Báo cáo khoa học: Two overlapping antiparallel genes encoding the iron regulator DmdR1 and the Adm proteins control sidephore and antibiotic biosynthesis in Streptomyces coelicolor A3(2) pdf

Two mutants with stop codons resulting in arrested translation of either DmdR1 or Adm were obtained by gene replacement and compared with a deletion mutant DdmdR1⁄ adm that was defective

regulator DmdR1 and the Adm proteins control sidephore and antibiotic biosynthesis in Streptomyces coelicolor

A3(2)

Sedef Tunca1,*, Carlos Barreiro1, Juan-Jose´ R Coque1,2and Juan F Martı´n1,2

1 Institute of Biotechnology of Leo´n, INBIOTEC, Parque Cientı´fico de Leo´n, Avenida Real no 1, Spain

2 Area of Microbiology, Faculty of Environmental and Biological Sciences, University of Leo´n, Spain

Introduction

Iron is an essential nutrient that is required for many

key metabolic processes in microorganisms, such as

electron transport, energy metabolism and DNA

synthesis [1] Recent studies have indicated that iron

metabolism in bacteria is directly connected to

oxidative stress [2,3] Iron binding and transport by

siderophores play important roles in iron availability

in the cell [4–6] However, the concentration of free iron in the cells must be strictly regulated, as high levels of free intracellular ferric iron are toxic to the cells In Gram-negative bacteria, iron metabolism is controlled by the global regulatory protein Fur [7,8]

Keywords

antibiotics; desferrioxamines; iron regulation;

siderophores; Streptomyces

Correspondence

J F Martı´n, Instituto de Biotecnologı´a,

INBIOTEC, Parque Cientı´fico de Leo´n,

Avenida Real no 1, 24006 Leo´n, Spain

Fax: +34 987 210 388

Tel: 34 987 210 308

E-mail: jf.martin@unileon.es

*Present address

Biology Department, Faculty of Science,

Gebze Institute of Technology, Kocaeli,

Turkey

(Received 13 March 2009, revised 26 May

2009, accepted 29 June 2009)

doi:10.1111/j.1742-4658.2009.07182.x

The dmdR1 gene of Streptomyces coelicolor encodes an important regulator

of iron metabolism An antiparallel gene (adm) homologous to a develop-ment-regulated gene of Streptomyces aureofaciens has been found to over-lap with dmdR1 Both proteins DmdR1 and Adm are formed in solid and liquid cultures of S coelicolor A3(2) The purpose of this study was to assess possible interaction between the products of these two antiparallel genes Two mutants with stop codons resulting in arrested translation of either DmdR1 or Adm were obtained by gene replacement and compared with a deletion mutant (DdmdR1⁄ adm) that was defective in both genes The deletion mutant was unable to form either protein, did not sporulate and lacked desferrioxamine, actinorhodin and undecylprodigiosin biosyn-thesis; biosynthesis of these compounds was recovered by complementation with dmdR1⁄ adm genes The mutant in which formation of Adm protein was arrested showed normal levels of DmdR1, lacked Adm and over-produced the antibiotics undecylprodigiosin and actinorhodin (in MS medium), suggesting that Adm plays an important role in secondary metabolism The mutant in which DmdR1 formation was arrested synthe-sized desferrioxamines in a constitutive (deregulated) manner, and pro-duced relatively normal levels of antibiotics In conclusion, our results suggest that there is a fine interplay of expression of these antiparallel genes, as observed for other genes that encode lethal proteins such as the toxin⁄ antitoxin systems The Adm protein seems to have a major effect on the control of secondary metabolism, and its formation is probably tightly controlled, as expected for a key regulator

Abbreviations

Adm, antiparallel to DmdR; DmdR, divalent metal-dependent gene ⁄ protein; R/T, room temperature.

and in Gram-positive bacteria is regulated mainly by

the DmdR (divalent metal-dependent) family of

regula-tory proteins, which includes DtxR of

Corynebacte-rium diphtheriae [9], DmdR of Corynebacterium

(Brevibacterium) lactofermentum [10], IdeR of

myco-bacteria [11] and the DmdR1 and DmdR2 proteins of

Streptomyces coelicolor [12] After interaction with

Fe2+ or other divalent metals, these proteins act as

transcriptional regulators, usually as repressors of

genes involved in iron metabolism In some cases,

these iron regulators can also function as

transcrip-tional activators [7,13,14]

