Transcriptional Regulation of Congocidine (Netropsin) Biosynthesis and Resistance

ABSTRACT

The production of specialized metabolites by Streptomyces bacteria is usually temporally regulated. This regulation is complex and frequently involves both global and pathway-specific mechanisms. Streptomyces ambofaciens ATCC23877 produces several specialized metabolites, including spiramycins, stambomycins, kinamycins and congocidine. The production of the first three molecules has been shown to be controlled by one or several cluster-situated transcriptional regulators. However, nothing is known regarding the regulation of congocidine biosynthesis. Congocidine (netropsin) belongs to the family of pyrrolamide metabolites, which also includes distamycin and anthelvencins. Most pyrrolamides bind into the minor groove of DNA, specifically in A/T-rich regions, which gives them numerous biological activities, such as antimicrobial and antitumoral activities. We previously reported the characterization of the pyrrolamide biosynthetic gene clusters of congocidine (cgc) in S. ambofaciens ATCC23877, distamycin (dst) in Streptomyces netropsis DSM40846, and anthelvencins (ant) in Streptomyces venezuelae ATCC14583. The three gene clusters contain a gene encoding a putative transcriptional regulator, cgc1, dst1, and ant1, respectively. Cgc1, Dst1, and Ant1 present a high percentage of amino acid sequence similarity. We demonstrate here that Cgc1, an atypical orphan response regulator, activates the transcription of all cgc genes in the stationary phase of S. ambofaciens growth. We also show that the cgc cluster is constituted of eight main transcriptional units. Finally, we show that congocidine induces the expression of the transcriptional regulator Cgc1 and of the operon containing the resistance genes (cgc20 and cgc21, coding for an ABC transporter), and propose a model for the transcriptional regulation of the cgc gene cluster.

IMPORTANCE Understanding the mechanisms of regulation of specialized metabolite production can have important implications both at the level of specialized metabolism study (expression of silent gene clusters) and at the biotechnological level (increase of the production of a metabolite of interest). We report here a study on the regulation of the biosynthesis of a metabolite from the pyrrolamide family, congocidine. We show that congocidine biosynthesis and resistance are controlled by Cgc1, a cluster-situated regulator. As the gene clusters directing the biosynthesis of the pyrrolamides distamycin and anthelvencin encode a homolog of Cgc1, our findings may be relevant for the biosynthesis of other pyrrolamides. In addition, our results reveal a new type of feed-forward induction mechanism, in which congocidine induces its own biosynthesis through the induction of the transcription of cgc1.

INTRODUCTION

The control of specialized metabolite production in Streptomyces is a complex process that often involves multiple levels of regulation, including transcriptional and translational regulations and precursor availability (1). At the transcriptional level only, a large diversity of regulation mechanisms may be involved. The transcription of a specialized metabolite biosynthetic gene cluster can be controlled by “global” regulators sensing and responding to the nature of the nutrient source (for example DasR) or to nutrient limitation (GlnR for nitrogen or PhoP for phosphate limitations) (1). Small signaling molecules (γ-butyrolactones, furans, ppGpp) may also be required, participating to the regulatory cascade leading to the activation or repression of cluster-situated regulators (CSRs, also called pathway-specific regulators), encoded by genes located within biosynthetic gene clusters and generally dedicated to their regulation (1, 2). In some cases, specialized metabolites themselves (e.g., actinorhodin or simocyclinone D8) or their late biosynthetic intermediates (simocyclinone C4) can also contribute to the control of the transcription of their biosynthetic genes, through interactions with various transcription factors (2–5). Understanding the regulation mechanisms that govern the biosynthesis of a metabolite have important implications. In addition to participating to a better knowledge of the biology of the producer, it may help to determine conditions favorable to this metabolite production or give means to increase this production (6). As specialized metabolites remain an important source of drugs, yield improvement can be economically valuable.

We previously isolated and characterized the biosynthetic gene clusters of three pyrrolamide antibiotics, congocidine (also called netropsin, Fig. 1) in Streptomyces ambofaciens ATCC23877, distamycin in Streptomyces netropsis DSM40846, and anthelvencins in Streptomyces venezuelae ATCC14583 (7–9). Pyrrolamides are characterized by the presence of 4-aminopyrrole-2-carboxylate groups linked by amide bonds. They exhibit a large range of biological activities (antiviral, antimicrobial or anthelmintic activities), mainly due to their ability to bind into the minor groove of the DNA (10). Indeed, congocidine and distamycin have been extensively studied for the specificity of their binding to DNA. Physical studies (e.g., X-ray structure determination and circular dichroism measurements) have shown that they exhibit a marked preference for A/T rich DNA regions (11, 12). Many analogs and hybrids of congocidine and distamycin have been synthesized chemically, in attempts to modify or increase the sequence specificity of the DNA binding for gene expression regulation applications (10), or to create new nontoxic antimicrobial molecules (13).

FIG 1.

Genetic organization of the congocidine gene cluster in S. ambofaciens (A). The gene functions are indicated by various filling patterns. Promoter regions in the cgc gene cluster are indicated by arrows. (B) Chemical structure of congocidine. Colors indicate the different types of congocidine substructures.

The regulation of pyrrolamide production by Streptomyces species has not been explored so far. Our previous studies have revealed the presence of one gene, cgc1 in the S. ambofaciens congocidine gene cluster, dst1 in the S. netropsis distamycin/congocidine/disgocidine gene cluster and ant1 in the S. venezuelae anthelvencin gene cluster, coding for three homologous putative transcriptional response regulators (7–9). This, combined with the strikingly similar genetic organization of the three pyrrolamide gene clusters, suggests that the transcriptional regulation of these three gene clusters could be similar.

In this study, we show that Cgc1, an atypical orphan response regulator, activates the transcription of the congocidine biosynthetic and resistance genes during the stationary phase in S. ambofaciens. We determine the transcriptional organization of the cgc gene cluster. Finally, we show that congocidine induces the expression of cgc1 encoding the transcriptional regulator Cgc1 and of the operon containing the resistance genes, cgc20 and cgc21; and we propose a model for the transcriptional regulation of the cgc gene cluster.

RESULTS

Cgc1 activates the transcription of the cgc genes.

We analyzed the sequence of S. ambofaciens Cgc1 using BLAST (14) to search for homologous protein and CD search (15) to look for conserved domains. These analyses suggested that cgc1 encodes a transcriptional regulator. This led us, in a previous study, to propose that Cgc1 may control the transcription of the cgc genes (7). To examine this hypothesis, cgc1 was deleted in the SPM110 reference strain, yielding the CGCA021 strain. S. ambofaciens SPM110 is a strain derived from S. ambofaciens ATCC23877 that produces congocidine but does not produce spiramycins (16). The production of congocidine and the transcription of the cgc genes were then analyzed. Comparative LC analyses of SPM110 and CGCA021 culture supernatants (120 h of culture in MP5 production medium) showed that production of congocidine is abolished in CGCA021 (Fig. 2A). RT-PCR analyses were carried out on RNA extracted after 43 h of culture in MP5 medium, a time at which congocidine production has started. These analyses showed that all cgc genes are transcribed in the SPM110 reference strain whereas transcription of all of them is abolished in CGCA021 (Fig. 2B). This suggested that Cgc1 acts as an activator of the transcription of the cgc gene cluster.

FIG 2.

Determination of the function of Cgc1. (A) LC analysis of congocidine production in the SPM110 reference strain (upper panel), the cgc1 deletion mutant CGCA021 (middle panel) and the genetically complemented CGCA031 strain (lower panel). (B) RT-PCR analysis of the expression of the cgc genes after 43 h of culture in MP5 medium in the SPM110 reference strain strain (upper panel), the cgc1 deletion mutant CGCA021 (middle panel) and the genetically complemented CGCA031 strain (lower panel). hrdB transcript was used as a control.

To confirm this result, we genetically complemented CGCA021 (cgc1 deletion mutant). We cloned cgc1 under the control of the strong ermE* promoter in the pOSV554 integrative vector and introduced the resulting pCGC081 plasmid in CGCA021 (yielding CGCA031 strain). LC analysis showed that congocidine production is restored into CGCA031 (Fig. 2A) and RT-PCR analyses at 43 h confirmed that all cgc genes are expressed in the complemented strain (Fig. 2B). Altogether, these data demonstrate that Cgc1 is a pathway specific activator of the transcription of the cgc genes in S. ambofaciens.

Cgc1 is an atypical orphan response regulator belonging to the NarL family.

BLAST and CD search analyses (14, 15) of the amino acid sequence of Cgc1 showed that this protein belongs to the family of two-component response regulators. More specifically, Cgc1 belongs to the NarL family of transcriptional regulators, constituted of a N-terminal receiver (REC) domain coupled with a C-terminal LuxR-type DNA-binding Helix Turn Helix (HTH) domain (17). In bacterial genomes, genes encoding response regulators are typically found clustered with genes coding for histidine kinases, with which they form the classical two-component signal transduction systems (18). No histidine kinase is encoded within the cgc gene cluster, indicating that Cgc1 is an orphan response regulator. Such regulators may rely on kinases encoded elsewhere in the genome for their activation. However, some orphan response regulators, called atypical regulators, do not rely on phosphorylation (and thus on a histidine kinase) for their activation and their receiver domain lacks some of the conserved amino acids of the phosphorylation pocket (19).

