A gene encoding a homologue of the Escherichia coli GidA protein (glucose-inhibited division protein A) lies immediately upstream of aglU, a gene encoding a WD-repeat protein required for motility and development in Myxococcus xanthus. The GidA protein of M. xanthus shares about 48% identity overall with the small (≈ 450 amino acid) form of GidA from eubacteria and about 24% identity overall with the large (≈ 620 amino acid) form of GidA from eubacteria and eukaryotes. Each of these proteins has a conserved dinucleotide-binding motif at the N-terminus. To determine if GidA binds dinucleotide, the M. xanthus gene was expressed with a His6 tag in E. coli cells. Purified rGidA is a yellow protein that absorbs maximally at 374 and 450 nm, consistent with FAD or FMN. Thin-layer chromatography (TLC) showed that rGidA contains an FAD cofactor. Fractionation and immunocytochemical localization show that full length GidA protein is present in the cytoplasm and transported to the periplasm of vegetative-grown M. xanthus cells. In cells that have been starved for nutrients, GidA is found in the cytoplasm. Although GidA lacks an obvious signal sequence, it contains a twin arginine transport (Tat) motif, which is conserved among proteins that bind cofactors in the cytoplasm and are transported to the periplasm as folded proteins. To determine if GidA, like AglU, is involved in motility and development, the gidA gene was disrupted. The gidA– mutant has wild-type gliding motility and initially is able to form fruiting bodies like the wild type when starved for nutrients. However, after several generations, a stable derivative arises, gidA*, which is indistinguishable from the gidA– parent on vegetative medium, but is no longer able to form fruiting bodies. The gidA* mutant releases a heat-stable, protease-resistant, small molecular weight molecule that acts in trans to inhibit aggregation and gene expression of wild-type cells during development.
Myxococcus xanthus is a Gram-negative, rod-shaped soil bacterium that can undergo a complex life cycle, ultimately producing heat- and desiccation-resistant spores, when starved for nutrients. During development, high densities of cells (> 100 000) co-operate to produce a simple mound structure in which cells differentiate into spores. To engage this developmental programme, cells must determine that there are insufficient nutrients for growth and share this information with neighbouring cells. Although cells undergo many changes at the molecular level during the first 15 h of development, microscopic changes in cell morphology are not obvious at this stage. The morphogenesis of rod-shaped cells into spherical cells within the mound occurs after about 20 h. Although spherical cells initially are sensitive to heat and sonication, they become increasingly resistant after the initial 24 h starvation period.
Early events in the development of myxospores require the intracellular production of (p)ppGpp (Manoil and Kaiser, 1980; Singer and Kaiser, 1995) and production of extracellular signal molecules (Hagen et al., 1978). A M. xanthus relA– mutant that fails to produce (p)ppGpp is unable to aggregate or express early developmental genes (Harris et al., 1998). Mutations that affect the production or transmission of the extracellular signals, Asg, Csg, and Esg, abolish spore formation and fruiting body morphogenesis. Although these mutants are unable to develop, their defects can be rescued if mixed with a strain that can provide the missing signal molecule. Hence, cells that are defective in signal production are still able to respond to the signal.
Cell alignment and aggregation are critical for cell–cell communication events required for sporogenesis and fruiting body development. Aggregation and alignment depend on gliding motility, a type of surface translocation that, in M. xanthus, is enabled by two distinct systems: social (S-gliding) and adventurous (A-gliding) (Hodgkin and Kaiser, 1979). A strain carrying a mutation in both an A gene and an S gene is non-motile and unable to develop. In contrast, single mutations in A-gliding or S-gliding genes may affect or delay development, but do not completely block development.
A-gliding genes control the ability of an individual cell to glide over a solid surface. Among these are homologues of the TolB, TolR, TolQ and TolA proteins (White and Hartzell, 2000; P. Youderian and P. Hartzell, unpublished results). In addition to their role in gliding, the Tol homologues also are required for a terminal stage of spore differentiation (White and Hartzell, 2000). By analogy with E. coli, the Tol proteins are predicted to form a transport complex spanning the inner and outer membranes of M. xanthus. The aglU (adventurous gliding unknown) gene encodes a homologue of TolB and is predicted to be an outer membrane lipoprotein. aglU is transcribed as the second gene in an operon with gidA, a homologue of GidA (glucose-inhibited division protein) from E. coli (von Meyenburg and Hansen, 1980; von Meyenburg et al., 1982). Although GidA is conserved among a wide range of prokaryotes, its role remains unclear. Prokaryotes including E. coli, Pseudomonas putida and Thermus thermophilus have a version of gidA that is located near the origin of replication (Ogasawara and Yoshikawa, 1992; Nardmann and Messer, 2000). The 70 kDa protein encoded by the E. coli gidA gene is thought to play a role in cell replication and division because gidA mutants form long filamentous cells when grown in the presence of glucose (von Meyenburg and Hansen, 1980). Transcription of the gidA gene near the origin of replication has a stimulatory affect on initiation of replication (Ogawa and Okazaki, 1991). A subset of organisms have a second gidA gene, which encodes a smaller GidA protein of unknown function.
Here we describe the characterization of the small GidA protein in M. xanthus and show that GidA is a periplasmic protein that binds FAD. The start codon for aglU overlaps the stop codon for the upstream gidA gene, suggesting that the products of these genes function in a common pathway. However, unlike aglU, which is required for A-gliding motility, a mutation in the gidA gene does not affect motility. Unlike aglU, which is required for differentiation of spores during development, the gidA mutant initially does not exhibit a developmental phenotype. However, after several passages a stable derivative, gidA*, appears which can no longer develop. The gidA* mutant produces a compound that inhibits the development of the wild-type strain. The phenotype of a gidA mutant can be suppressed by a second mutation in the aglU gene, suggesting that the inhibitor may be transported through the AglU membrane complex.
Identification of a GidA homologue in M. xanthus
In a previous report, we described the characterization of the aglU gene in M. xanthus, which is essential for adventurous gliding motility. The aglU mutation abolishes the ability of cells to glide away from the colony edge as isolated cells during growth on rich medium. When starved for nutrients, the aglU mutant forms fruiting bodies similar to those of the wild-type strain, but unlike the wild type, which produces refractile, heat-resistant spores, differentiation of cells into heat-resistant spores is blocked in the mutant. Sequence analysis shows that the aglU gene encodes a protein related to the WD-repeat proteins, which form β-propeller platforms that facilitate protein–protein interactions.
Within the 5 kb fragment that complements the aglU mutant, a second gene, gidA, was identified immediately upstream of aglU. The translation stop codon of gidA overlaps the translation start of aglU, and both genes are expressed from a promoter upstream of gidA (White and Hartzell, 2000). Translation of gidA is predicted to begin with the ATG at bp 374 and terminate with TGA at bp 1747 (GenBank accession number # AF162663).
This ATG is preceded by a purine-rich region, GAGGAA, that is a good match for the 3′-end of the 16S RNA (Weisburg et al., 1985) and probably serves as a ribosome binding site. Beginning with this ATG is an open reading frame (ORF) of 457 codons with 89% G + C in the third codon position, which is characteristic of ORFs from organisms with a high G + C content, such as M. xanthus (Bibb et al., 1984).
