Identification of metE as a Second Target of the sRNA scr5239 in Streptomyces coelicolor

While transcriptional regulation of the primary and secondary metabolism of the model organism Streptomyces coelicolor is well studied, little is still known about the role small noncoding RNAs (sRNAs) play in regulating gene expression in this organism. Here, we report the identification of a second target of the sRNA scr5239, an sRNA highly conserved in streptomycetes. The 159 nt long sRNA binds its target, the mRNA of the cobalamin independent methionine synthase metE (SCO0985), at the 5’ end of its open reading frame thereby repressing translation. We show that a high methionine level induces expression of scr5239 itself. This leads, in a negative feedback loop, to the repression of methionine biosynthesis. In contrast to the first reported target of this sRNA, the agarase dagA, this interaction seems to be conserved in a wide number of streptomycetes.


Introduction
Small noncoding RNAs (sRNAs) represent a class of molecules that add a further level of posttranscriptional regulation to the regulatory networks of bacteria. They usually are not translated and most arise from the intergenic regions of the genome where they are transcribed from their own promoters. These often are induced by specific stress conditions or environmental and morphological changes, which leads to a very efficient and defined regulation of sRNA expression [1,2]. Most sRNAs act by binding directly to or nearby the translation initiation region of their target mRNAs thereby influencing the initial steps of translation [3]. Using different experimental approaches, sRNAs have been found in a vast number of bacteria and archaea [4][5][6][7][8]. Most knowledge about sRNAs, their targets as well as their regulatory interactions that is available today, however, has been gathered from Gram-negative bacteria such as Escherichia coli and Salmonella typhimurium [9,10].
Streptomycetes are Gram-positive, GC-rich, soil dwelling bacteria, with some existent marine isolates. The members of this large class of bacteria usually undergo extensive morphological changes throughout their development, starting from a single spore to filamentous mycelia and finally aerial hyphae. Streptomyces coelicolor is the model organism for this group of bacteria. It has a 8.7 Mb linear genome with approximately 8000 annotated genes and about 100 predicted or verified sRNA genes (reviewed in [11]).

Plasmid construction
Construction of the plasmids pUWL-scr5239+ (overexpression), pUWL-asscr5239 (knockdown) and pHDV_scr5239 (in vitro transcription) was described previously [12]. For in vitro transcription of the metE mRNA fragments M1 or M2 the respective fragment was amplified from chromosomal DNA (M1: 1,038,088-1,038,302 nt, M2: 1,038,088-1,038,245 nt). The PCR products were cloned into the plasmid pSP64 [20] using the restriction sites EcoRI and HindIII. The resulting plasmids were named pSP64_M1 and pSP64_M2. For cloning of the scr5239 gusA promoter fusion the region 100 nt upstream of the sRNAs transcription start site to 20 nt downstream of it (5,699,761 to 5,699,880 nt in the S. coelicolor genome) was amplified from chromosomal DNA. The PCR product was cloned into the plasmid pGusA [21] using the restriction sites BglII and Acc65I giving the plasmid pGusA_scr5239p. To create metE reporter gene fusions the M1/M2 fragments were PCR-amplified from the respective pSP64 variant and cloned into the plasmid pGusA using the restriction sites Acc56I and AgeI. The metH 5'UTR and the first 30 nt of the ORF were PCR amplified from the genome (1,776,304-1,776,414 nt). The metH reporter gene fusions were then created as described for metE. Expression of the fusion protein and the control was driven by the constitutive promoter of the ccpA gene SCO4158 (4,576,135-4,576,184 nt). All oligonucleotide sequences can be found in supplementary S1 Table, plasmid maps are available upon request.

Deletion of scr5239
To delete scr5239 including its promoter (5,699,776-5,700,019 nt), the 1,000 nt up-and downstream of the sRNA were PCR-amplified from the genome and fused by overlap PCR to flank a kanamycin (km) resistance cassette. The resulting fragment was cloned into the single XbaI site of the suicide plasmid pKCLP2 [22] carrying an apramycin (apr) resistance cassette giving pKCLP2_dscr5239. The plasmid was transferred into S. coelicolor M145 by intergeneric conjugation from E. coli ET12567/pUZ8002. scr5239 deletion mutants were identified by replica plating and screening for apr S /km R clones. The deletion of scr5239 was confirmed by colony PCR, sequencing of the PCR product and northern blot (S2b Fig and S1 Table).

