Interplay between CRP-cAMP and PII-Ntr systems forms novel regulatory network between carbon metabolism and nitrogen assimilation in Escherichia coli.

In Escherichia coli, utilization of carbon sources is regulated by the phosphoenolpyruvate-dependent phosphotransferase system (PTS), which modulates the intracellular levels of cAMP. The cAMP receptor protein (CRP) controls the transcription of many catabolic genes. The availability of nitrogen is sensed by the PII protein at the level of intracellular glutamine. Glutamine is transported mainly by GlnHPQ, and synthesized by glutamine synthetase (GS) encoded by glnA. Previous studies suggest that CRP affects nitrogen assimilation. Here we showed that at least two mechanisms are involved. First, CRP activates glnHp1 via synergistic binding with sigma 70 RNA polymerase (Esigma(70)) and represses glnHp2. As a consequence, in the presence of glutamine, the overall enhancement of glnHPQ expression alters GlnB signalling and de-activates glnAp2. Second, in vitro studies show that CRP can be recruited by sigma 54 holoenzyme (Esigma(54)) to a site centred at -51.5 upstream of glnAp2. CRP-induced DNA-bending prevents the nitrogen regulation protein C (NtrC) activator from approaching the activator-accessible face of the promoter-bound Esigma(54) closed complex, and inhibits glnAp2. Therefore, as the major transcriptional effector of the 'glucose effect', CRP affects both the signal transduction pathway and the overall geometry of the transcriptional machinery of components of the nitrogen regulon.

In Escherichia coli and related bacteria, the presence of glucose in the growth medium prevents the utilization of other carbohydrates, the so-called classic ‘glucose effect’ (1). Preferential utilization of different carbon sources is regulated by the phosphoenolpyruvate-dependent phosphotransferase system (PTS) which modulates the intracellular levels of cAMP (2). cAMP levels are high when cells are grown on non-PTS carbon sources such as glycerol. When grown on carbohydrates such as glucose, cAMP levels are low (2,3). The activity of CRP can only be triggered by binding to cAMP (3), and therefore, in this article in order to simplify the text, the liganded CRP will be further designated as CRP, not CRP-cAMP. CRP contains three transcription activation regions and is considered as a proximal activator interacting over short distances with the major form of σ70-RNA polymerase (Eσ70), often as the promoters of sugar catabolic genes. As a CRP point mutation defective in its activation region I (ARI), CRPH159L is unable to interact with the αCTD of Eσ70, but is still able to bind and bend DNA in a manner similar to its wild-type counterpart (4). Location of CRP-binding sites upstream of CRP-dependent promoters, their affinity for CRP and the intrinsic promoter strength determine the extent of activation (3). Detailed three-dimensional models of CRP–Eσ70–promoter complexes constructed by Lawson et al. (5) have yielded insights into how CRP binds DNA and activates transcription.

Assimilation of ammonia and acquisition of various nitrogen sources is tightly controlled in E. coli. A key operon, glnALG, encodes a key enzyme of nitrogen assimilation, glutamine synthetase (GS), and two regulatory proteins that control expression of the Ntr regulon, NtrB (also called NRII) and NtrC (also called NRI) (6). NtrC is phosphorylated by NtrB to form a hexamer (7,8) of NtrC-phosphate (NtrC-P) under nitrogen-deficient growth conditions. The expression of the glnALG operon is controlled by tandem promoters (glnAp1 and glnAp2). glnAp1, located at 115 bp upstream of glnAp2, is a σ70-dependent weak promoter. Its transcription can be activated by CRP and blocked by NtrC-P (9). glnAp2 is transcribed by another form of RNA polymerase (Eσ54) and is activated by NtrC-P (6,10,11). Taken together, glnAp2 is responsible for increasing the levels of glnA transcription under nitrogen-deficient growth conditions (11).

Glutamine transport in E. coli has been shown to occur through a well-characterized high-affinity-binding-protein-dependent system (GlnHPQ) and poorly characterized low-affinity system(s). The glnHPQ operon is under the control of tandem promoters (glnHp1p2). glnHp1 is σ70-dependent, while glnHp2 is σ54- and NtrC-P-dependent (12,13).

