Mapping the DNA-binding domain and target sequences of the Streptomyces peucetius daunorubicin biosynthesis regulatory protein, DnrI


  • Kosan Biosciences, Inc., 3832 Bay Center Place, Hayward, CA 94545, USA.

  • *For correspondence. E-mail; Tel. (+1) 510 732 8400, ext. 219; Fax (+1) 510 732 8401. Present addresses:


Streptomyces antibiotic regulatory proteins (SARPs) constitute a novel family of transcriptional activators that control the expression of several diverse anti-biotic biosynthetic gene clusters. The Streptomyces peucetius DnrI protein, one of only a handful of these proteins yet discovered, controls the biosynthesis of the polyketide antitumour antibiotics daunorubicin and doxorubicin. Recently, comparative analyses have revealed significant similarities among the predicted DNA-binding domains of the SARPs and the C-terminal DNA-binding domain of the OmpR family of regulatory proteins. Using the crystal structure of the OmpR-binding domain as a template, DnrI was mapped by truncation and site-directed mutagenesis. Several highly conserved residues within the N-terminus are crucial for DNA binding and protein function. Tandemly arranged heptameric imperfect repeat sequences are found within the −35 promoter regions of target genes. Substitutions for each nucleotide within the repeats of the dnrG–dpsABCD promoter were performed by site-directed mutage-nesis. The mutant promoter fragments were found to have modified binding characteristics in gel mobility shift assays. The spacing between the repeat target sequences is also critical for successful occupation by DnrI and, therefore, competent transcriptional activation of the dnrG–dpsABCD operon.


Inducibility of gene expression is predicted by the ‘demand’ theory of gene regulation, which relates the mode of regulation of a pathway to the level of demand for expression of that pathway, as dictated by the natural environment of the cell. A prime example of an organism that rigorously adheres to this paradigm is the free-living filamentous soil bacterium of the genus Streptomyces. Bioactive secondary metabolite production within these microorganisms is under the precise control of an intricate cascade of regulatory events. A large number of these bioactive chemical agents have been developed into pharmaceutically useful drugs. Thus, there is considerable interest in understanding the molecular mechanisms involved in the control of the biosynthetic genetic pathways. In many species of streptomycetes that produce diverse chemical compounds, the final checkpoint in transcriptional regulation is controlled by a family of proteins recently coined Streptomyces antibiotic regulatory proteins (SARPs) (Wietzorrek and Bibb, 1997). This family of proteins has been characterized as transcriptional activators, and their overproduction has often been associated with a concomitant increase in titres of the correspond-ing antibiotic. Until very recently, little was known about the mechanism by which these proteins exerted their effect on gene expression, except that they bound DNA sequences within the promoters of antibiotic biosynthetic pathway genes and operons. Bioinformatical analyses using the primary amino acid sequences of the SARPs failed to reveal any significant similarities to the canonical helix–turn–helix (HTH) motifs commonly found in microbial DNA-binding regulatory proteins. Recent database queries have now revealed that the SARPs share sequence similarity with several members of the OmpR family of DNA-binding domains (Wietzorrek and Bibb, 1997). OmpR is a protein that regulates the expression of outer membrane porin proteins in enteric bacteria and belongs to a large family of transcriptional factors that bind DNA and interact productively with RNA polymerase (RNAP) (Makino et al., 1996; Kondo et al., 1997). Interestingly, the OmpR family of DNA-binding domains also fails to contain the canonical HTH motif (Mizuno and Tanaka, 1997). Regarding the DNA recognition domain(s) of the OmpR family, the recently determined X-ray structure reveals two helices creating a structure that superimposes well on those of other canonical HTH motifs (Martinez-Hackert and Stock, 1997a). However, this presumed structure is unique, in that the intervening loop sequence of the two helices is inordinately long and may be the cause of previous searches failing to detect a putative HTH motif in the OmpR family and the novel SARP family of DNA-binding regulatory proteins.

Here, we concentrate on the daunorubicin/doxorubicin (DNR/DXR) pathway-specific SARP, DnrI. The DnrI protein is the final, and most significant, connection of a co-ordinate cascade composed of three pathway-specific regulatory proteins that together maintain precise control of DNR/DXR biosynthetic gene cluster expression. The first player within the regulatory network is the DnrO repressor/activator. DnrO binds to the divergent promoter separating the dnrO and dnrN genes and maintains repression of dnrO while activating dnrN expression (Otten et al., 2000; H. Jiang, personal communication). In turn, DnrN activates the expression of dnrI in a process thought to be comparable with that of redZ–redD (Furuya and Hutchinson, 1996), the pathway-specific regulatory proteins of undecylprodigiosin production in Streptomyces coelicolor (Narva and Feitelson, 1990; Guthrie et al., 1998). We show that the DNA-binding domain of DnrI resides within the N-terminus of the protein. More precisely, a region of residues has been mapped by site-directed mutagenesis (SDM), which appears to be critical for recognition and/or interaction with the target DNA. In addition, the stoichiometry and kinetics of DnrI–DNA interaction are characterized. Together, these findings suggest a functional three-dimensional structural make-up that is comparable with the OmpR family.

