Mutation in the rel gene of Sorangium cellulosum affects morphological and physiological differentiation

Authors

  • Tina Knauber,

    1. Department of Microbiology and Molecular Biology, University of Giessen, Heinrich-Buff-Ring 26–32, 35392 Giessen, Germany.
    Search for more papers by this author
  • Sabrina D. Doss,

    1. Department of Microbiology and Molecular Biology, University of Giessen, Heinrich-Buff-Ring 26–32, 35392 Giessen, Germany.
    Search for more papers by this author
  • Klaus Gerth,

    1. Helmholtz Centre for Infection Research, Inhoffenstraße 7, 38124 Braunschweig, Germany.
    Search for more papers by this author
  • Olena Perlova,

    1. Department of Pharmaceutical Biotechnology, Saarland University, PO Box 151150, 66041 Saarbrücken, Germany.
    Search for more papers by this author
  • Rolf Müller,

    1. Department of Pharmaceutical Biotechnology, Saarland University, PO Box 151150, 66041 Saarbrücken, Germany.
    Search for more papers by this author
  • Anke Treuner-Lange

    Corresponding author
    1. Department of Microbiology and Molecular Biology, University of Giessen, Heinrich-Buff-Ring 26–32, 35392 Giessen, Germany.
      *E-mail Anke.Treuner-Lange@mikro.bio.uni-giessen.de; Tel. (+49) 6419935544; Fax (+49) 6419935549.
    Search for more papers by this author

*E-mail Anke.Treuner-Lange@mikro.bio.uni-giessen.de; Tel. (+49) 6419935544; Fax (+49) 6419935549.

Summary

Interruption of the (p)ppGpp synthetase gene (rel) of Sorangium cellulosum So ce56 resulted in loss of ppGpp accumulation after norvaline treatment during exponential growth phase. The rel mutant failed to produce wild-type levels of the polyketides chivosazol and etnangien in production media. In wild-type cells expression of the chivosazol biosynthetic operon can be significantly increased by norvaline or α-methylglucoside. This induction does not occur in the rel mutant. The rel mutant also lost the capability to form multicellular fruiting bodies under nutrient starvation.

Introduction

The immense diversity of secondary metabolites produced by microbial species in nature is a rich source of valuable compounds that find applications as pharmaceuticals and agrochemicals (Saxena and Pandey, 2001; Newman and Cragg, 2005). The most important secondary metabolite-producing bacteria are able to undergo morphological differentiation processes. This correlation has been observed for several soil-living bacteria, such as actinomycetes, bacilli and myxobacteria (Stone and Williams, 1992; Vining, 1992; Reichenbach and Höfle, 1993; Champness, 2000).

For myxobacteria it has been proposed that production of bioactive compounds is used as a strategy to kill prey microorganisms and to defend their niche in the habitat (Reichenbach and Höfle, 1993; Reichenbach, 1999). Most known myxobacterial secondary metabolites are produced by the genus Sorangium (Gerth et al., 2003; 2008). Therefore, a functional genome project to enable detailed investigation of a Sorangium cellulosum strain was performed (Gerth et al., 2003; Schneiker et al., 2007). The selected strain So ce56 produces the secondary metabolites chivosazol and etnangien and forms multicellular fruiting bodies. Recently, the biosynthetic gene cluster for chivosazol was identified (Perlova et al., 2006) and ChiR was described as a pleiotropic regulator positively influencing the production level of this secondary metabolite as well as required for fruiting body formation (Rachid et al., 2007).

Myxobacteria are specialized in the degradation of biomacromolecules. Most species are bacteriolytic and lyse whole cells of other microorganisms (bacteria and yeasts). Also, the degradation of xylan and chitin by myxobacterial species has been reported (Reichenbach, 1993). In contrast, to the predominantly bacteriolytic members of the suborder Cystobacterineae, degradation of cellulose is restricted to members of the suborder Sorangineae (Reichenbach, 1993; Brenner et al., 2005). Composition of growth media for cellulose-degrading S. cellulosum strains differ significantly from media used for growth of bacteriolytic myxobacteria. Besides some essential salts S. cellulosum can grow exclusively on cellulose as carbon source with nitrate or ammonia as nitrogen sources. Other carbon sources like glucose, mannose and maltose are also utilized (Müller and Gerth, 2006). In contrast, the bacteriolytic Myxococcus xanthus grows on microbial lysis products or on simple peptone-based media (Reichenbach and Dworkin, 1992; Reichenbach and Höfle, 1993) and carbohydrates do not significantly stimulate growth (Dworkin, 1963). M. xanthus, which belongs to the suborder Cystobacterineae, is the best studied myxobacterium in regard to growth and development, but its regulation of metabolism and the initiation of differentiation might be quite different compared to S. cellulosum.

Much effort has been devoted to understand the regulation of secondary metabolism in different Streptomyces species (Bibb, 2005). In streptomycetes, secondary metabolite production coincides, or slightly precedes, the onset of morphological differentiation in surface-grown cultures. In liquid-grown cultures, production occurs while cells are growing with a reduced growth rate (Woodruff, 1980; Bibb, 2005), causing several secondary metabolites to accumulate in the fermentation broth during stationary phase. Secondary metabolite production is often described as a physiological differentiation process as it results from differential or even spatial gene expression under nutrient limiting conditions (Khetan et al., 2000). Nitrogen-, carbon and/or phosphate limiting conditions trigger production of different secondary metabolites (Woodruff, 1980; Bibb, 2005). Nitrogen and phosphate deprivation was also shown to be an effective trigger of morphological differentiation in Streptomyces griseus (Kendrick and Ensign, 1983). The accumulation of (p)ppGpp provoked by amino acid or nitrogen starvation is involved in triggering antibiotic production as well as development in Streptomyces upon nitrogen starvation (Chakraburtty and Bibb, 1997; Jin et al., 2004; Bibb, 2005).

