Characterization of the Widely Distributed Novel ECF42 Group of Extracytoplasmic Function σ Factors in Streptomyces venezuelae

In contrast to the one- and two-component signal transduction systems in bacteria, the importance and diversity of ECFs have only recently been recognized in the course of comprehensive phylogenetic and comparative genomics studies. Thus, most of the ECFs still have not been experimentally characterized regarding their physiological functions and regulation mechanisms so far. The physiological roles, target promoter, and target regulons of a novel group of ECFs, ECF42, in S. venezuelae have been investigated in this work. More importantly, members of this group are characterized by a C-terminal extension, which has been verified to harbor a regulatory role in ECF42s. Hence, our work provides an important source for further research on such C-terminal extension containing ECFs.

ABSTRACT

Extracytoplasmic function σ factors (ECFs) represent the third most abundant fundamental principle of bacterial signal transduction, outranked only by one- and two-component systems. A recent census of ECFs revealed a large number of novel groups whose functions and regulatory mechanisms have not yet been elucidated. Here, we report the characterization of members of the novel group ECF42. ECF42 is a highly abundant and widely distributed ECF group that is present in 11 phyla but is predominantly found in Actinobacteria. Analysis of the genomic context conservation did not identify a putative anti-σ factor. Instead, ECF42 genes are cotranscribed with genes encoding a conserved DGPF protein. We have experimentally verified the target promoter of these ECFs (TGTCGA in the −35 region and CGA/TC in the −10 region), which was found upstream of the ECF42-encoding operons in Streptomyces venezuelae, suggesting that ECF42s are positively autoregulated. RNA sequencing (RNA-seq) was performed to define the regulons of the three ECF42 proteins in S. venezuelae, which identified mostly genes encoding DGPF proteins. In contrast to typical ECFs, ECF42 proteins harbor a long C-terminal extension, which is crucial for their activity. Our work provides the first analysis of the function and regulatory mechanism of this novel ECF group that contains a regulatory C-terminal extension.

IMPORTANCE In contrast to the one- and two-component signal transduction systems in bacteria, the importance and diversity of ECFs have only recently been recognized in the course of comprehensive phylogenetic and comparative genomics studies. Thus, most of the ECFs still have not been experimentally characterized regarding their physiological functions and regulation mechanisms so far. The physiological roles, target promoter, and target regulons of a novel group of ECFs, ECF42, in S. venezuelae have been investigated in this work. More importantly, members of this group are characterized by a C-terminal extension, which has been verified to harbor a regulatory role in ECF42s. Hence, our work provides an important source for further research on such C-terminal extension containing ECFs.

INTRODUCTION

The ability of bacteria to accurately sense and respond to changing environments and lurking competitors is a prerequisite to survive the struggle for suitable habitats. To mediate such an adaptation process, bacteria possess different means to connect an extracellular input with an appropriate cellular response. Extracytoplasmic function σ factors (ECFs) are the third most abundant fundamental principle of bacterial signal transduction, outranked only by one- and two-component systems (1).

These alternative σ factors, which regulate diverse processes, such as stress responses, differentiation, secondary metabolism, or virulence, belong to the σ⁷⁰ protein family that also includes the housekeeping σ factors (2–4). However, in contrast to the housekeeping σ factors that contain four highly conserved protein domains, ECFs harbor only two of the four conserved domains (σ₂ and σ₄), which are sufficient for both promoter recognition and binding to the RNA polymerase (3). The activity of the ECFs is generally regulated by their cognate anti-σ factors, which are often membrane-anchored proteins encoded in the same operon as the σ factors. In the absence of a signal, the anti-σ factor tightly binds the σ factor, thereby keeping it inactive. In the presence of an inducing stimulus, the anti-σ factor releases the σ factor, which then recruits the RNA polymerase core and redirects transcription initiation to its target promoters (2). The presence of group-specific target promoters upstream of the ECF-encoding operon leads to the positive autoregulation of most ECFs, thereby enhancing their activating effect as long as inducing conditions prevail.

On average, bacterial genomes harbor about six ECFs, but their distribution does not correlate linearly with genome size. ECFs seem to be underrepresented and often absent in smaller genomes, while they are overrepresented in more complex bacteria, with some genomes harboring more than 100 ECFs (5). In contrast to the extensive knowledge on one- and two-component systems, the importance and diversity of ECFs have been recognized only recently in the course of comprehensive phylogenetic and comparative genomics studies (1, 5–7). Currently, 62 major and 32 minor groups can be discriminated based on sequence similarity of the σ and anti-σ pair, genomic context conservation, and target promoter motif (6). Some are well characterized, including (i) RpoE or σ^w-like σ factors of groups ECF01 to ECF04, which are activated by regulated proteolysis of their membrane-anchored anti-σ factors, in response to cell envelope stress (8), (ii) ECFs of groups ECF11 to ECF14, such as SigR in Streptomyces coelicolor, which are associated with soluble anti-σ factors that release the σ factor after a conformational change induced by redox stress (9), (iii) the FecIR-like transmembrane signaling system of groups ECF05 to ECF07, which are involved in iron-siderophore uptake (10), and (iv) group ECF15, whose activity is controlled by a mechanism of phosphorylation-dependent competitive partner switching (11, 12). In contrast to these few well-characterized examples, most ECFs, including all members of many widely distributed and abundant ECF groups, have not been investigated experimentally.

In contrast to the classical ECFs, which only contain the σ₂ and σ₄ domains, a number of novel groups (ECF41, ECF42, ECF44, ECF45, ECF48, ECF52, ECF53, ECF56, and ECF57) have been found that harbor a group-specific C-terminal extension domain (6). The regulatory role of the additional C-terminal extension in ECFs has so far only been studied in two of these ECF groups. While the C-terminal extension of ECF44 proteins was shown to be necessary for the activity of these σ factors (13), the C-terminal extension in ECF41 plays both a positive and a negative role in ECF41-dependent gene regulation (14).