It has been recognized that the DNA-binding domains of OmpR family members bind to direct repeat sequences within the regulatory regions of target genes (Aoyama and Oka, 1990; Pratt and Lilhavy, 1995; Harrison-McMonagle et al., 1999; Okamura et al., 2000). Interestingly, gene promoter elements within several antibiotic biosynthetic gene clusters contain similar direct repeat sequences proposed to be important for DNA recognition and binding by their cognate SARPs (Tang et al., 1996; Wietzorrek and Bibb, 1997; McDowall et al., 1999). Using the DNR/DXR biosynthetic gene cluster as a model system, we have explored the effect of mutating the putative DnrI target tandem repeats found within the dnrG–dpsABCD promoter. This work demonstrates that these specific nucleotides play a critical role in maintaining precise transcriptional control of antibiotic biosynthetic genes. In addition, it has been revealed that proper spacing between the tandem repeats must be preserved to sustain adequate DNA–DnrI interaction and, consequently, successful transcriptional complex formation.


In vivo complementation of S. peucetius strain WMH1445 (dnrI) with truncated and alanine-substituted forms of DnrI

Based upon recent insight that SARPs may contain analogous structures to those found in the three-dimensional DNA-binding fold of OmpR (Wietzorrek and Bibb, 1997), we were interested in first mapping the region of DnrI involved in DNA recognition and interaction. To narrow the search for specific amino acids involved in DNA binding, a broad mutagenesis approach was used initially involving the truncation of varying stretches of polypeptide from the N- and C-terminal ends of the protein. After deletion of the corresponding spans of DNA, the truncated plasmid constructs (see Table 1) were transformed into a dnrI mutant strain (WMH1445), which lacks the ability to produce DNR/DXR (Stutzman-Engwall et al., 1992). When wild-type dnrI is supplied in trans to strain WMH1445, the culture gains the ability to produce the DNR biosynthetic intermediate ɛ-rhodomycinone (RHO) (Stutzman-Engwall et al., 1992) (Table 2). This inter-mediate is readily monitored by thin-layer chromato-graphy, with levels of detection in the nanogram range (the chromophore having a red/orange colour). The presence of dnrI in high copy number within the mutant strain led to the overproduction of RHO, at levels comparable with those observed previously (Stutzman-Engwall et al., 1992). Perhaps this observation is interesting from the perspective of strain development, where it is known that one can increase the copy number of DnrI and stimulate the overexpression of most, if not all, of the biosynthetic pathway genes, including the resistance genes (Madduri and Hutchinson, 1995a).

Table 1. Strains and plasmids used in this study.
Strain or plasmidRelevant characteristic(s)Source or reference
  1. Amp, ampicillin resistance; tsr, thiostrepton resistance.

Bacterial strains
S. peucetius
WMH1445 dnrI::aphII mutant Stutzman-Engwall et al. (1992)
E. coli
DH5αFrecA f80 dlacZ ΔM15Gibco BRL
BL21(DE3)FompT hsdS gal dcm (DE3)Novagen
pUC19High-copy E. coli vector; Ampr Sambrook et al. (1989)
pT7SCProtein expression vector; Ampr Brown and Campbell (1993)
pMalc2XMBP fusion vector – cytoplasmic expressionNEB
pMalp2XMBP fusion vector – periplasmic expressionNEB
pWHM1102N-terminal his-tagged dnrI in pET16b Tang et al. (1996)
pWHM1115N-terminal his-tagged dnrI under control of ermE* promoterThis work
pWHM1116N27 10% N-terminal truncation of DnrIThis work
pWHM1117N54 20% N-terminal truncation of DnrIThis work
pWHM1118N81 30% N-terminal truncation of DnrIThis work
pWHM1119N108 40% N-terminal truncation of DnrIThis work
pWHM1120C246 10% C-terminal truncation his-tagged DnrIThis work
pWHM1121C219 20% C-terminal truncation his-tagged DnrIThis work
pWHM1122C191 30% C-terminal truncation his-tagged DnrIThis work
pWHM1123MBP–DnrI fusion under control of ermE* promoterThis work
pWHM11241.2 kb XbaI–HindIII fragment from pWHM1102 in pUC19This work
pWHM1125T64A DnrIThis work
pWHM1126R71A DnrIThis work
pWHM1127G95A DnrIThis work
pWHM1128Y96A DnrIThis work
pWHM1129R181A DnrIThis work
pWHM1130L202A DnrIThis work
pWHM1131Native MBP–DnrI fusionThis work
pWHM1250High-copy-number shuttle vector containing ermE* promoter, tsr Madduri et al. (1998)
Table 2. Complementation of dnrI mutant strain.
StrainPhenotype/mutation% RHO production
pWHM1115Wild-type his-tagged DnrI100
pWHM111610% N-terminal truncate DnrI0
pWHM111720% N-terminal truncate DnrI0
pWHM111830% N-terminal truncate DnrI0
pWHM111940% N-terminal truncate DnrI0
pWHM112010% C-terminal truncate his-tagged DnrI0
pWHM112120% C-terminal truncate his-tagged DnrI0
pWHM112230% C-terminal truncate his-tagged DnrI0
pWHM1123MBP–DnrI fusion0
pWHM1125DnrI T64A0
pWHM1126DnrI R71A0
pWHM1127DnrI G96A0
pWHM1128DnrI Y97A0
pWHM1129DnrI R181A98
pWHM1130DnrI L202A71