Under nutrient limitation, accumulation of (p)ppGpp initiates a global change in the cellular metabolism, which is called the stringent response. The alarmone (p)ppGpp changes the activity of RNA polymerase and is involved in growth rate control (Wagner, 2002). It is also involved in triggering antibiotic production as well as development in Streptomyces upon nitrogen starvation (Chakraburtty and Bibb, 1997; Jin et al., 2004; Bibb, 2005). In M. xanthus, amino acid limitation causes synthesis of (p)ppGpp and initiates the differentiation process (Harris et al., 1998). In contrast to streptomycetes, S. cellulosum So ce56 and several other myxobacterial species produce secondary metabolites even during exponential growth phase (Gerth et al., 1982; Kunze et al., 1984; Reichenbach and Höfle, 1993; Kegler et al., 2006). These observations raised the question whether secondary metabolism in S. cellulosum is affected by the stringent response. Whether (p)ppGpp is involved in triggering secondary metabolism in myxobacteria has to our knowledge not been investigated yet.

In this manuscript, we describe the effect of impaired (p)ppGpp synthesis on morphological as well as physiological differentiation in S. cellulosum So ce56. The phenotypic analysis of a constructed rel mutant clearly indicates that rel is required for morphological and physiological differentiation in S. cellulosum.

Results and discussion

Identification of Rel and features of the rel gene region of S. cellulosum

Using tblastn (Altschul et al., 1990) the genome sequence of S. cellulosum So ce56 was screened for (p)ppGpp synthetase encoding genes. Only one deduced protein sequence revealed significant homology to bacterial (p)ppGpp synthetases over the whole length (e-values from 0-e-130; Table 1).

Table 1.  Selected Rel orthologues.
OrganismLengthe-valueIdentities (%)
  1. The sequence of the Rel protein of S. cellulosum was used as a query sequence in the blast searches. Data obtained with sequences from delta proteobacteria, Gram-positive bacteria and Escherichia coli are shown.

Anaeromyxobacter dehalogenans9450.056
Myxococcus xanthus7570.055
Geobacter sulfurreduceus7160.054
Bdellovibrio bacteriovorus7380.049
Desulfovibrio desulfuricans7150.048
Desulfovibrio vulgaris7170.047
Bacillus subtilis7340.043
Streptomyces coelicolor8477e-16843
Escherichia coli RelA702e-15340
Escherichia coli SpoT744e-13036

Three gene products Rel, RelA and SpoT are known as ppGpp synthetases, which catalyse the transfer of a pyrophosphoryl group from ATP or GTP to the 3′-hydroxyl group of GTP. In several beta and gamma proteobacteria two different proteins, SpoT and RelA, are involved in the stringent response-dependent ppGpp-synthesis. In contrast, many other Gram-negative as well as several Gram-positive bacteria, for example Streptomyces coelicolor and Bacillus subtilis, harbour only one RelA/SpoT paralogue named Rel. All three different proteins share significant sequence homology in their central region. They differ solely in their N- and C-terminal regions. The RelA and Rel proteins are ribosome-bound (p)ppGpp synthetases. These proteins are believed to be activated by uncharged tRNA, binding to the ribosomes as a cellular signal for amino acid starvation (Lipmann and Sy, 1976; Wendrich et al., 2002). This activity is associated with the ACT-domain at the C-terminal part of these proteins. This ligand-binding domain is found in a wide range of enzymes that are regulated by amino acid concentration (Chipman and Shaanan, 2001).

In contrast, the SpoT proteins lack such ACT-domains. SpoT and Rel proteins share another domain, a HDc-domain in their N-terminal part (Aravind and Koonin, 1998). This domain seems to be responsible for the (p)ppGpp-degrading activity of these proteins (Gentry and Cashel, 1996). According to these domains (p)ppGpp synthetases were described and classified as follows (Wendrich et al., 2000): (p)ppGpp synthetases I (RelA) are monofunctional and synthesize (p)ppGpp after amino acid stress; (p)ppGpp synthetases II (SpoT) are bifunctional and synthesize (p)ppGpp after carbon starvation and hydrolyse it; (p)ppGpp synthetases III (Rel) are trifunctional and synthesize (p)ppGpp after amino acid stress, hydrolyse (p)ppGpp and synthesize it after carbon starvation.

The domain organization of (p)ppGpp synthetases I, II, III and the S. cellulosum protein were analysed using CD search (Marchler-Bauer et al., 2005). The domain structure of the protein from S. cellulosum indicated it to be a (p)ppGpp synthetase type III (Fig. 1). Therefore, we called the (p)ppGpp synthetase gene from S. cellulosum rel. The deduced Rel protein is highly similar to myxobacterial (p)ppGpp synthetases (55–56% identity) and proteins from other delta proteobacteria (47–54% identity) (Table 1). There is also significant homology to other eubacterial proteins [36–43% identity (Table 1)].

Figure 1.

Domain organization of the three different types of (p)ppGpp synthetases. The selected protein sequences for a CD search (Marchler-Bauer et al., 2005) were SpoT [(p)ppGpp synthetase I] and RelA [(p)ppGpp synthetase II] from E. coli (AAC76674, AAC75826) and Rel [(p)ppGpp synthetase III] from B. subtilis (O54408). The domain ‘mutated HD’ in the (p)ppGpp synthetase II was added per hand. This domain is not recognized by CD search because of two mutations (F82P83 instead of the crucial residues HD). It was suggested that this domain is inactivated but still retains its native structure (Aravind and Koonin, 1998). The numbers below the domains of Rel S. cellulosum indicate start and end positions of the corresponding domain. The e-values calculated by CD search for the domains of the Rel protein from S. cellulosum were the following: HDc 2e-05, RelA_SpoT 2e-41, TGS 5e-24, ACT 1e-12. The white box above the TGS domain of RelS. cellulosum indicates the truncation point in the mutant protein. The bottom of the figure shows the genetical organization of the rel gene from S. cellulosum. The graphic represents the 6 kb gene region deposited in GenBank (AY626562).

