Two novel homologous proteins of Streptomyces coelicolor and Streptomyces lividans are involved in the formation of the rodlet layer and mediate attachment to a hydrophobic surface

Authors


Summary

The filamentous bacteria Streptomyces coelicolor and Streptomyces lividans exhibit a complex life cycle. After a branched submerged mycelium has been established, aerial hyphae are formed that may septate to form chains of spores. The aerial structures possess several surface layers of unknown nature that make them hydrophobic, one of which is the rodlet layer. We have identified two homologous proteins, RdlA and RdlB, that are involved in the formation of the rodlet layer in both streptomycetes. The rdl genes are expressed in growing aerial hyphae but not in spores. Immunolocalization showed that RdlA and RdlB are present at surfaces of aerial structures, where they form a highly insoluble layer. Disruption of both rdlA and rdlB in S. coelicolor and S. lividans (ΔrdlAB strains) did not affect the formation and differentiation of aerial hyphae. However, the characteristic rodlet layer was absent. Genes rdlA and rdlB were also expressed in submerged hyphae that were in contact with a hydrophobic solid. Attachment to this substratum was greatly reduced in the ΔrdlAB strains. Sequences homologous to rdlA and rdlB occur in a number of streptomycetes representing the phylogenetic diversity of this group of bacteria, indicating a general role for these proteins in rodlet formation and attachment.

Introduction

Streptomycetes are Gram-positive soil bacteria that colonize moist substrates by forming a branched network of multinucleoid hyphae. At some stage during their life cycle, these bacteria are confronted with a hydrophobic environment. For instance, after a feeding substrate mycelium has been established, hyphae leave the aqueous environment to grow into the hydrophobic air. These aerial hyphae differentiate further by forming chains of uninucleoid cells, which metamorphose into pigmented spores. Spores or hyphae of streptomycetes may also encounter hydrophobic solids such as surfaces of dead or living organisms. When streptomycete hyphae leave their aqueous environment, they change their surface. Hyphae in a moist substrate are hydrophilic, whereas the surfaces of aerial hyphae and spores are hydrophobic.

Formation of aerial structures has been best studied in Streptomyces coelicolor (for recent reviews, see Chater, 1998; 2001; Kelemen and Buttner, 1998; Wösten and Willey, 2000). Bald (bld) mutants of S. coelicolor were isolated that, when grown on rich medium, are affected in the formation of aerial structures and in the production of a small surface-active peptide called SapB (Willey et al., 1991). Many of these mutants appear to be affected in an extracellular signalling cascade involved in the erection of aerial hyphae (Willey et al., 1993; Nodwell et al., 1996; 1999). Experimental evidence suggests the existence of at least five signalling molecules. It was hypothesized that each signal triggers the synthesis and release of the next signal, ultimately leading to the production and secretion of SapB (Willey et al., 1993; Nodwell et al., 1996). By lowering the water surface tension from 72 to 32 mJ m−2, SapB enables hyphae to breach the water–air interface to grow into the air (Tillotson et al., 1998).

Aerial hyphae and spores of S. coelicolor have several surface layers that make them hydrophobic. One surface layer, called the rodlet layer, has a typical ultrastructure of a mosaic of 8- to 10-nm-wide parallel rods (Wildermuth et al., 1971; Smucker and Pfister, 1978). The nature of the surface layers is not known. SapB is not expected to form one of these layers, as this peptide was localized in the culture medium but could not be detected at the surfaces of aerial structures (Wösten and Willey, 2000).

The life cycle of filamentous fungi is very similar to that of the streptomycetes. They also form hydrophobic reproductive structures (e.g. aerial hyphae, fruiting bodies such as mushrooms, and spores) with rodlet-decorated surfaces. In this case, a film of highly insoluble self-assembled class I hydrophobin is responsible for this typical ultrastructure and the surface hydrophobicity (Wösten et al., 1993; 1994). These proteins also mediate attachment to hydrophobic solids (Wösten et al., 1993; 1994), such as to the surface of a host of a pathogenic fungus (Talbot et al., 1993; 1996). The latter is important during the initial stages of infection.