Here, we report on a third such novel group, ECF42, which is characterized by a long C-terminal extension that contains a tetratricopeptide repeat (TPR) domain (1), which is postulated to be important for protein-protein interaction (15). Genes encoding ECF42 proteins are not genomically associated with apparent anti-σ factor genes. Instead, the large majority are associated with genes encoding the so-far-uncharacterized DGPF proteins, named after their most conserved Asp-Gly-Pro-Phe motif. So far, none of the σ factors belonging to ECF42 has been experimentally studied, with the exception of ECF-10 from Pseudomonas putida, which was found to be involved in stress resistance and biofilm formation (16). Since ECF42s are highly abundant in Actinobacteria, especially in the genus Streptomyces, we investigated ECF42s in S. venezuelae with regard to target promoters, physiological role, and regulatory mechanism. We identified the target genes of ECF42 and demonstrated that the C-terminal extension of ECF42 was essential for its activity.

RESULTS

Phylogenetic distribution and genomic context conservation of ECF42s.

The initial analysis of phylogenetic distribution and genomic context conservation on group ECF42 (1) in 2009 was based on a data set containing only 111 ECF42 protein sequences. To account for the huge increase in bacterial genomes sequenced within the last decade, we reanalyzed group ECF42 based on 2,661 ECF42 protein sequences that were extracted from the NCBI database.

An unrooted phylogenetic tree was constructed based on the gapless multiple-sequence alignment of 2,661 ECF42s to analyze the phylogenetic distribution of ECF42s in bacteria (Fig. 1A). ECF42s show a wide taxonomic distribution, and proteins of this group can be found in 11 different phyla but predominantly in the Actinobacteria (75.5%) and Proteobacteria (19.4%) (Table 1). In these two phyla, ECF42 proteins subdivide into eight and four separated clusters in the tree, respectively. Except for one branch of ECF42s from Proteobacteria that is included in a branch formed by ECF42s of Actinobacteria, the remaining ECF42s from Bacteroidetes, Acidobacteria, Firmicutes, Planctomycetes, Cyanobacteria, Verrucomicrobia, Spirochaetes, Chloroflexi, and Gemmatimonadetes formed separate branches (Fig. 1A).

FIG 1.

Phylogenetic distribution and genomic context conservation of ECF42s. (A) The phylogenetic tree of ECF42s is based on a gapless multiple-sequence alignment of 2,661 ECF42 protein sequences constructed using ClustalW, as implemented in the CLC Main Workbench. The resulting tree was calculated using the neighbor-joining method and Jukes-Cantor protein distance model. Assignment of sequences to bacterial phyla is indicated by color. The three ECF42s encoded in S. venezuelae (Sven_0747, Sven_4377, and Sven_7131) are highlighted with a red dot. (B) Conserved genomic context of ECF42. Genes are color-coded according to their COG family. The number of analyzed species per phylum is indicated in parentheses. On the right side of each genomic context, the illustration indicates the ratio between the number of genes with that conserved genomic context and the number of ECF42-coding genes identified in all the species of that phylum.

TABLE 1.

Phylogenetic distribution of ECF42 σ factors

Phylum	No. of ECF42s in each phylum	% of ECF42 proteins	No. of species containing ECF42 proteins	Avg no. of ECF42s/species
Actinobacteria	2,011	75.5	544	3.7
Proteobacteria	517	19.4	353	1.5
Bacteroidetes	36	1.4	33	1.1
Acidobacteria	20	0.8	11	1.8
Firmicutes	20	0.8	19	1.1
Planctomycetes	18	0.7	15	1.2
Cyanobacteria	12	0.5	11	1.1
Verrucomicrobia	9	0.3	7	1.3
Spirochaetes	7	0.3	3	2.3
Chloroflexi	6	0.2	4	1.5
Gemmatimonadetes	5	0.2	3	1.7

The 2,661 ECF42s derive from 1,003 different species, suggesting that several of these species harbor more than one copy of ECF42 genes in their genome. This is particularly obvious in the phylum Actinobacteria, in which a species harbors more than three copies of ECF42s on average. However, members of other phyla harbor only one to two ECF42s per species (Table 1). In contrast to most ECF groups defined to date (6), ECF42 genes lack neighboring anti-σ factor genes. Instead, more than 90% of ECF42 genes are genomically associated with genes encoding DGPF proteins (YCII superfamily, COG3795) (Fig. 1B). In almost 90% of these cases, DGPF genes are located upstream and oriented in the same direction as ECF42 genes with only a short intergenic region between them, indicative of cotranscription. In addition to the DGPF proteins, in a few cases, proteins from other families are also found, such as glyoxalase/bleomycin resistance proteins/dioxygenases (COG-PhnB), acyl coenzyme A (acyl-CoA) dehydrogenases (COG-CaiA), uncharacterized conserved proteins (COG-3832), or ketosteroid isomerase homologs (COG-4319) (Fig. 1B).

Based on the predominant distribution of the ECF42s in the Actinobacteria, ECF42s from S. venezuelae, a model organism for the streptomycetes, were chosen for our study on ECF42s regarding the determination of target promoters, physiological roles, and regulatory mechanisms.

Prediction and verification of ECF42-dependent target promoters in S. venezuelae.

To accurately redirect gene expression, ECFs select promoters with high stringency by combining sequence-specific interactions with the −10 and −35 promoter elements, and hence, each ECF group has its own group-specific target promoter. Identification of the target promoters of ECFs enables us to define their target regulons, thereby providing direct access to the physiological role of ECF-dependent regulation. Generally, the target promoters of ECFs can be found at the intergenic regions directly upstream of the operon encoding ECFs, which leads to the positive autoregulation of most ECFs. Therefore, the putative target promoters of ECF42s were predicted by using the MEME Suite (17) to search for a conserved promoter motif upstream of the operons encoding ECF42s from different bacterial species belonging to different phyla. As expected, a conserved putative promoter motif was identified with TGTCGA in the −35 region and CGTC in the −10 region (Fig. 2A). This predicted ECF42 promoter motif was also present upstream of the three operons encoding ECF42s in S. venezuelae (Fig. 2B).

FIG 2.