The M. xanthus gidA gene is predicted to encode a 50 kDa protein that shares identity with a large family of GidA (glucose inhibited division protein) proteins over its entire length. GidA is most closely related to a group of small GidA proteins that range in size from 48 kDa in Bacillus subtilis to 53 kDa in Deinococcus radiodurans. GidA has 48% identity with the B. subtilis Gid protein. The small forms of GidA share about 25% identity with the N-terminal 450 amino acids of the large GidA proteins in eubacteria, which range in size from 59 kDa in Lactococcus lactis to 69 kDa in P. putida. Yeast encode a homologue of GidA that is about 74 kDa, and a gene predicted to encode a 50-kDa GidA homologue is found on the mitochondrial genome of Pinus contorta.
GidA is a flavoprotein
Alignment of large and small GidA proteins in the database reveals a highly conserved dinucleotide-binding motif near the N-terminus (Fig. 1A). To determine if GidA binds dinucleotide, the M. xanthus gidA gene was cloned into the pET24 expression vector to produce pDW432, which was introduced into E. coli strain BL21. The gidA gene in pDW432 has a C-terminal fusion with a His6 tag and is predicted to produce a 51 kDa protein upon induction with IPTG. A 63 kDa protein was detected in the strain that carried pDW432 that was not present in the pET24 control. The 63 kDa recombinant GidA protein (rGidA) was purified to apparent homogeneity from the soluble fraction by affinity chromatography (Fig. 1B). The purified rGidA is yellow in colour, which is characteristic of flavoproteins. UV-visible spectral analysis of the purified protein shows that GidA absorbs at 370 and 450 nm, with a shoulder at 480 nm (Fig. 1C), suggesting that the protein binds FAD or FMN.
To determine whether the rGidA protein contains FAD or FMN, the cofactor was removed from the protein by TCA precipitation and compared with FAD and FMN standards by thin-layer chromatography (TLC). Three different solvent systems were used to compare the Rf values of the cofactor extracted from rGidA with commercially available FAD and FMN controls. The Rf value for the rGidA cofactor is very similar to that of FAD in all three solvent systems tested (FAD = 0.27; rGidA cofactor = 0.27, respectively, in Na2HPO4; FAD = 0.14, rGidA cofactor = 0.18 in 12:3:5 n-butanol:acetic acid:H2O; and FAD = 0.06; rGidA = 0.06 in 4:1:5 n-butanol:acetic acid:H2O), confirming that rGidA contains a bound FAD moiety (Table 1). These results also show that, although the cofactor is tightly associated with the enzyme, it is not covalently attached. To determine the number of FAD molecules in GidA, the concentration of FAD bound to a known amount of protein was estimated using the extinction coefficient ε450 (M−1 cm−1) = 11 300. It was found that 0.8 mol of FAD was present per mol of rGidA protein purified from E. coli. Because FAD often is limiting relative to production of apoprotein in an overexpression system, we predict that the stoichiometry is 1FAD:1GidA under standard conditions.
Table 1. Rf values of flavins by ascending paper chromatography.
Samples were separated on Whatman no. 1 paper as described in the text. The rGidA cofactor was released from purified protein by TCA precipitation. Apoportein was removed by centrifugation. Solvent system A, 5% disodium hydrogen phosphate in H2O; solvent system B, n-butanol-acetic acid-H2O (12:3:5); solvent system C, n-butanol-acetic acid- H2O (4:1:5).
GidA is equally distributed between the cytoplasm and periplasm of M. xanthus cells during growth
To determine where the GidA protein is located in M. xanthus cells, antibody against rGidA was generated in a chicken. IgY, purified from the yolk, cross-reacted with the 63 kDa rGidA-His6 protein from E. coli and the slightly smaller, native GidA protein in whole cell lysates from M. xanthus. Commercially available antibody against the His6 tag reacted with the 63 kDa protein, confirming that this protein is the product of the gidA gene (Fig. 2A, lane 9). The anti-GidA antibody also reacts slightly with a 75 kDa protein present in the wild-type strain and in the MxH1171 mutant. This corresponds in size with a 615-amino-acid protein that is encoded by a second gidA gene on the M. xanthus chromosome. The 457-amino-acid GidA shares 24% identity with the 615-amino-acid GidA protein, which also has an FAD-binding motif at the N-terminus (Fig. 1).
In contrast with the AglU protein, which is predicted to be an outer membrane lipoprotein, the GidA protein does not have any regions that score well as membrane-spanning segments, nor does it have an N-terminal sequence that resembles a signal peptide. To determine the cellular location of GidA, cell fractions were recovered and probed with anti-GidA antibody. Myxococcus xanthus cells, which had been equilibrated in sucrose and EDTA, were transferred to MgSO4 to release the contents of the periplasm (Thomas et al., 2001). The spheroplasts were isolated by centrifugation and the membrane fraction was separated from the cytoplasmic contents by ultracentrifugation. Cell fractions were separated by SDS–PAGE, transferred to membranes, and probed with antibody against GidA and FrzCD. GidA protein was equally distributed between the cytoplasmic fraction and the periplasmic fraction of vegetatively grown wild-type cells and in the MxH1777 (ΔaglU) mutant (Fig. 2A). The 63 kDa GidA was not detected in the MxH1171 (gidA–) mutant. The antibody reacted with a protein of Mr 75 000, which we predict is the form of GidA made by the second gidA gene in M. xanthus. The 75 kDa protein was found only in the periplasm of the wild-type strain and the MxH1171 mutant. Surprisingly, none of the 75 kDa protein was detected in strain MxH1777, which lacks the membrane lipoprotein AglU, which is transcribed with the gene encoding the small GidA protein. Antibody against FrzCD protein was used to show the integrity of the periplasmic fraction. FrzCD was not detected in the periplasmic fraction, but was found in the cytoplasmic and membrane fractions, which is consistent with the published data for FrzCD (McBride et al., 1992).
The cellular location of GidA during development was determined by immunoblot analysis of cell fractions obtained from cells that had been allowed to starve for 2 h. In contrast with the vegetatively grown wild-type cells, which contain periplasmic and cytoplasmic GidA, starved cells contain only cytoplasmic GidA. The 75 kDa protein (the second GidA) was present in the periplasmic fraction of the wild-type strain and the MxH1171 mutant (Fig. 2A).
The location of GidA was confirmed by immunocytochemical analysis of thin sections. As shown in Fig. 2B, sections of the wild-type and the gidA mutant were probed with anti-GidA antibody, then stained with secondary antibody conjugated with 18 nm gold particles. In the wild-type strain, 57% (309 out of 544 particles examined in seven micrographs, containing a total of about 50 cells) of the gold particles were found in a region corresponding to the periplasm at the periphery of the cell, 30% were associated with the cellular debris that was outside the cell (periplasmic content or secreted material), 8% of the label was associated with the cytoplasm and 5% did not appear to be bound to cell material. In contrast, very few gold particles were detected in thin sections of the gidA mutant (MxH1171) that were prepared in parallel with the wild-type samples. In a total of 52 gold particles, 58% were periplasmic, 8% were cytoplasmic and 35% were unassociated with cellular material. When thin sections were stained with secondary antibody alone, fewer that 1% of gold particles were detected in fields similar to the samples treated with both primary and secondary antibodies.
