Insight into the Molecular Mechanism of the Transcriptional Regulation of amtB Operon in Streptomyces coelicolor

In Streptomyces coelicolor, amtB transcription is promptly regulated by the global nitrogen regulator GlnR. Although the GlnR binding cis-element has been characterized in amtB promoter, consisting of three GlnR boxes of a3-b3, a1-b1, and a2-b2, its role in GlnR-mediated transcriptional regulation remains unclear. Here, we showed that GlnR had different binding affinity against each pair of GlnR binding sites in amtB promoter (i.e., a3-b3, a1-b1, and a2-b2 sites), and GlnR was able to bind a3-b3 and a1-b1, respectively, but not a2-b2 alone. Consistently, a2 was not a typical GlnR binding site and further experiments showed that a2 was non-essential for GlnR-mediated binding in vitro and transcriptional regulation in vivo. To uncover the physiological role of the three GlnR boxes, we then mutated the wild-type amtB promoter to a typical GlnR-binding motif containing two GlnR boxes (a3-b3–a2-b2), and found although the transcription of the mutated promoter could still be activated by GlnR, its increasing rate was less than that of the wild-type. Based on these findings, one could conclude that the three GlnR boxes assisted GlnR in more promptly activating amtB transcription in response to nitrogen limitation, facilitating bacterial growth under nitrogen stresses.


INTRODUCTION
Soil-dwelling actinomycetes produce a large number of bioactive secondary metabolites, including antibiotics, immunosuppressants, and antitumor agents (Berdy, 2005). Biosynthesis of these metabolites has been found to be regulated by the availability of carbon, nitrogen, and phosphate sources (Martin and Demain, 1980;Hodgson, 2000), among which the concentration of utilizable nitrogen is particularly pivotal (Reuther and Wohlleben, 2007).
Nitrogen is an essential element for bacterial growth and an important component of cellular biological macromolecules such as amino acids, nucleotides, and cell wall components (Merrick and Edwards, 1995). Consequently, bacteria have developed complex transcriptional regulatory systems to allow them to sense internal and external nitrogen levels, coordinating the expression of genes involved in nitrogen metabolism and helping bacteria adapt to environmental stress (Leigh and Dodsworth, 2007). In enteric bacteria, the regulation of nitrogen metabolism is mediated by the NtrB-NtrC two-component system (Reitzer, 2003). However, in Streptomyces coelicolor, a model actinomycete, the global nitrogen metabolism is stringently regulated by the OmpR-type regulator, GlnR (Fink et al., 2002;Tiffert et al., 2008Tiffert et al., , 2011. Under nitrogen-limited conditions, GlnR activates the transcription of its targets and thus promotes nitrogen metabolisms (Tiffert et al., 2008), which may help bacteria cope with the nitrogen deficiency.
As an orphan response regulator, GlnR has no cognate kinase and is post-translationally modified by phosphorylation and acetylation (Wray and Fisher, 1993;Lin et al., 2014;Amin et al., 2016). GlnR homologs are widely distributed and the GlnR-mediated global regulatory system is highly conserved in actinomycetes (Amon et al., 2008;Tiffert et al., 2008). The GlnR binding cis-element was first defined by Tiffert et al. (2008), which was comprised of two 22-bp GlnR boxes, each consisting of one "a site" of "gTnAc" and one b site of "GaAAc" separated by six nucleotides ("5-nt a site-n6-5-nt b site-n6"). However, with more and more GlnR targets characterized, more complicated GlnR-binding sequences are found, which may consist of either triple 22-bp GlnR boxes in the amtB promoter (Wang et al., 2012) or merely two a sites, separated by variable length in the promoters of nasA (Wang and Zhao, 2009) and SCO5163 (Lewis et al., 2011) in S. coelicolor. Although information-based models were later established to describe GlnR box as consisting of two 11-nt direct repeats (Sola-Landa et al., 2013), the new models were still unable to explain the GlnR-binding cis-elements observed in nasA and SCO5163 promoters in S. coelicolor. Besides, in Streptomyces venezuelae, the most common GlnR binding motif is composed of double a sites (GTnAC-n6-GTnAC), and in some cases, no conserved GlnR binding consensus sequences can be identified (Pullan et al., 2011). For the nas operon in the Amycolatopsis mediterranei, three GlnR binding sites (a1-b1-b2 sites) were required for GlnR-mediated activation of its transcription, whereas the a2 site was non-essential (Wang et al., 2013). Therefore, the GlnR binding cis-elements are complicated, and more work needs to be done for better understanding of the molecular mechanism of GlnR-mediated transcriptional regulation, especially for targets with non-standard GlnR binding cis-elements.
In S. coelicolor, GlnR has been shown to directly regulate the transcription of the amtB-glnK-glnD operon (amtB operon) through binding to its promoter region (Wang et al., 2012;Sola-Landa et al., 2013). Besides of the two typical 22-bp GlnR boxes, Wang et al. (2012) ever characterized an extra GlnR box consisting of a3-b3 sites. However, the biological function and the advantage of the three GlnR boxes in activation of amtB transcription remained unclear. We here systematically studied the GlnR binding activities against the three GlnR boxes in amtB promoter and investigated the roles of the three GlnR boxes in the transcriptional regulation of amtB.

