Social motility in Myxococcus xanthus requires FrzS, a protein with an extensive coiled-coil domain



Gliding motility in the developmental bacterium Myxococcus xanthus involves two genetically distinct motility systems, designated adventurous (A) and social (S). Directed motility responses, which facilitate both vegetative swarming and developmental aggregation, additionally require the ‘frizzy’ (Frz) signal transduction pathway. In this study, we have analysed a new gene (frzS), which is positioned upstream of the frzA–F operon. Insertion mutations in frzS caused both vegetative spreading and developmental defects, including ‘frizzy’ aggregates in the FB strain background. The ‘frizzy’ phenotype was previously considered to result only from defective directed motility responses. However, deletion of the frzS gene in an AS+ motility background demonstrated that FrzS is a new component of the S-motility system, as the AfrzS double mutant was non-spreading (AS). Compared with known S-motility mutants, the frzS mutants appear similar to pilT mutants, in that both produce type IV pili, extracellular fibrils and lipopolysaccharide (LPS) O-antigen, and both agglutinate rapidly in a cohesion assay. The FrzS protein has an unusual domain composition for a bacterial protein. The N-terminal domain shows similarity to the receiver domains of the two-component response regulator proteins. The C-terminal domain is composed of up to 38 heptad repeats (a b c d e f g)38, in which residues at positions a and d are predominantly hydrophobic, whereas residues at positions e and g are predominantly charged. This periodic disposition of specific residues suggests that the domain forms a long coiled-coil structure, similar to those found in the α-fibrous proteins, such as myosin. Overexpression of this domain in Escherichia coli resulted in the formation of an unusual striated protein lattice that filled the cells. We speculate on the role that this novel protein could play in gliding motility.


Myxococcus xanthus is a Gram-negative bacterium that has a complex life cycle, which involves vegetative spreading under nutrient-rich conditions and fruiting body formation when starved. Both vegetative spreading and developmental aggregation, whereby cells stream into fruiting bodies, require co-ordinated cell movements (Ward and Zusman, 1997). Movement in M. xanthus takes place by gliding motility, a poorly understood process that occurs only on solid surfaces and requires two genetically distinct systems (Hodgkin and Kaiser, 1979a,b). The adventurous (A) motility system facilitates movement of individual isolated cells, although it is also involved in the movement of cells within groups. In contrast, the social (S) motility system requires cell–cell contact and is therefore involved only in the movement of groups of cells (Kaiser and Crosby, 1983; Kroos et al., 1990). Currently, no mechanism for A motility has been firmly established. S motility has been more fully characterized and is known to require polar type IV pili (Kaiser, 1979). These structures are also involved in twitching motility, another surface-associated translocation mechanism found in Pseudomonas aeruginosa and certain other Gram-negative bacteria (Darzins and Russell, 1997). However, it is still not clear how pili are involved in either translocation process, although extrusion and retraction of these structures has been hypothesized to provide the mechanical basis for movement (Bradley, 1980; Wall and Kaiser, 1999). S motility in M. xanthus has also been shown to require an extracellular polysaccharide–protein matrix, termed fibrils (Behmlander and Dworkin, 1994a,b), and the O-antigen of lipopolysaccharide (Bowden and Kaplan, 1998).

Currently, two signal transduction systems (Frz and Dif) showing homology to the chemotactic (Che) directed motility systems of the flagellated enteric bacteria Escherichia coli and Salmonella typhimurium have been identified in M. xanthus (McBride et al., 1989; McCleary and Zusman, 1990; Yang et al., 1998). The dif system is specifically required for S motility (Yang et al., 1998). In contrast, the frz system regulates the reversal frequency of cell gliding, whether the cells are moving as isolated individuals (Blackhart and Zusman, 1985) or in groups (Shi et al., 1996). In this study, we have analysed a new gene (frzS) that is located upstream of the frz operon. We propose frzS to be a new S-motility gene and show that the frzS mutant phenotype is similar to the pilT mutant phenotype (Wu et al., 1997).


Cloning and sequencing of the frzS gene

Previously, we reported the cloning and sequencing of two complete genes, rpoE1 and orf5, and the 5′ end of a partial gene, orf6, which we proposed to be transcribed together as an operon (Ward et al., 1998a). In this study, we have renamed the orf6 gene frzS (see later). To characterize the complete operon, we identified a cosmid from within a library (kindly provided by Ron Gill, University of Colorado) using the previously sequenced 5′ end of the frzS gene as a hybridization probe. The remainder of the frzS gene and the region directly downstream were subcloned from this cosmid on a 0.5 kb PstI fragment (as pP0.5) that overlapped the known sequence and a 1.2 kb PstI fragment (as pP1.2) that lies downstream of the 0.5 kb fragment. The inserts from these new clones were sequenced on both strands. The junction between the two PstI fragments was sequenced on a 2.0 kb SalI fragment (cloned as pYSl2.0) that overlaps both PstI fragments (see Fig. 1). The new sequence identified in this study has been submitted to GenEMBL as an update of our previous submission (accession number AF049107).

Figure 1.

Map of the frz operon and upstream region showing the location of the frzS gene. Arrows indicate the direction of transcription of the genes. Genes that, when mutated in the FB background, result in the ‘frizzy’ aggregation phenotype are shown in black. A potential σA promoter (P) site upstream of rpoE1 is shown. DNA fragments cloned in pUC18 and sequenced during this study are indicated (see Table 2).

Analysis of the sequence data suggested the complete frzS gene to be 1686 bp in length, with appropriate codon usage for M. xanthus over the entire gene. Specific codon usage is highly indicative of the translational potential of proposed open reading frames (ORFs) in M. xanthus, because of the high G+C (67%) content of the DNA (Shimkets, 1993). The proposed ATG start codon lies 176 bp downstream of the orf5 TGA stop codon and is 81 bp downstream of an ‘in frame’ TAG stop codon. No alternative start codons are present in the intervening sequence. A potential ribosome binding site (AGGA) lies 14 bp upstream of the ATG start codon. No potential promoter sites were identified in the orf5–frzS intergenic region, and we were unable to locate a transcriptional start site upstream of the frzS gene using primer extension analysis (not shown), suggesting that frzS might be transcribed as part of an operon.

The TAG stop codon for frzS was followed 16 bp downstream by a probable transcriptional terminator structure with the sequence 5′-CCCCGGGCCACCCGTGCCCGGGG-3′. Analysis of the sequence downstream of this potential terminator structure identified no likely ORFs in the same transcriptional orientation as frzS. However, a gene (orf7) with appropriate codon usage for M. xanthus, but transcribed in the opposite orientation to frzS, was identified (Fig. 1). The stop codon of this gene was positioned 12 bp upstream of the proposed frzS transcriptional terminator. The translated Orf7 protein showed no similarity to other proteins in the GenEMBL databank, and we have not studied the orf7 gene further.

