ABSTRACT
We recently showed that type II signal peptidase (SPaseII) encoded by lspA is the target of an antibiotic called TA (myxovirescin), which is made by Myxococcus xanthus. SPaseII cleaves the signal peptide during bacterial lipoprotein processing. Bacteria typically contain one lspA gene; however, strikingly, the M. xanthus DK1622 genome contains four (lspA1 to lspA4). Since two of these genes, lspA3 and lspA4, are located in the giant TA biosynthetic gene cluster, we hypothesized they may play a role in TA resistance. To investigate the functions of the four M. xanthus lspA (lspAMx) genes, we conducted sequence comparisons and found that they contained nearly all the conserved residues characteristic of SPaseII family members. Genetic studies found that an Escherichia coli ΔlspA mutation could be complemented by any of the lspAMx genes in an lpp mutant background, but not in an E. coli lpp+ background. Because Lpp is the most abundant E. coli lipoprotein, these results suggest the M. xanthus proteins do not function as efficiently as the host enzyme. In E. coli, overexpression of each of the LspAMx proteins conferred TA and globomycin resistance, although LspA3 conferred the highest degree of resistance. In M. xanthus, each lspAMx gene could be deleted and was therefore dispensable for growth. However, lspA3 or lspA4 deletion mutants each exhibited a tan phase variation bias, which likely accounts for their reduced-swarming and delayed-development phenotypes. In summary, we propose that all four LspAMx proteins function as SPaseIIs and that LspA3 and LspA4 might also have roles in TA resistance and regulation, respectively.
INTRODUCTION
Myxobacteria are a group of common soil-dwelling deltaproteobacteria. Whole-genome sequencing of several myxobacterium species has revealed that they contain a large reservoir of secondary-metabolite-biosynthetic gene clusters. For instance, ∼10% of their genomes can be dedicated to this function (1). Thus, myxobacteria produce a prolific number of secondary metabolites and in this regard are similar to actinomycetes and certain fungi. Importantly, myxobacterial secondary metabolites frequently have novel structures and distinct modes of action (MOAs) (1). The antibiotic TA, also known as myxovirescin, is one such small-molecule natural product produced by the common laboratory strain Myxococcus xanthus DK1622 (2–4). TA is a macrocyclic antibiotic with rapid bactericidal activity and is active against many Gram-negative and some Gram-positive bacteria (3–7). We have recently shown that TA functions as a weapon in microbial predation and that its MOA is as an inhibitor of type II signal peptidase (SPaseII), which plays an essential role in bacterial lipoprotein maturation (2).
Lipoproteins are ubiquitously found in bacteria and are covalently modified at their N termini with a lipid, which anchors the proteins into the membrane. Lipoprotein maturation requires a group of three enzymes that act sequentially on the periplasmic side of the cytoplasmic membrane (8). Precursor lipoproteins contain four consensus amino acids, called the lipobox, in their signal sequences (9). After secretion through the Sec (or in some cases Tat) pathway, the lipobox serves as a recognition sequence for lipid attachment (10, 11). Here, the invariant Cys residue is covalently modified with the addition of a diacylglycerol group by the prolipoprotein diacylglyceryl transferase (Lgt) in the first enzymatic step. Next, the signal peptide is cleaved by LspA, so that the Cys becomes the N-terminal +1 residue. Once freed from the signal peptide, this N-terminal Cys residue can be further acylated at the α-amino position by the apolipoprotein N-acyltransferase (Lnt) in the final maturation step.
The lipoprotein-processing system (Lgt, LspA, and Lnt) is widely present and is essential in all Gram-negative bacteria analyzed to date, whereas in Gram-positive bacteria, this system can be conditionally essential (8, 12–14). In addition to playing an important role in bacterial viability, LspA also plays a key role in pathogenesis, because many virulence factors are lipoproteins or require lipoprotein functions (13, 15). In contrast, LspA is absent from eukaryotes. For these reasons, coupled with the discoveries of two inhibitors, TA and globomycin (GLM) (2, 16), LspA makes an attractive target for antibiotic discovery.
Escherichia coli LspA contains 164 amino acids with a molecular mass of 18 kDa. It is an integral inner membrane protein with four transmembrane helixes, and both the N and C termini are localized in the cytoplasm (17). Using Bacillus subtilis LspA as a model, Tjalsma and collaborators predicted that it belongs to a novel family of aspartic proteases (18). They also showed that two conserved Asp residues are essential and likely serve a catalytic function during signal peptide cleavage.
Bacteria typically contain only one lspA gene (12). However, in some bacteria, such as Listeria monocytogenes and Staphylococcus spp., a second lspA gene is found. Strikingly, the M. xanthus DK1622 genome contains four lspA genes (2, 19). Two of the lspA genes are in tandem and reside in an apparent operon with lgt. The other two lspA genes reside within the biosynthetic gene cluster that makes the antibiotic TA. Here, we undertook genetic approaches to understand the biological functions of the four M. xanthus LspA proteins.