The rel gene from S. cellulosum is transcribed as a monocistronic unit, the hemN gene upstream and the two genes downstream of rel (dor1 and dor2) are organized in opposing directions (Fig. 1). The rel gene has a length of 2175 nt and encodes a protein comprising 724 amino acids. The GC-content of the gene is 65.8% which is slightly less than the GC-content of the neighbouring genes ranging between 71.9% and 75.8%. A GC-content of myxobacterial genomes of 67–71 mol% has been reported earlier (Mandel and Leadbetter, 1965; McCurdy and Wolf, 1967; Johnson and Ordal, 1968).

In the genome of S. cellulosum two additional genes were found that encode proteins with a HDc domain. These two proteins consist of only 182 and 172 amino acids and have homology to several eukaryotic phosphodiesterases. As they lack the central RelA_SpoT domain, these proteins are not expected to be (p)ppGpp synthetases. However, they might be involved in the degradation of (p)ppGpp or nucleoside 3′,5′-cyclic phosphates as reported for HDc-comprising phosphodiesterases from eukaryotes. Therefore, the genome of S. cellulosum encodes only one homologue of (p)ppGpp synthetases, Rel.

This finding is in agreement with a comparative genomic analysis done on (p)ppGpp synthetase/hydrolase encoding genes, which revealed that coexistence of the RelA and SpoT proteins in one organism is only found in beta and gamma proteobacteria (Mittenhuber, 2001). In this comparative genomic analysis it was also suggested that the trifunctional Rel proteins like the one found in S. cellulosum or M. xanthus resemble a more ancestral Rel protein, whereas SpoT and RelA evolved from duplication of such an Rel precursor protein (Mittenhuber, 2001).

Complementation of E. coli CF1693 with S. cellulosum rel

The cloned S. cellulosum rel gene was tested for function and ability to complement Escherichia coli strains CF1651 (relA251::kan-r) and CF1693 (relA251::kan-r, spoT203::cat) which have deletions in only relA or in relA and spoT, resulting in phenotypes unable to grow on SMG minimal agar (Sarubbi et al., 1989; Xiao et al., 1991). We transformed the vector pASKIBA15 as well as the construct pIBA15-Rel3 into the E. coli mutant strains and selected for ampicillin resistance. Additionally, we verified the presence of the rel gene in the transformed clones by polymerase chain reaction (PCR). Clones were then checked for growth on SMG plates with (2 μg ml−1−2 ng ml−1) and without inductor AHT (anhydrotetracycline). Using the transformants of CF1651 no significant growth on SMG plates could be observed. However, significant complementation could be achieved in CF1693 (Fig. 2). In the presence of quite low concentrations of AHT (2–20 ng μl−1) good growth of two different CF1693pIBA15-Rel3 clones could be observed on SMG plates compared with two transformants of CF1693pASKIBA15 (Fig. 2).

Figure 2.

Growth of CF1693 transformants on LB (left) and SMG (right). A and D are two transformants of CF1693pIBA15-Rel3 and B and C are transformants of CF1693pASKIBA15. Both plates contained AHT (20 ng ml−1) for induction of the tet-promoter and ampicillin.

These data show that the rel gene from S. cellulosum is functional in E. coli to provoke a stringent response. Interestingly, the presence of SpoTE. coli and RelS. cellulosum in one E. coli cell seems to be less functional for a proper stringent response as if only RelS. cellulosum is present. This could be due to a putative imbalanced ratio of (p)ppGpp-synthesis and -hydrolysis in the CF1651pIBA-Rel3 transformants.

Disruption of rel prevents accumulation of (p)ppGpp

Because rel is located in a monocystronic unit (Fig. 1) and therefore polar effects on transcription of downstream genes could be ruled out when disrupting the rel gene, we inactivated the rel gene by insertion of the plasmid pSUPHygrelint based on pSUPHyg vector (Pradella et al., 2002). This mobilizable plasmid carries an internal 827 bp fragment of the rel gene (covering nucleotide positions 433–1260 bp) and confers resistance to hygromycin. A successful construction of S. cellulosum mutants using internal fragments of approximately 700 bp cloned into the pSUP102-derivative pSUPHyg was described earlier (Simon et al., 1986; Pradella et al., 2002). Constructs based on pSUP102 can be conjugated into S. cellulosum, and because the plasmid cannot replicate in S. cellulosum, hygromycin-resistant colonies contain the plasmid integrated into the chromosome after homologous recombination. The disruption of the rel gene was confirmed by PCR using gene and vector specific primers in different combinations for the amplification of DNA fragments from chromosomal DNA of putative mutants as templates. Several attempts to conjugate the plasmid pSUPHygrelint2, harbouring a shorter internal rel-fragment (covering 433–1188 bp), failed. The mutant constructed by integration of pSUPHygrelint was called Srel.

If the truncated rel gene were transcribed and translated a truncated Rel protein comprising the first 420 N-terminal amino acids of Rel and 47 additional vector-derived amino acids might be produced. This protein would contain a HDc- and RelA_SpoT-domain but lack a complete TGS and ACT domain (Fig. 1). We tested if such a truncated rel-mRNA is made in the rel mutant. Using Northern-blot analysis neither a complete nor a truncated rel-mRNA could be detected in the Srel mutant (Fig. 3). As anticipated from the genome annotation a rel-transcript with a length of approximately 2400 nucleotides could be detected using RNA prepared from growing and starving wild-type cells (Fig. 3). We therefore conclude that in the mutant neither a complete nor a truncated rel transcript is produced.

Figure 3.

A. Transcript of the rel gene during vegetative growth (lane 1) and 3 h and 6 h post starvation (lanes 2 and 3) of S. cellulosum. Lanes 4–6 contain the corresponding RNA samples from the Srel mutant.
B. The 16S RNA hybridization was used as a positive control.
C. The amount and integrity of the used RNA samples were controlled by ethidium-bromide staining of the gel prior to the transfer.

Growth comparisons using the complex M-medium and the minimal SM-medium did not show any significant growth defects in the mutant. Nevertheless, longer periods in stationary phase were not equally well survived by the mutant in comparison with the wild type. When plating out and counting serial dilutions of M-media grown cells we observed a decreased viability of Srel cells when an optical density of 7 was exceeded. For example, after 142 h in M-media 4862 ± 1988 cfu from the wild type and only 238 ± 37 Srel mutant cells could be counted. After 160 h, 350 ± 150 wild-type colonies and only 82 ± 18 colonies Srel colonies were obtained.