For the GidA protein to be delivered to the periplasm, it must be translocated through the inner membrane. Because GidA protein does not contain a signal peptide at the N-terminus that is characteristic in Sec-dependent pathways, we looked for other motifs that might suggest a mechanism for translocation. A subset of periplasmic proteins that bind cofactors, such as FAD, are transported using the twin-arginine translocation system (Tat) (Berks et al., 2000). Proteins that are recognized by this system have a conserved motif with invariant twin arginines, S/TRR, that identifies prefolded proteins for transport. Consistent with the pattern for Tat proteins, GidA has a SRR sequence at residues 26–28 near the N-terminus and contains an FAD cofactor. These data suggest that export of GidA with its bound cofactor to the periplasm may involve a Tat system. A homologue of TatD has been identified in M. xanthus, which suggests that a Tat-like transport system may exist in Myxococcus (data not shown).
gidA is not required for adventurous motility
To determine if gidA, as aglU, is required for adventurous motility, a strain carrying a disruption of gidA was constructed. Strain MxH1171 (gidA::nptII) was constructed by homologous recombination between an internal fragment of gidA carried on plasmid pDW301 and the chromosomal copy of gidA in the wild-type strain, which results in a partial gidA merodiploid (Fig. 3A). The polymerase chain reaction (PCR) was used to confirm the disruption of gidA. Primer C will anneal only with the M13 region on the integrating plasmid and primer D will anneal only with a region 380 bp upstream of gidA on the M. xanthus chromosome. Thus, neither the wild-type gidA nor the plasmid alone would give a product using primers C and D. The 1400-bp PCR product, obtained using chromosomal DNA from strain MxH1171, showed that plasmid pDW301 integrated into the gidA gene.
The motility phenotype of the MxH1171 mutant on 1.5% and 0.3% agar was compared with the wild-type strain and the aglU mutant to determine if GidA is involved in motility. The colony morphology of the gidA mutant is nearly identical to that of the wild-type strain and both isolated cells and rafts of cells are present at the edge of the colony. Colonies of the gidA mutant spread at the same rate as wild-type cells on 0.3% agar, a condition in which social motility predominates. Moreover, immunoblot analysis shows that fibrils and pili, the components that are required for S-motility, are present at normal levels (data not shown). Hence, GidA is not required for S-motility. Cells of the mutant were able to move as isolated cells at the periphery of the colony, showing that gidA is not involved directly in A-motility. However, the spreading rate on 1.5% agar, a condition where A-motility predominates, is slightly lower in the mutant than the wild-type strain. Disruption of gidA may have a mild affect on expression of aglU, the A motility gene downstream, or GidA may play a subtle role in motility.
Although MxH1171 grows well on solid medium, producing colonies that look identical to the wild type, the mutant does not grow dispersed in liquid like the wild type. When liquid-grown cultures are examined under the microscope, the cells aggregate into clumps that can be dispersed only by sonication.
The phenotype of the gidA mutant is unstable
When starved for nutrients, the wild-type strain responds by forming multicellular aggregates, called fruiting bodies, within which rod-shaped cells differentiate into spheres that develop into heat-resistant spores after about 30 h (Fig. 4A and B). The MxH1171 strain initially forms fruiting bodies similar to the wild type and produces a wild-type complement of heat-resistant spores (Fig. 4C and D). However, after several transfers of vegetative cells on solid medium, the MxH1171 cells are no longer able to develop when starved for nutrients (Fig. 4E and F). This Dev– derivative, MxH1171*, is stable and does not revert to MxH1171. MxH1171* is indistinguishable from MxH1171 on solid rich medium. However, in contrast with MxH1171, which produces clumps when grown in rich broth, the MxH1171* strain grows dispersed in liquid culture like the wild type. The switch from MxH1171 to MxH1171* usually occurred after the newly constructed strain MxH1171 had been passed several times on rich medium (within two to three weeks). In contrast, the wild-type strain always continued to develop on starvation medium despite many transfers on rich medium.
MxH1171* produces heat-resistant rods
When starved for nutrients, the MxH1171* fails to ripple or aggregate and fruiting bodies do not form even upon extended incubation (Fig. 4E). After 5 days, a sample of MxH1171* was examined microscopically to determine if cells had differentiated into spores. Unlike the aglU mutant, which is able to undergo the initial steps of differentiation to form non-refractile spheres, cells of the MxH1171* remained rod-shaped (Fig. 4F).
To determine if the MxH1171* produces heat-resistant cells, samples that had been allowed to starve for 5 days on TPM agar (10 mM Tris, 1 mM potassium phosphate and 5 mM MgSO4, at pH 7.6) were heated at 50°C for 2 h to kill heat-sensitive (undifferentiated) cells, then plated on rich medium. Surprisingly, although spherical cells that are characteristic of myxospores were never detected by microscopic examination, cells of the mutant were able to germinate on rich medium, producing colonies equivalent to about 0.1% of the complement of heat-resistant spores as the wild type. When vegetative cells of strain MxH1171* were heated at 50°C for 30 min, then plated on rich medium, no colonies formed. Hence, the ability of starved MxH1171* cells to survive heat treatment for 2 h at 50°C suggests that some cells have differentiated into a heat-resistant cell type, even in the absence of spore formation. Surprisingly, when incubated with 0.5 M glycerol, strain MxH1171* formed spherical spores that were indistinguishable microscopically from glycerol spores produced by the wild-type strain.
MxH1171* inhibits wild-type development
The phenotype of MxH1171* is similar to strains that carry mutations in genes involved in the production of extracellular signals during development. Mutants defective in signal production can be distinguished by extracellular complementation experiments in which a wild-type strain provides a missing signal that rescues a sporulation defect transiently. To determine if the developmental defect of strain MxH1171* could be complemented by wild-type cells, mixtures containing equal amounts of wild type (DK1622) and the MxH1171* mutant were incubated on starvation agar at 32°C. In contrast with the samples containing only the wild-type strain, cells in the mixtures failed to aggregate and fruiting body formation was inhibited. After several days, a small number of fruiting bodies formed beyond the perimeter of the original inoculum. This surprising result shows that the gidA– mutant strain MxH1171* produces an inhibitor that poisons the development of the wild-type strain (Fig. 5B).
The inhibitory effect of MxH1171* was less pronounced in mixtures that contained an excess of wild-type cells relative to the mutant. At a ratio of 5:1 (wild-type:mutant), fruiting body formation was inhibited, but at a ratio of 10:1, fruiting bodies were produced (data not shown). This result shows that inhibition of neighbouring cells requires a minimum concentration of inhibitor and may require cell–cell contact. MxH1171* also inhibited fruiting body formation of other strains of M. xanthus, including MxH1176 (ΔgidA-aglU), MxH1223 (sglK::Tn5lac-1222) and MxH1777 (ΔaglU). In each of these mixtures, the developmental phenotype was similar to that of the gidA mutant (MxH1171*) alone.
The inhibitor affects transcription of genes expressed early in development
Aggregation is one of the first visible signs of development, yet it occurs several hours after initiation of developmental-specific transcription. The ability of the MxH1171* to abolish aggregation of the wild type suggested that the inhibitor blocks events early in development. To determine if the inhibitor affects transcription of genes expressed early in development, the activity of several development-specific Tn5-lac insertions was compared in mixtures that contained a Tn5-lac marker strain plus either wild-type (unmarked) cells or MxH1171* cells. Myxococcus xanthus strains MxH1813, MxH1809 and MxH1812 carry insertions Ω4521, Ω4455 and Ω4469, which begin to express β-galactosidase at 1.5, 3 and 5 h after starvation, respectively. When each of these strains was developed in a mixture with equal numbers of MxH1171* cells, expression of β-galactosidase was inhibited 45% in MxH1813, 95% in MxH1809 and 66% in MxH1812, relative to a mixture containing equal numbers of each Tn5-lac strain plus wild-type cells (Table 2). These results show that the inhibitor produced by the MxH1171* acts early in development and at the level of gene expression. When vegetative MxH1171* cells were heated at 55°C for 20 min before mixing, fruiting body morphogenesis of the wild-type strain was not inhibited, suggesting that either the inhibitor is not produced in vegetative cells or that the inhibitor is heat labile (data not shown).
