Recent taxonomic studies have shown that B. amyloliquefaciens is closely related to B. subtilis and B. licheniformis, based on comparisons of their 16S rDNA and 16S-23S internal transcribed spacer (ITS) nucleotide sequences [261]. However, the genome size of B. amyloliquefaciens FZB42 is 3,9 Mb and thus significantly smaller than the genomes of all other sequenced Bacilli, including that of its close relatives B. subtilis (4,214 Mb; [7] and B. licheniformis (4,222 Mb; [11, 260]; Table 12). Furthermore, preliminary data revealed the presence of 3931 genes in the B. amyloliquefaciens FZB42 genome, whereas 4112 genes and 4286 genes are present in the genomes of B. subtilis 168 and B. licheniformis ATCC 14580 [7, 11], respectively. Almost 80% of the B. amyloliquefaciens FZB42 genes show more than 50% homology at amino acid level to genes of B. subtilis 168.

B. subtilis 168 [7], B. licheniformis ATCC 14580 [11, 260], B. halodurans C-125 [258], B. anthracis Ames [10] and B. cereus ATCC 14579 [9]

Feature

B. amyloliquefaciens FZB42

B. subtilis 168

B. licheniformis ATCC 14580

B. halodurans C-125

B. anthracis Ames

B. cereus ATCC 14579

Chromosome size (kb)

3916

4214

4222

4202

5227

5427

G+C content (%)

46

43,5

46,2

43,7

35,4

35,4

Number of genes

3931

4112

4286

4006

5508

5642

rRNA operons

11

10

7

8

11

12

Transposase genes

5

0

10

93

18

10

Table 12: Features of the B. amyloliquefaciens FZB42 genome and comparison with genomes of other Bacillus species

The unique genes of B. amyloliquefaciens FZB42 were found distributed in at least 14 DNA islands and islets around the whole genome. Interestingly, most of the DNA islands are situated in the same genetic locus where prophage-like elements are found in the B. subtilis 168 genome, indicating that those regions are susceptible to genetic rearrangements. This is the case for the first two DNA islands present in B. amyloliquefaciens FZB42. Both of them are inserted in the position of prophage-like elements 1 and 2 from B. subtilis 168, where they have replaced the majority of genes. The genome of B. amyloliquefaciens FZB42 includes only part of the prophage-like elements found in B. subtilis 168, but in addition it contains three transposases in different copy numbers (see Table 7). Interestingly, an insertion sequence element showing 51% homology to the IS3Bli1 element of B. licheniformis ATCC 14580, appears in three copies in the genome of B. amyloliquefaciens FZB42. In contrast, B. licheniformis ATCC 14580 has nine copies of the IS3Bli1 element and one more putative transposase that is closely related to a transposase previously identified in the Thermoanaerobacter tengcogensis genome (Table 12, [260]). B. subtilis 168 does not contain any transposases, and it is assumed that horizontal gene transfer is mainly achieved by bacteriophages [7]. Obviously, this is not the case for B. amyloliquefaciens FZB42 and the other sequenced Bacillus species. An extreme case is that of B. halodurans, which contains 93 transposase genes of IS elements (Table 12, [258]). On the other hand, the bacteriophages or the bacteriophage-like elements (SPβ, PBSX and the skin element) present in the genome of B. subtilis 168 are not found in the genomes of any other sequenced Bacilli, and that is also the case for B. amyloliquefaciens FZB42.

One of the most important and divergent features of microorganisms is their ability to receive and respondto different environmental signals. This ability to sense fluctuations in their environment defines the capacity of the microorganism to adapt and proliferate in its natural habitat.Signal transduction systems are responsible for such processes. Two-component regulatory systems (TCS) comprise a large family of signaltransducing proteins that accomplish the task of monitoring, processing,and responding to a plethora of divergent environmental stimuli [262]. Theyusually consist of a membrane-bound sensor kinase, which sensingan environmental stimulus autophosphorylatesat a specific histidine residue. Subsequently, this phosphateis transferred to a specific aspartate residue in the secondcomponent of the system, the cytoplasmatic response regulatorprotein [263, 264]. Phosphoryl transfer to the Asp residue in the N-terminal receiver (REC) domain of the response regulator affects the properties of its C-terminal [265, 266, 267]. The activated response regulator initiatesadaptive changes in behaviour, structure, or physiology ofthe cell, with most response regulators acting as transcriptionalrepressors or activators [263, 268].

TCS are widespread among prokaryotes. Thirty kinase-regulator pairs, that reside in an operon, have been found in B. subtilis 168 [7, 269]. Several of those two-component systems have been thoroughly characterized. BceRS, LiaRS, YxdJK and YvcPQ, located next to ABC transporter or transmembrane proteins, are the only TCS that control the cell envelope stress response in B. subtilis 168 [270, 271]. Only the biological function of BceRS and its neighbouring ABC transporter, BceAB, is known and that is to control the bacterium’s resistance against the cell wall antibiotic bacitracin [270]. BceRS, senses the presence of the toxic compound and activates the expression of the ABC transporter BceAB, which consequently facilitates the removal of the antibiotic [270, 272].

The sensor histidine kinases BceS, LiaS, YxdK and YvcQ of B. subtilis 168 belong to the recently-introduced subfamily of intramembrane-sensing histidine kinases (IM-HK) [273]. These proteins have striking similarities in their overall domain organisation: they are relatively small (less than 400aa) and their N-terminal sensing domain consists of two deduced transmembrane helices with a spacing of less than 25 amino acids. Therefore the N-terminal domain is almost entirely buried in the cytoplasmic membrane, indicating that no extracellular stimulus is detected [270]. Moreover, the cytoplasmic transmitter domain harbors only the standard features characteristic for all histidine-kinases (HisKA, HATPase_c for kinase activity and in some cases the dimerization domain HAMP), but it lacks any additional domains that would allow signal detection within the cytoplasm [270]. A very recent screen, [273], searching for this group of histidine kinases in completely sequenced microbial genomes, revealed 147 intramembrane-sensing histidine kinases (out of 5000 sensor kinases) with the majority of them found in the Firmicutes phylum (110). One striking feature of all studied IM-HKs is their common physiological role: they all seem to sense cell envelope stress and regulate genes important for the cell membrane organisation and integrity, detoxification and virulence [273]. Furthermore, most of those IM-HKs are located, together with their partner response regulator, adjacent to genes encoding ABC transporters or conserved transmembrane proteins [273].

The genome sequence of B. amyloliquefaciens FZB42 harbours thirty-one gene pairs encoding classical TCS. Twenty-one of them are orthologues to respective systems in B. subtilis 168, whereas ten are novel TCS that exhibit high similarity to respective systems of other bacteria from the Firmicutes phylum (Table 13). B. amyloliquefaciens FZB42 lacks nine two-component regulatory systems encoded by B. subtilis 168 (YbdKJ, YcbLM, YccHG, YesMN, YfiJK, YkoGH, YvcPQ, YvfUT, YxjML). Among them, only YvcPQ is of known function, and that is associated with the sensing of cell envelope stress, as already mentioned above.

The analysis of the thirty-one sensor kinases present in the genome of B. amyloliquefaciens FZB42, using the simple modular architecture research tool (SMART, http://smart.embl-heidelberg.de; [274]), accompanied by searches for genomic context conservation and sequence homology, revealed that the bacterium encodes five potential IM-HK. Three of them (BceS, YxdK and LiaS) are direct orthologues to corresponding proteins encoded in B. subtilis 168 (70%, 97% and 75% similarity on amino acid level respectively) and two (RBAM00197, RBAM03294) are novel proteins. bceS, yxdK, liaS and their cognate response regulators are localized in the same genomic context in B. amyloliquefaciens FZB42 as in B. subtilis 168, i.e. next to genes encoding ABC transporters (ytsCD, yxdLM) or a transmembrane protein (yvqF). The high degree of conservation of those TCS between B. amyloliquefaciens FZB42, B. subtilis 168 and B. licheniformis ATCC 14580, implies that their role in B. amyloliquefaciens FZB42 has also to do with sensing the cell envelope stress as it is the case for the other two organisms [270, 271, 275].

