Essential role of DivIVA in polar growth and morphogenesis in Streptomyces coelicolor A3(2)

Authors


E-mail klas.flardh@icm.uu.se; Tel.(+46) 18 4714058; Fax (+46) 18 530396.

Summary

Streptomycetes grow by cell wall extension at hyphal tips. The molecular basis for such polar growth in prokaryotes is largely unknown. It is reported here that DivIVASC, the Streptomyces coelicolor homologue of the Bacillus subtilis protein DivIVA, is essential and directly involved in hyphal tip growth and morphogenesis. A DivIVASC-EGFP hybrid was distinctively localized to hyphal tips and lateral branches. Reduction of divIVASC expression to about 10% of the normal level produced a phenotype strikingly similar to that of many tip growth mutants in fungi, including irregular curly hyphae and apical branching. Overexpression of the gene dramatically perturbed determination of cell shape at the growing tips. Furthermore, staining of nascent peptidoglycan with a fluorescent vancomycin conjugate revealed that induction of overexpression in normal hyphae disturbed tip growth, and gave rise to several new sites of cell wall assembly, effectively causing hyperbranching. The results show that DivIVASC is a novel bacterial morphogene, and it is localized at or very close to the apical sites of peptidoglycan assembly in Streptomyces hyphae.

Introduction

Simple rod-shaped bacteria such as Escherichia coli elongate through isotropic insertion of new cell wall material throughout the length of lateral walls (de Pedro et al., 1997). This is periodically interrupted when new cell poles are created by division at the cell midpoint. Once formed, the cell poles appear to be inert and stable (de Pedro et al., 1997). This is in sharp contrast to the situation in Streptomyces, which differs from most bacteria by growing as branching hyphae in a manner highly analogous to that of filamentous fungi: growth is extremely asymmetrical, and cell wall extension occurs by de novo incorporation of peptidoglycan precursors at the tips of hyphae (Braña et al., 1982; Gray et al., 1990; Prosser and Tough, 1991; Miguélez et al., 1992). On the other hand, DNA is replicated in a much greater portion of the hyphal length (Kummer and Kretschmer, 1986; Prosser and Tough, 1991). Cell division leads to formation of a new hyphal crosswall, often around the middle of the long apical cell (Kretschmer, 1982; 1989), but cell poles created by this division are not the sites for cell elongation. A new lateral branch has to be created in the subapical daughter cell before it can enlarge its dimensions again. Such pronounced apical growth depends on cell polarization, targeting of cell wall synthesis systems to the growing cell pole, and some means of moving DNA and other constituents towards the growing tip and into new branches. Filamentous fungi face similar cell biological problems of polarizing the cell, transporting material, and moving nuclei towards the growing pole. Genetic studies in fungi have identified genes involved in these processes, and shown that microtubules, F-actin and a number of other proteins (including kinesin and dynein) are critical for tip growth and nuclear migration (Fischer, 1999; Xiang and Morris, 1999; Momany, 2002). There are also clear parallels to polar growth in other eukaryotes, and homologues of several genes identified in filamentous fungi are conserved and have important functions also in other cell types (Morris, 2000). In comparison to the situation in fungi, the mechanism underlying apical growth in Streptomyces has received surprisingly little attention, despite the fact that it could shed light on important features of the prokaryotic cell.

This paper describes the first protein to be specifically localized at the tips of growing hyphae of the model organism Streptomyces coelicolor A3(2). This protein, which is a homologue of Bacillus subtilis DivIVA, was found to have profound impact on tip growth. In B. subtilis, two functions for DivIVA have been recognized. In vegetatively growing cells, it is localized at sites of ongoing cell division and at cell poles, where it appears to sequester the cell division inhibitors MinC and MinD, thereby allowing division only at central positions (Cha and Stewart, 1997; Edwards and Errington, 1997; Marston et al., 1998; Marston and Errington, 1999; Edwards et al., 2000). In the absence of divIVA, the uncontrolled activity of the MinCD division inhibitor prevents most septation. Thus, divIVA mutants are defective in cell division, whereas double minC divIVA or minD divIVA mutants grow and have the same minicell phenotype as minC or minD single mutants, with no overt impact of divIVA (Cha and Stewart, 1997; Edwards and Errington, 1997). In addition, overexpression of B. subtilis divIVA interfered with cell division in a way dependent on an active MinCD division inhibitor (Cha and Stewart, 1997). Thus, B. subtilis DivIVA is a functional analogue of the MinE protein in E. coli (Margolin, 2001; Rothfield et al., 2001). A second function of DivIVA in B. subtilis was recently discovered. DivIVA mutants produce prespore compartments devoid of DNA with a high frequency, indicating that DivIVA has a role in localizing or attaching the oriC region of the chromosome to the cell pole during sporulation (Errington, 2001; Thomaides et al., 2001). Proteins related to DivIVA are encoded by all Gram-positive bacterial genomes sequenced so far, with the possible exception of the mycoplasmas which have divIVA-like genes of more limited similarity. One member of the DivIVA family is antigen 84 (Ag84) from Mycobacterium tuberculosis, but the function of this protein has not been studied (Hermans et al., 1995). In this paper, evidence is presented that Streptomyces DivIVA is not directly involved in cell division or nucleoid localization in vegetatively growing hyphae. Instead, the results demonstrate a crucial and hitherto unknown role in hyphal polar growth and cell shape determination.

Results

The S. coelicolor homologue of divIVA is an essential gene

In the S. coelicolor A3(2) genome sequence (Bentley et al., 2002), the homologue of B. subtilis DivIVA is encoded by the gene SCO2077, which hereafter is referred to as divIVASC. As pointed out by Edwards et al. (2000), the DivIVA homologues share sequence similarity in the N-terminal part, and this is followed by sequences with a lower degree of conservation, but sharing a high likelihood of adopting coiled-coil conformations. Among the actinomycete members of DivIVA-like proteins present in the databases, the N-terminal region is highly conserved, and they contain one additional well conserved segment around the end of the region with a high likelihood of forming a coiled-coil structure (positions 280–302 in divIVASC; Fig. 1A). However, in all actinomycete DivIVAs the predicted coiled coil region is interrupted by a highly variable stretch of sequence. In DivIVASC, the corresponding region differs markedly from that in the other proteins, is of relatively low complexity, and consists of 30% glycine, 25% proline and 18% glutamine (residues 69–201; underlined in Fig. 1A).

Figure 1.

Figure 1.

