GlnR Negatively Regulates the Transcription of the Alanine Dehydrogenase Encoding Gene ald in Amycolatopsis mediterranei U32 under Nitrogen Limited Conditions via Specific Binding to Its Major Transcription Initiation Site

Ammonium assimilation is catalyzed by two enzymatic pathways, i.e., glutamine synthetase/glutamate synthase (GS/GOGAT) and alanine dehydrogenase (AlaDH) in Amycolatopsis mediterranei U32. Under nitrogen-rich conditions, the AlaDH pathway is the major route for ammonium assimilation, while the GS/GOGAT pathway takes over when the extracellular nitrogen supply is limited. The global nitrogen regulator GlnR was previously characterized to activate the transcription of the GS encoding gene glnA in response to nitrogen limitation and is demonstrated in this study as a repressor for the transcription of the AlaDH encoding gene ald, whose regulation is consistent with the switch of the ammonium assimilation pathways from AlaDH to GS/GOGAT responding to nitrogen limitation. Three transcription initiation sites (TISs) of ald were determined with primer extension assay, among which transcription from aldP2 contributed the major transcripts under nitrogen-rich conditions but was repressed to an undetectable level in response to nitrogen limitation. Through DNase I footprinting assay, two separate regions were found to be protected by GlnR within ald promoter, within which three GlnR binding sites (a1, b1 sites in region I and a2 site in region II) were defined. Interestingly, the major TIS aldP2 is located in the middle of a2 site within region II. Therefore, one may easily conclude that GlnR represses the transcription of ald via specific binding to the GlnR binding sites, which obviously blocks the transcription initiation from aldP2 and therefore reduces ald transcripts.


Introduction
Nitrogen is present in nearly all bacterial metabolites and can be assimilated from inorganic or organic sources. To cope with variations of nitrogen sources (excess versus limitation), bacteria have developed sophisticated regulation systems for nitrogen assimilation. For most bacteria, like Escherichia coli, Corynebacterium glutamicum and Streptomyces coelicolor, the glutamate dehydrogenase (GDH) pathway is the major route for ammonium assimilation when bacteria are growing under ammonium-excess conditions, and the glutamine synthetase/glutamate synthase (GS/ GOGAT) pathway then takes over when the environmental nitrogen supply is poor [1]. However, in Amycolatopsis mediterranei, an important actinomycete for industrial production of rifamycin, although the assimilation pathway is the same as most bacteria under nitrogen limitation, i.e. using the GS/GOGAT pathway, the alanine dehydrogenase (AlaDH) pathway, instead of GDH pathway, is responsible for ammonium assimilation under high ammonium conditions [2]. AlaDH catalyzes the reaction between pyruvate and ammonium to generate alanine without any expense of ATP. Similar to GDH, AlaDH also has a high K m value for ammonium, and thus the AlaDH pathway is less efficient than the energy-consuming GS/GOGAT pathway in ammonium assimilation and is only suitable for bacteria growing in ammonium-excess conditions [2].
In enteric bacteria, two major regulators, NtrC and Nac, regulate the nitrogen assimilation in response to changes of ammonia concentrations [3][4][5]. Unlikely, in Gram-positive bacteria, global regulators are often employed to govern the genes involved in global nitrogen metabolisms, e.g. GlnR and AmtR. Take GlnR for example, in the model actinomycete S. coelicolor, Tiffert et al. [6] found that GlnR is a global regulator for nitrogen assimilation, which recognizes a 22-bp consensus sequence of gTnAc-n 6 -GaAAc-n 6 that is comprised of an ''a-site'' of ''gTnAc'' and a ''b-site'' of ''GaAAc''. When the nitrogen supply is limited, GlnR acts as both an activator and a repressor, e.g. GlnR switches on the GS/GOGAT pathway via activating the transcription of glnA and switchs off the GDH pathway at the same time through repressing the gdhA transcription. Similarly, in Streptomyces venezuelae and Mycobacterium smegmatis, GlnR is also found to be a global regulator for the nitrogen metabolisms [7][8][9]. In Amycolatopsis mediterranei, enzymes involved in nitrogen metabolism with poor nitrogen sources, e.g. nitrate/nitrite redutases [10] and glutamine synthetase [11] are all stringently regulated by GlnR; however, little is known about AlaDH, which is the only known enzyme responsible for ammonium assimilation under nitrogen-rich conditions. Here in this study, we prove that the transcription of the AlaDH-encoding gene ald is directly negatively regulated by GlnR in A. mediterranei under nitrogenlimited conditions. Moreover, the in-depth mechanism for this transcriptional repression is also proposed.