Cells with insertion mutations in the frzS gene show developmental defects

An internal fragment of the frzS gene was amplified by polymerase chain reaction (PCR), cloned into the vector pZErO-2 and the resultant kanamycin-resistant construct (pZOrf6) electroporated into both a fully motile, wild-type strain (DZ2) and a reduced motility, fruiting body (FB) strain (DZF1). As the plasmid cannot replicate in M. xanthus, integration into the chromosome was indicated by maintenance of kanamycin resistance in the electroporants. Insertion of the plasmid within the frzS gene was confirmed by Southern blotting (not shown).

The resultant mutant strains, DZ4219 (DZ2 frzS) and DZF4220 (DZF1 frzS), were analysed for developmental defects by spotting cells at 2 × 109 cfu ml−1 on CF fruiting agar (Hagen et al., 1978) and observing the formation of aggregates and fruiting bodies over a period of 4 days. After development for 4 days, the DZF1 frzS insertion mutant had formed tangled, swirling aggregates (Fig. 2C) characteristic of frz mutants (Fig. 2B), which resulted in the renaming of the orf6 gene as frzS. Sporulation efficiencies were measured after sonication of the 4-day-old fruiting bodies. Spore counts in the frzS‘frizzy’ aggregates showed that sporulation was reduced to approximately 10% of that found in the parent strain, which is slightly lower than the sporulation level found in other DZF1 frz mutants (Blackhart and Zusman, 1985; Kashefi and Hartzell, 1995).

Figure 2.

Developmental aggregation and vegetative swarming phenotypes. Developmental aggregation in (A) DZF1 (FB strain); (B) DZF1 frzE; (C) DZF1 frzS (insertion mutant); (D) DZF1 frzS (deletion mutant); (E) DZ2 (wild type); (F) DZ2 frzE; (G) DZ2 frzS (insertion mutant); and (H) DZ2 frzS (deletion mutant), after 96 h on CF starvation agar. Photographs were taken using a Zeiss dissecting microscope at × 12 magnification. Vegetative swarming after 48 h on CYE agar of (I) DZ2 (wild-type); (J) DZ2 frzE; (K) DZ2 frzS (insertion mutant); and (L) DZ2 frzS (deletion mutant). Photographs were taken at × 8 magnification.

A stronger developmental defect was observed for the DZ2 frzS insertion mutant after 4 days on CF agar. Unlike either the parent DZ2 (Fig. 2E) or an frz mutant in this background (Fig. 2F), the DZ2 frzS mutant did not form mounds, aggregates or fruiting bodies (Fig. 2G). Sporulation in this mutant was also found to be severely defective (less than 0.1% of wild type).

Deletion of the frzS gene does not result in defective development

As the frzS insertion mutants could potentially express a truncated but still partially active form of the FrzS protein (see below), frzS‘in frame’ deletion mutants were constructed in both DZ2 and DZF1 backgrounds. PCR was used to amplify a 500 bp fragment covering from the 3′ end of the orf5 gene to the 5′ end of the frzS gene. A second PCR product was amplified from the 3′ end of the frzS gene. The two PCR fragments were ligated together using KpnI linkers designed on the free ends of the internal PCR primers to create a single fragment containing a 1.1 kb ‘in frame’ deletion of the 1.7 kb frzS gene. This fragment was then digested with restriction enzymes specific for sites designed on the outside ends of the external PCR primers to give a linear fragment with non-complementary ‘sticky’ ends. This fragment was then ligated into appropriately double-digested pKY480 (vector kindly provided by Kyungyun Cho, our laboratory), which encodes for resistance to kanamycin and sensitivity to sucrose (Cho and Zusman, 1999). The frzS‘in frame’ deletion was cloned in the same transcriptional orientation as the sacB gene on pKY480, as suggested previously (Wu and Kaiser, 1996). The resultant plasmid (pKYS1S2) was electroporated into both M. xanthus DZ2 and DZF1, and cells containing a chromosomally integrated copy of the plasmid (which cannot replicate in M. xanthus) were selected on plates containing kanamycin. The kanamycin-resistant strains were then screened for sucrose sensitivity by plating on CYE agar containing 7% sucrose. Cells resistant to kanamycin but sensitive to sucrose (KmR SacB+) were grown without selective pressure in liquid culture for 3 days, diluting the cells back to maintain growth in constant log phase. These cells were then plated onto CYE agar containing 7% sucrose. Spontaneous excision of the plasmid pKYS1S2 from the chromosome would allow cells to grow on these plates. Complete excision of the plasmid would also result in the cells becoming sensitive to kanamycin. The SacB KmS cells were screened for deletion of the frzS gene by Southern blotting (not shown), as excision of the plasmid could result in either deletion of the gene or reconstruction of the complete gene. The DZ2 ΔfrzS mutant (DZ4335) and the DZF1 ΔfrzS mutant (DZF4345) were assayed for developmental defects as described above. However, unlike the DZ2 and DZF1 frzS insertion mutants, the DZ2 and DZF1 frzS deletion mutants produced fruiting bodies (Fig. 2H and D respectively) containing a full complement of spores. However, it should be noted that the fruiting bodies formed by the DZ2 and DZF1 deletion mutants showed slight differences from the parental forms. In the DZF1 background, the mutant appeared to produce more fruiting bodies, whereas in the DZ2 background, the mutant made fruiting bodies with poorly defined or messy edges.

Vegetative swarming is reduced in frzS mutants

To determine whether any of the frzS mutants displayed reduced vegetative spreading, which would result from a motility defect rather than a specific developmental defect, concentrated cells were spotted onto nutrient-rich CYE plates. Plates containing either 1.5% agar (which facilitates effective movement involving the A-motility system) or 0.3% agar (which allows movement principally through the S-motility system; Shi and Zusman, 1993) were used, and translocation of the cells was measured for a period of up to 96 h. In a typical experiment (assays were repeated at least three times), spreading of the wild-type strain DZ2 after 48 h on 0.3% agar (Fig. 2I) had increased from an initial spot diameter of 5 mm to a final diameter of 22 mm. During the same time period, a DZ2 frzE mutant had increased in diameter from 6 mm to 12 mm (Fig. 2J). However, both the DZ2 frzS mutants showed severely reduced spreading: the DZ2 frzS insertion mutant increased in diameter from 5 mm to 8 mm in 48 h (Fig. 2K), whereas the DZ2 ΔfrzS mutant increased in diameter from 7 mm to 10 mm (Fig. 2L).

Differences in the spreading efficiency of the frzS mutants were less obvious when measured on 1.5% agar. After 96 h incubation, starting spots (all 5 mm in diameter) for DZ2, DZ2 frzE, DZ2 frzS and DZ2 ΔfrzS swarms had increased in size to 21 mm, 17 mm, 16 mm and 15 mm respectively (not shown). Although the generally reduced spreading of the DZ2 frzS mutants suggested that the frzS gene could be a motility gene, the difference in spreading efficiencies noted between 0.3% and 1.5% agar surfaces indicated that the motility defect might specifically involve the S-motility system. Surface translocation efficiencies of the DZF1, DZF1 frzE, DZF1 frzS and DZF1 ΔfrzS mutants were not much different on either 0.3% or 1.5% agar surfaces (data not shown).