Electrophoretic Mobility Shift Assay
Recombinant S. coelicolor GlnR was expressed in E. coli BL21(DE3) and purification process was the same as previously described (Shao et al., 2015). To prepare 6-carboxyfluorescein (FAM)-labeled probes for electrophoretic mobility shift assay (EMSA), primer FAM-M13F-47 was annealed with synthesized amtB oligos, followed by strand extension with Taq DNA polymerase (Tolo Biotech). Because the PCR products were so short that were difficult to be purified, we added the probes in the binding reactions based on the concentration of synthesized oligos used in PCR which were quantified with NanoDrop 2000c. The binding of His-tagged GlnR to FAM-labeled probes was performed at room temperature in a total volume of 20 µl containing 50 mM Tris-HCl (pH 8.0), 100 mM KCl, 2.5 mM MgCl 2 , 1 mM dithiothreitol, and 10% glycerol. In all reactions, sheared salmon sperm DNA was added to a final concentration of 100 ng/µl to prevent non-specific binding of the probes by GlnR. After 20-min incubation, probes were separated by a 5% nondenaturing polyacrylamide gel buffered with 1× Tris-acetate-EDTA. Gels were scanned with ImageQuant LAS 4000 mini (GE Healthcare).

The Hygromycin Reporter System
The schematic chart of plasmid construction was shown in Supplementary Figure S1. Specifically, the coding sequence of hygromycin B resistance gene hyg was amplified from pML814 (Amon et al., 2008) with primers Rshyg-4 and Rshyg-7, which was then digested with XbaI and SpeI before being introduced into the XbaI site of pSET152 (Bierman et al., 1992). The inserted hyg was checked via SpeI digestion, and only plasmids with hyg gene inserted in an inverted direction with the int gene were selected and verified by DNA sequencing with the correct plasmid named as pCMG201. With two pairs of primers, 201T1-F/201T1-A and 201ter-F/201ter-A, two terminators of rrb-T1 and T4-g32 were then added to the 5 and 3 termini of hyg, respectively, employing the site-directed mutation method described before (Wang et al., 2012). The obtained plasmid was designated as pCMG202, which can be used as a reporter system in ϕC31 attB containing strains, e.g., S. coelicolor.
Streptomyces spores were quantitated through serial dilution and colony count before being plated onto plates. The same amount of spores were then serially diluted and dotted onto the nitrogen-limited N-Evans medium with the 5 mM nitrate. Either apramycin (50 µg/ml) or hygromycin B (50 µg/ml) was supplemented, which was indicated in the figure legends.