We adopted here the protocol used previously to extract class I hydrophobins selectively from fungal aerial structures (Wessels et al., 1991a,b; de Vries et al., 1993). This protocol is based on the insolubility of self-assembled class I hydrophobins in hot 2% SDS and their solubility in trifluoroacetic acid (TFA). Using this method, we identified two homologous proteins, designated rodlins (Rdl), that are involved in the formation of the rodlet layer on aerial structures of S. lividans and S. coelicolor and that also mediate attachment to hydrophobic surfaces.

Results

Identification of an abundant SDS-insoluble cell wall protein specifically present in aerial structures of S. coelicolor and S. lividans

Cell walls from 5-day-old sporulating cultures of S. coelicolor and S. lividans grown on solid medium were treated with 2% SDS at 100°C. The SDS-extracted cell walls were washed with water, lyophilized and extracted with TFA. SDS–PAGE of the SDS-soluble fraction showed a complex pattern of polypeptides (results not shown). Among the cell wall proteins of S. coelicolor and S. lividans that were insoluble in hot SDS but soluble in TFA, an abundant polypeptide, called Rdl, was observed with an apparent molecular weight of 18 kDa (Fig. 1A, lane 2). This protein was absent in a TFA extract of SDS-treated cell walls from a 3-day-old liquid shaken culture of S. lividans (Fig. 1A, lane 1). Western analysis with anti-bodies raised against Rdl showed the absence of Rdl in the SDS-soluble fraction of cultures of both streptomycetes grown in liquid or on solid medium (results not shown).

Figure 1.

Rdl is a protein present in cultures of S. coelicolor and S. lividans forming aerial structures.

A. SDS–PAGE analysis of TFA extracts of SDS-treated cell walls from a 3-day-old liquid shaken culture (lane 1) and a 5-day-old sporulating culture of S. lividans (lane 2). Rdl extracted with TFA from the latter culture is the main protein that is soluble in water (lane 3). Arrow indicates the position of Rdl. Molecular weights are given in kDa.

B. Western analysis of TFA extracts of SDS-treated walls from 1-day-old (lane 1), 2-day-old (lane 2), 3-day-old (lane 3), 4-day-old (lane 4) and 7-day-old (lane 5) cultures of S. lividans grown on solid medium using antibodies raised against Rdl. Cultures formed aerial hyphae at day 2 and started to sporulate at day 3. Similar results were obtained with S. coelicolor.

The presence of Rdl correlated with the presence of aerial hyphae. Rdl was absent in cell walls from 1-day-old surface-grown cultures that had not yet formed aerial hyphae (Fig. 1B). Similarly, Rdl was absent in cell walls from 1- to 7-day-old cultures of the bld mutants bld261, bldD and bldH of S. coelicolor grown on solid or liquid medium (results not shown). In contrast, the protein was abundantly present in cultures of S. lividans TK23 that had formed a confluent layer of aerial hyphae on solid medium after 2 days of growth. The amount of Rdl did not change during the following 5 days when aerial hyphae differentiated further to form chains of spores (Fig. 1B). Similar results were obtained with S. coelicolor M145 and S. coelicolor whiG, although aerial hyphae formation was delayed compared with S. lividans (see below).

When water was added instead of 2% SDS to a TFA extract of SDS-treated cell walls from a culture forming aerial hyphae, Rdl was the main protein that dissolved (Fig. 1A, lane 3). However, the protein formed an SDS-insoluble complex upon shaking. The complex could be dissociated with TFA. These data indicate that, under physiological conditions, Rdl is an SDS-insoluble cell wall protein present in cultures of S. coelicolor and S. lividans forming aerial structures.

