The GlnD and GlnK homologues of Streptomyces coelicolor A3(2) are functionally dissimilar to their nitrogen regulatory system counterparts from enteric bacteria

Authors


  • This paper is dedicated to the memory of Professor Dr Werner Klipp, a great scientist and colleague.

Summary

Glutamine synthetase I (GSI) enzyme activity in Streptomyces coelicolor is controlled post-translationally by the adenylyltransferase (GlnE) as in enteric bacteria. Although other homologues of the Escherichia coli Ntr system (glnK, coding for a PII family protein; and glnD, coding for an uridylyltransferase) are found in the S. coelicolor genome, the regulation of the GSI activity was found to be different. The functions of glnK and glnD were analysed by specific mutants. Surprisingly, biochemical assay and two-dimensional PAGE analysis showed that modification of GSI by GlnE occurs normally in all mutant strains, and neither GlnK nor GlnD are required for the regulation of GlnE in response to nitrogen stimuli. Analysis of the post-translational regulation of GlnK in vivo by two-dimensional PAGE and mass spectrometry indicated that it is subject to both a reversible and a non-reversible modification in a direct response to nitrogen availability. The irreversible modification was identified as removal of the first three N-terminal amino acid residues of the protein, and the reversible modification as adenylylation of the conserved tyro-sine 51 residue that is known to be uridylylated in E. coli. The glnD insertion mutant expressing only the N-terminal half of GlnD was capable of adenylylating GlnK, but was unable to perform the reverse deadenylylation reaction in response to excess ammonium. The glnD null mutant completely lacked the ability to adenylylate GlnK. This work provides the first example of a PII protein that is modified by adenylylation, and demonstrates that this reaction is performed by a homologue of GlnD, previously described only as a uridylyltransferase enzyme.

Introduction

All rapidly growing organisms need to adapt to changes in their nutritional environment. This requires the existence of global regulatory networks that can sense and initiate immediate responses to these changes. In Enterobacteria, the nitrogen regulatory (Ntr) system regulates nitrogen metabolism by integrating signals that reflect cellular nitrogen and carbon status, and modulates enzyme activity and gene expression accordingly (for a review, see Merrick and Edwards, 1995; Magasanik, 1996; Reitzer, 1996a). Of particular importance in this context is the control of the glutamine synthetase I enzyme (GSI) encoded by glnA. GS enzymes assimilate ammonia via glutamate, generating glutamine. The low Km value of GSI for ammonia means that it plays a central role in nitrogen assimilation (Reitzer, 1996b). Furthermore, coupled to the glutamate synthase enzyme (GOGAT), which converts glutamine and α-ketoglutarate to two molecules of glutamate, the GS/GOGAT pathway supplies the two major precursors for synthesis of all nitrogenous compounds (Magasanik, 1996).

In enteric bacteria, the Ntr system co-ordinates both GSI activity and the expression of the GSI structural gene glnA (Fig. 1). At both levels, this control is mediated by the activity of the small, trimeric signal transmitter protein PII. The PII protein transduces the cellular nitrogen status to GlnE, the adenylyltransferase enzyme (ATase) responsible for modification of GSI, and to the sensor component NtrB of the two-component transcriptional regulatory system NtrB/NtrC. In Escherichia coli, there are two PII proteins, GlnB and GlnK, both of which are involved in this regulation, and both are subject to modification by the uridylyltransferase (UTase) enzyme GlnD (Atkinson and Ninfa, 1999), a sensor–regulator that responds to the intracellular glutamine concentration (Atkinson and Ninfa, 1998; Jiang et al., 1998a). Under conditions of nitrogen limitation (glutamine low), GlnD activity causes the PII proteins to become highly uridylylated, whereas during nitrogen excess (glutamine high), the proteins remain unuridylylated or are specifically deuridylylated by GlnD. The uridylylated PII proteins stimulate the deadenylylation of GSI, and the deuridylylated proteins stimulate GSI adenylylation. GlnB and GlnK themselves can also directly perceive nitrogen limitation via binding of α-ketoglutarate (Jiang et al., 1998a,b). Although GlnB and GlnK in E. coli are structural homologues, their functions appear to overlap only partly (Bueno et al., 1985; Atkinson and Ninfa, 1998; 1999), and the full physiological importance of the second PII protein in E. coli is not clearly understood. It has been reported that GlnK-UMP is not as readily deuridylylated by GlnD as GlnB-UMP, and so GlnB rather than GlnK is responsible for the rapid inactivation of GSI by adenylylation (Atkinson and Ninfa, 1998). A distinct function has also been described for the GlnK protein in Klebsiella pneumoniae, where it is required to relieve NifL inhibition of NifA under conditions of nitrogen limitation (He et al., 1998; Jack et al., 1999).

Figure 1.

Schematic model based on Merrick and Edwards (1995 ) showing the regulation of the activities of GSI and NtrC in response to nitrogen status in E. coli . GlnD catalyses the uridylylation of both the GlnK and GlnB PII proteins, which in turn influence the activities of the DNA-binding protein NtrC, and the GSI-modifying enzyme GlnE.

In the PII nomenclature suggested by Merrick et al., GlnK describes those PII proteins that are genetically coupled to ammonium transporter genes (amtB) (Thomas et al., 2000). It has been proposed that this conserved organization may reflect a physical interaction of GlnK and AmtB. The glnB structural genes are either monocistronically transcribed or linked to glnA or nadE (encoding for ammonia-dependent NAD synthetase enzymes). A third distinct class of paralogous PII proteins is found in nitrogen-fixing microorganisms, and these structural genes are co-localized with the nifH nitrogenase gene (Arcondéguy et al., 2001). PII proteins have been identified in archaebacteria, cyanobacteria, proteobacteria, actinobacteria, firmibacteria and even higher plants (Arcondéguy et al., 2001).