A. Alignment of selected homologues of DivIVA from Gram-positive bacteria. The alignment was generated using t-coffee, and edited using genedoc software. The top sequence is B. subtilis DivIVA (TrEMBL acc. no. P71021). The bottom three are examples from the actinomycete branch of Gram-positives: Antigen 84 from Mycobacterium tuberculosis strain H37Rv (P71021), Thermobifida fusca (ZP_00057769), and S. coelicolor (Q9S2X4). Filled bullets indicate regions in the B. subtilis (above the sequences) and S. coelicolor DivIVA (below the sequences) that have a likelihood of> 0.5 to adopt a coiled-coil structure according to coils version 2.1 (Lupas et al., 1991), using the MTIDK matrix, no weights, and a 28 aa window. Open bullets indicate the first and fourth positions in the coiled-coil heptad motifs. Asterisks are inserted at positions where the phase of the heptad repeats may be shifting.
B. Map of the ftsZ-divIVASC region on the S. coelicolor genome (Bentley et al., 2002). The ftsZ gene is followed by four open reading frames (SCO2081; SCO2080; SCO2079; SCO2078) of unknown function, divIVASC (SCO2077), and a probable isoleucyl-tRNA synthetase (SCO2076). Only the 3′-end of the latter gene is shown. Gene orientations are indicated by thin arrows. The inserts in relevant plasmids are shown, although only a part of the large insert in cosmid pKF56 is drawn. The cross-hatched box symbolises the ΩaacC4 cassette, and the vertically striped boxes denote egfp. The dotted lines indicate vector DNA. The bent arrow symbolises the tipAp promoter.
C. Schematic drawing of the integration and excision of cosmid pKF56 that gave rise to the allelic exchange at the divIVASC locus.

Figure 1.

Figure 1.

A. Alignment of selected homologues of DivIVA from Gram-positive bacteria. The alignment was generated using t-coffee, and edited using genedoc software. The top sequence is B. subtilis DivIVA (TrEMBL acc. no. P71021). The bottom three are examples from the actinomycete branch of Gram-positives: Antigen 84 from Mycobacterium tuberculosis strain H37Rv (P71021), Thermobifida fusca (ZP_00057769), and S. coelicolor (Q9S2X4). Filled bullets indicate regions in the B. subtilis (above the sequences) and S. coelicolor DivIVA (below the sequences) that have a likelihood of> 0.5 to adopt a coiled-coil structure according to coils version 2.1 (Lupas et al., 1991), using the MTIDK matrix, no weights, and a 28 aa window. Open bullets indicate the first and fourth positions in the coiled-coil heptad motifs. Asterisks are inserted at positions where the phase of the heptad repeats may be shifting.
B. Map of the ftsZ-divIVASC region on the S. coelicolor genome (Bentley et al., 2002). The ftsZ gene is followed by four open reading frames (SCO2081; SCO2080; SCO2079; SCO2078) of unknown function, divIVASC (SCO2077), and a probable isoleucyl-tRNA synthetase (SCO2076). Only the 3′-end of the latter gene is shown. Gene orientations are indicated by thin arrows. The inserts in relevant plasmids are shown, although only a part of the large insert in cosmid pKF56 is drawn. The cross-hatched box symbolises the ΩaacC4 cassette, and the vertically striped boxes denote egfp. The dotted lines indicate vector DNA. The bent arrow symbolises the tipAp promoter.
C. Schematic drawing of the integration and excision of cosmid pKF56 that gave rise to the allelic exchange at the divIVASC locus.

Figure 1.

Figure 1.

A. Alignment of selected homologues of DivIVA from Gram-positive bacteria. The alignment was generated using t-coffee, and edited using genedoc software. The top sequence is B. subtilis DivIVA (TrEMBL acc. no. P71021). The bottom three are examples from the actinomycete branch of Gram-positives: Antigen 84 from Mycobacterium tuberculosis strain H37Rv (P71021), Thermobifida fusca (ZP_00057769), and S. coelicolor (Q9S2X4). Filled bullets indicate regions in the B. subtilis (above the sequences) and S. coelicolor DivIVA (below the sequences) that have a likelihood of> 0.5 to adopt a coiled-coil structure according to coils version 2.1 (Lupas et al., 1991), using the MTIDK matrix, no weights, and a 28 aa window. Open bullets indicate the first and fourth positions in the coiled-coil heptad motifs. Asterisks are inserted at positions where the phase of the heptad repeats may be shifting.
B. Map of the ftsZ-divIVASC region on the S. coelicolor genome (Bentley et al., 2002). The ftsZ gene is followed by four open reading frames (SCO2081; SCO2080; SCO2079; SCO2078) of unknown function, divIVASC (SCO2077), and a probable isoleucyl-tRNA synthetase (SCO2076). Only the 3′-end of the latter gene is shown. Gene orientations are indicated by thin arrows. The inserts in relevant plasmids are shown, although only a part of the large insert in cosmid pKF56 is drawn. The cross-hatched box symbolises the ΩaacC4 cassette, and the vertically striped boxes denote egfp. The dotted lines indicate vector DNA. The bent arrow symbolises the tipAp promoter.
C. Schematic drawing of the integration and excision of cosmid pKF56 that gave rise to the allelic exchange at the divIVASC locus.

As in other Gram-positives (Cha and Stewart, 1997; Massidda et al., 1998), S. coelicolor divIVASC is located downstream from ftsZ (Fig. 1B). DivIVASC is immediately followed by a stem–loop-forming sequence that may be a transcriptional terminator. The next gene is transcribed in the opposite direction, and encodes a putative tRNA synthetase (SCO2076). In order to examine the function of the S. coelicolor divIVA homologue, a knockout mutation (ΔdivIVA::ΩaacC4) conferring apramycin resistance (ApraR) was exchanged for divIVA+ on a cosmid carrying this region of the chromosome and vector-encoded resistance to kanamycin (KmR) and ampicillin (AmpR). The resulting cosmid pKF56 was integrated into the chromosome of S. coelicolor A3(2) strain M145 by homologous recombination (Fig. 1C). The co-integrant strains, which were ApraR KmR and partial diploids for divIVA+/ΔdivIVA::ΩaacC4, were relatively unstable, and the cosmid was lost with high frequency after sporulation in the absence of selection. Haploid strains that had lost the integrated cosmid in such a way that only the ΔdivIVA::ΩaacC4 allele remained on the chromosome should be ApraR KmS (Fig. 1C). Of 152 tested KmS derivatives, all except one were ApraS. When only ApraR spores were selected from the co-integrants, 175 of 178 tested strains also retained KmR. The four ApraR KmS strains that were obtained from these experiments were shown by PCR to contain both the divIVASC+ and ΔdivIVA::ΩaacC4 alleles, and had probably lost only parts of the cosmid, including the KmR marker. These results indicated that either allelic exchange could not easily occur, or only divIVASC+ strains were able to grow. By recovery and analysis of excised cosmids from a pKF56 co-integrant, it was found that two of the 10 analysed cosmids had picked up the chromosomal divIVASC+ allele, showing that allelic exchange between chromosome and the pKF56 cosmid could occur. This strongly suggested that divIVASC is an essential gene.