ECF42 promoter determination and cross talk in S. venezuelae. (A) Conserved motif of ECF42 target promoters. The sequence logo was generated using the MEME motif discovery tool in MEME Suite (http://meme-suite.org/). The sequence logo graphically represents the ECF42 promoter position weight matrix and illustrates the degree of sequence conservation for each nucleotide position. The matrix is based on 18 putative promoter sequences found upstream of DGPF- and ECF42-encoding operons. (B) Genomic context of the three ECF42-coding genes of S. venezuelae, sven_0747, sven_4377, and sven_7131. DGPF and ECF42 genes are shown in light gray and dark gray, respectively. The orange square represents the location of the putative ECF42 target promoter. The putative −35 and −10 regions are underlined. (C) Evaluation of putative ECF42 target promoters' activities in strains overexpressing different ECF42s. The putative promoter-containing DNA fragments, which start 80 bp upstream of the putative −35 region and end 20 bp downstream of the putative −10 region, were inserted into pGUS, which carries a gus reporter gene, generating the plasmids pQLS007 (P_{sven_0747}), pQLS008 (P_{sven_4377}), and pQLS009 (P_{sven_7131}). The integrative plasmid was transformed into S. venezuelae strains individually overexpressing the ECF42s as well as in the wild-type (WT) strain. The activities of each promoter and of the empty pGUS vector were measured. The activity is expressed as Miller units per microgram proteins. (***, P < 0.001; **, P < 0.01 compared to the WT).

Next, we wanted to verify that these putative promoters were indeed targets of ECF42-dependent transcription initiation in S. venezuelae. Toward this goal, the corresponding promoter fragments were fused to the β-glucuronidase (Gus) reporter gene, and the activities of these promoters were assessed by measuring the Gus activity in S. venezuelae strains individually overexpressing three ECF42s from the constitutive promoter ermE*p. As shown in Fig. 2C, the activity of the sven_0747 promoter (P_{sven_0747}) was about 10-fold higher in the strain overexpressing Sven_0747 (TMS0112) than in the wild-type strain, while the sven_4377 promoter had a 5-fold increased activity in the strain overexpressing Sven_4377 (TMS0113) relative to the wild type, confirming our ECF42 predicted target promoters. In contrast, no increased activity was detected for the putative sven_7131 promoter in any of the ECF42 overexpression strains, indicating that this is not an ECF42 target promoter. In fact, it shows an approximately 3- to 5-fold reduced activity in strains overexpressing sven_0747 and sven_4377.

Since the genome of S. venezuelae encodes three putative ECF42s, we next wanted to investigate a potential cross talk between different ECF42s and their target promoters. The activity of each ECF42 target promoter was also evaluated in strains overexpressing the other two ECF42s. No cross talk was observed between any of the noncognate ECF-promoter pairs except for the Sven_4377-P_{sven_0747} pair (Fig. 2C).

Role of ECF42s in S. venezuelae.

ECF42s are highly abundant in Actinobacteria (with 1 to 4 copies per genome), especially in the genus Streptomyces, in which members encode up to 4 copies per genome. This suggests that they may play an important physiological role in these organisms. For this reason, the biological role of ECF42s in S. venezuelae was investigated. First, we checked the effect of ECF42 deletion on the development and morphology in S. venezuelae, but no significant differences were observed between the wild type and the deletion strain (data not shown). The same holds true for the strains individually overexpressing the ECF42s.

Subsequently, we performed Phenotype MicroArray (PM) analysis (Biolog), which is a high-throughput approach that allows testing of hundreds of different physiological conditions in parallel in order to identify phenotypes associated with genetic alterations (18). Our PM analysis included 960 assays for carbon, nitrogen, phosphorus, and sulfur utilization, nutrient stimulation, pH, and osmotic stress, as well as chemical sensitivity test covering 240 different compounds. We compared the S. venezuelae wild type (TMS0001) to the ECF42 triple-deletion mutant (TMS0050) and observed that they behave in a similar way under all tested conditions except for one, as follows: in the presence of the dipeptide methionine-histidine (Met-His) as the sole nitrogen source (see Fig. S1 in the supplemental material), the triple-deletion mutant showed a slightly higher metabolic activity than that of the wild type. However, this behavior was not observed in the presence of other dipeptides containing methionine or histidine or in the presence of the single amino acids as sole nitrogen sources in the Phenotype MicroArray assay. Taken together, these results indicate that the ECF42 triple-deletion mutant did not show any reproducible growth differences compared to the wild type under all the conditions tested. We were therefore not able to identify any ECF42-related phenotype.

Defining the regulon of ECF42s by RNA-seq.

Since the phenotypic analysis was nonconclusive, we next aimed at unraveling the physiological role of ECF42s by identifying their target genes in S. venezuelae. Toward this goal, RNA sequencing (RNA-seq) analyses were performed on strains overexpressing each of the ECF42s in comparison to the ECF42 triple-deletion mutant. The three ECF42s were overexpressed individually from an integrated vector with a constitutive promoter in the ECF42 triple-deletion mutant background. The production of the ECF42s during the growth cycle was confirmed by Western blotting (Fig. S2), and cells from mid-exponential cultures (10 h of incubation) were selected for RNA extraction, since these contained the highest level of ECF42 proteins. The RNA-seq data from the three ECF42s overexpression strains were analyzed and compared to those from the ECF42 triple-deletion mutant. The complete list of differentially expressed genes (P ≤ 0.05) in each of the ECF42 overexpression strains can be found in File S2, and the upregulated genes (2-fold) are listed in Table 2. The DGPF genes sven_0748 and sven_4376, located immediately upstream of ECF42 genes sven_0747 and sven_4377, were upregulated more than 16-fold. In addition to genes encoding DGPF proteins or unknown hypotheticals, genes whose products are involved in iron utilization, or genes that act as molecular chaperones, aminotransferases were also upregulated in the ECF42 overexpression strains (Table 2), although to a lesser degree (i.e., with lower fold changes). Putative ECF42 target promoters were identified in the upstream regions of all of these genes, suggesting that they are direct targets of ECF42s (Fig. 3A and B).

TABLE 2.

