Complex extracellular interactions of proteases and a protease inhibitor influence multicellular development of Streptomyces coelicolor



Streptomyces coelicolor produces an extracellular protease inhibitor protein, STI (Streptomyces trypsin inhibitor). We show that post-growth elimination of STI is needed for colonies to develop aerial mycelium efficiently. Inactivation of STI, and thus the normal progression of colony development, at least partly involves an extracellular protease specified by gene SCO5913. Two-hybrid analysis identified two possible targets of STI inhibition (the products of SCO1355 and SCO5447), both extracellular proteases containing a domain homologous with the P-domain of eukaryotic convertases, proteases that mediate the processing of many precursors with important cellular or developmental roles. At least the SCO1355 protease is needed for the normal progression of development. Two components of the proposed cascade are dependent on the tRNA for the rare UUA (leucine) codon, which is specified by the developmental gene bldA. A model is presented that links intracellular regulatory events with an extracellular protease cascade to facilitate normal development.


Streptomycetes are bacteria in which substrate mycelium gives rise to sporulating aerial mycelium, typically associated with the production of diverse secondary metabolites. This complex differentiation is controlled by a network involving bld and whi regulatory genes, and morphology-conferring proteins that aid the growth of reproductive hyphae into the air (Kelemen and Buttner, 1998; Chater, 2001; 2006; Elliot et al., 2003; Claessen et al., 2004; 2006; Kodani et al., 2004; Chater and Chandra, 2006; Capstick et al., 2007). Streptomycetes are adapted to grow in the soil on complex, often insoluble, organic polymers accessible only to organisms having suitable extracellular enzymes such as proteases, amylases, chitinases, etc. Here we focus on a role of proteases in differentiation.

The morphological development of streptomycetes is believed to be influenced by an extracellular proteolytic cascade, including the interaction of extracellular protease inhibitors with target serine proteases. For example, in Streptomyces exfoliatus SMF13, sporulation is regulated by a trypsin-like protease (TLP), whose activity is inhibited by the peptide leupeptin until late in development, when leupeptin is degraded by a leupeptin inactivation enzyme (Kim and Lee, 1995; 1996; Kim et al., 1998). In Streptomyces albidoflavus SMF301, one of a minority of streptomycetes that sporulates in submerged culture, mycelial growth is linked with the production of a chymotrypsin-like protease, while submerged spore formation is accompanied by the production of TLP and metalloprotease (MTP) (Kang et al., 1995). Furthermore, aerial mycelium growth and sporulation are supported nutritionally by the utilization of substrate mycelium, involving both MTP and TLP (Kang et al., 1995). In Streptomyces antibioticus, sporulation is regulated in part by nucleases, one of which is activated through processing by a serine protease (Nicieza et al., 1999). In Streptomyces griseus, several proteases and a protease inhibitor (SgiA) are regulated by the A factor-dependent protein AdpA, a key regulator of morphological and physiological differentiation (Kato et al., 2002; 2005a; Tomono et al., 2005; Hirano et al., 2006). Of these, the MTP SgmA partially contributes to development, while several trypsin and chymotrypsin proteases could be disrupted with no obvious morphological consequences. Despite all these intriguing observations, the complex roles and interactions of proteases and protease inhibitors during Streptomyces differentiation have not been characterized at the molecular level, and the connection between intracellular and extracellular events is not well defined.

In the model species Streptomyces coelicolor, expression of a secreted protease inhibitor STI (Streptomyces trypsin inhibitor) is completely dependent on bldA (Kim et al., 2005). Although dispensable for vegetative growth, bldA is required for morphological and physiological differentiation, encoding the only tRNA that can translate efficiently the rare leucine codon UUA (Lawlor et al., 1987; Leskiw et al., 1991). The dependence of both STI and morphological differentiation on bldA is mainly mediated via a TTA codon in the S. coelicolor orthologue of adpA (originally called bldH) (Nguyen et al., 2003; Takano et al., 2003). STI is homologous to other conserved Streptomyces subtilisin inhibitor family proteins, including SgiA of S. griseus (Hirano et al., 2006).

In this work, we have used genetic manipulation, fermentation and protein interaction studies to characterize the role of STI in the regulation of differentiation in S. coelicolor, and to find secreted proteases that form a part of the same extracellular regulatory pathway with STI. A model in which STI and the proteases participate in the regulation of morphological and physiological differentiation in S. coelicolor is proposed, representing a new scenario in the extracellular biology of Streptomyces spp.


STI production depends on bldA, is growth-associated and controls mycelium degradation

In S. coelicolor, SCO0762 (sti) encodes a putative extracellular protease inhibitory protein STI, predicted to be active against serine proteases. Expression of sti depends on the regulatory protein AdpA, which in turn is translationally dependent on the UUA-reading tRNA encoded by bldA (Kim et al., 2005). This suggested that the morphological defects of bldA mutants might be at least partially attributed to changes in the abundance and activities of extracellular proteases. As a starting point for the investigation of this possibility, the production of STI and proteases by a bldA null mutant and its parent strain (M600) were analysed in submerged culture using jar fermentors. Glucose consumption and mycelium growth under these conditions were more rapid in the bldA null mutant than in the parent strain (Fig. 1): the specific glucose uptake rate (qs) and specific mycelium growth rate (μ) were about four times higher in the mutant (Fig. S1, Table S1). In the parent, STI activity increased from the start of mycelium growth, peaking at 20 h, the specific rate of STI production (qSTI) being inversely related to the specific growth rate (μ) (Fig. S1). As mycelial growth ceased, STI began to decrease, becoming undetectable by 90 h. No significant mycelial degradation was observed. In contrast, production of STI was abolished in the bldA null mutant (Fig. 1C), and rapid degradation of the mycelium immediately followed glucose depletion (Fig. 1): the specific mycelium death rate (kd) was 10 times higher for the mutant than for the parent (Table S1). Total extracellular protease activity was higher in the bldA null mutant than the parent strain (Fig. 1D). These results suggested that the high extracellular protease activity in the bldA mutant might contribute to mycelial degradation, as well as to the low level of extracellular proteins in this strain (Kim et al., 2005).