Cloning and characterization of the rdl genes of S. coelicolor and S. lividans

N-terminal sequencing revealed that the Rdl protein band running at the 18 kDa position was in fact a mixture of two similar proteins, called RdlA and RdlB, with slightly different N-termini. In addition, the N-termini of two internal peptides were determined that resulted from a tryptic digestion of a mixture of RdlA and RdlB. A radioactive degenerated oligonucleotide based on one of the peptides was used to screen the cosmid library of S. coelicolor (Redenbach et al., 1996). The oligonucleotide hybridized to the overlapping cosmids C46 and C61. The hybridizing fragment of C46 was contained on a 4.5 kb SalI fragment. This fragment was cloned in the pBluescript KS+ SalI site and partially sequenced. An open reading frame (ORF) was identified that encodes a putative polypeptide of 131 amino acids, starting with a putative signal sequence for secretion of 28 amino acids followed by a sequence corresponding to the determined N-terminal sequence of RdlA as found in the cell wall (mature RdlA). The N-terminal sequences of both internal peptides were also identified in the ORF. The rdlB gene, divergently transcribed from rdlA, was identified 262 bp upstream of the start codon of rdlA. It encodes a protein very similar to that encoded by rdlA (68.7% identity, 83.2% similarity; see accession numbers AJ315950 and AJ315951) and contains the determined N-terminus of mature RdlB preceded by a putative signal sequence of 28 amino acids.

The coding sequences of rdlA and rdlB hybridized to the same unique fragments of genomic DNA of S. coelicolor and S. lividans digested with a variety of enzymes. For instance, a 4.5 kb SalI fragment of genomic DNA from S. coelicolor and S. lividans hybridized to both rdlA and rdlB. A slightly larger genomic fragment hybridized after digestion with BlpI, whereas digestion with PstI resulted in a fragment of about 8 kb (data not shown). The complete genome sequence of S. coelicolor (http://www.sanger.ac.uk/Projects/S_coelicolor/) did not reveal other sequences homologous to rdlA and rdlB.

Using polymerase chain reaction (PCR) and primers based on rdlA and rdlB of S. coelicolor, the homologues of S. lividans were isolated. Their sequences were identical to those of S. coelicolor. Genomic DNA from Streptomyces tendae, Streptomyces griseus and the potato pathogen Streptomyces scabies hybridized with probes directed against the coding sequences of rdlA and rdlB (data not shown). Genomic DNA from actinomycetes not belonging to the streptomycetes, i.e. Amycolatopsis mediterranei and Rhodococcus erythropolis, did not hybridize with either probe, even under low stringency (data not shown).

rdlA and rdlB are expressed in aerial hyphae

Total RNA was isolated from cultures of S. coelicolor M145 and S. lividans TK23 grown in liquid or solid medium. After separation on a formaldehyde gel and blotting to a nylon membrane, RNA was hybridized with a probe representing the coding sequence of rdlA (Fig. 2) or rdlB (not shown).

Figure 2.

Temporal expression of rdlA in S. lividans and S. coelicolor grown on solid medium. Northern blots were rehybridized with 16S rDNA to confirm that lanes contained equal amounts of RNA. Identical results were obtained using the coding sequence of rdlB as a probe.

Accumulation of mRNA from the rdl genes in S. lividans was only observed at day 2, coinciding with the formation of a confluent layer of aerial hyphae (Fig. 2). No accumulation of rdl mRNAs was observed in 1-day-old cultures growing submerged only or in 3-day-old sporulating cultures. mRNA from the rdl genes in S. coelicolor accumulated at days 2–4. Formation of aerial hyphae in this streptomycete was delayed compared with that in S. lividans. It started at day 2, but a confluent layer was not observed until day 4. As a consequence, the formation of aerial hyphae and sporulation partially overlapped (Fig. 2). From these data and the fact that accumulation of rdl mRNA was not observed in liquid shaken cultures throughout growth (not shown), we conclude that rdl expression correlated with the formation of aerial hyphae.