Thin-layer chromatography (TLC) analyses were performed to test if (p)ppGpp accumulation still occurs in the Srel mutant (Fig. 4). Some chemicals are known to artificially induce the stringent response (Tosa and Pizer, 1971a,b; Hansen et al., 1975; Beaman et al., 1983). We used 1% α-methylglycoside to induce glucose starvation, 1.25 mg ml−1 serine hydroxamate to inhibit the charging of seryl-tRNAs and 0.5 mg ml−1 DL-norvaline, which induces leucine and isoleucine starvation (Eymann et al., 2002). In these concentrations addition of these chemicals during exponential growth phase caused an immediate growth arrest of S. cellulosum (data not shown). By addition of DL-norvaline (p)ppGpp-production in the wild type was observed already after 10 min. After 20 and 30 min of treatment ppGpp was still detectable in the wild-type cells but the level decreased over time. However, the GTP levels in the wild-type remained more or less unaffected (Fig. 4). In extracts of the Srel mutant neither an accumulation of ppGpp nor significant GTP level changes did occur.

Figure 4.

Detection of ppGpp in formic acid extracts of H332PO4-labelled S. cellulosum cells by PEI TLC and phosphoimaging. Wild type and mutant cells were labelled, stressed with norvaline and extracted for (p)ppGpp after 0, 10, 20 and 30 min of treatment. Quantification was done using the software Bio-Rad Quantity One”. Levels of ppGpp are shown in boxes (▪ wild type, □ Srel mutant) and GTP levels are shown in circles (● wild-type, ○ Srel mutant).

The TLC experiments revealed an interesting difference between the M. xanthus and S. cellulosum. The bacteriolytic M. xanthus hydrolyses proteins and obtains peptides and amino acids, which are used as carbon as well as nitrogen source. Serine hydroxamate was also successfully used in M. xanthus to cause (p)ppGpp production (Manoil and Kaiser, 1980; Singer and Kaiser, 1995). However, α-methylglycoside had no effect on (p)ppGpp production in M. xanthus (M. Singer, pers. comm.). In contrast, S. cellulosum, as an efficient cellulose degrader, seems to respond also to glucose starvation by (p)ppGpp production. Therefore, although the protein similarity of the corresponding Rel proteins is quite high (Table 1), the mechanism causing (p)ppGpp production might be different in these two myxobacterial species. However, we cannot rule out that the observed difference is just due to different uptake mechanisms. As α-methylglycoside, a non-metabolizable glucose analogue, is used to decrease uptake of glucose by competition for the transport of this sugar (Hansen et al., 1975), M. xanthus might just not have such a glucose transporter. Nevertheless, a different glucose analogue 2-deoxyglucose (2dGlc) was shown to inhibit the growth and multicellular development of M. xanthus (Youderian et al., 1999).

(p)ppGpp accumulation is required for physiological differentiation of S. cellulosum

For M. xanthus, a direct correlation between (p)ppGpp accumulation and initiation of the development pathway has been clearly observed (Singer and Kaiser, 1995; Harris et al., 1998). Also, a rel-deletion mutant of S. coelicolor has been described to be defective in morphological differentiation and production of the antibiotics actinorhodin (Act) and undecylprodigiosin (Red) under nitrogen limitation (Chakraburtty and Bibb, 1997).

S. cellulosum So ce56 has been shown to produce the secondary metabolites chivosazol and etnangien (Pradella et al., 2002; Gerth et al., 2003) in batch cultures grown in the presence of the adsorber resin XAD (Pradella et al., 2002). An initial qualitative assay using Hansenula anomala as a chivosazol-sensitive and Micrococcus luteus as an etnangien-sensitive bioindicator showed severely reduced levels of these polyketides to be produced by the Srel mutant when grown in XAD-containing production medium (PD-medium), without peptone as a nitrogen source (Fig. 5A). Therefore, a quantitative high-performance liquid chromatography (HPLC) analysis was performed with XAD-extracts from wild type and Srel mutant growing in a batch fermentation with PD-medium. Already after 2 days the wild type produced significant levels of chivosazol and etnangien (Fig. 5B). At this time the cells are still growing. According to measurements of CO2 and O2 levels the cells entered stationary phase after 4–5 days. This analysis indicated a reduced level of chivosazol and etnangien production (25% and 22.5% of wild type respectively) in the Srel mutant (Fig. 5B). The fact that low levels of both polyketides were synthesized in the Srel mutant suggest a basal (p)ppGpp-independent production of these compounds in S. cellulosum. For the bioassay experiment peptone (0.02%) was omitted from the PD-medium to provoke a stronger nitrogen limitation. Nitrogen limitation seems to be an important trigger for chivosazol production in S. cellulosum So ce56 and peptone, which is a good nitrogen source for So ce56, has been already shown to decrease the production of chivosazol drastically (Müller and Gerth, 2006). Nevertheless, low concentrations of peptone (0.05%) have been reported to stimulate growth without any effect on secondary metabolite production (Müller and Gerth, 2006).

Figure 5.

Secondary metabolite production.
A. Qualitative bioassay to test for chivosazol and etnangien production. Growth inhibition of the chivosazol-sensitive H. anomala yeast (left) and the etnangien-sensitive M. luteus strain (right) by XAD-extracts prepared from wild type (top) and Srel mutant (bottom). The inhibition zone of M. luteus in the upper picture was larger than the frame of the image.
B. Chivosazol (left) and etnangien (right) production in a 15 l bioreactor with production medium. Levels produced by the wild type are shown with filled boxes (▪) and levels produced by the Srel mutant are shown with open symbols (◊). MAU means milli-absorption unit as a standard for the peak area in the HPLC chromatogram).