Figure 1.

Changes in glucose concentration (A), mycelium growth (B), STI production (C) and total activity of extracellular proteases (D) in submerged cultures of S. coelicolor M600 (open circles) and M600 ΔbldA (closed circles). The strains were cultured using SMM-CA medium in aerated (1 v.v.m.) and stirred (250 r.p.m.) jar fermentors.

STI activity depends on post-translational regulation by SCO5913 protease

In other Streptomyces spp., an extracellular protease inhibitor is inactivated by an extracellular protease at the time of morphological differentiation (Kang et al., 1995; Kim et al., 1998). In order to identify protease(s) that cause the loss of STI in S. coelicolor, six of 56 genes encoding predicted extracellular proteases in the S. coelicolor genome (ScoDB, were initially chosen for disruption, based on the resemblance of their gene products to proteases previously found to have developmental significance in other streptomycetes, or on the possession of a TTA codon in the case of SCO5913 (Table S2). One mutant, carrying a deletion in SCO5913 (SMF5434), showed a slight delay in differentiation. To find out if SCO5913 protease affected STI activity in vivo, we compared fermentation kinetics in submerged cultures of the M600 parental strain, the sti null mutant (SMF5428), and the SCO5913 null mutant. The patterns for glucose uptake and mycelium growth were similar in all three strains to the results shown in Fig. 1 for M600 (data not shown). STI activity was monitored in submerged cultures and solid cultures. No STI activity could be detected in the sti null mutant, confirming that STI is the sole source of the protease inhibitory activity under these growth conditions, while STI activity was higher and longer-lived in the SCO5913 null mutant (Fig. 2) (these results were confirmed and extended in subsequent experiments – see Fig. 4 below). These results indicate that SCO5913 protease is responsible for some inactivation of STI. However, STI eventually disappeared even from the SCO5913 null mutant, suggesting that some other protease(s) also contributed to STI inactivation. Despite the disappearance of STI activity in the later samples of both strains, sti mRNA was readily detected in both strains right up to 144 h, indicating post-translational regulation of STI. No sti mRNA was detectable in the sti null mutant, confirming that the transcript analysis was specific for sti (Fig. 2).

Figure 2.

sti transcription and STI activity during development of surface-grown cultures. Strains M600 (parent strain), sti null mutant and SCO5913 null mutant were cultured at 30°C on SMM-CA agar plates with and without cellophane overlays. sti transcript abundance was analysed by reverse transcriptase PCR using purified total RNA isolated from mycelium, and hrdB was used as the reference gene. STI activity was measured from extracts of total mycelium grown on the plates.

Figure 4.

STI activity inhibits actinorhodin production in strains of S. coelicolor. M600 (parent strain: open circles), the sti null mutant (closed circles), the SCO5913 null mutant (open squares) and the STI-overproducing strain (closed squares) were grown at 30°C in SMM-CS medium using jar fermentors with aeration (1 v.v.m.) and agitation (250 r.p.m.).

The disappearance of STI activity in the parent strain after 48 h coincided with a switch from vegetative to aerial growth. The confluent areas of the SCO5913 null mutant did not begin to differentiate until after 96 h, when the elevated levels of STI in this strain had subsided. This suggests that SCO5913 protease inactivates STI, and that high STI abundance is associated (possibly causally) with delayed aerial growth. A role for SCO5913 protease in STI inactivation was further indicated in an experiment in which expression of SCO5913 protease was induced in cultures of SMF5402 (harbouring ermEp-sti and tipAp-SCO5913) by adding 25 μg ml−1 thiostrepton after 18 h of growth: highly expressed STI activity of this strain disappeared within 30 h, whereas it remained high in the uninduced cultures (Fig. 3A). Degradation of STI by SCO5913 was confirmed by in vitro analysis using purified His-tagged STI and an untagged version of SCO5913, prepared by purifying glutathione S-transferase (GST)-tagged protein followed by selective thrombin cleavage of the tag. Incubation of STI with SCO5913 induced degradation of STI such that over a 3 h period, 1.5 μg of STI was degraded by 0.1 μg of SCO5913 to levels undetectable by Coomassie blue brilliant staining (Fig. 3B). Trace amounts of thrombin at levels expected to be carried over from removal of the GST-tag from SCO5913 did not degrade STI, indicating that SCO5913 activity alone is responsible for the observed degradation (data not shown).

Figure 3.

In vivo and in vitro analysis of STI inactivation by SCO5913 protease.
A. Induction of SCO5913 expression in vivo results in inactivation of STI. Plasmid pSMF5401 was constructed for overproduction of STI under the control of the strong and constitutive promoter (ermEp) and plasmid pSMF5402 was constructed for induction of gene SCO5913 expression by the thiostrepton-inducible promoter (tipAp). The strain SMF5402 (M600 carrying pSMF5401 and pSMF5402) was cultured at 30°C in SMM-CA using baffled flasks on rotary shaking incubators (230 r.p.m.). STI activity was measured in cultures that had been induced (closed circles) or not induced (open circles) by thiostrepton. Thiostrepton was added at 18 h to a final concentration of 25 μg ml−1.
B. SCO5913 degrades STI in vitro. His-tagged STI (1.5 μg) was incubated with purified SCO5913 (0.1 μg) at 30°C for the times shown. SDS-PAGE analysis shows the disappearance of STI protein from the samples with time.