Genes upregulated in the ECF42 overexpression strains

Strain and gene locus	Fold change	P value	Description
Mutant strain overexpressing sven_0747: sven_0748	16.34	1.00945E−23	DGPFAETKE family protein
Mutant strain overexpressing sven_4377
sven_3806	643.59	0	DGPFAETKE family protein
sven_4376	20.97	0	DGPFAETKE family protein
sven_6046	7.21	5.51978E−25	Molybdate-binding domain of ModE
sven_3126	2.91	1.98672E−07	Hypothetical protein
sven_4575	2.83	2.09087E−05	Putative RNA polymerase ECF39 group
sven_1017	2.36	1.52519E−12	Hypothetical protein
sven_0690	2.33	1.53116E−10	Putative lyase
sven_4614	2.19	5.28467E−16	Hypothetical protein
sven_1219	2.04	4.78931E−25	Iron utilization protein
Mutant strain overexpressing sven_7131
sven_2864	4.86	1.75073E−05	Phage major capsid protein
sven_1613	3.94	1.76361E−08	Aspartate or tyrosine or aromatic aminotransferase
sven_3169	2.28	1.94133E−14	Molecular chaperone IbpA, HSP20 family
sven_5840	2	4.73082E−06	Hypothetical protein

FIG 3.

ECF42 regulon in S. venezuelae. (A) Putative promoter sequences in the upstream region of the upregulated genes. The promoter sequences harbor the ECF42 target promoter motif and are shown according to individual ECF42 overexpression strains. The length of the spacer between −35 and −10 regions is indicated for each promoter sequence. (B) Sequence logo generated based on the promoter sequences listed in panel A. We note that this motif is very similar to the one shown in Fig. 2A, although the conservation in positions 7, 8, and 29 is lower. (C) Expression patterns of all DGPF protein-coding genes identified in the genome of S. venezuelae. In the graph are shown the fold changes of the expression of the DGPF protein-coding genes (sven_0748, sven_4376, sven_7130, sven_3806, sven_5447, and sven_6405) in the S. venezuelae ECF42 overexpression strains TMS0012 (Sven_0747), TMS0113 (Sven_4377), and TMS0114 (Sven_7131) in reference to the triple-deletion mutant TMS0050. ***, P < 0.001.

We next plotted the expression levels of the six DGPF-encoding genes identified in the genome of S. venezuelae to gain further insight into their expression patterns (Fig. 3C). Two of the three DGPF proteins associated with ECF42s were overexpressed in the strain containing their associated ECF42 genes, sven_0748 and sven_4376. This is in agreement with our data concerning the identification of ECF42 target promoters in S. venezuelae (Fig. 2C), since sven_0748 and sven_4376, but not sven_7130, were active upon overproducing the cognate ECF42 protein. Additionally, we observe that the DGPF protein-encoding gene sven_0748 is overexpressed in the strain overexpressing Sven_4377. Again, this observation is in agreement with our previous observation that the promoter of sven_0748 is active in the strain overproducing Sven_4377 (Fig. 2C). One additional gene coding for a DGPF protein that is not genomically associated with any ECF42-encoding gene (sven_3806) was strongly upregulated (643-fold) in the Sven_4377 overexpression strain (Fig. 3C). Collectively, our data suggest that regulating the expression of the DGPF proteins is the major role of the ECF42s in S. venezuelae. Unfortunately, in the absence of a clear phenotype, this result does not help unravel the physiological role of ECF42 proteins in S. venezuelae.

ECF42's C-terminal extension is necessary for activity.

In contrast to the majority of ECF42s, which are regulated by cognate anti-σ factors, ECF42 proteins lack such a partner protein and instead contain a conserved C-terminal extension that is hypothesized to play a regulatory role, as has previously been demonstrated for members of groups ECF41 and ECF44 (13, 14). We therefore aimed to investigate the postulated regulatory role of the C-terminal extension of ECF42 proteins on their activity. Typically, ECFs contain only the σ₂ and σ₄ conserved domains characteristic of the σ⁷⁰ protein family, which are sufficient to ensure interaction with the RNA polymerase and promoter recognition. A multiple-protein-sequence alignment of classical ECFs from different organisms and those of group ECF42 revealed the presence of a large C-terminal extension containing a conserved tetratricopeptide repeat (TPR) protein domain (Fig. 4A). Such TPR domains are known to mediate protein-protein interactions (15).

FIG 4.

Role of the C-terminal extension in the activity of the ECF42 Sven_4377 of S. venezuelae. (A) The multiple-sequence alignment of selected classical and ECF42s was constructed using DNAMAN. Identical amino acids at the same position are shaded in black, and similar amino acids are in pink or blue. The σ₂ and σ₄ domains and the C-terminal extension, which includes the TPR domain, are marked. The exact site of the truncations of Sven_4377 is indicated by arrows below the alignment. (B) Schematic representation of protein domain architecture of classical and ECF42s and positions of the constructed truncations. (C) Activity of the sven_4377 promoter in strains overexpressing the different alleles of Sven_4377. The sven_4377 promoter was cloned into the integrative pGus reporter vector and transformed into S. venezuelae strains overexpressing truncated FLAG-tagged ECF42 T1 (amino acids [aa] 1 to 179), T2 (aa 1 to 230), T3 (aa 1 to 287), and T4 (aa 1 to 353) (x axis). The activity of the promoter in the different strains was determined by β-glucuronidase assays. The activity is expressed as Miller units per microgram of protein. The negative control (Ng) corresponds to the empty pGUS vector in the strain overexpressing the full-length ECF42 (***, P ≤ 0.001 compared to the wild-type allele). (D) Production of soluble ECF42 alleles was verified by Western blotting using a FLAG tag-specific antibody.