A conditional role of the (p)ppGpp synthetase gene in antibiotic production and morphological differentiation was reported for S. coelicolor (Chakraburtty and Bibb, 1997). Whereas under nitrogen limitation a rel mutant of S. coelicolor failed to produce the antibiotics Red and Act it produced wild type-like levels of Red under phosphate limitation but production of the calcium-dependent antibiotic was not affected at all (Chakraburtty and Bibb, 1997). These observations suggest a complex network of regulatory mechanisms that can act independently to activate expression of secondary metabolite pathways.

In S. cellulosum the production of the polyketides chivosazol and etnangien is strongly affected by a rel mutation. Nevertheless, as suggested by the abundance of polyketide synthetase gene cluster in the genome of So ce56 (Gerth et al., 2003; 2008; Schneiker et al., 2007), the organism might be able to produce further secondary metabolites in a stringent-independent manner.

(p)ppGpp accumulation is required for morphological differentiation of S. cellulosum

Under appropriate starvation conditions So ce56 cells accumulate initially in flat masses resulting in more or less round structures which are still connected (Treuner-Lange et al., 2008) (Fig. 6A). At a later stage circular to oval sporangia with solid walls are formed within this cell aggregates (Fig. 6B). The brown fruiting bodies consist of anything from a few sporangia to great masses of hundreds. Fruiting body formation was initiated by nutrient limitation after growth in complex media. Nutrient limitation means a 200-fold reduction of the soy peptone content, and a 500-fold reduction of the carbon source content (mannose and starch respectively). In our hands, S. cellulosum does not survive total starvation by washing cells taken from growing cultures and suspending them in buffer. Differentiation occurs reproducibly only when high cell numbers (1–5 × 1011 cells) are spotted onto nutrient-limited agar surfaces. Under these conditions the wild type produces fruiting bodies after 3–4 days (Fig. 6C–F). The Srel mutant does not produce fruiting bodies (Fig. 6G–J). After longer incubation periods (5–10 days) the mutant obviously died, as we could not use these cells as inoculum for new vegetative cultures.

Figure 6.

Development of fruiting bodies of So ce56 under starvation on P-Diff. Under starvation conditions S. cellulosum accumulate first in flat masses resulting in more or less round structures (A). At a later stage circular to oval sporangia with solid walls are formed (B). The brown fruiting bodies consist of anything from a few sporangia to great masses of hundreds. Under the described conditions the wild type produced fruiting bodies after 3–4 days (C–F). The Srel mutant does not produce fruiting bodies (G–J).

One could argue that it is not the elevation of (p)ppGpp but the drop of GTP which actually causes the cells to start the developmental process under nutrient starvation, as in the case of B. subtilis endospore formation (Lopez et al., 1979; Ochi et al., 1982). Sporulation of B. subtilis was induced by the addition of decoyinine (0.2 mM), and inhibitor of GMP synthetase, which caused an extensive and long-lasting decrease in GTP (Ochi et al., 1982). We tested if addition of decoyinine (0.1–0.3 mM) to nutrient-rich or nutrient poor agar surfaces would initiate or accelerate the developmental process of So ce56 but could not observe any effect (data not shown).

To rule out that the differentiation defect was not due to the lack of the secondary metabolites chivosazol and etnangien, we tested the described chivosazol-negative mutant So ce56SB4 (Perlova et al., 2006) and an unpublished etnangien-negative mutant for fruiting body formation under nutrient starvation. Although the fruiting bodies looked somewhat more irregular than wild-type fruiting bodies the mutants were not significantly impaired in fruiting body formation.

rel is required for transcription of the chivosazol biosynthetic genes and the development-specific asgA gene under conditions of nutrient starvation

In S. coelicolor it has been shown that rel is required for transcription of the Red pathway-specific RedD transcriptional regulator under conditions of nitrogen starvation (Chakraburtty and Bibb, 1997). The same positive correlation was observed between appearance of (p)ppGpp synthesis and transcription of actII-ORF4, an Act pathway-specific activator protein (Chakraburtty and Bibb, 1997). Except for the direct regulator of chivosazol biosynthesis ChiR (Rachid et al., 2007) in S. cellulosum pathway-specific activators for chivosazol and etnangien are not known. ChiR was identified as DNA binding protein which specifically interacts with the promoter of the chivosazol biosynthetic gene cluster and triggers expression leading to increased production of chivosazol in a mutant overproducing ChiR (Rachid et al., 2007). We wanted to know whether the reduced levels of chivosazol production in the Srel mutant are due to reduced transcription of the gene cluster.

A huge operon coding for proteins required for chivosazol production could be identified (Pradella et al., 2002; Perlova et al., 2006). Primers which bind to the 3′ end of the chiB gene of the chiA–F operon (Kegler et al., 2006) were used for quantitative real-time RT-PCR (qRT-PCR) analyses. Total RNA was isolated from wild type and Srel mutant cells before and after addition of d-norvaline and α-methylglycoside.

The data showed that in the wild type the transcript levels of the chiB gene increased 6–7-fold after treatment with d-norvaline and α-methylglycoside (Fig. 7). The Srel mutant does not show such an increase, the transcription levels remain more or less unaffected.

Figure 7.

Quantitative real-time RT-PCR. Relative change of expression of the target gene chiB in wild type compared to the Srel mutant after artificial induction of the stringent response. The obtained data using RNA from untreated wild-type and mutant cells were arbitrarily set to one.

A strong transcriptional induction of the chivosazol gene cluster during the logarithmic growth phase in synthetic media (1% mannose, 0.5% asparagine) followed by a second transcriptional increase in early stationary phase has been already reported (Kegler et al., 2006). Our data show that more severe starvation conditions do increase the level of ppGpp in So ce56, so transcriptional activation of the chivosazol gene cluster might predominantly depend on (p)ppGpp under these conditions.

The observation that chivosazol is produced already during exponential growth phase in SM media (Kegler et al., 2006) might be explained by limitations due to the minimal medium and the decreased growth rate compared with growth in complex media, probably provoking already elevated ppGpp levels. Other kinds of growth limitations have already been reported to stimulate myxobacterial secondary metabolite production. Thus, significant levels of myxothiazol are only produced under oxygen limitation (Gerth et al., 1980). Also myxobacterial production of myxovirescin and stigmatellin seems to be supported by nitrogen-limiting conditions (Reichenbach and Höfle, 1993). Product formation of an ambruticin-producing S. cellulosum strain was investigated in amino acid-rich media, and only after consumption of aspartate and glutamate the growth became limited and ambruticin production occurred (Hopf et al., 1990).

