2007, 8: 229-10.1186/1471-2164-8-229.PubMed CentralPubMedView ArticleGoogle Scholar
- Kim CC, Joyce EA, Chan K, Falkow S: Improved analytical methods for microarray-based genome-composition analysis. Genome Biol. 2002, 3: RESEARCH0065-10.1186/gb-2002-3-11-research0065.PubMed CentralPubMedView ArticleGoogle Scholar
- Bucca G, Laing E, Mersinias V, Allenby N, Hurd D, Holdstock J, Brenner V, Harrison M, Smith CP: Development and application of versatile high density microarrays for genome wide analysis of Streptomyces coelicolor: characterization of the HspR regulon. Genome Biol. 2009, 10: R5-10.1186/gb-2009-10-1-r5.PubMed CentralPubMedView ArticleGoogle Scholar
- Hojati Z, Milne C, Harvey B, Gordon L, Borg M, Flett F, Wilkinson B, Sidebottom PJ, Rudd BAM, Hayes MA, Smith CP, Micklefield J: Structure, biosynthetic origin and engineered biosynthesis of calcium-dependant antibiotics from Streptomyces coelicolor. Chemistry & Biology. 2002, 9: 1175-1187.View ArticleGoogle Scholar
- Stachelhaus T, Mootz HD, Marahiel MA: The specificity-conferring code of adenylation domains in nonribosomal peptide synthetases. Chemistry & Biology. 1999, 6: 493-505.View ArticleGoogle Scholar
- Ball LJ, Goult CM, Donarski JA, Micklefield J, Ramesh V: NMR structure determination and calcium binding effects of lipopeptide antibiotic daptomycin. Organic and Biomolecular Chemistry. 2004, 2: 1872-1878. 10.1039/b402722a.PubMedView ArticleGoogle Scholar
- Grunewald J, Sieber SA, Mahlert C, Linne U, Marahiel MA: Synthesis and derivatization of daptomycin: a chemoenzymatic route to acidic lipopeptide antibiotics. J Am Chem Soc. 2004, 126: 17025-17031. 10.1021/ja045455t.PubMedView ArticleGoogle Scholar
- Weaver D, Karoonuthaisiri N, Tsai HH, Huang CH, Ho ML, Gai S, Patel KG, Huang J, Cohen SN, Hopwood DA, Chen CW, Kao CM: Genome plasticity in Streptomyces: identification of 1 Mb TIRs in the S. coelicolor A3(2) chromosome. Mol Microbiol. 2004, 51: 1535-1550. 10.1111/j.1365-2958.2003.03920.x.PubMedView ArticleGoogle Scholar
- Bey S-J, Tsou M-F, Huang C-H, Yang C-C, Chen CW: The homologous terminal sequence of the Streptomyces lividans chromosome and SLP2 plasmid. Microbiology. 2000, 146: 911-922.PubMedView ArticleGoogle Scholar
- Huang C-H, Chen C-Y, Tsai H-H, Chen C, Lin Y-S, Chen CW: Linear plasmid SLP2 of Streptomyces lividans is a composite replicon. Mol Microbiol. 2003, 47: 1563-1576. 10.1046/j.1365-2958.2003.03403.x.PubMedView ArticleGoogle Scholar
- Chen CW, Yu T-W, Chung H-M, Chou C-F: Discovery and characterization of a new transposable element, Tn4811, in Streptomyces lividans 66. J Bacteriol. 1992, 174: 7762-7769.PubMed CentralPubMedGoogle Scholar
- Eichenseer C, Altenbucher J: The very large amplifiable element AUD2 from Streptomyces lividans 66 has insertion sequence-like repeats at its ends. J Bacteriol. 1994, 176: 7107-7112.PubMed CentralPubMedGoogle Scholar
- Gehring A, Wang ST, Kearns DB, Storer NY, Losick R: Novel genes that influence development in Streptomyces coelicolor. J Bacteriol. 2004, 186: 3570-3577. 10.1128/JB.186.11.3570-3577.2004.PubMed CentralPubMedView ArticleGoogle Scholar
