Defining the networks that connect RNase III and RNase J-mediated regulation of primary and specialized metabolism in Streptomyces venezuelae

TL;DR

This study explores how two ribonucleases, RNase III and RNase J, regulate gene expression and metabolism in Streptomyces venezuelae, revealing their roles in development and antibiotic production.

Contribution

The paper identifies distinct and cooperative roles of RNase III and RNase J in regulating gene expression and metabolism in Streptomyces.

Findings

RNase III and RNase J mutants show dysregulation of the BldD regulon and altered expression of cyclic-di-GMP-associated enzymes.

RNase III directly targets PhoP regulon members, linking RNA metabolism to phosphate regulation and antibiotic production.

RNase J targets ribosomal protein transcripts, impacting translation and gene expression dynamics.

Abstract

RNA metabolism involves coordinating RNA synthesis with RNA processing and degradation. Ribonucleases play fundamental roles within the cell, contributing to the cleavage, modification, and degradation of RNA molecules, with these actions ensuring appropriate gene regulation and cellular homeostasis. Here, we employed RNA sequencing to explore the impact of RNase III and RNase J on the transcriptome of Streptomyces venezuelae. Differential expression analysis comparing wild-type and RNase mutant strains at distinct developmental stages revealed significant changes in transcript abundance, particularly in pathways related to multicellular development, nutrient acquisition, and specialized metabolism. Both RNase mutants exhibited dysregulation of the BldD regulon, including altered expression of many cyclic-di-GMP-associated enzymes. We also observed precocious chloramphenicol production in these RNase mutants and found that in the RNase III mutant, this was associated with PhoP-mediated regulation. We further found that RNase III directly targeted members of the PhoP regulon, suggesting a link between RNA metabolism and a regulator that bridges primary and specialized metabolism. We connected RNase J function with translation through the observation that RNase J directly targets multiple ribosomal protein transcripts for degradation. These findings establish distinct but complementary roles for RNase III and RNase J in coordinating the gene expression dynamics critical for S. venezuelae development and specialized metabolism.

RNA processing and metabolism are mediated by ribonucleases and are fundamental processes in all cells. In the morphologically complex and metabolically sophisticated Streptomyces bacteria, RNase III and RNase J influence both development and metabolism through poorly understood mechanisms. Here, we show that both ribonucleases are required for the proper expression of the BldD developmental pathway and contribute to the control of chloramphenicol production, with an interesting connection to phosphate regulation for RNase III. Additionally, we show that both RNases have the potential to impact translation through distinct mechanisms and can function cooperatively in degrading specific transcripts. This study advances our understanding of RNases in Streptomyces biology by providing insight into distinct contributions made by these enzymes and the intriguing interplay between them.

Figures

Global shifts in transcript abundance in RNase III and RNase J mutant strains. (A) Euler plot representing the number of significantly differentially expressed genes shared between RNase III and RNase J mutants during vegetative and fragmentation stages of development. (B) COG analysis for significantly differentially expressed genes, where these genes are sorted into functional categories. The Y-axis is ordered by average abundance across all samples.

Regulators of multicellular development are differentially expressed in RNase mutants. (A) Schematic of the BldD regulon, colored by stage of development where the associated gene product is typically active. Red arrows indicate genes that are significantly upregulated in both the RNase III and RNase J mutant strains, relative to the wild-type. (B) Differential expression of diguanylate cyclases and phosphodiesterases in S. venezuelae. Normalized counts for each strain (averaged across biological replicates) are presented on the left, and Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation. (C) Phenotypic comparisons between wild-type and rnc mutant strains (top left and right, respectively), relative to the rnc mutant strain overexpressing the chaplin/rodlin genes (rsbN mutant; bottom left), and the rnc mutant strain lacking several chaplin and rodlin genes (bottom right).

Dysregulation of nitrogen assimilation and phosphate uptake-related genes in RNase mutants. (A) Normalized counts (averaged across biological replicates) for GlnR regulon members are presented for RNase mutants at three stages of development and for wild type at two stages of development (left). Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation; S, sporulation. (B) Left: normalized counts for the PhoRP two-component system-encoding genes and their associated regulon for each strain (averaged across biological replicates). Right: Log2(fold change) for these counts between the wild-type and either RNase mutant. (C) Schematic illustrating the known crosstalk between GlnR and PhoP. (D) Overlapping GlnR binding sites (orange boxes) and PhoP binding sites (blue line) for the glnA promoter region in S. coelicolor and their conservation in S. venezuelae (putative binding sites in dashed boxes/lines). Non-conserved residues are coloured in red. (E) Fold change was calculated for each biological replicate, where the difference between promoter activity in a strain carrying an empty vector and a strain carrying the phoP phosphomimetic construct was calculated and log2 transformed, for each of the two promoters (glnA and nirB), in wild-type, rnc, and rnj mutant strain backgrounds.

Upregulation of the chloramphenicol biosynthetic cluster is associated with precocious chloramphenicol production in both RNase mutant strains. (A) Chloramphenicol biosynthetic gene expression was assessed in Δrnc and Δrnj mutants relative to the wild-type during vegetative growth. Normalized counts for each strain (averaged across biological replicates) are presented on the left, and the Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation. (B) Quantification of chloramphenicol production levels in wild-type (carrying an empty vector [EV] as a control), compared with RNase mutants carrying either an EV or vector with an overexpression phoP phosphomimetic variant (+phoP).

The PhoU transcript is directly targeted by RNase III. (A) Reads aligning to the phoRP and phoU region from the wild-type and RNase III mutant samples (during fragmentation), including a pileup of reads aligning to the 3′ UTR of phoU (black arrow), and a low level of antisense reads aligning to the phoR gene in the RNase III mutant background (red arrow). (B) Schematic for the transcriptionally and translationally acting inducible system. The addition of theophylline relieves translational repression, where the ribosome binding site upstream of tipA becomes accessible. The subsequent addition of thiostrepton increases the affinity of TipA for its own promoter sequence, which is upstream of rnc fused to a 3×FLAG tag N-terminal sequence. (C) Reads aligning to phoU, before and after RNase III induction. Normalized depth of coverage per nucleotide is plotted over the phoU coding region and into the 3′ UTR in the experimental sample (carrying the 3×FLAG tagged RNase III; left) and the control sample (lacking the inducible RNase III construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here). (D) The phoU transcript has a highly structured 3′ UTR based on RNAfold predictions (right), and this region is associated with lower entropy compared with the coding sequence (left). (E) Schematic indicating a role for PhoU in a negative feedback loop within the PhoP regulon, where it represses the expression of several PhoP-activated genes, including the pst operon.

RNase J is involved in regulating translational machinery. vnz_18105 (encoding 30S ribosomal protein S6) is a likely RNase J target gene. Normalized depth of coverage per nucleotide is plotted over the vnz_18105 coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here).

RNase III and RNase J may cooperatively degrade specific transcripts. (A) Schematic for mechanism of RNase III and RNase J cooperative degradation. Initial endoribonucleolytic cleavage by RNase III yields a 5′ monophosphorylated transcript, which is preferentially recognized by RNase J for exoribonucleolytic degradation. (B) Reads aligning to rnc are presented as normalized depth of coverage per nucleotide plotted over the rnc coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here). (C) Reads aligning to the rnc coding region from the wild-type and RNase J mutant samples (during fragmentation). The red arrow indicates a predicted internal cleavage site that could then serve as a substrate for RNase J (given the read pileup aligning downstream of this site in the RNase J mutant but not wild-type). (D) Reads are aligned to rspO and pnp, with normalized depth of coverage per nucleotide plotted over the rspO-pnp coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right).