The histidine kinases RBAM00197 and RBAM03294 display all the structural characteristics of IM-HK. RBAM00197 (340 aa) and its cognate response regulator RBAM00196 are located next to genes that display similarity to ABC transporters (RBAM00198/00199) (Table 13). In addition, the RBAM00196/00197-RBAM00198/00199 system displays similarity to the BceRS-BceAB, YxdJK- YxdLM systems of B. subtilis 168, strengthening our prediction that it also plays a role in cell envelope stress response. Interestingly, these sequences comprise part of a 22 kb size DNA island inserted in the genome of B. amyloliquefaciens FZB42 at the position where the prophage-like element 1 is located in the B. subtilis 168 genome.

RBAM name/position of HK-RR

Protein name/accession number/ identities of closest homologue

RBAM name of neighbou-ring ABC transpo-rters

Protein name/accession number/identities of closest homologue to ABC transporters

Organism

00197 (IM-HK)/210-00196/211

BceS/O35044/26%-BceR/O34951/42%

00198-00199

BceA/O34697/46%-BceB/O34741/23%

B. subtilis 168

03294 (IM-HK)/332-003295/3321

AAP08928/39%-AAP08927/56%

03295-03296

AAP08927/85%

B. cereus ATCC14579/B. subtilis 168

00207/221-00208/222

NP_348870/28%-NP_346816/56%

00209-00210-00211

NP_346812/66%-NP_346813/37%-NP_346814/32%

Clostridium acetobutylicum ATCC 824

00546/559-00545/558

EAO53032/64%-EAO53033/86%

-

-

B. thuringiensis serovar israelensis ATCC 35646

01839/1866-01840/1867

EAT23926/46%-EAT23927/55%

-

-

Clostridium phytofermentans ISDg

02015/2088-02014/2087

AAU24296/78%-AAU24295/75%

-

-

B. licheniformis ATCC 14580

03132/3166-03131/ 3165

NP_978181/50%-NP_978180/76%

03133-03134-03135

NP_978182/74%-NP_978183/48%-NP_978184/48%

B. cereus ATCC 10987

03180/3211-03181/3212

SpaK/AAB91593/54%-SpaR/AAB9159480%

03182-03183-03184

SpaG/AAB91595/49%-SpaE/AB91596/50%-SpaF/AAB91597/74%

B. subtilis ATCC 6633

03606/3610-03605/3610

YP_085861/51%-YP_085862/58%

03607-03608-03609

YP_085858/39%-YP_085859/59%-YP_085860/86%

B. cereus E33L

03780/3771-03781/3772

MrsK2/CAB60253/98%-MrsR2/CAB60254/99%

03782-03783-03784

MrsF/CAB60255/99%-MrsG/CAB60256/98%-MrsE/CAB60257/99%

Bacillus sp. HIL-Y85

Table 13: Novel two-component regulatory systems in B. amyloliquefaciens FZB42

RBAM name is the protein name in B. amyloliquefaciens FZB42. The name of the proteins (omitting the RBAM prefix) and their positions, indicated in kb, are given in the first column. The two-component sensor histidine kinase protein (HK) is written first, followed by the two-component response regulator (RR). Similarities to the closest homologue are derived by BLASTX alignment and are indicated on amino acid level for the overall protein length. The closest homologue’s protein name is given if it is known. Minus indicates the absence of a neighbouring ABC transporter to the two-component system. IM-HK; intramembrane histidine kinase

On the other hand, the two-component system RBAM03294/03295 comprises a two-gene insertion in a region that is conserved between the genomes of B. amyloliquefaciens FZB42 and B. subtilis 168. It is located next to bmrA, a multidrug ABC transporter that is functionally active in B. subtilis and is constitutively expressed throughout growth [276]. Therefore, it would be intriguing to check if the inserted two-component system has a functional link with the multidrug ABC transporter, and whether it alters the regulation of bmrA in B. amyloliquefaciens FZB42. Furthermore, a more general function of the IM-HK RBAM03294 in the cell envelope stress response should not be excluded.

The cell envelope is the first and major line of defence againstthreats from the environment. It gives the cell its shape, counteracts the high inner osmotic pressure and providesan important sensory interface and molecular sieve between abacterial cell and its surroundings, mediating both informationflow and controlled transport of solutes. Therefore, monitoring the cell envelope integrityand adequately changing its composition is critical for survival. B. amyloliquefaciens FZB42 possesses five candidate two-component systems involved in the cell envelope stress response, two of which are novel members among the sequenced bacteria, as mentioned above. The closely related bacterium B. licheniformis ATCC 14580 possesses only three TCS for the same scope, BceRS, LiaRS and YxdJK, shared by B. subtilis 168 and B. amyloliquefaciens FZB42 [275]. B. subtilis 168 has an additional TCS, YvcPQ, that is not found in the two other organisms. All these data indicate substantial overlap, but also a degree of differentiation between the three closely related bacteria in respect with their response to envelope stress. Different environmental cues trigger presumably distinct responses in the three bacteria, which allow them to adopt different strategies to survive in their natural habitat, the soil.

Apart from RBAM00196/00197 and RBAM03294/03295, B. amyloliquefaciens FZB42 has eight more novel two-component regulatory systems that show similarity to systems present in other Bacilli or Clostridia (Table 13). Five of them are located adjacent to novel putative ABC transporters. For example, the TCS RBAM03780/03781 is highly homologous to the MrsK2R2 proteins of Bacillus sp.HIL-Y85 (92% and 99% similarity on amino acid level respectively), which control the immunity against the lantibiotic mersacidin, produced by the strain [36, 37]. In parallel, the adjacent putative ABC transporter proteins RBAM03782-03784 are almost identical to the MrsFGE transport system (99%, 87% and 84% similarity on amino acid level respectively), which confers self-protection against mersacidin to the producer bacterium [36, 37]. Thereby, it can be assumed that B. amyloliquefaciens FZB42 is also immune to mersacidin. Taking into consideration that the combination of two-component and ABC transporter systems is characteristic of detoxification units, which can selectively sense a harmful for the cell compound and export it into the extracellular space [277, 278], we can postulate that the respective systems in B. amyloliquefaciens FZB42 control immunity against various antibiotics, produced either by B. amyloliquefaciens FZB42 itself or by other competing microorganisms. Thereby, these detoxification units provide FZB42 with defensive mechanisms in order to survive in a highly competitive environment such as the soil. This resistant capacity of B. amyloliquefaciens FZB42 makes the bacterium more competent of surviving in the plant roots, where it exerts its biocontrol activity.

A key issue for the proper functioning of a signal transduction system is its ability to balance the input signalling with the output response. This is thought to occur through regulation of the overall phosphorylation state of the system and/or through regulation of the activity of the output domain of the response regulator. The Rap (response regulator aspartate phosphatase) phosphatases are a conserved family of regulatory proteins that negatively influence many response regulators [234]. B. subtilis 168 encodes 11 Rap proteins, eight of which constitute operons with downstream phr genes [249]. However, the expression of phr genes is usually controlled by an additional σ^H-dependent promoter [234]. Pre-Phr is synthesized as a small protein with a putative signal peptide, which is cleaved and secreted as a pentapeptide to the external milieu [250]. The Phr pentapeptide is imported again into the cells by an oligopeptide permease [251], and inhibits the activity of its cognate Rap protein [166]. The Rap proteins inhibit the action of the target response regulators either by dephosphorylating them [279] or by binding to the DNA-binding domain of the response regulator [167, 246].