To study the role of this C-terminal extension, we generated a series of C-terminal truncations of the ECF42 Sven_4377 from S. venezuelae and evaluated the activity of its target promoter P_{sven_4377}. Four different C-terminally truncated alleles were constructed and designated T1 to T4, with T1 being the longest truncation (i.e., the shortest allele) and T4 the shortest truncation (i.e., the longest allele) (Fig. 4B). All our truncations resulted in a severe decrease in the activity of the ECF42 relative to its full-length allele, as judged by its target promoter activity (Fig. 4C). This behavior was not due to a reduced production level or aggregation of the truncated alleles, since the production of soluble ECF42 alleles was verified by Western blotting (Fig. 4D). These observations suggest that the C-terminal extension of ECF42 is necessary for its activity, as seen previously for ECF41 (14). However, in contrast to the situation for ECF41, the complete ECF42 C-terminal extension is necessary for ECF activity, since even the shortest truncation already led to a loss of σ factor activity.

DISCUSSION

In this report, we provide the very first detailed characterization of members of the novel group of ECF42 proteins with regard to their phylogenetic distribution, physiological role, target promoter sequence, regulons, and the regulatory role of the C-terminal extension in S. venezuelae.

Our search for overrepresented motifs in the upstream regions of operons encoding ECF42s identified a motif for ECF42 target promoters (Fig. 2A), which we subsequently confirmed for two of the three operons encoding ECF42s of S. venezuelae (Fig. 2B and C). Only the ECF-promoter pair Sven_7131-P_{sven_7131} was not active. This ECF42 is longer than the other two and contains an additional phage integrase N-terminal SAM-like domain (Pfam database, entry PF02899), which is known to be involved in DNA recombination (19). Remarkably, the overproduction of Sven_7131 only resulted in reduced fold change differences, that is, not a single gene was overexpressed in this strain (File S2). Together, these data suggest that Sven_7131 does not act as a σ factor but has gained a different function during evolution that still needs to be uncovered.

The bioinformatics analysis of the phylogenetic distribution of ECF42s revealed a vast overrepresentation of these ECFs in actinobacterial genomes (Fig. 1A). These organisms in fact encode several copies of these σ factors per genome on average (Table 1). While this might argue for a functional redundancy, our data seem to support the hypothesis that they work independently. On the one hand, we have observed little cross talk between the ECF-promoter pairs, with the minor exception of the noncognate pair Sven_4377-P_{sven_0747} (Fig. 2C). On the other hand, our RNA-seq data showed that each of the three ECF42 proteins activates a different set of genes without any overlap between them. The analysis of the putative ECF42 binding sites in the upstream regions of the genes controlled by each ECF42 did not reveal any obvious sequence determinants in the core −10 and −35 promoter regions. However, we cannot exclude that less-conserved residues in the spacer region will not contribute to or determine this specificity, as demonstrated for other ECF promoters in Bacillus subtilis (20). Unfortunately, the limited number of ECF42 target genes in S. venezuelae, indicative of very small regulons, currently prevents the identification of such residues.

Another salient aspect of our bioinformatics analysis was the almost-complete cooccurrence of genes encoding ECF42s and those coding for DGPF proteins (Fig. 1B). The significance of this linkage became obvious by the finding that genes coding for DGPF proteins, either linked to the ECF42s or not, are the major targets of these ECF42s, at least in S. venezuelae. The exact function of the DGPF (or YCII-related) proteins is currently unknown, although the structure of members of the same protein family has been determined (21). This protein family is defined by sequence similarity and belongs to the Dim_A_B_barrel clan, whose proteins possess a ferredoxin-like fold. One member of this protein family has been characterized as a dehydrochlorinase (22). Additionally, the Escherichia coli YCII protein was found to interact with RibD, a deaminase/reductase of the riboflavin biosynthesis pathway (23).

YCII/DGPF domains are in their large majority found isolated but in some cases have been reported to be fused to other proteins (or domains). In two cases, they are located N-terminally to sigma70_R2 domain-containing proteins (in Arthrobacter spp. and Mycobacterium ulcerans); in another case, it has been found N-terminally fused to a sigma70_R2 and a sigma70_R4 domain (in Caulobacter crescentus CC_1329). All of these proteins are also ECF42s, which suggests this domain might play a regulatory role in ECF42 activity. However, our unpublished data have shown that the activity of the ECF42 from Xanthomonas campestris was not affected by the presence of the cognate DGPF protein.

The slight increase in metabolic activity of the ECF42 triple-deletion mutant in the presence of the Met-His dipeptide as a sole nitrogen source was surprising to us, since it implies a negative effect of these ECFs on metabolism. There are two ways in which we can envision this mechanism. On one hand, we could hypothesize a direct effect of the ECF42s on genes whose products are involved in the import or utilization of this dipeptide. Indeed, we have found the conserved ECF42 promoter signature in the antisense orientation of the 5′ untranslated region of the operon encoding the methionine ABC transporter (sven_1158, sven_1157, and sven_1156), as shown in Fig. S3. This would suggest that ECF42s negatively regulate the expression of this operon by transcriptional interference. However, our RNA-seq data show only a very small downregulation of genes in this operon in the strains overexpressing Sven_0747 or Sven_4377 (between 0.8- and 0.57-fold). On the other hand, we could hypothesize an indirect effect of the ECF42s mediated by their target, the DGPF proteins. Proteins of this family have been shown to interact with RibD, a deaminase/reductase of the riboflavin biosynthesis pathway (23). Riboflavin is involved in the production of methionine from homocysteine via folate. The interaction of the DGPF with RibD could to some extent interfere with this process. However, we could not see a similar decrease in metabolic activity of the ECF42 triple-deletion mutant in the presence of only methionine or other methionine-containing dipeptides.

The role of C-terminal extensions in controlling ECFs has already been investigated for two ECF groups, ECF41 (14) and ECF44 (13). In the case of ECF41s, we have demonstrated that the extension has both an activating role and a repressing role on σ factor activity, while it has been demonstrated for ECF44 that the extension coordinates a metal ion and determines the specificity of the σ factor. Here, we show that the full C-terminal extension of ECF42 is essential for its activity. We hypothesize that any disturbance of the TPR protein-protein interaction domain will render the ECF inactive, but the exact mechanism by which this effect is exerted will require further investigation.