To determine the spatial expression of rdlA and rdlB, both orientations of the 262 bp intergenic region of the rdl coding sequences were cloned in vector pIJ8630 in front of the coding sequence of an enhanced green fluorescent protein (eGFP) with an adapted codon usage for S. coelicolor and S. lividans (Sun et al., 1999). This resulted in plasmids pIJ8630a and pIJ8630b. Spores from wild-type strains of S. coelicolor and S. lividans and transformants containing either construct were inoculated as a lawn on an object glass with a thin layer of agar medium. It appeared that wild-type strains of S. coelicolor were highly autofluorescent, but autofluorescence of S. lividans was negligible. Fluorescence in colonies of S. lividans transformed with either eGFP construct was observed at the outer part of the colony after 2 days of growth, correlating with the area in which aerial hyphae were formed (Fig. 3A), and was absent in wild type (Fig. 3B). When growth was prolonged, fluorescence in this zone decreased to wild-type levels but increased in the central zone, coinciding with the formation of aerial hyphae. At higher magnification, it was observed that aerial hyphae but not submerged hyphae were fluorescent (Fig. 3E). No fluorescence was observed in the wild-type strain at this magnification (Fig. 3F). These results show that the rdl genes are expressed in developing aerial hyphae.

Figure 3.

Spatial expression of rdlA in 2-day-old-cultures of S. lividans grown on solid medium visualized using eGFP as a reporter. Fluorescence of GFP was observed in transformants in a zone where aerial hyphae were formed (A), but not in a wild-type colony (B). At higher magnification, light microscopy showed a dense mycelium in the zone where the transformant (C) and the wild type (D) formed aerial hyphae. However, fluorescence of GFP was restricted to aerial hyphae of transformants being focused in the upper plane of the sample (E), wild-type aerial hyphae serving as a negative control (F). Identical results were obtained with an S. lividans strain expressing eGFP under the control of the rdlB promoter. Bar represents 500 μm (A and B) and 25 μm (C and D).

RdlA and RdlB are localized at the outer surface of aerial hyphae and spores

RdlA and RdlB were localized using an antiserum raised against a mixture of RdlA and RdlB from S. lividans. Immunolabelling was observed at the outer surface of aerial hyphae and spores of S. lividans and S. coelicolor (Fig. 4B). Some label was also found within the cell walls of the aerial structures. The reactive layer at the outer surface was sometimes detached, indicating that it is a discrete layer. The antiserum reacted neither with submerged hyphae of wild-type strains of S. coelicolor and S. lividans (Fig. 4A) nor with hyphae of the bld261, bldD and bldH mutants of S. coelicolor. In contrast, aerial hyphae of a whiG mutant of S. coelicolor were labelled (data not shown).

Figure 4.

Immunolocalization of RdlA and RdlB in submerged (A) and aerial (B) hyphae of S. coelicolor. Bars represent 0.5 μm.

Disruption of rdlA and rdlB does not affect the formation of aerial structures but does affect the formation of the rodlet layer

As expression profiles of rdlA and rdlB were similar, these genes may be redundant. Therefore, both genes were inactivated in S. coelicolor M145 and S. lividans TK23 using deletion construct pC46d. The complete coding sequence of rdlA, most of the coding sequence of rdlB and the intergenic region were replaced by a hygromycin B resistance cassette. Gene replacement was confirmed by Southern analysis. To exclude interference from the replacement of rdlA and rdlB with transcription of upstream and downstream genes, Northerns were probed with ORF SCC46.02c located 288 bp upstream of rdlA and rdlB and ORF SCC46.05c located 89 bp downstream of these genes. Accumulation of mRNA was similar in wild-type and disruptant strains grown on solid media.

Germination of spores, growth rates and differentiation of aerial hyphae into spores were similar in wild-type and ΔrdlAB strains using a variety of media and culture conditions (data not shown). In addition, no difference could be observed in the viability of spores after freeze-drying or drying spores in the air. Surface hydrophobicity was also unaffected (van der Mei et al., 1991). Wild-type strains of S. coelicolor and S. lividans showed water contact angles of 124 ± 5°, whereas those of disruptant strains were 133 ± 6° and 126 ± 3° respectively.