MATERIALS AND METHODS
Bacterial strains and growth conditions.The strains and plasmids used in this study are listed in Table 1. M. xanthus was routinely grown in the dark at 33°C in CTT (1% Casitone, 10 mM Tris-HCl, pH 7.6, 8 mM MgSO4, 1 mM KH2PO4, pH 7.6). After initial growth, plates were sometimes stored in the dark at room temperature to prolong the viability of the strains. E. coli was routinely grown in LB medium at 37°C. The antibiotic concentrations used were as follows: kanamycin (Km) at 50 μg/ml, chloramphenicol (Cm) at 25 μg/ml, and ampicillin (Amp) at 100 μg/ml. When needed, arabinose or glucose was supplemented at 0.2% (wt/vol) or as indicated, and galactose (Gal) was supplemented at 1%.
Strains and plasmids
Strain and plasmid construction.Markerless in-frame deletions of the four M. xanthus lspA (lspAMx) genes were constructed in strain DK1622. These mutations were made by a gene replacement method with the pBJ113 vector, which contains a positive-negative (Kmr-galK) selection cassette (20). To construct the lspA1 single-gene deletion, an upstream fragment (722 bp) and a downstream fragment (734 bp) of lspA1 were PCR amplified and cloned into pBJ113 at the BamHI and HindIII restriction sites to generate pYX341. This plasmid was then electroporated into DK1622, and homologous recombinants were selected based on Km resistance (Kmr). Subsequent plasmid resolution and loss were counterselected on 1% galactose CTT agar (1.5%) plates. This markerless in-frame deletion, which generated strain DW1224, removed 579 bp of the lspA1 sequence and was verified by PCR with flanking primers and primers that were internal to the deleted region. Markerless in-frame deletions in the other three lspAMx genes were constructed in a similar manner. The primers are listed in Table S1 in the supplemental material.
The complete open reading frames of each M. xanthus lspA gene and the single E. coli lspA gene were PCR amplified and cloned into the pBAD30 vector (Ampr) at the EcoRI and SalI sites (21). The same ribosomal binding site (underlined) and spacing sequence (5′-AGGAGGTCTGCCTG-3′) were engineered upstream of the initiation codons of the open reading frames. The resulting constructs were verified by restriction digestion and DNA sequencing. The primers used are listed in Table S1 in the supplemental material.
A chromosomal E. coli lspA (lspAEc) deletion allele was constructed using the λ-Red system in strain YX231 (MG1655 pKOBEG pBAD30-lspAEc), which expresses the lspAEc gene from a plasmid-carried PBAD promoter (Table 1) (21–23). Expression of the Red recombinase was induced with arabinose from pKOBEG (Cmr) at 30°C. Linear PCR products containing a Kmr cassette and homologous sequences flanking the lspAEc gene were transformed into YX231 and recombined into the genome by selecting for Kmr on LB agar plates supplemented with Km, Amp, and arabinose at 37°C, under which the temperature-sensitive pKOBEG plasmid is cured. Recombinants were then screened by colony PCR and were also assayed for arabinose-dependent growth. For the latter test, isolated colonies were inoculated in LB containing Km and Amp with no arabinose and were grown for 6 h at 37°C. Next, a 10-fold serial dilution of each culture was carried out, followed by replica transfer onto LB Km plates with or without arabinose. Bacterial growth was scored after overnight incubation at 37°C. Primers EclspA-KO-F and EclspA-KO-R were used to verify the deletion of the wild-type (WT) allele, and primers K1, K2, and Kt were used to verify the presence of the insertion cassette (23). A strain containing a verified chromosomal lspA deletion/insertion mutant was identified and named YX238.
Complementation analysis.To test if M. xanthus LspA proteins could substitute for the function of E. coli LspA, we moved the ΔlspA::Kan mutation from YX238 into E. coli strains that expressed individual M. xanthus lspAMx clones from a PBAD plasmid. To ensure efficient homologous recombination of linear chromosomal DNA, the pKOBEG plasmid was first transformed into each test strain. Subsequently, genomic DNA from YX238 (isolated with a PureLink Genomic DNA Mini Kit from Invitrogen) was electroporated into the test strains by selecting for Kmr. Strains expressing lspAEc (YX231 and YX354) were used as positive controls, and strains containing the pBAD30 vector only (YX349 and YX353) were used as negative controls. The competency of E. coli cells was tested with control genomic DNA isolated from DW150 (malF3180::Tn10-Kan). Electrocompetent cells were prepared at 30°C with arabinose to allow expression of Red recombinase. To optimize this protocol for linear genomic DNA transformation, pilot experiments were first done using control strains and DNA at concentrations of 100, 200, 500, 1,000, and 1,500 ng per transformation (data not shown). These results showed that 500 ng of DNA yielded a maximum number of transformants (hundreds), which was used in subsequent transformations. In total, 40 μl of competent cells was mixed with 500 ng of genomic DNA in a 1-mm cuvette and electroporated at 1.8 kV.