Streptomyces coelicolor A3(2), a well-known model

actinomycete that is able to synthesize several types of

secondary metabolites, contains two copies of the iron

regulator gene, dmdR1 and dmdR2 [12,15] We

reported previously that disruption of dmdR1 resulted

in significant changes in the S coelicolor A3(2)

prote-ome, particularly in enzymes related to iron

metabo-lism [12], although changes in some key glycolysis

proteins were also observed; these changes were

diffi-cult to explain at that time, unless dmdR1 plays a

fur-ther role in S coelicolor A3(2) in addition to

controlling iron metabolism During our studies on the

deletion mutant S coelicolor DdmdR1, we observed

that it was defective with respect to sporulation

Strik-ingly, we found that the antisense strand of dmdR1

gene contains an ORF that almost fully overlaps

dmdR1 and shows 87% identity with a

development-regulated gene from Streptomyces aureofaciens [16]

(Fig 1) This S aureofaciens gene was found in a

search for promoters that are recognized by the RpoZ

sigma factor, a member of the WhiG family [17] of

sigma factors [18,19] Nova´kova´ et al [16] provided

evidence showing that this (unnamed) gene is tran-scribed in S aureofaciens The role of this putative

‘development’ gene is unknown, and its possible involvement as a putative antisense regulator of dmdR1 is intriguing, as it may modulate iron meta-bolism and therefore expression of many other genes

in the cell

Overlapping antiparallel genes are rare [20], and have not been described in actinomycetes, but recent evidence suggests that, in Escherichia coli, RNA formed from intragenic reverse promoters may explain the tight balance required to prevent a toxic protein from attacking the producer cell (toxin⁄ antitoxin sys-tem) [21,22] It was therefore of great interest to deter-mine whether proteins encoded by the antiparallel genes are translated, and to study the involvement of dmdR1 and the antisense gene (adm) in the regulation

of S coelicolor A(3)2 sporulation and antibiotic and siderophore biosynthesis

Results

The Adm protein is present in S coelicolor A3(2) cells grown in solid medium and in liquid cultures

Analysis of the S coelicolor dmdR1 gene revealed the presence in the opposite orientation (antisense strand)

of a putative gene overlapping dmdR1 (Fig S1) The protein predicted to be encoded by this gene showed 87% identity at amino acid level, to that predicted to

be encoded by a gene reported in S aureofaciens [16] that has been named adm (antiparallel to dmdR1) Streptomyces genes have a biased codon usage, with a

A

B

Fig 1 Nonsense mutations introduced into the coding sequence of the dmdR1 and adm genes (A) Organization of the ORFs in the

S coelicolor A3(2) DNA region corresponding to the dmdR1 gene Note the antiparallel organization of dmdR1 and adm (B) The introduced nonsense mutations are in bold and underlined, and indicated as STOP The promoter regions are indicated by solid arrows Only relevant parts of the sequence are shown The numbers in the sequence indicate the nucleotide positions numbered from the translation start codon

of the dmdR1 or adm gene Note that the stop codons do not affect the amino acids encoded by the complementary strands.

high GC content in the third position of the codons.

The GC content in this position was 96% for the

dmdR1 gene and 94% for the adm gene, supporting

the conclusion that both ORFs encode translated

Streptomycesgenes

Using rabbit antibodies raised against the C-terminal

region of the Adm protein, we observed significant

lev-els of Adm protein in liquid S coelicolor A3(2)

cul-tures between 24 and 96 h in TSB medium (Fig 2)

Adm was detected at a lower concentration in cells

grown in MS solid medium (days 1–7) using the same

antibodies at the same dilution (Fig 2) These results

indicate that the Adm protein is more abundant in

liquid cultures, while its lower level in solid medium

may favour sporulation (see below) The formation of

Adm protein and its expression pattern support the

initial data obtained by Nova´kova´ et al [16] on

tran-scription from the adm promoter (designated PREN 40

by those authors)