Table 2. The MxH1171 * inhibitor affects expression of genes during development.
Strains that carry Tn5-lac markers were allowed to develop in mixtures with an equal amount of DK1622 (wild-type), MxH1171*, or self. Samples were harvested as described in the text and assayed for β-galactosidase activity (calculated as µg ONPG hydrolysed min−1 mg−1 protein). The total activity was adjusted to correct for the presence of DK 1622 and MxH1171*, which lack the Tn5-lac marker. Time refers to the time when the lacZ marker is expressed after the onset of starvation.
The inhibitor is a small molecular weight, heat-stable, protease-resistant molecule that is secreted during development
Strain MxH1171* was induced to starve in tissue culture wells and aliquots of the buffer surrounding the cells were removed at various times to determine if the inhibitor is released during development. The spent buffer was used to suspend a pellet of wild-type cells to a high cell density, then placed on starvation agar. Buffer harvested from MxH1171* cells starved for 2 h delayed, but did not completely inhibit, the fruiting body formation of the wild-type strain. Buffer harvested from MxH1171* cells after 8 h (spent buffer) inhibited fruiting body formation, with peak yield of inhibitor occurring about 18 h after the onset of starvation. In contrast, when the spent buffer harvested from developing wild-type cells was added to fresh wild-type cells, fruiting bodies formed more quickly and were larger in size than fruiting bodies formed by the wild-type strain in TPM buffer.
Spent buffer harvested from the mutant after 18 h was passed through a series of membranes to enrich and estimate the molecular mass of the inhibitor. The wild-type strain was used as a bioassay to demonstrate that the inhibitor was present in the flow-through after ultrafiltration with 10 000, 3000 and 1000 molecular weight cut-off (MWCO) membranes. The inhibitor was partially retained by a 500 MWCO filter, which suggests that the inhibitor is about 500 Da. To determine if the inhibitor was a small peptide, the partially purified sample was incubated with proteinase K, then subjected to ultrafiltration using a YM10 (10 000 MWCO) filter. The filtrate, containing the small molecular weight compound, free of proteinase K, was added to a pellet containing wild-type cells and allowed to develop on starvation medium. After 48 h, fruiting bodies had not formed, which shows that the inhibitor is unaffected by proteinase treatment. The inhibitor also is resistant to heat treatment. Heating the partially purified inhibitor at 80°C for 10 min before adding it to wild-type cells did not affect the ability of the compound to inhibit development. These results show that the inhibitor is a small, heat-stable molecule that is released from the mutant after the onset of development.
Disruption of aglU suppresses the inhibitory effect of the gidA mutant
During development, both gidA and aglU are transcribed together from a promoter upstream of gidA, which hints that these genes may be functionally related during development. Based on the homology between AglU and the E. coli TolB protein, we predict that AglU is a lipoprotein involved in transport of macromolecules. Hence, AglU may affect export of the inhibitor produced by strain MxH1171*. To test this idea, cells of a mutant carrying a deletion of both gidA and aglU (MxH1176) were mixed with the wild-type strain and incubated at 32°C on starvation medium. As shown in Fig. 5C, the double mutant does not affect fruiting body development of wild-type cells. This result suggests that AglU is required for release or processing of the inhibitor.
Ectopic expression of gidA in strain MxH1171* does not complement the developmental defect
The delayed appearance of the MxH1171* phenotype may be as a result of depletion of GidA from the MxH1171 parent, or acquisition of a second mutation. To distinguish between these possibilities, plasmid pDW236T, which carries an intact copy of gidA and approximately 300 bp of DNA upstream of gidA, was integrated at the Mx8 attachment site in strain MxH1171* to produce MxH1832. When starved for nutrients, MxH1832 fails to ripple or aggregate and is indistinguishable from the MxH1171* parent. This result suggests that strain MxH1171* carries a second mutation, outside gidA, that abolishes development.
Myxococcus xanthus strain MxH1176 (ΔgidA aglU) is an A– mutant that ripples and produces fruiting bodies that contain differentiated, heat-sensitive cells when starved for nutrients. Unlike MxH1171 (gidA–) which is unstable, MxH1176 has a stable phenotype and has been passaged for many generations without incident. Because GidA has been depleted in this strain, it should serve as an appropriate host to determine if the amount of GidA in the cell is critical for the Fru– phenotype. Plasmid pDW232 was introduced into strain MxH1176 to make strain MxH1830. Plasmid pDW232, which carries a good copy of aglU but only a portion of the gidA gene, integrates at the native chromosomal location by homologous recombination. Hence, the genotype remains gidA– but becomes aglU+. Plasmid pDW232 restores A-motility to strain MxH1176 and during development, the mutant behaves like the wild-type strain. This shows that addition of aglU alone was sufficient to restore both motility and fruiting body formation to a strain that lacks both GidA and AglU. Thus, the transition from MxH1171 (Fru+gidA–) to MxH1171* (Fru–gidA–) is not solely the result of depletion of GidA during growth. If the phenotype of MxH1171* (Fru– Spo–) was due to dilution of GidA over time, then the Fru– Spo+ phenotype should have been apparent immediately upon introduction of pDW232 into the ΔgidA aglU double mutant, because GidA has already been depleted in this strain. Consistent with the MxH1171 strain, strain MxH1830 was able to aggregate and make fruiting bodies initially, but lost the ability to develop after 10–20 days.
To determine if a second mutation in MxH1171* was linked with gidA, Mx8 phage lysate grown on MxH1171* was used to infect the wild-type strain. All of the KanR transformants (6/6) aggregated and produced dark fruiting bodies like the wild-type on starvation medium, indicating that the Fru– phenotype of MxH1171* is not the result of genetic changes linked with gidA. Within a few generations, transformants derived from MxH1171* were no longer able to develop, consistent with the results obtained with the original gidA disruption mutant. The Fru– MxH1171* strain appears to be stable and has not given rise to revertants that are able to develop.
The GidA protein is conserved among Eubacteria and can be divided into two groups based on size. A large version of the GidA protein (GidAL), ranging in size from 611 to 679 amino acids, found in eukaryotes and eubacteria, was first described in E. coli. Disruption of gidA in E. coli affects cell division when cells are grown on glucose, but has no obvious affect on cell division in the absence of glucose. A smaller protein (GidAS), ranging in size from 435 to 482 amino acids, that shares identity with GidA, has been identified among the genomes of Bacillus subtilis, Aquifex aeolicus, Thermatoga maritima, Rhodobacter capsulatus, Deinococcus radiodurans and Synechocystis sp. The small GidA protein lacks the C-terminal domain present in the E. coli GidA. In eubacteria, the gene encoding GidAL is usually located near the replication origin (oriC) whereas the chromosomal position of the gene encoding GidAS is variable. Myxococcus xanthus carries genes encoding the large GidA (615 amino acids) and the small GidA (457 amino acids) proteins. In addition to M. xanthus, the eubacteria B. subtilis, A. aeolicus and T. maritima also carry genes encoding both GidAL and GidAS.
Alignment of both the large and small forms of GidA reveals a conserved, dinucleotide-binding motif at the N-terminus. To determine if GidA binds cofactor, the M. xanthus GidA protein was produced in E. coli. Purified GidA is a yellow protein that absorbs at 375 and 450 nm, which are characteristic of FAD. The peaks at 375 and 450 nm are broader than the peaks observed with purified FAD (Fig. 1C), which is common for cofactors when bound to their cognate protein. When cofactor was extracted from GidA and compared with FAD and FMN by TLC, it co-migrated with FAD.