Bioinformatics analysis.The amino acid sequence alignments were done using MUSCLE (24), and phylogenetic trees were constructed using PhyML 3.0 (25).
Arabinose dose-response growth assay.E. coli YX238 was grown overnight in LB medium supplemented with Km, Amp, and arabinose. The culture was then diluted 1:100 in the same medium, except arabinose was replaced with glucose and shaken for 2.5 h at 37°C. A 2.5-h incubation in glucose was empirically determined to significantly deplete cells of LspA without causing cell lysis (data not shown). The optical density at 600 nm (OD600) of this culture was adjusted to 0.1, and then the culture was divided into seven cultures, which were supplemented with arabinose at concentrations of 0, 0.02%, 0.05%, 0.08%, 0.1%, 0.15%, and 0.2% (wt/vol), followed by shaking at 37°C. Culture turbidity was monitored, and growth curves were plotted.
MICs.To determine MIC values, 2-fold serial dilutions of TA or globomycin (GLM) were carried out in 96-well microtiter dishes. DK1622 was tested over a dilution range of 0.5 to 128 μg/ml in CTT broth. The inocula contained 5 × 105 CFU/ml in a final volume of 180 μl. Both antibiotic-only and strain-only aliquots were included as controls. Microtiter dishes were incubated at 33°C for 2.5 days, and the MIC was visually scored as the lowest antibiotic concentration that resulted in no growth. MIC values for E. coli strains were determined in a similar manner, except LB medium with Amp and arabinose was used. Microtiter dishes were scored after incubation at 33°C for 18 h.
Zone of inhibition (ZOI) assay.Overnight M. xanthus cultures were collected by centrifugation and resuspended in TPM buffer (10 mM Tris-HCl, pH 7.6, 1 mM KH2PO4, 8 mM MgSO4) to a density of 3 × 108 CFU/ml. A 5-μl aliquot of cells was spotted on CTT 1% agar plates. The plates were incubated at 33°C for 48 h prior to the addition of a soft agar (SA) overlay with E. coli as the indicator strain (100 μl of cells at an OD600 of 1 mixed with 3 ml of molten LB 0.7% agar). The plates were then incubated overnight at 33°C before observation.
Development and motility assays.Fruiting body assays were conducted on TPM 1.5% agar plates, and motility assays were conducted on CTT hard agar (HA) (1%) and CTT soft agar (0.5%) plates. In all cases, overnight M. xanthus cultures were collected by centrifugation and resuspended in TPM buffer to a calculated cell density of 3 × 109 CFU/ml. A 5-μl aliquot of cells was then spotted on the respective plates. The plates were observed, and micrographs were taken with an Olympus SZX10 stereomicroscope coupled with a digital imaging system (QCapture Pro 6.0; QImaging) at the indicated times. All assays were done in triplicate and were repeated independently at least twice.
RESULTS
The Myxococcus fulvus and M. xanthus genomes each carry four lspA genes.We first determined that the M. xanthus DK1622 genome contains four lspA genes, which we named numerically. As described below, we found that other myxobacterial genomes contain only one or two LspA homologs, so we named the most widely distributed ortholog LspA1 and the second most widely distributed one LspA2. This nomenclature differs from our prior use (2). lspA1 (MXAN_0368) and lspA2 (MXAN_0369) are located in the same apparent operon with lgt (MXAN_0370) (Fig. 1A). Because these genes are all predicted to function in lipoprotein processing, we predict they perform housekeeping functions. In contrast, the other two genes, lspA3 (MXAN_3944) and lspA4 (MXAN_3930), may play redundant and/or alternative roles, as they are located in the ta biosynthetic gene cluster (Fig. 1A).
Genetic map of the ta biosynthetic gene cluster and the four lspA genes. (A) In M. xanthus, lspA1 and lspA2 are separate from the ta gene cluster. The map is not drawn to scale. (B) In M. fulvus, the ta gene cluster is absent, but the lspA3 and lspA4 genes are retained. The dashed gray arrows indicate homologous genes found between M. xanthus and M. fulvus. The hatched boxes represent predicted promoter regions.
Next, we surveyed other sequenced myxobacterial genomes to see how many lspA genes they might contain. Interestingly, the myxobacterium M. fulvus strain HW-1 also carries four lspA genes (26), which we similarly named lspA1 (LILAB_06825), lspA2 (LILAB_06820), lspA3 (LILAB_27850), and lspA4 (LILAB_27845) (Fig. 1B). These orthologs show similar genome arrangements (Fig. 1) and share high sequence identity with the respective M. xanthus genes. At the amino acid level, identities between these orthologs are as follows: LspA1, 90%; LspA2, 89%; LspA3, 83%; and LspA4, 84%. Similar to M. xanthus, the M. fulvus genes lspA3 and lspA4 are in close proximity to each other and are divergently transcribed. In M. fulvus, however, the ta biosynthetic gene cluster is absent, and therefore, lspA3 and lspA4 are adjacent in the M. fulvus genome. Sequence comparisons of the loci near lspA3 and lspA4 showed that the upstream and downstream genes in M. fulvus are similar to the corresponding regions around the ta cluster in M. xanthus (Fig. 1B, dashed arrows). One possible explanation for the M. fulvus genomic arrangement is that in an ancestor strain(s), the remainder of the ta gene cluster was deleted, which likely would have involved at least two deletion events, as the presumed internal lspA3 gene remains. Alternatively, the ta gene cluster could have inserted into an ancestral M. xanthus strain, again presumably in a two-step process.