The frzS insertion mutation has no effect on the motility behaviour of isolated individual cells

As motility defects associated with A-motility, S-motility or directed motility genes are difficult to define based only on developmental aggregation and vegetative spreading phenotypes (see Hartzell and Youderian, 1995), we performed an analysis of isolated single cell movements on 1.5% CYE agar using time-lapse video microscopy. Analysis of reversal frequency (Fig. 3) and speed of movement of the DZ2 frzS insertion mutant (which displayed both strong vegetative spreading and developmental defects) showed the mutant to behave similarly to wild type when observed as isolated cells on 1.5% CYE agar. The movement of isolated cells in this assay suggests that the A-motility system is unaffected by this mutation. The difference in reversal frequency between the wild-type DZ2 (7.3 ± 4.0 reversals h−1) and the DZ2 frzS insertion mutant (10.8 ± 4.1 reversals h−1) show an overlapping distribution. These new data, in conjunction with the previously reported reversal frequency for DZ2 cells (approximately 10 reversals h−1; Blackhart and Zusman, 1985), indicates that there is no significant difference between the reversing behaviour of DZ2 and DZ2 frzS cells under these conditions. The average speed of cell movement for DZ2 (0.77 ± 0.36 µm min−1) and the DZ2 frzS insertion mutant (0.65 ± 0.29 µm min−1) also suggests that there is no significant difference in the speed of movement of isolated individual cells between these strains.

Figure 3.

Reversal frequency of isolated DZ2 and DZ2 frzS (insertion mutant DZ4219) cells filmed for 30 min on 1.5% CYE agar. Cells with mutations in frz genes usually display no reversals over an equivalent period of filming.

The frzS gene is an S-motility gene

As vegetative spreading of the DZ2 frzS (DZ4219) and DZ2 ΔfrzS (DZ4335) mutants on 0.3% agar, which favours movement using the S-motility system, was severely reduced, we considered two likely possibilities with respect to FrzS function; first, that the frzS gene could be an S-motility gene and/or, secondly, that the FrzS protein could be involved specifically with regulating S-motility behaviour. To analyse these possibilities, the frzS insertion and deletion mutations were reconstructed in strain DK1217, which has a mutation in the A-motility system (aglB1) but has normal S-motility (AS+). Although DK1217 still shows some translocation on 1.5% agar, both the DK1217 frzS insertion and deletion mutants (DZ4221 and DZ4337 respectively) were non-spreading (not shown). As cells with mutations in both A-motility and S-motility genes (AS) are non-spreading (Hodgkin and Kaiser, 1979a,b), the non-spreading phenotype of both the DK1217 frzS mutants shows the frzS gene to be an S-motility gene.

To confirm that the frzS gene was not additionally a component of the A-motility system, the frzS deletion mutation was introduced into strain DK1300, which has a mutation in an S-motility gene (sglG1), but has normal A motility (A+S). The resultant double mutant DK1300 ΔfrzS (DZ4338) retained normal adventurous motility, i.e. single cells and small groups of cells were seen moving away from the colony edge under vegetative conditions. This movement of cells away from the colony edge appeared to be identical in both parent and mutant strains (not shown), thereby showing that frzS is not an A-motility gene.

The frzS insertion mutants do not show defects in the production of cell surface-associated S-motility components

S motility has been shown to require the presence of type IV polar pili (Kaiser, 1979), an extracellular polysaccharide–protein matrix termed fibrils (Behmlander and Dworkin, 1994a,b) and LPS O-antigen (Bowden and Kaplan, 1998). Consequently, we analysed the frzS insertion mutants for the presence of these components. Cells were negatively stained with uranyl acetate and examined by electron microscopy for the presence of polar pili. An analysis of 20 cells showed that polar pili were present on 50% of the DZ2 cells and on 37% of the DZ2 frzS insertion mutant cells. Levels of pili on the DZF1 strain were not examined, as FB strains are known to produce reduced numbers of pili (Kaiser, 1979). Those pili identified on DZ2 and DZ2 frzS cells were only seen localized to one pole of each cell. DZ2 was shown to have an average of 2.9 pili per piliated cell end, whereas the DZ2 frzS mutant had an average of 2.0 pili per piliated cell end. However, it should be noted that pili are fragile structures, and the numbers represented here are the minimum number of pili per cell.

End-point dilution colony dot-blot analyses were performed to determine the levels of fibrillar material and LPS O-antigen produced by the frzS insertion mutants in relation to the parent strains. Colony blots were probed with monoclonal antibody (mAb) 2105 for the presence of fibrils on the cell surface, and with mAb 783 for the presence of LPS O-antigen (monoclonal antibodies were generated by Marty Dworkin, University of Minnesota, and Heidi Kaplan, University of Texas respectively). The dilutions at which the cells no longer reacted with either antibody were seen to be very similar between all the strains tested (no more than a twofold difference; data not shown), suggesting that the parent strains and frzS mutants produce similar amounts of fibrils and O-antigen.

The loss of S motility in the frzS mutants, in combination with unaffected type IV pili, fibril and O-antigen biogenesis, appeared to be similar to the previously reported phenotype of pilT mutants (Wu et al., 1997). These PilT mutants (DK2111, DK2112, DK2116 and DK2117) were also shown to retain normal intercellular cohesion profiles in a clumping assay. However, a second group of mutants (DK2145 and DK2161), which also possess both pili and fibrils, showed no clumping (Wu et al., 1997). Therefore, we performed this assay on the DZ2 frzS mutants. Although control strains lacking either pili (DK1300) or fibrils (SW505) showed no agglutination, both the wild-type strain and the DZ2 frzS insertion mutant showed similar clumping profiles (Fig. 4). The clumping profile for the DZ2 ΔfrzS mutant was also similar to the wild type (not shown).

Figure 4.

Clumping assay. Cells were grown to a concentration of 100–110 Klett units and allowed to agglutinate at room temperature (Wu et al., 1997). The OD550 was recorded every 20 min for 5 h. The average for three experiments for each strain is shown with standard deviation error bars. Control strain DK1300 (sglG1;Hodgkin and Kaiser, 1979b) lacks polar type IV pili, whereas SW505 (difA) lacks fibrils (Z. Yang, personal communication).