In the actinomycete Streptomyces coelicolor A3(2), a single PII protein encoded by a glnK locus was identified by the genome sequencing project (http:www.sanger.ac.ukProjectsScoelicolor). This glnK gene, although translationally coupled to a putative ammonium transporter gene (amtB), is also located immediately upstream of a glnD homologue. These three genes have recently been shown to be co-transcribed (Fink et al., 2002). A similar genetic organization has been described for Corynebacterium glutamicum (Jakoby et al., 1999) and Mycobacterium tuberculosis (Cole et al., 1998), and this genetic linkage may be a conserved feature of actinomycetes. S. coelicolor undergoes a complex differentiation process, in which aerial hyphae emerge from substrate mycelium and differentiate further to form chains of unigenomic spores. Morphological differentiation is accompanied by a physiological differentiation requiring complex regulatory networks that are still only poorly understood. We have reported previously on the Ntr homologue glnE, which encodes the ATase enzyme-regulating GSI activity (Fink et al., 1999). In this paper, we describe the effect of mutations in glnK and glnD on the regulation of GSI activity in S. coelicolor and the use of proteomics tools [two-dimensional PAGE coupled with matrix-assisted laser desorption ionization time-of-flight (MALDI-TOF) mass spectrometry] to investigate the post-translational regulation of GlnK and GSI in the parent and mutant strains. Two novel modifications of GlnK are reported.

Results

Construction of S. coelicolor SCglnD-1 and SCglnK-3 mutants shows that neither glnK nor glnD is essential for nitrogen source utilization in S. coelicolor

The promoter of the S. coelicolor amtB–glnK–glnD operon is recognized by two OmpR-like regulator proteins, GlnR and GlnRII, which both also interact with the promoter of the GSI gene, glnA (Fink et al., 2002). To analyse whether GlnK and GlnD are involved in the post-translational regulation of GSI activity in S. coelicolor, glnD was initially insertionally inactivated, and glnK replaced, by a kanamycin resistance cassette (Table 1). The resultant SCglnK-3 and SCglnD-1 mutants grew equally as well as the parent strain on surface-grown and in liquid cultures using defined minimal media containing asparagine, aspartate, glutamine, glutamate, histidine, serine, nitrate or ammonia as sole nitrogen source. SCglnK-3 sporulated precociously on surface-grown cultures of all the media tested, whereas no morphological differences were observed between the parent and SCglnD-1 (Fig. 2).

Table 1. . Strains and plasmids used in this study.
Strains/plasmidsGenotype/characteristicsReference
E. coli XL1-Blue recA1 hsdR17 relA1 lac [ F′ lacI q ZM15 Tn 10(Tet r ) ] Bullock et al. (1987)
E. coli ET12567 F dam 13:: Tn 9 dcm-6 hsdM hsdR lacY1 MacNeil et al. (1992)
S. coelicolor M145 S. coelicolor A3(2), plasmid-free derivative Kieser et al. (2000 )
SCglnD-1Mutant of S. coelicolor M145 with insertional inactivation of glnD at nucleotide
 position 1457 by an aphII cassette, kanr
This work
SCglnD-3Mutant of S. coelicolor M145 with glnD replaced by an aac(3)IV cassette, aprarThis work
SCglnK-3Mutant of S. coelicolor M145 with glnK replaced by an aphII cassette, kanrThis work
pUC18/19 bla , lacZ′ α-complementation system Vieira and Messing (1982)
pUC19aphIIpUC19 derivative with aphII; blaC. Bohrmann
pUC21 bla , lacZ′ α-complementation system Vieira and Messing (1991)
pWMH3 E. coli origin, Streptomyces origin; bla , tsr Vara et al. (1989 )
pUC21glnD2.6 kb PCR fragment of S. coelicolor glnD in pUC21 via NdeI and BamHIThis work
pRX1A 2.1 kb Kpnl–HincII fragment from pUC21 glnD in pUC18 via KpnI–HincIIThis work
pRX1a+pRX1 cleaved at the SmaI site in glnD and ligated with a 1.4 kb aphII cassette
 from pUC19aphII, cut with SmaI–HincII
This work
pWHM3glnD::kan-1A 3.5 kb EcoRI–HindIII fragment from pRX1a+, ligated to EcoRI–HindIII-
 digested pWHM3
This work
pUC18glnKup1 kb PCR fragment with 9 bp of the glnK 5′ region ligated EcoRI–BamHI to pUC18This work
pUC18glnKlow1 kb PCR fragment with 39 bp of the glnK 3′ region ligated BamHI–HindIII in pUC18This work
pUC18glnKdelThe 1 kb EcoRI–BamH1 fragment from pUC18-glnKupin EcoRI–BamHI-digested
 pUC18glnKlow
This work
pUC18glnK::kan-3pUC18glnKdel cleaved with BamHI and ligated with a 1.4 kb aphII cassetteThis work
pWHM3glnK::kan-3pWHM3 with the 3 kb EcoRI–HindIII fragment from pUC18glnK::kan-3This work
Figure 2.

Phenotypic comparison of S. coelicolor M145 with SC glnD -1, SC glnK -3 and SC glnE -1. Strains were tested for growth behaviour on different minimal media supplemented with 1% (w/v) casamino acids (A) , 5 mM ammonium (B) and 20 mM nitrate (C) . Cultures are 3 days old.