In conclusion, we have demonstrated that two out of the three putative ECF42s of S. venezuelae are indeed functional nonredundant ECFs since they activate different sets of genes (Fig. 3A and Table 2). Moreover, we have shown that the main target genes of these σ factors are those coding for DGPF proteins (Table 2). Finally, we have determined that the full-length C-terminal extension is necessary for ECF42 activity. Together, these data provide the first insights into the function and mechanism of one of the most abundant, previously untouched groups of ECFs. We hope that our results will inspire subsequent studies on ECF42 from different bacterial species to ultimately unravel both the physiological role and the exact mechanism of ECF42-dependent stimulus perception and σ factor activation.

MATERIALS AND METHODS

Bioinformatics analysis.

An ECF42 protein sequence from S. venezuelae (NCBI RefSeq accession no. WP_015035572.1) was submitted to NCBI blastp and run against the nonredundant protein sequence database. The protein sequences of the complete 10,010 hits were extracted in December 2017. False positives (i.e., proteins that did not belong to the ECF42 group) and proteins from more than one sequenced strain per species were removed, leaving 2,661 sequences for further analysis. Multiple-sequence alignments were performed using ClustalW (24), and the phylogenetic tree was generated from the gapless multiple alignments using the neighbor-joining method and Jukes-Cantor protein distance model, as implemented in CLC Main Workbench (Qiagen). Genomic context analysis of the ECF42 family (COG4941) was performed using the database MicrobesOnline (25) (http://www.microbesonline.org/). Eighteen operons encoding ECF42 and DGPF proteins were selected, and the 250 bp upstream of the starting codon of the first gene of the operon was extracted for promoter motif analysis. The presence of a putative promoter motif was investigated with the MEME motif discovery tool implemented in the MEME Suite (17) (http://meme-suite.org/). The parameter settings for the MEME analysis were distribution, zero or one; width, minimum 3 and maximum 30 (this width allows the identification of either the complete promoter or one part of the bipartite target promoter); and optional parameter, search given strand only. Protein domain architecture and alignment of ECF42s with classical ECFs were analyzed using the DNAMAN software package (Lynnon BioSoft, Vaudreuil, Quebec, Canada).

Bacterial strains and culture conditions.

All E. coli and S. venezuelae strains used in this study are listed in Table 3. E. coli DH10β was used for plasmid propagation. E. coli ET12567/pUZ8002 was used for transferring plasmids to S. venezuelae by conjugation. All E. coli strains were grown in lysogeny broth (LB medium) at 37°C with agitation. Conjugation mixtures were plated on MS agar medium (20 g of mannitol, 20 g of soya flour, and 20 g of agar in 1 liter tap water) and incubated for 3 to 4 days at 30°C (26, 27). S. venezuelae was grown either in liquid (with shaking) or on solid MYM (4 g of maltose, 4 g of yeast extract, and 10 g of malt extract per liter) medium at 30°C. MYM medium was prepared using 50% tap water and 50% ultrapure water and after autoclaving was supplemented with 200 μl trace elements stock solution [20 g of ZnCl₂, 100 g of FeCl₃·6H₂O, 5 g of CuCl₂·2H₂O, 5 g of MnCl₂·4H₂O, 5 g of Na₂B₄O₇·10H₂O, 5 g of (NH₄)₆Mo₇O₂₄·6H₂O in 1 liter ultrapure water] per 100 ml. The following antibiotics were added to select for plasmid-bearing and mutant strains: apramycin (50 μg/ml), thiostrepton (25 μg/ml), chloramphenicol (25 μg/ml), hygromycin (50 μg/ml), kanamycin (50 μg/ml), and nalidixic acid (25 μg/ml).

TABLE 3.

Bacterial strains used in this study

Strain	Genotype or descriptiona	Reference or source
E. coli
DH10β	F− mcrA Δ(mrr-hsdRMS-mcrBC) ϕ80lacZΔM15 ΔlacX74 recA1 endA1 araD139 Δ(ara leu)7697 galU galK rpsL nupG λ−	Invitrogen
ET12567	dam-13::Tn9 dcm-6 hsdM hsdR	29
ET12567/pUZ8002	ET12567 harboring pUZ8002, a non-self-transmissible plasmid that can mobilize oriT-containing plasmids by conjugation	30
S. venezuelae
TMS0001	ATCC 10712	Lab strain
TMS0039	ATCC 10712 Δsven_7131	This work
TMS0044	ATCC 10712 Δsven_7131 Δsven_4377	This work
TMS0050	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131	This work
TMS0112	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_0747	This work
TMS0113	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_4377	This work
TMS0114	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_7131	This work
TMS0174	ATCC 10712 attBϕC31::pGUS	This work
TMS0175	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕC31::pGUS	This work
TMS0176	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_0747 attBϕC31::pGUS	This work
TMS0177	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_4377 attBϕC31::pGUS	This work
TMS0178	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_7131 attBϕC31::pGUS	This work
TMS0179	ATCC 10712 attBϕC31::pGUS-NP_0747	This work
TMS0180	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕC31::pGUS-NP_0747	This work
TMS0181	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_0747 attBϕC31::pGUS-NP_0747	This work
TMS0182	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_4377 attBϕC31::pGUS-NP_0747	This work
TMS0183	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_7131 attBϕC31::pGUS-NP_0747	This work
TMS0184	ATCC 10712 attBϕC31::pGUS-NP_4377	This work
TMS0185	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕC31::pGUS-NP_4377	This work
TMS0186	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_0747 attBϕC31::pGUS-NP_4377	This work
TMS0187	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_4377 attBϕC31::pGUS-NP_4377	This work
TMS0188	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_7131 attBϕC31::pGUS-NP_4377	This work
TMS0189	ATCC 10712 attBϕC31::pGUS-NP_7131	This work
TMS0190	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕC31::pGUS-NP_7131	This work
TMS0191	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_0747 attBϕC31::pGUS-NP_7131	This work
TMS0192	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_4377 attBϕC31::pGUS-NP_7131	This work
TMS0193	ATCC 10712 Δsven_0747 Δsven_4377 Δsven_7131 attBϕBT1::pIJ10257-N_3×FLAG_sven_7131 attBϕC31::pGUS-NP_7131	This work
TMS0036	ATCC 10712 Δsven_4377::apr pGUS-NP_4377	This work
TMS0051	ATCC 10712 Δsven_4377::apr attBϕC31::pGUS-NP_4377 attBϕBT1::pIJ10257-3×FLAG _sven_4377_T1 (aa 1–179)	This work
TMS0052	ATCC 10712 Δsven_4377::apr attBϕC31::pGUS-NP_4377 attBϕBT1::pIJ10257-3×FLAG_sven_4377_T2 (aa 1–230)	This work
TMS0053	ATCC 10712 Δsven_4377::apr attBϕC31::pGUS-NP_4377 attBϕBT1::pIJ10257-3×FLAG_sven_4377_T3 (aa 1–287)	This work
TMS0054	ATCC 10712 Δsven_4377::apr attBϕC31::pGUS-NP_4377 attBϕBT1::pIJ10257-3×FLAG_sven_4377_T4 (aa 1–353)	This work

^aaa, amino acids.