When the truncated forms of dnrI were introduced into strain WMH1445, not one maintained the ability to complement the dnrI mutation. Indeed, even minimal truncations from the N- or C-terminal ends resulted in the disruption of DnrI function (Table 2). In an effort to ensure that the failure to restore DNR intermediate production was not the result of a lack of mutant protein expression, Western blot analyses of soluble cell fractions were conducted using a histidine tag-specific monoclonal antibody. The wild-type and C-terminal truncates should exist as his-tag fusions of DnrI (see Experimental procedures). It was found that the cultures expressed the expected wild-type and truncated forms of DnrI (data not shown). Additionally, it was observed that the maltose-binding protein (MBP)–DnrI wild-type fusion protein (construct pWHM1123) fails to support RHO production in strain WMH1445 (Table 2) (see Discussion). In contrast to the MBP–DnrI fusion, the N-terminal his-tagged wild-type DnrI (construct pWHM1115) retains the ability to complement strain WMH1445 (Table 2). It appears that the presence of the smaller, less obtrusive histidine tag does not interfere with DNA recognition and binding or with interactions with other transcriptional factors.

To determine which amino acid residues may be critical for DNA interaction and binding, individual alanine substitutions were made at targeted positions. As the predicted secondary structure of the N-terminal half of DnrI, and most SARPs, is highly similar to the winged-helix DNA-binding proteins of the OmpR family, a guide for generating precise changes was readily available (see Fig. 1) (Martinez-Hackert and Stock, 1997a; Wietzorrek and Bibb, 1997). It was observed that changes to T64, R71, G95 and Y96 caused a serious defect in the ability of the protein to stimulate RHO production in strain WMH1445 (Table 2). These results support the prediction that these amino acids make major contributions to protein function through participation in specific DNA interactions. If the winged-helix DNA-binding motif is located within the N-terminal half of the protein, mutations generated in the C-terminal end of DnrI should not inhibit transcriptional activation. To test this idea, alanine mutations were created within the C-terminal end of the protein. Changes to R181 and L202 did not abrogate RHO production, as had the N-terminal mutations. Interestingly, quantification revealed that R181A maintained nearly complete complementation of RHO production, whereas L202A resulted in less than complete restoration (Table 2).

Figure 1.

Alignment of the N-terminal amino acid sequence of DnrI with the C-terminal DNA-binding domain of OmpR. Amino acid residues that have been exchanged for alanines are highlighted above the DnrI sequence. Regions of OmpR sequence known to make specific contacts with DNA (double underline) and holo-RNAP (dashed underline) are marked. The long intervening sequence separating alpha-helices 2 and 3 is denoted.

MBP–DnrI fusion protein isolated as a stable soluble protein

In an effort to avoid the problems of low protein yield and instability encountered previously during the purification and storage of DnrI (Tang et al., 1996), we pursued a strategy to clone and express DnrI as an N-terminal MBP fusion protein. It has been demonstrated that obtaining a target protein in a soluble, biologically active form can be enhanced by fusing the aggregation-prone polypeptide to a highly soluble partner such as MBP (Kapust and Waugh, 1999). From Escherichia coli BL21(DE3), an N-terminal MBP-tagged DnrI fusion protein was overexpressed and purified via one-step amylose column affinity chromatography (Fig. 2), as were all the mutant proteins generated in this work. The wild-type DnrI fusion protein was found to behave similarly to native DnrI (Tang et al., 1996) in DNA-binding assays. Surprisingly, the fusion protein exhibited exceptional stability, maintaining DNA-binding activity for a period of up to 6 months upon storage in numerous buffer compositions at only −20°C.

Figure 2.

Representative SDS–PAGE of MBP–DnrI fusion proteins isolated from E. coli. Lane 1, MBP–DnrI (wild type); lane 2 MBP–DnrI N27; lane 3, MBP–DnrI N54; lane 4, MBP–DnrI N81; lane 5, MBP–DnrI N108; lane 6, MBP–DnrI C246; lane 7, MBP–DnrI C219; lane 8, MBP–DnrI C191. Molecular weight marker is shown on the left.