Rapid GSI adenylylation does not require glnK or glnD

Streptomyces coelicolor M145 mycelium from nitrogen-limited cultures responded to an ammonium shock with a rapid loss of GSI activity. This reduction in activity was recovered after treatment with snake venom phosphodiesterase and does not occur in the previously charac-terized ATase mutant, E4 ( Fink et al., 1999 ). This is consistent with inactivation of GSI by adenylylation in response to the increased ammonium availability. The E4 mutant is now designated SC glnE -1 for future consistency. As the S. coelicolor GlnD and GlnK proteins are potentially involved in relaying nitrogen-regulatory stimuli to the ATase GlnE, ammonium shock experiments coupled to GSI activity measurements were performed using the mutants SC glnK -3 and SC glnD -1. Strain SC glnE -1 was studied in parallel for comparison. Parent and mutant strains were grown to mid-exponential phase (OD 450 0.6) in a minimal medium supplemented with 0.2% (w/v) casamino acids as sole nitrogen source (SMM; Strauch et al., 1991 ) and then subjected to a sudden increase in extracellular ammonium concentration to 20 mM. Samples from each culture were taken before ammonium addition (time 0) and at 5, 30 and 120 min thereafter. A dramatic reduction in GSI activity was observed after ammonium shock in the parent as well as in the mutants SC glnK -3 and SC glnD -1, although the responses of these three strains differed in detail ( Fig. 3 ).

Figure 3.

GSI activity of S. coelicolor M145, the SC glnD -1, SC glnE- 1 and SC glnK -3 mutants after ammonium shock (see text for details). Mid-exponential cultures grown in supplemented minimal medium (SMM) were subjected to ammonium shock by addition to a final concentration of 20 mM NH 4 Cl. Cultures were harvested at time point 0 (before shock) and 5, 30 and 120 min after shock. GS activity was determined by the γ-glutamyltransferase assay. One unit corresponds to the formation of 1 µmol of γ-glutamylhydroxamate min −1 from 20 mM glutamine at pH 7.0 and 37°C. The experiments were carried out at least in triplicate, and standard errors are indicated.

In vivo modification of GSI is intact in SCglnD-1 and SCglnK-3

The ammonium shock assay data with SCglnD-1 and SCglnK-3 (see Fig. 3) indicated that the modification of the GSI enzyme in response to a nitrogen stimulus was intact in both mutants. In order to examine this further, GSI protein levels in cells before and after treatment with ammonium were examined using two-dimensional PAGE (Fig. 4). Protein spots 1 and 2, which are located at the same apparent molecular weight but have slightly different isoelectric point values, were both identified as GSI by tryptic digestion and MALDI-TOF peptide mass fingerprint analysis (data not shown). Only a single GSI protein spot, spot 1, was detected in the ATase mutant SCglnE-1 (Fig. 4C, 1 and 2), and this therefore corresponds to the unmodified form of GSI. Spot 2 is the result of adenylylation of GSI (i.e. GSI-AMP) by GlnE (Fig. 4A, B and D). This was confirmed by the peptide mass fingerprint data. Peptide 395–420 (containing the Y-397 residue believed to be adenylylated, by analogy with GSI in E. coli) showed the expected mass of 2807.4 Da in spot 1, but shifted to 3136.4 Da in spot 2 (data not shown). This increase of 329.0 Da corresponds to the addition of an adenylyl group to this peptide. In cultures of the parental strain during transition phase, the majority of GSI is in the unmodified form (part A1 in Fig. 4). After ammonium shock treatment, GSI is rapidly converted to GSI-AMP, as illustrated by the changes in intensities of spots 1 and 2 in parts A1 and A2 (Fig. 4). Compared with the parent strain, the pattern of modification of GSI by adenylylation was similar in both SCglnK-3 (Fig. 4D, 1 and 2) and SCglnD-1 mutants (Fig. 4B, 1 and 2), thus confirming that neither GlnK nor GlnD is directly required for the rapid inactivation of GSI activity by the ATase enzyme (GlnE).

Figure 4.

Two-dimensional PAGE analysis showing conversion of GSI to GSI-AMP by adenylylation in response to ammonium shock. Cultures were grown to early transition phase (13 h) in SMM and subjected to ammonium shock treatment by addition to a final concentration of 20 mM NH 4 Cl. Total-protein extracts from harvested cells were separated by two-dimensional PAGE using 18 cm IPG strips covering isoelectric point range pH 4.5–5.5 for the first dimension and 24 cm × 24 cm 12.5% Duracryl slab gels for the second. Subsections (≈ 4 cm × 4 cm) of silver-stained gels are shown. Spots 1 and 2 were subsequently identified by MALDI-TOF peptide mass fingerprint analysis and are GSI and GSI-AMP respectively.

GlnK in S. coelicolor is modified by adenylylation and not uridylylation

In E. coli, both GlnB and GlnK PII proteins are modified by uridylylation at tyrosine 51 (Y-51) (Rhee, 1984; van Heeswijk et al., 1996). This tyrosine residue is in a region in which PII proteins have high local identities, and the motif Y-R-G-A-E-Y (position 46–51) is conserved in the E. coli PII proteins GlnB and GlnK and also in GlnK of S. coelicolor (Fig. 5). In cyanobacteria, modification of PII occurs via phosphorylation at the serine 49 residue in the Y-R-G-S-E-Y motif (Forchhammer and de Marsac, 1994). Analysis of the proteome of S. coelicolor M145 during growth in SMM using two-dimensional PAGE showed that the GlnK protein exists in two forms that have the same apparent molecular weights, but significantly different isoelectric points (Fig. 6A). Both these protein spots were absent from the glnK null mutant strain SCglnK-3 (data not shown).

Figure 5.

Alignment of the S. coelicolor GlnK with the PII proteins GlnB and GlnK from E. coli . Black shading depicts identical amino acid residues or conserved exchanges in all three sequences, grey shading is identical residues or conserved exchanges in two of the three sequences. The arrow indicates tyrosine 51 (Y-51), the uridylylation target site in E. coli .

Figure 6.