RNA Isolation
Total RNA was isolated using the hot phenol method described by Mattatall et al. [23]. Five ml cultures were harvested and resuspended in 300 μl lysis buffer (10 mM sodium acetate, 150 mM sucrose, pH 4.8). Glass beads (200 μl, 0.4 mm diameter) and 300 μl hot phenol (65°C ) were added. Cells were disrupted using a FastPrep-24 instrument (MP Biomedicals) for 5x 1 min at 6.5 m/s. After phenol/chlorophorm extraction and ethanol precipitation the RNA was resuspended in 500 μl ddH 2 O and the concentration was determined (usually 2-4 μg/μl). One hundred mg of total RNA were incubated with 30 U Turbo DNase (Ambion) for 1 h to remove residual DNA, subsequently precipitated and resuspended in 50 μl ddH 2 O. Usually a concentration of 1-1.5 μg/μl was yielded and 1 μg was quality checked on a 1% agarose gel.

In vitro transcription
For in vitro transcription the template plasmid pHDV_scr5239 or pSP64_M1 was linearized using the restriction enzyme HindIII. After phenol/chlorophorm extraction and isopropanol precipitation the linear plasmid was resuspended in ddH 2 O. Transcription was carried out over night at 37°C in a 2 ml reaction containing 200 μg linear plasmid, 25 mM magnesium acetate, 200 mM Tris-HCl pH 8.0, 20 mM DTT, 2 mM spermidine, 4 mM of each NTP and 2.5 μl T7 RNA polymerase (home-made). After the incubation 1 ml 0.5 M EDTA pH 8.0 was added to dissolve pyrophosphate. The mixture was subsequently precipitated, dissolved in 3 ml RNA loading dye (deionized formamide with 25 mM EDTA) and loaded onto a 6% denaturing PAA gel with 8 M urea. After the run, the transcribed RNA was visualized in the gel by UV-shadowing, excised and eluted overnight at 4°C in 300 mM sodium acetate pH 6.5. After ethanol precipitation the RNA was dissolved in ddH 2 O.
Electrophoretic mobility shift assay to detect RNA-RNA complex formation Ten pmol in vitro transcribed RNA were 5' labeled with γ-32 P-ATP. Per lane, 0.5 μM unlabelled scr5239 RNA mixed with 50 kcpm of the 32 P-labeled scr5239 RNA were incubated with 7 or 14 μM unlabeled M1 RNA in a 10 μl reaction containing 10 mM Tris pH 7.0, 100 mM KCl, 10 mM MgCl 2 and 1 μg yeast tRNA (Invitrogen) to prevent unspecific binding. After 15 min incubation at 28°C each sample was mixed with 5 μl 5x native loading buffer (50% glycerol, 0.2% bromophenol blue, 0.5x TB) and loaded onto a running 6% native PAA gel-precooled to 4°Cwith 0.5x TB as running buffer. After the run, the gel was dried for 30 min at 80°C. Radioactivity was detected using a FUJI FLA-5000 Phosphoimager. For competition assays 10 μM of the oligonucleotide 1-E (S1 Table) were added prior to the incubation.