To determine whether Cgc1 is an atypical orphan response regulator, we aligned Cgc1 REC domain with REC domains from either classical (SCO3008; SCO2281; SCO3818; SCO1654; BldM and Aor1) or atypical (RedZ, WhiI, JadR1, SCO3134; SCO4144; PapR6 and VmsT) orphan response regulators from Streptomyces species and with the prototypical member of the family, CheY (Escherichia coli) (Fig. S1 in Supplemental material). REC domains from Dst1, homolog of Cgc1 for the biosynthesis of congocidine, disgocidine and distamycin in S. netropsis DSM40846 (8), and from Ant1, homolog of Cgc1 for the biosynthesis of anthelvencins in S. venezuelae ATCC14583 (9) were also included. The alignment shows that the residues constituting the phosphorylation pocket of REC domains, conserved in typical response regulators, are absent in Cgc1, Dst1 and Ant1, as in other atypical orphan response regulators (Fig. S1). This strongly suggests that Cgc1 and its homologs encoded in other pyrrolamide biosynthetic gene clusters are atypical orphan response regulators.

The cgc gene cluster is constituted of eight transcriptional units.

Having demonstrated that Cgc1 controls the transcription of the cgc genes, we established the transcriptional organization of the cgc cluster. The 22 cgc genes are clustered on the chromosome in a 29 kb region. Given their respective orientation (Fig. 1), the minimal number of transcriptional units is two. We undertook the study of the transcriptional organization of the cgc gene cluster. The absence or small size of intergenic regions between several genes of the cgc cluster means that the promoter of a gene is located in the coding sequence of the upstream gene. For this reason, we chose the approach consisting in the insertion of a transcriptional terminator as described by Dangel and colleagues (20). To identify the transcriptional units, we replaced the first gene of a putative unit by a cassette containing the apramycin resistance gene flanked with transcription terminators (Ωaac) (21). The transcription of the downstream genes was then studied by RT-PCR. However, this strategy could not be used to determine whether cgc1 is cotranscribed with cgc2 and the other downstream genes in the SPM110 strain, as Cgc1 controls the transcription of all cgc genes. Similarly, replacing cgc20 with the apramycin resistance cassette was problematic as (i) cgc20 codes for a subunit of the ABC transporter conferring congocidine resistance, and (ii) export of congocidine by this ABC transporter is likely the only resistance mechanism of S. ambofaciens against congocidine (7). Thus, deleting cgc20 in SPM110 could result in a congocidine-sensitive strain still producing congocidine (if cgc22 is not cotranscribed with cgc20 and cgc21), a situation that would most likely be lethal. In these two cases, cotranscription was investigated by amplifying various large regions encompassing these adjacent genes by RT-PCR, using RNA extracted from the reference strain SPM110 as a matrix. Total RNA extraction for the different strains was performed after 43 h of cultivation, a time at which all the cgc genes are transcribed.

The transcriptional analysis of the cgc gene cluster is detailed in the Materials and Methods section and in Fig. S2. We failed to amplify the cgc1-cgc2 intergenic region using primers located in cgc1 and cgc2, suggesting that cgc1 and cgc2 are not cotranscribed (data not shown). We observed that the insertion of a transcriptional terminator into cgc2, cgc3, cgc6, and cgc19 had no influence on the transcription of the downstream genes, indicating that they are transcribed alone. On the contrary, insertion of the same terminator into cgc4 abolished the transcription of cgc5 and insertion into cgc7 abolished transcription of all genes from cgc8 to cgc18. We detected a transcript overlapping the three genes cgc20, cgc21, and cgc22. Altogether, our results indicate that the cluster is constituted of eight main transcriptional units, as shown in Fig. 1.

The cgc gene cluster expression is induced during metabolic differentiation.

In a previous, more general study, we performed RNA-seq experiments with S. ambofaciens grown for different time periods in MP5 medium (GEO accession number GSE162865) (22). Focusing on the cgc gene cluster, we noticed that the current NCBI annotation of S. ambofaciens ATCC 23877 genome (NZ_CP012382.1) incorrectly defines the cgc1 start codon, the most probable start codon being located 450 bp upstream the defined one. We therefore conducted a new transcriptome analysis on the modified annotation. Comparing the transcriptome of 24 h old cultures, when congocidine is not yet produced, with the transcriptome of 48 h old cultures, at the onset of congocidine production, showed important changes in gene expression (Fig. 3A). About two-thirds of genes are regulated in a statistically significant manner (30% being induced and 27% repressed in stationary phase). Remarkably, the cgc genes are among the most expressed and upregulated genes in stationary phase (Fig. 3A). During exponential phase of growth (24 h), all cgc genes are poorly transcribed, except cgc1 (Fig. 3B). On the contrary, during stationary phase, the whole cluster is induced in a statistically significant manner. In both conditions, cgc1 is the most expressed gene of the cgc cluster. This emphasizes the importance of Cgc1 as a regulator involved both, in the initial induction and in the maintenance of the cgc gene cluster expression.

FIG 3.

Congocidine gene cluster transcriptome dynamics overgrowth in MP5 medium. Bacteria were grown in MP5 liquid medium and harvested in either exponential (24 h) or stationary (48 h) phase. (A) Plot of the log ratio of differential expression as a function of gene expression in stationary phase. The differentially expressed genes (adjusted P value < 0.05) are colored in red circles and yellow triangles for cgc and non-cgc genes, respectively. (B) Focus on the RNA-seq analysis of cgc gene cluster in exponential and stationary phase. Each dot corresponds to the number of reads (normalized by DESeq2 and on gene size, log₂) for one gene in the different RNA-seq condition performed in triplicates. A circle (in the color of the condition of interest) has been added to indicate overlapping dots. Of note, the expression of the cgc genes was statistically induced in stationary phase (adjusted P value ≤ 0.01, Supplemental Data set S1).

When the reads from the RNA-seq experiment performed after 48h of growth were mapped on the DNA sequence of the cgc gene cluster, the results were in agreement with the transcriptional organization determined by genetic approaches (Fig. S3). In particular, cgc1 and cgc2 do not constitute an operon, confirming the results obtained previously. Interestingly, the relative level of gene expression decreased from the beginning to the end of an operon, so that cgc22 (last gene in the cgc20-cgc22 operon) and cgc10 to cgc18 (at the end of the cgc7-cgc18 operon) were expressed at a relatively low level compared to the other genes of the congocidine cluster (Fig. 3B). This result suggests that the relative abundance of the Cgc proteins may greatly vary even if they are encoded by genes from the same operon.

Congocidine induces the expression of the congocidine resistance genes in S. ambofaciens.

The cgc20 and cgc21 genes encode the subunits of an ABC transporter that is sufficient to confer congocidine resistance to a Streptomyces sensitive strain (Streptomyces lividans TK23) (7). Moreover, a S. ambofaciens strain deleted for the entire cgc gene cluster is as sensitive to congocidine as S. lividans TK23, suggesting that no gene outside the cluster is involved in congocidine resistance (7). We observed in the present study that cgc1 appears to control the transcription of all cgc genes (Fig. 2), yet the cgc1 deletion mutant (CGCA021) is resistant to congocidine at a level comparable to the one of the parental SPM110 strain, while a strain deleted for the complete cgc gene cluster is sensitive to congocidine under the same conditions (Fig. S4 in supplemental material).

To explain the observed resistance of CGCA021 (Δcgc1) to congocidine, we hypothesized that the resistance genes cgc20 and cgc21 might be expressed independently of Cgc1 in the presence of congocidine. To test this hypothesis, we cultivated CGCA021 in the presence of 5 μg/ml (11.6 μM) congocidine in MP5 medium. Total RNAs were extracted at 43 h and the expression of cgc20 and cgc21, together with the expression of a few biosynthetic genes (cgc22, cotranscribed with cgc20 and cgc21, cgc2, cgc3 and cgc19), were determined by RT-PCR (Fig. 4).

FIG 4.

RT-PCR analysis of the expression of cgc20 and cgc21 (resistance genes), of cgc1 (transcriptional regulator gene) and of cgc22, cgc2, cgc3 and cgc19 (biosynthetic genes) in the reference strain SPM110 and in the cgc1 deletion mutant (CGCA021) in the presence of 11.6 μM congocidine. The transcript of hrdB was used as a control. It should be noted that the primers used do not allow the detection of cgc1 transcripts for the CGCA021 (Δcgc1) strain as they are complementary to the deleted region.

While in the absence of congocidine, no cgc gene is transcribed in CGCA021 (Fig. 2B), in the presence of exogenous congocidine, transcription of cgc20, cgc21, cgc22 (encoding an acyl-CoA synthetase) and, to a lesser extent, of cgc2 is observed. This indicates that the presence of congocidine triggers the transcription of the cgc20-cgc22 operon, but not the transcription of other biosynthetic genes (with the possible exception of cgc2), by a mechanism independent of Cgc1. As Cgc1 is the only transcriptional regulator encoded with the cgc gene cluster, the transcription of cgc20 and cgc21 in the absence of cgc1 and presence of congocidine could involve a transcriptional regulator gene located elsewhere in the genome. In this case, the expression of the resistance genes may not be induced by congocidine in a heterologous host.

Congocidine induces the expression of the congocidine resistance genes in the heterologous host S. lividans TK23.