The molecular mass of Adm deduced from the

SDS–PAGE gels was approximately 28 kDa, in

agree-ment with the expected size according to the ORF that

starts with the TTG proposed by Nova´kova´ et al [16]

Evidence for expression of the dmdR1 gene, resulting

in formation of the DmdR1 protein, was also obtained

(see below), confirming previous results [12]

Inactivation of the dmdR1 and/or adm genes: phenotypic analyses

A S coelicolor A3(2) mutant with a deletion of the dmdR1 gene showed significant changes in the prote-ome [12] Deletion of an internal fragment of the dmdR1 gene inactivated both the dmdR1 and adm open reading frames (hereafter named DdmdR1⁄ adm), and caused a deficiency in sporulation in the disrupted mutant

To assess the in vivo role of the dmdR1- and adm-encoded proteins in S coelicolor A3(2), the chromo-somal dmdR1 and adm genes were replaced separately with mutated dmdR1 and adm genes that cannot be translated, as described in Experimental procedures These nonsense codon mutations (two stop codons in one or the other ORF) were introduced in such a way that the mutations preventing translation of the dmdR1 mRNA do not affect the amino acid sequence of the protein encoded by the adm gene, and vice versa (Fig 1)

Among 652 recombinants tested, eight putative adm mutants containing the two stop codons in the adm ORF were found Similarly, two putative dmdR1 mutants were obtained among 678 recombinants carry-ing plasmid pSKMdmdR1 with the mutant allele with the two stop codons in the dmdR1 ORF The gene replacement in each of these mutants was confirmed

by DNA sequencing As clones with the same site-directed mutations showed identical morphology and growth behaviour, one of the clones for each of these two mutations was chosen at random for further experiments and named TAdmdR1 (strain in which translation was arrested in the sense strand of dmdR1) and TAadm (strain in which translation of the adm mRNA was arrested)

To compare their morphology, sporulation and physiological properties, strains DdmdR1⁄ adm, TAdmdR1, TAadm and the parental strain S coelicolor A3(2) were grown on several types of solid medium (Fig 3) The parental strain S coelicolor A3(2) and the TAadm mutant sporulate normally on TBO and

MS sporulation media However, TAdmdR1 does not sporulate well on the same sporulation media (Fig 3) The deletion mutant DdmdR1⁄ adm showed no sporula-tion at all on MS medium, and did not produce pigments in any of the media tested (Fig 3)

The TAadm strain sporulated efficiently on all spor-ulation media tested, and produced far more pig-mented antibiotics than the other strains on several media (Fig 3) These results suggest that removal of the Adm protein stimulates sporulation and pigmented antibiotic formation, whereas elimination of DmdR1

Fig 2 Western blot analysis of the Adm protein using antibodies

against Adm Samples were taken from cultures in TSB liquid

med-ium (upper panel) or MS solid medmed-ium (lower panel) at various

times Adm*, ovalbumin-conjugated Adm peptide; M, molecular

mass markers (sizes in kDa are shown on the left) Six micrograms

of total protein were loaded in each lane except for the control

Adm* (5 lg) Note that the ovalbumin-conjugated Adm peptide

(Adm*) is larger than the Adm protein of S coelicolor A3(2)

(28 kDa).

decreases sporulation and pigmentation in solid media.

The phenotype of strains TAadm and TAdmdR1 was

partially medium-dependent, as reported for antibiotic

biosynthesis in many Streptomyces strains [26,28]

Pig-mentation of the TAdmdR1 in R5 or TBO solid media

was still intense, in contrast to the results for MS

med-ium, in which the pigmentation was much lower

Actinorhodin and undecylprodigiosin production in

these strains was quantified in liquid MS and R5

cultures, which supported high levels of actinorhodin and undecylprodigiosin (Fig 4) Growth in these media

of the three mutant strains was similar or even higher than that of the parental strain S coelicolor A3(2) The deletion mutant DdmdR1⁄ adm does not produce actinorhodin or undecylprodigiosin in liquid MS med-ium (Fig 4A–C) The same result was observed in liquid R5 medium (Fig 4D–F), confirming the impor-tant role of these genes in biosynthesis of secondary