Cells with insertion mutations in frzS express a truncated form of the FrzS protein

Translation of the DNA sequence of frzS suggests the FrzS protein to be 562 amino acids in length, with a calculated Mr of 60 239 Da. The protein can be conveniently divided into two parts (Fig. 5A): an N-terminal domain (amino acids 1–237) and a C-terminal domain (amino acids 282–562), which are connected by a 44-amino-acid linker region, rich in alanine and proline residues (34.1% Ala, 40.9% Pro). A gapped blast analysis (Altschul et al., 1997) of the N-terminal domain showed it to have similarity to the receiver modules of the two-component signal transduction response regulator proteins. The most similar proteins in the databank were the PhoB homologue of Vibrio cholerae (31% identity and 53% similarity over the first 139 amino acids) and MtrA of Mycobacterium tuberculosis (32% identity and 47% similarity over the first 160 amino acids). Alignment of the N-terminal domain of FrzS with similar response regulator proteins suggests that FrzS may retain a potential phosphorylation site at residue Asp-55. Other residues known to be important for phosphorylation (Asp-11 and Lys-109; Stock et al., 1989) are also conserved. However, no obvious DNA-binding motif (typical of most response regulator proteins; Parkinson and Kofoid, 1992) was identified from the FrzS sequence.

Figure 5.

FrzS analysis.

A. The full length FrzS protein can be divided into three sections: the N-terminal response regulator domain (grey), joined by an Ala-Pro-rich linker to the C-terminal coiled-coil domain (black). The first 286 amino acids of FrzS, covering the response regulator domain, the Ala-Pro linker and the first few residues of the coiled-coil, would be expressed in frzS insertion mutants as the truncated form of FrzS. The majority of the coiled-coil domain in this truncated form of FrzS would be replaced by a small fusion to amino acids derived from the mutagenesis vector pZErO-2. In the ‘in frame’frzS deletion mutants, the majority of the FrzS protein has been deleted. The 6xHis-tagged fusion proteins purified during this study and covering either the N- or the C-terminal domains of FrzS are shown.

B. SDS–PAGE analysis of purified FrzS(N) showing the 40 kDa monomer and the 80 kDa dimer in the presence or absence of the reducing agent DTT.

C. Western blot analysis of whole-cell extracts from DZ2 and the DZ2 frzS mutants (insertion and deletion), showing a single strong signal at ≈ 100 kDa in the wild-type strain and a truncated version of the protein (≈ 50 kDa) in the insertion mutant.

D. Output from the coiled-coil prediction program macstripe 2.0b1 showing the high probability of coiled-coil formation in the C-terminal domain of FrzS.

The 5′ end of the frzS gene, corresponding to the entire N-terminal portion of the FrzS protein, was PCR amplified and cloned into the expression vector pQE30 (Qiagen) to create the expression construct pQEfrzSN. The pQEfrzSN insert was sequenced to ensure that no errors were introduced during the amplification process and that the cloning site was maintained intact, thus ensuring the production of an N-terminally 6xHis-tagged FrzS(N) fusion protein (Fig. 5A). The construct was transformed into E. coli JM109, and overexpression of the FrzS(N) fusion protein was induced with 1 mM IPTG. The His-tagged FrzS(N) protein was then purified under native conditions using Ni-NTA superflow resin (Qiagen). The purified protein was analysed by SDS–PAGE and shown to run as a single band with an approximate molecular weight of 40 kDa (expected Mr 27 958) in the presence of the reducing agent dithiothreitol (DTT) (Fig. 5B). We presume that the reduced mobility of the FrzS(N) protein resulted from aberrant binding of SDS. In the absence of a reducing agent, the protein was shown to form dimers, presumably by spontaneous oxidation of cysteine residues. Purified protein (1 mg ml−1) was injected into rabbits for the production of polyclonal anti-FrzS antibodies (Cocalico Biologicals).

A Western blot analysis of whole-cell extracts was performed on DZ2, DZ2 frzS and DZ2 ΔfrzS strains to determine the specificity of the antibody. A single band, corresponding to a protein of approximately 100 kDa, was detected using the anti-FrzS antibody in the parent strain cell extract (Fig. 5C). This band was missing in both the DZ2 frzS mutants, suggesting that it corresponds to FrzS, even though it runs on SDS–PAGE gels at a significantly higher molecular weight than expected for the 60 kDa FrzS protein. Interestingly, the DZ2 frzS insertion mutant extract showed a faint new band at ≈ 55 kDa, which was not present in either the parent or the deletion mutant. This band may correspond to a truncated version of FrzS (N-terminal) expressed in these cells (see Fig. 5A). Further analysis showed that the parent strains (DZF1, DK1217 and DK1300) used in this study all expressed the 100 kDa protein that interacted with the anti-FrzS antibody. Additionally, the insertion mutants DZF1 frzS (DZF4220) and DK1217 frzS (DZ4221) both expressed the truncated 55 kDa form of this protein, whereas the deletion mutants DZF1 ΔfrzS (DZF4345), DK1217 ΔfrzS (DZ4337) and DK1300 ΔfrzS (DZ4338) expressed neither protein (not shown).

The FrzS protein is predicted to have an extensive C-terminal coiled-coil domain

A gapped blast search using the C-terminal domain of FrzS indicated that this domain has similarity to the heavy chain of the eukaryotic protein myosin (20% identity and 43% similarity), whereas a secondary structure prediction suggested it to be composed entirely of α-helices. As myosin heavy chain contains a long α-helical, coiled-coil region, we analysed the C-terminal domain of FrzS for the possibility that it might also form a coiled-coil and that this structure might be the basis for the similarity to myosin. Our preliminary analysis showed that the C-terminal domain of FrzS is composed of up to 38 heptad repeats (a b c d e f g)38, where positions a and d were most likely to be occupied by hydrophobic residues, particularly leucine (39% at both sites) or alanine (26% in position a and 24% in position d), whereas positions e and g were most likely to be occupied by charged residues, either basic or acidic in nature (position e: 18% Arg, 15% Asp and 15% Glu; position g: 42% Glu, 15% Arg). This periodic disposition of specific residues is highly indicative that the domain could fold into an amphipathic α-helix as a coiled-coil structure (Cohen and Parry, 1990). The predicted length of this coiled-coil would be approximately 40 nm, as the rod regions of known coiled-coil proteins extend approximately 15 nm per 100 amino acid residues (Bourne, 1991). Therefore, we analysed FrzS using the coiled-coil prediction program macstripe 2.0b1 [created by Alex Knight, York University, based on the coils algorithm of Andrei Lupas (Lupas et al., 1991)]. The output prediction (Fig. 5D) suggests with an extremely high probability that almost the entire C-terminal domain of FrzS is folded into a coiled-coil structure. This analysis also predicts that a single break may be present in the coiled-coil, around amino acids 318–322, and that the terminal 14 amino acids (or more) are unlikely to maintain the coiled-coil. As it is possible to predict the oligomerization states of coiled-coil proteins (Woolfson and Alber, 1995), the program multicoil (Wolf et al., 1997) was used to analyse the oligomerization state of FrzS. Output from this program suggests that the FrzS protein may form a homodimer using the rod-like coiled-coil domain (not shown).

