Exopolysaccharide biosynthesis genes required for social motility in Myxococcus xanthus

Identification and phenotypic characterization of genetic loci involved in EPS biogenesis

Random mutagenesis with the transposon Tn5 has been commonly used in M. xanthus to generate polar insertion mutants (Kuner and Kaiser, 1981; Torti and Zusman, 1981; Zusman, 1982; Chang and Dworkin, 1996; Rodriguez-Soto and Kaiser, 1997; Bowden and Kaplan, 1998; Yang et al., 1998b; Sun and Shi, 2001; Cusick et al., 2002). As Tn5 transposons have ‘hot spots’ and ‘cold spots’ for insertion, we screened a library previously created using another method for random mutagenesis developed based on recombination and integration of gene fragments cloned in a plasmid library (Cho and Zusman, 1999). Specifically, sheared genomic DNA fragments of ≈ 500 base pairs in length were cloned into a plasmid that encodes kanamycin resistance, as described (Cho and Zusman, 1999). The plasmids, which cannot replicate in M. xanthus, were then electroporated into wild-type cells. The Kan^R electroporants that resulted from homologous recombination between the fragments in the plasmids and chromosomal DNA were then screened for potential indicators of mutations in EPS biogenesis, specifically, defects in S-motility and fruiting body formation. Among the 5000 colonies screened, 68 showed phenotypes suggesting altered motility and fruiting body formation. These 68 mutants were further tested for phenotypes indicative of the loss of EPS, such as defects in cellular agglutination and CFW binding while retaining functioning A-motility and normal LPS. The sites of plasmid integration were then mapped and four previously unknown loci were identified for further study.

The four strains containing insertions, SW801–SW804 (Tables 1 and 2; Fig. 1), exhibited the defects expected of EPS biogenesis mutants. They were unable to form fruiting bodies, bind CFW or spread on soft agar (Table 3). These insertions mapped to four genes in two regions, eps and eas, which had not been previously characterized (Fig. 1; Table 2).

Table 3. Summary of insertion mutations in eps region.

	Agglutination	CFW binding	S-motility	Fruiting body formation
IGS	+	+	+	+
epsA	–	–	–	–
epsE	+/–	+	–	–
epsF	+	+	+/–	+
epsH	–	–	–	–
epsI	+/–	–	–	–
epsK	+	+	+	–
epsN	+	+	+	+
epsO	+	+	+	–
epsU	–	–	–	–
epsV	–	–	–	–
epsY	–	–	–	–
epsZ	–	–	–	–

None of these four mutations was located in any of the known pil genes, which encode proteins required for TFP biosynthesis, assembly and function. Furthermore, Western immunoblot analysis, using anti-PilA serum, confirmed that PilA was present in whole-cell lysates of the mutants and on their cell surfaces (data not shown). The phenotypes of the mutants were similar to the dif mutants, known to lack EPS (Bellenger et al., 2002; Li et al., 2003; Yang et al., 1998a): mutant strains SW801–SW804 spread poorly in soft agar (Table 3) and show elevated levels of sheared surface pilin (data not shown). It should be noted that the SW801–SW804 strains exhibited normal A-motility, as observed by examining the movement of individual cells near the colony edges (data not shown). To confirm these motility phenotypes, the mutations were transduced into strains containing only S-motility (A^–S⁺), and resulted in non-motile transductants. These mutations were also transduced into A-motility (A⁺S^–) strains resulting in cells with a phenotype similar to the parent strains. The results, listed in Table 4, indicate that these eps and eas mutations localize to genes specifically required for S-motility, but not A-motility.

Mutations in LPS O-antigen were previously shown to result in defective S-motility and fruiting body formation (Bowden and Kaplan, 1998), but EPS from these cells was normal. To rule out the possibility of an LPS mutation, the LPS of mutant strains SW801–SW804 was therefore analysed by silver staining of a deoxycholate (DOC)-gel that separated the LPS prepared from the mutant strains using the hot phenol–water method (Apicella et al., 1994). In all four mutants, the LPS appeared indistinguishable from the wild-type strain DK1622 and the dif mutant SW504 (data not shown). Thus, the eps and eas mutations appear to be unrelated to LPS biogenesis.