The intracellular glutamine status is integrated and transduced by the PII signalling protein. Escherichia coli possesses two closely related PII paralogues, GlnB and GlnK. GlnB is produced constitutively, and it regulates the NtrB/NtrC two-component system (14,15). Its non-modified form can bind NtrB, which dephosphorylates the response regulator NtrC, thus switching off transcription of glnAp2 (16–18), and of other NtrC-dependent promoters including glnHp2 and the glnK promoter (19). Low intracellular glutamine levels, signalling nitrogen deficiency, leads to uridylylation of GlnB (20). The modified complex GlnB–UMP is not able to combine with NtrB. In this state, by phosphoryl transfer, NtrC is phosphorylated by NtrB and so activates transcription (17).

For glnAp2, NtrC-P binds to enhancer-like sequences centred at −108 and −140 and forms an oligomer (7) to activate transcription via direct interaction with promoter bound Eσ54. These NtrC-P binding sites are needed for activation at low intracellular NtrC-P concentration, while high concentrations of NtrC-P can stimulate transcription at glnAp2 without specific binding sites (6,21). In both cases, the NtrC-P must approach promoter bound Eσ54 from the opposite side of the DNA to which Eσ54 is bound (22).

Previously, we have shown that CRP down-regulates glnAp2 in E. coli (9). The latest studies suggest that this down-regulation may work through different mechanisms for different σ54-dependent promoters. In the case of glnAp2, the effect of carbon source on NtrC-dependent gene expression can be mediated through CRP-dependent glutamine uptake, thereby affecting the glutamine-sensitive UTase–PII signalling system (16). However, the mechanisms behind the CRP-dependent glutamine uptake remain unknown.

In this study, we establish that CRP influences nitrogen regulation in E. coli by at least two mechanisms. First, with glutamine as the nitrogen source, glnHp1 is activated by CRP in vivo and in vitro. Through glnHPQ-dependent GlnB signalling, CRP acts to decrease the amount of the phosphorylated NtrC activator, which in turn causes a decrease in glnAp2 expression. Second, in vitro CRP is recruited to a sequence upstream of the core glnAp2 promoter through direct interaction between the ARI of CRP and the αCTD of Eσ54 resulting in a reduced activity of glnAp2. The latter direct down-regulatory effect is evident in vivo when a limiting amount of ammonium is used as the nitrogen source.

MATERIALS AND METHODS

Bacterial strains and plasmids

Bacterial strains and plasmids used in this study are listed in Table 1. The entire glnHPQ operon in TP2339-1 was replaced by bla gene using pKO3 system (24). This results in strain BD4000.

Table 1.

Bacterial strains and plasmids used in this work

Strain/plasmid	Relevant characteristics	Source/reference
E. coli strains
TP2339-1	F−, xyl, cya, crp-39, lacΔX74, argH1, glp	Tian et al. (9)
BD4000	F−, xyl, cya, crp-39, lacΔX74, argH1, glp, glnHPQ	This article
Plasmids
pKU101	glnAp2mCRP::lacZYA fusion in pGD926	Tian et al. (9)
pKU101H	entire glnHPQ operon inserted into pKU101	This article
pKU550	glnHp1p2::lacZYA fusion in pGD926	This article
pLG339CRP	pLG339 carrying crp under the control of the crp promoter	Kolb. A (23)
pLG339CRPH159L	H159L mutant in pLG339CRP	Busby. S (28)
pLG339ΔRS	pLG339 with EcoRI/SalI internal deletion	Busby. S (28)

mCRP, mutated CRP-binding site; ::, novel joint; pLG339ΔRS is the crp deleted version of plasmid pLG339CRP, as the negative control for CRP and its ARI mutant CRPH159L.

For construction of plasmid pKU550, the entire glnHp1p2 region together with the first 7 codons of the glnHPQ open reading frame (ORF) (from −289 to +63 of glnHp2) was amplified as a 364-bp DNA fragment by PCR using E. coli strain MG1655 chromosome DNA as template and 5′-CCCAAGCTTGCAGCCCACCTCAACGCAC-3′ (p#1) and 5′-CGGGATCCACTTTTAATACAGACTTCATAG-3′ (p#2) as primers. The 364-bp DNA fragment was digested with HindIII and BamHI, cloned into pUC18 and verified by DNA sequencing. The 364-bp HindIII–BamHI DNA fragment was cloned into pGD926, and the glnA ORF was fused in-frame with the eigth codon of the lacZ reporter gene. This results in plasmid pKU550.

The entire glnHPQ operon was amplified by PCR using p#1 and 5′-CCCAAGCTTAATAAGACACATTGCCTG-3′ (p#3) as primers and E. coli strain MG1655 chromosome DNA as template. The PCR product was restricted with HindIII, cloned into pUC18 and verified by DNA sequencing. The HindIII DNA fragment was cloned into the glnAp2 reporter plasmid pKU101. This results in plasmid pKU101H, which complements strain BD4100 for glnHPQ.