Tables

TABLE 1. Genes whose transcripts are directly targeted by RNase J post-induction

Gene name and associated product	Fold change pre- to 40 min
Ribosomal protein-encoding genes
30S ribosomal protein S4 (vnz_05370)	−9.2
50S ribosomal protein L20 (vnz_05835)	−14.2
30S ribosomal protein S1 (vnz_07970)	−3.8
30S ribosomal protein S20 (vnz_11455)	−9.3
30S ribosomal protein S6 (vnz_18105)	−15.9
50S ribosomal protein L11 (vnz_21445)	−9.3
30S ribosomal protein S12 (vnz_21600)	−14.5
30S ribosomal protein S17 (vnz_21685)	−31.3
30S ribosomal protein S11 (vnz_21775)	−10.1
50S ribosomal protein L13 (vnz_21805)	−16.5
50S ribosomal protein L32 (vnz_26035)	−3.8
50S ribosomal protein L19 (vnz_26165)	−9.3
30S ribosomal protein S2 (vnz_26230)	−4.9
30S ribosomal protein S15 (rpsO; vnz_26635)	−9.6
Other targets
Elongation factor P (vnz_05295)	−4.5
Peptidylprolyl isomerase (vnz_06015)	−3.0
Hypothetical protein (vnz_07005)	−10.4
Leucine tRNA ligase (vnz_11510)	−2.3
Hypothetical protein (vnz_15735)	−16.0
Aspartate tRNA ligase (vnz_17530)	−2.8
Translation initiation factor IF-1 (vnz_21760)	−6.1
Hypothetical protein (vnz_24805)	−3.6
Acetolactate synthase large subunit (vnz_25640)	−2.1
Hypothetical protein (vnz_26030)	−7.9
RNase III (rnc; vnz_26040)	−3.3
Translation elongation factor Ts (vnz_26235)	−4.6
Hypothetical protein (vnz_26610)	−3.2
PNPase (pnp; vnz_26440)	−2.4

Equations

Full text

INTRODUCTION

Post-transcriptional regulation is a critical component of cellular control in bacteria. The constant turnover of RNA can facilitate rapid changes in gene expression, and the stability of a given transcript is largely governed by its accessibility to ribonucleases (RNases) (1). RNases can degrade RNA from within a transcript (endoribonucleases) or from a terminal end (exoribonuclease). In many bacteria, RNA turnover is mediated by degradosome complexes (2–5). The degradosome has been best studied in the Gram-negative bacterium Escherichia coli, where the complex is assembled around the large endoribonuclease RNase E (2, 6). In this complex, RNase E associates with the exoribonuclease polynucleotide phosphorylase (PNPase), the glycolytic enzyme enolase, and the DEAD-box RNA helicase RhlB. RNA degradation in Gram-positive bacteria involves a different suite of RNases, as most Gram-positive bacteria do not encode RNase E homologs. In Bacillus subtilis, the membrane-anchored endoribonuclease RNase Y is the primary decay-initiating enzyme (7–11). It has been suggested that RNase Y makes transient interactions with several “degradosome-like” proteins (12–15), in addition to scaffolding a “Y-complex” that includes YlbF, YmcA, and YaaT (16–21). RNase E and RNase Y both preferentially cleave 5ʹ monophosphorylated RNA molecules (5, 9). Thus, RNA decay is typically initiated by a preliminary endonucleolytic cleavage event that generates monophosphorylated 5ʹ ends or by a pyrophosphohydrolase (e.g., RppH) that converts 5ʹ triphosphates to monophosphates (5, 9). In addition to a suite of diverse single-strand-specific RNases, most bacteria also encode the double-stranded RNA-specific RNase III (22). This enzyme is highly conserved across bacteria and eukaryotes, and its capacity to cleave double-stranded RNA means it contributes to the metabolism and/or processing of both mRNAs bound by non-coding small RNAs and highly structured RNAs like ribosomal RNAs (rRNAs) (5, 22–24).

As investigations into RNase activity in bacteria have extended beyond the E. coli and B. subtilis model systems, it has become evident that there is diversity in the number and type of RNases employed by any given bacterium (2, 25–30). The actinobacteria are a phylum of Gram-positive bacteria that includes Streptomyces and Mycobacterium. These bacteria encode a set of RNases that includes RNase III, RNase J, and contrary to what is observed in most other Gram-positive genera, RNase E (31). Streptomyces have been studied extensively as a model of multicellular growth in bacteria and for their production of medically relevant specialized metabolites. The Streptomyces classical life cycle is defined by morphologically and metabolically distinct developmental stages (32). The cycle begins with spore germination, which involves the emergence of one or two hyphal filaments that grow via tip extension and branching to form a dense hyphal network. On solid media, reproductive growth initiates with the raising of aerial hyphae and culminates with the subdivision of the aerial cells into chains of unigenomic spores. In Streptomyces species that can differentiate in a liquid medium (e.g., Streptomyces venezuelae), the growth stage that is analogous to aerial development is termed “fragmentation,” where the vegetative hyphae break apart before forming short chains of spores. It is during aerial development/fragmentation that Streptomyces typically initiate specialized metabolism. The cooperative regulation of multicellular development and specialized metabolism has been extensively studied in Streptomyces, but the contribution of RNases to the regulation of these processes has not been studied in detail.

We have previously observed that deleting rnc (RNase III) or rnj (RNase J) from S. venezuelae affects development, antibiotic production, normal sporulation, and ribosome assembly (33). The loss of these enzymes leads to decreased production of the specialized metabolite jadomycin and the accumulation of 100S inactive ribosomes. Deleting rnc also causes the growing colony to peel away from the solid medium, while the Δrnj mutant develops more slowly in a liquid medium than its wild-type parent and is delayed in sporulation (33). While the phenotypes of the rnc and rnj mutants are clear in S. venezuelae, the pathways by which their associated RNases exert their effects remain unknown. Here, we took a transcriptomic approach to understanding the changes in gene expression and transcript abundance over time within these two mutants. We observed that regulatory networks governing the progression of multicellular development, nitrogen assimilation, phosphate uptake, and specialized metabolism were significantly differentially expressed. The altered transcription profiles observed were similar for both mutants and were significantly different compared with their wild-type parent strain; however, these parallel transcriptional changes yielded different phenotypic outcomes for the two mutant strains. We determined that RNase III, but not RNase J, likely contributes directly to the control of phosphate uptake in S. venezuelae. Our data also suggest that RNase III and RNase J affect the biogenesis of the translational machinery in distinct ways. Overall, our work emphasizes the complexity of RNase-based regulation and provides insights into how these enzymes affect the temporal regulation of development and specialized metabolism in Streptomyces.

RESULTS

RNA-sequencing analysis reveals a global impact of rnc and rnj deletion

To better understand how RNase III and RNase J affect S. venezuelae development and specialized metabolism, we undertook a transcriptomic approach. We isolated RNA from wild-type and RNase mutant strains at distinct stages of classical development in a liquid medium (vegetative growth, hyphal fragmentation, and—for the mutant strains—sporulation; we were unable to successfully extract intact RNA from sporulating wild-type samples). We subjected the vegetative and hyphal fragmentation samples to differential expression analysis, comparing mutant and wild-type strains. We also analyzed changes in the transcriptome of each mutant over time to assess trends in gene expression across developmental stages. For our differential expression analysis, we focused our attention on genes that were reasonably well expressed (baseMean >50, representing the normalized average expression level of a gene across all samples) and that had significantly altered transcript abundance when comparing the wild-type strain to either mutant (log_2_ fold change > |2|, with an adjusted P-value < 0.05).