To further test the requirement of divIVASC for viability, pKF58, a single-copy integrative plasmid containing divIVASC under control of the tipAp promoter, was introduced into cosmid co-integrant strains. Analysis of the resulting strains, and controls with the empty vector pPM927, showed that the native chromosomal allele of divIVASC could be deleted, but only if the gene product was provided in trans (Table 1). All tested ApraR KmS strains containing pKF58 were found to have lost the native divIVA+ allele and retained only ΔdivIVA::ΩaacC4 at the normal locus. The ΔdivIVASC mutants were obtained even in the absence of the tipAp inducer thiostrepton. When the experiment was performed with a low level of thiostrepton (1 µg ml−1) in all plates, strains retaining only ΔdivIVA::ΩaacC4 at the native locus arose with a higher frequency and grew better than in the absence of thiostrepton. Probably, the known low level of tipAp activity even in the absence of thiostrepton (Ali et al., 2002) gave sufficient divIVASC expression for survival, but a higher level was needed for good growth (see below).

Table 1. . Segregation of integrated cosmid pKF56 (KmR) and the ΔdivIVA::ΩaacC4 allele (ApraR) in the presence and absence of divIVASC+in trans.
Original cosmidco-integrantaPlasmidin attBpSAM2aNumber of ApraR colonies testedbFrequency of
ApraR KmS divIVA)bApraR KmR (divIVA ±/ΔdivIVA)b
  • a

    . Cosmid pKF56 co-integrants are partial diploids for the region surrounding divIVASC, and heterozygous for this locus (divIVA+/ΔdivIVA::ΩaacC4). In addition, they contained pPM927 (empty vector) or pKF58 (providing tipAp-divIVASC+in trans).

  • b . After sporulation of the original strains, derivatives retaining the ΔdivIVA::ΩaacC4 deletion allele [conferring apramycin resistance (Apra R)] were tested for resistance to kanamycin (KmR), which is indicative of the presence of the cosmid vector and that the strain was still a partial diploid, heterozygous for the divIVASC locus.

  • c

    . Genotype at divIVA locus confirmed by PCR for randomly chosen strains.

M145 pKF56pPM927 72 0%100%
M145 pKF56pKF5811621%c 79%

Partial depletion of DivIVASC causes defects in hyphal growth

As no divIVASC null-mutant could be obtained, the function of the gene was investigated by partial depletion of DivIVASC. Spores of strains K114 (divIVASC+/tipAp-divIVASC+) and K115 (ΔdivIVASC/tipAp-divIVASC+) were germinated in TSB liquid medium in the absence of thiostrepton and examined after 17 h of growth. As indicated by Western blot analysis, K114 had a somewhat higher relative content of DivIVASC (Fig. 2, lane 5) than the wild-type strain M145 (Fig. 2, lane 1), whereas strain K115, which lacked the native chromosomal copy of the gene, contained around 10-fold less of the protein (Fig. 2, lane 3). The hyphae formed by K114 were indistinguishable from those formed by the parent strain M145 during these conditions. On the other hand, K115 had a distinctive phenotype. Hyphae were irregularly shaped, kinky and curly. Many lysed cells could be observed. Branching was irregular and branches were often seen close to hyphal tips (Fig. 3). This overall phenotype is strikingly reminiscent of that produced in some fungal mutants defective in tip growth or nuclear migration (see Discussion).

Figure 2.

Overexpression and partial depletion of DivIVASC. Western blot analysis using monoclonal antibody FM126-2 raised against M. kansasii antigen 84, which is a mycobacterial homologue of DivIVA. Equal amounts of total cell protein were loaded in each well from S. coelicolor strains grown in the absence (lanes 1, 3, 5) or presence (lanes 2, 4, 6) of 10 µg ml−1 thiostrepton. The strains were M145 (lane 1); K116 (lane 2); K115 (lanes 3, 4); K114 (lanes 5, 6). Plasmid pKF58 carries divIVASC inserted under control of the thiostrepton-inducible tipAp promoter in the vector pPM927. Lane 7 contained an extract of M. tuberculosis strain Harlingen. The total protein content loaded in lane 7 was not the same as in the others. The positions of molecular weight size markers (kDa) are indicated.

Figure 3.

Effect of DivIVASC depletion on hyphal morphology. Strain K114 (divIVASC+/pKF58[tipAp-divIVASC]) (A, C, E), and strain K115 (ΔdivIVA::ΩaacC4/pKF58[tipAp-divIVASC]) (B, D, F), were grown in the absence of thiostrepton, before being prepared for microscopy. Representative phase-contrast microscopy images are shown (A and B). DNA was stained with 7-aminoactinomycin D, and observed in a fluorescence microscope (E and F). To better visual the distribution of DNA in hyphae, overlays of the fluorescence signals on phase-contrast images were generated (C and D). Size bar, 10 µm (G) Histogram of distances between tips and primary lateral branches in cultures grown 24 h in YEME. Strain K114 was grown in the absence of thiostrepton (filled columns), and K115 in the absence (open columns) or presence of 0.01 µg ml−1 thiostrepton (crosshatched columns). For around 200 primary branch-points per strain, the length was measured between the tip and the first lateral branch. For each branch, only the distance to the tip furthest away was measured.

The apical branching phenotype was further analysed after growth in YEME medium, which reduces clumping of mycelium and gives more dispersed growth. Although somewhat less severely affected than in TSB, K115 had similarly altered morphology and branching patterns in YEME and TSB. The average distance from tips to the first lateral branchpoints was 6.4 ± 3.9 µm for K115 compared to 14.7 ± 4.7 µm for K114 (Fig. 3G). In the presence of 0.01 µg ml−1 thiostrepton, the distance between tips and first branches increased to 15.2 ± 6.4 µm for K115 (Fig. 3G) and overall morphology was restored to normal, whereas this concentration had no significant effect on K114 (data not shown). Thus, induction of tipAp-divIVASC expression with a low level of thiostrepton was sufficient to suppress most of the growth defect in K115.