A CRP consensus-binding site was introduced into position −50.5 nucleotides upstream of the glnAp2 by Takara Biotechnology (Dalian) and the promoter was designed as CC-50.5 according to the location of the centre of the CRP consensus-binding site (22).

Growth media and enzyme assays

M63 minimal medium was prepared as previously described (25) with glycerol (0.4% w/v) as the sole carbon source, and either glutamine (0.2% w/v) or 0.2 mM ammonium as nitrogen source, in the absence/presence of exogenous cAMP (2 mM), as indicated. Cells were grown at 30°C under aerobic conditions. The growth of the cells was measured spectrophotometrically by following the optical density of the culture (OD600). The cultures that were used for measurements were initiated by diluting a cell suspension from an overnight-grown culture in the appropriate medium to an initial OD600 of 0.04–0.05. When the day-culture reached an OD600 between 0.8 and 1.0, β-galactosidase assays were performed as described before (25).

Primer extension and in vivo KMnO4 footprinting experiments

As described before (9), two primers (5′-CCCAGTCACGACGTTGTAAAACG-3′ (p#4) and 5′-CAACCGTTCCGCCAGTTTGCGT-3′ (p#5)) hybridize with the structural gene of lacZ (p#4, for probing transcripts from glnHp1p2) and tetR (p#5, for the control) on pGD926-derived plasmids, respectively, in primer extension experiments.

Protein purification, in vitro transcription assays and DNase I footprinting

As described before (22).

Real time RT-PCR assays

Total RNA was isolated from 5 ml E. coli cells (TP2339-1 with pKU101 and pLG339CRP/pLG339CRPH159L) at OD600 0.8 using RNeasy® Mini Kit (QIAGEN). About 1.5 μg of total DNase I-treated RNA were reverse-transcribed using SuperScriptII™ RT (Invitrogen) according to the manufacturer's instructions. A real-time PCR was performed with each pair of primers (5′-CGGATACCGCCTTCGTTC-3′ (p#6) and 5′-ACCGCTCTTCACAGCAACC-3′ (p#7) that were designed for quantifying transcripts from glnHp1p2; 5′-CCAGAATGGTGCATCTTCAGG-3′ (p#8) and 5′-TCAGCTCTTTAGCGATGGCAG-3′ (p#9) were designed for quantifying transcripts from glnHp1; 5′-ACGGCCCCAACAGTGAAGTA-3′ (p#10) and 5′-ACGGTCTGACGACACGCAA-3′ (p#11) were designed for quantifying transcripts of tetR from pKU101 as internal control) on a MJ OPTICON® 2 using SYBR Green PCR Master Mix (Applied Biosystems). Data analyses for a relative quantification of gene expression were performed by the comparative Ct (threshold cycle) method according to the manufacturer's instructions. The parameter Ct is defined as the cycle number at which intensity of fluorescence (which is proportional to the quantity of DNA present in the tube during the exponential phase of the PCR) passes a fixed threshold value. The relative amount of target in two preparations is AE(Ct condition A internal control−Ct condition A sample)–(Ct condition B internal control−Ct condition B sample), where conditions A and B were in the presence/absence of exogenous cAMP (2 mM), respectively, sample is glnHp1 or glnHp1p2 product, internal control is tetR, AE = 1.8. Experiments were performed three times.

Determination of the modification status of GlnB

The modification status of GlnB was determined by non-denaturing polyacrylamide gel electrophoresis followed by immunodetection of GlnB using GlnB-specific antibodies as described before (26). The cell pellet was resuspended in 50 µl of ice-cold buffer (50 mM Tris-HCl pH7.4; 5 mM EDTA, 1 mM DTT and 1 mM benzamidin), added with glass beads (0.11 mm) to about the same volume as the cell pellet, and broken in a FastPrep (Thermo). Cell debris and glass beads were removed by centrifugation and the protein concentration in the supernatant was estimated as described before (27). A total of 2.5 µg protein was loaded per lane for non-denaturing PAGE. The proteins were blotted onto nitrocellulose membranes (Hybond-C Extra, Amersham) and the blots were probed with specific anti-GlnB (provided by K. Forchhammer), which were subsequently visualized using peroxidase-conjugated secondary antibodies (Promega) and DAB (Pierce) diluted 1:10 in stable peroxide substrate buffer (Pierce) as substrate.