Fig 3 Growth, formation of aerial mycelia,

sporulation and pigment formation of

various strains Mutants TAdmdR1,

DdmdR1 ⁄ adm, TAadm and the parental

strain S coelicolor A3(2) were grown in

TBO, MS and R5 media Photographs were

taken after 10 days of culture from the top

(left column) or bottom of the plates (right

column) Note the intense pigmentation in

most tested media by the TAadm strain,

and the lack of pigment formation by the

DdmdR1 ⁄ adm deletion mutant This latter

mutant forms aerial mycelia, but does not

sporulate.

Fig 4 Growth (mg dry weight per mL) and specific production (nmolÆmg)1dry weight) of actinorhodin and undecylprodigiosin Liquid cul-tures of the various strains were produced in MS (A–C) or R5 media (D–F) Vertical bars indicate standard deviation from the mean value Note the high levels of actinorhodin and undecylprodigiosin produced by the TAadm mutant, and the lack of production by the DdmdR1 ⁄ adm mutant in MS medium Undecylprodigiosin and actinorhodin were quantified spectrophotometrically using standard procedures [38].

metabolites The TAadm strains produced more

unde-cylprodigiosin (about sevenfold) and actinorhodin

(about twofold) compared with the other strains in

MS medium, and these results were consistent in

repeated experiments This stimulation was

medium-dependent In R5 medium, the TAadm strain also

pro-duced very high levels of undecylprodigiosin, but no

significant increase of actinorhodin compared to the

parental strain was found

On the other hand, the strain TAdmdR1, defective

in DmdR1 translation, produced consistently less

anti-biotic than the wild-type in liquid MS medium These

results confirmed the observations in solid medium,

supporting the conclusion that arrest of Adm

forma-tion at the translaforma-tional level has a stimulatory effect

on undecylprodigiosin production, i.e the Adm

protein appears to act as a negative regulator of the

biosynthesis of undecylprodigiosin The opposite

behaviour of TAadm and TAdmdR1 strains regarding

production of antibiotic suggests interaction between

the actions of the Adm and DmdR1 proteins (see

below)

Lack of both DmdR1 and Adm proteins prevents desferrioxamine production

Production of the siderophore desferrioxamine in the mutants and wild-type strains was quantified by HPLC analysis in the supernatants of cultures grown in mini-mal medium alone or minimini-mal medium supplemented with 35 lm iron The results of HPLC analyses showed that strain TAadm produces desferrioxamines only in the absence of iron, like the wild-type strain (Fig 5A)

As expected, addition of 35 lm iron to the culture medium completely suppressed desferrioxamine pro-duction in the wild-type and TAadm strains These results prove that a functional DmdR1 represses the formation of desferrioxamines in the presence of iron, whereas biosynthesis of this siderophore occurs in iron-starved cultures

It is noteworthy that production of desferrioxamines

in the TAdmdR1 mutant was similar in the presence or absence of iron (compare Fig 5A,B) DmdR1 is known to be the major iron regulator in S coelicolor A3(2), acting as a repressor in presence of iron

A

B

Fig 5 HPLC analyses of the formation of

desferrioxamines B and E (labelled as DesB

and DesE) and complementation of

desferri-oxamine production (A) Cultures of S

coeli-color A3(2) and the TAdmdR1, TAadm and

DdmdR1 ⁄ adm mutants were produced in

minimal medium without iron

supplementa-tion (left) or minimal medium supplemented

with 35 l M iron (right) Formation of

desfer-rioxamines is absent in the DdmdR1 ⁄ adm

mutant and is inhibited by iron addition in all

strains except in TAdmdR1, which lacks the

DmdR1 iron regulator The chemical

struc-tures of desferrioxamines E and B are

shown in Challis [6] and Tunca et al [23].