DNase I Footprinting Assay
To analyze the GlnR-protected regions in both wild-type and mutated promoter regions of S. coelicolor amtB operon, DNase I footprinting assay was employed, which was performed by Tolo Biotech. Besides of pMZ-amtBp-a2m, a2 site was also mutated to other sequences, including poly As, poly Ts and poly Cs, respectively. Briefly, plasmid pMZ-amtBP was taken as the template for PCR amplification, using paired primers of amtB-a2m-5a-F, amtB-a2m-5t-F, amtB-a2m-5c-F, and amtB-a2m-R, respectively. The PCR amplicons were then purified and self-ligated at the presence of T4PNK (NEB) and T4 DNA ligase (Tolo Biotech). After that, the ligation products were transformed into E. coli DH10B, and the transformants were verified by DNA sequencing with correct plasmids named as pMZ-amtBP-5a, pMZ-amtBP-5t, and pMZ-amtBP-5c, respectively.
To prepare DNA probes for DNase I footprinting assay, both wild-type and mutated promoter regions of amtB was first PCR amplified with the primers of SCamtBFP(M13F) and amtB-XbaI-R, and the amplicon was then used as the template for further preparation of FAM-labeled probes with primer pairs of amtB-XbaI-R and M13F-FAM. PCR products were purified with Wizard SV gel and the PCR Clean-Up system (Promega) and quantified with NanoDrop 2000c. The probe (300 ng) was incubated with different amounts of recombinant GlnR protein in a total volume of 40 µl in the buffer same as described in EMSA. Further random digestion by DNase I, electrophoresis with ABI3130, preparation of the sequencing ladder, and data analysis were carried out following the same procedure as previously described (Wang et al., 2012). sites were shown in bold. Three fragments for EMSA were labeled and indicated by horizontal solid lines. The translation start site (atg) was underlined, and both the transcription initiation site (P1) and the predicted -10 box were labeled. (B) EMSA with GlnR and probes shown in (A). Three different amounts of GlnR were employed, and sheared salmon sperm DNA was added to prevent non-specific binding of GlnR to the probes.

Construction of the S. coelicolor Mutant SCamtBp-m3-gn
Mutation of GlnR boxes in the genome of S. coelicolor M145 was achieved by the way of one-step high-efficiency CRISPR/Cas9mediated genome editing system . In brief, the sgRNA sequence was amplified with Overlap-m3-sgRNA and Overlap-gTEMDN, employing pKCcas9 as the template . Then, two pairs of primers Overlap-m3up-F/Overlap-m3-up-R and Overlap-m3-down-F/Overlap-m3down-R were used to amplify the upstream and downstream homologous arms, using M145 genomic DNA as the template. The three fragments were purified and used as templates for subsequent overlapping PCR amplification, using primers of Overlap-m3-sgRNA and Overlap-m3-down-R. The amplicon was then digested with SpeI and HindIII, before being inserted into the same sites of pKCcas9. The obtained plasmid was first verified by DNA sequencing and was then conjugated into S. coelicolor M145 from E. coli ET12567/pUZ8002. The conjugants were verified by PCR analysis and subsequent DNA sequencing. Correct mutant, where a1-b1 sites were deleted, was named as SCamtBp-m3-gn.