STI has an inhibitory effect on production of a polyketide antibiotic

A comparison of submerged fermentations of the M600 parent strain with the sti and SCO5913 null mutants indicated that production of the blue antibiotic actinorhodin (Act) is in some way inhibited by STI activity in the cultures (Fig. 4). Act production was twofold higher in the sti null mutant (which lacks STI activity) than in the parent, but production was delayed by over 20 h in the SCO5913 null mutant, which exhibited abnormally high STI activity early in growth. In order to confirm the suggested inhibitory effect of STI activity on Act production, an STI-overproducing strain (SMF5401), which overexpresses STI from a strong and constitutive promoter (ermEp), was analysed. STI activity was significantly higher and Act production was markedly decreased at every time point compared with the parent strain M600 (Fig. 4).

Screening for a target protein interacting with STI by two-hybrid analysis

To search for proteases that may be inhibited by STI during the early development of colonies, two-hybrid analysis was performed. A cDNA library of S. coelicolor constructed in Saccharomyces cerevisiae strain AH109 was mated with a bait strain, Y187 harbouring pGBKT7-sti. Auxotroph selection and a colony-lift filter assay were used to select clones exhibiting evidence of a strong interaction. After quantification of the interactions using β-galactosidase assays with ortho-nitrophenyl-β-D-galactopyranoside (ONPG) as a substrate, 41 clones were selected that showed > 10% of the signal obtained with a known positive control interaction (pGADT7-T with pGBKT7-53). The cDNA insert sequences of the 41 pGADT7-cDNA clones were determined, and a total of nine candidates were characterized as potential interaction partners of STI. These included a putative neutral zinc MTP SCO5447, which possesses three predicted functional motifs/domains related to the function of thermolysin metallopeptidase (peptidase M4 propeptide motif, peptidase M4 and M4C domains), and a fourth C-terminal proprotein convertase P-domain that is characteristic of eukaryotic subtilisin-like proprotein convertase (Fig. 5A). Interestingly, the cDNA clone identified as positively interacting with STI in the two-hybrid screen encoded only amino acids 575–684 and thus contained only the P-domain of SCO5447, suggesting that this domain is by itself sufficient for the interaction. Figure 6A shows the interaction intensity between STI and the P-domain of SCO5447 by β-galactosidase assay. In the S. coelicolor genome, only SCO5447 and SCO1355 encode proteases annotated as containing a P-domain, and in each case the domain is located at the C-terminus (Fig. 5A). SCO1355 encodes a putative secreted serine protease with two catalytic domains (subtilisin N and peptidase S8) and a P-domain. Alignment of SCO1355 with SCO5447 shows significant similarity only in the highly homologous P-domain (55% identity and 72% similarity, Fig. 5B).

Figure 5.

A comparison of the proteases encoded by genes SCO5447 and SCO1355 showing the conserved C-terminal P-domain.
A. The proteases encoded by SCO5447 and SCO1335 both contain the conserved C-terminal P-domain that is the site of interaction with STI protein.
B. Similarity of the P-domains of SCO5447 and SCO1355 shown by amino acid alignment.

Figure 6.

Interaction between STI and proteases containing a P-domain.
A. ONPG was used as a substrate in a β-galactosidase assay to monitor expression of a lacZ reporter gene in a two-hybrid system. The interaction obtained when using the STI bait plasmid pGBKT7-sti, and the positive cDNA clone in pGADT7 containing a partial fragment of SCO5447 corresponding to the P-domain (aa 575–684) is shown to the right of the controls. The interaction between plasmid pGADT7-T and pGBKT7-53 is shown as positive control, with pGADT7-T and pGBKT7-lam being used as a negative control.
B. Interaction analysis between STI and SCO1355 was performed by auxotroph analysis of the two-hybrid system. Full-length SCO1355 (1355, aa 1–537), catalytic domain (CD, aa 1–426) and P-domain (PD, aa 427–537) were cloned into pGADT7-Rec (plasmid containing activation domain, AD) and pGBKT7 (plasmid containing binding domain, BD), and then co-transformed into S. cerevisiae AH109 strain with pGBKT7-sti and pGADT7-sti respectively. Interaction was examined in synthetic dropout media without leucine, tryptophan, histidine and adenine.
C. 1.5 μg of purified STI (His-tagged) was incubated with 0.1, 0.3 and 1.0 μg SCO1355 protein (GST-tagged), and analysed using native gel electrophoresis (left panel). Complex formation is indicated by the arrow. Incubation with GST-tagged SCO4677 protein (1.0 μg) was used as a negative control, and lanes analysing only 1.0 μg GST-tagged SCO1355 protein or 1.5 μg of His-tagged STI are also shown to indicate the absence in these samples of any protein corresponding to the position of the SCO1355-STI complex band. SDS-PAGE analysis of the excised complex band (right band) indicates that it is composed of the STI and SCO1355 proteins.
D. Interaction analysis of STI and SCO1355 protein by co-immunoprecipitation. His-tagged STI was mixed with GST-tagged SCO1355 protein, then immunoprecipitation performed using anti-GST antibody, as detailed in Experimental procedures. Western blot analysis of SDS-PAGE separated samples using anti-His antibody (top panel) and anti-GST antibody (bottom panel) indicates that His-STI co-purifies with GST-SCO1355. Analysis of His-STI and GST-SCO1355 standards is shown, and the immunoprecipitated sample is indicated as Co-IP. In a negative control experiment, His-tagged STI did not co-immunoprecipitate with only GST (data not shown).