As studying the biology of S. cellulosum on the molecular level just started, no development-specific genes were known so far. Based on its homology to a development-specific gene product of the myxobacterium M. xanthus we could identify a homologue of a hybrid histidine kinase named AsgA in S. cellulosum. Based on phenotypic analysis of the corresponding mutant, AsgA seems to be only required for the fruiting body formation whereas secondary metabolite production was not impaired. We consider this gene therefore as a development-specific gene and tested its expression in the Srel mutant. As observed for chiB, asgA expression was also decreased in the Srel mutant compared with wild-type (not shown). For M. xanthus a rel-dependent expression of early A-signal dependent genes has been reported as well as insufficient production of A-factor in the relA mutant strain DK527 (Harris et al., 1998). Although the genetic repertoire of S. cellulosum suggests similarities to M. xanthus in regard to development- and sporulation-specific genes (Schneiker et al., 2007), we are far away from knowing if and how extracellular signalling occurs in S. cellulosum. We could not find A-factor like activities in the supernatant of starving So ce56 cells using the A-factor bioassay of M. xanthus (Kuspa et al., 1986). Also the csgA gene responsible for extracellular C-signalling in M. xanthus is not conserved in the genome of S. cellulosum, suggesting that in regard to extracellular signalling the organisms might differ significantly.

However, the qRT-PCR data suggest that crucial transcriptional regulators specific for the two differentiation processes are not produced in the Srel mutant after provoking a stringent response. We are in the process to find such transcriptional regulators, especially those which are involved in regulating development-specific genes, such as asgA.

Using MALDI-TOF we identified a NtcA-like protein as a putative regulator of asgA expression after DNA-chromatography with the hypothetical promoter region of asgA and proteins extracts from So ce56. This NtcA-like protein was also found as a chivosazol promoter-binding protein (Rachid et al., 2007). We are in the process to explore NtcA activity in the Srel mutant as we think that NtcA is an important transcriptional regulator for the physiological as well as the developmental processes in S. cellulosum.

Experimental procedures

Bacterial strains, culture conditions and microbiological procedures

As complex media for S. cellulosum So ce56 (stock culture, GBF, Braunschweig, Germany) either P-medium (0.2% peptone (Marcor), 0.5% starch (Roth, Karlsruhe, Germany), 0.1% probion (Gerth et al., 1984), 0.05% CaCl2 × 2H2O, 0.05% MgSO4 × 7H2O, 1.2% HEPES) or M-medium (1.0% soy-peptone, 1.0% maltose, 0.1% CaCl2 × 2H2O, 0.1% MgSO4 × 7H2O, 1.2% HEPES, and 0.0008% Na-Fe-EDTA) were used. Both media were adjusted to pH 7.2.

The synthetic medium SM was used as a minimal media (0.25% aspartate, 0.006% K2HPO4, 0.0008% Na-Fe-EDTA, 0.05% CaCl2 × 2H2O, 0.05% MgSO4 × 7H2O, 2.4% HEPES). The medium was adjusted to a pH of 7.2 and after autoclaving mannose from a 20% filter-sterilized stock solution was added to a final concentration of 1%.

Secondary metabolite production was tested using production (PD) medium (0.8% potato starch (Maizena, Germany), 0.2% glucose, 0.02% probion, 0.02% peptone, 0.1% CaCl2 × 2H2O, 0.1% MgSO4 × 7H2O, 1.2% HEPES, 0.0008% Na-Fe-EDTA, pH 7.4 titrated with (KOH). Cells were grown at 32°C. Liquid cultures were shaken at 180 r.p.m. If required, media were supplemented with 60 μg hygromycin B ml−1 (Invitrogen, Leek). E. coli cells were grown in Luria–Bertani (LB) media (Sambrook et al., 1989) or SMG plates [M9 media (Sambrook et al., 1989)] supplemented with 100 μg ml−1 serine, glycine and methionine (Uzan and Danchin, 1976), supplemented, when required, with 100 μg hygromycin B, 50 μg tetracycline, 50 μg ampicillin and or 50 μg kanamycin. AHT was used in the range of 2 ng−2 μg per ml for induction of the tetA promoter. Before transformation of the E. coli mutants CF1651 and CF1693 single colonies were tested for growth on LB and SMG plates. SMG-negative colonies were taken from the corresponding LB plate and used to start a culture in medium C for transformation following the protocol of the TransformAid™ bacterial transformation kit from Fermentas.

tblast n search

Protein sequences from M. xanthus (O52177), E. coli (AAC76674, AAC75826), B. subtilis (O54408) and S. coelicolor (CAA60717) were used to find ppGpp synthetase encoding genes in the genome of S. cellulosum So ce56.

Conjugation

The conjugation experiments were basically performed as already described (Pradella et al., 2002; Kopp et al., 2004). An overnight culture of E. coli ET12567 (MacNeil et al., 1992) transformed with pUB307 (Bennett et al., 1977) and pSUPHyg:relint (HygR, this study) grown in LB media, was used to inoculate 10 ml of LB media. The cells were harvested at optical density at 0.8 OD600, washed with LB media and adjusted to a cell density of 1010 cells ml−1. S. cellulosum So ce56 was grown in 50 ml of P-medium or M-medium for 3 days. The cells were harvested at a cell density of 5 × 108 cells ml−1 to 109 cells ml−1. The pellet was resuspended and washed twice with P- or M-media and afterwards suspended in approximately 3 ml of P-medium to a titer of 1010 cells ml−1. Then 100 μl of E. coli, ET12567pUB307 including pSUPHyg:relint and S. cellulosum So ce56 was carefully mixed and spotted onto P-medium agar plates. The cells were conjugated for approximately 40 h at 37°C. Afterwards the cell mixture was scraped from the agar and resuspended in 1 ml of P-medium and plated at 350 μl aliquots on P-medium agar plates containing hygromycin. Intensive orange-coloured colonies arose after 10–16 days of incubation at 32°C. The transconjugants were picked and incubated in 5 ml M-media. The successful integration of pSUPHyg:relint into the chromosome of the transconjugants was tested by PCR using the following primers (position in the sequence deposited in GenBank AY626562): relint- (nt 2790–2773) 5′-CTCGCCGACCTGCGTGTG-3′, relint+ (nt 1963–80) 5′-CGGACGCTCGAGTTCATG-3′, rel4 (nt 2950–2934) 5′-AGTTGCGGATCTTCGAG-3′, rel2 (nt 1803–1819) 5′-GGAGGTGGCGTTCCTCG as well as vector specific primers were used to confirm the interruption of the rel gene in the Srel mutant.