Surveys of other publically available complete myxobacterial genomes, as of July 2013, revealed that they all contained orthologs of LspA1 and that orthologs of LspA2 were present in four other myxobacterial genomes. In contrast, there were no orthologs of LspA3 and LspA4 in the other myxobacterial genomes. A phylogenetic tree was constructed to investigate sequence relationships among select LspA family members (Fig. 2A). This analysis showed that the LspA1-LspA3 and LspA2-LspA4 protein pairs are, respectively, more similar to each other than the other proteins, possibly suggesting an ancient duplication event followed by sequence divergence. In support of this hypothesis, the lspA1-lspA2 and lspA3-lspA4 gene pairs are adjacent to each other in the M. fulvus genome; although in the latter case, the genes are divergently transcribed. However, given that the four lspA genes are significantly divergent from each other and from homologs outside the myxobacteria group, it is not clear if they represent paralogs or xenologs of each other.
M. xanthus LspA protein sequences are divergent. (A) Phylogenetic tree of selected LspA homologs. (B) Amino acid sequence alignment (MUSCLE) of M. xanthus LspA1 to LspA4 with homologs from E. coli and B. subtilis. The height and darkness of the bars indicate the degree of conservation among amino acid residues, with the black bars corresponding to the previously identified 15 invariant amino acids (18). M. xanthus LspAs contain almost all the conserved amino acid residues except for the natural Gly-to-Ala substitutions in LspA3 and LspA4, which are boxed. The stars above the sequence indicate putative catalytic residues for protease activity. Sequence identity relative to E. coli LspA (percent identity, the number of identities/total number of nongap positions) is shown at the right.
LspAMx proteins contain conserved residues, but the sequences are divergent.An analysis of the deduced amino acid sequences for 19 known SPaseII enzymes revealed 15 conserved residues (18) (Fig. 2B). Systematic alanine scanning mutagenesis studies in B. subtilis showed that residue Asp-14 is required for LspA stability and that Asp-102 and Asp-129 are putative catalytic sites for protease function (18). Therefore, we included both the E. coli and B. subtilis prototypic LspA sequences in our sequence analysis of the M. xanthus proteins (Fig. 2B). Although these six protein sequences are divergent, the previously reported 15 conserved residues are nearly all retained in the M. xanthus proteins (Fig. 2B, black bars). LspA1 and LspA2 indeed contained all 15 conserved residues, whereas LspA3 and LspA4 contained all the conserved residues except for a Gly-to-Ala substitution in LspA3 at position 58 and the same substitution in LspA4 at position 107. These two changes are not, however, critical, as identical substitutions were made in B. subtilis SPaseII and were functional (18).
Construction of a PBAD-lspAEc allele results in conditional E. coli growth.To investigate the functionality of LspAMx proteins, we engineered a tool strain of E. coli that would conditionally express LspA. This step was necessary because LspA is essential for E. coli growth (8). The resulting strain, YX238, had the chromosomal copy of lspAEc deleted, and a plasmid copy of lspAEc was expressed from the PBAD promoter (21). As predicted, strain YX238 showed arabinose-dependent growth in independent assays. First, in an assay for growth dynamics, YX238 failed to grow in the absence of arabinose (Fig. 3). However, when arabinose was added to the culture medium at 0.02 to 0.2%, there was a corresponding increase in growth, and 0.15% and 0.2% arabinose supported the best growth (Fig. 3). In an earlier study, we showed that when LspA activity was inhibited by the antibiotic TA or GLM, the morphology of E. coli cells changed, so that regions in the cell became translucent (plasmolysis), indicating that patches of the inner membrane had become detached from the cell wall (2). As a second measure of a conditional phenotype, when YX238 cells were grown in the absence of arabinose (4 h) to deplete LspA, it resulted in the aforementioned morphological changes that phenocopied TA/GLM antibiotic treatment (Fig. 3, inset) (2). Third, on LB agar plates, an analogously constructed E. coli strain (YX354; chromosomal ΔlspA::Kan allele with complementing PBAD-lspAEc plasmid) also showed arabinose-dependent growth (Fig. 4).
Growth of strain YX238 (PBAD-lspAEc ΔlspA::Kan) is arabinose dependent. Shown are arabinose dose-response growth curves after LspA depletion (arabinose starvation for 2.5 h). The inset shows the morphology of YX238 after 4 h of arabinose (LspA) depletion. The arrow highlights one cell in which plasmolysis has occurred.