The FAD moiety of GidA and the conservation of the FAD binding site among all GidA proteins suggest that these proteins catalyse oxidation–reduction reactions or that they act as sensors for the redox state of the cell. In B. subtilis, nitrogen metabolism is regulated by NifA and NifL, in response to elevated levels of oxygen and fixed nitrogen (Bauer et al., 1999). The mechanism by which NifL controls the NifA regulator does not involve phosphorylation of NifL, but rather a change in the redox state of the FAD cofactor bound to NifL. In E. coli, the regulatory flavoprotein Aer couples information about the redox state of the cell to aerotaxis (Bibikov et al., 2000). Like NifL and Aer, GidA may sense changes in proton motive force or redox late in stationary phase and relay this information to regulatory proteins in M. xanthus.
The cellular location of flavoproteins provides clues to their function. The flavoprotein Aer is located in the cytoplasmic membrane where it can act as a sensor. FAD proteins involved in oxidation–reduction reactions required for energy production are associated with the membrane or located in the periplasm (Dym et al., 2000). The small GidA protein was detected in the cytoplasmic and periplasmic fractions of M. xanthus cells harvested from vegetative medium and small amounts of GidA were detected in the membrane fraction. When antibody was used as probe in thin sections by electron microscopy, GidA was found primarily at the periphery of the cell and very little GidA was detected in the cytoplasm. Greater than 90% of the gold particles outside the cell in thin sections of the wild-type strain were associated with material similar to cell debris that may have leaked from the periplasm. Taken together, these results suggest that cytoplasmic GidA is associated with the inner membrane, and is transported to the periplasm. A 75 kDa protein that cross-reacted with the GidA antibody was detected in the periplasmic and membrane fractions of wild-type cells and the gidA mutant. A protein of the same size in E. coli also reacted with antibody (data not shown). The 75 kDa protein is similar to the predicted size of the large GidA protein in M. xanthus and E. coli, which suggests that the cross-reacting material is GidAL. Hence, both of the M. xanthus GidA proteins reside primarily in the periplasm or in association with the inner membrane.
Each of the GidA proteins in M. xanthus lacks an obvious signal sequence and therefore must be transported across the inner membrane by an alternative mechanism. An SRR motif was found in the N-terminus of the small GidA protein, within the dinucleotide-binding motif, making GidA a candidate for a twin-arginine-translocation-like (Tat) system. The Tat system is involved in export of folded proteins to the periplasm, all of which bind a cofactor. In these systems, binding to a cofactor appears to be critical for export because it positions the S/TRR motif for recognition by the Tat export complex. Myxococcus xanthus has a gene encoding a protein with 40% identity with the E. coli TatD and probably has other proteins that comprise a homologous Tat complex. If M. xanthus transports GidA using a Tat-like complex, the protein probably is folded in the cytoplasm and binds to the FAD cofactor before export to the periplasm.
Previously, we reported that aglU, the gene downstream of gidA, is predicted to encode an outer membrane lipoprotein required for adventurous gliding motility and maturation of spores during development of M. xanthus. Because the aglU gene is transcribed with gidA, it seemed probable that the GidA protein might also be involved in A motility (White and Hartzell, 2000). However, gidA is not required for A motility because a gidA mutant spreads at a rate of about 100 mm2 per day on 1.5% agar, nearly identical to that of the wild-type strain.
Newly constructed gidA– mutants could be passed for several generations without losing the ability to form mature, spore-filled fruiting bodies under starvation conditions. However, as cells aged, a stable derivative, MxH1171*, was obtained. In the wild-type strain, GidA may be needed to oxidize or reduce a material in the periplasm. If the absence of GidA results in an accumulation of an oxidized or reduced material, which is mildly toxic, a second mutation, which yields MxH1171*, may be needed for long-term survival of the gidA mutant. The conversion of Fru+ MxH1171 to Fru– MxH1171* is not the result of genetic instability of the mutation in MxH1171 because different types of gidA mutants also give rise to the same Fru– phenotype. When the gidA mutation from MxH1171* was transduced into a wild-type background, development was normal, similar to development of the original gidA mutant. Finally, the change in phenotype is not the result of depletion of GidA during passage because expression of gidA from the Mx8 attachment site does not complement the developmental defect of the MxH1171* mutant. Expression of genes integrated at the Mx8 attachment site is not so robust as expression from the native chromosomal site (Fisseha et al., 1996). Reduced expression of gidA might explain why pDW236T failed to complement the MxH1171* mutant.
GidA is located in the cytoplasm during development, in contrast with its cytoplasmic and periplasmic distribution in growing cells. This hints that, upon starvation, the transport of GidA to the periplasm is abolished or that GidA is quickly removed from the periplasm in starving cells. Whereas motility of the MxH1171* is identical to that of the MxH1171 parent during growth, when starved for nutrients, the MxH1171* strain does not form fruiting bodies or produce spherical-shaped spores. When aliquots of MxH1171* were examined under the microscope, only misshapen rods were detected after prolonged incubated on starvation medium. We have noticed that the wild-type strain also produces misshapen rods at various stages during development, hinting that a misshapen rod is a normal part of the transition from a rigid rod to a sphere. Surprisingly, although the morphological change from rods to spherical spores did not occur in strain MxH1171*, a significant number of heat-resistant cells were produced. This shows that internal events critical for conferring heat resistance have occurred in the mutant. Although the production of heat-resistant rods is uncharacteristic of M. xanthus, other closely related species of myxobacteria develop heat-resistant cells in the absence of morphogical conversion to spheres (Reichenbach, 1984). Hence, the pathways that control components critical for heat-resistance may be distinct from pathways for cell morphogenesis.
The MxH1171* mutant inhibits development of wild-type cells when the two are allowed to develop as a mixture. The inhibition occurred early in development and affected the transcription of genes that are expressed as early as 1.5 h after the onset of development. We noticed that the inhibitory affect was generally confined to the space where the initial sample was placed because small fruiting bodies were able to develop at the periphery of the spot. This hinted that inhibition required cell–cell contact or that the inhibitor was poorly diffusible. The ability of MxH1171* to inhibit the wild-type strain is not because of interference from the mutant cells, but was the result of an inhibitor that is released from the mutant. Spent-buffer harvest from the mutant during development was sufficient to inhibit the development of the wild-type strain. A mutation in the aglU gene suppresses the phenotype of a gidA mutant and prevents the release of the inhibitor. Because AglU is predicted to form of a Tol-like transport complex in the membrane, these data suggest that AglU is involved in export of the inhibitor. Initial characterization shows that the inhibitor is a heat-stable, protease-resistant molecule of about 500 Da. Hence, it is improbable that the inhibitor is a peptide, but a small compound that may interfere with a receptor that is required for signalling during development.