Bacterial strains and media and primers
A. mediterranei strains, including the wild type, the glnR null mutant strain Rk [11] and the glnR complementation strain LRS987 (named as Rk/pVKER in [11]), were grown at 30uC in either Bennet rich medium [12] or minimal medium [13] with appropriate nitrogen sources supplemented [10]. When needed, erythromycin (200 mg/ml) and ampicillin (100 mg/ml) were added to the media. Primers were listed in Table 1.

Total RNA isolation, Reverse Transcription-PCR (RT-PCR) and Northern Blot
A. mediterranei strains were all grown on Bennet agar supplemented with either 120 mM ammonium or 80 mM nitrate for 4 to 5 days. Total RNA was extracted from the mycelia using TRIzol reagent, following the manufacture's instructions, while RT-PCR assay was performed as described before [14]. Northern analyses were carried out as described [15], using a 1% agarose formaldehyde gel. Total RNA was extracted from U32 that was grown in minimal medium with different extracellular nitrogen sources supplemented. Fifty microgram total RNA was loaded onto each gel slot after heating at 65uC for 10 min and quickly chilling on ice to destroy the secondary structure. A DNA fragment containing the whole ald gene (1.1 kb) was used as a template to prepare probes by random-primed a-32 P-dCTP labeling. Hybridization and washing procedures were performed at 42uC at the presence of 50% de-ionized formamide.

Primer extension analysis
A. mediterranei U32 was grown in Bennet liquid medium supplemented with different extra nitrogen sources and total RNA was extracted for primer extension assay. According to the method previously mentioned [16], 20 mg total RNA was used in each assay, using the c-32 P-labeled primer ald_pe, which was complementary to the 23th to the 17th nucleotides of the ald protein coding sequence (CDS). At the same time, radio-labeled ald_pe was also used for preparation of the sequencing ladder, with the usage of the fmol DNA Cycle Sequencing System (Promega). The reverse transcribed products together with the sequencing ladders were analyzed on 6% ployacrylamide sequencing gels containing 7 M urea and analyzed with a phoshorimager.

Electrophoretic Mobility Shift Assay (EMSA) and DNase I footprinting assay
The ald promoter region was PCR amplified, purified with gel purification kit and then labeled with c-32 P-ATP using T4 polynucleotide kinase (PNK) (NEB). Production of purified recombinant A. mediterranei GlnR as well as the EMSA conditions were the same as reported [10]. For DNase I footprinting experiment, primer ald_pe was firstly end-labeled with c-32 P-ATP using T4 PNK, and then a 527-bp DNA fragment was PCR amplified using primer U32aldp_F1 and c-32 P-labeled primer ald_pe. Proper probe (about 120,000 cpm) was incubated at 30uC for 30 min with different amounts of purified His 6 -GlnR, 1 mg sheared salmon sperm DNA in a total volume of 40 ml in the same buffer as EMSA and 0.3 Unit DNase I (TaKaRa, Shiga, Japan) was used for digestion at room temperature for 1 min. The preparation of DNA sequencing ladders and analyses of the digests were the same as reported before [16].

Atomic force microscopy (AFM) imaging
The A. mediterranei U32 chromosomal DNA was used as a template for PCR amplification with paired primers aldbs_F2 and ald_r and the products (fragment AFM_Ald) were then purified for further AFM experiments. In a volume of 15 ml, 600 ng of fragment AFM_Ald was incubated for 15 min at room temperature with 1 mg of purified His 6 _GlnR in the same binding buffer as used in EMSA. The reaction mixture was firstly diluted by 15 to 20 times with deionized water, and was then deposited on a piece of mica modified by Ni 2+ and incubated for about 2 min. After gently washed with deionized water, the resulting sample was imaged with AFM. All images were performed in tapping-mode TM using a Multi-mode AFM with a Nanoscope IIIa controller (Digital Instruments (D.I.) Co., Santa Barbara, CA, USA). AFM images were collected using commercialized silicon tip (NASC18, MicroMasch) with a spring constant of 3.5 N/m and a resonant frequency of 75 kHz. All AFM experiments were conducted in an ambient environment.