B. amyloliquefaciens FZB42 has six Rap proteins that are also present in B. subtilis 168; RapA, RapC, RapF, RapB, RapD and RapJ (also the cognate Phr of the first three are conserved). Due to high similarity of the Rap proteins and their target response regulators in B. amyloliquefaciens FZB42 and B. subtilis 168, it can be assumed that the function of the Rap proteins is conserved in the two bacteria. Therefore, RapA, RapB probably have a negative influence on the initiation of sporulation, by dephosphorylating Spo0F [279], while RapC and RapF probably inhibit binding of ComA to its target genes [167, 246]. It is noteworthy that B. amyloliquefaciens FZB42 lacks orthologues of RapG and RapH that negatively regulate DegU in B. subtilis 168.

In addition, B. amyloliquefaciens FZB42 possesses three novel putative Rap proteins (Table 14). The novel Rap proteins contain tetratricopeptide (TRP) domains, similarly to other members of the Rap family [280]. The TRP domains are thought to be directly implicated in protein-protein interactions [281]. It is considered that TRP domains play an important role in the interaction between the Rap protein and its cognate Phr [250] and it is speculated that Rap proteins, whose inhibitory function is not associated with dephosphorylation of their target response regulators, bind to the target regulator through their TRP domains [167, 248]. Moreover, no cognate phr genes were identified downstream of the three novel rap genes of B. amyloliquefaciens FZB42. Interestingly, RBAM03282 is situated at the same genetic locus where RapG is in B. subtilis 168. However, RBAM03282 shows very low homology (maximum 27% on amino acid level) to all Rap proteins of B. subtilis 168 but relatively high homology to a putative Rap protein of Bacillus licheniformis ATCC 14580 (YP_080123), which has not been studied so far (48% on amino acid level). We have shown that RBAM03282 is involved in the regulation of the bmyD operon and, therefore, designated it with a new name, RapX (see chapter 3.4 and later on in Discussion). The functions of the two other novel Rap proteins remain to be identified.

The position of the proteins is given in kb and the closest homologue is presented, as derived by BLASTX alignment. Similarities on amino acid level are indicated for the aligned part of the sequences. * The orthologous RapA protein of B. subtilis 168 is present in another position on the genome of B. amyloliquefaciens FZB42

Protein name

Position

Size (aa)

Protein name/accession number of closest homologue

Identities (aa level)

Organism

RBAM00430

462

358

RapH/P40771

152/353 (43%)

B. subtilis 168

RBAM02010

2082

378

RapA/Q00828*

189/379 (49%)

B. subtilis 168

RBAM03832 (RapX)

3830

371

Putative response regulator aspartate phosphatase/YP_080123

173/358 (48%)

B. licheniformis ATCC 14580

Table 14: Novel Rap (response regulator aspartate phosphatase) proteins in the genome of B. amyloliquefaciens FZB42

The model Gram-positive bacterium B. subtilis 168 has 17 σ factors [7], seven of which deal with extracytoplasmic functions and therefore are designated as ECF σ factors (σ^M, σ^V, σ^W, σ^X, σ^Y, σ^Z,σ^ylaC; [282]). On the other hand, B. amyloliquefaciens FZB42 possesses 16 σ factors with six of them being ECF σ factors. The two organisms retain conserved all their non-ECF σ factors. Five of the ECF σ factors are common between B. subtilis 168 and B. amyloliquefaciens FZB42 (σ^M, σ^V, σ^W, σ^X,σ^ylaC), whereas the latter lacks σ^Yand σ^Z, but has in addition a novel putative ECF σ factor, Sig01.

ECF σ factors are typically regulated by a co-transcribed membrane-bound anti-sigma factor that keeps the sigma factor inactive, bound in the cell membrane [283]. Sig01 is not an exception of this rule, since a putative anti-sigma factor is located downstream of its coding region. σ⁰¹ and anti-σ⁰¹ of B. amyloliquefaciens FZB42 display low similarity on amino acid level to a novel ECF σ factor (Bli04171) and its cognate anti-sigma factor (Bli04170)(40% and 27%, respectively) found in B.licheniformis strain ATCC 14580 [11]. Recently it was shown that Bli04171 ECFσ factor (designated σ^ecfH hereafter) is part of the regulatory network that controls the cell envelope stress response in B. licheniformis ATCC 14580, since its expression was induced seven- and five-fold after vancomycin and bacitracin treatment, respectively [275]. These results indicate that σ⁰¹ could be also involved in the cell envelope stress response of B. amyloliquefaciens FZB42. Until now, the only knowledge we have about the function of σ⁰¹ is that it does not control the expression of bacillomycin D (see chapter 3.4) and most probably the expression of all lipopeptides and polyketides produced by the strain (data not shown). Therefore, it would be intriguing to find out whether and how this novel ECF σ factor contributes to cell envelope stress response.

It is noteworthy that a core of five ECF sigma factors are conserved in B. amyloliquefaciens FZB42, B. licheniformis ATCC 14580 and B. subtilis 168 (σ^M, σ^V, σ^W, σ^X,σ^ylaC). B. amyloliquefaciens FZB42 has one additional ECF σ factor (Sig01), B. subtilis 168 two (σ^Y and σ^Z) [282] and B. licheniformis ATCC 14580 has three (σ^Y, σ^efcG and σ^ecfH) [275]. These findings indicate, once more again, regulatory divergence, but also a partial overlap between the three Bacilli in respect with their response to envelope stress. Interestingly, B. halodurans strain C-125 has 20 σ factors with only half of them conserved in B. subtilis 168 [258]. Eleven σ factors belong to the ECF family, but only one (σ^W) is homologous to the ECF σ factors of B. subtilis 168, indicating that its unique ECF σ factors regulate special mechanisms that allow the bacterium to live in an alkaline environment [258].

Genetic or natural competence is a physiological differentiation state in which bacteria are able to take up exogenous DNA from the medium. The molecular processes involved in the competence development in the model gram-positive bacterium B. subtilis have been studied extensively over the last decades. The establishment of competence requires at least 25 different genes, acting together in a finely intertwined cascade of signal transduction pathways and regulatory circuits, reviewed in [284]. B. amyloliquefaciens FZB42 is a natural competent strain (deviating from the transformation protocol published for B. subtilis 168 [205])was developed in this study, see Materials and Methods) and its genome contains orthologs of all genes involved in the development of competence in B. subtilis 168. In contrast, B. licheniformis ATCC 14580 is not naturally transformable due to the lack of a comS homologue and to a transposon insertion into the comP gene [260].

Despite that the majority of competence genes in B. amyloliquefaciens FZB42 are highly homologous to their counterparts of B. subtilis 168, the genes that control the competence quorum-sensing system of B. amyloliquefaciens FZB42 (comQ, comX, comP) exhibit low similarity to the respective genes of B. subtilis 168 (36%, 31% and 55%, respectively). Such low sequence similarity of the competence quorum-sensing system has been already observed among various Bacillus isolates [285]. The genetic polymorphism extends through comQ, comX and the 5' two-thirds of comP [285], as it is the case in B. amyloliquefaciens FZB42. Furthermore, it was exhibited that this genetic variability is correlated with specificity in the quorum-sensing response, so that each pheromone is sensed only by its cognate receptor [286]. The quorum-sensing locus may have been introduced by horizontal transmission into a common ancestor of Bacillus strains and thereafter subjected to strong positive selection, which resulted into a dramatic sequence polymorphism and pheromone specificity [287].