RESULTS

CRP activates the transcription of glnHp1 in vitro through its synergistic binding with Eσ70

Previous studies suggested that glutamine uptake is increased by the presence of CRP and increased extracellular glutamine elicits a nitrogen-excess signal (16). The glnHPQ operon encodes the high-affinity glutamine uptake system, and is under the control of σ 70-dependent glnHp1 and σ54- and NtrC-P-dependent glnHp2 (12,13). Under in vivo conditions, glnHp2 is repressed by CRP (Li et al., unpublished data, see also Figure 1). However, in a macroarray assay, we have found that the overall transcription of glnHPQ is activated by the presence of exogenous cAMP (2 mM, data not shown). Therefore whether glnHp1 transcription and the overall transcription of the glnHPQ operon are activated by CRP in ARI-dependent manner was studied.

Figure 1.

Effect of CRP and its ARI mutant CRPH159L on the glnHp1p2 expression in the E. coli cya crp double mutant TP2339-1 detected by primer extension analysis. Lane 5 is G marker ladder. Cells were grown in M63 minimal medium with glycerol (0.4% w/v) as the sole carbon source, and glutamine (0.2% w/v) as nitrogen source, in the absence/presence of exogenous cAMP (2 mM).

Sequence analysis indicates that no obvious CRP-binding site can be found upstream of glnHp1, and DNA fragment containing glnHp1 cannot be retarded by CRP alone in gel shift assays (data not shown). In order to investigate the biochemical behaviour of CRP and Eσ70 on glnHp1, DNase I footprinting assays were carried out in vitro. The results show that in the absence of Eσ70, CRP cannot occupy the upstream regulatory sequences of glnHp1 even at high concentrations (lanes 1–5 of Figure 2A). Also, in the absence of CRP, only weak occupancy of Eσ70 at the core promoter of glnHp1 is evident (lanes 6 and 10 of Figure 2A; lanes 2 and 9 of Figure 2B). In contrast, when wild-type CRP is present together with Eσ70, the binding of Eσ70 is strongly increased (Figure 2A and B). Also, wild-type CRP is recruited by Eσ70 to a site centred at −61.5, organized as Class I σ70-promoters (28). Moreover, the above synergistic binding is abolished when wild-type CRP is replaced by its ARI mutant CRPH159L (Figure 2B).

Figure 2.

DNase I footprints of Eσ70 in the presence or absence of CRP and CRPH159L on glnHp1 (non template strand). (A) Titration with increasing concentrations of CRP (lanes 2 and 7, 30 nM; lanes 3 and 8, 100 nM; lanes 4 and 9, 300 nM) were performed in the absence (2–4) or presence of 50 nM Eσ70 (lanes 7–9). Lane 11 is A + G marker ladder. The protected regions were monitored by adding increasing concentrations of CRP in the presence or absence of Eσ70. (B) Titration with increasing concentrations of CRP (lane 3, 33 nM; lane 4, 100 nM; lane 5, 300 nM) and CRPH159L (lane 10, 33 nM; lane 11, 100 nM; lane 12, 300 nM) were performed in the presence of 25 nM Eσ70 (lanes 3–5 and 10–12). Lane 7 is A + G marker ladder. The limits of protected regions are indicated. Note that wild-type CRP is recruited to a site centred at −61.5 by Eσ70-RNA polymerase, which suggests a higher affinity for DNA binding of Eσ70 than that of CRP.

In order to investigate whether CRP can activate the transcription of glnHp1 in an ARI-dependent manner in vitro, run-off and multiple-round transcriptional assays were carried out with different concentrations of CRP or its ARI mutant CRPH159L. Results showed that, regardless of the concentration of Eσ70, wild-type CRP, but not its ARI mutant CRPH159L, can activate glnHp1 by a factor up to 6- and 14-fold in single- and multiple-round transcriptional assays, respectively. The activation factor increased when the concentration of CRP was increased (Figure 3).

Figure 3.

Effect of CRP and its ARI mutant CRPH159L on the glnHp1 expression in vitro. (A) multiple-round run-off assay; (B) single-round run-off assay.

Taken together, we have identified a naturally occurring case, where Eσ70 RNA polymerase binds very weakly to glnHp1. Binding of Eσ70 to glnHp1 recruits CRP to a degenerate site at −61.5 (‘ttgTATCCacatcaTCACAcaa’, which has a score of 1.44 bits, according to methods described before ((29), in which 17 bits was used as the cutoff score for strong CRP-binding sites; and 10 bits as the cutoff score for weak CRP-binding sites)), and the synergistic binding between CRP and Eσ70 clearly helps both partners in a reciprocal manner, which results in transcriptional activation of glnHp1.