Real-Time Reverse Transcription PCR
Streptomyces coelicolor strains were first cultured at 30 • C shaker in liquid S medium for 18 h till the cells entered middleexponential phase. Then, cells were collected and washed twice with Evans minimal medium without any nitrogen sources, followed by being inoculated into fresh Evans medium with 5 mM sodium nitrate as the sole nitrogen source. Cells were collected by centrifugation at 4 • C at the time points of 0 and 120 min, respectively, and then quickly stored in liquid nitrogen.
Total RNA was extracted by using TRIzol (Thermo Fisher Scientific) and SV Total RNA Isolation System (Promega). DNA contamination was removed by digestion with recombinant FIGURE 2 | Analysis of the in vivo transcriptional activities of both the wild-type and mutated amtB promoters with the hygromycin reporter system. M145 derivatives were grown on N-Evans medium with 5 mM sodium nitrate as the sole nitrogen source, and either (A) 50 µg/ml apramycin or (B) 50 µg/ml hygromycin B was supplemented. The aac(3)IV gene on the reporter plasmid enabled the conjugants to grow under the apramycin conditions, which was employed as a positive control. The ability to grow on hygromycin B medium reflected the expression level of the hyg gene as well as the transcriptional activities of the tested promoters. Quantification and gradient dilution of spores were described in Section "Materials and Methods." Photos were taken after 5 days' incubation at 30 • C.
Frontiers in Microbiology | www.frontiersin.org DNase I (TaKaRa). One microgram of total RNA was used for cDNA synthesis with 6-bp random primers using the PrimeScript TM 1st Strand cDNA Synthesis Kit (TaKaRa) in a 20 µl reaction system, following the manufacturer's protocol. Real-time PCR assay was carried out with SYBR R Premix Ex Taq TM II (TaKaRa) according to the manufacturer's instructions. At least three independent samples were tested, and the fold change of relative transcript levels of gene amtB and glnA in response to deficient nitrogen source in 120 to 0 min between the WT and SCamtB-m3-gn strains was expressed as mean ± standard deviation (SD), employing the transcription of hrdB as an internal control. t-Test was performed to test the difference.

GlnR Shows Different Binding Affinities against the Three GlnR Boxes in amtB Promoter
Three GlnR boxes were found in the promoter of amtB operon, each consisting of one a site and one b site (a3-b3, a1-b1, and a2-b2; Figure 1A). This three-GlnR-box model was conserved in amtB promoter among Streptomyces, which was distinct from most other GlnR targets, where usually two GlnR boxes were identified (Supplementary Figure S2). To reveal the role of each GlnR box in GlnR binding of amtB promoter, we measured the GlnR binding affinity against each box. Three synthesized DNA fragments, each of which contained an individual GlnR box (i.e., a3-b3, a1-b1, and a2-b2), were employed for EMSA with GlnR. The results showed that GlnR was able to bind a3-b3 and a1-b1, and the binding affinity against a3-b3 was higher. However, a2-b2 alone could hardly be bound by GlnR (Figure 1B).
To reveal the cause of the low binding affinity between GlnR and a2-b2 sites, we compared the upstream regions of the amtB operon among Streptomyces through sequence alignment analysis (Supplementary Figure S2A), and found that the a2 site was not a typical GlnR binding site, which also indicated that the six GlnR binding sites might show different binding affinities with GlnR.
a2 Is Dispensable for GlnR-Mediated Activation of amtB Transcription in Vivo and GlnR Binding in Vitro To further characterize the six GlnR binding sites, base transition mutation of each site was individually performed, generating six mutated promoters (Figure 2A). We then developed a reporter system to test the in vivo functions of these six GlnR binding sites, employing hygromycin B resistance gene (hyg), which encodes a phosphotransferase and inactivates hygromycin through phosphorylating a hydroxyl group, as the reporter gene. Mutated amtB promoters together with the wild-type promoter were inserted into the reporter system, and the reporter plasmids were then conjugated into S. coelicolor to test the bacterial resistance to hygromycin B, which could reflect the FIGURE 3 | Characterization of the GlnR binding sites in amtB promoter through in vitro DNase I footprinting assay. DNase I footprinting analyses of the GlnR-protected DNA sequences in both the wild-type and a2-mutated amtB promoter regions. Different amounts of GlnR protein were used, including both 10 pmol (A) and 40 pmol (B). The GlnR binding boxes were marked with dashed boxes and labeled in red in the sequencing panel below each figure. For each assay, the GlnR-protected sites were indicated in the bracket below the name of the probe. Red lines: assays without GlnR; blue lines: assays with GlnR.
Frontiers in Microbiology | www.frontiersin.org promoter activities. The results clearly showed that only a2 mutation had no effect on bacterial growth, while the other five mutations severely impeded bacterial growth in medium with hygromycin ( Figure 2B), indicating that a2 site was dispensable.
Furthermore, we also employed the DNase I footprinting assay to accurately determine the binding affinity of GlnR to the wild-type and a2-mutated promoters, with two different concentrations of GlnR assayed (Figure 3). For wild-type amtB promoter, a3-b3 and a1-b1 could be protected by GlnR under a low concentration (0.2 pmol/µl), while the protection of a2-b2 required an increased concentration of GlnR (0.8 pmol/µl). This result was consistent with those of EMSA, where GlnR bound a2-b2 with the least binding affinity (Figure 1B). When a2 was mutated (Figure 3 and Supplementary Figure S3), the GlnR-protected region was the same as that in the wild-type promoter, which was highly consistent with the in vivo reporter assay, indicating that a2 was dispensable for GlnR binding of amtB promoter in vitro.