Sequence analysis of mutations

The sites of the SW801–SW804 mutations were determined by sequencing. The sequences were analysed using blast (Altschul et al., 1990; 1997; Gish and States, 1993) and the nearly completed TIGR genome sequence of M. xanthus (http://www.tigr.org/tdb/mdb/mdbinprogress.html). The possible open reading frames (ORFs) were then aligned against the NCBI database, yielding the following results: strains SW801 (epsZ) and SW802 (epsY) have insertions in two different genes at the distal end of the eps locus (Fig. 1; Table 2). EpsZ is a homologue of a Clostridium perfringens capsular polysaccharide biosynthesis protein with glycosyltransferase activity. There are two overlapping ORFs in the location of the epsY gene, both of which are disrupted by the insertion. One ORF is 37% identical and 51% similar to a Geobacter metallireducens periplasmic protein involved in polysaccharide export. The other ORF is 26% identical and 47% similar to a Chloroflexus aurantiacus heteropolysaccharide repeat unit export protein. The insertion in strain SW803 is located in epsA, at the proximal end of the eps locus. EpsA is 39% identical and 59% similar to EpsP from Methylobacillus sp. 12S containing a conserved domain with UDP-N-acetyl-mannosaminuronic acid transferase/UDP-N-acetyl- d-mannosamine transferase activity. According to their presumed functions, the genes found in the eps region are likely to be involved in EPS biosynthesis, assembly or transport.

The insertion in strain SW804 contains partial sequences of two ORFs, easA and easB. EasA is similar to conserved hypothetical proteins and EasB has no significant similarity to anything in the databases. The conserved hypothetical proteins similar to EasA contain a conserved domain of unknown function (DUF574) and this domain is found in Chloroflexus aurantiacus, Streptomyces coelicolor, Thermobifida fusca, Streptomyces collinus and Streptomyces avermitilis MA-4680. Sequence analysis of the eas region shows that the insertion could have been incorporated into either one of two hypothetical proteins. It is likely that this insertion had a polar effect on the second gene of the operon, effectively inactivating it. This analysis did not yield any information regarding the potential function of these genes in EPS biogenesis.

Genomic analysis of the eps region

The eps genes disrupted in strains SW801–SW803 were further analysed with respect to their relative positions within the M. xanthus genome and were found to localize to either end of a 37 kb region containing 26 putative ORFs, which we call the eps region (Fig. 1). This region contains genes similar to those involved in EPS biogenesis in other bacteria, and many genes of unknown function (Table 2). Including the genes disrupted in strains SW801–SW803, the eps region appears to encode at least six glycosyltransferase homologues, four homologues to regulatory proteins, one polysaccharide transporter homologue, one (endo)glucanase homologue and two ORFs that encode homologues to other proteins known to be involved in EPS biogenesis.

The original insertions were found in the following ORFs: MXORF06813 (epsZ), MXORF06810/12 (epsY ) and MXORFA00949 (epsA). epsZ appears to be a single gene operon with a strong hairpin loop terminator sequence between it and epsY. epsY, on the other hand, appears to be the first gene in five-gene operon suggesting that the downstream operon is involved in EPS biogenesis. epsZ and epsY are located at the distal end of this 37 kb region whereas epsA appears to be the last gene of a four-gene operon at the proximal end of this locus (Fig. 1). Analysis of the regions downstream of epsA and upstream of epsZ shows no ORFs. These regions appear to be large intergenic spaces with the nearby genes having no suspected EPS-related function. An insertion in the intergenic region downstream of epsA appeared to be a normal, EPS⁺ strain (Table 3). Insertional mutations introduced throughout the region into the first ORF of each operon supports the hypothesis that the whole region is involved in the biosynthesis of fibrils and fibril EPS (Table 3).

Sequence analysis of the operon containing epsA revealed three upstream genes corresponding to homologues of a glycosyltransferase (ORF06778, epsD), a serine acetyltransferase (ORF06776, epsC) and an endoglucanase (ORF06774, epsB) (Fig. 1). As insertion mutations have the possibility of introducing polar effects, this operon was analysed further using in-frame deletion analysis to determine the role of each individual ORF in EPS biogenesis, as described below.