Analysis of MBP–DnrI DNA-binding properties

With the stable fusion protein in hand, more detailed DNA-binding assays were performed. The stoichiometry of DNA binding by the MBP–DnrI fusion was characterized by conducting gel mobility shift (GMS) assays. The DNA substrate was a 90 bp DNA fragment from the intergenic region separating the dpsEF and dnrG–dpsABCD divergent operons. This sequence includes one intact pair of direct repeat elements known to be critical for DNA recognition and binding (Fig. 3) (see Discussion). After precise determination of protein and DNA stock concentrations, varying amounts of protein were reacted with constant concentrations of labelled DNA substrate. As the ratio of protein to DNA substrate approached 2 : 1, nearly all labelled DNA shifted into the bound complex (Fig. 4). This result does not preclude the idea that DnrI may bind as a monomer to each of the tandem heptameric repeats. Indeed, at 0.5 : 1 and 1 : 1 ratios of protein to DNA, there are clearly bandshifts that represent an intermediate complex (Fig. 4). Perhaps this complex represents a single molecule of DnrI binding to a single repeat target within the −35 region of the dnrG–dpsABCD operon (see Discussion).

Figure 3.

. Schematic representation of the intergenic region between the dpsEF and dnrG–dpsABCD operons within the DNR/DXR biosynthetic gene cluster. The bold underlines detail the tandem heptameric ‘target’ sequences within the −35 promoter regions, and the dashed underline highlights the spacer nucleotides separating the repeats. The 90 bp XhoI–EagI and 203 bp FspI–EagI fragments used as probes for the GMS assays are shown.

Figure 4.

Stoichiometric analysis of DnrI DNA binding by gel mobility shift assay. Lane 1, free DNA; lane 2, 3.5 ng of DnrI; lane 3, 7.0 ng of DnrI; lane 4, 10.5 ng of DnrI; lane 5, 14.0 ng of DnrI. Free and protein-bound DNA label are indicated by arrows. Ratios of protein and DNA are indicated underneath.

As gel shift assays are readily quantifiable, the results from the experiment used to determine the stoichiometry of protein–DNA binding were also used to estimate the dissociation constant (Kd) for MBP–DnrI dnrG promoter affinity. The Kd is defined as the concentration of MBP–DnrI required to shift 50% of the radiolabelled substrate DNA. Densitometry of the autoradiogram revealed that a shift of 50% was observed with 7.0 ng of protein (Fig. 4), resulting in an apparent Kd of 2.4 × 10−9 M, a measure that is consistent with DNA-binding proteins within the genus Streptomyces (Rother et al., 1999).

Determination of the DNA-binding ability of DnrI mutant proteins

From the complementation studies using both truncated and site-directed mutant forms of DnrI, it was obvious that certain manipulations caused a disruption of protein function. Disruption of either transactivation or DNA binding as a cause for the abolition of complementation remained as two distinct possibilities. DNA-binding assays were carried out using purified MBP–DnrI fusion proteins of the various truncated and N-terminal alanine-substituted mutants that failed to complement strain WMH1445. The DNA substrate was a 203 bp FspI–EagI fragment that contained the entire intergenic region separating the dpsEF and dnrG–dpsABCD operons and included two intact sets of tandem heptameric repeat sequences (Fig. 3). In reaction mixtures that contained a protein concentration at least 10-fold greater than DNA substrate, none of the DnrI truncates maintained the ability to bind DNA (Fig. 5A). In similar experiments, the N-terminal alanine-substituted mutant proteins did not bind DNA (Fig. 5B). These results clearly demonstrate that the failure of the DnrI mutants to complement strain WMH1445 is directly related to the inability of the proteins to recognize and bind the specific DNA targets. DnrI contains a DNA-binding domain within its N-terminus that includes specific residues that may be comparable in function to those in the OmpR family of winged-helix DNA-binding proteins.

Figure 5.

Determination of DNA-binding ability of mutant forms of DnrI by gel mobility shift assay.

A. GMS using truncated forms of DnrI. Lane 1, free DNA; lane 2, 500 ng (same amount for all lanes) of wild-type MBP–DnrI; lane 3, MBP–DnrI N27; lane 4, MBP–DnrI N54; lane 5, MBP–DnrI N81; lane 6, MBP–DnrI N108; lane 7, wild-type MBP–DnrI; lane 8, MBP–DnrI C246; lane 9, MBP–DnrI C219; lane 10, MBP–DnrI C191.

B. GMS using SDM forms of DnrI. Lane 1, free DNA; lane 2, 500 ng of wild-type MBP–DnrI; lane 3, MBP–DnrI T64A; lane 4, MBP–DnrI R71A; lane 5, wild-type MBP–DnrI; lane 6, MBP–DnrI G95A; lane 7, MBP–DnrI Y96A; lane 8, free DNA.