When considering comparisons between wild-type and RNase mutants during vegetative growth and hyphal fragmentation, we found 739 unique genes had significantly altered transcript levels in the Δrnc mutant relative to the wild-type, representing 9.8% of all annotated genes. In the Δrnj mutant, 688 unique genes were significantly changed, representing 9.2% of all annotated genes. The Δrnc and Δrnj mutants shared 367 differentially expressed genes, which included 242 genes during vegetative growth, 171 genes during hyphal fragmentation, and 46 genes represented at both stages (Fig. 1A). Intriguingly, we noted that these 367 shared genes tended to be similarly affected (either upregulated or downregulated in the same way, relative to wild-type). In assessing the biological functions impacted by the loss of these RNases, we sorted all significantly differentially expressed genes into functional categories defined by the Clusters of Orthologous Genes (COG) database (34). For both the Δrnc and Δrnj mutants, among the COG categories that were most highly represented were those involved in transcription and signal transduction mechanisms (Fig. 1B). This could, in part, explain the global shift in transcript abundance observed for these two mutants relative to their wild-type parent, as altered levels of key transcriptional regulators could affect the transcript abundance of their associated regulons. We, therefore, sought to identify regulatory networks of interest within the signal transduction and transcription COG categories, where changing the abundance of a few members could be expected to have broad transcriptional effects.

Global shifts in transcript abundance in RNase III and RNase J mutant strains. (A) Euler plot representing the number of significantly differentially expressed genes shared between RNase III and RNase J mutants during vegetative and fragmentation stages of development. (B) COG analysis for significantly differentially expressed genes, where these genes are sorted into functional categories. The Y-axis is ordered by average abundance across all samples.

Regulators of multicellular development are differentially expressed in RNase mutants

RNases have been previously implicated in the regulation of morphological differentiation in Streptomyces. For example, AdpA is a transcriptional regulator that controls the expression of hundreds of genes involved in reproductive growth. In Streptomyces coelicolor, the adpA transcript is significantly upregulated in an RNase III mutant and is cleaved by this enzyme in vitro (35). Contrary to what was observed in S. coelicolor, we found that adpA transcript levels in S. venezuelae were less abundant in the Δrnc mutant compared with the wild-type during vegetative growth and hyphal fragmentation (although these differences were not statistically significant), suggesting that adpA is unlikely to be a direct target of RNase III in S. venezuelae. This suggests that RNase III targets may differ between species.

While we observed no significant change in adpA transcript levels, additional analyses revealed that many other developmental genes were significantly differentially expressed in both the ∆rnc and ∆rnj mutants. For example, we found that bldM and bldN were highly upregulated during vegetative growth with respect to the wild-type. These are both known targets of the master regulator BldD (36), and indeed, most members of the BldD regulon were upregulated during vegetative growth in both mutant strains, although the expression of bldD itself was not significantly altered (Fig. 2A). BldD activity is controlled by cyclic-di-GMP (c-di-GMP), where this molecule promotes BldD dimerization, enabling DNA binding and target gene repression (37–39). Since we observed changes in the BldD regulon but not bldD levels, we therefore asked whether any diguanylate cyclase- (responsible for c-di-GMP synthesis) or phosphodiesterase- (responsible for c-di-GMP degradation) encoding genes were differentially expressed in either the ∆rnc or ∆rnj mutant strains. We found that many of these genes had altered transcript abundances (Fig. 2B). During vegetative growth, the diguanylate cyclase-encoding genes cdgE and cdgC were upregulated in the ∆rnc and ∆rnj mutants, respectively. During hyphal fragmentation, the diguanylate cyclase-encoding gene cdgC and the phosphodiesterase-encoding gene rmdB were downregulated in both mutant strains, whereas cdgF and vnz_36075 (encoding a bifunctional diguanylate cyclase/phosphodiesterase and putative diguanylate cyclase, respectively) were upregulated in both mutant strains. cdgE and cdgD were also upregulated during hyphal fragmentation in the ∆rnc and ∆rnj mutants, respectively (40). This collectively suggested that c-di-GMP pools may be altered in the RNase mutant strains. The observed upregulation of BldD regulon members suggests that this repressor is less active, implying that there may be reduced concentrations of c-di-GMP available (either globally or locally) to promote its dimerization and repressive function.

Regulators of multicellular development are differentially expressed in RNase mutants. (A) Schematic of the BldD regulon, colored by stage of development where the associated gene product is typically active. Red arrows indicate genes that are significantly upregulated in both the RNase III and RNase J mutant strains, relative to the wild-type. (B) Differential expression of diguanylate cyclases and phosphodiesterases in S. venezuelae. Normalized counts for each strain (averaged across biological replicates) are presented on the left, and Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation. (C) Phenotypic comparisons between wild-type and rnc mutant strains (top left and right, respectively), relative to the rnc mutant strain overexpressing the chaplin/rodlin genes (rsbN mutant; bottom left), and the rnc mutant strain lacking several chaplin and rodlin genes (bottom right).

Included amongst the differentially expressed genes within the (indirect) BldD regulon were the rodlin and chaplin genes, which were highly upregulated in both ∆rnc and ∆rnj mutant strains. The function of the rodlin and chaplin proteins is to coat the surface of the aerial hyphae, where they confer these hyphae with hydrophobic properties that promote their escape of surface tension and enable their extension into the air (41–46); these genes are under the direct control of sigma BldN (σ^BldN^). The precocious upregulation of these morphogenetic protein-encoding genes during vegetative growth did not lead to premature reproductive growth for either the ∆rnc or ∆rnj mutants. Instead, we observed that the ∆rnj mutant was significantly delayed in sporulation. We did, however, note that the rnc mutant had a tendency to “peel” away from the growth substrate on solid medium, and we wondered if this phenotype may be a result of enhanced chaplin/rodlin production, given the surfactant capabilities of both these protein family members. To investigate this possibility, we generated two S. venezuelae strains: one where the rodlins and chaplins were expected to be overexpressed and another where most of the chaplins and rodlins were deleted in the ∆rnc mutant background. To achieve overexpression, we generated an ∆rsbN mutant, where rsbN encodes the σ^BldN^-specific anti-sigma factor. rsbN deletion leads to hyperactivity of σ^BldN^, which in turn increases the expression of the rodlin and chaplin genes (47). In parallel, we generated a ∆rnc∆chpDFG∆rdlAB mutant strain, where many of the rodlin (two of three) and chaplin (three of seven) genes were deleted in the ∆rnc mutant background. We had predicted that if the rodlins and chaplins were responsible for the rnc peeling phenotype, then the “peeling” aspect of the ∆rnc mutant phenotype would be recapitulated in the ∆rsbN strain and resolved in the ∆rnc∆chpDFG∆rdlAB strain. Instead, we found that deleting select rodlin and chaplin genes did not alter the peeling characteristics of the ∆rnc mutant phenotype (Fig. 2C), nor was an equivalent peeling phenotype seen for the chaplin/rodlin overexpressing ∆rsbN strain. These observations suggested that this phenotype was unlikely to be the result of precocious rodlin and chaplin production.

Regulators of nutrient acquisition are differentially expressed in RNase mutants

Given that the initiation of Streptomyces’s reproductive growth is thought to occur in response to nutrient starvation, we considered whether the RNase mutants may be dysregulated in their nutrient uptake or metabolism. In looking at different nutrient acquisition pathways, we found that many genes involved in nitrogen assimilation and phosphate uptake were differentially expressed in both the ∆rnc and ∆rnj mutant strains relative to wild-type.

