Abstract
Bacterial lipoproteins are a class of extracellular proteins tethered to cell membranes by covalently attached lipids. Deleting the lipoprotein signal peptidase (lsp) gene in Streptomyces coelicolor results in growth and developmental defects that cannot be restored by reintroducing the lsp. We report resequencing of the genomes of the wild-type M145 and the cis-complemented Δlsp mutant (BJT1004), mapping and identifying secondary mutations, including an insertion into a novel putative small RNA, scr6809. Disruption of scr6809 led to a range of developmental phenotypes. However, these secondary mutations do not increase the efficiency of disrupting lsp suggesting they are not lsp specific suppressors. Instead we suggest that these were induced by introducing the cosmid St4A10Δlsp as part of the Redirect mutagenesis protocol, which transiently duplicates a number of important cell division genes. Disruption of lsp using no gene duplication resulted in the previously observed phenotype. We conclude that lsp is not essential in S. coelicolor but loss of lsp does lead to developmental defects due to the loss of lipoproteins from the cell. Significantly, our results indicate the use of cosmid libraries for the genetic manipulation of bacteria can lead to unexpected phenotypes not necessarily linked to the gene or pathway of interest.
Introduction
Bacterial lipoproteins are essential for building and maintaining the cell envelope and also provide a key interface with the external environment 1–3 Most lipoprotein precursors are exported as unfolded polypeptides via the Sec (general secretory) pathway but others can be exported via the twin arginine transport (Tat) pathway, which is typically utilised for the transport of fully folded proteins 4–6 The signal peptides of lipoproteins closely resemble other types of bacterial Sec and Tat signal peptide but they contain a characteristic lipobox motif, typically L−3-A/S−2-G/A−1-C+1, relative to the signal cleavage site, in which the cysteine residue is essential and invariant. The lipobox motif allows putative lipoproteins to be easily identified in bacterial genome sequences3,7.
Following translocation, lipoprotein precursors are firstly modified by covalent attachment of a diacylglycerol molecule, derived from a membrane phospholipid, to the thiol of the conserved lipobox cysteine residue via a thioether linkage. This reaction is catalysed by an enzyme named Lgt (Lipoprotein diacylglycerol transferase) and results in a diacylated lipoprotein. Lsp (Lipoprotein signal peptidase) then cleaves the signal sequence immediately upstream of the lipidated cysteine to leave it at the +1 position. These early steps in lipoprotein biogenesis are highly conserved and unique to bacteria making them potential targets for antibacterial drug development 2,8. In Gram-negative bacteria and Gram-positive Actinobacteria, lipoproteins can be further modified by addition of an amide-linked fatty acid to the amino group of the diacylated cysteine residue at the mature N-terminus. This final step is catalysed by the enzyme Lnt (Lipoprotein n-acyltransferase) and results in triacylated lipoproteins. In Gram-negative proteobacteria, Lnt modification is a pre-requisite for the recognition of lipoproteins by the Lol machinery, which transports lipoproteins to the outer membrane 2,9 but its function in monoderm Gram-positive bacteria is not known. Members of the Gram-positive phyla Firmicutes and Mollicutes also N-acylate lipoproteins despite lacking Lnt homologues and S. aureus can diacylate or triacylate individual lipoproteins in an environmentally dependent manner 10–14. These studies suggest that triacylation of lipoproteins in Gram-positive bacteria has an important role in their natural environment but is dispensable in vitro. Loss of Lnt activity in Streptomyces bacteria has no obvious effect on fitness or lipoprotein localisation in vitro but it does have a moderate effect on virulence in the plant pathogen Streptomyces scabies, supporting the idea that it has environmental importance 15.