Despite the conservation of the gidA gene among the bacteria, the phenotypes of different gidA mutants do not point to a common mechanism for GidA function in these organisms. Moreover, a detailed understanding of GidA is complicated by the fact that homologues of GidA come in two sizes, and are transcribed from genes that are not linked genetically. Mutations that affect the small form of GidA have not been described before now. In contrast, some information is available about the large form of GidA. A mutation in the E. coli gidA gene causes cells to form long filaments when grown on medium with glucose (von Meyenburg and Hansen, 1980; von Meyenburg et al., 1982). Studies of oriC in E. coli have shown that active transcription of the gidA gene, which is located immediately to the left of oriC, is required for efficient initiation of replication. The filamentous phenotype of the gidA mutant in E. coli and the proximity of gidA to oriC suggest that GidA may couple replication with cell division. This is consistent with the fact that Proteus mirabilis gidA mutants are unable to differentiate from swimmer to swarmer cells, a process involving cell division, when placed on solid medium (Belas et al., 1995). Like the small, periplasmic, FAD-binding GidA protein in M. xanthus, we predict that the large and small forms of GidA among the eubacteria bind FAD and that a subset of GidA proteins are periplasmic. This emerging family of proteins may play a critical role in sensing redox changes, or in catalysing redox reactions that couple replication and growth with cell division and differentiation.
Growth of bacterial strains
Bacterial strains, plasmids, oligonucleotide primers and phage strains are listed in Table 3. Escherichia coli strains JM107 (Yanisch-Perron et al., 1985) and BL21 (Novagen), grown at 37°C in Luria–Bertani (LB) broth or agar (Sambrook et al., 1989), were used for the construction of plasmids and the preparation of plasmid DNA, and expression of GidA respectively. Escherichia coli derivatives with plasmids were grown in LB supplemented with kanamycin sulphate (Kan; 40 µg ml−1) or tetracycline (Tet; 30 µg ml−1).
gidA region amplified using oligonucleotide primers A + B
gidA gene in pET24B
Mx8 att, nptII
Ectopic complementation of M. xanthus mutants
TA cloning vector
nptII, T7 promoter
Expression of GidA
Upstream of gidA
DK1622 is the yellow, fully motile, strain that we use as wild-type (Wall et al., 1999). MxH1176 is an A–S+ strain that carries a deletion of the last 430/457 codons of the gidA coding sequence and the first 252/476 codons of the aglU coding sequence in strain, DK1622 (White and Hartzell, 2000). MxH1777 is an A–S+ strain that carries a deletion of codons 74–378 of the aglU coding sequence in DK1622. Because tan variants have a reduced A gliding motility, only yellow variant colonies of M. xanthus were used in these studies. Construction and characterization of strains containing Tn5-lac insertions in genes expressed at different stages of development have been described (Kroos et al., 1986). Tn5-lac insertions Ω4521, Ω4455 and Ω4435, in a DK101 background, were a gift from Heidi Kaplan. Transposons were transduced from strain DK101 to strain DK1622 and KanR colonies were selected.
Myxococcus xanthus strains were grown at 32°C in CTPM (1% casitone, 10 mM Tris, 1 mM potassium phosphate and 5 mM MgSO4 at pH 7.6) broth with vigorous shaking or on CTPM with 1.5% agar. CTPM medium was supplemented with kanamycin sulphate (40 µg ml−1) to select for plasmids or strains carrying resistance determinants. For selections involving tetracycline (8 µg ml−1), CTP (CTPM medium lacking MgSO4) medium was used. TPM starvation medium is CTPM without casitone. Enzymes from NEB, Gibco/BRL or Promega, were used in accordance with the manufacturers' instructions. Other chemicals were from Sigma or Fisher.
Bacteriophage and plasmids
The generalized transducing phage Mx8 cp2 (Martin et al., 1978) was used to move Tn5-lac and other marker regions among M. xanthus strains. A series of 10-fold dilutions (10−1-10−5) of the Mx8 cp2 phage were each added to 200 µl of exponentially growing donor strains (5 × 108 cells ml−1), incubated at room temperature for 20 min and plated on CTPM plates using 3 ml of CTPM with 0.35% agar. Phage was harvested from each dilution plate that contained near complete lysis, after incubation for 2 days at 32°C, by adding 3 ml of TPM broth to the top agar to suspend phage particles. The supernatant was harvested from the plate after incubation for 4 h at 4°C, treated with 50 µl of CHCl3 and centrifuged at 12,000 g for 10 min at 4°C to pellet cell debris. Transduction of the markers into recipient strains was performed using 100 µl of phage suspension and 500 µl of exponentially growing recipient cells. Mixtures were incubated for 20 min at room temperature, with gentle agitation, and plated on CTPM plates containing the appropriate antibiotic using 3 ml of CTPM or CTP containing 1.5% agar.
Plasmids were introduced by electroporation into E. coli (Taketo, 1988), and M. xanthus (Kashefi and Hartzell, 1995). PCR amplification reactions consisted of: DNA (5 ng of chromosomal DNA or 1 ng of plasmid DNA) in a 25 µl reaction, using 10 µM of each oligonucleotide primer, 25 µM each dNTP, polymerase buffer, 1.5 µM MgCl2 and 1 unit of Taq polymerase (Gibco/BRL). Then, 30 cycles of 95°C for 30 s, 55°C for 30 s and 72°C for 1 min were used to amplify DNA.
Cloning and expression of M. xanthus genes
The gidA gene was amplified from the DK1622 chromosome using oligonucleotide primers A and B, listed in Table 3, and ligated to plasmid pTOPO2.1 (Clontech) to make plasmid pDW430. Plasmid pDW430 was digested with NdeI and XhoI and ligated with pET24b which contains the T7-promoter driven expression system (Novagen) to make plasmid pDW432 and electroporated into E. coli strain BL21. Escherichia coli strain BL21 carries a copy of T7 RNA polymerase, controlled by lacUV5. The engineered 5′NdeI and 3′XhoI sites allowed for IPTG-inducible production of GidA-His6.
Purification of the rGidA-His6 in E. coli
To induce the expression of the rGidA-His6 protein, a single colony of BL21 carrying pDW432 was grown in LB medium with 10 mM glucose and kanamycin to an OD600 = 0.4, then harvested by centrifugation at 5500 g and suspended in the same volume of LB medium with 10 mM IPTG and kanamycin. The culture was grown for 3 h at 30°C, then chilled on ice and harvested by centrifugation at 5500 g for 10 m at 4°C. The cell pellet was stored at −20°C. Frozen cells were suspended in 10 ml of phosphate buffer saline at pH 7.6 with 0.5% n-octyl-β-d-glucopyranoside (buffer A), then lysed by passage through a French pressure cell at 19 000 psi and centrifuged at 245 000 g for 1 h. The soluble fraction containing rGidA was affinity-purified using Talon resin (Clontech, Palo Alto), washed with 10 vol. of buffer A and eluted with 10 ml of buffer A containing 10 mM imidazole. Samples were fractionated by SDS–PAGE gel to detect the 50 kDa GidA protein. Fractions containing GidA were yellow in colour. Spectral data for these fractions were collected with a Perkin-Elmer UV/VIS Lambda-12 spectrophotometer scanning from 300 to 700 nm.
Spectral analysis showed that GidA contained a flavin cofactor. To further characterize the cofactor, 5 µg of recombinant protein was precipitated using 5% TCA to extract the cofactor. The rGidA cofactor, FAD, and FMN were spotted on Whatman no. 1 paper, air-dried, and chromatographed in the dark, using three solvent systems: 5% aqueous Na2HPO4n-butanol-acetic acid-H2O at ratios of 4:1:5 and 12:3:5. The position of each spot was visualized using a 365 nm-UV (Huennekens and Felton, 1957). The stoichiometry of FAD in a known amount of GidA was estimated from the molar extinction coefficient of 11 300 M−1 for FAD at 450 nm.