Genes related to nitrogen assimilation were upregulated during reproductive growth, particularly in the ∆rnj mutant. Notably, nearly all these differentially affected genes were part of the GlnR regulon (Fig. 3A). Under nitrogen-limited growth conditions, GlnR activates the expression of genes involved in nitrogen assimilation, including those encoding nitrate and ammonium transporters, assimilatory nitrate/nitrite reductases, and glutamine synthetases (48, 49). Comparing transcript abundance within each mutant over time showed that these higher levels of gene expression continued into the sporulation phase for both ∆rnc and ∆rnj mutants (Fig. 3A). These increased transcript levels were unlikely to be the result of increased glnR expression, as its transcript levels did not change significantly in either mutant relative to the wild-type, at any time point (Fig. 3A). One possible explanation could be due to lack of competition with PhoP. In contrast to the nitrogen-associated regulon, the genes encoding the PhoRP two-component system were downregulated during vegetative growth (Fig. 3B). This system responds to phosphate limitation by increasing the expression of genes involved in phosphate uptake (50–53). Under low phosphate growth conditions, the PhoR kinase phosphorylates and activates the PhoP response regulator. We observed that several PhoP target genes were expressed at lower levels in the RNase mutants compared with the wild-type during vegetative growth (and fragmentation for the rnj mutant), including the pst operon, which encodes a phosphate ABC transporter; phoD, which encodes a secreted alkaline phosphatase; and phoU, which encodes a phosphate transport system regulatory protein (Fig. 3B).

Dysregulation of nitrogen assimilation and phosphate uptake-related genes in RNase mutants. (A) Normalized counts (averaged across biological replicates) for GlnR regulon members are presented for RNase mutants at three stages of development and for wild type at two stages of development (left). Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation; S, sporulation. (B) Left: normalized counts for the PhoRP two-component system-encoding genes and their associated regulon for each strain (averaged across biological replicates). Right: Log2(fold change) for these counts between the wild-type and either RNase mutant. (C) Schematic illustrating the known crosstalk between GlnR and PhoP. (D) Overlapping GlnR binding sites (orange boxes) and PhoP binding sites (blue line) for the glnA promoter region in S. coelicolor and their conservation in S. venezuelae (putative binding sites in dashed boxes/lines). Non-conserved residues are coloured in red. (E) Fold change was calculated for each biological replicate, where the difference between promoter activity in a strain carrying an empty vector and a strain carrying the phoP phosphomimetic construct was calculated and log2 transformed, for each of the two promoters (glnA and nirB), in wild-type, rnc, and rnj mutant strain backgrounds.

The phosphate uptake and the nitrogen assimilation regulatory networks are interconnected, with PhoP competing with GlnR for binding to the promoters of several nitrogen assimilation genes (Fig. 3C) (52, 54, 55). The regulatory effects of GlnR and PhoP on the nitrogen assimilation genes have also been reported to be inversely correlated, with GlnR activating their expression and PhoP repressing them. In S. coelicolor, shared GlnR/PhoP target promoters include glnR, glnA, and glnII (glutamine synthetases), as well as amtB (ammonium transporter) (55). If PhoP was less abundant in the S. venezuelae RNase mutants, as suggested by the RNA-sequencing data, then GlnR may have less competition in binding their shared targets, leading to a concomitant increase in transcription of these shared target genes.

We, therefore, sought to investigate the contribution of phoRP downregulation to the ∆rnc and ∆rnj mutant transcriptomes. We reasoned that if reduced competition between GlnR and PhoP was responsible for the upregulation of genes (e.g., glnII, glnA, and amtB), then overexpressing phoP should return transcript levels of these genes in the mutants to near wild-type levels. To test this, we generated a phoP phosphomimetic overexpression construct (53, 56) and introduced it into ∆rnc, ∆rnj, and wild-type S. venezuelae strains. To monitor the expression of nitrogen-assimilation-related genes, we employed a GusA reporter system and introduced the reporter constructs into our phoP overexpression and control strains. We monitored the activity of the nirB promoter as a negative control (not reported to be co-regulated by GlnR and PhoP) and the glnA promoter as our candidate regulon member, as it is targeted by both PhoP and GlnR in S. coelicolor (55) and has conserved binding sites in S. venezuelae (Fig. 3D). We measured the promoter activity of nirB and glnA in strains with an empty plasmid control and compared this to the same strain with the phoP phosphomimetic overexpression plasmid. As expected, nirB reporter activity was unaffected by phoP overexpression (Fig. 3E). Conversely, glnA promoter activity was reproducibly reduced in the ∆rnc mutant carrying the phoP phosphomimetic overexpression construct compared with its control strain (empty plasmid-carrying ∆rnc mutant) (Fig. 3E). The phoP overexpression construct had no effect; however, on glnA promoter activity in the ∆rnj mutant or wild-type strains. This suggested that the ∆rnc mutant may be more sensitive to changes in PhoP-mediated regulation than either the wild-type or the ∆rnj mutant.

PhoP contributes to precocious chloramphenicol production in ∆rnc but not ∆rnj

Beyond the effect of PhoP on nitrogen metabolism, there can also be significant crosstalk between PhoP and specialized metabolic regulators. This has been seen in S. coelicolor, where PhoP and AfsR compete for binding to the afsS promoter, and AfsS, in turn, impacts the expression of the actinorhodin and undecylprodigiosin biosynthetic clusters (57). Given the disruptions we found in the PhoRP regulon in our RNase mutants, we sought to assess whether specialized metabolic gene expression in the ∆rnc and ∆rnj mutant strains differed from that of the wild-type.

In both RNase mutant strains, we identified genes having altered transcript abundance in five biosynthetic gene clusters. These clusters were predicted to direct the production of chloramphenicol, melanin, an uncharacterized lanthipeptide, an unknown non-ribosomal peptide, and a terpene (based on antiSMASH detection [58]). For the affected genes in each cluster, precocious expression was seen in the two RNase mutant strains, with transcript levels being significantly higher than in the wild-type during vegetative growth. We set out to test if the increased expression of biosynthetic gene clusters also resulted in increased cluster product levels, and if so, whether this increased specialized metabolite production stemmed from phoRP downregulation. For this, we focused on the chloramphenicol biosynthetic cluster due to the strong upregulation of its genes in both mutants (Fig. 4A) and its well-characterized biosynthetic pathway (59–62). We quantified chloramphenicol abundance during vegetative growth in the wild-type, ∆rnc mutant, and ∆rnj mutant strains of S. venezuelae, with each carrying either our phoP phosphomimetic overexpression construct or an empty vector as a control. We found that both ∆rnc and ∆rnj mutants produced significantly higher levels of chloramphenicol compared with the wild-type, and that chloramphenicol levels were higher in the ∆rnj mutant than the ∆rnc mutant (Fig. 4B). We further observed that in the ∆rnc mutant strain, phoP overexpression reduced chloramphenicol production to roughly wild-type levels, indicating that phoRP downregulation likely contributed to the increased antibiotic production observed in the ∆rnc mutant strain. Our findings here mirrored our observation that PhoP affected glnA promoter activity in the ∆rnc mutant but not the ∆rnj mutant. Indeed, overexpression of phosphomimetic PhoP did not significantly impact chloramphenicol levels in the ∆rnj mutant relative to its control strain (Fig. 4B). While there was a trend observed for less chloramphenicol production in the ∆rnj mutant following phoP overexpression, this reduction did not approach wild-type levels; therefore, we could not attribute the increased antibiotic production in the ∆rnj mutant solely to phoP downregulation. It seems that while the ∆rnc and ∆rnj mutants have similar transcriptional and metabolic profiles, different mechanisms appear to drive the observed effects.