Effect of promoter mutations on the in vitro binding activity of DnrI

In the dnrG–dpsABCD promoter region, DNA sequences bound by DnrI have been located by DNase I footprinting and gel mobility shift assay (Tang et al., 1996). Tandemly arrayed repeat sequences have been discovered within this sequence, which includes the region surrounding the −35 hexamer of the promoter (Tang et al., 1996; Wietzorrek and Bibb, 1997). These sequences are predicted to be critical for DNA recognition and binding by DnrI. To determine the importance of nucleotides within the tandem repeats, the DNA sequence of the direct repeats and the spacer region separating the repeats found within the dnrG–dpsABCD promoter (Fig. 3) was subjected to site-directed mutagenesis. Base changes, substituting the alternative purine or pyrimidine in each position, were created within the direct repeats. In an effort to characterize the effect of the site-directed mutations, the binding efficiency of DnrI to the mutant promoter fragments was tested by GMS assays. Each promoter fragment was excised from the mutant constructs in a process that preserved the dnrG–dpsABCD tandem repeats while disrupting the upstream dpsEF repeat sequences. Base substitutions within the 5′ repeat box resulted in a decrease in binding efficiency by DnrI (Fig. 6A), characterized by non-discrete bandshifts and smearing of label, with the one exception being at position 6 (Fig. 6A, lane 8). Substituting A for G (Fig. 3) at that position does not result in inhibition of binding and may perhaps enhance DnrI–DNA interaction. Every substitution within the 3′ repeat led to diminished DnrI binding efficiency (Fig. 6B). However, a quantitative analysis of the gel image revealed that binding was disrupted the least where an A residue replaced G at position 4 (Fig. 6B, lane 5). An addition of one nucleotide to the spacer sequence resulted in nearly complete disruption of DnrI binding (Fig. 6C, lane 3). Diminution of DnrI binding through mutation of conserved nucleotides within the putative target repeats clearly indicates the importance of these sequences and is in good agreement with similar studies performed with OmpR target sequences (Harlocker et al., 1995). Deleting 1 bp from within the linker significantly disrupts DnrI binding, but not to the extent observed where there was an addition of one nucleotide (Fig. 6C, lane 4). Again, these results are comparable with observations made after changing OmpR family target sequences (Pratt and Lilhavy, 1995), where the proper spacing between the repeats is crucial for promoter activity.

Figure 6.

Gel mobility shift assays using the mutated promoter DNA fragments.

A. Lane 1, no protein – free DNA; lane 2, wild-type DNA fragment; lanes 3–9, mutant promoters G1.1–G1.7; lane 10, double mutant promoter G1.11/G1.4.

B. Lane 1, no protein – free DNA; lane 2, wild-type DNA fragment; lanes 3 and 4, mutant promoters G2.1–G2.2; lanes 5–8, mutant promoters G2.4–G2.7.

C. Lane 1, no protein – free DNA; lane 2, wild-type DNA fragment; lane 3, mutant promoter, one nucleotide addition; lane 4, mutant promoter, one nucleotide deletion. Arrows shown on the left mark the positions of DnrI-free and -bound DNA.


The SARP family of transcriptional activators appears to be genetically and functionally related to the winged-helix family of transcriptional regulators, of which the E. coli regulatory protein OmpR is a model member. The DNA-binding domain of the SARPs is located within the N-terminus, whereas the comparable binding motif is located in the C-terminus of the OmpR family. This particular structural orientation is not unprecedented, as other members of the winged-helix OmpR–PhoB family contain the DNA-binding domain within the N-terminus (Martinez-Hackert and Stock, 1997b). OmpR belongs to a family of response regulators for which signal transduction occurs in combination with a sensor kinase partner through a phosphorelay mechanism, where the phosphorylation domain resides in the N-terminus. In contrast, DnrI is the concluding link of a pathway-specific transcriptional cascade and does not require phosphorylation to render it an active transcriptional activator.

The amino acid sequences of the DnrI (Stutzman-Engwall et al., 1992), ActII-ORF4 (Fernandez-Moreno et al., 1991), RedD (Narva and Feitelson, 1990), CcaR (Perez-Llarena et al., 1997) and a handful of other pathway-specific activators (SARPs) are quite similar. Thus, it is reasonable to predict that the tertiary structure of each protein contributes to a related mechanism of DNA binding and transcriptional activation. Interestingly, expression of dnrI can complement an actII-ORF4 mutant strain (Stutzman-Engwall et al., 1992), whereas expression of redD (Stutzman-Engwall et al., 1992) or ccaR (Perez-Llarena et al., 1997) do not. Accordingly, it has been proposed that DnrI and ActII-ORF4 recognize remarkably similar DNA repeat sequences within the promoter regions of the corresponding DNR and ACT biosynthetic genes respectively (Wietzorrek and Bibb, 1997). Indeed, upon inspection, one can find these direct repeats within the gene promoter regions bound by DnrI (Tang et al., 1996) and ActII-ORF4 (Arias et al., 1999). From the stoichiometry analysis, it is apparent that two molecules of MBP–DnrI bind to the DNA substrate containing one intact pair of heptameric repeat target sequences. Although three repeat sequences are present within the DNA fragment (Fig. 3), it is thought that productive binding by DnrI requires co-operation between monomers that recognize intact adjacent direct repeats, similar to the essential OmpR–OmpR co-operative interactions required for binding cognate target sequences (Harlocker et al., 1995). DnrI binding to all three repeat target sequences should result in the formation of two intermediate complexes. A single intermediate complex is observed in Fig. 4. Recently, the three-dimensional structure of the DNA-binding/transactivation domain of PhoB, an OmpR family member, bound to its target DNA was determined via nuclear magnetic resonance (NMR) imaging (Okamura et al., 2000). The data suggest a structure in which one PhoB monomer is bound to each direct repeat comprising the target sites. Hence, we predict that a DnrI monomer binds to each heptameric repeat within the promoters of DNR/DXR biosynthetic genes. Where discrete DNA target sequences have been clearly defined for transcriptional regulators, such as direct or inverted repeats, the stoichiometry of binding protein to DNA is ordinarily found to be 1:1 (Cicero et al., 1998).