The overall transcription of the glnHPQ operon is activated by CRP in vivo

To investigate the effect of CRP on the expression of glnHPQ in vivo, pKU550 carrying glnHp1p2 fused with the lacZ reporter gene was transformed into cya crp double mutant TP2339-1 containing either pLG339CRP or pLG339CRPH159L. The cells were grown in M63 medium in the absence or presence of exogenous cAMP. The results showed that in the presence of exogenous cAMP, glnHp1p2 can be activated 3-fold by wild-type CRP. This activation was abolished when ARI mutant CRPH159L replaced wild-type CRP (Table 2a).

Table 2.

Effect of CRP and its ARI mutant CRPH159L on the expression of glnHp1p2 and glnAp2 in the E. coli cya crp double mutant TP2339-1 and cya crp glnHPQ triple mutant BD4000

	Strain	CRP plasmid	Promoter (plasmid)	β-gal
				−cAMP	+cAMP
a	TP2339-1	pLG339ΔRS	glnHp1p2(pKU550)	6533 ± 350	6245 ± 347
	TP2339-1	pLG339CRP	glnHp1p2(pKU550)	6378 ± 287	20190 ± 1851
	TP2339-1	pLG339CRPH159L	glnHp1p2(pKU550)	6829 ± 393	6416 ± 334
b	TP2339-1	pLG339ΔRS	glnAp2(pKU101)	6650 ± 276	6188 ± 265
	TP2339-1	pLG339CRP	glnAp2(pKU101)	6588 ± 355	207 ± 43
	TP2339-1	pLG339CRPH159L	glnAp2(pKU101)	6423 ± 287	5582 ± 255
	BD4000	pLG339ΔRS	glnAp2(pKU101)	6478 ± 312	6033 ± 283
	BD4000	pLG339CRP	glnAp2(pKU101)	6431 ± 305	3509 ± 167
	BD4000	pLG339CRPH159L	glnAp2(pKU101)	6249 ± 203	5163 ± 179
	BD4000	pLG339ΔRS	glnAp2(pKU101H)	912 ± 124	712 ± 76
	BD4000	pLG339CRP	glnAp2(pKU101H)	896 ± 130	20 ± 17
	BD4000	pLG339CRPH159L	glnAp2(pKU101H)	869 ± 117	578 ± 50
c	TP2339-1	pLG339ΔRS	glnAp2(pKU101)	7106 ± 312	6779 ± 259
	TP2339-1	pLG339CRP	glnAp2(pKU101)	6907 ± 296	3542 ± 229
	TP2339-1	pLG339CRPH159L	glnAp2(pKU101)	6891 ± 355	6682 ± 203
	BD4000	pLG339ΔRS	glnAp2(pKU101)	6485 ± 278	6135 ± 264
	BD4000	pLG339CRP	glnAp2(pKU101)	6431 ± 257	3509 ± 305
	BD4000	pLG339CRPH159L	glnAp2(pKU101)	6249 ± 219	5163 ± 249
	Strain	CRP plasmid	Promoter	Relative expression
				− cAMP	+ cAMP
d	TP2339-1	pLG339ΔRS	glnHp1	1	1.8 ± 0.7
	TP2339-1	pLG339CRP	glnHp1	1	56.6 ± 2.1
	TP2339-1	pLG339CRPH159L	glnHp1	1	3.0 ± 1.5
	TP2339-1	pLG339ΔRS	glnHp1p2	1	1.0 ± 0.3
	TP2339-1	pLG339CRP	glnHp1p2	1	3.1 ± 0.5
	TP2339-1	pLG339CRPH159L	glnHp1p2	1	0.9 ± 0.2

All strains were grown in M63 minimal medium with glycerol (0.4% w/v) as the sole carbon source, and 0.2% w/v glutamine as nitrogen source (except c, 0.2 mM ammonium as nitrogen source), in the absence/presence of exogenous cAMP (2 mM). pLG339ΔRS is the negative control for CRP and its ARI mutant CRPH159L (see Table 1). a–c: Promoter expression is measured by β-galactosidase assays. d: level of the glnHp1 and glnHp1p2 expression from chromosome measured by real time RT-PCR in TP2339-1.