Partial depletion of DivIVASC appeared to cause branching of apical cells, which is unusual in Streptomyces. However, it was also possible that underexpression of divIVASC could cause alterations of cell division patterns, such that crosswalls would be present close to tips. In that case, branching of the subapical cell would be close to the apex, and might give the appearance of apical branching. To resolve this, hyphal crosswalls were visualized by staining the membrane of young mycelia. This clearly showed that K115 frequently had a branch in the apical cell (examples shown in Fig. 4), whereas this was not observed in the wild-type M145 (example in Fig. 4) or K114. No overt differences between K115 and K114 were observed in the frequency or spacing of septa. However, exact quantitative comparisons of septation were not meaningful because of the different hyphal morphologies of the strains.

Figure 4.

Examples of branching of apical hyphal cells caused by reduction in divIVASC expression. Strains M145 (A) and K115 (M145 ΔdivIVA::ΩaacC4 pKF58[tipAp-divIVASC]) (B, C, D) were grown in the absence of thiostrepton, before being harvested, and stained with FM4-64 to visualize hyphal crosswalls. Crosswalls are indicated by arrowheads. Size bar, 10 µm.

Because nuclear migration defects in fungi could give a phenotype similar to that caused by partial DivIVASC depletion, and as B. subtilis DivIVA appears to have a role in targeting the chromosomal oriC region to the cell pole during sporulation, it was relevant to look at distribution of DNA in the DivIVASC-underexpressing hyphae. Staining of DNA with 7-aminoactinomycin D (Fig. 3C–F) or DAPI (data not shown) did not reveal any overt problems in the distribution of DNA into branches and tips in K115, and the general staining patterns were similar to those of K114. In both cases, nucleoids of variable sizes and shapes were uniformly distributed in hyphae and branches, except for the very tip where no DNA staining was seen (Fig. 3C–F)

Overexpression of DivIVASC dramatically alters cell shape

By adding 10 µg ml−1 of thiostrepton to the medium, the tipAp promoter on plasmid pKF58 could be strongly induced, leading to around 25-fold overexpression of divIVASC (Fig. 2, lanes 4 and 6), compared to when strains M145 or K114 (M145/pKF58) were grown in the absence of thiostrepton, or when strain K116 (M145/pPM927) was grown in the presence of thiostrepton (Fig. 2, lanes 1, 2, and 5). The additional bands of truncated divIVASC products that are visible in lanes 4 and 6 constituted only a minor fraction of the overproduced protein, and they were not detected when the sample was diluted to a level where the main band had the same intensity as that of the wild-type strain in lane 1. However, it can not be ruled out that the truncated forms of the protein may contribute to the phenotypic consequences of overexpression that are described below. Whereas K114 grown in the absence or K116 grown in the presence of thiostrepton were indistinguishable from the normal hyphae produced by the parent strain M145, overexpression of divIVASC gave rise to a very different cell type (Fig. 5). These cells had an oval and swollen shape and were much shorter and thicker than normal hyphae (a field of representative cells is shown in Fig. 5C–E). They were able to grow, branch, and form relatively dense cultures, although these cultures grew more slowly than normal and contained many lysed cells. The cells appeared to retain a marked polarity with the wider and rounded end always pointing away from branch points. In similarity to the very tips of normal hyphae, there was a polar zone apparently devoid of DNA at these cell ends (Fig. 5E).

Figure 5.

Effect of divIVASC overexpression on hyphal morphology. Cultures were grown in the absence (A) and presence (B–E) of 10 µg ml−1 of the inducer thiostrepton. Strain K114 carrying pKF58[tipAp-divIVASC] (A, C, D, E) or K116 carrying the empty vector pPM927 (B) were used. Fixed samples were stained with FM4-64 to visualize the cell membrane (D), or DAPI to visualize DNA (E). C–E show exactly the same field viewed with different fluorescence filters. A–C show phase-contrast images. In C, the thin black arrow indicates a branched cell, and the thicker white arrow a lysed ghost cell retaining the shape of living cells. Size bar, 10 µm.

The swollen cell shape (Fig. 5C) suggested that divIVASC overexpression could weaken the cell wall by disturbing peptidoglycan synthesis or stimulating autolytic processes. To observe whether the cell wall appeared degraded, cross-sections of such cells were analysed by transmission electron microscopy (TEM). This showed that the large oval cells were surrounded by an intact and apparently normal peptidoglycan layer (Fig. 6). In addition, many empty ghost cells were observed, which retained cell walls with shapes similar to those of living cells (arrow in Fig. 6C). Similar ghost cells were observed by phase-contrast microscopy (thick arrow in Fig. 5C). The TEM cross-sections also showed a large electron-dense region at the thick rounded poles of divIVASC overexpressing cells. This region was clearly distinct from the appearance of the rest of the cytoplasm (Fig. 6), and corresponded in shape and position to the DNA-free space at these cell poles observed after DAPI staining (Fig. 5E).

Figure 6.

Transmission electron micrographs of thin sections of normal hyphae formed by strain M145 (A), and the characteristic rounded cells of strain K114 overexpressing DivIVASC (B and C). The arrow indicates an empty ghost cell. Size bar, 1 µm.

Overexpression of DivIVASC affects tip extension and can cause hyperbranching

To observe the effects of overexpression of divIVASC on preformed hyphae, spores from K114 were allowed to germinate and grow in the absence of thiostrepton for 12 h, before the inducer was added. Figure 7 shows that 5 h after induction of tipAp, hyphal tips had become swollen and rounded, and multiple rounded outgrowths had appeared along the length of hyphae. Using a recently developed method for staining of nascent, not yet cross-linked peptidoglycan with fluorescently labelled vancomycin (Daniel and Errington, 2003), it was possible to visualize the effect of divIVASC overexpression on tip growth. Without induction of tipAp, hyphae of strain K114 showed strong staining at the tips (typical example in Fig. 7E). Three hours after addition of the inducer, the overall fluorescence was weaker, and vancomycin stained a broader region at the swelling tips (arrows in Fig. 7F and H). In addition, there were multiple foci of fluorescence along the hyphae, which often coincided with the outgrowths of the lateral wall (Fig. 7F). A smaller fraction of the hyphae retained an apparently normal shape 3 h after induction and did not show any clear swellings or outgrowths. Invariably, such hyphae did not stain at the tips (not shown), indicating that apical extension was inhibited by overexpression of divIVASC in some cells.