Maltose-binding protein fused to the N-terminus disrupts the ability of DnrI to complement strain WMH1445, although it does not inhibit DNA substrate binding in GMS assays. It has been suggested that the alpha-subunit of RNAP contacts the long loop region separating alpha-helices 2 and 3 (the turn portion of this HTH variant) of the OmpR DNA-binding domain (Kondo et al., 1997). Additionally, evidence has been found suggesting that the long turn region of PhoB acts as a putative interaction site for the RNAP sigma-subunit (Makino et al., 1996; Okamura et al., 2000). Perhaps the existence of MBP on the N-terminus of DnrI disrupts potential interactions with other transcriptional factors such as RNAP.

From the structural studies of OmpR and PhoB, it has become apparent that the DNA-binding domain architecture of the winged-helix family of transcriptional activators (including the SARPs) consists of a central three-helical bundle containing the HTH variant separated by a long loop (see Fig. 1). The predicted long loop residues are not well conserved, and it has been proposed that each individual OmpR family protein takes on a characteristic conformation within the HTH based upon the make-up of the long loop region (Okamura et al., 2000). Thus, the particular loop conformation may contribute to the specificity for transactivation and DNA binding of each individual SARP. Failure to engineer the Streptomyces clavuligerus SARP CcaR into a dnrI-complementing protein by substituting residues within helix three of the HTH variant, and not within the long separating loop (P. J. Sheldon and C. R. Hutchinson, unpublished data), supports this argument. The function of the C-terminal half of the protein remains unknown at this time. It may be involved in intra- or intermolecular interactions required for oligomerization and association with other transcriptional complex proteins respectively. These predictions may explain the abrogation of protein activity upon deletion of just a few amino acids or the drop in RHO production when certain residues within the C-terminus are mutated.

The dnrG–dpsABCD promoter was chosen as the target for mutational analysis because it is an early acting and important promoter in the biosynthesis of aklanonic acid, the first detectable intermediate of daunorubicin biosynthesis. Substitutions within the heptameric repeats resulted in differential affinity of DnrI to the mutant repeat sequences, which is attested to by the variable binding of the protein in the gel mobility shift assays. In natural conditions, variations in the intracellular environment may contribute to variations in DnrI binding. OmpR differentially associates with individual promoter elements depending on the state of osmolarity surrounding and within cells (Harlocker et al., 1995; Bergstrom et al., 1998). The heptameric repeats are separated by a specific number of nucleotides; in the case of the dnrG promoter, this number is four. As the addition or deletion of the spacing nucleotides had such a dramatic effect on DNA binding, the spacing is critical for DnrI–DNA interaction. Correct spacing appears to enhance the co-operativity of protein monomers while they are binding to the target sequence(s). Changing the spacing between adjacent heptameric repeat target sequences could either disrupt the ability of DnrI monomers to interact with the DNA (induced by steric hindrance) or may adversely affect the spacing of the −35 and −10 hexamers, in a manner that prevents a competent transcription complex from forming. In support of the former idea, gel mobility shift assays clearly show that binding efficiency is dramatically diminished when an addition or deletion occurs within the sequence that separates the direct repeat binding sites. Increasing the space separating the repeats caused greater interference, suggesting that co-operativity among protein monomers is significant.

The results reported here suggest that targeted changes to the repeat sequences recognized by DnrI may be useful in the engineering and development of more effective DNR/DXR-producing strains. As there are data indicating that DnrI controls transcriptional activation of most, if not all, the DNR/DXR biosynthetic genes (Madduri and Hutchinson, 1995b; Tang et al., 1996), metabolic engineering of DnrI could potentially increase DNR and DXR yield. These predictions might also be extended to the SARP family as a whole. SARP genes isolated from various organisms or from within different regions of the genome (Culebras et al., 1999) could be exploited in the development of valuable production strains. Significantly, this family of regulators has been found to be a key component within the biosynthetic gene clusters of many commercially notable Actinomycete-derived pharmaceuticals such as cephamycin (Perez-Llarena et al., 1997), tetracycline (McDowall et al., 1999) and tylosin (Bate et al., 1999).