(B) Complementation of the DdmdR1 ⁄ adm

mutant in minimal medium without iron was

achieved by transformation with plasmids

pHZBH9 (multiple-copy) and pRAdmdR1

(single-copy) containing the dmdR1 ⁄ adm

genes The numbers adjacent to the DesB

and DesE peaks indicate their retention

time.

[6,12,15]; lack of the iron regulatory protein in the

TAdmdR1 mutant prevents iron repression, thus

con-verting this strain to a deregulated desferrioxamine

producer

The deletion mutant DdmdR1⁄ adm was defective in

desferrioxamine production in the presence or absence

of iron No desferrioxamines could be detected in

cul-tures of the deleted strain under the same conditions

used for the other strains (Fig 5A) This result is

inter-esting and suggests an important role for Adm in iron

metabolism, in addition to that for DmdR1 As

indi-cated above, the simple absence of DmdR1 does not

prevent desferrioxamine biosynthesis, but simultaneous

lack of Adm prevents deferrioxamine formation;

there-fore, the mutant DdmdR1⁄ adm is defective in iron

scavenging and transport, which may explain its very

poor or null growth in some media The Adm protein

may serve as an activator of desferrioxamine synthesis,

and the DmdR1⁄ Adm ‘tandem’ proteins may act as a

fine modulator system of iron regulation As indicated

above, the DadmR1⁄ adm mutant does not produce

ac-tinorhodin and undecylprodigiosin, suggesting that

interaction of these two proteins plays an important

role in the control of secondary metabolism (including

production of desferrioxamines and antibiotics)

Complementation of the DdmdR1/adm mutant

restored desferrioxamine production

Complementation of the DdmdR1⁄ adm strain was

performed using a 9233 bp BamHI⁄ HindIII fragment

containing dmdR1 and adjacent regions in a multicopy

pHZ1351-derived vector named pHZBH9, or using a

smaller plasmid pRAdmdR1 (constructed by cloning a

2.2 kb BclI fragment of pSKdmdR1 into a BamHI site

of the single-copy plasmid pRAKn)

Kanamycin-(pRAdmdR1) or thiostrepton- (pHZBH9) resistant

transformants carrying these plasmids were assayed for

their siderophore and antibiotic production

Comple-mentation of the deletion in the DdmdRI-adm strain by

the wild-type gene (either one copy or multiple copies)

restored desferrioxamine production (Fig 5B)

Func-tional complementation of the deleted mutant was the

result of either one single integrated copy of the gene

(in pRAdmdR1) or multiple copies of the original gene

(in pHZBH9) As shown in Fig 5B, as expected,

com-plementation with the single-copy pRAdmdR1 resulted

in lower levels of desferrioxamines B and E than

complementation with the multiple-copy plasmid,

which restored production of desferrioxamines B and

E to the levels in the wild-type strain

Desferrioxam-ines B and E are two products of the desferrioxamine

biosynthetic pathway that differ with respect to the

presence of an acetyl or succinyl group modifying the N-hydroxycadaverine residue [23]

Complementation of the DdmdR1/adm deletion mutant with multiple copies of dmdR1/adm restored actinorhodin production

Complementation of the dmdR1⁄ adm deletion with the multiple-copy plasmid pHZBH9 restored actinorhodin production in TSB liquid medium and solid TSA medium (Fig S2); on the other hand, the single-copy plasmid pRAdmdR1 complemented the antibiotic pro-duction deficiency in TSB and R5 liquid media but not

in TSA solid medium This latter plasmid contains a smaller insert (2.2 kb including the complete dmdR1⁄ admgenes) In order to clarify whether the lack of com-plementation of antibiotic production in TSA solid medium by pRAdmdR1 was due to the small insert size

or the single-copy number of the dmdR1⁄ adm genes, we cloned the 2.2 kb BclI fragment (the same as in pRAdmdR1) into the BamHI site of the multiple-copy vector pHZ1351, to give pHZdmdR1 Multiple copies of the 2.2 kb fragment restored actinorhodin production

in TSA solid medium, as occurred with pHZBH9 which carries a 9.2 kb insert (Fig S2B) Those results indicate that the single copy of the dmdR1⁄ adm genes in pRAdmdR1 was insufficient for full restoration of anti-biotic biosynthesis in some media Similar observations have been made using pRA-derived plasmids in Strepto-myces natalensis with respect to pimaricin production (J.F Aparicio, Institute of Biotechnology, Leon, Spain, personal communication)