Recombinant DNA techniques and construction of plasmids

Genomic DNA, plasmids and DNA fragments were isolated and purified with appropriate isolation kits (Qiagen GmbH, Hilden, Germany or Peqlab Biotechnology GmbH, Erlangen, Germany). PCRs were employed according to the instructions of the polymerase supplier. An 887 bp internal fragment of the So ce56 rel gene was amplified by PCR, using chromosomal DNA from So ce56 as template, Fail safe polymerase (Biozym, USA) and the primers relint+ and relint-. The PCR product was directly ligated into the pGem® T-vector from Promega This construct was designated T-vector:relint. The T-vector:relint was cleaved with the enzymes SphI and PstI and the fragment including an internal fragment of the rel gene from So ce56 was treated with Klenow polymerase. The blunt-ended relint-fragment was then ligated into pSUPHyg (HygR, pSUP102 derivative; Pradella et al., 2002). Before ligation the vector was digested with the enzymes BamHI and HindIII and treated with Klenow polymerase to form blunt ends. The transformants were selected on hygromycin-containing LB agar plates and checked by PCR and sequencing.

To construct the plasmid pASK-IBa15-Rel we amplified by PCR the rel gene using the primers relstart (5′-ATGCTGACGTTGACGGAGCT-3′) and relstop (5′GCCGCTCAGGTCCGCTCG3′) and AccuPrime™ DNA-Polymerase (Invitrogen). The 2179 bp PCR-fragment was cloned into the vector pCR®-XL-Topo® (Invitrogen). After sequencing the construct pXL-rel12 was cut with EcoRI and the 2198 bp fragment was ligated into the EcoRI site of pASKIBA15. With the resulting construct pIBA15-Rel3 the AHT-inducible expression of the S. cellulosum rel gene is possible in E. coli. The resulting Rel protein comprises 754 amino acids and includes an N-terminal Strep-tag II peptide (WSHPQFEK) as well as an enterokinase cleavage site.

Thin-layer chromatography

Accumulation of pp(p)Gpp was measured using cells growing in P-medium. At an OD600 of 0.09–0.11 the cells were labelled with [32P]-H3PO4 (Amersham Biosciences) overnight. In two control flasks wild type and mutant cells were further incubated without labelling. The next day (around 18 h later) the cells in the control flasks had reached an OD600 of 0.7–0.9. Cells were taken before and after they were stressed with either norvaline (0.5 μg ml−1; Fluka, Steinheim, Switzerland). Before stress and after 15 and 30 and 60 min treatment 500 μl portions of the cultures were centrifuged. The pellets were suspended in 20 μl 1 M formic acid subjected to three freeze-thaw cycles. The samples were harvested (10 000 g for 5 min) and 20 μl of the supernatant were spotted on a 20 × 20 cm polyethyleneimine-cellulose (PEI) plate (Polygram Cel300PEI/UV254, Macherey-Nagel, Düren, Germany) for separation by TLC of the phosphorylated guanosine nucleotides in 1.5 M KH2PO4 (pH 5). Separated from the radioactive samples, cold standards as ppGpp (kindly provided by Prof Rolf Wagner) and GTP were also applied. The standard lanes were cut off after the TLC and position of the standards was visualized and marked under UV light. Serine hydroxamate (1.25 mg ml−1) and α-methylglucoside (1%) could also be used to induce a growth arrest and (p)ppGpp synthesis (data not shown). The TLC experiments were repeated several times with independently grown cells and visualized by either autoradiography or phosphoimaging. In every case cell volume equivalents were added to each TLC lane.

Total RNA extraction

Total RNA from stringent-response activated cells was prepared as follows: Two 50 ml M-medium cultures were grown up to an OD600 of 0.7–0.8. At that time point a 15 ml aliquot was harvested for 10 min at 8000 g. The pellet was immediately frozen in liquid nitrogen. The one culture was further incubated with α-methylglucoside (1%) and the other with norvaline (0.5 μg ml−1). After 30 min, 15 ml aliquots of both cultures were harvested. The RNA was isolated using hot phenol. After isolation, RNA was solved in DEPC water and purity and concentration were determined by measuring absorbance at 260 and 280 nm. Two independent sets of RNA samples were prepared.

Northern blot analyses

For the vegetative sample total RNAs from mutant and wild-type cells were isolated from cells grown in M-medium (OD 0.7–0.8). Afterwards the cells were washed in M-medium and suspended in M-diff media (M-medium with a 100-fold reduction of soy-peptone and maltose). After 3 h and 6 h incubation in M-diff the cells were harvested and RNA was isolated as described above. For each time 15 μg of total RNAs was fractionated on denaturant 1% agarose/formaldehyde gel, transferred to a nylon membrane and fixed by baking (80°C for 0.5 h) and cross-linking. The probes were obtained by PCR using the following primers: relint+, relint-, 16S–(5′-GACGGGCGGTGTGTACAAGG) and 16S+2 (5′-GGTAGTCCACGCCGTAAACG). The fragments were labelled using a nick translation kit and P32-dCTP. After hybridization the blots were analysed by phosphoimaging.