To analyse whether the disruption of rdlA and rdlB affects the formation of the rodlet layer at surfaces of aerial hyphae and spores, wild-type and ΔrdlAB strains were analysed using scanning electron microscopy. In contrast to the wild-type strains, no rodlets were observed at the surfaces of aerial hyphae and spores from S. coelicolorΔrdlAB6 (Fig. 5) and S. lividansΔrdlAB3 (data not shown). Integration of the 4.5 kb SalI fragment encompassing both rdl genes into the genomic attP site of the null mutants of S. coelicolor and S. lividans restored rodlet formation (see Fig. 5).

Figure 5.

Scanning electron microscopy revealed that rodlets are present on spores of a wild-type strain of S. coelicolor (A) but are absent in S. coelicolorΔrdlAB6 (B). Rodlet formation was restored in a genetically complemented strain of S. coelicolorΔrdlAB6 (C). Similar results were obtained with S. lividansΔrdlAB3. Bars represent 100 nm.

Disruption of the rdl genes affects the attachment of hyphae to polystyrene

Expression of rdlA and rdlB in hyphae confronted with a hydrophobic solid was studied by growing S. lividans strains transformed with plasmid pIJ8630a or pIJ8630b (see above) in 96-well plates in liquid medium without shaking, followed by analysis of GFP expression. Under this condition, no autofluorescence was observed. Hyphae not in contact with the hydrophobic surface of the microtitre plate were not fluorescent throughout culturing (Fig. 6C). In contrast, hyphae in contact with the solid did express eGFP (Fig. 6D).

Figure 6.

S. coelicolor wild-type strain M145 attaches more strongly to a polystyrene microtitre plate (A) than strain ΔrdlAB6 (B), as visualized by staining attached hyphae with crystal violet. Similar results were obtained with S. lividans. Expression of rdlA and rdlB in S. lividans during attachment was studied using eGFP as a reporter. Fluorescence was not observed in hyphae that were loosely, if at all, attached (C), whereas GFP accumulated in hyphae firmly attached to the polystyrene surface (D). Bar represents 25 μm.

A role for RdlA and RdlB in attachment was studied by growing cultures in microtitre plates, followed by staining with crystal violet and thorough washing to remove all unattached cells. Throughout culturing, attachment of S. coelicolorΔrdlAB6 was only 10–50% compared with that of the wild-type strain (Fig. 6A and B). Similar results were obtained with S. lividans (data not shown). Attachment of the ΔrdlAB strains could not be restored by integrating the 4.5 kb SalI fragment encompassing both rdl genes into the genomic attP site. As the reason for this was not clear, two additional independent null mutants were analysed for their capacity to adhere to the microtitre plate. Similar results were obtained to those with ΔrdlAB6 and ΔrdlAB3 confirming that the rodlins are involved in attachment.

Discussion

The lifecycle, the mode of growth and the ecological niches of streptomycetes are remarkably similar to those of filamentous fungi. Yet, these microbes belong to different kingdoms that diverged early in evolution. Spores of both groups germinate and form a mycelium that colonizes moist substrates. This mycelium consists of filaments that are surrounded by rigid walls and grow at their apices. After a submerged feeding mycelium has been established, filaments may leave the substrate to form spore-bearing aerial structures. The aerial structures of most species are hydrophobic and characterized by rodlet-decorated surfaces.