STI interacts with SCO1355 in both in vivo and in vitro conditions

To find out if SCO1355 protease also interacts with STI via its P-domain, further two-hybrid analyses were performed. Full-length SCO1355 and a clone carrying only the P-domain gave a positive signal, while a clone carrying only the catalytic domain showed only weak evidence of an interaction signal (Fig. 6B). Thus, STI interacts with the P-domains of both SCO1355 and SCO5447, implicating the P-domain as a critical target of STI.

Interaction of STI and SCO1355 was confirmed by in vitro analysis. Purified STI (His-tagged) and SCO1355 (GST-tagged) were analysed in a gel binding assay using native PAGE. The formation of a complex of STI and SCO1355 was observed, and this increased as the concentration of SCO1355 was increased, keeping the concentration of STI fixed (Fig. 6C). STI did not interact with a negative control protein, GST-tagged SCO4677, indicating that the GST-tag plays no part in the observed interaction between STI and SCO1355. The presence of the two proteins in the complex band was verified by excising it from the gel, and subjecting it to analysis under denaturing conditions using SDS-PAGE (Fig. 6C). The in vitro interaction of STI and SCO1355 was also confirmed using a co-immunoprecipitation method. His-tagged STI co-purified with GST-tagged SCO1355 following immunoprecipitation using anti-GST antibody (Fig. 6D). As a negative control, it was shown that His-tagged STI was not similarly co-immunoprecipitated when GST was used in place of GST-SCO1355 (data not shown).

STI and proteases SCO5913 and SCO1355 influence differentiation

Isolated colonies of M600 (parent strain), sti null mutant, STI-overproducing strain and SCO5913 null mutant showed differences in the timing and pattern of switching from vegetative growth to the formation of aerial hyphae (Fig. 7). In the parent strain, aerial mycelium growth initiated from the centre of the colony after 3 days, and extended progressively to cover the entire colony by 5 days. However, in the sti null mutant, aerial mycelium covered the vast majority of the colony after only 3 days, while in the STI-overproducing strain, aerial mycelium formation was markedly delayed, covering less than 50% of the surface even after 9 days. The control strain harbouring only the empty pWHM3A-emEp vector differentiated normally (data not shown). The SCO5913 null mutant phenotype was intermediate between the parent strain and the STI-overproducing strain, reflecting elevated STI activity compared with M600, but lower activity than in the STI-overproducing strain (see Fig. 4). Aerial mycelium formation of the SCO5913 null mutant was largely restored by complementation with gene SCO5913 (data not shown).

Figure 7.

Phenotypic analysis of single colonies of M600 and its derivatives during growth on SMM-CA. Parent strain (M600), sti null mutant, STI-overproducing strain, SCO5913 null mutant and SCO1355 null mutant single colonies that were not influenced by any neighbouring colonies were selected and observed using an optical microscope.

Deletion of SCO1355 (SMF5435) caused markedly delayed aerial mycelium formation, especially of the outer region of colonies, similar to the STI-overproducing strain (Fig. 7). This implies that SCO1355 protease is a potential key extracellular target of STI in the earlier growth stages, and that an absence of SCO1355 activity resulting from mutational elimination of SCO1355 or its prolonged inhibition by STI delays aerial mycelium formation in the outer parts of the colony. Elimination of the other protease containing a P-domain by deletion of SCO5447 [which encodes an extracellular protease homologous to SgmA of S. griseus (Kato et al., 2002)] did not cause an obvious phenotypic change (Table S2). Interestingly, the S. coelicolor genome contains a neighbouring gene, SCO5446, encoding a homologue of SgmA that lacks the C-terminal P-domain. It is possible that the SCO5447 mutation was functionally complemented by SCO5446 under the conditions tested.


A serine protease, SCO5913, contributes significantly to a decline in STI activity at the onset of development

Streptomyces trypsin inhibitor was produced throughout growth, at a production rate (qSTI) directly related to the mycelium growth rate (μ). In stationary phase, its activity decreased in a manner indicating that it was being degraded (Fig. 1, Fig. S1). Proteomic analysis also showed a decline in stationary phase of the two protein spots corresponding to STI (Kim et al., 2005). It was reasonable to assume that one or more extracellular proteases might be involved in its inactivation. We selected and inactivated six of the 56 chromosomal genes for proteases (Table S2), including SCO5913 protease, which has low similarity to other Streptomyces proteases (Kelemen and Buttner, 1998; Kim et al., 2006). Only the SCO5913 null mutant showed a phenotypic change, leading us to investigate the in vivo effects of SCO5913 protease on STI abundance during growth and development. Inactivation of STI by SCO5913 protease was indicated by both increased STI activity in the SCO5913 null mutant and diminished STI in an SCO5913-overproducing strain. These effects appeared to be post-transcriptional, as sti mRNA was equally readily detected at all time points and independently of SCO5913 status [in a previous study, Li et al. (2007) also deleted SCO5913 and they reported that there was no specific phenotype change. This difference may be attributable to the use of slightly different media and the fairly subtle phenotype, which was not obvious in crowded areas] Consistent with these results, in vitro experiments clearly demonstrated that purified SCO5913 protease was capable of degrading STI. Our findings of retarded development in the SCO5913 null mutant and STI-overproducing strain, and accelerated development in the sti null mutant, strongly indicate that the inhibition of proteases by STI has significant consequences for the timing and progression of development. At least one other protease must also be involved in STI inactivation, as a gradual loss of STI was seen even in the SCO5913 null mutant, and aerial mycelium did eventually form in the mutant.