The GlnR Binding Affinity Is Important for GlnR-Mediated Activation of amtB Transcription
Typical GlnR target promoters contain two GlnR boxes (i.e., the two-GlnR-box motif), such as glnA, glnII, and nirB, but amtB promoter has an extra GlnR box (Supplementary Figure S2B). To further study the biological roles of triple GlnR boxes in amtB promoter, we mutated the amtB promoter and reconstructed it as a typical GlnR target promoter, which had only two GlnR boxes ( Figure 4A). Then, the transcriptional activities of the mutated promoters were tested by the hyg reporter system (Figure 4B).
We first deleted a3-b3 to generate the amtB promoter mutant 1 (m1), and found the reporter gene failed to be activated in m1. This finding was consistent with the results we observed FIGURE 4 | Engineering of an amtB promoter with two GlnR boxes. (A) Illustration of the construction of three two-GlnR-box amtB promoters (i.e., m1 to m3), and deleted regions and mutated sequences were labeled in each mutated promoter. DNA sequences used for EMSA in (C) were indicated with black horizontal lines.
(B) Analysis of the promoter activities of the three mutated amtB promoters. The wild-type promoter (WT) was employed as a positive control, while the reporter plasmid with no promoter (NC) was employed as a negative control. Serially diluted spores were dotted on N-Evans medium with 5 mM sodium nitrate as the sole nitrogen source, and either 50 µg/ml apramycin or 50 µg/ml hygromycin B was supplemented. Photos were taken after 5 days' incubation at 30 • C. (C) EMSA analysis of the GlnR binding affinity against each mutated promoter. Different amounts of GlnR protein were used, and sheared salmon sperm DNA was added in each reaction system to prevent non-specific binding. (D) DNase I footprinting analysis of the GlnR-protected DNA sequences in the wild-type and m3 promoter regions of amtB. The GlnR binding sites were marked with black dashed boxes and labeled either above or below the boxes.
FIGURE 5 | Time-course analysis of the relative transcriptional levels of amtB in both wild-type and SCamtB-m3-gn (m3 mutant). From time 0 min, cells were transferred to N-Evans medium with 5 mM sodium nitrate as the sole nitrogen source, and cells were harvested for transcriptional analysis at the time 120 min. As a representative of two-GlnR-box promoters, glnA was employed as a control. The transcriptional level of amtB and glnA at time 0 min was normalized to 1.0, and SD as well as error bars were calculated with data from three independent experiments. * * P < 0.01; n.s., no significant; t-test.
before (Wang et al., 2012), where mutation of a3-b3 in amtB promoter resulted in the failure in GlnR-mediated transcriptional activation of amtB. When a3-b3 was mutated and meanwhile a1-b1 was mutated to a3-b3, the obtained promoter was designated as amtB promoter mutant 2 (m2), which had only a3-b3 and a2-b2. Interestingly, m2 could be moderately activated by GlnR. However, when we deleted a region containing a1-b1 sites in the wild-type promoter, which exactly moved a3-b3 to the position of a1-b1, the generated amtB promoter mutant 3 (m3) could also be activated by GlnR to an extent comparable to the wild-type. Notably, in both m2 and m3 promoters, the distance between b3 site and a2 site was 6 nt, a previously well-defined distance between GlnR binding sites (Tiffert et al., 2008;Wang et al., 2015), making a3-b3-a2-b2 a typical GlnR box.
We further confirmed the GlnR binding affinity against each mutated promoter with EMSA, and found GlnR had the highest binding affinity for m3 while the lowest for m1 (Figure 4C), which was in high accordance with the in vivo promoter activities (Figure 4B), suggesting the GlnR binding affinity to the target promoter was the key requirement for GlnR-mediated transcriptional activation. Moreover, with DNase I footprinting assay, we found the GlnR-protected region in m3 fully covered a3-b3-a2-b2 sites, which was correspondent with our expectation (Figure 4D).