We previously characterised all four steps of the lipoprotein biogenesis pathway in Streptomyces spp. (Figure 1) 5,15, which is one of the best studied genera in the Gram-positive phylum Actinobacteria. Our key findings are (i) that Tat exports ~20% of lipoprotein precursors in streptomycetes; (ii) they N-acylate lipoproteins using two non-essential Lnt enzymes; (iii) Streptomyces coelicolor encodes two functional copies of Lgt which cannot be removed in the same strain; (iv) lsp mutants can be isolated at low frequencies but they acquire spontaneous secondary mutations which might be lsp suppressors. It was recently reported that Lgt is essential in Mycobacterium tuberculosis, which is also a member of the phylum Actinobacteria, and that lgt deletion in the fast-growing species Mycobacterium smegmatis is accompanied by spontaneous secondary mutations 16. Natural product antibiotics that target the lipoprotein biogenesis pathway include globomycin, made by Streptomyces globisporus 2 and antibiotic TA made by Myxococcus xanthus 1,16. Both inhibit Lsp activity and are lethal to Escherichia coli but TA resistance arises through spontaneous IS3 insertion into the lpp gene, which encodes an abundant lipoprotein that attaches the E. coli outer membrane to the peptidoglycan cell wall 16,17. Over-expressing lsp also confers TA resistance in both E. coli and M. xanthus, and the latter encodes additional Lsp homologues within the TA biosynthetic gene cluster 17.
Deletion of S. coelicolor lsp results in very small and flat colonies that are delayed in sporulation and these lsp mutants could not be fully complemented even by reintroducing the lsp gene to its native locus. Although both cis and in trans complementation restored lipoprotein biogenesis and sporulation it did not restore the wild-type growth rate 5. There are two likely reasons for this: either lsp is essential and the mutant strains acquire secondary suppressor mutations, or the Redirect PCR targeting method that we used to delete the lsp gene resulted in chromosomal rearrangements and mutations independent of lsp. Here we provide evidence to support the second hypothesis and we demonstrate that introduction of the cosmid carrying an ~40 kb region of the S. coelicolor chromosome, including lsp, from E. coli to S. coelicolor transiently duplicates cell division and cell wall biosynthesis genes which leads to secondary mutations including disruption of a putative small RNA. We further confirm that lsp is non-essential but deletion of the lsp gene does lead to growth and developmental delays and the over-production of the antibiotic actinorhodin in S. coelicolor, as observed previously. These phenotypes must therefore be due to the loss of lipoproteins from its cytoplasmic membrane.
Results
Mapping secondary mutations in the cis complemented Δlsp strain BJT1004
We previously reported that the S. coelicolor Δlsp mutant BJT1001 cannot be complemented even by restoring lsp to its native locus 5. Since cis complementation should effectively restore the genome to wild-type this suggests that other spontaneous mutations have occurred during the genetic manipulations. To test this we Illumina sequenced the genomes of the parent strain S. coelicolor M145 and the cis-complemented Δlsp strain BJT1004 using two independent companies (GATC Biotech and The Genome Analysis Centre). Across the four sequence samples a total of 51 unique single nucleotide polymorphisms (SNPs) were detected (Supplementary Table S1 online) as well as a chromosomal rearrangement in BJT1004 that is not present in the parent strain M145 (Figure 2(A-B) and Supplementary Text S1 and S2 online). Of the 51 SNPS, 13 are unique to one of the BJT1004 sequences, with 4 residing inside coding regions. However of all of these only one SNP occurs in both BJT1004 sequences and this is in the intergenic region between sco5331 and sco5332. In the single chromosomal rearrangement, the IS21 insertion element (genes sco6393 and sco6394) has inserted into the intergenic region between the sco6808 and sco6809 genes and this was confirmed by PCR (Figure 2(A-C)). Although this might affect the downstream promoter of sco6808, which encodes a putative regulator, deletion of sco6808 (using vector pJM010) had no effect on growth or development under standard laboratory conditions (Figure 3(A)). The intergenic position of IS21 in BJT1004 suggested it might disrupt a non-coding RNA and examination of RNA sequence data for S. coelicolor M145 confirmed the presence of a 189 nt transcript initiating 107 bp upstream of sco6808 and reading into the last 82 nucleotides of the sco6809 gene (data from GSM1121652 and GSM1121655 RNA sequencing; Supplementary Text S1 and 2 online; Figure 2(A-B)). Following convention we named this putative small RNA scr6809 for S. coelicolor RNA 6809. Deletion of the scr6809 (pJM012) sequence (without disrupting