In addition, B. amyloliquefaciens FZB42 was found to be competent in an earlier stage of growth than its closely related B. subtilis 168; the former showed increased transformation rates during mid to late exponential phase (see Materials and Methods), whereas the latter is known to become competent upon entry into stationary phase [204].It is tempting tospeculate that B. amyloliquefaciens FZB42 exhibits a distinct temporal regulation of its competence gene circuit from its sibling B. subtilis 168, apart from maintaining its specific pheromone (ComX)-modificator (ComQ) pair for initiating the competence process. Identifying the differentially regulated competence genes between the two organisms would be a future challenge, since it will permit the genetic manipulation of two organisms in order to modify/improve their DNA uptake, both in terms of yield and of chronological occurrence.

B. amyloliquefaciens FZB42 encodes eight gene clusters which are responsible for the nonribosomal synthesis of secondary metabolites. These operons comprise 8% of the bacterium’s genome and encode for peptide/polyketide antibiotics and a siderophore. We have verified the functionality of all eight gene clusters and we believe that the secondary metabolites produced enable B. amyloliquefaciens FZB42 to dominate over competing organisms within its natural environment and/or serve as signals that trigger cellular responses to the receiving organisms in the surrounding [288, 289].

In detail, B. amyloliquefaciens FZB42 is able to produce three distinct lipopeptide antibiotics: surfactin, fengycin and bacillomycin D. All three lipopeptides are synthesized nonribosomally according to the multicarrier thiotemplate mechanism (see introduction; review by [48]. Surfactin isencoded by the srf operon, which is also found in the genome of B. subtilis 168, a strain unable to produce lipopeptides or polyketides due to frameshift mutation on the sfp gene [174]. The chromosomal locus, as well as the organisation of the genes and modules within the srf, operon are identical among the two bacteria; only the downstream-flanking genes of the srf operon vary. One of these genes is aat, a putative transcriptional regulator. No dramatic change in the production of surfactin or of other lipopeptides was observed when aat was deleted in strain FZB42 (data not shown). Moreover, the aat deletion had no effect on the transcriptional regulation of bacillomycin D (see chapter 3.4). However, we have no data about the effects caused by the aat deletion on the transcriptional regulation of surfactin, fengycin and the polyketides. Therefore, the putative function of aat in the regulation of lipopeptides and polyketides should be more closely examined.

Surfactin is known to provide antibacterial activity to the producer strain, since it can penetrate bacterial membranes and disturb their function [290]. In addition, it is essential for the swarming motility of the microorganism [132, 133, 134, 135], as well as for the formation of biofilms [136, 137]. Thereby, surfactin controls colonization of surfaces and can aid in acquisition of nutrients though its surface-wetting and detergent properties [291]. Recently, it was shown that surfactin is required for the development of aerial structures in the biofilms produced by B. subtilis, which resemble the fruiting-body formation by myxobacteria [288, 292]. Moreover, it was shown that the surfactin produced by B. subtilis acts antagonistically against Streptomyces coelicolor by inhibiting its development of aerial hyphae and spores [288]. Interestingly, surfactin did not inhibit the vegetative growth of Streptomyces coelicolor, as a typical antibiotic would do, but prevented a specific developmental process of Streptomyces coelicolor [288]. Therefore, surfactin protects B. amyloliquefaciens FZB42 against bacteria [197] and enables it to form biofilms, equipping thus the bacterium with powerful antagonistic advantages during surface colonization.

The bmy and fen operons are responsible for the biosynthesis of bacillomycin D and fengycin in B. amyloliquefaciens FZB42, respectively. These gene clusters are located at the same chromosomal locus with a distance of about 25 kb between them. Interestingly, the gene clusters directing the biosynthesis of bacillomycin L in B. subtilis A1/3 and iturin A (a lipopeptide with similar structure as bacillomycin D) in B. subtilis RB14 are situated at the same position as the bmy operon in B. amyloliquefaciens FZB42. In addition, the pps operon in B. subtilis 168, which is assigned to fengycin biosynthesis (despite of the strain’s inability to produce it), as well as the fen operon in the producer B. subtilis strains F29-3 [222] and A1/3 [140], are located at the same genetic locus as the fen operon in B. amyloliquefaciens FZB42. On the other hand, the genome of B. subtilis ATCC 6633 contains the myc operon (directing the biosynthesis of mycosubtilin, an iturin-like lipopeptide) at the same position that the fen operon occupies in strains F29-3 and A1/3 [63]. These findings indicate high degree of genetic flexibility in this region and suggest that additional nonribosomal peptide synthetases (NRPS) can be integrated in it either as an insertion or as a substitution of already existing NRPS operons.

Synthesis of bacillomycin D occurs according to the multicarrier thiotemplate mechanism. We have tried to verify the biosynthetic pathway of bacillomycin D by disrupting one by one the last six modules (in specific the respective adenylation domains) of the nonribosomal peptide synthetase and then by trying to identify the intermediate elongation variants (see Fig. 20 and chapter 3.3). However, the expected products could not be detected by MALDI-TOF MS analysis of neither culture filtrate extracts nor sonificated cell extracts. This indicates that only the full length lipopeptide is exerted from the cell, whereas the intermediate products are covalently attached to the multienzyme system, from which they can not be completely detached, even after sonification. A possible way to achieve detachment of the products from the enzymes would be reaction with a suitable thiol- compound, such as cysteine or cysteamine. Reaction with such a compound, under the appropriate conditions, could lead in the transfer of the thioester bound product onto the free thiol-group, rendering thus possible the identification of the obtained intermediate variants of bacillomycin D by MALDI-TOF MS. We are currently pursuing this issue further, in collaboration with Dr. J. Vater.

Bacillomycin D and fengycin inhibit the growth of various phytopathogenic fungi. Abolishment of each antibiotic led to decreased inhibition of the fungal growth, compared to the wild type strain; the effect of fengycin was smaller than that of bacillomycin D. Deletion of both antibiotics deprived B. amyloliquefaciens FZB42 of its antifungal abilities (see chapter 3.3). Thereby, we have demonstrated a synergistic action of both lipopeptide antibiotics against the target microorganism, a phenomenon previously described for secondary metabolites produced by actinomycetes. The synergistic activity of the antibiotics had been interpreted as an evolved adaptation mechanism of the producer organism in order to compete with other microorganisms and maintain its sessile lifestyle [293]. In the case of B. amyloliquefaciens FZB42, the level of fengycin production is considerably lower than that of bacillomycin D and thus the observed synergistic effect of the antifungal compounds was unexpected.

Interestingly, several of the mutant derivatives of B. amyloliquefaciens FZB42 have opposing effects on the production of bacillomycin D and fengycin. In particular, we have shown that the comA, sigH and sigB mutations reduce bmy expression by several-fold (Fig. 28A-D), whereas preliminary data obtained by MALDI-TOF MS analysis of the respective mutant strains show enhanced production of fengycin (see chapter 3.5). On the other hand, the bmyD (AK1) and degU (TF1) mutant strains, that completely lack bacillomycin D, did not display an elevated production of fengycin. These results suggest that the same regulatory pathways (and not itself the production of bacillomycin D) may opposingly direct the regulation of both antifungal compounds. The bacterium can, thereby, enhance the expression of fengycin in conditions where the expression of bacillomycin D is low. In this way, any single fengycin or bacillomycin D mutant retains a considerable inhibitory effect on fungal growth compared to the double mutant.

It is noteworthy that bacillomycin D and fengycin, in contrast to surfactin, have no effect on biofilm formation (data not shown; [137]). Recently, iturin A (that belongs to the same family of peptide antibiotics like bacillomycin D) was shown to inhibit sporulation of Streptomyces scabies, but not its growth [289]. This suggests that bacillomycin D and fengycin might have additional roles as secondary messengers.