DNA manipulations, plasmids, and oligonucleotides.

The plasmids used or generated in this study are listed in Table 4. Restriction of DNA molecules and DNA cloning were carried out using enzymes purchased from New England BioLabs (NEB), according to the manufacturer's instructions. DNA fragments required for plasmid or strain construction were amplified from the genomic DNA of S. venezuelae ATCC 10712 by PCR with Q5 high-fidelity DNA polymerase (NEB) using the primers listed in Table S1 in the supplemental material. DNA fragments were cloned into the relevant vectors, and all generated plasmids were verified by DNA sequencing.

TABLE 4.

Plasmids used in this study

Plasmid	Descriptiona	Reference or source
pIJ10257	Integrated into ϕBT1 attachment site of Streptomyces sp. ermE*p promoter cloned into pMS82 (Hygr)	31
pGUS	Conjugative and ϕ31-integrative vector, Gus reporter gene without promoter (Aprr)	32
pIJ12738	pKC1132 with multiple-cloning site and I-SceI site from pUC57-Simple_SceI	28
pIJ12742	pGM1190 with ermE*p-I-SceI gene (Thir)	28
pQLS001	Up- and downstream regions of sven_0747 were ligated and cloned into pIJ12738 with HindIII and KpnI	This work
pQLS002	Up- and downstream regions of sven_4377 were ligated and cloned into pIJ12738 with HindIII and KpnI	This work
pQLS003	Up- and downstream regions of sven_7131 were ligated and cloned into pIJ12738 with HindIII and KpnI	This work
pQLS004	3×FLAG_sven_0747 was introduced into multiple-cloning site of pIJ10257 with NdeI and HindIII	This work
pQLS005	3×FLAG_sven_4377 was introduced into multiple-cloning site of pIJ10257 with NdeI and HindIII	This work
pQLS006	3×FLAG_sven_7131 was introduced into multiple-cloning site of pIJ10257 with NdeI and HindIII	This work
pQLS007	Predicted target promoter region of sven_0747 was introduced into multiple-cloning site of pGUS with XbaI and KpnI	This work
pQLS008	Predicted target promoter region of sven_4377 was introduced into multiple-cloning site of pGUS with XbaI and KpnI	This work
pQLS009	Predicted target promoter region of sven_7131 was introduced into multiple-cloning site of pGUS with XbaI and KpnI	This work
pQLS010	3×FLAG_sven_4377_T1 was introduced into multiple-cloning site of pIJ10257 with HindIII and KpnI	This work
pQLS011	3×FLAG_sven_4377_T2 was introduced into multiple-cloning site of pIJ10257 with HindIII and KpnI	This work
pQLS012	3×FLAG_sven_4377_T3 was introduced into multiple-cloning site of pIJ10257 with HindIII and KpnI	This work
pQLS013	3×FLAG_sven_4377_T4 was introduced into multiple-cloning site of pIJ10257 with HindIII and KpnI	This work

^aHyg^r, hygromycin resistant; Apr^r, apramycin resistant; Thi^r, thiostrepton resistant.

I-SceI meganuclease-mediated triple deletion of ECF42s.

S. venezuelae encodes three ECF42s (Sven_0747, Sven_4377, and Sven_7131). The markerless triple-deletion mutant (Δsven_0747 Δsven_4377 Δsven_7131) was generated using the I-SceI meganuclease-mediated method (28). Approximately 2-kb sequences up- and downstream of each gene were amplified using the appropriate primers (Table S1) and pairwise cloned into the delivery vector pIJ12738 (that contains the I-SceI recognition site) with HindIII and KpnI, resulting in the constructs pQLS001, pQLS002, and pQLS003 (Table 4). First, plasmid pQLS003 was introduced into wild-type S. venezuelae by conjugation to generate the single-crossover mutant of sven_7131. Single-crossover mutants were selected on the basis of apramycin resistance and confirmed by colony PCR. A plasmid encoding the I-SceI meganuclease (pIJ2742) was then conjugated into the single-crossover strain and selected on the basis of thiostrepton resistance. Exconjugants were grown on medium containing 50 μg/ml thiostrepton at 30°C. The constitutively expressed I-SceI caused the double-strand breaks at its introduced recognition site in the chromosome. After homologous recombination, the double-crossover mutants were counterselected on the basis of apramycin sensitivity before confirmation by colony PCR. Since pIJ2742 has a temperature-sensitive origin of replication, selected Δsven_7131 mutants were therefore grown at 37°C to promote loss of the plasmid. Once a markerless ECF42 single-deletion mutant, TMS0039 (Δsven_7131), was obtained, the ECF42 double-deletion mutant TMS0044 (Δsven_7131 Δsven_4377) was subsequently obtained by deleting sven_4377 using the same method described above based on the single-deletion mutant (Δsven_7131). The triple-deletion mutant of ECF42, TMS0050 (Δsven_0747 Δsven_4377 Δsven_7131), was then generated by the deletion of the third ECF42 gene, sven_0747, based on the double-deletion mutant TMS0044 in the same way.

Phenotype MicroArrays.