Two-dimensional PAGE analysis of the expression and post-translational modification of GlnK in (A) S. coelicolor M145; (B) SCglnD-1 (insertion mutant); (C) SCglnD-3 (deletion mutant); (D) SCglnE-1. Total-protein extracts prepared from mycelium grown to mid- exponential phase (1) or early transition phase (2) in SMM were separated by two-dimensional PAGE (using pH 5.5–6.7 for the first dimension and 15% Duracryl for the second) and silver stained. Aliquots of the 13 h cultures were subjected to a NH4Cl shock for 5 min and then treated similarly (3). Subsections (≈ 6 cm × 8 cm) of gels are shown. Spots were subsequently identified by MALDI-TOF peptide mass fingerprint analysis (see Fig. 7) and are: spot 1, unmodified GlnK; spot 2, adenylylated GlnK (GlnK-AMP); and spot 3, GlnK modified by removal of the first three N-terminal amino acid residues (GlnK (4–112).

Tryptic digestion followed by MALDI-TOF peptide mass fingerprint analysis confirmed that the spot with the observed isoelectric point of ≈ 5.9 (spot 1 in Fig. 6) is the unmodified form of the protein (Fig. 7B). The theoretical isoelectric point for GlnK is 5.90. The tryptic peptide fragment containing the conserved Y-51 residue, peptide d in Fig. 7A, appears at the expected mass of 1191.6 Da, and peptides a–c, e and f are also present at their predicted positions. In contrast, the peptide mass fingerprint of the GlnK spot with the more acidic isoelectric point of 5.6 (spot 2 in Fig. 6) lacks peptide d at 1191.6 Da, but contains a new tryptic fragment at 1520.6 Da (peak *1 in Fig. 7C).

Figure 7.

MALDI-TOF peptide mass fingerprint analysis of tryptic digests of GlnK protein spots excised from two-dimensional gels.

A. Annotated amino acid sequence of GlnK showing the expected peptides and their calculated masses (Da) after theoretical digestion with trypsin (only peptides with masses>800 Da have been considered).

B. Peptide mass fingerprint of unmodified GlnK (spot 1 in Fig. 6 ). Peaks annotated a–f correspond to the peptides in (A) above plus 1 Da caused by ionization by protonation. Observed masses are accurate to within 50 p.p.m. Peak g is peptide c plus amino acids Q-39 and R-40 (i.e. +284.1 Da) and is the result of incomplete digestion by trypsin at the site on the C-terminal side of R-40. Peak h is peptide f, but tryptophan W-92 has been converted to formylkyneurenine by reaction with formic acid causing a mass increase of 32.0 Da. Peak T is a peptide resulting from autodigestion of trypsin, and peak M is a sample matrix signal.

C. Peptide mass fingerprint of GlnK modified by adenylylation (GlnK-AMP; spot 2 in Fig. 6 ). Peaks are annotated as in (B), except that *1 corresponds to d plus AMP (i.e. +329.0 Da).

D. Peptide mass fingerprint of GlnK modified by removal of the first three N-terminal amino acid residues [GlnK (4–112); spot 3 in Fig. 6 ]. Peaks are annotated as in (B), except that *2 corresponds to peptide a lacking the L-3 amino acid (i.e. minus 113.0 Da).

Amino acid sequence analysis of this peak using Q-TOF mass spectrometry produced a sequence identical to peptide d (data not shown). The increase in mass of peptide d by 329.0 Da in the modified GlnK protein is diagnostic of the addition of an adenylyl group (the addition of a uridylyl group would cause a mass increase of 306.2 Da and a peak shift to 1497.8 Da). The Q-TOF fragmentation pattern indicated that the modification occurs on amino acid Y-51 and is also entirely consistent with modification by adenylylation rather than uridylylation. Adenylylation of GlnK in S. coelicolor is a novel form of PII modification.

GlnK is adenylylated in response to nitrogen availability

During exponential growth in SMM (8 h culture), unmodified GlnK was present in M145 at low levels, and GlnK-AMP could hardly be detected (Fig. 6A, 1). However, when culture growth slowed on entry into transition phase (Fig. 6A, 2; 13 h culture), the overall amount of GlnK, as calculated from densitometry measurements, increased 29-fold. At transition phase, GlnK-AMP was the more abundant of the two species. Spot densitometry measurements estimate the ratio of GlnK-AMP to GlnK at this point as 85:15. Little change was observed for this ratio in early stationary phase cultures (data not shown).

The addition of 20 mM NH4Cl to the 13 h culture caused a rapid and dramatic decrease in GlnK-AMP and a proportional increase in unmodified GlnK (Fig. 6A, 3). The ratio of GlnK-AMP to GlnK was reversed from 85:15 immediately before shock to 5:95 5 min after shock. This indicates that GlnK-AMP is actively deadenylylated in response to the sudden increase in the availability of ammonium.

GlnK is adenylylated in the insertion mutant SCglnD-1, but is not rapidly deadenylylated in response to excess ammonium

Adenylylation of PII proteins has not been described previously. In E. coli and other Ntr-regulated bacteria, PII proteins are uridylylated by the UTase enzyme GlnD. In S. coelicolor, there is a glnD homologue located immediately downstream of glnK that is co-transcribed as part of the same operon. In order to determine whether the glnD homologue is responsible for the novel adenylyla-tion of GlnK, the pattern of post-translational modification of GlnK was analysed in strain SCglnD-1 using two-dimensional PAGE (Fig. 6B, 1–3). The mutant SCglnE-1 was studied similarly, as GlnE is a known ATase enzyme involved in the regulation of nitrogen metabolism (Fig. 6D, 1–3).