Surprisingly, the extent of STI accumulation in the different strains appeared also to be inversely associated with the level of Act production (Fig. 4). A large number of different genes and signals have previously been found to affect production of this antibiotic (Chater and Bibb, 1997), and our previous proteomic studies had shown that some of the enzymes of Act biosynthesis are extracellular, and are subject to proteolytic processing (Hesketh and Chater, 2003; Kim et al., 2005), but the physiological significance of these observations remains unclear.

A eukaryotic-type proprotein convertase is essential for STI-dependent differentiation

Two-hybrid analysis revealed that the C-terminal convertase P-domains of two proteases could interact with STI. These were an extracellular neutral zinc MTP encoded by SCO5447, and a probable serine protease encoded by SCO1355. Interaction between STI and SCO1355 was confirmed by in vitro analysis. It is therefore likely that these proteases are targets of STI, and although inhibition of their activity by STI has not yet been experimentally demonstrated, the inhibition of proteases by STI-type proteins is well established in the literatures (Takeuchi et al., 1992; Ueda et al., 1992; Kumazaki et al., 1993; Taguchi et al., 1998; Oda et al., 2001).

In eukaryote cells, proprotein convertase, containing a P-domain, is involved in the processing of precursors of many hormones, peptides and specific enzymes, including extracellular MTP (Ueda et al., 2003; Henrich et al., 2005). Often, such processing is important for the accurate regulation of cellular events (Rockwell et al., 2002). In yeast, convertase Kex2 generates mature α-mating factor and killer toxin from their precursors (Fuller et al., 1988). Kex2 and its mammalian analogues have a subtilisin-like catalytic domain and form a distinct subfamily (Kex2 endoproteinases) within the subtilisin superfamily (Nakayama, 1997). In mammalian cells, furin family proteases of the proprotein convertase type are also involved in tumour progression, disease and bacterial/viral infection (Bassi et al., 2005). In the mouse model, P-domain folding of proprotein convertase (PCSK5) was recently shown to be critical for its activity, and abnormal folding of the P-domain induced serious cellular defects in development, showing the importance of the P-domain in cellular processing (Szumska et al., 2008). Because of the importance of convertases, inhibitors of these enzymes have been studied as potential therapeutic agents for various diseases (Bontemps et al., 2007), and many synthetic convertase inhibitors have been evaluated (Basak, 2005). However, relatively few natural convertase inhibitors have been reported. In addition to inhibitors of mammalian convertase (furin) identified in humans (PI8) and Drosophila (Serpin4) (Dahlen et al., 1998; Osterwalder et al., 2004), another inhibitor, kexstatin I, which inhibits yeast kexin (Kex2), was found in the culture filtrate of Streptomyces platensis Q26. Kexstatin shows high similarity to STI (Oda et al., 2001), strongly supporting the general idea that the P-domain of proteases such as SCO5447 or SCO1355 might mark them for targeting by Streptomyces subtilisin inhibitor family protease inhibitors such as STI. In S. coelicolor, SCO5447 or SCO1355 proteases may process target proteins required for differentiation and Act production. Thus, in streptomycetes, as with yeast Kex2, a protease containing a P-domain may be involved in processing proteins needed for cellular events. In this context, it is interesting to note that SapB, a secreted morphogenetic peptide that is essential for aerial mycelium formation in streptomycetes under certain growth conditions, is derived from the ramS gene product by post-translational modification of specific amino acid residues, and proteolytic processing of a leader peptide sequence (Kodani et al., 2004). The ramC gene product is believed to encode the modification enzyme, while the protease has yet to be identified. Furthermore, in current work a screen for processing targets of SCO1355 protease has been undertaken using the exo-site scanning strategy developed for protease substrate screening (Overall et al., 1999; McQuibban et al., 2000). This has revealed a putative lipoprotein whose gene is closely linked to genes encoding morphogenetic proteins (rodlins and chaplins), suggesting a relationship between the protease cascade and morphogenesis (data not shown). These morphology-conferring proteins could potentially be the targets of the prokaryote convertase.

The somewhat similar colony phenotypes of the SCO1355 mutant and a STI-overproducing strain strengthen the idea that STI and SCO1355 protease are part of an extracellular cascade important for the formation of aerial mycelium. Inactivation of SCO5447, on the other hand, had no obvious morphological effects, possibly because of functional redundancy with the adjacent gene SCO5446. In addition to any involvement in processing of specific proprotein targets, the SCO5447 or SCO1355 proteases could also be involved in the reuse of vegetative biomass to generate nutrients supporting the growth of the aerial mycelium. Perhaps the increased death rate of stationary phase mycelium of a bldA mutant reflects a loss of control of the protease cascade.

Production of STI is growth-associated, yet it depends on a tRNA associated with stationary phase physiology

The fact that STI production is both bldA-dependent and growth-associated was surprising, as most effects of bldA mutation are associated with stationary growth phase processes rather than vegetative mycelial growth (Kim et al., 2005; Hesketh et al., 2007). In line with this unexpected growth-associated effect, the bldA mutation also had a decisive effect on the mycelium growth rate (μ) and glucose uptake rate (qs). A first hint of the change in growth rate was noticed by Hesketh et al. (2007), who also found some growth-associated transcriptome and proteome changes in the bldA mutant.