Figure 7.

Development of divIVASC overexpression phenotype. S. coelicolor strains were grown from spores in YEME medium for 12 h to form young hyphae, and then 5 h in the absence (A) or presence of 10 µg ml−1 of thiostrepton (B–D) to induce overproduction of DivIVASC. The strains were K114 carrying pKF58 (tipAp-divIVASC) (A–C), and K116 carrying the empty vector pPM927 (D). E–H show typical hyphae from strain K114 sampled after 3 h in the absence (E,G) or presence (F,H) of thiostrepton. They have been stained with BODIPY FL vancomycin to visualize sites of ongoing peptidoglycan assembly. Fluorescence images with inverted greyscale (E and F) and phase-contrast images (G and H) are shown. Size bars, 5 µm.

DivIVA is specifically localized at growing hyphal tips

Plasmid pKF59 had the entire divIVASC fused in frame to egfp, and was integrated via a single homologous recombination event at the divIVASC locus. Plasmid pKF61 was identical except for an N-terminal truncation of divIVASC, such that upon integration it disrupted the chromosomal copy of divIVASC, leaving only the divIVASC-egfp fusion active. In immunoblots, it was confirmed that the fusion protein was expressed at the same level as divIVASC in wild-type strains, and that DivIVASC was absent from the pKF61-co-integrant K117 (Fig. 8L). Because the strain carrying pKF61 grew normally, it can be concluded that the DivIVASC-EGFP hybrid protein retained most of the function of DivIVASC. However, small irregularities in branching and shape of hyphal tips indicated that the EGFP tag was not completely neutral for DivIVASC activity.

Figure 8.

Subcellular localization of DivIVASC-EGFP. The strains were M145 (A and B), K113 carrying an ftsZ-egfp fusion integrated at attBφC31 (C and D), K112 carrying a divIVASC-egfp fusion integrated at the divIVASC locus (E–H), and K117 carrying an divIVASC-egfp fusion integrated at the divIVASC locus in such a way that the native allele was disrupted (I–J). A, C, E, G, and I show overlays of fluorescence images (green) on the phase-contrast images (greyscale). B, D, F, H, and J show the fluorescence images with inverted greyscale to clearly visualize the signals. Small red arrows highlight the strong fluorescence at each hyphal tip. Small white arrows indicate examples of fluorescent foci at putative nascent branch points, White arrowheads point to cytokinetic FtsZ rings, and black arrowheads indicate examples of DivIVASC-EGFP at possible division sites. K is a montage showing spores of strain K112 at different stages of germination. Phase-contrast images to the left have an overlay showing the localization of DivIVASC-EGFP in green. The corresponding fluorescence signals are shown to the right with inverted greyscale. L shows an immunoblot where DivIVASC and DivIVASC-EGFP has been detected in cell extracts of stains M145 (lane 1), K112 (lane 2), and K117 (lane 3), exactly as described for Fig. 2. An irrelevant lane between lanes 1 and 2 has been deleted in the image. Size bar, 10 µm.

A background of faint fluorescence was often seen in vegetative hyphae of M145 carrying no EGFP fusion, and exposure conditions were adjusted so that this was barely visible (Fig. 8A and B). Strikingly, both strains K112 (with pKF59) and K117 (with pKF61) showed distinctive regions of very strong fluorescence at the tips of all vegetative hyphae (Fig. 8E–J, red arrows). In addition, some smaller foci of fluorescence were observed at positions along the cell periphery. In many cases these were consistent with putative positions of branch initials (examples indicated by white arrows in Fig. 8E–J). Many scattered spots of very weak fluorescence were also detected, but these could not be distinguished from inherent autofluorescence in the hyphae of strain M145.

In strain K113, which carries an ftsZ-egfp fusion, cytokinetic FtsZ rings were clearly seen, indicating the sites of ongoing cell division (arrowheads in Fig. 8D). Faint signals in the shape of bands of fluorescence were occasionally observed in K112 and K117 (both expressing DivIVASC-EGFP) at subapical positions that could resemble sites of cell division or hyphal crosswalls (examples indicated by black arrowheads in Fig. 8F), but such signals were weak and inconsistent in comparison to the strong fluorescence at hyphal tips. They were only seen in some hyphae, and most hyphal crosswalls did not have such signals. This suggested that Streptomyces DivIVASC may transiently or weakly associate with division sites, but does not normally remain at the crosswalls. In summary, DivIVASC was predominantly located at hyphal tips and nascent branches, which is clearly different from the subcellular distribution of B. subtilis DivIVA.

DivIVASC-EGFP localization was also monitored during spore germination. Spores did not contain a clear signal (Fig. 8K), although it has previously been shown that EGFP can readily be observed in mature S. coelicolor spores when expressed from promoters active during sporulation (Sun et al., 1999). During germination, spores swelled to become large, round and dark in phase-contrast illumination. At this stage, distinctive foci of DivIVASC-EGFP fluorescence became visible, and in all examined spores with any sign of germ tube emergence, there was a fluorescent focus at the site of tip outgrowth (examples in Fig. 8K).

Discussion

DivIVASC is the first molecular marker of hyphal tips that has been identified in a prokaryotic mycelial organism. Whereas a number of proteins, for example some penicillin-binding proteins, are expected to be targeted to these cell poles, DivIVASC is particularly interesting as it is an essential protein with a new and unknown function that is shown here to have a profound impact on tip growth. Partial depletion of DivIVASC interfered with polar growth and lead to abnormal hyphal shapes. The crooked, curled, and distorted hyphae, with irregular and often apical branching, were strikingly similar to the overall phenotype of several fungal mutants with defects in, for example, cytoplasmic dynein, kinesin, actin-related Arp1, vacuolar ATPase, and other genes related to tip growth or nuclear migration (Plamann et al., 1994; Xiang et al., 1995; Seiler et al., 1997; Inoue et al., 1998; Bowman et al., 2000; Momany, 2002). A similar phenotype has to our knowledge not previously been described in Streptomyces. Furthermore, overexpression of DivIVASC strongly perturbed cell shape determination, which must largely take place at the tips in apically growing hyphae. Finally, overexpression of the protein was shown to affect cell wall assembly at the tips, and gave rise to many new sites of de novo incorporation of peptidoglycan, as revealed by staining with fluorescently labelled vancomycin. These observations strongly suggest that DivIVASC is required for mechanisms involved in apical extension in S. coelicolor. Also the subcellular localization of the protein is consistent with such function.