The genome of B. amyloliquefaciens FZB42 contains three giant modular polyketide gene clusters (for details see [197]. The bae operon is responsible for the biosynthesis of bacillaene, a conjugated hexaene with a linear structure [294], whose chemical structure is still unknown. The dif gene cluster is devoted to the synthesis of difficidin and oxydifficidin, which are highly unsaturated 22-member macrolides with a rare phosphate group [295]. The third polyketide gene cluster is designated pks2 and is involved in the synthesis of macrolactin (K.Schneider and Xiao-Hua Chen, unpublished results). Notably, this is the first time that the complete gene clusters involved in the biosynthesis of bacillaene and difficidin/oxydifficidin are defined. Modular organisation of the three pks clusters in B. amyloliquefaciens FZB42 revealed an unusual trans-AT architecture, which indicates that all PKS modules lack an AT domain and are complemented by ATs encoded on isolated genes [197]. This unusual trans-AT architecture was recently described for a polyketide synthase-peptide synthetase gene cluster of an uncultured bacterial symbiont of Paederus beetles [296].

B. subtilis 168 possesses only one large polyketide gene cluster, designated pksX. However, this strain is unable to produce the respective polyketide, due to a mutation in the sfp (4'-phosphopantetheinyl transferase) gene [174]. Therefore, until recently it was not known which polyketide is synthesized by the pksX cluster. We have demonstrated that B. subtilis OKB105, a sfp ⁺ derivative of B. subtilis 168, is able to produce bacillaene indicating that pksX directs synthesis of this polyketide [197].

Bacillaene and difficidin/oxydifficidin exhibit strong antibacterial activities, whereas macrolactin inhibits the growth of B. megaterium and E. carotovora only weakly [197]. Interestingly, bioautographs of the wild type strain and the sfp (CH3) mutant derivative of B. amyloliquefaciens FZB42 (that is deficient in lipopeptide and polyketide synthesis) on B. megaterium lawn, revealed the production of an antibacterial compound with unknown structure [197].

In addition, the genome of B. amyloliquefaciens FZB42 contains the bac operon that controls the synthesis of the dipeptide bacilysin [217]. Organisation and localization of the bac operon in the genomes of B. amyloliquefaciens FZB42 and B. subtilis 168 are identical. Recently it was shown in B. subtilis 168 that genes bacDE are involved in amino acid ligation and bacilysin immunity, respectively [297].

Bacilysin is active against a wide range of bacteria [218]. It was suggested that its antibacterial spectrum overlaps with that of the polyketide compounds bacillaene and difficidin [137]. However, bacilysin does not account for the remaining antibacterial compound detected in the sfp mutant derivative of B. amyloliquefaciens FZB42, as observed in bioautographs on B. megaterium lawn [298]. This is the first evidence for an additional ribosomally produced, antibacterial compound of B. amyloliquefaciens FZB42.

The last operon involved in the nonribosomal synthesis of a compound in B. amyloliquefaciens FZB42 is that of dhb. The dhbACEBF operon is involved in the synthesis of 2,3-dihydroxybenzoate (DHB) as well as its modification and esterification to the iron siderophore bacillibactin [78] that enables microorganisms to efficiently scavenge iron [220, 221]. Bacillibactin was detected in the culture filtrate extracts of B. amyloliquefaciens FZB42, verifying the functionality of the dhb operon (Fig. 37). Both the organisation and the localization of the operon are conserved between B. amyloliquefaciens FZB42 and B. subtilis 168.

Iron is an essential trace element for all bacteria [299]. In many aerobic, neutral or alkaline environments, Fe⁺² is present in only suboptimal concentrations due to its low solubility. Microorganisms have therefore developed elaborate systems for scavenging iron from environmental sources. These systems frequently involve the synthesis of high-affinity chelators, their excertion into the environment, and the recapturing of the iron-loaded chelator via affinity transport systems [300, 301]. Similarly, iron limitation triggers the production of bacillibactin [255] in Bacilli, which is then secreted from the cell to act as an iron scavenger and then is re-imported into the cell, where its hydrolysis leads to release of cytosolic iron [256]. In a highly competitive environment, such as the plant rhizosphere, the microorganisms that can make use of the environmental iron are more likely to survive. Therefore, it is possible that B. amyloliquefaciens FZB42 enhances plant growth by depriving soil pathogenic microorganims of iron, like already proposed for other plant growth promoting rhizobacteria (PGPR) [302].

In conclusion, the genome of B. amyloliquefaciens FZB42 contains eight operons that direct nonribosomal synthesis of three lipopeptides, three polyketides, one dipeptide and a siderophore. These compounds exhibit strong antifungal and antibacterial activities and enable the bacterium to survive in its natural environment. As B. amyloliquefaciens FZB42 colonizes the plant roots, it inhibits growth of phytopathogenic bacteria or fungi either by depriving them of the essential iron (through the action of bacillibactin) or by directly inhibiting their growth and/or certain of their developmental processes (through the actions of lipopeptides and polyketides). We must note that antibiotic activity is possibly not the only function of lipopeptides and polyketides produced by B. amyloliquefaciens FZB42. Surfactin is involved in intercellular signalling [288] and may be other secondary metabolites play also a role in interspecies communication and thereby affect the developmental pathway of a bacterium without influencing its vegetative growth. Until now, only preliminary studies have been performed with cocultivated bacteria, a situation that resembles more the natural settings.

Interestingly, B. amyloliquefaciens FZB42 does not produce most of the ribosomally synthesized peptide antibiotics that B. subtilis 168 does. The genome of B. amyloliquefaciens FZB42 does not contain the gene clusters of bacteriocins subtilosin (sbo-alb) and the SPβ proghage-encoded sublancin (see chapter 1.3.1). Moreover, the bacterium does not produce the antibiotic-like killing factor Skf (sporulation killing factor) or the toxic protein SdpC (sporulation delay protein) [303]. Notably, SdpC is present only in B. subtilis strains and orthologues of it have not been identified in other bacteria including all Bacillus species sequenced to date [223].

Recently, it was reported that the sfp derivative of B. amyloliquefaciens FZB42 inhibits the growth of a sigW deficient strain of B. subtilis as strongly as the wild-type strain FZB42 [223]. B. amyloliquefaciens FZB42 exhibited one of the strongest inhibitory effects on a sigW mutant of B. subtilis, among several members of the Bacilli family tested. This indicates that B. amyloliquefaciens FZB42 encodes ribosomally synthesized peptide(s) or toxic protein(s) with antibacterial function, as observed in experiments on B. megaterium lawn performed in our lab. The ydbST and fosB (yndN) genes, present also in the genome of B. amyloliquefaciens FZB42, contribute to resistance against these antimicrobial compound(s), albeit to a smaller extent than σ^W [223]. Other members of the σ^W regulon could be also involved in promoting resistance against the ribosomally synthesized antibacterial compounds of B. amyloliquefaciens FZB42.

DegU is a two-component system response regulator of the LuxR-FixJ family, whose members have a helix-turn-helix (HTH) structure at their C-terminus [306]. It is known to control many cellular processes, including exoprotease production, competence development, motility and to trigger post-exponential-phase responses under growth limiting conditions [245, 307]. Recently, two genome-wide transcriptional profiling studies have been published for the role of DegU in B. subtilis [253, 257]. Although none of them directly compared the gene expression in the wild-type strain versus that of the isogenic degU mutant, an extensive regulon was identified for DegU. In addition, DegU has been associated with response of B. subtilis to high salinity [205, 254].