The Phenotype MicroArray assay for phenotypic characterization of microbial cells (29, 30) was performed by Biolog for the S. venezuelae wild type and ECF42 triple-deletion mutant strains to find the ECF42-related phenotypes. Briefly, S. venezuelae was grown overnight on MYM agar medium at 30°C. After subculture for a second time, cells from the agar plate were transferred into a sterile capped tube containing 25 ml of IF-0a medium (Biolog, Hayward, CA) to obtain a uniform suspension with transmittance of 80%. One hundred microliters of cell suspension with a metal ion cocktail was then distributed into the 96-well microplates PM1 to PM20 with different medium compositions in each well, such as different carbon sources, nitrogen sources, chemical reagents, and antibiotics (more information concerning the plate layouts and medium composition can be obtained at https://biolog.com/products-portfolio-overview/phenotype-microarrays-for-microbial-cells/). All PM plates were incubated at 30°C for 48 h, and cell growth was continuously monitored in all wells of the arrays.

Overexpression of ECF42s.

Three ECF42 genes (sven_0747, sven_4377, and sven_7131) with an N-terminal 3×FLAG tag were amplified with primer pairs TM5631/TM4056, TM5632/TM4059, and TM5633/TM4062 and then cloned into the vector pIJ10257 (31) with NdeI and HindIII to generate ECF42-overexpressing plasmids pQLS004, pQLS005, and pQLS006, respectively (Table 4). The overexpression plasmid was conjugated into S. venezuelae ECF42 triple-deletion mutant TMS0050 to obtain three different ECF42 overexpression strains, TMS0112 (3×FLAG_sven_0747), TMS0113 (3×FLAG_sven_4377), and TMS0114 (3×FLAG_sven_7131). The overexpression of ECF42 in S. venezuelae was confirmed by Western blotting using an anti-FLAG antibody, as described previously (32).

RNA isolation.

For RNA isolation, three 30-ml cultures of S. venezuelae were grown in MYM supplemented with trace element solution at 30°C with shaking (initial optical density at 450 nm [OD₄₅₀], 0.01). After 10 h of incubation, cultures were mixed with 20 ml of precooled (−20°C) solution of 60% glycerol and 34% of sodium chloride. The cells were then harvested by centrifugation at 8,000 × g and −20°C for 30 min. Cells were resuspended in 1 ml of RNA TRI Reagent (Zymo Research), transferred to a beat beater tube containing glass beads, and homogenized with the help of a bead beater (FastPrep FP120; Thermo) for 3 cycles of 30 s. Two hundred microliters of chloroform was then added to the homogenized samples, thoroughly mixed by vortexing, and centrifuged at 4°C and 12,000 × g for 20 min. The supernatant was transferred to a new tube, mixed with 2.5 volumes of pure ethanol, and frozen for 12 h. Afterwards, the sample was centrifuged at 4°C and 12,000 × g for 20 min. The pellet was washed twice with 80% ethanol. The crude nucleic acid preparation was then dissolved in 200 μl diethyl pyrocarbonate (DEPC)-treated water and further purified with the Direct-zol RNA MiniPrep Plus kit (Zymo Research). After that, rRNA was removed with the Ribo-Zero rRNA removal kit (Gram-positive bacteria; Illumina). The remaining RNA was concentrated in RNase-free water by using the RNA Clean & Concentrator-25 kit (Zymo Research).

RNA-seq and data analysis.

The RNA-seq and data analysis were performed at the biotechnology center of Technische Universität Dresden. Briefly, the mRNA was quantified with TapeStation, and 100 ng was chemically fragmented using the Ultra II directional RNA library prep kit (NEB). Next, reverse transcription was performed with random hexamers, and second-strand synthesis was performed using a dUTP-primer mix. After an XP-bead purification (1.8×) step, the cDNA was subjected to end repair and A-tailing before custom adapters were ligated. After ligation, adapters were removed by an XP bead purification (Beckman Coulter), adding beads at a ratio of 0.9:1. Indexing was done by PCR enrichment (with 6 cycles) using custom amplification primers carrying the index sequence. After two more XP bead purifications (0.9×), libraries were quantified using the Fragment Analyzer (AATI). For Illumina flow cell production, samples were equimolarly pooled and distributed on all lanes used for 75-bp single-read sequencing on an Illumina NextSeq 500 platform. After sequencing, FastQC (version 0.11.3; http://www.bioinformatics.babraham.ac.uk/) was used to perform a basic quality control check on the sequence data. The reference genome (ASM25323v1) and annotation (v36) for S. venezuelae ATCC 10712 were obtained from Ensembl Bacteria. Reads were mapped to the S. venezuelae genome using GSNAP (v2017-08-15). Further quality control on mapped reads, rRNA content, and coverage of coding genes was performed with RNA-SeQC (version 1.1.8). A table of raw read counts per gene was created based on the overlap of the uniquely mapped reads with the S. venezuelae gene annotation using featureCounts (version 1.5.3). Normalization of the raw read counts based on the library size was performed with the DESeq2 R package (version 1.16.1). Principal-component analysis, sample-to-sample Euclidean distance, and Pearson's and Spearman's correlation coefficients were computed based on the normalized gene expression level. For testing differential expression with DESeq2, the count data were fitted to the negative binomial distribution, and the P values for the statistical significance of the fold change were adjusted for multiple testing with the Benjamini-Hochberg correction for controlling the false-discovery rate, accepting a maximum of 5% false discoveries (adjusted P value ≤ 0.05).

Gus assay.

S. venezuelae strains were grown in liquid MYM medium with shaking for 48 h at 30°C. Cells were harvested by centrifugation, resuspended in lysis buffer (50 mM phosphate buffer [pH 7.0], 0.1% Triton X-100, and 0.27% β-mercaptoethanol), and lysed by sonication. The β-glucuronidase (GusA) activity in the supernatant of the cell lysate was measured as described previously (33, 34) and expressed as Miller units per milligram of protein calculated using the following formula: 1,000 × (OD₄₂₀ − 1.75 × OD₅₅₀)/time of reaction (in minutes) × volume of culture assayed × protein concentration (in milligrams per milliliter). Protein concentration was measured by using the Pierce bicinchoninic acid (BCA) protein assay kit (catalog no. 23225) from Thermo Fisher.