To test whether the induction of the expression of the cgc20-21 resistance genes would also occur in a heterologous context, we constructed a S. lividans strain expressing the cgc gene cluster except for cgc1. By PCR targeting, we replaced cgc1 by an att2aac cassette in the BAC (pCGC002) that we previously used for the heterologous expression of the cgc gene cluster (7). After excision of the att2aac cassette, the resulting BAC (pCGC313), lacking cgc1, was introduced in S. lividans TK23 by intergeneric conjugation and the resulting strain was named CGCL095. The strain CGCL095 was cultivated in the presence or in the absence of 11.6 μM congocidine in MP5 medium, respectively. Total RNAs were extracted at 43 h and the expression of cgc20 and cgc21, together with the expression of a few biosynthetic genes (cgc22, cotranscribed with cgc20 and cgc21, and cgc2, cgc3, cgc4, cgc7 and cgc19), were determined by RT-qPCR (Fig. 5).

FIG 5.

RT-qPCR analysis of the expression of the cgc genes by the S. lividans CGCL095 strain cultivated in presence or absence of 11.6 μM exogenous congocidine. S. lividans CGCL095 strain was cultured 43 h in 50 ml liquid medium supplemented or not with 11.6 μM congocidine. The results correspond to the mean fold change (± standard error) of RNA transcripts quantified in the presence of congocidine over the control condition (no congocidine) in three independent clones. The quantifications were performed in triplicates and normalized on the mean of four reference genes (aspS, rpoB, gyrA, gyrB). Significant differences between the level of expression in presence or in the absence of congocidine are indicated with asterisks (based on a Welch two sample t test, *P < 0.05, **P < 0.01).

Addition of 11.6 μM congocidine in the culture medium resulted in a statistically significant increase of the expression of the genes from the cgc20-cgc22 operon (between 2.8- and 3.9-fold, Fig. 5). The expression of the biosynthetic genes tested did not change in a statistically significant manner (cgc2, cgc3 and cgc19) or decreased slightly (cgc4, cgc6 and cgc7) (Fig. 5). Thus, in a congocidine nonproducing host, as already observed in the natural producer, S. ambofaciens, the presence of congocidine induces the expression of the cgc20-cgc22 operon comprising the resistance genes, suggesting that congocidine may exert a direct effect on the promoter of these genes.

Exogenous congocidine only affects the transcription of a subset of cgc genes in S. ambofaciens.

We next studied the effect of congocidine on S. ambofaciens ATCC23877 in the absence of endogenous production. For this purpose, we grew S. ambofaciens ATCC23877 in YEME10 medium, a medium in which limited production of congocidine is detected in culture supernatants after metabolic differentiation (22). The strain was grown for 24 h in medium supplemented or not with 2.3 μM (1 μg/ml) congocidine, respectively. We used this concentration of congocidine as it alters the growth of a Streptomyces-sensitive strain while having no impact on the growth of the native producer (22).

The addition of congocidine had very little impact on gene expression levels, which were modified in a statistically significant manner in the presence of congocidine for six genes only (Fig. 6, Data set S1). Among them, only one gene, SAM23877_RS20590 (encoding a hypothetical protein), is not located within the cgc cluster. Its expression is only slightly repressed in the presence of congocidine (0.49-fold, adjusted P value 0.0042). The largest changes in gene expression, however, were observed in the cgc gene cluster. The expression of cgc20, cgc21 and cgc22, for which an induction in the presence of congocidine had already been observed, was strongly induced (50.3-fold, 57.1-fold and 40.5-fold induction, respectively). In addition, an induction of the expression was observed for cgc1 (9-fold induction) and, at a lower level, for cgc2 (3.4-fold induction). Thus, at low concentration (less than 1% of the concentration usually reached in culture supernatants), congocidine appears to exert a very specific effect on the cgc20-cgc1 intergenic region that contains the promoters for the transcription of cgc1 and of the cgc20-cgc22 operon.

FIG 6.

Congocidine gene cluster transcriptome dynamics in the presence of exogenous congocidine. Bacteria were grown in YEME10 liquid medium in presence or absence of 2.3 μM congocidine and harvested in exponential (24 h) phase. Each dot corresponds to the number of reads (normalized by DESeq2 and on gene size) for one gene in the different RNA-seq condition performed in triplicates. A circle (in the color of the condition of interest) has been added to indicate overlapping dots. Genes whose expression is statistically induced (adjusted P value ≤ 0.01, supplemental Data set 1) are framed in dotted lines, and their fold change (considering unlogged numbers of reads) indicated.

Congocidine increases transcription driven by the cgc1 and cgc20 promoters in a heterologous host.

To further investigate the effect of congocidine on the promoters of cgc1 and cgc20, we cloned the cgc20-cgc1 intergenic region in both orientations (P_cgc20 and P_cgc1) upstream of gusA (see Materials and Methods section for details). The gusA gene codes for a β-glucuronidase and is commonly used as a reporter gene (23). The resulting plasmids (pCGC310 and pCGC311; see Table 1), together with a control one not containing any promoter region upstream of the gusA reporter gene (pCGC309), were introduced by intergeneric conjugation into the S. lividans TK23 heterologous host. The resulting exconjugants were verified by PCR (see Materials and Methods section for details). The reporter strains were first cultivated at 28°C in 10 ml of TSB medium for 24h, a time at which 1 μg/ml (2.3 μM) or 5 μg/ml (11.6 μM) congocidine was added to the cultures. Cultures were continued for 5 h and cells were harvested and lysed. GusA activity in the crude soluble protein fraction was measured spectrophotometrically for each strain using p-nitophenyl-β-d-glucuronide as the substrate. The results from four independent experiments are presented in Fig. 7. In the absence of congocidine, a low level of GusA activity is observed when gusA is under the control of P_cgc1 whereas no activity is observed when gusA is under the control of P_cgc20. Addition of 2.3 μM or 11.6 μM congocidine to the culture medium leads to a significant increase of GusA activity (about 11.5-fold when gusA expression is under the control of P_cgc1, between 12- and 68-fold when gusA expression is under the control of P_cgc20). Thus, in a congocidine nonproducing host, congocidine acts as an inducer of the expression of genes under the control of both the P_cgc20 and P_cgc1 promoters. A difference of GusA activity between the strain expressing gusA under the control of P_cgc1 and the one expressing gusA under the control of P_cgc20 is observed. This difference may be due to various factors, including a differential effect of congocidine on the two promoters or different levels of GusA protein expression due to different Ribosome Binding Sites (RBS). Nevertheless, RNA-seq analyses (Fig. 3 and 6) also show similar trends in basal levels and expression induction for cgc1 and cgc20. This suggests more likely a difference in the strength of the divergent promoters that control the expression of these genes.

TABLE 1.

Strains and plasmids used in this study

Name	Description	Reference
E. coli strains
S17.1	Host strain for conjugation from E. coli to Streptomyces	(57)
ET12567/pUZ8002	Host strain for conjugation from E. coli to Streptomyces	(49)
DH5α	General cloning host	Promega
DY330	Strain used for λRED-mediated recombination	(51)
BW25113  pIJ790	Strain used for λ RED-mediated recombination	(49)
Streptomyces strains
S. ambofaciens ATCC 23877	Wild-type S. ambofaciens strain (type strain)	ATCC
SPM110	Derived from S. ambofaciens OSC2, spiramycin nonproducer	(16)
CGCA013	SPM110-cgc1 :: att2aac	This work
CGCA017	SPM110-cgc2 :: att2Ωaac	This work
CGCA019	SPM110-cgc22/19:: att3	(7)
CGCA021	SPM110-cgc1 :: att2	This work
CGCA031	CGCA021 containing pCGC081	This work
CGCA039	SPM110-cgc3 :: att1Ωaac	This work
CGCA040	SPM110-cgc6 :: att1Ωaac	This work
CGCA041	SPM110-cgc7:: att1Ωaac	This work
CGCA043	SPM110-cgc18 :: att1Ωaac	This work
CGCA044	SPM110-cgc4:: att1Ωaac	This work
CGCL095	S. lividans TK23 harboring pCGC313 (cgc gene cluster deleted for cgc1)	This work
CGCL101	S. lividans TK23 harboring pCGC309 (gusA reporter cassette, no promoter)	This work
CGCL102	S. lividans TK23 harboring pCGC310 (gusA under the control of Pcgc1)	This work
CGCL103	S. lividans TK23 harboring pCGC311 (gusA under the control of Pcgc20)	This work
CGCL104	S. lividans CGCL107 harboring pCGC309 (gusA reporter cassette, no promoter)	This work
CGCL105	S. lividans CGCL107 harboring pCGC310 (gusA under the control of Pcgc1)	This work
CGCL106	S. lividans CGCL107 harboring pCGC311 (gusA under the control of Pcgc20)	This work
CGCL107	S. lividans TK23 harboring pCGC090 (expressing cgc20 and cgc21 under their own promoter, Pcgc20)	This work
Plasmids
pGEMT-easy	E. coli cloning vector	Promega
pOSV010	Source of the Ωhyg-oriT fragment	(48)
pOSV221	Source of the att2Ωaac cassette	(16)
pOSV230	Source of the att1Ωaac cassette	(21)
pOSV232	Source of the att2aac cassette	(21)
pOSV235	Same as pOSV236 (16) with the pac-oriT insert in the opposite orientation	(21)
pOSV408	E. coli/Streptomyces shuttle vector integrative in Streptomyces	(8)
pOSV554	Streptomyces expression vector	(16)
pUC19	E. coli cloning vector	(58)
pOSInt3	Plasmid bearing the pSAM2 xis/int genes, used to excise the att2aac cassette	(59)
pGUS	pSET152 derivate containing omega addA cassette and gusA	(23)
pOSV751	pGUS derivative with a NotI-cat-NdeI gene cassette cloned upstream of gusA and a Tfd terminator downstream of gusA	This work
pOSV808	Conjugative vector containing hygromycin resistance, VWB integrase and amilCP	(55)
pCGC001	pBeloBAC11 containing a 43.4 kb S. ambofaciens DNA fragment carrying the entire cgc gene cluster	(7)
pCGC002	Integrative and conjugative BAC carrying the complete cgc gene cluster	(4)
pCGC006	Ωhyg-oriT HindIII/XbaI/Klenow fragment from pOSV010 cloned in pCGC001/HpaI	This work
pCGC041	pUC19 containing the 7.3 kb (BstBI/BsiBW) fragment from pCGC001	This work
pCGC043	pCGC041-cgc1 :: att2aac	This work
pCGC044	pCGC041-cgc2 :: att2Ωaac	This work
pCGC051	pCGC006-cgc1 :: att2aac	This work
pCGC052	pCGC006-cgc2 :: att2Ωaac	This work
pCGC071	cgc1 cloned into pGEM-T Easy	This work
pCGC079	pCGC001 containing neo-oriT from pOSV408	This work
pCGC081	cgc1 under the control or ermE*p cloned in pOSV554	This work
pCGC084	pCGC079-cgc3 :: att1Ωaac	This work
pCGC085	pCGC079-cgc6 :: att1Ωaac	This work
pCGC086	pCGC079-cgc7 :: att1Ωaac	This work
pCGC088	pCGC079-cgc18 :: att1Ωaac	This work
pCGC089	pCGC079-cgc4 :: att1Ωaac	This work
pCGC090	pSET152 derivative with cgc20 and cgc21 under the control of their native promoter Pcgc20	This work
pCGC306	pOSV751 derivative with a deletion of the cat cassette	This work
pCGC307	pOSV751 derivative with gusA under the control of Pcgc1	This work
pCGC308	pOSV751 derivative with gusA under the control of Pcgc20	This work
pCGC309	pOSV808 derivative, amilCP replaced by gusA without promoter sequence	This work
pCGC310	pOSV808 derivative, amilCP replaced by Pcgc1_gusA	This work
pCGC311	pOSV808 derivative, amilCP replaced by Pcgc20_gusA	This work
pCGC312	pCGC002-cgc1:: att2aac	This work
pCGC313	pCGC002-cgc1:: att2	This work