Figure 6B , 2 and D, 2 clearly shows GlnK-AMP in both SC glnD -1 and SC glnE -1 mutants. Although the pattern of GlnK and GlnK-AMP expression in SMM cultures of SC glnE -1 is similar to that seen in the parent strain (compare Fig. 6D , 1 and 2 with Fig. 6A , 1 and 2), the transition phase sample of SC glnD -1 shows a very high level of unmodified GlnK (compare Fig. 6B , 2 with Fig. 6A , 2), and the ratio of GlnK-AMP to unmodified GlnK does not change significantly after ammonium shock treatment, remaining at ≈ 50:50 in both Fig. 6B , 2 and 3. Thus, although adenylylation of GlnK still occurs in strain SC glnD -1, the adaptive deadenylylation reaction detected in the parent strain has been lost. Surprisingly, this is also observed, albeit less pronounced, in SC glnE -1: the ratio of GlnK-AMP to GlnK only changed from ≈ 90:10 before shock ( Fig. 6D , 2) to 60:40 5 min thereafter ( Fig. 6D, 3).

The glnD null mutant SCglnD-3 cannot adenylylate GlnK, although GSI modification is still intact

The disruption of glnD in SCglnD-1 leaves approximately half the gene intact, encoding the first 482 amino acids. The observed deregulated adenylylation of GlnK in strain SCglnD-1 implies that the truncated GlnD possesses some catalytic activity, but that it is not capable of responding appropriately to changes in nitrogen status. Alternatively, there is the formal possibility that GlnD is not responsible for the adenylylation of GlnK, but is involved in the regulation of the ATase activity. To investigate this further, we constructed a glnD null mutant using a polymerase chain reaction (PCR)-targeted mutagenesis method (see Experimental procedures). The null mutation in strain SCglnD-3 was verified by Southern analysis and PCR (data not shown). A renewed analysis of the post-translational modification of GlnK revealed a complete lack of adenylylation in strain SCglnD-3 (Fig. 6C, 1–3).

The biochemical and phenotypic analysis of the glnD insertion mutant SCglnD-1 was repeated using the SCglnD-3 null mutant strain. SCglnD-3 grew normally on minimal media containing asparagine, aspartate, glutamine, glutamate, histidine, serine, nitrate or ammonia as the sole nitrogen source, similar to SCglnD-1 and the parental strain, and no effect on sporulation was observed in the glnD-3 mutant (data not shown). GSI activity was downregulated normally in SCglnD-3 upon ammonium shock treatment (Fig. 3), and two-dimensional PAGE analysis confirmed that this was associated with conversion of GSI to GSI-AMP (data not shown).

GlnK in S. coelicolor is proteolytically processed in response to ammonium shock treatment

Two-dimensional PAGE analyses of the cultures subjected to ammonium shock treatment (Fig. 6A, 3, B, 3 and D, 3) revealed the presence of a new protein spot (spot 3 in Fig. 6) located immediately below GlnK-AMP (spot 2 in Fig. 6). This was slightly more pronounced in M145 and the SCglnD-1 mutant than in SCglnE-1. MALDI-TOF peptide mass fingerprint analysis revealed that spot 3 is another modified form of GlnK (Fig. 7D). Peptide d (with the tyrosine 51 residue) is present at 1191.6 Da, indicating that this form of GlnK is not adenylylated. Peptides b, c, e and f are again all present at their expected masses, but there is no peak corresponding to the N-terminal peptide a, which should be present at 1133.7 Da. However, a new peak is present at 1020.6 Da (*2 in Fig. 8D), which corresponds accurately to the predicted mass of peptide a, after removal of the three N-terminal residues MKL from GlnK (predicted difference 113.08 Da; ob-served difference 113.10 Da). The theoretical values for the isoelectric point and molecular weight of GlnK protein N-terminally truncated in this way are 5.59 and 11.8 kDa, in good agreement with the position of spot 3 on the two-dimensional gel. Truncated GlnK was not observed in the SCglnD-3 proteome (Fig. 6C, 3), and two-dimensional PAGE analysis of extracts from M145 subjected to ammonium shock at different points during growth indicated that it is generally only produced in cells from cultures that have already entered stationary phase.

Figure 8.

Two-dimensional PAGE analysis showing the induction of a truncated form of glyceraldehyde-3-phosphate dehydrogenase (arrow) on ammonium shock. Different subsections of the two-dimensional PAGE gels produced for the 13 h (before shock) and 13 h + 5 min NH 4 Cl shock (after shock) samples shown in Fig. 6 are presented.

Glyceraldehyde-3-phosphate dehydrogenase is also processed proteolytically in response to ammonium shock treatment

Image analysis of the complete two-dimensional PAGE gels produced for studying the expression and post-translational modification of GlnK identified a number of other interesting differences in the protein expression patterns. Figure 8 shows one example of a protein spot present at low levels in M145 but markedly increased after ammonium shock treatment (compare Fig. 8A, 1 with Fig. 8A, 2). In the SCglnD-1 and SCglnE-1 mutants, the same spot appears after ammonium shock, but here it is significantly more abundant before the addition of ammonium compared with the parental strain (compare Fig. 8B, 1 and Fig. 8C, 1 with Fig. 8A, 1). Tryptic digestion followed by MALDI-TOF peptide mass fingerprint analysis identified this protein as glyceraldehyde-3-phosphate dehydrogenase (GAPDH), a key enzyme in central carbon metabolism responsible for the first step in the second phase of glycolysis. From its deduced amino acid sequence (http:www.sanger.ac.ukProjectsScoelicolor), GAPDH was predicted to appear at isoelectric point pH 5.25 and with a molecular weight of 36.2 kDa and, indeed, it has been identified previously at this position (data not shown). These data clearly differ from those documented in Fig. 8. The full-length protein is 336 amino acids long, but the MALDI-TOF data only show peptides corresponding to amino acids 2–225, suggesting that a C-terminal portion of the protein has been removed. The protein detected here at pI and molecular weight co-ordinates pH 6.5 and 27 kDa, respectively, therefore appears to be a truncated derivative of GAPDH.