Overexpression of the coiled-coil domain of FrzS in E. coli results in the formation of a striated protein lattice

The coiled-coil encoding region of the frzS gene was amplified by PCR and cloned into the expression vector pQE30 to create the construct pQEfrzSC. The insert and cloning site were sequenced to ensure that the correct fusion protein would be produced. Expression of the 6xHis-tag FrzS(C) fusion protein (Fig. 5A) was then induced in E. coli M15 with 1 mM IPTG. The E. coli M15 cells also contained the plasmid pREP4 to prevent expression of the fusion protein before induction (Qiagen). The cells were harvested after 4 h, and the overexpressed protein was shown to be insoluble by SDS–PAGE analysis. As the TlpA protein of S. typhimurium has previously been shown to encode a protein with extensive coiled-coil domains, which formed lamellae when overexpressed in E. coli (Hurme et al., 1994), we sectioned E. coli cells containing either pQE30 or pQEfrzSC after 4 h induction with IPTG. Examination of these sections by electron microscopy showed that the proposed coiled-coil domain of FrzS also forms a striated lattice structure when overexpressed in E. coli(Fig. 6A). This structure was not present in induced cells containing only the expression vector pQE30 (Fig. 6B). Measurement of the repeating pattern formed by the C-terminal domain of FrzS showed it to have a periodicity of 44 nm (Fig. 6C), which is close to the length predicted for the FrzS coiled-coil.

Figure 6.

Electron microscopic analysis of sections of E. coli cells containing (A) pQEfrzSC or (B) pQE30, after 4 h induction with 1 mM IPTG. Photographs were taken at × 16 000 magnification.

C. Further magnification of the striated pattern produced by overexpression of FrzS(C), demonstrating that the pattern repeats every 44 nm.

The C-terminal domain of FrzS forms a homodimeric coiled-coil

The His-tagged FrzS(C) fusion protein was purified under denaturing conditions using Ni-NTA resin (Qiagen), then refolded by dialysis against several changes of dH2O (pH 4.0). The protein was maintained at pH 4.0, as it was shown to precipitate into a gel under more alkaline conditions. Samples of the protein were analysed on SDS–PAGE gels, and the 35 kDa fusion protein was shown to run as a single band at approximately 50 kDa (in the presence or absence of DTT). Using CD spectra analysis, refolded FrzS(C) showed the characteristic twin minima at 222 nm ([θ] = −4.3 × 104 deg cm2 dmol−1) and 208 nm of α-helical proteins, which were not present in a denatured sample of the protein (not shown). To determine the oligomerization state of the protein, 1 µg of refolded FrzS(C) was cross-linked using 1 µl of glutaraldehyde (25% solution) in a 20 µl final volume. After 10 min at room temperature, 20 µl of loading buffer was added, the sample was boiled for 5 min, and a 10 µl volume was run on an 8% denaturing polyacrylamide gel. After cross-linking, the FrzS(C) protein appeared on the stained gel as two bands, the monomer at ≈ 50 kDa and a dimer at ≈ 100 kDa.


Motility and the regulation of directed motility responses are complex processes that are required for the completion of the different stages of the M. xanthus multicellular life cycle (see Hartzell and Youderian, 1995). Cells use two genetically distinct mechanisms for motility, the A- and S-motility systems (Spormann, 1999). Self-generated vegetative spreading or developmental aggregation signals are presumed to be relayed via signal transduction pathways, such as the Frz pathway, to these motility systems to facilitate appropriate directed motility responses (Ward and Zusman, 1997). Currently, many important components of both the motility and the directed motility systems remain unknown (MacNeil et al., 1994a,b). In this study, we have identified a new component of the S-motility system by cloning and sequencing the frzS gene, which is located approximately 6 kb upstream of the frz signal transduction locus at the end of the rpoE1 operon.

FrzS was identified as a potential component of one of the gliding motility systems because of the reduced vegetative spreading phenotypes of the DZ2 frzS mutants. This reduction in surface translocation was particularly apparent on soft agar, which has previously been shown to support S-motility-associated movements (Shi and Zusman, 1993). In the FB strain (DZF1) background, the frzS insertion mutation had only a minor effect on spreading. However, FB mutants have been shown to have a leaky S-motility defect (pilQ1;Wall et al., 1999), and this mutation could partially mask a secondary S-motility defect. The possibility that frzS could be an S-motility gene was confirmed by deleting the frzS gene in an AS+ motility background. The AS+ΔfrzS double mutant was non-spreading on 1.5% agar (AS), proving that frzS is an S-motility gene. However, it should be noted that the non-spreading phenotype of the double mutant was not associated with a complete loss of motility. Microscopic observations, made at the colony edges of these swarms, showed occasional short and jerky movements of a few individual cells (data not shown), suggesting that a low level of motility is retained in these cells. Normal A motility in the A+SΔfrzS double mutant showed that frzS is not also an A-motility gene.

Many S-motility mutants have been analysed previously, and various mutants have been grouped together depending on specific phenotypic properties and/or the production of known cell surface components. Therefore, the frzS mutants were analysed with respect to these known groups of mutants. Electron microscopic observation of the DZ2 frzS insertion mutant showed that the cells produce almost normal levels of polar type IV pili, thereby indicating that FrzS is not required for pilus biogenesis or export. Other S-motility mutants that produce pili (dsp mutants or mutants in the sasA locus) show a distinctive non-spreading phenotype when introduced into the AS+ mutant background (Shimkets, 1986; Bowden and Kaplan, 1998). Although these AS double mutants are classified as non-spreading, a small fringe of cells surrounds the colony edge on 1.5% agar, suggesting that they are not completely non-motile under these conditions. While the AfrzS mutants were also not completely non-motile, no fringe of cells was seen surrounding the colonies on 1.5% CYE agar. In addition, the frzS mutant was shown to express normal levels of fibrils, which are absent in dsp cells (Arnold and Shimkets, 1988), and O-antigen, which is missing in cells with mutations in the sasA locus (Bowden and Kaplan, 1998). The presence of the cell surface-associated pili, fibrils and O-antigen on frzS mutants suggests that the FrzS protein could either be a new surface component required for S motility or it could interact, perhaps in a regulatory manner, with one of the known components.

The PilT protein has been proposed to play a role in the regulation of type IV pilus function, as pilT mutants lack S motility, but have structurally normal pili (Wu et al., 1997). The similarities identified in this study between frzS and pilT mutants suggest that FrzS might also play a role in type IV pilus function. Both mutant strains lack S motility but express normal levels of type IV pili, fibrils and O-antigen. Both mutants also show similar clumping profiles in a cohesion assay. However, whereas the PilT protein is conserved in other organisms that move by twitching motility, e.g. P. aeruginosa and Neisseria gonorrhoeae, we could not find a gene similar to frzS in the genomes of either species, suggesting that FrzS is not an integral part of the type IV pilus.