To our knowledge, this is the first time that DegU is demonstrated to play a central role in the regulation of a nonribosomally synthesized antibiotic. A series of in-vivo and in-vitro data demonstrated that DegU directly activates the expression of P_bmy (see also Figs. 26, 27 and 30). In addition, the results from the EMSA and the DNase I footprinting experiments (see also Figs. 30 and 31) coincided and pointed out that DegU retains two distinct DNA binding-sites at the bmy promoter. The first site, Site I, is located relative near to the transcriptional start, between bps -123 and -99 (relative to the transcriptional start), whereas the second one, Site II, is situated further upstream between -201 and -172 (see Figs 25 and 31). Binding of DegU to the latter upstream site is absolutely essential for the optimal activation of the promoter (see Figs. 22B and 26A). The existence of a third DegU binding-site that is located more upstream than –230 bps should not be excluded, since our in-vitro footprint data do not provide conclusive evidence for this region.

This is the second study to date, which has directly monitored the binding of DegU to a promoter by footprinting analysis. The protection that DegU offers to the DNA at its two binding-sites is quite weak, similarly to that exhibited in the previous study by Hamoen et al. (2000). On the contrary, strong hypersensitive sites can be observed adjacently to the two DNA binding-sites, implying that the binding of DegU to its sites rearranges the local DNA architecture, probably by inducing strong DNA-bending, constraint or even unwinding, which makes the DNA more accessible to DNase I attack. This correlates well to the role of DegU in the activation of the comK promoter [231]. Based on a series of data, Hamoen et al. proposed that DegU alters the shape of the ~ 4 DNA helixes that separate the tandem ComK boxes (possibly by unwinding and/or bending the DNA), and, thereby, facilitates the binding of ComK to them; ComK can then stimulate the transcription of its own promoter.

There are several reasons why the binding of DegU to the DNA only weakly protects the latter against DNase I attack. First, both in our experiments and the Hamoen et al. (2000) study, unphosphorylated DegU was used for the footprinting analysis. Although in many studies response regulators are used in their unphosphorylated form in order to demonstrate DNA-binding, the use of the phosphorylated response regulator can often result in more distinct/extended regions being protected against DNAse I cleavage [162, 308]. I also performed the footprinitng analysis with phosphorylated DegU (after incubation with “cold” acetyl phosphate) and obtained very similar results to those of the unphosphorylated DegU. Even though incubation of a response regulator with acetyl phosphate should result in its phosphorylation, no direct proof can be provided whether transphosphorylation actually took place, without using radioactive acetyl phosphate. Nevertheless, experiments with unphosphorylated response regulator can provide important information on its DNA-binding ability as seen before in many cases, such as that of UhpA, ComA and Spo0A [162, 308].

Another reason for the weak protection patterns of DegU is the nature of its binding-sites. A/T-rich DNA regions, such as the DNA-binding-sites of DegU, are more curved and therefore less accessible to DNase I, even when the DNA is naked without any protein bound to it. Thus, the A/T-rich DNA-binding-sites appear protected even in the absence of their binding partner. Hydroxyl radical footprinting has given more clear results in such cases, and should be considered as an alternative method in future studies.

Despite the fact that in-vitro assays monitoring the binding of DegU to DNA promoter regions are limited [231, 244, 248], two possible motifs have been suggested as putative DegU recognition-sites [244, 245]. Shimane et al (2004) based on in-vivo data from the aprE and comK promoters proposed that DegU recognises an A/T-rich motif (either a tandem repeat of a 5-nucloetide sequence TAAAT or an inverted repeat of ATTTA-N7-TAAAT), whereas Dartois et al. (1998) based on in-vivo studies in the wapA promoter and an alignment of DegU-regulated promoters, proposed AGAA-N₁₁-TTCAG as the recognition site for DegU. Although none of these studies provides conclusive evidence and they are contradicting to each other, degenerate forms of the latter motif could be identified in the DegU protected regions at the bmy promoter region (both sites I and II; see also Fig. 25), whereas the A/T-rich motifs proposed by Shimane et al (2004) were part of the hypersensitive sites that were generated at the bmy promoter region upon addition of DegU in the DNase I footprints. In any case, further experimental evidence, involving extensive site-directed mutagenesis, will be required to identify the consensus sequence recognised by DegU in P_bmy and/or other promoters.

All previous studies which have carefully assessed the binding of DegU (always unphosphorylated DegU used) to different promoter regions (comK and aprE; [231, 244, 248]) have shown a picture similar to the one exhibited in this study (Fig. 30). Increasing amounts of DegU cause a gradual shift of the DNA fragment. In most cases, DNA binding-proteins that recognise defined motifs at the DNA and bind tightly to it produce distinct shifts that their number reflects how many binding-sites are present at this DNA fragment. If there is not enough protein in the assay to fully occupy the DNA binding-site(s) then the bound (shifted) and the unbound DNA are in equilibrium. The pattern of the band-shift assays produced by DegU at the bmy promoter raise interesting mechanistic scenarios in respect with how DegU binds to the promoter and activates transcription. It seems plausible that initial DegU binding serves as an anchor to further recruit DegU molecules to the promoter. However till now, little is known about the multimerisation state of DegU when it binds to its target sites, or what the helix-turn-helix of each DegU molecule recognises as DNA binding-motif.

As mentioned above, the in-vivo data (see also Figs. 22 and 26) of this study pinpoint an upstream regulatory region as absolutely essential for the maximal activation of the P_bmy promoter by DegU and the rest of the global regulators identified here to be involved in the bmy expression (see also below; most of them are shown or proposed to mediate their effects indirectly, via DegU). Nevertheless, the in-vitro data (Figs. 30 and 31) suggest that DegU retains at least two DNA binding-sites. The first of them (site I) is centred in a region that is not shown to be able to activate per se the P_bmy promoter, i.e site I is included in AK10, which does not show a significant difference in its activity from AK11, which lacks the DegU recognition site I (Figs. 21, 22 and 25). On the contrary site II of DegU is located within the upstream DNA region that is necessary for the promoter activation (Figs. 21, 22 and 25). We propose that the binding of DegU to Site I triggers a sharp DNA bend directly downstream of it and thus enables the DegU bound to Site II to activate the promoter (Fig. 38). This is a rather common transcriptional activation mechanism. Figure 38: Proposed mechanism of action of DegU on the P_bmy promoter.

This figure illustrates how DegU might activate the function of the vegetative RNA polymerase (RNAP) on the P_bmy promoter. DegU is shown to bind to its identified DNA sites in dimers for presentational reasons (direct information for this is missing). Site I and site II are located between -123 and -99 bps, and between -201 and -177 bps (relative to the transcriptional start), respectively. The DNA U-turn, shown directly downstream of site I, reflects to the strong hypersensitivity that this site exhibits in the DNase I footrpinting experiments. The DNA bending that possibly occurs directly upstream of site II (see DNase I footprints) is not shown in this picture for presentational reasons. The two C-terminal domains of the α subunit (designated as αCTD) are tethered with a flexible linker to the RNAP-bound N-terminal domains of the α subunit (designated as αNTD). Stars indicate possible interactions between RNAP and DegU. The binding of DegU to its site I alters the relative location of the DegU, bound to the site II, towards RNAP and renders the two in position to interact with each other.