Discussion

In previous work, we have identified a homologue of the Ntr system ATase enzyme GlnE in S. coelicolor and demonstrated that it is involved in the regulation of GSI enzyme activity in this organism (Fink et al., 1999). In this study, we investigated whether post-translational regulation of GSI is also dependent on the Ntr system homologues GlnD and GlnK in a manner reminiscent of their roles in enteric bacteria. In E. coli, GlnD functions as the predominant sensor of nitrogen metabolic status, transmitting the signal via uridylylation of the PII proteins GlnB and GlnK, which in turn influence, among other things, the adenylylation of GSI by GlnE. Inactivation of glnD or a double mutation in the glnB and glnK genes results in the so-called NtrC phenotype characterized by severely impaired growth on poorer nitrogen sources such as arginine, a failure to utilize nitrate as the sole nitrogen source and an altered requirement for glutamine (Foor et al., 1978; Bueno et al., 1985; Edwards and Merrick, 1995; Atkinson and Ninfa, 1998). In contrast, the S. coelicolor glnD and glnK mutants described here (SCglnD-1, SCglnD-3 and SCglnK-3) do not exhibit these phenotypes, and all grew as vigorously as the parental strain on (and in) the defined minimal media tested, independent of the nitrogen source. The function of GlnD in S. coelicolor therefore appears not to be analogous to that of GlnD in E. coli.

More detailed characterization of the mutants confirmed that the S. coelicolor GlnK and GlnD proteins are not essential for the transfer of sensory information about nitrogen status to GSI via GlnE. Biochemical assays of the GSI enzyme activity in cells harvested before and after administration of an ammonium shock showed a dramatic decrease in activity in the parental strain and in the SCglnD-1, SCglnD-3 and SCglnK-3 mutant strains (see Fig. 3). Disruption of the sensory relay, by for instance mutation in glnE, results in GSI activity being unaffected by ammonium addition (Fink et al., 1999). These biochemical data were supported by the results of two-dimensional PAGE analysis of the post-translational adenylylation of GSI (producing GSI-AMP) in a parallel set of ammonium shock experiments (see Fig. 4). SCglnD-1, SCglnD-3 and SCglnK-3 all showed similar patterns of GSI and GSI-AMP expression to the parent strain, whereas SCglnE-1 completely lacked any GSI-AMP under any conditions. This contrasts with results in E. coli, where mutation in glnD causes deregulation of GSI activity because the enzyme is constantly adenylylated as a result of a permanent interaction with the unmodified PII proteins (Atkinson and Ninfa, 1998).

Like S. coelicolor, Azotobacter vinelandii also possesses a single PII family protein that is a GlnK homologue. However, disruption of glnK or glnD (Meletzus et al., 1998; Colnaghi et al., 2001) in this strain is lethal. Colnaghi et al. (2001) have proposed that the dependency on GlnD and GlnK for cell viability is a consequence of an absolute requirement for GlnK-UMP in stimulating the deadenylylation of GSI-AMP by the GlnE ATase enzyme. Our study indicates that the GlnE enzyme in S. coelicolor is not dependent on either GlnD or GlnK for its function, and the question arises as to how GlnE-dependent GSI modification is regulated. Either GlnE senses the nutritional status directly through allosteric interaction with effector molecules such as glutamine, or GlnE regulation requires an unknown mechanism of post-translational control. Independent observations suggest that ATase enzymes can be regulated without the involvement of GlnK or GlnB. In Azospirillum brasilense, PII proteins are not essential for either adenylylation or deadenylyation of GSI (de Zamaroczy, 1998). In vitro experiments with the isolated C-terminal adenylylase domain of the E. coli ATase have shown that this truncated protein is capable of GSI adenylylation in a glutamine-dependent and GlnB-independent manner (Jaggi et al., 1997).

To our knowledge, GlnD proteins have always been described as sensor proteins responsible for the addition and removal of uridylyl groups to PII regulatory proteins in response to changes in cellular nitrogen status (for example, see Arcondéguy et al., 2001). Comparison of the post-translational modification of GlnK on administration of an ammonium shock to cultures of the glnD null mutant SCglnD-3 and the parental strain M145 (see Fig. 4) clearly shows that GlnD is required for the modification and demodification of GlnK. However, MALDI-TOF and Q-TOF mass spectrometry analysis of the modified form of GlnK clearly identifies the modification as adenylylation and not uridylylation. The kinetic properties of the purified GlnD UTase enzyme from E. coli, including its nucleotide substrate specificity, have been extensively characterized by Jiang et al. (1998c). In vitro studies indicated that the nucleotide specificity was not strict, and PII covalently containing AMP (or CMP, GMP and even dAMP, dCMP) could be formed readily. However, the enzyme strongly favours UTP as its substrate and, when all nucleotides are available at approximately their in vivo concentrations, PII-UMP constituted 94% of the modified PII, but 3% PII-AMP was also detected. Jiang et al. (1998c) speculated that the minor forms of modified PII could therefore be formed in vivo and that PII-AMP could be functionally different from PII-UMP. In S. coelicolor, in which the covalently modified PII protein species is GlnK-AMP, the absence of an effect on control of the adenylylation status of GSI by GlnE in the glnK mutant and the glnD null mutant strain supports this proposal.

Although the role of GlnD and GlnK in the regulation of nitrogen metabolism in S. coelicolor remains unclear, it is evident that these proteins are distinct from all their previously described counterparts. An amtB–glnK linkage has been described in a number of species (Thomas et al., 2000), but the genetic co-localization of amtB, glnK and glnD appears to be a characteristic unique to actinobacteria (see Arcondéguy et al., 2001). Studies in S. coelicolor (Fink et al., 2002) and in the related actinomycete Corynebacterium glutamicum (Nolden et al., 2001) have shown that these operons are regulated by nitrogen availability, and glnD expression is therefore not constitutive and nitrogen independent as in the enteric Ntr model (van Heeswijk et al., 1996). There is a further distinction, however, even within the actinobacteria, as the amtB–glnK–glnD operon is controlled on the one hand by repression in C. glutamicum (Jakoby et al., 2000) and, on the other, by activation in S. coelicolor (Fink et al., 2002). Analysis of the genome sequence indicates that, in addition to glnA, S. coelicolor possesses a second known glutamine synthetase gene (glnII) that resembles the eukaryotic synthetases and also three putative glutamine synthetase genes. The regulation of nitrogen metabolism is therefore likely to be complex.