The swollen tips seen in hyphae after a few hours of divIVASC induction (Fig. 7) were superficially reminiscent of the effects of lysozyme or certain antibiotics on hyphal actinomycetes (Locci, 1980; Gray et al., 1990). In addition, many lysed cells were observed after prolonged divIVASC overexpression (Fig. 5). However, the DivIVASC-induced effect on morphology was caused by interference with cell shape determination, rather than being swellings due to a weakened or degraded cell wall. This conclusion is based on the following observations: (i) TEM cross-sections showed a cell wall of apparently normal thickness; (ii) the empty ghost cells retained the shape of living cells and did typically not show signs of breakage at the tips; (iii) DivIVASC-overexpressing cells were not easier to lyse by sonication than normal hyphae (K. Flärdh, unpubl. obs.); (iv) the shape of the overexpressing cells was not dependent on medium osmolarity; (v) the wall in the overexpressing cells was stained with peptidoglycan-binding fluore-scently labelled wheat germ agglutinin (K. Flärdh, unpubl. data). Thus, DivIVASC is a novel cell shape determinant in bacteria, and it is localized at or very close to the sites of peptidoglycan assembly in Streptomyces hyphae.

The B. subtilis DivIVA protein is involved in selection of cell division sites, and is required for proper segregation of chromosomes into developing spores. In this paper, it is shown that the DivIVA homologue in S. coelicolor has a different function related to asymmetrical growth and cell shape determination. On the following grounds, I conclude that Streptomyces DivIVA does not primarily affect cell division, and does not act through the MinCD system: (i) the overwhelming effects of altered divIVASC-expression were on cell morphology and hyphal growth rather than on cell division; (ii) both DivIVASC-depleted and overexpressing cells formed abundant septa; (iii) no homologue of MinC has been detected in the S. coelicolor genome (Bentley et al., 2002; K. Flärdh, unpubl. obs.). The data presented in this paper do not support any role for Streptomyces DivIVA in the distribution of DNA within hyphae, although it can not be excluded that the small amount of DivIVASC left in the underexpressing strains may be sufficient to fulfil a role in DNA positioning.

The polar targeting of Bacillus and Streptomyces DivIVA show both similarities and differences. In B. subtilis, the cell poles are created by cell division, and DivIVA localizes to these sites once the septation process has been initiated at midcell, and then remains at the poles after division. Typically, DivIVA seems to be more abundant at the septation sites than at the poles in B. subtilis (Edwards and Errington, 1997; Marston et al., 1998). In contrast, the hyphal apices in S. coelicolor, where DivIVASC primarily was found, are created de novo as lateral hyphal branches or as germ tubes emanating from spores, and are not dependent on cell division. The cell poles generated by division in S. coelicolor, i.e. those at hyphal crosswalls, seemed only occasionally or transiently to contain detectable DivIVASC. Thus, DivIVASC is able to specifically recognize hyphal tips independently of septation. Interestingly, in the first cell cycle after spore germination, also B. subtilis DivIVA appears to find the new cell poles independently of cell division (Hamoen and Errington, 2003; Harry and Lewis, 2003). Therefore, despite their different functions, it can not be excluded that Streptomyces and Bacillus DivIVA proteins might share similar polar targeting mechanisms, at least during spore germination. This could involve some general feature of cell poles as the B. subtilis protein can target poles in both E. coli and Schizosaccharomyces pombe, none of which contain DivIVA-like proteins (Edwards et al., 2000). On the other hand, the presence of DivIVASC-EGFP foci at the earliest recognizable stages of germ-tube emergence, and possibly at sites where lateral branches will emerge, suggest that the Streptomyces protein could target sites that are not yet cell poles. Only when apical growth is established at such sites (perhaps dependent on DivIVASC) will they become new hyphal tips.

At least three hypothetical roles can be envisioned for DivIVASC. First, it could be required for creating or maintaining the cell polarity that is needed for tip extension. Because DivIVASC can specifically localize to the tips, it might act by recruiting other proteins needed for growth polarity to these sites. The assembly of DivIVASC foci before any other visible signs of growth polarization (Fig. 8K), and the apparent hyperbranching caused by DivIVASC overexpression (Fig. 7) are consistent with such a model. However, if DivIVASC is required for tip growth, it appears contradictory that decreased expression caused emergence of branches in apical cells. It should be noted that many different mutations that interfere with tip extension in filamentous fungi also give rise to apical branching. A plausible explanation is that poor, sluggish or unstable extension of the apical cell may generate signals for branching in the same way as the absence of tip extension in subapical cells in an unknown way may trigger branching. The residual level of DivIVASC present in the partial depletion experiment could be sufficient for a new branch to arise in the apical cell. A second possible function for DivIVASC is to act as a structural scaffold or be involved in cytoskeleton-like functions at the growing cell pole. The architecture of the zone of localized peptidoglycan polymerization at the tip is not clear, but may be organized by cytoplasmic elements. It is also likely that the cell wall at the tip exhibits some degree of flexibility and may be both osmotically susceptible and prone to autolysis (Prosser and Tough, 1991; Miguélez et al., 1992), and DivIVASC could have a role in maintaining the shape or integrity of the tip. The abundance of lysed hyphae upon partial depletion of the protein would be consistent with this model. In addition, the size and intensity of the fluorescent foci of DivIVASC-EGFP at the tips indicated that a large number of molecules were concentrated there, possibly as oligomeric structures, as suggested by the oligomerization of B. subtilis DivIVA (Muchováet al., 2002). Third, DivIVASC could be involved in apical dominance or suppression of branching in the vicinity of growing tips. This would bear analogy to how DivIVA acts via MinCD to inhibit septation near cell poles in B. subtilis. Although the apical branching phenotype fits well in this model, it is not obvious how it could explain that depletion of the protein also caused aberrant hyphal shape and poor extension.