To determine the cellular location of GidA in M. xanthus, wild-type (DK1622) and gidA mutant (MxH1171) cells harvested from vegetative and developing cells were analysed in Western blots, probed with anti-GidA antibody. Cells were grown to density of about 5 × 108 cells ml−1, harvested by centrifugation, and suspended in TPM buffer to 5 × 109 cells ml−1. One half of the sample was dispensed in 400 µl aliquots into a 24-well polystyrene plate and incubated at 32°C for 2 h (developing sample), whereas the other half of the cell suspension was immediately fractionated (vegetative sample). After 2 h, the cells in the 24-well plate were harvested and fractionated as described elsewhere (Thomas et al., 2001). Cells were collected by centrifugation at 2900 g, suspended in 20% sucrose, 10 mM Tris 0.5 mM EDTA at pH 7.6, and incubated at 25°C for 10 min. The sample was centrifuged at 2900 g at 24°C for 10 min and the cell pellet was suspended in ice-cold 5 mM MgSO4 and incubated on ice for 20 min to release the periplasmic contents. The spheroplasts were separated from periplasmic material by centrifugation at 4400 g at 4°C for 15 min. The spheroplast pellet was suspended in ice-cold MgSO4 and sonicated on ice (5 × 10 s pulses) and the cell debris and whole cells were isolated by centrifugation at 10 000 g. The supernatant fraction containing bacterial cell membranes and cytoplasmic contents was separated by centrifugation at 252 000 g for 30 m at 4°C using a Sorvall RCM120EX. The supernatant that contained cytoplasmic contents was removed and the membrane pellet was suspended in 10 mM Tris (pH 8) containing 0.1% SDS. Aliquots of each fraction were separated by 10% SDS–PAGE and transferred to PVDF membranes (Amersham) using a mini-trans-blot apparatus (Bio-Rad).
Polyclonal IgY antibody was generated in chicken immunized with purified GidA and purified from the yolk using the Eggcellent Chicken IgY purification kit (Pierce). GidA proteins were detected using a GidA primary antibody generated in chicken and secondary antibody raised against chicken antibodies (Aveslabs) cross-linked with horseradish peroxidase using the ECL detection kit (Amersham), following the manufacturer's instructions, and imaged. Polyclonal antibody from rabbits against M. xanthus cytoplasmic protein, FrzCD, was kindly provided by David Zusman and John Kirby.
The production of extracellular fibrils and pili by gidA mutants was compared with wild-type cells by immunoblot analysis using monoclonal antibody 2105 that reacts with fibril material (provided by Marty Dworkin, Behmlander and Dworkin, 1994) and polyclonal antibody against PilA (provided by Dale Kaiser, Wu and Kaiser, 1997) respectively (data not shown). Cells were prepared as described previously (White and Hartzell, 2000).
Electron microscopy of vegetative cells
To prepare samples for transmission electron microscopy, wild-type and gidA mutant cells were grown in CTPM liquid, concentrated fivefold by centrifugation (10 000 g) and fixed in 2.5% glutaraldehyde (v/v), 2% paraformaldehyde (v/v) in 50 mM PIPES (pH 7.0) overnight at 4°C. Cells were washed with 50 mM PIPES, dehydrated with ethanol and embedded in London Resin white (Polysciences). Thin sections were placed on Ni-200-mesh grids coated with formvar for use in immunolocalization studies.
Immunolocalization of GidA
Antibody generated against GidA was used to localize GidA in wild-type and gidA-cells. Formvar-coated grids containing wild-type or gidA mutant cells were blocked in TBST (10 mM Tris, 250 mM NaCl, 0.3% Tween-20 at pH 7.6) for 2.5 min using a Pelco 3450 lab microwave processor (Redding) at 37°C, cooled to 25°C and washed (2 × 15 s) in BST. Cells were probed with anti-GidA antibody at a dilution of 1:10 for 2.5 min in the microwave processor, allowed to cool and washed (2 × 15 s) with BST. Cells were probed with secondary antibody (rabbit α chicken, 1:50) conjugated to colloidal gold (18 nM) for 2.5 min in the microwave processor, allowed to cool, washed in 2.25 M NaCl2 (1 × 15 s), H2O (2 × 15 s) and allowed to dry. Samples were stained with potassium permangenate:uranyl acetate (1:3) for 5 min, rinsed with 3 vol. H2O and dried and visualized using a JEOL 1200 EX transmission electron microscope.
Construction of gidA mutants
Myxococcus xanthus strain MxH1171 is a KanR mutant generated by insertion of plasmid pDW301 in the gidA gene in an otherwise isogenic background. Because plasmid pDW301 integrates by homologous recombination between an internal fragment of gidA on the plasmid and the gidA locus, the mutant carried two partial copies of gidA separated by the pBGS18 plasmid. To make pDW301, a 928-bp internal Eco47III–BglII fragment (bp 630–1450) of the gidA gene from plasmid pDW120 was ligated with plasmid pBGS18 and digested with BamHI and SmaI. Plasmid pDW301 was electroporated into DK1622 and KanR colonies were selected. To confirm that integration had occurred at the gidA site, the size of each partial copy of gidA was analysed by PCR. Primer C (M13 forward, Table 3) and primer D (5′-end the gidA on the chromosome) were used to amplify the integrated fragment from DNA prepared the KanR electroporants (White and Hartzell, 2000). When wild-type DNA was used as template for primers C and D, no product was detected.
Myxococcus xanthus strain MxH1830 is a gidA– mutant reconstructed by adding plasmid pDW232, which has a good copy of aglU, to a ΔgidA aglU mutant. Plasmid pDW232 is a pBGS18 derivative with a BglII–SalI fragment from pDW120 that contains the 3′ of gidA (corresponding to residues 341–454) and all of aglU. Plasmid pDW232 was electroporated into MxH1176 (ΔgidA aglU) and KanR colonies, resulting from a homologous recombination event between the partial copy of the aglU gene on the chromosome and the complete copy of aglU on pDW232, were recovered.
Myxococcus xanthus wild-type strain DK1622 and gidA mutants MxH1171 and MxH1830 were grown on CTPM plates containing 40 µgml−1 kanamycin sulphate at 32°C in the dark. Every 5 days an isolated yellow colony was used to inoculate a fresh CTPM plate with kanamycin and a 5 ml broth culture of the same media. The broth culture was grown overnight with shaking at 32°C, concentrated to 5 × 109 cells ml−1 and incubated on starvation agar at 32°C to monitor development. This screening procedure was continued weekly until single colonies from both MxH1171 and MxH1830 were unable to develop (Fru–). The MxH1171* strain is the stable Fru– version of MxH1171. Wild-type strain DK1622 was grown and screened under identical conditions. Repeated attempts to isolate a Fru+ revertant of MxH1171* were unsuccessful. Oligonucleotide primers C and D were again used in a PCR reaction with template DNA from the Fru-MxH1171 strain to verify that the gidA disruption was maintained (data not shown).
Adventurous and social gliding capabilities of strain MxH1171 (gidA::pDW301) and MxH1830 (gidA::pDW232) were compared with the fully motile strain DK1622. Strains were grown in CTPM broth, concentrated to 5 × 109 cells ml−1 and 2 µl aliquots were spotted onto CTPM agar containing 0.3% and 1.5% agar. The area of the spot was measured approximately every 24 h for 5 days. The formation of A and S motility flares on 1.5% agar was examined with a Nikon FXA microscope at 25X. The spreading ratio was determined from the average area of six samples incubated over the five-day period (Shi and Zusman, 1993).