Upregulation of the chloramphenicol biosynthetic cluster is associated with precocious chloramphenicol production in both RNase mutant strains. (A) Chloramphenicol biosynthetic gene expression was assessed in Δrnc and Δrnj mutants relative to the wild-type during vegetative growth. Normalized counts for each strain (averaged across biological replicates) are presented on the left, and the Log2(fold change) for these counts between the wild-type and either RNase mutant is presented on the right. V, vegetative; F, fragmentation. (B) Quantification of chloramphenicol production levels in wild-type (carrying an empty vector [EV] as a control), compared with RNase mutants carrying either an EV or vector with an overexpression phoP phosphomimetic variant (+phoP).

phoU is directly targeted by RNase III

Having established that multiple aspects of the ∆rnc mutant phenotype were due to reduced expression of the phoRP genes, we next wanted to better understand the connection between RNase III and this operon. In E. coli, the phoP transcript can be bound by a small, *cis-*encoded antisense RNA, with the resulting double-stranded complex being targeted for degradation by RNase III (63). In S. venezuelae, we noted low-level antisense reads mapping to the phoRP operon (Fig. 5A; red arrow), and so we wondered if phoP may be targeted by RNase III in S. venezuelae in a similar way as in E. coli.

The PhoU transcript is directly targeted by RNase III. (A) Reads aligning to the phoRP and phoU region from the wild-type and RNase III mutant samples (during fragmentation), including a pileup of reads aligning to the 3′ UTR of phoU (black arrow), and a low level of antisense reads aligning to the phoR gene in the RNase III mutant background (red arrow). (B) Schematic for the transcriptionally and translationally acting inducible system. The addition of theophylline relieves translational repression, where the ribosome binding site upstream of tipA becomes accessible. The subsequent addition of thiostrepton increases the affinity of TipA for its own promoter sequence, which is upstream of rnc fused to a 3×FLAG tag N-terminal sequence. (C) Reads aligning to phoU, before and after RNase III induction. Normalized depth of coverage per nucleotide is plotted over the phoU coding region and into the 3′ UTR in the experimental sample (carrying the 3×FLAG tagged RNase III; left) and the control sample (lacking the inducible RNase III construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here). (D) The phoU transcript has a highly structured 3′ UTR based on RNAfold predictions (right), and this region is associated with lower entropy compared with the coding sequence (left). (E) Schematic indicating a role for PhoU in a negative feedback loop within the PhoP regulon, where it represses the expression of several PhoP-activated genes, including the pst operon.

To identify transcripts that were directly targeted for degradation by RNase III, we designed an inducible system that allowed us to control rnc expression, reasoning that direct targets would be degraded more rapidly than indirect targets. To minimize leaky expression from our uninduced construct, we employed a two-tiered inducible system, encompassing a translationally-acting theophylline-responsive riboswitch controlling the tipA gene and a thiostrepton-inducible promoter that could be induced by thiostrepton-bound TipA (Fig. 5B). We then cloned a FLAG-tagged rnc gene downstream of the thiostrepton-inducible promoter, introduced the resulting construct into the rnc mutant strain, and confirmed robust 3×FLAG-RNase III production after adding the theophylline and thiostrepton inducers (Fig. S1A). We collected cells prior to induction and at 20 and 40 min post-induction (when we observed strong induction of RNase III; Fig. S1A), and isolated RNA from these samples, alongside a ∆rnc strain bearing an empty inducible vector to control for the effects of theophylline and thiostrepton addition. These samples were then subjected to RNA-sequencing. We focused on genes that were both significantly downregulated in the induced samples relative to the uninduced and whose transcript levels were unaffected by inducer addition in the control strain. While we did not see any significant change in phoP levels (or those of its associated antisense RNA), of the RNase III targets we identified, phoU was of particular interest. Its transcript levels decreased dramatically post-RNase III induction and were generally unchanged in the control when comparing uninduced and induced samples (Fig. 5C). In S. venezuelae, phoU is expressed divergently from the phoRP operon (Fig. 5A; bottom). We observed that its transcription continued beyond the end of the phoU coding sequence, yielding a long 3′ UTR. A segment of this UTR was more abundant in the wild-type compared with the ∆rnc mutant, suggesting that this may be a stable product that accumulates following RNase III cleavage (Fig. 5A; black arrow). Whether this product has any function in the cell remains to be determined. In analyzing the structure of the phoU transcript using the RNAfold program (64), we found the 3′ UTR was predicted to be highly structured (Fig. 5D), suggesting that the many predicted stem-loops could serve as substrates for direct RNase III-mediated cleavage.

Ribosomal protein transcripts are directly targeted by RNase J

While the PhoP-mediated effects appeared to be specific to RNase III and not RNase J, we remained intrigued by the congruence of the transcription profiles for both ∆rnc and ∆rnj mutant strains. We wondered what transcripts were directly targeted by RNase J and whether these could provide insight into the affected pathways that differ when compared with the ∆rnc mutant (e.g., chloramphenicol biosynthesis being upregulated in both mutants, but via apparently different pathways). After having confirmed that a 3×FLAG-RNase J expressed from its native promoter could complement the sporulation defect of the mutant strain, we generated an inducible construct using the theophylline/thiostrepton-inducible system described above and introduced this (alongside an empty plasmid control) into the ∆rnj mutant in S. venezuelae. We first confirmed that the 3×FLAG-RNase J was effectively induced (Fig. S1B). We then conducted an equivalent induction experiment as described above for RNase III. In focusing on transcripts that decreased post-induction in the RNase J experimental samples but not in the control, we found no target transcripts that could easily explain the ∆rnj mutant phenotypes described so far. Considering the list of RNase J target transcripts, it was instead immediately striking that half of them (14/28) encoded ribosomal proteins (Table 1). These transcripts all decreased dramatically (between 3- and 30-fold) post-RNase J induction compared to the control samples, where no change in abundance was detected (Table 1; Fig. 6; Fig. S3). Notably, this effect was not seen following RNase III induction, indicating that it is specific to RNase J. Ribosomal proteins are generally understood to autogenously regulate their own mRNA transcripts by binding to structured regions in the associated 5′ UTRs (65, 66). Our observations here suggest a potentially novel mechanism for ribosomal protein control, whereby RNase J functions to process these transcripts.

RNase J is involved in regulating translational machinery. vnz_18105 (encoding 30S ribosomal protein S6) is a likely RNase J target gene. Normalized depth of coverage per nucleotide is plotted over the vnz_18105 coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here).

Co-regulation of specific transcripts by RNase III and RNase J

Reflecting on the role that RNase J may have in degrading ribosomal protein mRNA transcripts, we noted that this was another pathway where there was congruence between RNase III and RNase J, having previously shown that RNase III was involved in 16S rRNA maturation (33). Beyond the effects of these ribonucleases on discrete ribosome components, we also observed that ssrA (a transfer-messenger RNA [tmRNA]) was upregulated in the ∆rnc mutant, although not in the ∆rnj mutant (Tables S1 to S4). tmRNA molecules function to rescue stalled ribosomes (67). The increase in ssrA transcripts could suggest that these messages are direct targets of RNase III (although they were not identified in our RNase III induction experiments), and/or it could be the result of an RNase III-specific cellular stress response, given that studies have shown tmRNA levels to increase in abundance in response to diverse stresses in Streptomyces (68–70). We propose that the shared transcriptional profiles observed for our ∆rnc and ∆rnj mutant strains, which were markedly distinct from that of the wild-type strain, may at least in part stem from (independent) defects in translation, given the close coupling of transcription and translation in many bacteria (71–77).