either the sco6808 or sco6809 coding sequences) resulted in a range of phenotypes from colonies that look like wild-type to non-sporulating bald and white mutants defective in aerial hyphae formation and sporulation, respectively, antibiotic overproducers and small slow growing colonies. Restreaking the Δscr6809 colonies (double crossovers) with wild-type appearance again gave rise within the next generation to a range of colonies with different morphologies, including growth and developmental defects (Figure 3(B)). Colonies with mutated morphologies would retain that morphology in subsequent generations indicating another situation where spontaneous secondary mutations are arising. A previous report showed that a sco6808 deletion mutant had accelerated production of actinorhodin and undecylprodigiosin as well as precocious spore formation on R5 medium 18. There was no observable difference between the wild-type and Δsco6808 strains under the growth conditions used here but disruption of sco6808 in strain BJT1004 resulted in an improvement in sporulation (Figure 3(A)). We suspect this difference is based on the recovery of the scr6809 loci to wildtype as result of the double crossovers between the chromosome of BJT1004 and the Δsco6808 deletion cosmid St1A2Δsco6808. This was also seen for St1A2Δsco6811 disruptions in each background (not shown).
To determine whether IS21 insertion into scr6809 is induced by deletion of lsp, we isolated ten more non-clonal lsp mutants by introducing cosmid St4A10Δlsp (pJM014) into wild-type strain M145 and then PCR amplified the intergenic region between sco6808 and sco6809. The size of the PCR products matched the predicted wild-type size and indicated that none of these lsp mutants contain an IS21 insertion suggesting that the original observation is not specific to lsp mutants (Figure 4). Consistent with this conclusion, the frequency with which lsp mutants could be isolated was not increased in BJT1004 relative to M145 suggesting that none of the mapped mutations in BJT1004 suppress fitness defects that arise from deleting Δlsp. Attempts at over-expressing scr6809 using pJM017 in S. coelicolor M145, S. scabies 87-22 and S. venezualae ATCC 10712 resulted in no observable phenotype but as the same vectors failed to prevent accumulation of developmental phenotypes in the Δscr6809 strain, this suggests that functional scr6809 is not expressed from these vectors.
Cumulatively these results suggested that deletion of lsp does not result in secondary mutations and prompted us to hypothesise that these accumulate as a result of duplicating cell division genes on cosmid St4A10 which was used to delete lsp. These results further suggest a role for scr6809 in S. coelicolor differentiation, although there is no obvious link to lsp and so this was not pursued further here.
Introduction of wild-type St4A10 results in a pleiotropic phenotype
The Redirect PCR-targeting method uses E. coli as a host strain for an S. coelicolor cosmid library which can be used to make targeted deletions 19,20. The Redirect method was used to PCR-target the lsp gene sco2074 on cosmid St4A10, which contains a ~40 kb region of the S. coelicolor genome spanning genes sco2069-2104 (Supplementary Table S2 online). Conjugation of St4A10Δlsp into S. coelicolor transiently duplicates all the genes on that cosmid (except lsp) and because this region includes many important cell division genes (ftsZ, ftsQ, ftsW, ftsI and ftsL) and essential cell wall synthesis genes (murG, murD, murX, murF and murE) we reasoned that over-expression of these genes, rather than deletion of lsp, is responsible for the spontaneous secondary mutations and the resulting pleiotropic phenotype. To test this idea we introduced an origin of transfer into the wild-type St4A10 cosmid backbone and then conjugated this cosmid into strain M145 and selected for single cross-over events where the whole cosmid is integrated into the chromosome, thus duplicating the S. coelicolor genes on St4A10. Analysis of these single crossover strains, maintained on kanamycin to select for the cosmid, revealed them to be genetically unstable, with many initially appearing similar to the observed Δlsp phenotype, i.e. small and delayed in sporulation (Figure 5). However, they lack the characteristic Δlsp overproduction of the blue antibiotic actinorhodin and colonies from this M145::St4A10 strain also acquired more significant developmental issues upon prolonged maintenance and restreaking on MS agar containing kanamycin (not shown). This suggests they accumulate spontaneous secondary mutations as a direct result of carrying St4A10 and that the observed Δlsp phenotype is at least in part due to duplication of the genes on cosmid St4A10. This is consistent with the fact that complementation of Δlsp restored lipoprotein biogenesis but did not restore wild-type colony morphology 5.