Three GlnR Boxes Assist GlnR in More Promptly Activating amtB Transcription
Based on the results of the above reporter assay, m3 could be normally activated by GlnR to an extent comparable to the wild-type. However, its transcriptional activity was not quantitatively determined. To precisely measure the m3 transcriptional activity and compare the differences between m3 and the wild-type, we in situ constructed the m3 mutation in the genome of S. coelicolor M145, generating strain SCamtB-m3-gn (m3 mutant in genome). Then, we measured the amtB transcription in both m3 mutant and M145, responding to nitrogen limitation.
Strains were first cultured in rich S medium and then transferred to minimal Evans medium with 5 mM nitrate, mimicking nitrogen-limited conditions. With the employment of real-time reverse transcription PCR, we found that although amtB transcription rapidly increased in both strains after the addition of nitrate, the increase at the time of 60 min relative to 0 min was much higher in the wild-type than in m3 mutant (i.e., 300 folds in M145 versus 45 folds in m3 mutant; Figure 5). As a control, the transcriptional level of the unmodified glnA was almost the same in both strains. Notably, as a representative of the typical GlnR targets with a two-GlnR-box motif, glnA was moderately regulated (i.e., twofold of activation; Figure 5). Taken together, the above results proved that three GlnR boxes enabled more prompt and greater activation of amtB transcription in response to nitrogen limitation.