- Okamoto S, Lezhava A, Hosaka T, Okamoto-Hosoya Y, Ochi K: Enhanced expression of S-Adenosylmethionine synthetase causes overproduction of actinorhodin in Streptomyces coelicolor A3(2). J Bacteriol. 2003, 185: 601-609. 10.1128/JB.185.2.601-609.2003.PubMed CentralPubMedView ArticleGoogle Scholar
- Kim D-J, Huh J-H, Yang Y-Y, Kang C-M, Lee I-H, Hyun C-G, Hong S-K, Suh J-W: Accumulation of S-Adenosyl-L-methionine enhances production of actinorhodin but inhibits sporulation in Streptomyces lividans TK23. J Bacteriol. 2003, 185: 592-600. 10.1128/JB.185.2.592-600.2003.PubMed CentralPubMedView ArticleGoogle Scholar
- Xu D, Kwon H-J, Suh J-W: S-Adenosylmethionine induces BldH and activates secondary metabolism by involving the TTA-codon control of bldH expression in Streptomyces lividans. Arch Microbiol. 2008, 189: 419-426. 10.1007/s00203-007-0336-4.PubMedView ArticleGoogle Scholar
- Park H-S, Shin S-K, Yang Y-Y, Kwon H-J, Suh J-W: Accumulation of S-adenosylmethionine induced oligopeptide transporters including BldK to regulate differentiation events in Streptomyces coelicolor M145. F.E.M.S. Microbiol Lett. 2005, 249: 199-206. 10.1016/j.femsle.2005.05.047.View ArticleGoogle Scholar
- Shin S-K, Park H-S, Kwon H-J, Yoon H-J, Suh J-W: Genetic characterization of two S-adenosylmethionine induced ABC transporters reveals their roles in modulations of secondary metabolism and sporulation in Streptomyces coelicolor M145. J Microbiol Biotechnol. 2007, 17: 1818-1825.PubMedGoogle Scholar
- Pope MK, Green B, Westpheling J: The bldB gene encodes a small protein required for morphogenesis, antibiotic production, and catabolite control in Streptomyces coelicolor. J Bacteriol. 1998, 180: 1556-1562.PubMed CentralPubMedGoogle Scholar
- Mishig-Ochiriin T, Won H-S, Lee S-O, Lee B-J: Biophysical and structural property of the putative DNA-binding protein, BldB, from Streptomyces lividans. Biopolymers. 2003, 69: 343-350. 10.1002/bip.10387.PubMedView ArticleGoogle Scholar
- Pope MK, Green BD, Westpheling J: The bld mutants of Streptomyces coelicolor are defective in the regulation of carbon utilization, morphogenesis and cell-cell signalling. Mol Microbiol. 1996, 19: 747-756. 10.1046/j.1365-2958.1996.414933.x.PubMedView ArticleGoogle Scholar
- Eccleston M, Ali RA, Seyler R, Westpheling J, Nodwell J: Structural and genetic analysis of the BldB protein of Streptomyces coelicolor. J Bacteriol. 2002, 184: 4270-4276. 10.1128/JB.184.15.4270-4276.2002.PubMed CentralPubMedView ArticleGoogle Scholar
- Price DL, Manatunga AK: Modelling relative survival using transformation methods. Statistics in Medicine. 2004, 23: 2463-2477. 10.1002/sim.1832.PubMedView ArticleGoogle Scholar
- Borodina I, Krabben P, Nielsen J: Genome-scale analysis of Streptomyces coelicolor A3(2) metabolism. Genome Research. 2005, 6: 820-829. 10.1101/gr.3364705.View ArticleGoogle Scholar