FIG 7.

GusA activity in S. lividans strains with gusA expression under the control of P_cgc1 and P_cgc20 promoters in the presence or absence of congocidine. Results correspond to the mean values (± standard errors) of four independent experiments. Statistical differences were determined by using a Welch Two Sample t-tests (one symbol: P value < 0.05, two symbols: P value < 0.01). Black asterisks correspond to statistical differences in the GusA activity of reporter strains compared to the control strain grown in the same condition, blue asterisks indicate statistical differences in the GusA activity of a given strain after growth in the absence versus in the presence of congocidine.

The cgc20/cgc21 genes code for the ABC transporter responsible for the export of and resistance to congocidine. Therefore, we introduced the plasmids pCGC309, pCGC310 and pCGC311 described above in S. lividans strains expressing the cgc20/cgc21 genes under the control of P_cgc20. Similar to what was observed in strains not harboring the cgc20/cgc21 genes, addition of 2.3 μM and 11.6 μM congocidine resulted in an increase of GusA activity (Fig. S5). However, the levels of GusA activity were lower than those obtained from strains not expressing the cgc20/cgc21 genes. This reduced activity reflects a weaker transcription from the P_cgc20 and P_cgc1promoters, consistent with a reduction in the intracellular concentration of congocidine, under the action of the ABC transporter Cgc20-Cgc21.

The cgc20-cgc1 intergenic region is rich in strong congocidine binding sites.

Congocidine is a minor groove DNA binder that binds at A/T-rich regions, more precisely a succession of four or more A/T bases (24). Its binding into DNA minor groove is known to perturb DNA topology locally, narrowing the minor groove, bending the DNA or introducing negative supercoiling, in a sequence-dependent manner (25, 26). Thus, we considered the possibility that the transcription of cgc20-cgc22 and cgc1 in the presence of congocidine could be a direct consequence of congocidine DNA binding in the cgc20-cgc1 intergenic region. This region (740 nt) contains 11 sequences of four or more consecutive A/T bases (five of four A/T bases, four of five A/T bases, one of six A/T bases and one of nine A/T bases). Studies have shown that the strength of congocidine binding to (A/T)₄ motifs depends on the sequence of the motif. In particular, 5′-AAAA-3′, 5′-TTTT-3′ and 5′-AATT-3′ have been identified as strong congocidine binding motifs (27, 28). On the contrary, motifs that contain the dinucleotide 5′-TA-3′ sequence have significantly lower binding constants. There are eight non-overlapping strong congocidine binding motifs in the cgc20-cgc1 intergenic region (Fig. S6 in Supplemental Materials). When we searched for non-overlapping strong congocidine binding motifs in all S. ambofaciens ATCC23877 intergenic regions of more than 4 bp, we found two intergenic regions, both duplicated as they are located in the terminal inverted repeats, much larger than the cgc20-cgc1 intergenic region (7300 nt and 2268 nt versus 740 nt) that contain more motifs than the cgc20-cgc1 intergenic region (Fig. S7 in Supplemental Materials). The occurrence of multiple strong congocidine binding motifs in the cgc20-cgc1 bidirectional promoter region may therefore explain the effect of congocidine on the transcription of the cgc20-cgc22 and cgc1 genes in the absence of Cgc1, possibly by modifying the topology of the DNA.

The expression of the cgc20 and cgc21 resistance genes is also controlled by Cgc1.

The induction of the expression of the cgc20-cgc22 operon by congocidine raised the question of the control of cgc20-cgc22 expression by Cgc1. Indeed, in the absence of cgc1, the lack of expression of cgc20-cgc22 could be due to an absence of congocidine. However, we noticed that the E. coli ET12567/pUZ8002 strain used for intergeneric conjugation became resistant to 2 μg/ml (4.6 μM) congocidine after the introduction of the BAC pCGC002 containing the complete cgc gene cluster, while not producing congocidine (data not shown). This prompted us to compare the growth of three E. coli strains: ET12567/pUZ8002 (control strain), ET12567/pUZ8002/pCGC0002 containing the complete cgc cluster and ET12567/pUZ8002/pCGC313 containing the cgc gene cluster deleted for cgc1. Results presented in Fig. 8 show that the strain containing the complete cgc gene cluster is much more resistant to congocidine than the control strain, showing that cgc20-cgc21 confer congocidine resistance to E. coli. The strain ET12567/pUZ8002/pCGC313 containing the cluster deleted for cgc1 presents an intermediate level of congocidine resistance. Thus, the induction of the expression of the cgc20-cgc21 genes by congocidine observed in Streptomyces also occurs in E. coli and confers congocidine resistance at low concentrations (5 μg/ml, 11.6 μM). However, Cgc1 is required to obtain higher levels of resistance (10 μg/ml, 23.2 μM) in E. coli. This demonstrates that Cgc1 also controls the expression of the cgc20-cgc21 resistance genes.

FIG 8.

Growth of E.coli ET1267/pUZ8002 strains harboring the complete or cgc1-deleted congocidine biosynthetic gene cluster in presence or absence of congocidine. Black triangles: ET12567/pUZ8002/pCGC002 (complete cgc gene cluster); dark gray squares: ET12567/pUZ8002/pCGC313 (cgc gene cluster deleted for cgc1); light gray disks: ET12567/pUZ8002 (control). Results correspond to the mean values (± standard errors) of at least 4 independent experiments performed in triplicates.

DISCUSSION

In this study, we aimed at characterizing the transcriptional regulation of the genes constituting the congocidine biosynthetic gene cluster. The sequence analysis of Cgc1, the predicted transcriptional regulator encoded within the cgc gene cluster, indicated that Cgc1 belongs to the NarL family of response regulators (RR). Cgc1 is an orphan response regulator, as cgc1 is not clustered with a gene encoding a histidine kinase. In addition, the receiver domain of Cgc1 lacks most of the residues constituting the phosphorylation pocket. In particular, the phosphorylatable aspartate residue is replaced by a histidine. Thus, in addition to being an orphan RR, Cgc1 is also an atypical RR. Several of such regulators have been studied and characterized, notably in Streptomyces species, for which some of them are involved in specialized metabolism regulation (29). Thus, JadR1, the jadomycin pathway-specific transcriptional activator, and RedZ, the undecylprodiginin pathway-specific transcriptional activator, lack most of the conserved amino acids of the phosphorylation pocket (30, 31). It has notably been reported that JadR1 is not phosphorylated when incubated in the presence of small phosphodonor molecule such as phosphoramidate (31).

The deletion of cgc1 in the reference strain S. ambofaciens SPM110 (not producing spiramycin) and the genetic complementation of the mutant strain carried out in this study demonstrate that similarly to JadR1 and RedZ, Cgc1 activates the transcription of the cgc genes. Determination of the transcriptional organization of the cgc gene cluster showed that this cluster is constituted of eight main transcriptional units. Upstream of four of these transcriptional units, the intergenic regions are extremely small or inexistent (upstream of cgc4, cgc6, cgc7 and cgc19), suggesting that in these situations the promoter and the Cgc1 operator sites are located within the preceding coding sequence. Using MEME (32) to search for potential motifs for Cgc1 DNA binding did not allow to identify any convincing motif. It is possible that as observed for JadR1, the Cgc1 recognition sequence is poorly conserved (31).