Results
GlnR negatively regulates the transcription of ald in A. mediterranei U32 To examine the expression of ald in response to extracellular nitrogen supplies, U32 was cultured in minimal medium supplemented with different nitrogen sources, including increasing concentrations of ammonium and nitrate, and total RNA was then extracted for Northern blot analysis (Fig. 1A). According to the Northern blot results, the transcription of ald increased with the increase of extracellular ammonium concentrations while was stringently repressed by nitrate, which usually represents a poor nitrogen source, demonstrating that ald transcription was either activated in nitrogen-rich conditions or repressed in nitrogenlimited conditions. Similar results were obtained with RT-PCR assays when U32 was grown in rich Bennet medium supplemented with either 120 mM ammonium or 80 mM nitrate, i.e. transcriptional level of ald was high in ammonium conditions while no transcripts could be detected in the presence of nitrate. When reverse transcriptase was omitted from the RT reaction mixture, no visible DNA bands of PCR products were detected, indicating no contamination of DNA in the RNA samples (data not shown).
Besides ald, the transcriptional regulation of other nitrogen metabolism-related genes, including glnA, nas operon [17] and amtB (data not shown), in Bennet medium with ammonium (or nitrate) is the same as that in minimal medium with ammonium (or nitrate). Therefore, Bennet medium, which is easier to prepare, is usually used for analysis of gene transcription in U32 and Bennet medium supplemented with 120 mM ammonium is designated  ''nitrogen-rich'' medium for U32 while supplemented with 80 mM nitrate is designated ''nitrogen-limited'' medium for U32.
In glnR null mutant Rk, the transcription of ald remained at a high level under nitrogen-limited conditions ( Figure 1B, lanes 3  and 4). When glnR gene with its original promoter was introduced back into Rk, the transcriptional repression by nitrate supplementation reappeared again ( Figure 1B, lanes 5 and 6), which thus indicated GlnR as a repressor for ald transcription in response to limitation of extracellular nitrogen sources.
ald has three transcription initiation sites Primer extension assay identified three transcription initiation sites (TISs) for ald, the 2173G (P1), the 246A (P2) and 267G (P3) (relative to translation start site, Figure 2), among which P1 and P2 were main TISs while transcription from P3 was extremely weak and was only detectable under nitrogen-rich conditions (Figure 2). Under the nitrogen-rich condition, transcripts initiated from P2 contributed to the majority of ald mRNA, while the transcript from P2 was basically undetectable under nitrogen-limited condition (Figure 2), suggesting a stringent regulation of the transcription from P2 in response to extracellular nitrogen supplies. Compared to P2, the level of transcription from P1 was much weaker under the nitrogen-rich condition and obviously decreased, but to a detectable level under the nitrogen-limited condition. Therefore, P1 seems to be responsible for the basal level expression of ald that may be important for normal growth.

GlnR protects two separate regions within the promoter region of ald
With purified recombinant N-terminal His-Tagged GlnR and the ald promoter region, EMSA was employed to find that GlnR could specifically bind to the ald promoter in vitro (Figure 3). DNase I footprinting assay was further performed to characterize the precise DNA sequences that GlnR protected in ald promoter region, with two separate regions of RI and RII characterized (Figure 4). During the increase of GlnR concentrations, RI was apparently protected prior to RII, indicating GlnR has a higher binding affinity for RI (Figure 4).
Three GlnR binding sites (a1, b1 sites in RI and a2 site in RII) were manually defined within the ald promoter region (Figure 5A), among which the DNA sequences of a1 and inverted b1 were identical to those of a1 and b1 sites in the promoter region of nas operon ( Figure 5B) [10], and a2 site of ''GTAAC'' followed the rule of ''gTnAc'' for an a-site as previously defined by Tiffert et al. [6]. The arrangement of the GlnR binding sites is similar in both promoters ( Figure 5B), i.e. a1-b1 sites are close in distance, while the third site (ald_a2 or nas_b2) is separated from the a1-b1 sites. In nas promoter, the internal distance between b1 and b2 sites is 16 bps [10], while b1 and a2 sites are separated by 76 bps in ald promoter.
Because the ald major transcription initiation site P2 locates in the middle of the a2 site of ''GTAAC'', where the italicized ''A'' represents P2 ( Figure 5A), one may easily conclude that GlnR negatively regulates ald transcription via specific binding to P2 and acts as a roadblock to shut down the majority of ald transcription. In addition, binding of GlnR to a2 site, which is downstream of P1 and P3 ( Figure 5A), in theory blocks the transcription from P1 and P3 sites to a certain extent, which might well explain the reduction of ald transcription from P1 as well as P3 in response to nitrogen limitation.