- Khannapho C, Zhao H, Bonde BK, Kierzek AM, Avignone-Rossa CA, Bushell ME: Selection of objective function in genome scale flux balance analysis for process feed development in antibiotic production. Metabolic Engineering. 2008, 10: 227-233. 10.1016/j.ymben.2008.06.003.PubMedView ArticleGoogle Scholar
- Akiopiants K, Florova G, Li C, Reynolds KA: Multiple pathways for acetate assimilation in Streptomyces cinnamonensis. J Ind Microbiol Biotechnol. 2006, 33: 141-150. 10.1007/s10295-005-0029-4.View ArticleGoogle Scholar
- Han L, Reynolds KA: A novel alternative anaplerotic pathway to the glyoxylate cycle in Streptomycetes. J Bacteriol. 1997, 179: 5157-5164.PubMed CentralPubMedGoogle Scholar
- Large KP, Mirjalili N, Osborne M, Peacock LL, Zormaidis V, Walsh M, Cavanagh ME, Leadlay PF, Ison AP: Lipase activity in Streptomycetes. Enzyme and Microbial Technology. 1999, 25: 569-575. 10.1016/S0141-0229(99)00080-0.View ArticleGoogle Scholar
- Peacock L, Ward J, Ratledge C, Dickinson FM, Ison A: How Streptomyces lividans uses oils and sugars as mixed substrates. Enzyme and Microbial Technology. 2003, 32: 157-166. 10.1016/S0141-0229(02)00278-8.View ArticleGoogle Scholar
- Ensign SA: Revisiting the glyoxylate cycle: alternative pathways for microbial acetate assimilation. Mol Microbiol. 2006, 61: 274-276. 10.1111/j.1365-2958.2006.05247.x.PubMedView ArticleGoogle Scholar
- Erb TJ, Frerichs-Revermann L, Fuchs G, Alber BE: The apparent malate synthase activity of Rhodobacter sphaeroides is due to two paralogous enzymes, (3S)-malyl-coenzyme A (CoA)/β-methylmalyl-CoA lyase and (3S)-malyl-CoA thioesterase. J Bacteriol. 2010, 192: 1249-1258. 10.1128/JB.01267-09.PubMed CentralPubMedView ArticleGoogle Scholar
- Erb TJ, Fuchs G, Alber BE: (2S)-methylsuccinyl-CoA dehydrogenase closes the ethylmalonyl-CoA pathway for acetyl-CoA assimilation. Mol Microbiol. 2009, 73: 992-1008. 10.1111/j.1365-2958.2009.06837.x.PubMedView ArticleGoogle Scholar
- Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber W, Iacus S, Irizarry R, Leisch F, Li C, Maechler M, Rossini AJ, Sawitzki G, Smith C, Smyth G, Tierney L, Yang JY, Zhang J: Bioconductor: open software development for computational biology and bioinformatics. Genome Biol. 2004, 5: R80-10.1186/gb-2004-5-10-r80.PubMed CentralPubMedView ArticleGoogle Scholar
- Smyth GK, Speed TP: Normalization of cDNA microarray data. Methods. 2003, 31: 265-273. 10.1016/S1046-2023(03)00155-5.PubMedView ArticleGoogle Scholar
- R. [http://www.R-project.org]
- Rice P, Longed I, Bleasby A: EMBOSS: The European Molecular Biology Open Software Suite. Trends in Genetics. 2000, 16: 276-277. 10.1016/S0168-9525(00)02024-2.PubMedView ArticleGoogle Scholar
- Codon Usage Database. [http://www.kazusa.or.jp/codon/]
- Nakamura Y, Gojobori T, Ikemura T: Codon usage tabulated from the international DNA sequence databases: status for the year 2000. Nucl Acids Res. 2000, 28: 292-10.1093/nar/28.1.292.PubMed CentralPubMedView ArticleGoogle Scholar
- scoFBA. [http://sysbio.sbs.surrey.ac.uk/ScoFBA/index.html]
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (<url>http://creativecommons.org/licenses/by/2.0</url>), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.