Transcription of sti was readily detected in rapidly growing cultures (Kim et al., 2005; Fig. 2). As transcription of sti is dependent on AdpA, which is itself dependent on bldA because of a UUA codon in adpA mRNA (Nguyen et al., 2003; Takano et al., 2003; Kim et al., 2005), rapidly growing liquid cultures must be expressing adpA and bldA at biologically significant levels, even though most of the processes dependent on these genes are manifested during stationary phase. The absence of STI activity from a bldA mutant indicates either that S. coelicolor produces no other STI-like activity, or that if it does, any other such activity is also dependent on bldA. The former possibility is more likely as, although there is one more homologue of STI in S. coelicolor, its expression level and inhibition activity are very low compared with STI (Kato et al., 2005b). The bldA dependence of STI-like protease inhibitors seems to be a general aspect of Streptomyces spp., as production of a protease inhibitor (SgiA) by S. griseus, a phylogenetically distant species, was also dependent on a TTA-containing adpA gene (Hirano et al., 2006).

As gene SCO5913 contains a TTA codon (unlike any of the other genes for proteases), the production of SCO5913 protease presumably depends on bldA. The suppression of the morphological deficiency of a bldA mutant by changing the TTA codon in adpA (Nguyen et al., 2003; Takano et al., 2003) should therefore not have been complete. In fact, the colonies of the adpA TTA codon conversion mutant did appear to be somewhat delayed in development, consistent with this expectation (Nguyen et al., 2003).

Extracellular regulation of differentiation in S. coelicolor

The interactions described in this paper are summarized in Fig. 8 in the form of a cascade model that links intracellular regulatory events with a multi-step extracellular cascade. Although the action of bldA is needed for the expression of both STI (via AdpA) and the SCO5913 protease, likely differences in transcriptional regulation will mean that the production and secretion of these two proteins do not necessarily occur at the same time nor to the same extent. There is also no information yet to indicate how transcription of the majority of the genes presented in the model is controlled, making it difficult at present to rationalize the timing of the events depicted. Previously, a cascade of uncharacterized extracellular signals was revealed by extracellular cross-complementation of aerial mycelium deficiencies among a collection of bld mutant, including bldA and adpA (= bldH) (Willey et al., 1993). It is an interesting possibility that the two cascades might overlap or even be the same. In that case, some of the protease genes, such as SCO1355 and SCO5447, could turn out to be regulated by bld genes downstream of bldA in the cascade of Willey et al. (1993) (e.g., bldG, bldC, bldD). Interestingly, the only previously identified signal in the cascade is an oligopeptide, whose origin is unknown, but which might itself be generated by the action of a protease (Nodwell and Losick, 1998). The extracellular biology of S. coelicolor is proving to be very complex, and also includes hormone-like quorum-sensing lactones (Takano et al., 2001; Chater and Horinouchi, 2003), morphogenetic lanthionine-containing peptides and hydrophobin-like proteins that form a morphogenetic outer layer that coats aerial hyphae (Elliot et al., 2003; Claessen et al., 2004; 2006; Kodani et al., 2004; Chater, 2006; Chater and Chandra, 2006; Capstick et al., 2007), and the production of cellulose at aerial hyphal tips (Xu et al., 2008). Proteases and protease inhibitors clearly also have a role to play, and the bldA dependence of SCO5913 protease and the protease inhibitor STI establishes a significant connection of intracellular and extracellular developmental regulation that represents a new aspect of the extracellular biology of streptomycetes.

Figure 8.

A model for the extracellular regulation of differentiation by proteases and the STI protease inhibitor in S. coelicolor. This model is based on the results of the present study and on previous studies referred to in the text, and is intended to illustrate the extent of the network of possible regulatory switches and interactions rather than representing a temporal sequence of events. Nevertheless, we have used stippled shading to indicate events that should occur late in order to activate development. Aspects in the regulation that are yet to be defined are represented by a question mark.

Experimental procedures

Strains, plasmids and media

Streptomyces coelicolor M600, a plasmid-free S. coelicolor A(3)2 derivative, was the parent strain for construction of null mutants. Escherichia coli DH5α was used for general plasmid preparation. E. coli BW25113/pIJ790 was the host for λ Red-mediated PCR-targeted mutagenesis (Gust et al., 2003). E. coli ET12567/pUZ8002 was used to prepare unmethylated cosmid and plasmid DNA, and for conjugational transfer to S. coelicolor strains (Gust et al., 2003). Maintenance of S. coelicolor and E. coli and conditions for submerged culture were as reported previously (Kim et al., 2005). S. coelicolor was cultured on MS agar or in supplemented minimal medium (SMM), the nitrogen source being 0.2% casamino acids (SMM-CA) or 0.2% sodium caseinate (SMM-CS) depending on the experimental objective (Kieser et al., 2000).

Clontech's yeast two-hybrid system 3 (Clontech) was used to construct a S. coelicolor cDNA library and to screen for STI-interacting proteins. S. cerevisiae AH109 and Y187 strains were used with pGADT7-Rec and pGBKT7, respectively, for interaction screening, with pGADT7-T (SV40 large T-antigen) and pGBKT7-53 (murine p53) as positive control for interaction, and pGADT7-T and pGBKT7-lam (human lamin C) as negative control. For yeast culture, YPD (10 g of yeast extract, 20 g of Bacto peptone and 20 g of glucose per litre; pH 6.5) and synthetic dropout (minimal synthetic dropout base plus amino acid dropout supplement; Clontech) were used. Bacterial strains, plasmids and primers used are listed in Tables S3 and S4.