An alternative/additional explanation for the similar transcriptomic shifts observed in the ∆rnc and ∆rnj mutant strains could be that some transcripts are directly targeted by both RNase III and RNase J. Mechanistically, there is the potential for these ribonucleases to cooperatively degrade target transcripts: RNase J preferentially recognizes and degrades transcripts having a 5′ monophosphate, which can arise through RNase III cleavage (Fig. 7A). There is at least one previously reported instance of their coordinated activity in Staphylococcus aureus, where these ribonucleases act to cooperatively cleave the methionine biosynthesis T-box riboswitch (78). Given this precedent, we sought to investigate whether any target transcripts were direct targets for degradation by both RNase III and RNase J in S. venezuelae.

RNase III and RNase J may cooperatively degrade specific transcripts. (A) Schematic for mechanism of RNase III and RNase J cooperative degradation. Initial endoribonucleolytic cleavage by RNase III yields a 5′ monophosphorylated transcript, which is preferentially recognized by RNase J for exoribonucleolytic degradation. (B) Reads aligning to rnc are presented as normalized depth of coverage per nucleotide plotted over the rnc coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right). Depth of coverage/nucleotide was normalized by library size, and a loess regression trendline is presented (span = 0.1 was used to account for high/nucleotide variability; representative biological replicate is presented here). (C) Reads aligning to the rnc coding region from the wild-type and RNase J mutant samples (during fragmentation). The red arrow indicates a predicted internal cleavage site that could then serve as a substrate for RNase J (given the read pileup aligning downstream of this site in the RNase J mutant but not wild-type). (D) Reads are aligned to rspO and pnp, with normalized depth of coverage per nucleotide plotted over the rspO-pnp coding region in the experimental sample (carrying the 3×FLAG tagged RNase J; left) and the control sample (lacking the inducible RNase J construct; right).

Our RNase III- and RNase J-induction experiments did not identify any shared target transcripts; these experiments may have been confounded by an inability to eliminate leaky expression from our inducible construct. We did, however, observe that two previously described RNase III targets (79, 80) appeared to be directly cleaved by RNase J. In many bacteria, including E. coli and S. coelicolor, RNase III exerts autoregulatory control by cleaving a stem-loop structure within its rnc transcript (80). We found that the rnc coding region was rapidly degraded in S. venezuelae in our RNase J-induced samples (Fig. 7B). Our original RNA sequencing data also revealed that the 3′ half of the rnc transcript increased in abundance in the ∆rnj mutant, suggesting that this enzyme may function to degrade the truncated transcript after internal RNase III-mediated cleavage (Fig. 7C). It has also been reported previously in S. coelicolor that RNase III can degrade the rspO-pnp transcript (22, 79). This operon encodes a ribosomal protein and a polynucleotide phosphorylase (PNPase) that has both exoribonuclease and RNA 3′-polyribonucleotide polymerase activity. The rspO-pnp operon was not identified as an RNase III target in our induced experiment, but it was upregulated in the ∆rnc mutant strain to levels that were just below our significance threshold. This region too was identified as a direct target of RNase J (Fig. 7D), suggesting that this operon may also be under the cooperative control of RNase III and RNase J.

DISCUSSION

The effect of RNase III and RNase J on Streptomyces development

Our transcriptional analyses of the ∆rnc and ∆rnj mutant strains revealed that these enzymes had important but distinct roles in regulating morphological development and metabolism. These mutants shared similar gene expression trends when compared with the wild-type, particularly concerning the dysregulation of genes whose products were involved in reproductive growth and specialized metabolism. Our work suggested that the master regulator BldD was less active in the ∆rnc and ∆rnj mutant strains, given the strong upregulation of the BldD regulon. While there was differential expression observed for multiple diguanylate cyclases (c-di-GMP “makers”) and phosphodiesterases (c-di-GMP “breakers”), the redundancy of these enzymes makes it challenging to predict the net effect of these changes on global c-di-GMP pools. There is, however, increasing evidence suggesting that c-di-GMP signaling frequently functions on a more local scale (81), and thus the change in expression seen for the c-di-GMP synthases and phosphodiesterases here may serve to alter the local pools of c-di-GMP relative to wild-type, with those impacting BldD appearing to be amongst the most profoundly affected.

Our transcriptional data also suggested that, unlike in S. coelicolor, adpA was not a direct target of RNase III in S. venezuelae, at least under the conditions tested here. The precise site of RNase III cleavage has not been mapped for the adpA transcript in S. coelicolor, but in vitro experiments in which adpA transcripts were incubated with purified RNase III yielded multiple products (35), suggesting that there may be several RNase III cleavage sites within the adpA transcript. RNase activity can be impacted by both the sequence and structure of its target transcripts and by target accessibility within the cell (e.g., protection by translating ribosomes). The difference in adpA targeting between these two Streptomyces species suggests that either the S. venezuelae transcript adopts an alternative structure compared with its S. coelicolor counterpart and/or is subject to different translational or accessibility controls.

Connections between RNase III and phosphate uptake

The phoRP operon was among the most significantly differentially expressed genes in the ∆rnc and ∆rnj mutant strains, relative to wild-type. In Streptomyces species, there is known crosstalk between the PhoRP and the GlnR regulons (82–85), a phenomenon that was also observed here. In probing this connection, we noted that the ∆rnc mutant was more responsive to changes in PhoP levels than the ∆rnj mutant, with both chloramphenicol abundance and glnA expression being impacted by altered PhoP levels in the RNase III mutant strain. These observations may be explained in part by the direct targeting of phoU transcripts by RNase III.

In S. coelicolor, PhoU represses several PhoP-activated genes, including glpQ1 and pstS (86). Given that phoU transcription is activated by PhoP binding to the phoU-phoRP intergenic region, this is thought to create a negative feedback loop that can dampen the otherwise strong positive effect of phoRP expression (i.e., genes that are initially upregulated by PhoP are then repressed by PhoU; Fig. 5E) (86). While phoRP and the associated regulon were downregulated in both mutant strains during vegetative growth, we observed that phoU was upregulated in the ∆rnc mutant relative to wild-type during reproductive growth, which would be consistent with phoU being a direct target of RNase III (Fig. 3B). No upregulation was observed in the ∆rnj mutant at this same developmental stage, and the connection between phoRP downregulation and RNase J activity remains unclear. These data suggest that despite observing similar shifts in transcript abundance following the deletion of either rnc or rnj in S. venezuelae, there seem to be fundamental differences in the regulatory networks that contribute to these shared effects.

Beyond our observations for the phoRP operon and the connection between RNase III and phoU, we did not always see congruence between our different transcriptomic experiments (e.g., developmental time course versus inducible time course). We had expected that transcripts identified as direct RNase targets would be more abundant in the corresponding RNase mutant strain, where those transcripts would presumably be protected from degradation. This was not, however, consistently observed upon the cross-referencing of these data sets. Our inducible system informed which transcripts decreased in abundance upon exposure to increasing concentrations of RNase III or RNase J; any transcript whose abundance decreased under these conditions represented a putative direct target of RNase III or RNase J. Considering the more global ∆rnc and ∆rnj mutant RNA sequencing, we are unable to rule out the possibility that some transcripts were both directly and indirectly affected by these RNases; we were also unable to exclude the possibility that they were subject to sequential cleavage by multiple ribonucleases. We would further note that we identified the PNPase-encoding transcript as a potential target of both RNase III and RNase J, where PNPase is a promiscuous ribonuclease that contributes to bulk mRNA decay in Streptomyces (87). Our observations here emphasize the complexity associated with RNA turnover in S. venezuelae and the interplay of the many enzymes that influence transcript stability.