It is generally accepted that DegU has two modes of action: phosphorylated DegU directly activates degradative enzyme production and represses motility, whereas unphosphorylated DegU directly stimulates competence [307] through binding to the comK promoter [231]. The belief that only unphosphorylated DegU is required for competence was supported by the observation that hyperphosphorylation of DegU (degU32(Hy) shows a 7-fold increase in the stability of the phosphorylated form of DegU) or inactivation of the degSU operon decreased competence, whereas inactivation of degS alone left competence unaffected. Moreover, a DegU mutant with an impaired phosphorylation site had no effect in competence [307]. However there are alternative explanations why the hyperactive form of DegU, or the complete absence of DegU, hinder competence, whereas the modest activity of the unphosphorylated DegU is enough to activate competence. DegU has opposing effects to different members of the DNA uptake gene-cascade. On one hand, it co-activates with ComK the comK promoter [231], but on the other hand it represses the srf operon [257] and therefore, also inhibits the expression of comS. Reduced ComS levels result into an enhanced MecA/ClpCP-mediated degradation of ComK [309]. Thus, it may well be that the final output of DegU on ComK is only positive, when the levels of DegU and/or its DNA binding affinity are relatively low (remember that the unphosphorylated form of response regulators has usually weaker binding affinity to its DNA targets; see above). On the contrary, when the cellular amounts or activity of DegU increase then the negative effect on comS expression prevails. Further evidence for such a scenario can be deduced by the genome-wide transcriptional profiling by Ogura et al (2001). In this study, the DegU regulon was identified by comparing a degS mutant strain with its isogenic strain (degS mutant too), having though DegU overexpressed from a plasmid. comK was not part of the induced genes, whereas all the phosphorylated DegU-dependent genes were. This insinuated that the phosphorylation state of DegU alone does not dictate the targets of DegU. It is probably the DegU amounts and relative activity (which can be modulated by the phosphorylation state of the protein) that do so.

In our case, DegU activates the expression of the bmy operon during stationary phase growth, and therefore it seems plausible that the phosphorylated form of DegU is more suitable for optimal promoter binding and activation. However more direct evidence would be required for verifying this suggestion (see also above).

Finally, DegU seems to have a pronounced role in the synthesis of bacillomycin D since in its absence, B. amyloliquefaciens FZB42 is defected solely in the production of this peptide antibiotic (Fig. 28). This detrimental effect on bacillomycin D production cannot be only due to the reported effect of DegU on the P_bmy promoter, since several other regulators exert effects of similar extent on the promoter activity, but do not completely inhibit the synthesis of the antibiotic (Figs. 26-28). Two scenarios can explain this situation. First, it is possible that DegU also controls the activity of a second, yet unidentified, internal promoter in the bmy operon. In this case, the mutation of degU would be deleterious for bacillomycin D biosynthesis, since more than one promoter responsible for the expression of the bmy operon would be strongly hindered. However, till now, there are no reports about internal promoters regulating the expression of gene clusters encoding nonribosomal peptide synthetases. A second more plausible scenario would be that DegU is involved in the post-transcriptional regulation of bacillomycin D. DegU would then have to control the expression of a protein involved in the synthesis of bacillomycin D, but not in its export (no bacillomycin D was detected in sonificated cell extracts of the degU ^– mutant). The possibility that this protein is Sfp should be ruled out, since DegU exerts a specific effect only on bacillomycin D. In contrast, production of surfactin, fengycin and the three polyketides was not impaired in the degU mutant stain. Moreover, DegU is not involved in the transcriptional regulation of yczE, which also controls the production of bacillomycin D in a post-transcriptional manner (see later and Fig. 36). Therefore, the putative post-transcriptional effect of DegU on the synthesis of bacillomycin D should be mediated through pathways independent of Sfp and YczE, and prior to the antibiotic’s export out of the cell.

DegQ is a small pleiotropic regulatory protein, which consists of 46 amino acids and controls the expression of degradative enzymes, intracellular proteases and several other secreted enzymes (levansucrase, β-glucanase, xylanase, subtilisin and α-amylase) [171, 224]. Lately it was also shown to stimulate the expression of several peptide antibiotics [172, 173]. DegQ shares no homology to typical transcriptional regulators, i.e. DNA-binding proteins. It may be located adjacently to the competence genes in the chromosome of different Bacilli organisms, but its function has been associated with that of DegU, with which they exhibit a significant target overlap [171]. In the absence of DegU, DegQ ceases to control the expression of sacB (encoding a levansucrase), implying that the effects of DegQ on sacB expression are indirect and mediated through DegU [171]. Our results show also that the effects of DegU are epistatic to those of DegQ on bacillomycin D production, since DegQ overexpression cannot complement for the loss of DegU in terms of bacillomycin D synthesis (see also Fig. 32). Thus, it seems plausible that DegQ regulates the transcription of its target genes only in an indirect manner, via DegU. DegQ possibly modulates the activity of DegU, via a yet unidentified mechanism. It is worth mentioning that DegQ shows homology to a region of the eukaryotic A-kinase anchor proteins (Dransfield et al., 1997), and therefore a plausible role of it would be that it anchors DegS and facilitates the transphosphorylation to DegU.

Earlier studies had shown that Bacilli harbour two different versions of the σ^A-dependent promoter that is responsible for the transcription of degQ. B. subtilis 168 (and its derivative MO1099 used here) carry the degenerated promoter version, whereas B. amyloliquefaciens FZB42 possesses the optimised promoter version with a more consensus-like -10 hexamer, designated as degQ36(Hy) (for more details see corresponding text in results). Consistently, strains that carry the degQ36(Hy) show more prominent production of the enzymes DegQ regulates [225]. I have shown here that supplying the defected on degQ expression, B. subtilis MO1099, with ectopically produced DegQ, results into a 3-fold increase in the activity of the bacillomycin D promoter (see also Fig. 23). Moreover, this increase could be observed only when both DegU recognition sites were intact in the promoter region, verifying that DegQ exerts its role on the promoter activity via the action of DegU. In addition, this has been the first time that degQ was demonstrated to have an effect on the transcriptional regulation of a nonribosomally synthesized antibiotic. However, this effect was not as pronounced as the effect of DegQ on the overall production levels of iturin A or plipastatin, where an increase of 8- to 10-fold was observed [172, 173]. This insinuates that DegQ has an additional post-transcriptional role on lipopeptide synthesis. Consistently, DegU seems to exert a post-transcriptional effect on bmy expression (see above), and therefore, the two proteins may act again as a “pair” in the post-transcriptional control of the bacillomycin D synthesis.

A further player that positively influenced the expression of bacillomycin D was the two-component system response regulator ComA (see also Figs 26 and 27). ComA is known to be involved in the regulation of several central developmental processes in the cell. Phosphorylated ComA activates the promoter of the srf operon [162], which encodes the enzyme complex that catalyzes the synthesis of the surfactin and also the competence regulation factor ComS, that lies within and out-of-frame in the srfAB gene. Consequently, ComS destabilizes the ternary ComK/MecA/ClpC complex with which ComK is degraded [309], releasing, thereby, the competence transcription factor that acts as a key regulator element in the development of competence [310]. Thus, ComA triggers the expression of surfactin and that of late competence genes. In addition, ComA controls the expression of rapA [311], a phosphatase which negatively regulates the initiation of sporulation by dephosphorylating Spo0F [279]. rapC and rapF, are also activated by ComA, creating thus a negative feedback loop, since both Rap proteins inhibit the function of ComA [167, 311]. Finally, ComA has a crucial role in the activation of degQ, along with DegU, which shows a more subtle effect on this process [171].

Here, we have shown that ComA exhibits similar effects in the activity of P_bmy as DegU (Figs. 26 and 27). The effects of ComA were mostly dependent on the presence of an upstream DNA region (-342 to -126 bp, relative to the transcriptional start), again similarly to those of DegU, raising the possibility that the two proteins mediate their effects on P_bmy through the same pathway. Since DegU is shown to directly bind to a DNA-site within this region, and ComA controls the expression of DegQ [171], which presumably serves as an auxiliary factor to DegU (see above), it would be plausible that the effects of ComA on the expression of the bmy operon are indirect and mediated through the DegQ-DegU system (Fig. 39). It is noteworthy that two ComA-boxes are located upstream of the degQ gene in B. amyloliquefaciens FZB42, similarly to the situation in B. subtilis 168 (data not shown). In addition, ComA activates both degQ promoter versions [171], and therefore it would be expected to promote the expression of DegQ in B. amyloliquefaciens FZB42, too. Unfortunately, our attempts to verify the proposed indirect role of ComA, by constructing a comA deficient strain of B. amyloliquefaciens FZB42, with degQ being expressed from an IPTG-inducible promoter, were unsuccessful (see corresponding section in the Results).