In this study, we have identified two novel forms of GlnK modification not previously reported for PII family proteins. An alternative mechanism for PII modification, phosphorylation, has so far only been described for cyanobacteria (Forchhammer and de Marsac, 1994), in which GlnB does not act as a post-translational regulator of GSI, but rather co-ordinates carbon and nitrogen metabolism (Hisbergues et al., 1999). It will be interesting to discover whether the nucleotide preference of the GlnD enzyme in S. coelicolor for ATP is Streptomyces specific or more widely distributed, occurring for instance in other actinomycetes. An additional GlnK regulation mechanism in S. coelicolor appears to involve specific cleavage of the first three N-terminal amino acids of the protein. Amino acid residues 3, 5, 60 and 62 have been reported to interact in the cavity formed by PII protein trimerization (Xu et al., 1998). Amino acid residues 3 and 5 (K-3 and D-5) in GlnB proteins are charged residues, whereas neutral residues (L-3 and T-5) characterize these positions in GlnK proteins (Xu et al., 1998; Jack et al., 1999). The observation that GlnK is cleaved after its third N-terminal residue upon ammonium shock therefore raises the question as to whether this is specific only for GlnK-like PII proteins, and whether this can directly affect trimer formation or stability.

Experimental procedures

Bacterial strains, plasmids, growth conditions

Strains and plasmids used in this study are listed in Table 1. Streptomyces coelicolor strains were cultivated on R2YE (Kieser et al., 2000), HA (Schwartz et al., 1996), SFM (Kieser et al., 2000) and S-medium (Okanishi et al., 1974) as required. If necessary, 25 µg ml−1 kanamycin or 25 µg ml−1 thiostrepton was added to the growth medium. For physiological studies, nitrogen-limited N-Evans medium was used (Fink et al., 2002), or nitrate in N-Evans was substituted for 10 mM ammonium, 20 mM glutamate, 20 mM glutamine or 1% (w/v) casamino acids (Difco).

Escherichia coli was cultivated at 37°C in LB medium or on LB agar ( Miller, 1972 ). Ampicillin (150 µg ml −1 ) or kanamycin (50 µg ml −1 ) was added if necessary.

Construction of mutants SCglnK-3 and SCglnD-1

For the construction of SCglnK-3, the glnK upstream and downstream regions were amplified by PCR with Pwo polymerase enzyme (Peqlab). A 1 kb PCR fragment, encompassing the upstream region and 9 bp of glnK, was amplified with primers 5′-TTGAATTCTGGTCTTCCAGCTGATGTT-3′ and 5′-TTGGATCCGAGCTTCATGCGTCCACCT-3′, and a 1 kb downstream fragment with 39 bp of glnK was amplified with primers 5′-AAAGGATCCACCGTCGACACGGCCGTACG-3′ and 5′-TTTAAGCTTGTACGGACCGCGAGCGGAGC-3′. The PCR products were cloned by their respective EcoRI–BamHI and BamHI–HindIII primer sites to pUC18 to give pUC18glnKup and pUC18glnKlow. The EcoRI–BamHI fragment from pUC18glnKup was excised and ligated to pUC18glnKlow, yielding pUC18glnKdel. A 1.4 kb aphII cassette with BamHI cleaved ends was introduced into the BamHI site of pUC18glnKdel, generating pUC18glnK::kan-3 (where ‘-3’ specifies the full replacement of glnK by the aphII cassette). The 3 kb disruption construct was transferred as an EcoRI–HindIII fragment to the Streptomyces suicide vector pWHM3. The transformation and selection procedure for mutation of S. coelicolor was as described by Fink et al. (1999).

For construction of SCglnD-1, a 2.6 kb glnD fragment was amplified with Pwo polymerase (Peqlab) and the primer combination 5′-AAAAAAAACATATGACGAGTACGGGCGTGACG GAC-3′ and 5′-TTTGGATCCCACGGCCGCGTAGGTCCTCG CGT C-3′. This fragment was ligated to an NdeI- and BamHI-restricted pUC21 plasmid. A 2.1 kb partial glnD fragment was removed from pUC21glnD by KpnI and HincII digestion and ligated to a likewise restricted pUC18, giving rise to pRX1. The SmaI site in pRX1 (1436 bp downstream of the glnD translational start) was used for insertion of a SmaI–HincII-restricted aphII cassette, generating the plasmid pRX1a+. The 3.5 kb disruption cassette in pRX1+ was transferred as an EcoRI–BamHI fragment to the suicide vector pWHM3, generating pWHM3glnD::kan-1 (where ‘-1’ specifies disruption of the gene via insertional inactivation). Transformation and selection for the mutation were as described by Fink et al. (1999).