Expression of the overlapping genes: Western blot analysis using antibodies to DmdR1 and Adm

To determine whether the lack of translation of one gene affected expression of the antiparallel one, western blot analyses of formation of the DmdR1 protein were performed for the various mutants with either deletion

or arrested translation of dmdR1 or adm The results showed unequivocally that mutants TAdmdR1 and DdmdR1⁄ adm completely lack the DmdR1 protein, which was partially replaced by increased formation of the homologous protein DmdR2 (Fig 6A,B) The lack

of DmdR1 protein formation in the TAdmdR1 mutants correlated well with the lack of iron regulation of desferrioxamine biosynthesis (Fig 5A) Iron regulation

of desferrioxamine biosynthesis requires the DmdR1 protein and high levels of iron

Experiments to assess the level of Adm protein in the various strains confirmed that the Adm protein is

indeed abundant in the wild-type strain and that it is

absent in the TAadm and DdmdR1⁄ adm strains, as

expected (Fig 6C) It is noteworthy that the Adm

pro-tein levels were low in the TAdmdR1 mutant,

suggest-ing that lack of DmdR1 in this mutant reduces

synthesis of the Adm protein A putative

DmdR1-binding sequence (iron box) has been found in the

upstream region of the adm gene, which may explain

this effect (see Discussion)

Discussion

Streptomyces species produce a plethora of secondary

metabolites, including antibiotics [24] and siderophores

[25], and undergo a complex developmental cycle

Many factors appear to influence the onset of

antibi-otic production in actinomycetes [26] Despite the

iden-tification and characterization of several genes that

affect antibiotic production, there is still no overall

understanding of the network that integrates the

vari-ous environmental and nutritional signals that bring

about changes in the expression of biosynthetic genes

[3,27] Imbalances in metabolism lead to physiological

stress and influence the onset of antibiotic production

(reviewed by Bibb [26] and Martı´n [28])

The intracellular level of iron in the soil-dewelling

Streptomyces species must be strictly regulated to

adjust their metabolism to the iron concentration in

diverse habitats Numerous studies have shown that the DtxR (DmdR) and Fur proteins are pleiotropic regulators that control iron uptake and also other pro-cesses related to iron metabolism, including oxidative stress [1,2] Camacho et al [29] reported that, in addi-tion to genes involved in siderophore producaddi-tion and iron storage, the Mycobacterium tuberculosis DmdR homologue, named IdeR, controls genes that encode putative transporters, transcriptional regulators, pro-teins involved in general metabolism, members of the

PE⁄ PPE family of conserved mycobacterial proteins, and the virulence determinant MmpL4 Our results suggest that DmdR1, like IdeR and DtxR, may also

be a wide-domain regulator controlling not only iron metabolism [12] but also other processes such as spor-ulation, although some of these other functions may

be due to the antiparallel adm gene (see below) The iron regulatory proteins may negatively or posi-tively regulate transcription of various genes [7,13,14] The results presented here indicate that DmdR1 acts

as a negative regulator of desferrioxamine biosynthesis

in the presence of iron Lack of DmdR1 or deprivation

of iron leads to deregulated (constitutive) formation of desferrioxamines However, expression of adm is required for overall function of the iron regulatory control, as the DdmdR1⁄ adm mutant cannot produce detectable levels of desferrioxamines The Adm protein may act as a positive regulator of desferrioxamine