Genetic analysis of the epsD-A operon

After reverse transcription polymerase chain reaction (RT-PCR) analysis confirmed that epsD-A constitutes a single transcriptional unit (data not shown), we generated in-frame mutations in each of the genes in the epsD-A operon to determine their effect on the production of M. xanthus EPS. Mutants SW810 (ΔepsA) and SW813 (ΔepsD), which carry deletions in genes that encode glycosyltransferase homologues, were found to have Eps^– phenotypes, including deficiencies in CFW binding, agglutination, S-motility and fruiting body formation 2-4). They also react poorly with Mab2105, a monoclonal antibody generated against FibA protease, a major protein component of the fibril material (Fig. 6). SW812 (ΔepsC), which has an in-frame deletion in the gene encoding a serine acetyltransferase homologue, exhibits a less severe EPS^– phenotype. This mutant binds CFW (Fig. 3D), but has a delay in developmental aggregation (data not shown) and is unable to form mature wild-type fruiting bodies (Fig. 2D). SW811 (ΔepsB), which contains an in-frame deletion in a gene encoding an endoglucanase homologue, did not show a clear EPS^– phenotype, but still presented an overall colony morphology different from wild type 2-45, 6). A modified trypan blue assay was used to determine relative amounts of EPS in each of these mutant strains (Black and Yang, 2004). The results show a significant different in the binding of trypan blue to DK1622 cells versus the mutants (Fig. 7). Notably, SW812 (ΔepsB) shows less than 40% of the binding seen in wild-type cells, even though the other assays show no difference between the two strains. The gene identified in strain SW811 was independently discovered by the Hartzell/Youdarian labs (NCBI Accession No. AAF00919) to have a possible role in S-motility; however, they performed no further analysis of the region. In summary, this operon appears to function in M. xanthus EPS biogenesis and mutations in these genes have noticeable effects on the bacterium's ability to produce wild-type EPS or perform EPS-related functions, including fruiting body formation and S-motility.

Transcriptional regulation of epsD-A during development

It has been demonstrated by several laboratories that EPS production in M. xanthus increases under high cell density and during starvation-induced development (Bacon et al., 1975; Sutherland and Thomson, 1975; Kim et al., 1999). The discovery of EPS biogenesis genes permitted us to test whether these conditions would increase transcription of the EPS biosynthesis genes. We constructed an epsD-A–lacZ transcriptional fusion strain (SW830) and assessed the expression patterns of epsD-A under conditions known to effect EPS production. As shown in Fig. 8, the β-galactosidase activities of the epsD-A–lacZ fusion strain did not show statistically significant increases under these conditions. In a similar manner, the β-galactosidase activities of the epsD-A–lacZ transcriptional fusion were indistinguishable between cells in high density (5 × 10⁹ cells per ml) and low density (1 × 10⁸ cells per ml) (data not shown). These data suggest that this operon is not regulated at the level of transcription under the conditions tested.

Do the dif genes control eps expression?

Strains with mutations in the dif genes, which are defective in a set of chemotaxis-like proteins (Yang et al., 1998a), display the Eps^– phenotype (Yang et al., 2000; Bellenger et al., 2002). As chemotaxis-like pathways generally perform regulatory functions and the dif operon does not encode enzymes likely to be involved in EPS biogenesis, we wondered whether the dif genes play a direct or indirect role in the regulation of the eps genes. We used real-time RT-PCR to examine such a possibility. As shown in Fig. 9, expression levels of various eps genes are similar in dif mutant strains and wild-type DK1622, indicating that the transcription of these eps genes is not dependent on dif.

Bacterial strains, plasmids, phages and media

Bacterial strains used in this study are listed in Table 1. Escherichia coli TOP10 (Invitrogen) was used for in vitro DNA cloning. M. xanthus strains were grown and maintained at 32°C in CYE (Casitone-yeast extract medium) culture with appropriate antibiotics. Liquid cultures were grown to OD₆₀₀ = 0.6–1 (3–5 × 10⁸ cells per ml). MCM buffer [10 mM MOPS (pH 6.8), 10 mM MgCl₂, 1 mM CaCl₂] and MOPS (10 mM, pH 7.0) medium were used for fruiting body formation assays. Mutant strains were confirmed using Southern blots. E. coli cells were grown in Luria–Bertani (LB) with appropriate antibiotics at 37°C (Sambrook et al., 1989).

Generation of random insertional mutants

Mutants were generated as described previously (Cho and Zusman, 1999). Briefly, random pieces of genomic DNA from wild-type M. xanthus DZ2, ≈500 base pairs each, were cloned into plasmid pZErO-2 (Invitrogen), which carries a kanamycin resistance gene (Cho and Zusman, 1999). The resulting plasmid library was then electroporated into M. xanthus DZ2 (Kashefi and Hartzell, 1995), selected for Kan^R on CYE plates containing 100 µg ml⁻¹ kanamycin and further screened for the phenotype of interest.

Generation of insertional mutants

Insertional mutants were created by electroporating vector pCR2.1 (Invitrogen) carrying internal fragments of the genes of interest into DK1622 (Table 5).

Generation of in-frame deletion mutants

Genomic DNA was used as template for PCR amplification of fragments used in cloning for all mutants. The vector pBJ113 was used for cloning in-frame constructs.