In order to verify the CRP effect on glnHp1 and glnHp2 respectively, two parallel experiments were carried out. First, in vivo primer extension assays were carried out on both promoters from plasmid pKU550 (Figure 1). Second, a more quantifiable real time RT-PCR was carried out to verify the CRP effect on both glnHp1 and glnHp1p2 transcribed directly from chromosome (Table 2d). In both cases, the results showed that wild-type CRP activates glnHp1, but represses glnHp2. In contrast, CRPH159L does not affect the expression of glnHp1 or glnHp2. Quantitatively, the results obtained from real time RT-PCR correlate well with the results obtained from β-galactosidase assays (compare Table 2d with 2a). Therefore, it seems that the CRP effect on glnHp1p2 is somewhat similar to that obtained with glnAp1p2, where CRP also activates σ70-dependent glnAp1, and represses σ54-dependent glnAp2 (9). However, they are different in at least two ways. First, the overall output is different. In this case, CRP causes a 3–4-fold decrease of overall expression for glnA (9), but a 3-fold increase for glnH (Table 2a and d). Second, it is known that a classical CRP-binding site is located upstream of the glnAp1 promoter (10,30), essential for CRP-dependent activation of glnAp1 (9). In contrast, no obvious CRP-binding site can be found upstream of glnHp1. CRP activates the transcription of glnHp1 in vitro through its synergistic binding with Eσ70 (Figure 2).

The CRP effect is maximized through glnHPQ-mediated glutamine uptake

The above results, together with the results obtained from Maheswaran and Forchhammer (16), suggest that CRP increases expression of the glnHPQ operon, which in turn, increases glutamine uptake. It has been proposed that, as a consequence, the intracellular glutamine level is increased, and so elicits a nitrogen-excess signal. In this case, the GlnB signalling process leads to the dephosphorylation of NtrC, and subsequent de-activation of NtrC-P-dependent promoters such as glnAp2 and glnHp2 (16,31).

To further investigate the above proposal, the glnHPQ operon was deleted from cya and crp double mutant TP2339-1 to give a triple mutant strain designated BD4000. The expression of glnAp2 (pKU101) was measured in the carbon poor medium in the presence of glutamine, in the presence/absence of cAMP. The results showed that the repression effect of wild-type CRP on the expression of glnAp2 is reduced from 30-fold to 2-fold upon deletion of glnHPQ. As control, the ARI mutant CRPH159L has no such repression effect (Table 2b). The glnHPQ mutant could still grow in the carbon-poor medium in the presence of glutamine, where glutamine uptake might be carried out by the low-affinity systems (32) (also see Figure 4).

Figure 4.

Effect of CRP and its ARI mutant CRPH159L on the uridylylation status of GlnB in the E. coli cya crp double mutant TP2339-1 and cya crp glnHPQ triple mutant BD4000. All strains (lanes 1–8) were grown in M63 minimal medium with glycerol (0.4% w/v) as the sole carbon source, and glutamine (0.2% w/v) as nitrogen source, in the absence/presence of exogenous cAMP (2 mM). TP2339-1 was grown in nitrogen-sufficient medium with 40 mM ammonium as control (lane 9).

To complement BD4000 for the glnHPQ operon, and minimize the artificial effects from genetic manipulation, a new glnAp2 reporter low copy number plasmid carrying the entire glnHPQ operon was constructed (designed as pKU101H), and used for the following assays. In this case, the CRP/cAMP effect on the expression of glnAp2 was investigated. Results indicated that when BD4000 was complemented by pKU101H, the CRP-mediated repression effect could be restored from 2-fold to 50-fold by wild-type CRP, while its ARI mutant CRPH159L had no such effect (Table 2b). We also noticed that such restoration could only be observed when cells grown in M63-minimum medium supplemented with glutamine, but not with ammonium, as the sole nitrogen source, indicating that the restoration effect requires physical transport of glutamine into the cells, rather than the sole enhancement of glnHPQ expression (Table 2c).

Maheswaran and Forchhammer (16) showed that deuridylylation of GlnB is a direct consequence of glutamine addition to cAMP-treated cells, which agrees with their proposal that cAMP increases the intracellular glutamine level when glutamine is used as the sole nitrogen source. In this study, similar assays were done with both TP2339-1 and BD4000 strains (Figure 4). The cells were grown in the presence of glutamine as the sole nitrogen source except the control (lane 9). As shown in the figure, in the absence of wild-type CRP (lanes 1, 3 and 4) or in the absence of GlnHPQ (lanes 5–8), GlnB is in its highly uridylylated state. By contrast, more than 80% of GlnB is in its deuridylylated state in the presence of both wild-type CRP and GlnHPQ (lane 2). As control, we showed that GlnB is mostly deuridylylated in the presence of 40 mM ammonium (lane 9). Taken together, the above results show that the GlnHPQ-mediated GlnB signalling is regulated by CRP. The intracellular concentration of glutamine is increased to a nitrogen-excess level by the GlnHPQ transport system in the presence of CRP and GlnB is deuridylylated in this case. As a consequence, NtrC is dephosphorylated and the expression of glnAp2 is decreased in the presence of CRP.