The formation of aerial hyphae has been described as a two-step process (Wösten et al., 1999). Although oversimplified, given the genetic complexity of this differentiation process, this model is a means to begin to understand aerial growth. In the first step, the water surface tension is dramatically reduced from 72 to 32 mJ m−2, enabling hyphae to breach the colony surface–air interface (Wösten et al., 1999). In the second step, the aerial hyphae are coated with a hydrophobic rodlet layer. In fila-mentous fungi it has been shown that hydrophobins both lower the surface tension and form the rodlet-decorated hydrophobic coating (Wösten et al., 1993; 1994a; 1999; 2001). Filamentous bacteria appear to have evolved different molecules to lower the surface tension and to coat the aerial hyphae (Wösten and Willey, 2000). S. coelicolor lowers the water surface tension by secreting a small surface-active peptide called SapB (Tillotson et al., 1998; Willey et al., 1991; 1993). We identified here the proteins that form the rodlet layer. So far, this is the first example of structural proteins coating aerial structures of filamentous bacteria. These proteins, called rodlins, were isolated adopting the procedure used selectively to extract the hydrophobins from cell walls of fungal aerial structures. Despite their remarkable resemblance in solubility characteristics, rodlins are not related to the fungal hydrophobins. Apparently, distinct proteins can form a surface layer with a similar ultrastructural appearance. The rodlet layers found in streptomycetes are probably all formed by rodlins, as the encoding genes in S. lividans and S. coelicolor, rdlA and rdlB, hybridized to genomic DNA from five different streptomycetes representing the phylogenetic diversity of this group of bacteria.

Inactivation of the SC3 hydrophobin gene in S. commune affected the formation of aerial hyphae. Those aerial hyphae formed were hydrophilic (van Wetter et al., 1996). In contrast, deletion of both rdl genes in S. coelicolor and S. lividans neither affected the formation of aerial hyphae nor surface hydrophobicity. Apparently, the rodlet layer is not involved in the formation of aerial hyphae. This can be explained by the fact that SapB mediates escape of hyphae into the air by lowering the water surface tension whereas, apart from the rodlin layer, other layers render aerial hyphae hydrophobic. Hydrophobins have been shown to mediate the attachment of fungal hyphae to hydrophobic surfaces (Wösten et al., 1994b), such as the hydrophobic surface of a host of a plant pathogen (Talbot et al., 1993; 1996). Attachment to a hydrophobic surface was also strongly decreased in ΔrdlAB strains of S. coelicolor and S. lividans. Yet, by expressing the rdl genes at the attP locus in the genome, attachment could not be complemented. The reason for this is not yet clear but may result from different expression levels at this ectopic site, interfering with the proper formation of the attaching layer. Adhesion of streptomycetes to hydrophobic surfaces may play a role during invasive growth of wood being rich in hydrophobic lignin. In pathogenic streptomycetes (e.g. the potato pathogen Streptomyces scabies), homologues of RdlA and RdlB may be instrumental in pathogenicity by attaching the pathogen to the host.

Experimental procedures

Strains and plasmids

Escherichia coli strains DH5α or JM110 were used for cloning purposes. S. coelicolor strains M145 (Kieser et al., 2000), J1700 (hisA1, uraA1, strA1, bldA39, Pgl) (Piret and Chater, 1985), J774 (mthB2, cysD18, pheA1, agaA7, strA1, bldD, NF, SCP2) (Merrick, 1976), C109 (hisA1, uraA1, strA1, bldH109, Pgl) (Champness, 1988) and J1820 (hisA1, uraA1, strA1, whiG71, Pgl) (Méndez and Chater, 1987) were used as well as S. lividans TK23 (Kieser et al., 2000), Streptomyces tendae Tü901/8c (Richter et al., 1998) and Streptomyces griseus (DSMZ 40236). Chromosomal DNA from Streptomyces scabies was kindly provided by Professor E. M. H. Wellington (University of Warwick, UK). Vectors and constructs are summarized in Table 1.