Trypan blue binding assay

Cells (100 µl), resuspended in MCM to 5 × 10⁸ cells per ml, were added to 900 µl of 0.005% trypan blue in MCM. Sample were vortexed, then allowed to incubate undisturbed at room temperature for 30 min. The samples were then spun down and the OD₅₉₅ of each supernatant was recorded. Assays were performed in triplicate (Black and Yang, 2004).

Generation of a lacZ fusion mutant in the epsD-A operon

The following primers and constructs were used for creating lacZ fusion to the epsD-A operon: an insert that includes the region 600 bp upstream of the ATG (putative start codon) of epsD, as well as the codons for the first 17 amino acids of epsD, was generated using the oligonucleotides 5′-GGTACC ATCTTCTGCACACTGGATGG-3′ and 5′-GGATCCTTGTA CGTGGCGATGACG-3′ and cloned into pKY481, a plasmid that carries a promoterless lacZ.

Swarm plates

Myxococcus xanthus cells (2 µl) concentrated to OD₆₀₀ = 10 (≈ 5 × 10⁹ cells per ml) were spotted onto the centre of a CYE plate containing 0.3% agar (Shi and Zusman, 1993) and incubated for 3 days at 32°C. S-motility was analysed by observing the colonies for expansion. Colony morphology was documented using the GelDoc system (Bio-Rad).

Calcofluor white binding assay on plates

Myxococcus xanthus cells (10 µl) concentrated to OD₆₀₀ = 10 were spotted onto CYE plates containing 1.5% agar and 5 µg ml⁻¹ CFW. The plates were incubated for 6 days at 32°C. CFW binding was documented under exposure to long-wavelength UV light using a hand-held digital camera (Canon Powershot S200).

Generalized transduction

Generalized transduction was performed as previously described using phage Mx4 (Campos et al., 1978; Geisselsoder et al., 1978).

Western blots

Whole-cell lysates were made by pelleting cells and resuspending cells to 5 × 10⁸ cells per ml in protein lysis buffer with SDS. These lysates were electrophoresed through SDS-PAGE gels and transferred to nitrocellulose membrane. MAb2105 (Behmlander and Dworkin, 1991) against FibA, an EPS-associated protein, and rabbit polyclonal anti-PilA antibody (Wu and Kaiser, 1997) against pilin subunits were used as primary antibodies. Cell surface pilin was isolated and assayed as described previously (Wu and Kaiser, 1997).

Agglutination assay

The cohesion of M. xanthus cells was measured with an agglutination assay described by Shimkets (1986a,b).

Fruiting body formation on MOPS plates

Cells (5 µl) harvested by centrifugation (3000 g for 6 min at 4°C) and resuspended in MOPS buffer to 5 × 10⁹ cells per ml were spotted on MOPS/1.5% agar. Fruiting body development was documented after 48 h.

LPS analysis

LPS was isolated from 10 ml of logarithmically growing liquid cultures of M. xanthus cells using the modified hot phenol-water method (Apicella et al., 1994). Approximately 1–5 µg of each preparation was separated by electrophoresis through a DOC-15% polyacrylamide gel (Laemmli, 1970), and visualized by silver staining (Tsai and Frasch, 1982).

β-Galactosidase assay

Samples were collected at different time points during fruiting body formation in submerged culture. Specifically, the cells, incubated at 2.5 × 10⁸ cells per well in 24-well culture plates with 400 µl of MCM buffer, were harvested and 40 µl of 500 mM phosphate buffer (pH 7.2) was added to bring the final concentration to 50 mM phosphate buffer. The cells were frozen at −20°C for at least 12 h. Cells were thawed, 500 µl of each of the cultures was combined with 10 µl of toluene, vortexed for 30 s and incubated at 37°C for 5 min. β-ONPG (20 mM; 50 µl) was added and samples were incubated until yellow colour develops, at which time 700 µl of NaCO₃ was added to stop the reaction. The samples were spun down to remove the debris and they were read on the spectrophotometer at a wavelength of 450 nm. β-Galactosidase assays were presented in specific activity per mg of protein.

Real-time RT-PCR

RNA was isolated from vegetative cells using Trizol (Invitrogen) extraction. cDNA was made using Stratascript reverse transcriptase (Stratagene) and primed with random hexamers. The real-time PCR was performed with the iCycler from Bio-Rad using the primer pairs listed in Table 5. SYBR Green I (Molecular Probes) was used for detection of the amplified products. For each gene being measured, a sequence-specific standard curve was generated and expression of each gene in the dif genetic background was compared with wild-type expression.