DISCUSSION
A classic GlnR binding cis-element is comprised of two GlnR boxes, each consisting of one "a site" and one "b site." However, with the characterization of more and more GlnR targets, the GlnR binding features are now well-known to be complicated. Among GlnR targets, the amtB promoter contains three pairs of GlnR boxes, i.e., six GlnR binding sites, which is so far the largest number of GlnR sites found within one promoter. In this study, we demonstrated that a2 site was not essential for GlnR-mediated binding of amtB promoter in vitro and transcriptional activation of amtB in vivo. Although probe a2-b2 was almost the same as probe a1-b1, it could not be normally bound by GlnR, which was distinct from a1-b1. Therefore, we concluded that the non-conservative a2 site led to the failure in binding of a2-b2 alone by GlnR. As b2 site was demonstrated as an essential site, the above results were therefore not contradictory with our previous findings that the GlnR box2 comprised of a2-b2 sites was essential for amtB transcription, where mutation of both a2 and b2 sites led to inactivation of amtB promoter (Wang et al., 2012). Moreover, in our previous study, the a2 site in the promoter of nas operon in A. mediterranei U32 was also proved to be non-essential for both GlnR binding in vitro and GlnR-mediated transcriptional regulation in vivo (Wang et al., 2013).
In amtB promoter, a2 site locates just at the position of a classical −35 element (Figure 1). Based on the classic model of transcriptional regulation, occupation of a2 by GlnR may prevent the σ subunit of the RNA polymerase (RNAP) from recognizing the −35 element. Therefore, GlnR may interact with the σ subunit to recruit RNAP to activate the amtB transcription (Browning and Busby, 2004). However, there exist 67 σ factors in S. coelicolor, and it is still unclear which σ factor is responsible for transcriptional regulation of nitrogen-related genes. Therefore, further characterization of the σ factor that participates in the transcription of amtB operon may facilitate better understanding of the GlnR-mediated transcriptional regulation of amtB transcription.
Besides, as a3-b3 sites are competitively bound by GlnR and PhoP (a global regulator for phosphate metabolism) in S. coelicolor, amtB operon is therefore regulated by both regulators responding to both signals of nitrogen and phosphate availabilities (Wang et al., 2012;Sola-Landa et al., 2013). GlnR functions as a transcriptional activator (e.g., for most GlnR targets) or repressor (e.g., for gdhA), and regulates the transcription of GlnR regulon under nitrogen-limited conditions (Fink et al., 2002;Tiffert et al., 2008Tiffert et al., , 2011. While PhoP governs the phosphate metabolisms and regulates the transcription of Pho regulon under phosphate limitation (Sola-Landa et al., 2005Apel et al., 2007;Rodriguez-Garcia et al., 2009;Martin et al., 2017). Because a3-b3 sites are essential for the activation of amtB transcription (Wang et al., 2012), the competitive binding of a3-b3 between GlnR and PhoP determines the transcriptional level of amtB operon, and only when GlnR occupies a3-b3, the transcription of amtB is activated.
Although the influence of phosphate supply on amtB transcriptional regulation was not tested in this study, the results here clearly indicated the cross-regulation between phosphate and nitrogen metabolisms in S. coelicolor. For example, when phosphate is in excess but nitrogen is limited, GlnR activates amtB transcription to increase the intracellular nitrogen supply, helping bacteria cope with the stress of nitrogen limitation. Otherwise, when S. coelicolor is cultured in phosphate-limited medium, PhoP may compete against GlnR to occupy a3-b3, which will repress the expression of amtB operon to slow down the nitrogen metabolisms, obtaining intracellular nitrogen/phosphate balance.
Taken together, the existence of the extra GlnR box (a3-b3) in amtB promoter not only builds a perfect connection between nitrogen and phosphate metabolisms but also assists in more prompt activation of amtB transcription in response to nitrogen availability.

AUTHOR CONTRIBUTIONS
ZL and XL performed most of the experiments. JzW performed the DNase I footprinting assay. YW constructed the hygromycin reporter system. GsZ helped in the construction of mutant SCamtBp-m3-gn. ZL drafted the manuscript. YL, GpZ, and JW analyzed the data and revised the manuscript. JW supervised the study.

FUNDING
This work was supported by grants from the National Natural Science Foundation of China (31430004 and 31421061) and the Youth Innovation Promotion Association CAS (No. 2017322).

ACKNOWLEDGMENTS
We acknowledge Yuanyuan Liu (Tolo Biotech, Shanghai, China) for her assistance in the DNase I footprinting assay.

SUPPLEMENTARY MATERIAL
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fmicb. 2018.00264/full#supplementary-material FIGURE S1 | Schematic chart for construction of pCMG202. The detailed procedure could be found in the Section "Materials and Methods." and b2 sites were indicated by black boxes, while the a3-b3 and a2 sites were not conserved and were indicated by gray boxes.
FIGURE S3 | Characterization of the GlnR-protected regions in both wild-type and a2-mutated amtB promoters. The a2 site was mutated to "AAAAA," "CCCCC," or "TTTTT," respectively. For DNase I footprinting assays, mutated promoters were incubated with GlnR protein, employing the wild-type promoter as a positive control. As a2 site ("GGGCT") contained several Gs, the site was thus not mutated to poly Gs. Assays without GlnR were shown in red lines, and blue lines indicated assays with 40 pmol GlnR.