However, the recognition sequences of ComA have already been identified and consist of a palindromic segments, termed as ComA-box, i.e. TTGCGG-N₄-CCGCAA [162, 312]. The centres of dyad symmetry of the ComA-boxes are separated by about 45 bp. A screen for the above-mentioned motif did not reveal any putative ComA-binding sites at the bmyD upstream region. This supports our suggestion that ComA only indirectly controls the transcriptional regulation of the bmy operon. Nevertheless, further experimental proof has to be provided for this statement. Either band-shift assays or assaying the role of ComA (in the presence of degQ being expressed from an IPTG-inducible promoter) on bmy expression in B. subtilis MO1099 would tackle the problem.

It is worth mentioning that the effect of ComA on the final production of bacillomycin D was not as devastating as that of DegU (Fig. 28), indicating that the transcriptional control of DegU and ComA on bmy expression might be exerted through the same pathway, but this is not the case for the post-transcriptional effects on bacillomycin D production (Fig. 39).

Two sigma factors were shown to positively regulate bacillomycin D transcription: σ ^H , the sporulation sigma factor [233] and regulator of late-growth activities [162], and σ ^B , the general stress sigma factor in Bacilli [237, 238]. Both of them stimulate the activity of the σ^A-dependent P_bmy promoter (Figs. 26 and 27). Their effects on bmy expression are of the same magnitude to those of DegU and ComA (Figs. 26 and 27), and are most probably exerted in an indirect manner, since there are no sequences in the bmyD promoter region that resemble the promoter consensus sequences of σ^H (AGGANNT-15-17bp-GAAT; [234]) and σ^B (GTTT-15-17bp-GGGWAW, where W stands for A/T; [239]).

Bacillomycin D production was not silenced in the absence of σ^H and σ^B, similarly to the comA deletion. This indicates that either these sigma factors act principally on ComA or that they just moderately modulate the activity of DegU and do not completely abolish it. Based on our results and on former studies, we propose that the effects of σ^H and σ^B are mediated through various Rap proteins that control the activities of ComA and DegU.

RapC/RapF/RapK and RapG/RapH have been shown to inhibit ComA and DegU, respectively, from binding to their target sequences, in B. subtilis 168 [167, 246, 247, 248, 249]. These Rap proteins directly bind to the C-terminally located DNA-binding domain of the two response regulators and, thereby, hinder their transcriptional regulatory function [279]. The activity of the above-mentioned five Rap proteins is inhibited by specific, adjacently encoded, Phr pentapeptides (see also section 4.1.3). Interestingly, rap and phr genes are co-transcribed by a σ^A-dependent promoter [312], while the phr genes are additionally controlled by a σ^H-dependent promoter [234].

B. amyloliquefaciens FZB42 also encodes rapC, rapF, and their cognate phr genes, but lacks orthologues of rapK, rapG, rapH and/or their cognate phr genes (see also section 4.1.3). In addition, the bacterium possesses three novel Rap proteins (see also Table 14), which do not have a cognate Phr partner. Based on studies performed in B. subtilis 168 [167, 246], and our results demonstrating that ComA positively regulates expression of bacillomycin D, it is very likely that the effect of σ^H on the transcriptional regulation of the bmy operon is mediated through RapC and RapF. Deletion of σ^H decreases expression of PhrC and PhrF, and, thereby, RapC and RapF can more efficiently inhibit ComA from activating the expression of bacillomycin D (Fig. 39).

In B. subtilis 168, the σ^B–controlled RghR [252] was recently shown to specifically repress rapG and rapH by directly binding to their promoter regions [249]. RghR has no effect on other Rap proteins of B. subtilis 168 [249]. Although B. amyloliquefaciens FZB42 lacks rapG and rapH orthologues, RghR binding-sites were found upstream of one of its novel rap members, rapX (see also section 3.4.5). We have shown that a deletion of rapX results in enhancement of the P_bmy promoter activity (Fig. 33), which indicates the participation of RapX in the antibiotic’s complex regulatory circuit (Fig. 39). However, due to its low homology to any of the Rap proteins of B. subtilis 168, it remains unclear whether RapX inhibits ComA or DegU or both of them. If the target of RapX is DegU, then the presence of increased amounts of RapX (in a sigB mutant) do not completely silence the activity of DegU since the sigB deficient strain can still produce bacillomycin D (see also Fig. 28). RapX could either dephosphorylate its target response regulator(s) or bind to its DNA-binding site and inhibit its function. Even though there is no direct evidence, we postulate that ComA and DegU are inhibited by Rap proteins via the same mechanisms in B. subtilis 168 and in B. amyloliquefaciens FZB42, i.e. the Rap protein binds to the DNA-binding site of the response regulator and blocks its action.

Furthermore, a double sigB rapX mutation clearly derepressed the expression of bacillomycin D, which was defected in the sigB single mutant (Fig. 33). This indicates that the effect of σ^B is mediated through RapX. We presume that the intermediate link is RghR (Fig. 39), since rghR (and its promoter region) is highly conserved between B. subtilis 168 and B. amyloliquefaciens FZB42 and the promoter region of rapX carries optimal DNA binding-sites for RghR. Further experimental evidence will be required for our assumption to be verified. In addition, it seems plausible that RghR might repress further Rap proteins (that inhibit the function of DegU or ComA), since the derepression effect on bmy expression observed after introducing a rapX mutation on the sigB mutant strain was not complete. A good candidate would be RBAM00430, which shows 43% similarity on amino acid level to RapH of B.subtilis 168, but preliminary searches for RghR binding-sites on its promoter region revealed only relatively degenerate motifs in comparison to the published consensus sequence of the RghR DNA binding-site [249].

Sfp and YczE were both shown to post-transcriptionally regulate the expression of bacillomycin D. The essentiality of Sfp on nonribosomal synthesis is already known and thereby, the strain’s deficiency to produce lipopeptides and polyketides in a sfp ^- strain was expected. Surprisingly, the deletion of the adjacently located gene, yczE, encoding for a predicted membrane protein, specifically abolished the production of bacillomycin D (Fig. 34.C), even though the activity of the P_bmy promoter was not impaired (Fig. 35). YczE is not involved in the export of the lipopeptide into the external milieu, similarly to DegU, and it exerts its effects through a separate pathway than that of DegU (Fig. 36). Both DegU (see also above) and YczE exert distinct control over the expression and the synthesis of bacillomycin D than Sfp, and therefore their mechanism of action remains an issue for further research. Figure 39: A complex regulatory network governs bacillomycin D production in Bacillus amyloliquefaciens strain FZB42

Boxes and cycles indicate ORFs and proteins respectively. Arrows and T-bars indicate activation and repression respectively. Interactions that have not been proven are represented by the dotted lines. σ^A and σ^H represent the promoters of the corresponding genes. Sites I and II are binding sites of DegU at the upstream region of bmyD.

Finally, the genes governing the export of bacillomycin D or securing the organism’s immunity against the lipopeptide have not been identified yet. A novel TCS located upstream of the bmy operon (RBAM01839/RBAM01840) was investigated for its role on export of bacillomycin D (or other peptide antibiotics) and/or on self-resistance to B. amyloliquefaciens FZB42 against the antibacterial compounds produced by the strain. Deletion of this TCS did not impair the export of lipopeptides/polyketides, nor did the mutant strain show growth disadvantages when mixed with equal amounts of wild-type cells and let grow for several generations (data not shown). This indicates that the TCS RBAM01839/RBAM01840 is not involved in the release of lipopeptides/polyketides to the external milieu or in the self-resistance mechanisms.