Measurement of ß-glucuronidase activity 10 8 pre-germinated spores of S. coelicolor strains were inoculated in 50 ml Evans medium containing amino acids as depicted. Mycelia were harvested after 72 h of cultivation, washed with gus-buffer (50 mM NaHPO 4 pH 7.0, 0.1% Triton X100, 5 mM DTT) and centrifuged again. The subsequent disruption of the cells was done using 200 μl glass beads and a FastPrep-24 instrument (MP Biomedicals) for 6x 30 s at 6.0 m/s. Mycelial debris were removed by centrifugation at 4°C. Protein concentration of the supernatant was determined by Bradford assay [24]. For measurement of GusA activity 25-50 μg protein in a total volume of 750 μl gus-buffer was incubated at 37°C for 15 min. The reaction was started by the addition of 80 μl 200 mM pnitrophenyl-β-D-glucuronide (Glycosynth) and stopped with 200 μl 1 M Na 2 CO 3 upon appearance of a yellowish coloration. Absorption at 415 nm was measured and divided by the reaction time to calculate Gus activity [A 415 min -1 ] and further divided by the amount of protein used to calculate specific Gus activity [A 415 mg -1 min -1 ], which was used herein to represent Gus units (GU). All assays were conducted in triplicate and repeated at least once. Data are presented as the mean +/-standard deviation. 10 8 pre-germinated spores of S. coelicolor strains were inoculated in 200 ml TSB medium and grown at 28°C under continuous shaking. After 3 days, 50 ml samples were harvested, washed twice in Evans medium and resuspended in Evans medium containing amino acids as indicated. The strains were incubated as before for another 24 h, harvested and washed twice in 1x ZAP (50 mM NaCl, 50 mM Tris-HCl pH 8.0, 10% glycerol, 10 mM PMSF). Cell disruption was done as described for the GUS measurements only in 800 μl 1x ZAP. Fifty μg of crude extract were separated on a 20 x 20 cm 8% PAA SDS gel [25] overnight at 4°C and 40 mA. For silver staining the gel was fixed in 50% methanol, 12% acetic acid and 0.18% CH 2 O overnight. The next day, the gel was washed three times for 20 s in 50% ethanol. Next, it was incubated in solution 1 (0.5% Na 2 S 2 O 3 ) for one min followed by three washing steps in ddH 2 O for 20 s each. After 20 min incubation in solution 2 (1.25% AgNO 3 , 0.18% CH 2 O), the gel again was washed three times for 20 s in ddH 2 O and finally developed in solution 3 (6% Na 2 CO 3 , 0.12% CH 2 O, 2% solution 1). Once the protein bands were clearly visible, the reaction was stopped by addition of acetic acid to a final concentration of~10%.

SDS-PAGE and silver staining
Protein bands showing a differential expression pattern were excised using a clean scalpel. Protein identification was done in the mass-spec facility of the Institute of Pharmaceutical Chemistry, Goethe University Frankfurt, Germany.

Results
The sRNA scr5239 can be found in approximately two-thirds of the available sequenced Streptomyces genomes. However the only validated target of scr5239, the agarase dagA is present merely in S. coelicolor. This led us to the hypothesis, that scr5239 had to have further-possibly more conserved-target mRNAs.
Previously we have shown, that the expression of scr5239 decreases under nitrogen limiting conditions [12]. So, to get a lead on other targets of scr5239 we further investigated this phenotype. First, we analysed the expression of scr5239 in the presence of the key components of nitrogen storage, glutamine and glutamate. Fig. 1a shows a northern blot of S. coelicolor wild type cells grown in minimal medium supplemented with glutamine or glutamate. These two amino acids have a contrary effect; glutamine leads to a repression, glutamate to an induction of scr5239 expression (see S1 Fig for quantification of the northern blot). This finding prompted us to analyse the proteome of S. coelicolor under these conditions. Fig. 1b shows a gel slice, where overexpression (OX), wild type (wt) and knockdown (KD) strains of scr5239 [12] had been grown in the presence of glutamine. One prominent band specifically is induced in this medium in the scr5239 knockdown strain. This band was identified by mass spectrometry to be the B12-independent methionine synthase MetE (SCO0985).
MetE is one of two isoenzymes that catalyse the S-methylation of homocysteine to methionine. The other isoenzyme, MetH, is vitamin B12-dependent; it requires methylcobalamin (derived from B12) as a cofactor to catalyse the de novo formation of methionine. In bacteria that carry functionally equivalent B12-dependent and B12-independent enzymes, the activity of the B12-independent one often is regulated by a B12 riboswitch in the 5'UTR of the respective mRNA [26]. By comparing a large variety of bacterial genomes Vitreschak et al. [26] predicted the existence of a B12 riboswitch in the intergenic region 382-155 nt upstream of the metE ORF.
A transcriptome analysis of S. coelicolor grown on solid medium done in our lab supports this prediction. It places the metE transcription start point at 1,037,861 nt, 382 nt upstream of the GTG start codon (S2 Fig).