Construction of strains

The PCR-targeted deletion of genes was confirmed by Southern blotting using the digoxigenin method (Roche). SCO0762, the sti gene, was amplified by PCR using the sti PF and sti PR primers and genomic DNA as template. The amplified sti gene was cloned into the KpnI site downstream of the erythromycin-resistance gene start codon in plasmid pZErO-2 (Invitrogen). A HindIII plus EcoRI fragment of this construct, containing sti downstream of the ermE promoter, was then cloned between the HindIII and EcoRI sites of pWHM3A (pWHM3 with apramycin-resistance gene), to yield pSMF5401. pSMF5401 was introduced into S. coelicolor M600 by protoplast transformation (Kieser et al., 2000) to give strain SMF5401. Gene SCO5913 was amplified from cosmid SC10A5 using primers 5913F and 5913R, and cloned between the NdeI and BglII sites of pIJ8600, to yield pSMF5402 containing gene SCO5913 under the control of the thiostrepton-inducible promoter tipA. This plasmid was conjugated into SMF5401, producing SMF5402, in which overproduction of SCO5913 can be induced by adding thiostrepton (25 μg ml−1).

Analysis of sti transcript abundance and STI activity in agar-grown cultures

After pre-germination in 2× YT liquid medium (6 h, 30°C), 3 × 106 spores were used to inoculate SMM-CA agar with and without cellophane overlays. Mycelium harvested from cellophane-covered agar media for RNA extraction was immediately soaked in protector solution (Qiagen) to stabilize transcripts. Total RNA was then prepared using an RNeasy midi kit (Qiagen) according to the manufacturer's instructions but with additional phenol/chloroform extractions. Contamination by chromosomal DNA was prevented by two treatments with DNase I (Qiagen): an on-column digestion and a final solution digestion with the RNA clean-up method (Qiagen). First-stranded cDNA and double-stranded DNA were synthesized by Superscript II RT (Invitrogen) and Ex-Taq premix (Takara) respectively. sti TF and sti TR primers were used to quantify sti transcripts. Twenty-five cycles were used for PCR amplification. As a control for contamination with chromosomal DNA, standard PCR was also performed with purified total RNA. Expression analysis of hrdB was used as an internal control, employing primers hrdB TF and hrdB TR. For determination of STI activity, colonies developed on agar plates without cellophane were harvested in 10 ml of Tris-HCl buffer (pH 7.6) containing glass beads, and then vortexed vigorously for 10 min at room temperature. The mixture was centrifuged at 10 000 g for 10 min and the supernatant used for the determination of STI activity.

Fermentation conditions and kinetic analysis

Measurement of growth rate, glucose concentration, STI activity and Act levels in duplicate flask and jar fermentor cultures was as in Kim et al. (2005) and Ryu et al. (2006). STI activity was defined as the leupeptin unit (Kim and Lee, 1995; Kim et al., 2005). Measurement of total protease activity was as in Kang et al. (1995), and fermentation kinetic parameters were calculated as in Pirt (1975).

Two-hybrid screening

A BD Matchmaker library construction kit (Clontech) was used to construct a cDNA library to screen for proteins interacting with STI. cDNA was synthesized from 1 μg of RNA isolated from S. coelicolor mycelium grown on R2YE agar after 3 and 6 days. Synthesized ds cDNA (20 μl from one reaction) and linearized pGADT7-Rec (3 μg) were used in the co-transformation of S. cerevisiae AH109 for pGADT7-cDNA construction by homologous recombination. After 4 days, clones harbouring pGADT7-cDNA were selected on leucine dropout synthetic solid media. Transformants were harvested by using freezing medium [YPD medium containing 25% (v/v) glycerol], yielding 2 × 107 cells ml−1 as the S. coelicolor cDNA library pool.

To make a bait construct, PCR primers for sti gene amplification were designed so as to exclude the signal peptide (aa 1–35). PCR was performed with sti F and sti R primers from genomic DNA. The resultant PCR product was cloned into pGBKT7 between the NdeI and EcoRI sites, yielding pGBKT7-sti. Interaction screening was performed according to the manufacturer's instructions (Matchmaker library screening kit, Clontech). Briefly, a 1 ml aliquot of AH109 cells containing the cDNA library was combined with cultured Y187 cells (1 × 109 cells ml−1) harbouring the bait pGBKT7-sti plasmid in a sterile 2 l flask, and 45 ml 2× YPDA supplemented with kanamycin (50 μg ml−1) was added. Cells were incubated at 30°C for 24 h with gentle swirling (30–50 r.p.m.). After mating, the mating mixture was washed and harvested. The cell pellets were re-suspended in 10 ml of 0.5× YPDA with kanamycin (50 μg ml−1). For selection of yeast diploids expressing interacting proteins, the mating mixture was inoculated to leucine, tryptophan, histidine and adenine dropout synthetic medium (QDO). Well-grown colonies were chosen as positive colonies for isolation of cDNA library vectors, and insert cDNA genes were identified by nucleotide sequencing (Bionics, Korea). The intensity of interactions was quantitatively compared by colony-lift filter assay and ONPG assay according to the manufacturer's instructions (yeast protocol handbook, Clontech). One unit of β-galactosidase activity is defined as the amount that hydrolyses 1 μmol of ONPG per min per cell.

Two-hybrid analysis of interaction between STI and SCO1355

CD F and CD R primers were used to amplify the catalytic domain coding region (aa 1–426) of SCO1355; PD F and PD R primers to amplify the P-domain coding region (aa 427–537); and CD F and PD R primers to amplify full-length SCO1355 (aa 1–537). These segments were cloned between the NdeI and EcoRI sites of pGADT7 (containing the GAL4 activation domain), producing pGADT7-1355 (full-length SCO1355), pGADT7-CD (catalytic domain) and pGADT7-PD (P-domain), taking care that the reading frame was preserved through the fusion points. Each plasmid was used with pGBKT7-sti to co-transform S. cerevisiae AH109 for two-hybrid analysis. To control for false positive signals, the reciprocal interactions were also analysed by cloning these fragments into pGBKT7 (containing the GAL4 binding domain), and introducing the resultant plasmids (pGBKT7-1355, pGBKT7-CD and pGBKT7-PD) into S. cerevisiae AH109 carrying pGADT7-sti. Self-activation of the transcriptional machinery in the two-hybrid system was examined by auxotroph assay in histidine, adenine and leucine (or tryptophan) dropout solid media for each vector construct.