Targeted deletion of lsp results in a small colony phenotype
To test how much deletion of lsp contributed to the phenotype of BJT1001 (the Δlsp strain generated using Redirect) we undertook a targeted disruption of lsp in wild-type strain M145 using a suicide vector, which does not duplicate or affect any other coding sequences. The lsp suicide vector, pJM016 (Table 1), was introduced into wild-type S. coelicolor by conjugation and ex-conjugants were selected by growing on MS agar plates containing apramycin. Following introduction of the pJM016, two colony types were observed (Figure 6), one with wild-type appearance while the others were small colonies that over-produce actinorhodin, reminiscent of the lsp mutant BJT1001. PCR testing of the genomic DNA of both morphotypes revealed that those with the wild-type colony morphology have a wild-type fully functioning lsp gene whereas those with a small colony phenotype have disruptions in lsp caused by pJM016. PCR amplification followed by sequencing of the loci in the small colony variant revealed an interesting and unexpected recombination event had occurred: the vector and almost all the lsp gene have been removed such that all that remains is the apramycin resistance cassette (Supplementary Text S4 and S5 online). These data confirm that lsp is not essential in S. coelicolor but loss of Lsp does result in a growth and developmental defect and overproduction of the blue antibiotic actinorhodin as observed previously 5.
Discussion
The pleiotropic nature of the original S. coelicolor Δlsp strain BJT1001 resulted primarily from the introduction of cosmid St4A10, most likely caused by the over-expression of cell division and cell wall biosynthesis genes carried on that cosmid (Supplementary Table S2 online). It seems likely, but is not proven, that this led to the secondary mutations we observed in this strain. These secondary mutations do not make it easier to delete lsp suggesting they are not lsp-specific suppressors. Genetic manipulation has always been challenging in Streptomyces bacteria and the Redirect PCR targeting method has been a significant development but this work should be a cautionary tale to others to consider the effects of using large insert cosmid libraries in the genetic manipulation of bacteria. Recent advances in CRISPR/Cas9 editing of Streptomyces genomes 21 negate the need for a cosmid library and these techniques will accelerate research into the basic biology of Streptomyces and other filamentous actinomycetes. This is vital because the secondary metabolites derived from these bacteria still represent a major underutilised reservoir from which new antibiotics and other bioactive natural products can be discovered. Moreover, the identification here of the novel small RNA scr6809 and demonstration that its deletion results in a range of growth and developmental defects add to the growing appreciation 22–24 of the significance of small RNAs in streptomycetes.
Materials and Methods
Bacterial strains and culture conditions
All primers, plasmids and strains used are listed in Table 1. Strains were routinely grown as previously described 5 following the recipes of Kieser et al., (2000). E. coli was grown in LB or LB–NaCl for Hygromycin selection and S. coelicolor M145 and its derivatives were grown on Soya Flour Mannitol (SFM) medium to study growth and development or LB culture for genomic isolations.