A

C B

Fig 6 Immunodetection of DmdR1 and Adm proteins (A,B) Western blot reactions of wild-type S coelicolor A3(2) (lane 2) and the TAadm (lane 3), TAdmdR1 (lane 4) and DdmdR1 ⁄ adm (lane 5) mutants Lane 1, pure DmdR1 protein; lane 6, pure DmdR2 protein (A) Reaction with antibodies against DmdR1 (B) Reaction with antibodies against DmdR2 Antibodies against DmdR1 react with both DmdR1 and DmdR2, but antibodies against DmdR2 are largely specific for DmdR2 [15] Note the lack of DmdR1 protein in TAdmdR1 and DdmdR1 ⁄ adm strains (C) The central panel shows extracts of wild-type S coelicolor A3(2) (WT) and the TAadm, TAdmdR1 and DdmdR1 ⁄ adm mutants The left and right panels show control reactions of Adm* (a 15 amino acid peptide coupled to ovalbumin) and pure DmdR1 revealed using antibodies against Adm (left panel) and antibodies against DmdR1 (right panel) Note the absence of Adm protein in both TAadm and DdmdR1 ⁄ adm, and the lower reaction intensity in TAdmdR1 strains Six micrograms of total protein were loaded in each lane.

biosynthesis, thus compensating for the negative effect

of DmdR1

Unlike dmdR1 of S coelicolor A3(2), which is not

essential (results presented here) [12], ideR of

M tuberculosis is an essential gene; an ideR null

mutant cannot be generated without incorporation of

a second copy of the gene [11] A reason for this

dif-ference is that in Streptomyces species there are two

copies (dmdR1 and dmdR2) of the dmdR gene [15],

as also shown here (Fig 6A,B) We have confirmed

by computer searches that the genomes of

Streptomy-ces avermitilis, StreptomyStreptomy-ces griseus, StreptomyStreptomy-ces

livi-dans and Saccharopolyspora erythreae also contain

two copies of the iron regulator The antiparallel

gene adm is present and overlaps with dmdR1, but

there is no similar ORF in the antisense strand of

dmdR2

DmdR1 is known to bind to an iron box located in

the upstream region of the desABCD cluster that

encodes enzymes for desferrioxamine biosynthesis [6]

At least nine other iron boxes were identified by

bioin-formatic analyses upstream of other genes in S

coeli-colorA3(2) [12,15]

It is noteworthy that there is a significant increase in

undecylprodigiosin production in the TAadm mutant,

which lacks the Adm protein and has a normal

DmdR1 content The Adm protein appears to act as a

negative regulator of the biosynthesis of actinorhodin

and undecylprodigiosin in MS medium cultures of

S coelicolor A3(2) The response of the TAadm null

mutant is highly dependent on the culture medium, as

expected because of their differences in iron and

phos-phate content This hypothesis of a strong regulatory

effect of Adm on various antibiotic pathways is

consis-tent with the lack of antibiotic and desferrioxamine

formation and the absence of sporulation in the

dele-tion DdmdR1⁄ adm strain Indeed, our previous results

on the proteomics of the DdmdR1⁄ adm mutant showed

that several enzymes of the primary metabolism (e.g

fructose-1,6-bisphosphate aldolase, a key glycolysis

enzyme) and iron-related pathways are over- or

under-expressed in this mutant compared to the parental

strain It may be concluded that some of these effects

are probably due to the simultaneous disruption of

admin the deletion mutant

The finding that the dmdR1 gene overlaps with an

antiparallel gene (adm) that shows 87% identity with a

development-regulated gene of S aureofaciens [16] is

really intriguing The S aureofaciens adm gene was

found in a search for promoters that are recognized

by the RpoZ sigma factor, a member of the WhiG

family [17,18] Nova´kova´ et al [16] reported that the

promoter of the adm gene is induced at the time of

aerial mycelium formation and switched off in a rpoZ-defective strain Using high-resolution S1 mapping, Nova´kova´ et al [16] showed that this gene is expressed in S aureofaciens, and identified its pro-moter region

An important question is whether both proteins DmdR1 and Adm can be translated from the antipar-allel genes Our results show that the Adm protein is clearly seen in liquid TSB S coelicolor A3(2) cultures for up to 96 h, at which time the secondary metabo-lites have already formed It was also detected in cells collected from solid medium until day 7 There is very little information on expression of antiparallel genes in bacteria [20] Recently, negative regulation of the EcoRI-encoding gene by two intragenic reverse pro-moters has been described in E coli [22] These intra-genic reverse promoters are functional, and mutations

of their conserved sequences led to decreased expres-sion of the region of the gene lying downstream of the reverse promoter This has been interpreted as a mech-anism for tight control of expression of genes encoding proteins that may be lethal, e.g the toxin⁄ antitoxin proteins [21] and other so-called ‘genetic addiction’ systems [30,31]