The genetic data show that divIVASC is essential, but it is not yet clear why cells can not grow without this gene. As discussed above, the effects of partial depletion of the protein to about 10% of the normal level may be ascribed to problems with tip extension. In this view, we speculate that DivIVASC is required for growth or integrity of hyphal tips. However, a tight system for conditional inactivation or complete depletion of the protein is needed to clarify this. A recent survey of genes required for growth of M. tuberculosis indicated that the mycobacterial divIVA homologue Rv2145c (gene wag31 encoding Antigen 84) is essential (Sassetti et al., 2003). Furthermore, J. A. Gil and co-workers recently found that the divIVA homologue in the rod-shaped or pleiomorphic actinomycete Brevibacterium lactofermentum could not be disrupted, that the gene product was polarly localized, and that overexpression of this gene resulted in oval swollen cells, reminiscent of S. coelicolor cells overexpressing divIVASC (A. Ramos, P. Honrubia, N. Valbuena, J. Vaquera, L. M. Mateos and J. A. Gil, manuscript in preparation). It has also recently been demonstrated that the very closely related Corynebacterium glutamicum grows in a polar fashion (Daniel and Errington, 2003). In conjunction with the results presented here, this suggests that DivIVA-like proteins could have essential roles in polar growth and morphogenesis in many actinomycetes, not only those growing as hyphae. It also indicates that the long Gly-Pro-Gln-rich sequence that separates the predicted coiled coil segments in the S. coelicolor protein may not be required for a function in growth polarity, as this segment is not present in B. lactofermentum and other corynebacteria.

Experimental procedures

Bacterial strains and media

The Streptomyces coelicolor A3(2) and Escherichia coli strains that were used are listed in Table 2. Culture conditions, antibiotic concentrations, genetic manipulations, and recombinant DNA work followed in general previously described procedures for E. coli (Sambrook et al., 1989) and Streptomyces (Kieser et al., 2000). Streptomyces coelicolor strains were cultivated in TSB or YEME liquid medium, or on MS agar plates (Kieser et al., 2000).

Table 2. . Bacterial strains used in this study.
StrainGenotypeReference
Streptomyces coelicolor A3(2) strains
K112M145 divIVASC+::pKF59[Φ(divIVASC-egfp)]This work
K113M145 attBφC31::pKF41[Φ(ftsZ-egfp)] Grantcharova et al. (2003)
K114M145 attBpSAM2::pKF58[tipAp-divIVASC]This work
K115M145 ΔdivIVA::ΩaacC4 attBpSAM2::pKF58[tipAp-divIVASC]This work
K116M145 attBpSAM2::pPM927This work
K117M145 divIVASC::pKF61[Φ(divIVASC-egfp)]This work
M145prototrophic, SCP1 SCP2 Pgl+ Kieser et al. (2000)
Escherichia coli strains
DH5α supE44 ΔlacU169 (φ80 lacZΔM15) hsdR17 recA1 endA1 gyrA96 thi-1 relA1Our lab stock
DY380DH10B λcI857 Δ(cro-bio) < > tet Lee et al. (2001)
ET12567GM2929 hsdMhsdRzjj-202 MacNeil et al. (1992)
GM2929 dam-13::Tn9 dcm-6 hsdR2 recF143 galK2 galT22 ara-14 lacY1 xyl-5 thi-1 tonA31 rpsL136 hisG4 tsx-78 mtl-1 glnV44 leuB6 rfbDM. Marinus
XL1-Blue recA1 endA1 gyrA96 thi-1 hsdR17 supE44 relA1 lac[F′proAB lacIqZΔM15 Tn10 (TetR)]Our lab stock

Construction of plasmids

The divIVA homologue of S. coelicolor A3(2) was obtained from cosmid C91 (provided by H. Kieser and D. A. Hopwood, John Innes Centre, Norwich, UK). This corresponds to cosmid SC4A10 used in the genome sequencing project (Redenbach et al., 1996; Bentley et al., 2002), although the exact endpoints of the insert in cosmid C91 are unknown. The divIVASC gene from C91 proved to be fully functional, as, after subcloning, it complemented a null-mutation in the chromosomal divIVASC locus (see Results).

Plasmid pKF58 was constructed by amplifying divIVASC from cosmid C91 using Pfu DNA polymerase and primers KF86 (ggtcgacggcgagacggtca) and KF87 (cgtatcgcagagggcgggttc), digesting the product with BclI, and cloning the 1844 bp fragment in the BamHI site of pPM927 (Smokvina et al., 1990), oriented so that divIVASC was placed under the control of the thiostrepton-inducible promoter tipAp. By conjugation from E. coli strain ET12567 containing the driver plasmid pUZ8002 (Kieser et al., 2000), pKF58 was transferred into S. coelicolor strains, where it integrated into the pSAM2 attachment site.

To construct a translational fusion of divIVASC to the gene for EGFP, the PCR product generated using primers KF86 and KF87 was digested with BclI and BsiWI, and cloned between the BglII and Asp718 sites of pEGFP-1 (Clontech Laboratories). This generated plasmid pKF54, which contained divIVASC placed in front of the egfp gene. Primers KF100 (gtcgtcctcgtcgatcaggaac) and KF101 (ccggtcgccaccatggtgag) were 5′-phosphorylated and used to amplify most of pKF54 with Pfu DNA polymerase, except the region between divIVASC and egfp. The resulting PCR product was religated to generate plasmid pKF59, in which the last codon (Asn) and the stop codon of divIVASC were replaced by in-frame codons for Pro-Val-Ala-Thr, immediately followed by the ATG start codon of egfp, thus generating a translational fusion denoted Φ(divIVASC-egfp). Plasmid pKF61 was made by digestion of pKF59 with AccI and BssHI, end-filling with Klenow polymerase, and religation. This deleted a 233 bp fragment, including the start codon and the first 144 bp of divIVASC.

Creation of a divIVASC null-allele

The PCR-targeting procedure described by Yu et al. (2000) was used to replace divIVASC on cosmid C91 with a resistance marker. Oligonucleotides KF78 (AAACCCGGGG CCCGCCCTGA ACGTTGCCGA AGGCTACGCC GTACTA CAGA cagaggtttt caccgtcatc ac) and KF79 (GAGAGAGATA CGGGCTTGCC GAATGCCGAC GACAACGTTG AGGT GAAGAG gccctttcgt cttcaagaat tcc) were designed so that the 3′-ends (written in small letters above) would prime amplification of antibiotic resistance cassettes from the pHP45Ω plasmids (Blondelet-Rouault et al., 1997), whereas the 5′-ends were 50 bp tails that would be homologous to sequences flanking divIVASC(written in capital letters above). These primers were used to amplify the ΩaacC4 cassette, which confers resistance to apramycin (Blondelet-Rouault et al., 1997). The PCR product was transformed into E. coli strain DY380 (Lee et al., 2001) carrying cosmid C91 and the λRed system induced as described previously (Yu et al., 2000). Among apramycin-resistant transformants, one was identified which had the expected gene replacement, as confirmed by restriction mapping and PCR amplifications. This cosmid carrying ΔdivIVA::ΩaacC4 was designated pKF56.