It is interesting to note that when the glnHPQ operon is deleted, the carbon source/cAMP effect remains, although it is reduced dramatically (Table 2b). This result indicates that there is another repression mechanism, which might not be related to the glutamine uptake. To further investigate this possibility, ammonium was used to replace the glutamine as the sole nitrogen source. Results showed that wild-type CRP, but not its ARI mutant CRPH159L, still represses glnAp2 by a factor of 2-fold in both TP2339-1 and BD4000 (Table 2c). This result supports the notion that alternative mechanism(s) may exist for CRP-mediated repression on glnAp2.

CRP repressed the expression of glnAp2 directly in vitro

To investigate whether CRP can repress glnAp2 directly, multiple-round transcriptional assays were carried out with different concentrations of CRP or its ARI mutant CRPH159L. Results showed that wild-type CRP can repress the transcriptional activity of glnAp2 by a factor of 2-fold. Moreover, when wild-type CRP was replaced by its ARI mutant CRPH159L, the repression effect was abolished (Figure 5). Also, this repression effect can be increased to 10-fold, when NtrCS160F (a constitutive active form of NtrC with a lower activity compared to NtrC-P) was used to replace wild-type NtrC-P (Figure 5), indicating that repression would be greater when a partially active form of NtrC exists in the cell (33).

Figure 5.

Effect of CRP and its ARI mutant CRPH159L on the glnAp2 expression in vitro. Wild-type CRP and NtrCS160F (▴) CRP and NtrC-P (▪), and CRPH159L and NtrC-P (♦), were used in the multiple-round transcriptional assays. The activity of glnAp2 with Eσ54+NtrC-P/NtrCS160F in the absence of CRP or CRPH159L was taken as 100%.

To investigate the mechanism of the ARI-dependent repression effect on glnAp2, DNase I footprinting assays were carried out in vitro. In the absence of Eσ54, CRP or its ARI mutant CRPH159L cannot occupy the upstream sequence of glnAp2 (lanes 9 and 10 of Figure 6). On the other hand, in the absence of CRP, full occupancy of Eσ54 on the core promoter of glnAp2 was observed (lane 13 of Figure 6). When CRP is present together with Eσ54, two additional protection regions were observed upstream of the binding site of Eσ54. In contrast, when ARI mutant CRPH159L was used in the assay, these additional protections were not evident (compare lanes 11 and 12 of Figure 6).

Figure 6.

DNase I footprints of Eσ54 in the presence or absence of CRP and CRPH159L on CC-50.5 (A CRP consensus-binding site was introduced into position −50.5 nucleotides upstream of the glnAp2) or wild-type glnAp2. CRP, CRPH159L and Eσ54, when added, were present at a final concentration of 50 nM. Lanes 1 and 8 are A+G marker ladders. The limits of protected regions are indicated. On glnAp2, only wild-type CRP can bind co-operatively with Eσ54 and the protection region is comparable to the one on CC-50.5. The locations of specific protected or hypersensitive bands are different on the two templates due to the difference of the sequence of CRP-binding sites. The hypersensitive band due to the CRP-binding on glnAp2 is marked by arrow. The protection of αCTD binding, which is co-operative only with wild-type CRP, is similar on both glnAp2 and CC-50.5 (lanes 5 and 12). These results strongly support that the relative orientation between CRP and Eσ54 is similar on glnAp2 and CC-50.5.

The two additional protection regions are from −42 to −61 and from −63 to −71. According to the location and size of the protected regions, it is possible that CRP binds and bends DNA from −42 to −61 and αCTD of Eσ54 binds DNA upstream. The binding and bending centre for CRP is at −51.5. Previous research by Huo et al. (22) had shown that when CRP was located at −50.5 on glnAp2, CRP and Eσ54 lie on the same face of DNA helix. Presumably the DNA bending induced by CRP, places NtrC-P far from the activator-accessible face of the closed complex and in turn represses the expression of glnAp2. The contact between αCTD and the upstream DNA region supports the view that DNA bending exists. Taken together, CRP was recruited by Eσ54 in an ARI-dependent manner and repressed NtrC-P mediated activation of glnAp2 directly.