Discussion
So far, GlnR has been well characterized as a central regulator for governing global nitrogen metabolisms in several important actinomycetes, including S. coelicolor, S. venezuelae, M. smegmatis, etc. In most cases, GlnR exerts positive regulation on its target genes in response to nitrogen limitation; however, GlnR may sometimes act as a negative regulator (e.g. for gdhA) to coordinate the global metabolisms, helping the bacteria to cope with the environmental stress. A. mediterranei is an important actinomycete for rifamycin production and has been studied for decades in our lab. In A. mediterranei, GlnR has been characterized to positively regulate the transcription of glnA [11], nas operon [10] and amtB (unpublished data) under nitrogen-limited conditions; however, because the GDH activities do not change significantly in A. mediterranei (data not shown), transcription of gdhA is unlikely to be regulated. Interestingly, unlike most bacteria, A. mediterranei uses AlaDH pathway instead of the GDH pathway for ammonium fixation when the extracellular ammonium concentration is high. Therefore, a simple question raised is whether GlnR plays a repressor role in ald transcription in A. mediterranei, just as its homologues do in many other phylogenetically close actinomycetes? Based on the experiments exemplified in this study, we have proved that GlnR negatively regulates ald transcription and switches the ammonium assimilation route from the AlaDH pathway to GS/GOGAT pathway in response to nitrogen limitation, facilitating the bacterial adaption to the nutritional stress. Therefore, GlnR may be considered a global nitrogen regulator in A. mediterranei.
The GlnR binding cis-elements have been studied for years by several groups around the world. Through aligning several promoter sequences of the under-regulated genes, Fink et al. firstly proposed a 44-bp S. coelicolor GlnR binding motif [18]. Based on Fink's work, Tiffert et al. [6] later deduced a much shorter consensus sequence specific for GlnR binding, which is 22bp, comprising of an ''a-site'' of gTnAc and a ''b-site'' of GaAAc with a fixed distance of 6 bases in between [6]. However, the GlnR binding cis-elements have been found to be extremely complicated and can be comprised of varied GlnR binding sites (from 0 to 6) in different target genes of different species [7,14,16,19,20]. A. mediterranei GlnR is not only structurally homologous to that of S. coelicolor GlnR [11] but also able to complement the glutamine auxotrophic phenotype of the S. coelicolor glnR null mutant in trans [21], and the two GlnRs are proposed to share much homogeneity in their binding DNA consensus sequences [10]. We ever characterized three essential GlnR binding sites (a1, b1 and b2) in the nas promoter, which are of high similarity to the GlnR binding sites characterized in S. coelicolor [10]. Here, once again, we predict three GlnR binding sites in the promoter region of ald for further investigation. Notably, these sites in ald promoter are highly similar to those in nas promoter, in not only their primary DNA sequences but also their configurations. Although the mechanism for negative regulation of ald transcription by GlnR is obvious and clear, the in-depth depiction of the action of GlnR (e.g. how GlnR binds to these three sites) seems a mission impossible at present.
In the case of nas regulation, two complexes of CI and CII were found during the process of GlnR binding to nas promoter and two distinct steps were proposed. First, GlnR binds to a1 and b1 sites to form CI; then, on the basis of CI formation, GlnR further binds b2 site to produce CII, which is required for the activation of nas transcription [10]. Although it has been known that b1 and b2 are required for the formation of CII, it is unclear whether GlnR binds to a1 site in CII. Recently, based on the results of DSScrosslinking assay and crystal structural analysis, Lin et al. demonstrated that GlnR mainly works as a homodimer [22], indicating only b1 and b2 sites are bound by a homodimer GlnR in the complex of CII. Due to the high similarities between the GlnR binding cis-elements in nas and ald promoters, we presume here a similar procedure for GlnR binding to the ald promoter, and GlnR may first bind to a1-b1 sites and then to b1-b2 sites with a final release of a1 site ( Figure 6E). As b1 and b2 sites are separated by 76 bps, the binding of a homodimer GlnR to b1-b2 will therefore form a DNA looping ( Figure 6E), which has been found widely involved in regulation of gene transcription, DNA recombination and replication [23][24][25].
To further test this hypothesis, we used Atomic Force Microscopy (AFM) to directly examine the change of DNA configuration after binding of GlnR to the ald promoter region ( Figure 6A) in vitro. Probably due to the short distance between b1 and b2, which in turn results in the formation of a small loop, and the aggregation of purified recombinant GlnR in solution, we failed to observe clear DNA loops after addition of GlnR ( Figure 6B). However, dramatic change of the angles of the DNA molecules was detected ( Figure 6B and 6C), indicating the existence of possible DNA loops. Besides, the position of bound GlnR on the DNA molecules was measured ( Figure 6D) and a Gaussian fit was performed. The histogram was shown by a pink line, whose peak located at ,100 nm (,300 bps) from one end of the DNA molecules, in consistent very well with the predicted position of GlnR binding sites on the DNA molecules ( = 310 bps to one end).
Double-stranded DNA is semi-flexible and DNA fragments shorter than 150 bps are difficult to form a DNA looping [26]. As the space between b1 and a2 in ald promoter is only 76 bps, other protein(s) may probably participate in bending of the DNA between the sites and facilitate GlnR binding. Although possible loops were observed with merely GlnR addition in the in vitro AFM assay, the concentration of GlnR added was obviously higher than in vivo and thus may be unable to fully reflect the in vivo situation. And the suspects include a group of Nucleoid-Associated Proteins (NAPs), e.g. Fis (Factor for Inversion Stimulation), IHF (Integration Host Factor) and HU (Heat-Unstable Nucleoid Protein), which have been shown to be able to bend DNAs [27,28] and are worth of further investigation.