Analysis of the transcriptome of S. ambofaciens ATCC23877 carried out in the congocidine production medium MP5 showed that during the exponential growth phase, cgc1 is the only cgc gene expressed, whereas all cgc genes are induced in a statistically significant manner in stationary phase. The presence of transcripts of cgc1 but not of biosynthetic/resistance genes in exponential phase raised the question of the existence of another level of regulation, possibly at the translational level. Indeed, the cgc1 sequence contains two TTA codons separated by two codons. TTA codons are rarely used in Streptomyces, which have GC-rich genomes (around 72%), but they are often found in genes encoding cluster-situated transcriptional activators (33, 34). In several Streptomyces species, it has been proposed that the presence of a TTA codon in a gene could constitute a temporal regulation mechanism, as bldA, the only tRNA that recognizes the UUA codon, is more abundant during the stationary phase than during the growth phase (33, 34). In S. ambofaciens, however, the abundance of bldA (SAM23877_RS14785) transcript is constantly decreasing during metabolic differentiation (transcriptomes analyzed in MP5 and YEME10 media at 24h, 30h, 36h, 48h and 72h (see Lioy et al. [22], Table S5), indicating that bldA transcript abundance cannot explain the delay between the transcription of cgc1 and that of the other cgc genes. Yet, the bldA tRNA may still be involved in some translational regulation of Cgc1 production as posttranscriptional modifications of tRNAs are often needed for them to be fully functional (35, 36).

During the course of our study, we noticed that a cgc1 deletion mutant is resistant to congocidine, while a strain deleted of the entire cgc gene cluster is as sensitive to congocidine than a congocidine nonproducing Streptomyces strain (7). Analysis of the expression of cgc genes in the cgc1 deletion mutant showed that the cgc20-cgc22 operon is expressed in the presence of congocidine, even in the absence of Cgc1. This operon contains the genes coding for the ABC transporter responsible for resistance to congocidine (7). Repeating this experiment in a heterologous host context (S. lividans TK23) yielded the same observation, the induction of the expression of the cgc20-cgc22 operon in the presence of congocidine and absence of Cgc1. Furthermore, the analysis of S. ambofaciens ATCC23877 transcriptome in the presence or absence of low levels of congocidine (1 μg/ml, 2.3 μM) in a condition of limited congocidine production showed a modulation of the expression of only six genes, all but one related to the cgc gene cluster (Fig. 6). The expression of the cgc20-cgc22 operon and to small extent of cgc2, already observed in the cgc1 deletion mutant (Fig. 4) was induced. In addition, the induction of the expression of the transcriptional activator Cgc1 was observed. Altogether, this data suggests that congocidine exerts a very specific effect on the cgc20-cgc1 intergenic region, which contains the cgc1 and cgc20 promoters. To further explore this effect, we expressed the gusA reporter gene under the control of the P_cgc1 and P_cgc20 promoters, in presence or absence of the resistance genes cgc20-cgc21. In both cases, an increase of GusA activity was observed in the presence of congocidine, albeit at a lower level when the cgc20-cgc21 resistance genes were present (Fig. 7, Fig. S5). This confirmed the capacity of congocidine to induce the expression of genes under the control of the P_cgc1 and P_cgc20 promoters. The lower GusA activity observed in the presence of cgc20-cgc21 may be due to the partial exclusion of congocidine from the cells, comforting their proposed role in congocidine efflux (7).

Induction of resistance mechanisms by the product of a gene cluster is not unprecedented (3, 37–39). For example, in the actinorhodin gene cluster, the TetR family member ActR controls the expression of the actinorhodin export pumps ActA/ActB. The repression of actA/actB expression by ActR is relieved by the binding of actinorhodin or some of its late biosynthetic intermediates to ActR (39). Similarly, the repression of the expression of the SimX efflux pump by SimR (TetR family) is lifted by the binding of simocyclinone D8 and its biosynthetic intermediate simocyclinone C4 (3). In these two cases, ActR and SimR specifically control the expression of the resistance genes, while other(s) transcriptional regulator(s) control the transcription of the biosynthetic genes (ActII-orf4 in the actinorhodin cluster (4) and SimReg1 in the simocyclinone D8 cluster (5)). In the cgc gene cluster, Cgc1 is the sole transcriptional regulator encoded within the cluster. The experiments presented in this study show that it controls both, the expression of the biosynthetic as well as resistance genes. In contrast to what has been reported for actinorhodin and simocyclinone D8, the induction of the congocidine resistance genes does not seem to rely exclusively on a transcriptional regulator and may involve a direct binding of congocidine to the P_cgc20 promoter. Indeed, the induction of the resistance genes by congocidine occurs in the absence of cgc1 and in heterologous hosts as distantly related as E. coli. Furthermore, the cgc20-cgc1 intergenic region, which contains the promoters for the cgc20-cgc22 operon and for cgc1, is rich in congocidine strong binding sites (Fig. S6). Thus, the binding of congocidine into the DNA minor groove could perturb DNA topology locally and favor the transcription of cgc20-cgc22 and of cgc1. In fact, it has been shown that congocidine can activate transcription initiation in E. coli from promoters that have been engineered to include a congocidine binding site in the spacer sequence between the −35 and −10 regions (40). Interestingly, to our knowledge, only one other DNA-binding metabolite, daunorubicin (intercaling agent, binding to GC-rich motifs), has been suggested to be directly involved in the regulation of its own biosynthesis. In this case, binding of daunorubicin to DNA is proposed to exert negative feedback on its production by inhibiting the transcription of two of the three transcriptional activators of the regulatory cascade leading to daunorubicin biosynthesis (41, 42). In addition to the direct effect of congocidine on the cgc20-cgc1 region, our results show that Cgc1 is functional in E. coli, which is phylogenetically distant from S. ambofaciens. This suggests that the Cgc1 dependent or independent regulation of the divergent cgc20-cgc1 promoters does not rely on cis or trans elements other than the intergenic region of cgc20-cgc1, the congocidine and/or Cgc1.

Altogether, the data obtained in this study lead us to propose a model for the regulation of the cgc gene cluster (Fig. 9). During exponential growth, before the onset of congocidine production, the regulatory gene cgc1 is the only cgc gene transcribed at significant but relatively low level. At this stage, the amount of Cgc1 may not be sufficient for the efficient activation of the transcription of the cgc genes and/or additional levels of regulation may be required. Later, in response to signal(s) yet to be determined, metabolic differentiation occurs and Cgc1 activates the transcription of the cgc genes, leading to congocidine production and efflux. Results from experiments performed with S. ambofaciens ATCC23877 or with heterologous hosts demonstrate that congocidine induces the expression of the resistance genes cgc20-cgc21 and of the pathway-specific transcriptional activator gene, cgc1. This induction can occur in the absence of Cgc1, probably by the direct binding of congocidine to the promoter region of the cgc20-cgc22 operon and of cgc1, rich in strong congocidine binding sites. In a cell population, in which the onset of congocidine production is not synchronized, such induction will first offer protection against congocidine to cells that are not yet producing it. It will also increase the synthesis of Cgc1, first step toward the biosynthesis of congocidine, thus affording the concerted production of this metabolite among the cell population. Such feed-forward mechanism, in which a metabolite is inducing its own production, has already been observed in Actinobacteria (planosporicin, Planomonospora alba) (43) or in Bacilli (nisin, Lactococcus lactis; subtilin, Bacillus subtilis) (44). It has been proposed that such concerted biosynthesis may be required to reach effective levels of antibiotics. In B. subtilis and in L. lactis, induction of the gene clusters by subtilin and nisin, respectively, is mediated through a two-component system. In P. alba, planosporicin has been proposed to interact with PspW, an anti-σ factor, releasing the ECF-σ factor that controls the transcription of the biosynthetic genes (43). The mechanism by which congocidine induces its own biosynthesis constitutes a new type of feed-forward induction in which the metabolite itself induces the transcription of the pathway-specific transcriptional regulator.

FIG 9.

Proposed model for the transcriptional regulation of the cgc gene cluster.

MATERIALS AND METHODS

Bacterial strains, plasmids, and growth conditions.

The strains and plasmids used in this study are listed in Table 1. Escherichia coli strains were grown in LB medium with appropriate antibiotics when necessary (ampicillin 50 μg/ml; apramycin 50 μg/ml; chloramphenicol 35 μg/ml; hygromycin 100 μg/ml). Streptomyces strains were grown at 28°C on solid medium SFM (Soya Flour Mannitol) for genetic manipulations and spore stock preparations (45). They were grown in liquid MP5 medium (46) for the analysis of congocidine production and RNA extraction. For RNA-seq analysis, S. ambofaciens ATCC23877 strain was grown in MP5 or YEME10 medium 10.3% saccharose (3 g/liter yeast extract, 5 g/liter bactotryptone, 3 g/liter malt extract, 10 g/liter glucose, 103 g/liter saccharose; pH 7.0–7.2) with or without 2.3 μM (1 μg/ml) congocidine at 30°C in a shaking agitator (220 rpm, INFORS HT multitron standard) as previously described in (22).

Preparation and DNA manipulations.