Experimental procedures

Bacterial strains, culture conditions and media

Escherichia coli DH5α used as a host for the generation of double-stranded plasmid DNA was grown at 37°C on Luria–Bertani (LB) medium (Sambrook et al., 1989). E. coli BL21(DE3) (Novagen), used as host for protein expression and was grown at 37°C in LB medium. S. peucetius strain WMH1445 (Stutzman-Engwall et al., 1992) was grown on R2YE medium containing 0.5% glycine at 30°C for the preparation of protoplasts (Hopwood et al., 1985). Strain WHM1445 transformants were grown on ISP medium 4 agar plates amended with 0.2% yeast extract and 25 μg of thiostrepton ml−1 medium at 30°C to produce spores for culture stock preservation.

Biochemicals and other reagents

All commonly used biochemicals were purchased from Sigma. Thiostrepton, used for antibiotic selection of Streptomyces transformations, was kindly provided as a gift from S. Lucania, Bristol-Myers-Squibb, Princeton, NJ, USA. All restriction endonucleases and DNA-modifying enzymes were purchased from either New England Biolabs (NEB), Gibco BRL or Promega.

Complementation of strain WMH1445

From pWHM1102, a construct that produces a his6x-tagged version of DnrI (Tang et al., 1996), a 1.2 kb XbaI–HindIII fragment was extracted and ligated to the same sites in vector pWHM1250 (Madduri et al., 1998). This construct (pWHM1115) results in dnrI expression under the control of the strongly constitutive Streptomyces promoter ermE*p (Bibb et al., 1994). The dnrI N-terminal truncations were fabricated using the polymerase chain reaction (PCR) to synthesize DNA fragments that incorporated an NdeI restriction endonuclease site at positions within the 5′ coding sequence of the dnrI gene and a HindIII site downstream of the dnrI stop codon. The DNA fragments were then ligated to the same sites in plasmid vector pT7SC (Brown and Campbell, 1993). From the resulting constructs, XbaI–HindIII fragments were isolated and ligated to the same sites in plasmid vector pWHM1250 to form plasmids pWHM1116–1119 (see Table 1). To construct the dnrI C-terminal truncations, PCR was used to generate DNA fragments that would incorporate a native XbaI site upstream of the start codon and a TGA stop codon followed by a HindIII site at various positions within the C-terminal coding sequence of dnrI. The PCR-derived fragments also contained coding sequence that would incorporate a histidine tag on the N-terminal end of the protein. The DNA fragments were digested with XbaI–HindIII and ligated to the same sites in vector pWHM1250 creating plasmids pWHM1120–1122 (see Table 1). An MBP–DnrI (wild-type DnrI) fusion under the control of ermE*p was constructed using PCR to amplify dnrI containing EcoRI and HindIII sites at the 5′ and 3′ ends, respectively, using pWHM1102 as template. The EcoRI–HindIII fragment was then cloned into the same sites in vector pMalp2X (NEB) to form the MBP–DnrI fusion. From the resulting plasmid, an NdeI–HindIII fragment containing the MBP–DnrI fusion was ligated to the same sites in pT7SC to add restriction sites and a ribosome binding site (preceding the MBP start site). An XbaI–HindIII fragment was excised from this intermediate construct and cloned into the same sites in pWHM1250 to give plasmid pWHM1123. The DnrI mutant constructs were then introduced into S. peucetius strain WMH1445. Transformants were selected for resistance to thiostrepton (25 μg of drug ml−1 medium) and kanamycin (10 μg of drug μl−1 medium). To analyse metabolite production, transformants were grown in APM medium according to previously published protocols (Stutzman-Engwall et al., 1992) and analysed by thin-layer chromatography (TLC) for the presence of ɛ-rhodomycinone (RHO), a doxorubicin biosynthetic intermediate. Quantification of RHO was made by extracting normalized culture volumes with CHCl3, drying the extracted compound under vacuum and determining the mass. As the expression of PKS genes involved in the construction of the molecular core of doxorubicin has been shown to result in an accurate representation of pathway regulation by DnrI (Stutzman-Engwall et al., 1992), the use of the validated WMH1445 strain for measurement of RHO was warranted. For the oligonucleotide primer sequences used for the PCRs, see Table 3.