Construction of SCglnD-3 by PCR targeted mutagenesis

For the construction of SCglnD-3, a PCR targeting approach originally described by Datsenko and Wanner (2000) and adapted for use in S. coelicolor was carried out (B. Gust et al., manuscript in preparation). A 1370 bp apramycin resistance (aac(3)IV; Blondelet-Rouault et al., 1997) disruption cassette flanked by 39 bp sequences homologous to glnD (underlined) was amplified with primer combination 5′-GGCCACC G A CGCCGCCTCGTCCCCGGGCAGCGGCGT ACCTGTAG GCTGGAGCTGCTTC-3′ and 5′-CGACAACGGAACGGGAG TCGCCGGGTGACG A GT A CGGGCA ATCCGGGGATCCGT CGACC-3′. This fragment was integrated by λ RED-mediated recombination into E. coli BW25113 containing the S. coelicolor cosmid SC7A1 (Redenbach et al., 1996), generating a glnD deletion at basepair position 1–2439 (amino acids 1–813). The disruption cassette also contained an origin of transfer (oriT) allowing conjugation to be used to introduce the PCR-targeted cosmid DNA into S. coelicolor. Replacement of the parental allele by double cross-over and screening for the mutant SCglnD-3 were as described by Kieser et al. (2000).

Ammonium shock and GS assays

Dense spore suspensions of S. coelicolor M145 and mutants SCglnK-2 and SCglnD-1 were pregerminated as in procedure 1 in Kieser et al. (2000). Cultures were grown in SMM to an OD450 of 0.6. Ammonium shock was performed by the addition of NH4Cl to a final concentration of 20 mM. Cell harvesting, breakage, crude extract preparation and GS assays were as described by Fink et al. (1999).

Protein extraction and two-dimensional gel electrophoresis

Mycelium for protein extraction was harvested from cultures by brief centrifugation (30 s at 4000 r.p.m.) at room temperature and immediately frozen in liquid nitrogen. Typically, mycelium from 10 ml culture aliquots was collected, and the transfer time from culture flask to frozen sample was 1.5 min. Frozen cells were thawed on ice in 5 ml of washing buffer (40 mM Tris, pH 9.0, 1 mM EGTA, 1 mM EDTA) then pelleted by centrifugation. Washed cells were resuspended in 400 µl of denaturing isoelectric focusing (IEF) buffer [7 M urea, 2 M thiourea, 4% CHAPS, 40 mM Tris, pH 9.0, 1 mM EDTA, 50 mM dithiothreitol (DTT), 4 mM Pefabloc SC protease inhibitor (Roche)] and disrupted by sonication (Sanyo Soniprep 150; 10 × 2 s bursts at amplitude 7.5 microns) while cooling in an ethanol–ice bath. Cell debris was removed by centrifugation (10 min, 13000 r.p.m., 4°C), and the protein extract was stored frozen in aliquots at −80°C until use. For first-dimension IEF, 18 cm IPG strips, pH 4.5–5.5 or 5.5–6.7 (Amersham Biosciences), were rehydrated overnight in IEF buffer containing 1% ampholytes according to the manufacturer's instructions using a Phaser isoelectric focusing unit (Genomic Solutions) set at 20 V. Protein samples to be separated were applied to the rehydrated strips at the anodic end, soaked in a 5–10 mm section of IEF electrode strip (Amersham Biosciences). Separation was performed for 120 000 volt-hours with a maximum voltage of 5000 V. After IEF, IPG strips were equilibrated for the second dimension for 15 min in IPG equilibration buffer (50 mM Tris, pH 6.8, 6 M urea, 30% glycerol, 1% SDS and 0.01% bromophenol blue) plus 80 mM DTT, then for 10 min in IPG equilibration buffer plus 135 mM iodoacetamide. Approximately 1 cm was removed from the anodic end of each equilibrated strip before application to the top of a vertical 12.5% SDS-PAGE gel for second-dimension separation. Gels were stained with colloidal Coomassie G-250 (Neuhoff et al., 1988) or silver nitrate (Rabilloud, 1992) and scanned in a ProXPRESS proteomic imaging system (Perkin-Elmer). Image analysis was performed using phoretix 2D version 5.1 (NonLinear Dynamics).

Protein identification using mass spectrometry

Protein spots of interest were excised from colloidal Coomassie-stained gels, in-gel digested with trypsin and identified by MALDI-TOF peptide mass fingerprint analysis. Excised gel pieces (≈ 1-mm-diameter circles cut from 1-mm-thick gels) were washed twice with 100 µl of 50 mM ammonium bicarbonate for 15 min, once with 100 µl of 20% acetonitrile−40 mM ammonium bicarbonate for 15 min and once with 100% acetonitrile. Washed gel pieces were allowed to air dry for 10 min before being rehydrated in 5 µl of 10 µg ml−1 trypsin (sequencing grade modified; Promega) in 10 mM ammonium bicarbonate. Digestion was performed at 37°C for 4 h, then stopped by the addition of 5 µl of 5% formic acid. Extraction of peptides into this solution was encouraged by a sonic water bath treatment for 20 min. Peptide extract (0.5 µl) was spotted on to a thin layer of α-cyano-4-hydroxycinnnamic acid applied to a MALDI-TOF sample template and analysed by MALDI-TOF mass spectrometry (Bruker Reflex III) using an accelerating voltage of 25 kV. Samples were externally calibrated using a standard mixture of six peptides ranging in mass from 1046.5423 Da to 3494.6500 Da. Identification of proteins from MALDI-TOF peptide mass fingerprint data was performed using the ‘Mascot’ search engine at http:www.matrixscience.com. Q-TOF mass spectrometry for the further characterization of peptides of interest was performed using a Micromass Q-TOF-2 mass spectrometer.

Acknowledgements

We thank R. Ort-Winklbauer for excellent technical assistance, and Mike Naldrett and Andrew Bottrill for help with mass spectrometry. D.F. is grateful for a scholarship from the Studienstiftung des Deutschen Volkes. H.U.R. is supported by grant 99NR068 (Fachagentur für Nachwachsende Rohstoffe). A.H. was supported by grant 208/FGT11408 from the BBSRC's Technologies for Functional Genomics Initiative; B.G. was supported by grant 208/IGF12434 from the BBSRC's Investigating Gene Function Initiative; and K.C. was supported by a competitive strategic grant from the BBSRC to the John Innes Centre.

Ancillary