M. xanthus LspA proteins can substitute for E. coli LspA function. The E609L (lpp mutant) parental strain was constructed with the chromosomal ΔlspA::Kmr allele, and strains contained the indicated lspA genes cloned into pBAD30 complementing plasmids. The strains were 10-fold serially diluted on LB plates with and without arabinose. The strains are listed in Table 1.
M. xanthus LspA proteins complement an E. coli ΔlspA mutant in the absence of Lpp.To test LspAMx functionality, cross-species complementation experiments were conducted by moving, via electroporation, the ΔlspA::Kan allele from YX238 into E. coli strains that expressed the various lspAMx alleles. Thus, if the ΔlspA::Kan allele was successfully transferred into the recipient strain, it would suggest that the LspAMx protein substitutes for the essential function of LspAEc and therefore functions as a SPaseII. For these studies, genomic DNA was purified from YX238 and from DW150, which has a Tn10-Kan insertion in the nonessential malF gene. The latter DNA was used as a control to monitor transformation/recombination efficiencies in each strain. Recipient E. coli strains harboring pBAD30-lspAEc or the pBAD30 vector only were used as positive and negative controls, respectively. To optimize homologous-recombination efficiencies, recipient strains expressed λ-Red.
Control transformations of the six MG1655-based (pKOBEG) E. coli strains (YX280, YX286, YX282, YX326, YX285, and YX284) with malF::Tn10-Kan genomic DNA yielded similar numbers of Kmr colonies for each strain, indicating that all six strains were equally competent. In contrast, in the experimental group, only the positive-control strain harboring pBAD30-lspAEc (YX286) yielded a high number of transformants (thousands). The test and negative-control strains yielded few transformants (<20), suggesting that the four lspAMx genes did not complement an E. coli lspA deletion. Indeed, when the transformants from the experimental group were inoculated into LB broth containing Km, Amp, and arabinose, they grew poorly or failed to grow in an arabinose-dependent manner (data not shown), indicating that they were aberrant recombinants. These transformation experiments were repeated at least three independent times with 500 ng or more of genomic DNA. The amount of arabinose was also increased to 0.4%; however, in all cases, the four lspAMx genes failed to complement (Table 2).
Complementation of ΔlspA::Kan transfer
We hypothesized that one reason complementation failed was that the M. xanthus LspA enzymes might not function efficiently and/or be expressed efficiently in the heterologous E. coli host. Since we and others have found that both TA and GLM resistance can partly be conferred in E. coli by inactivating the highly abundant Lpp (Braun's) lipoprotein (2, 27), which is present at nearly a million copies per cell (28), we reasoned that removing Lpp from the cells may ease the burden of LspA function in lipoprotein processing and thus allow less efficient LspAMx alleles to possibly complement an E. coli ΔlspA mutant. In this regard, we changed the E. coli parental strain from WT to an lpp mutant strain (E609L) (2, 29), with the idea that the toxic effect of the accumulation of unprocessed/mislocalized Lpp in the cell occurs when LspA function is inhibited or is reduced (2, 27). For clarity, it should also be noted that E. coli contains essential lipoproteins, which are in lower abundance than Lpp, and apparently, by a second mechanism, E. coli can be killed by blocking their proper processing/localization. However, this second mechanism requires more complete LspA inhibition by higher concentrations of TA or GLM (2, 30).
To test the above hypothesis, transformation experiments were repeated in the E609L(pKOBEG) background. The control malF::Tn10-Kan genomic DNA again yielded a high number of transformants (hundreds) for all six strains. As expected, when using the ΔlspA::Kan genomic DNA, hundreds of colonies were obtained for the positive control, whereas the negative control yielded only a few colonies (<20). Importantly, E609L-derived E. coli cells that expressed M. xanthus lspA2 and lspA3 also yielded hundreds of transformants, similar to the positive control, whereas lspA1- and lspA4-expressing strains yielded only tens of transformants. Next, we tested if the transformants exhibited arabinose dependence for growth. A rare transformant from the negative control (pBAD30 vector) failed to show arabinose-dependent growth, thus indicating it was an aberrant recombinant that did not have the chromosomal copy of lspAEc deleted (data not shown). In contrast, the four lspAMx test strains all showed bona fide arabinose dependence for growth, indicating the chromosomal copy of lspAEc had indeed been deleted (Fig. 4). To confirm these results, the arabinose-dependent growth assay was repeated three times and was reproducible (data not shown). We therefore conclude that the four different LspAMx proteins can substitute for LspAEc function when the apparent burden of Lpp processing is removed. Of the four M. xanthus genes, lspA1 exhibited the poorest ability to rescue E. coli growth (Fig. 4) and, as noted above, resulted in few transformants of the chromosomal ΔlspA::Kan allele. Given the low efficiency of plating on an arabinose plate of the selected lspA1 clone (Fig. 4), it is possible that the emerging colonies developed suppressor mutations. Nevertheless, for growth to occur in these lspA1 clones, arabinose was required, indicating LspA1 was functional in E. coli (Fig. 4).