We addressed the question whether the expression of scr5239 was dependent on methionine. We cultivated the wild type in minimal medium +/-methionine and checked for sRNA expression in a northern blot. Quantification of the northern blots showed, that the presence of methionine in the medium leads to an almost 2-fold increase in the amount of scr5239 ( Fig. 1c; p-value < 0.05). Fusion of the scr5239 promoter region to a gusA reporter gene led to the same increase showing that this effect is due to a higher transcription rate rather than a stabilization of the sRNA in the presence of methionine (Fig. 1d).
In the next step, we tested whether scr5239 affects metE expression in a direct or indirect manner. Most sRNAs that repress translation of their targets bind them in or around the translation initiation region [1]. Using RNAhybrid [27] for sequence alignment we identified a potential scr5239 binding site on the metE mRNA spanning the area from nucleotide +7 to +30 in the ORF ( Fig. 2a and b). On the sRNA side the binding site covers the nucleotides 114 to 130 which are almost exactly the same nucleotides as for the previously identified target dagA (nt 113 to 129) [12].
To test if there is a direct interaction between scr5239 and the metE mRNA we performed mobility shift assays with scr5239 and the metE mRNA. To omit possible interference of the B12 riboswitch during in vitro transcription or during refolding of the RNA after its purification, we used a 215 nt fragment of the metE mRNA (M1) covering the nucleotides-155 to +60. The mobility shift assay was performed using radiolabeled scr5239 RNA mixed with rising amounts of M1 RNA. Fig. 2c-left panel-shows that there is a direct and specific interaction between the two RNAs. The same effect can be seen when radiolabeled M1 RNA is titrated with scr5239 RNA (S3 Fig). To check, whether we predicted the correct binding site on the sRNA side, we performed another mobility shift assay where we added an excess of the DNA oligonucleotide 1-E thereby blocking the proposed binding side for access from M1 RNA (Fig. 2c, right panel). 1-E was able to abolish the interaction between scr5239 and M1, validating that the interaction site was predicted correctly. The multiple bands that can be seen in the assay arise from several conformations of the M1 RNA. In a native gel, the free M1 RNA adopts at least three visible conformations all of which are bound by scr5239 (see S3 Fig; first lane). We also verified, that scr5239 is able to bind to the full-length metE 5'UTR including the  riboswitch. For that we performed another gel retardation assay using the complete metE 5'UTR spanning the nucleotides-382 to +60 (S4 Fig).  Fig. 2b shows the 2D structure of free scr5239 as validated by enzymatic and inline-probing as well as a model of the metE-scr5239 complex. Like for dagA the binding site of metE almost solely consisted of the nucleotides forming stem P4 that has to unfurl to allow for the interaction to take place.
In the next step, we wanted to confirm the interaction in vivo. The standard method to analyse sRNA-target interactions in vivo was developed for E. coli by Urban et al. in 2007 [28]. There, the sRNA of interest is expressed from a high-copy plasmid while a target-mRNA-GFP fusion is expressed from a low-copy plasmid to enhance the regulatory effect and the sensitivity of the system. Mutations in the sRNA and/or the mRNA fragment serve to validate the interaction. Since its publication, this system has been adapted for a wide range of organisms like Sinorhizobium [29] or Vibrio [30]. We redesigned this system for the use in Streptomyces to measure the scr5239-metE interaction. Due to the high intrinsic fluorescence of S. coelicolor we chose to use gusA instead of gfp as reporter gene. Further, we used the integrating plasmid pGusA to gain a reliable and stable copy-number of the reporter fusion. In addition, we decided to create a scr5239 deletion strain (Δscr5239), where the sRNA was replaced by a kanamycin resistance cassette (S5a and S5b Fig.). We chose to delete scr5239 because the previously used antisense-mediated knockdown strain [12] showed considerable variation in knockdown efficiency depending on the medium conditions used (data not shown).
We used the same RNA sequence as for the mobility shift experiments for the in vivo measurement of the scr5239-metE interaction (construct M1). As a control we cloned a shorter fragment covering the same 5'UTR part but only the first six nucleotides of the metE ORF thereby omitting the scr5239 binding site (construct M2, Fig. 3a). As a second control we used the original pGusA plasmid without any metE fragments attached to the gusA mRNA. Measurement of reporter gene expression of S. coelicolor wt vs. Δscr5239 shows a 2.6-fold induction of metE expression in the deletion strain. This induction is lost in the M2 construct where the binding site is missing (Fig. 3b). The overall lower expression level of the M2 construct may be due to an altered translation efficiency and/or stability of the fusion protein. Scr5239 seems to regulate metE expression on the translational level as the mRNA abundance of metE does not change in the Δscr5239 strain compared to the wild type (S6 Fig).