Type IV pili have been suggested to play two potential motility-associated roles in M. xanthus (see Wall and Kaiser, 1999). First, the pilus could play a mechanical role in force generation. In this model, the pili would be extruded from the cell, attach to a site on another cell or the substratum, then retract, thereby pulling the cells along. Mutants that produce pili but are unable to move by S motility would be considered to be ‘paralyzed’ in this model. The PilT protein has been proposed to be specifically involved in pilus retraction (Wall and Kaiser, 1999), and it seems possible that the FrzS protein might also be involved in this retraction process. The alternative model proposes that the pili might act as a motility signal transduction apparatus. In this case, the pili would be extruded from the cell as described previously. However, when the pili contact another cell, a signal would be passed between the cells. This signal would subsequently be transduced to the actual motility apparatus and stimulate S motility from a currently unidentified motor. In this case, piliated cells that do not display S motility might have a signal reception defect. Could FrzS play a role in either the reception or the transduction of such a signal? In future studies, we would be interested in exploring this question further. Preliminary studies, using the anti-FrzS antibody to determine the location of FrzS in the cell by immunofluorescence, have suggested FrzS to be localized to the cell poles (data not shown), adding weight to our proposal that FrzS might be involved with type IV pilus function. In addition to continuing these localization studies, we hope to be able to analyse retraction of the type IV pili in both frzS and pilT mutants.

The developmental phenotypes of the frzS mutants generated in this study can perhaps be rationalized by considering the structure and possible function of the FrzS protein. Deletion of the frzS gene in the fully motile DZ2 background, or in the reduced motility FB background, appeared to have little effect on either developmental aggregation or sporulation. However, insertion mutants in the same backgrounds resulted in either loss of aggregation and a severe sporulation defect in DZ2, or the ‘frizzy’ phenotype in DZF1. One key difference between the insertion and deletion mutants was made apparent after Western blot analysis of whole-cell extracts using the anti-FrzS antibody. Whereas the FrzS protein was absent in the deletion mutants, all the insertion mutants were shown to produce a truncated form of the protein. This truncated form of FrzS would include the entire N-terminal domain of the protein (see Fig. 5A), fused to a short region of non-FrzS amino acids (from the plasmid pZOrf6 that was used for mutagenesis). The difference in the developmental phenotypes of the frzS insertion and deletion mutants suggests that the truncated form of FrzS retains some activity. But what effect could this truncated form of FrzS have on development? The first possibility is that the truncated form of FrzS might partially restore S motility and that this partial restoration of S motility somehow interrupts development. This appears unlikely, as the DK1217 frzS insertion mutant showed the same non-spreading (AS) phenotype as the DK1217 ΔfrzS mutant. A second possibility is that the truncated form of FrzS, which has homology to response regulator proteins and contains a potential phosphorylation site (Asp-55), could be interfering with a developmental signal transduction pathway. Similar interruptions, which attenuate communication between a sensor histidine protein kinase and its cognate response regulator, have been termed quenching (Parkinson, 1995), as signal flow is halted when the receiver domain is unable to exert control over output domain activity. The ‘frizzy’ phenotype of the DFZ1 frzS insertion mutant suggests that it might be the flow of phosphate through the Frz signal transduction system that is being quenched. During this study, we attempted to address this possibility by labelling frzS mutants with a tetrazolium dye and filming cell behaviour during early development (as directed motility behaviour was unaffected in isolated cells moving on nutrient-rich agar). This technique has been used previously to track single cell movements within groups of cells resulting from the presence of dye crystals within a proportion of the labelled cell population (Shi et al., 1996). However, using the previously determined labelling conditions, we found many spores within the labelled population of cells. As the labelling technique resulted in development, we considered this particular analysis unsuitable for our studies. In the future, we hope to identify a more suitable method for tracking single cells within populations in order to explore the possible link between the Frz signal transduction pathway and S motility further.

Perhaps the most interesting aspect of the FrzS protein is the presence of an extensive C-terminal coiled-coil domain. Long coiled-coil domains are unusual components of bacterial proteins, although they are common structural elements in eukaryotic cells. In our preliminary analysis of the coiled-coil domain of FrzS, we have shown that purified, refolded His-tagged FrzS(C) is α-helical. We have also analysed the oligomerization state of FrzS(C) using cross-linking studies and shown that the domain oligomerizes to form a homodimer. Although the C-terminal domain of FrzS forms an extensive homodimeric coiled-coil, a shorter version of this domain would presumably act similarly, suggesting that the coiled-coil probably plays a role in addition to acting as a dimerization interface. Still, dimerization of the C-terminal domain of FrzS could play an important role in FrzS function. Phosphorylation-induced dimerization of the receiver domain of the response regulator protein FixJ has been shown to be required for activation of the protein (Da Re et al., 1999), and the coiled-coil domain of FrzS might influence the oligomerization state of the N-terminal response regulator domain. Conversely, the phosphorylation state of the receiver domain could affect the oligomerization state of the coiled-coil, as is seen in the eukaryotic lamin proteins during mitosis (McKeon et al., 1986). Receptor dimerization by coiled-coil interaction has also been suggested to play an important role in signalling and adaptation during bacterial chemotaxis (Surette and Stock, 1996). Alternative roles for the domain are also possible. The coiled-coil of TlpA from S. typhimurium has been suggested to act as a thermometer for temperature-dependent regulation of transcription (Hurme et al., 1997) by falling apart and reassembling in a precisely regulated manner. Another extensive bacterial coiled-coil protein, Ompα from Thermatoga maritima, has been suggested to act simply as a stiff structural periplasmic spacer, which specifies the distance between the outer membrane and the inner cell body (Engel et al., 1992).

The ability of the His-tagged FrzS(C) fusion protein to form a striated lattice, resembling those formed by intermediate filament proteins, when overexpressed in E. coli is also unusual for a bacterial protein. The periodicity of the striations, which reoccur every 44 nm, are very similar to the predicted length of the coiled-coil domain (40 nm), and we presume that the lattice may be formed by end-on-end and side-by-side alignments of the homodimeric coils. However, we currently have no evidence that the protein forms this type of extensive network in M. xanthus, although other homodimeric coiled-coils, for example the intermediate filament proteins, are known to form higher order oligomeric configurations.

During this study, we have identified a new S-motility gene, frzS, and performed a preliminary characterization of the FrzS protein. Our analysis of frzS mutants leads us to believe that the homodimeric FrzS protein could be associated with type IV pilus function. However, the difference in the developmental aggregation phenotypes of the insertion and deletion mutants, in conjunction with the similarity of the N-terminal domain of FrzS to other response regulator proteins, indicates that FrzS is most probably a component of a signal transduction pathway that could either regulate S motility or transduce S-motility-associated signals. Currently, a significant role for the extensive coiled-coil domain of FrzS in S motility remains to be elucidated. Future studies will be directed to determining a more precise role for the FrzS protein in the S-motility process.