Bioassay

Cells were grown in 10 ml production medium without peptone containing 1% of the adsorber resin XAD-16 (Rohm and Haas, Frankfurt/Main, Germany). After the cultures have shaken for 14 days at 32°C the XAD was harvested by filtering. The XAD resin was eluted with 1 ml methanol and incubated for 2 h. The supernatant was further concentrated in a Speed-Vac concentrator to a final volume of 250 μl. For every supernatant to test for chivosazol production two small filter discs (0.5 cm) of Whatman paper were placed onto a Myc-plate (1% peptone, 1% glucose, 1.5% agar) covered with a layer of Myc-soft agar (0.6% agar) freshly inoculated with the chivosazol-sensitive H. anomala yeast strain (GBF, Braunschweig, Germany). For the qualitative etnangien assay M. luteus was used as a bioindicator strain using freshly inoculated LB plates. The inhibition zones around the filter discs were measured and photographed.

Quantification of the secondary metabolites chivosazol and etnangien

Liquid cultures of S. cellulosum So ce56 and the Srel mutant were started by inoculating the bacteria into 250 ml flasks containing 100 ml production medium. A 15 l bioreactor with 10 l of the production medium as described previously (but without HEPES) was inoculated with 1 l of a 4-day-old pre-culture grown under shaking (160 r.p.m., 30°C) in the same medium. The fermentation was run at 30°C with an aeration of 0.1 vvm. A pO2 of 40% was maintained by regulation of the stirrer speed. The pH was maintained with 10% of KOH. For continuous adsorption of the produced secondary metabolites 2% of adsorber resin XAD-16 was added to the bioreactor before autoclaving. The resin form samples of the fermentation taken every day were harvested by filtration and eluted with methanol. The concentrated eluates were analysed as previously (Müller and Gerth, 2006).

Starvation-induced fruiting body formation

S. cellulosum cells were grown in M-medium. Aggregation was monitored on P-Diff-agar (0.002% peptone (Marcor), 0.005% starch, 0.001% probion, 0.05% CaCl2 × 2H2O, 0.05% MgSO4 × 7H2O, 1.2% HEPES). The medium was adjusted to pH 7.2. Cells were grown to an OD600 4.0 harvested by centrifugation (10 000 g, 4 min) and washed once with M-media. Afterwards the cells were concentrated to an OD600 of 25. The day before spotting 3 ml of P-Diff agar was filled in the wells of a 12-well multiplate. Fifty microlitres of the concentrated cell suspensions was spotted onto P-Diff-agar. Incubation of the multiplate occurred at 32°C for several days. Aggregation was monitored visually by using a Nikon SMZ800 stereomicroscope equipped with a Nikon digital camera.

Quantitative real-time RT-PCR analysis

The one-step RT-PCR kit (Qiagen, Hilden) was used for reverse transcription and real-time PCR in one single tube. This kit uses an optimized combination of Omniscript reverse transcriptase, Sensiscript Reverse Transcriptase, and HotStarTaq DNA Polymerase and recommends the use of gene-specific primers rather than random oligomers. The reaction conditions were the following: 30 min at 50°C for reverse transcription, 15 min at 96°C for activation of the Taq polymerase and inactivation of the reverse transcriptases, 45 cycles of 15 s denaturation at 94°C, 20 s annealing at primer-specific temperatures, and 30 s of extension at 72°C. Reactions were done in duplicates. For every target gene to be analysed the specific primer pair was tested on the RNA without reverse transcription before. DNA contaminations were removed by DNase treatments as often as necessary. Also the primer pair efficiencies were tested prior to the analyses using sequential dilutions of the RNA sample and running standard curve analyses using the Rotor-Gene 6.0.23 software. For every target gene a master-mix was prepared and reactions were done in a total volume of 15 μl containing 4 ng total RNA, 0.6 μM primers, 4 U ribonuclease inhibitor (Fermentas) and SYBR green. The analyses were run on a Rotor-Gene 3000 machine. The obtained data were used to calculate the relative expression of the target genes versus the reference 16S RNA using the method of Pfaffl (2001). The relative expression ratio (R) of a target gene is calculated based on its PCR efficiency (E) and the crossing point (CP) deviation of the treated sample versus the untreated sample, and expressed in comparison with the 16S RNA reference. (ratio = (Etarget)ΔCP target (untreated − treated)/(E16S RNA reference)ΔCPreference (untreated − treated). In our case the relative expression ratio of the chiB and asgA gene is calculated based on their E-value and their CP value deviations of a treated sample (norvaline, etc.) versus the untreated control and expressed in comparison with the 16S reference gene. The values of the untreated control are set to one. Previously, we tested if 16S RNA can be used as a reference transcript under our conditions. Northern blot hybridization using a 16S-specific probe revealed that total 16S levels are stable in S. cellulosum wild-type and the Srel mutant within 1 h after addition of d-norvaline and α-methylglycosid. This was confirmed by the qRT-PCR analyses showing that in all used RNA samples the CP values (crossing point in exponential phase) for 16S RNA did not alleviate more than 1.9 cycle. This indicated that indeed comparable amounts of total RNA were used. The RT-PCR amplification products were separated on a 3% agarose gel and were proofed to be single products with the desired length. The used primer pairs 16s2 and chivo156 were designed by Kegler et al. (2006). Additionally we used the primers asgAint–: 5′-GCAGATCGACAGGCCGAGCC-3′ and asgAint+: 5′-ACGGGCACGGAGTTCTGCGC-3′ (869 nt amplicon). Generally, the qRT-PCRs were done on two separately isolated sets of RNA and principally gave similar results. Although the n-fold changes differed slightly, the reduced transcriptions levels in Srel were reproduced. Shown are experiments done on one set of RNA.

Acknowledgements

The authors would like to thank Prof K. Forchhammer for critical reading of the manuscript. We also thank Prof Wagner for kindly providing ppGpp and for sending the strains CF1693 and CF1651 with the permission of Prof J. Hernandez. This work was funded by the Bundesministerium für Bildung und Forschung within the Genomik network and the Graduiertenkolleg ‘Biochemie von Nukleoproteinkomplexen’. T. Knauber and S. Doss were members of the Graduiertenkolleg ‘Biochemie von Nukleoproteinkomplexen’.

Ancillary