Table 3. Oligonucleotides used for DnrI mutagenesis.
N-terminal truncates pWHM1116-1119
C-terminal truncates pWHM1120-1122
 HindIII DnrI stop
MBP fusion oligonucleotides
  XmnI site DnrI start
 HindIII DnrI stop
 EcoRI DnrI start
N-terminal truncates
MBP C-terminal truncations were constructed using the native MBP–DnrI forward primer and the C246-191 reverse primers listed above
Site-directed mutations

Site-directed mutagenesis of DnrI

To produce DnrI mutants, the Stratagene Quickchange site-directed mutagenesis (SDM) kit was used according to the manufacturer’s protocol. Specific nucleotide substitutions were introduced using mutagenic oligonucleotides synthesized at the University of Wisconsin–Madison Biotechnology Center. For the PCR-based reactions, pWHM1124 was used as the template DNA. This construct contains the 1.2 kb XbaI–HindIII his-tagged dnrI fragment from pWHM1102 cloned into the same sites in pUC19. For the dnrI complementation experiments, an XbaI–HindIII fragment from SDM constructs was extracted and ligated to the same sites in pWHM1250, resulting in plasmids pWHM1125–1130 (see Table 1). The SDM DnrI mutants were confirmed to contain only the desired changes by nucleotide sequence analysis. These constructs were transformed into S. peucetius strain WMH1445, and the resulting transformants were analysed as described above. The nucleotide sequences of the site-directed mutagenic oligonucleotides are found in Table 3.

Site-directed mutagenesis of the dnrG–dpsABCD promoter

Site-directed mutations within the dnrG–dpsABCD promoter were made using the Quickchange kit as described above. For the PCR-based reactions, pWHM1105 (Tang et al., 1996) was used as the template DNA. This construct contains the dpsEF–dnrGdpsABCD intergenic region cloned into vector pUC18 (Sambrook et al., 1989). After determining that the correct mutations had been made by sequence analysis, the constructs were digested with SalI and BamHI, and the resulting 450 bp fragments, containing the mutant dnrG promoters, were cloned into the same sites in plasmid pGEM11zf (Promega).

Construction and expression of MBP–DnrI fusions

Fusion of DnrI to the MBP was accomplished using PCR to generate DNA fragments from template vector pWHM1102 containing dnrI wth XmnI and HindIII sites before the start and downstream of the translational stop codons respectively. The resulting DNA fragments were ligated to the XmnI–HindIII sites of plasmid vector pMalc2X (NEB). To construct the N-terminally truncated forms of MBP–DnrI, PCR was used to amplify DNA fragments that incorporated an XmnI restriction site at positions within the 5′ coding sequence of the dnrI gene and a HindIII site downstream of the dnrI stop codon. The resulting DNA fragments were then treated as described above. C-terminal DnrI–MBP truncates were constructed using PCR to generate DNA fragments that included an XmnI site upstream of the native start codon and a TGA stop codon followed by a HindIII site within the C-terminal coding sequence of dnrI. The resulting DNA fragments were treated in the same manner as above to give the final DnrI C-terminally truncated MBP fusion constructs. Using the same oligonucleotide primers as those for the construction of the wild-type DnrI fusion, fragments from the site-directed mutant constructs (pWHM1126–1130) were amplified and cloned in the same manner as described above to form MBP–DnrI SDM fusion plasmids. For expression of MBP–DnrI fusion proteins, the above constructs were transformed into E. coli strain BL21(DE3), grown at 37°C to an OD600 of ≈ 0.6 and protein expression induced with the addition of IPTG to a final concentration of 1 mM. After disruption of cells via sonication, the cytosolic fraction was passed over an amylose–agarose resin (NEB), resulting in the isolation of MBP-tagged proteins. After release of the MBP–DnrI proteins from the amylose column using maltose, the maltose was removed by filtration concentration using an Amicon microcon filter apparatus and repeated washing with storage buffer (10 mM Tris, pH 7.2, 20 mM NaCl).

Mobility shift DNA-binding assays

Either a 203 bp FspI–EagI or a 90 bp XhoI–EagI DNA fragment containing the dpsEF–dnrGdpsABCD intergenic promoter region was end labelled with [α-32P]-dCTP (Amersham) using the Klenow large fragment of DNA polymerase. Binding reactions (40 μl) contained 7.5% glycerol, 1.5 μg of poly-(dI–dC), purified protein of varying concentrations and a constant concentration of labelled substrate DNA (10 000–15 000 c.p.m.) in a buffer composed of 90 mM Tris-HCl, pH 8.5, 90 mM boric acid and 20 mM EDTA. After incubation at ambient temperature for 30 min, protein-bound and free DNA were separated by electrophoresis at ambient temperature on a 5% native polyacrylamide gel running at ≈ 25 mAmp. The gels were developed and analysed using a PhosphorImager SI (Molecular Dynamics) and the IMAGEQUANT software respectively. Densitometry of the developed phosphorimages was used in the stoichiometric and kinetic calculations of DnrI–DNA interactions. The mutant promoter constructs were examined as follows. PCR was used to amplify a fragment from the dnrGdpsABCD mutagenic promoter regions. After digestion with XhoI and EagI, the fragments were end labelled as described above. Binding reactions (20 μl) contained 7.5% glycerol, 1.5 μg of poly-(dI–dC), 48 ng of purified MBP–DnrI protein and a normalized concentration of labelled substrate DNA (10 000–15 000 c.p.m.) in the same buffer described above.


This research was supported in part by a grant from the National Institutes of Health (CA64161).