Experimental procedures

Strains and culture conditions

The strains used in this study are listed in Table 1. Plasmids are listed in Table 2. M. xanthus strains were grown in CYE medium (Campos et al., 1978), and developmental assays were performed on CF fruiting media (Hagen et al., 1978). E. coli strains were grown in LB media.

Table 1. Strains.
StrainGenotype or relevant propertiesSource/reference
M. xanthus strains
 DZ2Wild type Campos and Zusman (1975)
 DZF1 pilQ1 Wall et al. (1999)
 DZF3377 pilQ1, frzE::Tn5Ω226 Shi et al. (1993)
 DZ4148 frzE::Tn5W226 Shi et al. (1993)
 DZ4219 frzS::pZOrf6This study
 DZF4220 pilQ1, frzS::pZOrf6This study
 DZ4335ΔfrzSThis study
 DZF4345 pilQ1, ΔfrzSThis study
 DK1217 aglB1 Hodgkin and Kaiser (1979b)
 DZ4221 aglB1, frzS::pZOrf6This study
 DZ4337 aglB1, ΔfrzSThis study
 DK1300 sglG1 Hodgkin and Kaiser (1979b)
 DZ4338 sglG1, ΔfrzSThis study
 SW505 difA::Tn5kan90Ω101 Yang et al. (1998)
E. coli strains
 XL1-BlueGeneral cloning strainStratagene
 Top10Cloning strain for pZErO-2, does not contain lacIq gene,
allowing constitutive expression of ccdB
 JM109Contains lacIq for regulated expression from IPTG-inducible promoters Yanisch-Perron et al. (1985)
 M15 (pREP4)Regulated overexpression from IPTG inducible promoters resulting from pREP4Qiagen
Table 2. Plasmids.
PlasmidRelevant characteristicsSource/reference
pUC18General cloning vector Yanisch-Perron et al. (1985)
pZErO-2Zero background cloning vector, KmRInvitrogen
pZOrf6Internal fragment of frzS (orf6) cloned in pZErO-2, KmRThis study
pP0.50.5 kb PstI fragment of frzS cloned in pUC18This study
pP1.21.2 kb PstI fragment of frzS cloned in pUC18This study
pYSl2.02.0 kb SalI fragment of frzS and orf7 cloned in pUC18This study
pKY480Vector for construction of deletion mutations by
counterselection against sucrose resistance, KmR
Cho and Zusman (1999)
pKYS1S2PCR fragments of frzS ligated together to give a
1.1 kb deletion of the gene, cloned in pKY480, KmR
This study
pQE30His-tag expression construct, AmpRQiagen
pREP4Contains lacIq to prevent early expression of genes
cloned in pQE30, KmR
pQEfrzSNN-terminal domain of frzS cloned in pQE30This study
pQEfrzSCC-terminal domain of frzS cloned in pQE30This study

DNA manipulations and PCR

All plasmids used in this study were prepared using the QIAprep spin miniprep kit (Qiagen). Chromosomal DNA was also prepared using a miniprep protocol utilizing cetyltrimethylammonium bromide-NaCl and phenol–chloroform extraction as detailed by Ward et al. (1998a). Restriction enzyme digests were performed as specified by the manufacturers. PCR optimizations and cycling parameters were identified using the protocol of Kramer and Coen (1995). In general, glycerol concentrations of 20% were required for high yields of pure products. Taq polymerase (Promega) and Pfu polymerase (Stratagene) were used in amplifications requiring low and high fidelity respectively.

Cloning and sequencing

A cosmid containing the entire frz operon, the rpoE1 operon and the intervening sequences was identified from a cosmid library (kindly provided by Ron Gill, University of Colorado) using a fragment of the previously sequenced orf6 (frzS) gene (Ward et al., 1998a) as a probe. The probe was constructed by PCR, incorporating the hapten digoxigenin as digoxigenin-11-dUTP (Boehringer Mannheim), and detection was performed by enzyme immunoassay and an enzyme-catalysed colour reaction. A single isolate from the library (tray 3, position B5) was shown to hybridize with this probe. A 0.5 kb PstI fragment, which overlaps the known sequence, was subcloned from the cosmid into pUC18 as pP0.5. The downstream 1.2 kb PstI fragment was also cloned into pUC18 as pP1.2. The inserts within both pP0.5 and pP1.2 were sequenced on both stands at the University of California Davis sequencing facility. To confirm the junction between the two PstI fragments, a 2.0 kb SalI fragment was cloned into pUC18 as pYSl2.0 and used for sequencing of this region.

Construction of insertion mutants

Insertion mutants in the frzS gene were constructed by cloning an internal fragment of the gene into the vector pZErO-2 (Invitrogen), which cannot replicate in M. xanthus. The internal fragment of frzS was prepared using PCR with the following primer pair. The forward primer, 5′-ATGAATTCGACGAGTACGTCGCCAAGCC, is positioned approximately 300 bp downstream of the ATG translational start for frzS. The reverse primer, 5′-ATGAATTCTGCAGTTCGGCCACC TTCGC, is positioned approximately 860 bp downstream of this point, thereby producing in combination a 560 bp fragment for cloning. Cloning was performed using the EcoRI sites engineered onto the primers (shown underlined) to allow direct ligation into the EcoRI site of pZErO-2. The resultant construct, pZOrf6, was transformed into E. coli Top10 cells, which allow Zero Background cloning (Invitrogen). Purified pZOrf6 was then introduced into M. xanthus by electroporation (Ward et al., 1998b), and cells that had integrated the plasmid into the chromosome were selected by growth on kanamycin (25 µg ml−1). Chromosomal DNA was prepared from the mutant strains, and the insertion site of the plasmid was checked by Southern blotting. Probes were constructed by PCR incorporating the hapten digoxigenin, and detection was performed as stated above.

Construction of deletion mutants

The following primers were used in the amplification of two fragments, an upstream fragment covering the orf5frzS genes and a downstream fragment within the 3′ end of the frzS gene. The downstream fragment was designed within the frzS gene because of difficulties encountered with obtaining PCR amplification of DNA over the strong frzS terminator structure. Upstream fragment (external end): 5′-C CGGTCGACCGTTGGTGGTGGTGACGGTGG-3′; (internal end): 5′-CCGGTACCGGAGAGCGCGGTGTCGCTTTC-3′. Downstream fragment (internal end): 5′-CCGGTACCGCCGAGTCCGAGCTCGCCCAG-3′; (external end): 5′-CCG GCTAGCGCGGCTTCGCTGGCGTCGTCC-3′. The XhoI, KpnI and NheI restriction sites used for ligating the two fragments together (to create a 1.1 kb ‘in frame’ deletion of frzS) and for ligating this fragment into the vector pKY480 are shown underlined. The resultant plasmid pKYS1S2 was electroporated into M. xanthus strains and selected for as described in the text. As the two PCR fragments used for the generation of pKYS1S2 were ligated together using a KpnI linker, the resultant FrzS deletion protein would contain two non-FrzS amino acids (GT) at the ligation site. All mutants constructed using this technique were analysed by Southern blotting for deletion of the frzS gene and by Western blotting to show loss of the FrzS protein.