To initiate development of fruiting bodies and spores, cells in logarithic phase were harvested by centrifugation and suspended in TPM buffer to a density of 5 × 109 cells ml−1. Aliquots (20 µl) were spotted on TPM agar and incubated at 32°C for 5 days. Developing samples were examined every 12–24 h under a Nikon SMZ-U stereomicroscope to monitor the progress of rippling, mound formation and darkening of the fruiting body. On days 2 and 5, samples were removed and examined by diffraction interference contrast at 800X to determine if rod-shaped M. xanthus cells had differentiated to spheres, and to monitor the appearance of refractile spores. On day 5, plates containing developing cultures were incubated at 50°C for 2 h. Five spots were scraped into piles, suspended in 100 µl TPM buffer and sonicated briefly to disperse the aggregates within the fruiting body. Samples were serially diluted in TPM, then spotted on CTPM agar to quantify germination (MacNeil et al., 1994).
To determine if the developmental phenotype of the MxH1171* mutant could be rescued by mixing with the wild-type strain, 100 µl volumes of wild-type and mutant cells at 5 × 109 cells ml−1 were mixed in Eppendorf tubes, then spotted in 20 µl aliquots on TPM starvation agar. Homogeneous 20 µl suspensions of both wild-type and the gidA mutant cells were spotted on the TPM plates for comparison. Mixtures were incubated at 32°C and monitored daily under the microscope to detect rippling, aggregation and mound and spore formation.
To determine if the gidA gene is able to complement the Fru– phenotype in strain MxH1171* (gidA::pDW301), plasmid pDW236T, which carries a 3.5 kb gidA aglU fragment, tetA and the Mx8 att-int region, was constructed. pDW236T was constructed by cloning the 1.4 kb tetA gene from pBR323 into EcoR1 digested pDW236 (White and Hartzell, 2000). pDW236T was electroporated into MxH1171* to make strain MxH1832. Transformants were selected on CTP plates containing 8 µg ml−1 tetracycline, and integration at the attB site was verified by PCR as described previously (White and Hartzell, 2000).
Expression of β-galactosidase from strains with insertions of Tn5-lac
Tn5-lac insertions were used as markers to determine if the inhibitor produced by strain MxH1171* inhibits expression of genes induced during development. Strains MxH1813, MxH1809 and MxH1812 carry insertions Ω4521, Ω4455 and Ω4435 in an otherwise wild-type background. Expression of lacZ from Ω4521, Ω4455 and Ω4435 begins at 1.5, 3 and 5 h, respectively, after the onset of development. Each Tn5-lac marker strain was mixed in equal amounts with MxH1171* (gidA–), itself, and the wild-type strain DK1622. Strains were grown in CTPM medium and concentrated to 5 × 109 cells ml−1. Then, 40 µl aliquots were mixed in a single well of a 96-well microtiter plate and incubated at 32°C. After 6 h, SDS and EDTA were added to each 80 µl sample to final concentrations of 1% and 1 mM, respectively. Lysates were stored at 4°C.
β-Galactosidase assays were performed as described elsewhere (Kroos et al., 1986) with the following modifications. To determine β-galactosidase activity from each strain during development, 20 µl of 1:25 dilution of each cell lysate was mixed with 80 µl of Z buffer (100 mM Na2HPO4 pH 7, 10 mM KCL, 1 mM MgSO4 and 50 mM β-mercaptoethanol) containing 1 mg ml−1o-nitrophenyl-β-d-galactoside (ONPG) in a microtiter plate. Samples were incubated at 37°C until sufficient yellow colour developed or 1.5 h, at which time 100 µl of stop buffer (1 M Na2CO3) was added and the absorbance at 420 nm was recorded on a Titertek Multiskan 340. All samples were assayed in triplicate. The blank for each sample contained the same volume of diluted, lysed cells in Z buffer without ONPG. Protein concentration was determined using the Bradford reagent (Bio-Rad). Developmental cell samples were lysed by addition of 1% SDS and 20 µl of a 1:125 dilution of cell lysate was mixed with 20 µl of undiluted Bradford reagent and 60 µl of H2O. All samples were assayed in triplicate and compared with a standard assay using bovine serum albumin (BSA) and γ-globulin. Samples were incubated at 37°C for 20 min, and the absorbance at 595 nm was monitored in a microtiter plate using a Titertek Multiskan 340 (Labsystems).
Isolation and partial purification of the inhibitor produced by MxH1171*
Strain MxH1171 produces an extracellular inhibitor of development during the first 24 h after the onset of starvation. To isolate the inhibitor produced by strain MxH1171, vegetative cells were concentrated to a density of 5 × 109 cells ml−1 in TPM and 400 µl aliquots were dispensed into a 24-well plate. After incubation at 32°C for 18 h, the cells and spent TPM were harvested and separated by centrifugation at 6500 g for 10 min. Spent TPM from starving wild-type cells (DK1622) was isolated at 18 h and tested for the presence of inhibitor. The spent TPM was filtered by passage through an ultrafiltration apparatus (Amicon) containing a YM10 filter at a pressure of 60 psi. To assay the inhibitor, vegetative-grown, wild-type cells were harvested by centrifugation and suspended to 5 × 109 cells ml−1 in either fresh TPM, spent TPM from DK1622 (wild-type), spent, filtered TPM from MxH1171 or spent, unfiltered TPM from MxH117. Cells were spotted in 20 µl aliquots on TPM plates and incubated at 32°C.
Characterization of the inhibitor
To determine if the inhibitor is sensitive to heat, 100 µl of the spent, filtered TPM containing the inhibitor was heated to 80°C for 10 min. The sample was allowed to cool, then centrifuged at 13 000 g for 5 min to remove debris. Then, 100 µl of the supernatant was used to suspend 5 × 108 wild-type cells that were spotted in 20 µl aliquots on TPM agar. An aliquot of the same TPM sample containing inhibitor, which was not heated, was used to suspend wild-type cells for comparison. Cells were allowed to develop for 5 days at 32°C and monitored daily under the microscope.
To determine if the inhibitor was sensitive to protease, proteinase K (final concentration 250 µg ml−1) and CaCl2 (final concentration 5 mM) were added to 1.5 ml of the spent, filtered TPM from strain MxH1171 and incubated at 37°C for 2 h. Proteinase K was separated from the small molecular weight inhibitor by ultrafiltration using a YM10 filter, which retains the proteinase K, but not the inhibitor. The flow-through material was used to suspend wild-type cells to a concentration of 5 × 109 cells ml−1 and allowed to develop at 32°C.
Samples were separated on Whatman no. 1 paper as described in the text. The rGidA cofactor was released from purified protein by TCA precipitation. Apoprotein was removed by centrifugation. Solvent system A: 5% disodium hydrogen phosphate in H2O; solvent system B: n-butanol-acetic acid-H2O (12:3:5); and solvent system C: n-butanol-acetic acid-H2O (4:1:5).
Strains that carry Tn5-lac markers were allowed to develop in mixtures with an equal amount of DK1622 (wild type), MxH1171* or self. Samples were harvested as described in the text and assayed for β-galactosidase activity (calculated as µg of ONPG hydrolysed min−1mg−1 protein). The total activity was adjusted to correct for the presence of DK1622 and MxH1171*, which lack the Tn5-lac marker. Time refers to the time at the lacZ marker is expressed after the onset of starvation.
We wish to thank the staff at the electron microscopy facility at Washington State University for help with preparation of samples for immunocytochemistry and Phil Youderian for helpful discussions and sequence data. The sequence of the gene encoding the large GidA protein was obtained from Cereon Microbial Sequence Database. This work was supported by grants from the National Institutes of Health (GM50962) and National Science Foundation (MCB0094635) to PLH.