M. xanthus LspA3 confers TA resistance.Previously, we showed that when LspAEc is overexpressed in E. coli, it confers TA and GLM resistance (2). Therefore, we tested whether LspAMx overexpression could also confer antibiotic resistance in E. coli. MIC tests were done in a WT strain (MG1655) and in imp4213 (DW37; bamA4213), a sensitized E. coli strain in which the outer membrane is more permeable (31), and thus, small amounts of TA and GLM can be used for testing (2). For clarity, the aforementioned E. coli strains contain their endogenous lspA genes. We note that after repeated attempts, transformation of pBAD30-lspA1 into the imp4213 strain was, however, unsuccessful, presumably because of an undefined synthetic-lethal interaction with the imp4213 allele. The lspAEc clone and vector only were used as controls. All four LspAMx proteins conferred some degree of resistance to TA and/or GLM in the WT E. coli background (Table 3). Interestingly, lspA3 conferred the highest resistance among all LspAMx proteins. Compared with the vector-only control, lspA3 conferred a 4-fold increase in TAr and a 2-fold increase in GLMr (Table 3). MIC results in the imp4213 background again showed that lspA3 conferred the highest degree of resistance among tested lspAMx clones (Table 3). In contrast, the lspA4 clone did not confer resistance in the imp4213 background and may have slightly increased TA/GLM sensitivity. We also tested these imp4213 strains against a panel of antibiotics (bacitracin, polymyxin B, rifampin, and Cm) that have different MOAs and found no cross-resistance (data not shown), indicating that lspA expression confers resistance that is specific to TA/GLM. These results substantiate the above-mentioned findings that the LspAMx proteins function in E. coli (Table 2).
Effects of lspA genes on antibiotic susceptibility
The M. xanthus lspA genes are genetically redundant.The above-mentioned E. coli complementation studies indicate that all four lspAMx genes encode functional SPaseII enzymes. To determine whether these genes have redundant functions in M. xanthus, we sought to delete each lspA gene from the M. xanthus genome. All markerless in-frame single-deletion mutants were successfully constructed, demonstrating that each lspA gene is dispensable for growth. Next, we attempted to create lspA double-, triple-, and even quadruple-deletion mutants. All double-deletion mutants except ΔlspA1 ΔlspA2 (no positive clones from screening 120 Galr Kms colonies by PCR) were successfully made (Table 4), showing that the resulting strains are viable. In attempts to construct triple deletions and a quadruple deletion, only a ΔlspA1 ΔlspA3 ΔlspA4 mutant (DW1226) was successfully made. From these studies we conclude that either LspA1 or LspA2 function is required for viability, and interestingly, a triple lspA mutant could be made without a significant impact on growth under laboratory conditions (data not shown).
Summary of M. xanthus double-deletion construction
lspA mutants produce different levels of TA.Because the lspA3 and lspA4 genes reside within the ta gene cluster, we tested whether inactivation of any of the lspA genes affected TA production. To examine this, ZOI assays were used with the lspA mutants against an E. coli (MG1655) indicator strain, because we have previously shown that TA is the major diffusible factor produced by strain DK1622 that blocks the growth of E. coli (32). Unlike a Δta1 mutant, all single-deletion mutants, including lspA3, were able to produce an inhibition zone (Fig. 5A), showing that the LspA proteins are not individually required for TA biosynthesis, contrary to an earlier suggestion (33), and that no polar effects blocked TA production. The size of the ZOIs did, however, vary, indicating that each mutant has different capabilities to produce TA. In particular, even though both ΔlspA3 and ΔlspA4 showed the same phase variation phenotype (see below), they yielded opposite phenotypes in the ZOI assay. In the case of ΔlspA3, there was a substantial decrease in the ZOI, whereas the ΔlspA4 strain reproducibly produced a ZOI that was slightly larger than that of the DK1622 control (Fig. 5A). These results suggest that LspA3 and LspA4 might play opposing roles in regulating TA levels.
Phenotypes of lspA mutants in M. xanthus. (A) ZOIs of isogenic DK1622 (TA+), DW1034 (TA− Δta1), DW1224 (ΔlspA1), DW1225 (ΔlspA2), DW1220 (ΔlspA3), and DW1221 (ΔlspA4) on CTT agar. The E. coli indicator strain was MG1655. Note that for the ΔlspA3 and ΔlspA4 mutants, care was taken to select the yellowest colonies for ZOI assays. (B) Fruiting body development after 12 h on TPM starvation agar and motility on CTT 1% agar (HA) and on CTT 0.5% agar (SA). The phase properties of the five strains shown are typical under these assay conditions. The colony swarm pictures were taken at 58 h.
ΔlspA3 and ΔlspA4 mutations cause social behavior defects and result in a tan phase bias.Certain lipoproteins play important roles in M. xanthus social behaviors (34, 35). To determine whether inactivation of SPaseII enzymes affects social behaviors, we tested the following phenotypes: fruiting body formation and motility on soft and hard agar. Single deletions of lspA1 or lspA2 showed no overt defects in these behaviors (Fig. 5B). In contrast, ΔlspA3 and ΔlspA4 revealed different mutant phenotypes compared with the DK1622 control. First, they were delayed in fruiting body formation (Fig. 5B and data not shown). Second, on soft agar, they had a reduced swarm expansion rate, and on hard agar, these mutants showed a distinct smooth colony texture (Fig. 5B).