Phenotypic analyses

Developmental defects were screened for on CF starvation agar by plating 5 µl of cells at 2 × 109 cells ml−1 directly onto the plate and incubating for 96 h. Spore counts were performed on these aggregates after 4 days of incubation. Cells were removed from the agar and resuspended in TM buffer (10 mM Tris-HCl, 8 mM MgSO4, pH 7.6). Spore clumps were dispersed by sonication, and appropriate dilutions were placed in a Petroff–Hausser counting chamber for counting under magnification.

Vegetative spreading analyses (previously termed swarming) were performed on CYE agar containing either 1.5% agar or 0.3% agar. Cells were spotted directly onto the plates, and colony diameters were measured at 12 h intervals. Time-lapse motion analysis of cell movements was performed on cultures diluted to ≈ 107 cells ml−1, with 10 µl volumes spread over CYE agar thinly layered on microscope slides. Isolated individual cells were filmed for periods of 30 min. Fields of 10–30 cells were documented at a time, and analyses were performed on over 100 cells. Cells were observed with a Nikon Labphot-2 microscope with a 40 × objective. Images were recorded with a Dage-MTI CCD-72 series camera and a time-lapse video cassette recorder (120 h speed setting; model GYYR TLC 1800). Filming was started 10–15 min after inoculation and continued for up to 3 h. Longer periods of filming were avoided to minimize cell–cell interactions. Data were analysed manually by tracing the movement of cells during playback. Cells for video microscopic analyses within groups were grown to ≈ 1 × 109 cells ml−1, then left overnight unshaken on the bench in 0.002% 2,3,5-triphenyltetrazolium chloride (Sigma), until the cells turned very red (Shi et al., 1996).

Colony dot blots

Cells for colony blot analysis were grown on CYE agar plates to facilitate the production of extracellular fibrils. The cells were harvested and diluted to 100 Klett, then 5 µl volumes were spotted onto Immobilon-P membrane (Millipore). Cells were serially diluted 1:1, and dilutions were also spotted onto the membrane to facilitate an end-point dilution analysis. Blots were incubated overnight in TBS (20 mM Tris, 500 mM NaCl, pH 7.5) containing 5% skimmed milk, then incubated for 1 h in the appropriate primary antibody (diluted 1:500). Blots were washed three times using TTBS (TBS containing 0.05% Tween 20). A 1:5000 dilution of the secondary antibody (Protein A–G-CIP conjugate; Pierce) was added to the blots for 1 h, and the blots were again washed three times in TTBS, followed by once in TBS. Detection was performed using NBT–BCIP in an enzyme-catalysed colour reaction.

Electron microscopy

Cells were grown for type IV pilus analysis to approximately 5 × 108 cells ml−1, and an aliquot was washed gently in 10 mM Tris (pH 7.4). Cells were resuspended and spotted onto electron microscopy grids, then allowed to settle for 2 min. The grids were washed three times with 1% uranyl acetate and air dried. Both ends of 20 negatively stained cells were then observed, and the number of pili present counted. For analysis of E. coli cells overexpressing the C-terminal domain of FrzS, the cells were prepared by high-pressure freezing (McDonald, 1999), sectioned, stained and photographed by Kent McDonald at the University of California Berkeley electron microscopy facility.

Overexpression of the N-terminal and C-terminal domains of FrzS

The 5′ end of the frzS gene, corresponding to the entire N-terminal domain and part of the Ala–Pro linker region, was amplified by PCR and cloned into the expression vector pQE30 (Qiagen). Forward primer: 5′-CGGGTACCGGCCGT CCCTCCATGTCGAAGAAAATC-3′; reverse primer: 5′-CG AAGCTTAATGAAGCCGATCACCGT CTTCTCGTC-3′. The KpnI and HindIII restriction enzyme sites used for cloning are shown underlined. The insert was sequenced and a correct construct transformed into E. coli JM109 for overexpression. The overexpressed protein was purified under native conditions using Ni-NTA Superflow resin (Qiagen), concentrated and used for the production of a polyclonal anti-FrzS antibody.

The 3′ end of the frzS gene, corresponding to part of the Ala–Pro linker region and the entire C-terminal domain of the protein, was PCR amplified and cloned into pQE30. Forward primer: 5′-CGGGATCCCCCGCGGCCGACGCCGCGGAGC TGCGCAGCCTG-3′; reverse primer: 5′-CGGGTACCCGGG TGGCCCGGGGAGCGCGGCCCGTGCGACTA-3′. The insert was sequenced and a correct clone transformed into E. coli M15 (pREP4) for overexpression. The protein was purified under denaturing conditions using Ni-NTA Superflow resin (Qiagen), concentrated, then refolded by dialysis against dH2O (pH 4.0).

Protein analyses

Protein concentrations were determined using the BCA protein assay kit (Pierce). Purified proteins or whole-cell extracts (10 µg of total protein) were run on SDS–polyacrylamide gels prepared using ProtoGel components (National Diagnostics) to determine protein purity, oligomerization states or before Western transfer. Rainbow-coloured protein molecular weight markers (Amersham) were used on all gels. Protein samples were visualized by silver staining where appropriate. Western blotting was performed using a Bio-Rad semi-dry transfer cell using Bjerrum and Schefer-Nielson buffer (48 mM Tris base, 39 mM glycine, 0.0375% SDS, 20% methanol, pH 9.2) and Immobilon-P membranes (Millipore). Western detections were performed using the SuperSignal chemiluminescent kit (Pierce). The primary antibody (anti-FrzS) was diluted 1:500, and the secondary antibody (goat anti-rabbit horseradish peroxidase conjugate; Pierce) was diluted 1:5000 before use.

CD spectra were collected on an Aviv 62 DS circular dichroism spectrometer in a 1 cm pathlength cuvette. Recordings were taken every second at 1 nm intervals in the range 200–300 nm. The unit of CD measurement [θ] (deg cm2 dmol−1) was calculated using the following equation: [θ] = ψMW/(104lcn), where Ψ is the observable signal (millidegrees), l is the pathlength in cm, c is the concentration of the protein in g ml−1, n is the number of amino acids and MW is the molecular weight in g mol−1.

Computer analyses

DNA and protein sequence analyses were performed using the program dnastar.


We would like to thank Ron Gill for his cosmid library, Heidi Kaplan for monoclonal antibodies, Kyungyun Cho for the vector pKY480, Srebrenka Robic for CD spectra analysis, Tsz Lau for technical assistance in the construction of pQEfrzSC, and Tom Alber for advice on the handling of a coiled-coil protein. We would particularly like to thank Kent McDonald at the University of California Berkeley electron microscopy facility for his interest and assistance in this project. We would also like to thank the University of California Davis sequencing facility. Research in our laboratory was supported by Public Health Service grant GM20509 from the National Institutes of Health.