These findings show that there is a negative regulation of metE expression by the methionine inducible sRNA scr5239. We also tested whether the second methionine synthase, metH, is regulated in the same way. A reporter gene fusion of the protein, however, showed no increase of MetH expression upon scr5239 deletion (S7 Fig).

Discussion
In this study, we identified a second target of the sRNA scr5239 in S. coelicolor. This finding fits to the current understanding of sRNA function in bacteria: that sRNAs, like the eukaryotic microRNAs, usually do not regulate just one target but act as posttranscriptional control instances for whole networks of target genes [31]. One example for this is the S. typhimurium sRNA GcvB that is a major regulator of amino acid metabolism in this organism [32].
The binding site of scr5239 to its second target again is located within the ORF of the mRNA. In the best-studied examples, E. coli and S. typhimurium, it is known but rather unusual for an sRNA to bind its target in its ORF. The majority of sRNAs in these organisms regulate the expression of their targets by blocking access to the ribosomal binding site and/or start codon thereby hindering translation initiation. However, recent studies have shown that binding of an sRNA up to 5 codons downstream of the start codon still is sufficiently close to the translation initiation region to interfere with 30S subunit binding [3]. The binding site of scr5239 on the metE mRNA fits this rule as it covers the codons 3-10 (Fig. 2b). Interestingly, the binding site of scr5239 on dagA covers the codons 11-18 on the mRNA [12]. This supports our current proposal, that the regulation of dagA-a rather recent acquisition to the S. coelicolor genome-by scr5239 is still in the progress of being established.
In contrast to dagA, the metE gene including its binding site to scr5239 is highly conserved throughout the streptomycetes indicating that this regulatory interaction is much more widespread (S8 and S9 Figs). On the side of scr5239 a seven nucleotide core binding motif (GCUGCGG) is highly conserved in all homologs we identified so far. In about half the homologs this motif occurs-with one G to C variation-in tandem (GCUCCGG-GCUGCGG, see S8a Fig top five sequences). Being in the coding region of the metE gene, the binding site on the mRNA is even more highly conserved (S8b Fig). But a few variations occur: the core motif complementary to scr5239s is CCGCAGC. In S. hygroscopicus there is a silent A to G mutation that increases the possible binding site by three nucleotides (S9 Fig). An additional CGC insertion in the ORF three nucleotides downstream of the conserved motif also got integrated into the binding site on the sRNAs side.
Little is known about the regulation of amino acid metabolism in Streptomyces. To date, three conserved transcription factors involved in amino acid biosynthesis have been characterized: GlnR (glutamine, glutamate), ArgR (arginine) and NdgR (leucine, cysteine and methionine) [18,33,34]. NdgR has recently been shown to act as a transcriptional activator on at least two genes of the methionine synthesis system: metH and the methylenetetrahydrofolate reductase metF that catalyzes the second to last step of methionine synthesis. In this study, Kim et al. used the DNA sequences of the presumed promoter regions of the respective genes as bait and screened for NdgR binding. The metE probe, however, failed to capture any NdgR even though they used the whole intergenic region upstream of metE as bait. Kim et al. identified two possible binding motifs for NdgR within the probes that were captured by the protein. However, both of them are absent in the intergenic region upstream of the metE gene. This implies, that the two enzymes MetE and MetH are regulated by different means although they perform the same reaction: MetH expression is dependent on the presence of the transcriptional activator NdgR. The expression of MetE, on the other hand, depends both, on the absence of coenzyme B12 and of the sRNA scr5239. Fig. 4 shows a model of our current understanding of the role scr5239 plays in the regulation of metE. If coenzyme B12 is available the B12 dependent methionine synthase MetH is expressed while transcription of metE is terminated by the B12 riboswitch in its 5'UTR. If B12 is scarce, transcription of the metE mRNA is permitted and MetE is expressed. It is still unknown, if and how MetH expression is downregulated under these conditions. NdgR activates its transcription upon nutrient downshift [18], but there is no data showing if the availability of coenzyme B12 influences the activity NdgR and/or MetH.