A noted property of M. xanthus is the ability to vary phases between tan and yellow colonies, which differ in their abilities to produce the yellow pigment DKxanthene, to swarm, and to develop; in colony texture; and in other phenotypes (36–38). Under standard laboratory conditions, DK1622 is generally biased toward the yellow phase; however, importantly, when lspA3 or lspA4 was deleted, these mutants produced predominantly tan colonies (Fig. 5B). To examine this trait in more detail, yellow colonies of DK1622 and each lspA deletion mutant were individually inoculated into liquid cultures. Cells were harvested after overnight growth and concentrated by centrifugation. The WT, ΔlspA1, and ΔlspA2 cultures were clearly yellow, whereas the ΔlspA3 and ΔlspA4 mutants were pale yellow or tan, even though the inoculating colonies were yellow (Fig. 6A). In support of this, Fig. 5B shows that the ΔlspA3 and ΔlspA4 mutant colonies were smooth and tan, whereas DK1622, ΔlspA1, and ΔlspA2 colonies were rough and yellow. Therefore, the described development and motility phenotypes of ΔlspA3 and ΔlspA4 mutants could be an indirect result of a bias toward the tan phase (36, 39, 40).
Phase variation and TA production. (A) Phase colors of concentrated 3 × 109-CFU/ml liquid cultures of DK1622 (WT) and the four isogenic lspA deletion mutants. (B) Comparisons of DK1622 yellow (Y) and tan (T) colonies (top) and their corresponding ZOIs (bottom) on CTT agar. The E. coli indicator strain was MG1655. (C) Exogenous addition of TA causes DK1622 cultures (CTT medium in microtiter wells) to vary phase from yellow (left) to tan (right).
Based on a number of indirect observations, it appeared to us that tan variants produced less TA than did yellow variants. We sought to test this directly. When performing the ZOI assays, we found that when the DK1622 colony that was used for liquid inoculation was older (e.g., 10 to 14 days old), the resulting culture predominantly produced tan colonies when aliquots were transferred to CTT agar plates (Fig. 6B, top), a finding that is consistent with tan variants being more fit or enriched in stationary phase (36). Thus, we directly compared yellow and tan colonies for TA production by overlaying them with E. coli (2). In contrast to a yellow colony, which forms an inhibition zone, a tan colony variant of the same strain did not produce an inhibition halo (Fig. 6B, bottom). This phenotype was reproducible and indicates that TA production was markedly reduced or abolished in tan variants compared with yellow ones.
The above-mentioned result is consistent with prior findings that the ta biosynthetic gene cluster is transcriptionally repressed in tan variants and that expression of DKxanthene (yellow pigment) and that of TA are transcriptionally coupled (36, 41, 42). From these findings, we hypothesized that the tan phase bias of the ΔlspA3 mutant may explain why it produced a smaller ZOI (Fig. 5A). Similarly, the ΔlspA4 mutant produces a large ZOI when a colony is in a yellow phase, but tan-biased colonies formed smaller ZOIs (data not shown). To better test this, we attempted to isolate yellow colonies from the ΔlspA3 and the ΔlspA4 mutants. Occasionally, a relatively yellow colony was identified; however, when it was grown in liquid medium, the resulting culture inevitably turned tan/pale yellow (data not shown). In addition, all four lspA single-deletion mutants were tested for TA susceptibility and were found to have the same MIC value as DK1622 (64 μg/ml), indicating no overt change in TA sensitivity. However, we found that at higher concentrations of TA (e.g., ≥8 μg/ml), the M. xanthus cultures turned tan (Fig. 6C), suggesting that exogenous addition of TA triggers the cells to vary the phase. Phenotypic tests were also initiated with the double and triple lspA mutants; however, they were not pursued because the phase variation phenotype of the ΔlspA3 and ΔlspA4 mutants was not stable.
In WT M. xanthus isolates, lspA3, but not lspA4, correlates with the presence of the ta gene cluster.Because the M. fulvus HW-1 genome contains all four lspA genes but lacks a ta gene cluster, we sought to determine whether environmental M. xanthus isolates could also contain this genomic arrangement. First, we screened by PCR with diagnostic primer sets four M. xanthus isolates reported not to make TA (43). As suggested from the prior results, these isolates did not contain the ta biosynthetic genes (Fig. 7). In contrast to M. fulvus HW-1, these strains also did not contain lspA3; however, they did contain orthologs of lspA1, lspA2, and lspA4 (Fig. 7). We also screened four environmental isolates that produce TA (43), and they all showed a genomic arrangement similar to that of DK1622 with respect to the four lspA orthologs and the ta gene cluster (Fig. 7). This data set suggests that although lspA3 could be a characteristic feature of M. fulvus species that do not produce TA, M. xanthus isolates that lack the ta gene cluster also lack lspA3.