Table 1. Vectors and constructs.
PlasmidDescriptionReference
pIJ8600 E. coli–Streptomyces shuttle vector with pUC18 ori, oriT and attP site. It integrates at the ϕC31
attachment site in S. coelicolor and S. lividans
Sun et al. (1999)
pIJ8630 E. coli–Streptomyces shuttle vector with pUC18 ori, oriT and attP site containing an eGFP gene
adapted for codon usage in streptomycetes. It integrates at the ϕC31 attachment site in
S. coelicolor and S. lividans
Sun et al. (1999)
pIJ8630apIJ8630 containing the 262 bp S. coelicolor promoter region of rdlA with an NdeI site at the 3′ end
allowing translational fusions
This work
pIJ8630bAs pIJ8630a but with the promoter region of rdlBThis work
pC46apBluescript KS+ (Stratagene) containing a 4.5 kb SalI fragment of cosmid C46 (Redenbach et al.,
1996) of S. coelicolor encompassing the coding sequences of rdlA and rdlB, their interspersed
promoter region and flanking regions of 0.8 kb (rdlA) and 2.5 kb (rdlB)
This work
pC46bpZErO-2.1 (Invitrogen) containing the SalI fragment described for pC46aThis work
pC46cpC46b derivative carrying a 1.4 kb SmaI fragment containing a hygromycin resistance cassette
(Zalacaín et al., 1986) replacing a 0.8 kb BlpI–ScaI fragment encompassing rdlA, the 5′ end of
the coding sequence of rdlB as well as their interspersed promoter region
This work
pC46dpC46c derivative with a 1.8 kb SmaI fragment containing an apramycin resistance cassette
(Prentki and Krisch, 1984) cloned in the XbaI site of pC46c
This work
pSET-C46apSET152 derivative (Bierman et al., 1992) containing the 4.5 kb SalI fragment described for pC46aThis work

Growth conditions and media

Streptomyces strains were grown at 30°C on solid MS agar medium or in YEME medium as liquid shaken cultures (Kieser et al., 2000). The solid medium R2YE (Kieser et al., 2000) was used for regenerating protoplasts. To assess attachment to a hydrophobic solid, S. coelicolor and S. lividans were grown in NMMP (Kieser et al., 2000) in the absence of PEG 6000 and using 50 mM glucose as a carbon source in 96-well flat-bottomed microtitre plates (Costar, Corning). Before inoculation, spores, stored at –20°C in 20% glycerol, were taken up in NMMP to a final concentration of 5 × 106 spores ml−1. Flat-bottomed 96-well microtitre plates were filled with 200 μl of spore suspension per well.

Molecular techniques

Standard molecular techniques followed the methods described by Sambrook et al. (1989). Protoplast preparation and transformation were performed as described by Kieser et al. (2000) using alkali-denatured DNA (Oh and Chater, 1997). Chromosomal DNA of S. coelicolor and S. lividans was isolated according to the method of Verhasselt et al. (1989) and modified by the method of Nagy et al. (1995). Total RNA of S. coelicolor and S. lividans was isolated using the SV Total RNA isolation system (Promega) according to the method of Veenendaal and Wösten (1998). DNA and RNA were blotted on nylon filters (Boehringer Mannheim) and hybridized under conditions described by Church and Gilbert (1984) at 60°C. Under these conditions, rdlA and rdlB do not cross-hybridize. Radioactively labelled probes were made using the oligolabelling kit (Pharmacia).

Isolation of the rdlA and rdlB genes from S. coelicolor and S. lividans

To isolate rdl genes from S. coelicolor M145 and S. lividans TK23, a degenerate oligonucleotide (SGCSGASAGSACS GASAGGTCCTCSAGSACGTGSGASAGSGCGCCGTC) representing the N-terminal sequence of the carboxy-terminal internal peptide of RdlA of S. lividans (see Results) was radioactively labelled and hybridized to the cosmid library of S. coelicolor (Redenbach et al., 1996). Accession numbers for rdlA and rdlB are AJ315950 and AJ315951 respectively.

Construction of the ΔrdlAB gene deletion plasmid pC46d

The rdl genes were deleted by replacing a 0.8 kb BlpI–ScaI fragment of pC46b (see Table 1), containing the entire coding sequence of rdlA and 136 bp of that of rdlB as well as the interspersed promoter region, with a 1.4 kb SmaI fragment encompassing the hygromycin B resistance cassette (Zalacaín et al., 1986). This resulted in vector pC46c. To select for double cross-over events, plasmid pC46d was made by introducing a 1.8 kb SmaI fragment containing an apramycin resistance cassette (Prentki and Krisch, 1984) in the XbaI site of pC46c.