Expression of MetE leads to an increased methionine level, which-by a yet unknown mechanism-induces the expression of scr5239. The sRNA then represses the translation of the metE mRNA thereby reducing methionine synthesis.
The S-methylation of homocysteine is the final step in methionine biosynthesis. While there are several alternative ways leading to homocysteine, this last step is highly conserved. The two methionine synthases MetH and MetE can be found in parallel in a large number of bacteria, although there are species that carry one but not the other [15]. The transcription of metE is regulated by a B12 riboswitch in its 5'UTR in most bacteria. This allows the cell to make the decision if coenzyme B12 is available and if therefore the expression of a cobalamin-independent methionine synthase is necessary. Some bacteria, like Bacillus clausii, additionally have a SAM riboswitch in tandem to the B12 riboswitch in their metE 5'UTR. Also the expression of metH and metK is controlled by a SAM riboswitch in this organism. This setup generates a feedback inhibition where high SAM levels directly repress the expression of these three enzymes [35]. Streptomycetes lack a SAM riboswitch in metE (also in metH and metK), therefore we propose that here the conserved sRNA scr5239 takes over its regulatory function: to give a feedback about the cells methionine (and SAM) supply. MetH which uses adenosylcobalamin as methyl donor catalyses the reaction with high turnover rate when coenzyme B12 is present. In the absence of B12, a B12-dependent riboswitch allows transcription of metE mRNA leading to increased methionine production by MetE. High methionine concentration results in feed back regulation of both MetH and MetE. MetH is controlled by the transcriptional regulator NdgR [18]. In addition, high methionin concentration also induces the expression of scr5239, which in turn represses metE translation. A possible link between nitrogen and methionine metabolism via the sRNA is speculative. Nature has come up with diverse solutions to tackle the problem of multiple signal input. In B. subtilis, the glmS riboswitch probably represents the simplest solution. The ligand binding pocket of the aptamer domain not only binds intracellular Glucosamine-6-phosphate as an activating agent. It also binds Glucose-6-phosphate, which acts as an inhibitor of ribozyme activity. The riboswitch, thus, integrates metabolite information to sense the overall metabolic state of the cell [36]. A second solitary element that senses and integrates two different input signals is the adenine riboswitch encoded by the add gene from Vibrio vulnificus. This translational regulatory element senses temperature and ligand concentration to confer efficient regulation over the range of physiologically relevant temperatures, allowing gene expression control to adapt to the differing temperatures experienced in the free-living and in the infectious environments [37]. As already discussed above, in B. clausii, two different riboswitches respond independently to SAM and vitamin B12, respectively. This system's output fits a Boolean NOR gate, because transcription is terminated in the presence of only one or both ligands [35]. In the work presented here, not two riboswitches but a riboswitch combined with a regulatory small RNA mediate the input of the two signals. A combination of sRNAs and riboswitches regulation was also reported for SroA and SroG, two sRNA processed out of a riboswitch [38] or, even more intriguingly, two riboswitches which respond to SAM in Listeria monocytogenes which control the expression of the virulence regulator PrfA act as sRNA themselves [39]. On the contrary, a recent report describes a B12 riboswitch that is part of the sRNA EutX controlling ethanolamin utilization in Enterococcus faecalis. This riboswitch can lead to premature termination of the sRNA-target interaction therefore regulating the sRNA function. Obviously there are plenty of possibilities to perform multiple input integration, however to our knowledge; metE is the first example of an sRNA regulating gene expression in combination with a riboswitch.
What is the connection of the regulation of methionine synthesis to the glutamine/glutamate complex? Glutamine acts as nitrogen storage in the cell. Feeding the cells with glutamine shifts the balance towards nitrogen surplus and leads to repression of scr5239 expression (Fig. 1a). It is not clear, why methionine production would have to be upregulated under these conditions. Generally, nitrogen limitation is regarded as the 'normal' state for bacteria growing in soil [40,41]. This would be represented by a high glutamate over glutamine ratio leading to an induction of scr5239 and thereby a repression of methionine synthesis. A good nitrogen supply, however, can induce a period of fast growth where protein synthesis and thereby the demand for amino acids increases. Therefore scr5239 regulation possibly connects the nitrogen supply status of the cell to amino acid synthesis.