E. coli transformations and E. coli/Streptomyces conjugations were performed under standard conditions (45, 47). High-fidelity DNA polymerase Phusion (ThermoFisher) was used to amplify DNA cassettes and cgc1, and Taq polymerase (Qiagen) was used for PCR verification of gene deletions in BACs or Streptomyces strains. DNA fragments and PCR products were purified using the Nucleospin Gel and PCR cleanup kit from Macherey-Nagel. All oligonucleotides used in this work are listed in Table 2.

TABLE 2.

Oligonucleotides used in this study

Name	Sequence	Description
cgc(amb)1F	AAGCTTGAGAAGGGAGCGGACATATGAGTCAGACGCTGTGCCA	Amplification cgc1, HindIII site underlined
cgc(amb)1R	AGGCCTTCACGCGGCCGCGTGGTGGTGGTGGTGGTGCGCGGCCGCTCAGCAGGTGATGAGACCGT	Amplification cgc1, StuI site underlined
cmj4F	GGGCGCACGGACGGGGACGTCCCCCGGCCGCTGAGGCCCGATCTACCTCTTCGTCCCGAAGCAACT	cgc1 in frame deletion or replacement
cmj17	GGGGGGCCAGAGGCGCAAGGCGGCAGTGAGGAGGAGTACGATCGGCGCGCTTCGTTCGGGACGAA
cmj12F	CGAGACGGACGGCACGGTGAC	Verification cgc1 deletion
cmj12R	TGGCAACTCGGCCGCACTCG
cmj13F	GTGCCCTCGGCCGTGTCAGCCCACCCGGGGACGAGGCCCGATCTACCTCTTCGTCCCGAAGCAACT	cgc2 in frame deletion or replacement
cmj13R	TACGGTGAGCGGGGCGTGACAGCCCCGGAGAGGAGCGAGCATCGGCGCGCTTCGTTCGGGACGAAG
cmj14F	TCGGAGCCTTCGGCCACAG	Verification cgc2 deletion
cmj14R	CGGGCAGGGCCATATGTACC
cgc3-dis-T-F	CGAGTTCCGTCTCGACCGCGTGCCGCAGCGAACGGGGTGATCTACCTCTTCGTCCCGAAG	Replacement of cgc3
cgc3-dis-T-R	CGCCGGCCACCGTGGTGTCCCTCATGACATCTCCCGATCATCGCGCGCTTCGTTCGGGAC
cgc3-SC-FOR	CGGCGTCGATGATGTCCC	Verification of cgc3 replacement
cgc3-SC-REV	ACCGCTGACCGTCCAC
cgc4-dis-T-F	GCCGAGTTCGACGTGGGTGCCGTCGGTGCGCATCAGCTATCTACCTCTTCGTCCCGAAG	Replacement of cgc4
cgc4-dis-T-R	GGACGGAGGGCCTCGACATGACCGGAGTCAGTGTCTTCATCGCGCGCTTCGTTCGGGAC
cgc4-SC-FOR	GGCGTGACGACGACGATGGG	Verification of cgc4 replacement
cgc4-SC-REV	GAGGTCGTGGGTTCAAATCC
cgc6-dis-T-F	GCTCCGCGCGCACGACCTCGGGGCGCAGGTACATCATCAATCTACCTCTTCGTCCCGAAG	Replacement of cgc6
cgc6-dis-T-R	GACCTGGTCGGACGCAGTGTCAAGAGGTCATGAGGTGAGATCGCGCGCTTCGTTCGGGAC
cgc6-SC-FOR	GACGGGATGCCGGTAGACAC	Verification of cgc6 replacement
cgc6-SC-REV	CACTGCACCGGGAGATGCAC
cgc7-dis-T-F2	CTCGGGAAAGCGCATGGTCATCGTGCAGTAGAGCGAGGCATCTACCTCTTCGTCCCGAAG	Replacement of cgc7
cgc7-dis-T-R2	ACTCGTGCACCGCGCGCCTTCGACCAGAGGAGAACCCATGATCGCGCGCTTCGTTCGGGA
cgc7-SC-FOR2	CCGGTGGGATCCTTCAGGAC	Verification of cgc7 replacement
cgc7-SC-REV	GGGAGCGCCTGTTCTGGCAC
cgc18-dis-T-F	CCGCGTCCTCCTCGAGCATCGCGGTGGCGTCGAACTTGGATCTACCTCTTCGTCCCGAAG	Replacement of cgc18
cgc18-dis-T-R	GCGGCCGGCAAGGTGGTGCTGATCCCGTGAGCATCTCCAATCGCGCGCTTCGTTCGGGAC
cgc18-SC-FOR	CTCGAGGACACCCGAGAAGC	Verification of cgc18 replacement
cgc18-SC-REV	CGTTCCGGTTGTTCGACTCC
cmj31F	CCGACGGCCCAAGTTACTGC	RT-PCR cgc1, 307 bp
cmj31R	CCGGGCGATCTGTTTGTTGC
cmj32F	TCCGTCGCCGGTTGAAGTAG	RT-PCR cgc20, 302 bp
cmj32R	GCCGTACATCGCCCTCCATC
cgc22F	TCGTGTGCCGCGACGAACAG	RT-PCR cgc22, 311 bp
cgc22R	CTCGACGACCTGCTCGACAG
cmj34F	GTTTCGCACCGTGTTCCGTTC	RT-PCR cgc2, 326 bp
cmj34R	TGCAGCACGCTCTCCTTGTC
cmj35F	GATGGGCGACACCTTCAGGG	RT-PCR cgc3, 317 bp
cmj35R	GGTGCACGTAGACGCGCTTG
cmj36F	GACGAACCATGCGGACAACC	RT-PCR cgc19, 303 bp
cmj36R	GGAGTGGCGTTCTCGAAGAGAC
cmj40F	GACGACGGTGTGCTGGAGTTC	RT-PCR cgc18, 302 bp
cmj40R	TCGACCTTGCCGTTGGGATTC
cmj43F	GCGCATGGTCATCGTGCAGTAG	RT-PCR cgc7, 333 bp
cmj43R	TTCCGCCGGTTCGAGAACTTCC
cmj62F	GAAGGAGGCGATGCTGGTGAG	RT-PCR cgc21, 312 bp
cmj62R	GCCGAGGTCAACGGCTATGTG
cmj63F	CTCCGCTCATGTCGCCGAGAAG	RT-PCR cgc4, 292 bp
cmj63R	GGCCAGGTCGAAGACCGTCAC
cmj64F	CACCGGCATGCTCAACAGCAC	RT-PCR cgc5, 296 bp
cmj64R	CGGCCAGCTTGACGCTGAAC
cmj65F	CGTGATCAACGGGCACTACGAG	RT-PCR cgc6, 316 bp
cmj65R	CACGCCCTGGTTGGAGATCTTC
cmj66F	GACGCCCGGATCCTGCTCTC	RT-PCR cgc8, 313 bp
cmj66R	GGACCCGCCAGGTGTCGTAG
cmj67F	CCACCTCCTCGACTGGCTCTC	RT-PCR cgc9, 301 bp
cmj67R	CTCGACGAACTGCGGGATCAC
cmj68F	GTGAAGGTCCAGCCGTTCCC	RT-PCR cgc10, 296 bp
cmj68R	GGTCCCTGGCCGATGATGTG
cmj69F	CCTGTGGTCCCACCACAAGAAG	RT-PCR cgc11, 285 bp
cmj69R	CAGTCGCCCTCGATGACGTAG
cmj70F	TGGCCCTGATCGAGGACTGC	RT-PCR cgc12, 306 bp
cmj70R	CGAGCTGGACACGTCCGATG
cmj55F	CGTCTTCTGGGCCGACTTTG	RT-PCR cgc13, 300 bp
cmj55R	GAGTCCGCGTGGATGATCTC
cmj71F	GGGACTGAGCGGACTGAAGAC	RT-PCR cgc14, 312 bp
cmj71R	GGCTGGTACGAGCCGAAGATG
cgc15F	ACGTCGCCGTCCTCAGCAAC	RT-PCR cgc15, 308 bp
cgc15R	CGACTTGACGCGGGAGAATC	RT-PCR cgc15, 308 bp
cgc16F	GGGAACTGGAACGGCTCTAC	RT-PCR cgc16, 319 bp
ccg16R	GCGAGCAGCAGCATGAACAC
cmj74F	CGTGGTCCTGCTCCCGATCC	RT-PCR cgc17, 298 bp
cmj74R	GGTCAGGTCGGCGATCTCCC
cAV08	AGGCCCACGACTTCATCACG	RT-PCR cgc22-21-20, 2412 bp
cAV09	CGCACCCTTGCGTGCAGTTC
cAV010	GGCGGACGCAATCTGTCCAC	RT-PCR cgc22-cgc21, 1391 bp
cAV11	GCCTCCGTGCAGCCGTAGAC
cgc1_dis_F	TCCCAGTCACGACGTTGTAAAACGACGGCCAGTGCCAAGGCGAGGCAAGCTTATCGATG	Replacement of cgc1
cgc1_dis_R	GCAAGGCGGCAGTGAGGAGGAGTACGAGTGAGTCAGACGCACGAGGCCCTTTCGTCTTC
cgc1_SC_F	AGAAGCTGGGCGAACAAACG	Verification of cgc1 replacement
cgc1_SC_R	GCGGATGTTGCGATTACTTC
SBM21	CCAAGAACAGCATCGACGAG	RT-qPCR, rpoB, 234 bp
SBM22	TGTCGATGGAGGAGTCGAAG
AspS_RT_F1	CTGCTGATGATCTCGGGCTT	RT-qPCR, aspS, 104 bp
AspS_RT_R1	TCGAGCTGGTAGAACTCGCC
GyrA_RT_F1	GGCGACTCCTCCATCTACGA	RT-qPCR, gyrA, 166 bp
GyrA_RT_R1	GGACCATCTCCATCGACAGC
GyrB_RT_F1	TCGAGACCACCGACTACTCCTT	RT-qPCR, gyrB, 163 bp
GyrB_RT_R1	CTCTTGACCTCGTGCTTCTCG
cgc2RT-F3	ACGGCACCCTCGTCCAA	RT-qPCR, cgc2, 66 bp
cgc2RT-R3	GACCGATGGCCCACACTTC
cgc3RT-F3	GATCGCGCCCTCCAACTAC	RT-qPCR, cgc3, 125 bp
cgc3RT-R3	GAAGGTGTCGCCCATCATCA
cgc4RT-F1	GCCTCGACATGACCGGAGT	RT-qPCR, cgc4, 125 bp
cgc4RT-R1	GCGACATGAGCGGAGAAGAC
cgc6RT-F2	TGATCAACGGGCACTACGAGA	RT-qPCR, cgc6, 115 bp
cgc6RT-R2	ACCAGGCTCCACCAGCTGAA
cgc7_RT F1	TCCCCATGAACGCGATGAGT	RT-qPCR, cgc7, 113 bp
cgc7_RT R1	CGAACGACCCGTCCTTGAAT
cgc19RT-F2	CGCGGTGACTTCTTCGAACT	RT-qPCR, cgc19, 77 bp
cgc19RT-R2	ATCCTGCCTATGGCTTGCAG
cgc20RT-F4	CAGGTCTTCGCCAGTCTGCTC	RT-qPCR, cgc20, 110 bp
cgc20RT-R4	ACCATCAGGTTCAGGGACTCG
cgc21_RT F1	GCCTGGCTGTTCACCTACCTC	RT-qPCR, cgc21, 122 bp
cgc21_RT R1	GTGTCGTGGAACGTCATGTCC
cgc22RT-F1	TACGGGGAACTGCACGCAAG	RT-qPCR, cgc22, 245 bp
cgc22RT-R2	CGGAGTCTCGCCTCGTGTG
LS-RP1	CATATGTCGTACTCCTCCTCACTGC	Amplification of the cgc20-cgc1 intergenic region, NdeI site underlined
LS-RP2	GCGGCCGCGGCACTCCTTCTCT	Amplification of the cgc20-cgc1 intergenic region, NotI site underlined
LS-RP3	CATATGCGGCACTCCTTCTCTGG	Amplification of the cgc1-cgc20 intergenic region, NdeI site underlined
LS-RP4	GCGGCCGCTCGTACTCCTCCTCA	Amplification of the cgc1-cgc20 intergenic region, NotI site underlined
LS04	TGCCACAGGGATGCCACAAC	Verification of the CGC101-CGCL106 strains, used with CEA31
LS25	GATGTTACCCGAGAGCTTGG	Verification of the CGC101-CGCL106 strains, used with LS26 or Cmj12R
LS26	GCTGGTCGATACCGCAGTTC	Verification of the CGC101-CGCL106 strains, used with LS25 or Cmj12R
LS63	CATCCGGGTACGACAACGGC	Verification of the CGC101-CGCL106 strains, used with CEA29
CEA29	AAACGAGAACGGGCTCCACTG	Verification of the CGC101-CGCL106 strains, used with LS04
CEA31	ACGCTTGGACAGCACGAGAC	Verification of the CGC101-CGCL106 strains, used with LS63