A role for RNase III and RNase J in regulating translational machinery

Our work here and previously (33) has suggested that RNase III and RNase J both contribute to the expression/processing/function of translational machinery but in different ways. The loss of each RNase has been associated with an accumulation of inactive 100S ribosome dimers (33), with the ∆rnc mutant having a further increase in free 30S and 50S subunits (and fewer 70S ribosomes) compared with the wild-type and ∆rnj mutant strains (33). For the ∆rnc mutant, these ribosome assembly defects may be consistent with increased tmRNA levels. tmRNA functions within the trans-translation pathway to recycle stalled ribosomes by promoting their disassociation, and this activity has the potential to increase the relative levels of 30S and 50S ribosomal subunits (88–91).

While RNase III is known to be involved in the pre-processing of 16S rRNA, our data here indicated that RNase J impacts the turnover of ribosomal protein transcripts. It is well-established in E. coli (65) and, more recently, in the Streptomyces relatives Mycolicibacterium smegmatis and Mycobacterium tuberculosis (66) that many ribosomal protein-encoding genes and operons are subject to autoregulatory feedback inhibition at the post-transcriptional level. Our observations here may imply that this autoregulatory feedback could include direct RNA-degrading mechanisms via the recruitment of RNase J.

RNase mutant phenotypes have qualities of the stringent response

The effect of mutating rnc and rnj in S. venezuelae has broad phenotypic consequences, including alterations in translational capabilities, nitrogen metabolism, and antibiotic production. When considering these impacted processes, we noted many parallels with the initiation of the stringent response (92, 93). The stringent response is mediated by the accumulation of the second messenger (p)ppGpp and initiates in response to nitrogen limitation and a corresponding drop in both amino acid levels and charged tRNA pools, ultimately leading to an accumulation of stalled ribosomes (92, 94). We observed strong upregulation of the GlnR regulon in the ∆rnc and ∆rnj mutants, which may suggest that these strains are experiencing nitrogen limitation. In addition, we observed upregulation of tmRNA in the ∆rnc mutant, where tmRNA contributes to the rescue of stalled ribosomes.

Interestingly, in S. coelicolor and other Streptomyces species, the accumulation of (p)ppGpp is also necessary for the production of some antibiotics (93, 95–97). This is reminiscent of our observations here, where our RNase mutants exhibited both significant upregulation of multiple biosynthetic gene clusters and increased antibiotic production. This trend was particularly evident in the RNase J mutant, where the greatest upregulation of nitrogen-assimilation-related genes and the greatest increase in chloramphenicol (antibiotic) production were both observed. It will be interesting to explore a possible connection between these RNases and the stringent response in the future.

Concluding remarks

We have demonstrated that RNase III and RNase J are integral to the coordinated regulation of primary and specialized metabolism in S. venezuelae. These ribonucleases have important roles in ensuring the appropriate timing of key developmental and metabolic processes: loss of either RNase resulted in the dysregulation of genes whose products were involved in development, nitrogen assimilation, phosphate uptake, and specialized metabolism. We identified direct targets of these RNases, provided insight into how their catalytic activity contributed to the regulation of phosphate uptake (and consequent antibiotic production), and revealed differences in their contribution to the production and assembly of ribosomal components. We further revealed the potential for RNase III and RNase J to cooperatively degrade specific transcripts in the case of the rspO-pnp operon. This operon is highly conserved, and this mechanism of regulation may be similarly conserved across bacteria encoding both RNase III and RNase J. Our work underscores the complex nature of RNase-based regulation and how these enzymes function to collectively enable the appropriate control of essential cellular processes in bacteria.

MATERIALS AND METHODS

Strains, plasmids, media, and culture conditions

Strains, plasmids, and primers used in this study are listed in Tables S6 to S8, respectively. S. venezuelae NRRL B-65442 was grown in liquid MYM (1% malt extract, 0.4% yeast extract, 0.4% maltose) for overnight cultivation and on solid MYM (2% agar) for spore stock generation. All Streptomyces cultures were grown at 30°C. E. coli DH5α was used for general cloning and plasmid preparation. All E. coli strains were grown at 37°C on lysogeny broth (LB) or Difco nutrient agar plates, or in LB or super optimal broth liquid medium, supplemented with antibiotics where appropriate to maintain plasmid selection.

RNA isolation and sequencing to identify differentially expressed genes and direct RNase targets

RNA was isolated, as described previously (98), from S. venezuelae cultures grown in a liquid MYM medium. For wild-type and RNase mutant transcriptomic analyses, two biological replicates were collected for each strain and stage of growth (with the timing of growth phase harvest optimized for each strain). Early/vegetative samples were collected at 10 h for wild-type and the ∆rnc strain and at 12 h for the ∆rnj strain. Mid-stage/fragmenting samples were collected at 14 h for the wild-type and the ∆rnc strain and at 18 h for the ∆rnj strain. Late-stage/sporulating samples were collected at 20 h for the ∆rnc strain and at 24 h for the ∆rnj strain.

Strains carrying our thiostrepton-theophylline inducible system (or the empty plasmid control) were sub-cultured to 0.1 OD_600_ and grown at 30°C for 8 h. Theophylline was then added to a concentration of 4 mM to induce TipA translation, after which cultures were grown for an additional 2 h at 30°C before thiostrepton was added to a final concentration of 50 µg/mL to promote expression of FLAG-tagged RNases. Cells were collected at 0 min (pre-induction) and at 20 and 40 min post-thiostrepton addition. Aliquots were then collected for RNA isolation and western blotting.

RNA sequencing was conducted by the MOBIX facility at McMaster University using paired-end technology on the Illumina MiSeq v3 platform. The data discussed in this publication have been deposited in NCBI’s Gene Expression Omnibus (99) and are accessible through GEO Series accession numbers GSE285999 and GSE286000.

Differential expression analysis for RNA sequencing samples

In all instances, reads were aligned to the S. venezuelae NRRL B-65442 genome (NCBI accession: NZ_CP018074.1) using the BowTie2 program (100), and the aligned files were sorted using SamTools (101). The number of reads aligning to genomic features was counted using the HTSeq-count program (102) for each replicate. The HTSeq-count tables were then normalized and used for differential transcript level analysis with the program DESeq2 (103).

For transcriptomic analysis of the RNase mutants compared to wild-type S. venezuelae, these data were filtered by significance (adjusted P-value < 0.05), expression (BaseMean > 50), and differential expression (log_2_ fold change > |2|). For our inducible experiments, we filtered these data by significance (adjusted P-value < 0.05), expression (BaseMean > 50), and differential expression (fold change < −2). We also used these criteria to exclude any genes whose transcript abundance was significantly altered in the plasmid controls between the 0 min and the 40 min samples and excluded any genes whose transcript abundance increased post-induction.