Gene deletions and complementation
Gene deletions were carried out following the Redirect method of PCR-targeting 26 as previously described Hutchings et al. (2006). Disruption of lsp (sco2074::apr) on cosmid St4A10 (pJM013, St4A10Δlsp) using the pIJ773 apramycin disruption-cassette and sco6808 (sco6808::apr) and scr6809 (scr6809::apr) on cosmid St1A2 (pJM010 - St1A2Δsco6808 and pJM012 - St1A2Δsco6808 respectively) using primers JM0101-2, JM0083-84 and JM0091-2 respectively were confirmed by PCR using primers JM0150-1, JM0085-6 and JM0093-4 respectively. Introduction of the wild-type cosmid St4A10 was facilitated by introducing an oriT by disruption and replacement of the Supercos-1 backbone bla resistance gene (pJM014 – St4A10bla::hyg) using primers JM0095-6 and the hygromycin disruption cassette from pIJ10701, confirmed using primers JM0099-100. The lsp suicide vector pJM016 was produced by introducing a 411 bp fragment of the lsp gene with an N-terminal BamHI site, amplified with primers JM0117-8 and cloned into pGEM T-Eazy. The BamHI site was then used to subclone the BamHI fragment from a pIJ773 digest, containing an apr disruption cassette. An overexpression construct, pJM017 was synthesised by Genscript to include the pMC500 MCS and terminators 28 with scr6809 (sequence included in Supplementary Text S5 online). All constructs were subsequently conjugated into S. coelicolor following the method described by Gust et al. (2002).
Genomic DNA isolation
Genomic DNA was isolated from M145 and BJT1004 following the Pospiech and Neumann (1995) salting out method as described by Keiser et al. (2000). Mycelium from a 30 ml culture was resuspended in 5 ml SET buffer containing 1 mg/ml lysozyme and incubated at 37°C 30-60 min. To this lysate, 140 μl of proteinase K solution (20 mg ml−1) was added, mixed, then 600 μl of 10% SDS added, mixed and incubated at 55°C for 2 h, with occasional mixing throughout. After this incubation 2 ml of 5 M NaCl was added, mixed and left to cool to 37°C before adding 5 ml chloroform, mixed at 20°C for 30 min. Samples were centrifuged at 4500 x g for 15 min at 20°C. The supernatant was removed to a fresh tube and DNA precipitated by adding 0.6 volumes of 100% isopropanol. Tubes were mixed by inversion and after at least 3 min DNA spooled out using a sterile Pasteur pipette. The DNA was rinced in 70% ethanol, air dried and dissolved in 1-2 ml TE buffer (10 mM Tris-HCl pH 7.8, 1 mM EDTA) at 55°C.
Genome resequencing and secondary mutation identification
The isolated DNA from our wild-type S. coelicolor M145 parent strain and BJT1004 were sent to both GATC Biotech and The Genome Analysis Centre (TGAC) for 35bp paired end HiSeq Illumina sequencing. Assembly mapping and SNP identification was carried out with MIRA (Chevreux et al., 2004) using the reference genome NC_003888 (Bentley et al., 2002) as a scaffold for mapping each of the resequenced genomes. Putative SNPs were detected in each sample independently reporting the SNP position, the nucleotide change, the number of reads that sequence the region, those containing wild-type or mutated nucleotides and a percentage change. Each set of results was then compared by eye to determine the likely hood that a SNP was real by number of reads containing the mutation and its presence in each sample. Larger mutations (rearrangements) were identified in the same fashion.
Microscopy
Brightfield images were acquired using a Zeiss M2 Bio Quad SV11 stereomicroscope. Samples were illuminated from above using a halogen lamp images captured with an AxioCam HRc CCD camera. The AxioVision software (Carl Zeiss, Welwyn Garden City, UK) was used for image capture and processing.
Acknowledgments
We thank Marie Elliot for sharing unpublished RNA sequence data for Streptomyces coelicolor M145. This work was supported by a NERC PhD studentship to JTM and BBSRC grants BB/F009429/1 and BB/F009224/1 to MIH and TP, respectively.
Author contribution
JTM designed and carried out the experiments, JTM ICS TP MIH designed experiments, JTM DAW ICS TP MIH analysed data and wrote the manuscript, DAW prepared DNA for sequencing, GC analysed sequencing data.
Additional information
Supplementary information accompanies this paper at:
Competing financial interests: The authors declare no competing financial interests.