Construction of the cgc1 and cgc2 mutants in S. ambofaciens SPM110.

To facilitate the construction of the cgc1 and cgc2 deletion mutants, the pCGC001 BAC (7) was engineered to allow conjugational transfer from E. coli ET12567/pUZ8002 or S17.1. A 3 kb HindIII/XbaI/Klenow fragment from pOSV010 (48), containing the hygromycin resistance gene and the RK2 origin of transfer was cloned into the HpaI-digested pCGC001 BAC, yielding pCGC006.

Attempts to replace cgc1 and cgc2 by the att2aac or att2Ωaac cassettes (16, 21) directly in pCGC006 by PCR-targeting failed. Therefore, a 7.3 kb BstBI/BsiWI/Klenow fragment from pCGC001, containing cgc1 and cgc2 was cloned into a SmaI digested pUC19, yielding pCGC041 (Fig. S8). The cgc1 and cgc2 genes were replaced by the att2aac (from pOSV232) and the att2Ωaac (from pOSV221) cassettes, respectively, by PCR-targeting (49), using the oligonucleotide pairs cmj4F/cmj17 and cmj13F/cmj13R. This resulted in plasmids pCGC043 and pCGC044. The cgc1 gene was then replaced by the att2aac cassette in pCGC006 by λ-RED recombination using the 8.5 kb PvuII fragment from pCGC043, yielding pCGC051 (Fig. S8). The same protocol was followed to replace cgc2 with the att2Ωaac cassette (using pCGC044), resulting in pCGC052.

The BACs pCGC051 and pCGC052 were introduced into S. ambofaciens SPM110 by conjugation from E. coli S17.1 and exconjugants were selected for with apramycin. Conjugants were screened for sensitivity to hygromycin, indicating a double-crossover allelic exchange. Replacements of cgc1 and cgc2 by the apramycin resistance cassettes in S. ambofaciens SPM110 were verified by PCR using the primer pairs cmj12F/cmj12R and cmj14F/cmj14R, respectively, and the mutant strains were named CGCA013 and CGCA017, respectively. To excise the att2aac cassette in CGCA013, pOSV235 (50) was introduced into CGCA013 by conjugation from E. coli S17.1 and conjugants were selected for with thiostrepton. Mutant strains for which the att2aac cassette had been excised by site specific recombination were screened for apramycin sensitivity. After the loss of the unstable pOSV235 plasmid, in-frame deletion of cgc1 was verified by PCR using the primer pair cmj12F/cmj12R and sequencing of the PCR product and the strain was named CGCA021.

Genetic complementation of the cgc1 mutant.

To verify that the abolition of the transcription of the cgc genes in CGCA021 was due to the deletion of cgc1, a plasmid expressing cgc1 under the control of the strong constitutive promoter P_ermE* was constructed. The cgc1 gene was amplified using the primers cgc(amb)1F (containing a HindIII site) and cgc(amb)1R (containing a StuI site). The PCR product was purified, incubated 10 min at 72°C with the Taq polymerase and cloned into the pGEMT-Easy vector, yielding pCGC071. The insert was verified by DNA sequencing. The 0.75 kb HindIII/StuI fragment from pCGC071 was then cloned into the conjugative and integrative vector pOSV554 (16) digested by the same enzymes. The resulting plasmid, pCGC081, was introduced by conjugation from E. coli S17.1 into the S. ambofaciens mutant strain deleted for cgc1, CGCA021. Conjugants were selected on SFM containing apramycin and verified by PCR using the primers cgc(amb)1F/R. The resulting strain was called CGCA031.

Construction of the cgc3, cgc4, cgc6, cgc7 and cgc18 replacement mutants.

To study the transcriptional organization of the cgc cluster, we replaced several cgc genes by the att1Ωaac cassette (containing an apramycin resistance gene flanked by transcription terminators [21]). The pCGC006 BAC contains a hygromycin gene whose promoter is identical to the promoter of the apramycin resistance gene in the att1Ωaac cassette. Consequently, the BAC pCGC006 could not be used to construct the mutants by PCR targeting. Thus, the pCGC001 BAC was reengineered to allow conjugational transfer from E. coli using pOSV408 (8) as the source of a cassette containing a kanamycin resistance gene and the RK2 origin of transfer. The cassette was amplified by PCR and integrated in pCGC001 by PCR targeting (51), yielding the BAC pCGC079.

The same strategy was used to replace each of the cgc3, cgc4, cgc6, cgc7 and cgc18 genes by the att1Ωaac cassette by PCR-targeting in pCGC079 (see Table S1 for the names of the strains and plasmids and Table S2 for the oligonucleotides used). Briefly, for the replacement of cgc3, the att1Ωaac cassette apramycin was amplified from pOSV230 using the cgc3-dis-T-F/R primers and used for the replacement of cgc3 in pCGC079, yielding pCGC084. Correct replacement of cgc3 was verified by PCR using the primers cgc3-SC-F/R and pCGC084 was subsequently introduced into S. ambofaciens SMP110 by conjugal transfer via S17.1. Conjugants were selected for apramycin resistance and then screened for kanamycin sensitivity to isolate double cross-over recombinants. The resulting mutant strain was verified by PCR and named CGCA039. Replacement of cgc4, cgc6, cgc7 and cgc18 by the att1Ωaac cassette yielded the strains CGCA044, CGCA040, CGCA041 and CGCA043, respectively.

Construction of a S. lividans TK23 strain expressing the cgc gene cluster deleted for cgc1.