Presence of lspA and ta genes in environmental isolates.
DISCUSSION
The M. xanthus DK1622 genome includes four lspA genes. Based on sequence comparisons, they all appeared to be functional because they contain the conserved SPaseII residues, including the putative catalytic aspartate residues (Fig. 2B) (18). We showed that the four LspAMx proteins could substitute for the essential SPaseII function in E. coli when the abundant Lpp lipoprotein was removed (Table 2). These results suggest that the M. xanthus clones function less efficiently than lspAEc in E. coli. Plausible explanations for why lspAMx alleles do not complement well in E. coli include poor translation efficiencies because of differences in DNA GC contents, differences in substrate specificities, and/or innate enzymatic activities.
Our MIC results for heterologous LspAMx expression in E. coli showed that the lspA3 allele conferred the highest level of TAr (Table 3). The mechanism of resistance might be due to reduced affinities of TA and GLM for binding to LspA3. In this regard, LspA3 has a notable change of a conserved Gly at position 58 to Ala (Fig. 2B). In a regulatory context, the lspA3 locus is located near the middle of the ta gene cluster, and thus, when the expression of the TA biosynthetic pathway is induced, such as by yellow phase variants or by development (36, 41, 42), LspA3 will also be expressed. Based on these findings, it is plausible that the function of LspA3 is to help confer producer strain resistance, perhaps mediated by reduced affinity for TA, while still providing necessary SPaseII function. In support of this, when lspA3 was deleted, the resulting M. xanthus mutant phase varied to tan, which we postulate is a stress response (36, 44), because the cells are producing TA in the absence of LspA3. Consistent with this, exogenous addition of TA also causes M. xanthus to vary the phase to tan (Fig. 6C). In turn, tan cells may be more resistant to TA and other stresses than yellow cells.
SPaseII function is inhibited by TA (2). Therefore, TA is likely to bind to LspA, and as such, these proteins have the capacity to sense cellular levels of TA. Our finding that the ΔlspA4 mutant overproduces TA, even though the cells are phase biased toward tan, suggests that one function of LspA4 could be sensing intracellular TA levels. In turn, a feedback pathway could regulate TA biosynthesis. In contrast to this activity, we suggest the primary function of LspA1 and LspA2 is in “housekeeping” lipoprotein processing, as orthologs of these proteins are commonly found in other myxobacterial genomes. Because M. xanthus is predicted to express ∼400 different lipoproteins (19, 34), the cells might also need multiple SPaseIIs to efficiently process a large and diverse set of substrates where the recognition motifs for cleavage would vary (45).
The inclusion of the target of a secondary metabolite antibiotic within the biosynthetic gene cluster is striking (Fig. 1). We suggest that inspecting other or cryptic polyketide synthetase and nonribosomal peptide synthetase gene clusters for genes that are generally known to be essential for cell function could be used as a bioinformatic approach to identify targets of the resulting natural product. Given the large number of sequenced bacterial genomes, other examples likely exist.
Interestingly, all lspA single-deletion mutants were viable. These findings, in conjunction with the E. coli complementation results, suggest that the lspA genes have redundant functions in M. xanthus, although either lspA1 or lspA2 must be retained for viability. In this work, we provide evidence that the reasons for redundancy may be related to expanded functions of LspA3 and LspA4 in TA resistance and regulation, and the genes themselves can be differentially regulated (36, 41, 42). Other possible reasons for redundancy include alternative routes of lipoprotein secretion (11), modification, and perhaps sorting in the cell. Indeed, it has been shown in Gram-positive bacteria in which a buildup of lipoprotein precursors has occurred in the membranes because of lgt and lspA mutations that alternative processing by other peptidases, such as Eep and type I signal peptidases, can occur (13, 46, 47). Moreover, in M. xanthus, it has been suggested that distinctive amino acid signals for sorting lipoproteins to the extracellular matrix could be recognized by alternative sorting pathways other than the Lol pathway (34, 48). Future studies are needed to understand the precise roles that the four SPaseII enzymes play in lipoprotein processing and in the complex biology of M. xanthus.
ACKNOWLEDGMENTS
We are grateful to Masatoshi Inukai for providing GLM and to Klaus Gerth and Rolf Müller for providing TA. We thank Mark Gomelsky for helpful discussion and for providing a strain.
This work was supported by grants AES WYO-474-12 and NCRR and the Wyoming NIH INBRE 2P20RR016474.
FOOTNOTES
- Received 16 November 2013.
- Accepted 26 December 2013.
- Accepted manuscript posted online 3 January 2014.
- Address correspondence to Daniel Wall, dwall2{at}uwyo.edu.
↵* Present address: Yao Xiao, Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, Michigan, USA.
Supplemental material for this article may be found at http://dx.doi.org/10.1128/JB.01361-13.
REFERENCES
- Copyright © 2014, American Society for Microbiology. All Rights Reserved.