In this work, the results of immunodetection studies showed that both proteins DmdR1 and Adm are formed in the wild-type strain, although their protein levels are probably cross-influenced, as shown in the western analysis of extracts of mutants defective in the translation of DmdR1.These analyses confirmed that the TAdmdR1 mutant lacks DmdR1 protein and instead show increased levels of the ‘substitute’ ana-logue DmdR2, as observed previously [15] The cells therefore tend to compensate for lack of the important DmdR1 regulator by switching on DmdR2 as a ‘sub-stitute’ iron regulator

When the Adm protein levels were compared in the various strains, it was found to be abundant in the wild-type and less so in the TAdmdR1 The molecular mechanism by which formation of Adm responds to the level of DmdR1 protein is still unknown; expres-sion of the adm gene is probably under the control of the iron regulation mechanism, as a putative DmdR1-binding sequence (iron box) has been found in the upstream region of the adm gene Alternatively, the adm transcript may act (if not properly loaded with ribosomes) as an antisense RNA, preventing dmdR1 mRNA translation, and vice versa

An important question is why the TAadm strain overproduces antibiotics As the DmdR1 protein in this transformant is not significantly altered with respect to the parental wild-type (Fig 6A,B), the dras-tic effect on pigmentation must be due to the absence

of Adm protein in this strain, i.e Adm may act as a

negative regulator of antibiotic biosynthesis

Expres-sion of the adm gene is known to be under the control

of the RpoZ sigma factor [16], suggesting involvement

of a sigma-factor mediated cascade in the expression

of secondary metabolites and differentiation genes

The antiparallel adm ORF occurs in all Streptomyces

genomes known so far [32–34], but not in those of

Mycobacteriumor Corynebacterium species or

Sacchar-ophlyspora erythrea [35] (Fig S3), in good correlation

with the ability of the Streptomyces species to produce

secondary metabolites and to sporulate

Experimental procedures

Microorganisms, plasmids and growth conditions

The bacterial strains and plasmids used in this study are

listed in Table 1 Streptomyces coelicolor A3(2) cultures

were grown in YEME medium (yeast extract 10 gÆL)1, malt

extract 10 gÆL)1) to achieve disperse growth, iron-limited

minimal medium (ILMM) [36] for siderophore production experiments, TBO medium [37] for spore preparations, MS solid medium [28] and TSB liquid medium or TSA solid medium [39] for protein isolation or antibiotic production, and R5 medium [38] for phenotypic analysis

Streptomyces lividans 1326 was used as the host for Streptomyces plasmid contructions E coli cultures were grown in Luria–Bertani broth alone or Luria–Bertani broth supplemented with glucose (20 mm) E coli DH5a (Strata-gene) was used as the host for routine plasmid construc-tions Ampicillin (100 lgÆmL)1), apramycin (50 lgÆmL)1), chloramphenicol (25 lgÆmL)1), kanamycin (50 lgÆmL)1) or thiostrepton (50 lgÆmL)1 in solid medium; 25 lgÆmL)1 in liquid medium) were added to growth media as necessary

DNA manipulations Isolation of plasmid and bacterial chromosomal DNA, restriction enzyme digestions, agarose gel electrophoresis and Southern analysis were performed according to stan-dard molecular biology techniques [39] Plasmids were

Table 1 Strains and plasmids used in this study.

Cosmids ⁄ plasmids

(1996) [42]

gene replacement in Streptomyces

Kieser et al (2000) [38]

derived from pSET152

R Pe´rez-Redondo (this laboratory)

resistance gene (neo)

This study

E coli strains

k+) sup44relA1k) (u80dlacZDM15)

D(lacZYA-argF) U169

Stratagene

Streptomyces strains

Norwich, UK

S coelicolor TAdmdR1 dmdR1 with two translational stop codons

in the sense strand

This study

in the antisense strand

This study