Disruption of the chromosomal divIVASC locus

The cosmid pKF56 was passaged through E. coli strain GM2929, and the non-methylated and alkali-denatured DNA was used to transform S. coelicolor A3(2) strain M145 to apramycin resistance as described (Oh and Chater, 1997; Kieser et al., 2000). All transformants tested were also kanamycin resistant, and four such cosmid co-integrants were grown in the absence of antibiotic selection for making spore preparations (Kieser et al., 2000). Spores were plated with or without apramycin selection to obtain single colonies, which were examined for antibiotic resistances. To determine which divIVA allele was present in some of the resulting strains, total DNA was extracted using the CTAB procedure (Kieser et al., 2000), and used as template for PCR amplification with primers KF86 and KF87, which gave characteristic products from the two different divIVA alleles.

The experiments testing for segregation of cosmids and divIVA alleles were repeated using cosmid pKF56 co-integrants into which pPM927 or pKF58 had been introduced by conjugation. The absence of the native chromosomal divIVASC allele from all eight randomly chosen ApraR KmS strains that arose from the co-integrants carrying pKF58 was confirmed using PCR as described above, and in Southern blots and hybridizations with a digoxygenin-labelled divIVASC probe (DIG system, Roche Diagnostics).

Cosmids that were excised by homologous recombination from the chromosome of S. coelicolor strain M145 carrying integrated cosmid pKF56 were recovered by transformation of E. coli strain XL1-Blue (Oh and Chater, 1997). The identity of the divIVASC allele on recovered cosmids was indicated by resistance or sensitivity to apramycin of the KmR AmpRE. coli transformants, and then confirmed by restriction mapping of extracted cosmid DNA.

Overexpression and partial depletion of DivIVA SC

Spores of S. coelicolor strains carrying pKF58 or pPM927 were inoculated into TSB medium (if not stated otherwise) containing spectinomycin at 100 µg ml−1, and incubated with shaking at 30°C. For overexpression of divIVASC, thiostrepton was included at 10 µg ml−1.

The relative cellular contents of DivIVASC were estimated by immunoblotting. Mycelium was harvested after 17 h of growth, washed twice in ice-cold Tris-HCl, pH 7.4; 150 mM NaCl; 1 mM EDTA, and resuspended in the same buffer containing 1% Triton X-100 and Protease Inhibitor Cocktail for Mammalian Tissues (Sigma-Aldrich). Cells were broken by sonication. Total protein concentrations in the crude lysates were determined with a modification of the Lowry method (Peterson, 1977). Crude cell lysates were mixed with an equal volume of 2 × sample buffer (100 mM Tris-HCl, pH 6,8; 4% SDS; 20% glycerol; 200 mM dithiothreitol; 0.25 mg ml−1 bromophenol blue), and boiled for 3 min. Equal amounts of total protein were separated on SDS-polyacrylamide gels. Proteins were transferred electrophoretically to Immobilon-P membranes (Millipore Corporation) using a Mini Trans-Blot cell according to the manufacturer's instructions (Bio-Rad Laboratories). DivIVASC was detected using a 1 : 5000 dilution of monoclonal antibody F126-2 raised against Ag84 from Mycobacterium kansasii (Hermans et al., 1995), peroxidase-linked sheep anti-mouse IgG secondary antibodies (Amersham Pharmacia Biotech), and the ECL Western blotting detection reagents (Amersham Pharmacia Biotech). To estimate the difference in DivIVASC content between strains, samples were diluted until they gave rise to a band of equal intensity as that in a control strain (e.g. the wild-type parent).

Fluorescence microscopy

Living hyphae were mounted for microscopy and observation of EGFP by spotting about 20 µl liquid culture directly on microscope slides coated with 1% agarose in phosphate-buffered saline (PBS), and covering with a coverslip.

For staining procedures, samples of culture were fixed, stained, and allowed to air-dry on poly l-lysine-coated slides as described previously (Grantcharova et al., 2003). Cells were mounted under coverslips in PBS containing 50% glycerol, 1 µg ml−1 FM4-64, and 1.25 µg ml−1 DAPI. Alternatively, the fixed and dried cells on poly l-lysine-coated slides were incubated with 10 µg ml−1 7-aminoactinomycin D (Molecular Probes) in PBS for 1 h, washed five times with PBS, and mounted in PBS containing 50% glycerol. Staining of nascent peptidoglycan was done according to a method developed by Daniel and Errington (2003). BODIPY FL vancomycin (Molecular Probes) and unlabelled vancomycin (Sigma) were incubated with the growing culture for 5 min, both at final concentrations of 1 µg ml−1. Cells were then fixed in 2.3% formaldehyde, washed twice in PBS and mounted on slides as described above.

Phase-contrast microscopic images in Fig. 7A–D were captured using a Nikon Optiphot-2 microscope (Åkerlund et al., 1995). All other light microscopy samples were viewed and digital images captured as in Grantcharova et al. (2003).

Transmission electron microscopy

Cells collected by centrifugation were fixed by resuspension in freshly made 2.5% glutaraldehyde in phosphate buffer, and post-fixed in 1% osmium tetroxide. The cells were embedded in Epon, and post-staining was done with 1% uranyl acetate. Sections were made approximately 60 nm thick. Specimens were analysed with a Zeiss CEM 902 electron microscope, equipped with a spectrometer to enhance image contrast, at an accelerating voltage of 80 kV. A liquid nitrogen cooling trap of the specimen holder was used throughout. Series of electron micrographs were examined to recognise representative morphologies.

Acknowledgements

Stefan Höglund is gratefully acknowledged for excellent electron microscopy work. I would like to thank Mark Buttner, Keith Chater, Santanu Dasgupta and Kurt Nordström for valuable comments on the manuscript, José A. Gil for communication of results prior to publication, R. A. Daniel for advice on the vancomycin staining, and Nina Grantcharova and Håkan Lund for discussions and assistance in the laboratory. I am very grateful to Arend Kolk for gift of anti-Ag84 antibodies, Thomas Åkerlund for gift of mycobacterial cell extract, and Don Court, Martin Marinus, and Jean-Luc Pernodet for strains and plasmids. This work was supported by grants from The Swedish Research Council, Åke Wibergs Stiftelse, and Magn. Bergvalls Stiftelse. The Zeiss Axioplan II Imaging fluorescence microscope was purchased with support from the Swedish Natural Sciences Research Council.

Ancillary