DISCUSSION

In this study, we have shown that there are two novel mechanisms for CRP to mediate nitrogen regulation. In one mechanism CRP can affect the PII-mediated signal transduction pathway, hence causing deactivation of σ54-dependent gln promoters (for instance, glnAp2). Specifically, in vitro experiments show that glnHp1 promoter is a Class I-like CRP-dependent promoter. Through ARI–αCTD interaction, CRP and Eσ70 bind synergistically to glnHp1 to form the closed complex (Figure 2). The synergy between CRP and Eσ70 activates glnHp1 (Table 2d). Despite glnHp2 repression (Figure 1), the overall enhanced glnHPQ expression (∼3-fold, see Table 2a and d) leads to increased glutamine uptake, leading to PII-UMP deuridylylation (Figure 4) and shuts down the Ntr response in consequence. The other mechanism is that CRP can change the topological arrangement of Eσ54, NtrC-phosphate and glnAp2 promoter, thus directly repressing the glnAp2 promoter (Figure 5). Specifically, CRP can be recruited to the glnAp2 promoter at position −51.5 through ARI–αCTD interactions. This recruitment leads to DNA bending by CRP at −51.5, and hence inhibits the glnAp2 promoter activity (Figure 6). Taken together, our study reveals the mechanisms of CRP-mediated nitrogen regulation and presents two novel regulatory linkages between carbon metabolism and nitrogen assimilation in E. coli. The signal transduction pathway shown in our study illuminates the regulatory link between PII and carbon regulation (31). Although glutamine can be used as a sole carbon source supporting slow growth (34), it is surprising to find that the transport of glutamine is controlled by both global-carbon and global-nitrogen signal transduction pathways.

The dynamic range of expression of glnHPQ from the ‘off’ state to the ‘on’ state is greater when it has one σ54-dependent promoter and one σ70 promoter. The σ54-dependent promoter may be able to give higher levels of transcription than the σ70-dependent one could, but still is repressible by carbon status through action of CRP. In the presence of glutamine, the product of these two tandem promoters provides a novel feed-back to the glnAp2 expression, through PII-NtrBC in the genetic cascade controlling nitrogen assimilation, other than GlnK-AmtB (35). The complicated regulatory mechanisms described above are involved in a related physiological pathway. To manage the nitrogen deficiency, E. coli uses NtrC/Nac system to scavenge for nitrogen-containing compounds as a first line of defense against nitrogen starvation (36). To manage the carbon deficiency, E. coli uses CRP to activate various carbon source transporter systems (37–39) (You et al. unpublished data). Results obtained from this study indicate that CRP also accelerates transport of glutamine, even in the absence of glutamine in the medium, and represses the synthesis of glutamine to efficiently balance the carbon metabolism and nitrogen assimilation. When the relevant key signals in the promoter sequences are searched for in silico, the CRP inhibitory effect on glutamine biosynthesis appears to be also encountered in other enterobacteria (such as Klebsiella pneumoniae, Shigella flexneri, Enterobacter gergoviae, Erwinia carotovora and Salmonella enterica serovar Typhimurium), although the details of the regulation might differ. Interestingly in S. typhimurium, a bona fide CRP-binding site (‘caaTGTGAaagttgGCACAgat’, which has a score of 10.99 bits, according to methods described before (29)) is found centred at −27.5 upstream of glnAp2. This suggests that competition may occur between CRP and Eσ54 for binding to the promoter region. As there is no Nac protein found yet in S. typhimurium (40), it would be interesting to know if a tighter regulatory linkage exists.

As a global regulator, CRP is estimated to interact at ≈200 regulatory regions in E. coli. Zheng et al. (41) had revealed 152 hitherto unknown CRP regulons involved in energy production, amino acid metabolism, nucleotide metabolism and ion transport systems by run-off transcription/microarray analysis (ROMA). Moreover, Grainger et al. (42) had shown that CRP also interacts with thousands of weaker sites across the whole E. coli chromosome other than CRP-binding sites at known CRP regulated promoters by chromatin immunoprecipitation and high-density microarrays (ChIP-chip). We demonstrate that, physiologically, CRP tends to use weak affinity DNA-binding sites for transcriptional activation or repression by a regulated recruitment mechanism (43). It would be interesting to see if cross-talk between carbon metabolism and the other major areas of metabolism are mediated through mechanisms similar to those identified in our studies.