Western blotting to detect FLAG-tagged RNases

We used western blotting to confirm the expression of FLAG-tagged proteins in S. venezuelae following induction of their expression. Between 40 and 80 µg of total protein from each sample were run on a 12% SDS-denaturing polyacrylamide gel for 1 h at 150 V. The proteins were transferred to a methanol-activated PVDF membrane (Amersham Bioscience) using the Bio-Rad Trans-Blot Turbo Transfer System with 1× transfer buffer (48 mM Tris, 39 mM glycine, 1.28 mM SDS, 20% (vol/vol) MeOH) for 25 min at 25 V. The membrane was blocked for at least 1 h with 6% blocking solution (6 g of skim milk powder in 100 mL of Tris-buffered saline with 0.1% (vol/vol) Tween [TBS-T]). The membrane was washed twice with TBS-T for 10 s, after which 3.3 µL of Anti-FLAG Rb antibody (Sigma-Aldrich) were added to 10 mL of blocking solution, and the membrane was left shaking in the solution overnight at 4°C. The membrane was then washed several times with TBS-T before 3.3 µL of secondary antibody (PierceTM Gt anti-Rb IgG SuperclonalTM secondary antibody HRP conjugate) were added to 10 mL of blocking solution. The membrane was then incubated in the solution for at least 2 h while shaking at room temperature. The membrane was again washed several times with TBS-T, after which proteins were detected using the Bio-Rad Clarity Western ECL substrate.

Creation of mutant/overexpression/inducible strains

The in-frame deletion of rsbN was created using ReDirect technology. The coding sequence of rsbN in cosmid sv-6-D01 was replaced with the aac(3)IV-oriT apramycin resistance cassette, and the mutant cosmid was confirmed by PCR before being introduced into the non-methylating E. coli strain ET12567/pUZ8002 and conjugated into wild-type S. venezuelae. The in-frame deletion of chpDFG and rdlAB involved replacing their coding sequences (and intervening genes) in cosmid SV-4-H10 with the aac(3)IV-oriT apramycin resistance cassette. As above, the mutant was confirmed with PCR and then conjugated into wild-type S. venezuelae. Deleting rnc from cosmid sv-3-B07 involved replacement with a hyg-oriT hygromycin resistance marker. Again, the resulting mutant was confirmed by PCR before being conjugated into the ΔchpDFGΔrdlAB S. venezuelae strain to generate the ΔrncΔchpDFGΔrdlAB S. venezuelae strain. In all instances, PCR was used to confirm that the desired mutation had been created in S. venezuelae.

The phoP phosphomimetic overexpression plasmid was made by overlap extension-mediated site-directed mutagenesis and subsequent amplification of the resulting phoP variant (encoding PhoP D52E). This amplified product was then digested with NdeI and XhoI and cloned into the integrating plasmid pMS82, under the control of the constitutive ermE* promoter. The integrity of the resulting construct was confirmed by sequencing prior to conjugation into wild-type, rnc, and rnj mutant strains of S. venezuelae.

Our thiostrepton-theophylline inducible system was generated by amplifying the tipA gene from Streptomyces coelicolor genomic DNA and the tipA promoter sequence from the pIJ6902 vector. The primers used to amplify these fragments contained the theophylline-responsive riboswitch sequence, such that it was situated between the tipA promoter and gene following overlap extension PCR. The overlap extension PCR product (p*tipA-*riboswitch-tipA) was digested with BglII and cloned into the integrating plasmid pIJ6902. Constructs with rnc and rnj fused to an N-terminal 3×FLAG tag were synthesized by GenScript. The sequences for FLAG-tagged rnc and rnj were PCR amplified, digested with NdeI and KpnI, and cloned behind the original tipA promoter in pIJ6902. The resulting constructs had two tipA promoters: one driving the expression of tipA (with the associated riboswitch) and the other driving the expression of FLAG-tagged rnc or rnj.

Transcriptional reporter assays for differentially expressed genes in rnc and rnj mutants

To test the promoter activity of glnA and nirB, their promoter sequences were amplified (using primers detailed in Table S8) before being digested with XbaI and KpnI and cloned into pGus (Tables S7 and S8). The resulting constructs were confirmed by sequencing and were introduced into wild-type S. venezuelae and the RNase mutant strains by conjugation, alongside a promoterless pGUS control. The resulting pGUS-containing strains were inoculated into 10 mL MYM medium and grown overnight. Cultures were normalized to an optical density at 600 nm (OD_600_) of 0.1, and these subcultures were then grown at 30°C for 18 h, after which 1 mL of culture was removed, cells were pelleted, and then washed with 1 mL of sterile water.

Washed cell pellets were resuspended in lysis buffer (50 mM phosphate buffer [pH 7.0], 0.27% [vol/vol] β-mercaptoethanol, 0.1% [vol/vol] Triton X-100, 1 mg/mL lysozyme) and incubated at 37°C for 30 min. The resulting cell lysate was then centrifuged, and the supernatant was used in the β-glucuronidase assay. In a 96-well plate, 150 µL of Z-buffer (600 mM Na_2_HPO_4_, 400 mM NaH_2_PO_4_, 0.1 mM MgSO_4_, 10 mM KCl, 2.7 mL β-mercaptoethanol), 16.7 µL of cell lysate, and 33.3 µL of PNPG (4 mg/mL p-nitrophenyl-β-D-glucuronide dissolved in Z-buffer) were loaded into each well. For blanks, 166.7 µL Z-buffer and 33.3 µL of PNPG were loaded. OD_420_ was measured at 1 min intervals for 1 h in a microplate reader (CytationTM3 Multi-Mode Microplate Reader and Cell Imager). Miller units were calculated for each replicate at the endpoint (1 h or at the time of signal overflow) reading using the following formula:

$$[eqn]$$

In this calculation, Abs_420_ corresponds to the yellow colour of the o-nitrophenol, v represents volume assayed in milliliters, t represents the reaction time in minutes, and OD_600_ is the optical density at 600 nm of the starting culture and represents cell density. For each of the four biological replicates, this measurement was averaged across three technical replicates. The signal associated with the pGUS plasmid control in a respective parent strain (wild-type or RNase mutant with/without phoP) was subtracted from the experimental glnA or nirB sample. Subsequently, the fold change between the parent (RNase mutant or wild-type) and the same strain carrying the phoP phosphomimetic overexpression construct was calculated and plotted.

Chloramphenicol quantification by LC-MS

Cultures were lyophilized, and the resulting lyophiles were resuspended in 10 mL methanol and shaken overnight on a rotary shaker at 4°C. After centrifugation to remove particulate matter, the soluble samples were used for liquid chromatography-mass spectrometry (LC-MS) analyses. The extracts were analyzed using an Agilent 1290 Infinity II LC coupled to an Agilent 6495C triple quadrupole, where 5 µL of the injected extracts were separated on a Zorbax SB-C18 column (100 mm by 2.1 mm by 1.8 µm) at a flow rate of 0.3 mL/min. The draw speed was 50 µL/min, and the eject speed was 200 µL/min with a 2.4 s wait time. Extracted metabolite separation was achieved using a gradient of 0 to 8 min from 95% to 5% A, 8 to 10 min isocratic 5% A, a gradient of 10 to 10.1 min from 5% to 95% A, and 10.1 to 12 min isocratic 95% A, where A is water with 0.1% formic acid (FA) and B is acetonitrile with 0.1% FA. Chloramphenicol was detected using the negative ionization mode, where 321–152 is the quantifying transition and 321–35.2 is the qualifying transition (considering precursor ion–product ion). Chloramphenicol was quantified from cell extracts by comparison to a ten-point standard curve of 10–2,000 ppb chloramphenicol in technical duplicate.