Preparation of cell walls and protein extracts

Filaments of S. coelicolor and S. lividans were fragmented at 20 000 p.s.i. using an SLM French pressure cell press. The homogenate was treated with 2% SDS for 10 min at 100°C, after which the cell walls were fractionated from the cytoplasmic content by centrifugation at 10 000 g for 10 min. The cell wall fraction was extracted with hot 2% SDS once more, washed extensively with water and freeze dried. SDS-treated cell walls were then extracted with TFA (Wösten et al., 1993). After evaporating the solvent by a stream of air, extracts were taken up in SDS sample buffer (2% SDS, 20% glycerol, 0.02% bromophenol blue, 0.1 M Tris-HCl, pH 6.8, and 5%β-mercaptoethanol) and subjected to SDS–PAGE. If necessary, adjustments in pH were done by the addition of 25% ammonia. RdlA and RdlB were purified by taking up TFA extracts of SDS-treated cell walls in water without shaking. Insolubles were removed by centrifugation at 10 000 g for 15 min.

Gel electrophoresis and Western blotting

SDS–PAGE was performed in 16% gels according to the method of Laemmli (1970). Prestained broad-range molecular weight markers from Bio-Rad were used. After separation, proteins were stained with 0.25% Coomassie brilliant blue G-250 (CBB) or blotted onto a polyvinylidene difluoride (PVDF) membrane using semi-dry blotting. For N-terminal sequencing, a PVDF membrane was stained with CBB, and a slice of the membrane containing the protein was excised. After destaining with 30% methanol, the N-terminal sequence was determined using a pulse liquid sequencer on line connected to a PTH analyser (Eurosequence). To determine N-terminal sequences of internal peptides, the protein was eluted from the SDS–PAA gel followed by tryptic digestion. Peptides were sequenced after separation on a C18 reversed phase high-performance liquid chromatography (HPLC) column.

Polyclonal antibodies were raised against a mixture of the S. lividans RdlA and RdlB proteins, eluted from an SDS–PAA gel. Antibodies were purified with an acetone powder of mycelium from a liquid shaken culture (Harlow and Lane, 1988). PVDF membranes were treated with diluted anti-RdlA/RdlB serum (1:1000) as described previously (Harlow and Lane, 1988).

Immunolocalization

Fixation, embedding and immunolabelling of cultures were performed as described previously (Wösten et al., 1994) with the modification that K4M was substituted for Unicryl. Sectioned material was examined in a Philips CM10 electron microscope. Photographs were made on FGP Kodak film.

Other electron microscopic techniques

For cryo scanning electron microscopy (SEM), sporulating cultures grown on solid MS medium were frozen in a mixture of liquid and solid nitrogen and sputter coated with gold– palladium. Examination was done at 5.0 kV in a Jeol field emission scanning electron microscope type 6301F.

Attachment assay

To quantify attachment of S. coelicolor and S. lividans to the surface of polystyrene microtitre plates, 25 μl of 0.5% crystal violet (Acros Organics) was added to each well and incubated for 10 min to stain cell material. Wells were washed vigorously with water using a Vaccu-Pette/96 (Sigma), removing all non-adherent cells. After drying overnight at 30°C, the crystal violet associated with the attached material was dissolved in 200 μl of 10% SDS (Reynolds and Fink, 2001) for 30 min under shaking conditions (900 r.p.m.). A sample of 100 μl was transferred to a new well to determine the OD570 in a microtitre plate reader. If necessary, dilutions were made in 10% SDS. Total biomass was determined using the DC protein assay (Bio-Rad) after treating the material at 100°C for 30 min in 0.2% SDS–1 M NaOH. Bovine serum albumin (BSA) was used as a standard.

Acknowledgements

We are indebted to Ietse Stokroos for performing the scanning electron microscopy, and Heine Deelstra, Jan de Jong and Nynke Penninga for their work on the presence of rodlins in developmental mutants of S. coelicolor and in other streptomycetes and actinomycetes. Moreover, we are indented to Dr O.M.H. de Vries for expert advice.

Ancillary