To study the expression of the resistance genes in a heterologous host and in a genetic context similar to the one in S. ambofaciens, a S. lividans TK23 strain expressing the cgc gene cluster deleted for cgc1 was constructed. The pCGC043 plasmid previously constructed was digested by PstI/EcoRI and the 8.5 kb fragment (containing cgc1 replaced by the att2aac cassette) was purified and used for the replacement of cgc1 by att2aac in pCGC002 by PCR targeting. The resulting BAC was verified by PCR and by enzymatic digestions and named pCGC312. The aac(3)IV gene was next excised using the pOSInt3 plasmid bearing the pSAM2 int and xis genes (21). The correct excision was verified by PCR and the resulting BAC named pCGC313. This BAC was introduced into S. lividans TK23 by intergeneric conjugation using the E. coli ET12567/pUZ8002 as the host strain. The resulting strain was named CGCL095.

RNA extraction from Streptomyces liquid culture and gene expression analysis by RT-PCR and RT-qPCR.

RNA was isolated from 2 ml of 43 h-old liquid cultures carried out in MP5 medium complemented with congocidine 5 μg/ml when necessary, using the NucleoSpin RNAII and NucleoSpin RNA/DNA Buffer Set kits (Macherey-Nagel). The RNA samples were treated with RNase-free DNase (Ambion or Qiagen) followed by purification using the NucleoSpin RNA Clean-up kit (Macherey-Nagel). Quality and quantity of RNAs were assessed by UV spectroscopy and agarose gel electrophoresis.

For RT-PCR analyses, the purified RNA samples were tested for potential DNA contamination by 27 cycles of PCR using primers for hrdB amplification with Taq polymerase (Qiagen). They were then subjected to reverse transcription by ThermoScript™ RT-PCR System (Invitrogen) 25°C for 10 min followed by 50°C for 50 min and 85°C for 5 min. 10% of the resulting cDNA was used as the template for 27 cycles of PCR amplification with appropriate primers (20- to 22-mers with average melting temperature from 65°C to 70°C) designed to amplify products of about 300 bp (see Table S2). The PCR amplifications were performed using the following program: 6 cycles at 94°C for 1 min, 55°C for 30 sec, and 72°C for 30 sec, followed by 21 cycles at 94°C for 1 min, 58°C for 30 sec, and 72°C for 30 sec. 1 The hrdB gene (encoding the major sigma factor) was used as a control, as it expressed at constant level (22). PCR products were analyzed by 1.2% agarose gel electrophoresis.

For RT-qPCR analyses, 200 ng of total RNA was reverse-transcribed in 20 μl final reaction volume using the Thermoscientific Revertaid Reverse Transcription kit (K1622) following the manufacturer's instructions. For each sample, negative reverse transcription reaction was done to verify the absence of genomic contamination in subsequent q-PCR. Primer sequences (see Table S2) were designed using NCBI primer-blast software online. SYBR green q-PCRs were performed using the Bio-Rad CFX Connect real-time PCR detection system in 96-well optical reaction plates. One μl of cDNA (env. 3.3 ng/reaction), standard or water (no-template control) were used as the template for q-PCRs with LightCycler(R)FastStart DNA Master SYBR green I kit (Roche) and primers at 500 nM final concentration. Real-time q-PCR amplifications were carried out (95°C for 10 min, followed by 6 cycles of 95°C for 10 sec, 55°C for 30 sec and 72°C for 30 sec, then 39 cycles of 95°C for 10 sec, 58°C for 30 sec and 72°C for 30 sec and a final dissociation curve analysis step from 65°C to 95°C). Technical triplicate experiments were performed for each biological sample (3 independent clones). The amplification efficiency of each probe was generated using the slopes of the standard curves obtained by 10-fold dilution series of either genomic DNA or cDNA mix. The efficiency of the q-PCR amplifications for all the genes tested was higher than 90% (except for cgc19 for which the efficiency was at 80%). Amplification specificity for each q-PCR was confirmed by the dissociation curve analysis and by gel electrophoresis. Determined Ct values were then exploited for further analysis (52). The values were normalized on four reference genes (aspS, rpoB, gyrA, gyrB), the mean value being used to calculate the fold change and performed statistical analyses of the results. Data were analyzed with R software (53) and statistical significance was assessed by means of Welch two sample t tests. Asterisks depict statistical significance (*P < 0.05, **P < 0.01, ***P < 0.001).

Bioinformatic analyses and code availability.

We previously described RNA extraction and sequencing from S. ambofaciens ATCC 23877 strain grown in MP5 and YEME10 media supplemented or not with congocidine (22). Accordingly we analyzed the data deposited in the NCBI Gene Expression Omnibus (GEO, https://www.ncbi.nlm.nih.gov/geo/) under the accession number GSE162865. We manually curated the annotation of the start and end position of some cgc genes (Data set S1) to generate a new annotation file. The scripts used for RNA-seq data analyses and for strong congocidine motif counting are available on the following Github link: https://github.com/Bury-Mone/Congomics (supplemental methods). The Integrative Genomics Viewer (IGV) tool was used to visualize simultaneously RNAseq data and genomic annotations (54).

Liquid chromatography analyses.

After 5 days of culture at 28°C in MP5 medium, supernatants were filtered through Mini-UniPrep syringeless filter devices (0.2 μm, Whatman) and analyzed on an Atlantis T3 column (250 mm by 4.6 mm, 5 μm, column temperature 30°C) using an Agilent 1200 HPLC instrument equipped with a quaternary pump. Samples were eluted with isocratic 0.1% HCOOH in H₂O (solvent A)/0.1% HCOOH in CH₃CN (solvent B) (95:5) at 1 ml ^. min⁻¹ for 7 min, followed by a gradient to 40:60 A/B over 23 min. Congocidine was detected by monitoring absorbance at 297 nm.

E. coli ET1267/pUZ8002 growth curves.

E. coli ET1267/pUZ8002 strain was transformed with pCGC002 (complete cgc gene cluster) or pCGC313 (cgc gene cluster deleted for cgc1) and selected on LB medium with hygromycin (50 μg/ml) and kanamycin (50 μg/ml). Bacteria were cultivated overnight in 2 ml of LB supplemented with kanamycin (50 μg/ml) at 37°C under agitation (180 rpm). The samples were then diluted to an adjusted OD600nm of 0.02 in LB medium supplemented or not with congocidine (11.6 or 23.2 μM). 200 μl of these dilutions were inoculated in triplicates into a 96-well plate (655161 GBO), sealed with an adhesive (4Ti – 0516/96). Bacteria were grown at 37°C in a CLARIOstar microplate reader (BMG LABTECH), under agitation (400 rpm), set with an automatically repeating protocol of cell density (OD600nm) readings every 20 min for each well during 16 h. The data obtained from the plate reader measurements were analyzed with Excel (Microsoft) and R software.

Study of the cgc20-cgc1 intergenic region using the reporter gene gusA in S. lividans TK23.

The complete cgc20-cgc1 intergenic region was amplified by PCR using LS-RP1 and LS-RP2, or LS-RP3 and LS-RP4 primers to introduce NotI and NdeI restriction sites in both orientations. The amplicons were cloned in pCR-blunt (Invitrogen) and sequenced. These plasmids were digested by NdeI and NotI, and the fragments containing the cgc20-cgc1 intergenic region were cloned upstream of gusA in the pOSV751 plasmid linearized by the same enzymes, yielding pCGC306 (gusA under the control of the cgc1 promoter, P_cgc1) and pCGC307 (gusA under the control of the cgc20 promoter, P_cgc20). The negative control, pCGC308, was obtained after religation of NdeI/NotI/Klenow-digested pOSV751. The three plasmids were next digested by SphI and PvuII and the fragments encompassing the three reporter cassettes (gusA without promoter or under the control of P_cgc1 or P_cgc20 promoters) surrounded by terminators were cloned into SphI/PvuII-digested pOVS808 (conjugative and integrative plasmid) (55). The resulting plasmids were named pCGC309 (gusA without promoter), pCGC310 (gusA under the control of P_cgc1), and pCGC311 (gusA under the control of P_cgc20), respectively, and introduced in S. lividans TK23 and CGCL0107 (expressing the cgc20-cgc21 resistance genes) by intergeneric conjugation, yielding the CGCL101 to CGCL106 strains. Genomic DNA was extracted from the exconjugants and analyzed by PCR using primers CEA31+LS04 and CEA29+LS63 for the proper integration at the VWB attB site and primers LS25, LS26 and Cmj12R for the presence of the appropriate reporter cassette.

Spectrophotometric measurement of β-Glucuronidase activity in cell lysates.

For each strain, 10⁵ spores were inoculated into 10 ml of TSB medium at 28°C under agitation (180 rpm, INFORS HT Unitron). After 24 h, the cultures were supplemented or not with congocidine (2.3 or 11.6 μM) and incubated under the same conditions for five additional hours. The mycelia were harvested by centrifugation, washed once with sodium phosphate buffer (100 mM, pH 7) and resuspended in 1 ml of buffer. Cells were disrupted using a One Shot homogenizer (Constant Systems Ltd.) at 2.5 bar. Lysates were centrifuged at 16,000 × g and 4°C for 15 min. The supernatants were used to determine the protein concentration using the Bradford assay (56). GusA enzymatic activity was measured as follows: 100 μg of proteins were diluted in 950 μl (final volume) of sodium phosphate buffer (100 mM, at pH 7) and placed in a spectrophotometer (UV-1800 Shimadzu) with temperature control and connected to a computer. Reactions were performed at 37°C and started by addition of 50 μl of 20 mM p-nitophenyl-β-d-glucuronide (Megazyme). The optical density at 415 nm was automatically measured all 30 sec during 30 min. The data obtained from spectrophotometer measurements were analyzed with Excel (Microsoft) and R software.