Purification of STI, SCO5913 and SCO1355

Plasmid pET19b was used for purification of His-tagged STI, and pGEX4T-1 was used to make GST-tagged SCO5913 and SCO1355. PCR primers for each gene were designed so as to exclude the signal peptide (aa 1–35 of STI, aa 1–27 of SCO5913, and aa 1–33 of SCO1355). PCR reactions were performed with genomic DNA as template, using primers sti 19F and sti 19R for sti, 5913 GF and 5913 GR for SCO5913, and 1355 GF and 1355 GR for SCO1355. The resultant PCR product of sti was cloned into pET19b between the NdeI and XhoI sites, SCO5913 and SCO1355 were inserted into pGEX4T-1 between the BamHI and EcoRI sites, and BamHI and XhoI sites respectively. Each plasmid was introduced into E. coli BL21 (DE3) pLysS strain for expression of the STI, SCO5913 and SCO1355 proteins. The proteins were quite insoluble in normal heterologous expression conditions, so induction was performed by adding IPTG (1 mM) to 2 l fermentation cultures in 5 l jar fermentors at 25°C with aeration (1 v.v.m.) and agitation (170 r.p.m.) for 6 h. Expressed proteins present in the soluble fractions were purified using a Ni-NTA column (Novagen) for His-tagged STI, and a GST-column (ELPIS Biotech, Korea) for GST-tagged SCO5913 and GST-tagged SCO1355. Partially purified proteins were concentrated using an Amicon ultra filter device (Millipore) with 2× storage buffer (0.4 M Tris, pH 7.9, 0.2 M NaCl, 0.1 M EDTA). The GST-tag of SCO5913 was removed by in-column digestion using thrombin as follows: 2 μg of GST-tagged SCO5913 was bound to a GST column, and then treated by incubation with 0.04 unit of thrombin (Novagen) for 12 h at 20°C. The untagged protein liberated was eluted in phosphate buffer (20 mM , pH 7.0, 1.0 mM EDTA).

In vitro proteolysis of STI by SCO5913 protein

1.5 μg of purified His-tagged STI and 0.1 μg of SCO5913 protein (GST-tag removed) were incubated in phosphate buffer (20 mM, pH 7.0, 1.0 mM EDTA, 1.0 mM DTT) at 30°C for 0.25, 0.5 and 3 h. The incubated mixtures were subjected to SDS-PAGE analysis to monitor any changes in abundance of STI. Protein bands were visualized by Coomassie brilliant blue staining.

Native gel binding assay

Purified His-tagged STI (1.5 μg) and GST-tagged SCO1355 protein (0.1, 0.3 and 1.0 μg) were incubated in 50 μl of binding buffer (10 mM Tris , pH 7.6, 0.1 M NaCl, 1 mM DTT) at 30°C for 30 min. 1.0 μg of GST-tagged SCO4677 protein (Kim et al., 2008) was used as a negative control. After incubation, mixtures were loaded onto a pre-run 7.5% native polyacrylamide gel. Protein bands were visualized by Coomassie brilliant blue staining. Complex formation was verified by excising the protein band of interest from the gel, homogenizing and incubating overnight in SDS-equilibration buffer (50 μl) and SDS-sample buffer (50 μl) sequentially, and analysing the incubated sample using 15% SDS-PAGE.


His-tagged STI (0.5 μg) was incubated with GST-tagged SCO1355 protein (0.2 μg) in binding buffer (50 mM Tris, pH 7.5, 50 mM KCl, 3 mM MgCl2, 1 mM DTT) for 1 h at room temperature. Anti-GST antibody (0.025 mg, GE Healthcare) was added, and the mixture was incubated for a further 1 h at room temperature. Ten microlitres of nProtein A beads (GE Healthcare) was then added to the complex of proteins and antibodies and the incubation continued for 1 h. The beads were then collected by centrifugation (7000 r.p.m., 30 s at 25°C), and washed 10 times with binding buffer (500 μl), vortexing the sample for 10 s at each washing step. Proteins bound to the beads were finally eluted by boiling with SDS-sample buffer (total 50 μl), and then separated on two identical SDS-PAGE gels. The separated proteins were transferred to PVDF membranes by Western blotting using Bio-rad transfer kit with transfer buffer (25 mM Tris, 192 mM glycine, 20% methnaol) at 80 V for 1 h. One membrane was incubated with anti-His antibody (Qiagen), and the other with anti-GST antibody. Western signals were detected using the ECL system (GE Healthcare).

Characterization of growth on agar media

Diluted spores were used to inoculate SMM-CA agar plates to give a few colonies per plate. Three to five well-isolated colonies were observed and photographed over 3 days through a low-power photomicroscope (Stemi 2000-C, Zeiss).


This work was supported by the KICOS through a grant provided by the Korean Ministry of Science and Technology in 2007 (K20726000001-07B0100-00110), for the collaboration researches with the ActinoGEN (IP005224) in EU. The authors D.W.K., E.S.K. and J.Y.S. were supported by the second stage of the Brain Korea 21 Project.